{
  "best_metric": 0.056979671120643616,
  "best_model_checkpoint": "autotrain-st-pair-class/checkpoint-339505",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 339505,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003681830900870385,
      "grad_norm": 0.8678207397460938,
      "learning_rate": 1.472710671261524e-08,
      "loss": 0.7265,
      "step": 25
    },
    {
      "epoch": 0.000736366180174077,
      "grad_norm": 1.7817554473876953,
      "learning_rate": 2.945421342523048e-08,
      "loss": 0.7232,
      "step": 50
    },
    {
      "epoch": 0.0011045492702611154,
      "grad_norm": 1.143256425857544,
      "learning_rate": 4.418132013784572e-08,
      "loss": 0.7232,
      "step": 75
    },
    {
      "epoch": 0.001472732360348154,
      "grad_norm": 1.4593238830566406,
      "learning_rate": 5.890842685046096e-08,
      "loss": 0.7206,
      "step": 100
    },
    {
      "epoch": 0.0018409154504351924,
      "grad_norm": 1.088563084602356,
      "learning_rate": 7.36355335630762e-08,
      "loss": 0.7212,
      "step": 125
    },
    {
      "epoch": 0.0022090985405222308,
      "grad_norm": 1.2012892961502075,
      "learning_rate": 8.836264027569144e-08,
      "loss": 0.7245,
      "step": 150
    },
    {
      "epoch": 0.0025772816306092695,
      "grad_norm": 0.7117802500724792,
      "learning_rate": 1.0308974698830668e-07,
      "loss": 0.7222,
      "step": 175
    },
    {
      "epoch": 0.002945464720696308,
      "grad_norm": 0.8568280935287476,
      "learning_rate": 1.1781685370092192e-07,
      "loss": 0.7207,
      "step": 200
    },
    {
      "epoch": 0.003313647810783346,
      "grad_norm": 0.6288392543792725,
      "learning_rate": 1.3254396041353717e-07,
      "loss": 0.7234,
      "step": 225
    },
    {
      "epoch": 0.003681830900870385,
      "grad_norm": 1.2614713907241821,
      "learning_rate": 1.472710671261524e-07,
      "loss": 0.717,
      "step": 250
    },
    {
      "epoch": 0.004050013990957424,
      "grad_norm": 0.634227454662323,
      "learning_rate": 1.6199817383876764e-07,
      "loss": 0.7217,
      "step": 275
    },
    {
      "epoch": 0.0044181970810444615,
      "grad_norm": 0.5872971415519714,
      "learning_rate": 1.7672528055138287e-07,
      "loss": 0.7196,
      "step": 300
    },
    {
      "epoch": 0.0047863801711315,
      "grad_norm": 0.9929088354110718,
      "learning_rate": 1.914523872639981e-07,
      "loss": 0.72,
      "step": 325
    },
    {
      "epoch": 0.005154563261218539,
      "grad_norm": 1.1321295499801636,
      "learning_rate": 2.0617949397661337e-07,
      "loss": 0.7167,
      "step": 350
    },
    {
      "epoch": 0.005522746351305577,
      "grad_norm": 0.7045577764511108,
      "learning_rate": 2.209066006892286e-07,
      "loss": 0.7169,
      "step": 375
    },
    {
      "epoch": 0.005890929441392616,
      "grad_norm": 0.9594325423240662,
      "learning_rate": 2.3563370740184384e-07,
      "loss": 0.7167,
      "step": 400
    },
    {
      "epoch": 0.006259112531479654,
      "grad_norm": 0.6934572458267212,
      "learning_rate": 2.503608141144591e-07,
      "loss": 0.7171,
      "step": 425
    },
    {
      "epoch": 0.006627295621566692,
      "grad_norm": 0.6293821334838867,
      "learning_rate": 2.6508792082707434e-07,
      "loss": 0.7148,
      "step": 450
    },
    {
      "epoch": 0.006995478711653731,
      "grad_norm": 0.725359320640564,
      "learning_rate": 2.7981502753968957e-07,
      "loss": 0.7142,
      "step": 475
    },
    {
      "epoch": 0.00736366180174077,
      "grad_norm": 0.6588715314865112,
      "learning_rate": 2.945421342523048e-07,
      "loss": 0.7168,
      "step": 500
    },
    {
      "epoch": 0.0077318448918278085,
      "grad_norm": 0.7691717743873596,
      "learning_rate": 3.0926924096492004e-07,
      "loss": 0.7133,
      "step": 525
    },
    {
      "epoch": 0.008100027981914847,
      "grad_norm": 0.7171897292137146,
      "learning_rate": 3.239963476775353e-07,
      "loss": 0.7119,
      "step": 550
    },
    {
      "epoch": 0.008468211072001885,
      "grad_norm": 0.6589468121528625,
      "learning_rate": 3.3872345439015056e-07,
      "loss": 0.7134,
      "step": 575
    },
    {
      "epoch": 0.008836394162088923,
      "grad_norm": 0.5842499136924744,
      "learning_rate": 3.5345056110276574e-07,
      "loss": 0.711,
      "step": 600
    },
    {
      "epoch": 0.009204577252175963,
      "grad_norm": 0.732860267162323,
      "learning_rate": 3.6817766781538103e-07,
      "loss": 0.7122,
      "step": 625
    },
    {
      "epoch": 0.009572760342263,
      "grad_norm": 0.9822880625724792,
      "learning_rate": 3.829047745279962e-07,
      "loss": 0.7116,
      "step": 650
    },
    {
      "epoch": 0.009940943432350038,
      "grad_norm": 0.7484005093574524,
      "learning_rate": 3.976318812406115e-07,
      "loss": 0.7116,
      "step": 675
    },
    {
      "epoch": 0.010309126522437078,
      "grad_norm": 1.1163777112960815,
      "learning_rate": 4.1235898795322674e-07,
      "loss": 0.7066,
      "step": 700
    },
    {
      "epoch": 0.010677309612524116,
      "grad_norm": 1.132763385772705,
      "learning_rate": 4.27086094665842e-07,
      "loss": 0.7095,
      "step": 725
    },
    {
      "epoch": 0.011045492702611154,
      "grad_norm": 1.3809318542480469,
      "learning_rate": 4.418132013784572e-07,
      "loss": 0.7047,
      "step": 750
    },
    {
      "epoch": 0.011413675792698193,
      "grad_norm": 0.5831483602523804,
      "learning_rate": 4.565403080910725e-07,
      "loss": 0.7103,
      "step": 775
    },
    {
      "epoch": 0.011781858882785231,
      "grad_norm": 0.5990871787071228,
      "learning_rate": 4.712674148036877e-07,
      "loss": 0.7076,
      "step": 800
    },
    {
      "epoch": 0.01215004197287227,
      "grad_norm": 0.9539962410926819,
      "learning_rate": 4.85994521516303e-07,
      "loss": 0.7065,
      "step": 825
    },
    {
      "epoch": 0.012518225062959309,
      "grad_norm": 1.093916893005371,
      "learning_rate": 5.007216282289182e-07,
      "loss": 0.7073,
      "step": 850
    },
    {
      "epoch": 0.012886408153046347,
      "grad_norm": 0.7179368734359741,
      "learning_rate": 5.154487349415335e-07,
      "loss": 0.7045,
      "step": 875
    },
    {
      "epoch": 0.013254591243133385,
      "grad_norm": 0.6126222610473633,
      "learning_rate": 5.301758416541487e-07,
      "loss": 0.7022,
      "step": 900
    },
    {
      "epoch": 0.013622774333220424,
      "grad_norm": 0.7177501916885376,
      "learning_rate": 5.44902948366764e-07,
      "loss": 0.7036,
      "step": 925
    },
    {
      "epoch": 0.013990957423307462,
      "grad_norm": 0.8863348960876465,
      "learning_rate": 5.596300550793791e-07,
      "loss": 0.6987,
      "step": 950
    },
    {
      "epoch": 0.014359140513394502,
      "grad_norm": 0.8881144523620605,
      "learning_rate": 5.743571617919944e-07,
      "loss": 0.701,
      "step": 975
    },
    {
      "epoch": 0.01472732360348154,
      "grad_norm": 1.149389386177063,
      "learning_rate": 5.890842685046096e-07,
      "loss": 0.7,
      "step": 1000
    },
    {
      "epoch": 0.015095506693568577,
      "grad_norm": 0.6700385212898254,
      "learning_rate": 6.038113752172248e-07,
      "loss": 0.6989,
      "step": 1025
    },
    {
      "epoch": 0.015463689783655617,
      "grad_norm": 0.7707194685935974,
      "learning_rate": 6.185384819298401e-07,
      "loss": 0.6975,
      "step": 1050
    },
    {
      "epoch": 0.015831872873742655,
      "grad_norm": 0.6434314846992493,
      "learning_rate": 6.332655886424554e-07,
      "loss": 0.6999,
      "step": 1075
    },
    {
      "epoch": 0.016200055963829695,
      "grad_norm": 0.575720489025116,
      "learning_rate": 6.479926953550705e-07,
      "loss": 0.6978,
      "step": 1100
    },
    {
      "epoch": 0.01656823905391673,
      "grad_norm": 1.1561897993087769,
      "learning_rate": 6.627198020676858e-07,
      "loss": 0.696,
      "step": 1125
    },
    {
      "epoch": 0.01693642214400377,
      "grad_norm": 1.0465784072875977,
      "learning_rate": 6.774469087803011e-07,
      "loss": 0.6939,
      "step": 1150
    },
    {
      "epoch": 0.01730460523409081,
      "grad_norm": 0.6008294224739075,
      "learning_rate": 6.921740154929163e-07,
      "loss": 0.6914,
      "step": 1175
    },
    {
      "epoch": 0.017672788324177846,
      "grad_norm": 1.0956335067749023,
      "learning_rate": 7.069011222055315e-07,
      "loss": 0.688,
      "step": 1200
    },
    {
      "epoch": 0.018040971414264886,
      "grad_norm": 0.6495553255081177,
      "learning_rate": 7.216282289181469e-07,
      "loss": 0.6899,
      "step": 1225
    },
    {
      "epoch": 0.018409154504351925,
      "grad_norm": 0.5202560424804688,
      "learning_rate": 7.363553356307621e-07,
      "loss": 0.6894,
      "step": 1250
    },
    {
      "epoch": 0.01877733759443896,
      "grad_norm": 0.7886903882026672,
      "learning_rate": 7.510824423433772e-07,
      "loss": 0.6863,
      "step": 1275
    },
    {
      "epoch": 0.019145520684526,
      "grad_norm": 0.6157426834106445,
      "learning_rate": 7.658095490559924e-07,
      "loss": 0.6868,
      "step": 1300
    },
    {
      "epoch": 0.01951370377461304,
      "grad_norm": 0.8816072344779968,
      "learning_rate": 7.805366557686078e-07,
      "loss": 0.6886,
      "step": 1325
    },
    {
      "epoch": 0.019881886864700077,
      "grad_norm": 0.7643646597862244,
      "learning_rate": 7.95263762481223e-07,
      "loss": 0.684,
      "step": 1350
    },
    {
      "epoch": 0.020250069954787116,
      "grad_norm": 0.7116807103157043,
      "learning_rate": 8.099908691938382e-07,
      "loss": 0.6852,
      "step": 1375
    },
    {
      "epoch": 0.020618253044874156,
      "grad_norm": 0.8170478343963623,
      "learning_rate": 8.247179759064535e-07,
      "loss": 0.6824,
      "step": 1400
    },
    {
      "epoch": 0.020986436134961192,
      "grad_norm": 0.7388837933540344,
      "learning_rate": 8.394450826190688e-07,
      "loss": 0.6823,
      "step": 1425
    },
    {
      "epoch": 0.021354619225048232,
      "grad_norm": 0.5734406113624573,
      "learning_rate": 8.54172189331684e-07,
      "loss": 0.6806,
      "step": 1450
    },
    {
      "epoch": 0.02172280231513527,
      "grad_norm": 0.7802960276603699,
      "learning_rate": 8.688992960442992e-07,
      "loss": 0.6791,
      "step": 1475
    },
    {
      "epoch": 0.022090985405222308,
      "grad_norm": 0.6324337720870972,
      "learning_rate": 8.836264027569144e-07,
      "loss": 0.675,
      "step": 1500
    },
    {
      "epoch": 0.022459168495309347,
      "grad_norm": 0.8283900618553162,
      "learning_rate": 8.983535094695296e-07,
      "loss": 0.6787,
      "step": 1525
    },
    {
      "epoch": 0.022827351585396387,
      "grad_norm": 1.1485595703125,
      "learning_rate": 9.13080616182145e-07,
      "loss": 0.677,
      "step": 1550
    },
    {
      "epoch": 0.023195534675483423,
      "grad_norm": 0.918366551399231,
      "learning_rate": 9.278077228947602e-07,
      "loss": 0.6717,
      "step": 1575
    },
    {
      "epoch": 0.023563717765570463,
      "grad_norm": 0.6480444073677063,
      "learning_rate": 9.425348296073754e-07,
      "loss": 0.6697,
      "step": 1600
    },
    {
      "epoch": 0.023931900855657502,
      "grad_norm": 0.6028603315353394,
      "learning_rate": 9.572619363199905e-07,
      "loss": 0.6708,
      "step": 1625
    },
    {
      "epoch": 0.02430008394574454,
      "grad_norm": 1.0693798065185547,
      "learning_rate": 9.71989043032606e-07,
      "loss": 0.6684,
      "step": 1650
    },
    {
      "epoch": 0.024668267035831578,
      "grad_norm": 0.8398106694221497,
      "learning_rate": 9.867161497452211e-07,
      "loss": 0.6721,
      "step": 1675
    },
    {
      "epoch": 0.025036450125918618,
      "grad_norm": 0.6863375902175903,
      "learning_rate": 1.0014432564578364e-06,
      "loss": 0.6653,
      "step": 1700
    },
    {
      "epoch": 0.025404633216005654,
      "grad_norm": 0.9351726174354553,
      "learning_rate": 1.0161703631704515e-06,
      "loss": 0.6677,
      "step": 1725
    },
    {
      "epoch": 0.025772816306092693,
      "grad_norm": 0.6590425372123718,
      "learning_rate": 1.030897469883067e-06,
      "loss": 0.6611,
      "step": 1750
    },
    {
      "epoch": 0.026140999396179733,
      "grad_norm": 0.5982798933982849,
      "learning_rate": 1.045624576595682e-06,
      "loss": 0.6593,
      "step": 1775
    },
    {
      "epoch": 0.02650918248626677,
      "grad_norm": 1.231074333190918,
      "learning_rate": 1.0603516833082973e-06,
      "loss": 0.6609,
      "step": 1800
    },
    {
      "epoch": 0.02687736557635381,
      "grad_norm": 0.562767744064331,
      "learning_rate": 1.0750787900209126e-06,
      "loss": 0.6549,
      "step": 1825
    },
    {
      "epoch": 0.02724554866644085,
      "grad_norm": 0.5797823667526245,
      "learning_rate": 1.089805896733528e-06,
      "loss": 0.6584,
      "step": 1850
    },
    {
      "epoch": 0.027613731756527884,
      "grad_norm": 0.6395525932312012,
      "learning_rate": 1.104533003446143e-06,
      "loss": 0.6496,
      "step": 1875
    },
    {
      "epoch": 0.027981914846614924,
      "grad_norm": 0.7141817808151245,
      "learning_rate": 1.1192601101587583e-06,
      "loss": 0.654,
      "step": 1900
    },
    {
      "epoch": 0.028350097936701964,
      "grad_norm": 0.5764775276184082,
      "learning_rate": 1.1339872168713736e-06,
      "loss": 0.6527,
      "step": 1925
    },
    {
      "epoch": 0.028718281026789003,
      "grad_norm": 0.5608025193214417,
      "learning_rate": 1.1487143235839889e-06,
      "loss": 0.6462,
      "step": 1950
    },
    {
      "epoch": 0.02908646411687604,
      "grad_norm": 1.2220577001571655,
      "learning_rate": 1.1634414302966041e-06,
      "loss": 0.6511,
      "step": 1975
    },
    {
      "epoch": 0.02945464720696308,
      "grad_norm": 0.5681129097938538,
      "learning_rate": 1.1781685370092192e-06,
      "loss": 0.6443,
      "step": 2000
    },
    {
      "epoch": 0.02982283029705012,
      "grad_norm": 1.0518550872802734,
      "learning_rate": 1.1928956437218345e-06,
      "loss": 0.644,
      "step": 2025
    },
    {
      "epoch": 0.030191013387137155,
      "grad_norm": 0.5548666715621948,
      "learning_rate": 1.2076227504344496e-06,
      "loss": 0.6385,
      "step": 2050
    },
    {
      "epoch": 0.030559196477224194,
      "grad_norm": 0.7215397953987122,
      "learning_rate": 1.222349857147065e-06,
      "loss": 0.6379,
      "step": 2075
    },
    {
      "epoch": 0.030927379567311234,
      "grad_norm": 0.5453612208366394,
      "learning_rate": 1.2370769638596802e-06,
      "loss": 0.6319,
      "step": 2100
    },
    {
      "epoch": 0.031295562657398274,
      "grad_norm": 0.8071221113204956,
      "learning_rate": 1.2518040705722957e-06,
      "loss": 0.631,
      "step": 2125
    },
    {
      "epoch": 0.03166374574748531,
      "grad_norm": 0.5668207406997681,
      "learning_rate": 1.2665311772849107e-06,
      "loss": 0.6291,
      "step": 2150
    },
    {
      "epoch": 0.032031928837572346,
      "grad_norm": 1.0014539957046509,
      "learning_rate": 1.281258283997526e-06,
      "loss": 0.6269,
      "step": 2175
    },
    {
      "epoch": 0.03240011192765939,
      "grad_norm": 0.7369861006736755,
      "learning_rate": 1.295985390710141e-06,
      "loss": 0.6242,
      "step": 2200
    },
    {
      "epoch": 0.032768295017746425,
      "grad_norm": 0.6795928478240967,
      "learning_rate": 1.3107124974227564e-06,
      "loss": 0.6245,
      "step": 2225
    },
    {
      "epoch": 0.03313647810783346,
      "grad_norm": 0.648115873336792,
      "learning_rate": 1.3254396041353717e-06,
      "loss": 0.6223,
      "step": 2250
    },
    {
      "epoch": 0.033504661197920504,
      "grad_norm": 0.5944338440895081,
      "learning_rate": 1.3401667108479868e-06,
      "loss": 0.6179,
      "step": 2275
    },
    {
      "epoch": 0.03387284428800754,
      "grad_norm": 0.6033785343170166,
      "learning_rate": 1.3548938175606023e-06,
      "loss": 0.6145,
      "step": 2300
    },
    {
      "epoch": 0.03424102737809458,
      "grad_norm": 0.8906118273735046,
      "learning_rate": 1.3696209242732175e-06,
      "loss": 0.6069,
      "step": 2325
    },
    {
      "epoch": 0.03460921046818162,
      "grad_norm": 0.6462284326553345,
      "learning_rate": 1.3843480309858326e-06,
      "loss": 0.6062,
      "step": 2350
    },
    {
      "epoch": 0.034977393558268656,
      "grad_norm": 0.5796801447868347,
      "learning_rate": 1.399075137698448e-06,
      "loss": 0.6062,
      "step": 2375
    },
    {
      "epoch": 0.03534557664835569,
      "grad_norm": 1.3210780620574951,
      "learning_rate": 1.413802244411063e-06,
      "loss": 0.6066,
      "step": 2400
    },
    {
      "epoch": 0.035713759738442735,
      "grad_norm": 0.9987689256668091,
      "learning_rate": 1.4285293511236783e-06,
      "loss": 0.6045,
      "step": 2425
    },
    {
      "epoch": 0.03608194282852977,
      "grad_norm": 0.5557406544685364,
      "learning_rate": 1.4432564578362938e-06,
      "loss": 0.6007,
      "step": 2450
    },
    {
      "epoch": 0.03645012591861681,
      "grad_norm": 0.7247636914253235,
      "learning_rate": 1.4579835645489088e-06,
      "loss": 0.5916,
      "step": 2475
    },
    {
      "epoch": 0.03681830900870385,
      "grad_norm": 0.8754281997680664,
      "learning_rate": 1.4727106712615241e-06,
      "loss": 0.594,
      "step": 2500
    },
    {
      "epoch": 0.03718649209879089,
      "grad_norm": 0.6612271070480347,
      "learning_rate": 1.4874377779741392e-06,
      "loss": 0.5903,
      "step": 2525
    },
    {
      "epoch": 0.03755467518887792,
      "grad_norm": 0.6568219661712646,
      "learning_rate": 1.5021648846867545e-06,
      "loss": 0.584,
      "step": 2550
    },
    {
      "epoch": 0.037922858278964966,
      "grad_norm": 0.7208539843559265,
      "learning_rate": 1.5168919913993698e-06,
      "loss": 0.5881,
      "step": 2575
    },
    {
      "epoch": 0.038291041369052,
      "grad_norm": 0.6969273090362549,
      "learning_rate": 1.5316190981119849e-06,
      "loss": 0.5826,
      "step": 2600
    },
    {
      "epoch": 0.03865922445913904,
      "grad_norm": 0.7232186198234558,
      "learning_rate": 1.5463462048246004e-06,
      "loss": 0.5789,
      "step": 2625
    },
    {
      "epoch": 0.03902740754922608,
      "grad_norm": 0.5670535564422607,
      "learning_rate": 1.5610733115372156e-06,
      "loss": 0.5786,
      "step": 2650
    },
    {
      "epoch": 0.03939559063931312,
      "grad_norm": 0.5489992499351501,
      "learning_rate": 1.5758004182498307e-06,
      "loss": 0.5755,
      "step": 2675
    },
    {
      "epoch": 0.039763773729400154,
      "grad_norm": 0.6829290390014648,
      "learning_rate": 1.590527524962446e-06,
      "loss": 0.5749,
      "step": 2700
    },
    {
      "epoch": 0.0401319568194872,
      "grad_norm": 0.5697892308235168,
      "learning_rate": 1.605254631675061e-06,
      "loss": 0.5666,
      "step": 2725
    },
    {
      "epoch": 0.04050013990957423,
      "grad_norm": 0.5284579992294312,
      "learning_rate": 1.6199817383876764e-06,
      "loss": 0.558,
      "step": 2750
    },
    {
      "epoch": 0.04086832299966127,
      "grad_norm": 0.6674126386642456,
      "learning_rate": 1.6347088451002919e-06,
      "loss": 0.5617,
      "step": 2775
    },
    {
      "epoch": 0.04123650608974831,
      "grad_norm": 0.6509599089622498,
      "learning_rate": 1.649435951812907e-06,
      "loss": 0.5595,
      "step": 2800
    },
    {
      "epoch": 0.04160468917983535,
      "grad_norm": 0.5450132489204407,
      "learning_rate": 1.6641630585255222e-06,
      "loss": 0.5586,
      "step": 2825
    },
    {
      "epoch": 0.041972872269922384,
      "grad_norm": 0.5366148352622986,
      "learning_rate": 1.6788901652381375e-06,
      "loss": 0.5531,
      "step": 2850
    },
    {
      "epoch": 0.04234105536000943,
      "grad_norm": 0.5460155010223389,
      "learning_rate": 1.6936172719507526e-06,
      "loss": 0.5549,
      "step": 2875
    },
    {
      "epoch": 0.042709238450096464,
      "grad_norm": 0.5339788794517517,
      "learning_rate": 1.708344378663368e-06,
      "loss": 0.5486,
      "step": 2900
    },
    {
      "epoch": 0.0430774215401835,
      "grad_norm": 0.5116156339645386,
      "learning_rate": 1.723071485375983e-06,
      "loss": 0.5528,
      "step": 2925
    },
    {
      "epoch": 0.04344560463027054,
      "grad_norm": 1.195949673652649,
      "learning_rate": 1.7377985920885985e-06,
      "loss": 0.5507,
      "step": 2950
    },
    {
      "epoch": 0.04381378772035758,
      "grad_norm": 0.5309670567512512,
      "learning_rate": 1.7525256988012138e-06,
      "loss": 0.5407,
      "step": 2975
    },
    {
      "epoch": 0.044181970810444615,
      "grad_norm": 0.7596344351768494,
      "learning_rate": 1.7672528055138288e-06,
      "loss": 0.5418,
      "step": 3000
    },
    {
      "epoch": 0.04455015390053166,
      "grad_norm": 0.47329920530319214,
      "learning_rate": 1.7819799122264441e-06,
      "loss": 0.5357,
      "step": 3025
    },
    {
      "epoch": 0.044918336990618694,
      "grad_norm": 0.5916969776153564,
      "learning_rate": 1.7967070189390592e-06,
      "loss": 0.5359,
      "step": 3050
    },
    {
      "epoch": 0.04528652008070573,
      "grad_norm": 0.6137320399284363,
      "learning_rate": 1.8114341256516747e-06,
      "loss": 0.529,
      "step": 3075
    },
    {
      "epoch": 0.045654703170792774,
      "grad_norm": 0.6124922633171082,
      "learning_rate": 1.82616123236429e-06,
      "loss": 0.5292,
      "step": 3100
    },
    {
      "epoch": 0.04602288626087981,
      "grad_norm": 0.4892748296260834,
      "learning_rate": 1.840888339076905e-06,
      "loss": 0.5356,
      "step": 3125
    },
    {
      "epoch": 0.046391069350966846,
      "grad_norm": 0.6712266206741333,
      "learning_rate": 1.8556154457895203e-06,
      "loss": 0.531,
      "step": 3150
    },
    {
      "epoch": 0.04675925244105389,
      "grad_norm": 0.5134905576705933,
      "learning_rate": 1.8703425525021356e-06,
      "loss": 0.533,
      "step": 3175
    },
    {
      "epoch": 0.047127435531140925,
      "grad_norm": 0.5121038556098938,
      "learning_rate": 1.8850696592147507e-06,
      "loss": 0.5311,
      "step": 3200
    },
    {
      "epoch": 0.04749561862122796,
      "grad_norm": 1.0264848470687866,
      "learning_rate": 1.8997967659273662e-06,
      "loss": 0.5273,
      "step": 3225
    },
    {
      "epoch": 0.047863801711315004,
      "grad_norm": 0.8714808225631714,
      "learning_rate": 1.914523872639981e-06,
      "loss": 0.5226,
      "step": 3250
    },
    {
      "epoch": 0.04823198480140204,
      "grad_norm": 0.7779112458229065,
      "learning_rate": 1.9292509793525966e-06,
      "loss": 0.5259,
      "step": 3275
    },
    {
      "epoch": 0.04860016789148908,
      "grad_norm": 0.5716907978057861,
      "learning_rate": 1.943978086065212e-06,
      "loss": 0.5243,
      "step": 3300
    },
    {
      "epoch": 0.04896835098157612,
      "grad_norm": 0.478712797164917,
      "learning_rate": 1.958705192777827e-06,
      "loss": 0.5235,
      "step": 3325
    },
    {
      "epoch": 0.049336534071663156,
      "grad_norm": 0.43825459480285645,
      "learning_rate": 1.9734322994904422e-06,
      "loss": 0.5224,
      "step": 3350
    },
    {
      "epoch": 0.04970471716175019,
      "grad_norm": 0.5753782987594604,
      "learning_rate": 1.9881594062030577e-06,
      "loss": 0.5248,
      "step": 3375
    },
    {
      "epoch": 0.050072900251837235,
      "grad_norm": 0.6924416422843933,
      "learning_rate": 2.002886512915673e-06,
      "loss": 0.5166,
      "step": 3400
    },
    {
      "epoch": 0.05044108334192427,
      "grad_norm": 0.48913905024528503,
      "learning_rate": 2.017613619628288e-06,
      "loss": 0.515,
      "step": 3425
    },
    {
      "epoch": 0.05080926643201131,
      "grad_norm": 0.8264048099517822,
      "learning_rate": 2.032340726340903e-06,
      "loss": 0.5145,
      "step": 3450
    },
    {
      "epoch": 0.05117744952209835,
      "grad_norm": 0.6421396732330322,
      "learning_rate": 2.0470678330535185e-06,
      "loss": 0.5093,
      "step": 3475
    },
    {
      "epoch": 0.05154563261218539,
      "grad_norm": 0.5583409667015076,
      "learning_rate": 2.061794939766134e-06,
      "loss": 0.5123,
      "step": 3500
    },
    {
      "epoch": 0.05191381570227242,
      "grad_norm": 0.5377383232116699,
      "learning_rate": 2.076522046478749e-06,
      "loss": 0.5169,
      "step": 3525
    },
    {
      "epoch": 0.052281998792359466,
      "grad_norm": 0.5501702427864075,
      "learning_rate": 2.091249153191364e-06,
      "loss": 0.509,
      "step": 3550
    },
    {
      "epoch": 0.0526501818824465,
      "grad_norm": 0.46207118034362793,
      "learning_rate": 2.105976259903979e-06,
      "loss": 0.5133,
      "step": 3575
    },
    {
      "epoch": 0.05301836497253354,
      "grad_norm": 0.5295949578285217,
      "learning_rate": 2.1207033666165947e-06,
      "loss": 0.5038,
      "step": 3600
    },
    {
      "epoch": 0.05338654806262058,
      "grad_norm": 0.526362419128418,
      "learning_rate": 2.13543047332921e-06,
      "loss": 0.5199,
      "step": 3625
    },
    {
      "epoch": 0.05375473115270762,
      "grad_norm": 0.5102420449256897,
      "learning_rate": 2.1501575800418253e-06,
      "loss": 0.5078,
      "step": 3650
    },
    {
      "epoch": 0.054122914242794654,
      "grad_norm": 0.5729849934577942,
      "learning_rate": 2.1648846867544403e-06,
      "loss": 0.5052,
      "step": 3675
    },
    {
      "epoch": 0.0544910973328817,
      "grad_norm": 0.6406521797180176,
      "learning_rate": 2.179611793467056e-06,
      "loss": 0.5071,
      "step": 3700
    },
    {
      "epoch": 0.05485928042296873,
      "grad_norm": 0.6915680766105652,
      "learning_rate": 2.194338900179671e-06,
      "loss": 0.5058,
      "step": 3725
    },
    {
      "epoch": 0.05522746351305577,
      "grad_norm": 0.8066368103027344,
      "learning_rate": 2.209066006892286e-06,
      "loss": 0.5045,
      "step": 3750
    },
    {
      "epoch": 0.05559564660314281,
      "grad_norm": 0.739680290222168,
      "learning_rate": 2.223793113604901e-06,
      "loss": 0.4987,
      "step": 3775
    },
    {
      "epoch": 0.05596382969322985,
      "grad_norm": 0.6398206949234009,
      "learning_rate": 2.2385202203175166e-06,
      "loss": 0.504,
      "step": 3800
    },
    {
      "epoch": 0.05633201278331689,
      "grad_norm": 0.640600860118866,
      "learning_rate": 2.253247327030132e-06,
      "loss": 0.505,
      "step": 3825
    },
    {
      "epoch": 0.05670019587340393,
      "grad_norm": 0.49086707830429077,
      "learning_rate": 2.267974433742747e-06,
      "loss": 0.5073,
      "step": 3850
    },
    {
      "epoch": 0.057068378963490964,
      "grad_norm": 0.6551923155784607,
      "learning_rate": 2.2827015404553622e-06,
      "loss": 0.5027,
      "step": 3875
    },
    {
      "epoch": 0.05743656205357801,
      "grad_norm": 0.5222965478897095,
      "learning_rate": 2.2974286471679777e-06,
      "loss": 0.4987,
      "step": 3900
    },
    {
      "epoch": 0.05780474514366504,
      "grad_norm": 0.4598608911037445,
      "learning_rate": 2.312155753880593e-06,
      "loss": 0.5047,
      "step": 3925
    },
    {
      "epoch": 0.05817292823375208,
      "grad_norm": 0.46309804916381836,
      "learning_rate": 2.3268828605932083e-06,
      "loss": 0.5031,
      "step": 3950
    },
    {
      "epoch": 0.05854111132383912,
      "grad_norm": 0.49093905091285706,
      "learning_rate": 2.3416099673058234e-06,
      "loss": 0.496,
      "step": 3975
    },
    {
      "epoch": 0.05890929441392616,
      "grad_norm": 0.49821043014526367,
      "learning_rate": 2.3563370740184384e-06,
      "loss": 0.5071,
      "step": 4000
    },
    {
      "epoch": 0.059277477504013194,
      "grad_norm": 0.5469359755516052,
      "learning_rate": 2.371064180731054e-06,
      "loss": 0.502,
      "step": 4025
    },
    {
      "epoch": 0.05964566059410024,
      "grad_norm": 0.493945449590683,
      "learning_rate": 2.385791287443669e-06,
      "loss": 0.4993,
      "step": 4050
    },
    {
      "epoch": 0.060013843684187274,
      "grad_norm": 0.6475746631622314,
      "learning_rate": 2.400518394156284e-06,
      "loss": 0.4977,
      "step": 4075
    },
    {
      "epoch": 0.06038202677427431,
      "grad_norm": 0.9257048964500427,
      "learning_rate": 2.415245500868899e-06,
      "loss": 0.5011,
      "step": 4100
    },
    {
      "epoch": 0.06075020986436135,
      "grad_norm": 0.687039852142334,
      "learning_rate": 2.4299726075815147e-06,
      "loss": 0.5029,
      "step": 4125
    },
    {
      "epoch": 0.06111839295444839,
      "grad_norm": 0.48104432225227356,
      "learning_rate": 2.44469971429413e-06,
      "loss": 0.505,
      "step": 4150
    },
    {
      "epoch": 0.061486576044535425,
      "grad_norm": 0.5689713954925537,
      "learning_rate": 2.4594268210067452e-06,
      "loss": 0.498,
      "step": 4175
    },
    {
      "epoch": 0.06185475913462247,
      "grad_norm": 0.5599949955940247,
      "learning_rate": 2.4741539277193603e-06,
      "loss": 0.4928,
      "step": 4200
    },
    {
      "epoch": 0.062222942224709504,
      "grad_norm": 0.5400413870811462,
      "learning_rate": 2.488881034431976e-06,
      "loss": 0.4963,
      "step": 4225
    },
    {
      "epoch": 0.06259112531479655,
      "grad_norm": 0.5699523687362671,
      "learning_rate": 2.5036081411445913e-06,
      "loss": 0.4955,
      "step": 4250
    },
    {
      "epoch": 0.06295930840488358,
      "grad_norm": 0.7645552158355713,
      "learning_rate": 2.518335247857206e-06,
      "loss": 0.5051,
      "step": 4275
    },
    {
      "epoch": 0.06332749149497062,
      "grad_norm": 1.205398440361023,
      "learning_rate": 2.5330623545698215e-06,
      "loss": 0.4913,
      "step": 4300
    },
    {
      "epoch": 0.06369567458505766,
      "grad_norm": 0.5377015471458435,
      "learning_rate": 2.5477894612824365e-06,
      "loss": 0.5012,
      "step": 4325
    },
    {
      "epoch": 0.06406385767514469,
      "grad_norm": 0.541007399559021,
      "learning_rate": 2.562516567995052e-06,
      "loss": 0.4927,
      "step": 4350
    },
    {
      "epoch": 0.06443204076523174,
      "grad_norm": 0.7172959446907043,
      "learning_rate": 2.5772436747076675e-06,
      "loss": 0.4879,
      "step": 4375
    },
    {
      "epoch": 0.06480022385531878,
      "grad_norm": 0.5034739971160889,
      "learning_rate": 2.591970781420282e-06,
      "loss": 0.4913,
      "step": 4400
    },
    {
      "epoch": 0.06516840694540581,
      "grad_norm": 0.4795535206794739,
      "learning_rate": 2.6066978881328973e-06,
      "loss": 0.4928,
      "step": 4425
    },
    {
      "epoch": 0.06553659003549285,
      "grad_norm": 0.4326831102371216,
      "learning_rate": 2.6214249948455128e-06,
      "loss": 0.484,
      "step": 4450
    },
    {
      "epoch": 0.0659047731255799,
      "grad_norm": 0.472896546125412,
      "learning_rate": 2.6361521015581283e-06,
      "loss": 0.4953,
      "step": 4475
    },
    {
      "epoch": 0.06627295621566692,
      "grad_norm": 0.5761368870735168,
      "learning_rate": 2.6508792082707434e-06,
      "loss": 0.4945,
      "step": 4500
    },
    {
      "epoch": 0.06664113930575397,
      "grad_norm": 0.6462481021881104,
      "learning_rate": 2.6656063149833584e-06,
      "loss": 0.4884,
      "step": 4525
    },
    {
      "epoch": 0.06700932239584101,
      "grad_norm": 0.6638084650039673,
      "learning_rate": 2.6803334216959735e-06,
      "loss": 0.4921,
      "step": 4550
    },
    {
      "epoch": 0.06737750548592804,
      "grad_norm": 0.49843668937683105,
      "learning_rate": 2.695060528408589e-06,
      "loss": 0.4914,
      "step": 4575
    },
    {
      "epoch": 0.06774568857601508,
      "grad_norm": 0.5445410013198853,
      "learning_rate": 2.7097876351212045e-06,
      "loss": 0.485,
      "step": 4600
    },
    {
      "epoch": 0.06811387166610212,
      "grad_norm": 0.8247371912002563,
      "learning_rate": 2.7245147418338196e-06,
      "loss": 0.4984,
      "step": 4625
    },
    {
      "epoch": 0.06848205475618915,
      "grad_norm": 0.4787178039550781,
      "learning_rate": 2.739241848546435e-06,
      "loss": 0.4832,
      "step": 4650
    },
    {
      "epoch": 0.0688502378462762,
      "grad_norm": 0.6253044605255127,
      "learning_rate": 2.7539689552590497e-06,
      "loss": 0.4852,
      "step": 4675
    },
    {
      "epoch": 0.06921842093636324,
      "grad_norm": 0.7872785329818726,
      "learning_rate": 2.7686960619716652e-06,
      "loss": 0.5008,
      "step": 4700
    },
    {
      "epoch": 0.06958660402645027,
      "grad_norm": 0.5098380446434021,
      "learning_rate": 2.7834231686842807e-06,
      "loss": 0.4815,
      "step": 4725
    },
    {
      "epoch": 0.06995478711653731,
      "grad_norm": 0.5085175633430481,
      "learning_rate": 2.798150275396896e-06,
      "loss": 0.4871,
      "step": 4750
    },
    {
      "epoch": 0.07032297020662436,
      "grad_norm": 0.48235610127449036,
      "learning_rate": 2.8128773821095113e-06,
      "loss": 0.4981,
      "step": 4775
    },
    {
      "epoch": 0.07069115329671138,
      "grad_norm": 0.5846980214118958,
      "learning_rate": 2.827604488822126e-06,
      "loss": 0.4837,
      "step": 4800
    },
    {
      "epoch": 0.07105933638679843,
      "grad_norm": 0.6444407105445862,
      "learning_rate": 2.8423315955347415e-06,
      "loss": 0.4879,
      "step": 4825
    },
    {
      "epoch": 0.07142751947688547,
      "grad_norm": 0.47602659463882446,
      "learning_rate": 2.8570587022473565e-06,
      "loss": 0.4861,
      "step": 4850
    },
    {
      "epoch": 0.0717957025669725,
      "grad_norm": 0.5946886539459229,
      "learning_rate": 2.871785808959972e-06,
      "loss": 0.4788,
      "step": 4875
    },
    {
      "epoch": 0.07216388565705954,
      "grad_norm": 0.9483232498168945,
      "learning_rate": 2.8865129156725875e-06,
      "loss": 0.4787,
      "step": 4900
    },
    {
      "epoch": 0.07253206874714659,
      "grad_norm": 0.7961587309837341,
      "learning_rate": 2.901240022385202e-06,
      "loss": 0.4865,
      "step": 4925
    },
    {
      "epoch": 0.07290025183723362,
      "grad_norm": 0.7431612610816956,
      "learning_rate": 2.9159671290978177e-06,
      "loss": 0.488,
      "step": 4950
    },
    {
      "epoch": 0.07326843492732066,
      "grad_norm": 0.6214255094528198,
      "learning_rate": 2.9306942358104328e-06,
      "loss": 0.4821,
      "step": 4975
    },
    {
      "epoch": 0.0736366180174077,
      "grad_norm": 0.6615573763847351,
      "learning_rate": 2.9454213425230483e-06,
      "loss": 0.4805,
      "step": 5000
    },
    {
      "epoch": 0.07400480110749473,
      "grad_norm": 0.6024646759033203,
      "learning_rate": 2.9601484492356638e-06,
      "loss": 0.4809,
      "step": 5025
    },
    {
      "epoch": 0.07437298419758177,
      "grad_norm": 0.5407711267471313,
      "learning_rate": 2.9748755559482784e-06,
      "loss": 0.4758,
      "step": 5050
    },
    {
      "epoch": 0.07474116728766882,
      "grad_norm": 0.6843119263648987,
      "learning_rate": 2.989602662660894e-06,
      "loss": 0.4798,
      "step": 5075
    },
    {
      "epoch": 0.07510935037775585,
      "grad_norm": 0.5549395084381104,
      "learning_rate": 3.004329769373509e-06,
      "loss": 0.4789,
      "step": 5100
    },
    {
      "epoch": 0.07547753346784289,
      "grad_norm": 0.818297803401947,
      "learning_rate": 3.0190568760861245e-06,
      "loss": 0.4801,
      "step": 5125
    },
    {
      "epoch": 0.07584571655792993,
      "grad_norm": 0.8120235800743103,
      "learning_rate": 3.0337839827987396e-06,
      "loss": 0.4761,
      "step": 5150
    },
    {
      "epoch": 0.07621389964801696,
      "grad_norm": 0.7820422053337097,
      "learning_rate": 3.048511089511355e-06,
      "loss": 0.4821,
      "step": 5175
    },
    {
      "epoch": 0.076582082738104,
      "grad_norm": 0.7186234593391418,
      "learning_rate": 3.0632381962239697e-06,
      "loss": 0.4833,
      "step": 5200
    },
    {
      "epoch": 0.07695026582819105,
      "grad_norm": 0.6120290160179138,
      "learning_rate": 3.0779653029365852e-06,
      "loss": 0.481,
      "step": 5225
    },
    {
      "epoch": 0.07731844891827808,
      "grad_norm": 0.6122280955314636,
      "learning_rate": 3.0926924096492007e-06,
      "loss": 0.4862,
      "step": 5250
    },
    {
      "epoch": 0.07768663200836512,
      "grad_norm": 0.5714207887649536,
      "learning_rate": 3.107419516361816e-06,
      "loss": 0.481,
      "step": 5275
    },
    {
      "epoch": 0.07805481509845216,
      "grad_norm": 0.609611451625824,
      "learning_rate": 3.1221466230744313e-06,
      "loss": 0.4891,
      "step": 5300
    },
    {
      "epoch": 0.07842299818853919,
      "grad_norm": 0.5937544107437134,
      "learning_rate": 3.136873729787046e-06,
      "loss": 0.4778,
      "step": 5325
    },
    {
      "epoch": 0.07879118127862623,
      "grad_norm": 0.698113739490509,
      "learning_rate": 3.1516008364996614e-06,
      "loss": 0.4771,
      "step": 5350
    },
    {
      "epoch": 0.07915936436871328,
      "grad_norm": 0.6822773814201355,
      "learning_rate": 3.166327943212277e-06,
      "loss": 0.4756,
      "step": 5375
    },
    {
      "epoch": 0.07952754745880031,
      "grad_norm": 0.610352635383606,
      "learning_rate": 3.181055049924892e-06,
      "loss": 0.4759,
      "step": 5400
    },
    {
      "epoch": 0.07989573054888735,
      "grad_norm": 0.6700140237808228,
      "learning_rate": 3.1957821566375075e-06,
      "loss": 0.4813,
      "step": 5425
    },
    {
      "epoch": 0.0802639136389744,
      "grad_norm": 0.909814178943634,
      "learning_rate": 3.210509263350122e-06,
      "loss": 0.4826,
      "step": 5450
    },
    {
      "epoch": 0.08063209672906142,
      "grad_norm": 0.4589587450027466,
      "learning_rate": 3.2252363700627377e-06,
      "loss": 0.4706,
      "step": 5475
    },
    {
      "epoch": 0.08100027981914847,
      "grad_norm": 0.5017442107200623,
      "learning_rate": 3.2399634767753527e-06,
      "loss": 0.4756,
      "step": 5500
    },
    {
      "epoch": 0.08136846290923551,
      "grad_norm": 0.5886111259460449,
      "learning_rate": 3.2546905834879682e-06,
      "loss": 0.4689,
      "step": 5525
    },
    {
      "epoch": 0.08173664599932254,
      "grad_norm": 0.5248635411262512,
      "learning_rate": 3.2694176902005837e-06,
      "loss": 0.4736,
      "step": 5550
    },
    {
      "epoch": 0.08210482908940958,
      "grad_norm": 0.5123960375785828,
      "learning_rate": 3.2841447969131984e-06,
      "loss": 0.4703,
      "step": 5575
    },
    {
      "epoch": 0.08247301217949662,
      "grad_norm": 0.5010682344436646,
      "learning_rate": 3.298871903625814e-06,
      "loss": 0.4782,
      "step": 5600
    },
    {
      "epoch": 0.08284119526958365,
      "grad_norm": 0.6653745770454407,
      "learning_rate": 3.313599010338429e-06,
      "loss": 0.4701,
      "step": 5625
    },
    {
      "epoch": 0.0832093783596707,
      "grad_norm": 0.5438437461853027,
      "learning_rate": 3.3283261170510445e-06,
      "loss": 0.4714,
      "step": 5650
    },
    {
      "epoch": 0.08357756144975774,
      "grad_norm": 0.5668756365776062,
      "learning_rate": 3.34305322376366e-06,
      "loss": 0.4681,
      "step": 5675
    },
    {
      "epoch": 0.08394574453984477,
      "grad_norm": 0.7274172902107239,
      "learning_rate": 3.357780330476275e-06,
      "loss": 0.4638,
      "step": 5700
    },
    {
      "epoch": 0.08431392762993181,
      "grad_norm": 0.6077244877815247,
      "learning_rate": 3.37250743718889e-06,
      "loss": 0.4781,
      "step": 5725
    },
    {
      "epoch": 0.08468211072001885,
      "grad_norm": 0.4999208152294159,
      "learning_rate": 3.387234543901505e-06,
      "loss": 0.4618,
      "step": 5750
    },
    {
      "epoch": 0.08505029381010588,
      "grad_norm": 0.5622535347938538,
      "learning_rate": 3.4019616506141207e-06,
      "loss": 0.4706,
      "step": 5775
    },
    {
      "epoch": 0.08541847690019293,
      "grad_norm": 0.6131230592727661,
      "learning_rate": 3.416688757326736e-06,
      "loss": 0.4711,
      "step": 5800
    },
    {
      "epoch": 0.08578665999027997,
      "grad_norm": 0.811798095703125,
      "learning_rate": 3.4314158640393513e-06,
      "loss": 0.4626,
      "step": 5825
    },
    {
      "epoch": 0.086154843080367,
      "grad_norm": 0.5276443362236023,
      "learning_rate": 3.446142970751966e-06,
      "loss": 0.4589,
      "step": 5850
    },
    {
      "epoch": 0.08652302617045404,
      "grad_norm": 0.590082049369812,
      "learning_rate": 3.4608700774645814e-06,
      "loss": 0.4653,
      "step": 5875
    },
    {
      "epoch": 0.08689120926054109,
      "grad_norm": 0.5512232780456543,
      "learning_rate": 3.475597184177197e-06,
      "loss": 0.462,
      "step": 5900
    },
    {
      "epoch": 0.08725939235062811,
      "grad_norm": 0.4878486096858978,
      "learning_rate": 3.490324290889812e-06,
      "loss": 0.4628,
      "step": 5925
    },
    {
      "epoch": 0.08762757544071516,
      "grad_norm": 0.7966400384902954,
      "learning_rate": 3.5050513976024275e-06,
      "loss": 0.4622,
      "step": 5950
    },
    {
      "epoch": 0.0879957585308022,
      "grad_norm": 0.5125811696052551,
      "learning_rate": 3.519778504315042e-06,
      "loss": 0.4636,
      "step": 5975
    },
    {
      "epoch": 0.08836394162088923,
      "grad_norm": 0.883798360824585,
      "learning_rate": 3.5345056110276577e-06,
      "loss": 0.4696,
      "step": 6000
    },
    {
      "epoch": 0.08873212471097627,
      "grad_norm": 0.5936166048049927,
      "learning_rate": 3.549232717740273e-06,
      "loss": 0.4726,
      "step": 6025
    },
    {
      "epoch": 0.08910030780106332,
      "grad_norm": 0.9113532900810242,
      "learning_rate": 3.5639598244528882e-06,
      "loss": 0.4624,
      "step": 6050
    },
    {
      "epoch": 0.08946849089115035,
      "grad_norm": 0.8539900779724121,
      "learning_rate": 3.5786869311655037e-06,
      "loss": 0.4648,
      "step": 6075
    },
    {
      "epoch": 0.08983667398123739,
      "grad_norm": 0.7228135466575623,
      "learning_rate": 3.5934140378781184e-06,
      "loss": 0.472,
      "step": 6100
    },
    {
      "epoch": 0.09020485707132443,
      "grad_norm": 0.6169723868370056,
      "learning_rate": 3.608141144590734e-06,
      "loss": 0.4591,
      "step": 6125
    },
    {
      "epoch": 0.09057304016141146,
      "grad_norm": 0.7366740703582764,
      "learning_rate": 3.6228682513033494e-06,
      "loss": 0.4636,
      "step": 6150
    },
    {
      "epoch": 0.0909412232514985,
      "grad_norm": 1.0246922969818115,
      "learning_rate": 3.6375953580159645e-06,
      "loss": 0.4601,
      "step": 6175
    },
    {
      "epoch": 0.09130940634158555,
      "grad_norm": 0.603578507900238,
      "learning_rate": 3.65232246472858e-06,
      "loss": 0.4681,
      "step": 6200
    },
    {
      "epoch": 0.09167758943167258,
      "grad_norm": 0.6169431805610657,
      "learning_rate": 3.667049571441195e-06,
      "loss": 0.4709,
      "step": 6225
    },
    {
      "epoch": 0.09204577252175962,
      "grad_norm": 0.6031911969184875,
      "learning_rate": 3.68177667815381e-06,
      "loss": 0.4548,
      "step": 6250
    },
    {
      "epoch": 0.09241395561184666,
      "grad_norm": 0.7893526554107666,
      "learning_rate": 3.696503784866425e-06,
      "loss": 0.4542,
      "step": 6275
    },
    {
      "epoch": 0.09278213870193369,
      "grad_norm": 0.738535463809967,
      "learning_rate": 3.7112308915790407e-06,
      "loss": 0.4577,
      "step": 6300
    },
    {
      "epoch": 0.09315032179202073,
      "grad_norm": 0.8681890964508057,
      "learning_rate": 3.725957998291656e-06,
      "loss": 0.4612,
      "step": 6325
    },
    {
      "epoch": 0.09351850488210778,
      "grad_norm": 0.6814743876457214,
      "learning_rate": 3.7406851050042713e-06,
      "loss": 0.4621,
      "step": 6350
    },
    {
      "epoch": 0.09388668797219481,
      "grad_norm": 0.5325171947479248,
      "learning_rate": 3.7554122117168863e-06,
      "loss": 0.4646,
      "step": 6375
    },
    {
      "epoch": 0.09425487106228185,
      "grad_norm": 0.6647558808326721,
      "learning_rate": 3.7701393184295014e-06,
      "loss": 0.4549,
      "step": 6400
    },
    {
      "epoch": 0.0946230541523689,
      "grad_norm": 0.5672147870063782,
      "learning_rate": 3.784866425142117e-06,
      "loss": 0.4528,
      "step": 6425
    },
    {
      "epoch": 0.09499123724245592,
      "grad_norm": 0.7314438223838806,
      "learning_rate": 3.7995935318547324e-06,
      "loss": 0.46,
      "step": 6450
    },
    {
      "epoch": 0.09535942033254297,
      "grad_norm": 1.0649642944335938,
      "learning_rate": 3.8143206385673475e-06,
      "loss": 0.447,
      "step": 6475
    },
    {
      "epoch": 0.09572760342263001,
      "grad_norm": 0.47005271911621094,
      "learning_rate": 3.829047745279962e-06,
      "loss": 0.4602,
      "step": 6500
    },
    {
      "epoch": 0.09609578651271704,
      "grad_norm": 0.678763210773468,
      "learning_rate": 3.843774851992578e-06,
      "loss": 0.4521,
      "step": 6525
    },
    {
      "epoch": 0.09646396960280408,
      "grad_norm": 0.4841667413711548,
      "learning_rate": 3.858501958705193e-06,
      "loss": 0.4439,
      "step": 6550
    },
    {
      "epoch": 0.09683215269289112,
      "grad_norm": 0.5574337244033813,
      "learning_rate": 3.873229065417809e-06,
      "loss": 0.4469,
      "step": 6575
    },
    {
      "epoch": 0.09720033578297815,
      "grad_norm": 0.5939049124717712,
      "learning_rate": 3.887956172130424e-06,
      "loss": 0.4527,
      "step": 6600
    },
    {
      "epoch": 0.0975685188730652,
      "grad_norm": 0.4860667288303375,
      "learning_rate": 3.902683278843039e-06,
      "loss": 0.4379,
      "step": 6625
    },
    {
      "epoch": 0.09793670196315224,
      "grad_norm": 0.6220310926437378,
      "learning_rate": 3.917410385555654e-06,
      "loss": 0.4429,
      "step": 6650
    },
    {
      "epoch": 0.09830488505323927,
      "grad_norm": 0.6435657143592834,
      "learning_rate": 3.932137492268269e-06,
      "loss": 0.4479,
      "step": 6675
    },
    {
      "epoch": 0.09867306814332631,
      "grad_norm": 0.5766171813011169,
      "learning_rate": 3.9468645989808845e-06,
      "loss": 0.4528,
      "step": 6700
    },
    {
      "epoch": 0.09904125123341335,
      "grad_norm": 0.7544565796852112,
      "learning_rate": 3.9615917056935e-06,
      "loss": 0.4544,
      "step": 6725
    },
    {
      "epoch": 0.09940943432350038,
      "grad_norm": 0.514340877532959,
      "learning_rate": 3.9763188124061155e-06,
      "loss": 0.4394,
      "step": 6750
    },
    {
      "epoch": 0.09977761741358743,
      "grad_norm": 0.5275327563285828,
      "learning_rate": 3.99104591911873e-06,
      "loss": 0.45,
      "step": 6775
    },
    {
      "epoch": 0.10014580050367447,
      "grad_norm": 0.5144765377044678,
      "learning_rate": 4.005773025831346e-06,
      "loss": 0.4442,
      "step": 6800
    },
    {
      "epoch": 0.1005139835937615,
      "grad_norm": 0.6812426447868347,
      "learning_rate": 4.020500132543961e-06,
      "loss": 0.4471,
      "step": 6825
    },
    {
      "epoch": 0.10088216668384854,
      "grad_norm": 0.5578859448432922,
      "learning_rate": 4.035227239256576e-06,
      "loss": 0.4438,
      "step": 6850
    },
    {
      "epoch": 0.10125034977393559,
      "grad_norm": 0.5542449951171875,
      "learning_rate": 4.049954345969191e-06,
      "loss": 0.4504,
      "step": 6875
    },
    {
      "epoch": 0.10161853286402261,
      "grad_norm": 0.47460320591926575,
      "learning_rate": 4.064681452681806e-06,
      "loss": 0.449,
      "step": 6900
    },
    {
      "epoch": 0.10198671595410966,
      "grad_norm": 0.5776277184486389,
      "learning_rate": 4.079408559394421e-06,
      "loss": 0.4435,
      "step": 6925
    },
    {
      "epoch": 0.1023548990441967,
      "grad_norm": 0.6199057102203369,
      "learning_rate": 4.094135666107037e-06,
      "loss": 0.4394,
      "step": 6950
    },
    {
      "epoch": 0.10272308213428373,
      "grad_norm": 0.5915837287902832,
      "learning_rate": 4.108862772819652e-06,
      "loss": 0.4424,
      "step": 6975
    },
    {
      "epoch": 0.10309126522437077,
      "grad_norm": 0.6803795695304871,
      "learning_rate": 4.123589879532268e-06,
      "loss": 0.4354,
      "step": 7000
    },
    {
      "epoch": 0.10345944831445782,
      "grad_norm": 0.6849611401557922,
      "learning_rate": 4.1383169862448826e-06,
      "loss": 0.4321,
      "step": 7025
    },
    {
      "epoch": 0.10382763140454485,
      "grad_norm": 0.7074750661849976,
      "learning_rate": 4.153044092957498e-06,
      "loss": 0.441,
      "step": 7050
    },
    {
      "epoch": 0.10419581449463189,
      "grad_norm": 0.5996087789535522,
      "learning_rate": 4.167771199670113e-06,
      "loss": 0.4315,
      "step": 7075
    },
    {
      "epoch": 0.10456399758471893,
      "grad_norm": 0.7005184888839722,
      "learning_rate": 4.182498306382728e-06,
      "loss": 0.437,
      "step": 7100
    },
    {
      "epoch": 0.10493218067480596,
      "grad_norm": 0.6087741255760193,
      "learning_rate": 4.197225413095344e-06,
      "loss": 0.4332,
      "step": 7125
    },
    {
      "epoch": 0.105300363764893,
      "grad_norm": 0.6142122149467468,
      "learning_rate": 4.211952519807958e-06,
      "loss": 0.4306,
      "step": 7150
    },
    {
      "epoch": 0.10566854685498005,
      "grad_norm": 0.5826869010925293,
      "learning_rate": 4.226679626520574e-06,
      "loss": 0.437,
      "step": 7175
    },
    {
      "epoch": 0.10603672994506708,
      "grad_norm": 0.6333126425743103,
      "learning_rate": 4.241406733233189e-06,
      "loss": 0.4268,
      "step": 7200
    },
    {
      "epoch": 0.10640491303515412,
      "grad_norm": 0.6748629808425903,
      "learning_rate": 4.256133839945805e-06,
      "loss": 0.4382,
      "step": 7225
    },
    {
      "epoch": 0.10677309612524116,
      "grad_norm": 0.5124508738517761,
      "learning_rate": 4.27086094665842e-06,
      "loss": 0.4228,
      "step": 7250
    },
    {
      "epoch": 0.10714127921532819,
      "grad_norm": 0.5307368636131287,
      "learning_rate": 4.285588053371035e-06,
      "loss": 0.4263,
      "step": 7275
    },
    {
      "epoch": 0.10750946230541523,
      "grad_norm": 0.572047770023346,
      "learning_rate": 4.3003151600836505e-06,
      "loss": 0.4298,
      "step": 7300
    },
    {
      "epoch": 0.10787764539550228,
      "grad_norm": 0.7120387554168701,
      "learning_rate": 4.315042266796265e-06,
      "loss": 0.4186,
      "step": 7325
    },
    {
      "epoch": 0.10824582848558931,
      "grad_norm": 0.5541498064994812,
      "learning_rate": 4.329769373508881e-06,
      "loss": 0.427,
      "step": 7350
    },
    {
      "epoch": 0.10861401157567635,
      "grad_norm": 0.6520849466323853,
      "learning_rate": 4.344496480221496e-06,
      "loss": 0.432,
      "step": 7375
    },
    {
      "epoch": 0.1089821946657634,
      "grad_norm": 0.5903097987174988,
      "learning_rate": 4.359223586934112e-06,
      "loss": 0.425,
      "step": 7400
    },
    {
      "epoch": 0.10935037775585042,
      "grad_norm": 0.7111144661903381,
      "learning_rate": 4.373950693646726e-06,
      "loss": 0.4266,
      "step": 7425
    },
    {
      "epoch": 0.10971856084593747,
      "grad_norm": 0.7609113454818726,
      "learning_rate": 4.388677800359342e-06,
      "loss": 0.4227,
      "step": 7450
    },
    {
      "epoch": 0.11008674393602451,
      "grad_norm": 0.6072840690612793,
      "learning_rate": 4.403404907071957e-06,
      "loss": 0.4217,
      "step": 7475
    },
    {
      "epoch": 0.11045492702611154,
      "grad_norm": 0.5471121668815613,
      "learning_rate": 4.418132013784572e-06,
      "loss": 0.4202,
      "step": 7500
    },
    {
      "epoch": 0.11082311011619858,
      "grad_norm": 0.7271543741226196,
      "learning_rate": 4.4328591204971875e-06,
      "loss": 0.422,
      "step": 7525
    },
    {
      "epoch": 0.11119129320628562,
      "grad_norm": 0.6403235793113708,
      "learning_rate": 4.447586227209802e-06,
      "loss": 0.421,
      "step": 7550
    },
    {
      "epoch": 0.11155947629637265,
      "grad_norm": 0.6067976355552673,
      "learning_rate": 4.462313333922418e-06,
      "loss": 0.42,
      "step": 7575
    },
    {
      "epoch": 0.1119276593864597,
      "grad_norm": 0.5522935390472412,
      "learning_rate": 4.477040440635033e-06,
      "loss": 0.4218,
      "step": 7600
    },
    {
      "epoch": 0.11229584247654674,
      "grad_norm": 0.5163924694061279,
      "learning_rate": 4.491767547347649e-06,
      "loss": 0.4217,
      "step": 7625
    },
    {
      "epoch": 0.11266402556663378,
      "grad_norm": 0.5037286877632141,
      "learning_rate": 4.506494654060264e-06,
      "loss": 0.4151,
      "step": 7650
    },
    {
      "epoch": 0.11303220865672081,
      "grad_norm": 0.7009169459342957,
      "learning_rate": 4.521221760772879e-06,
      "loss": 0.4181,
      "step": 7675
    },
    {
      "epoch": 0.11340039174680785,
      "grad_norm": 0.6608708500862122,
      "learning_rate": 4.535948867485494e-06,
      "loss": 0.4206,
      "step": 7700
    },
    {
      "epoch": 0.1137685748368949,
      "grad_norm": 0.5974350571632385,
      "learning_rate": 4.55067597419811e-06,
      "loss": 0.4079,
      "step": 7725
    },
    {
      "epoch": 0.11413675792698193,
      "grad_norm": 0.5736297369003296,
      "learning_rate": 4.5654030809107244e-06,
      "loss": 0.4073,
      "step": 7750
    },
    {
      "epoch": 0.11450494101706897,
      "grad_norm": 0.6583185791969299,
      "learning_rate": 4.58013018762334e-06,
      "loss": 0.4115,
      "step": 7775
    },
    {
      "epoch": 0.11487312410715601,
      "grad_norm": 0.5225375294685364,
      "learning_rate": 4.5948572943359554e-06,
      "loss": 0.4137,
      "step": 7800
    },
    {
      "epoch": 0.11524130719724304,
      "grad_norm": 0.6312311291694641,
      "learning_rate": 4.60958440104857e-06,
      "loss": 0.4096,
      "step": 7825
    },
    {
      "epoch": 0.11560949028733009,
      "grad_norm": 0.5658524036407471,
      "learning_rate": 4.624311507761186e-06,
      "loss": 0.4067,
      "step": 7850
    },
    {
      "epoch": 0.11597767337741713,
      "grad_norm": 0.5291056036949158,
      "learning_rate": 4.639038614473801e-06,
      "loss": 0.4019,
      "step": 7875
    },
    {
      "epoch": 0.11634585646750416,
      "grad_norm": 0.638701856136322,
      "learning_rate": 4.653765721186417e-06,
      "loss": 0.3978,
      "step": 7900
    },
    {
      "epoch": 0.1167140395575912,
      "grad_norm": 0.5484588146209717,
      "learning_rate": 4.668492827899031e-06,
      "loss": 0.4032,
      "step": 7925
    },
    {
      "epoch": 0.11708222264767824,
      "grad_norm": 0.6266508102416992,
      "learning_rate": 4.683219934611647e-06,
      "loss": 0.4034,
      "step": 7950
    },
    {
      "epoch": 0.11745040573776527,
      "grad_norm": 0.7253883481025696,
      "learning_rate": 4.697947041324261e-06,
      "loss": 0.3979,
      "step": 7975
    },
    {
      "epoch": 0.11781858882785232,
      "grad_norm": 0.6128943562507629,
      "learning_rate": 4.712674148036877e-06,
      "loss": 0.4051,
      "step": 8000
    },
    {
      "epoch": 0.11818677191793936,
      "grad_norm": 0.8992940187454224,
      "learning_rate": 4.727401254749492e-06,
      "loss": 0.4065,
      "step": 8025
    },
    {
      "epoch": 0.11855495500802639,
      "grad_norm": 0.6067450046539307,
      "learning_rate": 4.742128361462108e-06,
      "loss": 0.3956,
      "step": 8050
    },
    {
      "epoch": 0.11892313809811343,
      "grad_norm": 0.5823917984962463,
      "learning_rate": 4.7568554681747225e-06,
      "loss": 0.3936,
      "step": 8075
    },
    {
      "epoch": 0.11929132118820047,
      "grad_norm": 0.7057197093963623,
      "learning_rate": 4.771582574887338e-06,
      "loss": 0.4008,
      "step": 8100
    },
    {
      "epoch": 0.1196595042782875,
      "grad_norm": 0.780258297920227,
      "learning_rate": 4.7863096815999535e-06,
      "loss": 0.3969,
      "step": 8125
    },
    {
      "epoch": 0.12002768736837455,
      "grad_norm": 0.8426298499107361,
      "learning_rate": 4.801036788312568e-06,
      "loss": 0.4089,
      "step": 8150
    },
    {
      "epoch": 0.12039587045846159,
      "grad_norm": 0.679779589176178,
      "learning_rate": 4.815763895025184e-06,
      "loss": 0.4062,
      "step": 8175
    },
    {
      "epoch": 0.12076405354854862,
      "grad_norm": 0.5497132539749146,
      "learning_rate": 4.830491001737798e-06,
      "loss": 0.3992,
      "step": 8200
    },
    {
      "epoch": 0.12113223663863566,
      "grad_norm": 0.7404468059539795,
      "learning_rate": 4.845218108450414e-06,
      "loss": 0.3931,
      "step": 8225
    },
    {
      "epoch": 0.1215004197287227,
      "grad_norm": 0.647087574005127,
      "learning_rate": 4.859945215163029e-06,
      "loss": 0.3994,
      "step": 8250
    },
    {
      "epoch": 0.12186860281880973,
      "grad_norm": 0.7880210280418396,
      "learning_rate": 4.874672321875645e-06,
      "loss": 0.3974,
      "step": 8275
    },
    {
      "epoch": 0.12223678590889678,
      "grad_norm": 0.8280056715011597,
      "learning_rate": 4.88939942858826e-06,
      "loss": 0.3905,
      "step": 8300
    },
    {
      "epoch": 0.12260496899898382,
      "grad_norm": 0.6425456404685974,
      "learning_rate": 4.904126535300876e-06,
      "loss": 0.3957,
      "step": 8325
    },
    {
      "epoch": 0.12297315208907085,
      "grad_norm": 0.6932951211929321,
      "learning_rate": 4.9188536420134905e-06,
      "loss": 0.3937,
      "step": 8350
    },
    {
      "epoch": 0.1233413351791579,
      "grad_norm": 0.7406250238418579,
      "learning_rate": 4.933580748726106e-06,
      "loss": 0.3852,
      "step": 8375
    },
    {
      "epoch": 0.12370951826924494,
      "grad_norm": 0.6257508397102356,
      "learning_rate": 4.948307855438721e-06,
      "loss": 0.3925,
      "step": 8400
    },
    {
      "epoch": 0.12407770135933197,
      "grad_norm": 0.749832272529602,
      "learning_rate": 4.963034962151336e-06,
      "loss": 0.3916,
      "step": 8425
    },
    {
      "epoch": 0.12444588444941901,
      "grad_norm": 0.6868317723274231,
      "learning_rate": 4.977762068863952e-06,
      "loss": 0.392,
      "step": 8450
    },
    {
      "epoch": 0.12481406753950605,
      "grad_norm": 0.8427602648735046,
      "learning_rate": 4.992489175576566e-06,
      "loss": 0.4006,
      "step": 8475
    },
    {
      "epoch": 0.1251822506295931,
      "grad_norm": 0.6355247497558594,
      "learning_rate": 5.007216282289183e-06,
      "loss": 0.3928,
      "step": 8500
    },
    {
      "epoch": 0.1255504337196801,
      "grad_norm": 0.7556335926055908,
      "learning_rate": 5.0219433890017964e-06,
      "loss": 0.393,
      "step": 8525
    },
    {
      "epoch": 0.12591861680976715,
      "grad_norm": 0.6544154286384583,
      "learning_rate": 5.036670495714412e-06,
      "loss": 0.388,
      "step": 8550
    },
    {
      "epoch": 0.1262867998998542,
      "grad_norm": 0.7124810218811035,
      "learning_rate": 5.0513976024270274e-06,
      "loss": 0.3793,
      "step": 8575
    },
    {
      "epoch": 0.12665498298994124,
      "grad_norm": 1.151803731918335,
      "learning_rate": 5.066124709139643e-06,
      "loss": 0.3837,
      "step": 8600
    },
    {
      "epoch": 0.12702316608002828,
      "grad_norm": 0.8259851932525635,
      "learning_rate": 5.080851815852258e-06,
      "loss": 0.3858,
      "step": 8625
    },
    {
      "epoch": 0.12739134917011533,
      "grad_norm": 0.9365520477294922,
      "learning_rate": 5.095578922564873e-06,
      "loss": 0.3875,
      "step": 8650
    },
    {
      "epoch": 0.12775953226020234,
      "grad_norm": 0.6092934012413025,
      "learning_rate": 5.110306029277489e-06,
      "loss": 0.3819,
      "step": 8675
    },
    {
      "epoch": 0.12812771535028938,
      "grad_norm": 0.665541410446167,
      "learning_rate": 5.125033135990104e-06,
      "loss": 0.3822,
      "step": 8700
    },
    {
      "epoch": 0.12849589844037643,
      "grad_norm": 0.5750928521156311,
      "learning_rate": 5.13976024270272e-06,
      "loss": 0.3794,
      "step": 8725
    },
    {
      "epoch": 0.12886408153046347,
      "grad_norm": 0.8719878196716309,
      "learning_rate": 5.154487349415335e-06,
      "loss": 0.3839,
      "step": 8750
    },
    {
      "epoch": 0.1292322646205505,
      "grad_norm": 0.5725120902061462,
      "learning_rate": 5.16921445612795e-06,
      "loss": 0.3827,
      "step": 8775
    },
    {
      "epoch": 0.12960044771063756,
      "grad_norm": 0.6080131530761719,
      "learning_rate": 5.183941562840564e-06,
      "loss": 0.3784,
      "step": 8800
    },
    {
      "epoch": 0.12996863080072457,
      "grad_norm": 0.7401121854782104,
      "learning_rate": 5.19866866955318e-06,
      "loss": 0.3785,
      "step": 8825
    },
    {
      "epoch": 0.13033681389081161,
      "grad_norm": 0.637549877166748,
      "learning_rate": 5.2133957762657946e-06,
      "loss": 0.369,
      "step": 8850
    },
    {
      "epoch": 0.13070499698089866,
      "grad_norm": 0.6860336065292358,
      "learning_rate": 5.22812288297841e-06,
      "loss": 0.3767,
      "step": 8875
    },
    {
      "epoch": 0.1310731800709857,
      "grad_norm": 0.6861122846603394,
      "learning_rate": 5.2428499896910256e-06,
      "loss": 0.3764,
      "step": 8900
    },
    {
      "epoch": 0.13144136316107274,
      "grad_norm": 0.608595073223114,
      "learning_rate": 5.257577096403641e-06,
      "loss": 0.3771,
      "step": 8925
    },
    {
      "epoch": 0.1318095462511598,
      "grad_norm": 0.6960775852203369,
      "learning_rate": 5.2723042031162566e-06,
      "loss": 0.3757,
      "step": 8950
    },
    {
      "epoch": 0.1321777293412468,
      "grad_norm": 0.9528156518936157,
      "learning_rate": 5.287031309828872e-06,
      "loss": 0.3786,
      "step": 8975
    },
    {
      "epoch": 0.13254591243133385,
      "grad_norm": 1.1325275897979736,
      "learning_rate": 5.301758416541487e-06,
      "loss": 0.3728,
      "step": 9000
    },
    {
      "epoch": 0.1329140955214209,
      "grad_norm": 0.6594205498695374,
      "learning_rate": 5.316485523254102e-06,
      "loss": 0.3839,
      "step": 9025
    },
    {
      "epoch": 0.13328227861150793,
      "grad_norm": 0.6903712749481201,
      "learning_rate": 5.331212629966717e-06,
      "loss": 0.3671,
      "step": 9050
    },
    {
      "epoch": 0.13365046170159497,
      "grad_norm": 0.8763689398765564,
      "learning_rate": 5.345939736679332e-06,
      "loss": 0.3657,
      "step": 9075
    },
    {
      "epoch": 0.13401864479168202,
      "grad_norm": 0.7336785197257996,
      "learning_rate": 5.360666843391947e-06,
      "loss": 0.3725,
      "step": 9100
    },
    {
      "epoch": 0.13438682788176903,
      "grad_norm": 0.7568676471710205,
      "learning_rate": 5.3753939501045625e-06,
      "loss": 0.3696,
      "step": 9125
    },
    {
      "epoch": 0.13475501097185608,
      "grad_norm": 0.7238675951957703,
      "learning_rate": 5.390121056817178e-06,
      "loss": 0.3695,
      "step": 9150
    },
    {
      "epoch": 0.13512319406194312,
      "grad_norm": 0.6972078084945679,
      "learning_rate": 5.4048481635297935e-06,
      "loss": 0.3648,
      "step": 9175
    },
    {
      "epoch": 0.13549137715203016,
      "grad_norm": 0.6761888861656189,
      "learning_rate": 5.419575270242409e-06,
      "loss": 0.3586,
      "step": 9200
    },
    {
      "epoch": 0.1358595602421172,
      "grad_norm": 0.7237305045127869,
      "learning_rate": 5.434302376955024e-06,
      "loss": 0.373,
      "step": 9225
    },
    {
      "epoch": 0.13622774333220425,
      "grad_norm": 1.0163944959640503,
      "learning_rate": 5.449029483667639e-06,
      "loss": 0.3635,
      "step": 9250
    },
    {
      "epoch": 0.13659592642229126,
      "grad_norm": 0.7047263979911804,
      "learning_rate": 5.463756590380255e-06,
      "loss": 0.366,
      "step": 9275
    },
    {
      "epoch": 0.1369641095123783,
      "grad_norm": 0.8444983959197998,
      "learning_rate": 5.47848369709287e-06,
      "loss": 0.378,
      "step": 9300
    },
    {
      "epoch": 0.13733229260246535,
      "grad_norm": 0.7295616865158081,
      "learning_rate": 5.493210803805484e-06,
      "loss": 0.3649,
      "step": 9325
    },
    {
      "epoch": 0.1377004756925524,
      "grad_norm": 0.6368780136108398,
      "learning_rate": 5.5079379105180995e-06,
      "loss": 0.3678,
      "step": 9350
    },
    {
      "epoch": 0.13806865878263944,
      "grad_norm": 0.8212118148803711,
      "learning_rate": 5.522665017230715e-06,
      "loss": 0.365,
      "step": 9375
    },
    {
      "epoch": 0.13843684187272648,
      "grad_norm": 0.7287224531173706,
      "learning_rate": 5.5373921239433305e-06,
      "loss": 0.3691,
      "step": 9400
    },
    {
      "epoch": 0.1388050249628135,
      "grad_norm": 0.641190767288208,
      "learning_rate": 5.552119230655946e-06,
      "loss": 0.365,
      "step": 9425
    },
    {
      "epoch": 0.13917320805290054,
      "grad_norm": 0.6722202897071838,
      "learning_rate": 5.5668463373685615e-06,
      "loss": 0.3496,
      "step": 9450
    },
    {
      "epoch": 0.13954139114298758,
      "grad_norm": 0.5809231996536255,
      "learning_rate": 5.581573444081176e-06,
      "loss": 0.3659,
      "step": 9475
    },
    {
      "epoch": 0.13990957423307462,
      "grad_norm": 1.0955806970596313,
      "learning_rate": 5.596300550793792e-06,
      "loss": 0.3623,
      "step": 9500
    },
    {
      "epoch": 0.14027775732316167,
      "grad_norm": 0.7431578040122986,
      "learning_rate": 5.611027657506407e-06,
      "loss": 0.3474,
      "step": 9525
    },
    {
      "epoch": 0.1406459404132487,
      "grad_norm": 0.7483206987380981,
      "learning_rate": 5.625754764219023e-06,
      "loss": 0.3574,
      "step": 9550
    },
    {
      "epoch": 0.14101412350333573,
      "grad_norm": 0.8954830169677734,
      "learning_rate": 5.640481870931636e-06,
      "loss": 0.3637,
      "step": 9575
    },
    {
      "epoch": 0.14138230659342277,
      "grad_norm": 0.6910772323608398,
      "learning_rate": 5.655208977644252e-06,
      "loss": 0.3559,
      "step": 9600
    },
    {
      "epoch": 0.1417504896835098,
      "grad_norm": 0.6799713969230652,
      "learning_rate": 5.669936084356867e-06,
      "loss": 0.3558,
      "step": 9625
    },
    {
      "epoch": 0.14211867277359685,
      "grad_norm": 0.7354669570922852,
      "learning_rate": 5.684663191069483e-06,
      "loss": 0.3578,
      "step": 9650
    },
    {
      "epoch": 0.1424868558636839,
      "grad_norm": 0.792984664440155,
      "learning_rate": 5.699390297782098e-06,
      "loss": 0.3544,
      "step": 9675
    },
    {
      "epoch": 0.14285503895377094,
      "grad_norm": 0.6234838962554932,
      "learning_rate": 5.714117404494713e-06,
      "loss": 0.3529,
      "step": 9700
    },
    {
      "epoch": 0.14322322204385796,
      "grad_norm": 0.8698832392692566,
      "learning_rate": 5.7288445112073286e-06,
      "loss": 0.3512,
      "step": 9725
    },
    {
      "epoch": 0.143591405133945,
      "grad_norm": 0.685492753982544,
      "learning_rate": 5.743571617919944e-06,
      "loss": 0.3587,
      "step": 9750
    },
    {
      "epoch": 0.14395958822403204,
      "grad_norm": 0.8978196382522583,
      "learning_rate": 5.7582987246325596e-06,
      "loss": 0.3445,
      "step": 9775
    },
    {
      "epoch": 0.14432777131411909,
      "grad_norm": 0.6405984163284302,
      "learning_rate": 5.773025831345175e-06,
      "loss": 0.3541,
      "step": 9800
    },
    {
      "epoch": 0.14469595440420613,
      "grad_norm": 0.7188529968261719,
      "learning_rate": 5.7877529380577906e-06,
      "loss": 0.3471,
      "step": 9825
    },
    {
      "epoch": 0.14506413749429317,
      "grad_norm": 0.7249335050582886,
      "learning_rate": 5.802480044770404e-06,
      "loss": 0.3468,
      "step": 9850
    },
    {
      "epoch": 0.1454323205843802,
      "grad_norm": 0.7496221661567688,
      "learning_rate": 5.81720715148302e-06,
      "loss": 0.3546,
      "step": 9875
    },
    {
      "epoch": 0.14580050367446723,
      "grad_norm": 0.745561420917511,
      "learning_rate": 5.831934258195635e-06,
      "loss": 0.3511,
      "step": 9900
    },
    {
      "epoch": 0.14616868676455427,
      "grad_norm": 0.7024303078651428,
      "learning_rate": 5.84666136490825e-06,
      "loss": 0.3505,
      "step": 9925
    },
    {
      "epoch": 0.14653686985464132,
      "grad_norm": 0.7497286796569824,
      "learning_rate": 5.8613884716208655e-06,
      "loss": 0.3472,
      "step": 9950
    },
    {
      "epoch": 0.14690505294472836,
      "grad_norm": 0.6580822467803955,
      "learning_rate": 5.876115578333481e-06,
      "loss": 0.3498,
      "step": 9975
    },
    {
      "epoch": 0.1472732360348154,
      "grad_norm": 0.7973480224609375,
      "learning_rate": 5.8908426850460965e-06,
      "loss": 0.3454,
      "step": 10000
    },
    {
      "epoch": 0.14764141912490242,
      "grad_norm": 0.7136325836181641,
      "learning_rate": 5.9049807074902066e-06,
      "loss": 0.3511,
      "step": 10025
    },
    {
      "epoch": 0.14800960221498946,
      "grad_norm": 1.0403891801834106,
      "learning_rate": 5.919707814202822e-06,
      "loss": 0.3442,
      "step": 10050
    },
    {
      "epoch": 0.1483777853050765,
      "grad_norm": 0.9081152677536011,
      "learning_rate": 5.9344349209154376e-06,
      "loss": 0.3388,
      "step": 10075
    },
    {
      "epoch": 0.14874596839516355,
      "grad_norm": 0.7664965391159058,
      "learning_rate": 5.949162027628052e-06,
      "loss": 0.3445,
      "step": 10100
    },
    {
      "epoch": 0.1491141514852506,
      "grad_norm": 0.6614516973495483,
      "learning_rate": 5.963889134340668e-06,
      "loss": 0.3441,
      "step": 10125
    },
    {
      "epoch": 0.14948233457533763,
      "grad_norm": 0.7747263312339783,
      "learning_rate": 5.978616241053283e-06,
      "loss": 0.3414,
      "step": 10150
    },
    {
      "epoch": 0.14985051766542468,
      "grad_norm": 0.7055206894874573,
      "learning_rate": 5.993343347765899e-06,
      "loss": 0.3478,
      "step": 10175
    },
    {
      "epoch": 0.1502187007555117,
      "grad_norm": 1.1476428508758545,
      "learning_rate": 6.008070454478514e-06,
      "loss": 0.3476,
      "step": 10200
    },
    {
      "epoch": 0.15058688384559873,
      "grad_norm": 0.6901858448982239,
      "learning_rate": 6.02279756119113e-06,
      "loss": 0.3403,
      "step": 10225
    },
    {
      "epoch": 0.15095506693568578,
      "grad_norm": 0.8119340538978577,
      "learning_rate": 6.0375246679037435e-06,
      "loss": 0.3345,
      "step": 10250
    },
    {
      "epoch": 0.15132325002577282,
      "grad_norm": 0.7182418704032898,
      "learning_rate": 6.052251774616359e-06,
      "loss": 0.3444,
      "step": 10275
    },
    {
      "epoch": 0.15169143311585986,
      "grad_norm": 0.6877521872520447,
      "learning_rate": 6.0669788813289745e-06,
      "loss": 0.3368,
      "step": 10300
    },
    {
      "epoch": 0.1520596162059469,
      "grad_norm": 0.7715916633605957,
      "learning_rate": 6.08170598804159e-06,
      "loss": 0.3411,
      "step": 10325
    },
    {
      "epoch": 0.15242779929603392,
      "grad_norm": 0.866104006767273,
      "learning_rate": 6.096433094754205e-06,
      "loss": 0.3423,
      "step": 10350
    },
    {
      "epoch": 0.15279598238612097,
      "grad_norm": 1.0410808324813843,
      "learning_rate": 6.11116020146682e-06,
      "loss": 0.3347,
      "step": 10375
    },
    {
      "epoch": 0.153164165476208,
      "grad_norm": 0.8808249831199646,
      "learning_rate": 6.125887308179436e-06,
      "loss": 0.3439,
      "step": 10400
    },
    {
      "epoch": 0.15353234856629505,
      "grad_norm": 0.7519041895866394,
      "learning_rate": 6.140614414892051e-06,
      "loss": 0.3384,
      "step": 10425
    },
    {
      "epoch": 0.1539005316563821,
      "grad_norm": 0.7106051445007324,
      "learning_rate": 6.155341521604667e-06,
      "loss": 0.3342,
      "step": 10450
    },
    {
      "epoch": 0.15426871474646914,
      "grad_norm": 0.8196017742156982,
      "learning_rate": 6.170068628317281e-06,
      "loss": 0.3318,
      "step": 10475
    },
    {
      "epoch": 0.15463689783655615,
      "grad_norm": 0.8981484770774841,
      "learning_rate": 6.184795735029897e-06,
      "loss": 0.3303,
      "step": 10500
    },
    {
      "epoch": 0.1550050809266432,
      "grad_norm": 0.8908175826072693,
      "learning_rate": 6.1995228417425115e-06,
      "loss": 0.3337,
      "step": 10525
    },
    {
      "epoch": 0.15537326401673024,
      "grad_norm": 1.051892876625061,
      "learning_rate": 6.214249948455127e-06,
      "loss": 0.3453,
      "step": 10550
    },
    {
      "epoch": 0.15574144710681728,
      "grad_norm": 0.760026216506958,
      "learning_rate": 6.228977055167742e-06,
      "loss": 0.3345,
      "step": 10575
    },
    {
      "epoch": 0.15610963019690433,
      "grad_norm": 0.7725528478622437,
      "learning_rate": 6.243704161880357e-06,
      "loss": 0.3324,
      "step": 10600
    },
    {
      "epoch": 0.15647781328699137,
      "grad_norm": 0.7681843042373657,
      "learning_rate": 6.258431268592973e-06,
      "loss": 0.3337,
      "step": 10625
    },
    {
      "epoch": 0.15684599637707838,
      "grad_norm": 0.8469594717025757,
      "learning_rate": 6.273158375305588e-06,
      "loss": 0.3313,
      "step": 10650
    },
    {
      "epoch": 0.15721417946716543,
      "grad_norm": 0.9619236588478088,
      "learning_rate": 6.287885482018204e-06,
      "loss": 0.3312,
      "step": 10675
    },
    {
      "epoch": 0.15758236255725247,
      "grad_norm": 0.9958358407020569,
      "learning_rate": 6.302612588730819e-06,
      "loss": 0.3319,
      "step": 10700
    },
    {
      "epoch": 0.1579505456473395,
      "grad_norm": 0.8480785489082336,
      "learning_rate": 6.317339695443434e-06,
      "loss": 0.3169,
      "step": 10725
    },
    {
      "epoch": 0.15831872873742656,
      "grad_norm": 0.7890802025794983,
      "learning_rate": 6.332066802156049e-06,
      "loss": 0.3231,
      "step": 10750
    },
    {
      "epoch": 0.1586869118275136,
      "grad_norm": 0.735247015953064,
      "learning_rate": 6.346793908868664e-06,
      "loss": 0.3224,
      "step": 10775
    },
    {
      "epoch": 0.15905509491760061,
      "grad_norm": 0.7928955554962158,
      "learning_rate": 6.361521015581279e-06,
      "loss": 0.3273,
      "step": 10800
    },
    {
      "epoch": 0.15942327800768766,
      "grad_norm": 0.8884017467498779,
      "learning_rate": 6.376248122293894e-06,
      "loss": 0.3168,
      "step": 10825
    },
    {
      "epoch": 0.1597914610977747,
      "grad_norm": 0.8429638147354126,
      "learning_rate": 6.3909752290065096e-06,
      "loss": 0.3295,
      "step": 10850
    },
    {
      "epoch": 0.16015964418786174,
      "grad_norm": 0.8525298237800598,
      "learning_rate": 6.405702335719125e-06,
      "loss": 0.3272,
      "step": 10875
    },
    {
      "epoch": 0.1605278272779488,
      "grad_norm": 0.806425154209137,
      "learning_rate": 6.4204294424317406e-06,
      "loss": 0.3165,
      "step": 10900
    },
    {
      "epoch": 0.16089601036803583,
      "grad_norm": 0.7422736287117004,
      "learning_rate": 6.435156549144356e-06,
      "loss": 0.3215,
      "step": 10925
    },
    {
      "epoch": 0.16126419345812285,
      "grad_norm": 1.050544261932373,
      "learning_rate": 6.449883655856971e-06,
      "loss": 0.3145,
      "step": 10950
    },
    {
      "epoch": 0.1616323765482099,
      "grad_norm": 0.7095434069633484,
      "learning_rate": 6.464610762569586e-06,
      "loss": 0.3243,
      "step": 10975
    },
    {
      "epoch": 0.16200055963829693,
      "grad_norm": 0.8569414019584656,
      "learning_rate": 6.479337869282202e-06,
      "loss": 0.329,
      "step": 11000
    },
    {
      "epoch": 0.16236874272838397,
      "grad_norm": 0.704474687576294,
      "learning_rate": 6.494064975994817e-06,
      "loss": 0.3165,
      "step": 11025
    },
    {
      "epoch": 0.16273692581847102,
      "grad_norm": 1.3744553327560425,
      "learning_rate": 6.508792082707431e-06,
      "loss": 0.3207,
      "step": 11050
    },
    {
      "epoch": 0.16310510890855806,
      "grad_norm": 0.9816151261329651,
      "learning_rate": 6.5235191894200465e-06,
      "loss": 0.3131,
      "step": 11075
    },
    {
      "epoch": 0.16347329199864508,
      "grad_norm": 1.0555229187011719,
      "learning_rate": 6.538246296132662e-06,
      "loss": 0.3264,
      "step": 11100
    },
    {
      "epoch": 0.16384147508873212,
      "grad_norm": 0.816231906414032,
      "learning_rate": 6.5529734028452775e-06,
      "loss": 0.3153,
      "step": 11125
    },
    {
      "epoch": 0.16420965817881916,
      "grad_norm": 0.8610765933990479,
      "learning_rate": 6.567700509557893e-06,
      "loss": 0.3156,
      "step": 11150
    },
    {
      "epoch": 0.1645778412689062,
      "grad_norm": 0.8233363628387451,
      "learning_rate": 6.582427616270508e-06,
      "loss": 0.3165,
      "step": 11175
    },
    {
      "epoch": 0.16494602435899325,
      "grad_norm": 0.8406914472579956,
      "learning_rate": 6.597154722983123e-06,
      "loss": 0.3234,
      "step": 11200
    },
    {
      "epoch": 0.1653142074490803,
      "grad_norm": 0.7772729992866516,
      "learning_rate": 6.611881829695739e-06,
      "loss": 0.32,
      "step": 11225
    },
    {
      "epoch": 0.1656823905391673,
      "grad_norm": 0.9047859311103821,
      "learning_rate": 6.626608936408354e-06,
      "loss": 0.3127,
      "step": 11250
    },
    {
      "epoch": 0.16605057362925435,
      "grad_norm": 0.6934922933578491,
      "learning_rate": 6.64133604312097e-06,
      "loss": 0.3125,
      "step": 11275
    },
    {
      "epoch": 0.1664187567193414,
      "grad_norm": 0.8364558219909668,
      "learning_rate": 6.6560631498335835e-06,
      "loss": 0.319,
      "step": 11300
    },
    {
      "epoch": 0.16678693980942844,
      "grad_norm": 0.8287280797958374,
      "learning_rate": 6.670790256546199e-06,
      "loss": 0.3076,
      "step": 11325
    },
    {
      "epoch": 0.16715512289951548,
      "grad_norm": 0.8386410474777222,
      "learning_rate": 6.6855173632588145e-06,
      "loss": 0.3178,
      "step": 11350
    },
    {
      "epoch": 0.16752330598960252,
      "grad_norm": 1.0043636560440063,
      "learning_rate": 6.70024446997143e-06,
      "loss": 0.3279,
      "step": 11375
    },
    {
      "epoch": 0.16789148907968954,
      "grad_norm": 0.7347243428230286,
      "learning_rate": 6.7149715766840455e-06,
      "loss": 0.3041,
      "step": 11400
    },
    {
      "epoch": 0.16825967216977658,
      "grad_norm": 0.7245997786521912,
      "learning_rate": 6.72969868339666e-06,
      "loss": 0.3117,
      "step": 11425
    },
    {
      "epoch": 0.16862785525986362,
      "grad_norm": 1.2553895711898804,
      "learning_rate": 6.744425790109276e-06,
      "loss": 0.3149,
      "step": 11450
    },
    {
      "epoch": 0.16899603834995067,
      "grad_norm": 0.9861347079277039,
      "learning_rate": 6.759152896821891e-06,
      "loss": 0.3043,
      "step": 11475
    },
    {
      "epoch": 0.1693642214400377,
      "grad_norm": 0.8539161682128906,
      "learning_rate": 6.773880003534507e-06,
      "loss": 0.3157,
      "step": 11500
    },
    {
      "epoch": 0.16973240453012475,
      "grad_norm": 0.939351499080658,
      "learning_rate": 6.788607110247122e-06,
      "loss": 0.3093,
      "step": 11525
    },
    {
      "epoch": 0.17010058762021177,
      "grad_norm": 0.8767924904823303,
      "learning_rate": 6.803334216959737e-06,
      "loss": 0.3046,
      "step": 11550
    },
    {
      "epoch": 0.1704687707102988,
      "grad_norm": 0.8383442163467407,
      "learning_rate": 6.8180613236723514e-06,
      "loss": 0.3143,
      "step": 11575
    },
    {
      "epoch": 0.17083695380038585,
      "grad_norm": 0.8109354972839355,
      "learning_rate": 6.832788430384967e-06,
      "loss": 0.2992,
      "step": 11600
    },
    {
      "epoch": 0.1712051368904729,
      "grad_norm": 0.8896241188049316,
      "learning_rate": 6.8475155370975824e-06,
      "loss": 0.3104,
      "step": 11625
    },
    {
      "epoch": 0.17157331998055994,
      "grad_norm": 0.8169327974319458,
      "learning_rate": 6.862242643810197e-06,
      "loss": 0.308,
      "step": 11650
    },
    {
      "epoch": 0.17194150307064698,
      "grad_norm": 1.13442862033844,
      "learning_rate": 6.876969750522813e-06,
      "loss": 0.3081,
      "step": 11675
    },
    {
      "epoch": 0.172309686160734,
      "grad_norm": 0.8486407995223999,
      "learning_rate": 6.891696857235428e-06,
      "loss": 0.3046,
      "step": 11700
    },
    {
      "epoch": 0.17267786925082104,
      "grad_norm": 1.0129576921463013,
      "learning_rate": 6.906423963948044e-06,
      "loss": 0.3037,
      "step": 11725
    },
    {
      "epoch": 0.17304605234090809,
      "grad_norm": 0.9492967128753662,
      "learning_rate": 6.921151070660659e-06,
      "loss": 0.3046,
      "step": 11750
    },
    {
      "epoch": 0.17341423543099513,
      "grad_norm": 0.9998227953910828,
      "learning_rate": 6.935878177373275e-06,
      "loss": 0.294,
      "step": 11775
    },
    {
      "epoch": 0.17378241852108217,
      "grad_norm": 0.7796307802200317,
      "learning_rate": 6.950605284085889e-06,
      "loss": 0.2962,
      "step": 11800
    },
    {
      "epoch": 0.17415060161116921,
      "grad_norm": 0.8251902461051941,
      "learning_rate": 6.965332390798504e-06,
      "loss": 0.2982,
      "step": 11825
    },
    {
      "epoch": 0.17451878470125623,
      "grad_norm": 0.7870060801506042,
      "learning_rate": 6.980059497511119e-06,
      "loss": 0.2966,
      "step": 11850
    },
    {
      "epoch": 0.17488696779134327,
      "grad_norm": 0.9390984177589417,
      "learning_rate": 6.994786604223734e-06,
      "loss": 0.3103,
      "step": 11875
    },
    {
      "epoch": 0.17525515088143032,
      "grad_norm": 0.9660308957099915,
      "learning_rate": 7.0095137109363496e-06,
      "loss": 0.3009,
      "step": 11900
    },
    {
      "epoch": 0.17562333397151736,
      "grad_norm": 1.0213398933410645,
      "learning_rate": 7.024240817648965e-06,
      "loss": 0.306,
      "step": 11925
    },
    {
      "epoch": 0.1759915170616044,
      "grad_norm": 1.5115578174591064,
      "learning_rate": 7.0389679243615805e-06,
      "loss": 0.2988,
      "step": 11950
    },
    {
      "epoch": 0.17635970015169145,
      "grad_norm": 0.902195930480957,
      "learning_rate": 7.053695031074196e-06,
      "loss": 0.3002,
      "step": 11975
    },
    {
      "epoch": 0.17672788324177846,
      "grad_norm": 0.783263087272644,
      "learning_rate": 7.0684221377868115e-06,
      "loss": 0.2987,
      "step": 12000
    },
    {
      "epoch": 0.1770960663318655,
      "grad_norm": 0.7162420153617859,
      "learning_rate": 7.082560160230922e-06,
      "loss": 0.2949,
      "step": 12025
    },
    {
      "epoch": 0.17746424942195255,
      "grad_norm": 0.9869486093521118,
      "learning_rate": 7.097287266943536e-06,
      "loss": 0.2955,
      "step": 12050
    },
    {
      "epoch": 0.1778324325120396,
      "grad_norm": 0.9826450943946838,
      "learning_rate": 7.112014373656152e-06,
      "loss": 0.3006,
      "step": 12075
    },
    {
      "epoch": 0.17820061560212663,
      "grad_norm": 0.9040399789810181,
      "learning_rate": 7.126741480368767e-06,
      "loss": 0.2947,
      "step": 12100
    },
    {
      "epoch": 0.17856879869221368,
      "grad_norm": 1.1333189010620117,
      "learning_rate": 7.141468587081383e-06,
      "loss": 0.2933,
      "step": 12125
    },
    {
      "epoch": 0.1789369817823007,
      "grad_norm": 0.8149101138114929,
      "learning_rate": 7.156195693793998e-06,
      "loss": 0.298,
      "step": 12150
    },
    {
      "epoch": 0.17930516487238773,
      "grad_norm": 0.7296385765075684,
      "learning_rate": 7.170922800506614e-06,
      "loss": 0.2944,
      "step": 12175
    },
    {
      "epoch": 0.17967334796247478,
      "grad_norm": 0.8775405883789062,
      "learning_rate": 7.185649907219228e-06,
      "loss": 0.2978,
      "step": 12200
    },
    {
      "epoch": 0.18004153105256182,
      "grad_norm": 0.8729538321495056,
      "learning_rate": 7.200377013931844e-06,
      "loss": 0.2932,
      "step": 12225
    },
    {
      "epoch": 0.18040971414264886,
      "grad_norm": 0.8459117412567139,
      "learning_rate": 7.2151041206444585e-06,
      "loss": 0.2961,
      "step": 12250
    },
    {
      "epoch": 0.1807778972327359,
      "grad_norm": 0.9352242946624756,
      "learning_rate": 7.229831227357074e-06,
      "loss": 0.307,
      "step": 12275
    },
    {
      "epoch": 0.18114608032282292,
      "grad_norm": 1.170359492301941,
      "learning_rate": 7.244558334069689e-06,
      "loss": 0.2913,
      "step": 12300
    },
    {
      "epoch": 0.18151426341290997,
      "grad_norm": 0.7829805612564087,
      "learning_rate": 7.259285440782304e-06,
      "loss": 0.2972,
      "step": 12325
    },
    {
      "epoch": 0.181882446502997,
      "grad_norm": 0.9527215361595154,
      "learning_rate": 7.27401254749492e-06,
      "loss": 0.2876,
      "step": 12350
    },
    {
      "epoch": 0.18225062959308405,
      "grad_norm": 0.7842198014259338,
      "learning_rate": 7.288739654207535e-06,
      "loss": 0.2925,
      "step": 12375
    },
    {
      "epoch": 0.1826188126831711,
      "grad_norm": 0.9339554905891418,
      "learning_rate": 7.303466760920151e-06,
      "loss": 0.2872,
      "step": 12400
    },
    {
      "epoch": 0.18298699577325814,
      "grad_norm": 0.7690834403038025,
      "learning_rate": 7.318193867632765e-06,
      "loss": 0.2916,
      "step": 12425
    },
    {
      "epoch": 0.18335517886334515,
      "grad_norm": 0.8643265962600708,
      "learning_rate": 7.332920974345381e-06,
      "loss": 0.2932,
      "step": 12450
    },
    {
      "epoch": 0.1837233619534322,
      "grad_norm": 0.9393995404243469,
      "learning_rate": 7.347648081057996e-06,
      "loss": 0.2958,
      "step": 12475
    },
    {
      "epoch": 0.18409154504351924,
      "grad_norm": 0.9852946400642395,
      "learning_rate": 7.362375187770611e-06,
      "loss": 0.2893,
      "step": 12500
    },
    {
      "epoch": 0.18445972813360628,
      "grad_norm": 1.0065406560897827,
      "learning_rate": 7.377102294483226e-06,
      "loss": 0.2961,
      "step": 12525
    },
    {
      "epoch": 0.18482791122369333,
      "grad_norm": 0.8727032542228699,
      "learning_rate": 7.391829401195841e-06,
      "loss": 0.2928,
      "step": 12550
    },
    {
      "epoch": 0.18519609431378037,
      "grad_norm": 1.0887058973312378,
      "learning_rate": 7.406556507908457e-06,
      "loss": 0.2936,
      "step": 12575
    },
    {
      "epoch": 0.18556427740386738,
      "grad_norm": 0.8902204632759094,
      "learning_rate": 7.421283614621072e-06,
      "loss": 0.2928,
      "step": 12600
    },
    {
      "epoch": 0.18593246049395443,
      "grad_norm": 0.9425715208053589,
      "learning_rate": 7.436010721333688e-06,
      "loss": 0.2857,
      "step": 12625
    },
    {
      "epoch": 0.18630064358404147,
      "grad_norm": 1.170000433921814,
      "learning_rate": 7.450737828046303e-06,
      "loss": 0.2958,
      "step": 12650
    },
    {
      "epoch": 0.1866688266741285,
      "grad_norm": 1.0390965938568115,
      "learning_rate": 7.465464934758918e-06,
      "loss": 0.2812,
      "step": 12675
    },
    {
      "epoch": 0.18703700976421556,
      "grad_norm": 1.0281293392181396,
      "learning_rate": 7.480192041471533e-06,
      "loss": 0.2878,
      "step": 12700
    },
    {
      "epoch": 0.1874051928543026,
      "grad_norm": 1.0428563356399536,
      "learning_rate": 7.494919148184149e-06,
      "loss": 0.2854,
      "step": 12725
    },
    {
      "epoch": 0.18777337594438961,
      "grad_norm": 0.9071276187896729,
      "learning_rate": 7.509646254896764e-06,
      "loss": 0.2808,
      "step": 12750
    },
    {
      "epoch": 0.18814155903447666,
      "grad_norm": 0.9674225449562073,
      "learning_rate": 7.524373361609378e-06,
      "loss": 0.2914,
      "step": 12775
    },
    {
      "epoch": 0.1885097421245637,
      "grad_norm": 0.8867351412773132,
      "learning_rate": 7.539100468321994e-06,
      "loss": 0.2809,
      "step": 12800
    },
    {
      "epoch": 0.18887792521465074,
      "grad_norm": 1.048401951789856,
      "learning_rate": 7.553827575034609e-06,
      "loss": 0.2844,
      "step": 12825
    },
    {
      "epoch": 0.1892461083047378,
      "grad_norm": 1.1923837661743164,
      "learning_rate": 7.568554681747225e-06,
      "loss": 0.2846,
      "step": 12850
    },
    {
      "epoch": 0.18961429139482483,
      "grad_norm": 1.000769853591919,
      "learning_rate": 7.58328178845984e-06,
      "loss": 0.2825,
      "step": 12875
    },
    {
      "epoch": 0.18998247448491185,
      "grad_norm": 0.9111427664756775,
      "learning_rate": 7.598008895172455e-06,
      "loss": 0.2867,
      "step": 12900
    },
    {
      "epoch": 0.1903506575749989,
      "grad_norm": 0.8391136527061462,
      "learning_rate": 7.61273600188507e-06,
      "loss": 0.2773,
      "step": 12925
    },
    {
      "epoch": 0.19071884066508593,
      "grad_norm": 1.2547825574874878,
      "learning_rate": 7.627463108597686e-06,
      "loss": 0.2799,
      "step": 12950
    },
    {
      "epoch": 0.19108702375517297,
      "grad_norm": 0.8818742632865906,
      "learning_rate": 7.6421902153103e-06,
      "loss": 0.2744,
      "step": 12975
    },
    {
      "epoch": 0.19145520684526002,
      "grad_norm": 0.884929358959198,
      "learning_rate": 7.656917322022917e-06,
      "loss": 0.2858,
      "step": 13000
    },
    {
      "epoch": 0.19182338993534706,
      "grad_norm": 1.0774343013763428,
      "learning_rate": 7.671644428735531e-06,
      "loss": 0.2904,
      "step": 13025
    },
    {
      "epoch": 0.19219157302543408,
      "grad_norm": 1.2477922439575195,
      "learning_rate": 7.686371535448146e-06,
      "loss": 0.2886,
      "step": 13050
    },
    {
      "epoch": 0.19255975611552112,
      "grad_norm": 1.028517246246338,
      "learning_rate": 7.70109864216076e-06,
      "loss": 0.2866,
      "step": 13075
    },
    {
      "epoch": 0.19292793920560816,
      "grad_norm": 0.9460515975952148,
      "learning_rate": 7.715825748873377e-06,
      "loss": 0.2791,
      "step": 13100
    },
    {
      "epoch": 0.1932961222956952,
      "grad_norm": 0.9999874830245972,
      "learning_rate": 7.730552855585992e-06,
      "loss": 0.2752,
      "step": 13125
    },
    {
      "epoch": 0.19366430538578225,
      "grad_norm": 0.9967283606529236,
      "learning_rate": 7.745279962298608e-06,
      "loss": 0.2722,
      "step": 13150
    },
    {
      "epoch": 0.1940324884758693,
      "grad_norm": 0.9238536357879639,
      "learning_rate": 7.760007069011223e-06,
      "loss": 0.2767,
      "step": 13175
    },
    {
      "epoch": 0.1944006715659563,
      "grad_norm": 1.0233261585235596,
      "learning_rate": 7.774734175723837e-06,
      "loss": 0.2783,
      "step": 13200
    },
    {
      "epoch": 0.19476885465604335,
      "grad_norm": 0.9148591756820679,
      "learning_rate": 7.789461282436454e-06,
      "loss": 0.2791,
      "step": 13225
    },
    {
      "epoch": 0.1951370377461304,
      "grad_norm": 1.153035044670105,
      "learning_rate": 7.804188389149068e-06,
      "loss": 0.2749,
      "step": 13250
    },
    {
      "epoch": 0.19550522083621744,
      "grad_norm": 1.205528736114502,
      "learning_rate": 7.818915495861685e-06,
      "loss": 0.2766,
      "step": 13275
    },
    {
      "epoch": 0.19587340392630448,
      "grad_norm": 0.8279467821121216,
      "learning_rate": 7.833642602574298e-06,
      "loss": 0.2867,
      "step": 13300
    },
    {
      "epoch": 0.19624158701639152,
      "grad_norm": 1.032379150390625,
      "learning_rate": 7.848369709286914e-06,
      "loss": 0.275,
      "step": 13325
    },
    {
      "epoch": 0.19660977010647854,
      "grad_norm": 1.5636012554168701,
      "learning_rate": 7.863096815999529e-06,
      "loss": 0.2791,
      "step": 13350
    },
    {
      "epoch": 0.19697795319656558,
      "grad_norm": 0.8588216304779053,
      "learning_rate": 7.877823922712145e-06,
      "loss": 0.2852,
      "step": 13375
    },
    {
      "epoch": 0.19734613628665262,
      "grad_norm": 1.1563206911087036,
      "learning_rate": 7.89255102942476e-06,
      "loss": 0.2761,
      "step": 13400
    },
    {
      "epoch": 0.19771431937673967,
      "grad_norm": 0.9631919264793396,
      "learning_rate": 7.907278136137374e-06,
      "loss": 0.2813,
      "step": 13425
    },
    {
      "epoch": 0.1980825024668267,
      "grad_norm": 0.9245600700378418,
      "learning_rate": 7.92200524284999e-06,
      "loss": 0.2779,
      "step": 13450
    },
    {
      "epoch": 0.19845068555691375,
      "grad_norm": 1.0227192640304565,
      "learning_rate": 7.936732349562605e-06,
      "loss": 0.2778,
      "step": 13475
    },
    {
      "epoch": 0.19881886864700077,
      "grad_norm": 0.9280269145965576,
      "learning_rate": 7.951459456275222e-06,
      "loss": 0.2751,
      "step": 13500
    },
    {
      "epoch": 0.1991870517370878,
      "grad_norm": 0.945601761341095,
      "learning_rate": 7.966186562987836e-06,
      "loss": 0.2758,
      "step": 13525
    },
    {
      "epoch": 0.19955523482717485,
      "grad_norm": 0.8174274563789368,
      "learning_rate": 7.980913669700451e-06,
      "loss": 0.2754,
      "step": 13550
    },
    {
      "epoch": 0.1999234179172619,
      "grad_norm": 0.8976578712463379,
      "learning_rate": 7.995640776413066e-06,
      "loss": 0.2741,
      "step": 13575
    },
    {
      "epoch": 0.20029160100734894,
      "grad_norm": 1.0173979997634888,
      "learning_rate": 8.010367883125682e-06,
      "loss": 0.269,
      "step": 13600
    },
    {
      "epoch": 0.20065978409743598,
      "grad_norm": 0.9142144918441772,
      "learning_rate": 8.025094989838297e-06,
      "loss": 0.2748,
      "step": 13625
    },
    {
      "epoch": 0.201027967187523,
      "grad_norm": 1.0898303985595703,
      "learning_rate": 8.039822096550913e-06,
      "loss": 0.2695,
      "step": 13650
    },
    {
      "epoch": 0.20139615027761004,
      "grad_norm": 0.9205647706985474,
      "learning_rate": 8.054549203263528e-06,
      "loss": 0.2717,
      "step": 13675
    },
    {
      "epoch": 0.20176433336769709,
      "grad_norm": 0.9046348333358765,
      "learning_rate": 8.069276309976142e-06,
      "loss": 0.2689,
      "step": 13700
    },
    {
      "epoch": 0.20213251645778413,
      "grad_norm": 1.0964778661727905,
      "learning_rate": 8.084003416688759e-06,
      "loss": 0.2745,
      "step": 13725
    },
    {
      "epoch": 0.20250069954787117,
      "grad_norm": 0.9805202484130859,
      "learning_rate": 8.098730523401373e-06,
      "loss": 0.2739,
      "step": 13750
    },
    {
      "epoch": 0.20286888263795821,
      "grad_norm": 0.9941619038581848,
      "learning_rate": 8.11345763011399e-06,
      "loss": 0.2667,
      "step": 13775
    },
    {
      "epoch": 0.20323706572804523,
      "grad_norm": 1.1271330118179321,
      "learning_rate": 8.128184736826604e-06,
      "loss": 0.2647,
      "step": 13800
    },
    {
      "epoch": 0.20360524881813227,
      "grad_norm": 0.9569908380508423,
      "learning_rate": 8.142911843539219e-06,
      "loss": 0.2674,
      "step": 13825
    },
    {
      "epoch": 0.20397343190821932,
      "grad_norm": 0.942244827747345,
      "learning_rate": 8.157638950251834e-06,
      "loss": 0.268,
      "step": 13850
    },
    {
      "epoch": 0.20434161499830636,
      "grad_norm": 1.2185558080673218,
      "learning_rate": 8.17236605696445e-06,
      "loss": 0.2651,
      "step": 13875
    },
    {
      "epoch": 0.2047097980883934,
      "grad_norm": 0.8764684200286865,
      "learning_rate": 8.187093163677065e-06,
      "loss": 0.2614,
      "step": 13900
    },
    {
      "epoch": 0.20507798117848045,
      "grad_norm": 1.1464710235595703,
      "learning_rate": 8.20182027038968e-06,
      "loss": 0.2749,
      "step": 13925
    },
    {
      "epoch": 0.20544616426856746,
      "grad_norm": 0.9970076680183411,
      "learning_rate": 8.216547377102296e-06,
      "loss": 0.2661,
      "step": 13950
    },
    {
      "epoch": 0.2058143473586545,
      "grad_norm": 1.0001120567321777,
      "learning_rate": 8.23127448381491e-06,
      "loss": 0.262,
      "step": 13975
    },
    {
      "epoch": 0.20618253044874155,
      "grad_norm": 1.0340194702148438,
      "learning_rate": 8.246001590527527e-06,
      "loss": 0.2638,
      "step": 14000
    },
    {
      "epoch": 0.2065507135388286,
      "grad_norm": 0.8450446128845215,
      "learning_rate": 8.260139612971636e-06,
      "loss": 0.2626,
      "step": 14025
    },
    {
      "epoch": 0.20691889662891563,
      "grad_norm": 0.991098165512085,
      "learning_rate": 8.274866719684252e-06,
      "loss": 0.2613,
      "step": 14050
    },
    {
      "epoch": 0.20728707971900268,
      "grad_norm": 0.9824557900428772,
      "learning_rate": 8.289004742128363e-06,
      "loss": 0.2676,
      "step": 14075
    },
    {
      "epoch": 0.2076552628090897,
      "grad_norm": 0.9877932667732239,
      "learning_rate": 8.303731848840978e-06,
      "loss": 0.2686,
      "step": 14100
    },
    {
      "epoch": 0.20802344589917673,
      "grad_norm": 1.0560307502746582,
      "learning_rate": 8.318458955553592e-06,
      "loss": 0.255,
      "step": 14125
    },
    {
      "epoch": 0.20839162898926378,
      "grad_norm": 0.992840588092804,
      "learning_rate": 8.333186062266207e-06,
      "loss": 0.2613,
      "step": 14150
    },
    {
      "epoch": 0.20875981207935082,
      "grad_norm": 0.8604615926742554,
      "learning_rate": 8.347913168978823e-06,
      "loss": 0.2656,
      "step": 14175
    },
    {
      "epoch": 0.20912799516943786,
      "grad_norm": 1.2327070236206055,
      "learning_rate": 8.362640275691438e-06,
      "loss": 0.26,
      "step": 14200
    },
    {
      "epoch": 0.2094961782595249,
      "grad_norm": 0.932135820388794,
      "learning_rate": 8.377367382404054e-06,
      "loss": 0.2601,
      "step": 14225
    },
    {
      "epoch": 0.20986436134961192,
      "grad_norm": 0.9523219466209412,
      "learning_rate": 8.392094489116669e-06,
      "loss": 0.2667,
      "step": 14250
    },
    {
      "epoch": 0.21023254443969897,
      "grad_norm": 1.3120263814926147,
      "learning_rate": 8.406821595829284e-06,
      "loss": 0.26,
      "step": 14275
    },
    {
      "epoch": 0.210600727529786,
      "grad_norm": 1.3057265281677246,
      "learning_rate": 8.4215487025419e-06,
      "loss": 0.258,
      "step": 14300
    },
    {
      "epoch": 0.21096891061987305,
      "grad_norm": 0.9782459139823914,
      "learning_rate": 8.436275809254515e-06,
      "loss": 0.2697,
      "step": 14325
    },
    {
      "epoch": 0.2113370937099601,
      "grad_norm": 1.1482174396514893,
      "learning_rate": 8.451002915967131e-06,
      "loss": 0.2546,
      "step": 14350
    },
    {
      "epoch": 0.21170527680004714,
      "grad_norm": 1.288827657699585,
      "learning_rate": 8.465730022679744e-06,
      "loss": 0.2614,
      "step": 14375
    },
    {
      "epoch": 0.21207345989013415,
      "grad_norm": 0.9124887585639954,
      "learning_rate": 8.48045712939236e-06,
      "loss": 0.2565,
      "step": 14400
    },
    {
      "epoch": 0.2124416429802212,
      "grad_norm": 1.1914868354797363,
      "learning_rate": 8.495184236104975e-06,
      "loss": 0.2567,
      "step": 14425
    },
    {
      "epoch": 0.21280982607030824,
      "grad_norm": 1.1137144565582275,
      "learning_rate": 8.509911342817591e-06,
      "loss": 0.2591,
      "step": 14450
    },
    {
      "epoch": 0.21317800916039528,
      "grad_norm": 0.965062141418457,
      "learning_rate": 8.524638449530206e-06,
      "loss": 0.2517,
      "step": 14475
    },
    {
      "epoch": 0.21354619225048233,
      "grad_norm": 1.0583897829055786,
      "learning_rate": 8.53936555624282e-06,
      "loss": 0.2547,
      "step": 14500
    },
    {
      "epoch": 0.21391437534056937,
      "grad_norm": 1.2841545343399048,
      "learning_rate": 8.554092662955437e-06,
      "loss": 0.2568,
      "step": 14525
    },
    {
      "epoch": 0.21428255843065638,
      "grad_norm": 1.2085353136062622,
      "learning_rate": 8.568819769668052e-06,
      "loss": 0.2583,
      "step": 14550
    },
    {
      "epoch": 0.21465074152074343,
      "grad_norm": 0.9443012475967407,
      "learning_rate": 8.583546876380668e-06,
      "loss": 0.253,
      "step": 14575
    },
    {
      "epoch": 0.21501892461083047,
      "grad_norm": 1.041521668434143,
      "learning_rate": 8.598273983093283e-06,
      "loss": 0.2521,
      "step": 14600
    },
    {
      "epoch": 0.2153871077009175,
      "grad_norm": 0.9093597531318665,
      "learning_rate": 8.613001089805897e-06,
      "loss": 0.2551,
      "step": 14625
    },
    {
      "epoch": 0.21575529079100456,
      "grad_norm": 1.1667312383651733,
      "learning_rate": 8.627728196518512e-06,
      "loss": 0.2583,
      "step": 14650
    },
    {
      "epoch": 0.2161234738810916,
      "grad_norm": 0.9899322390556335,
      "learning_rate": 8.642455303231128e-06,
      "loss": 0.257,
      "step": 14675
    },
    {
      "epoch": 0.21649165697117861,
      "grad_norm": 0.9446772933006287,
      "learning_rate": 8.657182409943743e-06,
      "loss": 0.2501,
      "step": 14700
    },
    {
      "epoch": 0.21685984006126566,
      "grad_norm": 1.0645805597305298,
      "learning_rate": 8.671909516656358e-06,
      "loss": 0.2559,
      "step": 14725
    },
    {
      "epoch": 0.2172280231513527,
      "grad_norm": 0.973878026008606,
      "learning_rate": 8.686636623368974e-06,
      "loss": 0.2505,
      "step": 14750
    },
    {
      "epoch": 0.21759620624143974,
      "grad_norm": 1.1371912956237793,
      "learning_rate": 8.701363730081589e-06,
      "loss": 0.2491,
      "step": 14775
    },
    {
      "epoch": 0.2179643893315268,
      "grad_norm": 1.156105875968933,
      "learning_rate": 8.716090836794205e-06,
      "loss": 0.253,
      "step": 14800
    },
    {
      "epoch": 0.21833257242161383,
      "grad_norm": 1.585997223854065,
      "learning_rate": 8.73081794350682e-06,
      "loss": 0.2458,
      "step": 14825
    },
    {
      "epoch": 0.21870075551170085,
      "grad_norm": 1.114240288734436,
      "learning_rate": 8.745545050219436e-06,
      "loss": 0.2484,
      "step": 14850
    },
    {
      "epoch": 0.2190689386017879,
      "grad_norm": 0.9762231707572937,
      "learning_rate": 8.76027215693205e-06,
      "loss": 0.2544,
      "step": 14875
    },
    {
      "epoch": 0.21943712169187493,
      "grad_norm": 1.1107852458953857,
      "learning_rate": 8.774999263644665e-06,
      "loss": 0.2491,
      "step": 14900
    },
    {
      "epoch": 0.21980530478196197,
      "grad_norm": 0.7990845441818237,
      "learning_rate": 8.78972637035728e-06,
      "loss": 0.2509,
      "step": 14925
    },
    {
      "epoch": 0.22017348787204902,
      "grad_norm": 1.110044002532959,
      "learning_rate": 8.804453477069894e-06,
      "loss": 0.257,
      "step": 14950
    },
    {
      "epoch": 0.22054167096213606,
      "grad_norm": 1.1343072652816772,
      "learning_rate": 8.81918058378251e-06,
      "loss": 0.2537,
      "step": 14975
    },
    {
      "epoch": 0.22090985405222308,
      "grad_norm": 1.0467344522476196,
      "learning_rate": 8.833907690495125e-06,
      "loss": 0.2541,
      "step": 15000
    },
    {
      "epoch": 0.22127803714231012,
      "grad_norm": 0.992863655090332,
      "learning_rate": 8.848634797207742e-06,
      "loss": 0.2457,
      "step": 15025
    },
    {
      "epoch": 0.22164622023239716,
      "grad_norm": 0.9078419208526611,
      "learning_rate": 8.863361903920356e-06,
      "loss": 0.2611,
      "step": 15050
    },
    {
      "epoch": 0.2220144033224842,
      "grad_norm": 1.1889991760253906,
      "learning_rate": 8.878089010632973e-06,
      "loss": 0.2475,
      "step": 15075
    },
    {
      "epoch": 0.22238258641257125,
      "grad_norm": 1.1085773706436157,
      "learning_rate": 8.892816117345587e-06,
      "loss": 0.2517,
      "step": 15100
    },
    {
      "epoch": 0.2227507695026583,
      "grad_norm": 1.073651909828186,
      "learning_rate": 8.907543224058202e-06,
      "loss": 0.2575,
      "step": 15125
    },
    {
      "epoch": 0.2231189525927453,
      "grad_norm": 1.0204280614852905,
      "learning_rate": 8.922270330770817e-06,
      "loss": 0.2504,
      "step": 15150
    },
    {
      "epoch": 0.22348713568283235,
      "grad_norm": 1.0358093976974487,
      "learning_rate": 8.936997437483433e-06,
      "loss": 0.2429,
      "step": 15175
    },
    {
      "epoch": 0.2238553187729194,
      "grad_norm": 1.191390037536621,
      "learning_rate": 8.951724544196048e-06,
      "loss": 0.2471,
      "step": 15200
    },
    {
      "epoch": 0.22422350186300644,
      "grad_norm": 1.0816316604614258,
      "learning_rate": 8.966451650908662e-06,
      "loss": 0.249,
      "step": 15225
    },
    {
      "epoch": 0.22459168495309348,
      "grad_norm": 0.9376477599143982,
      "learning_rate": 8.981178757621279e-06,
      "loss": 0.2486,
      "step": 15250
    },
    {
      "epoch": 0.22495986804318052,
      "grad_norm": 1.1276350021362305,
      "learning_rate": 8.995905864333893e-06,
      "loss": 0.2455,
      "step": 15275
    },
    {
      "epoch": 0.22532805113326757,
      "grad_norm": 1.0941764116287231,
      "learning_rate": 9.01063297104651e-06,
      "loss": 0.2482,
      "step": 15300
    },
    {
      "epoch": 0.22569623422335458,
      "grad_norm": 1.093008279800415,
      "learning_rate": 9.025360077759124e-06,
      "loss": 0.2453,
      "step": 15325
    },
    {
      "epoch": 0.22606441731344162,
      "grad_norm": 0.9502381086349487,
      "learning_rate": 9.040087184471739e-06,
      "loss": 0.2458,
      "step": 15350
    },
    {
      "epoch": 0.22643260040352867,
      "grad_norm": 1.2488805055618286,
      "learning_rate": 9.054814291184355e-06,
      "loss": 0.2417,
      "step": 15375
    },
    {
      "epoch": 0.2268007834936157,
      "grad_norm": 1.0725972652435303,
      "learning_rate": 9.06954139789697e-06,
      "loss": 0.2453,
      "step": 15400
    },
    {
      "epoch": 0.22716896658370275,
      "grad_norm": 1.4769214391708374,
      "learning_rate": 9.084268504609585e-06,
      "loss": 0.2414,
      "step": 15425
    },
    {
      "epoch": 0.2275371496737898,
      "grad_norm": 1.148898959159851,
      "learning_rate": 9.0989956113222e-06,
      "loss": 0.2511,
      "step": 15450
    },
    {
      "epoch": 0.2279053327638768,
      "grad_norm": 1.0634573698043823,
      "learning_rate": 9.113722718034816e-06,
      "loss": 0.2396,
      "step": 15475
    },
    {
      "epoch": 0.22827351585396385,
      "grad_norm": 1.1080799102783203,
      "learning_rate": 9.12844982474743e-06,
      "loss": 0.2565,
      "step": 15500
    },
    {
      "epoch": 0.2286416989440509,
      "grad_norm": 1.3626021146774292,
      "learning_rate": 9.143176931460047e-06,
      "loss": 0.248,
      "step": 15525
    },
    {
      "epoch": 0.22900988203413794,
      "grad_norm": 1.0013554096221924,
      "learning_rate": 9.157904038172661e-06,
      "loss": 0.2393,
      "step": 15550
    },
    {
      "epoch": 0.22937806512422498,
      "grad_norm": 1.0677698850631714,
      "learning_rate": 9.172631144885276e-06,
      "loss": 0.2425,
      "step": 15575
    },
    {
      "epoch": 0.22974624821431203,
      "grad_norm": 0.9467232823371887,
      "learning_rate": 9.187358251597892e-06,
      "loss": 0.2328,
      "step": 15600
    },
    {
      "epoch": 0.23011443130439904,
      "grad_norm": 1.2118295431137085,
      "learning_rate": 9.202085358310507e-06,
      "loss": 0.2442,
      "step": 15625
    },
    {
      "epoch": 0.23048261439448609,
      "grad_norm": 1.2218703031539917,
      "learning_rate": 9.216812465023123e-06,
      "loss": 0.2312,
      "step": 15650
    },
    {
      "epoch": 0.23085079748457313,
      "grad_norm": 1.2991282939910889,
      "learning_rate": 9.231539571735736e-06,
      "loss": 0.2456,
      "step": 15675
    },
    {
      "epoch": 0.23121898057466017,
      "grad_norm": 1.34254789352417,
      "learning_rate": 9.246266678448353e-06,
      "loss": 0.2426,
      "step": 15700
    },
    {
      "epoch": 0.23158716366474721,
      "grad_norm": 1.237637996673584,
      "learning_rate": 9.260993785160967e-06,
      "loss": 0.2461,
      "step": 15725
    },
    {
      "epoch": 0.23195534675483426,
      "grad_norm": 1.3032089471817017,
      "learning_rate": 9.275720891873584e-06,
      "loss": 0.2367,
      "step": 15750
    },
    {
      "epoch": 0.23232352984492127,
      "grad_norm": 1.2366684675216675,
      "learning_rate": 9.290447998586198e-06,
      "loss": 0.2374,
      "step": 15775
    },
    {
      "epoch": 0.23269171293500832,
      "grad_norm": 1.2915616035461426,
      "learning_rate": 9.305175105298813e-06,
      "loss": 0.2401,
      "step": 15800
    },
    {
      "epoch": 0.23305989602509536,
      "grad_norm": 1.1223536729812622,
      "learning_rate": 9.31990221201143e-06,
      "loss": 0.2409,
      "step": 15825
    },
    {
      "epoch": 0.2334280791151824,
      "grad_norm": 1.1354379653930664,
      "learning_rate": 9.334629318724044e-06,
      "loss": 0.2441,
      "step": 15850
    },
    {
      "epoch": 0.23379626220526945,
      "grad_norm": 1.1635915040969849,
      "learning_rate": 9.34935642543666e-06,
      "loss": 0.2401,
      "step": 15875
    },
    {
      "epoch": 0.2341644452953565,
      "grad_norm": 1.0397142171859741,
      "learning_rate": 9.364083532149275e-06,
      "loss": 0.2406,
      "step": 15900
    },
    {
      "epoch": 0.2345326283854435,
      "grad_norm": 1.171473741531372,
      "learning_rate": 9.37881063886189e-06,
      "loss": 0.2399,
      "step": 15925
    },
    {
      "epoch": 0.23490081147553055,
      "grad_norm": 1.1377612352371216,
      "learning_rate": 9.393537745574504e-06,
      "loss": 0.24,
      "step": 15950
    },
    {
      "epoch": 0.2352689945656176,
      "grad_norm": 1.2155762910842896,
      "learning_rate": 9.40826485228712e-06,
      "loss": 0.2397,
      "step": 15975
    },
    {
      "epoch": 0.23563717765570463,
      "grad_norm": 1.1246799230575562,
      "learning_rate": 9.422991958999735e-06,
      "loss": 0.2472,
      "step": 16000
    },
    {
      "epoch": 0.23600536074579168,
      "grad_norm": 1.2643800973892212,
      "learning_rate": 9.43771906571235e-06,
      "loss": 0.2445,
      "step": 16025
    },
    {
      "epoch": 0.23637354383587872,
      "grad_norm": 0.971194863319397,
      "learning_rate": 9.452446172424966e-06,
      "loss": 0.2441,
      "step": 16050
    },
    {
      "epoch": 0.23674172692596573,
      "grad_norm": 1.0318214893341064,
      "learning_rate": 9.467173279137581e-06,
      "loss": 0.2379,
      "step": 16075
    },
    {
      "epoch": 0.23710991001605278,
      "grad_norm": 0.9700537919998169,
      "learning_rate": 9.481900385850197e-06,
      "loss": 0.2447,
      "step": 16100
    },
    {
      "epoch": 0.23747809310613982,
      "grad_norm": 1.1324304342269897,
      "learning_rate": 9.496627492562812e-06,
      "loss": 0.233,
      "step": 16125
    },
    {
      "epoch": 0.23784627619622686,
      "grad_norm": 1.808837652206421,
      "learning_rate": 9.511354599275428e-06,
      "loss": 0.2326,
      "step": 16150
    },
    {
      "epoch": 0.2382144592863139,
      "grad_norm": 1.1952983140945435,
      "learning_rate": 9.526081705988043e-06,
      "loss": 0.238,
      "step": 16175
    },
    {
      "epoch": 0.23858264237640095,
      "grad_norm": 1.0128469467163086,
      "learning_rate": 9.540808812700658e-06,
      "loss": 0.2368,
      "step": 16200
    },
    {
      "epoch": 0.23895082546648796,
      "grad_norm": 1.2425153255462646,
      "learning_rate": 9.555535919413272e-06,
      "loss": 0.2339,
      "step": 16225
    },
    {
      "epoch": 0.239319008556575,
      "grad_norm": 1.3166993856430054,
      "learning_rate": 9.570263026125889e-06,
      "loss": 0.2289,
      "step": 16250
    },
    {
      "epoch": 0.23968719164666205,
      "grad_norm": 1.0847030878067017,
      "learning_rate": 9.584990132838503e-06,
      "loss": 0.2375,
      "step": 16275
    },
    {
      "epoch": 0.2400553747367491,
      "grad_norm": 1.2066354751586914,
      "learning_rate": 9.599717239551118e-06,
      "loss": 0.2348,
      "step": 16300
    },
    {
      "epoch": 0.24042355782683614,
      "grad_norm": 1.2283575534820557,
      "learning_rate": 9.614444346263734e-06,
      "loss": 0.2279,
      "step": 16325
    },
    {
      "epoch": 0.24079174091692318,
      "grad_norm": 1.2124601602554321,
      "learning_rate": 9.629171452976349e-06,
      "loss": 0.2387,
      "step": 16350
    },
    {
      "epoch": 0.2411599240070102,
      "grad_norm": 1.1417065858840942,
      "learning_rate": 9.643898559688965e-06,
      "loss": 0.2366,
      "step": 16375
    },
    {
      "epoch": 0.24152810709709724,
      "grad_norm": 1.068881630897522,
      "learning_rate": 9.65862566640158e-06,
      "loss": 0.2297,
      "step": 16400
    },
    {
      "epoch": 0.24189629018718428,
      "grad_norm": 0.9088355302810669,
      "learning_rate": 9.673352773114195e-06,
      "loss": 0.2348,
      "step": 16425
    },
    {
      "epoch": 0.24226447327727132,
      "grad_norm": 1.6773308515548706,
      "learning_rate": 9.688079879826811e-06,
      "loss": 0.2396,
      "step": 16450
    },
    {
      "epoch": 0.24263265636735837,
      "grad_norm": 1.2640100717544556,
      "learning_rate": 9.702806986539426e-06,
      "loss": 0.2298,
      "step": 16475
    },
    {
      "epoch": 0.2430008394574454,
      "grad_norm": 1.1822272539138794,
      "learning_rate": 9.71753409325204e-06,
      "loss": 0.231,
      "step": 16500
    },
    {
      "epoch": 0.24336902254753243,
      "grad_norm": 1.2628257274627686,
      "learning_rate": 9.732261199964655e-06,
      "loss": 0.2366,
      "step": 16525
    },
    {
      "epoch": 0.24373720563761947,
      "grad_norm": 1.206954002380371,
      "learning_rate": 9.746988306677271e-06,
      "loss": 0.2287,
      "step": 16550
    },
    {
      "epoch": 0.2441053887277065,
      "grad_norm": 1.3411674499511719,
      "learning_rate": 9.761126329121382e-06,
      "loss": 0.2417,
      "step": 16575
    },
    {
      "epoch": 0.24447357181779356,
      "grad_norm": 1.1334586143493652,
      "learning_rate": 9.775853435833997e-06,
      "loss": 0.224,
      "step": 16600
    },
    {
      "epoch": 0.2448417549078806,
      "grad_norm": 1.0482016801834106,
      "learning_rate": 9.790580542546611e-06,
      "loss": 0.2342,
      "step": 16625
    },
    {
      "epoch": 0.24520993799796764,
      "grad_norm": 0.9900887608528137,
      "learning_rate": 9.805307649259228e-06,
      "loss": 0.2396,
      "step": 16650
    },
    {
      "epoch": 0.24557812108805466,
      "grad_norm": 1.1004738807678223,
      "learning_rate": 9.820034755971842e-06,
      "loss": 0.2219,
      "step": 16675
    },
    {
      "epoch": 0.2459463041781417,
      "grad_norm": 1.236008644104004,
      "learning_rate": 9.834761862684457e-06,
      "loss": 0.2248,
      "step": 16700
    },
    {
      "epoch": 0.24631448726822874,
      "grad_norm": 1.283540964126587,
      "learning_rate": 9.849488969397073e-06,
      "loss": 0.2269,
      "step": 16725
    },
    {
      "epoch": 0.2466826703583158,
      "grad_norm": 1.0043452978134155,
      "learning_rate": 9.864216076109688e-06,
      "loss": 0.232,
      "step": 16750
    },
    {
      "epoch": 0.24705085344840283,
      "grad_norm": 1.3650057315826416,
      "learning_rate": 9.878943182822304e-06,
      "loss": 0.2273,
      "step": 16775
    },
    {
      "epoch": 0.24741903653848987,
      "grad_norm": 0.9510201811790466,
      "learning_rate": 9.893670289534919e-06,
      "loss": 0.226,
      "step": 16800
    },
    {
      "epoch": 0.2477872196285769,
      "grad_norm": 1.5678138732910156,
      "learning_rate": 9.908397396247534e-06,
      "loss": 0.224,
      "step": 16825
    },
    {
      "epoch": 0.24815540271866393,
      "grad_norm": 1.2678954601287842,
      "learning_rate": 9.92312450296015e-06,
      "loss": 0.2114,
      "step": 16850
    },
    {
      "epoch": 0.24852358580875097,
      "grad_norm": 1.2294344902038574,
      "learning_rate": 9.937851609672765e-06,
      "loss": 0.2325,
      "step": 16875
    },
    {
      "epoch": 0.24889176889883802,
      "grad_norm": 1.084308385848999,
      "learning_rate": 9.95257871638538e-06,
      "loss": 0.2284,
      "step": 16900
    },
    {
      "epoch": 0.24925995198892506,
      "grad_norm": 1.3282686471939087,
      "learning_rate": 9.967305823097994e-06,
      "loss": 0.2193,
      "step": 16925
    },
    {
      "epoch": 0.2496281350790121,
      "grad_norm": 1.0947468280792236,
      "learning_rate": 9.98203292981061e-06,
      "loss": 0.2252,
      "step": 16950
    },
    {
      "epoch": 0.24999631816909912,
      "grad_norm": 1.2787188291549683,
      "learning_rate": 9.996760036523225e-06,
      "loss": 0.2271,
      "step": 16975
    },
    {
      "epoch": 0.2503645012591862,
      "grad_norm": 1.179134726524353,
      "learning_rate": 1.0011487143235841e-05,
      "loss": 0.2223,
      "step": 17000
    },
    {
      "epoch": 0.2507326843492732,
      "grad_norm": 1.1341197490692139,
      "learning_rate": 1.0026214249948456e-05,
      "loss": 0.22,
      "step": 17025
    },
    {
      "epoch": 0.2511008674393602,
      "grad_norm": 0.9111505150794983,
      "learning_rate": 1.004094135666107e-05,
      "loss": 0.2275,
      "step": 17050
    },
    {
      "epoch": 0.2514690505294473,
      "grad_norm": 1.2634352445602417,
      "learning_rate": 1.0055668463373687e-05,
      "loss": 0.2317,
      "step": 17075
    },
    {
      "epoch": 0.2518372336195343,
      "grad_norm": 1.4474118947982788,
      "learning_rate": 1.0070395570086302e-05,
      "loss": 0.2208,
      "step": 17100
    },
    {
      "epoch": 0.2522054167096214,
      "grad_norm": 1.125046968460083,
      "learning_rate": 1.0085122676798918e-05,
      "loss": 0.2181,
      "step": 17125
    },
    {
      "epoch": 0.2525735997997084,
      "grad_norm": 1.0893241167068481,
      "learning_rate": 1.0099849783511533e-05,
      "loss": 0.2251,
      "step": 17150
    },
    {
      "epoch": 0.25294178288979546,
      "grad_norm": 1.244471549987793,
      "learning_rate": 1.0114576890224147e-05,
      "loss": 0.2225,
      "step": 17175
    },
    {
      "epoch": 0.2533099659798825,
      "grad_norm": 1.1279207468032837,
      "learning_rate": 1.0129303996936764e-05,
      "loss": 0.2277,
      "step": 17200
    },
    {
      "epoch": 0.2536781490699695,
      "grad_norm": 1.1421986818313599,
      "learning_rate": 1.0144031103649378e-05,
      "loss": 0.2207,
      "step": 17225
    },
    {
      "epoch": 0.25404633216005656,
      "grad_norm": 1.43101966381073,
      "learning_rate": 1.0158758210361995e-05,
      "loss": 0.2279,
      "step": 17250
    },
    {
      "epoch": 0.2544145152501436,
      "grad_norm": 1.2407532930374146,
      "learning_rate": 1.0173485317074608e-05,
      "loss": 0.2271,
      "step": 17275
    },
    {
      "epoch": 0.25478269834023065,
      "grad_norm": 1.3361932039260864,
      "learning_rate": 1.0188212423787222e-05,
      "loss": 0.2239,
      "step": 17300
    },
    {
      "epoch": 0.25515088143031767,
      "grad_norm": 1.0362130403518677,
      "learning_rate": 1.0202939530499839e-05,
      "loss": 0.2252,
      "step": 17325
    },
    {
      "epoch": 0.2555190645204047,
      "grad_norm": 0.9963209629058838,
      "learning_rate": 1.0217666637212453e-05,
      "loss": 0.221,
      "step": 17350
    },
    {
      "epoch": 0.25588724761049175,
      "grad_norm": 0.906834065914154,
      "learning_rate": 1.0232393743925068e-05,
      "loss": 0.2217,
      "step": 17375
    },
    {
      "epoch": 0.25625543070057877,
      "grad_norm": 1.3475807905197144,
      "learning_rate": 1.0247120850637684e-05,
      "loss": 0.2226,
      "step": 17400
    },
    {
      "epoch": 0.25662361379066584,
      "grad_norm": 1.1310131549835205,
      "learning_rate": 1.0261847957350299e-05,
      "loss": 0.2259,
      "step": 17425
    },
    {
      "epoch": 0.25699179688075285,
      "grad_norm": 1.1497154235839844,
      "learning_rate": 1.0276575064062915e-05,
      "loss": 0.2163,
      "step": 17450
    },
    {
      "epoch": 0.2573599799708399,
      "grad_norm": 1.5212485790252686,
      "learning_rate": 1.029130217077553e-05,
      "loss": 0.2284,
      "step": 17475
    },
    {
      "epoch": 0.25772816306092694,
      "grad_norm": 1.4213180541992188,
      "learning_rate": 1.0306029277488146e-05,
      "loss": 0.2152,
      "step": 17500
    },
    {
      "epoch": 0.25809634615101396,
      "grad_norm": 1.1019536256790161,
      "learning_rate": 1.0320756384200761e-05,
      "loss": 0.2236,
      "step": 17525
    },
    {
      "epoch": 0.258464529241101,
      "grad_norm": 1.2373889684677124,
      "learning_rate": 1.0335483490913376e-05,
      "loss": 0.2205,
      "step": 17550
    },
    {
      "epoch": 0.25883271233118804,
      "grad_norm": 1.1303311586380005,
      "learning_rate": 1.0350210597625992e-05,
      "loss": 0.2163,
      "step": 17575
    },
    {
      "epoch": 0.2592008954212751,
      "grad_norm": 1.8989242315292358,
      "learning_rate": 1.0364937704338607e-05,
      "loss": 0.2155,
      "step": 17600
    },
    {
      "epoch": 0.25956907851136213,
      "grad_norm": 1.081231951713562,
      "learning_rate": 1.0379664811051223e-05,
      "loss": 0.2182,
      "step": 17625
    },
    {
      "epoch": 0.25993726160144914,
      "grad_norm": 1.2433748245239258,
      "learning_rate": 1.0394391917763838e-05,
      "loss": 0.2191,
      "step": 17650
    },
    {
      "epoch": 0.2603054446915362,
      "grad_norm": 1.1952991485595703,
      "learning_rate": 1.0409119024476452e-05,
      "loss": 0.2188,
      "step": 17675
    },
    {
      "epoch": 0.26067362778162323,
      "grad_norm": 1.2529122829437256,
      "learning_rate": 1.0423846131189069e-05,
      "loss": 0.2223,
      "step": 17700
    },
    {
      "epoch": 0.2610418108717103,
      "grad_norm": 1.153318166732788,
      "learning_rate": 1.0438573237901683e-05,
      "loss": 0.214,
      "step": 17725
    },
    {
      "epoch": 0.2614099939617973,
      "grad_norm": 1.440818190574646,
      "learning_rate": 1.04533003446143e-05,
      "loss": 0.2119,
      "step": 17750
    },
    {
      "epoch": 0.2617781770518844,
      "grad_norm": 1.3393558263778687,
      "learning_rate": 1.0468027451326914e-05,
      "loss": 0.2198,
      "step": 17775
    },
    {
      "epoch": 0.2621463601419714,
      "grad_norm": 1.0429470539093018,
      "learning_rate": 1.0482754558039527e-05,
      "loss": 0.2162,
      "step": 17800
    },
    {
      "epoch": 0.2625145432320584,
      "grad_norm": 1.112441062927246,
      "learning_rate": 1.0497481664752144e-05,
      "loss": 0.2195,
      "step": 17825
    },
    {
      "epoch": 0.2628827263221455,
      "grad_norm": 1.2542566061019897,
      "learning_rate": 1.0512208771464758e-05,
      "loss": 0.2164,
      "step": 17850
    },
    {
      "epoch": 0.2632509094122325,
      "grad_norm": 1.3128899335861206,
      "learning_rate": 1.0526935878177373e-05,
      "loss": 0.2223,
      "step": 17875
    },
    {
      "epoch": 0.2636190925023196,
      "grad_norm": 1.1940679550170898,
      "learning_rate": 1.054166298488999e-05,
      "loss": 0.2173,
      "step": 17900
    },
    {
      "epoch": 0.2639872755924066,
      "grad_norm": 1.258396029472351,
      "learning_rate": 1.0556390091602604e-05,
      "loss": 0.2074,
      "step": 17925
    },
    {
      "epoch": 0.2643554586824936,
      "grad_norm": 1.1842387914657593,
      "learning_rate": 1.057111719831522e-05,
      "loss": 0.2157,
      "step": 17950
    },
    {
      "epoch": 0.2647236417725807,
      "grad_norm": 1.0509520769119263,
      "learning_rate": 1.0585844305027835e-05,
      "loss": 0.2129,
      "step": 17975
    },
    {
      "epoch": 0.2650918248626677,
      "grad_norm": 1.2387256622314453,
      "learning_rate": 1.060057141174045e-05,
      "loss": 0.2239,
      "step": 18000
    },
    {
      "epoch": 0.26546000795275476,
      "grad_norm": 0.9733960628509521,
      "learning_rate": 1.0615298518453066e-05,
      "loss": 0.2179,
      "step": 18025
    },
    {
      "epoch": 0.2658281910428418,
      "grad_norm": 1.2037309408187866,
      "learning_rate": 1.063002562516568e-05,
      "loss": 0.215,
      "step": 18050
    },
    {
      "epoch": 0.26619637413292885,
      "grad_norm": 1.305167555809021,
      "learning_rate": 1.0644752731878297e-05,
      "loss": 0.2156,
      "step": 18075
    },
    {
      "epoch": 0.26656455722301586,
      "grad_norm": 1.1345758438110352,
      "learning_rate": 1.0659479838590911e-05,
      "loss": 0.2119,
      "step": 18100
    },
    {
      "epoch": 0.2669327403131029,
      "grad_norm": 1.0518168210983276,
      "learning_rate": 1.0674206945303526e-05,
      "loss": 0.2077,
      "step": 18125
    },
    {
      "epoch": 0.26730092340318995,
      "grad_norm": 1.3454447984695435,
      "learning_rate": 1.0688934052016142e-05,
      "loss": 0.2117,
      "step": 18150
    },
    {
      "epoch": 0.26766910649327696,
      "grad_norm": 1.540737509727478,
      "learning_rate": 1.0703661158728757e-05,
      "loss": 0.2098,
      "step": 18175
    },
    {
      "epoch": 0.26803728958336404,
      "grad_norm": 1.0186264514923096,
      "learning_rate": 1.0718388265441373e-05,
      "loss": 0.2106,
      "step": 18200
    },
    {
      "epoch": 0.26840547267345105,
      "grad_norm": 1.2026820182800293,
      "learning_rate": 1.0733115372153988e-05,
      "loss": 0.206,
      "step": 18225
    },
    {
      "epoch": 0.26877365576353807,
      "grad_norm": 0.9871214628219604,
      "learning_rate": 1.0747842478866603e-05,
      "loss": 0.2063,
      "step": 18250
    },
    {
      "epoch": 0.26914183885362514,
      "grad_norm": 1.1501942873001099,
      "learning_rate": 1.0762569585579219e-05,
      "loss": 0.2125,
      "step": 18275
    },
    {
      "epoch": 0.26951002194371215,
      "grad_norm": 1.422052025794983,
      "learning_rate": 1.0777296692291834e-05,
      "loss": 0.2144,
      "step": 18300
    },
    {
      "epoch": 0.2698782050337992,
      "grad_norm": 1.3499372005462646,
      "learning_rate": 1.0792023799004447e-05,
      "loss": 0.2091,
      "step": 18325
    },
    {
      "epoch": 0.27024638812388624,
      "grad_norm": 1.1054781675338745,
      "learning_rate": 1.0806750905717063e-05,
      "loss": 0.2116,
      "step": 18350
    },
    {
      "epoch": 0.2706145712139733,
      "grad_norm": 1.160321831703186,
      "learning_rate": 1.0821478012429678e-05,
      "loss": 0.2091,
      "step": 18375
    },
    {
      "epoch": 0.2709827543040603,
      "grad_norm": 0.9781479835510254,
      "learning_rate": 1.0836205119142294e-05,
      "loss": 0.2124,
      "step": 18400
    },
    {
      "epoch": 0.27135093739414734,
      "grad_norm": 1.2315099239349365,
      "learning_rate": 1.0850932225854909e-05,
      "loss": 0.212,
      "step": 18425
    },
    {
      "epoch": 0.2717191204842344,
      "grad_norm": 1.3107922077178955,
      "learning_rate": 1.0865659332567523e-05,
      "loss": 0.2042,
      "step": 18450
    },
    {
      "epoch": 0.2720873035743214,
      "grad_norm": 1.2262879610061646,
      "learning_rate": 1.088038643928014e-05,
      "loss": 0.2135,
      "step": 18475
    },
    {
      "epoch": 0.2724554866644085,
      "grad_norm": 1.409114956855774,
      "learning_rate": 1.0895113545992754e-05,
      "loss": 0.2138,
      "step": 18500
    },
    {
      "epoch": 0.2728236697544955,
      "grad_norm": 1.0485094785690308,
      "learning_rate": 1.090984065270537e-05,
      "loss": 0.2039,
      "step": 18525
    },
    {
      "epoch": 0.27319185284458253,
      "grad_norm": 1.374588131904602,
      "learning_rate": 1.0924567759417985e-05,
      "loss": 0.204,
      "step": 18550
    },
    {
      "epoch": 0.2735600359346696,
      "grad_norm": 0.9809952974319458,
      "learning_rate": 1.0939294866130602e-05,
      "loss": 0.2088,
      "step": 18575
    },
    {
      "epoch": 0.2739282190247566,
      "grad_norm": 1.3115181922912598,
      "learning_rate": 1.0954021972843216e-05,
      "loss": 0.1981,
      "step": 18600
    },
    {
      "epoch": 0.2742964021148437,
      "grad_norm": 1.44467031955719,
      "learning_rate": 1.0968749079555831e-05,
      "loss": 0.2063,
      "step": 18625
    },
    {
      "epoch": 0.2746645852049307,
      "grad_norm": 0.9407679438591003,
      "learning_rate": 1.0982887101999942e-05,
      "loss": 0.2032,
      "step": 18650
    },
    {
      "epoch": 0.27503276829501777,
      "grad_norm": 1.440572738647461,
      "learning_rate": 1.0997614208712557e-05,
      "loss": 0.2056,
      "step": 18675
    },
    {
      "epoch": 0.2754009513851048,
      "grad_norm": 1.6290857791900635,
      "learning_rate": 1.1012341315425173e-05,
      "loss": 0.2078,
      "step": 18700
    },
    {
      "epoch": 0.2757691344751918,
      "grad_norm": 1.3203442096710205,
      "learning_rate": 1.1027068422137788e-05,
      "loss": 0.2107,
      "step": 18725
    },
    {
      "epoch": 0.2761373175652789,
      "grad_norm": 1.2521476745605469,
      "learning_rate": 1.1041795528850404e-05,
      "loss": 0.207,
      "step": 18750
    },
    {
      "epoch": 0.2765055006553659,
      "grad_norm": 1.4565852880477905,
      "learning_rate": 1.1056522635563019e-05,
      "loss": 0.2045,
      "step": 18775
    },
    {
      "epoch": 0.27687368374545296,
      "grad_norm": 1.0191683769226074,
      "learning_rate": 1.1071249742275633e-05,
      "loss": 0.2008,
      "step": 18800
    },
    {
      "epoch": 0.27724186683554,
      "grad_norm": 1.4495227336883545,
      "learning_rate": 1.108597684898825e-05,
      "loss": 0.2013,
      "step": 18825
    },
    {
      "epoch": 0.277610049925627,
      "grad_norm": 1.3123514652252197,
      "learning_rate": 1.1100703955700864e-05,
      "loss": 0.2046,
      "step": 18850
    },
    {
      "epoch": 0.27797823301571406,
      "grad_norm": 1.053816318511963,
      "learning_rate": 1.111543106241348e-05,
      "loss": 0.2096,
      "step": 18875
    },
    {
      "epoch": 0.2783464161058011,
      "grad_norm": 0.9499918222427368,
      "learning_rate": 1.1130158169126095e-05,
      "loss": 0.1954,
      "step": 18900
    },
    {
      "epoch": 0.27871459919588815,
      "grad_norm": 1.004050850868225,
      "learning_rate": 1.114488527583871e-05,
      "loss": 0.2021,
      "step": 18925
    },
    {
      "epoch": 0.27908278228597516,
      "grad_norm": 1.1039396524429321,
      "learning_rate": 1.1159612382551326e-05,
      "loss": 0.2081,
      "step": 18950
    },
    {
      "epoch": 0.27945096537606223,
      "grad_norm": 1.31987726688385,
      "learning_rate": 1.1174339489263941e-05,
      "loss": 0.1875,
      "step": 18975
    },
    {
      "epoch": 0.27981914846614925,
      "grad_norm": 1.3621447086334229,
      "learning_rate": 1.1189066595976554e-05,
      "loss": 0.2082,
      "step": 19000
    },
    {
      "epoch": 0.28018733155623626,
      "grad_norm": 1.6277425289154053,
      "learning_rate": 1.120379370268917e-05,
      "loss": 0.207,
      "step": 19025
    },
    {
      "epoch": 0.28055551464632333,
      "grad_norm": 1.0901118516921997,
      "learning_rate": 1.1218520809401785e-05,
      "loss": 0.2028,
      "step": 19050
    },
    {
      "epoch": 0.28092369773641035,
      "grad_norm": 1.1058247089385986,
      "learning_rate": 1.1233247916114401e-05,
      "loss": 0.2122,
      "step": 19075
    },
    {
      "epoch": 0.2812918808264974,
      "grad_norm": 1.084421157836914,
      "learning_rate": 1.1247975022827016e-05,
      "loss": 0.2068,
      "step": 19100
    },
    {
      "epoch": 0.28166006391658444,
      "grad_norm": 1.2624268531799316,
      "learning_rate": 1.126270212953963e-05,
      "loss": 0.1984,
      "step": 19125
    },
    {
      "epoch": 0.28202824700667145,
      "grad_norm": 1.3556143045425415,
      "learning_rate": 1.1277429236252247e-05,
      "loss": 0.2064,
      "step": 19150
    },
    {
      "epoch": 0.2823964300967585,
      "grad_norm": 1.1427866220474243,
      "learning_rate": 1.1292156342964861e-05,
      "loss": 0.2007,
      "step": 19175
    },
    {
      "epoch": 0.28276461318684554,
      "grad_norm": 1.3310309648513794,
      "learning_rate": 1.1306883449677478e-05,
      "loss": 0.2041,
      "step": 19200
    },
    {
      "epoch": 0.2831327962769326,
      "grad_norm": 1.4609546661376953,
      "learning_rate": 1.1321610556390092e-05,
      "loss": 0.2063,
      "step": 19225
    },
    {
      "epoch": 0.2835009793670196,
      "grad_norm": 1.464328408241272,
      "learning_rate": 1.1336337663102707e-05,
      "loss": 0.2053,
      "step": 19250
    },
    {
      "epoch": 0.2838691624571067,
      "grad_norm": 1.2487759590148926,
      "learning_rate": 1.1351064769815323e-05,
      "loss": 0.2109,
      "step": 19275
    },
    {
      "epoch": 0.2842373455471937,
      "grad_norm": 1.0267369747161865,
      "learning_rate": 1.1365791876527938e-05,
      "loss": 0.1941,
      "step": 19300
    },
    {
      "epoch": 0.2846055286372807,
      "grad_norm": 1.5034706592559814,
      "learning_rate": 1.1380518983240554e-05,
      "loss": 0.2062,
      "step": 19325
    },
    {
      "epoch": 0.2849737117273678,
      "grad_norm": 1.3053853511810303,
      "learning_rate": 1.1395246089953169e-05,
      "loss": 0.2088,
      "step": 19350
    },
    {
      "epoch": 0.2853418948174548,
      "grad_norm": 1.1365464925765991,
      "learning_rate": 1.1409973196665784e-05,
      "loss": 0.2091,
      "step": 19375
    },
    {
      "epoch": 0.2857100779075419,
      "grad_norm": 1.1613692045211792,
      "learning_rate": 1.14247003033784e-05,
      "loss": 0.2008,
      "step": 19400
    },
    {
      "epoch": 0.2860782609976289,
      "grad_norm": 1.2385408878326416,
      "learning_rate": 1.1439427410091015e-05,
      "loss": 0.2009,
      "step": 19425
    },
    {
      "epoch": 0.2864464440877159,
      "grad_norm": 1.4164599180221558,
      "learning_rate": 1.1454154516803631e-05,
      "loss": 0.196,
      "step": 19450
    },
    {
      "epoch": 0.286814627177803,
      "grad_norm": 1.0749568939208984,
      "learning_rate": 1.1468881623516246e-05,
      "loss": 0.2048,
      "step": 19475
    },
    {
      "epoch": 0.28718281026789,
      "grad_norm": 1.2188838720321655,
      "learning_rate": 1.148360873022886e-05,
      "loss": 0.2047,
      "step": 19500
    },
    {
      "epoch": 0.28755099335797707,
      "grad_norm": 1.2900882959365845,
      "learning_rate": 1.1498335836941475e-05,
      "loss": 0.2046,
      "step": 19525
    },
    {
      "epoch": 0.2879191764480641,
      "grad_norm": 1.3515424728393555,
      "learning_rate": 1.151306294365409e-05,
      "loss": 0.2013,
      "step": 19550
    },
    {
      "epoch": 0.28828735953815116,
      "grad_norm": 1.1831262111663818,
      "learning_rate": 1.1527790050366704e-05,
      "loss": 0.196,
      "step": 19575
    },
    {
      "epoch": 0.28865554262823817,
      "grad_norm": 1.2154779434204102,
      "learning_rate": 1.154251715707932e-05,
      "loss": 0.2001,
      "step": 19600
    },
    {
      "epoch": 0.2890237257183252,
      "grad_norm": 1.0575876235961914,
      "learning_rate": 1.1557244263791935e-05,
      "loss": 0.198,
      "step": 19625
    },
    {
      "epoch": 0.28939190880841226,
      "grad_norm": 1.2004785537719727,
      "learning_rate": 1.1571971370504552e-05,
      "loss": 0.1928,
      "step": 19650
    },
    {
      "epoch": 0.2897600918984993,
      "grad_norm": 1.2369384765625,
      "learning_rate": 1.1586698477217166e-05,
      "loss": 0.1973,
      "step": 19675
    },
    {
      "epoch": 0.29012827498858634,
      "grad_norm": 1.194305419921875,
      "learning_rate": 1.1601425583929781e-05,
      "loss": 0.1994,
      "step": 19700
    },
    {
      "epoch": 0.29049645807867336,
      "grad_norm": 1.2108451128005981,
      "learning_rate": 1.1616152690642397e-05,
      "loss": 0.1964,
      "step": 19725
    },
    {
      "epoch": 0.2908646411687604,
      "grad_norm": 1.3389203548431396,
      "learning_rate": 1.1630879797355012e-05,
      "loss": 0.2059,
      "step": 19750
    },
    {
      "epoch": 0.29123282425884744,
      "grad_norm": 1.416451334953308,
      "learning_rate": 1.1645606904067628e-05,
      "loss": 0.2053,
      "step": 19775
    },
    {
      "epoch": 0.29160100734893446,
      "grad_norm": 1.76418137550354,
      "learning_rate": 1.1660334010780243e-05,
      "loss": 0.1958,
      "step": 19800
    },
    {
      "epoch": 0.29196919043902153,
      "grad_norm": 1.3602815866470337,
      "learning_rate": 1.167506111749286e-05,
      "loss": 0.2086,
      "step": 19825
    },
    {
      "epoch": 0.29233737352910855,
      "grad_norm": 1.4478594064712524,
      "learning_rate": 1.1689788224205474e-05,
      "loss": 0.1931,
      "step": 19850
    },
    {
      "epoch": 0.2927055566191956,
      "grad_norm": 1.269147276878357,
      "learning_rate": 1.1704515330918089e-05,
      "loss": 0.1954,
      "step": 19875
    },
    {
      "epoch": 0.29307373970928263,
      "grad_norm": 1.711710810661316,
      "learning_rate": 1.1719242437630705e-05,
      "loss": 0.1935,
      "step": 19900
    },
    {
      "epoch": 0.29344192279936965,
      "grad_norm": 1.4398845434188843,
      "learning_rate": 1.173396954434332e-05,
      "loss": 0.1955,
      "step": 19925
    },
    {
      "epoch": 0.2938101058894567,
      "grad_norm": 1.256732702255249,
      "learning_rate": 1.1748696651055936e-05,
      "loss": 0.1964,
      "step": 19950
    },
    {
      "epoch": 0.29417828897954373,
      "grad_norm": 1.09647798538208,
      "learning_rate": 1.176342375776855e-05,
      "loss": 0.1977,
      "step": 19975
    },
    {
      "epoch": 0.2945464720696308,
      "grad_norm": 1.3230574131011963,
      "learning_rate": 1.1778150864481165e-05,
      "loss": 0.1933,
      "step": 20000
    },
    {
      "epoch": 0.2949146551597178,
      "grad_norm": 1.3102632761001587,
      "learning_rate": 1.1792877971193782e-05,
      "loss": 0.1913,
      "step": 20025
    },
    {
      "epoch": 0.29528283824980484,
      "grad_norm": 1.1603052616119385,
      "learning_rate": 1.1807605077906395e-05,
      "loss": 0.1923,
      "step": 20050
    },
    {
      "epoch": 0.2956510213398919,
      "grad_norm": 1.3793220520019531,
      "learning_rate": 1.182233218461901e-05,
      "loss": 0.193,
      "step": 20075
    },
    {
      "epoch": 0.2960192044299789,
      "grad_norm": 1.0161463022232056,
      "learning_rate": 1.1837059291331626e-05,
      "loss": 0.2002,
      "step": 20100
    },
    {
      "epoch": 0.296387387520066,
      "grad_norm": 1.4899036884307861,
      "learning_rate": 1.185178639804424e-05,
      "loss": 0.2016,
      "step": 20125
    },
    {
      "epoch": 0.296755570610153,
      "grad_norm": 1.4574910402297974,
      "learning_rate": 1.1866513504756857e-05,
      "loss": 0.197,
      "step": 20150
    },
    {
      "epoch": 0.2971237537002401,
      "grad_norm": 1.5206778049468994,
      "learning_rate": 1.1881240611469471e-05,
      "loss": 0.1973,
      "step": 20175
    },
    {
      "epoch": 0.2974919367903271,
      "grad_norm": 1.0904589891433716,
      "learning_rate": 1.1895967718182086e-05,
      "loss": 0.1965,
      "step": 20200
    },
    {
      "epoch": 0.2978601198804141,
      "grad_norm": 1.367783546447754,
      "learning_rate": 1.1910694824894702e-05,
      "loss": 0.2014,
      "step": 20225
    },
    {
      "epoch": 0.2982283029705012,
      "grad_norm": 1.2343593835830688,
      "learning_rate": 1.1925421931607317e-05,
      "loss": 0.1982,
      "step": 20250
    },
    {
      "epoch": 0.2985964860605882,
      "grad_norm": 1.2906816005706787,
      "learning_rate": 1.1940149038319933e-05,
      "loss": 0.2008,
      "step": 20275
    },
    {
      "epoch": 0.29896466915067527,
      "grad_norm": 1.178275227546692,
      "learning_rate": 1.1954876145032548e-05,
      "loss": 0.1955,
      "step": 20300
    },
    {
      "epoch": 0.2993328522407623,
      "grad_norm": 1.4055285453796387,
      "learning_rate": 1.1969603251745163e-05,
      "loss": 0.2023,
      "step": 20325
    },
    {
      "epoch": 0.29970103533084935,
      "grad_norm": 1.0455254316329956,
      "learning_rate": 1.1984330358457779e-05,
      "loss": 0.1942,
      "step": 20350
    },
    {
      "epoch": 0.30006921842093637,
      "grad_norm": 1.0922812223434448,
      "learning_rate": 1.1999057465170394e-05,
      "loss": 0.1988,
      "step": 20375
    },
    {
      "epoch": 0.3004374015110234,
      "grad_norm": 0.9248048067092896,
      "learning_rate": 1.201378457188301e-05,
      "loss": 0.1909,
      "step": 20400
    },
    {
      "epoch": 0.30080558460111045,
      "grad_norm": 1.0108305215835571,
      "learning_rate": 1.2028511678595625e-05,
      "loss": 0.1946,
      "step": 20425
    },
    {
      "epoch": 0.30117376769119747,
      "grad_norm": 1.4391227960586548,
      "learning_rate": 1.204323878530824e-05,
      "loss": 0.1943,
      "step": 20450
    },
    {
      "epoch": 0.30154195078128454,
      "grad_norm": 1.42784583568573,
      "learning_rate": 1.2057965892020856e-05,
      "loss": 0.1824,
      "step": 20475
    },
    {
      "epoch": 0.30191013387137156,
      "grad_norm": 1.3134140968322754,
      "learning_rate": 1.207269299873347e-05,
      "loss": 0.197,
      "step": 20500
    },
    {
      "epoch": 0.30227831696145857,
      "grad_norm": 1.3688997030258179,
      "learning_rate": 1.2087420105446087e-05,
      "loss": 0.1912,
      "step": 20525
    },
    {
      "epoch": 0.30264650005154564,
      "grad_norm": 1.3083784580230713,
      "learning_rate": 1.2102147212158701e-05,
      "loss": 0.1928,
      "step": 20550
    },
    {
      "epoch": 0.30301468314163266,
      "grad_norm": 1.4008431434631348,
      "learning_rate": 1.2116874318871314e-05,
      "loss": 0.1973,
      "step": 20575
    },
    {
      "epoch": 0.30338286623171973,
      "grad_norm": 1.101962685585022,
      "learning_rate": 1.213160142558393e-05,
      "loss": 0.1925,
      "step": 20600
    },
    {
      "epoch": 0.30375104932180674,
      "grad_norm": 1.5726611614227295,
      "learning_rate": 1.2146328532296545e-05,
      "loss": 0.194,
      "step": 20625
    },
    {
      "epoch": 0.3041192324118938,
      "grad_norm": 1.2495619058609009,
      "learning_rate": 1.216105563900916e-05,
      "loss": 0.1933,
      "step": 20650
    },
    {
      "epoch": 0.30448741550198083,
      "grad_norm": 1.1473134756088257,
      "learning_rate": 1.2175782745721776e-05,
      "loss": 0.1854,
      "step": 20675
    },
    {
      "epoch": 0.30485559859206784,
      "grad_norm": 1.277658462524414,
      "learning_rate": 1.2190509852434391e-05,
      "loss": 0.188,
      "step": 20700
    },
    {
      "epoch": 0.3052237816821549,
      "grad_norm": 1.6464298963546753,
      "learning_rate": 1.2205236959147007e-05,
      "loss": 0.1902,
      "step": 20725
    },
    {
      "epoch": 0.30559196477224193,
      "grad_norm": 1.2113037109375,
      "learning_rate": 1.2219374981591116e-05,
      "loss": 0.1941,
      "step": 20750
    },
    {
      "epoch": 0.305960147862329,
      "grad_norm": 0.9931922554969788,
      "learning_rate": 1.2234102088303733e-05,
      "loss": 0.1877,
      "step": 20775
    },
    {
      "epoch": 0.306328330952416,
      "grad_norm": 1.262056827545166,
      "learning_rate": 1.2248829195016347e-05,
      "loss": 0.1939,
      "step": 20800
    },
    {
      "epoch": 0.30669651404250303,
      "grad_norm": 1.4179117679595947,
      "learning_rate": 1.2263556301728962e-05,
      "loss": 0.1945,
      "step": 20825
    },
    {
      "epoch": 0.3070646971325901,
      "grad_norm": 1.4320852756500244,
      "learning_rate": 1.2278283408441578e-05,
      "loss": 0.194,
      "step": 20850
    },
    {
      "epoch": 0.3074328802226771,
      "grad_norm": 1.748063325881958,
      "learning_rate": 1.2293010515154193e-05,
      "loss": 0.192,
      "step": 20875
    },
    {
      "epoch": 0.3078010633127642,
      "grad_norm": 1.6274995803833008,
      "learning_rate": 1.230773762186681e-05,
      "loss": 0.1889,
      "step": 20900
    },
    {
      "epoch": 0.3081692464028512,
      "grad_norm": 1.4107574224472046,
      "learning_rate": 1.2322464728579424e-05,
      "loss": 0.201,
      "step": 20925
    },
    {
      "epoch": 0.3085374294929383,
      "grad_norm": 1.573684811592102,
      "learning_rate": 1.2337191835292039e-05,
      "loss": 0.1819,
      "step": 20950
    },
    {
      "epoch": 0.3089056125830253,
      "grad_norm": 1.3869659900665283,
      "learning_rate": 1.2351918942004655e-05,
      "loss": 0.1945,
      "step": 20975
    },
    {
      "epoch": 0.3092737956731123,
      "grad_norm": 1.1741231679916382,
      "learning_rate": 1.236664604871727e-05,
      "loss": 0.1852,
      "step": 21000
    },
    {
      "epoch": 0.3096419787631994,
      "grad_norm": 1.3392157554626465,
      "learning_rate": 1.2381373155429886e-05,
      "loss": 0.1851,
      "step": 21025
    },
    {
      "epoch": 0.3100101618532864,
      "grad_norm": 1.2029756307601929,
      "learning_rate": 1.23961002621425e-05,
      "loss": 0.1836,
      "step": 21050
    },
    {
      "epoch": 0.31037834494337346,
      "grad_norm": 1.9749929904937744,
      "learning_rate": 1.2410827368855117e-05,
      "loss": 0.1973,
      "step": 21075
    },
    {
      "epoch": 0.3107465280334605,
      "grad_norm": 1.1220290660858154,
      "learning_rate": 1.2425554475567732e-05,
      "loss": 0.1964,
      "step": 21100
    },
    {
      "epoch": 0.3111147111235475,
      "grad_norm": 1.6141853332519531,
      "learning_rate": 1.2440281582280346e-05,
      "loss": 0.1899,
      "step": 21125
    },
    {
      "epoch": 0.31148289421363456,
      "grad_norm": 1.1342378854751587,
      "learning_rate": 1.2455008688992963e-05,
      "loss": 0.1882,
      "step": 21150
    },
    {
      "epoch": 0.3118510773037216,
      "grad_norm": 1.2282766103744507,
      "learning_rate": 1.2469735795705577e-05,
      "loss": 0.1916,
      "step": 21175
    },
    {
      "epoch": 0.31221926039380865,
      "grad_norm": 1.3788354396820068,
      "learning_rate": 1.2484462902418194e-05,
      "loss": 0.1826,
      "step": 21200
    },
    {
      "epoch": 0.31258744348389567,
      "grad_norm": 1.207901120185852,
      "learning_rate": 1.2499190009130808e-05,
      "loss": 0.1897,
      "step": 21225
    },
    {
      "epoch": 0.31295562657398274,
      "grad_norm": 1.5424796342849731,
      "learning_rate": 1.2513917115843421e-05,
      "loss": 0.1829,
      "step": 21250
    },
    {
      "epoch": 0.31332380966406975,
      "grad_norm": 1.495710015296936,
      "learning_rate": 1.2528644222556036e-05,
      "loss": 0.1794,
      "step": 21275
    },
    {
      "epoch": 0.31369199275415677,
      "grad_norm": 1.404462218284607,
      "learning_rate": 1.2543371329268652e-05,
      "loss": 0.1861,
      "step": 21300
    },
    {
      "epoch": 0.31406017584424384,
      "grad_norm": 1.2983025312423706,
      "learning_rate": 1.2558098435981267e-05,
      "loss": 0.1898,
      "step": 21325
    },
    {
      "epoch": 0.31442835893433085,
      "grad_norm": 1.0166761875152588,
      "learning_rate": 1.2572825542693883e-05,
      "loss": 0.191,
      "step": 21350
    },
    {
      "epoch": 0.3147965420244179,
      "grad_norm": 1.436242699623108,
      "learning_rate": 1.2587552649406498e-05,
      "loss": 0.1896,
      "step": 21375
    },
    {
      "epoch": 0.31516472511450494,
      "grad_norm": 1.688125729560852,
      "learning_rate": 1.2602279756119114e-05,
      "loss": 0.1924,
      "step": 21400
    },
    {
      "epoch": 0.31553290820459196,
      "grad_norm": 1.1041502952575684,
      "learning_rate": 1.2617006862831729e-05,
      "loss": 0.1804,
      "step": 21425
    },
    {
      "epoch": 0.315901091294679,
      "grad_norm": 1.085607647895813,
      "learning_rate": 1.2631733969544344e-05,
      "loss": 0.1848,
      "step": 21450
    },
    {
      "epoch": 0.31626927438476604,
      "grad_norm": 1.2319964170455933,
      "learning_rate": 1.264646107625696e-05,
      "loss": 0.1897,
      "step": 21475
    },
    {
      "epoch": 0.3166374574748531,
      "grad_norm": 1.2102179527282715,
      "learning_rate": 1.2661188182969575e-05,
      "loss": 0.1864,
      "step": 21500
    },
    {
      "epoch": 0.3170056405649401,
      "grad_norm": 1.4258583784103394,
      "learning_rate": 1.2675915289682191e-05,
      "loss": 0.1896,
      "step": 21525
    },
    {
      "epoch": 0.3173738236550272,
      "grad_norm": 1.3039108514785767,
      "learning_rate": 1.2690642396394806e-05,
      "loss": 0.1858,
      "step": 21550
    },
    {
      "epoch": 0.3177420067451142,
      "grad_norm": 1.050047516822815,
      "learning_rate": 1.270536950310742e-05,
      "loss": 0.1763,
      "step": 21575
    },
    {
      "epoch": 0.31811018983520123,
      "grad_norm": 1.3864538669586182,
      "learning_rate": 1.2720096609820037e-05,
      "loss": 0.1932,
      "step": 21600
    },
    {
      "epoch": 0.3184783729252883,
      "grad_norm": 1.6021428108215332,
      "learning_rate": 1.2734823716532651e-05,
      "loss": 0.1841,
      "step": 21625
    },
    {
      "epoch": 0.3188465560153753,
      "grad_norm": 1.1374555826187134,
      "learning_rate": 1.2749550823245268e-05,
      "loss": 0.1835,
      "step": 21650
    },
    {
      "epoch": 0.3192147391054624,
      "grad_norm": 1.4656171798706055,
      "learning_rate": 1.2764277929957882e-05,
      "loss": 0.1758,
      "step": 21675
    },
    {
      "epoch": 0.3195829221955494,
      "grad_norm": 1.4196912050247192,
      "learning_rate": 1.2779005036670497e-05,
      "loss": 0.1816,
      "step": 21700
    },
    {
      "epoch": 0.3199511052856364,
      "grad_norm": 1.0034286975860596,
      "learning_rate": 1.2793732143383113e-05,
      "loss": 0.1879,
      "step": 21725
    },
    {
      "epoch": 0.3203192883757235,
      "grad_norm": 1.1203784942626953,
      "learning_rate": 1.2808459250095728e-05,
      "loss": 0.178,
      "step": 21750
    },
    {
      "epoch": 0.3206874714658105,
      "grad_norm": 1.4380009174346924,
      "learning_rate": 1.2823186356808341e-05,
      "loss": 0.1786,
      "step": 21775
    },
    {
      "epoch": 0.3210556545558976,
      "grad_norm": 1.2680689096450806,
      "learning_rate": 1.2837913463520957e-05,
      "loss": 0.1851,
      "step": 21800
    },
    {
      "epoch": 0.3214238376459846,
      "grad_norm": 1.1762382984161377,
      "learning_rate": 1.2852640570233572e-05,
      "loss": 0.1825,
      "step": 21825
    },
    {
      "epoch": 0.32179202073607166,
      "grad_norm": 1.2245060205459595,
      "learning_rate": 1.2867367676946188e-05,
      "loss": 0.1885,
      "step": 21850
    },
    {
      "epoch": 0.3221602038261587,
      "grad_norm": 1.1323437690734863,
      "learning_rate": 1.2882094783658803e-05,
      "loss": 0.1836,
      "step": 21875
    },
    {
      "epoch": 0.3225283869162457,
      "grad_norm": 1.123909831047058,
      "learning_rate": 1.2896821890371418e-05,
      "loss": 0.1821,
      "step": 21900
    },
    {
      "epoch": 0.32289657000633276,
      "grad_norm": 1.4582033157348633,
      "learning_rate": 1.2911548997084034e-05,
      "loss": 0.1846,
      "step": 21925
    },
    {
      "epoch": 0.3232647530964198,
      "grad_norm": 1.174345850944519,
      "learning_rate": 1.2926276103796649e-05,
      "loss": 0.1808,
      "step": 21950
    },
    {
      "epoch": 0.32363293618650685,
      "grad_norm": 1.231270432472229,
      "learning_rate": 1.2941003210509265e-05,
      "loss": 0.1713,
      "step": 21975
    },
    {
      "epoch": 0.32400111927659386,
      "grad_norm": 1.2960635423660278,
      "learning_rate": 1.295573031722188e-05,
      "loss": 0.1857,
      "step": 22000
    },
    {
      "epoch": 0.3243693023666809,
      "grad_norm": 0.9982641935348511,
      "learning_rate": 1.2970457423934494e-05,
      "loss": 0.1825,
      "step": 22025
    },
    {
      "epoch": 0.32473748545676795,
      "grad_norm": 1.3632073402404785,
      "learning_rate": 1.298518453064711e-05,
      "loss": 0.1869,
      "step": 22050
    },
    {
      "epoch": 0.32510566854685496,
      "grad_norm": 1.3916586637496948,
      "learning_rate": 1.2999911637359725e-05,
      "loss": 0.1823,
      "step": 22075
    },
    {
      "epoch": 0.32547385163694204,
      "grad_norm": 1.2210173606872559,
      "learning_rate": 1.3014638744072342e-05,
      "loss": 0.1729,
      "step": 22100
    },
    {
      "epoch": 0.32584203472702905,
      "grad_norm": 1.3650914430618286,
      "learning_rate": 1.3029365850784956e-05,
      "loss": 0.1811,
      "step": 22125
    },
    {
      "epoch": 0.3262102178171161,
      "grad_norm": 1.2728363275527954,
      "learning_rate": 1.304409295749757e-05,
      "loss": 0.1845,
      "step": 22150
    },
    {
      "epoch": 0.32657840090720314,
      "grad_norm": 1.3349742889404297,
      "learning_rate": 1.3058820064210187e-05,
      "loss": 0.1858,
      "step": 22175
    },
    {
      "epoch": 0.32694658399729015,
      "grad_norm": 1.4142699241638184,
      "learning_rate": 1.3073547170922802e-05,
      "loss": 0.1769,
      "step": 22200
    },
    {
      "epoch": 0.3273147670873772,
      "grad_norm": 1.687619924545288,
      "learning_rate": 1.3088274277635418e-05,
      "loss": 0.1879,
      "step": 22225
    },
    {
      "epoch": 0.32768295017746424,
      "grad_norm": 1.2499809265136719,
      "learning_rate": 1.3103001384348033e-05,
      "loss": 0.1823,
      "step": 22250
    },
    {
      "epoch": 0.3280511332675513,
      "grad_norm": 1.4056971073150635,
      "learning_rate": 1.311772849106065e-05,
      "loss": 0.182,
      "step": 22275
    },
    {
      "epoch": 0.3284193163576383,
      "grad_norm": 1.1245547533035278,
      "learning_rate": 1.3132455597773262e-05,
      "loss": 0.1779,
      "step": 22300
    },
    {
      "epoch": 0.32878749944772534,
      "grad_norm": 1.2430375814437866,
      "learning_rate": 1.3147182704485877e-05,
      "loss": 0.1792,
      "step": 22325
    },
    {
      "epoch": 0.3291556825378124,
      "grad_norm": 1.1041802167892456,
      "learning_rate": 1.3161909811198491e-05,
      "loss": 0.1845,
      "step": 22350
    },
    {
      "epoch": 0.3295238656278994,
      "grad_norm": 1.1275758743286133,
      "learning_rate": 1.3176636917911108e-05,
      "loss": 0.1747,
      "step": 22375
    },
    {
      "epoch": 0.3298920487179865,
      "grad_norm": 1.3952943086624146,
      "learning_rate": 1.3191364024623722e-05,
      "loss": 0.1795,
      "step": 22400
    },
    {
      "epoch": 0.3302602318080735,
      "grad_norm": 1.4974526166915894,
      "learning_rate": 1.3206091131336339e-05,
      "loss": 0.1777,
      "step": 22425
    },
    {
      "epoch": 0.3306284148981606,
      "grad_norm": 1.3657947778701782,
      "learning_rate": 1.3220818238048953e-05,
      "loss": 0.1783,
      "step": 22450
    },
    {
      "epoch": 0.3309965979882476,
      "grad_norm": 1.5914496183395386,
      "learning_rate": 1.323554534476157e-05,
      "loss": 0.1726,
      "step": 22475
    },
    {
      "epoch": 0.3313647810783346,
      "grad_norm": 1.3611053228378296,
      "learning_rate": 1.3250272451474184e-05,
      "loss": 0.1827,
      "step": 22500
    },
    {
      "epoch": 0.3317329641684217,
      "grad_norm": 1.180898904800415,
      "learning_rate": 1.3264999558186799e-05,
      "loss": 0.1756,
      "step": 22525
    },
    {
      "epoch": 0.3321011472585087,
      "grad_norm": 1.2224446535110474,
      "learning_rate": 1.3279726664899415e-05,
      "loss": 0.1783,
      "step": 22550
    },
    {
      "epoch": 0.33246933034859577,
      "grad_norm": 1.2551331520080566,
      "learning_rate": 1.329445377161203e-05,
      "loss": 0.17,
      "step": 22575
    },
    {
      "epoch": 0.3328375134386828,
      "grad_norm": 1.4178850650787354,
      "learning_rate": 1.3309180878324646e-05,
      "loss": 0.1813,
      "step": 22600
    },
    {
      "epoch": 0.3332056965287698,
      "grad_norm": 1.0674101114273071,
      "learning_rate": 1.3323907985037261e-05,
      "loss": 0.1865,
      "step": 22625
    },
    {
      "epoch": 0.33357387961885687,
      "grad_norm": 1.501469373703003,
      "learning_rate": 1.3338635091749876e-05,
      "loss": 0.1925,
      "step": 22650
    },
    {
      "epoch": 0.3339420627089439,
      "grad_norm": 1.135437250137329,
      "learning_rate": 1.3353362198462492e-05,
      "loss": 0.167,
      "step": 22675
    },
    {
      "epoch": 0.33431024579903096,
      "grad_norm": 1.4651609659194946,
      "learning_rate": 1.3368089305175107e-05,
      "loss": 0.1777,
      "step": 22700
    },
    {
      "epoch": 0.334678428889118,
      "grad_norm": 1.3253328800201416,
      "learning_rate": 1.3382816411887723e-05,
      "loss": 0.1824,
      "step": 22725
    },
    {
      "epoch": 0.33504661197920504,
      "grad_norm": 1.2355338335037231,
      "learning_rate": 1.3397543518600338e-05,
      "loss": 0.1854,
      "step": 22750
    },
    {
      "epoch": 0.33541479506929206,
      "grad_norm": 1.2989897727966309,
      "learning_rate": 1.3412270625312952e-05,
      "loss": 0.1892,
      "step": 22775
    },
    {
      "epoch": 0.3357829781593791,
      "grad_norm": 1.2566183805465698,
      "learning_rate": 1.3426997732025569e-05,
      "loss": 0.1798,
      "step": 22800
    },
    {
      "epoch": 0.33615116124946615,
      "grad_norm": 1.4441183805465698,
      "learning_rate": 1.3441724838738182e-05,
      "loss": 0.178,
      "step": 22825
    },
    {
      "epoch": 0.33651934433955316,
      "grad_norm": 1.923704743385315,
      "learning_rate": 1.3455862861182294e-05,
      "loss": 0.1797,
      "step": 22850
    },
    {
      "epoch": 0.33688752742964023,
      "grad_norm": 1.0510995388031006,
      "learning_rate": 1.3470589967894909e-05,
      "loss": 0.1688,
      "step": 22875
    },
    {
      "epoch": 0.33725571051972725,
      "grad_norm": 1.6159011125564575,
      "learning_rate": 1.3485317074607525e-05,
      "loss": 0.1735,
      "step": 22900
    },
    {
      "epoch": 0.33762389360981426,
      "grad_norm": 1.45757257938385,
      "learning_rate": 1.350004418132014e-05,
      "loss": 0.1761,
      "step": 22925
    },
    {
      "epoch": 0.33799207669990133,
      "grad_norm": 1.2837272882461548,
      "learning_rate": 1.3514771288032755e-05,
      "loss": 0.1768,
      "step": 22950
    },
    {
      "epoch": 0.33836025978998835,
      "grad_norm": 1.4296029806137085,
      "learning_rate": 1.352949839474537e-05,
      "loss": 0.1791,
      "step": 22975
    },
    {
      "epoch": 0.3387284428800754,
      "grad_norm": 1.47047758102417,
      "learning_rate": 1.3544225501457984e-05,
      "loss": 0.1761,
      "step": 23000
    },
    {
      "epoch": 0.33909662597016244,
      "grad_norm": 1.6144695281982422,
      "learning_rate": 1.3558952608170599e-05,
      "loss": 0.1807,
      "step": 23025
    },
    {
      "epoch": 0.3394648090602495,
      "grad_norm": 1.231798768043518,
      "learning_rate": 1.3573679714883215e-05,
      "loss": 0.1762,
      "step": 23050
    },
    {
      "epoch": 0.3398329921503365,
      "grad_norm": 1.2265921831130981,
      "learning_rate": 1.358840682159583e-05,
      "loss": 0.168,
      "step": 23075
    },
    {
      "epoch": 0.34020117524042354,
      "grad_norm": 1.3389085531234741,
      "learning_rate": 1.3603133928308446e-05,
      "loss": 0.1675,
      "step": 23100
    },
    {
      "epoch": 0.3405693583305106,
      "grad_norm": 1.2000136375427246,
      "learning_rate": 1.361786103502106e-05,
      "loss": 0.1652,
      "step": 23125
    },
    {
      "epoch": 0.3409375414205976,
      "grad_norm": 1.2100751399993896,
      "learning_rate": 1.3632588141733675e-05,
      "loss": 0.1786,
      "step": 23150
    },
    {
      "epoch": 0.3413057245106847,
      "grad_norm": 1.1619250774383545,
      "learning_rate": 1.3647315248446292e-05,
      "loss": 0.1777,
      "step": 23175
    },
    {
      "epoch": 0.3416739076007717,
      "grad_norm": 1.4115016460418701,
      "learning_rate": 1.3662042355158906e-05,
      "loss": 0.1742,
      "step": 23200
    },
    {
      "epoch": 0.3420420906908587,
      "grad_norm": 1.1313114166259766,
      "learning_rate": 1.3676769461871523e-05,
      "loss": 0.1663,
      "step": 23225
    },
    {
      "epoch": 0.3424102737809458,
      "grad_norm": 1.1694964170455933,
      "learning_rate": 1.3691496568584137e-05,
      "loss": 0.1763,
      "step": 23250
    },
    {
      "epoch": 0.3427784568710328,
      "grad_norm": 1.2373450994491577,
      "learning_rate": 1.3706223675296752e-05,
      "loss": 0.1738,
      "step": 23275
    },
    {
      "epoch": 0.3431466399611199,
      "grad_norm": 1.5605484247207642,
      "learning_rate": 1.3720950782009368e-05,
      "loss": 0.1727,
      "step": 23300
    },
    {
      "epoch": 0.3435148230512069,
      "grad_norm": 1.084865927696228,
      "learning_rate": 1.3735677888721983e-05,
      "loss": 0.1763,
      "step": 23325
    },
    {
      "epoch": 0.34388300614129397,
      "grad_norm": 1.0823957920074463,
      "learning_rate": 1.37504049954346e-05,
      "loss": 0.1687,
      "step": 23350
    },
    {
      "epoch": 0.344251189231381,
      "grad_norm": 1.2936733961105347,
      "learning_rate": 1.3765132102147214e-05,
      "loss": 0.1698,
      "step": 23375
    },
    {
      "epoch": 0.344619372321468,
      "grad_norm": 1.125900387763977,
      "learning_rate": 1.3779859208859828e-05,
      "loss": 0.1714,
      "step": 23400
    },
    {
      "epoch": 0.34498755541155507,
      "grad_norm": 1.1072611808776855,
      "learning_rate": 1.3794586315572445e-05,
      "loss": 0.1757,
      "step": 23425
    },
    {
      "epoch": 0.3453557385016421,
      "grad_norm": 1.4236797094345093,
      "learning_rate": 1.380931342228506e-05,
      "loss": 0.1582,
      "step": 23450
    },
    {
      "epoch": 0.34572392159172916,
      "grad_norm": 1.2479220628738403,
      "learning_rate": 1.3824040528997676e-05,
      "loss": 0.1732,
      "step": 23475
    },
    {
      "epoch": 0.34609210468181617,
      "grad_norm": 1.3718407154083252,
      "learning_rate": 1.3838767635710289e-05,
      "loss": 0.1826,
      "step": 23500
    },
    {
      "epoch": 0.3464602877719032,
      "grad_norm": 1.2314164638519287,
      "learning_rate": 1.3853494742422903e-05,
      "loss": 0.1717,
      "step": 23525
    },
    {
      "epoch": 0.34682847086199026,
      "grad_norm": 1.364166498184204,
      "learning_rate": 1.386822184913552e-05,
      "loss": 0.1761,
      "step": 23550
    },
    {
      "epoch": 0.34719665395207727,
      "grad_norm": 1.3473548889160156,
      "learning_rate": 1.3882948955848134e-05,
      "loss": 0.169,
      "step": 23575
    },
    {
      "epoch": 0.34756483704216434,
      "grad_norm": 1.6499969959259033,
      "learning_rate": 1.3897676062560749e-05,
      "loss": 0.1673,
      "step": 23600
    },
    {
      "epoch": 0.34793302013225136,
      "grad_norm": 1.2165099382400513,
      "learning_rate": 1.3912403169273365e-05,
      "loss": 0.1699,
      "step": 23625
    },
    {
      "epoch": 0.34830120322233843,
      "grad_norm": 1.2824857234954834,
      "learning_rate": 1.392713027598598e-05,
      "loss": 0.173,
      "step": 23650
    },
    {
      "epoch": 0.34866938631242544,
      "grad_norm": 1.24893319606781,
      "learning_rate": 1.3941857382698596e-05,
      "loss": 0.1644,
      "step": 23675
    },
    {
      "epoch": 0.34903756940251246,
      "grad_norm": 1.295711636543274,
      "learning_rate": 1.3956584489411211e-05,
      "loss": 0.1757,
      "step": 23700
    },
    {
      "epoch": 0.34940575249259953,
      "grad_norm": 1.4278984069824219,
      "learning_rate": 1.3971311596123827e-05,
      "loss": 0.1616,
      "step": 23725
    },
    {
      "epoch": 0.34977393558268655,
      "grad_norm": 1.3779631853103638,
      "learning_rate": 1.3986038702836442e-05,
      "loss": 0.1764,
      "step": 23750
    },
    {
      "epoch": 0.3501421186727736,
      "grad_norm": 1.279738426208496,
      "learning_rate": 1.4000765809549057e-05,
      "loss": 0.1674,
      "step": 23775
    },
    {
      "epoch": 0.35051030176286063,
      "grad_norm": 1.2752861976623535,
      "learning_rate": 1.4015492916261673e-05,
      "loss": 0.1622,
      "step": 23800
    },
    {
      "epoch": 0.35087848485294765,
      "grad_norm": 1.615437626838684,
      "learning_rate": 1.4030220022974288e-05,
      "loss": 0.1751,
      "step": 23825
    },
    {
      "epoch": 0.3512466679430347,
      "grad_norm": 1.3879810571670532,
      "learning_rate": 1.4044947129686904e-05,
      "loss": 0.166,
      "step": 23850
    },
    {
      "epoch": 0.35161485103312173,
      "grad_norm": 1.3716177940368652,
      "learning_rate": 1.4059674236399519e-05,
      "loss": 0.1669,
      "step": 23875
    },
    {
      "epoch": 0.3519830341232088,
      "grad_norm": 1.2520028352737427,
      "learning_rate": 1.4074401343112133e-05,
      "loss": 0.1728,
      "step": 23900
    },
    {
      "epoch": 0.3523512172132958,
      "grad_norm": 1.7196623086929321,
      "learning_rate": 1.408912844982475e-05,
      "loss": 0.1766,
      "step": 23925
    },
    {
      "epoch": 0.3527194003033829,
      "grad_norm": 1.1781913042068481,
      "learning_rate": 1.4103855556537364e-05,
      "loss": 0.1736,
      "step": 23950
    },
    {
      "epoch": 0.3530875833934699,
      "grad_norm": 1.1760460138320923,
      "learning_rate": 1.411858266324998e-05,
      "loss": 0.1637,
      "step": 23975
    },
    {
      "epoch": 0.3534557664835569,
      "grad_norm": 1.2736172676086426,
      "learning_rate": 1.4133309769962595e-05,
      "loss": 0.1588,
      "step": 24000
    },
    {
      "epoch": 0.353823949573644,
      "grad_norm": 1.2447682619094849,
      "learning_rate": 1.4148036876675208e-05,
      "loss": 0.1635,
      "step": 24025
    },
    {
      "epoch": 0.354192132663731,
      "grad_norm": 1.9752280712127686,
      "learning_rate": 1.4162763983387825e-05,
      "loss": 0.1662,
      "step": 24050
    },
    {
      "epoch": 0.3545603157538181,
      "grad_norm": 1.1681700944900513,
      "learning_rate": 1.417749109010044e-05,
      "loss": 0.1707,
      "step": 24075
    },
    {
      "epoch": 0.3549284988439051,
      "grad_norm": 1.2221779823303223,
      "learning_rate": 1.4192218196813054e-05,
      "loss": 0.1748,
      "step": 24100
    },
    {
      "epoch": 0.3552966819339921,
      "grad_norm": 1.6404857635498047,
      "learning_rate": 1.420694530352567e-05,
      "loss": 0.1741,
      "step": 24125
    },
    {
      "epoch": 0.3556648650240792,
      "grad_norm": 1.340167760848999,
      "learning_rate": 1.4221672410238285e-05,
      "loss": 0.1669,
      "step": 24150
    },
    {
      "epoch": 0.3560330481141662,
      "grad_norm": 1.8098108768463135,
      "learning_rate": 1.4236399516950901e-05,
      "loss": 0.1582,
      "step": 24175
    },
    {
      "epoch": 0.35640123120425327,
      "grad_norm": 1.1971865892410278,
      "learning_rate": 1.4251126623663516e-05,
      "loss": 0.1637,
      "step": 24200
    },
    {
      "epoch": 0.3567694142943403,
      "grad_norm": 1.2659186124801636,
      "learning_rate": 1.426585373037613e-05,
      "loss": 0.1708,
      "step": 24225
    },
    {
      "epoch": 0.35713759738442735,
      "grad_norm": 1.5661580562591553,
      "learning_rate": 1.4280580837088747e-05,
      "loss": 0.1704,
      "step": 24250
    },
    {
      "epoch": 0.35750578047451437,
      "grad_norm": 1.0712963342666626,
      "learning_rate": 1.4295307943801362e-05,
      "loss": 0.1687,
      "step": 24275
    },
    {
      "epoch": 0.3578739635646014,
      "grad_norm": 1.2966463565826416,
      "learning_rate": 1.4310035050513978e-05,
      "loss": 0.1694,
      "step": 24300
    },
    {
      "epoch": 0.35824214665468845,
      "grad_norm": 1.4097758531570435,
      "learning_rate": 1.4324762157226593e-05,
      "loss": 0.1677,
      "step": 24325
    },
    {
      "epoch": 0.35861032974477547,
      "grad_norm": 1.2268353700637817,
      "learning_rate": 1.4339489263939207e-05,
      "loss": 0.1694,
      "step": 24350
    },
    {
      "epoch": 0.35897851283486254,
      "grad_norm": 1.4043735265731812,
      "learning_rate": 1.4354216370651824e-05,
      "loss": 0.1647,
      "step": 24375
    },
    {
      "epoch": 0.35934669592494956,
      "grad_norm": 1.1261810064315796,
      "learning_rate": 1.4368943477364438e-05,
      "loss": 0.168,
      "step": 24400
    },
    {
      "epoch": 0.35971487901503657,
      "grad_norm": 1.194634199142456,
      "learning_rate": 1.4383670584077055e-05,
      "loss": 0.1615,
      "step": 24425
    },
    {
      "epoch": 0.36008306210512364,
      "grad_norm": 1.2855753898620605,
      "learning_rate": 1.439839769078967e-05,
      "loss": 0.1694,
      "step": 24450
    },
    {
      "epoch": 0.36045124519521066,
      "grad_norm": 1.473142147064209,
      "learning_rate": 1.4413124797502284e-05,
      "loss": 0.1655,
      "step": 24475
    },
    {
      "epoch": 0.3608194282852977,
      "grad_norm": 1.2650806903839111,
      "learning_rate": 1.44278519042149e-05,
      "loss": 0.1721,
      "step": 24500
    },
    {
      "epoch": 0.36118761137538474,
      "grad_norm": 1.557308554649353,
      "learning_rate": 1.4442579010927515e-05,
      "loss": 0.1631,
      "step": 24525
    },
    {
      "epoch": 0.3615557944654718,
      "grad_norm": 2.086210012435913,
      "learning_rate": 1.4457306117640128e-05,
      "loss": 0.165,
      "step": 24550
    },
    {
      "epoch": 0.36192397755555883,
      "grad_norm": 1.1287481784820557,
      "learning_rate": 1.4472033224352744e-05,
      "loss": 0.1694,
      "step": 24575
    },
    {
      "epoch": 0.36229216064564584,
      "grad_norm": 1.2974869012832642,
      "learning_rate": 1.4486760331065359e-05,
      "loss": 0.1605,
      "step": 24600
    },
    {
      "epoch": 0.3626603437357329,
      "grad_norm": 1.2801960706710815,
      "learning_rate": 1.4501487437777975e-05,
      "loss": 0.168,
      "step": 24625
    },
    {
      "epoch": 0.36302852682581993,
      "grad_norm": 1.556368350982666,
      "learning_rate": 1.451621454449059e-05,
      "loss": 0.1715,
      "step": 24650
    },
    {
      "epoch": 0.363396709915907,
      "grad_norm": 1.3484870195388794,
      "learning_rate": 1.4530941651203205e-05,
      "loss": 0.16,
      "step": 24675
    },
    {
      "epoch": 0.363764893005994,
      "grad_norm": 1.9718481302261353,
      "learning_rate": 1.4545668757915821e-05,
      "loss": 0.1669,
      "step": 24700
    },
    {
      "epoch": 0.36413307609608103,
      "grad_norm": 1.3649804592132568,
      "learning_rate": 1.4560395864628436e-05,
      "loss": 0.165,
      "step": 24725
    },
    {
      "epoch": 0.3645012591861681,
      "grad_norm": 1.5112513303756714,
      "learning_rate": 1.4575122971341052e-05,
      "loss": 0.1633,
      "step": 24750
    },
    {
      "epoch": 0.3648694422762551,
      "grad_norm": 1.6509013175964355,
      "learning_rate": 1.4589850078053667e-05,
      "loss": 0.1677,
      "step": 24775
    },
    {
      "epoch": 0.3652376253663422,
      "grad_norm": 1.242371916770935,
      "learning_rate": 1.4604577184766283e-05,
      "loss": 0.1639,
      "step": 24800
    },
    {
      "epoch": 0.3656058084564292,
      "grad_norm": 1.3893762826919556,
      "learning_rate": 1.4619304291478898e-05,
      "loss": 0.1641,
      "step": 24825
    },
    {
      "epoch": 0.3659739915465163,
      "grad_norm": 1.8748382329940796,
      "learning_rate": 1.4633442313923007e-05,
      "loss": 0.1677,
      "step": 24850
    },
    {
      "epoch": 0.3663421746366033,
      "grad_norm": 1.4344875812530518,
      "learning_rate": 1.4648169420635623e-05,
      "loss": 0.1639,
      "step": 24875
    },
    {
      "epoch": 0.3667103577266903,
      "grad_norm": 1.229787826538086,
      "learning_rate": 1.4662896527348238e-05,
      "loss": 0.1669,
      "step": 24900
    },
    {
      "epoch": 0.3670785408167774,
      "grad_norm": 1.7821155786514282,
      "learning_rate": 1.4677623634060854e-05,
      "loss": 0.1694,
      "step": 24925
    },
    {
      "epoch": 0.3674467239068644,
      "grad_norm": 1.4540947675704956,
      "learning_rate": 1.4692350740773469e-05,
      "loss": 0.1632,
      "step": 24950
    },
    {
      "epoch": 0.36781490699695146,
      "grad_norm": 1.3375898599624634,
      "learning_rate": 1.4707077847486085e-05,
      "loss": 0.1614,
      "step": 24975
    },
    {
      "epoch": 0.3681830900870385,
      "grad_norm": 1.3651843070983887,
      "learning_rate": 1.47218049541987e-05,
      "loss": 0.1579,
      "step": 25000
    },
    {
      "epoch": 0.3685512731771255,
      "grad_norm": 1.2038607597351074,
      "learning_rate": 1.4736532060911314e-05,
      "loss": 0.1684,
      "step": 25025
    },
    {
      "epoch": 0.36891945626721256,
      "grad_norm": 1.362025260925293,
      "learning_rate": 1.475125916762393e-05,
      "loss": 0.155,
      "step": 25050
    },
    {
      "epoch": 0.3692876393572996,
      "grad_norm": 1.3822096586227417,
      "learning_rate": 1.4765986274336545e-05,
      "loss": 0.1624,
      "step": 25075
    },
    {
      "epoch": 0.36965582244738665,
      "grad_norm": 2.2989706993103027,
      "learning_rate": 1.4780713381049162e-05,
      "loss": 0.1651,
      "step": 25100
    },
    {
      "epoch": 0.37002400553747367,
      "grad_norm": 1.4963966608047485,
      "learning_rate": 1.4795440487761776e-05,
      "loss": 0.1634,
      "step": 25125
    },
    {
      "epoch": 0.37039218862756074,
      "grad_norm": 1.811550498008728,
      "learning_rate": 1.4810167594474391e-05,
      "loss": 0.1666,
      "step": 25150
    },
    {
      "epoch": 0.37076037171764775,
      "grad_norm": 1.3327304124832153,
      "learning_rate": 1.4824894701187007e-05,
      "loss": 0.1567,
      "step": 25175
    },
    {
      "epoch": 0.37112855480773477,
      "grad_norm": 1.1280299425125122,
      "learning_rate": 1.483962180789962e-05,
      "loss": 0.162,
      "step": 25200
    },
    {
      "epoch": 0.37149673789782184,
      "grad_norm": 1.8525338172912598,
      "learning_rate": 1.4854348914612235e-05,
      "loss": 0.1699,
      "step": 25225
    },
    {
      "epoch": 0.37186492098790885,
      "grad_norm": 1.4508144855499268,
      "learning_rate": 1.4869076021324851e-05,
      "loss": 0.1653,
      "step": 25250
    },
    {
      "epoch": 0.3722331040779959,
      "grad_norm": 1.5556083917617798,
      "learning_rate": 1.4883803128037466e-05,
      "loss": 0.1578,
      "step": 25275
    },
    {
      "epoch": 0.37260128716808294,
      "grad_norm": 1.4834439754486084,
      "learning_rate": 1.4898530234750082e-05,
      "loss": 0.1652,
      "step": 25300
    },
    {
      "epoch": 0.37296947025816996,
      "grad_norm": 1.4071025848388672,
      "learning_rate": 1.4913257341462697e-05,
      "loss": 0.1585,
      "step": 25325
    },
    {
      "epoch": 0.373337653348257,
      "grad_norm": 1.3072419166564941,
      "learning_rate": 1.4927984448175312e-05,
      "loss": 0.1604,
      "step": 25350
    },
    {
      "epoch": 0.37370583643834404,
      "grad_norm": 1.2802237272262573,
      "learning_rate": 1.4942711554887928e-05,
      "loss": 0.1575,
      "step": 25375
    },
    {
      "epoch": 0.3740740195284311,
      "grad_norm": 1.0141359567642212,
      "learning_rate": 1.4957438661600543e-05,
      "loss": 0.151,
      "step": 25400
    },
    {
      "epoch": 0.3744422026185181,
      "grad_norm": 1.3200985193252563,
      "learning_rate": 1.4972165768313159e-05,
      "loss": 0.1611,
      "step": 25425
    },
    {
      "epoch": 0.3748103857086052,
      "grad_norm": 1.478265643119812,
      "learning_rate": 1.4986892875025774e-05,
      "loss": 0.1681,
      "step": 25450
    },
    {
      "epoch": 0.3751785687986922,
      "grad_norm": 1.1375505924224854,
      "learning_rate": 1.5001619981738388e-05,
      "loss": 0.1581,
      "step": 25475
    },
    {
      "epoch": 0.37554675188877923,
      "grad_norm": 1.9286413192749023,
      "learning_rate": 1.5016347088451005e-05,
      "loss": 0.1711,
      "step": 25500
    },
    {
      "epoch": 0.3759149349788663,
      "grad_norm": 1.6296018362045288,
      "learning_rate": 1.503107419516362e-05,
      "loss": 0.1616,
      "step": 25525
    },
    {
      "epoch": 0.3762831180689533,
      "grad_norm": 1.482069492340088,
      "learning_rate": 1.5045801301876236e-05,
      "loss": 0.1619,
      "step": 25550
    },
    {
      "epoch": 0.3766513011590404,
      "grad_norm": 1.1911407709121704,
      "learning_rate": 1.506052840858885e-05,
      "loss": 0.1602,
      "step": 25575
    },
    {
      "epoch": 0.3770194842491274,
      "grad_norm": 1.1659958362579346,
      "learning_rate": 1.5075255515301465e-05,
      "loss": 0.1619,
      "step": 25600
    },
    {
      "epoch": 0.37738766733921447,
      "grad_norm": 1.192456841468811,
      "learning_rate": 1.5089982622014081e-05,
      "loss": 0.1591,
      "step": 25625
    },
    {
      "epoch": 0.3777558504293015,
      "grad_norm": 1.5322376489639282,
      "learning_rate": 1.5104709728726696e-05,
      "loss": 0.1645,
      "step": 25650
    },
    {
      "epoch": 0.3781240335193885,
      "grad_norm": 1.208443522453308,
      "learning_rate": 1.5119436835439312e-05,
      "loss": 0.1559,
      "step": 25675
    },
    {
      "epoch": 0.3784922166094756,
      "grad_norm": 1.7743768692016602,
      "learning_rate": 1.5134163942151927e-05,
      "loss": 0.1589,
      "step": 25700
    },
    {
      "epoch": 0.3788603996995626,
      "grad_norm": 1.4697054624557495,
      "learning_rate": 1.514889104886454e-05,
      "loss": 0.1511,
      "step": 25725
    },
    {
      "epoch": 0.37922858278964966,
      "grad_norm": 1.455379605293274,
      "learning_rate": 1.5163618155577156e-05,
      "loss": 0.165,
      "step": 25750
    },
    {
      "epoch": 0.3795967658797367,
      "grad_norm": 1.2325196266174316,
      "learning_rate": 1.5178345262289771e-05,
      "loss": 0.1546,
      "step": 25775
    },
    {
      "epoch": 0.3799649489698237,
      "grad_norm": 1.4367910623550415,
      "learning_rate": 1.5193072369002386e-05,
      "loss": 0.1687,
      "step": 25800
    },
    {
      "epoch": 0.38033313205991076,
      "grad_norm": 1.4400395154953003,
      "learning_rate": 1.5207799475715002e-05,
      "loss": 0.1519,
      "step": 25825
    },
    {
      "epoch": 0.3807013151499978,
      "grad_norm": 1.612979769706726,
      "learning_rate": 1.5222526582427617e-05,
      "loss": 0.1584,
      "step": 25850
    },
    {
      "epoch": 0.38106949824008485,
      "grad_norm": 1.6632273197174072,
      "learning_rate": 1.5237253689140233e-05,
      "loss": 0.159,
      "step": 25875
    },
    {
      "epoch": 0.38143768133017186,
      "grad_norm": 1.35444974899292,
      "learning_rate": 1.5251980795852848e-05,
      "loss": 0.1652,
      "step": 25900
    },
    {
      "epoch": 0.38180586442025893,
      "grad_norm": 1.7284669876098633,
      "learning_rate": 1.5266707902565464e-05,
      "loss": 0.1544,
      "step": 25925
    },
    {
      "epoch": 0.38217404751034595,
      "grad_norm": 1.4676893949508667,
      "learning_rate": 1.528143500927808e-05,
      "loss": 0.1624,
      "step": 25950
    },
    {
      "epoch": 0.38254223060043296,
      "grad_norm": 1.2698357105255127,
      "learning_rate": 1.5296162115990693e-05,
      "loss": 0.1599,
      "step": 25975
    },
    {
      "epoch": 0.38291041369052004,
      "grad_norm": 1.423574686050415,
      "learning_rate": 1.5310889222703308e-05,
      "loss": 0.1616,
      "step": 26000
    },
    {
      "epoch": 0.38327859678060705,
      "grad_norm": 1.264630913734436,
      "learning_rate": 1.5325616329415926e-05,
      "loss": 0.1554,
      "step": 26025
    },
    {
      "epoch": 0.3836467798706941,
      "grad_norm": 1.2193396091461182,
      "learning_rate": 1.534034343612854e-05,
      "loss": 0.1567,
      "step": 26050
    },
    {
      "epoch": 0.38401496296078114,
      "grad_norm": 1.2291114330291748,
      "learning_rate": 1.5355070542841155e-05,
      "loss": 0.16,
      "step": 26075
    },
    {
      "epoch": 0.38438314605086815,
      "grad_norm": 1.4714957475662231,
      "learning_rate": 1.536979764955377e-05,
      "loss": 0.1691,
      "step": 26100
    },
    {
      "epoch": 0.3847513291409552,
      "grad_norm": 1.1252244710922241,
      "learning_rate": 1.5384524756266385e-05,
      "loss": 0.1541,
      "step": 26125
    },
    {
      "epoch": 0.38511951223104224,
      "grad_norm": 1.645963430404663,
      "learning_rate": 1.5399251862979003e-05,
      "loss": 0.1505,
      "step": 26150
    },
    {
      "epoch": 0.3854876953211293,
      "grad_norm": 1.2302244901657104,
      "learning_rate": 1.5413978969691617e-05,
      "loss": 0.1609,
      "step": 26175
    },
    {
      "epoch": 0.3858558784112163,
      "grad_norm": 1.4861561059951782,
      "learning_rate": 1.5428706076404232e-05,
      "loss": 0.1588,
      "step": 26200
    },
    {
      "epoch": 0.3862240615013034,
      "grad_norm": 1.3448596000671387,
      "learning_rate": 1.5443433183116847e-05,
      "loss": 0.1595,
      "step": 26225
    },
    {
      "epoch": 0.3865922445913904,
      "grad_norm": 2.0465972423553467,
      "learning_rate": 1.545816028982946e-05,
      "loss": 0.1557,
      "step": 26250
    },
    {
      "epoch": 0.3869604276814774,
      "grad_norm": 1.3477450609207153,
      "learning_rate": 1.5472887396542076e-05,
      "loss": 0.1481,
      "step": 26275
    },
    {
      "epoch": 0.3873286107715645,
      "grad_norm": 1.526447057723999,
      "learning_rate": 1.548761450325469e-05,
      "loss": 0.1608,
      "step": 26300
    },
    {
      "epoch": 0.3876967938616515,
      "grad_norm": 1.4494584798812866,
      "learning_rate": 1.5502341609967305e-05,
      "loss": 0.1599,
      "step": 26325
    },
    {
      "epoch": 0.3880649769517386,
      "grad_norm": 1.4448504447937012,
      "learning_rate": 1.5517068716679923e-05,
      "loss": 0.1495,
      "step": 26350
    },
    {
      "epoch": 0.3884331600418256,
      "grad_norm": 1.3479735851287842,
      "learning_rate": 1.5531795823392538e-05,
      "loss": 0.1506,
      "step": 26375
    },
    {
      "epoch": 0.3888013431319126,
      "grad_norm": 2.5692548751831055,
      "learning_rate": 1.5546522930105152e-05,
      "loss": 0.1544,
      "step": 26400
    },
    {
      "epoch": 0.3891695262219997,
      "grad_norm": 1.4863978624343872,
      "learning_rate": 1.5561250036817767e-05,
      "loss": 0.1487,
      "step": 26425
    },
    {
      "epoch": 0.3895377093120867,
      "grad_norm": 1.1599884033203125,
      "learning_rate": 1.5575977143530382e-05,
      "loss": 0.1602,
      "step": 26450
    },
    {
      "epoch": 0.38990589240217377,
      "grad_norm": 1.4004433155059814,
      "learning_rate": 1.5590704250243e-05,
      "loss": 0.1558,
      "step": 26475
    },
    {
      "epoch": 0.3902740754922608,
      "grad_norm": 1.9732985496520996,
      "learning_rate": 1.5605431356955614e-05,
      "loss": 0.1543,
      "step": 26500
    },
    {
      "epoch": 0.39064225858234786,
      "grad_norm": 1.709584355354309,
      "learning_rate": 1.562015846366823e-05,
      "loss": 0.1523,
      "step": 26525
    },
    {
      "epoch": 0.39101044167243487,
      "grad_norm": 1.5066332817077637,
      "learning_rate": 1.5634885570380844e-05,
      "loss": 0.1413,
      "step": 26550
    },
    {
      "epoch": 0.3913786247625219,
      "grad_norm": 1.2560043334960938,
      "learning_rate": 1.564961267709346e-05,
      "loss": 0.1554,
      "step": 26575
    },
    {
      "epoch": 0.39174680785260896,
      "grad_norm": 1.3362622261047363,
      "learning_rate": 1.5664339783806076e-05,
      "loss": 0.1583,
      "step": 26600
    },
    {
      "epoch": 0.392114990942696,
      "grad_norm": 1.3887149095535278,
      "learning_rate": 1.567906689051869e-05,
      "loss": 0.1544,
      "step": 26625
    },
    {
      "epoch": 0.39248317403278304,
      "grad_norm": 1.496191143989563,
      "learning_rate": 1.5693793997231306e-05,
      "loss": 0.1525,
      "step": 26650
    },
    {
      "epoch": 0.39285135712287006,
      "grad_norm": 1.2821321487426758,
      "learning_rate": 1.570852110394392e-05,
      "loss": 0.1475,
      "step": 26675
    },
    {
      "epoch": 0.3932195402129571,
      "grad_norm": 1.2387585639953613,
      "learning_rate": 1.5723248210656535e-05,
      "loss": 0.1463,
      "step": 26700
    },
    {
      "epoch": 0.39358772330304415,
      "grad_norm": 1.3753855228424072,
      "learning_rate": 1.5737975317369153e-05,
      "loss": 0.1505,
      "step": 26725
    },
    {
      "epoch": 0.39395590639313116,
      "grad_norm": 1.6069930791854858,
      "learning_rate": 1.5752702424081768e-05,
      "loss": 0.1492,
      "step": 26750
    },
    {
      "epoch": 0.39432408948321823,
      "grad_norm": 1.6889283657073975,
      "learning_rate": 1.576742953079438e-05,
      "loss": 0.1538,
      "step": 26775
    },
    {
      "epoch": 0.39469227257330525,
      "grad_norm": 1.3301491737365723,
      "learning_rate": 1.5782156637506997e-05,
      "loss": 0.1477,
      "step": 26800
    },
    {
      "epoch": 0.3950604556633923,
      "grad_norm": 1.5237812995910645,
      "learning_rate": 1.5796883744219612e-05,
      "loss": 0.1395,
      "step": 26825
    },
    {
      "epoch": 0.39542863875347933,
      "grad_norm": 1.6008751392364502,
      "learning_rate": 1.5811610850932226e-05,
      "loss": 0.1577,
      "step": 26850
    },
    {
      "epoch": 0.39579682184356635,
      "grad_norm": 1.2232062816619873,
      "learning_rate": 1.5825748873376337e-05,
      "loss": 0.151,
      "step": 26875
    },
    {
      "epoch": 0.3961650049336534,
      "grad_norm": 1.3552442789077759,
      "learning_rate": 1.5840475980088955e-05,
      "loss": 0.1639,
      "step": 26900
    },
    {
      "epoch": 0.39653318802374043,
      "grad_norm": 1.5702840089797974,
      "learning_rate": 1.5855203086801567e-05,
      "loss": 0.156,
      "step": 26925
    },
    {
      "epoch": 0.3969013711138275,
      "grad_norm": 1.4246914386749268,
      "learning_rate": 1.586993019351418e-05,
      "loss": 0.1561,
      "step": 26950
    },
    {
      "epoch": 0.3972695542039145,
      "grad_norm": 1.04789137840271,
      "learning_rate": 1.58846573002268e-05,
      "loss": 0.1556,
      "step": 26975
    },
    {
      "epoch": 0.39763773729400154,
      "grad_norm": 1.5424654483795166,
      "learning_rate": 1.5899384406939414e-05,
      "loss": 0.1585,
      "step": 27000
    },
    {
      "epoch": 0.3980059203840886,
      "grad_norm": 1.3209010362625122,
      "learning_rate": 1.591411151365203e-05,
      "loss": 0.1482,
      "step": 27025
    },
    {
      "epoch": 0.3983741034741756,
      "grad_norm": 1.437687873840332,
      "learning_rate": 1.5928838620364643e-05,
      "loss": 0.1491,
      "step": 27050
    },
    {
      "epoch": 0.3987422865642627,
      "grad_norm": 1.3784931898117065,
      "learning_rate": 1.5943565727077258e-05,
      "loss": 0.1544,
      "step": 27075
    },
    {
      "epoch": 0.3991104696543497,
      "grad_norm": 1.365294098854065,
      "learning_rate": 1.5958292833789876e-05,
      "loss": 0.1527,
      "step": 27100
    },
    {
      "epoch": 0.3994786527444368,
      "grad_norm": 1.4586961269378662,
      "learning_rate": 1.597301994050249e-05,
      "loss": 0.1513,
      "step": 27125
    },
    {
      "epoch": 0.3998468358345238,
      "grad_norm": 1.2528156042099,
      "learning_rate": 1.5987747047215105e-05,
      "loss": 0.1491,
      "step": 27150
    },
    {
      "epoch": 0.4002150189246108,
      "grad_norm": 1.2851122617721558,
      "learning_rate": 1.600247415392772e-05,
      "loss": 0.1498,
      "step": 27175
    },
    {
      "epoch": 0.4005832020146979,
      "grad_norm": 1.3917914628982544,
      "learning_rate": 1.6017201260640335e-05,
      "loss": 0.1576,
      "step": 27200
    },
    {
      "epoch": 0.4009513851047849,
      "grad_norm": 1.2627981901168823,
      "learning_rate": 1.6031928367352953e-05,
      "loss": 0.1529,
      "step": 27225
    },
    {
      "epoch": 0.40131956819487197,
      "grad_norm": 1.2389428615570068,
      "learning_rate": 1.6046655474065567e-05,
      "loss": 0.1585,
      "step": 27250
    },
    {
      "epoch": 0.401687751284959,
      "grad_norm": 1.2482959032058716,
      "learning_rate": 1.6061382580778182e-05,
      "loss": 0.141,
      "step": 27275
    },
    {
      "epoch": 0.402055934375046,
      "grad_norm": 1.5105633735656738,
      "learning_rate": 1.6076109687490797e-05,
      "loss": 0.1356,
      "step": 27300
    },
    {
      "epoch": 0.40242411746513307,
      "grad_norm": 1.5747884511947632,
      "learning_rate": 1.609083679420341e-05,
      "loss": 0.1487,
      "step": 27325
    },
    {
      "epoch": 0.4027923005552201,
      "grad_norm": 1.2051383256912231,
      "learning_rate": 1.610556390091603e-05,
      "loss": 0.1556,
      "step": 27350
    },
    {
      "epoch": 0.40316048364530715,
      "grad_norm": 1.313347339630127,
      "learning_rate": 1.6120291007628644e-05,
      "loss": 0.1528,
      "step": 27375
    },
    {
      "epoch": 0.40352866673539417,
      "grad_norm": 1.3574309349060059,
      "learning_rate": 1.613501811434126e-05,
      "loss": 0.1531,
      "step": 27400
    },
    {
      "epoch": 0.40389684982548124,
      "grad_norm": 1.1611549854278564,
      "learning_rate": 1.6149745221053873e-05,
      "loss": 0.1355,
      "step": 27425
    },
    {
      "epoch": 0.40426503291556826,
      "grad_norm": 1.5017125606536865,
      "learning_rate": 1.6164472327766488e-05,
      "loss": 0.146,
      "step": 27450
    },
    {
      "epoch": 0.40463321600565527,
      "grad_norm": 2.055236577987671,
      "learning_rate": 1.6179199434479102e-05,
      "loss": 0.1408,
      "step": 27475
    },
    {
      "epoch": 0.40500139909574234,
      "grad_norm": 1.3194584846496582,
      "learning_rate": 1.6193926541191717e-05,
      "loss": 0.1505,
      "step": 27500
    },
    {
      "epoch": 0.40536958218582936,
      "grad_norm": 1.3802636861801147,
      "learning_rate": 1.6208653647904332e-05,
      "loss": 0.1548,
      "step": 27525
    },
    {
      "epoch": 0.40573776527591643,
      "grad_norm": 1.34214186668396,
      "learning_rate": 1.622338075461695e-05,
      "loss": 0.1363,
      "step": 27550
    },
    {
      "epoch": 0.40610594836600344,
      "grad_norm": 1.2384545803070068,
      "learning_rate": 1.6238107861329564e-05,
      "loss": 0.1528,
      "step": 27575
    },
    {
      "epoch": 0.40647413145609046,
      "grad_norm": 1.8981614112854004,
      "learning_rate": 1.625283496804218e-05,
      "loss": 0.1479,
      "step": 27600
    },
    {
      "epoch": 0.40684231454617753,
      "grad_norm": 1.6245635747909546,
      "learning_rate": 1.6267562074754794e-05,
      "loss": 0.1519,
      "step": 27625
    },
    {
      "epoch": 0.40721049763626455,
      "grad_norm": 1.3965609073638916,
      "learning_rate": 1.628228918146741e-05,
      "loss": 0.1538,
      "step": 27650
    },
    {
      "epoch": 0.4075786807263516,
      "grad_norm": 1.1727200746536255,
      "learning_rate": 1.6297016288180026e-05,
      "loss": 0.1458,
      "step": 27675
    },
    {
      "epoch": 0.40794686381643863,
      "grad_norm": 1.342296838760376,
      "learning_rate": 1.631174339489264e-05,
      "loss": 0.1502,
      "step": 27700
    },
    {
      "epoch": 0.4083150469065257,
      "grad_norm": 1.6224783658981323,
      "learning_rate": 1.6326470501605256e-05,
      "loss": 0.1457,
      "step": 27725
    },
    {
      "epoch": 0.4086832299966127,
      "grad_norm": 1.3962500095367432,
      "learning_rate": 1.634119760831787e-05,
      "loss": 0.1497,
      "step": 27750
    },
    {
      "epoch": 0.40905141308669973,
      "grad_norm": 1.6830774545669556,
      "learning_rate": 1.6355924715030485e-05,
      "loss": 0.1431,
      "step": 27775
    },
    {
      "epoch": 0.4094195961767868,
      "grad_norm": 1.2840871810913086,
      "learning_rate": 1.6370651821743103e-05,
      "loss": 0.1545,
      "step": 27800
    },
    {
      "epoch": 0.4097877792668738,
      "grad_norm": 1.1633930206298828,
      "learning_rate": 1.6385378928455718e-05,
      "loss": 0.1487,
      "step": 27825
    },
    {
      "epoch": 0.4101559623569609,
      "grad_norm": 1.2747087478637695,
      "learning_rate": 1.6400106035168332e-05,
      "loss": 0.1436,
      "step": 27850
    },
    {
      "epoch": 0.4105241454470479,
      "grad_norm": 1.332680344581604,
      "learning_rate": 1.6414833141880947e-05,
      "loss": 0.1437,
      "step": 27875
    },
    {
      "epoch": 0.4108923285371349,
      "grad_norm": 1.2956751585006714,
      "learning_rate": 1.6429560248593565e-05,
      "loss": 0.1411,
      "step": 27900
    },
    {
      "epoch": 0.411260511627222,
      "grad_norm": 1.2702312469482422,
      "learning_rate": 1.644428735530618e-05,
      "loss": 0.1435,
      "step": 27925
    },
    {
      "epoch": 0.411628694717309,
      "grad_norm": 1.5242955684661865,
      "learning_rate": 1.6459014462018794e-05,
      "loss": 0.1499,
      "step": 27950
    },
    {
      "epoch": 0.4119968778073961,
      "grad_norm": 1.279181718826294,
      "learning_rate": 1.6473741568731406e-05,
      "loss": 0.1473,
      "step": 27975
    },
    {
      "epoch": 0.4123650608974831,
      "grad_norm": 1.2853556871414185,
      "learning_rate": 1.6488468675444024e-05,
      "loss": 0.1484,
      "step": 28000
    },
    {
      "epoch": 0.41273324398757016,
      "grad_norm": 1.3675915002822876,
      "learning_rate": 1.650319578215664e-05,
      "loss": 0.1502,
      "step": 28025
    },
    {
      "epoch": 0.4131014270776572,
      "grad_norm": 1.477524757385254,
      "learning_rate": 1.6517922888869253e-05,
      "loss": 0.1409,
      "step": 28050
    },
    {
      "epoch": 0.4134696101677442,
      "grad_norm": 1.4106136560440063,
      "learning_rate": 1.6532649995581868e-05,
      "loss": 0.1453,
      "step": 28075
    },
    {
      "epoch": 0.41383779325783127,
      "grad_norm": 1.4092079401016235,
      "learning_rate": 1.6547377102294482e-05,
      "loss": 0.1502,
      "step": 28100
    },
    {
      "epoch": 0.4142059763479183,
      "grad_norm": 1.4397227764129639,
      "learning_rate": 1.65621042090071e-05,
      "loss": 0.146,
      "step": 28125
    },
    {
      "epoch": 0.41457415943800535,
      "grad_norm": 1.4936139583587646,
      "learning_rate": 1.6576831315719715e-05,
      "loss": 0.144,
      "step": 28150
    },
    {
      "epoch": 0.41494234252809237,
      "grad_norm": 1.263158917427063,
      "learning_rate": 1.659155842243233e-05,
      "loss": 0.1464,
      "step": 28175
    },
    {
      "epoch": 0.4153105256181794,
      "grad_norm": 1.577727198600769,
      "learning_rate": 1.6606285529144944e-05,
      "loss": 0.146,
      "step": 28200
    },
    {
      "epoch": 0.41567870870826645,
      "grad_norm": 1.1452049016952515,
      "learning_rate": 1.6621012635857562e-05,
      "loss": 0.1421,
      "step": 28225
    },
    {
      "epoch": 0.41604689179835347,
      "grad_norm": 1.2748959064483643,
      "learning_rate": 1.6635739742570177e-05,
      "loss": 0.1412,
      "step": 28250
    },
    {
      "epoch": 0.41641507488844054,
      "grad_norm": 1.3492980003356934,
      "learning_rate": 1.665046684928279e-05,
      "loss": 0.1524,
      "step": 28275
    },
    {
      "epoch": 0.41678325797852755,
      "grad_norm": 1.2692748308181763,
      "learning_rate": 1.6665193955995406e-05,
      "loss": 0.1499,
      "step": 28300
    },
    {
      "epoch": 0.4171514410686146,
      "grad_norm": 1.4279532432556152,
      "learning_rate": 1.667992106270802e-05,
      "loss": 0.1481,
      "step": 28325
    },
    {
      "epoch": 0.41751962415870164,
      "grad_norm": 1.5554064512252808,
      "learning_rate": 1.669464816942064e-05,
      "loss": 0.1521,
      "step": 28350
    },
    {
      "epoch": 0.41788780724878866,
      "grad_norm": 1.4604122638702393,
      "learning_rate": 1.6709375276133254e-05,
      "loss": 0.1502,
      "step": 28375
    },
    {
      "epoch": 0.4182559903388757,
      "grad_norm": 1.6504544019699097,
      "learning_rate": 1.672410238284587e-05,
      "loss": 0.1439,
      "step": 28400
    },
    {
      "epoch": 0.41862417342896274,
      "grad_norm": 1.6239420175552368,
      "learning_rate": 1.6738829489558483e-05,
      "loss": 0.1497,
      "step": 28425
    },
    {
      "epoch": 0.4189923565190498,
      "grad_norm": 1.3682606220245361,
      "learning_rate": 1.6753556596271098e-05,
      "loss": 0.1514,
      "step": 28450
    },
    {
      "epoch": 0.41936053960913683,
      "grad_norm": 1.132874846458435,
      "learning_rate": 1.6768283702983716e-05,
      "loss": 0.1442,
      "step": 28475
    },
    {
      "epoch": 0.41972872269922384,
      "grad_norm": 1.3476136922836304,
      "learning_rate": 1.6783010809696327e-05,
      "loss": 0.146,
      "step": 28500
    },
    {
      "epoch": 0.4200969057893109,
      "grad_norm": 1.2397880554199219,
      "learning_rate": 1.679773791640894e-05,
      "loss": 0.1438,
      "step": 28525
    },
    {
      "epoch": 0.42046508887939793,
      "grad_norm": 1.3864375352859497,
      "learning_rate": 1.681246502312156e-05,
      "loss": 0.1456,
      "step": 28550
    },
    {
      "epoch": 0.420833271969485,
      "grad_norm": 1.5281440019607544,
      "learning_rate": 1.6827192129834174e-05,
      "loss": 0.1536,
      "step": 28575
    },
    {
      "epoch": 0.421201455059572,
      "grad_norm": 1.2818355560302734,
      "learning_rate": 1.684191923654679e-05,
      "loss": 0.1432,
      "step": 28600
    },
    {
      "epoch": 0.4215696381496591,
      "grad_norm": 1.3338851928710938,
      "learning_rate": 1.6856646343259404e-05,
      "loss": 0.1395,
      "step": 28625
    },
    {
      "epoch": 0.4219378212397461,
      "grad_norm": 1.4486943483352661,
      "learning_rate": 1.6871373449972018e-05,
      "loss": 0.1412,
      "step": 28650
    },
    {
      "epoch": 0.4223060043298331,
      "grad_norm": 1.2234041690826416,
      "learning_rate": 1.6886100556684636e-05,
      "loss": 0.1526,
      "step": 28675
    },
    {
      "epoch": 0.4226741874199202,
      "grad_norm": 1.268106460571289,
      "learning_rate": 1.690082766339725e-05,
      "loss": 0.1439,
      "step": 28700
    },
    {
      "epoch": 0.4230423705100072,
      "grad_norm": 1.360742211341858,
      "learning_rate": 1.6915554770109866e-05,
      "loss": 0.1513,
      "step": 28725
    },
    {
      "epoch": 0.4234105536000943,
      "grad_norm": 1.2657285928726196,
      "learning_rate": 1.693028187682248e-05,
      "loss": 0.1518,
      "step": 28750
    },
    {
      "epoch": 0.4237787366901813,
      "grad_norm": 1.2721775770187378,
      "learning_rate": 1.6945008983535095e-05,
      "loss": 0.1415,
      "step": 28775
    },
    {
      "epoch": 0.4241469197802683,
      "grad_norm": 1.7750695943832397,
      "learning_rate": 1.6959736090247713e-05,
      "loss": 0.1366,
      "step": 28800
    },
    {
      "epoch": 0.4245151028703554,
      "grad_norm": 1.446153998374939,
      "learning_rate": 1.6974463196960328e-05,
      "loss": 0.1384,
      "step": 28825
    },
    {
      "epoch": 0.4248832859604424,
      "grad_norm": 1.5607752799987793,
      "learning_rate": 1.6989190303672942e-05,
      "loss": 0.1461,
      "step": 28850
    },
    {
      "epoch": 0.42525146905052946,
      "grad_norm": 1.1609091758728027,
      "learning_rate": 1.7003328326117053e-05,
      "loss": 0.1462,
      "step": 28875
    },
    {
      "epoch": 0.4256196521406165,
      "grad_norm": 1.1347569227218628,
      "learning_rate": 1.7018055432829668e-05,
      "loss": 0.14,
      "step": 28900
    },
    {
      "epoch": 0.42598783523070355,
      "grad_norm": 1.4356962442398071,
      "learning_rate": 1.7032782539542282e-05,
      "loss": 0.1479,
      "step": 28925
    },
    {
      "epoch": 0.42635601832079056,
      "grad_norm": 1.5667729377746582,
      "learning_rate": 1.7047509646254897e-05,
      "loss": 0.1491,
      "step": 28950
    },
    {
      "epoch": 0.4267242014108776,
      "grad_norm": 1.314302682876587,
      "learning_rate": 1.7062236752967515e-05,
      "loss": 0.1437,
      "step": 28975
    },
    {
      "epoch": 0.42709238450096465,
      "grad_norm": 1.2139393091201782,
      "learning_rate": 1.707696385968013e-05,
      "loss": 0.1379,
      "step": 29000
    },
    {
      "epoch": 0.42746056759105167,
      "grad_norm": 1.203644871711731,
      "learning_rate": 1.7091690966392744e-05,
      "loss": 0.1431,
      "step": 29025
    },
    {
      "epoch": 0.42782875068113874,
      "grad_norm": 1.437659740447998,
      "learning_rate": 1.710641807310536e-05,
      "loss": 0.1443,
      "step": 29050
    },
    {
      "epoch": 0.42819693377122575,
      "grad_norm": 1.3156840801239014,
      "learning_rate": 1.7121145179817974e-05,
      "loss": 0.1504,
      "step": 29075
    },
    {
      "epoch": 0.42856511686131277,
      "grad_norm": 1.430440068244934,
      "learning_rate": 1.7135872286530592e-05,
      "loss": 0.1444,
      "step": 29100
    },
    {
      "epoch": 0.42893329995139984,
      "grad_norm": 1.129550814628601,
      "learning_rate": 1.7150599393243206e-05,
      "loss": 0.1445,
      "step": 29125
    },
    {
      "epoch": 0.42930148304148685,
      "grad_norm": 1.8005977869033813,
      "learning_rate": 1.716532649995582e-05,
      "loss": 0.1463,
      "step": 29150
    },
    {
      "epoch": 0.4296696661315739,
      "grad_norm": 1.3299059867858887,
      "learning_rate": 1.7180053606668436e-05,
      "loss": 0.1461,
      "step": 29175
    },
    {
      "epoch": 0.43003784922166094,
      "grad_norm": 1.46122407913208,
      "learning_rate": 1.719478071338105e-05,
      "loss": 0.143,
      "step": 29200
    },
    {
      "epoch": 0.430406032311748,
      "grad_norm": 1.190050721168518,
      "learning_rate": 1.7209507820093665e-05,
      "loss": 0.1341,
      "step": 29225
    },
    {
      "epoch": 0.430774215401835,
      "grad_norm": 1.300050973892212,
      "learning_rate": 1.722423492680628e-05,
      "loss": 0.1422,
      "step": 29250
    },
    {
      "epoch": 0.43114239849192204,
      "grad_norm": 1.2624523639678955,
      "learning_rate": 1.7238962033518894e-05,
      "loss": 0.1407,
      "step": 29275
    },
    {
      "epoch": 0.4315105815820091,
      "grad_norm": 1.954111099243164,
      "learning_rate": 1.7253689140231512e-05,
      "loss": 0.1437,
      "step": 29300
    },
    {
      "epoch": 0.4318787646720961,
      "grad_norm": 1.4629451036453247,
      "learning_rate": 1.7268416246944127e-05,
      "loss": 0.1362,
      "step": 29325
    },
    {
      "epoch": 0.4322469477621832,
      "grad_norm": 1.2829498052597046,
      "learning_rate": 1.728314335365674e-05,
      "loss": 0.1434,
      "step": 29350
    },
    {
      "epoch": 0.4326151308522702,
      "grad_norm": 1.4281407594680786,
      "learning_rate": 1.7297870460369356e-05,
      "loss": 0.1308,
      "step": 29375
    },
    {
      "epoch": 0.43298331394235723,
      "grad_norm": 1.4115138053894043,
      "learning_rate": 1.731259756708197e-05,
      "loss": 0.1375,
      "step": 29400
    },
    {
      "epoch": 0.4333514970324443,
      "grad_norm": 1.5216060876846313,
      "learning_rate": 1.732732467379459e-05,
      "loss": 0.1405,
      "step": 29425
    },
    {
      "epoch": 0.4337196801225313,
      "grad_norm": 1.1875633001327515,
      "learning_rate": 1.7342051780507204e-05,
      "loss": 0.1456,
      "step": 29450
    },
    {
      "epoch": 0.4340878632126184,
      "grad_norm": 1.9678187370300293,
      "learning_rate": 1.735677888721982e-05,
      "loss": 0.1366,
      "step": 29475
    },
    {
      "epoch": 0.4344560463027054,
      "grad_norm": 1.3012348413467407,
      "learning_rate": 1.7371505993932433e-05,
      "loss": 0.1445,
      "step": 29500
    },
    {
      "epoch": 0.43482422939279247,
      "grad_norm": 1.2349598407745361,
      "learning_rate": 1.7386233100645048e-05,
      "loss": 0.1376,
      "step": 29525
    },
    {
      "epoch": 0.4351924124828795,
      "grad_norm": 1.517460584640503,
      "learning_rate": 1.7400960207357666e-05,
      "loss": 0.1379,
      "step": 29550
    },
    {
      "epoch": 0.4355605955729665,
      "grad_norm": 1.4243059158325195,
      "learning_rate": 1.741568731407028e-05,
      "loss": 0.1337,
      "step": 29575
    },
    {
      "epoch": 0.4359287786630536,
      "grad_norm": 1.931024432182312,
      "learning_rate": 1.7430414420782895e-05,
      "loss": 0.1378,
      "step": 29600
    },
    {
      "epoch": 0.4362969617531406,
      "grad_norm": 1.5100048780441284,
      "learning_rate": 1.744514152749551e-05,
      "loss": 0.1337,
      "step": 29625
    },
    {
      "epoch": 0.43666514484322766,
      "grad_norm": 1.2116131782531738,
      "learning_rate": 1.7459868634208124e-05,
      "loss": 0.1435,
      "step": 29650
    },
    {
      "epoch": 0.4370333279333147,
      "grad_norm": 1.585527777671814,
      "learning_rate": 1.7474595740920742e-05,
      "loss": 0.1425,
      "step": 29675
    },
    {
      "epoch": 0.4374015110234017,
      "grad_norm": 1.3392324447631836,
      "learning_rate": 1.7489322847633354e-05,
      "loss": 0.1447,
      "step": 29700
    },
    {
      "epoch": 0.43776969411348876,
      "grad_norm": 1.6167775392532349,
      "learning_rate": 1.7504049954345968e-05,
      "loss": 0.1507,
      "step": 29725
    },
    {
      "epoch": 0.4381378772035758,
      "grad_norm": 1.5940417051315308,
      "learning_rate": 1.7518777061058586e-05,
      "loss": 0.1463,
      "step": 29750
    },
    {
      "epoch": 0.43850606029366285,
      "grad_norm": 1.433103084564209,
      "learning_rate": 1.75335041677712e-05,
      "loss": 0.1474,
      "step": 29775
    },
    {
      "epoch": 0.43887424338374986,
      "grad_norm": 1.6639951467514038,
      "learning_rate": 1.7548231274483816e-05,
      "loss": 0.1386,
      "step": 29800
    },
    {
      "epoch": 0.43924242647383693,
      "grad_norm": 1.7384814023971558,
      "learning_rate": 1.756295838119643e-05,
      "loss": 0.1409,
      "step": 29825
    },
    {
      "epoch": 0.43961060956392395,
      "grad_norm": 1.450372576713562,
      "learning_rate": 1.7577685487909045e-05,
      "loss": 0.1461,
      "step": 29850
    },
    {
      "epoch": 0.43997879265401096,
      "grad_norm": 1.1789109706878662,
      "learning_rate": 1.7592412594621663e-05,
      "loss": 0.1322,
      "step": 29875
    },
    {
      "epoch": 0.44034697574409803,
      "grad_norm": 1.5792953968048096,
      "learning_rate": 1.7607139701334278e-05,
      "loss": 0.1403,
      "step": 29900
    },
    {
      "epoch": 0.44071515883418505,
      "grad_norm": 1.680922031402588,
      "learning_rate": 1.7621866808046892e-05,
      "loss": 0.1394,
      "step": 29925
    },
    {
      "epoch": 0.4410833419242721,
      "grad_norm": 1.2849785089492798,
      "learning_rate": 1.7636593914759507e-05,
      "loss": 0.1407,
      "step": 29950
    },
    {
      "epoch": 0.44145152501435914,
      "grad_norm": 1.3818351030349731,
      "learning_rate": 1.765132102147212e-05,
      "loss": 0.1399,
      "step": 29975
    },
    {
      "epoch": 0.44181970810444615,
      "grad_norm": 1.699486255645752,
      "learning_rate": 1.766604812818474e-05,
      "loss": 0.1446,
      "step": 30000
    },
    {
      "epoch": 0.4421878911945332,
      "grad_norm": 1.5980112552642822,
      "learning_rate": 1.7680775234897354e-05,
      "loss": 0.1363,
      "step": 30025
    },
    {
      "epoch": 0.44255607428462024,
      "grad_norm": 1.4802896976470947,
      "learning_rate": 1.769550234160997e-05,
      "loss": 0.1386,
      "step": 30050
    },
    {
      "epoch": 0.4429242573747073,
      "grad_norm": 1.1918233633041382,
      "learning_rate": 1.7710229448322584e-05,
      "loss": 0.1421,
      "step": 30075
    },
    {
      "epoch": 0.4432924404647943,
      "grad_norm": 1.561808466911316,
      "learning_rate": 1.7724956555035198e-05,
      "loss": 0.1468,
      "step": 30100
    },
    {
      "epoch": 0.4436606235548814,
      "grad_norm": 1.0813902616500854,
      "learning_rate": 1.7739683661747816e-05,
      "loss": 0.1427,
      "step": 30125
    },
    {
      "epoch": 0.4440288066449684,
      "grad_norm": 1.9768576622009277,
      "learning_rate": 1.775441076846043e-05,
      "loss": 0.1409,
      "step": 30150
    },
    {
      "epoch": 0.4443969897350554,
      "grad_norm": 1.3829275369644165,
      "learning_rate": 1.7769137875173046e-05,
      "loss": 0.1305,
      "step": 30175
    },
    {
      "epoch": 0.4447651728251425,
      "grad_norm": 1.5168200731277466,
      "learning_rate": 1.778386498188566e-05,
      "loss": 0.1402,
      "step": 30200
    },
    {
      "epoch": 0.4451333559152295,
      "grad_norm": 1.5185489654541016,
      "learning_rate": 1.7798592088598275e-05,
      "loss": 0.1445,
      "step": 30225
    },
    {
      "epoch": 0.4455015390053166,
      "grad_norm": 1.5915287733078003,
      "learning_rate": 1.781331919531089e-05,
      "loss": 0.1434,
      "step": 30250
    },
    {
      "epoch": 0.4458697220954036,
      "grad_norm": 1.6269174814224243,
      "learning_rate": 1.7828046302023504e-05,
      "loss": 0.1421,
      "step": 30275
    },
    {
      "epoch": 0.4462379051854906,
      "grad_norm": 2.000884771347046,
      "learning_rate": 1.784277340873612e-05,
      "loss": 0.138,
      "step": 30300
    },
    {
      "epoch": 0.4466060882755777,
      "grad_norm": 1.268903374671936,
      "learning_rate": 1.7857500515448737e-05,
      "loss": 0.1373,
      "step": 30325
    },
    {
      "epoch": 0.4469742713656647,
      "grad_norm": 1.5966827869415283,
      "learning_rate": 1.787222762216135e-05,
      "loss": 0.134,
      "step": 30350
    },
    {
      "epoch": 0.44734245445575177,
      "grad_norm": 1.3796306848526,
      "learning_rate": 1.7886954728873966e-05,
      "loss": 0.132,
      "step": 30375
    },
    {
      "epoch": 0.4477106375458388,
      "grad_norm": 1.3174057006835938,
      "learning_rate": 1.790168183558658e-05,
      "loss": 0.1447,
      "step": 30400
    },
    {
      "epoch": 0.44807882063592586,
      "grad_norm": 1.365686058998108,
      "learning_rate": 1.7916408942299195e-05,
      "loss": 0.1416,
      "step": 30425
    },
    {
      "epoch": 0.44844700372601287,
      "grad_norm": 1.4339334964752197,
      "learning_rate": 1.7931136049011814e-05,
      "loss": 0.131,
      "step": 30450
    },
    {
      "epoch": 0.4488151868160999,
      "grad_norm": 1.3043586015701294,
      "learning_rate": 1.7945863155724428e-05,
      "loss": 0.139,
      "step": 30475
    },
    {
      "epoch": 0.44918336990618696,
      "grad_norm": 1.7352689504623413,
      "learning_rate": 1.7960590262437043e-05,
      "loss": 0.1429,
      "step": 30500
    },
    {
      "epoch": 0.449551552996274,
      "grad_norm": 1.3499093055725098,
      "learning_rate": 1.7975317369149657e-05,
      "loss": 0.1322,
      "step": 30525
    },
    {
      "epoch": 0.44991973608636104,
      "grad_norm": 1.353043794631958,
      "learning_rate": 1.7990044475862276e-05,
      "loss": 0.1352,
      "step": 30550
    },
    {
      "epoch": 0.45028791917644806,
      "grad_norm": 1.6418284177780151,
      "learning_rate": 1.800477158257489e-05,
      "loss": 0.1377,
      "step": 30575
    },
    {
      "epoch": 0.45065610226653513,
      "grad_norm": 1.2903937101364136,
      "learning_rate": 1.8019498689287505e-05,
      "loss": 0.137,
      "step": 30600
    },
    {
      "epoch": 0.45102428535662215,
      "grad_norm": 1.2895569801330566,
      "learning_rate": 1.803422579600012e-05,
      "loss": 0.1354,
      "step": 30625
    },
    {
      "epoch": 0.45139246844670916,
      "grad_norm": 1.5892612934112549,
      "learning_rate": 1.8048952902712734e-05,
      "loss": 0.1364,
      "step": 30650
    },
    {
      "epoch": 0.45176065153679623,
      "grad_norm": 1.3265799283981323,
      "learning_rate": 1.8063680009425352e-05,
      "loss": 0.1387,
      "step": 30675
    },
    {
      "epoch": 0.45212883462688325,
      "grad_norm": 1.2973653078079224,
      "learning_rate": 1.8078407116137967e-05,
      "loss": 0.1317,
      "step": 30700
    },
    {
      "epoch": 0.4524970177169703,
      "grad_norm": 1.486402988433838,
      "learning_rate": 1.809313422285058e-05,
      "loss": 0.1358,
      "step": 30725
    },
    {
      "epoch": 0.45286520080705733,
      "grad_norm": 1.2491552829742432,
      "learning_rate": 1.8107861329563193e-05,
      "loss": 0.1383,
      "step": 30750
    },
    {
      "epoch": 0.45323338389714435,
      "grad_norm": 1.4163362979888916,
      "learning_rate": 1.812258843627581e-05,
      "loss": 0.1325,
      "step": 30775
    },
    {
      "epoch": 0.4536015669872314,
      "grad_norm": 1.8074849843978882,
      "learning_rate": 1.8137315542988425e-05,
      "loss": 0.1387,
      "step": 30800
    },
    {
      "epoch": 0.45396975007731843,
      "grad_norm": 1.5657782554626465,
      "learning_rate": 1.815204264970104e-05,
      "loss": 0.1315,
      "step": 30825
    },
    {
      "epoch": 0.4543379331674055,
      "grad_norm": 1.5195591449737549,
      "learning_rate": 1.8166769756413655e-05,
      "loss": 0.1306,
      "step": 30850
    },
    {
      "epoch": 0.4547061162574925,
      "grad_norm": 1.4572436809539795,
      "learning_rate": 1.8181496863126273e-05,
      "loss": 0.1458,
      "step": 30875
    },
    {
      "epoch": 0.4550742993475796,
      "grad_norm": 1.7552560567855835,
      "learning_rate": 1.819563488557038e-05,
      "loss": 0.1451,
      "step": 30900
    },
    {
      "epoch": 0.4554424824376666,
      "grad_norm": 1.2025951147079468,
      "learning_rate": 1.8210361992282998e-05,
      "loss": 0.1422,
      "step": 30925
    },
    {
      "epoch": 0.4558106655277536,
      "grad_norm": 1.4719462394714355,
      "learning_rate": 1.8225089098995613e-05,
      "loss": 0.1476,
      "step": 30950
    },
    {
      "epoch": 0.4561788486178407,
      "grad_norm": 1.5610344409942627,
      "learning_rate": 1.8239816205708228e-05,
      "loss": 0.1325,
      "step": 30975
    },
    {
      "epoch": 0.4565470317079277,
      "grad_norm": 1.1036796569824219,
      "learning_rate": 1.8254543312420842e-05,
      "loss": 0.1328,
      "step": 31000
    },
    {
      "epoch": 0.4569152147980148,
      "grad_norm": 1.2756305932998657,
      "learning_rate": 1.8269270419133457e-05,
      "loss": 0.1353,
      "step": 31025
    },
    {
      "epoch": 0.4572833978881018,
      "grad_norm": 1.3201026916503906,
      "learning_rate": 1.8283997525846075e-05,
      "loss": 0.1375,
      "step": 31050
    },
    {
      "epoch": 0.4576515809781888,
      "grad_norm": 1.4888620376586914,
      "learning_rate": 1.829872463255869e-05,
      "loss": 0.1423,
      "step": 31075
    },
    {
      "epoch": 0.4580197640682759,
      "grad_norm": 1.075620174407959,
      "learning_rate": 1.8313451739271304e-05,
      "loss": 0.1311,
      "step": 31100
    },
    {
      "epoch": 0.4583879471583629,
      "grad_norm": 1.1998538970947266,
      "learning_rate": 1.832817884598392e-05,
      "loss": 0.1335,
      "step": 31125
    },
    {
      "epoch": 0.45875613024844997,
      "grad_norm": 1.4449012279510498,
      "learning_rate": 1.8342905952696534e-05,
      "loss": 0.1344,
      "step": 31150
    },
    {
      "epoch": 0.459124313338537,
      "grad_norm": 1.598140835762024,
      "learning_rate": 1.835763305940915e-05,
      "loss": 0.1299,
      "step": 31175
    },
    {
      "epoch": 0.45949249642862405,
      "grad_norm": 1.2417925596237183,
      "learning_rate": 1.8372360166121766e-05,
      "loss": 0.1306,
      "step": 31200
    },
    {
      "epoch": 0.45986067951871107,
      "grad_norm": 1.6606899499893188,
      "learning_rate": 1.838708727283438e-05,
      "loss": 0.1334,
      "step": 31225
    },
    {
      "epoch": 0.4602288626087981,
      "grad_norm": 1.806178331375122,
      "learning_rate": 1.8401814379546996e-05,
      "loss": 0.1347,
      "step": 31250
    },
    {
      "epoch": 0.46059704569888515,
      "grad_norm": 1.320526361465454,
      "learning_rate": 1.841654148625961e-05,
      "loss": 0.1263,
      "step": 31275
    },
    {
      "epoch": 0.46096522878897217,
      "grad_norm": 1.7611198425292969,
      "learning_rate": 1.8431268592972228e-05,
      "loss": 0.1317,
      "step": 31300
    },
    {
      "epoch": 0.46133341187905924,
      "grad_norm": 1.796420693397522,
      "learning_rate": 1.8445995699684843e-05,
      "loss": 0.1355,
      "step": 31325
    },
    {
      "epoch": 0.46170159496914626,
      "grad_norm": 1.1251814365386963,
      "learning_rate": 1.8460722806397458e-05,
      "loss": 0.1225,
      "step": 31350
    },
    {
      "epoch": 0.46206977805923327,
      "grad_norm": 1.2082396745681763,
      "learning_rate": 1.8475449913110072e-05,
      "loss": 0.1419,
      "step": 31375
    },
    {
      "epoch": 0.46243796114932034,
      "grad_norm": 1.1356067657470703,
      "learning_rate": 1.8490177019822687e-05,
      "loss": 0.1277,
      "step": 31400
    },
    {
      "epoch": 0.46280614423940736,
      "grad_norm": 1.736794114112854,
      "learning_rate": 1.85049041265353e-05,
      "loss": 0.1285,
      "step": 31425
    },
    {
      "epoch": 0.46317432732949443,
      "grad_norm": 1.3372302055358887,
      "learning_rate": 1.8519631233247916e-05,
      "loss": 0.1392,
      "step": 31450
    },
    {
      "epoch": 0.46354251041958144,
      "grad_norm": 1.379860281944275,
      "learning_rate": 1.853435833996053e-05,
      "loss": 0.1362,
      "step": 31475
    },
    {
      "epoch": 0.4639106935096685,
      "grad_norm": 1.103456974029541,
      "learning_rate": 1.854908544667315e-05,
      "loss": 0.1299,
      "step": 31500
    },
    {
      "epoch": 0.46427887659975553,
      "grad_norm": 1.44719660282135,
      "learning_rate": 1.8563812553385764e-05,
      "loss": 0.1327,
      "step": 31525
    },
    {
      "epoch": 0.46464705968984255,
      "grad_norm": 1.378851294517517,
      "learning_rate": 1.8578539660098378e-05,
      "loss": 0.1323,
      "step": 31550
    },
    {
      "epoch": 0.4650152427799296,
      "grad_norm": 1.592476725578308,
      "learning_rate": 1.8593266766810993e-05,
      "loss": 0.1354,
      "step": 31575
    },
    {
      "epoch": 0.46538342587001663,
      "grad_norm": 1.4160147905349731,
      "learning_rate": 1.8607993873523607e-05,
      "loss": 0.1398,
      "step": 31600
    },
    {
      "epoch": 0.4657516089601037,
      "grad_norm": 1.1927225589752197,
      "learning_rate": 1.8622720980236226e-05,
      "loss": 0.1355,
      "step": 31625
    },
    {
      "epoch": 0.4661197920501907,
      "grad_norm": 1.4182437658309937,
      "learning_rate": 1.863744808694884e-05,
      "loss": 0.1315,
      "step": 31650
    },
    {
      "epoch": 0.46648797514027773,
      "grad_norm": 1.1829264163970947,
      "learning_rate": 1.8652175193661455e-05,
      "loss": 0.128,
      "step": 31675
    },
    {
      "epoch": 0.4668561582303648,
      "grad_norm": 1.4739160537719727,
      "learning_rate": 1.866690230037407e-05,
      "loss": 0.1281,
      "step": 31700
    },
    {
      "epoch": 0.4672243413204518,
      "grad_norm": 1.5583876371383667,
      "learning_rate": 1.8681629407086684e-05,
      "loss": 0.129,
      "step": 31725
    },
    {
      "epoch": 0.4675925244105389,
      "grad_norm": 1.4014379978179932,
      "learning_rate": 1.8696356513799302e-05,
      "loss": 0.135,
      "step": 31750
    },
    {
      "epoch": 0.4679607075006259,
      "grad_norm": 1.5254454612731934,
      "learning_rate": 1.8711083620511917e-05,
      "loss": 0.1316,
      "step": 31775
    },
    {
      "epoch": 0.468328890590713,
      "grad_norm": 1.4662139415740967,
      "learning_rate": 1.872581072722453e-05,
      "loss": 0.1335,
      "step": 31800
    },
    {
      "epoch": 0.4686970736808,
      "grad_norm": 1.3167014122009277,
      "learning_rate": 1.8740537833937146e-05,
      "loss": 0.1356,
      "step": 31825
    },
    {
      "epoch": 0.469065256770887,
      "grad_norm": 1.8181610107421875,
      "learning_rate": 1.875526494064976e-05,
      "loss": 0.137,
      "step": 31850
    },
    {
      "epoch": 0.4694334398609741,
      "grad_norm": 2.111936569213867,
      "learning_rate": 1.876999204736238e-05,
      "loss": 0.1347,
      "step": 31875
    },
    {
      "epoch": 0.4698016229510611,
      "grad_norm": 1.3071465492248535,
      "learning_rate": 1.8784719154074993e-05,
      "loss": 0.1395,
      "step": 31900
    },
    {
      "epoch": 0.47016980604114816,
      "grad_norm": 1.4029223918914795,
      "learning_rate": 1.8799446260787608e-05,
      "loss": 0.1362,
      "step": 31925
    },
    {
      "epoch": 0.4705379891312352,
      "grad_norm": 1.134627342224121,
      "learning_rate": 1.8814173367500223e-05,
      "loss": 0.1413,
      "step": 31950
    },
    {
      "epoch": 0.4709061722213222,
      "grad_norm": 1.1709901094436646,
      "learning_rate": 1.8828900474212837e-05,
      "loss": 0.1465,
      "step": 31975
    },
    {
      "epoch": 0.47127435531140927,
      "grad_norm": 1.2714532613754272,
      "learning_rate": 1.8843627580925452e-05,
      "loss": 0.1293,
      "step": 32000
    },
    {
      "epoch": 0.4716425384014963,
      "grad_norm": 1.1055426597595215,
      "learning_rate": 1.8858354687638067e-05,
      "loss": 0.1339,
      "step": 32025
    },
    {
      "epoch": 0.47201072149158335,
      "grad_norm": 1.716193437576294,
      "learning_rate": 1.887308179435068e-05,
      "loss": 0.1314,
      "step": 32050
    },
    {
      "epoch": 0.47237890458167037,
      "grad_norm": 1.7784597873687744,
      "learning_rate": 1.88878089010633e-05,
      "loss": 0.1343,
      "step": 32075
    },
    {
      "epoch": 0.47274708767175744,
      "grad_norm": 1.379332184791565,
      "learning_rate": 1.8902536007775914e-05,
      "loss": 0.13,
      "step": 32100
    },
    {
      "epoch": 0.47311527076184445,
      "grad_norm": 1.1867659091949463,
      "learning_rate": 1.891726311448853e-05,
      "loss": 0.1302,
      "step": 32125
    },
    {
      "epoch": 0.47348345385193147,
      "grad_norm": 1.303497314453125,
      "learning_rate": 1.8931990221201143e-05,
      "loss": 0.1169,
      "step": 32150
    },
    {
      "epoch": 0.47385163694201854,
      "grad_norm": 1.1655243635177612,
      "learning_rate": 1.8946717327913758e-05,
      "loss": 0.1317,
      "step": 32175
    },
    {
      "epoch": 0.47421982003210555,
      "grad_norm": 1.1606842279434204,
      "learning_rate": 1.8961444434626376e-05,
      "loss": 0.1255,
      "step": 32200
    },
    {
      "epoch": 0.4745880031221926,
      "grad_norm": 1.7602474689483643,
      "learning_rate": 1.897617154133899e-05,
      "loss": 0.1335,
      "step": 32225
    },
    {
      "epoch": 0.47495618621227964,
      "grad_norm": 1.8369065523147583,
      "learning_rate": 1.8990898648051605e-05,
      "loss": 0.1306,
      "step": 32250
    },
    {
      "epoch": 0.47532436930236666,
      "grad_norm": 1.317317247390747,
      "learning_rate": 1.900562575476422e-05,
      "loss": 0.123,
      "step": 32275
    },
    {
      "epoch": 0.4756925523924537,
      "grad_norm": 1.2774357795715332,
      "learning_rate": 1.9020352861476835e-05,
      "loss": 0.1307,
      "step": 32300
    },
    {
      "epoch": 0.47606073548254074,
      "grad_norm": 1.5253177881240845,
      "learning_rate": 1.9035079968189453e-05,
      "loss": 0.1326,
      "step": 32325
    },
    {
      "epoch": 0.4764289185726278,
      "grad_norm": 1.3377888202667236,
      "learning_rate": 1.9049807074902067e-05,
      "loss": 0.1253,
      "step": 32350
    },
    {
      "epoch": 0.47679710166271483,
      "grad_norm": 1.155227541923523,
      "learning_rate": 1.9064534181614682e-05,
      "loss": 0.1298,
      "step": 32375
    },
    {
      "epoch": 0.4771652847528019,
      "grad_norm": 1.1564701795578003,
      "learning_rate": 1.9079261288327297e-05,
      "loss": 0.1376,
      "step": 32400
    },
    {
      "epoch": 0.4775334678428889,
      "grad_norm": 1.2230069637298584,
      "learning_rate": 1.909398839503991e-05,
      "loss": 0.136,
      "step": 32425
    },
    {
      "epoch": 0.47790165093297593,
      "grad_norm": 1.498449444770813,
      "learning_rate": 1.910871550175253e-05,
      "loss": 0.124,
      "step": 32450
    },
    {
      "epoch": 0.478269834023063,
      "grad_norm": 1.4574726819992065,
      "learning_rate": 1.912344260846514e-05,
      "loss": 0.1262,
      "step": 32475
    },
    {
      "epoch": 0.47863801711315,
      "grad_norm": 1.7008484601974487,
      "learning_rate": 1.9138169715177755e-05,
      "loss": 0.1342,
      "step": 32500
    },
    {
      "epoch": 0.4790062002032371,
      "grad_norm": 1.3751330375671387,
      "learning_rate": 1.9152896821890373e-05,
      "loss": 0.1321,
      "step": 32525
    },
    {
      "epoch": 0.4793743832933241,
      "grad_norm": 1.2620458602905273,
      "learning_rate": 1.9167623928602988e-05,
      "loss": 0.128,
      "step": 32550
    },
    {
      "epoch": 0.4797425663834111,
      "grad_norm": 1.7352402210235596,
      "learning_rate": 1.9182351035315603e-05,
      "loss": 0.1294,
      "step": 32575
    },
    {
      "epoch": 0.4801107494734982,
      "grad_norm": 1.4355820417404175,
      "learning_rate": 1.9197078142028217e-05,
      "loss": 0.1292,
      "step": 32600
    },
    {
      "epoch": 0.4804789325635852,
      "grad_norm": 1.8195645809173584,
      "learning_rate": 1.9211805248740832e-05,
      "loss": 0.1385,
      "step": 32625
    },
    {
      "epoch": 0.4808471156536723,
      "grad_norm": 1.464348554611206,
      "learning_rate": 1.922653235545345e-05,
      "loss": 0.1327,
      "step": 32650
    },
    {
      "epoch": 0.4812152987437593,
      "grad_norm": 1.9561387300491333,
      "learning_rate": 1.9241259462166065e-05,
      "loss": 0.1349,
      "step": 32675
    },
    {
      "epoch": 0.48158348183384636,
      "grad_norm": 1.7696665525436401,
      "learning_rate": 1.925598656887868e-05,
      "loss": 0.143,
      "step": 32700
    },
    {
      "epoch": 0.4819516649239334,
      "grad_norm": 1.2237669229507446,
      "learning_rate": 1.9270713675591294e-05,
      "loss": 0.1333,
      "step": 32725
    },
    {
      "epoch": 0.4823198480140204,
      "grad_norm": 1.3987947702407837,
      "learning_rate": 1.928544078230391e-05,
      "loss": 0.1387,
      "step": 32750
    },
    {
      "epoch": 0.48268803110410746,
      "grad_norm": 1.272199273109436,
      "learning_rate": 1.9300167889016527e-05,
      "loss": 0.1315,
      "step": 32775
    },
    {
      "epoch": 0.4830562141941945,
      "grad_norm": 1.6304975748062134,
      "learning_rate": 1.931489499572914e-05,
      "loss": 0.1261,
      "step": 32800
    },
    {
      "epoch": 0.48342439728428155,
      "grad_norm": 1.2250480651855469,
      "learning_rate": 1.9329622102441756e-05,
      "loss": 0.1343,
      "step": 32825
    },
    {
      "epoch": 0.48379258037436856,
      "grad_norm": 1.7616349458694458,
      "learning_rate": 1.934434920915437e-05,
      "loss": 0.1247,
      "step": 32850
    },
    {
      "epoch": 0.4841607634644556,
      "grad_norm": 1.8865598440170288,
      "learning_rate": 1.935907631586699e-05,
      "loss": 0.1284,
      "step": 32875
    },
    {
      "epoch": 0.48452894655454265,
      "grad_norm": 1.4966330528259277,
      "learning_rate": 1.9373214338311096e-05,
      "loss": 0.1444,
      "step": 32900
    },
    {
      "epoch": 0.48489712964462967,
      "grad_norm": 1.5111415386199951,
      "learning_rate": 1.938794144502371e-05,
      "loss": 0.128,
      "step": 32925
    },
    {
      "epoch": 0.48526531273471674,
      "grad_norm": 1.2859002351760864,
      "learning_rate": 1.940266855173633e-05,
      "loss": 0.1303,
      "step": 32950
    },
    {
      "epoch": 0.48563349582480375,
      "grad_norm": 1.6504935026168823,
      "learning_rate": 1.9417395658448943e-05,
      "loss": 0.1244,
      "step": 32975
    },
    {
      "epoch": 0.4860016789148908,
      "grad_norm": 1.4879761934280396,
      "learning_rate": 1.9432122765161558e-05,
      "loss": 0.135,
      "step": 33000
    },
    {
      "epoch": 0.48636986200497784,
      "grad_norm": 1.4184648990631104,
      "learning_rate": 1.9446849871874173e-05,
      "loss": 0.1276,
      "step": 33025
    },
    {
      "epoch": 0.48673804509506485,
      "grad_norm": 1.7261390686035156,
      "learning_rate": 1.946157697858679e-05,
      "loss": 0.1285,
      "step": 33050
    },
    {
      "epoch": 0.4871062281851519,
      "grad_norm": 1.2350839376449585,
      "learning_rate": 1.9476304085299405e-05,
      "loss": 0.129,
      "step": 33075
    },
    {
      "epoch": 0.48747441127523894,
      "grad_norm": 1.2649427652359009,
      "learning_rate": 1.949103119201202e-05,
      "loss": 0.1131,
      "step": 33100
    },
    {
      "epoch": 0.487842594365326,
      "grad_norm": 1.2869795560836792,
      "learning_rate": 1.9505758298724635e-05,
      "loss": 0.1276,
      "step": 33125
    },
    {
      "epoch": 0.488210777455413,
      "grad_norm": 1.3838227987289429,
      "learning_rate": 1.952048540543725e-05,
      "loss": 0.123,
      "step": 33150
    },
    {
      "epoch": 0.48857896054550004,
      "grad_norm": 1.4343010187149048,
      "learning_rate": 1.9535212512149864e-05,
      "loss": 0.1311,
      "step": 33175
    },
    {
      "epoch": 0.4889471436355871,
      "grad_norm": 1.1730332374572754,
      "learning_rate": 1.954993961886248e-05,
      "loss": 0.1315,
      "step": 33200
    },
    {
      "epoch": 0.4893153267256741,
      "grad_norm": 1.2627941370010376,
      "learning_rate": 1.9564666725575093e-05,
      "loss": 0.1333,
      "step": 33225
    },
    {
      "epoch": 0.4896835098157612,
      "grad_norm": 1.5235569477081299,
      "learning_rate": 1.9579393832287708e-05,
      "loss": 0.1262,
      "step": 33250
    },
    {
      "epoch": 0.4900516929058482,
      "grad_norm": 1.6173646450042725,
      "learning_rate": 1.9594120939000326e-05,
      "loss": 0.1284,
      "step": 33275
    },
    {
      "epoch": 0.4904198759959353,
      "grad_norm": 2.243269205093384,
      "learning_rate": 1.960884804571294e-05,
      "loss": 0.1232,
      "step": 33300
    },
    {
      "epoch": 0.4907880590860223,
      "grad_norm": 1.573286771774292,
      "learning_rate": 1.9623575152425555e-05,
      "loss": 0.1235,
      "step": 33325
    },
    {
      "epoch": 0.4911562421761093,
      "grad_norm": 1.7263425588607788,
      "learning_rate": 1.963830225913817e-05,
      "loss": 0.1243,
      "step": 33350
    },
    {
      "epoch": 0.4915244252661964,
      "grad_norm": 1.7113111019134521,
      "learning_rate": 1.9653029365850788e-05,
      "loss": 0.134,
      "step": 33375
    },
    {
      "epoch": 0.4918926083562834,
      "grad_norm": 1.3466795682907104,
      "learning_rate": 1.9667756472563403e-05,
      "loss": 0.1298,
      "step": 33400
    },
    {
      "epoch": 0.49226079144637047,
      "grad_norm": 1.4281026124954224,
      "learning_rate": 1.9682483579276017e-05,
      "loss": 0.119,
      "step": 33425
    },
    {
      "epoch": 0.4926289745364575,
      "grad_norm": 1.7018014192581177,
      "learning_rate": 1.9697210685988632e-05,
      "loss": 0.1324,
      "step": 33450
    },
    {
      "epoch": 0.4929971576265445,
      "grad_norm": 1.7589343786239624,
      "learning_rate": 1.9711937792701247e-05,
      "loss": 0.128,
      "step": 33475
    },
    {
      "epoch": 0.4933653407166316,
      "grad_norm": 2.133329153060913,
      "learning_rate": 1.9726664899413865e-05,
      "loss": 0.125,
      "step": 33500
    },
    {
      "epoch": 0.4937335238067186,
      "grad_norm": 1.4540302753448486,
      "learning_rate": 1.974139200612648e-05,
      "loss": 0.127,
      "step": 33525
    },
    {
      "epoch": 0.49410170689680566,
      "grad_norm": 1.419734239578247,
      "learning_rate": 1.9756119112839094e-05,
      "loss": 0.1274,
      "step": 33550
    },
    {
      "epoch": 0.4944698899868927,
      "grad_norm": 1.654998779296875,
      "learning_rate": 1.977084621955171e-05,
      "loss": 0.1298,
      "step": 33575
    },
    {
      "epoch": 0.49483807307697975,
      "grad_norm": 1.9749138355255127,
      "learning_rate": 1.9785573326264323e-05,
      "loss": 0.1231,
      "step": 33600
    },
    {
      "epoch": 0.49520625616706676,
      "grad_norm": 1.4517003297805786,
      "learning_rate": 1.980030043297694e-05,
      "loss": 0.1294,
      "step": 33625
    },
    {
      "epoch": 0.4955744392571538,
      "grad_norm": 1.6205573081970215,
      "learning_rate": 1.9815027539689556e-05,
      "loss": 0.1378,
      "step": 33650
    },
    {
      "epoch": 0.49594262234724085,
      "grad_norm": 1.3572132587432861,
      "learning_rate": 1.9829754646402167e-05,
      "loss": 0.132,
      "step": 33675
    },
    {
      "epoch": 0.49631080543732786,
      "grad_norm": 1.6994099617004395,
      "learning_rate": 1.9844481753114785e-05,
      "loss": 0.1294,
      "step": 33700
    },
    {
      "epoch": 0.49667898852741493,
      "grad_norm": 1.5217084884643555,
      "learning_rate": 1.98592088598274e-05,
      "loss": 0.1239,
      "step": 33725
    },
    {
      "epoch": 0.49704717161750195,
      "grad_norm": 1.607015609741211,
      "learning_rate": 1.9873935966540015e-05,
      "loss": 0.1259,
      "step": 33750
    },
    {
      "epoch": 0.49741535470758896,
      "grad_norm": 1.6565525531768799,
      "learning_rate": 1.988866307325263e-05,
      "loss": 0.121,
      "step": 33775
    },
    {
      "epoch": 0.49778353779767603,
      "grad_norm": 1.2221335172653198,
      "learning_rate": 1.9903390179965244e-05,
      "loss": 0.1261,
      "step": 33800
    },
    {
      "epoch": 0.49815172088776305,
      "grad_norm": 1.582075595855713,
      "learning_rate": 1.9918117286677862e-05,
      "loss": 0.1249,
      "step": 33825
    },
    {
      "epoch": 0.4985199039778501,
      "grad_norm": 1.5876294374465942,
      "learning_rate": 1.9932844393390477e-05,
      "loss": 0.125,
      "step": 33850
    },
    {
      "epoch": 0.49888808706793714,
      "grad_norm": 1.7459077835083008,
      "learning_rate": 1.994757150010309e-05,
      "loss": 0.1276,
      "step": 33875
    },
    {
      "epoch": 0.4992562701580242,
      "grad_norm": 1.3217989206314087,
      "learning_rate": 1.9962298606815706e-05,
      "loss": 0.1277,
      "step": 33900
    },
    {
      "epoch": 0.4996244532481112,
      "grad_norm": 1.583194375038147,
      "learning_rate": 1.997702571352832e-05,
      "loss": 0.1263,
      "step": 33925
    },
    {
      "epoch": 0.49999263633819824,
      "grad_norm": 1.1127156019210815,
      "learning_rate": 1.999175282024094e-05,
      "loss": 0.1203,
      "step": 33950
    },
    {
      "epoch": 0.5003608194282853,
      "grad_norm": 1.244364619255066,
      "learning_rate": 1.9999279996334528e-05,
      "loss": 0.1366,
      "step": 33975
    },
    {
      "epoch": 0.5007290025183724,
      "grad_norm": 1.5604734420776367,
      "learning_rate": 1.9997643624367546e-05,
      "loss": 0.1304,
      "step": 34000
    },
    {
      "epoch": 0.5010971856084594,
      "grad_norm": 1.4373990297317505,
      "learning_rate": 1.999600725240056e-05,
      "loss": 0.1344,
      "step": 34025
    },
    {
      "epoch": 0.5014653686985464,
      "grad_norm": 1.628875970840454,
      "learning_rate": 1.9994370880433575e-05,
      "loss": 0.1258,
      "step": 34050
    },
    {
      "epoch": 0.5018335517886334,
      "grad_norm": 1.3895543813705444,
      "learning_rate": 1.999273450846659e-05,
      "loss": 0.1268,
      "step": 34075
    },
    {
      "epoch": 0.5022017348787204,
      "grad_norm": 1.3760876655578613,
      "learning_rate": 1.9991098136499604e-05,
      "loss": 0.1273,
      "step": 34100
    },
    {
      "epoch": 0.5025699179688076,
      "grad_norm": 1.3263821601867676,
      "learning_rate": 1.998946176453262e-05,
      "loss": 0.1279,
      "step": 34125
    },
    {
      "epoch": 0.5029381010588946,
      "grad_norm": 1.43476402759552,
      "learning_rate": 1.9987825392565636e-05,
      "loss": 0.1205,
      "step": 34150
    },
    {
      "epoch": 0.5033062841489816,
      "grad_norm": 1.4423213005065918,
      "learning_rate": 1.998618902059865e-05,
      "loss": 0.122,
      "step": 34175
    },
    {
      "epoch": 0.5036744672390686,
      "grad_norm": 1.5728139877319336,
      "learning_rate": 1.9984552648631668e-05,
      "loss": 0.1245,
      "step": 34200
    },
    {
      "epoch": 0.5040426503291556,
      "grad_norm": 1.6493754386901855,
      "learning_rate": 1.9982916276664683e-05,
      "loss": 0.1212,
      "step": 34225
    },
    {
      "epoch": 0.5044108334192428,
      "grad_norm": 1.5477712154388428,
      "learning_rate": 1.99812799046977e-05,
      "loss": 0.124,
      "step": 34250
    },
    {
      "epoch": 0.5047790165093298,
      "grad_norm": 1.4639281034469604,
      "learning_rate": 1.9979643532730715e-05,
      "loss": 0.1305,
      "step": 34275
    },
    {
      "epoch": 0.5051471995994168,
      "grad_norm": 1.7540819644927979,
      "learning_rate": 1.997800716076373e-05,
      "loss": 0.1218,
      "step": 34300
    },
    {
      "epoch": 0.5055153826895038,
      "grad_norm": 1.3233895301818848,
      "learning_rate": 1.9976370788796744e-05,
      "loss": 0.1277,
      "step": 34325
    },
    {
      "epoch": 0.5058835657795909,
      "grad_norm": 1.4739441871643066,
      "learning_rate": 1.997473441682976e-05,
      "loss": 0.1324,
      "step": 34350
    },
    {
      "epoch": 0.5062517488696779,
      "grad_norm": 1.3000911474227905,
      "learning_rate": 1.9973098044862776e-05,
      "loss": 0.1283,
      "step": 34375
    },
    {
      "epoch": 0.506619931959765,
      "grad_norm": 1.398351788520813,
      "learning_rate": 1.997146167289579e-05,
      "loss": 0.1205,
      "step": 34400
    },
    {
      "epoch": 0.506988115049852,
      "grad_norm": 1.6885526180267334,
      "learning_rate": 1.9969825300928805e-05,
      "loss": 0.1243,
      "step": 34425
    },
    {
      "epoch": 0.507356298139939,
      "grad_norm": 1.6618565320968628,
      "learning_rate": 1.9968188928961823e-05,
      "loss": 0.1235,
      "step": 34450
    },
    {
      "epoch": 0.5077244812300261,
      "grad_norm": 1.8180230855941772,
      "learning_rate": 1.9966552556994837e-05,
      "loss": 0.13,
      "step": 34475
    },
    {
      "epoch": 0.5080926643201131,
      "grad_norm": 1.3617326021194458,
      "learning_rate": 1.9964916185027852e-05,
      "loss": 0.1236,
      "step": 34500
    },
    {
      "epoch": 0.5084608474102001,
      "grad_norm": 1.6490955352783203,
      "learning_rate": 1.9963279813060866e-05,
      "loss": 0.1202,
      "step": 34525
    },
    {
      "epoch": 0.5088290305002872,
      "grad_norm": 1.1779961585998535,
      "learning_rate": 1.9961643441093884e-05,
      "loss": 0.1271,
      "step": 34550
    },
    {
      "epoch": 0.5091972135903742,
      "grad_norm": 1.934651494026184,
      "learning_rate": 1.99600070691269e-05,
      "loss": 0.121,
      "step": 34575
    },
    {
      "epoch": 0.5095653966804613,
      "grad_norm": 1.6155422925949097,
      "learning_rate": 1.9958370697159913e-05,
      "loss": 0.1213,
      "step": 34600
    },
    {
      "epoch": 0.5099335797705483,
      "grad_norm": 1.213598370552063,
      "learning_rate": 1.995673432519293e-05,
      "loss": 0.123,
      "step": 34625
    },
    {
      "epoch": 0.5103017628606353,
      "grad_norm": 1.6968715190887451,
      "learning_rate": 1.9955097953225945e-05,
      "loss": 0.1187,
      "step": 34650
    },
    {
      "epoch": 0.5106699459507223,
      "grad_norm": 1.2820351123809814,
      "learning_rate": 1.995346158125896e-05,
      "loss": 0.118,
      "step": 34675
    },
    {
      "epoch": 0.5110381290408094,
      "grad_norm": 1.6911979913711548,
      "learning_rate": 1.9951825209291978e-05,
      "loss": 0.1292,
      "step": 34700
    },
    {
      "epoch": 0.5114063121308965,
      "grad_norm": 1.2066638469696045,
      "learning_rate": 1.9950188837324992e-05,
      "loss": 0.1208,
      "step": 34725
    },
    {
      "epoch": 0.5117744952209835,
      "grad_norm": 2.2320609092712402,
      "learning_rate": 1.9948552465358007e-05,
      "loss": 0.1215,
      "step": 34750
    },
    {
      "epoch": 0.5121426783110705,
      "grad_norm": 1.6644694805145264,
      "learning_rate": 1.994691609339102e-05,
      "loss": 0.1247,
      "step": 34775
    },
    {
      "epoch": 0.5125108614011575,
      "grad_norm": 1.2053136825561523,
      "learning_rate": 1.994527972142404e-05,
      "loss": 0.1217,
      "step": 34800
    },
    {
      "epoch": 0.5128790444912446,
      "grad_norm": 1.4280292987823486,
      "learning_rate": 1.9943643349457053e-05,
      "loss": 0.1215,
      "step": 34825
    },
    {
      "epoch": 0.5132472275813317,
      "grad_norm": 1.1937901973724365,
      "learning_rate": 1.9942006977490068e-05,
      "loss": 0.1311,
      "step": 34850
    },
    {
      "epoch": 0.5136154106714187,
      "grad_norm": 1.4878497123718262,
      "learning_rate": 1.9940370605523086e-05,
      "loss": 0.1215,
      "step": 34875
    },
    {
      "epoch": 0.5139835937615057,
      "grad_norm": 1.559525966644287,
      "learning_rate": 1.99387342335561e-05,
      "loss": 0.1175,
      "step": 34900
    },
    {
      "epoch": 0.5143517768515927,
      "grad_norm": 1.5552871227264404,
      "learning_rate": 1.9937097861589115e-05,
      "loss": 0.1291,
      "step": 34925
    },
    {
      "epoch": 0.5147199599416798,
      "grad_norm": 1.4817222356796265,
      "learning_rate": 1.993546148962213e-05,
      "loss": 0.1179,
      "step": 34950
    },
    {
      "epoch": 0.5150881430317669,
      "grad_norm": 1.1439144611358643,
      "learning_rate": 1.9933890572533826e-05,
      "loss": 0.1237,
      "step": 34975
    },
    {
      "epoch": 0.5154563261218539,
      "grad_norm": 1.3597592115402222,
      "learning_rate": 1.993225420056684e-05,
      "loss": 0.1226,
      "step": 35000
    },
    {
      "epoch": 0.5158245092119409,
      "grad_norm": 1.6552609205245972,
      "learning_rate": 1.9930617828599855e-05,
      "loss": 0.1275,
      "step": 35025
    },
    {
      "epoch": 0.5161926923020279,
      "grad_norm": 1.4046581983566284,
      "learning_rate": 1.9928981456632873e-05,
      "loss": 0.1303,
      "step": 35050
    },
    {
      "epoch": 0.516560875392115,
      "grad_norm": 1.3805830478668213,
      "learning_rate": 1.9927345084665888e-05,
      "loss": 0.1208,
      "step": 35075
    },
    {
      "epoch": 0.516929058482202,
      "grad_norm": 1.6304214000701904,
      "learning_rate": 1.9925708712698902e-05,
      "loss": 0.1273,
      "step": 35100
    },
    {
      "epoch": 0.5172972415722891,
      "grad_norm": 1.558475136756897,
      "learning_rate": 1.9924072340731917e-05,
      "loss": 0.1203,
      "step": 35125
    },
    {
      "epoch": 0.5176654246623761,
      "grad_norm": 2.1353535652160645,
      "learning_rate": 1.9922435968764934e-05,
      "loss": 0.1229,
      "step": 35150
    },
    {
      "epoch": 0.5180336077524631,
      "grad_norm": 1.4988514184951782,
      "learning_rate": 1.992079959679795e-05,
      "loss": 0.1369,
      "step": 35175
    },
    {
      "epoch": 0.5184017908425502,
      "grad_norm": 1.7591900825500488,
      "learning_rate": 1.9919163224830963e-05,
      "loss": 0.1262,
      "step": 35200
    },
    {
      "epoch": 0.5187699739326372,
      "grad_norm": 1.1921674013137817,
      "learning_rate": 1.991752685286398e-05,
      "loss": 0.1204,
      "step": 35225
    },
    {
      "epoch": 0.5191381570227243,
      "grad_norm": 1.6571495532989502,
      "learning_rate": 1.9915890480896996e-05,
      "loss": 0.1244,
      "step": 35250
    },
    {
      "epoch": 0.5195063401128113,
      "grad_norm": 1.442743182182312,
      "learning_rate": 1.991425410893001e-05,
      "loss": 0.1264,
      "step": 35275
    },
    {
      "epoch": 0.5198745232028983,
      "grad_norm": 1.2768187522888184,
      "learning_rate": 1.9912617736963028e-05,
      "loss": 0.1311,
      "step": 35300
    },
    {
      "epoch": 0.5202427062929854,
      "grad_norm": 1.5320920944213867,
      "learning_rate": 1.9910981364996042e-05,
      "loss": 0.1273,
      "step": 35325
    },
    {
      "epoch": 0.5206108893830724,
      "grad_norm": 1.7867873907089233,
      "learning_rate": 1.9909344993029057e-05,
      "loss": 0.1244,
      "step": 35350
    },
    {
      "epoch": 0.5209790724731594,
      "grad_norm": 2.1170566082000732,
      "learning_rate": 1.990770862106207e-05,
      "loss": 0.1271,
      "step": 35375
    },
    {
      "epoch": 0.5213472555632465,
      "grad_norm": 1.2551827430725098,
      "learning_rate": 1.990607224909509e-05,
      "loss": 0.1083,
      "step": 35400
    },
    {
      "epoch": 0.5217154386533335,
      "grad_norm": 1.4103198051452637,
      "learning_rate": 1.9904435877128104e-05,
      "loss": 0.1198,
      "step": 35425
    },
    {
      "epoch": 0.5220836217434206,
      "grad_norm": 1.889062762260437,
      "learning_rate": 1.9902799505161118e-05,
      "loss": 0.1156,
      "step": 35450
    },
    {
      "epoch": 0.5224518048335076,
      "grad_norm": 1.1291379928588867,
      "learning_rate": 1.9901163133194136e-05,
      "loss": 0.1194,
      "step": 35475
    },
    {
      "epoch": 0.5228199879235946,
      "grad_norm": 1.7137246131896973,
      "learning_rate": 1.989952676122715e-05,
      "loss": 0.1299,
      "step": 35500
    },
    {
      "epoch": 0.5231881710136816,
      "grad_norm": 1.4984245300292969,
      "learning_rate": 1.9897890389260165e-05,
      "loss": 0.126,
      "step": 35525
    },
    {
      "epoch": 0.5235563541037688,
      "grad_norm": 1.8383368253707886,
      "learning_rate": 1.989625401729318e-05,
      "loss": 0.1202,
      "step": 35550
    },
    {
      "epoch": 0.5239245371938558,
      "grad_norm": 1.216751217842102,
      "learning_rate": 1.9894617645326194e-05,
      "loss": 0.1217,
      "step": 35575
    },
    {
      "epoch": 0.5242927202839428,
      "grad_norm": 1.2445589303970337,
      "learning_rate": 1.989298127335921e-05,
      "loss": 0.1271,
      "step": 35600
    },
    {
      "epoch": 0.5246609033740298,
      "grad_norm": 1.3670048713684082,
      "learning_rate": 1.9891344901392226e-05,
      "loss": 0.123,
      "step": 35625
    },
    {
      "epoch": 0.5250290864641168,
      "grad_norm": 1.6851133108139038,
      "learning_rate": 1.9889708529425244e-05,
      "loss": 0.1248,
      "step": 35650
    },
    {
      "epoch": 0.525397269554204,
      "grad_norm": 1.3160358667373657,
      "learning_rate": 1.9888072157458258e-05,
      "loss": 0.1241,
      "step": 35675
    },
    {
      "epoch": 0.525765452644291,
      "grad_norm": 1.607149362564087,
      "learning_rate": 1.9886435785491273e-05,
      "loss": 0.1278,
      "step": 35700
    },
    {
      "epoch": 0.526133635734378,
      "grad_norm": 1.6950891017913818,
      "learning_rate": 1.988479941352429e-05,
      "loss": 0.1233,
      "step": 35725
    },
    {
      "epoch": 0.526501818824465,
      "grad_norm": 1.227681040763855,
      "learning_rate": 1.9883163041557305e-05,
      "loss": 0.1216,
      "step": 35750
    },
    {
      "epoch": 0.526870001914552,
      "grad_norm": 1.3604323863983154,
      "learning_rate": 1.988152666959032e-05,
      "loss": 0.1126,
      "step": 35775
    },
    {
      "epoch": 0.5272381850046391,
      "grad_norm": 1.1941028833389282,
      "learning_rate": 1.9879890297623334e-05,
      "loss": 0.1206,
      "step": 35800
    },
    {
      "epoch": 0.5276063680947262,
      "grad_norm": 1.280265212059021,
      "learning_rate": 1.987825392565635e-05,
      "loss": 0.1251,
      "step": 35825
    },
    {
      "epoch": 0.5279745511848132,
      "grad_norm": 1.2790465354919434,
      "learning_rate": 1.9876617553689366e-05,
      "loss": 0.1224,
      "step": 35850
    },
    {
      "epoch": 0.5283427342749002,
      "grad_norm": 1.8109391927719116,
      "learning_rate": 1.987498118172238e-05,
      "loss": 0.1248,
      "step": 35875
    },
    {
      "epoch": 0.5287109173649872,
      "grad_norm": 1.5231170654296875,
      "learning_rate": 1.98733448097554e-05,
      "loss": 0.1223,
      "step": 35900
    },
    {
      "epoch": 0.5290791004550743,
      "grad_norm": 1.469890832901001,
      "learning_rate": 1.9871708437788413e-05,
      "loss": 0.1143,
      "step": 35925
    },
    {
      "epoch": 0.5294472835451614,
      "grad_norm": 1.472482442855835,
      "learning_rate": 1.9870072065821427e-05,
      "loss": 0.1156,
      "step": 35950
    },
    {
      "epoch": 0.5298154666352484,
      "grad_norm": 1.5676894187927246,
      "learning_rate": 1.9868435693854442e-05,
      "loss": 0.1184,
      "step": 35975
    },
    {
      "epoch": 0.5301836497253354,
      "grad_norm": 1.5137412548065186,
      "learning_rate": 1.9866799321887456e-05,
      "loss": 0.1203,
      "step": 36000
    },
    {
      "epoch": 0.5305518328154225,
      "grad_norm": 1.3111070394515991,
      "learning_rate": 1.9865162949920474e-05,
      "loss": 0.1177,
      "step": 36025
    },
    {
      "epoch": 0.5309200159055095,
      "grad_norm": 1.5890854597091675,
      "learning_rate": 1.986352657795349e-05,
      "loss": 0.1261,
      "step": 36050
    },
    {
      "epoch": 0.5312881989955965,
      "grad_norm": 0.9979857802391052,
      "learning_rate": 1.9861890205986507e-05,
      "loss": 0.1285,
      "step": 36075
    },
    {
      "epoch": 0.5316563820856836,
      "grad_norm": 1.7929030656814575,
      "learning_rate": 1.986025383401952e-05,
      "loss": 0.1184,
      "step": 36100
    },
    {
      "epoch": 0.5320245651757706,
      "grad_norm": 1.296306848526001,
      "learning_rate": 1.9858617462052535e-05,
      "loss": 0.114,
      "step": 36125
    },
    {
      "epoch": 0.5323927482658577,
      "grad_norm": 1.580112099647522,
      "learning_rate": 1.9856981090085553e-05,
      "loss": 0.1228,
      "step": 36150
    },
    {
      "epoch": 0.5327609313559447,
      "grad_norm": 1.4098597764968872,
      "learning_rate": 1.9855344718118568e-05,
      "loss": 0.1223,
      "step": 36175
    },
    {
      "epoch": 0.5331291144460317,
      "grad_norm": 1.383877158164978,
      "learning_rate": 1.9853708346151582e-05,
      "loss": 0.1224,
      "step": 36200
    },
    {
      "epoch": 0.5334972975361187,
      "grad_norm": 1.6452189683914185,
      "learning_rate": 1.9852071974184597e-05,
      "loss": 0.1195,
      "step": 36225
    },
    {
      "epoch": 0.5338654806262058,
      "grad_norm": 1.42545485496521,
      "learning_rate": 1.985043560221761e-05,
      "loss": 0.1148,
      "step": 36250
    },
    {
      "epoch": 0.5342336637162929,
      "grad_norm": 1.2486947774887085,
      "learning_rate": 1.984879923025063e-05,
      "loss": 0.1247,
      "step": 36275
    },
    {
      "epoch": 0.5346018468063799,
      "grad_norm": 1.4072372913360596,
      "learning_rate": 1.9847162858283643e-05,
      "loss": 0.1255,
      "step": 36300
    },
    {
      "epoch": 0.5349700298964669,
      "grad_norm": 1.4645627737045288,
      "learning_rate": 1.984552648631666e-05,
      "loss": 0.1229,
      "step": 36325
    },
    {
      "epoch": 0.5353382129865539,
      "grad_norm": 1.3357573747634888,
      "learning_rate": 1.9843890114349676e-05,
      "loss": 0.115,
      "step": 36350
    },
    {
      "epoch": 0.535706396076641,
      "grad_norm": 1.6706513166427612,
      "learning_rate": 1.984225374238269e-05,
      "loss": 0.1136,
      "step": 36375
    },
    {
      "epoch": 0.5360745791667281,
      "grad_norm": 1.2168182134628296,
      "learning_rate": 1.9840617370415705e-05,
      "loss": 0.1192,
      "step": 36400
    },
    {
      "epoch": 0.5364427622568151,
      "grad_norm": 1.2111717462539673,
      "learning_rate": 1.983898099844872e-05,
      "loss": 0.1244,
      "step": 36425
    },
    {
      "epoch": 0.5368109453469021,
      "grad_norm": 1.5248171091079712,
      "learning_rate": 1.9837344626481737e-05,
      "loss": 0.1286,
      "step": 36450
    },
    {
      "epoch": 0.5371791284369891,
      "grad_norm": 2.1165120601654053,
      "learning_rate": 1.983570825451475e-05,
      "loss": 0.1296,
      "step": 36475
    },
    {
      "epoch": 0.5375473115270761,
      "grad_norm": 1.4845212697982788,
      "learning_rate": 1.9834071882547766e-05,
      "loss": 0.1204,
      "step": 36500
    },
    {
      "epoch": 0.5379154946171633,
      "grad_norm": 1.6977769136428833,
      "learning_rate": 1.9832435510580784e-05,
      "loss": 0.1177,
      "step": 36525
    },
    {
      "epoch": 0.5382836777072503,
      "grad_norm": 1.3811136484146118,
      "learning_rate": 1.9830799138613798e-05,
      "loss": 0.124,
      "step": 36550
    },
    {
      "epoch": 0.5386518607973373,
      "grad_norm": 1.491607427597046,
      "learning_rate": 1.9829162766646816e-05,
      "loss": 0.1225,
      "step": 36575
    },
    {
      "epoch": 0.5390200438874243,
      "grad_norm": 1.7326167821884155,
      "learning_rate": 1.9827526394679827e-05,
      "loss": 0.1231,
      "step": 36600
    },
    {
      "epoch": 0.5393882269775114,
      "grad_norm": 1.3634510040283203,
      "learning_rate": 1.9825890022712845e-05,
      "loss": 0.1171,
      "step": 36625
    },
    {
      "epoch": 0.5397564100675984,
      "grad_norm": 1.6107808351516724,
      "learning_rate": 1.982425365074586e-05,
      "loss": 0.1237,
      "step": 36650
    },
    {
      "epoch": 0.5401245931576855,
      "grad_norm": 1.855141282081604,
      "learning_rate": 1.9822617278778874e-05,
      "loss": 0.1161,
      "step": 36675
    },
    {
      "epoch": 0.5404927762477725,
      "grad_norm": 1.6258479356765747,
      "learning_rate": 1.9820980906811892e-05,
      "loss": 0.118,
      "step": 36700
    },
    {
      "epoch": 0.5408609593378595,
      "grad_norm": 1.2577881813049316,
      "learning_rate": 1.9819344534844906e-05,
      "loss": 0.1243,
      "step": 36725
    },
    {
      "epoch": 0.5412291424279466,
      "grad_norm": 1.2613511085510254,
      "learning_rate": 1.981770816287792e-05,
      "loss": 0.1191,
      "step": 36750
    },
    {
      "epoch": 0.5415973255180336,
      "grad_norm": 1.5669910907745361,
      "learning_rate": 1.981607179091094e-05,
      "loss": 0.1271,
      "step": 36775
    },
    {
      "epoch": 0.5419655086081206,
      "grad_norm": 1.261261224746704,
      "learning_rate": 1.9814435418943953e-05,
      "loss": 0.1093,
      "step": 36800
    },
    {
      "epoch": 0.5423336916982077,
      "grad_norm": 1.5597432851791382,
      "learning_rate": 1.9812799046976967e-05,
      "loss": 0.1168,
      "step": 36825
    },
    {
      "epoch": 0.5427018747882947,
      "grad_norm": 1.5948351621627808,
      "learning_rate": 1.9811162675009982e-05,
      "loss": 0.1232,
      "step": 36850
    },
    {
      "epoch": 0.5430700578783818,
      "grad_norm": 1.305525779724121,
      "learning_rate": 1.9809526303043e-05,
      "loss": 0.1186,
      "step": 36875
    },
    {
      "epoch": 0.5434382409684688,
      "grad_norm": 1.4193530082702637,
      "learning_rate": 1.9807889931076014e-05,
      "loss": 0.1246,
      "step": 36900
    },
    {
      "epoch": 0.5438064240585558,
      "grad_norm": 1.3771121501922607,
      "learning_rate": 1.980625355910903e-05,
      "loss": 0.1221,
      "step": 36925
    },
    {
      "epoch": 0.5441746071486429,
      "grad_norm": 1.1653172969818115,
      "learning_rate": 1.9804617187142046e-05,
      "loss": 0.1185,
      "step": 36950
    },
    {
      "epoch": 0.5445427902387299,
      "grad_norm": 1.3549933433532715,
      "learning_rate": 1.980298081517506e-05,
      "loss": 0.1199,
      "step": 36975
    },
    {
      "epoch": 0.544910973328817,
      "grad_norm": 1.6777881383895874,
      "learning_rate": 1.9801344443208075e-05,
      "loss": 0.129,
      "step": 37000
    },
    {
      "epoch": 0.545279156418904,
      "grad_norm": 1.0790871381759644,
      "learning_rate": 1.979977352611977e-05,
      "loss": 0.1126,
      "step": 37025
    },
    {
      "epoch": 0.545647339508991,
      "grad_norm": 1.8433680534362793,
      "learning_rate": 1.9798137154152787e-05,
      "loss": 0.1162,
      "step": 37050
    },
    {
      "epoch": 0.546015522599078,
      "grad_norm": 2.001495838165283,
      "learning_rate": 1.97965007821858e-05,
      "loss": 0.121,
      "step": 37075
    },
    {
      "epoch": 0.5463837056891651,
      "grad_norm": 1.0270895957946777,
      "learning_rate": 1.9794864410218816e-05,
      "loss": 0.118,
      "step": 37100
    },
    {
      "epoch": 0.5467518887792522,
      "grad_norm": 1.172116756439209,
      "learning_rate": 1.9793228038251834e-05,
      "loss": 0.116,
      "step": 37125
    },
    {
      "epoch": 0.5471200718693392,
      "grad_norm": 1.4375494718551636,
      "learning_rate": 1.9791591666284848e-05,
      "loss": 0.1208,
      "step": 37150
    },
    {
      "epoch": 0.5474882549594262,
      "grad_norm": 1.8035261631011963,
      "learning_rate": 1.9789955294317863e-05,
      "loss": 0.125,
      "step": 37175
    },
    {
      "epoch": 0.5478564380495132,
      "grad_norm": 1.0941935777664185,
      "learning_rate": 1.978831892235088e-05,
      "loss": 0.1255,
      "step": 37200
    },
    {
      "epoch": 0.5482246211396004,
      "grad_norm": 1.4023436307907104,
      "learning_rate": 1.9786682550383895e-05,
      "loss": 0.1149,
      "step": 37225
    },
    {
      "epoch": 0.5485928042296874,
      "grad_norm": 1.587387204170227,
      "learning_rate": 1.978504617841691e-05,
      "loss": 0.1218,
      "step": 37250
    },
    {
      "epoch": 0.5489609873197744,
      "grad_norm": 1.603115439414978,
      "learning_rate": 1.9783409806449924e-05,
      "loss": 0.1238,
      "step": 37275
    },
    {
      "epoch": 0.5493291704098614,
      "grad_norm": 1.7399941682815552,
      "learning_rate": 1.9781773434482942e-05,
      "loss": 0.1158,
      "step": 37300
    },
    {
      "epoch": 0.5496973534999484,
      "grad_norm": 1.4176956415176392,
      "learning_rate": 1.9780137062515956e-05,
      "loss": 0.12,
      "step": 37325
    },
    {
      "epoch": 0.5500655365900355,
      "grad_norm": 1.8631614446640015,
      "learning_rate": 1.977850069054897e-05,
      "loss": 0.1212,
      "step": 37350
    },
    {
      "epoch": 0.5504337196801226,
      "grad_norm": 1.2380393743515015,
      "learning_rate": 1.977686431858199e-05,
      "loss": 0.1178,
      "step": 37375
    },
    {
      "epoch": 0.5508019027702096,
      "grad_norm": 1.3811695575714111,
      "learning_rate": 1.9775227946615003e-05,
      "loss": 0.1184,
      "step": 37400
    },
    {
      "epoch": 0.5511700858602966,
      "grad_norm": 1.2970376014709473,
      "learning_rate": 1.9773591574648017e-05,
      "loss": 0.1135,
      "step": 37425
    },
    {
      "epoch": 0.5515382689503836,
      "grad_norm": 1.7509592771530151,
      "learning_rate": 1.9771955202681032e-05,
      "loss": 0.1177,
      "step": 37450
    },
    {
      "epoch": 0.5519064520404707,
      "grad_norm": 1.4221574068069458,
      "learning_rate": 1.977031883071405e-05,
      "loss": 0.111,
      "step": 37475
    },
    {
      "epoch": 0.5522746351305577,
      "grad_norm": 1.5943177938461304,
      "learning_rate": 1.9768682458747064e-05,
      "loss": 0.1286,
      "step": 37500
    },
    {
      "epoch": 0.5526428182206448,
      "grad_norm": 1.7850430011749268,
      "learning_rate": 1.976704608678008e-05,
      "loss": 0.1244,
      "step": 37525
    },
    {
      "epoch": 0.5530110013107318,
      "grad_norm": 1.545997142791748,
      "learning_rate": 1.9765409714813097e-05,
      "loss": 0.1292,
      "step": 37550
    },
    {
      "epoch": 0.5533791844008188,
      "grad_norm": 2.268383264541626,
      "learning_rate": 1.976377334284611e-05,
      "loss": 0.1286,
      "step": 37575
    },
    {
      "epoch": 0.5537473674909059,
      "grad_norm": 1.4364897012710571,
      "learning_rate": 1.9762136970879125e-05,
      "loss": 0.1189,
      "step": 37600
    },
    {
      "epoch": 0.5541155505809929,
      "grad_norm": 1.4818049669265747,
      "learning_rate": 1.9760500598912143e-05,
      "loss": 0.1208,
      "step": 37625
    },
    {
      "epoch": 0.55448373367108,
      "grad_norm": 1.2418051958084106,
      "learning_rate": 1.9758864226945158e-05,
      "loss": 0.1205,
      "step": 37650
    },
    {
      "epoch": 0.554851916761167,
      "grad_norm": 1.3513381481170654,
      "learning_rate": 1.9757227854978172e-05,
      "loss": 0.121,
      "step": 37675
    },
    {
      "epoch": 0.555220099851254,
      "grad_norm": 1.4748746156692505,
      "learning_rate": 1.9755591483011187e-05,
      "loss": 0.1122,
      "step": 37700
    },
    {
      "epoch": 0.5555882829413411,
      "grad_norm": 1.6285781860351562,
      "learning_rate": 1.9753955111044205e-05,
      "loss": 0.1217,
      "step": 37725
    },
    {
      "epoch": 0.5559564660314281,
      "grad_norm": 1.184108853340149,
      "learning_rate": 1.975231873907722e-05,
      "loss": 0.1116,
      "step": 37750
    },
    {
      "epoch": 0.5563246491215151,
      "grad_norm": 1.7575715780258179,
      "learning_rate": 1.9750682367110233e-05,
      "loss": 0.114,
      "step": 37775
    },
    {
      "epoch": 0.5566928322116022,
      "grad_norm": 1.788015365600586,
      "learning_rate": 1.974904599514325e-05,
      "loss": 0.1172,
      "step": 37800
    },
    {
      "epoch": 0.5570610153016893,
      "grad_norm": 1.2830640077590942,
      "learning_rate": 1.9747409623176266e-05,
      "loss": 0.1234,
      "step": 37825
    },
    {
      "epoch": 0.5574291983917763,
      "grad_norm": 1.3260087966918945,
      "learning_rate": 1.974577325120928e-05,
      "loss": 0.1166,
      "step": 37850
    },
    {
      "epoch": 0.5577973814818633,
      "grad_norm": 1.5391265153884888,
      "learning_rate": 1.9744136879242295e-05,
      "loss": 0.124,
      "step": 37875
    },
    {
      "epoch": 0.5581655645719503,
      "grad_norm": 1.5043689012527466,
      "learning_rate": 1.974250050727531e-05,
      "loss": 0.1142,
      "step": 37900
    },
    {
      "epoch": 0.5585337476620373,
      "grad_norm": 1.7274752855300903,
      "learning_rate": 1.9740864135308327e-05,
      "loss": 0.1195,
      "step": 37925
    },
    {
      "epoch": 0.5589019307521245,
      "grad_norm": 1.496433973312378,
      "learning_rate": 1.973922776334134e-05,
      "loss": 0.1108,
      "step": 37950
    },
    {
      "epoch": 0.5592701138422115,
      "grad_norm": 1.4745001792907715,
      "learning_rate": 1.973759139137436e-05,
      "loss": 0.1136,
      "step": 37975
    },
    {
      "epoch": 0.5596382969322985,
      "grad_norm": 1.769690990447998,
      "learning_rate": 1.9735955019407374e-05,
      "loss": 0.1151,
      "step": 38000
    },
    {
      "epoch": 0.5600064800223855,
      "grad_norm": 1.2575122117996216,
      "learning_rate": 1.9734318647440388e-05,
      "loss": 0.1192,
      "step": 38025
    },
    {
      "epoch": 0.5603746631124725,
      "grad_norm": 1.5350185632705688,
      "learning_rate": 1.9732682275473406e-05,
      "loss": 0.1184,
      "step": 38050
    },
    {
      "epoch": 0.5607428462025597,
      "grad_norm": 1.7502778768539429,
      "learning_rate": 1.973104590350642e-05,
      "loss": 0.1367,
      "step": 38075
    },
    {
      "epoch": 0.5611110292926467,
      "grad_norm": 1.618236780166626,
      "learning_rate": 1.9729409531539435e-05,
      "loss": 0.114,
      "step": 38100
    },
    {
      "epoch": 0.5614792123827337,
      "grad_norm": 1.5210295915603638,
      "learning_rate": 1.972777315957245e-05,
      "loss": 0.1084,
      "step": 38125
    },
    {
      "epoch": 0.5618473954728207,
      "grad_norm": 1.673702597618103,
      "learning_rate": 1.9726136787605464e-05,
      "loss": 0.1231,
      "step": 38150
    },
    {
      "epoch": 0.5622155785629077,
      "grad_norm": 1.2953345775604248,
      "learning_rate": 1.9724500415638482e-05,
      "loss": 0.1099,
      "step": 38175
    },
    {
      "epoch": 0.5625837616529948,
      "grad_norm": 1.6855987310409546,
      "learning_rate": 1.9722864043671496e-05,
      "loss": 0.1179,
      "step": 38200
    },
    {
      "epoch": 0.5629519447430819,
      "grad_norm": 1.5913565158843994,
      "learning_rate": 1.9721227671704514e-05,
      "loss": 0.1143,
      "step": 38225
    },
    {
      "epoch": 0.5633201278331689,
      "grad_norm": 1.6520943641662598,
      "learning_rate": 1.971959129973753e-05,
      "loss": 0.1201,
      "step": 38250
    },
    {
      "epoch": 0.5636883109232559,
      "grad_norm": 1.2099671363830566,
      "learning_rate": 1.9717954927770543e-05,
      "loss": 0.1135,
      "step": 38275
    },
    {
      "epoch": 0.5640564940133429,
      "grad_norm": 1.32758629322052,
      "learning_rate": 1.9716318555803557e-05,
      "loss": 0.1127,
      "step": 38300
    },
    {
      "epoch": 0.56442467710343,
      "grad_norm": 1.3274176120758057,
      "learning_rate": 1.9714682183836572e-05,
      "loss": 0.1121,
      "step": 38325
    },
    {
      "epoch": 0.564792860193517,
      "grad_norm": 1.2761715650558472,
      "learning_rate": 1.971304581186959e-05,
      "loss": 0.1192,
      "step": 38350
    },
    {
      "epoch": 0.5651610432836041,
      "grad_norm": 1.3949676752090454,
      "learning_rate": 1.9711409439902604e-05,
      "loss": 0.1139,
      "step": 38375
    },
    {
      "epoch": 0.5655292263736911,
      "grad_norm": 1.7472232580184937,
      "learning_rate": 1.970977306793562e-05,
      "loss": 0.1234,
      "step": 38400
    },
    {
      "epoch": 0.5658974094637782,
      "grad_norm": 1.031726598739624,
      "learning_rate": 1.9708136695968636e-05,
      "loss": 0.1161,
      "step": 38425
    },
    {
      "epoch": 0.5662655925538652,
      "grad_norm": 1.216566562652588,
      "learning_rate": 1.970650032400165e-05,
      "loss": 0.1218,
      "step": 38450
    },
    {
      "epoch": 0.5666337756439522,
      "grad_norm": 1.6059186458587646,
      "learning_rate": 1.970486395203467e-05,
      "loss": 0.1126,
      "step": 38475
    },
    {
      "epoch": 0.5670019587340392,
      "grad_norm": 1.2862287759780884,
      "learning_rate": 1.970322758006768e-05,
      "loss": 0.1151,
      "step": 38500
    },
    {
      "epoch": 0.5673701418241263,
      "grad_norm": 1.1815801858901978,
      "learning_rate": 1.9701591208100698e-05,
      "loss": 0.1145,
      "step": 38525
    },
    {
      "epoch": 0.5677383249142134,
      "grad_norm": 1.4452340602874756,
      "learning_rate": 1.9699954836133712e-05,
      "loss": 0.1203,
      "step": 38550
    },
    {
      "epoch": 0.5681065080043004,
      "grad_norm": 1.355634331703186,
      "learning_rate": 1.9698318464166727e-05,
      "loss": 0.114,
      "step": 38575
    },
    {
      "epoch": 0.5684746910943874,
      "grad_norm": 1.4844880104064941,
      "learning_rate": 1.9696682092199744e-05,
      "loss": 0.1172,
      "step": 38600
    },
    {
      "epoch": 0.5688428741844744,
      "grad_norm": 2.0493764877319336,
      "learning_rate": 1.969504572023276e-05,
      "loss": 0.1178,
      "step": 38625
    },
    {
      "epoch": 0.5692110572745614,
      "grad_norm": 1.5760899782180786,
      "learning_rate": 1.9693409348265777e-05,
      "loss": 0.1172,
      "step": 38650
    },
    {
      "epoch": 0.5695792403646486,
      "grad_norm": 1.7074697017669678,
      "learning_rate": 1.969177297629879e-05,
      "loss": 0.1114,
      "step": 38675
    },
    {
      "epoch": 0.5699474234547356,
      "grad_norm": 1.3330516815185547,
      "learning_rate": 1.9690136604331806e-05,
      "loss": 0.1207,
      "step": 38700
    },
    {
      "epoch": 0.5703156065448226,
      "grad_norm": 1.3043917417526245,
      "learning_rate": 1.968850023236482e-05,
      "loss": 0.1239,
      "step": 38725
    },
    {
      "epoch": 0.5706837896349096,
      "grad_norm": 1.8919068574905396,
      "learning_rate": 1.9686863860397835e-05,
      "loss": 0.1256,
      "step": 38750
    },
    {
      "epoch": 0.5710519727249966,
      "grad_norm": 1.4486078023910522,
      "learning_rate": 1.9685227488430852e-05,
      "loss": 0.1255,
      "step": 38775
    },
    {
      "epoch": 0.5714201558150838,
      "grad_norm": 1.5534908771514893,
      "learning_rate": 1.9683591116463867e-05,
      "loss": 0.1135,
      "step": 38800
    },
    {
      "epoch": 0.5717883389051708,
      "grad_norm": 1.693839192390442,
      "learning_rate": 1.968195474449688e-05,
      "loss": 0.1161,
      "step": 38825
    },
    {
      "epoch": 0.5721565219952578,
      "grad_norm": 1.4074006080627441,
      "learning_rate": 1.96803183725299e-05,
      "loss": 0.1145,
      "step": 38850
    },
    {
      "epoch": 0.5725247050853448,
      "grad_norm": 1.6323904991149902,
      "learning_rate": 1.9678682000562914e-05,
      "loss": 0.1113,
      "step": 38875
    },
    {
      "epoch": 0.5728928881754318,
      "grad_norm": 1.2794816493988037,
      "learning_rate": 1.967704562859593e-05,
      "loss": 0.1098,
      "step": 38900
    },
    {
      "epoch": 0.573261071265519,
      "grad_norm": 1.4640345573425293,
      "learning_rate": 1.9675409256628943e-05,
      "loss": 0.1205,
      "step": 38925
    },
    {
      "epoch": 0.573629254355606,
      "grad_norm": 1.3224183320999146,
      "learning_rate": 1.967377288466196e-05,
      "loss": 0.1095,
      "step": 38950
    },
    {
      "epoch": 0.573997437445693,
      "grad_norm": 1.985844373703003,
      "learning_rate": 1.9672136512694975e-05,
      "loss": 0.1127,
      "step": 38975
    },
    {
      "epoch": 0.57436562053578,
      "grad_norm": 1.6131892204284668,
      "learning_rate": 1.967050014072799e-05,
      "loss": 0.1194,
      "step": 39000
    },
    {
      "epoch": 0.5747338036258671,
      "grad_norm": 1.5700842142105103,
      "learning_rate": 1.9668863768761007e-05,
      "loss": 0.1233,
      "step": 39025
    },
    {
      "epoch": 0.5751019867159541,
      "grad_norm": 1.1362309455871582,
      "learning_rate": 1.966722739679402e-05,
      "loss": 0.1098,
      "step": 39050
    },
    {
      "epoch": 0.5754701698060412,
      "grad_norm": 1.4811644554138184,
      "learning_rate": 1.9665656479705715e-05,
      "loss": 0.1199,
      "step": 39075
    },
    {
      "epoch": 0.5758383528961282,
      "grad_norm": 1.5162842273712158,
      "learning_rate": 1.9664020107738733e-05,
      "loss": 0.1218,
      "step": 39100
    },
    {
      "epoch": 0.5762065359862152,
      "grad_norm": 1.2654857635498047,
      "learning_rate": 1.9662383735771748e-05,
      "loss": 0.1098,
      "step": 39125
    },
    {
      "epoch": 0.5765747190763023,
      "grad_norm": 1.4312494993209839,
      "learning_rate": 1.9660747363804762e-05,
      "loss": 0.1145,
      "step": 39150
    },
    {
      "epoch": 0.5769429021663893,
      "grad_norm": 1.5518170595169067,
      "learning_rate": 1.9659110991837777e-05,
      "loss": 0.1133,
      "step": 39175
    },
    {
      "epoch": 0.5773110852564763,
      "grad_norm": 1.2797623872756958,
      "learning_rate": 1.9657474619870795e-05,
      "loss": 0.1138,
      "step": 39200
    },
    {
      "epoch": 0.5776792683465634,
      "grad_norm": 1.4248781204223633,
      "learning_rate": 1.965583824790381e-05,
      "loss": 0.1207,
      "step": 39225
    },
    {
      "epoch": 0.5780474514366504,
      "grad_norm": 2.2524561882019043,
      "learning_rate": 1.9654201875936823e-05,
      "loss": 0.1107,
      "step": 39250
    },
    {
      "epoch": 0.5784156345267375,
      "grad_norm": 1.7344497442245483,
      "learning_rate": 1.965256550396984e-05,
      "loss": 0.1127,
      "step": 39275
    },
    {
      "epoch": 0.5787838176168245,
      "grad_norm": 1.7456754446029663,
      "learning_rate": 1.9650929132002856e-05,
      "loss": 0.1112,
      "step": 39300
    },
    {
      "epoch": 0.5791520007069115,
      "grad_norm": 1.370923399925232,
      "learning_rate": 1.964929276003587e-05,
      "loss": 0.1093,
      "step": 39325
    },
    {
      "epoch": 0.5795201837969985,
      "grad_norm": 1.7654410600662231,
      "learning_rate": 1.9647656388068885e-05,
      "loss": 0.119,
      "step": 39350
    },
    {
      "epoch": 0.5798883668870856,
      "grad_norm": 1.5445888042449951,
      "learning_rate": 1.9646020016101903e-05,
      "loss": 0.1118,
      "step": 39375
    },
    {
      "epoch": 0.5802565499771727,
      "grad_norm": 1.3537577390670776,
      "learning_rate": 1.9644383644134917e-05,
      "loss": 0.1156,
      "step": 39400
    },
    {
      "epoch": 0.5806247330672597,
      "grad_norm": 1.781290054321289,
      "learning_rate": 1.964274727216793e-05,
      "loss": 0.11,
      "step": 39425
    },
    {
      "epoch": 0.5809929161573467,
      "grad_norm": 1.2838529348373413,
      "learning_rate": 1.964111090020095e-05,
      "loss": 0.1211,
      "step": 39450
    },
    {
      "epoch": 0.5813610992474337,
      "grad_norm": 1.7609925270080566,
      "learning_rate": 1.9639474528233964e-05,
      "loss": 0.1171,
      "step": 39475
    },
    {
      "epoch": 0.5817292823375207,
      "grad_norm": 1.424841046333313,
      "learning_rate": 1.9637838156266978e-05,
      "loss": 0.1137,
      "step": 39500
    },
    {
      "epoch": 0.5820974654276079,
      "grad_norm": 1.388629674911499,
      "learning_rate": 1.9636201784299996e-05,
      "loss": 0.1136,
      "step": 39525
    },
    {
      "epoch": 0.5824656485176949,
      "grad_norm": 1.6771466732025146,
      "learning_rate": 1.963456541233301e-05,
      "loss": 0.1054,
      "step": 39550
    },
    {
      "epoch": 0.5828338316077819,
      "grad_norm": 1.0926169157028198,
      "learning_rate": 1.9632929040366025e-05,
      "loss": 0.1115,
      "step": 39575
    },
    {
      "epoch": 0.5832020146978689,
      "grad_norm": 1.7161418199539185,
      "learning_rate": 1.963129266839904e-05,
      "loss": 0.1063,
      "step": 39600
    },
    {
      "epoch": 0.583570197787956,
      "grad_norm": 1.692179560661316,
      "learning_rate": 1.9629656296432057e-05,
      "loss": 0.114,
      "step": 39625
    },
    {
      "epoch": 0.5839383808780431,
      "grad_norm": 1.1478818655014038,
      "learning_rate": 1.9628019924465072e-05,
      "loss": 0.1219,
      "step": 39650
    },
    {
      "epoch": 0.5843065639681301,
      "grad_norm": 1.1141635179519653,
      "learning_rate": 1.9626383552498086e-05,
      "loss": 0.1134,
      "step": 39675
    },
    {
      "epoch": 0.5846747470582171,
      "grad_norm": 1.1304652690887451,
      "learning_rate": 1.9624747180531104e-05,
      "loss": 0.1189,
      "step": 39700
    },
    {
      "epoch": 0.5850429301483041,
      "grad_norm": 1.385841965675354,
      "learning_rate": 1.962311080856412e-05,
      "loss": 0.1131,
      "step": 39725
    },
    {
      "epoch": 0.5854111132383912,
      "grad_norm": 1.639119267463684,
      "learning_rate": 1.9621474436597133e-05,
      "loss": 0.1084,
      "step": 39750
    },
    {
      "epoch": 0.5857792963284782,
      "grad_norm": 1.5052881240844727,
      "learning_rate": 1.9619838064630147e-05,
      "loss": 0.1125,
      "step": 39775
    },
    {
      "epoch": 0.5861474794185653,
      "grad_norm": 1.258682131767273,
      "learning_rate": 1.9618201692663165e-05,
      "loss": 0.1214,
      "step": 39800
    },
    {
      "epoch": 0.5865156625086523,
      "grad_norm": 1.3739937543869019,
      "learning_rate": 1.961656532069618e-05,
      "loss": 0.1092,
      "step": 39825
    },
    {
      "epoch": 0.5868838455987393,
      "grad_norm": 1.277521014213562,
      "learning_rate": 1.9614928948729194e-05,
      "loss": 0.11,
      "step": 39850
    },
    {
      "epoch": 0.5872520286888264,
      "grad_norm": 2.2052865028381348,
      "learning_rate": 1.9613292576762212e-05,
      "loss": 0.1056,
      "step": 39875
    },
    {
      "epoch": 0.5876202117789134,
      "grad_norm": 1.716227412223816,
      "learning_rate": 1.9611656204795226e-05,
      "loss": 0.1065,
      "step": 39900
    },
    {
      "epoch": 0.5879883948690005,
      "grad_norm": 1.6410953998565674,
      "learning_rate": 1.961001983282824e-05,
      "loss": 0.1105,
      "step": 39925
    },
    {
      "epoch": 0.5883565779590875,
      "grad_norm": 1.2135493755340576,
      "learning_rate": 1.960838346086126e-05,
      "loss": 0.1213,
      "step": 39950
    },
    {
      "epoch": 0.5887247610491745,
      "grad_norm": 1.3212488889694214,
      "learning_rate": 1.960674708889427e-05,
      "loss": 0.1124,
      "step": 39975
    },
    {
      "epoch": 0.5890929441392616,
      "grad_norm": 1.5297878980636597,
      "learning_rate": 1.9605110716927288e-05,
      "loss": 0.1174,
      "step": 40000
    },
    {
      "epoch": 0.5894611272293486,
      "grad_norm": 1.4061139822006226,
      "learning_rate": 1.9603474344960302e-05,
      "loss": 0.1116,
      "step": 40025
    },
    {
      "epoch": 0.5898293103194356,
      "grad_norm": 1.1281315088272095,
      "learning_rate": 1.960183797299332e-05,
      "loss": 0.1169,
      "step": 40050
    },
    {
      "epoch": 0.5901974934095227,
      "grad_norm": 0.9868475794792175,
      "learning_rate": 1.9600201601026334e-05,
      "loss": 0.1128,
      "step": 40075
    },
    {
      "epoch": 0.5905656764996097,
      "grad_norm": 1.150622010231018,
      "learning_rate": 1.959856522905935e-05,
      "loss": 0.1105,
      "step": 40100
    },
    {
      "epoch": 0.5909338595896968,
      "grad_norm": 1.3219376802444458,
      "learning_rate": 1.9596928857092367e-05,
      "loss": 0.1072,
      "step": 40125
    },
    {
      "epoch": 0.5913020426797838,
      "grad_norm": 1.5164532661437988,
      "learning_rate": 1.959529248512538e-05,
      "loss": 0.1051,
      "step": 40150
    },
    {
      "epoch": 0.5916702257698708,
      "grad_norm": 1.4182822704315186,
      "learning_rate": 1.9593656113158396e-05,
      "loss": 0.1102,
      "step": 40175
    },
    {
      "epoch": 0.5920384088599578,
      "grad_norm": 1.4331551790237427,
      "learning_rate": 1.959201974119141e-05,
      "loss": 0.1224,
      "step": 40200
    },
    {
      "epoch": 0.592406591950045,
      "grad_norm": 1.3870913982391357,
      "learning_rate": 1.9590383369224425e-05,
      "loss": 0.1127,
      "step": 40225
    },
    {
      "epoch": 0.592774775040132,
      "grad_norm": 1.5275180339813232,
      "learning_rate": 1.9588746997257442e-05,
      "loss": 0.1148,
      "step": 40250
    },
    {
      "epoch": 0.593142958130219,
      "grad_norm": 1.587823510169983,
      "learning_rate": 1.9587110625290457e-05,
      "loss": 0.1141,
      "step": 40275
    },
    {
      "epoch": 0.593511141220306,
      "grad_norm": 0.9925780296325684,
      "learning_rate": 1.9585474253323475e-05,
      "loss": 0.118,
      "step": 40300
    },
    {
      "epoch": 0.593879324310393,
      "grad_norm": 1.0631872415542603,
      "learning_rate": 1.958383788135649e-05,
      "loss": 0.1128,
      "step": 40325
    },
    {
      "epoch": 0.5942475074004802,
      "grad_norm": 0.9391284584999084,
      "learning_rate": 1.9582201509389504e-05,
      "loss": 0.1134,
      "step": 40350
    },
    {
      "epoch": 0.5946156904905672,
      "grad_norm": 1.4708532094955444,
      "learning_rate": 1.958056513742252e-05,
      "loss": 0.1144,
      "step": 40375
    },
    {
      "epoch": 0.5949838735806542,
      "grad_norm": 1.280452847480774,
      "learning_rate": 1.9578928765455533e-05,
      "loss": 0.1134,
      "step": 40400
    },
    {
      "epoch": 0.5953520566707412,
      "grad_norm": 1.3945668935775757,
      "learning_rate": 1.957729239348855e-05,
      "loss": 0.1072,
      "step": 40425
    },
    {
      "epoch": 0.5957202397608282,
      "grad_norm": 1.4512226581573486,
      "learning_rate": 1.9575656021521565e-05,
      "loss": 0.1103,
      "step": 40450
    },
    {
      "epoch": 0.5960884228509153,
      "grad_norm": 1.2649255990982056,
      "learning_rate": 1.957401964955458e-05,
      "loss": 0.1174,
      "step": 40475
    },
    {
      "epoch": 0.5964566059410024,
      "grad_norm": 1.0409470796585083,
      "learning_rate": 1.9572383277587597e-05,
      "loss": 0.1145,
      "step": 40500
    },
    {
      "epoch": 0.5968247890310894,
      "grad_norm": 1.2053929567337036,
      "learning_rate": 1.957074690562061e-05,
      "loss": 0.1086,
      "step": 40525
    },
    {
      "epoch": 0.5971929721211764,
      "grad_norm": 1.389586091041565,
      "learning_rate": 1.956911053365363e-05,
      "loss": 0.1137,
      "step": 40550
    },
    {
      "epoch": 0.5975611552112634,
      "grad_norm": 1.229688048362732,
      "learning_rate": 1.9567474161686644e-05,
      "loss": 0.1157,
      "step": 40575
    },
    {
      "epoch": 0.5979293383013505,
      "grad_norm": 1.4029161930084229,
      "learning_rate": 1.956583778971966e-05,
      "loss": 0.1122,
      "step": 40600
    },
    {
      "epoch": 0.5982975213914375,
      "grad_norm": 1.3531118631362915,
      "learning_rate": 1.9564201417752673e-05,
      "loss": 0.1301,
      "step": 40625
    },
    {
      "epoch": 0.5986657044815246,
      "grad_norm": 1.492425560951233,
      "learning_rate": 1.9562565045785687e-05,
      "loss": 0.1067,
      "step": 40650
    },
    {
      "epoch": 0.5990338875716116,
      "grad_norm": 1.325297236442566,
      "learning_rate": 1.9560928673818705e-05,
      "loss": 0.1117,
      "step": 40675
    },
    {
      "epoch": 0.5994020706616987,
      "grad_norm": 1.4290852546691895,
      "learning_rate": 1.955929230185172e-05,
      "loss": 0.1132,
      "step": 40700
    },
    {
      "epoch": 0.5997702537517857,
      "grad_norm": 1.2368844747543335,
      "learning_rate": 1.9557655929884734e-05,
      "loss": 0.1104,
      "step": 40725
    },
    {
      "epoch": 0.6001384368418727,
      "grad_norm": 1.4443602561950684,
      "learning_rate": 1.9556019557917752e-05,
      "loss": 0.1171,
      "step": 40750
    },
    {
      "epoch": 0.6005066199319598,
      "grad_norm": 1.6459391117095947,
      "learning_rate": 1.9554383185950766e-05,
      "loss": 0.1158,
      "step": 40775
    },
    {
      "epoch": 0.6008748030220468,
      "grad_norm": 1.3893733024597168,
      "learning_rate": 1.9552746813983784e-05,
      "loss": 0.1063,
      "step": 40800
    },
    {
      "epoch": 0.6012429861121339,
      "grad_norm": 1.834304690361023,
      "learning_rate": 1.9551110442016795e-05,
      "loss": 0.1095,
      "step": 40825
    },
    {
      "epoch": 0.6016111692022209,
      "grad_norm": 1.7188986539840698,
      "learning_rate": 1.9549474070049813e-05,
      "loss": 0.1129,
      "step": 40850
    },
    {
      "epoch": 0.6019793522923079,
      "grad_norm": 1.3190722465515137,
      "learning_rate": 1.9547837698082828e-05,
      "loss": 0.1164,
      "step": 40875
    },
    {
      "epoch": 0.6023475353823949,
      "grad_norm": 1.718002200126648,
      "learning_rate": 1.9546201326115842e-05,
      "loss": 0.1153,
      "step": 40900
    },
    {
      "epoch": 0.602715718472482,
      "grad_norm": 1.3596863746643066,
      "learning_rate": 1.954456495414886e-05,
      "loss": 0.109,
      "step": 40925
    },
    {
      "epoch": 0.6030839015625691,
      "grad_norm": 1.914109230041504,
      "learning_rate": 1.9542928582181874e-05,
      "loss": 0.1141,
      "step": 40950
    },
    {
      "epoch": 0.6034520846526561,
      "grad_norm": 1.6325136423110962,
      "learning_rate": 1.9541292210214892e-05,
      "loss": 0.1193,
      "step": 40975
    },
    {
      "epoch": 0.6038202677427431,
      "grad_norm": 1.3778468370437622,
      "learning_rate": 1.9539655838247907e-05,
      "loss": 0.1121,
      "step": 41000
    },
    {
      "epoch": 0.6041884508328301,
      "grad_norm": 1.5594487190246582,
      "learning_rate": 1.953801946628092e-05,
      "loss": 0.1136,
      "step": 41025
    },
    {
      "epoch": 0.6045566339229171,
      "grad_norm": 1.9369747638702393,
      "learning_rate": 1.9536383094313936e-05,
      "loss": 0.11,
      "step": 41050
    },
    {
      "epoch": 0.6049248170130043,
      "grad_norm": 1.337169885635376,
      "learning_rate": 1.953474672234695e-05,
      "loss": 0.1077,
      "step": 41075
    },
    {
      "epoch": 0.6052930001030913,
      "grad_norm": 1.4679465293884277,
      "learning_rate": 1.9533110350379968e-05,
      "loss": 0.118,
      "step": 41100
    },
    {
      "epoch": 0.6056611831931783,
      "grad_norm": 1.0459349155426025,
      "learning_rate": 1.9531473978412982e-05,
      "loss": 0.1052,
      "step": 41125
    },
    {
      "epoch": 0.6060293662832653,
      "grad_norm": 1.3841328620910645,
      "learning_rate": 1.9529837606445997e-05,
      "loss": 0.1085,
      "step": 41150
    },
    {
      "epoch": 0.6063975493733523,
      "grad_norm": 1.6530568599700928,
      "learning_rate": 1.9528201234479015e-05,
      "loss": 0.1141,
      "step": 41175
    },
    {
      "epoch": 0.6067657324634395,
      "grad_norm": 1.4696321487426758,
      "learning_rate": 1.952656486251203e-05,
      "loss": 0.1042,
      "step": 41200
    },
    {
      "epoch": 0.6071339155535265,
      "grad_norm": 1.3700885772705078,
      "learning_rate": 1.9524928490545047e-05,
      "loss": 0.1157,
      "step": 41225
    },
    {
      "epoch": 0.6075020986436135,
      "grad_norm": 1.6159257888793945,
      "learning_rate": 1.9523292118578058e-05,
      "loss": 0.1095,
      "step": 41250
    },
    {
      "epoch": 0.6078702817337005,
      "grad_norm": 1.4576144218444824,
      "learning_rate": 1.9521655746611076e-05,
      "loss": 0.1094,
      "step": 41275
    },
    {
      "epoch": 0.6082384648237876,
      "grad_norm": 1.3515383005142212,
      "learning_rate": 1.952001937464409e-05,
      "loss": 0.1105,
      "step": 41300
    },
    {
      "epoch": 0.6086066479138746,
      "grad_norm": 2.048415184020996,
      "learning_rate": 1.9518383002677105e-05,
      "loss": 0.1126,
      "step": 41325
    },
    {
      "epoch": 0.6089748310039617,
      "grad_norm": 1.4868552684783936,
      "learning_rate": 1.9516746630710123e-05,
      "loss": 0.1131,
      "step": 41350
    },
    {
      "epoch": 0.6093430140940487,
      "grad_norm": 1.4398176670074463,
      "learning_rate": 1.9515110258743137e-05,
      "loss": 0.1109,
      "step": 41375
    },
    {
      "epoch": 0.6097111971841357,
      "grad_norm": 1.5747963190078735,
      "learning_rate": 1.951347388677615e-05,
      "loss": 0.1237,
      "step": 41400
    },
    {
      "epoch": 0.6100793802742228,
      "grad_norm": 1.4340903759002686,
      "learning_rate": 1.951183751480917e-05,
      "loss": 0.113,
      "step": 41425
    },
    {
      "epoch": 0.6104475633643098,
      "grad_norm": 1.3397008180618286,
      "learning_rate": 1.9510201142842184e-05,
      "loss": 0.1166,
      "step": 41450
    },
    {
      "epoch": 0.6108157464543968,
      "grad_norm": 1.5803062915802002,
      "learning_rate": 1.9508564770875198e-05,
      "loss": 0.1028,
      "step": 41475
    },
    {
      "epoch": 0.6111839295444839,
      "grad_norm": 1.4686222076416016,
      "learning_rate": 1.9506928398908213e-05,
      "loss": 0.1124,
      "step": 41500
    },
    {
      "epoch": 0.6115521126345709,
      "grad_norm": 1.6495213508605957,
      "learning_rate": 1.950529202694123e-05,
      "loss": 0.1091,
      "step": 41525
    },
    {
      "epoch": 0.611920295724658,
      "grad_norm": 1.6780121326446533,
      "learning_rate": 1.9503721109852924e-05,
      "loss": 0.1079,
      "step": 41550
    },
    {
      "epoch": 0.612288478814745,
      "grad_norm": 1.4636390209197998,
      "learning_rate": 1.950208473788594e-05,
      "loss": 0.1078,
      "step": 41575
    },
    {
      "epoch": 0.612656661904832,
      "grad_norm": 1.2767215967178345,
      "learning_rate": 1.9500448365918957e-05,
      "loss": 0.1039,
      "step": 41600
    },
    {
      "epoch": 0.613024844994919,
      "grad_norm": 1.3839603662490845,
      "learning_rate": 1.949881199395197e-05,
      "loss": 0.1116,
      "step": 41625
    },
    {
      "epoch": 0.6133930280850061,
      "grad_norm": 1.48479163646698,
      "learning_rate": 1.9497175621984986e-05,
      "loss": 0.1145,
      "step": 41650
    },
    {
      "epoch": 0.6137612111750932,
      "grad_norm": 1.3524249792099,
      "learning_rate": 1.9495539250018e-05,
      "loss": 0.1125,
      "step": 41675
    },
    {
      "epoch": 0.6141293942651802,
      "grad_norm": 1.472377061843872,
      "learning_rate": 1.9493902878051018e-05,
      "loss": 0.1158,
      "step": 41700
    },
    {
      "epoch": 0.6144975773552672,
      "grad_norm": 1.097522258758545,
      "learning_rate": 1.9492266506084032e-05,
      "loss": 0.1025,
      "step": 41725
    },
    {
      "epoch": 0.6148657604453542,
      "grad_norm": 1.3611419200897217,
      "learning_rate": 1.9490630134117047e-05,
      "loss": 0.1135,
      "step": 41750
    },
    {
      "epoch": 0.6152339435354413,
      "grad_norm": 1.1177406311035156,
      "learning_rate": 1.9488993762150065e-05,
      "loss": 0.1124,
      "step": 41775
    },
    {
      "epoch": 0.6156021266255284,
      "grad_norm": 1.864156723022461,
      "learning_rate": 1.948735739018308e-05,
      "loss": 0.1108,
      "step": 41800
    },
    {
      "epoch": 0.6159703097156154,
      "grad_norm": 1.9078421592712402,
      "learning_rate": 1.9485721018216094e-05,
      "loss": 0.1115,
      "step": 41825
    },
    {
      "epoch": 0.6163384928057024,
      "grad_norm": 1.9030513763427734,
      "learning_rate": 1.948408464624911e-05,
      "loss": 0.1153,
      "step": 41850
    },
    {
      "epoch": 0.6167066758957894,
      "grad_norm": 1.1852060556411743,
      "learning_rate": 1.9482448274282123e-05,
      "loss": 0.1108,
      "step": 41875
    },
    {
      "epoch": 0.6170748589858766,
      "grad_norm": 1.758246660232544,
      "learning_rate": 1.948081190231514e-05,
      "loss": 0.1056,
      "step": 41900
    },
    {
      "epoch": 0.6174430420759636,
      "grad_norm": 1.5357226133346558,
      "learning_rate": 1.9479175530348155e-05,
      "loss": 0.1132,
      "step": 41925
    },
    {
      "epoch": 0.6178112251660506,
      "grad_norm": 1.4268290996551514,
      "learning_rate": 1.9477539158381173e-05,
      "loss": 0.116,
      "step": 41950
    },
    {
      "epoch": 0.6181794082561376,
      "grad_norm": 1.5423840284347534,
      "learning_rate": 1.9475902786414187e-05,
      "loss": 0.1027,
      "step": 41975
    },
    {
      "epoch": 0.6185475913462246,
      "grad_norm": 1.2127351760864258,
      "learning_rate": 1.94742664144472e-05,
      "loss": 0.1033,
      "step": 42000
    },
    {
      "epoch": 0.6189157744363117,
      "grad_norm": 1.6671496629714966,
      "learning_rate": 1.947263004248022e-05,
      "loss": 0.1077,
      "step": 42025
    },
    {
      "epoch": 0.6192839575263988,
      "grad_norm": 1.988416314125061,
      "learning_rate": 1.9470993670513234e-05,
      "loss": 0.1102,
      "step": 42050
    },
    {
      "epoch": 0.6196521406164858,
      "grad_norm": 1.5522608757019043,
      "learning_rate": 1.946935729854625e-05,
      "loss": 0.1188,
      "step": 42075
    },
    {
      "epoch": 0.6200203237065728,
      "grad_norm": 1.4972020387649536,
      "learning_rate": 1.9467720926579263e-05,
      "loss": 0.1041,
      "step": 42100
    },
    {
      "epoch": 0.6203885067966598,
      "grad_norm": 1.4773261547088623,
      "learning_rate": 1.9466084554612277e-05,
      "loss": 0.1054,
      "step": 42125
    },
    {
      "epoch": 0.6207566898867469,
      "grad_norm": 1.4228787422180176,
      "learning_rate": 1.9464448182645295e-05,
      "loss": 0.1071,
      "step": 42150
    },
    {
      "epoch": 0.6211248729768339,
      "grad_norm": 2.1490607261657715,
      "learning_rate": 1.946281181067831e-05,
      "loss": 0.1134,
      "step": 42175
    },
    {
      "epoch": 0.621493056066921,
      "grad_norm": 1.4083261489868164,
      "learning_rate": 1.9461175438711327e-05,
      "loss": 0.1183,
      "step": 42200
    },
    {
      "epoch": 0.621861239157008,
      "grad_norm": 1.4812012910842896,
      "learning_rate": 1.9459539066744342e-05,
      "loss": 0.1102,
      "step": 42225
    },
    {
      "epoch": 0.622229422247095,
      "grad_norm": 1.5146194696426392,
      "learning_rate": 1.9457902694777356e-05,
      "loss": 0.1146,
      "step": 42250
    },
    {
      "epoch": 0.6225976053371821,
      "grad_norm": 1.5088244676589966,
      "learning_rate": 1.9456266322810374e-05,
      "loss": 0.1137,
      "step": 42275
    },
    {
      "epoch": 0.6229657884272691,
      "grad_norm": 1.7292484045028687,
      "learning_rate": 1.9454629950843385e-05,
      "loss": 0.1018,
      "step": 42300
    },
    {
      "epoch": 0.6233339715173561,
      "grad_norm": 1.335016131401062,
      "learning_rate": 1.9452993578876403e-05,
      "loss": 0.11,
      "step": 42325
    },
    {
      "epoch": 0.6237021546074432,
      "grad_norm": 1.0359456539154053,
      "learning_rate": 1.9451357206909418e-05,
      "loss": 0.1155,
      "step": 42350
    },
    {
      "epoch": 0.6240703376975302,
      "grad_norm": 1.4816323518753052,
      "learning_rate": 1.9449720834942435e-05,
      "loss": 0.1079,
      "step": 42375
    },
    {
      "epoch": 0.6244385207876173,
      "grad_norm": 1.6909608840942383,
      "learning_rate": 1.944808446297545e-05,
      "loss": 0.1039,
      "step": 42400
    },
    {
      "epoch": 0.6248067038777043,
      "grad_norm": 1.2284351587295532,
      "learning_rate": 1.9446448091008464e-05,
      "loss": 0.1059,
      "step": 42425
    },
    {
      "epoch": 0.6251748869677913,
      "grad_norm": 1.3990721702575684,
      "learning_rate": 1.9444811719041482e-05,
      "loss": 0.109,
      "step": 42450
    },
    {
      "epoch": 0.6255430700578783,
      "grad_norm": 2.34006929397583,
      "learning_rate": 1.9443175347074497e-05,
      "loss": 0.106,
      "step": 42475
    },
    {
      "epoch": 0.6259112531479655,
      "grad_norm": 1.3233106136322021,
      "learning_rate": 1.944153897510751e-05,
      "loss": 0.1112,
      "step": 42500
    },
    {
      "epoch": 0.6262794362380525,
      "grad_norm": 1.436488151550293,
      "learning_rate": 1.9439902603140526e-05,
      "loss": 0.1114,
      "step": 42525
    },
    {
      "epoch": 0.6266476193281395,
      "grad_norm": 1.1722458600997925,
      "learning_rate": 1.943826623117354e-05,
      "loss": 0.1062,
      "step": 42550
    },
    {
      "epoch": 0.6270158024182265,
      "grad_norm": 1.8988996744155884,
      "learning_rate": 1.9436629859206558e-05,
      "loss": 0.1202,
      "step": 42575
    },
    {
      "epoch": 0.6273839855083135,
      "grad_norm": 1.3709423542022705,
      "learning_rate": 1.9434993487239572e-05,
      "loss": 0.1151,
      "step": 42600
    },
    {
      "epoch": 0.6277521685984007,
      "grad_norm": 1.8680797815322876,
      "learning_rate": 1.943335711527259e-05,
      "loss": 0.1145,
      "step": 42625
    },
    {
      "epoch": 0.6281203516884877,
      "grad_norm": 1.2792071104049683,
      "learning_rate": 1.9431720743305605e-05,
      "loss": 0.1071,
      "step": 42650
    },
    {
      "epoch": 0.6284885347785747,
      "grad_norm": 1.495862364768982,
      "learning_rate": 1.943008437133862e-05,
      "loss": 0.1039,
      "step": 42675
    },
    {
      "epoch": 0.6288567178686617,
      "grad_norm": 1.2855013608932495,
      "learning_rate": 1.9428447999371637e-05,
      "loss": 0.1123,
      "step": 42700
    },
    {
      "epoch": 0.6292249009587487,
      "grad_norm": 1.80340576171875,
      "learning_rate": 1.9426811627404648e-05,
      "loss": 0.1017,
      "step": 42725
    },
    {
      "epoch": 0.6295930840488358,
      "grad_norm": 1.610538125038147,
      "learning_rate": 1.9425175255437666e-05,
      "loss": 0.1106,
      "step": 42750
    },
    {
      "epoch": 0.6299612671389229,
      "grad_norm": 1.4962716102600098,
      "learning_rate": 1.942353888347068e-05,
      "loss": 0.1064,
      "step": 42775
    },
    {
      "epoch": 0.6303294502290099,
      "grad_norm": 1.2373082637786865,
      "learning_rate": 1.9421902511503695e-05,
      "loss": 0.1069,
      "step": 42800
    },
    {
      "epoch": 0.6306976333190969,
      "grad_norm": 1.5830014944076538,
      "learning_rate": 1.9420266139536713e-05,
      "loss": 0.1079,
      "step": 42825
    },
    {
      "epoch": 0.6310658164091839,
      "grad_norm": 1.6252222061157227,
      "learning_rate": 1.9418629767569727e-05,
      "loss": 0.1062,
      "step": 42850
    },
    {
      "epoch": 0.631433999499271,
      "grad_norm": 1.5839202404022217,
      "learning_rate": 1.9416993395602745e-05,
      "loss": 0.1035,
      "step": 42875
    },
    {
      "epoch": 0.631802182589358,
      "grad_norm": 1.857698917388916,
      "learning_rate": 1.941535702363576e-05,
      "loss": 0.1091,
      "step": 42900
    },
    {
      "epoch": 0.6321703656794451,
      "grad_norm": 1.4146044254302979,
      "learning_rate": 1.9413720651668774e-05,
      "loss": 0.1031,
      "step": 42925
    },
    {
      "epoch": 0.6325385487695321,
      "grad_norm": 1.1775039434432983,
      "learning_rate": 1.9412084279701788e-05,
      "loss": 0.1153,
      "step": 42950
    },
    {
      "epoch": 0.6329067318596191,
      "grad_norm": 1.484903335571289,
      "learning_rate": 1.9410447907734803e-05,
      "loss": 0.1088,
      "step": 42975
    },
    {
      "epoch": 0.6332749149497062,
      "grad_norm": 1.5222270488739014,
      "learning_rate": 1.940881153576782e-05,
      "loss": 0.1147,
      "step": 43000
    },
    {
      "epoch": 0.6336430980397932,
      "grad_norm": 1.775303602218628,
      "learning_rate": 1.9407175163800835e-05,
      "loss": 0.0978,
      "step": 43025
    },
    {
      "epoch": 0.6340112811298803,
      "grad_norm": 1.9845086336135864,
      "learning_rate": 1.940553879183385e-05,
      "loss": 0.1163,
      "step": 43050
    },
    {
      "epoch": 0.6343794642199673,
      "grad_norm": 1.6547818183898926,
      "learning_rate": 1.9403902419866867e-05,
      "loss": 0.1053,
      "step": 43075
    },
    {
      "epoch": 0.6347476473100544,
      "grad_norm": 1.240424633026123,
      "learning_rate": 1.9402266047899882e-05,
      "loss": 0.0989,
      "step": 43100
    },
    {
      "epoch": 0.6351158304001414,
      "grad_norm": 1.1651711463928223,
      "learning_rate": 1.94006296759329e-05,
      "loss": 0.1086,
      "step": 43125
    },
    {
      "epoch": 0.6354840134902284,
      "grad_norm": 1.3533746004104614,
      "learning_rate": 1.939899330396591e-05,
      "loss": 0.1,
      "step": 43150
    },
    {
      "epoch": 0.6358521965803154,
      "grad_norm": 1.5015009641647339,
      "learning_rate": 1.939735693199893e-05,
      "loss": 0.1121,
      "step": 43175
    },
    {
      "epoch": 0.6362203796704025,
      "grad_norm": 1.7772409915924072,
      "learning_rate": 1.9395720560031943e-05,
      "loss": 0.1094,
      "step": 43200
    },
    {
      "epoch": 0.6365885627604896,
      "grad_norm": 1.4367293119430542,
      "learning_rate": 1.9394084188064957e-05,
      "loss": 0.1031,
      "step": 43225
    },
    {
      "epoch": 0.6369567458505766,
      "grad_norm": 1.2458570003509521,
      "learning_rate": 1.9392447816097975e-05,
      "loss": 0.1058,
      "step": 43250
    },
    {
      "epoch": 0.6373249289406636,
      "grad_norm": 1.1537954807281494,
      "learning_rate": 1.939081144413099e-05,
      "loss": 0.1131,
      "step": 43275
    },
    {
      "epoch": 0.6376931120307506,
      "grad_norm": 1.1713804006576538,
      "learning_rate": 1.9389175072164008e-05,
      "loss": 0.1085,
      "step": 43300
    },
    {
      "epoch": 0.6380612951208376,
      "grad_norm": 1.6194653511047363,
      "learning_rate": 1.9387538700197022e-05,
      "loss": 0.1141,
      "step": 43325
    },
    {
      "epoch": 0.6384294782109248,
      "grad_norm": 2.1495585441589355,
      "learning_rate": 1.9385902328230037e-05,
      "loss": 0.1127,
      "step": 43350
    },
    {
      "epoch": 0.6387976613010118,
      "grad_norm": 1.8170406818389893,
      "learning_rate": 1.938426595626305e-05,
      "loss": 0.1092,
      "step": 43375
    },
    {
      "epoch": 0.6391658443910988,
      "grad_norm": 1.5318728685379028,
      "learning_rate": 1.9382629584296065e-05,
      "loss": 0.1059,
      "step": 43400
    },
    {
      "epoch": 0.6395340274811858,
      "grad_norm": 1.771115779876709,
      "learning_rate": 1.9380993212329083e-05,
      "loss": 0.109,
      "step": 43425
    },
    {
      "epoch": 0.6399022105712728,
      "grad_norm": 1.321984887123108,
      "learning_rate": 1.9379356840362098e-05,
      "loss": 0.1135,
      "step": 43450
    },
    {
      "epoch": 0.64027039366136,
      "grad_norm": 1.3801411390304565,
      "learning_rate": 1.9377720468395112e-05,
      "loss": 0.1083,
      "step": 43475
    },
    {
      "epoch": 0.640638576751447,
      "grad_norm": 1.4694974422454834,
      "learning_rate": 1.937608409642813e-05,
      "loss": 0.1127,
      "step": 43500
    },
    {
      "epoch": 0.641006759841534,
      "grad_norm": 1.6573680639266968,
      "learning_rate": 1.9374447724461144e-05,
      "loss": 0.1092,
      "step": 43525
    },
    {
      "epoch": 0.641374942931621,
      "grad_norm": 1.3211015462875366,
      "learning_rate": 1.9372811352494162e-05,
      "loss": 0.1069,
      "step": 43550
    },
    {
      "epoch": 0.641743126021708,
      "grad_norm": 1.2467752695083618,
      "learning_rate": 1.9371174980527173e-05,
      "loss": 0.1052,
      "step": 43575
    },
    {
      "epoch": 0.6421113091117951,
      "grad_norm": 1.2694215774536133,
      "learning_rate": 1.936953860856019e-05,
      "loss": 0.1136,
      "step": 43600
    },
    {
      "epoch": 0.6424794922018822,
      "grad_norm": 1.6949684619903564,
      "learning_rate": 1.9367902236593206e-05,
      "loss": 0.1119,
      "step": 43625
    },
    {
      "epoch": 0.6428476752919692,
      "grad_norm": 1.5692588090896606,
      "learning_rate": 1.936626586462622e-05,
      "loss": 0.1033,
      "step": 43650
    },
    {
      "epoch": 0.6432158583820562,
      "grad_norm": 1.5174171924591064,
      "learning_rate": 1.9364694947537917e-05,
      "loss": 0.1063,
      "step": 43675
    },
    {
      "epoch": 0.6435840414721433,
      "grad_norm": 1.9864505529403687,
      "learning_rate": 1.9363058575570932e-05,
      "loss": 0.111,
      "step": 43700
    },
    {
      "epoch": 0.6439522245622303,
      "grad_norm": 1.7174837589263916,
      "learning_rate": 1.9361422203603946e-05,
      "loss": 0.114,
      "step": 43725
    },
    {
      "epoch": 0.6443204076523174,
      "grad_norm": 1.2995622158050537,
      "learning_rate": 1.9359785831636964e-05,
      "loss": 0.1177,
      "step": 43750
    },
    {
      "epoch": 0.6446885907424044,
      "grad_norm": 1.2390779256820679,
      "learning_rate": 1.935814945966998e-05,
      "loss": 0.1078,
      "step": 43775
    },
    {
      "epoch": 0.6450567738324914,
      "grad_norm": 1.2460603713989258,
      "learning_rate": 1.9356513087702993e-05,
      "loss": 0.1135,
      "step": 43800
    },
    {
      "epoch": 0.6454249569225785,
      "grad_norm": 1.3848339319229126,
      "learning_rate": 1.9354876715736008e-05,
      "loss": 0.1114,
      "step": 43825
    },
    {
      "epoch": 0.6457931400126655,
      "grad_norm": 1.4496217966079712,
      "learning_rate": 1.9353240343769025e-05,
      "loss": 0.1001,
      "step": 43850
    },
    {
      "epoch": 0.6461613231027525,
      "grad_norm": 1.855474591255188,
      "learning_rate": 1.935160397180204e-05,
      "loss": 0.1069,
      "step": 43875
    },
    {
      "epoch": 0.6465295061928396,
      "grad_norm": 1.4969669580459595,
      "learning_rate": 1.9349967599835054e-05,
      "loss": 0.1193,
      "step": 43900
    },
    {
      "epoch": 0.6468976892829266,
      "grad_norm": 1.3219048976898193,
      "learning_rate": 1.9348331227868072e-05,
      "loss": 0.0964,
      "step": 43925
    },
    {
      "epoch": 0.6472658723730137,
      "grad_norm": 1.2522467374801636,
      "learning_rate": 1.9346694855901087e-05,
      "loss": 0.0986,
      "step": 43950
    },
    {
      "epoch": 0.6476340554631007,
      "grad_norm": 1.5771262645721436,
      "learning_rate": 1.93450584839341e-05,
      "loss": 0.1143,
      "step": 43975
    },
    {
      "epoch": 0.6480022385531877,
      "grad_norm": 1.211543321609497,
      "learning_rate": 1.9343422111967116e-05,
      "loss": 0.1013,
      "step": 44000
    },
    {
      "epoch": 0.6483704216432747,
      "grad_norm": 1.8445452451705933,
      "learning_rate": 1.9341785740000133e-05,
      "loss": 0.1195,
      "step": 44025
    },
    {
      "epoch": 0.6487386047333618,
      "grad_norm": 1.344293475151062,
      "learning_rate": 1.9340149368033148e-05,
      "loss": 0.1063,
      "step": 44050
    },
    {
      "epoch": 0.6491067878234489,
      "grad_norm": 1.3796851634979248,
      "learning_rate": 1.9338512996066162e-05,
      "loss": 0.106,
      "step": 44075
    },
    {
      "epoch": 0.6494749709135359,
      "grad_norm": 1.2502880096435547,
      "learning_rate": 1.933687662409918e-05,
      "loss": 0.1091,
      "step": 44100
    },
    {
      "epoch": 0.6498431540036229,
      "grad_norm": 1.391850233078003,
      "learning_rate": 1.9335240252132195e-05,
      "loss": 0.1119,
      "step": 44125
    },
    {
      "epoch": 0.6502113370937099,
      "grad_norm": 1.270440936088562,
      "learning_rate": 1.933360388016521e-05,
      "loss": 0.1073,
      "step": 44150
    },
    {
      "epoch": 0.6505795201837969,
      "grad_norm": 1.4724406003952026,
      "learning_rate": 1.9331967508198227e-05,
      "loss": 0.1171,
      "step": 44175
    },
    {
      "epoch": 0.6509477032738841,
      "grad_norm": 0.945414662361145,
      "learning_rate": 1.9330331136231238e-05,
      "loss": 0.103,
      "step": 44200
    },
    {
      "epoch": 0.6513158863639711,
      "grad_norm": 1.268459439277649,
      "learning_rate": 1.9328694764264256e-05,
      "loss": 0.1011,
      "step": 44225
    },
    {
      "epoch": 0.6516840694540581,
      "grad_norm": 1.4114058017730713,
      "learning_rate": 1.932705839229727e-05,
      "loss": 0.1106,
      "step": 44250
    },
    {
      "epoch": 0.6520522525441451,
      "grad_norm": 1.3198155164718628,
      "learning_rate": 1.9325422020330288e-05,
      "loss": 0.0992,
      "step": 44275
    },
    {
      "epoch": 0.6524204356342322,
      "grad_norm": 1.4559319019317627,
      "learning_rate": 1.9323785648363303e-05,
      "loss": 0.1134,
      "step": 44300
    },
    {
      "epoch": 0.6527886187243193,
      "grad_norm": 1.640488624572754,
      "learning_rate": 1.9322149276396317e-05,
      "loss": 0.107,
      "step": 44325
    },
    {
      "epoch": 0.6531568018144063,
      "grad_norm": 1.2860337495803833,
      "learning_rate": 1.9320512904429335e-05,
      "loss": 0.1009,
      "step": 44350
    },
    {
      "epoch": 0.6535249849044933,
      "grad_norm": 1.3136343955993652,
      "learning_rate": 1.931887653246235e-05,
      "loss": 0.1106,
      "step": 44375
    },
    {
      "epoch": 0.6538931679945803,
      "grad_norm": 1.3006441593170166,
      "learning_rate": 1.9317240160495364e-05,
      "loss": 0.1101,
      "step": 44400
    },
    {
      "epoch": 0.6542613510846674,
      "grad_norm": 1.223717212677002,
      "learning_rate": 1.9315603788528378e-05,
      "loss": 0.1096,
      "step": 44425
    },
    {
      "epoch": 0.6546295341747544,
      "grad_norm": 2.2580654621124268,
      "learning_rate": 1.9313967416561393e-05,
      "loss": 0.1163,
      "step": 44450
    },
    {
      "epoch": 0.6549977172648415,
      "grad_norm": 1.3434696197509766,
      "learning_rate": 1.931233104459441e-05,
      "loss": 0.1031,
      "step": 44475
    },
    {
      "epoch": 0.6553659003549285,
      "grad_norm": 1.7967331409454346,
      "learning_rate": 1.9310694672627425e-05,
      "loss": 0.1071,
      "step": 44500
    },
    {
      "epoch": 0.6557340834450155,
      "grad_norm": 1.4282450675964355,
      "learning_rate": 1.9309058300660443e-05,
      "loss": 0.1068,
      "step": 44525
    },
    {
      "epoch": 0.6561022665351026,
      "grad_norm": 1.3770360946655273,
      "learning_rate": 1.9307421928693457e-05,
      "loss": 0.1076,
      "step": 44550
    },
    {
      "epoch": 0.6564704496251896,
      "grad_norm": 1.8066473007202148,
      "learning_rate": 1.9305785556726472e-05,
      "loss": 0.1057,
      "step": 44575
    },
    {
      "epoch": 0.6568386327152766,
      "grad_norm": 1.4721416234970093,
      "learning_rate": 1.930414918475949e-05,
      "loss": 0.1073,
      "step": 44600
    },
    {
      "epoch": 0.6572068158053637,
      "grad_norm": 1.9056334495544434,
      "learning_rate": 1.93025128127925e-05,
      "loss": 0.1033,
      "step": 44625
    },
    {
      "epoch": 0.6575749988954507,
      "grad_norm": 1.0975908041000366,
      "learning_rate": 1.930087644082552e-05,
      "loss": 0.1067,
      "step": 44650
    },
    {
      "epoch": 0.6579431819855378,
      "grad_norm": 2.040506362915039,
      "learning_rate": 1.9299240068858533e-05,
      "loss": 0.1071,
      "step": 44675
    },
    {
      "epoch": 0.6583113650756248,
      "grad_norm": 1.5115771293640137,
      "learning_rate": 1.929760369689155e-05,
      "loss": 0.1092,
      "step": 44700
    },
    {
      "epoch": 0.6586795481657118,
      "grad_norm": 1.4625195264816284,
      "learning_rate": 1.9295967324924565e-05,
      "loss": 0.1043,
      "step": 44725
    },
    {
      "epoch": 0.6590477312557989,
      "grad_norm": 1.707213282585144,
      "learning_rate": 1.929433095295758e-05,
      "loss": 0.1064,
      "step": 44750
    },
    {
      "epoch": 0.6594159143458859,
      "grad_norm": 1.4760093688964844,
      "learning_rate": 1.9292694580990598e-05,
      "loss": 0.1084,
      "step": 44775
    },
    {
      "epoch": 0.659784097435973,
      "grad_norm": 1.9389700889587402,
      "learning_rate": 1.9291058209023612e-05,
      "loss": 0.1023,
      "step": 44800
    },
    {
      "epoch": 0.66015228052606,
      "grad_norm": 1.11986243724823,
      "learning_rate": 1.9289421837056627e-05,
      "loss": 0.0964,
      "step": 44825
    },
    {
      "epoch": 0.660520463616147,
      "grad_norm": 1.4513438940048218,
      "learning_rate": 1.928778546508964e-05,
      "loss": 0.1132,
      "step": 44850
    },
    {
      "epoch": 0.660888646706234,
      "grad_norm": 1.8370686769485474,
      "learning_rate": 1.9286149093122655e-05,
      "loss": 0.1034,
      "step": 44875
    },
    {
      "epoch": 0.6612568297963212,
      "grad_norm": 1.4491779804229736,
      "learning_rate": 1.9284512721155673e-05,
      "loss": 0.1071,
      "step": 44900
    },
    {
      "epoch": 0.6616250128864082,
      "grad_norm": 1.2620149850845337,
      "learning_rate": 1.9282876349188688e-05,
      "loss": 0.1063,
      "step": 44925
    },
    {
      "epoch": 0.6619931959764952,
      "grad_norm": 1.2185320854187012,
      "learning_rate": 1.9281239977221706e-05,
      "loss": 0.1048,
      "step": 44950
    },
    {
      "epoch": 0.6623613790665822,
      "grad_norm": 1.1284700632095337,
      "learning_rate": 1.927960360525472e-05,
      "loss": 0.1087,
      "step": 44975
    },
    {
      "epoch": 0.6627295621566692,
      "grad_norm": 1.401711344718933,
      "learning_rate": 1.9277967233287735e-05,
      "loss": 0.1018,
      "step": 45000
    },
    {
      "epoch": 0.6630977452467564,
      "grad_norm": 1.722172737121582,
      "learning_rate": 1.9276330861320752e-05,
      "loss": 0.1013,
      "step": 45025
    },
    {
      "epoch": 0.6634659283368434,
      "grad_norm": 1.3857054710388184,
      "learning_rate": 1.9274694489353763e-05,
      "loss": 0.1008,
      "step": 45050
    },
    {
      "epoch": 0.6638341114269304,
      "grad_norm": 1.1007604598999023,
      "learning_rate": 1.927305811738678e-05,
      "loss": 0.103,
      "step": 45075
    },
    {
      "epoch": 0.6642022945170174,
      "grad_norm": 1.9387741088867188,
      "learning_rate": 1.9271421745419796e-05,
      "loss": 0.1052,
      "step": 45100
    },
    {
      "epoch": 0.6645704776071044,
      "grad_norm": 1.5912046432495117,
      "learning_rate": 1.926978537345281e-05,
      "loss": 0.1066,
      "step": 45125
    },
    {
      "epoch": 0.6649386606971915,
      "grad_norm": 1.3904691934585571,
      "learning_rate": 1.9268149001485828e-05,
      "loss": 0.1111,
      "step": 45150
    },
    {
      "epoch": 0.6653068437872786,
      "grad_norm": 1.140342354774475,
      "learning_rate": 1.9266512629518842e-05,
      "loss": 0.0987,
      "step": 45175
    },
    {
      "epoch": 0.6656750268773656,
      "grad_norm": 1.0333884954452515,
      "learning_rate": 1.926487625755186e-05,
      "loss": 0.1088,
      "step": 45200
    },
    {
      "epoch": 0.6660432099674526,
      "grad_norm": 1.5406895875930786,
      "learning_rate": 1.9263239885584875e-05,
      "loss": 0.1052,
      "step": 45225
    },
    {
      "epoch": 0.6664113930575396,
      "grad_norm": 1.225435495376587,
      "learning_rate": 1.926160351361789e-05,
      "loss": 0.0954,
      "step": 45250
    },
    {
      "epoch": 0.6667795761476267,
      "grad_norm": 1.4176855087280273,
      "learning_rate": 1.9259967141650904e-05,
      "loss": 0.0989,
      "step": 45275
    },
    {
      "epoch": 0.6671477592377137,
      "grad_norm": 1.8527737855911255,
      "learning_rate": 1.9258330769683918e-05,
      "loss": 0.1098,
      "step": 45300
    },
    {
      "epoch": 0.6675159423278008,
      "grad_norm": 1.5954591035842896,
      "learning_rate": 1.9256694397716936e-05,
      "loss": 0.1018,
      "step": 45325
    },
    {
      "epoch": 0.6678841254178878,
      "grad_norm": 1.2099926471710205,
      "learning_rate": 1.925505802574995e-05,
      "loss": 0.1029,
      "step": 45350
    },
    {
      "epoch": 0.6682523085079748,
      "grad_norm": 1.6647226810455322,
      "learning_rate": 1.9253421653782965e-05,
      "loss": 0.1101,
      "step": 45375
    },
    {
      "epoch": 0.6686204915980619,
      "grad_norm": 1.2445528507232666,
      "learning_rate": 1.9251785281815983e-05,
      "loss": 0.1045,
      "step": 45400
    },
    {
      "epoch": 0.6689886746881489,
      "grad_norm": 1.1713930368423462,
      "learning_rate": 1.9250148909848997e-05,
      "loss": 0.1051,
      "step": 45425
    },
    {
      "epoch": 0.669356857778236,
      "grad_norm": 1.7021634578704834,
      "learning_rate": 1.9248512537882015e-05,
      "loss": 0.0968,
      "step": 45450
    },
    {
      "epoch": 0.669725040868323,
      "grad_norm": 1.8190854787826538,
      "learning_rate": 1.9246876165915026e-05,
      "loss": 0.1001,
      "step": 45475
    },
    {
      "epoch": 0.6700932239584101,
      "grad_norm": 1.7377605438232422,
      "learning_rate": 1.9245239793948044e-05,
      "loss": 0.1045,
      "step": 45500
    },
    {
      "epoch": 0.6704614070484971,
      "grad_norm": 1.0050832033157349,
      "learning_rate": 1.924360342198106e-05,
      "loss": 0.1011,
      "step": 45525
    },
    {
      "epoch": 0.6708295901385841,
      "grad_norm": 1.3286410570144653,
      "learning_rate": 1.9241967050014073e-05,
      "loss": 0.116,
      "step": 45550
    },
    {
      "epoch": 0.6711977732286711,
      "grad_norm": 1.4806352853775024,
      "learning_rate": 1.924033067804709e-05,
      "loss": 0.1095,
      "step": 45575
    },
    {
      "epoch": 0.6715659563187582,
      "grad_norm": 1.5563225746154785,
      "learning_rate": 1.9238694306080105e-05,
      "loss": 0.1004,
      "step": 45600
    },
    {
      "epoch": 0.6719341394088453,
      "grad_norm": 1.7315696477890015,
      "learning_rate": 1.9237057934113123e-05,
      "loss": 0.0998,
      "step": 45625
    },
    {
      "epoch": 0.6723023224989323,
      "grad_norm": 1.3105772733688354,
      "learning_rate": 1.9235421562146137e-05,
      "loss": 0.0992,
      "step": 45650
    },
    {
      "epoch": 0.6726705055890193,
      "grad_norm": 1.2332693338394165,
      "learning_rate": 1.9233785190179152e-05,
      "loss": 0.1004,
      "step": 45675
    },
    {
      "epoch": 0.6730386886791063,
      "grad_norm": 1.423226237297058,
      "learning_rate": 1.9232148818212166e-05,
      "loss": 0.1114,
      "step": 45700
    },
    {
      "epoch": 0.6734068717691933,
      "grad_norm": 1.5766901969909668,
      "learning_rate": 1.923057790112386e-05,
      "loss": 0.1077,
      "step": 45725
    },
    {
      "epoch": 0.6737750548592805,
      "grad_norm": 1.4316579103469849,
      "learning_rate": 1.9228941529156878e-05,
      "loss": 0.1067,
      "step": 45750
    },
    {
      "epoch": 0.6741432379493675,
      "grad_norm": 1.2268834114074707,
      "learning_rate": 1.9227305157189893e-05,
      "loss": 0.1103,
      "step": 45775
    },
    {
      "epoch": 0.6745114210394545,
      "grad_norm": 1.6540658473968506,
      "learning_rate": 1.9225668785222907e-05,
      "loss": 0.1024,
      "step": 45800
    },
    {
      "epoch": 0.6748796041295415,
      "grad_norm": 1.6553481817245483,
      "learning_rate": 1.9224032413255925e-05,
      "loss": 0.1014,
      "step": 45825
    },
    {
      "epoch": 0.6752477872196285,
      "grad_norm": 1.6704182624816895,
      "learning_rate": 1.922239604128894e-05,
      "loss": 0.1028,
      "step": 45850
    },
    {
      "epoch": 0.6756159703097157,
      "grad_norm": 1.7540957927703857,
      "learning_rate": 1.9220759669321957e-05,
      "loss": 0.111,
      "step": 45875
    },
    {
      "epoch": 0.6759841533998027,
      "grad_norm": 1.5921074151992798,
      "learning_rate": 1.9219123297354968e-05,
      "loss": 0.1012,
      "step": 45900
    },
    {
      "epoch": 0.6763523364898897,
      "grad_norm": 1.7300814390182495,
      "learning_rate": 1.9217486925387986e-05,
      "loss": 0.102,
      "step": 45925
    },
    {
      "epoch": 0.6767205195799767,
      "grad_norm": 1.493948221206665,
      "learning_rate": 1.9215850553421e-05,
      "loss": 0.1036,
      "step": 45950
    },
    {
      "epoch": 0.6770887026700638,
      "grad_norm": 1.2351493835449219,
      "learning_rate": 1.9214214181454015e-05,
      "loss": 0.1038,
      "step": 45975
    },
    {
      "epoch": 0.6774568857601508,
      "grad_norm": 1.4516537189483643,
      "learning_rate": 1.9212577809487033e-05,
      "loss": 0.1038,
      "step": 46000
    },
    {
      "epoch": 0.6778250688502379,
      "grad_norm": 1.8888888359069824,
      "learning_rate": 1.9210941437520047e-05,
      "loss": 0.1044,
      "step": 46025
    },
    {
      "epoch": 0.6781932519403249,
      "grad_norm": 1.982967495918274,
      "learning_rate": 1.9209305065553062e-05,
      "loss": 0.0951,
      "step": 46050
    },
    {
      "epoch": 0.6785614350304119,
      "grad_norm": 1.325954794883728,
      "learning_rate": 1.920766869358608e-05,
      "loss": 0.1011,
      "step": 46075
    },
    {
      "epoch": 0.678929618120499,
      "grad_norm": 1.3796824216842651,
      "learning_rate": 1.9206032321619094e-05,
      "loss": 0.1051,
      "step": 46100
    },
    {
      "epoch": 0.679297801210586,
      "grad_norm": 1.883231520652771,
      "learning_rate": 1.920439594965211e-05,
      "loss": 0.1092,
      "step": 46125
    },
    {
      "epoch": 0.679665984300673,
      "grad_norm": 1.543330430984497,
      "learning_rate": 1.9202759577685123e-05,
      "loss": 0.1077,
      "step": 46150
    },
    {
      "epoch": 0.6800341673907601,
      "grad_norm": 1.438488245010376,
      "learning_rate": 1.920112320571814e-05,
      "loss": 0.1017,
      "step": 46175
    },
    {
      "epoch": 0.6804023504808471,
      "grad_norm": 1.5786123275756836,
      "learning_rate": 1.9199486833751155e-05,
      "loss": 0.1038,
      "step": 46200
    },
    {
      "epoch": 0.6807705335709342,
      "grad_norm": 1.271911382675171,
      "learning_rate": 1.919785046178417e-05,
      "loss": 0.1069,
      "step": 46225
    },
    {
      "epoch": 0.6811387166610212,
      "grad_norm": 1.5244625806808472,
      "learning_rate": 1.9196214089817188e-05,
      "loss": 0.0964,
      "step": 46250
    },
    {
      "epoch": 0.6815068997511082,
      "grad_norm": 1.725836157798767,
      "learning_rate": 1.9194577717850202e-05,
      "loss": 0.1022,
      "step": 46275
    },
    {
      "epoch": 0.6818750828411952,
      "grad_norm": 1.7017706632614136,
      "learning_rate": 1.9192941345883217e-05,
      "loss": 0.0946,
      "step": 46300
    },
    {
      "epoch": 0.6822432659312823,
      "grad_norm": 1.0990180969238281,
      "learning_rate": 1.919130497391623e-05,
      "loss": 0.0988,
      "step": 46325
    },
    {
      "epoch": 0.6826114490213694,
      "grad_norm": 1.1162875890731812,
      "learning_rate": 1.918966860194925e-05,
      "loss": 0.1001,
      "step": 46350
    },
    {
      "epoch": 0.6829796321114564,
      "grad_norm": 1.4849499464035034,
      "learning_rate": 1.9188032229982263e-05,
      "loss": 0.0977,
      "step": 46375
    },
    {
      "epoch": 0.6833478152015434,
      "grad_norm": 1.6886072158813477,
      "learning_rate": 1.9186395858015278e-05,
      "loss": 0.1129,
      "step": 46400
    },
    {
      "epoch": 0.6837159982916304,
      "grad_norm": 1.665539264678955,
      "learning_rate": 1.9184759486048296e-05,
      "loss": 0.1016,
      "step": 46425
    },
    {
      "epoch": 0.6840841813817174,
      "grad_norm": 1.4427030086517334,
      "learning_rate": 1.918312311408131e-05,
      "loss": 0.1056,
      "step": 46450
    },
    {
      "epoch": 0.6844523644718046,
      "grad_norm": 1.3449229001998901,
      "learning_rate": 1.9181486742114325e-05,
      "loss": 0.0998,
      "step": 46475
    },
    {
      "epoch": 0.6848205475618916,
      "grad_norm": 1.1819543838500977,
      "learning_rate": 1.9179850370147342e-05,
      "loss": 0.1079,
      "step": 46500
    },
    {
      "epoch": 0.6851887306519786,
      "grad_norm": 1.7344188690185547,
      "learning_rate": 1.9178213998180353e-05,
      "loss": 0.1093,
      "step": 46525
    },
    {
      "epoch": 0.6855569137420656,
      "grad_norm": 1.2552084922790527,
      "learning_rate": 1.917657762621337e-05,
      "loss": 0.1038,
      "step": 46550
    },
    {
      "epoch": 0.6859250968321527,
      "grad_norm": 1.4446220397949219,
      "learning_rate": 1.9174941254246386e-05,
      "loss": 0.1054,
      "step": 46575
    },
    {
      "epoch": 0.6862932799222398,
      "grad_norm": 1.3900601863861084,
      "learning_rate": 1.9173304882279404e-05,
      "loss": 0.1096,
      "step": 46600
    },
    {
      "epoch": 0.6866614630123268,
      "grad_norm": 1.7714508771896362,
      "learning_rate": 1.9171668510312418e-05,
      "loss": 0.1009,
      "step": 46625
    },
    {
      "epoch": 0.6870296461024138,
      "grad_norm": 1.361495852470398,
      "learning_rate": 1.9170032138345432e-05,
      "loss": 0.0987,
      "step": 46650
    },
    {
      "epoch": 0.6873978291925008,
      "grad_norm": 1.5515962839126587,
      "learning_rate": 1.916839576637845e-05,
      "loss": 0.1005,
      "step": 46675
    },
    {
      "epoch": 0.6877660122825879,
      "grad_norm": 1.361990213394165,
      "learning_rate": 1.9166759394411465e-05,
      "loss": 0.1097,
      "step": 46700
    },
    {
      "epoch": 0.688134195372675,
      "grad_norm": 1.5414332151412964,
      "learning_rate": 1.916512302244448e-05,
      "loss": 0.1074,
      "step": 46725
    },
    {
      "epoch": 0.688502378462762,
      "grad_norm": 1.8689407110214233,
      "learning_rate": 1.9163486650477494e-05,
      "loss": 0.0993,
      "step": 46750
    },
    {
      "epoch": 0.688870561552849,
      "grad_norm": 1.325414776802063,
      "learning_rate": 1.9161850278510508e-05,
      "loss": 0.1015,
      "step": 46775
    },
    {
      "epoch": 0.689238744642936,
      "grad_norm": 1.1635738611221313,
      "learning_rate": 1.9160213906543526e-05,
      "loss": 0.1052,
      "step": 46800
    },
    {
      "epoch": 0.6896069277330231,
      "grad_norm": 1.255697250366211,
      "learning_rate": 1.915857753457654e-05,
      "loss": 0.0973,
      "step": 46825
    },
    {
      "epoch": 0.6899751108231101,
      "grad_norm": 2.148343563079834,
      "learning_rate": 1.915694116260956e-05,
      "loss": 0.0998,
      "step": 46850
    },
    {
      "epoch": 0.6903432939131972,
      "grad_norm": 2.244750738143921,
      "learning_rate": 1.9155304790642573e-05,
      "loss": 0.1032,
      "step": 46875
    },
    {
      "epoch": 0.6907114770032842,
      "grad_norm": 1.555161476135254,
      "learning_rate": 1.9153668418675587e-05,
      "loss": 0.1129,
      "step": 46900
    },
    {
      "epoch": 0.6910796600933712,
      "grad_norm": 1.2285830974578857,
      "learning_rate": 1.9152032046708605e-05,
      "loss": 0.102,
      "step": 46925
    },
    {
      "epoch": 0.6914478431834583,
      "grad_norm": 1.63913893699646,
      "learning_rate": 1.9150395674741616e-05,
      "loss": 0.1069,
      "step": 46950
    },
    {
      "epoch": 0.6918160262735453,
      "grad_norm": 1.7224215269088745,
      "learning_rate": 1.9148759302774634e-05,
      "loss": 0.1048,
      "step": 46975
    },
    {
      "epoch": 0.6921842093636323,
      "grad_norm": 1.382535696029663,
      "learning_rate": 1.914712293080765e-05,
      "loss": 0.107,
      "step": 47000
    },
    {
      "epoch": 0.6925523924537194,
      "grad_norm": 1.5920780897140503,
      "learning_rate": 1.9145486558840666e-05,
      "loss": 0.1054,
      "step": 47025
    },
    {
      "epoch": 0.6929205755438064,
      "grad_norm": 1.670591950416565,
      "learning_rate": 1.914385018687368e-05,
      "loss": 0.1017,
      "step": 47050
    },
    {
      "epoch": 0.6932887586338935,
      "grad_norm": 1.547635793685913,
      "learning_rate": 1.9142213814906695e-05,
      "loss": 0.1063,
      "step": 47075
    },
    {
      "epoch": 0.6936569417239805,
      "grad_norm": 1.6910289525985718,
      "learning_rate": 1.9140577442939713e-05,
      "loss": 0.1062,
      "step": 47100
    },
    {
      "epoch": 0.6940251248140675,
      "grad_norm": 1.394062876701355,
      "learning_rate": 1.9138941070972728e-05,
      "loss": 0.1,
      "step": 47125
    },
    {
      "epoch": 0.6943933079041545,
      "grad_norm": 1.03938889503479,
      "learning_rate": 1.9137304699005742e-05,
      "loss": 0.0971,
      "step": 47150
    },
    {
      "epoch": 0.6947614909942417,
      "grad_norm": 1.3395599126815796,
      "learning_rate": 1.9135668327038756e-05,
      "loss": 0.0974,
      "step": 47175
    },
    {
      "epoch": 0.6951296740843287,
      "grad_norm": 1.4715617895126343,
      "learning_rate": 1.913403195507177e-05,
      "loss": 0.1013,
      "step": 47200
    },
    {
      "epoch": 0.6954978571744157,
      "grad_norm": 1.3387597799301147,
      "learning_rate": 1.913239558310479e-05,
      "loss": 0.1071,
      "step": 47225
    },
    {
      "epoch": 0.6958660402645027,
      "grad_norm": 1.8374826908111572,
      "learning_rate": 1.9130759211137803e-05,
      "loss": 0.1064,
      "step": 47250
    },
    {
      "epoch": 0.6962342233545897,
      "grad_norm": 1.387162446975708,
      "learning_rate": 1.912912283917082e-05,
      "loss": 0.1077,
      "step": 47275
    },
    {
      "epoch": 0.6966024064446769,
      "grad_norm": 1.6131870746612549,
      "learning_rate": 1.9127486467203835e-05,
      "loss": 0.0996,
      "step": 47300
    },
    {
      "epoch": 0.6969705895347639,
      "grad_norm": 1.4666141271591187,
      "learning_rate": 1.912585009523685e-05,
      "loss": 0.0977,
      "step": 47325
    },
    {
      "epoch": 0.6973387726248509,
      "grad_norm": 1.093837857246399,
      "learning_rate": 1.9124213723269868e-05,
      "loss": 0.0956,
      "step": 47350
    },
    {
      "epoch": 0.6977069557149379,
      "grad_norm": 1.4562652111053467,
      "learning_rate": 1.912257735130288e-05,
      "loss": 0.1075,
      "step": 47375
    },
    {
      "epoch": 0.6980751388050249,
      "grad_norm": 1.4729552268981934,
      "learning_rate": 1.9120940979335897e-05,
      "loss": 0.0974,
      "step": 47400
    },
    {
      "epoch": 0.698443321895112,
      "grad_norm": 2.071443557739258,
      "learning_rate": 1.911930460736891e-05,
      "loss": 0.0979,
      "step": 47425
    },
    {
      "epoch": 0.6988115049851991,
      "grad_norm": 1.8818188905715942,
      "learning_rate": 1.9117668235401926e-05,
      "loss": 0.1072,
      "step": 47450
    },
    {
      "epoch": 0.6991796880752861,
      "grad_norm": 1.4410918951034546,
      "learning_rate": 1.9116031863434943e-05,
      "loss": 0.099,
      "step": 47475
    },
    {
      "epoch": 0.6995478711653731,
      "grad_norm": 1.4807765483856201,
      "learning_rate": 1.9114395491467958e-05,
      "loss": 0.1082,
      "step": 47500
    },
    {
      "epoch": 0.6999160542554601,
      "grad_norm": 1.0479247570037842,
      "learning_rate": 1.9112759119500976e-05,
      "loss": 0.0945,
      "step": 47525
    },
    {
      "epoch": 0.7002842373455472,
      "grad_norm": 1.7598178386688232,
      "learning_rate": 1.911112274753399e-05,
      "loss": 0.0995,
      "step": 47550
    },
    {
      "epoch": 0.7006524204356342,
      "grad_norm": 0.9280902147293091,
      "learning_rate": 1.9109486375567005e-05,
      "loss": 0.0956,
      "step": 47575
    },
    {
      "epoch": 0.7010206035257213,
      "grad_norm": 1.8822145462036133,
      "learning_rate": 1.910785000360002e-05,
      "loss": 0.0997,
      "step": 47600
    },
    {
      "epoch": 0.7013887866158083,
      "grad_norm": 1.2990777492523193,
      "learning_rate": 1.9106213631633034e-05,
      "loss": 0.0977,
      "step": 47625
    },
    {
      "epoch": 0.7017569697058953,
      "grad_norm": 1.517279863357544,
      "learning_rate": 1.910457725966605e-05,
      "loss": 0.1051,
      "step": 47650
    },
    {
      "epoch": 0.7021251527959824,
      "grad_norm": 1.7701480388641357,
      "learning_rate": 1.9102940887699066e-05,
      "loss": 0.0956,
      "step": 47675
    },
    {
      "epoch": 0.7024933358860694,
      "grad_norm": 1.5461384057998657,
      "learning_rate": 1.910130451573208e-05,
      "loss": 0.1058,
      "step": 47700
    },
    {
      "epoch": 0.7028615189761565,
      "grad_norm": 2.013125419616699,
      "learning_rate": 1.9099668143765098e-05,
      "loss": 0.1029,
      "step": 47725
    },
    {
      "epoch": 0.7032297020662435,
      "grad_norm": 1.3453510999679565,
      "learning_rate": 1.9098031771798113e-05,
      "loss": 0.1037,
      "step": 47750
    },
    {
      "epoch": 0.7035978851563306,
      "grad_norm": 1.1628808975219727,
      "learning_rate": 1.9096395399831127e-05,
      "loss": 0.1009,
      "step": 47775
    },
    {
      "epoch": 0.7039660682464176,
      "grad_norm": 1.7610334157943726,
      "learning_rate": 1.909475902786414e-05,
      "loss": 0.0948,
      "step": 47800
    },
    {
      "epoch": 0.7043342513365046,
      "grad_norm": 1.581941843032837,
      "learning_rate": 1.909312265589716e-05,
      "loss": 0.1027,
      "step": 47825
    },
    {
      "epoch": 0.7047024344265916,
      "grad_norm": 1.7696298360824585,
      "learning_rate": 1.9091486283930174e-05,
      "loss": 0.0988,
      "step": 47850
    },
    {
      "epoch": 0.7050706175166787,
      "grad_norm": 1.5074560642242432,
      "learning_rate": 1.9089915366841868e-05,
      "loss": 0.1077,
      "step": 47875
    },
    {
      "epoch": 0.7054388006067658,
      "grad_norm": 1.5226465463638306,
      "learning_rate": 1.9088278994874886e-05,
      "loss": 0.1016,
      "step": 47900
    },
    {
      "epoch": 0.7058069836968528,
      "grad_norm": 1.6150729656219482,
      "learning_rate": 1.90866426229079e-05,
      "loss": 0.1051,
      "step": 47925
    },
    {
      "epoch": 0.7061751667869398,
      "grad_norm": 1.286678433418274,
      "learning_rate": 1.9085006250940915e-05,
      "loss": 0.0969,
      "step": 47950
    },
    {
      "epoch": 0.7065433498770268,
      "grad_norm": 1.6489728689193726,
      "learning_rate": 1.9083369878973932e-05,
      "loss": 0.1043,
      "step": 47975
    },
    {
      "epoch": 0.7069115329671138,
      "grad_norm": 1.15176260471344,
      "learning_rate": 1.9081733507006947e-05,
      "loss": 0.1039,
      "step": 48000
    },
    {
      "epoch": 0.707279716057201,
      "grad_norm": 1.4781898260116577,
      "learning_rate": 1.908009713503996e-05,
      "loss": 0.1004,
      "step": 48025
    },
    {
      "epoch": 0.707647899147288,
      "grad_norm": 1.9934347867965698,
      "learning_rate": 1.9078460763072976e-05,
      "loss": 0.0992,
      "step": 48050
    },
    {
      "epoch": 0.708016082237375,
      "grad_norm": 1.642067313194275,
      "learning_rate": 1.9076824391105994e-05,
      "loss": 0.1089,
      "step": 48075
    },
    {
      "epoch": 0.708384265327462,
      "grad_norm": 1.3216331005096436,
      "learning_rate": 1.9075188019139008e-05,
      "loss": 0.1035,
      "step": 48100
    },
    {
      "epoch": 0.708752448417549,
      "grad_norm": 1.0773968696594238,
      "learning_rate": 1.9073551647172022e-05,
      "loss": 0.1009,
      "step": 48125
    },
    {
      "epoch": 0.7091206315076362,
      "grad_norm": 1.0356942415237427,
      "learning_rate": 1.907191527520504e-05,
      "loss": 0.1127,
      "step": 48150
    },
    {
      "epoch": 0.7094888145977232,
      "grad_norm": 1.7541707754135132,
      "learning_rate": 1.9070278903238055e-05,
      "loss": 0.102,
      "step": 48175
    },
    {
      "epoch": 0.7098569976878102,
      "grad_norm": 1.1507172584533691,
      "learning_rate": 1.906864253127107e-05,
      "loss": 0.1051,
      "step": 48200
    },
    {
      "epoch": 0.7102251807778972,
      "grad_norm": 1.210410237312317,
      "learning_rate": 1.9067006159304084e-05,
      "loss": 0.0963,
      "step": 48225
    },
    {
      "epoch": 0.7105933638679842,
      "grad_norm": 1.0585200786590576,
      "learning_rate": 1.90653697873371e-05,
      "loss": 0.0939,
      "step": 48250
    },
    {
      "epoch": 0.7109615469580713,
      "grad_norm": 1.4461408853530884,
      "learning_rate": 1.9063733415370116e-05,
      "loss": 0.0969,
      "step": 48275
    },
    {
      "epoch": 0.7113297300481584,
      "grad_norm": 1.4013006687164307,
      "learning_rate": 1.906209704340313e-05,
      "loss": 0.1043,
      "step": 48300
    },
    {
      "epoch": 0.7116979131382454,
      "grad_norm": 1.6809771060943604,
      "learning_rate": 1.906046067143615e-05,
      "loss": 0.1049,
      "step": 48325
    },
    {
      "epoch": 0.7120660962283324,
      "grad_norm": 1.4544810056686401,
      "learning_rate": 1.9058824299469163e-05,
      "loss": 0.0983,
      "step": 48350
    },
    {
      "epoch": 0.7124342793184195,
      "grad_norm": 1.4573628902435303,
      "learning_rate": 1.9057187927502177e-05,
      "loss": 0.1028,
      "step": 48375
    },
    {
      "epoch": 0.7128024624085065,
      "grad_norm": 1.4733716249465942,
      "learning_rate": 1.9055551555535195e-05,
      "loss": 0.0979,
      "step": 48400
    },
    {
      "epoch": 0.7131706454985935,
      "grad_norm": 1.5174107551574707,
      "learning_rate": 1.905391518356821e-05,
      "loss": 0.1083,
      "step": 48425
    },
    {
      "epoch": 0.7135388285886806,
      "grad_norm": 1.47136652469635,
      "learning_rate": 1.9052278811601224e-05,
      "loss": 0.0999,
      "step": 48450
    },
    {
      "epoch": 0.7139070116787676,
      "grad_norm": 1.5804349184036255,
      "learning_rate": 1.905064243963424e-05,
      "loss": 0.0951,
      "step": 48475
    },
    {
      "epoch": 0.7142751947688547,
      "grad_norm": 1.93483567237854,
      "learning_rate": 1.9049006067667256e-05,
      "loss": 0.1065,
      "step": 48500
    },
    {
      "epoch": 0.7146433778589417,
      "grad_norm": 1.601036787033081,
      "learning_rate": 1.904736969570027e-05,
      "loss": 0.1056,
      "step": 48525
    },
    {
      "epoch": 0.7150115609490287,
      "grad_norm": 0.9350731372833252,
      "learning_rate": 1.9045733323733285e-05,
      "loss": 0.1049,
      "step": 48550
    },
    {
      "epoch": 0.7153797440391158,
      "grad_norm": 1.6524691581726074,
      "learning_rate": 1.9044096951766303e-05,
      "loss": 0.0961,
      "step": 48575
    },
    {
      "epoch": 0.7157479271292028,
      "grad_norm": 1.2378066778182983,
      "learning_rate": 1.9042460579799318e-05,
      "loss": 0.0932,
      "step": 48600
    },
    {
      "epoch": 0.7161161102192899,
      "grad_norm": 1.697403907775879,
      "learning_rate": 1.9040824207832332e-05,
      "loss": 0.106,
      "step": 48625
    },
    {
      "epoch": 0.7164842933093769,
      "grad_norm": 1.5412137508392334,
      "learning_rate": 1.9039187835865346e-05,
      "loss": 0.111,
      "step": 48650
    },
    {
      "epoch": 0.7168524763994639,
      "grad_norm": 1.4458122253417969,
      "learning_rate": 1.9037551463898364e-05,
      "loss": 0.101,
      "step": 48675
    },
    {
      "epoch": 0.7172206594895509,
      "grad_norm": 2.023613691329956,
      "learning_rate": 1.903591509193138e-05,
      "loss": 0.1081,
      "step": 48700
    },
    {
      "epoch": 0.717588842579638,
      "grad_norm": 1.2220475673675537,
      "learning_rate": 1.9034278719964393e-05,
      "loss": 0.0996,
      "step": 48725
    },
    {
      "epoch": 0.7179570256697251,
      "grad_norm": 1.5562247037887573,
      "learning_rate": 1.903264234799741e-05,
      "loss": 0.1014,
      "step": 48750
    },
    {
      "epoch": 0.7183252087598121,
      "grad_norm": 1.511494755744934,
      "learning_rate": 1.9031005976030425e-05,
      "loss": 0.1051,
      "step": 48775
    },
    {
      "epoch": 0.7186933918498991,
      "grad_norm": 1.2906818389892578,
      "learning_rate": 1.902936960406344e-05,
      "loss": 0.0984,
      "step": 48800
    },
    {
      "epoch": 0.7190615749399861,
      "grad_norm": 1.2554727792739868,
      "learning_rate": 1.9027733232096458e-05,
      "loss": 0.0995,
      "step": 48825
    },
    {
      "epoch": 0.7194297580300731,
      "grad_norm": 1.3066587448120117,
      "learning_rate": 1.902609686012947e-05,
      "loss": 0.1006,
      "step": 48850
    },
    {
      "epoch": 0.7197979411201603,
      "grad_norm": 1.3932998180389404,
      "learning_rate": 1.9024460488162487e-05,
      "loss": 0.0994,
      "step": 48875
    },
    {
      "epoch": 0.7201661242102473,
      "grad_norm": 1.5048588514328003,
      "learning_rate": 1.90228241161955e-05,
      "loss": 0.0983,
      "step": 48900
    },
    {
      "epoch": 0.7205343073003343,
      "grad_norm": 1.2415173053741455,
      "learning_rate": 1.902118774422852e-05,
      "loss": 0.1065,
      "step": 48925
    },
    {
      "epoch": 0.7209024903904213,
      "grad_norm": 1.327630877494812,
      "learning_rate": 1.9019551372261533e-05,
      "loss": 0.0943,
      "step": 48950
    },
    {
      "epoch": 0.7212706734805084,
      "grad_norm": 1.4476395845413208,
      "learning_rate": 1.9017915000294548e-05,
      "loss": 0.1102,
      "step": 48975
    },
    {
      "epoch": 0.7216388565705955,
      "grad_norm": 1.1207923889160156,
      "learning_rate": 1.9016278628327566e-05,
      "loss": 0.0969,
      "step": 49000
    },
    {
      "epoch": 0.7220070396606825,
      "grad_norm": 1.5516877174377441,
      "learning_rate": 1.901464225636058e-05,
      "loss": 0.0969,
      "step": 49025
    },
    {
      "epoch": 0.7223752227507695,
      "grad_norm": 1.5809789896011353,
      "learning_rate": 1.9013005884393595e-05,
      "loss": 0.1005,
      "step": 49050
    },
    {
      "epoch": 0.7227434058408565,
      "grad_norm": 1.2460857629776,
      "learning_rate": 1.901136951242661e-05,
      "loss": 0.0944,
      "step": 49075
    },
    {
      "epoch": 0.7231115889309436,
      "grad_norm": 1.2830605506896973,
      "learning_rate": 1.9009733140459624e-05,
      "loss": 0.0912,
      "step": 49100
    },
    {
      "epoch": 0.7234797720210306,
      "grad_norm": 1.442643642425537,
      "learning_rate": 1.900809676849264e-05,
      "loss": 0.1051,
      "step": 49125
    },
    {
      "epoch": 0.7238479551111177,
      "grad_norm": 1.4712581634521484,
      "learning_rate": 1.9006460396525656e-05,
      "loss": 0.1013,
      "step": 49150
    },
    {
      "epoch": 0.7242161382012047,
      "grad_norm": 1.5503424406051636,
      "learning_rate": 1.9004824024558674e-05,
      "loss": 0.1065,
      "step": 49175
    },
    {
      "epoch": 0.7245843212912917,
      "grad_norm": 1.4155282974243164,
      "learning_rate": 1.9003187652591688e-05,
      "loss": 0.1058,
      "step": 49200
    },
    {
      "epoch": 0.7249525043813788,
      "grad_norm": 1.432908535003662,
      "learning_rate": 1.9001551280624703e-05,
      "loss": 0.0947,
      "step": 49225
    },
    {
      "epoch": 0.7253206874714658,
      "grad_norm": 1.153944969177246,
      "learning_rate": 1.8999914908657717e-05,
      "loss": 0.1052,
      "step": 49250
    },
    {
      "epoch": 0.7256888705615528,
      "grad_norm": 1.7101420164108276,
      "learning_rate": 1.899827853669073e-05,
      "loss": 0.0985,
      "step": 49275
    },
    {
      "epoch": 0.7260570536516399,
      "grad_norm": 1.1989765167236328,
      "learning_rate": 1.899664216472375e-05,
      "loss": 0.0974,
      "step": 49300
    },
    {
      "epoch": 0.7264252367417269,
      "grad_norm": 1.0323476791381836,
      "learning_rate": 1.8995005792756764e-05,
      "loss": 0.0998,
      "step": 49325
    },
    {
      "epoch": 0.726793419831814,
      "grad_norm": 1.4167869091033936,
      "learning_rate": 1.8993369420789782e-05,
      "loss": 0.0997,
      "step": 49350
    },
    {
      "epoch": 0.727161602921901,
      "grad_norm": 1.1620646715164185,
      "learning_rate": 1.8991733048822796e-05,
      "loss": 0.1069,
      "step": 49375
    },
    {
      "epoch": 0.727529786011988,
      "grad_norm": 0.7772794961929321,
      "learning_rate": 1.899009667685581e-05,
      "loss": 0.0982,
      "step": 49400
    },
    {
      "epoch": 0.727897969102075,
      "grad_norm": 1.385764241218567,
      "learning_rate": 1.898846030488883e-05,
      "loss": 0.0987,
      "step": 49425
    },
    {
      "epoch": 0.7282661521921621,
      "grad_norm": 1.3250408172607422,
      "learning_rate": 1.8986823932921843e-05,
      "loss": 0.0984,
      "step": 49450
    },
    {
      "epoch": 0.7286343352822492,
      "grad_norm": 1.0105986595153809,
      "learning_rate": 1.8985187560954857e-05,
      "loss": 0.0967,
      "step": 49475
    },
    {
      "epoch": 0.7290025183723362,
      "grad_norm": 1.5108296871185303,
      "learning_rate": 1.8983551188987872e-05,
      "loss": 0.1031,
      "step": 49500
    },
    {
      "epoch": 0.7293707014624232,
      "grad_norm": 1.0454095602035522,
      "learning_rate": 1.8981914817020886e-05,
      "loss": 0.0982,
      "step": 49525
    },
    {
      "epoch": 0.7297388845525102,
      "grad_norm": 0.983040452003479,
      "learning_rate": 1.8980278445053904e-05,
      "loss": 0.1011,
      "step": 49550
    },
    {
      "epoch": 0.7301070676425974,
      "grad_norm": 1.607489824295044,
      "learning_rate": 1.897864207308692e-05,
      "loss": 0.0948,
      "step": 49575
    },
    {
      "epoch": 0.7304752507326844,
      "grad_norm": 1.830379605293274,
      "learning_rate": 1.8977005701119936e-05,
      "loss": 0.0978,
      "step": 49600
    },
    {
      "epoch": 0.7308434338227714,
      "grad_norm": 1.6186549663543701,
      "learning_rate": 1.897536932915295e-05,
      "loss": 0.0967,
      "step": 49625
    },
    {
      "epoch": 0.7312116169128584,
      "grad_norm": 1.5712846517562866,
      "learning_rate": 1.8973732957185965e-05,
      "loss": 0.105,
      "step": 49650
    },
    {
      "epoch": 0.7315798000029454,
      "grad_norm": 1.5042476654052734,
      "learning_rate": 1.897209658521898e-05,
      "loss": 0.1053,
      "step": 49675
    },
    {
      "epoch": 0.7319479830930326,
      "grad_norm": 1.6564544439315796,
      "learning_rate": 1.8970460213251994e-05,
      "loss": 0.0985,
      "step": 49700
    },
    {
      "epoch": 0.7323161661831196,
      "grad_norm": 1.3116837739944458,
      "learning_rate": 1.8968823841285012e-05,
      "loss": 0.0964,
      "step": 49725
    },
    {
      "epoch": 0.7326843492732066,
      "grad_norm": 2.3516788482666016,
      "learning_rate": 1.8967187469318027e-05,
      "loss": 0.0924,
      "step": 49750
    },
    {
      "epoch": 0.7330525323632936,
      "grad_norm": 1.4502614736557007,
      "learning_rate": 1.896555109735104e-05,
      "loss": 0.1021,
      "step": 49775
    },
    {
      "epoch": 0.7334207154533806,
      "grad_norm": 1.4912548065185547,
      "learning_rate": 1.896391472538406e-05,
      "loss": 0.1108,
      "step": 49800
    },
    {
      "epoch": 0.7337888985434677,
      "grad_norm": 1.382511019706726,
      "learning_rate": 1.8962278353417073e-05,
      "loss": 0.1022,
      "step": 49825
    },
    {
      "epoch": 0.7341570816335548,
      "grad_norm": 1.6152180433273315,
      "learning_rate": 1.896064198145009e-05,
      "loss": 0.1032,
      "step": 49850
    },
    {
      "epoch": 0.7345252647236418,
      "grad_norm": 1.4031436443328857,
      "learning_rate": 1.8959071064361785e-05,
      "loss": 0.0986,
      "step": 49875
    },
    {
      "epoch": 0.7348934478137288,
      "grad_norm": 1.2503541707992554,
      "learning_rate": 1.89574346923948e-05,
      "loss": 0.0975,
      "step": 49900
    },
    {
      "epoch": 0.7352616309038158,
      "grad_norm": 1.4993572235107422,
      "learning_rate": 1.8955798320427814e-05,
      "loss": 0.0995,
      "step": 49925
    },
    {
      "epoch": 0.7356298139939029,
      "grad_norm": 1.1862359046936035,
      "learning_rate": 1.895416194846083e-05,
      "loss": 0.1063,
      "step": 49950
    },
    {
      "epoch": 0.7359979970839899,
      "grad_norm": 1.5639508962631226,
      "learning_rate": 1.8952525576493846e-05,
      "loss": 0.1043,
      "step": 49975
    },
    {
      "epoch": 0.736366180174077,
      "grad_norm": 1.435240387916565,
      "learning_rate": 1.895088920452686e-05,
      "loss": 0.098,
      "step": 50000
    },
    {
      "epoch": 0.736734363264164,
      "grad_norm": 1.7050061225891113,
      "learning_rate": 1.8949252832559875e-05,
      "loss": 0.0956,
      "step": 50025
    },
    {
      "epoch": 0.737102546354251,
      "grad_norm": 1.7929081916809082,
      "learning_rate": 1.8947616460592893e-05,
      "loss": 0.1079,
      "step": 50050
    },
    {
      "epoch": 0.7374707294443381,
      "grad_norm": 1.2964180707931519,
      "learning_rate": 1.8945980088625908e-05,
      "loss": 0.0921,
      "step": 50075
    },
    {
      "epoch": 0.7378389125344251,
      "grad_norm": 1.3983927965164185,
      "learning_rate": 1.8944343716658922e-05,
      "loss": 0.098,
      "step": 50100
    },
    {
      "epoch": 0.7382070956245121,
      "grad_norm": 1.472080111503601,
      "learning_rate": 1.8942707344691936e-05,
      "loss": 0.0974,
      "step": 50125
    },
    {
      "epoch": 0.7385752787145992,
      "grad_norm": 1.0241786241531372,
      "learning_rate": 1.8941070972724954e-05,
      "loss": 0.099,
      "step": 50150
    },
    {
      "epoch": 0.7389434618046863,
      "grad_norm": 1.2100846767425537,
      "learning_rate": 1.893943460075797e-05,
      "loss": 0.0942,
      "step": 50175
    },
    {
      "epoch": 0.7393116448947733,
      "grad_norm": 1.2217756509780884,
      "learning_rate": 1.8937798228790983e-05,
      "loss": 0.1009,
      "step": 50200
    },
    {
      "epoch": 0.7396798279848603,
      "grad_norm": 1.4107080698013306,
      "learning_rate": 1.8936161856824e-05,
      "loss": 0.0926,
      "step": 50225
    },
    {
      "epoch": 0.7400480110749473,
      "grad_norm": 1.2571321725845337,
      "learning_rate": 1.8934525484857015e-05,
      "loss": 0.0994,
      "step": 50250
    },
    {
      "epoch": 0.7404161941650343,
      "grad_norm": 1.1018223762512207,
      "learning_rate": 1.893288911289003e-05,
      "loss": 0.0966,
      "step": 50275
    },
    {
      "epoch": 0.7407843772551215,
      "grad_norm": 1.0355523824691772,
      "learning_rate": 1.8931252740923048e-05,
      "loss": 0.0983,
      "step": 50300
    },
    {
      "epoch": 0.7411525603452085,
      "grad_norm": 1.5832747220993042,
      "learning_rate": 1.8929616368956062e-05,
      "loss": 0.1021,
      "step": 50325
    },
    {
      "epoch": 0.7415207434352955,
      "grad_norm": 1.28887939453125,
      "learning_rate": 1.8927979996989077e-05,
      "loss": 0.094,
      "step": 50350
    },
    {
      "epoch": 0.7418889265253825,
      "grad_norm": 1.5176652669906616,
      "learning_rate": 1.892634362502209e-05,
      "loss": 0.1066,
      "step": 50375
    },
    {
      "epoch": 0.7422571096154695,
      "grad_norm": 1.239126205444336,
      "learning_rate": 1.892470725305511e-05,
      "loss": 0.1003,
      "step": 50400
    },
    {
      "epoch": 0.7426252927055567,
      "grad_norm": 2.0967373847961426,
      "learning_rate": 1.8923070881088123e-05,
      "loss": 0.1038,
      "step": 50425
    },
    {
      "epoch": 0.7429934757956437,
      "grad_norm": 1.3348290920257568,
      "learning_rate": 1.8921434509121138e-05,
      "loss": 0.0968,
      "step": 50450
    },
    {
      "epoch": 0.7433616588857307,
      "grad_norm": 1.012868046760559,
      "learning_rate": 1.8919798137154156e-05,
      "loss": 0.1006,
      "step": 50475
    },
    {
      "epoch": 0.7437298419758177,
      "grad_norm": 1.712973713874817,
      "learning_rate": 1.891816176518717e-05,
      "loss": 0.1002,
      "step": 50500
    },
    {
      "epoch": 0.7440980250659047,
      "grad_norm": 1.1453803777694702,
      "learning_rate": 1.8916525393220185e-05,
      "loss": 0.0967,
      "step": 50525
    },
    {
      "epoch": 0.7444662081559918,
      "grad_norm": 1.5949763059616089,
      "learning_rate": 1.89148890212532e-05,
      "loss": 0.1106,
      "step": 50550
    },
    {
      "epoch": 0.7448343912460789,
      "grad_norm": 1.4714535474777222,
      "learning_rate": 1.8913252649286217e-05,
      "loss": 0.0939,
      "step": 50575
    },
    {
      "epoch": 0.7452025743361659,
      "grad_norm": 1.741516351699829,
      "learning_rate": 1.891161627731923e-05,
      "loss": 0.1047,
      "step": 50600
    },
    {
      "epoch": 0.7455707574262529,
      "grad_norm": 1.5519564151763916,
      "learning_rate": 1.8909979905352246e-05,
      "loss": 0.1025,
      "step": 50625
    },
    {
      "epoch": 0.7459389405163399,
      "grad_norm": 1.5566942691802979,
      "learning_rate": 1.8908343533385264e-05,
      "loss": 0.098,
      "step": 50650
    },
    {
      "epoch": 0.746307123606427,
      "grad_norm": 1.7024180889129639,
      "learning_rate": 1.8906707161418278e-05,
      "loss": 0.0984,
      "step": 50675
    },
    {
      "epoch": 0.746675306696514,
      "grad_norm": 1.2596818208694458,
      "learning_rate": 1.8905070789451293e-05,
      "loss": 0.101,
      "step": 50700
    },
    {
      "epoch": 0.7470434897866011,
      "grad_norm": 1.231046438217163,
      "learning_rate": 1.8903434417484307e-05,
      "loss": 0.0949,
      "step": 50725
    },
    {
      "epoch": 0.7474116728766881,
      "grad_norm": 1.4248147010803223,
      "learning_rate": 1.8901798045517325e-05,
      "loss": 0.0969,
      "step": 50750
    },
    {
      "epoch": 0.7477798559667752,
      "grad_norm": 1.8414579629898071,
      "learning_rate": 1.890016167355034e-05,
      "loss": 0.094,
      "step": 50775
    },
    {
      "epoch": 0.7481480390568622,
      "grad_norm": 1.1229593753814697,
      "learning_rate": 1.8898525301583354e-05,
      "loss": 0.1004,
      "step": 50800
    },
    {
      "epoch": 0.7485162221469492,
      "grad_norm": 1.872498631477356,
      "learning_rate": 1.8896888929616372e-05,
      "loss": 0.0978,
      "step": 50825
    },
    {
      "epoch": 0.7488844052370363,
      "grad_norm": 1.0315674543380737,
      "learning_rate": 1.8895252557649386e-05,
      "loss": 0.1044,
      "step": 50850
    },
    {
      "epoch": 0.7492525883271233,
      "grad_norm": 1.80376398563385,
      "learning_rate": 1.88936161856824e-05,
      "loss": 0.1019,
      "step": 50875
    },
    {
      "epoch": 0.7496207714172104,
      "grad_norm": 1.7327715158462524,
      "learning_rate": 1.889197981371542e-05,
      "loss": 0.0947,
      "step": 50900
    },
    {
      "epoch": 0.7499889545072974,
      "grad_norm": 1.1131285429000854,
      "learning_rate": 1.8890343441748433e-05,
      "loss": 0.0939,
      "step": 50925
    },
    {
      "epoch": 0.7503571375973844,
      "grad_norm": 1.7352478504180908,
      "learning_rate": 1.8888707069781447e-05,
      "loss": 0.0947,
      "step": 50950
    },
    {
      "epoch": 0.7507253206874714,
      "grad_norm": 1.7419734001159668,
      "learning_rate": 1.8887070697814462e-05,
      "loss": 0.0983,
      "step": 50975
    },
    {
      "epoch": 0.7510935037775585,
      "grad_norm": 1.6313399076461792,
      "learning_rate": 1.888543432584748e-05,
      "loss": 0.0972,
      "step": 51000
    },
    {
      "epoch": 0.7514616868676456,
      "grad_norm": 1.653120756149292,
      "learning_rate": 1.8883797953880494e-05,
      "loss": 0.1021,
      "step": 51025
    },
    {
      "epoch": 0.7518298699577326,
      "grad_norm": 1.4025392532348633,
      "learning_rate": 1.888216158191351e-05,
      "loss": 0.0995,
      "step": 51050
    },
    {
      "epoch": 0.7521980530478196,
      "grad_norm": 1.034682273864746,
      "learning_rate": 1.8880525209946526e-05,
      "loss": 0.0936,
      "step": 51075
    },
    {
      "epoch": 0.7525662361379066,
      "grad_norm": 1.1712440252304077,
      "learning_rate": 1.887888883797954e-05,
      "loss": 0.0963,
      "step": 51100
    },
    {
      "epoch": 0.7529344192279936,
      "grad_norm": 1.9316719770431519,
      "learning_rate": 1.8877252466012555e-05,
      "loss": 0.1021,
      "step": 51125
    },
    {
      "epoch": 0.7533026023180808,
      "grad_norm": 1.1776498556137085,
      "learning_rate": 1.887561609404557e-05,
      "loss": 0.0985,
      "step": 51150
    },
    {
      "epoch": 0.7536707854081678,
      "grad_norm": 1.3127164840698242,
      "learning_rate": 1.8873979722078584e-05,
      "loss": 0.0965,
      "step": 51175
    },
    {
      "epoch": 0.7540389684982548,
      "grad_norm": 1.6149221658706665,
      "learning_rate": 1.8872343350111602e-05,
      "loss": 0.0954,
      "step": 51200
    },
    {
      "epoch": 0.7544071515883418,
      "grad_norm": 1.319986343383789,
      "learning_rate": 1.8870706978144617e-05,
      "loss": 0.1009,
      "step": 51225
    },
    {
      "epoch": 0.7547753346784289,
      "grad_norm": 1.0375257730484009,
      "learning_rate": 1.8869070606177634e-05,
      "loss": 0.0988,
      "step": 51250
    },
    {
      "epoch": 0.755143517768516,
      "grad_norm": 1.3273632526397705,
      "learning_rate": 1.886743423421065e-05,
      "loss": 0.0969,
      "step": 51275
    },
    {
      "epoch": 0.755511700858603,
      "grad_norm": 1.7834882736206055,
      "learning_rate": 1.8865797862243663e-05,
      "loss": 0.1045,
      "step": 51300
    },
    {
      "epoch": 0.75587988394869,
      "grad_norm": 1.6718878746032715,
      "learning_rate": 1.886416149027668e-05,
      "loss": 0.0987,
      "step": 51325
    },
    {
      "epoch": 0.756248067038777,
      "grad_norm": 1.5869431495666504,
      "learning_rate": 1.8862525118309696e-05,
      "loss": 0.102,
      "step": 51350
    },
    {
      "epoch": 0.7566162501288641,
      "grad_norm": 1.3101059198379517,
      "learning_rate": 1.886088874634271e-05,
      "loss": 0.0975,
      "step": 51375
    },
    {
      "epoch": 0.7569844332189511,
      "grad_norm": 1.4751836061477661,
      "learning_rate": 1.8859252374375725e-05,
      "loss": 0.103,
      "step": 51400
    },
    {
      "epoch": 0.7573526163090382,
      "grad_norm": 1.408093810081482,
      "learning_rate": 1.885761600240874e-05,
      "loss": 0.0998,
      "step": 51425
    },
    {
      "epoch": 0.7577207993991252,
      "grad_norm": 1.8048471212387085,
      "learning_rate": 1.8855979630441757e-05,
      "loss": 0.0959,
      "step": 51450
    },
    {
      "epoch": 0.7580889824892122,
      "grad_norm": 1.5421810150146484,
      "learning_rate": 1.885434325847477e-05,
      "loss": 0.096,
      "step": 51475
    },
    {
      "epoch": 0.7584571655792993,
      "grad_norm": 1.2161544561386108,
      "learning_rate": 1.885270688650779e-05,
      "loss": 0.1043,
      "step": 51500
    },
    {
      "epoch": 0.7588253486693863,
      "grad_norm": 0.9103922247886658,
      "learning_rate": 1.8851070514540804e-05,
      "loss": 0.0863,
      "step": 51525
    },
    {
      "epoch": 0.7591935317594734,
      "grad_norm": 1.5881121158599854,
      "learning_rate": 1.8849434142573818e-05,
      "loss": 0.0995,
      "step": 51550
    },
    {
      "epoch": 0.7595617148495604,
      "grad_norm": 1.5382658243179321,
      "learning_rate": 1.8847797770606833e-05,
      "loss": 0.0954,
      "step": 51575
    },
    {
      "epoch": 0.7599298979396474,
      "grad_norm": 1.3792065382003784,
      "learning_rate": 1.8846161398639847e-05,
      "loss": 0.0971,
      "step": 51600
    },
    {
      "epoch": 0.7602980810297345,
      "grad_norm": 1.2182016372680664,
      "learning_rate": 1.8844525026672865e-05,
      "loss": 0.1004,
      "step": 51625
    },
    {
      "epoch": 0.7606662641198215,
      "grad_norm": 1.3744044303894043,
      "learning_rate": 1.884288865470588e-05,
      "loss": 0.1003,
      "step": 51650
    },
    {
      "epoch": 0.7610344472099085,
      "grad_norm": 1.3731317520141602,
      "learning_rate": 1.8841252282738897e-05,
      "loss": 0.101,
      "step": 51675
    },
    {
      "epoch": 0.7614026302999956,
      "grad_norm": 1.3473219871520996,
      "learning_rate": 1.883961591077191e-05,
      "loss": 0.1077,
      "step": 51700
    },
    {
      "epoch": 0.7617708133900826,
      "grad_norm": 1.207746148109436,
      "learning_rate": 1.8837979538804926e-05,
      "loss": 0.0994,
      "step": 51725
    },
    {
      "epoch": 0.7621389964801697,
      "grad_norm": 1.661965250968933,
      "learning_rate": 1.8836343166837944e-05,
      "loss": 0.1004,
      "step": 51750
    },
    {
      "epoch": 0.7625071795702567,
      "grad_norm": 1.4879807233810425,
      "learning_rate": 1.883470679487096e-05,
      "loss": 0.0915,
      "step": 51775
    },
    {
      "epoch": 0.7628753626603437,
      "grad_norm": 1.2649250030517578,
      "learning_rate": 1.8833070422903973e-05,
      "loss": 0.1006,
      "step": 51800
    },
    {
      "epoch": 0.7632435457504307,
      "grad_norm": 1.497623085975647,
      "learning_rate": 1.8831434050936987e-05,
      "loss": 0.097,
      "step": 51825
    },
    {
      "epoch": 0.7636117288405179,
      "grad_norm": 1.1468896865844727,
      "learning_rate": 1.8829797678970002e-05,
      "loss": 0.0939,
      "step": 51850
    },
    {
      "epoch": 0.7639799119306049,
      "grad_norm": 2.0533607006073,
      "learning_rate": 1.882816130700302e-05,
      "loss": 0.1044,
      "step": 51875
    },
    {
      "epoch": 0.7643480950206919,
      "grad_norm": 1.141491413116455,
      "learning_rate": 1.8826590389914713e-05,
      "loss": 0.1028,
      "step": 51900
    },
    {
      "epoch": 0.7647162781107789,
      "grad_norm": 1.504295825958252,
      "learning_rate": 1.882495401794773e-05,
      "loss": 0.0931,
      "step": 51925
    },
    {
      "epoch": 0.7650844612008659,
      "grad_norm": 1.668541431427002,
      "learning_rate": 1.8823317645980746e-05,
      "loss": 0.0998,
      "step": 51950
    },
    {
      "epoch": 0.765452644290953,
      "grad_norm": 1.099595308303833,
      "learning_rate": 1.882168127401376e-05,
      "loss": 0.0928,
      "step": 51975
    },
    {
      "epoch": 0.7658208273810401,
      "grad_norm": 2.686854839324951,
      "learning_rate": 1.8820044902046775e-05,
      "loss": 0.0979,
      "step": 52000
    },
    {
      "epoch": 0.7661890104711271,
      "grad_norm": 1.238408088684082,
      "learning_rate": 1.881840853007979e-05,
      "loss": 0.0923,
      "step": 52025
    },
    {
      "epoch": 0.7665571935612141,
      "grad_norm": 1.9959614276885986,
      "learning_rate": 1.8816772158112807e-05,
      "loss": 0.0916,
      "step": 52050
    },
    {
      "epoch": 0.7669253766513011,
      "grad_norm": 1.8930375576019287,
      "learning_rate": 1.881513578614582e-05,
      "loss": 0.0878,
      "step": 52075
    },
    {
      "epoch": 0.7672935597413882,
      "grad_norm": 1.0580706596374512,
      "learning_rate": 1.8813499414178836e-05,
      "loss": 0.0916,
      "step": 52100
    },
    {
      "epoch": 0.7676617428314753,
      "grad_norm": 1.445089340209961,
      "learning_rate": 1.8811863042211854e-05,
      "loss": 0.1049,
      "step": 52125
    },
    {
      "epoch": 0.7680299259215623,
      "grad_norm": 1.469886064529419,
      "learning_rate": 1.8810226670244868e-05,
      "loss": 0.0931,
      "step": 52150
    },
    {
      "epoch": 0.7683981090116493,
      "grad_norm": 1.663535714149475,
      "learning_rate": 1.8808590298277886e-05,
      "loss": 0.094,
      "step": 52175
    },
    {
      "epoch": 0.7687662921017363,
      "grad_norm": 0.9408849477767944,
      "learning_rate": 1.88069539263109e-05,
      "loss": 0.0922,
      "step": 52200
    },
    {
      "epoch": 0.7691344751918234,
      "grad_norm": 1.3737974166870117,
      "learning_rate": 1.8805317554343915e-05,
      "loss": 0.0944,
      "step": 52225
    },
    {
      "epoch": 0.7695026582819104,
      "grad_norm": 1.6405061483383179,
      "learning_rate": 1.880368118237693e-05,
      "loss": 0.0995,
      "step": 52250
    },
    {
      "epoch": 0.7698708413719975,
      "grad_norm": 1.3962156772613525,
      "learning_rate": 1.8802044810409944e-05,
      "loss": 0.0988,
      "step": 52275
    },
    {
      "epoch": 0.7702390244620845,
      "grad_norm": 2.0040688514709473,
      "learning_rate": 1.8800408438442962e-05,
      "loss": 0.1005,
      "step": 52300
    },
    {
      "epoch": 0.7706072075521715,
      "grad_norm": 1.5786962509155273,
      "learning_rate": 1.8798772066475976e-05,
      "loss": 0.1015,
      "step": 52325
    },
    {
      "epoch": 0.7709753906422586,
      "grad_norm": 1.340768575668335,
      "learning_rate": 1.879713569450899e-05,
      "loss": 0.1015,
      "step": 52350
    },
    {
      "epoch": 0.7713435737323456,
      "grad_norm": 1.3153142929077148,
      "learning_rate": 1.879549932254201e-05,
      "loss": 0.1,
      "step": 52375
    },
    {
      "epoch": 0.7717117568224326,
      "grad_norm": 1.3656054735183716,
      "learning_rate": 1.8793862950575023e-05,
      "loss": 0.093,
      "step": 52400
    },
    {
      "epoch": 0.7720799399125197,
      "grad_norm": 1.4795976877212524,
      "learning_rate": 1.8792226578608037e-05,
      "loss": 0.093,
      "step": 52425
    },
    {
      "epoch": 0.7724481230026068,
      "grad_norm": 1.3683342933654785,
      "learning_rate": 1.8790590206641052e-05,
      "loss": 0.0931,
      "step": 52450
    },
    {
      "epoch": 0.7728163060926938,
      "grad_norm": 1.2923591136932373,
      "learning_rate": 1.878895383467407e-05,
      "loss": 0.0998,
      "step": 52475
    },
    {
      "epoch": 0.7731844891827808,
      "grad_norm": 1.2481595277786255,
      "learning_rate": 1.8787317462707084e-05,
      "loss": 0.1047,
      "step": 52500
    },
    {
      "epoch": 0.7735526722728678,
      "grad_norm": 1.6275888681411743,
      "learning_rate": 1.87856810907401e-05,
      "loss": 0.095,
      "step": 52525
    },
    {
      "epoch": 0.7739208553629549,
      "grad_norm": 1.199249267578125,
      "learning_rate": 1.8784044718773116e-05,
      "loss": 0.104,
      "step": 52550
    },
    {
      "epoch": 0.774289038453042,
      "grad_norm": 1.2225192785263062,
      "learning_rate": 1.878240834680613e-05,
      "loss": 0.0986,
      "step": 52575
    },
    {
      "epoch": 0.774657221543129,
      "grad_norm": 1.8041224479675293,
      "learning_rate": 1.8780771974839145e-05,
      "loss": 0.0943,
      "step": 52600
    },
    {
      "epoch": 0.775025404633216,
      "grad_norm": 1.5905193090438843,
      "learning_rate": 1.877913560287216e-05,
      "loss": 0.0926,
      "step": 52625
    },
    {
      "epoch": 0.775393587723303,
      "grad_norm": 1.4913384914398193,
      "learning_rate": 1.8777499230905178e-05,
      "loss": 0.1008,
      "step": 52650
    },
    {
      "epoch": 0.77576177081339,
      "grad_norm": 1.650696039199829,
      "learning_rate": 1.8775862858938192e-05,
      "loss": 0.1046,
      "step": 52675
    },
    {
      "epoch": 0.7761299539034772,
      "grad_norm": 1.3778458833694458,
      "learning_rate": 1.8774226486971207e-05,
      "loss": 0.1011,
      "step": 52700
    },
    {
      "epoch": 0.7764981369935642,
      "grad_norm": 1.8180224895477295,
      "learning_rate": 1.8772590115004224e-05,
      "loss": 0.097,
      "step": 52725
    },
    {
      "epoch": 0.7768663200836512,
      "grad_norm": 1.3458949327468872,
      "learning_rate": 1.877095374303724e-05,
      "loss": 0.0948,
      "step": 52750
    },
    {
      "epoch": 0.7772345031737382,
      "grad_norm": 1.6592265367507935,
      "learning_rate": 1.8769317371070253e-05,
      "loss": 0.1001,
      "step": 52775
    },
    {
      "epoch": 0.7776026862638252,
      "grad_norm": 1.6483943462371826,
      "learning_rate": 1.876768099910327e-05,
      "loss": 0.1036,
      "step": 52800
    },
    {
      "epoch": 0.7779708693539124,
      "grad_norm": 1.3661013841629028,
      "learning_rate": 1.8766044627136286e-05,
      "loss": 0.0912,
      "step": 52825
    },
    {
      "epoch": 0.7783390524439994,
      "grad_norm": 1.1678917407989502,
      "learning_rate": 1.87644082551693e-05,
      "loss": 0.0945,
      "step": 52850
    },
    {
      "epoch": 0.7787072355340864,
      "grad_norm": 1.4729899168014526,
      "learning_rate": 1.8762771883202315e-05,
      "loss": 0.107,
      "step": 52875
    },
    {
      "epoch": 0.7790754186241734,
      "grad_norm": 1.327132225036621,
      "learning_rate": 1.8761135511235332e-05,
      "loss": 0.0947,
      "step": 52900
    },
    {
      "epoch": 0.7794436017142604,
      "grad_norm": 1.6538245677947998,
      "learning_rate": 1.8759499139268347e-05,
      "loss": 0.0962,
      "step": 52925
    },
    {
      "epoch": 0.7798117848043475,
      "grad_norm": 1.7244343757629395,
      "learning_rate": 1.875786276730136e-05,
      "loss": 0.0945,
      "step": 52950
    },
    {
      "epoch": 0.7801799678944346,
      "grad_norm": 1.4939213991165161,
      "learning_rate": 1.875622639533438e-05,
      "loss": 0.0957,
      "step": 52975
    },
    {
      "epoch": 0.7805481509845216,
      "grad_norm": 1.5427114963531494,
      "learning_rate": 1.8754590023367394e-05,
      "loss": 0.1036,
      "step": 53000
    },
    {
      "epoch": 0.7809163340746086,
      "grad_norm": 1.3049370050430298,
      "learning_rate": 1.8752953651400408e-05,
      "loss": 0.1012,
      "step": 53025
    },
    {
      "epoch": 0.7812845171646957,
      "grad_norm": 1.4951084852218628,
      "learning_rate": 1.8751317279433423e-05,
      "loss": 0.1063,
      "step": 53050
    },
    {
      "epoch": 0.7816527002547827,
      "grad_norm": 1.6251379251480103,
      "learning_rate": 1.874968090746644e-05,
      "loss": 0.0988,
      "step": 53075
    },
    {
      "epoch": 0.7820208833448697,
      "grad_norm": 1.550650954246521,
      "learning_rate": 1.8748044535499455e-05,
      "loss": 0.0948,
      "step": 53100
    },
    {
      "epoch": 0.7823890664349568,
      "grad_norm": 1.4532803297042847,
      "learning_rate": 1.874640816353247e-05,
      "loss": 0.0898,
      "step": 53125
    },
    {
      "epoch": 0.7827572495250438,
      "grad_norm": 1.425616979598999,
      "learning_rate": 1.8744771791565487e-05,
      "loss": 0.0985,
      "step": 53150
    },
    {
      "epoch": 0.7831254326151309,
      "grad_norm": 1.7752586603164673,
      "learning_rate": 1.87431354195985e-05,
      "loss": 0.1004,
      "step": 53175
    },
    {
      "epoch": 0.7834936157052179,
      "grad_norm": 1.4681975841522217,
      "learning_rate": 1.8741499047631516e-05,
      "loss": 0.0989,
      "step": 53200
    },
    {
      "epoch": 0.7838617987953049,
      "grad_norm": 1.60545814037323,
      "learning_rate": 1.8739862675664534e-05,
      "loss": 0.0974,
      "step": 53225
    },
    {
      "epoch": 0.784229981885392,
      "grad_norm": 1.7402600049972534,
      "learning_rate": 1.873822630369755e-05,
      "loss": 0.0988,
      "step": 53250
    },
    {
      "epoch": 0.784598164975479,
      "grad_norm": 1.5226898193359375,
      "learning_rate": 1.8736589931730563e-05,
      "loss": 0.0927,
      "step": 53275
    },
    {
      "epoch": 0.7849663480655661,
      "grad_norm": 1.093587040901184,
      "learning_rate": 1.8734953559763577e-05,
      "loss": 0.0971,
      "step": 53300
    },
    {
      "epoch": 0.7853345311556531,
      "grad_norm": 1.4167509078979492,
      "learning_rate": 1.8733317187796595e-05,
      "loss": 0.0947,
      "step": 53325
    },
    {
      "epoch": 0.7857027142457401,
      "grad_norm": 1.8050156831741333,
      "learning_rate": 1.873168081582961e-05,
      "loss": 0.0932,
      "step": 53350
    },
    {
      "epoch": 0.7860708973358271,
      "grad_norm": 1.195334792137146,
      "learning_rate": 1.8730044443862624e-05,
      "loss": 0.0969,
      "step": 53375
    },
    {
      "epoch": 0.7864390804259141,
      "grad_norm": 1.2866966724395752,
      "learning_rate": 1.8728408071895642e-05,
      "loss": 0.0972,
      "step": 53400
    },
    {
      "epoch": 0.7868072635160013,
      "grad_norm": 1.5800254344940186,
      "learning_rate": 1.8726771699928656e-05,
      "loss": 0.0934,
      "step": 53425
    },
    {
      "epoch": 0.7871754466060883,
      "grad_norm": 0.9870246052742004,
      "learning_rate": 1.872513532796167e-05,
      "loss": 0.0941,
      "step": 53450
    },
    {
      "epoch": 0.7875436296961753,
      "grad_norm": 1.4179213047027588,
      "learning_rate": 1.8723498955994685e-05,
      "loss": 0.0998,
      "step": 53475
    },
    {
      "epoch": 0.7879118127862623,
      "grad_norm": 1.460280179977417,
      "learning_rate": 1.87218625840277e-05,
      "loss": 0.0978,
      "step": 53500
    },
    {
      "epoch": 0.7882799958763493,
      "grad_norm": 1.501106858253479,
      "learning_rate": 1.8720226212060718e-05,
      "loss": 0.0951,
      "step": 53525
    },
    {
      "epoch": 0.7886481789664365,
      "grad_norm": 1.4157145023345947,
      "learning_rate": 1.8718589840093732e-05,
      "loss": 0.0982,
      "step": 53550
    },
    {
      "epoch": 0.7890163620565235,
      "grad_norm": 1.8089770078659058,
      "learning_rate": 1.871695346812675e-05,
      "loss": 0.0971,
      "step": 53575
    },
    {
      "epoch": 0.7893845451466105,
      "grad_norm": 1.740575909614563,
      "learning_rate": 1.8715317096159764e-05,
      "loss": 0.0985,
      "step": 53600
    },
    {
      "epoch": 0.7897527282366975,
      "grad_norm": 1.326170802116394,
      "learning_rate": 1.871368072419278e-05,
      "loss": 0.1007,
      "step": 53625
    },
    {
      "epoch": 0.7901209113267846,
      "grad_norm": 1.310464859008789,
      "learning_rate": 1.8712044352225797e-05,
      "loss": 0.0967,
      "step": 53650
    },
    {
      "epoch": 0.7904890944168717,
      "grad_norm": 1.653826355934143,
      "learning_rate": 1.871040798025881e-05,
      "loss": 0.0923,
      "step": 53675
    },
    {
      "epoch": 0.7908572775069587,
      "grad_norm": 1.535834789276123,
      "learning_rate": 1.8708771608291826e-05,
      "loss": 0.0947,
      "step": 53700
    },
    {
      "epoch": 0.7912254605970457,
      "grad_norm": 1.4991453886032104,
      "learning_rate": 1.870713523632484e-05,
      "loss": 0.0947,
      "step": 53725
    },
    {
      "epoch": 0.7915936436871327,
      "grad_norm": 1.3092700242996216,
      "learning_rate": 1.8705498864357854e-05,
      "loss": 0.1005,
      "step": 53750
    },
    {
      "epoch": 0.7919618267772198,
      "grad_norm": 1.345397710800171,
      "learning_rate": 1.8703862492390872e-05,
      "loss": 0.0856,
      "step": 53775
    },
    {
      "epoch": 0.7923300098673068,
      "grad_norm": 1.520815372467041,
      "learning_rate": 1.8702226120423887e-05,
      "loss": 0.0926,
      "step": 53800
    },
    {
      "epoch": 0.7926981929573939,
      "grad_norm": 1.2822909355163574,
      "learning_rate": 1.8700589748456905e-05,
      "loss": 0.0926,
      "step": 53825
    },
    {
      "epoch": 0.7930663760474809,
      "grad_norm": 1.1882009506225586,
      "learning_rate": 1.869895337648992e-05,
      "loss": 0.0886,
      "step": 53850
    },
    {
      "epoch": 0.7934345591375679,
      "grad_norm": 1.1720936298370361,
      "learning_rate": 1.8697317004522934e-05,
      "loss": 0.0972,
      "step": 53875
    },
    {
      "epoch": 0.793802742227655,
      "grad_norm": 1.4800114631652832,
      "learning_rate": 1.8695680632555948e-05,
      "loss": 0.1012,
      "step": 53900
    },
    {
      "epoch": 0.794170925317742,
      "grad_norm": 0.954601526260376,
      "learning_rate": 1.8694044260588962e-05,
      "loss": 0.0892,
      "step": 53925
    },
    {
      "epoch": 0.794539108407829,
      "grad_norm": 1.4864304065704346,
      "learning_rate": 1.869240788862198e-05,
      "loss": 0.1083,
      "step": 53950
    },
    {
      "epoch": 0.7949072914979161,
      "grad_norm": 2.026531219482422,
      "learning_rate": 1.8690836971533674e-05,
      "loss": 0.0989,
      "step": 53975
    },
    {
      "epoch": 0.7952754745880031,
      "grad_norm": 1.781833291053772,
      "learning_rate": 1.868920059956669e-05,
      "loss": 0.0949,
      "step": 54000
    },
    {
      "epoch": 0.7956436576780902,
      "grad_norm": 1.2690705060958862,
      "learning_rate": 1.8687564227599706e-05,
      "loss": 0.0969,
      "step": 54025
    },
    {
      "epoch": 0.7960118407681772,
      "grad_norm": 1.2814421653747559,
      "learning_rate": 1.868592785563272e-05,
      "loss": 0.096,
      "step": 54050
    },
    {
      "epoch": 0.7963800238582642,
      "grad_norm": 1.3262174129486084,
      "learning_rate": 1.868429148366574e-05,
      "loss": 0.0981,
      "step": 54075
    },
    {
      "epoch": 0.7967482069483512,
      "grad_norm": 1.3594454526901245,
      "learning_rate": 1.868265511169875e-05,
      "loss": 0.0977,
      "step": 54100
    },
    {
      "epoch": 0.7971163900384383,
      "grad_norm": 1.4144097566604614,
      "learning_rate": 1.8681018739731768e-05,
      "loss": 0.0927,
      "step": 54125
    },
    {
      "epoch": 0.7974845731285254,
      "grad_norm": 1.2568429708480835,
      "learning_rate": 1.8679382367764782e-05,
      "loss": 0.0937,
      "step": 54150
    },
    {
      "epoch": 0.7978527562186124,
      "grad_norm": 1.5641965866088867,
      "learning_rate": 1.8677745995797797e-05,
      "loss": 0.0999,
      "step": 54175
    },
    {
      "epoch": 0.7982209393086994,
      "grad_norm": 1.2955148220062256,
      "learning_rate": 1.8676109623830814e-05,
      "loss": 0.0846,
      "step": 54200
    },
    {
      "epoch": 0.7985891223987864,
      "grad_norm": 1.4151428937911987,
      "learning_rate": 1.867447325186383e-05,
      "loss": 0.089,
      "step": 54225
    },
    {
      "epoch": 0.7989573054888736,
      "grad_norm": 1.3893907070159912,
      "learning_rate": 1.8672836879896847e-05,
      "loss": 0.0923,
      "step": 54250
    },
    {
      "epoch": 0.7993254885789606,
      "grad_norm": 1.4878407716751099,
      "learning_rate": 1.867120050792986e-05,
      "loss": 0.0998,
      "step": 54275
    },
    {
      "epoch": 0.7996936716690476,
      "grad_norm": 1.5924643278121948,
      "learning_rate": 1.8669564135962876e-05,
      "loss": 0.0942,
      "step": 54300
    },
    {
      "epoch": 0.8000618547591346,
      "grad_norm": 1.259968876838684,
      "learning_rate": 1.866792776399589e-05,
      "loss": 0.0956,
      "step": 54325
    },
    {
      "epoch": 0.8004300378492216,
      "grad_norm": 1.595425009727478,
      "learning_rate": 1.8666291392028905e-05,
      "loss": 0.1024,
      "step": 54350
    },
    {
      "epoch": 0.8007982209393087,
      "grad_norm": 1.4488376379013062,
      "learning_rate": 1.8664655020061922e-05,
      "loss": 0.0957,
      "step": 54375
    },
    {
      "epoch": 0.8011664040293958,
      "grad_norm": 1.3736494779586792,
      "learning_rate": 1.8663018648094937e-05,
      "loss": 0.0925,
      "step": 54400
    },
    {
      "epoch": 0.8015345871194828,
      "grad_norm": 1.542526364326477,
      "learning_rate": 1.866138227612795e-05,
      "loss": 0.0972,
      "step": 54425
    },
    {
      "epoch": 0.8019027702095698,
      "grad_norm": 1.651017427444458,
      "learning_rate": 1.865974590416097e-05,
      "loss": 0.0982,
      "step": 54450
    },
    {
      "epoch": 0.8022709532996568,
      "grad_norm": 1.4605393409729004,
      "learning_rate": 1.8658109532193984e-05,
      "loss": 0.0995,
      "step": 54475
    },
    {
      "epoch": 0.8026391363897439,
      "grad_norm": 1.0872023105621338,
      "learning_rate": 1.8656473160227e-05,
      "loss": 0.0966,
      "step": 54500
    },
    {
      "epoch": 0.803007319479831,
      "grad_norm": 1.2862284183502197,
      "learning_rate": 1.8654836788260013e-05,
      "loss": 0.0917,
      "step": 54525
    },
    {
      "epoch": 0.803375502569918,
      "grad_norm": 1.5432548522949219,
      "learning_rate": 1.865320041629303e-05,
      "loss": 0.0954,
      "step": 54550
    },
    {
      "epoch": 0.803743685660005,
      "grad_norm": 1.3444406986236572,
      "learning_rate": 1.8651564044326045e-05,
      "loss": 0.0967,
      "step": 54575
    },
    {
      "epoch": 0.804111868750092,
      "grad_norm": 1.7558733224868774,
      "learning_rate": 1.864992767235906e-05,
      "loss": 0.1004,
      "step": 54600
    },
    {
      "epoch": 0.8044800518401791,
      "grad_norm": 1.575652837753296,
      "learning_rate": 1.8648291300392077e-05,
      "loss": 0.0996,
      "step": 54625
    },
    {
      "epoch": 0.8048482349302661,
      "grad_norm": 1.2382612228393555,
      "learning_rate": 1.864665492842509e-05,
      "loss": 0.0857,
      "step": 54650
    },
    {
      "epoch": 0.8052164180203532,
      "grad_norm": 1.3315802812576294,
      "learning_rate": 1.8645018556458106e-05,
      "loss": 0.0888,
      "step": 54675
    },
    {
      "epoch": 0.8055846011104402,
      "grad_norm": 2.0507583618164062,
      "learning_rate": 1.8643382184491124e-05,
      "loss": 0.0951,
      "step": 54700
    },
    {
      "epoch": 0.8059527842005272,
      "grad_norm": 1.2660714387893677,
      "learning_rate": 1.864174581252414e-05,
      "loss": 0.102,
      "step": 54725
    },
    {
      "epoch": 0.8063209672906143,
      "grad_norm": 1.1463576555252075,
      "learning_rate": 1.8640109440557153e-05,
      "loss": 0.0911,
      "step": 54750
    },
    {
      "epoch": 0.8066891503807013,
      "grad_norm": 1.4842114448547363,
      "learning_rate": 1.8638473068590167e-05,
      "loss": 0.0954,
      "step": 54775
    },
    {
      "epoch": 0.8070573334707883,
      "grad_norm": 1.8125190734863281,
      "learning_rate": 1.8636836696623185e-05,
      "loss": 0.092,
      "step": 54800
    },
    {
      "epoch": 0.8074255165608754,
      "grad_norm": 1.4588818550109863,
      "learning_rate": 1.86352003246562e-05,
      "loss": 0.0943,
      "step": 54825
    },
    {
      "epoch": 0.8077936996509625,
      "grad_norm": 1.6068251132965088,
      "learning_rate": 1.8633563952689214e-05,
      "loss": 0.0974,
      "step": 54850
    },
    {
      "epoch": 0.8081618827410495,
      "grad_norm": 1.197568416595459,
      "learning_rate": 1.8631927580722232e-05,
      "loss": 0.0875,
      "step": 54875
    },
    {
      "epoch": 0.8085300658311365,
      "grad_norm": 1.2989412546157837,
      "learning_rate": 1.8630291208755246e-05,
      "loss": 0.1026,
      "step": 54900
    },
    {
      "epoch": 0.8088982489212235,
      "grad_norm": 1.7455549240112305,
      "learning_rate": 1.862865483678826e-05,
      "loss": 0.0931,
      "step": 54925
    },
    {
      "epoch": 0.8092664320113105,
      "grad_norm": 1.8075650930404663,
      "learning_rate": 1.8627018464821275e-05,
      "loss": 0.0916,
      "step": 54950
    },
    {
      "epoch": 0.8096346151013977,
      "grad_norm": 1.4007229804992676,
      "learning_rate": 1.8625382092854293e-05,
      "loss": 0.1013,
      "step": 54975
    },
    {
      "epoch": 0.8100027981914847,
      "grad_norm": 1.1319169998168945,
      "learning_rate": 1.8623745720887308e-05,
      "loss": 0.0948,
      "step": 55000
    },
    {
      "epoch": 0.8103709812815717,
      "grad_norm": 1.3432068824768066,
      "learning_rate": 1.8622109348920322e-05,
      "loss": 0.0953,
      "step": 55025
    },
    {
      "epoch": 0.8107391643716587,
      "grad_norm": 1.4980003833770752,
      "learning_rate": 1.862047297695334e-05,
      "loss": 0.0942,
      "step": 55050
    },
    {
      "epoch": 0.8111073474617457,
      "grad_norm": 1.552869200706482,
      "learning_rate": 1.8618836604986354e-05,
      "loss": 0.096,
      "step": 55075
    },
    {
      "epoch": 0.8114755305518329,
      "grad_norm": 1.2944960594177246,
      "learning_rate": 1.861720023301937e-05,
      "loss": 0.0908,
      "step": 55100
    },
    {
      "epoch": 0.8118437136419199,
      "grad_norm": 1.6127983331680298,
      "learning_rate": 1.8615563861052387e-05,
      "loss": 0.0931,
      "step": 55125
    },
    {
      "epoch": 0.8122118967320069,
      "grad_norm": 2.118488073348999,
      "learning_rate": 1.86139274890854e-05,
      "loss": 0.0996,
      "step": 55150
    },
    {
      "epoch": 0.8125800798220939,
      "grad_norm": 1.4998732805252075,
      "learning_rate": 1.8612291117118416e-05,
      "loss": 0.0971,
      "step": 55175
    },
    {
      "epoch": 0.8129482629121809,
      "grad_norm": 1.4116191864013672,
      "learning_rate": 1.861065474515143e-05,
      "loss": 0.0931,
      "step": 55200
    },
    {
      "epoch": 0.813316446002268,
      "grad_norm": 0.9913114309310913,
      "learning_rate": 1.8609018373184448e-05,
      "loss": 0.0902,
      "step": 55225
    },
    {
      "epoch": 0.8136846290923551,
      "grad_norm": 1.3107503652572632,
      "learning_rate": 1.8607382001217462e-05,
      "loss": 0.0885,
      "step": 55250
    },
    {
      "epoch": 0.8140528121824421,
      "grad_norm": 1.0871694087982178,
      "learning_rate": 1.8605745629250477e-05,
      "loss": 0.0936,
      "step": 55275
    },
    {
      "epoch": 0.8144209952725291,
      "grad_norm": 1.364189863204956,
      "learning_rate": 1.8604109257283495e-05,
      "loss": 0.0962,
      "step": 55300
    },
    {
      "epoch": 0.8147891783626161,
      "grad_norm": 1.3079403638839722,
      "learning_rate": 1.860247288531651e-05,
      "loss": 0.0914,
      "step": 55325
    },
    {
      "epoch": 0.8151573614527032,
      "grad_norm": 1.3661223649978638,
      "learning_rate": 1.8600836513349524e-05,
      "loss": 0.0981,
      "step": 55350
    },
    {
      "epoch": 0.8155255445427902,
      "grad_norm": 1.3914251327514648,
      "learning_rate": 1.8599200141382538e-05,
      "loss": 0.0936,
      "step": 55375
    },
    {
      "epoch": 0.8158937276328773,
      "grad_norm": 1.5198974609375,
      "learning_rate": 1.8597563769415556e-05,
      "loss": 0.0973,
      "step": 55400
    },
    {
      "epoch": 0.8162619107229643,
      "grad_norm": 1.5449283123016357,
      "learning_rate": 1.859592739744857e-05,
      "loss": 0.0994,
      "step": 55425
    },
    {
      "epoch": 0.8166300938130514,
      "grad_norm": 1.245825171470642,
      "learning_rate": 1.8594291025481585e-05,
      "loss": 0.0895,
      "step": 55450
    },
    {
      "epoch": 0.8169982769031384,
      "grad_norm": 1.4883716106414795,
      "learning_rate": 1.8592654653514603e-05,
      "loss": 0.09,
      "step": 55475
    },
    {
      "epoch": 0.8173664599932254,
      "grad_norm": 1.524454116821289,
      "learning_rate": 1.8591018281547617e-05,
      "loss": 0.1025,
      "step": 55500
    },
    {
      "epoch": 0.8177346430833125,
      "grad_norm": 1.436755657196045,
      "learning_rate": 1.858938190958063e-05,
      "loss": 0.0985,
      "step": 55525
    },
    {
      "epoch": 0.8181028261733995,
      "grad_norm": 1.3007491827011108,
      "learning_rate": 1.858774553761365e-05,
      "loss": 0.1037,
      "step": 55550
    },
    {
      "epoch": 0.8184710092634866,
      "grad_norm": 1.233238935470581,
      "learning_rate": 1.858610916564666e-05,
      "loss": 0.0873,
      "step": 55575
    },
    {
      "epoch": 0.8188391923535736,
      "grad_norm": 1.7281906604766846,
      "learning_rate": 1.8584472793679678e-05,
      "loss": 0.0932,
      "step": 55600
    },
    {
      "epoch": 0.8192073754436606,
      "grad_norm": 1.6181998252868652,
      "learning_rate": 1.8582836421712693e-05,
      "loss": 0.0944,
      "step": 55625
    },
    {
      "epoch": 0.8195755585337476,
      "grad_norm": 1.7249391078948975,
      "learning_rate": 1.858120004974571e-05,
      "loss": 0.0988,
      "step": 55650
    },
    {
      "epoch": 0.8199437416238347,
      "grad_norm": 1.0093046426773071,
      "learning_rate": 1.8579563677778725e-05,
      "loss": 0.0903,
      "step": 55675
    },
    {
      "epoch": 0.8203119247139218,
      "grad_norm": 1.7590138912200928,
      "learning_rate": 1.857792730581174e-05,
      "loss": 0.0921,
      "step": 55700
    },
    {
      "epoch": 0.8206801078040088,
      "grad_norm": 1.2769867181777954,
      "learning_rate": 1.8576290933844757e-05,
      "loss": 0.0949,
      "step": 55725
    },
    {
      "epoch": 0.8210482908940958,
      "grad_norm": 1.095051884651184,
      "learning_rate": 1.8574654561877772e-05,
      "loss": 0.0954,
      "step": 55750
    },
    {
      "epoch": 0.8214164739841828,
      "grad_norm": 0.9720871448516846,
      "learning_rate": 1.8573018189910786e-05,
      "loss": 0.0897,
      "step": 55775
    },
    {
      "epoch": 0.8217846570742698,
      "grad_norm": 1.9454920291900635,
      "learning_rate": 1.85713818179438e-05,
      "loss": 0.0949,
      "step": 55800
    },
    {
      "epoch": 0.822152840164357,
      "grad_norm": 1.1506351232528687,
      "learning_rate": 1.8569745445976815e-05,
      "loss": 0.0969,
      "step": 55825
    },
    {
      "epoch": 0.822521023254444,
      "grad_norm": 1.3336644172668457,
      "learning_rate": 1.8568109074009833e-05,
      "loss": 0.0965,
      "step": 55850
    },
    {
      "epoch": 0.822889206344531,
      "grad_norm": 1.7930415868759155,
      "learning_rate": 1.8566472702042847e-05,
      "loss": 0.0942,
      "step": 55875
    },
    {
      "epoch": 0.823257389434618,
      "grad_norm": 1.3584586381912231,
      "learning_rate": 1.8564836330075865e-05,
      "loss": 0.0974,
      "step": 55900
    },
    {
      "epoch": 0.8236255725247051,
      "grad_norm": 1.6696362495422363,
      "learning_rate": 1.856319995810888e-05,
      "loss": 0.0976,
      "step": 55925
    },
    {
      "epoch": 0.8239937556147922,
      "grad_norm": 1.704949140548706,
      "learning_rate": 1.8561563586141894e-05,
      "loss": 0.0821,
      "step": 55950
    },
    {
      "epoch": 0.8243619387048792,
      "grad_norm": 1.5611543655395508,
      "learning_rate": 1.8559927214174912e-05,
      "loss": 0.0874,
      "step": 55975
    },
    {
      "epoch": 0.8247301217949662,
      "grad_norm": 1.1352851390838623,
      "learning_rate": 1.8558290842207923e-05,
      "loss": 0.096,
      "step": 56000
    },
    {
      "epoch": 0.8250983048850532,
      "grad_norm": 1.6469817161560059,
      "learning_rate": 1.855665447024094e-05,
      "loss": 0.0915,
      "step": 56025
    },
    {
      "epoch": 0.8254664879751403,
      "grad_norm": 1.9793018102645874,
      "learning_rate": 1.8555018098273955e-05,
      "loss": 0.0953,
      "step": 56050
    },
    {
      "epoch": 0.8258346710652273,
      "grad_norm": 1.0863748788833618,
      "learning_rate": 1.855344718118565e-05,
      "loss": 0.0988,
      "step": 56075
    },
    {
      "epoch": 0.8262028541553144,
      "grad_norm": 1.2504758834838867,
      "learning_rate": 1.8551810809218667e-05,
      "loss": 0.0904,
      "step": 56100
    },
    {
      "epoch": 0.8265710372454014,
      "grad_norm": 1.7556089162826538,
      "learning_rate": 1.855017443725168e-05,
      "loss": 0.1011,
      "step": 56125
    },
    {
      "epoch": 0.8269392203354884,
      "grad_norm": 1.510802984237671,
      "learning_rate": 1.85485380652847e-05,
      "loss": 0.0947,
      "step": 56150
    },
    {
      "epoch": 0.8273074034255755,
      "grad_norm": 1.0437312126159668,
      "learning_rate": 1.8546901693317714e-05,
      "loss": 0.0912,
      "step": 56175
    },
    {
      "epoch": 0.8276755865156625,
      "grad_norm": 1.0683832168579102,
      "learning_rate": 1.854526532135073e-05,
      "loss": 0.091,
      "step": 56200
    },
    {
      "epoch": 0.8280437696057495,
      "grad_norm": 1.1077921390533447,
      "learning_rate": 1.8543628949383743e-05,
      "loss": 0.0922,
      "step": 56225
    },
    {
      "epoch": 0.8284119526958366,
      "grad_norm": 1.2730482816696167,
      "learning_rate": 1.8541992577416757e-05,
      "loss": 0.0969,
      "step": 56250
    },
    {
      "epoch": 0.8287801357859236,
      "grad_norm": 1.4467557668685913,
      "learning_rate": 1.8540356205449775e-05,
      "loss": 0.0995,
      "step": 56275
    },
    {
      "epoch": 0.8291483188760107,
      "grad_norm": 1.4756429195404053,
      "learning_rate": 1.853871983348279e-05,
      "loss": 0.093,
      "step": 56300
    },
    {
      "epoch": 0.8295165019660977,
      "grad_norm": 1.5162990093231201,
      "learning_rate": 1.8537083461515804e-05,
      "loss": 0.0921,
      "step": 56325
    },
    {
      "epoch": 0.8298846850561847,
      "grad_norm": 1.324678897857666,
      "learning_rate": 1.8535447089548822e-05,
      "loss": 0.095,
      "step": 56350
    },
    {
      "epoch": 0.8302528681462717,
      "grad_norm": 1.3935545682907104,
      "learning_rate": 1.8533810717581836e-05,
      "loss": 0.0941,
      "step": 56375
    },
    {
      "epoch": 0.8306210512363588,
      "grad_norm": 1.6868776082992554,
      "learning_rate": 1.8532174345614854e-05,
      "loss": 0.0943,
      "step": 56400
    },
    {
      "epoch": 0.8309892343264459,
      "grad_norm": 1.6698260307312012,
      "learning_rate": 1.8530537973647865e-05,
      "loss": 0.0816,
      "step": 56425
    },
    {
      "epoch": 0.8313574174165329,
      "grad_norm": 1.8940362930297852,
      "learning_rate": 1.8528901601680883e-05,
      "loss": 0.093,
      "step": 56450
    },
    {
      "epoch": 0.8317256005066199,
      "grad_norm": 1.2623918056488037,
      "learning_rate": 1.8527265229713898e-05,
      "loss": 0.0872,
      "step": 56475
    },
    {
      "epoch": 0.8320937835967069,
      "grad_norm": 1.6204856634140015,
      "learning_rate": 1.8525628857746912e-05,
      "loss": 0.0943,
      "step": 56500
    },
    {
      "epoch": 0.8324619666867941,
      "grad_norm": 1.3996667861938477,
      "learning_rate": 1.852399248577993e-05,
      "loss": 0.0973,
      "step": 56525
    },
    {
      "epoch": 0.8328301497768811,
      "grad_norm": 1.9733930826187134,
      "learning_rate": 1.8522356113812944e-05,
      "loss": 0.0949,
      "step": 56550
    },
    {
      "epoch": 0.8331983328669681,
      "grad_norm": 1.4230526685714722,
      "learning_rate": 1.8520719741845962e-05,
      "loss": 0.1038,
      "step": 56575
    },
    {
      "epoch": 0.8335665159570551,
      "grad_norm": 1.498453974723816,
      "learning_rate": 1.8519083369878977e-05,
      "loss": 0.1003,
      "step": 56600
    },
    {
      "epoch": 0.8339346990471421,
      "grad_norm": 1.9248186349868774,
      "learning_rate": 1.851744699791199e-05,
      "loss": 0.0961,
      "step": 56625
    },
    {
      "epoch": 0.8343028821372293,
      "grad_norm": 1.9036245346069336,
      "learning_rate": 1.8515810625945006e-05,
      "loss": 0.1036,
      "step": 56650
    },
    {
      "epoch": 0.8346710652273163,
      "grad_norm": 1.4368208646774292,
      "learning_rate": 1.851417425397802e-05,
      "loss": 0.0963,
      "step": 56675
    },
    {
      "epoch": 0.8350392483174033,
      "grad_norm": 1.6156413555145264,
      "learning_rate": 1.8512537882011038e-05,
      "loss": 0.1007,
      "step": 56700
    },
    {
      "epoch": 0.8354074314074903,
      "grad_norm": 1.900478482246399,
      "learning_rate": 1.8510901510044052e-05,
      "loss": 0.0959,
      "step": 56725
    },
    {
      "epoch": 0.8357756144975773,
      "grad_norm": 1.5122090578079224,
      "learning_rate": 1.8509265138077067e-05,
      "loss": 0.0962,
      "step": 56750
    },
    {
      "epoch": 0.8361437975876644,
      "grad_norm": 1.2271615266799927,
      "learning_rate": 1.8507628766110085e-05,
      "loss": 0.0954,
      "step": 56775
    },
    {
      "epoch": 0.8365119806777515,
      "grad_norm": 1.4659217596054077,
      "learning_rate": 1.85059923941431e-05,
      "loss": 0.0922,
      "step": 56800
    },
    {
      "epoch": 0.8368801637678385,
      "grad_norm": 1.7371073961257935,
      "learning_rate": 1.8504356022176117e-05,
      "loss": 0.0893,
      "step": 56825
    },
    {
      "epoch": 0.8372483468579255,
      "grad_norm": 1.746718168258667,
      "learning_rate": 1.8502719650209128e-05,
      "loss": 0.0905,
      "step": 56850
    },
    {
      "epoch": 0.8376165299480125,
      "grad_norm": 1.482497215270996,
      "learning_rate": 1.8501083278242146e-05,
      "loss": 0.0869,
      "step": 56875
    },
    {
      "epoch": 0.8379847130380996,
      "grad_norm": 1.482007622718811,
      "learning_rate": 1.849944690627516e-05,
      "loss": 0.1014,
      "step": 56900
    },
    {
      "epoch": 0.8383528961281866,
      "grad_norm": 1.6660585403442383,
      "learning_rate": 1.8497810534308175e-05,
      "loss": 0.0959,
      "step": 56925
    },
    {
      "epoch": 0.8387210792182737,
      "grad_norm": 2.038933753967285,
      "learning_rate": 1.8496174162341193e-05,
      "loss": 0.0969,
      "step": 56950
    },
    {
      "epoch": 0.8390892623083607,
      "grad_norm": 1.5888947248458862,
      "learning_rate": 1.8494537790374207e-05,
      "loss": 0.0857,
      "step": 56975
    },
    {
      "epoch": 0.8394574453984477,
      "grad_norm": 1.071915864944458,
      "learning_rate": 1.849290141840722e-05,
      "loss": 0.1016,
      "step": 57000
    },
    {
      "epoch": 0.8398256284885348,
      "grad_norm": 1.7160131931304932,
      "learning_rate": 1.849126504644024e-05,
      "loss": 0.0933,
      "step": 57025
    },
    {
      "epoch": 0.8401938115786218,
      "grad_norm": 1.2102928161621094,
      "learning_rate": 1.8489628674473254e-05,
      "loss": 0.0905,
      "step": 57050
    },
    {
      "epoch": 0.8405619946687088,
      "grad_norm": 1.4412623643875122,
      "learning_rate": 1.8487992302506268e-05,
      "loss": 0.0948,
      "step": 57075
    },
    {
      "epoch": 0.8409301777587959,
      "grad_norm": 1.5016887187957764,
      "learning_rate": 1.8486355930539283e-05,
      "loss": 0.0866,
      "step": 57100
    },
    {
      "epoch": 0.841298360848883,
      "grad_norm": 1.4665857553482056,
      "learning_rate": 1.84847195585723e-05,
      "loss": 0.091,
      "step": 57125
    },
    {
      "epoch": 0.84166654393897,
      "grad_norm": 1.3634446859359741,
      "learning_rate": 1.8483083186605315e-05,
      "loss": 0.093,
      "step": 57150
    },
    {
      "epoch": 0.842034727029057,
      "grad_norm": 1.4443259239196777,
      "learning_rate": 1.848144681463833e-05,
      "loss": 0.092,
      "step": 57175
    },
    {
      "epoch": 0.842402910119144,
      "grad_norm": 1.4307743310928345,
      "learning_rate": 1.8479810442671347e-05,
      "loss": 0.0945,
      "step": 57200
    },
    {
      "epoch": 0.842771093209231,
      "grad_norm": 1.4008501768112183,
      "learning_rate": 1.8478174070704362e-05,
      "loss": 0.093,
      "step": 57225
    },
    {
      "epoch": 0.8431392762993182,
      "grad_norm": 1.025970697402954,
      "learning_rate": 1.8476537698737376e-05,
      "loss": 0.0928,
      "step": 57250
    },
    {
      "epoch": 0.8435074593894052,
      "grad_norm": 1.709593653678894,
      "learning_rate": 1.847490132677039e-05,
      "loss": 0.0987,
      "step": 57275
    },
    {
      "epoch": 0.8438756424794922,
      "grad_norm": 1.183369755744934,
      "learning_rate": 1.847326495480341e-05,
      "loss": 0.0893,
      "step": 57300
    },
    {
      "epoch": 0.8442438255695792,
      "grad_norm": 1.7387748956680298,
      "learning_rate": 1.8471628582836423e-05,
      "loss": 0.0922,
      "step": 57325
    },
    {
      "epoch": 0.8446120086596662,
      "grad_norm": 1.3709940910339355,
      "learning_rate": 1.8469992210869437e-05,
      "loss": 0.0957,
      "step": 57350
    },
    {
      "epoch": 0.8449801917497534,
      "grad_norm": 1.73492431640625,
      "learning_rate": 1.8468355838902455e-05,
      "loss": 0.0928,
      "step": 57375
    },
    {
      "epoch": 0.8453483748398404,
      "grad_norm": 1.300275206565857,
      "learning_rate": 1.846671946693547e-05,
      "loss": 0.095,
      "step": 57400
    },
    {
      "epoch": 0.8457165579299274,
      "grad_norm": 1.6966084241867065,
      "learning_rate": 1.8465083094968484e-05,
      "loss": 0.0977,
      "step": 57425
    },
    {
      "epoch": 0.8460847410200144,
      "grad_norm": 1.2545547485351562,
      "learning_rate": 1.8463446723001502e-05,
      "loss": 0.1018,
      "step": 57450
    },
    {
      "epoch": 0.8464529241101014,
      "grad_norm": 1.1827770471572876,
      "learning_rate": 1.8461810351034513e-05,
      "loss": 0.086,
      "step": 57475
    },
    {
      "epoch": 0.8468211072001885,
      "grad_norm": 1.3192706108093262,
      "learning_rate": 1.846017397906753e-05,
      "loss": 0.095,
      "step": 57500
    },
    {
      "epoch": 0.8471892902902756,
      "grad_norm": 1.3012096881866455,
      "learning_rate": 1.8458537607100545e-05,
      "loss": 0.0864,
      "step": 57525
    },
    {
      "epoch": 0.8475574733803626,
      "grad_norm": 1.1796921491622925,
      "learning_rate": 1.8456901235133563e-05,
      "loss": 0.0837,
      "step": 57550
    },
    {
      "epoch": 0.8479256564704496,
      "grad_norm": 1.5892575979232788,
      "learning_rate": 1.8455264863166578e-05,
      "loss": 0.0837,
      "step": 57575
    },
    {
      "epoch": 0.8482938395605366,
      "grad_norm": 1.2497235536575317,
      "learning_rate": 1.8453628491199592e-05,
      "loss": 0.0981,
      "step": 57600
    },
    {
      "epoch": 0.8486620226506237,
      "grad_norm": 1.4007341861724854,
      "learning_rate": 1.845199211923261e-05,
      "loss": 0.0823,
      "step": 57625
    },
    {
      "epoch": 0.8490302057407108,
      "grad_norm": 1.7249845266342163,
      "learning_rate": 1.8450355747265625e-05,
      "loss": 0.0919,
      "step": 57650
    },
    {
      "epoch": 0.8493983888307978,
      "grad_norm": 2.2453880310058594,
      "learning_rate": 1.844871937529864e-05,
      "loss": 0.0919,
      "step": 57675
    },
    {
      "epoch": 0.8497665719208848,
      "grad_norm": 1.5137193202972412,
      "learning_rate": 1.8447083003331653e-05,
      "loss": 0.0943,
      "step": 57700
    },
    {
      "epoch": 0.8501347550109719,
      "grad_norm": 1.403775691986084,
      "learning_rate": 1.844544663136467e-05,
      "loss": 0.0984,
      "step": 57725
    },
    {
      "epoch": 0.8505029381010589,
      "grad_norm": 2.0289063453674316,
      "learning_rate": 1.8443810259397686e-05,
      "loss": 0.0935,
      "step": 57750
    },
    {
      "epoch": 0.8508711211911459,
      "grad_norm": 0.902027428150177,
      "learning_rate": 1.84421738874307e-05,
      "loss": 0.089,
      "step": 57775
    },
    {
      "epoch": 0.851239304281233,
      "grad_norm": 1.2147347927093506,
      "learning_rate": 1.8440537515463718e-05,
      "loss": 0.0808,
      "step": 57800
    },
    {
      "epoch": 0.85160748737132,
      "grad_norm": 1.3671728372573853,
      "learning_rate": 1.8438901143496733e-05,
      "loss": 0.0875,
      "step": 57825
    },
    {
      "epoch": 0.8519756704614071,
      "grad_norm": 1.4228578805923462,
      "learning_rate": 1.8437264771529747e-05,
      "loss": 0.0954,
      "step": 57850
    },
    {
      "epoch": 0.8523438535514941,
      "grad_norm": 1.6377575397491455,
      "learning_rate": 1.8435628399562765e-05,
      "loss": 0.1018,
      "step": 57875
    },
    {
      "epoch": 0.8527120366415811,
      "grad_norm": 1.4888041019439697,
      "learning_rate": 1.8433992027595776e-05,
      "loss": 0.0933,
      "step": 57900
    },
    {
      "epoch": 0.8530802197316681,
      "grad_norm": 1.489107370376587,
      "learning_rate": 1.8432355655628794e-05,
      "loss": 0.0994,
      "step": 57925
    },
    {
      "epoch": 0.8534484028217552,
      "grad_norm": 1.6297394037246704,
      "learning_rate": 1.8430719283661808e-05,
      "loss": 0.0839,
      "step": 57950
    },
    {
      "epoch": 0.8538165859118423,
      "grad_norm": 1.50656259059906,
      "learning_rate": 1.8429082911694826e-05,
      "loss": 0.096,
      "step": 57975
    },
    {
      "epoch": 0.8541847690019293,
      "grad_norm": 1.3311855792999268,
      "learning_rate": 1.842744653972784e-05,
      "loss": 0.0905,
      "step": 58000
    },
    {
      "epoch": 0.8545529520920163,
      "grad_norm": 1.7467899322509766,
      "learning_rate": 1.8425810167760855e-05,
      "loss": 0.0911,
      "step": 58025
    },
    {
      "epoch": 0.8549211351821033,
      "grad_norm": 1.619917869567871,
      "learning_rate": 1.8424173795793873e-05,
      "loss": 0.101,
      "step": 58050
    },
    {
      "epoch": 0.8552893182721903,
      "grad_norm": 1.5799951553344727,
      "learning_rate": 1.8422537423826887e-05,
      "loss": 0.0836,
      "step": 58075
    },
    {
      "epoch": 0.8556575013622775,
      "grad_norm": 1.628247857093811,
      "learning_rate": 1.8420901051859902e-05,
      "loss": 0.0912,
      "step": 58100
    },
    {
      "epoch": 0.8560256844523645,
      "grad_norm": 1.4958336353302002,
      "learning_rate": 1.8419264679892916e-05,
      "loss": 0.0998,
      "step": 58125
    },
    {
      "epoch": 0.8563938675424515,
      "grad_norm": 1.626678466796875,
      "learning_rate": 1.841762830792593e-05,
      "loss": 0.0905,
      "step": 58150
    },
    {
      "epoch": 0.8567620506325385,
      "grad_norm": 1.1951794624328613,
      "learning_rate": 1.841599193595895e-05,
      "loss": 0.0943,
      "step": 58175
    },
    {
      "epoch": 0.8571302337226255,
      "grad_norm": 1.804368019104004,
      "learning_rate": 1.8414355563991963e-05,
      "loss": 0.0946,
      "step": 58200
    },
    {
      "epoch": 0.8574984168127127,
      "grad_norm": 1.451155662536621,
      "learning_rate": 1.841271919202498e-05,
      "loss": 0.0981,
      "step": 58225
    },
    {
      "epoch": 0.8578665999027997,
      "grad_norm": 1.9024156332015991,
      "learning_rate": 1.8411082820057995e-05,
      "loss": 0.0992,
      "step": 58250
    },
    {
      "epoch": 0.8582347829928867,
      "grad_norm": 1.3425402641296387,
      "learning_rate": 1.840944644809101e-05,
      "loss": 0.1,
      "step": 58275
    },
    {
      "epoch": 0.8586029660829737,
      "grad_norm": 1.6113873720169067,
      "learning_rate": 1.8407875531002707e-05,
      "loss": 0.0881,
      "step": 58300
    },
    {
      "epoch": 0.8589711491730608,
      "grad_norm": 1.4160832166671753,
      "learning_rate": 1.8406239159035718e-05,
      "loss": 0.1002,
      "step": 58325
    },
    {
      "epoch": 0.8593393322631478,
      "grad_norm": 1.4228044748306274,
      "learning_rate": 1.8404602787068736e-05,
      "loss": 0.0969,
      "step": 58350
    },
    {
      "epoch": 0.8597075153532349,
      "grad_norm": 1.7306244373321533,
      "learning_rate": 1.840296641510175e-05,
      "loss": 0.0903,
      "step": 58375
    },
    {
      "epoch": 0.8600756984433219,
      "grad_norm": 1.4546258449554443,
      "learning_rate": 1.8401330043134765e-05,
      "loss": 0.0914,
      "step": 58400
    },
    {
      "epoch": 0.8604438815334089,
      "grad_norm": 1.4770536422729492,
      "learning_rate": 1.8399693671167783e-05,
      "loss": 0.0957,
      "step": 58425
    },
    {
      "epoch": 0.860812064623496,
      "grad_norm": 1.6159794330596924,
      "learning_rate": 1.8398057299200797e-05,
      "loss": 0.0967,
      "step": 58450
    },
    {
      "epoch": 0.861180247713583,
      "grad_norm": 1.7649896144866943,
      "learning_rate": 1.8396420927233815e-05,
      "loss": 0.0976,
      "step": 58475
    },
    {
      "epoch": 0.86154843080367,
      "grad_norm": 1.5954879522323608,
      "learning_rate": 1.839478455526683e-05,
      "loss": 0.0905,
      "step": 58500
    },
    {
      "epoch": 0.8619166138937571,
      "grad_norm": 1.1200867891311646,
      "learning_rate": 1.8393148183299844e-05,
      "loss": 0.0925,
      "step": 58525
    },
    {
      "epoch": 0.8622847969838441,
      "grad_norm": 1.5213146209716797,
      "learning_rate": 1.8391511811332858e-05,
      "loss": 0.0998,
      "step": 58550
    },
    {
      "epoch": 0.8626529800739312,
      "grad_norm": 1.3924262523651123,
      "learning_rate": 1.8389875439365873e-05,
      "loss": 0.0902,
      "step": 58575
    },
    {
      "epoch": 0.8630211631640182,
      "grad_norm": 1.9037601947784424,
      "learning_rate": 1.838823906739889e-05,
      "loss": 0.09,
      "step": 58600
    },
    {
      "epoch": 0.8633893462541052,
      "grad_norm": 1.5457444190979004,
      "learning_rate": 1.8386602695431905e-05,
      "loss": 0.0881,
      "step": 58625
    },
    {
      "epoch": 0.8637575293441923,
      "grad_norm": 1.3769618272781372,
      "learning_rate": 1.838496632346492e-05,
      "loss": 0.0928,
      "step": 58650
    },
    {
      "epoch": 0.8641257124342793,
      "grad_norm": 1.835276484489441,
      "learning_rate": 1.8383329951497937e-05,
      "loss": 0.097,
      "step": 58675
    },
    {
      "epoch": 0.8644938955243664,
      "grad_norm": 1.8566334247589111,
      "learning_rate": 1.8381693579530952e-05,
      "loss": 0.0921,
      "step": 58700
    },
    {
      "epoch": 0.8648620786144534,
      "grad_norm": 1.4504879713058472,
      "learning_rate": 1.838005720756397e-05,
      "loss": 0.0866,
      "step": 58725
    },
    {
      "epoch": 0.8652302617045404,
      "grad_norm": 1.6636848449707031,
      "learning_rate": 1.837842083559698e-05,
      "loss": 0.0933,
      "step": 58750
    },
    {
      "epoch": 0.8655984447946274,
      "grad_norm": 1.262019395828247,
      "learning_rate": 1.837678446363e-05,
      "loss": 0.0924,
      "step": 58775
    },
    {
      "epoch": 0.8659666278847145,
      "grad_norm": 1.5972861051559448,
      "learning_rate": 1.8375148091663013e-05,
      "loss": 0.0924,
      "step": 58800
    },
    {
      "epoch": 0.8663348109748016,
      "grad_norm": 1.4554160833358765,
      "learning_rate": 1.8373511719696027e-05,
      "loss": 0.0894,
      "step": 58825
    },
    {
      "epoch": 0.8667029940648886,
      "grad_norm": 1.5403928756713867,
      "learning_rate": 1.8371875347729045e-05,
      "loss": 0.0868,
      "step": 58850
    },
    {
      "epoch": 0.8670711771549756,
      "grad_norm": 1.3819833993911743,
      "learning_rate": 1.837023897576206e-05,
      "loss": 0.0962,
      "step": 58875
    },
    {
      "epoch": 0.8674393602450626,
      "grad_norm": 1.4847444295883179,
      "learning_rate": 1.8368602603795078e-05,
      "loss": 0.0857,
      "step": 58900
    },
    {
      "epoch": 0.8678075433351498,
      "grad_norm": 1.8123759031295776,
      "learning_rate": 1.8366966231828092e-05,
      "loss": 0.0971,
      "step": 58925
    },
    {
      "epoch": 0.8681757264252368,
      "grad_norm": 1.2447463274002075,
      "learning_rate": 1.8365329859861107e-05,
      "loss": 0.0942,
      "step": 58950
    },
    {
      "epoch": 0.8685439095153238,
      "grad_norm": 1.2621221542358398,
      "learning_rate": 1.836369348789412e-05,
      "loss": 0.0934,
      "step": 58975
    },
    {
      "epoch": 0.8689120926054108,
      "grad_norm": 1.3399676084518433,
      "learning_rate": 1.8362057115927135e-05,
      "loss": 0.0898,
      "step": 59000
    },
    {
      "epoch": 0.8692802756954978,
      "grad_norm": 2.2415242195129395,
      "learning_rate": 1.8360420743960153e-05,
      "loss": 0.0867,
      "step": 59025
    },
    {
      "epoch": 0.8696484587855849,
      "grad_norm": 1.316906452178955,
      "learning_rate": 1.8358784371993168e-05,
      "loss": 0.0853,
      "step": 59050
    },
    {
      "epoch": 0.870016641875672,
      "grad_norm": 1.25469172000885,
      "learning_rate": 1.8357148000026182e-05,
      "loss": 0.0913,
      "step": 59075
    },
    {
      "epoch": 0.870384824965759,
      "grad_norm": 1.3025234937667847,
      "learning_rate": 1.83555116280592e-05,
      "loss": 0.0873,
      "step": 59100
    },
    {
      "epoch": 0.870753008055846,
      "grad_norm": 1.4825505018234253,
      "learning_rate": 1.8353875256092215e-05,
      "loss": 0.0951,
      "step": 59125
    },
    {
      "epoch": 0.871121191145933,
      "grad_norm": 1.3739885091781616,
      "learning_rate": 1.8352238884125232e-05,
      "loss": 0.0889,
      "step": 59150
    },
    {
      "epoch": 0.8714893742360201,
      "grad_norm": 1.4379063844680786,
      "learning_rate": 1.8350602512158243e-05,
      "loss": 0.0921,
      "step": 59175
    },
    {
      "epoch": 0.8718575573261071,
      "grad_norm": 1.3322168588638306,
      "learning_rate": 1.834896614019126e-05,
      "loss": 0.0977,
      "step": 59200
    },
    {
      "epoch": 0.8722257404161942,
      "grad_norm": 1.9719409942626953,
      "learning_rate": 1.8347329768224276e-05,
      "loss": 0.0874,
      "step": 59225
    },
    {
      "epoch": 0.8725939235062812,
      "grad_norm": 1.30929434299469,
      "learning_rate": 1.834569339625729e-05,
      "loss": 0.0981,
      "step": 59250
    },
    {
      "epoch": 0.8729621065963682,
      "grad_norm": 1.5915038585662842,
      "learning_rate": 1.8344057024290308e-05,
      "loss": 0.0909,
      "step": 59275
    },
    {
      "epoch": 0.8733302896864553,
      "grad_norm": 1.4955673217773438,
      "learning_rate": 1.8342420652323323e-05,
      "loss": 0.092,
      "step": 59300
    },
    {
      "epoch": 0.8736984727765423,
      "grad_norm": 1.8077608346939087,
      "learning_rate": 1.8340784280356337e-05,
      "loss": 0.0886,
      "step": 59325
    },
    {
      "epoch": 0.8740666558666293,
      "grad_norm": 1.2306609153747559,
      "learning_rate": 1.8339147908389355e-05,
      "loss": 0.0881,
      "step": 59350
    },
    {
      "epoch": 0.8744348389567164,
      "grad_norm": 1.1256755590438843,
      "learning_rate": 1.833751153642237e-05,
      "loss": 0.096,
      "step": 59375
    },
    {
      "epoch": 0.8748030220468034,
      "grad_norm": 1.2353475093841553,
      "learning_rate": 1.8335875164455384e-05,
      "loss": 0.0873,
      "step": 59400
    },
    {
      "epoch": 0.8751712051368905,
      "grad_norm": 1.5929750204086304,
      "learning_rate": 1.8334238792488398e-05,
      "loss": 0.0943,
      "step": 59425
    },
    {
      "epoch": 0.8755393882269775,
      "grad_norm": 1.3684848546981812,
      "learning_rate": 1.8332602420521416e-05,
      "loss": 0.0869,
      "step": 59450
    },
    {
      "epoch": 0.8759075713170645,
      "grad_norm": 1.2853327989578247,
      "learning_rate": 1.833096604855443e-05,
      "loss": 0.0947,
      "step": 59475
    },
    {
      "epoch": 0.8762757544071516,
      "grad_norm": 1.2716856002807617,
      "learning_rate": 1.8329329676587445e-05,
      "loss": 0.0946,
      "step": 59500
    },
    {
      "epoch": 0.8766439374972387,
      "grad_norm": 1.4013023376464844,
      "learning_rate": 1.8327693304620463e-05,
      "loss": 0.086,
      "step": 59525
    },
    {
      "epoch": 0.8770121205873257,
      "grad_norm": 1.4766559600830078,
      "learning_rate": 1.8326056932653477e-05,
      "loss": 0.1019,
      "step": 59550
    },
    {
      "epoch": 0.8773803036774127,
      "grad_norm": 1.2658036947250366,
      "learning_rate": 1.8324420560686492e-05,
      "loss": 0.0889,
      "step": 59575
    },
    {
      "epoch": 0.8777484867674997,
      "grad_norm": 1.474961757659912,
      "learning_rate": 1.8322784188719506e-05,
      "loss": 0.088,
      "step": 59600
    },
    {
      "epoch": 0.8781166698575867,
      "grad_norm": 1.3331751823425293,
      "learning_rate": 1.8321147816752524e-05,
      "loss": 0.0889,
      "step": 59625
    },
    {
      "epoch": 0.8784848529476739,
      "grad_norm": 1.0465377569198608,
      "learning_rate": 1.831951144478554e-05,
      "loss": 0.0819,
      "step": 59650
    },
    {
      "epoch": 0.8788530360377609,
      "grad_norm": 1.4545894861221313,
      "learning_rate": 1.8317875072818553e-05,
      "loss": 0.09,
      "step": 59675
    },
    {
      "epoch": 0.8792212191278479,
      "grad_norm": 1.7496753931045532,
      "learning_rate": 1.831623870085157e-05,
      "loss": 0.0936,
      "step": 59700
    },
    {
      "epoch": 0.8795894022179349,
      "grad_norm": 1.6359854936599731,
      "learning_rate": 1.8314602328884585e-05,
      "loss": 0.0936,
      "step": 59725
    },
    {
      "epoch": 0.8799575853080219,
      "grad_norm": 1.7177306413650513,
      "learning_rate": 1.83129659569176e-05,
      "loss": 0.0924,
      "step": 59750
    },
    {
      "epoch": 0.880325768398109,
      "grad_norm": 0.9038068652153015,
      "learning_rate": 1.8311329584950618e-05,
      "loss": 0.0893,
      "step": 59775
    },
    {
      "epoch": 0.8806939514881961,
      "grad_norm": 1.7128791809082031,
      "learning_rate": 1.830969321298363e-05,
      "loss": 0.0939,
      "step": 59800
    },
    {
      "epoch": 0.8810621345782831,
      "grad_norm": 1.707631230354309,
      "learning_rate": 1.8308056841016646e-05,
      "loss": 0.0925,
      "step": 59825
    },
    {
      "epoch": 0.8814303176683701,
      "grad_norm": 1.3236687183380127,
      "learning_rate": 1.830642046904966e-05,
      "loss": 0.0977,
      "step": 59850
    },
    {
      "epoch": 0.8817985007584571,
      "grad_norm": 1.399573564529419,
      "learning_rate": 1.830478409708268e-05,
      "loss": 0.0956,
      "step": 59875
    },
    {
      "epoch": 0.8821666838485442,
      "grad_norm": 1.2191760540008545,
      "learning_rate": 1.8303147725115693e-05,
      "loss": 0.0807,
      "step": 59900
    },
    {
      "epoch": 0.8825348669386313,
      "grad_norm": 1.6837016344070435,
      "learning_rate": 1.8301511353148708e-05,
      "loss": 0.0893,
      "step": 59925
    },
    {
      "epoch": 0.8829030500287183,
      "grad_norm": 1.6863495111465454,
      "learning_rate": 1.8299874981181726e-05,
      "loss": 0.0905,
      "step": 59950
    },
    {
      "epoch": 0.8832712331188053,
      "grad_norm": 1.422714114189148,
      "learning_rate": 1.829823860921474e-05,
      "loss": 0.0909,
      "step": 59975
    },
    {
      "epoch": 0.8836394162088923,
      "grad_norm": 1.6514418125152588,
      "learning_rate": 1.8296602237247754e-05,
      "loss": 0.0926,
      "step": 60000
    },
    {
      "epoch": 0.8840075992989794,
      "grad_norm": 1.4973467588424683,
      "learning_rate": 1.829496586528077e-05,
      "loss": 0.0851,
      "step": 60025
    },
    {
      "epoch": 0.8843757823890664,
      "grad_norm": 1.7477022409439087,
      "learning_rate": 1.8293329493313787e-05,
      "loss": 0.0945,
      "step": 60050
    },
    {
      "epoch": 0.8847439654791535,
      "grad_norm": 0.9304346442222595,
      "learning_rate": 1.82916931213468e-05,
      "loss": 0.0966,
      "step": 60075
    },
    {
      "epoch": 0.8851121485692405,
      "grad_norm": 1.712140679359436,
      "learning_rate": 1.8290056749379816e-05,
      "loss": 0.0943,
      "step": 60100
    },
    {
      "epoch": 0.8854803316593276,
      "grad_norm": 1.0455185174942017,
      "learning_rate": 1.8288420377412833e-05,
      "loss": 0.0914,
      "step": 60125
    },
    {
      "epoch": 0.8858485147494146,
      "grad_norm": 1.478262186050415,
      "learning_rate": 1.8286784005445848e-05,
      "loss": 0.0829,
      "step": 60150
    },
    {
      "epoch": 0.8862166978395016,
      "grad_norm": 1.6686688661575317,
      "learning_rate": 1.8285147633478862e-05,
      "loss": 0.0839,
      "step": 60175
    },
    {
      "epoch": 0.8865848809295886,
      "grad_norm": 1.6570348739624023,
      "learning_rate": 1.828351126151188e-05,
      "loss": 0.0978,
      "step": 60200
    },
    {
      "epoch": 0.8869530640196757,
      "grad_norm": 1.4632148742675781,
      "learning_rate": 1.828187488954489e-05,
      "loss": 0.0935,
      "step": 60225
    },
    {
      "epoch": 0.8873212471097628,
      "grad_norm": 1.9386918544769287,
      "learning_rate": 1.828023851757791e-05,
      "loss": 0.0916,
      "step": 60250
    },
    {
      "epoch": 0.8876894301998498,
      "grad_norm": 1.4522777795791626,
      "learning_rate": 1.8278602145610924e-05,
      "loss": 0.0889,
      "step": 60275
    },
    {
      "epoch": 0.8880576132899368,
      "grad_norm": 1.8300904035568237,
      "learning_rate": 1.827696577364394e-05,
      "loss": 0.0979,
      "step": 60300
    },
    {
      "epoch": 0.8884257963800238,
      "grad_norm": 1.9495863914489746,
      "learning_rate": 1.8275329401676956e-05,
      "loss": 0.1011,
      "step": 60325
    },
    {
      "epoch": 0.8887939794701109,
      "grad_norm": 1.876254916191101,
      "learning_rate": 1.827369302970997e-05,
      "loss": 0.0896,
      "step": 60350
    },
    {
      "epoch": 0.889162162560198,
      "grad_norm": 1.2851182222366333,
      "learning_rate": 1.8272056657742988e-05,
      "loss": 0.0931,
      "step": 60375
    },
    {
      "epoch": 0.889530345650285,
      "grad_norm": 1.205654263496399,
      "learning_rate": 1.8270420285776003e-05,
      "loss": 0.0948,
      "step": 60400
    },
    {
      "epoch": 0.889898528740372,
      "grad_norm": 1.613236665725708,
      "learning_rate": 1.8268783913809017e-05,
      "loss": 0.0928,
      "step": 60425
    },
    {
      "epoch": 0.890266711830459,
      "grad_norm": 1.8041343688964844,
      "learning_rate": 1.826714754184203e-05,
      "loss": 0.0892,
      "step": 60450
    },
    {
      "epoch": 0.890634894920546,
      "grad_norm": 0.9721527695655823,
      "learning_rate": 1.8265511169875046e-05,
      "loss": 0.0912,
      "step": 60475
    },
    {
      "epoch": 0.8910030780106332,
      "grad_norm": 1.0499447584152222,
      "learning_rate": 1.8263940252786743e-05,
      "loss": 0.0893,
      "step": 60500
    },
    {
      "epoch": 0.8913712611007202,
      "grad_norm": 1.4288532733917236,
      "learning_rate": 1.8262303880819758e-05,
      "loss": 0.0914,
      "step": 60525
    },
    {
      "epoch": 0.8917394441908072,
      "grad_norm": 1.3860613107681274,
      "learning_rate": 1.8260667508852776e-05,
      "loss": 0.0884,
      "step": 60550
    },
    {
      "epoch": 0.8921076272808942,
      "grad_norm": 1.6695598363876343,
      "learning_rate": 1.825903113688579e-05,
      "loss": 0.0911,
      "step": 60575
    },
    {
      "epoch": 0.8924758103709812,
      "grad_norm": 1.0595664978027344,
      "learning_rate": 1.8257394764918805e-05,
      "loss": 0.0926,
      "step": 60600
    },
    {
      "epoch": 0.8928439934610684,
      "grad_norm": 1.530721664428711,
      "learning_rate": 1.8255758392951822e-05,
      "loss": 0.0904,
      "step": 60625
    },
    {
      "epoch": 0.8932121765511554,
      "grad_norm": 1.4969135522842407,
      "learning_rate": 1.8254122020984833e-05,
      "loss": 0.0937,
      "step": 60650
    },
    {
      "epoch": 0.8935803596412424,
      "grad_norm": 1.4802945852279663,
      "learning_rate": 1.825248564901785e-05,
      "loss": 0.0884,
      "step": 60675
    },
    {
      "epoch": 0.8939485427313294,
      "grad_norm": 1.5618170499801636,
      "learning_rate": 1.8250849277050866e-05,
      "loss": 0.092,
      "step": 60700
    },
    {
      "epoch": 0.8943167258214165,
      "grad_norm": 1.3365405797958374,
      "learning_rate": 1.824921290508388e-05,
      "loss": 0.0916,
      "step": 60725
    },
    {
      "epoch": 0.8946849089115035,
      "grad_norm": 1.8510189056396484,
      "learning_rate": 1.8247576533116898e-05,
      "loss": 0.0898,
      "step": 60750
    },
    {
      "epoch": 0.8950530920015906,
      "grad_norm": 1.214875340461731,
      "learning_rate": 1.8245940161149913e-05,
      "loss": 0.0954,
      "step": 60775
    },
    {
      "epoch": 0.8954212750916776,
      "grad_norm": 1.3900574445724487,
      "learning_rate": 1.824430378918293e-05,
      "loss": 0.0936,
      "step": 60800
    },
    {
      "epoch": 0.8957894581817646,
      "grad_norm": 1.2594184875488281,
      "learning_rate": 1.8242667417215945e-05,
      "loss": 0.0927,
      "step": 60825
    },
    {
      "epoch": 0.8961576412718517,
      "grad_norm": 1.2305347919464111,
      "learning_rate": 1.824103104524896e-05,
      "loss": 0.0864,
      "step": 60850
    },
    {
      "epoch": 0.8965258243619387,
      "grad_norm": 1.9893605709075928,
      "learning_rate": 1.8239394673281974e-05,
      "loss": 0.0894,
      "step": 60875
    },
    {
      "epoch": 0.8968940074520257,
      "grad_norm": 1.546601414680481,
      "learning_rate": 1.8237758301314988e-05,
      "loss": 0.0947,
      "step": 60900
    },
    {
      "epoch": 0.8972621905421128,
      "grad_norm": 1.658313512802124,
      "learning_rate": 1.8236121929348006e-05,
      "loss": 0.0953,
      "step": 60925
    },
    {
      "epoch": 0.8976303736321998,
      "grad_norm": 1.5878702402114868,
      "learning_rate": 1.823448555738102e-05,
      "loss": 0.0864,
      "step": 60950
    },
    {
      "epoch": 0.8979985567222869,
      "grad_norm": 1.3052133321762085,
      "learning_rate": 1.8232849185414035e-05,
      "loss": 0.0827,
      "step": 60975
    },
    {
      "epoch": 0.8983667398123739,
      "grad_norm": 1.3506282567977905,
      "learning_rate": 1.8231212813447053e-05,
      "loss": 0.0958,
      "step": 61000
    },
    {
      "epoch": 0.8987349229024609,
      "grad_norm": 1.0250307321548462,
      "learning_rate": 1.8229576441480067e-05,
      "loss": 0.0844,
      "step": 61025
    },
    {
      "epoch": 0.899103105992548,
      "grad_norm": 1.356543779373169,
      "learning_rate": 1.8227940069513085e-05,
      "loss": 0.0944,
      "step": 61050
    },
    {
      "epoch": 0.899471289082635,
      "grad_norm": 1.3725234270095825,
      "learning_rate": 1.8226303697546096e-05,
      "loss": 0.0887,
      "step": 61075
    },
    {
      "epoch": 0.8998394721727221,
      "grad_norm": 1.2289153337478638,
      "learning_rate": 1.8224667325579114e-05,
      "loss": 0.0874,
      "step": 61100
    },
    {
      "epoch": 0.9002076552628091,
      "grad_norm": 1.7129337787628174,
      "learning_rate": 1.822303095361213e-05,
      "loss": 0.0996,
      "step": 61125
    },
    {
      "epoch": 0.9005758383528961,
      "grad_norm": 1.4440714120864868,
      "learning_rate": 1.8221394581645143e-05,
      "loss": 0.0911,
      "step": 61150
    },
    {
      "epoch": 0.9009440214429831,
      "grad_norm": 1.5119160413742065,
      "learning_rate": 1.821975820967816e-05,
      "loss": 0.0898,
      "step": 61175
    },
    {
      "epoch": 0.9013122045330703,
      "grad_norm": 1.3910434246063232,
      "learning_rate": 1.8218121837711175e-05,
      "loss": 0.0949,
      "step": 61200
    },
    {
      "epoch": 0.9016803876231573,
      "grad_norm": 1.4344723224639893,
      "learning_rate": 1.8216485465744193e-05,
      "loss": 0.0929,
      "step": 61225
    },
    {
      "epoch": 0.9020485707132443,
      "grad_norm": 1.3344783782958984,
      "learning_rate": 1.8214849093777208e-05,
      "loss": 0.0879,
      "step": 61250
    },
    {
      "epoch": 0.9024167538033313,
      "grad_norm": 2.065082311630249,
      "learning_rate": 1.8213212721810222e-05,
      "loss": 0.0874,
      "step": 61275
    },
    {
      "epoch": 0.9027849368934183,
      "grad_norm": 1.8886113166809082,
      "learning_rate": 1.8211576349843236e-05,
      "loss": 0.09,
      "step": 61300
    },
    {
      "epoch": 0.9031531199835054,
      "grad_norm": 1.7450238466262817,
      "learning_rate": 1.820993997787625e-05,
      "loss": 0.0908,
      "step": 61325
    },
    {
      "epoch": 0.9035213030735925,
      "grad_norm": 1.3400936126708984,
      "learning_rate": 1.820830360590927e-05,
      "loss": 0.0966,
      "step": 61350
    },
    {
      "epoch": 0.9038894861636795,
      "grad_norm": 1.3660224676132202,
      "learning_rate": 1.8206667233942283e-05,
      "loss": 0.0878,
      "step": 61375
    },
    {
      "epoch": 0.9042576692537665,
      "grad_norm": 1.0200551748275757,
      "learning_rate": 1.8205030861975298e-05,
      "loss": 0.0861,
      "step": 61400
    },
    {
      "epoch": 0.9046258523438535,
      "grad_norm": 1.4215823411941528,
      "learning_rate": 1.8203394490008316e-05,
      "loss": 0.0835,
      "step": 61425
    },
    {
      "epoch": 0.9049940354339406,
      "grad_norm": 1.767581820487976,
      "learning_rate": 1.820175811804133e-05,
      "loss": 0.0911,
      "step": 61450
    },
    {
      "epoch": 0.9053622185240277,
      "grad_norm": 1.8444174528121948,
      "learning_rate": 1.8200121746074348e-05,
      "loss": 0.0993,
      "step": 61475
    },
    {
      "epoch": 0.9057304016141147,
      "grad_norm": 1.2559212446212769,
      "learning_rate": 1.819848537410736e-05,
      "loss": 0.0895,
      "step": 61500
    },
    {
      "epoch": 0.9060985847042017,
      "grad_norm": 1.2072019577026367,
      "learning_rate": 1.8196849002140377e-05,
      "loss": 0.0901,
      "step": 61525
    },
    {
      "epoch": 0.9064667677942887,
      "grad_norm": 1.5533256530761719,
      "learning_rate": 1.819521263017339e-05,
      "loss": 0.0838,
      "step": 61550
    },
    {
      "epoch": 0.9068349508843758,
      "grad_norm": 1.1651979684829712,
      "learning_rate": 1.8193576258206406e-05,
      "loss": 0.0952,
      "step": 61575
    },
    {
      "epoch": 0.9072031339744628,
      "grad_norm": 1.433811068534851,
      "learning_rate": 1.8191939886239423e-05,
      "loss": 0.086,
      "step": 61600
    },
    {
      "epoch": 0.9075713170645499,
      "grad_norm": 1.8967841863632202,
      "learning_rate": 1.8190303514272438e-05,
      "loss": 0.0869,
      "step": 61625
    },
    {
      "epoch": 0.9079395001546369,
      "grad_norm": 1.5940126180648804,
      "learning_rate": 1.8188667142305452e-05,
      "loss": 0.0841,
      "step": 61650
    },
    {
      "epoch": 0.9083076832447239,
      "grad_norm": 1.2679402828216553,
      "learning_rate": 1.818703077033847e-05,
      "loss": 0.0844,
      "step": 61675
    },
    {
      "epoch": 0.908675866334811,
      "grad_norm": 1.1893339157104492,
      "learning_rate": 1.8185394398371485e-05,
      "loss": 0.0977,
      "step": 61700
    },
    {
      "epoch": 0.909044049424898,
      "grad_norm": 1.3891805410385132,
      "learning_rate": 1.81837580264045e-05,
      "loss": 0.0919,
      "step": 61725
    },
    {
      "epoch": 0.909412232514985,
      "grad_norm": 1.474007487297058,
      "learning_rate": 1.8182121654437514e-05,
      "loss": 0.0919,
      "step": 61750
    },
    {
      "epoch": 0.9097804156050721,
      "grad_norm": 1.4935157299041748,
      "learning_rate": 1.818048528247053e-05,
      "loss": 0.0823,
      "step": 61775
    },
    {
      "epoch": 0.9101485986951592,
      "grad_norm": 1.199958324432373,
      "learning_rate": 1.8178848910503546e-05,
      "loss": 0.0868,
      "step": 61800
    },
    {
      "epoch": 0.9105167817852462,
      "grad_norm": 1.1490696668624878,
      "learning_rate": 1.817721253853656e-05,
      "loss": 0.0945,
      "step": 61825
    },
    {
      "epoch": 0.9108849648753332,
      "grad_norm": 1.5686537027359009,
      "learning_rate": 1.8175576166569578e-05,
      "loss": 0.0968,
      "step": 61850
    },
    {
      "epoch": 0.9112531479654202,
      "grad_norm": 1.8495757579803467,
      "learning_rate": 1.8173939794602593e-05,
      "loss": 0.0936,
      "step": 61875
    },
    {
      "epoch": 0.9116213310555072,
      "grad_norm": 1.4681916236877441,
      "learning_rate": 1.8172303422635607e-05,
      "loss": 0.0918,
      "step": 61900
    },
    {
      "epoch": 0.9119895141455944,
      "grad_norm": 1.4203157424926758,
      "learning_rate": 1.817066705066862e-05,
      "loss": 0.09,
      "step": 61925
    },
    {
      "epoch": 0.9123576972356814,
      "grad_norm": 1.233629584312439,
      "learning_rate": 1.816903067870164e-05,
      "loss": 0.0806,
      "step": 61950
    },
    {
      "epoch": 0.9127258803257684,
      "grad_norm": 1.7074884176254272,
      "learning_rate": 1.8167394306734654e-05,
      "loss": 0.0925,
      "step": 61975
    },
    {
      "epoch": 0.9130940634158554,
      "grad_norm": 1.5755255222320557,
      "learning_rate": 1.816575793476767e-05,
      "loss": 0.0877,
      "step": 62000
    },
    {
      "epoch": 0.9134622465059424,
      "grad_norm": 1.721422553062439,
      "learning_rate": 1.8164121562800686e-05,
      "loss": 0.0889,
      "step": 62025
    },
    {
      "epoch": 0.9138304295960296,
      "grad_norm": 1.6015130281448364,
      "learning_rate": 1.81624851908337e-05,
      "loss": 0.091,
      "step": 62050
    },
    {
      "epoch": 0.9141986126861166,
      "grad_norm": 1.6153591871261597,
      "learning_rate": 1.8160848818866715e-05,
      "loss": 0.0935,
      "step": 62075
    },
    {
      "epoch": 0.9145667957762036,
      "grad_norm": 1.5466082096099854,
      "learning_rate": 1.8159212446899733e-05,
      "loss": 0.0951,
      "step": 62100
    },
    {
      "epoch": 0.9149349788662906,
      "grad_norm": 1.756897211074829,
      "learning_rate": 1.8157576074932744e-05,
      "loss": 0.0825,
      "step": 62125
    },
    {
      "epoch": 0.9153031619563776,
      "grad_norm": 1.7807000875473022,
      "learning_rate": 1.8155939702965762e-05,
      "loss": 0.1,
      "step": 62150
    },
    {
      "epoch": 0.9156713450464647,
      "grad_norm": 1.0893385410308838,
      "learning_rate": 1.8154303330998776e-05,
      "loss": 0.0848,
      "step": 62175
    },
    {
      "epoch": 0.9160395281365518,
      "grad_norm": 1.796811819076538,
      "learning_rate": 1.8152666959031794e-05,
      "loss": 0.0851,
      "step": 62200
    },
    {
      "epoch": 0.9164077112266388,
      "grad_norm": 1.8552054166793823,
      "learning_rate": 1.815103058706481e-05,
      "loss": 0.0944,
      "step": 62225
    },
    {
      "epoch": 0.9167758943167258,
      "grad_norm": 1.2951158285140991,
      "learning_rate": 1.8149394215097823e-05,
      "loss": 0.0847,
      "step": 62250
    },
    {
      "epoch": 0.9171440774068128,
      "grad_norm": 1.18398916721344,
      "learning_rate": 1.814775784313084e-05,
      "loss": 0.0954,
      "step": 62275
    },
    {
      "epoch": 0.9175122604968999,
      "grad_norm": 1.342115879058838,
      "learning_rate": 1.8146121471163855e-05,
      "loss": 0.0852,
      "step": 62300
    },
    {
      "epoch": 0.917880443586987,
      "grad_norm": 1.480775237083435,
      "learning_rate": 1.814448509919687e-05,
      "loss": 0.0837,
      "step": 62325
    },
    {
      "epoch": 0.918248626677074,
      "grad_norm": 1.4389697313308716,
      "learning_rate": 1.8142848727229884e-05,
      "loss": 0.0834,
      "step": 62350
    },
    {
      "epoch": 0.918616809767161,
      "grad_norm": 1.9070498943328857,
      "learning_rate": 1.8141212355262902e-05,
      "loss": 0.0931,
      "step": 62375
    },
    {
      "epoch": 0.9189849928572481,
      "grad_norm": 1.3322042226791382,
      "learning_rate": 1.8139575983295917e-05,
      "loss": 0.0942,
      "step": 62400
    },
    {
      "epoch": 0.9193531759473351,
      "grad_norm": 1.0659507513046265,
      "learning_rate": 1.813793961132893e-05,
      "loss": 0.0828,
      "step": 62425
    },
    {
      "epoch": 0.9197213590374221,
      "grad_norm": 1.1807823181152344,
      "learning_rate": 1.813630323936195e-05,
      "loss": 0.0868,
      "step": 62450
    },
    {
      "epoch": 0.9200895421275092,
      "grad_norm": 1.2817989587783813,
      "learning_rate": 1.8134666867394963e-05,
      "loss": 0.088,
      "step": 62475
    },
    {
      "epoch": 0.9204577252175962,
      "grad_norm": 1.4101269245147705,
      "learning_rate": 1.8133030495427978e-05,
      "loss": 0.0847,
      "step": 62500
    },
    {
      "epoch": 0.9208259083076833,
      "grad_norm": 1.565391182899475,
      "learning_rate": 1.8131394123460996e-05,
      "loss": 0.082,
      "step": 62525
    },
    {
      "epoch": 0.9211940913977703,
      "grad_norm": 1.62588369846344,
      "learning_rate": 1.8129757751494007e-05,
      "loss": 0.0911,
      "step": 62550
    },
    {
      "epoch": 0.9215622744878573,
      "grad_norm": 1.4808884859085083,
      "learning_rate": 1.8128121379527025e-05,
      "loss": 0.0896,
      "step": 62575
    },
    {
      "epoch": 0.9219304575779443,
      "grad_norm": 1.2906694412231445,
      "learning_rate": 1.812648500756004e-05,
      "loss": 0.0846,
      "step": 62600
    },
    {
      "epoch": 0.9222986406680314,
      "grad_norm": 1.2689462900161743,
      "learning_rate": 1.8124848635593057e-05,
      "loss": 0.0816,
      "step": 62625
    },
    {
      "epoch": 0.9226668237581185,
      "grad_norm": 1.4023360013961792,
      "learning_rate": 1.812321226362607e-05,
      "loss": 0.0858,
      "step": 62650
    },
    {
      "epoch": 0.9230350068482055,
      "grad_norm": 1.5279533863067627,
      "learning_rate": 1.8121575891659086e-05,
      "loss": 0.0918,
      "step": 62675
    },
    {
      "epoch": 0.9234031899382925,
      "grad_norm": 1.5480386018753052,
      "learning_rate": 1.8119939519692104e-05,
      "loss": 0.0972,
      "step": 62700
    },
    {
      "epoch": 0.9237713730283795,
      "grad_norm": 1.862874984741211,
      "learning_rate": 1.8118368602603798e-05,
      "loss": 0.088,
      "step": 62725
    },
    {
      "epoch": 0.9241395561184665,
      "grad_norm": 1.6180002689361572,
      "learning_rate": 1.8116732230636812e-05,
      "loss": 0.0937,
      "step": 62750
    },
    {
      "epoch": 0.9245077392085537,
      "grad_norm": 1.143884539604187,
      "learning_rate": 1.8115095858669826e-05,
      "loss": 0.0916,
      "step": 62775
    },
    {
      "epoch": 0.9248759222986407,
      "grad_norm": 1.1028547286987305,
      "learning_rate": 1.811345948670284e-05,
      "loss": 0.0871,
      "step": 62800
    },
    {
      "epoch": 0.9252441053887277,
      "grad_norm": 1.703330636024475,
      "learning_rate": 1.811182311473586e-05,
      "loss": 0.0888,
      "step": 62825
    },
    {
      "epoch": 0.9256122884788147,
      "grad_norm": 1.5463730096817017,
      "learning_rate": 1.8110186742768873e-05,
      "loss": 0.0924,
      "step": 62850
    },
    {
      "epoch": 0.9259804715689017,
      "grad_norm": 1.4528677463531494,
      "learning_rate": 1.810855037080189e-05,
      "loss": 0.0929,
      "step": 62875
    },
    {
      "epoch": 0.9263486546589889,
      "grad_norm": 1.3995440006256104,
      "learning_rate": 1.8106913998834906e-05,
      "loss": 0.0921,
      "step": 62900
    },
    {
      "epoch": 0.9267168377490759,
      "grad_norm": 1.1789085865020752,
      "learning_rate": 1.810527762686792e-05,
      "loss": 0.0845,
      "step": 62925
    },
    {
      "epoch": 0.9270850208391629,
      "grad_norm": 1.8213672637939453,
      "learning_rate": 1.8103641254900938e-05,
      "loss": 0.0909,
      "step": 62950
    },
    {
      "epoch": 0.9274532039292499,
      "grad_norm": 1.1578150987625122,
      "learning_rate": 1.810200488293395e-05,
      "loss": 0.0821,
      "step": 62975
    },
    {
      "epoch": 0.927821387019337,
      "grad_norm": 0.9961247444152832,
      "learning_rate": 1.8100368510966967e-05,
      "loss": 0.0891,
      "step": 63000
    },
    {
      "epoch": 0.928189570109424,
      "grad_norm": 1.5190428495407104,
      "learning_rate": 1.809873213899998e-05,
      "loss": 0.0993,
      "step": 63025
    },
    {
      "epoch": 0.9285577531995111,
      "grad_norm": 1.2491146326065063,
      "learning_rate": 1.8097095767032996e-05,
      "loss": 0.0835,
      "step": 63050
    },
    {
      "epoch": 0.9289259362895981,
      "grad_norm": 1.6360255479812622,
      "learning_rate": 1.8095459395066014e-05,
      "loss": 0.09,
      "step": 63075
    },
    {
      "epoch": 0.9292941193796851,
      "grad_norm": 1.5824953317642212,
      "learning_rate": 1.8093823023099028e-05,
      "loss": 0.0965,
      "step": 63100
    },
    {
      "epoch": 0.9296623024697722,
      "grad_norm": 1.808732509613037,
      "learning_rate": 1.8092186651132046e-05,
      "loss": 0.0821,
      "step": 63125
    },
    {
      "epoch": 0.9300304855598592,
      "grad_norm": 1.3798719644546509,
      "learning_rate": 1.809055027916506e-05,
      "loss": 0.0911,
      "step": 63150
    },
    {
      "epoch": 0.9303986686499462,
      "grad_norm": 1.413038730621338,
      "learning_rate": 1.8088913907198075e-05,
      "loss": 0.0941,
      "step": 63175
    },
    {
      "epoch": 0.9307668517400333,
      "grad_norm": 1.307576060295105,
      "learning_rate": 1.808727753523109e-05,
      "loss": 0.0899,
      "step": 63200
    },
    {
      "epoch": 0.9311350348301203,
      "grad_norm": 1.3909647464752197,
      "learning_rate": 1.8085641163264104e-05,
      "loss": 0.0864,
      "step": 63225
    },
    {
      "epoch": 0.9315032179202074,
      "grad_norm": 1.2756690979003906,
      "learning_rate": 1.808400479129712e-05,
      "loss": 0.0881,
      "step": 63250
    },
    {
      "epoch": 0.9318714010102944,
      "grad_norm": 1.1651579141616821,
      "learning_rate": 1.8082368419330136e-05,
      "loss": 0.0852,
      "step": 63275
    },
    {
      "epoch": 0.9322395841003814,
      "grad_norm": 1.597281813621521,
      "learning_rate": 1.808073204736315e-05,
      "loss": 0.0862,
      "step": 63300
    },
    {
      "epoch": 0.9326077671904685,
      "grad_norm": 1.5948517322540283,
      "learning_rate": 1.8079095675396168e-05,
      "loss": 0.0864,
      "step": 63325
    },
    {
      "epoch": 0.9329759502805555,
      "grad_norm": 1.3347128629684448,
      "learning_rate": 1.8077459303429183e-05,
      "loss": 0.0899,
      "step": 63350
    },
    {
      "epoch": 0.9333441333706426,
      "grad_norm": 1.7122323513031006,
      "learning_rate": 1.8075822931462197e-05,
      "loss": 0.0878,
      "step": 63375
    },
    {
      "epoch": 0.9337123164607296,
      "grad_norm": 0.9531201720237732,
      "learning_rate": 1.807418655949521e-05,
      "loss": 0.0792,
      "step": 63400
    },
    {
      "epoch": 0.9340804995508166,
      "grad_norm": 1.0709425210952759,
      "learning_rate": 1.807255018752823e-05,
      "loss": 0.0822,
      "step": 63425
    },
    {
      "epoch": 0.9344486826409036,
      "grad_norm": 1.5660427808761597,
      "learning_rate": 1.8070913815561244e-05,
      "loss": 0.0841,
      "step": 63450
    },
    {
      "epoch": 0.9348168657309907,
      "grad_norm": 1.3853843212127686,
      "learning_rate": 1.806927744359426e-05,
      "loss": 0.0807,
      "step": 63475
    },
    {
      "epoch": 0.9351850488210778,
      "grad_norm": 1.7013237476348877,
      "learning_rate": 1.8067641071627276e-05,
      "loss": 0.0869,
      "step": 63500
    },
    {
      "epoch": 0.9355532319111648,
      "grad_norm": 1.189096450805664,
      "learning_rate": 1.806600469966029e-05,
      "loss": 0.0831,
      "step": 63525
    },
    {
      "epoch": 0.9359214150012518,
      "grad_norm": 1.7477755546569824,
      "learning_rate": 1.806436832769331e-05,
      "loss": 0.0886,
      "step": 63550
    },
    {
      "epoch": 0.9362895980913388,
      "grad_norm": 1.5616755485534668,
      "learning_rate": 1.8062731955726323e-05,
      "loss": 0.0969,
      "step": 63575
    },
    {
      "epoch": 0.936657781181426,
      "grad_norm": 1.5929596424102783,
      "learning_rate": 1.8061095583759337e-05,
      "loss": 0.0848,
      "step": 63600
    },
    {
      "epoch": 0.937025964271513,
      "grad_norm": 1.4864095449447632,
      "learning_rate": 1.8059459211792352e-05,
      "loss": 0.0793,
      "step": 63625
    },
    {
      "epoch": 0.9373941473616,
      "grad_norm": 1.4461345672607422,
      "learning_rate": 1.8057822839825366e-05,
      "loss": 0.0941,
      "step": 63650
    },
    {
      "epoch": 0.937762330451687,
      "grad_norm": 1.7591949701309204,
      "learning_rate": 1.8056186467858384e-05,
      "loss": 0.0882,
      "step": 63675
    },
    {
      "epoch": 0.938130513541774,
      "grad_norm": 1.4366741180419922,
      "learning_rate": 1.80545500958914e-05,
      "loss": 0.0858,
      "step": 63700
    },
    {
      "epoch": 0.9384986966318611,
      "grad_norm": 1.7040621042251587,
      "learning_rate": 1.8052913723924413e-05,
      "loss": 0.0942,
      "step": 63725
    },
    {
      "epoch": 0.9388668797219482,
      "grad_norm": 1.5122658014297485,
      "learning_rate": 1.805127735195743e-05,
      "loss": 0.0897,
      "step": 63750
    },
    {
      "epoch": 0.9392350628120352,
      "grad_norm": 1.209884524345398,
      "learning_rate": 1.8049640979990445e-05,
      "loss": 0.0847,
      "step": 63775
    },
    {
      "epoch": 0.9396032459021222,
      "grad_norm": 1.0629122257232666,
      "learning_rate": 1.804800460802346e-05,
      "loss": 0.0868,
      "step": 63800
    },
    {
      "epoch": 0.9399714289922092,
      "grad_norm": 1.631815791130066,
      "learning_rate": 1.8046368236056474e-05,
      "loss": 0.0932,
      "step": 63825
    },
    {
      "epoch": 0.9403396120822963,
      "grad_norm": 1.504408597946167,
      "learning_rate": 1.8044731864089492e-05,
      "loss": 0.0873,
      "step": 63850
    },
    {
      "epoch": 0.9407077951723833,
      "grad_norm": 1.6363877058029175,
      "learning_rate": 1.8043095492122507e-05,
      "loss": 0.0885,
      "step": 63875
    },
    {
      "epoch": 0.9410759782624704,
      "grad_norm": 1.272243618965149,
      "learning_rate": 1.804145912015552e-05,
      "loss": 0.0945,
      "step": 63900
    },
    {
      "epoch": 0.9414441613525574,
      "grad_norm": 1.7437857389450073,
      "learning_rate": 1.803982274818854e-05,
      "loss": 0.0916,
      "step": 63925
    },
    {
      "epoch": 0.9418123444426444,
      "grad_norm": 1.6619237661361694,
      "learning_rate": 1.8038186376221553e-05,
      "loss": 0.093,
      "step": 63950
    },
    {
      "epoch": 0.9421805275327315,
      "grad_norm": 1.2458175420761108,
      "learning_rate": 1.8036550004254568e-05,
      "loss": 0.0816,
      "step": 63975
    },
    {
      "epoch": 0.9425487106228185,
      "grad_norm": 1.0379489660263062,
      "learning_rate": 1.8034913632287586e-05,
      "loss": 0.0911,
      "step": 64000
    },
    {
      "epoch": 0.9429168937129055,
      "grad_norm": 1.926479458808899,
      "learning_rate": 1.80332772603206e-05,
      "loss": 0.0919,
      "step": 64025
    },
    {
      "epoch": 0.9432850768029926,
      "grad_norm": 1.343337893486023,
      "learning_rate": 1.8031640888353615e-05,
      "loss": 0.0873,
      "step": 64050
    },
    {
      "epoch": 0.9436532598930796,
      "grad_norm": 1.640823245048523,
      "learning_rate": 1.803000451638663e-05,
      "loss": 0.0845,
      "step": 64075
    },
    {
      "epoch": 0.9440214429831667,
      "grad_norm": 0.8465896844863892,
      "learning_rate": 1.8028368144419647e-05,
      "loss": 0.0947,
      "step": 64100
    },
    {
      "epoch": 0.9443896260732537,
      "grad_norm": 1.8323887586593628,
      "learning_rate": 1.802673177245266e-05,
      "loss": 0.0913,
      "step": 64125
    },
    {
      "epoch": 0.9447578091633407,
      "grad_norm": 1.5387301445007324,
      "learning_rate": 1.8025095400485676e-05,
      "loss": 0.0935,
      "step": 64150
    },
    {
      "epoch": 0.9451259922534277,
      "grad_norm": 1.2456954717636108,
      "learning_rate": 1.8023459028518694e-05,
      "loss": 0.089,
      "step": 64175
    },
    {
      "epoch": 0.9454941753435149,
      "grad_norm": 1.3453234434127808,
      "learning_rate": 1.8021822656551708e-05,
      "loss": 0.0907,
      "step": 64200
    },
    {
      "epoch": 0.9458623584336019,
      "grad_norm": 1.6878015995025635,
      "learning_rate": 1.8020186284584723e-05,
      "loss": 0.093,
      "step": 64225
    },
    {
      "epoch": 0.9462305415236889,
      "grad_norm": 1.2218587398529053,
      "learning_rate": 1.8018549912617737e-05,
      "loss": 0.0941,
      "step": 64250
    },
    {
      "epoch": 0.9465987246137759,
      "grad_norm": 1.3658645153045654,
      "learning_rate": 1.8016913540650755e-05,
      "loss": 0.084,
      "step": 64275
    },
    {
      "epoch": 0.9469669077038629,
      "grad_norm": 1.2594447135925293,
      "learning_rate": 1.801527716868377e-05,
      "loss": 0.0889,
      "step": 64300
    },
    {
      "epoch": 0.9473350907939501,
      "grad_norm": 1.2579340934753418,
      "learning_rate": 1.8013640796716784e-05,
      "loss": 0.0823,
      "step": 64325
    },
    {
      "epoch": 0.9477032738840371,
      "grad_norm": 1.435950517654419,
      "learning_rate": 1.80120044247498e-05,
      "loss": 0.0898,
      "step": 64350
    },
    {
      "epoch": 0.9480714569741241,
      "grad_norm": 1.629874587059021,
      "learning_rate": 1.8010368052782816e-05,
      "loss": 0.0882,
      "step": 64375
    },
    {
      "epoch": 0.9484396400642111,
      "grad_norm": 1.4672939777374268,
      "learning_rate": 1.800873168081583e-05,
      "loss": 0.0854,
      "step": 64400
    },
    {
      "epoch": 0.9488078231542981,
      "grad_norm": 1.4122424125671387,
      "learning_rate": 1.800709530884885e-05,
      "loss": 0.09,
      "step": 64425
    },
    {
      "epoch": 0.9491760062443853,
      "grad_norm": 1.3469072580337524,
      "learning_rate": 1.800545893688186e-05,
      "loss": 0.0876,
      "step": 64450
    },
    {
      "epoch": 0.9495441893344723,
      "grad_norm": 1.363897442817688,
      "learning_rate": 1.8003822564914877e-05,
      "loss": 0.0828,
      "step": 64475
    },
    {
      "epoch": 0.9499123724245593,
      "grad_norm": 1.1692781448364258,
      "learning_rate": 1.8002186192947892e-05,
      "loss": 0.0847,
      "step": 64500
    },
    {
      "epoch": 0.9502805555146463,
      "grad_norm": 1.439585566520691,
      "learning_rate": 1.800054982098091e-05,
      "loss": 0.0803,
      "step": 64525
    },
    {
      "epoch": 0.9506487386047333,
      "grad_norm": 1.2381300926208496,
      "learning_rate": 1.7998913449013924e-05,
      "loss": 0.0916,
      "step": 64550
    },
    {
      "epoch": 0.9510169216948204,
      "grad_norm": 1.4574675559997559,
      "learning_rate": 1.799727707704694e-05,
      "loss": 0.0945,
      "step": 64575
    },
    {
      "epoch": 0.9513851047849075,
      "grad_norm": 1.155951738357544,
      "learning_rate": 1.7995640705079956e-05,
      "loss": 0.0871,
      "step": 64600
    },
    {
      "epoch": 0.9517532878749945,
      "grad_norm": 1.1017684936523438,
      "learning_rate": 1.799400433311297e-05,
      "loss": 0.0848,
      "step": 64625
    },
    {
      "epoch": 0.9521214709650815,
      "grad_norm": 1.5029999017715454,
      "learning_rate": 1.7992367961145985e-05,
      "loss": 0.0889,
      "step": 64650
    },
    {
      "epoch": 0.9524896540551685,
      "grad_norm": 1.666143536567688,
      "learning_rate": 1.7990731589179e-05,
      "loss": 0.0928,
      "step": 64675
    },
    {
      "epoch": 0.9528578371452556,
      "grad_norm": 1.2904329299926758,
      "learning_rate": 1.7989095217212014e-05,
      "loss": 0.0915,
      "step": 64700
    },
    {
      "epoch": 0.9532260202353426,
      "grad_norm": 1.4966198205947876,
      "learning_rate": 1.7987458845245032e-05,
      "loss": 0.0785,
      "step": 64725
    },
    {
      "epoch": 0.9535942033254297,
      "grad_norm": 1.3005931377410889,
      "learning_rate": 1.7985822473278047e-05,
      "loss": 0.0902,
      "step": 64750
    },
    {
      "epoch": 0.9539623864155167,
      "grad_norm": 1.4355655908584595,
      "learning_rate": 1.7984186101311064e-05,
      "loss": 0.089,
      "step": 64775
    },
    {
      "epoch": 0.9543305695056038,
      "grad_norm": 1.433249831199646,
      "learning_rate": 1.798254972934408e-05,
      "loss": 0.0867,
      "step": 64800
    },
    {
      "epoch": 0.9546987525956908,
      "grad_norm": 1.392223596572876,
      "learning_rate": 1.7980913357377093e-05,
      "loss": 0.0848,
      "step": 64825
    },
    {
      "epoch": 0.9550669356857778,
      "grad_norm": 1.0814814567565918,
      "learning_rate": 1.7979276985410108e-05,
      "loss": 0.0905,
      "step": 64850
    },
    {
      "epoch": 0.9554351187758648,
      "grad_norm": 1.419834852218628,
      "learning_rate": 1.7977640613443122e-05,
      "loss": 0.0867,
      "step": 64875
    },
    {
      "epoch": 0.9558033018659519,
      "grad_norm": 1.4989286661148071,
      "learning_rate": 1.797600424147614e-05,
      "loss": 0.091,
      "step": 64900
    },
    {
      "epoch": 0.956171484956039,
      "grad_norm": 1.759748101234436,
      "learning_rate": 1.7974367869509155e-05,
      "loss": 0.0906,
      "step": 64925
    },
    {
      "epoch": 0.956539668046126,
      "grad_norm": 1.4901723861694336,
      "learning_rate": 1.7972731497542172e-05,
      "loss": 0.0883,
      "step": 64950
    },
    {
      "epoch": 0.956907851136213,
      "grad_norm": 1.6070477962493896,
      "learning_rate": 1.7971095125575187e-05,
      "loss": 0.0933,
      "step": 64975
    },
    {
      "epoch": 0.9572760342263,
      "grad_norm": 1.7674330472946167,
      "learning_rate": 1.79694587536082e-05,
      "loss": 0.0873,
      "step": 65000
    },
    {
      "epoch": 0.957644217316387,
      "grad_norm": 1.4405423402786255,
      "learning_rate": 1.796782238164122e-05,
      "loss": 0.0852,
      "step": 65025
    },
    {
      "epoch": 0.9580124004064742,
      "grad_norm": 1.4775818586349487,
      "learning_rate": 1.7966186009674234e-05,
      "loss": 0.087,
      "step": 65050
    },
    {
      "epoch": 0.9583805834965612,
      "grad_norm": 1.1930218935012817,
      "learning_rate": 1.7964549637707248e-05,
      "loss": 0.0888,
      "step": 65075
    },
    {
      "epoch": 0.9587487665866482,
      "grad_norm": 1.4698896408081055,
      "learning_rate": 1.7962913265740262e-05,
      "loss": 0.0837,
      "step": 65100
    },
    {
      "epoch": 0.9591169496767352,
      "grad_norm": 1.6088793277740479,
      "learning_rate": 1.7961276893773277e-05,
      "loss": 0.0866,
      "step": 65125
    },
    {
      "epoch": 0.9594851327668222,
      "grad_norm": 1.456915259361267,
      "learning_rate": 1.7959640521806295e-05,
      "loss": 0.0943,
      "step": 65150
    },
    {
      "epoch": 0.9598533158569094,
      "grad_norm": 1.4380178451538086,
      "learning_rate": 1.795800414983931e-05,
      "loss": 0.0908,
      "step": 65175
    },
    {
      "epoch": 0.9602214989469964,
      "grad_norm": 1.2547969818115234,
      "learning_rate": 1.7956367777872327e-05,
      "loss": 0.0879,
      "step": 65200
    },
    {
      "epoch": 0.9605896820370834,
      "grad_norm": 1.3286045789718628,
      "learning_rate": 1.795473140590534e-05,
      "loss": 0.0828,
      "step": 65225
    },
    {
      "epoch": 0.9609578651271704,
      "grad_norm": 1.7631367444992065,
      "learning_rate": 1.7953095033938356e-05,
      "loss": 0.0906,
      "step": 65250
    },
    {
      "epoch": 0.9613260482172574,
      "grad_norm": 1.6213173866271973,
      "learning_rate": 1.795145866197137e-05,
      "loss": 0.096,
      "step": 65275
    },
    {
      "epoch": 0.9616942313073445,
      "grad_norm": 1.1831706762313843,
      "learning_rate": 1.7949887744883064e-05,
      "loss": 0.0909,
      "step": 65300
    },
    {
      "epoch": 0.9620624143974316,
      "grad_norm": 1.2344462871551514,
      "learning_rate": 1.7948251372916082e-05,
      "loss": 0.0874,
      "step": 65325
    },
    {
      "epoch": 0.9624305974875186,
      "grad_norm": 1.1486679315567017,
      "learning_rate": 1.7946615000949097e-05,
      "loss": 0.0888,
      "step": 65350
    },
    {
      "epoch": 0.9627987805776056,
      "grad_norm": 1.5928605794906616,
      "learning_rate": 1.794497862898211e-05,
      "loss": 0.0945,
      "step": 65375
    },
    {
      "epoch": 0.9631669636676927,
      "grad_norm": 1.2084425687789917,
      "learning_rate": 1.794334225701513e-05,
      "loss": 0.0838,
      "step": 65400
    },
    {
      "epoch": 0.9635351467577797,
      "grad_norm": 1.0286258459091187,
      "learning_rate": 1.7941705885048143e-05,
      "loss": 0.0815,
      "step": 65425
    },
    {
      "epoch": 0.9639033298478668,
      "grad_norm": 1.072543740272522,
      "learning_rate": 1.794006951308116e-05,
      "loss": 0.0802,
      "step": 65450
    },
    {
      "epoch": 0.9642715129379538,
      "grad_norm": 1.6432961225509644,
      "learning_rate": 1.7938433141114176e-05,
      "loss": 0.0818,
      "step": 65475
    },
    {
      "epoch": 0.9646396960280408,
      "grad_norm": 1.8096356391906738,
      "learning_rate": 1.793679676914719e-05,
      "loss": 0.0794,
      "step": 65500
    },
    {
      "epoch": 0.9650078791181279,
      "grad_norm": 1.352291226387024,
      "learning_rate": 1.7935160397180205e-05,
      "loss": 0.0908,
      "step": 65525
    },
    {
      "epoch": 0.9653760622082149,
      "grad_norm": 1.9157545566558838,
      "learning_rate": 1.793352402521322e-05,
      "loss": 0.0926,
      "step": 65550
    },
    {
      "epoch": 0.9657442452983019,
      "grad_norm": 1.394393801689148,
      "learning_rate": 1.7931887653246237e-05,
      "loss": 0.0832,
      "step": 65575
    },
    {
      "epoch": 0.966112428388389,
      "grad_norm": 1.070513129234314,
      "learning_rate": 1.793025128127925e-05,
      "loss": 0.0883,
      "step": 65600
    },
    {
      "epoch": 0.966480611478476,
      "grad_norm": 1.1968494653701782,
      "learning_rate": 1.7928614909312266e-05,
      "loss": 0.0914,
      "step": 65625
    },
    {
      "epoch": 0.9668487945685631,
      "grad_norm": 1.3151240348815918,
      "learning_rate": 1.7926978537345284e-05,
      "loss": 0.0809,
      "step": 65650
    },
    {
      "epoch": 0.9672169776586501,
      "grad_norm": 2.005269765853882,
      "learning_rate": 1.7925342165378298e-05,
      "loss": 0.08,
      "step": 65675
    },
    {
      "epoch": 0.9675851607487371,
      "grad_norm": 1.4291810989379883,
      "learning_rate": 1.7923705793411313e-05,
      "loss": 0.0909,
      "step": 65700
    },
    {
      "epoch": 0.9679533438388241,
      "grad_norm": 1.40546715259552,
      "learning_rate": 1.7922069421444327e-05,
      "loss": 0.0835,
      "step": 65725
    },
    {
      "epoch": 0.9683215269289112,
      "grad_norm": 1.1090855598449707,
      "learning_rate": 1.7920433049477345e-05,
      "loss": 0.0898,
      "step": 65750
    },
    {
      "epoch": 0.9686897100189983,
      "grad_norm": 1.6328848600387573,
      "learning_rate": 1.791879667751036e-05,
      "loss": 0.0844,
      "step": 65775
    },
    {
      "epoch": 0.9690578931090853,
      "grad_norm": 1.6949405670166016,
      "learning_rate": 1.7917160305543374e-05,
      "loss": 0.0921,
      "step": 65800
    },
    {
      "epoch": 0.9694260761991723,
      "grad_norm": 1.4263778924942017,
      "learning_rate": 1.791552393357639e-05,
      "loss": 0.091,
      "step": 65825
    },
    {
      "epoch": 0.9697942592892593,
      "grad_norm": 1.6204248666763306,
      "learning_rate": 1.7913887561609406e-05,
      "loss": 0.0946,
      "step": 65850
    },
    {
      "epoch": 0.9701624423793465,
      "grad_norm": 1.2282878160476685,
      "learning_rate": 1.791225118964242e-05,
      "loss": 0.0915,
      "step": 65875
    },
    {
      "epoch": 0.9705306254694335,
      "grad_norm": 1.5596424341201782,
      "learning_rate": 1.791061481767544e-05,
      "loss": 0.0858,
      "step": 65900
    },
    {
      "epoch": 0.9708988085595205,
      "grad_norm": 1.3790189027786255,
      "learning_rate": 1.7908978445708453e-05,
      "loss": 0.0818,
      "step": 65925
    },
    {
      "epoch": 0.9712669916496075,
      "grad_norm": 1.2927910089492798,
      "learning_rate": 1.7907342073741467e-05,
      "loss": 0.0882,
      "step": 65950
    },
    {
      "epoch": 0.9716351747396945,
      "grad_norm": 1.5115795135498047,
      "learning_rate": 1.7905705701774482e-05,
      "loss": 0.0839,
      "step": 65975
    },
    {
      "epoch": 0.9720033578297816,
      "grad_norm": 1.2746201753616333,
      "learning_rate": 1.79040693298075e-05,
      "loss": 0.0875,
      "step": 66000
    },
    {
      "epoch": 0.9723715409198687,
      "grad_norm": 1.374213695526123,
      "learning_rate": 1.7902432957840514e-05,
      "loss": 0.084,
      "step": 66025
    },
    {
      "epoch": 0.9727397240099557,
      "grad_norm": 1.3423460721969604,
      "learning_rate": 1.790079658587353e-05,
      "loss": 0.0831,
      "step": 66050
    },
    {
      "epoch": 0.9731079071000427,
      "grad_norm": 1.528391718864441,
      "learning_rate": 1.7899160213906546e-05,
      "loss": 0.093,
      "step": 66075
    },
    {
      "epoch": 0.9734760901901297,
      "grad_norm": 1.7538795471191406,
      "learning_rate": 1.789752384193956e-05,
      "loss": 0.0958,
      "step": 66100
    },
    {
      "epoch": 0.9738442732802168,
      "grad_norm": 1.7096960544586182,
      "learning_rate": 1.7895887469972575e-05,
      "loss": 0.0906,
      "step": 66125
    },
    {
      "epoch": 0.9742124563703038,
      "grad_norm": 0.8853684663772583,
      "learning_rate": 1.789425109800559e-05,
      "loss": 0.0814,
      "step": 66150
    },
    {
      "epoch": 0.9745806394603909,
      "grad_norm": 1.4343311786651611,
      "learning_rate": 1.7892614726038608e-05,
      "loss": 0.0886,
      "step": 66175
    },
    {
      "epoch": 0.9749488225504779,
      "grad_norm": 1.0935750007629395,
      "learning_rate": 1.7890978354071622e-05,
      "loss": 0.0778,
      "step": 66200
    },
    {
      "epoch": 0.9753170056405649,
      "grad_norm": 1.5821186304092407,
      "learning_rate": 1.7889341982104637e-05,
      "loss": 0.087,
      "step": 66225
    },
    {
      "epoch": 0.975685188730652,
      "grad_norm": 1.5097957849502563,
      "learning_rate": 1.7887705610137654e-05,
      "loss": 0.0859,
      "step": 66250
    },
    {
      "epoch": 0.976053371820739,
      "grad_norm": 1.6766763925552368,
      "learning_rate": 1.788606923817067e-05,
      "loss": 0.0918,
      "step": 66275
    },
    {
      "epoch": 0.976421554910826,
      "grad_norm": 1.2421185970306396,
      "learning_rate": 1.7884432866203683e-05,
      "loss": 0.088,
      "step": 66300
    },
    {
      "epoch": 0.9767897380009131,
      "grad_norm": 1.2647916078567505,
      "learning_rate": 1.78827964942367e-05,
      "loss": 0.0853,
      "step": 66325
    },
    {
      "epoch": 0.9771579210910001,
      "grad_norm": 1.9773876667022705,
      "learning_rate": 1.7881160122269716e-05,
      "loss": 0.0875,
      "step": 66350
    },
    {
      "epoch": 0.9775261041810872,
      "grad_norm": 1.2135019302368164,
      "learning_rate": 1.787952375030273e-05,
      "loss": 0.0867,
      "step": 66375
    },
    {
      "epoch": 0.9778942872711742,
      "grad_norm": 1.0465253591537476,
      "learning_rate": 1.7877887378335745e-05,
      "loss": 0.0843,
      "step": 66400
    },
    {
      "epoch": 0.9782624703612612,
      "grad_norm": 1.4281352758407593,
      "learning_rate": 1.7876251006368762e-05,
      "loss": 0.0912,
      "step": 66425
    },
    {
      "epoch": 0.9786306534513483,
      "grad_norm": 1.832780361175537,
      "learning_rate": 1.7874614634401777e-05,
      "loss": 0.0888,
      "step": 66450
    },
    {
      "epoch": 0.9789988365414354,
      "grad_norm": 1.7653343677520752,
      "learning_rate": 1.787297826243479e-05,
      "loss": 0.0858,
      "step": 66475
    },
    {
      "epoch": 0.9793670196315224,
      "grad_norm": 0.9594721794128418,
      "learning_rate": 1.787134189046781e-05,
      "loss": 0.0868,
      "step": 66500
    },
    {
      "epoch": 0.9797352027216094,
      "grad_norm": 1.0360686779022217,
      "learning_rate": 1.7869705518500824e-05,
      "loss": 0.0889,
      "step": 66525
    },
    {
      "epoch": 0.9801033858116964,
      "grad_norm": 1.1977566480636597,
      "learning_rate": 1.7868069146533838e-05,
      "loss": 0.0826,
      "step": 66550
    },
    {
      "epoch": 0.9804715689017834,
      "grad_norm": 1.925881266593933,
      "learning_rate": 1.7866432774566852e-05,
      "loss": 0.0841,
      "step": 66575
    },
    {
      "epoch": 0.9808397519918706,
      "grad_norm": 1.0635563135147095,
      "learning_rate": 1.786479640259987e-05,
      "loss": 0.0815,
      "step": 66600
    },
    {
      "epoch": 0.9812079350819576,
      "grad_norm": 1.1311936378479004,
      "learning_rate": 1.7863160030632885e-05,
      "loss": 0.089,
      "step": 66625
    },
    {
      "epoch": 0.9815761181720446,
      "grad_norm": 1.32826828956604,
      "learning_rate": 1.78615236586659e-05,
      "loss": 0.0815,
      "step": 66650
    },
    {
      "epoch": 0.9819443012621316,
      "grad_norm": 1.7389458417892456,
      "learning_rate": 1.7859887286698917e-05,
      "loss": 0.0937,
      "step": 66675
    },
    {
      "epoch": 0.9823124843522186,
      "grad_norm": 1.0361177921295166,
      "learning_rate": 1.785825091473193e-05,
      "loss": 0.0806,
      "step": 66700
    },
    {
      "epoch": 0.9826806674423058,
      "grad_norm": 2.10971736907959,
      "learning_rate": 1.7856614542764946e-05,
      "loss": 0.0875,
      "step": 66725
    },
    {
      "epoch": 0.9830488505323928,
      "grad_norm": 1.6492832899093628,
      "learning_rate": 1.785497817079796e-05,
      "loss": 0.0873,
      "step": 66750
    },
    {
      "epoch": 0.9834170336224798,
      "grad_norm": 1.559989333152771,
      "learning_rate": 1.7853341798830975e-05,
      "loss": 0.0802,
      "step": 66775
    },
    {
      "epoch": 0.9837852167125668,
      "grad_norm": 1.7486131191253662,
      "learning_rate": 1.7851705426863993e-05,
      "loss": 0.0851,
      "step": 66800
    },
    {
      "epoch": 0.9841533998026538,
      "grad_norm": 1.6348661184310913,
      "learning_rate": 1.7850069054897007e-05,
      "loss": 0.0916,
      "step": 66825
    },
    {
      "epoch": 0.9845215828927409,
      "grad_norm": 1.8323534727096558,
      "learning_rate": 1.7848432682930025e-05,
      "loss": 0.083,
      "step": 66850
    },
    {
      "epoch": 0.984889765982828,
      "grad_norm": 1.3669302463531494,
      "learning_rate": 1.784679631096304e-05,
      "loss": 0.0842,
      "step": 66875
    },
    {
      "epoch": 0.985257949072915,
      "grad_norm": 0.9307335019111633,
      "learning_rate": 1.7845159938996054e-05,
      "loss": 0.086,
      "step": 66900
    },
    {
      "epoch": 0.985626132163002,
      "grad_norm": 1.380886435508728,
      "learning_rate": 1.7843523567029072e-05,
      "loss": 0.0897,
      "step": 66925
    },
    {
      "epoch": 0.985994315253089,
      "grad_norm": 1.7508845329284668,
      "learning_rate": 1.7841887195062086e-05,
      "loss": 0.0929,
      "step": 66950
    },
    {
      "epoch": 0.9863624983431761,
      "grad_norm": 1.377454161643982,
      "learning_rate": 1.78402508230951e-05,
      "loss": 0.0928,
      "step": 66975
    },
    {
      "epoch": 0.9867306814332631,
      "grad_norm": 1.450796127319336,
      "learning_rate": 1.7838614451128115e-05,
      "loss": 0.0774,
      "step": 67000
    },
    {
      "epoch": 0.9870988645233502,
      "grad_norm": 1.3448824882507324,
      "learning_rate": 1.783697807916113e-05,
      "loss": 0.0833,
      "step": 67025
    },
    {
      "epoch": 0.9874670476134372,
      "grad_norm": 1.3844190835952759,
      "learning_rate": 1.7835341707194148e-05,
      "loss": 0.0846,
      "step": 67050
    },
    {
      "epoch": 0.9878352307035243,
      "grad_norm": 1.8527183532714844,
      "learning_rate": 1.7833705335227162e-05,
      "loss": 0.0908,
      "step": 67075
    },
    {
      "epoch": 0.9882034137936113,
      "grad_norm": 1.5654221773147583,
      "learning_rate": 1.783206896326018e-05,
      "loss": 0.0989,
      "step": 67100
    },
    {
      "epoch": 0.9885715968836983,
      "grad_norm": 1.2591010332107544,
      "learning_rate": 1.7830432591293194e-05,
      "loss": 0.0745,
      "step": 67125
    },
    {
      "epoch": 0.9889397799737853,
      "grad_norm": 1.2417330741882324,
      "learning_rate": 1.782879621932621e-05,
      "loss": 0.0917,
      "step": 67150
    },
    {
      "epoch": 0.9893079630638724,
      "grad_norm": 1.194953441619873,
      "learning_rate": 1.7827159847359223e-05,
      "loss": 0.087,
      "step": 67175
    },
    {
      "epoch": 0.9896761461539595,
      "grad_norm": 1.2350224256515503,
      "learning_rate": 1.7825523475392238e-05,
      "loss": 0.0908,
      "step": 67200
    },
    {
      "epoch": 0.9900443292440465,
      "grad_norm": 1.223141074180603,
      "learning_rate": 1.7823887103425255e-05,
      "loss": 0.0784,
      "step": 67225
    },
    {
      "epoch": 0.9904125123341335,
      "grad_norm": 1.784192442893982,
      "learning_rate": 1.782225073145827e-05,
      "loss": 0.0864,
      "step": 67250
    },
    {
      "epoch": 0.9907806954242205,
      "grad_norm": 1.5929571390151978,
      "learning_rate": 1.7820614359491288e-05,
      "loss": 0.0861,
      "step": 67275
    },
    {
      "epoch": 0.9911488785143076,
      "grad_norm": 1.3482542037963867,
      "learning_rate": 1.7818977987524302e-05,
      "loss": 0.0824,
      "step": 67300
    },
    {
      "epoch": 0.9915170616043947,
      "grad_norm": 1.104810118675232,
      "learning_rate": 1.7817341615557317e-05,
      "loss": 0.0865,
      "step": 67325
    },
    {
      "epoch": 0.9918852446944817,
      "grad_norm": 1.4847999811172485,
      "learning_rate": 1.7815705243590335e-05,
      "loss": 0.0799,
      "step": 67350
    },
    {
      "epoch": 0.9922534277845687,
      "grad_norm": 1.5398287773132324,
      "learning_rate": 1.781406887162335e-05,
      "loss": 0.0859,
      "step": 67375
    },
    {
      "epoch": 0.9926216108746557,
      "grad_norm": 1.618699312210083,
      "learning_rate": 1.7812432499656363e-05,
      "loss": 0.0857,
      "step": 67400
    },
    {
      "epoch": 0.9929897939647427,
      "grad_norm": 1.1916437149047852,
      "learning_rate": 1.7810796127689378e-05,
      "loss": 0.0919,
      "step": 67425
    },
    {
      "epoch": 0.9933579770548299,
      "grad_norm": 1.6624246835708618,
      "learning_rate": 1.7809159755722392e-05,
      "loss": 0.0919,
      "step": 67450
    },
    {
      "epoch": 0.9937261601449169,
      "grad_norm": 1.8165644407272339,
      "learning_rate": 1.780752338375541e-05,
      "loss": 0.0928,
      "step": 67475
    },
    {
      "epoch": 0.9940943432350039,
      "grad_norm": 1.5858564376831055,
      "learning_rate": 1.7805887011788425e-05,
      "loss": 0.0841,
      "step": 67500
    },
    {
      "epoch": 0.9944625263250909,
      "grad_norm": 2.407370090484619,
      "learning_rate": 1.7804316094700122e-05,
      "loss": 0.085,
      "step": 67525
    },
    {
      "epoch": 0.9948307094151779,
      "grad_norm": 1.5399054288864136,
      "learning_rate": 1.7802679722733136e-05,
      "loss": 0.0854,
      "step": 67550
    },
    {
      "epoch": 0.995198892505265,
      "grad_norm": 1.4991439580917358,
      "learning_rate": 1.780104335076615e-05,
      "loss": 0.0901,
      "step": 67575
    },
    {
      "epoch": 0.9955670755953521,
      "grad_norm": 1.1912407875061035,
      "learning_rate": 1.7799406978799165e-05,
      "loss": 0.0825,
      "step": 67600
    },
    {
      "epoch": 0.9959352586854391,
      "grad_norm": 1.625110387802124,
      "learning_rate": 1.779777060683218e-05,
      "loss": 0.0841,
      "step": 67625
    },
    {
      "epoch": 0.9963034417755261,
      "grad_norm": 1.3171745538711548,
      "learning_rate": 1.7796134234865198e-05,
      "loss": 0.0845,
      "step": 67650
    },
    {
      "epoch": 0.9966716248656132,
      "grad_norm": 1.5322037935256958,
      "learning_rate": 1.7794497862898212e-05,
      "loss": 0.0898,
      "step": 67675
    },
    {
      "epoch": 0.9970398079557002,
      "grad_norm": 1.6718095541000366,
      "learning_rate": 1.7792861490931227e-05,
      "loss": 0.0889,
      "step": 67700
    },
    {
      "epoch": 0.9974079910457873,
      "grad_norm": 1.542233943939209,
      "learning_rate": 1.7791225118964244e-05,
      "loss": 0.0843,
      "step": 67725
    },
    {
      "epoch": 0.9977761741358743,
      "grad_norm": 0.9952422976493835,
      "learning_rate": 1.778958874699726e-05,
      "loss": 0.0787,
      "step": 67750
    },
    {
      "epoch": 0.9981443572259613,
      "grad_norm": 1.5124484300613403,
      "learning_rate": 1.7787952375030277e-05,
      "loss": 0.0894,
      "step": 67775
    },
    {
      "epoch": 0.9985125403160484,
      "grad_norm": 1.3759278059005737,
      "learning_rate": 1.778631600306329e-05,
      "loss": 0.0848,
      "step": 67800
    },
    {
      "epoch": 0.9988807234061354,
      "grad_norm": 1.011905550956726,
      "learning_rate": 1.7784679631096306e-05,
      "loss": 0.0884,
      "step": 67825
    },
    {
      "epoch": 0.9992489064962224,
      "grad_norm": 1.3664641380310059,
      "learning_rate": 1.778304325912932e-05,
      "loss": 0.0914,
      "step": 67850
    },
    {
      "epoch": 0.9996170895863095,
      "grad_norm": 1.09990656375885,
      "learning_rate": 1.7781406887162335e-05,
      "loss": 0.0857,
      "step": 67875
    },
    {
      "epoch": 0.9999852726763965,
      "grad_norm": 1.1419481039047241,
      "learning_rate": 1.7779770515195352e-05,
      "loss": 0.0888,
      "step": 67900
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.0816650465130806,
      "eval_runtime": 118.5794,
      "eval_samples_per_second": 2991.651,
      "eval_steps_per_second": 5.844,
      "step": 67901
    },
    {
      "epoch": 1.0003534557664835,
      "grad_norm": 1.6040254831314087,
      "learning_rate": 1.7778134143228367e-05,
      "loss": 0.0793,
      "step": 67925
    },
    {
      "epoch": 1.0007216388565705,
      "grad_norm": 1.2348227500915527,
      "learning_rate": 1.777649777126138e-05,
      "loss": 0.0798,
      "step": 67950
    },
    {
      "epoch": 1.0010898219466577,
      "grad_norm": 1.301466703414917,
      "learning_rate": 1.77748613992944e-05,
      "loss": 0.0873,
      "step": 67975
    },
    {
      "epoch": 1.0014580050367448,
      "grad_norm": 1.350974202156067,
      "learning_rate": 1.7773225027327414e-05,
      "loss": 0.0802,
      "step": 68000
    },
    {
      "epoch": 1.0018261881268318,
      "grad_norm": 1.2824699878692627,
      "learning_rate": 1.7771588655360428e-05,
      "loss": 0.0835,
      "step": 68025
    },
    {
      "epoch": 1.0021943712169188,
      "grad_norm": 1.5551531314849854,
      "learning_rate": 1.7769952283393442e-05,
      "loss": 0.0801,
      "step": 68050
    },
    {
      "epoch": 1.0025625543070058,
      "grad_norm": 1.713180422782898,
      "learning_rate": 1.776831591142646e-05,
      "loss": 0.0776,
      "step": 68075
    },
    {
      "epoch": 1.0029307373970928,
      "grad_norm": 1.6707239151000977,
      "learning_rate": 1.7766679539459475e-05,
      "loss": 0.0783,
      "step": 68100
    },
    {
      "epoch": 1.0032989204871798,
      "grad_norm": 1.5839450359344482,
      "learning_rate": 1.776504316749249e-05,
      "loss": 0.0825,
      "step": 68125
    },
    {
      "epoch": 1.0036671035772669,
      "grad_norm": 1.555139422416687,
      "learning_rate": 1.7763406795525507e-05,
      "loss": 0.0818,
      "step": 68150
    },
    {
      "epoch": 1.0040352866673539,
      "grad_norm": 1.2419105768203735,
      "learning_rate": 1.776177042355852e-05,
      "loss": 0.088,
      "step": 68175
    },
    {
      "epoch": 1.0044034697574409,
      "grad_norm": 1.6057548522949219,
      "learning_rate": 1.7760134051591536e-05,
      "loss": 0.08,
      "step": 68200
    },
    {
      "epoch": 1.0047716528475281,
      "grad_norm": 1.7384859323501587,
      "learning_rate": 1.775849767962455e-05,
      "loss": 0.09,
      "step": 68225
    },
    {
      "epoch": 1.0051398359376151,
      "grad_norm": 1.2435907125473022,
      "learning_rate": 1.775686130765757e-05,
      "loss": 0.0886,
      "step": 68250
    },
    {
      "epoch": 1.0055080190277021,
      "grad_norm": 1.2714653015136719,
      "learning_rate": 1.7755224935690583e-05,
      "loss": 0.0875,
      "step": 68275
    },
    {
      "epoch": 1.0058762021177892,
      "grad_norm": 1.2328994274139404,
      "learning_rate": 1.7753588563723597e-05,
      "loss": 0.0823,
      "step": 68300
    },
    {
      "epoch": 1.0062443852078762,
      "grad_norm": 1.724279522895813,
      "learning_rate": 1.7751952191756615e-05,
      "loss": 0.0854,
      "step": 68325
    },
    {
      "epoch": 1.0066125682979632,
      "grad_norm": 1.153212547302246,
      "learning_rate": 1.775031581978963e-05,
      "loss": 0.0863,
      "step": 68350
    },
    {
      "epoch": 1.0069807513880502,
      "grad_norm": 2.0599777698516846,
      "learning_rate": 1.7748679447822644e-05,
      "loss": 0.0938,
      "step": 68375
    },
    {
      "epoch": 1.0073489344781372,
      "grad_norm": 1.7947320938110352,
      "learning_rate": 1.7747043075855662e-05,
      "loss": 0.0927,
      "step": 68400
    },
    {
      "epoch": 1.0077171175682242,
      "grad_norm": 1.0751878023147583,
      "learning_rate": 1.7745406703888676e-05,
      "loss": 0.0726,
      "step": 68425
    },
    {
      "epoch": 1.0080853006583113,
      "grad_norm": 1.325618863105774,
      "learning_rate": 1.774377033192169e-05,
      "loss": 0.0879,
      "step": 68450
    },
    {
      "epoch": 1.0084534837483985,
      "grad_norm": 1.3592146635055542,
      "learning_rate": 1.7742133959954705e-05,
      "loss": 0.0828,
      "step": 68475
    },
    {
      "epoch": 1.0088216668384855,
      "grad_norm": 1.5241844654083252,
      "learning_rate": 1.7740497587987723e-05,
      "loss": 0.0823,
      "step": 68500
    },
    {
      "epoch": 1.0091898499285725,
      "grad_norm": 1.4818546772003174,
      "learning_rate": 1.7738861216020738e-05,
      "loss": 0.0853,
      "step": 68525
    },
    {
      "epoch": 1.0095580330186595,
      "grad_norm": 1.7624226808547974,
      "learning_rate": 1.7737224844053752e-05,
      "loss": 0.0855,
      "step": 68550
    },
    {
      "epoch": 1.0099262161087466,
      "grad_norm": 1.298776388168335,
      "learning_rate": 1.773558847208677e-05,
      "loss": 0.0828,
      "step": 68575
    },
    {
      "epoch": 1.0102943991988336,
      "grad_norm": 1.4680594205856323,
      "learning_rate": 1.7733952100119784e-05,
      "loss": 0.0848,
      "step": 68600
    },
    {
      "epoch": 1.0106625822889206,
      "grad_norm": 1.138354778289795,
      "learning_rate": 1.77323157281528e-05,
      "loss": 0.0773,
      "step": 68625
    },
    {
      "epoch": 1.0110307653790076,
      "grad_norm": 1.0077418088912964,
      "learning_rate": 1.7730679356185813e-05,
      "loss": 0.0752,
      "step": 68650
    },
    {
      "epoch": 1.0113989484690946,
      "grad_norm": 1.3636548519134521,
      "learning_rate": 1.772904298421883e-05,
      "loss": 0.0873,
      "step": 68675
    },
    {
      "epoch": 1.0117671315591819,
      "grad_norm": 1.248867154121399,
      "learning_rate": 1.7727406612251845e-05,
      "loss": 0.0876,
      "step": 68700
    },
    {
      "epoch": 1.0121353146492689,
      "grad_norm": 1.4871373176574707,
      "learning_rate": 1.772577024028486e-05,
      "loss": 0.0833,
      "step": 68725
    },
    {
      "epoch": 1.0125034977393559,
      "grad_norm": 1.800140142440796,
      "learning_rate": 1.7724133868317878e-05,
      "loss": 0.0859,
      "step": 68750
    },
    {
      "epoch": 1.012871680829443,
      "grad_norm": 0.9257559180259705,
      "learning_rate": 1.7722497496350892e-05,
      "loss": 0.08,
      "step": 68775
    },
    {
      "epoch": 1.01323986391953,
      "grad_norm": 1.123623013496399,
      "learning_rate": 1.7720861124383907e-05,
      "loss": 0.0806,
      "step": 68800
    },
    {
      "epoch": 1.013608047009617,
      "grad_norm": 1.6497175693511963,
      "learning_rate": 1.7719224752416925e-05,
      "loss": 0.0835,
      "step": 68825
    },
    {
      "epoch": 1.013976230099704,
      "grad_norm": 1.2564496994018555,
      "learning_rate": 1.771758838044994e-05,
      "loss": 0.081,
      "step": 68850
    },
    {
      "epoch": 1.014344413189791,
      "grad_norm": 1.5004905462265015,
      "learning_rate": 1.7715952008482953e-05,
      "loss": 0.0909,
      "step": 68875
    },
    {
      "epoch": 1.014712596279878,
      "grad_norm": 1.2909660339355469,
      "learning_rate": 1.7714315636515968e-05,
      "loss": 0.0894,
      "step": 68900
    },
    {
      "epoch": 1.015080779369965,
      "grad_norm": 1.0051658153533936,
      "learning_rate": 1.7712679264548986e-05,
      "loss": 0.0817,
      "step": 68925
    },
    {
      "epoch": 1.0154489624600522,
      "grad_norm": 1.2211277484893799,
      "learning_rate": 1.7711042892582e-05,
      "loss": 0.0824,
      "step": 68950
    },
    {
      "epoch": 1.0158171455501392,
      "grad_norm": 1.0578739643096924,
      "learning_rate": 1.7709406520615015e-05,
      "loss": 0.0795,
      "step": 68975
    },
    {
      "epoch": 1.0161853286402263,
      "grad_norm": 1.3825381994247437,
      "learning_rate": 1.7707770148648033e-05,
      "loss": 0.0873,
      "step": 69000
    },
    {
      "epoch": 1.0165535117303133,
      "grad_norm": 1.2576650381088257,
      "learning_rate": 1.7706133776681047e-05,
      "loss": 0.0814,
      "step": 69025
    },
    {
      "epoch": 1.0169216948204003,
      "grad_norm": 1.559657335281372,
      "learning_rate": 1.770449740471406e-05,
      "loss": 0.0871,
      "step": 69050
    },
    {
      "epoch": 1.0172898779104873,
      "grad_norm": 1.4857251644134521,
      "learning_rate": 1.7702861032747076e-05,
      "loss": 0.0843,
      "step": 69075
    },
    {
      "epoch": 1.0176580610005743,
      "grad_norm": 1.4493496417999268,
      "learning_rate": 1.770122466078009e-05,
      "loss": 0.0783,
      "step": 69100
    },
    {
      "epoch": 1.0180262440906613,
      "grad_norm": 1.2584834098815918,
      "learning_rate": 1.7699588288813108e-05,
      "loss": 0.0825,
      "step": 69125
    },
    {
      "epoch": 1.0183944271807484,
      "grad_norm": 1.5866835117340088,
      "learning_rate": 1.7697951916846123e-05,
      "loss": 0.0836,
      "step": 69150
    },
    {
      "epoch": 1.0187626102708356,
      "grad_norm": 1.3111428022384644,
      "learning_rate": 1.769631554487914e-05,
      "loss": 0.0884,
      "step": 69175
    },
    {
      "epoch": 1.0191307933609226,
      "grad_norm": 1.3243120908737183,
      "learning_rate": 1.7694679172912155e-05,
      "loss": 0.0822,
      "step": 69200
    },
    {
      "epoch": 1.0194989764510096,
      "grad_norm": 1.3820977210998535,
      "learning_rate": 1.769304280094517e-05,
      "loss": 0.0844,
      "step": 69225
    },
    {
      "epoch": 1.0198671595410966,
      "grad_norm": 1.5885424613952637,
      "learning_rate": 1.7691406428978187e-05,
      "loss": 0.083,
      "step": 69250
    },
    {
      "epoch": 1.0202353426311837,
      "grad_norm": 1.6160551309585571,
      "learning_rate": 1.7689770057011202e-05,
      "loss": 0.0848,
      "step": 69275
    },
    {
      "epoch": 1.0206035257212707,
      "grad_norm": 1.4503554105758667,
      "learning_rate": 1.7688133685044216e-05,
      "loss": 0.0839,
      "step": 69300
    },
    {
      "epoch": 1.0209717088113577,
      "grad_norm": 1.1684181690216064,
      "learning_rate": 1.768649731307723e-05,
      "loss": 0.087,
      "step": 69325
    },
    {
      "epoch": 1.0213398919014447,
      "grad_norm": 1.3307864665985107,
      "learning_rate": 1.7684860941110245e-05,
      "loss": 0.0859,
      "step": 69350
    },
    {
      "epoch": 1.0217080749915317,
      "grad_norm": 1.3071322441101074,
      "learning_rate": 1.7683224569143263e-05,
      "loss": 0.0763,
      "step": 69375
    },
    {
      "epoch": 1.0220762580816187,
      "grad_norm": 1.0993708372116089,
      "learning_rate": 1.7681588197176277e-05,
      "loss": 0.0864,
      "step": 69400
    },
    {
      "epoch": 1.022444441171706,
      "grad_norm": 1.5368435382843018,
      "learning_rate": 1.7679951825209295e-05,
      "loss": 0.0864,
      "step": 69425
    },
    {
      "epoch": 1.022812624261793,
      "grad_norm": 1.384029507637024,
      "learning_rate": 1.767831545324231e-05,
      "loss": 0.089,
      "step": 69450
    },
    {
      "epoch": 1.02318080735188,
      "grad_norm": 1.5836247205734253,
      "learning_rate": 1.7676679081275324e-05,
      "loss": 0.0772,
      "step": 69475
    },
    {
      "epoch": 1.023548990441967,
      "grad_norm": 1.1067153215408325,
      "learning_rate": 1.767504270930834e-05,
      "loss": 0.085,
      "step": 69500
    },
    {
      "epoch": 1.023917173532054,
      "grad_norm": 1.1940453052520752,
      "learning_rate": 1.7673406337341353e-05,
      "loss": 0.0732,
      "step": 69525
    },
    {
      "epoch": 1.024285356622141,
      "grad_norm": 1.5667295455932617,
      "learning_rate": 1.767176996537437e-05,
      "loss": 0.0862,
      "step": 69550
    },
    {
      "epoch": 1.024653539712228,
      "grad_norm": 1.4569140672683716,
      "learning_rate": 1.7670133593407385e-05,
      "loss": 0.0738,
      "step": 69575
    },
    {
      "epoch": 1.025021722802315,
      "grad_norm": 1.644558072090149,
      "learning_rate": 1.7668497221440403e-05,
      "loss": 0.0834,
      "step": 69600
    },
    {
      "epoch": 1.025389905892402,
      "grad_norm": 1.1172680854797363,
      "learning_rate": 1.7666860849473418e-05,
      "loss": 0.0767,
      "step": 69625
    },
    {
      "epoch": 1.025758088982489,
      "grad_norm": 1.9643030166625977,
      "learning_rate": 1.7665224477506432e-05,
      "loss": 0.0872,
      "step": 69650
    },
    {
      "epoch": 1.0261262720725763,
      "grad_norm": 1.4155508279800415,
      "learning_rate": 1.766358810553945e-05,
      "loss": 0.0851,
      "step": 69675
    },
    {
      "epoch": 1.0264944551626634,
      "grad_norm": 1.1935136318206787,
      "learning_rate": 1.766195173357246e-05,
      "loss": 0.0912,
      "step": 69700
    },
    {
      "epoch": 1.0268626382527504,
      "grad_norm": 1.1054178476333618,
      "learning_rate": 1.766031536160548e-05,
      "loss": 0.0901,
      "step": 69725
    },
    {
      "epoch": 1.0272308213428374,
      "grad_norm": 1.4529372453689575,
      "learning_rate": 1.7658678989638493e-05,
      "loss": 0.0866,
      "step": 69750
    },
    {
      "epoch": 1.0275990044329244,
      "grad_norm": 1.525184988975525,
      "learning_rate": 1.7657042617671508e-05,
      "loss": 0.0797,
      "step": 69775
    },
    {
      "epoch": 1.0279671875230114,
      "grad_norm": 1.60329008102417,
      "learning_rate": 1.7655406245704526e-05,
      "loss": 0.0835,
      "step": 69800
    },
    {
      "epoch": 1.0283353706130984,
      "grad_norm": 1.1393709182739258,
      "learning_rate": 1.765376987373754e-05,
      "loss": 0.0791,
      "step": 69825
    },
    {
      "epoch": 1.0287035537031854,
      "grad_norm": 1.6377661228179932,
      "learning_rate": 1.7652133501770558e-05,
      "loss": 0.0926,
      "step": 69850
    },
    {
      "epoch": 1.0290717367932725,
      "grad_norm": 1.4320484399795532,
      "learning_rate": 1.7650497129803572e-05,
      "loss": 0.0875,
      "step": 69875
    },
    {
      "epoch": 1.0294399198833597,
      "grad_norm": 1.3752083778381348,
      "learning_rate": 1.7648860757836587e-05,
      "loss": 0.0879,
      "step": 69900
    },
    {
      "epoch": 1.0298081029734467,
      "grad_norm": 1.7365537881851196,
      "learning_rate": 1.76472243858696e-05,
      "loss": 0.0893,
      "step": 69925
    },
    {
      "epoch": 1.0301762860635337,
      "grad_norm": 1.254590392112732,
      "learning_rate": 1.7645588013902616e-05,
      "loss": 0.0884,
      "step": 69950
    },
    {
      "epoch": 1.0305444691536207,
      "grad_norm": 1.0306187868118286,
      "learning_rate": 1.7643951641935634e-05,
      "loss": 0.0848,
      "step": 69975
    },
    {
      "epoch": 1.0309126522437078,
      "grad_norm": 1.7979035377502441,
      "learning_rate": 1.7642315269968648e-05,
      "loss": 0.0847,
      "step": 70000
    },
    {
      "epoch": 1.0312808353337948,
      "grad_norm": 1.0098897218704224,
      "learning_rate": 1.7640678898001663e-05,
      "loss": 0.0787,
      "step": 70025
    },
    {
      "epoch": 1.0316490184238818,
      "grad_norm": 1.1822419166564941,
      "learning_rate": 1.763904252603468e-05,
      "loss": 0.0811,
      "step": 70050
    },
    {
      "epoch": 1.0320172015139688,
      "grad_norm": 1.4940446615219116,
      "learning_rate": 1.7637406154067695e-05,
      "loss": 0.0788,
      "step": 70075
    },
    {
      "epoch": 1.0323853846040558,
      "grad_norm": 1.5025551319122314,
      "learning_rate": 1.7635769782100713e-05,
      "loss": 0.0937,
      "step": 70100
    },
    {
      "epoch": 1.0327535676941428,
      "grad_norm": 1.3301732540130615,
      "learning_rate": 1.7634133410133724e-05,
      "loss": 0.0884,
      "step": 70125
    },
    {
      "epoch": 1.03312175078423,
      "grad_norm": 1.4251512289047241,
      "learning_rate": 1.763249703816674e-05,
      "loss": 0.0762,
      "step": 70150
    },
    {
      "epoch": 1.033489933874317,
      "grad_norm": 1.9587931632995605,
      "learning_rate": 1.7630926121078435e-05,
      "loss": 0.0818,
      "step": 70175
    },
    {
      "epoch": 1.033858116964404,
      "grad_norm": 1.5725241899490356,
      "learning_rate": 1.762928974911145e-05,
      "loss": 0.0823,
      "step": 70200
    },
    {
      "epoch": 1.0342263000544911,
      "grad_norm": 1.236335277557373,
      "learning_rate": 1.7627653377144468e-05,
      "loss": 0.0823,
      "step": 70225
    },
    {
      "epoch": 1.0345944831445781,
      "grad_norm": 1.218507170677185,
      "learning_rate": 1.7626017005177482e-05,
      "loss": 0.0901,
      "step": 70250
    },
    {
      "epoch": 1.0349626662346652,
      "grad_norm": 1.6564546823501587,
      "learning_rate": 1.7624380633210497e-05,
      "loss": 0.0835,
      "step": 70275
    },
    {
      "epoch": 1.0353308493247522,
      "grad_norm": 1.8725467920303345,
      "learning_rate": 1.7622744261243515e-05,
      "loss": 0.0946,
      "step": 70300
    },
    {
      "epoch": 1.0356990324148392,
      "grad_norm": 1.1960952281951904,
      "learning_rate": 1.762110788927653e-05,
      "loss": 0.0816,
      "step": 70325
    },
    {
      "epoch": 1.0360672155049262,
      "grad_norm": 2.004316568374634,
      "learning_rate": 1.7619471517309543e-05,
      "loss": 0.0818,
      "step": 70350
    },
    {
      "epoch": 1.0364353985950134,
      "grad_norm": 1.5343608856201172,
      "learning_rate": 1.7617835145342558e-05,
      "loss": 0.0787,
      "step": 70375
    },
    {
      "epoch": 1.0368035816851005,
      "grad_norm": 1.4142705202102661,
      "learning_rate": 1.7616198773375576e-05,
      "loss": 0.0727,
      "step": 70400
    },
    {
      "epoch": 1.0371717647751875,
      "grad_norm": 1.3772735595703125,
      "learning_rate": 1.761456240140859e-05,
      "loss": 0.0798,
      "step": 70425
    },
    {
      "epoch": 1.0375399478652745,
      "grad_norm": 1.4325580596923828,
      "learning_rate": 1.7612926029441605e-05,
      "loss": 0.0852,
      "step": 70450
    },
    {
      "epoch": 1.0379081309553615,
      "grad_norm": 2.0144031047821045,
      "learning_rate": 1.7611289657474623e-05,
      "loss": 0.0864,
      "step": 70475
    },
    {
      "epoch": 1.0382763140454485,
      "grad_norm": 1.111639142036438,
      "learning_rate": 1.7609653285507637e-05,
      "loss": 0.0891,
      "step": 70500
    },
    {
      "epoch": 1.0386444971355355,
      "grad_norm": 1.2016462087631226,
      "learning_rate": 1.760801691354065e-05,
      "loss": 0.0804,
      "step": 70525
    },
    {
      "epoch": 1.0390126802256225,
      "grad_norm": 1.6934276819229126,
      "learning_rate": 1.7606380541573666e-05,
      "loss": 0.0845,
      "step": 70550
    },
    {
      "epoch": 1.0393808633157096,
      "grad_norm": 1.9183427095413208,
      "learning_rate": 1.7604744169606684e-05,
      "loss": 0.0774,
      "step": 70575
    },
    {
      "epoch": 1.0397490464057966,
      "grad_norm": 1.5269534587860107,
      "learning_rate": 1.7603107797639698e-05,
      "loss": 0.0774,
      "step": 70600
    },
    {
      "epoch": 1.0401172294958838,
      "grad_norm": 1.3354527950286865,
      "learning_rate": 1.7601471425672713e-05,
      "loss": 0.0883,
      "step": 70625
    },
    {
      "epoch": 1.0404854125859708,
      "grad_norm": 1.8077341318130493,
      "learning_rate": 1.759983505370573e-05,
      "loss": 0.0903,
      "step": 70650
    },
    {
      "epoch": 1.0408535956760578,
      "grad_norm": 1.3553496599197388,
      "learning_rate": 1.7598198681738745e-05,
      "loss": 0.0838,
      "step": 70675
    },
    {
      "epoch": 1.0412217787661449,
      "grad_norm": 1.5021089315414429,
      "learning_rate": 1.759656230977176e-05,
      "loss": 0.0766,
      "step": 70700
    },
    {
      "epoch": 1.0415899618562319,
      "grad_norm": 1.4155638217926025,
      "learning_rate": 1.7594925937804777e-05,
      "loss": 0.0817,
      "step": 70725
    },
    {
      "epoch": 1.0419581449463189,
      "grad_norm": 1.2288072109222412,
      "learning_rate": 1.7593289565837792e-05,
      "loss": 0.0798,
      "step": 70750
    },
    {
      "epoch": 1.042326328036406,
      "grad_norm": 0.9162379503250122,
      "learning_rate": 1.7591653193870806e-05,
      "loss": 0.0872,
      "step": 70775
    },
    {
      "epoch": 1.042694511126493,
      "grad_norm": 1.4179303646087646,
      "learning_rate": 1.759001682190382e-05,
      "loss": 0.0843,
      "step": 70800
    },
    {
      "epoch": 1.04306269421658,
      "grad_norm": 1.0597056150436401,
      "learning_rate": 1.758838044993684e-05,
      "loss": 0.0728,
      "step": 70825
    },
    {
      "epoch": 1.0434308773066672,
      "grad_norm": 1.3661811351776123,
      "learning_rate": 1.7586744077969853e-05,
      "loss": 0.0889,
      "step": 70850
    },
    {
      "epoch": 1.0437990603967542,
      "grad_norm": 1.0863116979599,
      "learning_rate": 1.7585107706002867e-05,
      "loss": 0.0838,
      "step": 70875
    },
    {
      "epoch": 1.0441672434868412,
      "grad_norm": 1.4826573133468628,
      "learning_rate": 1.7583471334035885e-05,
      "loss": 0.0828,
      "step": 70900
    },
    {
      "epoch": 1.0445354265769282,
      "grad_norm": 1.5765396356582642,
      "learning_rate": 1.75818349620689e-05,
      "loss": 0.0752,
      "step": 70925
    },
    {
      "epoch": 1.0449036096670152,
      "grad_norm": 1.3515713214874268,
      "learning_rate": 1.7580198590101914e-05,
      "loss": 0.0849,
      "step": 70950
    },
    {
      "epoch": 1.0452717927571022,
      "grad_norm": 1.5682932138442993,
      "learning_rate": 1.757856221813493e-05,
      "loss": 0.0897,
      "step": 70975
    },
    {
      "epoch": 1.0456399758471893,
      "grad_norm": 1.2009916305541992,
      "learning_rate": 1.7576925846167946e-05,
      "loss": 0.0829,
      "step": 71000
    },
    {
      "epoch": 1.0460081589372763,
      "grad_norm": 1.3471386432647705,
      "learning_rate": 1.757528947420096e-05,
      "loss": 0.0868,
      "step": 71025
    },
    {
      "epoch": 1.0463763420273633,
      "grad_norm": 1.3757349252700806,
      "learning_rate": 1.7573653102233975e-05,
      "loss": 0.0831,
      "step": 71050
    },
    {
      "epoch": 1.0467445251174503,
      "grad_norm": 1.463050365447998,
      "learning_rate": 1.7572016730266993e-05,
      "loss": 0.0843,
      "step": 71075
    },
    {
      "epoch": 1.0471127082075375,
      "grad_norm": 1.5152462720870972,
      "learning_rate": 1.7570380358300008e-05,
      "loss": 0.0804,
      "step": 71100
    },
    {
      "epoch": 1.0474808912976246,
      "grad_norm": 1.9172595739364624,
      "learning_rate": 1.7568743986333022e-05,
      "loss": 0.0832,
      "step": 71125
    },
    {
      "epoch": 1.0478490743877116,
      "grad_norm": 1.480907678604126,
      "learning_rate": 1.756710761436604e-05,
      "loss": 0.0866,
      "step": 71150
    },
    {
      "epoch": 1.0482172574777986,
      "grad_norm": 1.302281379699707,
      "learning_rate": 1.756547124239905e-05,
      "loss": 0.0869,
      "step": 71175
    },
    {
      "epoch": 1.0485854405678856,
      "grad_norm": 1.1208454370498657,
      "learning_rate": 1.756383487043207e-05,
      "loss": 0.082,
      "step": 71200
    },
    {
      "epoch": 1.0489536236579726,
      "grad_norm": 1.2618451118469238,
      "learning_rate": 1.7562198498465083e-05,
      "loss": 0.0792,
      "step": 71225
    },
    {
      "epoch": 1.0493218067480596,
      "grad_norm": 1.8567328453063965,
      "learning_rate": 1.75605621264981e-05,
      "loss": 0.0776,
      "step": 71250
    },
    {
      "epoch": 1.0496899898381467,
      "grad_norm": 1.074665904045105,
      "learning_rate": 1.7558925754531116e-05,
      "loss": 0.0841,
      "step": 71275
    },
    {
      "epoch": 1.0500581729282337,
      "grad_norm": 1.6876140832901,
      "learning_rate": 1.755728938256413e-05,
      "loss": 0.0893,
      "step": 71300
    },
    {
      "epoch": 1.0504263560183207,
      "grad_norm": 1.3799030780792236,
      "learning_rate": 1.7555653010597148e-05,
      "loss": 0.0805,
      "step": 71325
    },
    {
      "epoch": 1.050794539108408,
      "grad_norm": 1.2006375789642334,
      "learning_rate": 1.7554016638630162e-05,
      "loss": 0.0786,
      "step": 71350
    },
    {
      "epoch": 1.051162722198495,
      "grad_norm": 1.3169827461242676,
      "learning_rate": 1.7552380266663177e-05,
      "loss": 0.0804,
      "step": 71375
    },
    {
      "epoch": 1.051530905288582,
      "grad_norm": 1.6053396463394165,
      "learning_rate": 1.755074389469619e-05,
      "loss": 0.0789,
      "step": 71400
    },
    {
      "epoch": 1.051899088378669,
      "grad_norm": 1.2583928108215332,
      "learning_rate": 1.7549107522729206e-05,
      "loss": 0.0794,
      "step": 71425
    },
    {
      "epoch": 1.052267271468756,
      "grad_norm": 0.9673905968666077,
      "learning_rate": 1.7547471150762224e-05,
      "loss": 0.0809,
      "step": 71450
    },
    {
      "epoch": 1.052635454558843,
      "grad_norm": 1.3871254920959473,
      "learning_rate": 1.7545834778795238e-05,
      "loss": 0.0757,
      "step": 71475
    },
    {
      "epoch": 1.05300363764893,
      "grad_norm": 1.766830325126648,
      "learning_rate": 1.7544198406828256e-05,
      "loss": 0.0769,
      "step": 71500
    },
    {
      "epoch": 1.053371820739017,
      "grad_norm": 1.335160732269287,
      "learning_rate": 1.754256203486127e-05,
      "loss": 0.0789,
      "step": 71525
    },
    {
      "epoch": 1.053740003829104,
      "grad_norm": 1.3315373659133911,
      "learning_rate": 1.7540925662894285e-05,
      "loss": 0.0766,
      "step": 71550
    },
    {
      "epoch": 1.0541081869191913,
      "grad_norm": 1.4730496406555176,
      "learning_rate": 1.7539289290927303e-05,
      "loss": 0.0855,
      "step": 71575
    },
    {
      "epoch": 1.0544763700092783,
      "grad_norm": 2.104713201522827,
      "learning_rate": 1.7537652918960314e-05,
      "loss": 0.0805,
      "step": 71600
    },
    {
      "epoch": 1.0548445530993653,
      "grad_norm": 1.2803919315338135,
      "learning_rate": 1.753601654699333e-05,
      "loss": 0.0746,
      "step": 71625
    },
    {
      "epoch": 1.0552127361894523,
      "grad_norm": 1.606890082359314,
      "learning_rate": 1.7534380175026346e-05,
      "loss": 0.0864,
      "step": 71650
    },
    {
      "epoch": 1.0555809192795393,
      "grad_norm": 1.3688819408416748,
      "learning_rate": 1.753274380305936e-05,
      "loss": 0.0784,
      "step": 71675
    },
    {
      "epoch": 1.0559491023696264,
      "grad_norm": 1.2413685321807861,
      "learning_rate": 1.753110743109238e-05,
      "loss": 0.0812,
      "step": 71700
    },
    {
      "epoch": 1.0563172854597134,
      "grad_norm": 1.475354552268982,
      "learning_rate": 1.7529471059125393e-05,
      "loss": 0.0817,
      "step": 71725
    },
    {
      "epoch": 1.0566854685498004,
      "grad_norm": 1.1470940113067627,
      "learning_rate": 1.752783468715841e-05,
      "loss": 0.0857,
      "step": 71750
    },
    {
      "epoch": 1.0570536516398874,
      "grad_norm": 1.1003742218017578,
      "learning_rate": 1.7526198315191425e-05,
      "loss": 0.0847,
      "step": 71775
    },
    {
      "epoch": 1.0574218347299744,
      "grad_norm": 1.3824670314788818,
      "learning_rate": 1.752456194322444e-05,
      "loss": 0.0797,
      "step": 71800
    },
    {
      "epoch": 1.0577900178200617,
      "grad_norm": 1.7201858758926392,
      "learning_rate": 1.7522925571257454e-05,
      "loss": 0.0772,
      "step": 71825
    },
    {
      "epoch": 1.0581582009101487,
      "grad_norm": 1.4946352243423462,
      "learning_rate": 1.752128919929047e-05,
      "loss": 0.0832,
      "step": 71850
    },
    {
      "epoch": 1.0585263840002357,
      "grad_norm": 1.4284608364105225,
      "learning_rate": 1.7519652827323486e-05,
      "loss": 0.0807,
      "step": 71875
    },
    {
      "epoch": 1.0588945670903227,
      "grad_norm": 1.5651804208755493,
      "learning_rate": 1.75180164553565e-05,
      "loss": 0.0783,
      "step": 71900
    },
    {
      "epoch": 1.0592627501804097,
      "grad_norm": 1.2496366500854492,
      "learning_rate": 1.751638008338952e-05,
      "loss": 0.0824,
      "step": 71925
    },
    {
      "epoch": 1.0596309332704967,
      "grad_norm": 1.242274284362793,
      "learning_rate": 1.7514743711422533e-05,
      "loss": 0.0781,
      "step": 71950
    },
    {
      "epoch": 1.0599991163605837,
      "grad_norm": 1.537037968635559,
      "learning_rate": 1.7513107339455548e-05,
      "loss": 0.0735,
      "step": 71975
    },
    {
      "epoch": 1.0603672994506708,
      "grad_norm": 1.0832239389419556,
      "learning_rate": 1.7511470967488565e-05,
      "loss": 0.0779,
      "step": 72000
    },
    {
      "epoch": 1.0607354825407578,
      "grad_norm": 1.5025373697280884,
      "learning_rate": 1.7509834595521576e-05,
      "loss": 0.0829,
      "step": 72025
    },
    {
      "epoch": 1.0611036656308448,
      "grad_norm": 1.574141025543213,
      "learning_rate": 1.7508198223554594e-05,
      "loss": 0.0818,
      "step": 72050
    },
    {
      "epoch": 1.061471848720932,
      "grad_norm": 1.5251744985580444,
      "learning_rate": 1.750656185158761e-05,
      "loss": 0.0754,
      "step": 72075
    },
    {
      "epoch": 1.061840031811019,
      "grad_norm": 1.8047915697097778,
      "learning_rate": 1.7504925479620623e-05,
      "loss": 0.0796,
      "step": 72100
    },
    {
      "epoch": 1.062208214901106,
      "grad_norm": 1.5582528114318848,
      "learning_rate": 1.750328910765364e-05,
      "loss": 0.0867,
      "step": 72125
    },
    {
      "epoch": 1.062576397991193,
      "grad_norm": 1.557234287261963,
      "learning_rate": 1.7501652735686656e-05,
      "loss": 0.081,
      "step": 72150
    },
    {
      "epoch": 1.06294458108128,
      "grad_norm": 2.2205381393432617,
      "learning_rate": 1.7500016363719673e-05,
      "loss": 0.0816,
      "step": 72175
    },
    {
      "epoch": 1.063312764171367,
      "grad_norm": 1.7665126323699951,
      "learning_rate": 1.7498379991752688e-05,
      "loss": 0.0752,
      "step": 72200
    },
    {
      "epoch": 1.0636809472614541,
      "grad_norm": 1.6299313306808472,
      "learning_rate": 1.7496743619785702e-05,
      "loss": 0.0829,
      "step": 72225
    },
    {
      "epoch": 1.0640491303515411,
      "grad_norm": 1.3616055250167847,
      "learning_rate": 1.7495107247818717e-05,
      "loss": 0.076,
      "step": 72250
    },
    {
      "epoch": 1.0644173134416282,
      "grad_norm": 1.0546361207962036,
      "learning_rate": 1.749347087585173e-05,
      "loss": 0.0764,
      "step": 72275
    },
    {
      "epoch": 1.0647854965317154,
      "grad_norm": 1.6121575832366943,
      "learning_rate": 1.749183450388475e-05,
      "loss": 0.0845,
      "step": 72300
    },
    {
      "epoch": 1.0651536796218024,
      "grad_norm": 1.368516206741333,
      "learning_rate": 1.7490198131917764e-05,
      "loss": 0.0795,
      "step": 72325
    },
    {
      "epoch": 1.0655218627118894,
      "grad_norm": 1.8296114206314087,
      "learning_rate": 1.7488561759950778e-05,
      "loss": 0.0835,
      "step": 72350
    },
    {
      "epoch": 1.0658900458019764,
      "grad_norm": 1.780468463897705,
      "learning_rate": 1.7486925387983796e-05,
      "loss": 0.0869,
      "step": 72375
    },
    {
      "epoch": 1.0662582288920635,
      "grad_norm": 1.5264418125152588,
      "learning_rate": 1.748528901601681e-05,
      "loss": 0.0816,
      "step": 72400
    },
    {
      "epoch": 1.0666264119821505,
      "grad_norm": 1.9545817375183105,
      "learning_rate": 1.7483718098928508e-05,
      "loss": 0.0909,
      "step": 72425
    },
    {
      "epoch": 1.0669945950722375,
      "grad_norm": 1.7828044891357422,
      "learning_rate": 1.748208172696152e-05,
      "loss": 0.08,
      "step": 72450
    },
    {
      "epoch": 1.0673627781623245,
      "grad_norm": 1.3515057563781738,
      "learning_rate": 1.7480445354994536e-05,
      "loss": 0.0847,
      "step": 72475
    },
    {
      "epoch": 1.0677309612524115,
      "grad_norm": 1.1792912483215332,
      "learning_rate": 1.747880898302755e-05,
      "loss": 0.0793,
      "step": 72500
    },
    {
      "epoch": 1.0680991443424985,
      "grad_norm": 1.4954952001571655,
      "learning_rate": 1.7477172611060565e-05,
      "loss": 0.0868,
      "step": 72525
    },
    {
      "epoch": 1.0684673274325858,
      "grad_norm": 1.7793201208114624,
      "learning_rate": 1.7475536239093583e-05,
      "loss": 0.0758,
      "step": 72550
    },
    {
      "epoch": 1.0688355105226728,
      "grad_norm": 1.258103847503662,
      "learning_rate": 1.7473899867126598e-05,
      "loss": 0.081,
      "step": 72575
    },
    {
      "epoch": 1.0692036936127598,
      "grad_norm": 0.9954408407211304,
      "learning_rate": 1.7472263495159612e-05,
      "loss": 0.0834,
      "step": 72600
    },
    {
      "epoch": 1.0695718767028468,
      "grad_norm": 1.409990668296814,
      "learning_rate": 1.747062712319263e-05,
      "loss": 0.0819,
      "step": 72625
    },
    {
      "epoch": 1.0699400597929338,
      "grad_norm": 0.9933326244354248,
      "learning_rate": 1.7468990751225644e-05,
      "loss": 0.0728,
      "step": 72650
    },
    {
      "epoch": 1.0703082428830208,
      "grad_norm": 1.8262516260147095,
      "learning_rate": 1.746735437925866e-05,
      "loss": 0.0837,
      "step": 72675
    },
    {
      "epoch": 1.0706764259731079,
      "grad_norm": 1.5246831178665161,
      "learning_rate": 1.7465718007291673e-05,
      "loss": 0.0867,
      "step": 72700
    },
    {
      "epoch": 1.0710446090631949,
      "grad_norm": 1.4583308696746826,
      "learning_rate": 1.746408163532469e-05,
      "loss": 0.0815,
      "step": 72725
    },
    {
      "epoch": 1.071412792153282,
      "grad_norm": 1.1462247371673584,
      "learning_rate": 1.7462445263357706e-05,
      "loss": 0.0849,
      "step": 72750
    },
    {
      "epoch": 1.0717809752433691,
      "grad_norm": 1.4642852544784546,
      "learning_rate": 1.746080889139072e-05,
      "loss": 0.0826,
      "step": 72775
    },
    {
      "epoch": 1.0721491583334561,
      "grad_norm": 1.5686918497085571,
      "learning_rate": 1.7459172519423738e-05,
      "loss": 0.0756,
      "step": 72800
    },
    {
      "epoch": 1.0725173414235432,
      "grad_norm": 1.3278517723083496,
      "learning_rate": 1.7457536147456752e-05,
      "loss": 0.0816,
      "step": 72825
    },
    {
      "epoch": 1.0728855245136302,
      "grad_norm": 1.6725236177444458,
      "learning_rate": 1.7455899775489767e-05,
      "loss": 0.0771,
      "step": 72850
    },
    {
      "epoch": 1.0732537076037172,
      "grad_norm": 1.555825114250183,
      "learning_rate": 1.745426340352278e-05,
      "loss": 0.08,
      "step": 72875
    },
    {
      "epoch": 1.0736218906938042,
      "grad_norm": 1.4804408550262451,
      "learning_rate": 1.74526270315558e-05,
      "loss": 0.081,
      "step": 72900
    },
    {
      "epoch": 1.0739900737838912,
      "grad_norm": 1.2189922332763672,
      "learning_rate": 1.7450990659588814e-05,
      "loss": 0.0677,
      "step": 72925
    },
    {
      "epoch": 1.0743582568739782,
      "grad_norm": 1.2289925813674927,
      "learning_rate": 1.7449354287621828e-05,
      "loss": 0.0749,
      "step": 72950
    },
    {
      "epoch": 1.0747264399640652,
      "grad_norm": 1.0798590183258057,
      "learning_rate": 1.7447717915654846e-05,
      "loss": 0.0846,
      "step": 72975
    },
    {
      "epoch": 1.0750946230541523,
      "grad_norm": 1.6045472621917725,
      "learning_rate": 1.744608154368786e-05,
      "loss": 0.0822,
      "step": 73000
    },
    {
      "epoch": 1.0754628061442395,
      "grad_norm": 1.045710563659668,
      "learning_rate": 1.7444445171720875e-05,
      "loss": 0.0757,
      "step": 73025
    },
    {
      "epoch": 1.0758309892343265,
      "grad_norm": 1.7396941184997559,
      "learning_rate": 1.7442808799753893e-05,
      "loss": 0.087,
      "step": 73050
    },
    {
      "epoch": 1.0761991723244135,
      "grad_norm": 1.4701290130615234,
      "learning_rate": 1.7441172427786904e-05,
      "loss": 0.0876,
      "step": 73075
    },
    {
      "epoch": 1.0765673554145005,
      "grad_norm": 1.0782220363616943,
      "learning_rate": 1.743953605581992e-05,
      "loss": 0.0839,
      "step": 73100
    },
    {
      "epoch": 1.0769355385045876,
      "grad_norm": 1.6876006126403809,
      "learning_rate": 1.7437899683852936e-05,
      "loss": 0.0782,
      "step": 73125
    },
    {
      "epoch": 1.0773037215946746,
      "grad_norm": 1.6829878091812134,
      "learning_rate": 1.7436263311885954e-05,
      "loss": 0.0791,
      "step": 73150
    },
    {
      "epoch": 1.0776719046847616,
      "grad_norm": 1.1190736293792725,
      "learning_rate": 1.743462693991897e-05,
      "loss": 0.0823,
      "step": 73175
    },
    {
      "epoch": 1.0780400877748486,
      "grad_norm": 1.127810001373291,
      "learning_rate": 1.7432990567951983e-05,
      "loss": 0.0718,
      "step": 73200
    },
    {
      "epoch": 1.0784082708649356,
      "grad_norm": 1.4211413860321045,
      "learning_rate": 1.7431354195985e-05,
      "loss": 0.0825,
      "step": 73225
    },
    {
      "epoch": 1.0787764539550229,
      "grad_norm": 1.359724760055542,
      "learning_rate": 1.7429717824018015e-05,
      "loss": 0.0859,
      "step": 73250
    },
    {
      "epoch": 1.0791446370451099,
      "grad_norm": 1.8260948657989502,
      "learning_rate": 1.742808145205103e-05,
      "loss": 0.0819,
      "step": 73275
    },
    {
      "epoch": 1.079512820135197,
      "grad_norm": 1.6171700954437256,
      "learning_rate": 1.7426445080084044e-05,
      "loss": 0.0749,
      "step": 73300
    },
    {
      "epoch": 1.079881003225284,
      "grad_norm": 1.9867669343948364,
      "learning_rate": 1.7424808708117062e-05,
      "loss": 0.0858,
      "step": 73325
    },
    {
      "epoch": 1.080249186315371,
      "grad_norm": 1.6073391437530518,
      "learning_rate": 1.7423172336150076e-05,
      "loss": 0.0897,
      "step": 73350
    },
    {
      "epoch": 1.080617369405458,
      "grad_norm": 1.0877889394760132,
      "learning_rate": 1.742153596418309e-05,
      "loss": 0.0748,
      "step": 73375
    },
    {
      "epoch": 1.080985552495545,
      "grad_norm": 1.6490763425827026,
      "learning_rate": 1.741989959221611e-05,
      "loss": 0.086,
      "step": 73400
    },
    {
      "epoch": 1.081353735585632,
      "grad_norm": 1.2874717712402344,
      "learning_rate": 1.7418263220249123e-05,
      "loss": 0.079,
      "step": 73425
    },
    {
      "epoch": 1.081721918675719,
      "grad_norm": 1.2430524826049805,
      "learning_rate": 1.7416626848282138e-05,
      "loss": 0.0806,
      "step": 73450
    },
    {
      "epoch": 1.082090101765806,
      "grad_norm": 1.2100410461425781,
      "learning_rate": 1.7414990476315155e-05,
      "loss": 0.0776,
      "step": 73475
    },
    {
      "epoch": 1.0824582848558932,
      "grad_norm": 1.1482142210006714,
      "learning_rate": 1.7413354104348167e-05,
      "loss": 0.0781,
      "step": 73500
    },
    {
      "epoch": 1.0828264679459803,
      "grad_norm": 1.4018893241882324,
      "learning_rate": 1.7411717732381184e-05,
      "loss": 0.0784,
      "step": 73525
    },
    {
      "epoch": 1.0831946510360673,
      "grad_norm": 1.5947740077972412,
      "learning_rate": 1.74100813604142e-05,
      "loss": 0.0759,
      "step": 73550
    },
    {
      "epoch": 1.0835628341261543,
      "grad_norm": 1.4407312870025635,
      "learning_rate": 1.7408444988447217e-05,
      "loss": 0.0718,
      "step": 73575
    },
    {
      "epoch": 1.0839310172162413,
      "grad_norm": 1.9396814107894897,
      "learning_rate": 1.740680861648023e-05,
      "loss": 0.0775,
      "step": 73600
    },
    {
      "epoch": 1.0842992003063283,
      "grad_norm": 1.4494119882583618,
      "learning_rate": 1.7405172244513246e-05,
      "loss": 0.081,
      "step": 73625
    },
    {
      "epoch": 1.0846673833964153,
      "grad_norm": 1.6604697704315186,
      "learning_rate": 1.7403535872546263e-05,
      "loss": 0.078,
      "step": 73650
    },
    {
      "epoch": 1.0850355664865023,
      "grad_norm": 1.3195675611495972,
      "learning_rate": 1.7401899500579278e-05,
      "loss": 0.0791,
      "step": 73675
    },
    {
      "epoch": 1.0854037495765894,
      "grad_norm": 1.4423450231552124,
      "learning_rate": 1.7400263128612292e-05,
      "loss": 0.0791,
      "step": 73700
    },
    {
      "epoch": 1.0857719326666766,
      "grad_norm": 1.245192289352417,
      "learning_rate": 1.7398626756645307e-05,
      "loss": 0.086,
      "step": 73725
    },
    {
      "epoch": 1.0861401157567636,
      "grad_norm": 1.1297852993011475,
      "learning_rate": 1.739699038467832e-05,
      "loss": 0.0861,
      "step": 73750
    },
    {
      "epoch": 1.0865082988468506,
      "grad_norm": 1.7305914163589478,
      "learning_rate": 1.739535401271134e-05,
      "loss": 0.0853,
      "step": 73775
    },
    {
      "epoch": 1.0868764819369376,
      "grad_norm": 1.5151560306549072,
      "learning_rate": 1.7393717640744354e-05,
      "loss": 0.0822,
      "step": 73800
    },
    {
      "epoch": 1.0872446650270247,
      "grad_norm": 1.3996646404266357,
      "learning_rate": 1.739208126877737e-05,
      "loss": 0.0816,
      "step": 73825
    },
    {
      "epoch": 1.0876128481171117,
      "grad_norm": 1.3975486755371094,
      "learning_rate": 1.7390444896810386e-05,
      "loss": 0.0831,
      "step": 73850
    },
    {
      "epoch": 1.0879810312071987,
      "grad_norm": 1.7937424182891846,
      "learning_rate": 1.73888085248434e-05,
      "loss": 0.0845,
      "step": 73875
    },
    {
      "epoch": 1.0883492142972857,
      "grad_norm": 1.3818455934524536,
      "learning_rate": 1.7387172152876418e-05,
      "loss": 0.0865,
      "step": 73900
    },
    {
      "epoch": 1.0887173973873727,
      "grad_norm": 1.3593707084655762,
      "learning_rate": 1.738553578090943e-05,
      "loss": 0.0877,
      "step": 73925
    },
    {
      "epoch": 1.0890855804774597,
      "grad_norm": 1.2562730312347412,
      "learning_rate": 1.7383899408942447e-05,
      "loss": 0.0815,
      "step": 73950
    },
    {
      "epoch": 1.089453763567547,
      "grad_norm": 1.4753555059432983,
      "learning_rate": 1.738226303697546e-05,
      "loss": 0.0756,
      "step": 73975
    },
    {
      "epoch": 1.089821946657634,
      "grad_norm": 0.9510453343391418,
      "learning_rate": 1.7380626665008476e-05,
      "loss": 0.085,
      "step": 74000
    },
    {
      "epoch": 1.090190129747721,
      "grad_norm": 1.5919740200042725,
      "learning_rate": 1.7378990293041494e-05,
      "loss": 0.0778,
      "step": 74025
    },
    {
      "epoch": 1.090558312837808,
      "grad_norm": 1.4968332052230835,
      "learning_rate": 1.7377353921074508e-05,
      "loss": 0.0733,
      "step": 74050
    },
    {
      "epoch": 1.090926495927895,
      "grad_norm": 1.4238145351409912,
      "learning_rate": 1.7375717549107526e-05,
      "loss": 0.0732,
      "step": 74075
    },
    {
      "epoch": 1.091294679017982,
      "grad_norm": 1.6327730417251587,
      "learning_rate": 1.737408117714054e-05,
      "loss": 0.0813,
      "step": 74100
    },
    {
      "epoch": 1.091662862108069,
      "grad_norm": 1.4643621444702148,
      "learning_rate": 1.7372444805173555e-05,
      "loss": 0.0828,
      "step": 74125
    },
    {
      "epoch": 1.092031045198156,
      "grad_norm": 1.0110392570495605,
      "learning_rate": 1.737080843320657e-05,
      "loss": 0.0856,
      "step": 74150
    },
    {
      "epoch": 1.092399228288243,
      "grad_norm": 1.255319356918335,
      "learning_rate": 1.7369172061239584e-05,
      "loss": 0.082,
      "step": 74175
    },
    {
      "epoch": 1.0927674113783303,
      "grad_norm": 1.1562066078186035,
      "learning_rate": 1.7367535689272602e-05,
      "loss": 0.0776,
      "step": 74200
    },
    {
      "epoch": 1.0931355944684173,
      "grad_norm": 1.4678603410720825,
      "learning_rate": 1.7365899317305616e-05,
      "loss": 0.0799,
      "step": 74225
    },
    {
      "epoch": 1.0935037775585044,
      "grad_norm": 1.4903764724731445,
      "learning_rate": 1.7364262945338634e-05,
      "loss": 0.0806,
      "step": 74250
    },
    {
      "epoch": 1.0938719606485914,
      "grad_norm": 1.6453461647033691,
      "learning_rate": 1.736262657337165e-05,
      "loss": 0.0857,
      "step": 74275
    },
    {
      "epoch": 1.0942401437386784,
      "grad_norm": 1.5547235012054443,
      "learning_rate": 1.7360990201404663e-05,
      "loss": 0.0788,
      "step": 74300
    },
    {
      "epoch": 1.0946083268287654,
      "grad_norm": 1.6805888414382935,
      "learning_rate": 1.735935382943768e-05,
      "loss": 0.0849,
      "step": 74325
    },
    {
      "epoch": 1.0949765099188524,
      "grad_norm": 1.1416271924972534,
      "learning_rate": 1.7357717457470692e-05,
      "loss": 0.0826,
      "step": 74350
    },
    {
      "epoch": 1.0953446930089394,
      "grad_norm": 1.2809503078460693,
      "learning_rate": 1.735608108550371e-05,
      "loss": 0.0852,
      "step": 74375
    },
    {
      "epoch": 1.0957128760990265,
      "grad_norm": 1.764859676361084,
      "learning_rate": 1.7354444713536724e-05,
      "loss": 0.0797,
      "step": 74400
    },
    {
      "epoch": 1.0960810591891135,
      "grad_norm": 1.3929708003997803,
      "learning_rate": 1.735280834156974e-05,
      "loss": 0.0701,
      "step": 74425
    },
    {
      "epoch": 1.0964492422792005,
      "grad_norm": 1.5949950218200684,
      "learning_rate": 1.7351171969602757e-05,
      "loss": 0.083,
      "step": 74450
    },
    {
      "epoch": 1.0968174253692877,
      "grad_norm": 1.3857530355453491,
      "learning_rate": 1.734953559763577e-05,
      "loss": 0.0809,
      "step": 74475
    },
    {
      "epoch": 1.0971856084593747,
      "grad_norm": 1.1845228672027588,
      "learning_rate": 1.734789922566879e-05,
      "loss": 0.0719,
      "step": 74500
    },
    {
      "epoch": 1.0975537915494618,
      "grad_norm": 1.6475591659545898,
      "learning_rate": 1.7346262853701803e-05,
      "loss": 0.0746,
      "step": 74525
    },
    {
      "epoch": 1.0979219746395488,
      "grad_norm": 1.65035080909729,
      "learning_rate": 1.7344691936613497e-05,
      "loss": 0.0803,
      "step": 74550
    },
    {
      "epoch": 1.0982901577296358,
      "grad_norm": 1.9141035079956055,
      "learning_rate": 1.734305556464651e-05,
      "loss": 0.0887,
      "step": 74575
    },
    {
      "epoch": 1.0986583408197228,
      "grad_norm": 1.7077194452285767,
      "learning_rate": 1.7341419192679526e-05,
      "loss": 0.0851,
      "step": 74600
    },
    {
      "epoch": 1.0990265239098098,
      "grad_norm": 1.180472493171692,
      "learning_rate": 1.7339782820712544e-05,
      "loss": 0.0837,
      "step": 74625
    },
    {
      "epoch": 1.0993947069998968,
      "grad_norm": 1.4680033922195435,
      "learning_rate": 1.733814644874556e-05,
      "loss": 0.0846,
      "step": 74650
    },
    {
      "epoch": 1.0997628900899838,
      "grad_norm": 1.3979564905166626,
      "learning_rate": 1.7336510076778573e-05,
      "loss": 0.0795,
      "step": 74675
    },
    {
      "epoch": 1.100131073180071,
      "grad_norm": 1.307181477546692,
      "learning_rate": 1.733487370481159e-05,
      "loss": 0.0826,
      "step": 74700
    },
    {
      "epoch": 1.100499256270158,
      "grad_norm": 1.5092790126800537,
      "learning_rate": 1.7333237332844605e-05,
      "loss": 0.0822,
      "step": 74725
    },
    {
      "epoch": 1.1008674393602451,
      "grad_norm": 1.431704044342041,
      "learning_rate": 1.7331600960877623e-05,
      "loss": 0.0891,
      "step": 74750
    },
    {
      "epoch": 1.1012356224503321,
      "grad_norm": 0.9238422513008118,
      "learning_rate": 1.7329964588910634e-05,
      "loss": 0.0797,
      "step": 74775
    },
    {
      "epoch": 1.1016038055404191,
      "grad_norm": 2.129912853240967,
      "learning_rate": 1.7328328216943652e-05,
      "loss": 0.0755,
      "step": 74800
    },
    {
      "epoch": 1.1019719886305062,
      "grad_norm": 1.3834251165390015,
      "learning_rate": 1.7326691844976666e-05,
      "loss": 0.0779,
      "step": 74825
    },
    {
      "epoch": 1.1023401717205932,
      "grad_norm": 1.7270833253860474,
      "learning_rate": 1.732505547300968e-05,
      "loss": 0.0762,
      "step": 74850
    },
    {
      "epoch": 1.1027083548106802,
      "grad_norm": 1.3919258117675781,
      "learning_rate": 1.73234191010427e-05,
      "loss": 0.0846,
      "step": 74875
    },
    {
      "epoch": 1.1030765379007672,
      "grad_norm": 1.6441198587417603,
      "learning_rate": 1.7321782729075713e-05,
      "loss": 0.0834,
      "step": 74900
    },
    {
      "epoch": 1.1034447209908542,
      "grad_norm": 1.5029547214508057,
      "learning_rate": 1.7320146357108728e-05,
      "loss": 0.0783,
      "step": 74925
    },
    {
      "epoch": 1.1038129040809415,
      "grad_norm": 2.0855712890625,
      "learning_rate": 1.7318509985141745e-05,
      "loss": 0.0766,
      "step": 74950
    },
    {
      "epoch": 1.1041810871710285,
      "grad_norm": 1.2368886470794678,
      "learning_rate": 1.731687361317476e-05,
      "loss": 0.0853,
      "step": 74975
    },
    {
      "epoch": 1.1045492702611155,
      "grad_norm": 1.6182918548583984,
      "learning_rate": 1.7315237241207774e-05,
      "loss": 0.0825,
      "step": 75000
    },
    {
      "epoch": 1.1049174533512025,
      "grad_norm": 1.0082389116287231,
      "learning_rate": 1.731360086924079e-05,
      "loss": 0.0822,
      "step": 75025
    },
    {
      "epoch": 1.1052856364412895,
      "grad_norm": 1.4351855516433716,
      "learning_rate": 1.7311964497273807e-05,
      "loss": 0.0837,
      "step": 75050
    },
    {
      "epoch": 1.1056538195313765,
      "grad_norm": 1.5235116481781006,
      "learning_rate": 1.731032812530682e-05,
      "loss": 0.0847,
      "step": 75075
    },
    {
      "epoch": 1.1060220026214636,
      "grad_norm": 1.17570161819458,
      "learning_rate": 1.7308691753339836e-05,
      "loss": 0.0774,
      "step": 75100
    },
    {
      "epoch": 1.1063901857115506,
      "grad_norm": 1.96731698513031,
      "learning_rate": 1.7307055381372853e-05,
      "loss": 0.0817,
      "step": 75125
    },
    {
      "epoch": 1.1067583688016376,
      "grad_norm": 1.8333358764648438,
      "learning_rate": 1.7305419009405868e-05,
      "loss": 0.0921,
      "step": 75150
    },
    {
      "epoch": 1.1071265518917248,
      "grad_norm": 1.3284287452697754,
      "learning_rate": 1.7303782637438882e-05,
      "loss": 0.0867,
      "step": 75175
    },
    {
      "epoch": 1.1074947349818118,
      "grad_norm": 1.3460052013397217,
      "learning_rate": 1.7302146265471897e-05,
      "loss": 0.0828,
      "step": 75200
    },
    {
      "epoch": 1.1078629180718989,
      "grad_norm": 1.5569969415664673,
      "learning_rate": 1.7300509893504915e-05,
      "loss": 0.0854,
      "step": 75225
    },
    {
      "epoch": 1.1082311011619859,
      "grad_norm": 1.6549465656280518,
      "learning_rate": 1.729887352153793e-05,
      "loss": 0.0867,
      "step": 75250
    },
    {
      "epoch": 1.1085992842520729,
      "grad_norm": 1.4125876426696777,
      "learning_rate": 1.7297237149570944e-05,
      "loss": 0.0756,
      "step": 75275
    },
    {
      "epoch": 1.10896746734216,
      "grad_norm": 1.3274352550506592,
      "learning_rate": 1.729560077760396e-05,
      "loss": 0.0843,
      "step": 75300
    },
    {
      "epoch": 1.109335650432247,
      "grad_norm": 1.4729095697402954,
      "learning_rate": 1.7293964405636976e-05,
      "loss": 0.081,
      "step": 75325
    },
    {
      "epoch": 1.109703833522334,
      "grad_norm": 1.2580407857894897,
      "learning_rate": 1.729232803366999e-05,
      "loss": 0.081,
      "step": 75350
    },
    {
      "epoch": 1.110072016612421,
      "grad_norm": 1.4386241436004639,
      "learning_rate": 1.7290691661703008e-05,
      "loss": 0.078,
      "step": 75375
    },
    {
      "epoch": 1.110440199702508,
      "grad_norm": 1.2250170707702637,
      "learning_rate": 1.728905528973602e-05,
      "loss": 0.0737,
      "step": 75400
    },
    {
      "epoch": 1.1108083827925952,
      "grad_norm": 1.6921952962875366,
      "learning_rate": 1.7287418917769037e-05,
      "loss": 0.0823,
      "step": 75425
    },
    {
      "epoch": 1.1111765658826822,
      "grad_norm": 1.3983107805252075,
      "learning_rate": 1.728578254580205e-05,
      "loss": 0.0847,
      "step": 75450
    },
    {
      "epoch": 1.1115447489727692,
      "grad_norm": 1.6122878789901733,
      "learning_rate": 1.728414617383507e-05,
      "loss": 0.0845,
      "step": 75475
    },
    {
      "epoch": 1.1119129320628562,
      "grad_norm": 1.3357850313186646,
      "learning_rate": 1.7282509801868084e-05,
      "loss": 0.083,
      "step": 75500
    },
    {
      "epoch": 1.1122811151529433,
      "grad_norm": 1.0580809116363525,
      "learning_rate": 1.7280873429901098e-05,
      "loss": 0.0815,
      "step": 75525
    },
    {
      "epoch": 1.1126492982430303,
      "grad_norm": 1.527647852897644,
      "learning_rate": 1.7279237057934116e-05,
      "loss": 0.0813,
      "step": 75550
    },
    {
      "epoch": 1.1130174813331173,
      "grad_norm": 0.9379507899284363,
      "learning_rate": 1.727760068596713e-05,
      "loss": 0.076,
      "step": 75575
    },
    {
      "epoch": 1.1133856644232043,
      "grad_norm": 1.5256074666976929,
      "learning_rate": 1.7275964314000145e-05,
      "loss": 0.0804,
      "step": 75600
    },
    {
      "epoch": 1.1137538475132913,
      "grad_norm": 1.0808801651000977,
      "learning_rate": 1.727432794203316e-05,
      "loss": 0.0784,
      "step": 75625
    },
    {
      "epoch": 1.1141220306033786,
      "grad_norm": 1.0583561658859253,
      "learning_rate": 1.7272691570066177e-05,
      "loss": 0.0717,
      "step": 75650
    },
    {
      "epoch": 1.1144902136934656,
      "grad_norm": 1.167378306388855,
      "learning_rate": 1.7271055198099192e-05,
      "loss": 0.0794,
      "step": 75675
    },
    {
      "epoch": 1.1148583967835526,
      "grad_norm": 1.806620478630066,
      "learning_rate": 1.7269418826132206e-05,
      "loss": 0.0878,
      "step": 75700
    },
    {
      "epoch": 1.1152265798736396,
      "grad_norm": 1.5262364149093628,
      "learning_rate": 1.7267782454165224e-05,
      "loss": 0.0824,
      "step": 75725
    },
    {
      "epoch": 1.1155947629637266,
      "grad_norm": 1.6135478019714355,
      "learning_rate": 1.726614608219824e-05,
      "loss": 0.084,
      "step": 75750
    },
    {
      "epoch": 1.1159629460538136,
      "grad_norm": 1.9469832181930542,
      "learning_rate": 1.7264509710231253e-05,
      "loss": 0.0846,
      "step": 75775
    },
    {
      "epoch": 1.1163311291439006,
      "grad_norm": 1.2898473739624023,
      "learning_rate": 1.726287333826427e-05,
      "loss": 0.0758,
      "step": 75800
    },
    {
      "epoch": 1.1166993122339877,
      "grad_norm": 1.1968735456466675,
      "learning_rate": 1.7261236966297282e-05,
      "loss": 0.0836,
      "step": 75825
    },
    {
      "epoch": 1.1170674953240747,
      "grad_norm": 1.5359967947006226,
      "learning_rate": 1.72596005943303e-05,
      "loss": 0.0792,
      "step": 75850
    },
    {
      "epoch": 1.1174356784141617,
      "grad_norm": 1.64653480052948,
      "learning_rate": 1.7257964222363314e-05,
      "loss": 0.0858,
      "step": 75875
    },
    {
      "epoch": 1.117803861504249,
      "grad_norm": 1.3337204456329346,
      "learning_rate": 1.7256327850396332e-05,
      "loss": 0.0883,
      "step": 75900
    },
    {
      "epoch": 1.118172044594336,
      "grad_norm": 1.4658782482147217,
      "learning_rate": 1.7254691478429347e-05,
      "loss": 0.0836,
      "step": 75925
    },
    {
      "epoch": 1.118540227684423,
      "grad_norm": 1.0437543392181396,
      "learning_rate": 1.725305510646236e-05,
      "loss": 0.082,
      "step": 75950
    },
    {
      "epoch": 1.11890841077451,
      "grad_norm": 1.3824712038040161,
      "learning_rate": 1.725141873449538e-05,
      "loss": 0.0894,
      "step": 75975
    },
    {
      "epoch": 1.119276593864597,
      "grad_norm": 1.7544254064559937,
      "learning_rate": 1.7249782362528393e-05,
      "loss": 0.077,
      "step": 76000
    },
    {
      "epoch": 1.119644776954684,
      "grad_norm": 1.0734038352966309,
      "learning_rate": 1.7248145990561408e-05,
      "loss": 0.0827,
      "step": 76025
    },
    {
      "epoch": 1.120012960044771,
      "grad_norm": 1.267118215560913,
      "learning_rate": 1.7246509618594422e-05,
      "loss": 0.0801,
      "step": 76050
    },
    {
      "epoch": 1.120381143134858,
      "grad_norm": 1.6631799936294556,
      "learning_rate": 1.7244873246627437e-05,
      "loss": 0.0821,
      "step": 76075
    },
    {
      "epoch": 1.120749326224945,
      "grad_norm": 1.2830431461334229,
      "learning_rate": 1.7243236874660455e-05,
      "loss": 0.0744,
      "step": 76100
    },
    {
      "epoch": 1.1211175093150323,
      "grad_norm": 1.292281150817871,
      "learning_rate": 1.724160050269347e-05,
      "loss": 0.0866,
      "step": 76125
    },
    {
      "epoch": 1.1214856924051193,
      "grad_norm": 1.574506163597107,
      "learning_rate": 1.7239964130726487e-05,
      "loss": 0.081,
      "step": 76150
    },
    {
      "epoch": 1.1218538754952063,
      "grad_norm": 1.2235420942306519,
      "learning_rate": 1.72383277587595e-05,
      "loss": 0.077,
      "step": 76175
    },
    {
      "epoch": 1.1222220585852933,
      "grad_norm": 1.723737359046936,
      "learning_rate": 1.7236691386792516e-05,
      "loss": 0.082,
      "step": 76200
    },
    {
      "epoch": 1.1225902416753804,
      "grad_norm": 1.4583799839019775,
      "learning_rate": 1.7235055014825534e-05,
      "loss": 0.0793,
      "step": 76225
    },
    {
      "epoch": 1.1229584247654674,
      "grad_norm": 1.1304832696914673,
      "learning_rate": 1.7233418642858545e-05,
      "loss": 0.0793,
      "step": 76250
    },
    {
      "epoch": 1.1233266078555544,
      "grad_norm": 0.9958282709121704,
      "learning_rate": 1.7231782270891563e-05,
      "loss": 0.0898,
      "step": 76275
    },
    {
      "epoch": 1.1236947909456414,
      "grad_norm": 1.4184434413909912,
      "learning_rate": 1.7230145898924577e-05,
      "loss": 0.0811,
      "step": 76300
    },
    {
      "epoch": 1.1240629740357284,
      "grad_norm": 1.0592827796936035,
      "learning_rate": 1.722850952695759e-05,
      "loss": 0.0754,
      "step": 76325
    },
    {
      "epoch": 1.1244311571258154,
      "grad_norm": 1.2309232950210571,
      "learning_rate": 1.722687315499061e-05,
      "loss": 0.0733,
      "step": 76350
    },
    {
      "epoch": 1.1247993402159027,
      "grad_norm": 1.6619325876235962,
      "learning_rate": 1.7225236783023624e-05,
      "loss": 0.0726,
      "step": 76375
    },
    {
      "epoch": 1.1251675233059897,
      "grad_norm": 0.8825359344482422,
      "learning_rate": 1.722360041105664e-05,
      "loss": 0.0762,
      "step": 76400
    },
    {
      "epoch": 1.1255357063960767,
      "grad_norm": 1.4367597103118896,
      "learning_rate": 1.7221964039089656e-05,
      "loss": 0.0835,
      "step": 76425
    },
    {
      "epoch": 1.1259038894861637,
      "grad_norm": 1.4315481185913086,
      "learning_rate": 1.722032766712267e-05,
      "loss": 0.0818,
      "step": 76450
    },
    {
      "epoch": 1.1262720725762507,
      "grad_norm": 1.7525913715362549,
      "learning_rate": 1.7218691295155685e-05,
      "loss": 0.0829,
      "step": 76475
    },
    {
      "epoch": 1.1266402556663377,
      "grad_norm": 1.1785986423492432,
      "learning_rate": 1.72170549231887e-05,
      "loss": 0.0756,
      "step": 76500
    },
    {
      "epoch": 1.1270084387564248,
      "grad_norm": 1.0775268077850342,
      "learning_rate": 1.7215418551221717e-05,
      "loss": 0.0831,
      "step": 76525
    },
    {
      "epoch": 1.1273766218465118,
      "grad_norm": 1.243326187133789,
      "learning_rate": 1.7213782179254732e-05,
      "loss": 0.0794,
      "step": 76550
    },
    {
      "epoch": 1.1277448049365988,
      "grad_norm": 1.255817174911499,
      "learning_rate": 1.721214580728775e-05,
      "loss": 0.0801,
      "step": 76575
    },
    {
      "epoch": 1.128112988026686,
      "grad_norm": 1.1211082935333252,
      "learning_rate": 1.7210509435320764e-05,
      "loss": 0.0827,
      "step": 76600
    },
    {
      "epoch": 1.128481171116773,
      "grad_norm": 1.6233681440353394,
      "learning_rate": 1.720887306335378e-05,
      "loss": 0.0792,
      "step": 76625
    },
    {
      "epoch": 1.12884935420686,
      "grad_norm": 1.3273680210113525,
      "learning_rate": 1.7207236691386796e-05,
      "loss": 0.0862,
      "step": 76650
    },
    {
      "epoch": 1.129217537296947,
      "grad_norm": 1.525869607925415,
      "learning_rate": 1.7205600319419807e-05,
      "loss": 0.0734,
      "step": 76675
    },
    {
      "epoch": 1.129585720387034,
      "grad_norm": 1.7137657403945923,
      "learning_rate": 1.7203963947452825e-05,
      "loss": 0.0803,
      "step": 76700
    },
    {
      "epoch": 1.129953903477121,
      "grad_norm": 1.5917319059371948,
      "learning_rate": 1.720232757548584e-05,
      "loss": 0.0847,
      "step": 76725
    },
    {
      "epoch": 1.1303220865672081,
      "grad_norm": 1.3040844202041626,
      "learning_rate": 1.7200691203518854e-05,
      "loss": 0.0687,
      "step": 76750
    },
    {
      "epoch": 1.1306902696572951,
      "grad_norm": 1.1933627128601074,
      "learning_rate": 1.7199054831551872e-05,
      "loss": 0.0811,
      "step": 76775
    },
    {
      "epoch": 1.1310584527473821,
      "grad_norm": 1.637626051902771,
      "learning_rate": 1.7197418459584886e-05,
      "loss": 0.084,
      "step": 76800
    },
    {
      "epoch": 1.1314266358374692,
      "grad_norm": 1.5711268186569214,
      "learning_rate": 1.7195782087617904e-05,
      "loss": 0.0797,
      "step": 76825
    },
    {
      "epoch": 1.1317948189275562,
      "grad_norm": 1.3153433799743652,
      "learning_rate": 1.719414571565092e-05,
      "loss": 0.0853,
      "step": 76850
    },
    {
      "epoch": 1.1321630020176434,
      "grad_norm": 1.6022580862045288,
      "learning_rate": 1.7192509343683933e-05,
      "loss": 0.0732,
      "step": 76875
    },
    {
      "epoch": 1.1325311851077304,
      "grad_norm": 1.388688564300537,
      "learning_rate": 1.7190872971716948e-05,
      "loss": 0.0836,
      "step": 76900
    },
    {
      "epoch": 1.1328993681978174,
      "grad_norm": 1.6517606973648071,
      "learning_rate": 1.718930205462864e-05,
      "loss": 0.088,
      "step": 76925
    },
    {
      "epoch": 1.1332675512879045,
      "grad_norm": 1.35842764377594,
      "learning_rate": 1.718766568266166e-05,
      "loss": 0.0839,
      "step": 76950
    },
    {
      "epoch": 1.1336357343779915,
      "grad_norm": 1.1529228687286377,
      "learning_rate": 1.7186029310694674e-05,
      "loss": 0.0804,
      "step": 76975
    },
    {
      "epoch": 1.1340039174680785,
      "grad_norm": 1.8080562353134155,
      "learning_rate": 1.7184392938727688e-05,
      "loss": 0.0748,
      "step": 77000
    },
    {
      "epoch": 1.1343721005581655,
      "grad_norm": 1.4784293174743652,
      "learning_rate": 1.7182756566760706e-05,
      "loss": 0.0808,
      "step": 77025
    },
    {
      "epoch": 1.1347402836482525,
      "grad_norm": 1.4406383037567139,
      "learning_rate": 1.718112019479372e-05,
      "loss": 0.0843,
      "step": 77050
    },
    {
      "epoch": 1.1351084667383398,
      "grad_norm": 1.3282133340835571,
      "learning_rate": 1.717948382282674e-05,
      "loss": 0.0777,
      "step": 77075
    },
    {
      "epoch": 1.1354766498284268,
      "grad_norm": 1.4475878477096558,
      "learning_rate": 1.717784745085975e-05,
      "loss": 0.0832,
      "step": 77100
    },
    {
      "epoch": 1.1358448329185138,
      "grad_norm": 1.152169108390808,
      "learning_rate": 1.7176211078892767e-05,
      "loss": 0.0846,
      "step": 77125
    },
    {
      "epoch": 1.1362130160086008,
      "grad_norm": 1.4110019207000732,
      "learning_rate": 1.7174574706925782e-05,
      "loss": 0.0781,
      "step": 77150
    },
    {
      "epoch": 1.1365811990986878,
      "grad_norm": 1.5411064624786377,
      "learning_rate": 1.7172938334958796e-05,
      "loss": 0.0828,
      "step": 77175
    },
    {
      "epoch": 1.1369493821887748,
      "grad_norm": 1.9236423969268799,
      "learning_rate": 1.7171301962991814e-05,
      "loss": 0.0763,
      "step": 77200
    },
    {
      "epoch": 1.1373175652788619,
      "grad_norm": 1.0170892477035522,
      "learning_rate": 1.716966559102483e-05,
      "loss": 0.0841,
      "step": 77225
    },
    {
      "epoch": 1.1376857483689489,
      "grad_norm": 1.2837483882904053,
      "learning_rate": 1.7168029219057843e-05,
      "loss": 0.0797,
      "step": 77250
    },
    {
      "epoch": 1.1380539314590359,
      "grad_norm": 1.3705393075942993,
      "learning_rate": 1.716639284709086e-05,
      "loss": 0.0709,
      "step": 77275
    },
    {
      "epoch": 1.138422114549123,
      "grad_norm": 1.2568047046661377,
      "learning_rate": 1.7164756475123875e-05,
      "loss": 0.0761,
      "step": 77300
    },
    {
      "epoch": 1.13879029763921,
      "grad_norm": 1.4926201105117798,
      "learning_rate": 1.716312010315689e-05,
      "loss": 0.0834,
      "step": 77325
    },
    {
      "epoch": 1.1391584807292972,
      "grad_norm": 1.5288939476013184,
      "learning_rate": 1.7161483731189904e-05,
      "loss": 0.0818,
      "step": 77350
    },
    {
      "epoch": 1.1395266638193842,
      "grad_norm": 1.317007303237915,
      "learning_rate": 1.7159847359222922e-05,
      "loss": 0.0751,
      "step": 77375
    },
    {
      "epoch": 1.1398948469094712,
      "grad_norm": 1.464195966720581,
      "learning_rate": 1.7158210987255937e-05,
      "loss": 0.0836,
      "step": 77400
    },
    {
      "epoch": 1.1402630299995582,
      "grad_norm": 1.6872813701629639,
      "learning_rate": 1.715657461528895e-05,
      "loss": 0.0775,
      "step": 77425
    },
    {
      "epoch": 1.1406312130896452,
      "grad_norm": 1.4622498750686646,
      "learning_rate": 1.715493824332197e-05,
      "loss": 0.0826,
      "step": 77450
    },
    {
      "epoch": 1.1409993961797322,
      "grad_norm": 1.582440972328186,
      "learning_rate": 1.7153301871354983e-05,
      "loss": 0.0773,
      "step": 77475
    },
    {
      "epoch": 1.1413675792698192,
      "grad_norm": 1.4376102685928345,
      "learning_rate": 1.7151665499387998e-05,
      "loss": 0.0787,
      "step": 77500
    },
    {
      "epoch": 1.1417357623599063,
      "grad_norm": 1.6611912250518799,
      "learning_rate": 1.7150029127421012e-05,
      "loss": 0.0797,
      "step": 77525
    },
    {
      "epoch": 1.1421039454499935,
      "grad_norm": 1.2961845397949219,
      "learning_rate": 1.714839275545403e-05,
      "loss": 0.0709,
      "step": 77550
    },
    {
      "epoch": 1.1424721285400805,
      "grad_norm": 1.6757700443267822,
      "learning_rate": 1.7146756383487045e-05,
      "loss": 0.0801,
      "step": 77575
    },
    {
      "epoch": 1.1428403116301675,
      "grad_norm": 1.1338918209075928,
      "learning_rate": 1.714512001152006e-05,
      "loss": 0.076,
      "step": 77600
    },
    {
      "epoch": 1.1432084947202545,
      "grad_norm": 1.7071967124938965,
      "learning_rate": 1.7143483639553077e-05,
      "loss": 0.079,
      "step": 77625
    },
    {
      "epoch": 1.1435766778103416,
      "grad_norm": 1.3826173543930054,
      "learning_rate": 1.714184726758609e-05,
      "loss": 0.0789,
      "step": 77650
    },
    {
      "epoch": 1.1439448609004286,
      "grad_norm": 0.780609130859375,
      "learning_rate": 1.7140210895619106e-05,
      "loss": 0.076,
      "step": 77675
    },
    {
      "epoch": 1.1443130439905156,
      "grad_norm": 1.3437656164169312,
      "learning_rate": 1.7138574523652124e-05,
      "loss": 0.0801,
      "step": 77700
    },
    {
      "epoch": 1.1446812270806026,
      "grad_norm": 1.351599931716919,
      "learning_rate": 1.7136938151685135e-05,
      "loss": 0.0804,
      "step": 77725
    },
    {
      "epoch": 1.1450494101706896,
      "grad_norm": 1.117048978805542,
      "learning_rate": 1.7135301779718153e-05,
      "loss": 0.08,
      "step": 77750
    },
    {
      "epoch": 1.1454175932607766,
      "grad_norm": 1.6592435836791992,
      "learning_rate": 1.7133665407751167e-05,
      "loss": 0.0825,
      "step": 77775
    },
    {
      "epoch": 1.1457857763508636,
      "grad_norm": 1.3648903369903564,
      "learning_rate": 1.7132029035784185e-05,
      "loss": 0.0677,
      "step": 77800
    },
    {
      "epoch": 1.1461539594409509,
      "grad_norm": 1.1514414548873901,
      "learning_rate": 1.71303926638172e-05,
      "loss": 0.0723,
      "step": 77825
    },
    {
      "epoch": 1.146522142531038,
      "grad_norm": 1.6848379373550415,
      "learning_rate": 1.7128756291850214e-05,
      "loss": 0.0843,
      "step": 77850
    },
    {
      "epoch": 1.146890325621125,
      "grad_norm": 1.2856078147888184,
      "learning_rate": 1.712711991988323e-05,
      "loss": 0.0747,
      "step": 77875
    },
    {
      "epoch": 1.147258508711212,
      "grad_norm": 1.5376570224761963,
      "learning_rate": 1.7125483547916246e-05,
      "loss": 0.0846,
      "step": 77900
    },
    {
      "epoch": 1.147626691801299,
      "grad_norm": 1.2517526149749756,
      "learning_rate": 1.712384717594926e-05,
      "loss": 0.0805,
      "step": 77925
    },
    {
      "epoch": 1.147994874891386,
      "grad_norm": 1.3925846815109253,
      "learning_rate": 1.7122210803982275e-05,
      "loss": 0.0845,
      "step": 77950
    },
    {
      "epoch": 1.148363057981473,
      "grad_norm": 1.7829298973083496,
      "learning_rate": 1.7120574432015293e-05,
      "loss": 0.0804,
      "step": 77975
    },
    {
      "epoch": 1.14873124107156,
      "grad_norm": 1.0492154359817505,
      "learning_rate": 1.7118938060048307e-05,
      "loss": 0.0744,
      "step": 78000
    },
    {
      "epoch": 1.149099424161647,
      "grad_norm": 1.22791588306427,
      "learning_rate": 1.7117301688081322e-05,
      "loss": 0.0798,
      "step": 78025
    },
    {
      "epoch": 1.1494676072517342,
      "grad_norm": 1.5338963270187378,
      "learning_rate": 1.711566531611434e-05,
      "loss": 0.0729,
      "step": 78050
    },
    {
      "epoch": 1.1498357903418213,
      "grad_norm": 1.3798317909240723,
      "learning_rate": 1.7114028944147354e-05,
      "loss": 0.082,
      "step": 78075
    },
    {
      "epoch": 1.1502039734319083,
      "grad_norm": 1.2627533674240112,
      "learning_rate": 1.711239257218037e-05,
      "loss": 0.0806,
      "step": 78100
    },
    {
      "epoch": 1.1505721565219953,
      "grad_norm": 1.3761323690414429,
      "learning_rate": 1.7110756200213386e-05,
      "loss": 0.079,
      "step": 78125
    },
    {
      "epoch": 1.1509403396120823,
      "grad_norm": 1.0720791816711426,
      "learning_rate": 1.7109119828246397e-05,
      "loss": 0.076,
      "step": 78150
    },
    {
      "epoch": 1.1513085227021693,
      "grad_norm": 1.3640034198760986,
      "learning_rate": 1.7107483456279415e-05,
      "loss": 0.0777,
      "step": 78175
    },
    {
      "epoch": 1.1516767057922563,
      "grad_norm": 1.560114860534668,
      "learning_rate": 1.710584708431243e-05,
      "loss": 0.0805,
      "step": 78200
    },
    {
      "epoch": 1.1520448888823434,
      "grad_norm": 1.338734745979309,
      "learning_rate": 1.7104210712345448e-05,
      "loss": 0.0737,
      "step": 78225
    },
    {
      "epoch": 1.1524130719724304,
      "grad_norm": 1.5766642093658447,
      "learning_rate": 1.7102574340378462e-05,
      "loss": 0.0846,
      "step": 78250
    },
    {
      "epoch": 1.1527812550625174,
      "grad_norm": 0.9298288822174072,
      "learning_rate": 1.7100937968411476e-05,
      "loss": 0.083,
      "step": 78275
    },
    {
      "epoch": 1.1531494381526046,
      "grad_norm": 1.1307923793792725,
      "learning_rate": 1.7099301596444494e-05,
      "loss": 0.0701,
      "step": 78300
    },
    {
      "epoch": 1.1535176212426916,
      "grad_norm": 1.4159973859786987,
      "learning_rate": 1.709766522447751e-05,
      "loss": 0.0827,
      "step": 78325
    },
    {
      "epoch": 1.1538858043327787,
      "grad_norm": 1.4030858278274536,
      "learning_rate": 1.7096028852510523e-05,
      "loss": 0.0818,
      "step": 78350
    },
    {
      "epoch": 1.1542539874228657,
      "grad_norm": 1.3141252994537354,
      "learning_rate": 1.7094392480543538e-05,
      "loss": 0.078,
      "step": 78375
    },
    {
      "epoch": 1.1546221705129527,
      "grad_norm": 1.3796144723892212,
      "learning_rate": 1.7092756108576552e-05,
      "loss": 0.0788,
      "step": 78400
    },
    {
      "epoch": 1.1549903536030397,
      "grad_norm": 1.6221325397491455,
      "learning_rate": 1.709111973660957e-05,
      "loss": 0.087,
      "step": 78425
    },
    {
      "epoch": 1.1553585366931267,
      "grad_norm": 1.1633723974227905,
      "learning_rate": 1.7089483364642584e-05,
      "loss": 0.0802,
      "step": 78450
    },
    {
      "epoch": 1.1557267197832137,
      "grad_norm": 0.8109989762306213,
      "learning_rate": 1.7087846992675602e-05,
      "loss": 0.0828,
      "step": 78475
    },
    {
      "epoch": 1.1560949028733007,
      "grad_norm": 1.7017933130264282,
      "learning_rate": 1.7086210620708617e-05,
      "loss": 0.08,
      "step": 78500
    },
    {
      "epoch": 1.156463085963388,
      "grad_norm": 1.5921801328659058,
      "learning_rate": 1.708457424874163e-05,
      "loss": 0.0789,
      "step": 78525
    },
    {
      "epoch": 1.156831269053475,
      "grad_norm": 1.5675138235092163,
      "learning_rate": 1.708293787677465e-05,
      "loss": 0.0795,
      "step": 78550
    },
    {
      "epoch": 1.157199452143562,
      "grad_norm": 1.6273598670959473,
      "learning_rate": 1.708130150480766e-05,
      "loss": 0.0745,
      "step": 78575
    },
    {
      "epoch": 1.157567635233649,
      "grad_norm": 1.4480699300765991,
      "learning_rate": 1.7079665132840678e-05,
      "loss": 0.0779,
      "step": 78600
    },
    {
      "epoch": 1.157935818323736,
      "grad_norm": 1.0813771486282349,
      "learning_rate": 1.7078028760873692e-05,
      "loss": 0.0761,
      "step": 78625
    },
    {
      "epoch": 1.158304001413823,
      "grad_norm": 1.6519935131072998,
      "learning_rate": 1.7076392388906707e-05,
      "loss": 0.0796,
      "step": 78650
    },
    {
      "epoch": 1.15867218450391,
      "grad_norm": 1.2597001791000366,
      "learning_rate": 1.7074756016939725e-05,
      "loss": 0.0781,
      "step": 78675
    },
    {
      "epoch": 1.159040367593997,
      "grad_norm": 1.4550304412841797,
      "learning_rate": 1.707311964497274e-05,
      "loss": 0.075,
      "step": 78700
    },
    {
      "epoch": 1.159408550684084,
      "grad_norm": 1.2241061925888062,
      "learning_rate": 1.7071483273005757e-05,
      "loss": 0.0777,
      "step": 78725
    },
    {
      "epoch": 1.1597767337741711,
      "grad_norm": 1.7053532600402832,
      "learning_rate": 1.706984690103877e-05,
      "loss": 0.0832,
      "step": 78750
    },
    {
      "epoch": 1.1601449168642581,
      "grad_norm": 1.2709039449691772,
      "learning_rate": 1.7068210529071786e-05,
      "loss": 0.0769,
      "step": 78775
    },
    {
      "epoch": 1.1605130999543454,
      "grad_norm": 1.2542306184768677,
      "learning_rate": 1.70665741571048e-05,
      "loss": 0.084,
      "step": 78800
    },
    {
      "epoch": 1.1608812830444324,
      "grad_norm": 1.642019271850586,
      "learning_rate": 1.7064937785137815e-05,
      "loss": 0.0748,
      "step": 78825
    },
    {
      "epoch": 1.1612494661345194,
      "grad_norm": 1.9131805896759033,
      "learning_rate": 1.7063301413170833e-05,
      "loss": 0.0777,
      "step": 78850
    },
    {
      "epoch": 1.1616176492246064,
      "grad_norm": 1.2671501636505127,
      "learning_rate": 1.7061665041203847e-05,
      "loss": 0.0753,
      "step": 78875
    },
    {
      "epoch": 1.1619858323146934,
      "grad_norm": 1.4796531200408936,
      "learning_rate": 1.7060028669236865e-05,
      "loss": 0.0837,
      "step": 78900
    },
    {
      "epoch": 1.1623540154047804,
      "grad_norm": 1.3053607940673828,
      "learning_rate": 1.705839229726988e-05,
      "loss": 0.0769,
      "step": 78925
    },
    {
      "epoch": 1.1627221984948675,
      "grad_norm": 1.5992932319641113,
      "learning_rate": 1.7056755925302894e-05,
      "loss": 0.0898,
      "step": 78950
    },
    {
      "epoch": 1.1630903815849545,
      "grad_norm": 1.2795113325119019,
      "learning_rate": 1.705511955333591e-05,
      "loss": 0.0839,
      "step": 78975
    },
    {
      "epoch": 1.1634585646750417,
      "grad_norm": 1.5682183504104614,
      "learning_rate": 1.7053483181368923e-05,
      "loss": 0.0806,
      "step": 79000
    },
    {
      "epoch": 1.1638267477651287,
      "grad_norm": 1.5702452659606934,
      "learning_rate": 1.705191226428062e-05,
      "loss": 0.0745,
      "step": 79025
    },
    {
      "epoch": 1.1641949308552157,
      "grad_norm": 1.4337961673736572,
      "learning_rate": 1.7050275892313635e-05,
      "loss": 0.0761,
      "step": 79050
    },
    {
      "epoch": 1.1645631139453028,
      "grad_norm": 1.2402743101119995,
      "learning_rate": 1.704863952034665e-05,
      "loss": 0.0868,
      "step": 79075
    },
    {
      "epoch": 1.1649312970353898,
      "grad_norm": 1.2124780416488647,
      "learning_rate": 1.7047003148379667e-05,
      "loss": 0.0879,
      "step": 79100
    },
    {
      "epoch": 1.1652994801254768,
      "grad_norm": 1.216910719871521,
      "learning_rate": 1.704536677641268e-05,
      "loss": 0.0734,
      "step": 79125
    },
    {
      "epoch": 1.1656676632155638,
      "grad_norm": 1.1891676187515259,
      "learning_rate": 1.70437304044457e-05,
      "loss": 0.0838,
      "step": 79150
    },
    {
      "epoch": 1.1660358463056508,
      "grad_norm": 1.1938279867172241,
      "learning_rate": 1.7042094032478714e-05,
      "loss": 0.0838,
      "step": 79175
    },
    {
      "epoch": 1.1664040293957378,
      "grad_norm": 1.1487699747085571,
      "learning_rate": 1.7040457660511728e-05,
      "loss": 0.0892,
      "step": 79200
    },
    {
      "epoch": 1.1667722124858249,
      "grad_norm": 1.526025652885437,
      "learning_rate": 1.7038821288544743e-05,
      "loss": 0.0788,
      "step": 79225
    },
    {
      "epoch": 1.1671403955759119,
      "grad_norm": 1.1601793766021729,
      "learning_rate": 1.7037184916577757e-05,
      "loss": 0.0819,
      "step": 79250
    },
    {
      "epoch": 1.167508578665999,
      "grad_norm": 0.9991488456726074,
      "learning_rate": 1.7035548544610775e-05,
      "loss": 0.0813,
      "step": 79275
    },
    {
      "epoch": 1.1678767617560861,
      "grad_norm": 1.7756083011627197,
      "learning_rate": 1.703391217264379e-05,
      "loss": 0.0766,
      "step": 79300
    },
    {
      "epoch": 1.1682449448461731,
      "grad_norm": 1.1006495952606201,
      "learning_rate": 1.7032275800676804e-05,
      "loss": 0.0721,
      "step": 79325
    },
    {
      "epoch": 1.1686131279362602,
      "grad_norm": 1.1231677532196045,
      "learning_rate": 1.703063942870982e-05,
      "loss": 0.0805,
      "step": 79350
    },
    {
      "epoch": 1.1689813110263472,
      "grad_norm": 1.5433217287063599,
      "learning_rate": 1.7029003056742836e-05,
      "loss": 0.0744,
      "step": 79375
    },
    {
      "epoch": 1.1693494941164342,
      "grad_norm": 1.6323466300964355,
      "learning_rate": 1.702736668477585e-05,
      "loss": 0.075,
      "step": 79400
    },
    {
      "epoch": 1.1697176772065212,
      "grad_norm": 1.1453145742416382,
      "learning_rate": 1.7025730312808865e-05,
      "loss": 0.0756,
      "step": 79425
    },
    {
      "epoch": 1.1700858602966082,
      "grad_norm": 1.7219243049621582,
      "learning_rate": 1.7024093940841883e-05,
      "loss": 0.0812,
      "step": 79450
    },
    {
      "epoch": 1.1704540433866955,
      "grad_norm": 1.5594786405563354,
      "learning_rate": 1.7022457568874897e-05,
      "loss": 0.0763,
      "step": 79475
    },
    {
      "epoch": 1.1708222264767825,
      "grad_norm": 1.3090931177139282,
      "learning_rate": 1.7020821196907912e-05,
      "loss": 0.0804,
      "step": 79500
    },
    {
      "epoch": 1.1711904095668695,
      "grad_norm": 1.0847352743148804,
      "learning_rate": 1.701918482494093e-05,
      "loss": 0.0759,
      "step": 79525
    },
    {
      "epoch": 1.1715585926569565,
      "grad_norm": 1.6062463521957397,
      "learning_rate": 1.7017548452973944e-05,
      "loss": 0.0796,
      "step": 79550
    },
    {
      "epoch": 1.1719267757470435,
      "grad_norm": 1.486147403717041,
      "learning_rate": 1.701591208100696e-05,
      "loss": 0.0778,
      "step": 79575
    },
    {
      "epoch": 1.1722949588371305,
      "grad_norm": 1.7430894374847412,
      "learning_rate": 1.7014275709039976e-05,
      "loss": 0.0825,
      "step": 79600
    },
    {
      "epoch": 1.1726631419272175,
      "grad_norm": 1.096242070198059,
      "learning_rate": 1.701263933707299e-05,
      "loss": 0.0761,
      "step": 79625
    },
    {
      "epoch": 1.1730313250173046,
      "grad_norm": 0.9886026382446289,
      "learning_rate": 1.7011002965106005e-05,
      "loss": 0.0805,
      "step": 79650
    },
    {
      "epoch": 1.1733995081073916,
      "grad_norm": 1.458372712135315,
      "learning_rate": 1.700936659313902e-05,
      "loss": 0.076,
      "step": 79675
    },
    {
      "epoch": 1.1737676911974786,
      "grad_norm": 1.2455376386642456,
      "learning_rate": 1.7007730221172038e-05,
      "loss": 0.0703,
      "step": 79700
    },
    {
      "epoch": 1.1741358742875656,
      "grad_norm": 1.4454127550125122,
      "learning_rate": 1.7006093849205052e-05,
      "loss": 0.0792,
      "step": 79725
    },
    {
      "epoch": 1.1745040573776528,
      "grad_norm": 2.116722345352173,
      "learning_rate": 1.7004457477238066e-05,
      "loss": 0.0859,
      "step": 79750
    },
    {
      "epoch": 1.1748722404677399,
      "grad_norm": 1.438284993171692,
      "learning_rate": 1.7002821105271084e-05,
      "loss": 0.0767,
      "step": 79775
    },
    {
      "epoch": 1.1752404235578269,
      "grad_norm": 1.1335705518722534,
      "learning_rate": 1.70011847333041e-05,
      "loss": 0.0725,
      "step": 79800
    },
    {
      "epoch": 1.175608606647914,
      "grad_norm": 0.9960853457450867,
      "learning_rate": 1.6999548361337113e-05,
      "loss": 0.0834,
      "step": 79825
    },
    {
      "epoch": 1.175976789738001,
      "grad_norm": 1.3727959394454956,
      "learning_rate": 1.6997911989370128e-05,
      "loss": 0.0745,
      "step": 79850
    },
    {
      "epoch": 1.176344972828088,
      "grad_norm": 1.1427618265151978,
      "learning_rate": 1.6996275617403146e-05,
      "loss": 0.0817,
      "step": 79875
    },
    {
      "epoch": 1.176713155918175,
      "grad_norm": 1.2711044549942017,
      "learning_rate": 1.699463924543616e-05,
      "loss": 0.0756,
      "step": 79900
    },
    {
      "epoch": 1.177081339008262,
      "grad_norm": 1.662916660308838,
      "learning_rate": 1.6993002873469174e-05,
      "loss": 0.0805,
      "step": 79925
    },
    {
      "epoch": 1.1774495220983492,
      "grad_norm": 1.486621618270874,
      "learning_rate": 1.6991366501502192e-05,
      "loss": 0.0776,
      "step": 79950
    },
    {
      "epoch": 1.1778177051884362,
      "grad_norm": 1.3024013042449951,
      "learning_rate": 1.6989730129535207e-05,
      "loss": 0.0795,
      "step": 79975
    },
    {
      "epoch": 1.1781858882785232,
      "grad_norm": 1.5854506492614746,
      "learning_rate": 1.698809375756822e-05,
      "loss": 0.0702,
      "step": 80000
    },
    {
      "epoch": 1.1785540713686102,
      "grad_norm": 1.459588885307312,
      "learning_rate": 1.698645738560124e-05,
      "loss": 0.0752,
      "step": 80025
    },
    {
      "epoch": 1.1789222544586972,
      "grad_norm": 1.313154697418213,
      "learning_rate": 1.698482101363425e-05,
      "loss": 0.0747,
      "step": 80050
    },
    {
      "epoch": 1.1792904375487843,
      "grad_norm": 1.4655460119247437,
      "learning_rate": 1.6983184641667268e-05,
      "loss": 0.0765,
      "step": 80075
    },
    {
      "epoch": 1.1796586206388713,
      "grad_norm": 2.191974401473999,
      "learning_rate": 1.6981548269700282e-05,
      "loss": 0.0862,
      "step": 80100
    },
    {
      "epoch": 1.1800268037289583,
      "grad_norm": 1.5848077535629272,
      "learning_rate": 1.69799118977333e-05,
      "loss": 0.0793,
      "step": 80125
    },
    {
      "epoch": 1.1803949868190453,
      "grad_norm": 1.5642589330673218,
      "learning_rate": 1.6978275525766315e-05,
      "loss": 0.0833,
      "step": 80150
    },
    {
      "epoch": 1.1807631699091323,
      "grad_norm": 1.6908992528915405,
      "learning_rate": 1.697663915379933e-05,
      "loss": 0.0823,
      "step": 80175
    },
    {
      "epoch": 1.1811313529992193,
      "grad_norm": 1.2724112272262573,
      "learning_rate": 1.6975002781832347e-05,
      "loss": 0.0803,
      "step": 80200
    },
    {
      "epoch": 1.1814995360893066,
      "grad_norm": 1.3824703693389893,
      "learning_rate": 1.697336640986536e-05,
      "loss": 0.0837,
      "step": 80225
    },
    {
      "epoch": 1.1818677191793936,
      "grad_norm": 1.66892671585083,
      "learning_rate": 1.6971730037898376e-05,
      "loss": 0.0775,
      "step": 80250
    },
    {
      "epoch": 1.1822359022694806,
      "grad_norm": 1.3801589012145996,
      "learning_rate": 1.697009366593139e-05,
      "loss": 0.0768,
      "step": 80275
    },
    {
      "epoch": 1.1826040853595676,
      "grad_norm": 1.526690125465393,
      "learning_rate": 1.6968457293964408e-05,
      "loss": 0.0818,
      "step": 80300
    },
    {
      "epoch": 1.1829722684496546,
      "grad_norm": 1.5594807863235474,
      "learning_rate": 1.6966820921997423e-05,
      "loss": 0.0797,
      "step": 80325
    },
    {
      "epoch": 1.1833404515397417,
      "grad_norm": 1.5034478902816772,
      "learning_rate": 1.6965184550030437e-05,
      "loss": 0.0815,
      "step": 80350
    },
    {
      "epoch": 1.1837086346298287,
      "grad_norm": 1.3898206949234009,
      "learning_rate": 1.6963548178063455e-05,
      "loss": 0.0819,
      "step": 80375
    },
    {
      "epoch": 1.1840768177199157,
      "grad_norm": 1.6028202772140503,
      "learning_rate": 1.696191180609647e-05,
      "loss": 0.0815,
      "step": 80400
    },
    {
      "epoch": 1.1844450008100027,
      "grad_norm": 1.446126103401184,
      "learning_rate": 1.6960275434129484e-05,
      "loss": 0.075,
      "step": 80425
    },
    {
      "epoch": 1.18481318390009,
      "grad_norm": 0.9129065275192261,
      "learning_rate": 1.6958639062162502e-05,
      "loss": 0.0761,
      "step": 80450
    },
    {
      "epoch": 1.185181366990177,
      "grad_norm": 1.6920312643051147,
      "learning_rate": 1.6957002690195513e-05,
      "loss": 0.0817,
      "step": 80475
    },
    {
      "epoch": 1.185549550080264,
      "grad_norm": 1.3888028860092163,
      "learning_rate": 1.695536631822853e-05,
      "loss": 0.0728,
      "step": 80500
    },
    {
      "epoch": 1.185917733170351,
      "grad_norm": 1.6679725646972656,
      "learning_rate": 1.6953729946261545e-05,
      "loss": 0.0908,
      "step": 80525
    },
    {
      "epoch": 1.186285916260438,
      "grad_norm": 1.0872584581375122,
      "learning_rate": 1.6952093574294563e-05,
      "loss": 0.0811,
      "step": 80550
    },
    {
      "epoch": 1.186654099350525,
      "grad_norm": 1.3815594911575317,
      "learning_rate": 1.6950457202327577e-05,
      "loss": 0.0748,
      "step": 80575
    },
    {
      "epoch": 1.187022282440612,
      "grad_norm": 1.1395231485366821,
      "learning_rate": 1.6948820830360592e-05,
      "loss": 0.083,
      "step": 80600
    },
    {
      "epoch": 1.187390465530699,
      "grad_norm": 1.7929774522781372,
      "learning_rate": 1.694718445839361e-05,
      "loss": 0.0777,
      "step": 80625
    },
    {
      "epoch": 1.187758648620786,
      "grad_norm": 1.1024836301803589,
      "learning_rate": 1.6945548086426624e-05,
      "loss": 0.0779,
      "step": 80650
    },
    {
      "epoch": 1.188126831710873,
      "grad_norm": 1.374184489250183,
      "learning_rate": 1.694391171445964e-05,
      "loss": 0.0776,
      "step": 80675
    },
    {
      "epoch": 1.1884950148009603,
      "grad_norm": 1.7337616682052612,
      "learning_rate": 1.6942275342492653e-05,
      "loss": 0.08,
      "step": 80700
    },
    {
      "epoch": 1.1888631978910473,
      "grad_norm": 1.402324914932251,
      "learning_rate": 1.6940638970525668e-05,
      "loss": 0.0796,
      "step": 80725
    },
    {
      "epoch": 1.1892313809811343,
      "grad_norm": 1.525404453277588,
      "learning_rate": 1.6939002598558685e-05,
      "loss": 0.0861,
      "step": 80750
    },
    {
      "epoch": 1.1895995640712214,
      "grad_norm": 1.3129658699035645,
      "learning_rate": 1.69373662265917e-05,
      "loss": 0.0761,
      "step": 80775
    },
    {
      "epoch": 1.1899677471613084,
      "grad_norm": 1.0589076280593872,
      "learning_rate": 1.6935729854624718e-05,
      "loss": 0.0813,
      "step": 80800
    },
    {
      "epoch": 1.1903359302513954,
      "grad_norm": 1.4902567863464355,
      "learning_rate": 1.6934093482657732e-05,
      "loss": 0.0741,
      "step": 80825
    },
    {
      "epoch": 1.1907041133414824,
      "grad_norm": 1.9505789279937744,
      "learning_rate": 1.6932457110690747e-05,
      "loss": 0.0757,
      "step": 80850
    },
    {
      "epoch": 1.1910722964315694,
      "grad_norm": 1.4526811838150024,
      "learning_rate": 1.693082073872376e-05,
      "loss": 0.0771,
      "step": 80875
    },
    {
      "epoch": 1.1914404795216564,
      "grad_norm": 1.4940719604492188,
      "learning_rate": 1.6929184366756776e-05,
      "loss": 0.0744,
      "step": 80900
    },
    {
      "epoch": 1.1918086626117437,
      "grad_norm": 1.354565978050232,
      "learning_rate": 1.6927547994789793e-05,
      "loss": 0.0773,
      "step": 80925
    },
    {
      "epoch": 1.1921768457018307,
      "grad_norm": 1.7248589992523193,
      "learning_rate": 1.6925911622822808e-05,
      "loss": 0.0785,
      "step": 80950
    },
    {
      "epoch": 1.1925450287919177,
      "grad_norm": 1.612230896949768,
      "learning_rate": 1.6924275250855822e-05,
      "loss": 0.0815,
      "step": 80975
    },
    {
      "epoch": 1.1929132118820047,
      "grad_norm": 1.3987089395523071,
      "learning_rate": 1.692263887888884e-05,
      "loss": 0.0812,
      "step": 81000
    },
    {
      "epoch": 1.1932813949720917,
      "grad_norm": 1.4682742357254028,
      "learning_rate": 1.6921002506921855e-05,
      "loss": 0.0703,
      "step": 81025
    },
    {
      "epoch": 1.1936495780621788,
      "grad_norm": 1.3240442276000977,
      "learning_rate": 1.6919366134954872e-05,
      "loss": 0.0781,
      "step": 81050
    },
    {
      "epoch": 1.1940177611522658,
      "grad_norm": 1.2376230955123901,
      "learning_rate": 1.6917729762987887e-05,
      "loss": 0.0774,
      "step": 81075
    },
    {
      "epoch": 1.1943859442423528,
      "grad_norm": 1.812394618988037,
      "learning_rate": 1.69160933910209e-05,
      "loss": 0.083,
      "step": 81100
    },
    {
      "epoch": 1.1947541273324398,
      "grad_norm": 1.4060970544815063,
      "learning_rate": 1.6914457019053916e-05,
      "loss": 0.0816,
      "step": 81125
    },
    {
      "epoch": 1.1951223104225268,
      "grad_norm": 1.2166340351104736,
      "learning_rate": 1.691282064708693e-05,
      "loss": 0.0738,
      "step": 81150
    },
    {
      "epoch": 1.195490493512614,
      "grad_norm": 1.5965485572814941,
      "learning_rate": 1.6911184275119948e-05,
      "loss": 0.0772,
      "step": 81175
    },
    {
      "epoch": 1.195858676602701,
      "grad_norm": 1.1460227966308594,
      "learning_rate": 1.6909547903152963e-05,
      "loss": 0.0852,
      "step": 81200
    },
    {
      "epoch": 1.196226859692788,
      "grad_norm": 1.0590081214904785,
      "learning_rate": 1.6907911531185977e-05,
      "loss": 0.0746,
      "step": 81225
    },
    {
      "epoch": 1.196595042782875,
      "grad_norm": 1.3020614385604858,
      "learning_rate": 1.6906275159218995e-05,
      "loss": 0.0705,
      "step": 81250
    },
    {
      "epoch": 1.1969632258729621,
      "grad_norm": 1.3257240056991577,
      "learning_rate": 1.690463878725201e-05,
      "loss": 0.0785,
      "step": 81275
    },
    {
      "epoch": 1.1973314089630491,
      "grad_norm": 1.8245090246200562,
      "learning_rate": 1.6903002415285024e-05,
      "loss": 0.0788,
      "step": 81300
    },
    {
      "epoch": 1.1976995920531361,
      "grad_norm": 1.2296271324157715,
      "learning_rate": 1.6901366043318038e-05,
      "loss": 0.0747,
      "step": 81325
    },
    {
      "epoch": 1.1980677751432232,
      "grad_norm": 1.7340651750564575,
      "learning_rate": 1.6899729671351056e-05,
      "loss": 0.0846,
      "step": 81350
    },
    {
      "epoch": 1.1984359582333102,
      "grad_norm": 1.7498055696487427,
      "learning_rate": 1.689809329938407e-05,
      "loss": 0.0787,
      "step": 81375
    },
    {
      "epoch": 1.1988041413233974,
      "grad_norm": 1.2333828210830688,
      "learning_rate": 1.6896456927417085e-05,
      "loss": 0.078,
      "step": 81400
    },
    {
      "epoch": 1.1991723244134844,
      "grad_norm": 1.2235738039016724,
      "learning_rate": 1.6894820555450103e-05,
      "loss": 0.0763,
      "step": 81425
    },
    {
      "epoch": 1.1995405075035714,
      "grad_norm": 1.2170321941375732,
      "learning_rate": 1.6893184183483117e-05,
      "loss": 0.0801,
      "step": 81450
    },
    {
      "epoch": 1.1999086905936585,
      "grad_norm": 1.5779669284820557,
      "learning_rate": 1.6891547811516135e-05,
      "loss": 0.0784,
      "step": 81475
    },
    {
      "epoch": 1.2002768736837455,
      "grad_norm": 0.9630232453346252,
      "learning_rate": 1.688991143954915e-05,
      "loss": 0.0748,
      "step": 81500
    },
    {
      "epoch": 1.2006450567738325,
      "grad_norm": 1.3928108215332031,
      "learning_rate": 1.6888275067582164e-05,
      "loss": 0.0679,
      "step": 81525
    },
    {
      "epoch": 1.2010132398639195,
      "grad_norm": 1.3765554428100586,
      "learning_rate": 1.688663869561518e-05,
      "loss": 0.0748,
      "step": 81550
    },
    {
      "epoch": 1.2013814229540065,
      "grad_norm": 1.31941556930542,
      "learning_rate": 1.6885067778526872e-05,
      "loss": 0.0749,
      "step": 81575
    },
    {
      "epoch": 1.2017496060440935,
      "grad_norm": 1.8676040172576904,
      "learning_rate": 1.688343140655989e-05,
      "loss": 0.0778,
      "step": 81600
    },
    {
      "epoch": 1.2021177891341805,
      "grad_norm": 1.0498555898666382,
      "learning_rate": 1.6881795034592905e-05,
      "loss": 0.075,
      "step": 81625
    },
    {
      "epoch": 1.2024859722242676,
      "grad_norm": 1.543021559715271,
      "learning_rate": 1.688015866262592e-05,
      "loss": 0.0791,
      "step": 81650
    },
    {
      "epoch": 1.2028541553143548,
      "grad_norm": 1.4632128477096558,
      "learning_rate": 1.6878522290658937e-05,
      "loss": 0.0677,
      "step": 81675
    },
    {
      "epoch": 1.2032223384044418,
      "grad_norm": 1.2469948530197144,
      "learning_rate": 1.687688591869195e-05,
      "loss": 0.0846,
      "step": 81700
    },
    {
      "epoch": 1.2035905214945288,
      "grad_norm": 1.282270908355713,
      "learning_rate": 1.6875249546724966e-05,
      "loss": 0.0826,
      "step": 81725
    },
    {
      "epoch": 1.2039587045846158,
      "grad_norm": 1.6557286977767944,
      "learning_rate": 1.687361317475798e-05,
      "loss": 0.074,
      "step": 81750
    },
    {
      "epoch": 1.2043268876747029,
      "grad_norm": 1.2044891119003296,
      "learning_rate": 1.6871976802790998e-05,
      "loss": 0.0768,
      "step": 81775
    },
    {
      "epoch": 1.2046950707647899,
      "grad_norm": 1.0072743892669678,
      "learning_rate": 1.6870340430824013e-05,
      "loss": 0.0724,
      "step": 81800
    },
    {
      "epoch": 1.205063253854877,
      "grad_norm": 1.2253731489181519,
      "learning_rate": 1.6868704058857027e-05,
      "loss": 0.0742,
      "step": 81825
    },
    {
      "epoch": 1.205431436944964,
      "grad_norm": 1.1172629594802856,
      "learning_rate": 1.6867067686890045e-05,
      "loss": 0.0746,
      "step": 81850
    },
    {
      "epoch": 1.2057996200350511,
      "grad_norm": 1.8046514987945557,
      "learning_rate": 1.686543131492306e-05,
      "loss": 0.0853,
      "step": 81875
    },
    {
      "epoch": 1.2061678031251382,
      "grad_norm": 1.565524935722351,
      "learning_rate": 1.6863794942956074e-05,
      "loss": 0.0923,
      "step": 81900
    },
    {
      "epoch": 1.2065359862152252,
      "grad_norm": 1.6693696975708008,
      "learning_rate": 1.6862158570989092e-05,
      "loss": 0.0756,
      "step": 81925
    },
    {
      "epoch": 1.2069041693053122,
      "grad_norm": 1.2145295143127441,
      "learning_rate": 1.6860522199022106e-05,
      "loss": 0.0802,
      "step": 81950
    },
    {
      "epoch": 1.2072723523953992,
      "grad_norm": 1.9580086469650269,
      "learning_rate": 1.685888582705512e-05,
      "loss": 0.0802,
      "step": 81975
    },
    {
      "epoch": 1.2076405354854862,
      "grad_norm": 1.2828919887542725,
      "learning_rate": 1.6857249455088135e-05,
      "loss": 0.0715,
      "step": 82000
    },
    {
      "epoch": 1.2080087185755732,
      "grad_norm": 1.1126285791397095,
      "learning_rate": 1.6855613083121153e-05,
      "loss": 0.0723,
      "step": 82025
    },
    {
      "epoch": 1.2083769016656603,
      "grad_norm": 1.360633373260498,
      "learning_rate": 1.6853976711154167e-05,
      "loss": 0.0749,
      "step": 82050
    },
    {
      "epoch": 1.2087450847557473,
      "grad_norm": 1.452627420425415,
      "learning_rate": 1.6852340339187182e-05,
      "loss": 0.0825,
      "step": 82075
    },
    {
      "epoch": 1.2091132678458343,
      "grad_norm": 1.4565470218658447,
      "learning_rate": 1.68507039672202e-05,
      "loss": 0.0739,
      "step": 82100
    },
    {
      "epoch": 1.2094814509359213,
      "grad_norm": 1.4982283115386963,
      "learning_rate": 1.6849067595253214e-05,
      "loss": 0.0731,
      "step": 82125
    },
    {
      "epoch": 1.2098496340260085,
      "grad_norm": 1.1828523874282837,
      "learning_rate": 1.684743122328623e-05,
      "loss": 0.0747,
      "step": 82150
    },
    {
      "epoch": 1.2102178171160956,
      "grad_norm": 0.8931102752685547,
      "learning_rate": 1.6845794851319243e-05,
      "loss": 0.0687,
      "step": 82175
    },
    {
      "epoch": 1.2105860002061826,
      "grad_norm": 1.536070704460144,
      "learning_rate": 1.684415847935226e-05,
      "loss": 0.0774,
      "step": 82200
    },
    {
      "epoch": 1.2109541832962696,
      "grad_norm": 0.6760857105255127,
      "learning_rate": 1.6842522107385275e-05,
      "loss": 0.0739,
      "step": 82225
    },
    {
      "epoch": 1.2113223663863566,
      "grad_norm": 1.686180830001831,
      "learning_rate": 1.684088573541829e-05,
      "loss": 0.0721,
      "step": 82250
    },
    {
      "epoch": 1.2116905494764436,
      "grad_norm": 1.1491360664367676,
      "learning_rate": 1.6839249363451308e-05,
      "loss": 0.075,
      "step": 82275
    },
    {
      "epoch": 1.2120587325665306,
      "grad_norm": 1.8029288053512573,
      "learning_rate": 1.6837612991484322e-05,
      "loss": 0.0759,
      "step": 82300
    },
    {
      "epoch": 1.2124269156566176,
      "grad_norm": 1.4963215589523315,
      "learning_rate": 1.6835976619517337e-05,
      "loss": 0.0798,
      "step": 82325
    },
    {
      "epoch": 1.2127950987467049,
      "grad_norm": 1.3388421535491943,
      "learning_rate": 1.683434024755035e-05,
      "loss": 0.0718,
      "step": 82350
    },
    {
      "epoch": 1.213163281836792,
      "grad_norm": 0.9847051501274109,
      "learning_rate": 1.6832703875583366e-05,
      "loss": 0.0791,
      "step": 82375
    },
    {
      "epoch": 1.213531464926879,
      "grad_norm": 1.8953324556350708,
      "learning_rate": 1.6831067503616383e-05,
      "loss": 0.0777,
      "step": 82400
    },
    {
      "epoch": 1.213899648016966,
      "grad_norm": 1.5321537256240845,
      "learning_rate": 1.6829431131649398e-05,
      "loss": 0.0768,
      "step": 82425
    },
    {
      "epoch": 1.214267831107053,
      "grad_norm": 1.463422417640686,
      "learning_rate": 1.6827794759682416e-05,
      "loss": 0.0672,
      "step": 82450
    },
    {
      "epoch": 1.21463601419714,
      "grad_norm": 1.3408867120742798,
      "learning_rate": 1.682615838771543e-05,
      "loss": 0.0805,
      "step": 82475
    },
    {
      "epoch": 1.215004197287227,
      "grad_norm": 1.4994094371795654,
      "learning_rate": 1.6824522015748445e-05,
      "loss": 0.0741,
      "step": 82500
    },
    {
      "epoch": 1.215372380377314,
      "grad_norm": 1.6190303564071655,
      "learning_rate": 1.6822885643781462e-05,
      "loss": 0.0759,
      "step": 82525
    },
    {
      "epoch": 1.215740563467401,
      "grad_norm": 1.7373073101043701,
      "learning_rate": 1.6821249271814477e-05,
      "loss": 0.0754,
      "step": 82550
    },
    {
      "epoch": 1.216108746557488,
      "grad_norm": 1.3519747257232666,
      "learning_rate": 1.681961289984749e-05,
      "loss": 0.0713,
      "step": 82575
    },
    {
      "epoch": 1.216476929647575,
      "grad_norm": 1.243880033493042,
      "learning_rate": 1.6817976527880506e-05,
      "loss": 0.0708,
      "step": 82600
    },
    {
      "epoch": 1.2168451127376623,
      "grad_norm": 1.8985295295715332,
      "learning_rate": 1.6816340155913524e-05,
      "loss": 0.0836,
      "step": 82625
    },
    {
      "epoch": 1.2172132958277493,
      "grad_norm": 1.1356552839279175,
      "learning_rate": 1.6814703783946538e-05,
      "loss": 0.0736,
      "step": 82650
    },
    {
      "epoch": 1.2175814789178363,
      "grad_norm": 1.2517777681350708,
      "learning_rate": 1.6813067411979553e-05,
      "loss": 0.0846,
      "step": 82675
    },
    {
      "epoch": 1.2179496620079233,
      "grad_norm": 1.986774206161499,
      "learning_rate": 1.681143104001257e-05,
      "loss": 0.0798,
      "step": 82700
    },
    {
      "epoch": 1.2183178450980103,
      "grad_norm": 1.293954849243164,
      "learning_rate": 1.6809794668045585e-05,
      "loss": 0.0786,
      "step": 82725
    },
    {
      "epoch": 1.2186860281880973,
      "grad_norm": 1.5015208721160889,
      "learning_rate": 1.68081582960786e-05,
      "loss": 0.0785,
      "step": 82750
    },
    {
      "epoch": 1.2190542112781844,
      "grad_norm": 1.105033040046692,
      "learning_rate": 1.6806521924111614e-05,
      "loss": 0.0769,
      "step": 82775
    },
    {
      "epoch": 1.2194223943682714,
      "grad_norm": 1.5143176317214966,
      "learning_rate": 1.6804885552144628e-05,
      "loss": 0.077,
      "step": 82800
    },
    {
      "epoch": 1.2197905774583586,
      "grad_norm": 1.419572353363037,
      "learning_rate": 1.6803249180177646e-05,
      "loss": 0.075,
      "step": 82825
    },
    {
      "epoch": 1.2201587605484456,
      "grad_norm": 2.0381019115448,
      "learning_rate": 1.680161280821066e-05,
      "loss": 0.0826,
      "step": 82850
    },
    {
      "epoch": 1.2205269436385326,
      "grad_norm": 1.2022831439971924,
      "learning_rate": 1.679997643624368e-05,
      "loss": 0.0801,
      "step": 82875
    },
    {
      "epoch": 1.2208951267286197,
      "grad_norm": 1.3925647735595703,
      "learning_rate": 1.6798340064276693e-05,
      "loss": 0.0734,
      "step": 82900
    },
    {
      "epoch": 1.2212633098187067,
      "grad_norm": 1.3878875970840454,
      "learning_rate": 1.6796703692309707e-05,
      "loss": 0.0789,
      "step": 82925
    },
    {
      "epoch": 1.2216314929087937,
      "grad_norm": 1.5642000436782837,
      "learning_rate": 1.6795067320342725e-05,
      "loss": 0.0822,
      "step": 82950
    },
    {
      "epoch": 1.2219996759988807,
      "grad_norm": 1.860133409500122,
      "learning_rate": 1.679343094837574e-05,
      "loss": 0.0789,
      "step": 82975
    },
    {
      "epoch": 1.2223678590889677,
      "grad_norm": 1.5566082000732422,
      "learning_rate": 1.6791794576408754e-05,
      "loss": 0.0701,
      "step": 83000
    },
    {
      "epoch": 1.2227360421790547,
      "grad_norm": 1.350318193435669,
      "learning_rate": 1.679015820444177e-05,
      "loss": 0.0747,
      "step": 83025
    },
    {
      "epoch": 1.2231042252691418,
      "grad_norm": 1.1439335346221924,
      "learning_rate": 1.6788521832474783e-05,
      "loss": 0.0764,
      "step": 83050
    },
    {
      "epoch": 1.2234724083592288,
      "grad_norm": 1.71803617477417,
      "learning_rate": 1.67868854605078e-05,
      "loss": 0.078,
      "step": 83075
    },
    {
      "epoch": 1.223840591449316,
      "grad_norm": 1.3117272853851318,
      "learning_rate": 1.6785249088540815e-05,
      "loss": 0.0759,
      "step": 83100
    },
    {
      "epoch": 1.224208774539403,
      "grad_norm": 1.5179363489151,
      "learning_rate": 1.6783612716573833e-05,
      "loss": 0.0764,
      "step": 83125
    },
    {
      "epoch": 1.22457695762949,
      "grad_norm": 1.1158641576766968,
      "learning_rate": 1.6781976344606848e-05,
      "loss": 0.0755,
      "step": 83150
    },
    {
      "epoch": 1.224945140719577,
      "grad_norm": 1.7928496599197388,
      "learning_rate": 1.6780339972639862e-05,
      "loss": 0.0787,
      "step": 83175
    },
    {
      "epoch": 1.225313323809664,
      "grad_norm": 1.5240285396575928,
      "learning_rate": 1.6778703600672877e-05,
      "loss": 0.0719,
      "step": 83200
    },
    {
      "epoch": 1.225681506899751,
      "grad_norm": 1.5038111209869385,
      "learning_rate": 1.677706722870589e-05,
      "loss": 0.0817,
      "step": 83225
    },
    {
      "epoch": 1.226049689989838,
      "grad_norm": 1.4744718074798584,
      "learning_rate": 1.677543085673891e-05,
      "loss": 0.0789,
      "step": 83250
    },
    {
      "epoch": 1.2264178730799251,
      "grad_norm": 1.0492841005325317,
      "learning_rate": 1.6773794484771923e-05,
      "loss": 0.0735,
      "step": 83275
    },
    {
      "epoch": 1.2267860561700121,
      "grad_norm": 1.5979945659637451,
      "learning_rate": 1.6772158112804938e-05,
      "loss": 0.079,
      "step": 83300
    },
    {
      "epoch": 1.2271542392600994,
      "grad_norm": 1.1743595600128174,
      "learning_rate": 1.6770521740837956e-05,
      "loss": 0.0766,
      "step": 83325
    },
    {
      "epoch": 1.2275224223501864,
      "grad_norm": 1.5478302240371704,
      "learning_rate": 1.676888536887097e-05,
      "loss": 0.0719,
      "step": 83350
    },
    {
      "epoch": 1.2278906054402734,
      "grad_norm": 1.2269262075424194,
      "learning_rate": 1.6767248996903988e-05,
      "loss": 0.075,
      "step": 83375
    },
    {
      "epoch": 1.2282587885303604,
      "grad_norm": 1.1477184295654297,
      "learning_rate": 1.6765612624937002e-05,
      "loss": 0.0697,
      "step": 83400
    },
    {
      "epoch": 1.2286269716204474,
      "grad_norm": 1.6375607252120972,
      "learning_rate": 1.6763976252970017e-05,
      "loss": 0.0815,
      "step": 83425
    },
    {
      "epoch": 1.2289951547105344,
      "grad_norm": 1.5812549591064453,
      "learning_rate": 1.676233988100303e-05,
      "loss": 0.0805,
      "step": 83450
    },
    {
      "epoch": 1.2293633378006215,
      "grad_norm": 1.5101075172424316,
      "learning_rate": 1.6760703509036046e-05,
      "loss": 0.0734,
      "step": 83475
    },
    {
      "epoch": 1.2297315208907085,
      "grad_norm": 1.4215960502624512,
      "learning_rate": 1.6759067137069064e-05,
      "loss": 0.0803,
      "step": 83500
    },
    {
      "epoch": 1.2300997039807955,
      "grad_norm": 1.4927319288253784,
      "learning_rate": 1.6757430765102078e-05,
      "loss": 0.0705,
      "step": 83525
    },
    {
      "epoch": 1.2304678870708825,
      "grad_norm": 1.6058988571166992,
      "learning_rate": 1.6755794393135092e-05,
      "loss": 0.0798,
      "step": 83550
    },
    {
      "epoch": 1.2308360701609697,
      "grad_norm": 1.5166833400726318,
      "learning_rate": 1.675415802116811e-05,
      "loss": 0.0788,
      "step": 83575
    },
    {
      "epoch": 1.2312042532510568,
      "grad_norm": 0.8335561752319336,
      "learning_rate": 1.6752521649201125e-05,
      "loss": 0.0774,
      "step": 83600
    },
    {
      "epoch": 1.2315724363411438,
      "grad_norm": 1.466392159461975,
      "learning_rate": 1.675088527723414e-05,
      "loss": 0.0749,
      "step": 83625
    },
    {
      "epoch": 1.2319406194312308,
      "grad_norm": 1.4224926233291626,
      "learning_rate": 1.6749248905267154e-05,
      "loss": 0.0791,
      "step": 83650
    },
    {
      "epoch": 1.2323088025213178,
      "grad_norm": 1.4799292087554932,
      "learning_rate": 1.674761253330017e-05,
      "loss": 0.0789,
      "step": 83675
    },
    {
      "epoch": 1.2326769856114048,
      "grad_norm": 0.98297119140625,
      "learning_rate": 1.6745976161333186e-05,
      "loss": 0.0749,
      "step": 83700
    },
    {
      "epoch": 1.2330451687014918,
      "grad_norm": 1.6584978103637695,
      "learning_rate": 1.67443397893662e-05,
      "loss": 0.0703,
      "step": 83725
    },
    {
      "epoch": 1.2334133517915788,
      "grad_norm": 1.3904367685317993,
      "learning_rate": 1.6742703417399218e-05,
      "loss": 0.0796,
      "step": 83750
    },
    {
      "epoch": 1.2337815348816659,
      "grad_norm": 1.6995155811309814,
      "learning_rate": 1.6741067045432233e-05,
      "loss": 0.0689,
      "step": 83775
    },
    {
      "epoch": 1.234149717971753,
      "grad_norm": 0.8751583695411682,
      "learning_rate": 1.673943067346525e-05,
      "loss": 0.0747,
      "step": 83800
    },
    {
      "epoch": 1.2345179010618401,
      "grad_norm": 1.6611430644989014,
      "learning_rate": 1.673785975637694e-05,
      "loss": 0.073,
      "step": 83825
    },
    {
      "epoch": 1.2348860841519271,
      "grad_norm": 1.1409391164779663,
      "learning_rate": 1.673622338440996e-05,
      "loss": 0.0754,
      "step": 83850
    },
    {
      "epoch": 1.2352542672420141,
      "grad_norm": 1.663902997970581,
      "learning_rate": 1.6734587012442973e-05,
      "loss": 0.0827,
      "step": 83875
    },
    {
      "epoch": 1.2356224503321012,
      "grad_norm": 1.592111349105835,
      "learning_rate": 1.6732950640475988e-05,
      "loss": 0.0795,
      "step": 83900
    },
    {
      "epoch": 1.2359906334221882,
      "grad_norm": 1.6744433641433716,
      "learning_rate": 1.6731314268509006e-05,
      "loss": 0.0761,
      "step": 83925
    },
    {
      "epoch": 1.2363588165122752,
      "grad_norm": 1.1632553339004517,
      "learning_rate": 1.672967789654202e-05,
      "loss": 0.0768,
      "step": 83950
    },
    {
      "epoch": 1.2367269996023622,
      "grad_norm": 1.0109707117080688,
      "learning_rate": 1.6728041524575035e-05,
      "loss": 0.0764,
      "step": 83975
    },
    {
      "epoch": 1.2370951826924492,
      "grad_norm": 1.5003042221069336,
      "learning_rate": 1.6726405152608052e-05,
      "loss": 0.0745,
      "step": 84000
    },
    {
      "epoch": 1.2374633657825362,
      "grad_norm": 1.63449227809906,
      "learning_rate": 1.6724768780641067e-05,
      "loss": 0.075,
      "step": 84025
    },
    {
      "epoch": 1.2378315488726233,
      "grad_norm": 1.4061245918273926,
      "learning_rate": 1.672313240867408e-05,
      "loss": 0.0745,
      "step": 84050
    },
    {
      "epoch": 1.2381997319627105,
      "grad_norm": 1.9697695970535278,
      "learning_rate": 1.6721496036707096e-05,
      "loss": 0.0803,
      "step": 84075
    },
    {
      "epoch": 1.2385679150527975,
      "grad_norm": 1.6412981748580933,
      "learning_rate": 1.6719859664740114e-05,
      "loss": 0.0777,
      "step": 84100
    },
    {
      "epoch": 1.2389360981428845,
      "grad_norm": 1.7949827909469604,
      "learning_rate": 1.6718223292773128e-05,
      "loss": 0.0775,
      "step": 84125
    },
    {
      "epoch": 1.2393042812329715,
      "grad_norm": 1.202728509902954,
      "learning_rate": 1.6716586920806143e-05,
      "loss": 0.0766,
      "step": 84150
    },
    {
      "epoch": 1.2396724643230586,
      "grad_norm": 1.2260409593582153,
      "learning_rate": 1.671495054883916e-05,
      "loss": 0.0745,
      "step": 84175
    },
    {
      "epoch": 1.2400406474131456,
      "grad_norm": 1.3268578052520752,
      "learning_rate": 1.6713314176872175e-05,
      "loss": 0.0744,
      "step": 84200
    },
    {
      "epoch": 1.2404088305032326,
      "grad_norm": 1.253363847732544,
      "learning_rate": 1.671167780490519e-05,
      "loss": 0.0689,
      "step": 84225
    },
    {
      "epoch": 1.2407770135933196,
      "grad_norm": 1.3394206762313843,
      "learning_rate": 1.6710041432938204e-05,
      "loss": 0.0818,
      "step": 84250
    },
    {
      "epoch": 1.2411451966834068,
      "grad_norm": 1.2930598258972168,
      "learning_rate": 1.670840506097122e-05,
      "loss": 0.0715,
      "step": 84275
    },
    {
      "epoch": 1.2415133797734939,
      "grad_norm": 1.5809293985366821,
      "learning_rate": 1.6706768689004236e-05,
      "loss": 0.0721,
      "step": 84300
    },
    {
      "epoch": 1.2418815628635809,
      "grad_norm": 1.3983064889907837,
      "learning_rate": 1.670513231703725e-05,
      "loss": 0.0745,
      "step": 84325
    },
    {
      "epoch": 1.2422497459536679,
      "grad_norm": 1.2547566890716553,
      "learning_rate": 1.670349594507027e-05,
      "loss": 0.0841,
      "step": 84350
    },
    {
      "epoch": 1.242617929043755,
      "grad_norm": 1.2029368877410889,
      "learning_rate": 1.6701859573103283e-05,
      "loss": 0.0753,
      "step": 84375
    },
    {
      "epoch": 1.242986112133842,
      "grad_norm": 1.7967641353607178,
      "learning_rate": 1.6700223201136297e-05,
      "loss": 0.0829,
      "step": 84400
    },
    {
      "epoch": 1.243354295223929,
      "grad_norm": 1.3531262874603271,
      "learning_rate": 1.6698586829169315e-05,
      "loss": 0.079,
      "step": 84425
    },
    {
      "epoch": 1.243722478314016,
      "grad_norm": 1.3004512786865234,
      "learning_rate": 1.669695045720233e-05,
      "loss": 0.0766,
      "step": 84450
    },
    {
      "epoch": 1.244090661404103,
      "grad_norm": 1.967198371887207,
      "learning_rate": 1.6695314085235344e-05,
      "loss": 0.0836,
      "step": 84475
    },
    {
      "epoch": 1.24445884449419,
      "grad_norm": 0.9700692296028137,
      "learning_rate": 1.669367771326836e-05,
      "loss": 0.0746,
      "step": 84500
    },
    {
      "epoch": 1.244827027584277,
      "grad_norm": 1.2876149415969849,
      "learning_rate": 1.6692041341301376e-05,
      "loss": 0.079,
      "step": 84525
    },
    {
      "epoch": 1.2451952106743642,
      "grad_norm": 1.5020794868469238,
      "learning_rate": 1.669040496933439e-05,
      "loss": 0.0699,
      "step": 84550
    },
    {
      "epoch": 1.2455633937644512,
      "grad_norm": 1.4075846672058105,
      "learning_rate": 1.6688768597367405e-05,
      "loss": 0.0793,
      "step": 84575
    },
    {
      "epoch": 1.2459315768545383,
      "grad_norm": 1.3767808675765991,
      "learning_rate": 1.6687132225400423e-05,
      "loss": 0.0792,
      "step": 84600
    },
    {
      "epoch": 1.2462997599446253,
      "grad_norm": 0.9941942691802979,
      "learning_rate": 1.6685495853433438e-05,
      "loss": 0.073,
      "step": 84625
    },
    {
      "epoch": 1.2466679430347123,
      "grad_norm": 1.1486207246780396,
      "learning_rate": 1.6683859481466452e-05,
      "loss": 0.076,
      "step": 84650
    },
    {
      "epoch": 1.2470361261247993,
      "grad_norm": 1.2689430713653564,
      "learning_rate": 1.6682223109499467e-05,
      "loss": 0.0777,
      "step": 84675
    },
    {
      "epoch": 1.2474043092148863,
      "grad_norm": 1.8343433141708374,
      "learning_rate": 1.668058673753248e-05,
      "loss": 0.0913,
      "step": 84700
    },
    {
      "epoch": 1.2477724923049733,
      "grad_norm": 1.5430928468704224,
      "learning_rate": 1.66789503655655e-05,
      "loss": 0.0836,
      "step": 84725
    },
    {
      "epoch": 1.2481406753950606,
      "grad_norm": 0.7241131663322449,
      "learning_rate": 1.6677313993598513e-05,
      "loss": 0.0722,
      "step": 84750
    },
    {
      "epoch": 1.2485088584851476,
      "grad_norm": 1.7381606101989746,
      "learning_rate": 1.667567762163153e-05,
      "loss": 0.0818,
      "step": 84775
    },
    {
      "epoch": 1.2488770415752346,
      "grad_norm": 1.4786512851715088,
      "learning_rate": 1.6674041249664546e-05,
      "loss": 0.071,
      "step": 84800
    },
    {
      "epoch": 1.2492452246653216,
      "grad_norm": 1.4238206148147583,
      "learning_rate": 1.667240487769756e-05,
      "loss": 0.0771,
      "step": 84825
    },
    {
      "epoch": 1.2496134077554086,
      "grad_norm": 1.4153484106063843,
      "learning_rate": 1.6670768505730578e-05,
      "loss": 0.0746,
      "step": 84850
    },
    {
      "epoch": 1.2499815908454956,
      "grad_norm": 1.505297064781189,
      "learning_rate": 1.6669132133763592e-05,
      "loss": 0.0743,
      "step": 84875
    },
    {
      "epoch": 1.2503497739355827,
      "grad_norm": 1.3245985507965088,
      "learning_rate": 1.6667495761796607e-05,
      "loss": 0.0695,
      "step": 84900
    },
    {
      "epoch": 1.2507179570256697,
      "grad_norm": 1.737091064453125,
      "learning_rate": 1.666585938982962e-05,
      "loss": 0.0752,
      "step": 84925
    },
    {
      "epoch": 1.2510861401157567,
      "grad_norm": 1.7985749244689941,
      "learning_rate": 1.6664223017862636e-05,
      "loss": 0.0742,
      "step": 84950
    },
    {
      "epoch": 1.2514543232058437,
      "grad_norm": 0.9943830966949463,
      "learning_rate": 1.6662586645895654e-05,
      "loss": 0.0744,
      "step": 84975
    },
    {
      "epoch": 1.2518225062959307,
      "grad_norm": 1.1452761888504028,
      "learning_rate": 1.6660950273928668e-05,
      "loss": 0.0845,
      "step": 85000
    },
    {
      "epoch": 1.2521906893860177,
      "grad_norm": 1.0950809717178345,
      "learning_rate": 1.6659313901961686e-05,
      "loss": 0.0722,
      "step": 85025
    },
    {
      "epoch": 1.252558872476105,
      "grad_norm": 1.2069015502929688,
      "learning_rate": 1.66576775299947e-05,
      "loss": 0.0816,
      "step": 85050
    },
    {
      "epoch": 1.252927055566192,
      "grad_norm": 1.7069686651229858,
      "learning_rate": 1.6656041158027715e-05,
      "loss": 0.0764,
      "step": 85075
    },
    {
      "epoch": 1.253295238656279,
      "grad_norm": 1.6437897682189941,
      "learning_rate": 1.665440478606073e-05,
      "loss": 0.0744,
      "step": 85100
    },
    {
      "epoch": 1.253663421746366,
      "grad_norm": 1.4827040433883667,
      "learning_rate": 1.6652768414093744e-05,
      "loss": 0.0774,
      "step": 85125
    },
    {
      "epoch": 1.254031604836453,
      "grad_norm": 1.4733189344406128,
      "learning_rate": 1.665113204212676e-05,
      "loss": 0.08,
      "step": 85150
    },
    {
      "epoch": 1.25439978792654,
      "grad_norm": 2.1713707447052,
      "learning_rate": 1.6649495670159776e-05,
      "loss": 0.0747,
      "step": 85175
    },
    {
      "epoch": 1.254767971016627,
      "grad_norm": 1.3723907470703125,
      "learning_rate": 1.6647859298192794e-05,
      "loss": 0.0778,
      "step": 85200
    },
    {
      "epoch": 1.2551361541067143,
      "grad_norm": 1.3837295770645142,
      "learning_rate": 1.6646222926225808e-05,
      "loss": 0.0764,
      "step": 85225
    },
    {
      "epoch": 1.2555043371968013,
      "grad_norm": 1.6274299621582031,
      "learning_rate": 1.6644586554258823e-05,
      "loss": 0.0764,
      "step": 85250
    },
    {
      "epoch": 1.2558725202868883,
      "grad_norm": 1.6906304359436035,
      "learning_rate": 1.664295018229184e-05,
      "loss": 0.0783,
      "step": 85275
    },
    {
      "epoch": 1.2562407033769754,
      "grad_norm": 1.3737411499023438,
      "learning_rate": 1.664131381032485e-05,
      "loss": 0.0753,
      "step": 85300
    },
    {
      "epoch": 1.2566088864670624,
      "grad_norm": 1.203748106956482,
      "learning_rate": 1.663967743835787e-05,
      "loss": 0.0721,
      "step": 85325
    },
    {
      "epoch": 1.2569770695571494,
      "grad_norm": 1.1847542524337769,
      "learning_rate": 1.6638041066390884e-05,
      "loss": 0.0771,
      "step": 85350
    },
    {
      "epoch": 1.2573452526472364,
      "grad_norm": 1.7089818716049194,
      "learning_rate": 1.66364046944239e-05,
      "loss": 0.0826,
      "step": 85375
    },
    {
      "epoch": 1.2577134357373234,
      "grad_norm": 1.32912278175354,
      "learning_rate": 1.6634768322456916e-05,
      "loss": 0.0795,
      "step": 85400
    },
    {
      "epoch": 1.2580816188274104,
      "grad_norm": 1.3430333137512207,
      "learning_rate": 1.663313195048993e-05,
      "loss": 0.0755,
      "step": 85425
    },
    {
      "epoch": 1.2584498019174974,
      "grad_norm": 1.7742334604263306,
      "learning_rate": 1.663149557852295e-05,
      "loss": 0.0813,
      "step": 85450
    },
    {
      "epoch": 1.2588179850075845,
      "grad_norm": 1.0110199451446533,
      "learning_rate": 1.6629859206555963e-05,
      "loss": 0.0759,
      "step": 85475
    },
    {
      "epoch": 1.2591861680976715,
      "grad_norm": 1.4277589321136475,
      "learning_rate": 1.6628222834588978e-05,
      "loss": 0.0724,
      "step": 85500
    },
    {
      "epoch": 1.2595543511877587,
      "grad_norm": 1.7643581628799438,
      "learning_rate": 1.6626586462621992e-05,
      "loss": 0.0808,
      "step": 85525
    },
    {
      "epoch": 1.2599225342778457,
      "grad_norm": 1.358424425125122,
      "learning_rate": 1.6624950090655006e-05,
      "loss": 0.0746,
      "step": 85550
    },
    {
      "epoch": 1.2602907173679327,
      "grad_norm": 1.502968192100525,
      "learning_rate": 1.6623313718688024e-05,
      "loss": 0.0787,
      "step": 85575
    },
    {
      "epoch": 1.2606589004580198,
      "grad_norm": 1.6026368141174316,
      "learning_rate": 1.662167734672104e-05,
      "loss": 0.0753,
      "step": 85600
    },
    {
      "epoch": 1.2610270835481068,
      "grad_norm": 1.2125557661056519,
      "learning_rate": 1.6620040974754053e-05,
      "loss": 0.0786,
      "step": 85625
    },
    {
      "epoch": 1.2613952666381938,
      "grad_norm": 1.624697208404541,
      "learning_rate": 1.661840460278707e-05,
      "loss": 0.0813,
      "step": 85650
    },
    {
      "epoch": 1.2617634497282808,
      "grad_norm": 1.8684204816818237,
      "learning_rate": 1.6616768230820085e-05,
      "loss": 0.0823,
      "step": 85675
    },
    {
      "epoch": 1.262131632818368,
      "grad_norm": 1.1944184303283691,
      "learning_rate": 1.6615131858853103e-05,
      "loss": 0.0773,
      "step": 85700
    },
    {
      "epoch": 1.262499815908455,
      "grad_norm": 1.3293205499649048,
      "learning_rate": 1.6613495486886114e-05,
      "loss": 0.0754,
      "step": 85725
    },
    {
      "epoch": 1.262867998998542,
      "grad_norm": 1.7817025184631348,
      "learning_rate": 1.6611859114919132e-05,
      "loss": 0.0836,
      "step": 85750
    },
    {
      "epoch": 1.263236182088629,
      "grad_norm": 1.3254010677337646,
      "learning_rate": 1.6610222742952147e-05,
      "loss": 0.0792,
      "step": 85775
    },
    {
      "epoch": 1.263604365178716,
      "grad_norm": 1.5083951950073242,
      "learning_rate": 1.660858637098516e-05,
      "loss": 0.0799,
      "step": 85800
    },
    {
      "epoch": 1.2639725482688031,
      "grad_norm": 1.5599150657653809,
      "learning_rate": 1.660694999901818e-05,
      "loss": 0.0726,
      "step": 85825
    },
    {
      "epoch": 1.2643407313588901,
      "grad_norm": 1.8269362449645996,
      "learning_rate": 1.6605313627051193e-05,
      "loss": 0.0835,
      "step": 85850
    },
    {
      "epoch": 1.2647089144489772,
      "grad_norm": 1.7932054996490479,
      "learning_rate": 1.6603677255084208e-05,
      "loss": 0.0668,
      "step": 85875
    },
    {
      "epoch": 1.2650770975390642,
      "grad_norm": 1.5463165044784546,
      "learning_rate": 1.6602040883117226e-05,
      "loss": 0.0743,
      "step": 85900
    },
    {
      "epoch": 1.2654452806291512,
      "grad_norm": 1.6358023881912231,
      "learning_rate": 1.660040451115024e-05,
      "loss": 0.0823,
      "step": 85925
    },
    {
      "epoch": 1.2658134637192382,
      "grad_norm": 1.7893600463867188,
      "learning_rate": 1.6598768139183255e-05,
      "loss": 0.076,
      "step": 85950
    },
    {
      "epoch": 1.2661816468093252,
      "grad_norm": 1.8993571996688843,
      "learning_rate": 1.659713176721627e-05,
      "loss": 0.082,
      "step": 85975
    },
    {
      "epoch": 1.2665498298994124,
      "grad_norm": 1.109728455543518,
      "learning_rate": 1.6595495395249287e-05,
      "loss": 0.0789,
      "step": 86000
    },
    {
      "epoch": 1.2669180129894995,
      "grad_norm": 1.1833776235580444,
      "learning_rate": 1.65938590232823e-05,
      "loss": 0.0756,
      "step": 86025
    },
    {
      "epoch": 1.2672861960795865,
      "grad_norm": 1.4657959938049316,
      "learning_rate": 1.6592222651315316e-05,
      "loss": 0.0772,
      "step": 86050
    },
    {
      "epoch": 1.2676543791696735,
      "grad_norm": 1.0066750049591064,
      "learning_rate": 1.6590586279348334e-05,
      "loss": 0.0813,
      "step": 86075
    },
    {
      "epoch": 1.2680225622597605,
      "grad_norm": 1.5638108253479004,
      "learning_rate": 1.6588949907381348e-05,
      "loss": 0.0755,
      "step": 86100
    },
    {
      "epoch": 1.2683907453498475,
      "grad_norm": 1.938494324684143,
      "learning_rate": 1.6587313535414366e-05,
      "loss": 0.0768,
      "step": 86125
    },
    {
      "epoch": 1.2687589284399345,
      "grad_norm": 1.5280470848083496,
      "learning_rate": 1.6585677163447377e-05,
      "loss": 0.0831,
      "step": 86150
    },
    {
      "epoch": 1.2691271115300218,
      "grad_norm": 1.5612305402755737,
      "learning_rate": 1.6584040791480395e-05,
      "loss": 0.085,
      "step": 86175
    },
    {
      "epoch": 1.2694952946201088,
      "grad_norm": 1.2244549989700317,
      "learning_rate": 1.658240441951341e-05,
      "loss": 0.0804,
      "step": 86200
    },
    {
      "epoch": 1.2698634777101958,
      "grad_norm": 1.543442726135254,
      "learning_rate": 1.6580768047546424e-05,
      "loss": 0.0783,
      "step": 86225
    },
    {
      "epoch": 1.2702316608002828,
      "grad_norm": 0.8403768539428711,
      "learning_rate": 1.6579131675579442e-05,
      "loss": 0.0834,
      "step": 86250
    },
    {
      "epoch": 1.2705998438903698,
      "grad_norm": 1.3470231294631958,
      "learning_rate": 1.6577495303612456e-05,
      "loss": 0.0807,
      "step": 86275
    },
    {
      "epoch": 1.2709680269804569,
      "grad_norm": 1.3456494808197021,
      "learning_rate": 1.657585893164547e-05,
      "loss": 0.0715,
      "step": 86300
    },
    {
      "epoch": 1.2713362100705439,
      "grad_norm": 1.604638934135437,
      "learning_rate": 1.657422255967849e-05,
      "loss": 0.0746,
      "step": 86325
    },
    {
      "epoch": 1.2717043931606309,
      "grad_norm": 1.5782653093338013,
      "learning_rate": 1.6572586187711503e-05,
      "loss": 0.0749,
      "step": 86350
    },
    {
      "epoch": 1.272072576250718,
      "grad_norm": 1.2369173765182495,
      "learning_rate": 1.6570949815744517e-05,
      "loss": 0.0734,
      "step": 86375
    },
    {
      "epoch": 1.272440759340805,
      "grad_norm": 1.6135135889053345,
      "learning_rate": 1.6569313443777532e-05,
      "loss": 0.0769,
      "step": 86400
    },
    {
      "epoch": 1.272808942430892,
      "grad_norm": 1.2767653465270996,
      "learning_rate": 1.656767707181055e-05,
      "loss": 0.0781,
      "step": 86425
    },
    {
      "epoch": 1.273177125520979,
      "grad_norm": 1.665022850036621,
      "learning_rate": 1.6566040699843564e-05,
      "loss": 0.0756,
      "step": 86450
    },
    {
      "epoch": 1.2735453086110662,
      "grad_norm": 1.2414312362670898,
      "learning_rate": 1.6564469782755258e-05,
      "loss": 0.0702,
      "step": 86475
    },
    {
      "epoch": 1.2739134917011532,
      "grad_norm": 1.4579577445983887,
      "learning_rate": 1.6562833410788276e-05,
      "loss": 0.0786,
      "step": 86500
    },
    {
      "epoch": 1.2742816747912402,
      "grad_norm": 1.1833775043487549,
      "learning_rate": 1.656119703882129e-05,
      "loss": 0.0758,
      "step": 86525
    },
    {
      "epoch": 1.2746498578813272,
      "grad_norm": 1.3675172328948975,
      "learning_rate": 1.6559560666854305e-05,
      "loss": 0.068,
      "step": 86550
    },
    {
      "epoch": 1.2750180409714142,
      "grad_norm": 1.6105674505233765,
      "learning_rate": 1.655792429488732e-05,
      "loss": 0.0749,
      "step": 86575
    },
    {
      "epoch": 1.2753862240615013,
      "grad_norm": 1.4817944765090942,
      "learning_rate": 1.6556287922920337e-05,
      "loss": 0.0767,
      "step": 86600
    },
    {
      "epoch": 1.2757544071515883,
      "grad_norm": 1.2722810506820679,
      "learning_rate": 1.655465155095335e-05,
      "loss": 0.0788,
      "step": 86625
    },
    {
      "epoch": 1.2761225902416755,
      "grad_norm": 1.7131518125534058,
      "learning_rate": 1.6553015178986366e-05,
      "loss": 0.0802,
      "step": 86650
    },
    {
      "epoch": 1.2764907733317625,
      "grad_norm": 0.9292106032371521,
      "learning_rate": 1.6551378807019384e-05,
      "loss": 0.0774,
      "step": 86675
    },
    {
      "epoch": 1.2768589564218495,
      "grad_norm": 1.4344921112060547,
      "learning_rate": 1.65497424350524e-05,
      "loss": 0.0754,
      "step": 86700
    },
    {
      "epoch": 1.2772271395119366,
      "grad_norm": 0.9840370416641235,
      "learning_rate": 1.6548106063085413e-05,
      "loss": 0.0745,
      "step": 86725
    },
    {
      "epoch": 1.2775953226020236,
      "grad_norm": 1.4911800622940063,
      "learning_rate": 1.654646969111843e-05,
      "loss": 0.0777,
      "step": 86750
    },
    {
      "epoch": 1.2779635056921106,
      "grad_norm": 1.4150290489196777,
      "learning_rate": 1.6544833319151445e-05,
      "loss": 0.077,
      "step": 86775
    },
    {
      "epoch": 1.2783316887821976,
      "grad_norm": 1.4077993631362915,
      "learning_rate": 1.654319694718446e-05,
      "loss": 0.0745,
      "step": 86800
    },
    {
      "epoch": 1.2786998718722846,
      "grad_norm": 1.453959345817566,
      "learning_rate": 1.6541560575217474e-05,
      "loss": 0.0801,
      "step": 86825
    },
    {
      "epoch": 1.2790680549623716,
      "grad_norm": 1.2002153396606445,
      "learning_rate": 1.6539924203250492e-05,
      "loss": 0.0646,
      "step": 86850
    },
    {
      "epoch": 1.2794362380524587,
      "grad_norm": 1.3084107637405396,
      "learning_rate": 1.6538287831283506e-05,
      "loss": 0.0697,
      "step": 86875
    },
    {
      "epoch": 1.2798044211425457,
      "grad_norm": 1.1693037748336792,
      "learning_rate": 1.653665145931652e-05,
      "loss": 0.0817,
      "step": 86900
    },
    {
      "epoch": 1.2801726042326327,
      "grad_norm": 1.7182427644729614,
      "learning_rate": 1.653501508734954e-05,
      "loss": 0.0742,
      "step": 86925
    },
    {
      "epoch": 1.28054078732272,
      "grad_norm": 1.6868401765823364,
      "learning_rate": 1.6533378715382553e-05,
      "loss": 0.0786,
      "step": 86950
    },
    {
      "epoch": 1.280908970412807,
      "grad_norm": 1.5034259557724,
      "learning_rate": 1.6531742343415568e-05,
      "loss": 0.0782,
      "step": 86975
    },
    {
      "epoch": 1.281277153502894,
      "grad_norm": 1.451076626777649,
      "learning_rate": 1.6530105971448582e-05,
      "loss": 0.0834,
      "step": 87000
    },
    {
      "epoch": 1.281645336592981,
      "grad_norm": 1.09324049949646,
      "learning_rate": 1.6528469599481596e-05,
      "loss": 0.0761,
      "step": 87025
    },
    {
      "epoch": 1.282013519683068,
      "grad_norm": 1.338484287261963,
      "learning_rate": 1.6526833227514614e-05,
      "loss": 0.0745,
      "step": 87050
    },
    {
      "epoch": 1.282381702773155,
      "grad_norm": 1.4144775867462158,
      "learning_rate": 1.652519685554763e-05,
      "loss": 0.0789,
      "step": 87075
    },
    {
      "epoch": 1.282749885863242,
      "grad_norm": 2.0029828548431396,
      "learning_rate": 1.6523560483580647e-05,
      "loss": 0.0768,
      "step": 87100
    },
    {
      "epoch": 1.2831180689533292,
      "grad_norm": 1.4517130851745605,
      "learning_rate": 1.652192411161366e-05,
      "loss": 0.0833,
      "step": 87125
    },
    {
      "epoch": 1.2834862520434163,
      "grad_norm": 1.5844374895095825,
      "learning_rate": 1.6520287739646675e-05,
      "loss": 0.0791,
      "step": 87150
    },
    {
      "epoch": 1.2838544351335033,
      "grad_norm": 0.8651666641235352,
      "learning_rate": 1.6518651367679693e-05,
      "loss": 0.0762,
      "step": 87175
    },
    {
      "epoch": 1.2842226182235903,
      "grad_norm": 0.9691341519355774,
      "learning_rate": 1.6517014995712704e-05,
      "loss": 0.0829,
      "step": 87200
    },
    {
      "epoch": 1.2845908013136773,
      "grad_norm": 1.4265964031219482,
      "learning_rate": 1.6515378623745722e-05,
      "loss": 0.0762,
      "step": 87225
    },
    {
      "epoch": 1.2849589844037643,
      "grad_norm": 1.5003539323806763,
      "learning_rate": 1.6513742251778737e-05,
      "loss": 0.0776,
      "step": 87250
    },
    {
      "epoch": 1.2853271674938513,
      "grad_norm": 1.250416874885559,
      "learning_rate": 1.651210587981175e-05,
      "loss": 0.0765,
      "step": 87275
    },
    {
      "epoch": 1.2856953505839384,
      "grad_norm": 1.2348967790603638,
      "learning_rate": 1.651046950784477e-05,
      "loss": 0.0771,
      "step": 87300
    },
    {
      "epoch": 1.2860635336740254,
      "grad_norm": 1.6736505031585693,
      "learning_rate": 1.6508833135877783e-05,
      "loss": 0.0869,
      "step": 87325
    },
    {
      "epoch": 1.2864317167641124,
      "grad_norm": 1.2044596672058105,
      "learning_rate": 1.65071967639108e-05,
      "loss": 0.0735,
      "step": 87350
    },
    {
      "epoch": 1.2867998998541994,
      "grad_norm": 1.6578843593597412,
      "learning_rate": 1.6505560391943816e-05,
      "loss": 0.0821,
      "step": 87375
    },
    {
      "epoch": 1.2871680829442864,
      "grad_norm": 1.4457515478134155,
      "learning_rate": 1.650392401997683e-05,
      "loss": 0.0755,
      "step": 87400
    },
    {
      "epoch": 1.2875362660343737,
      "grad_norm": 1.2665610313415527,
      "learning_rate": 1.6502287648009845e-05,
      "loss": 0.0786,
      "step": 87425
    },
    {
      "epoch": 1.2879044491244607,
      "grad_norm": 1.1844627857208252,
      "learning_rate": 1.650065127604286e-05,
      "loss": 0.0763,
      "step": 87450
    },
    {
      "epoch": 1.2882726322145477,
      "grad_norm": 1.7764798402786255,
      "learning_rate": 1.6499014904075877e-05,
      "loss": 0.0793,
      "step": 87475
    },
    {
      "epoch": 1.2886408153046347,
      "grad_norm": 1.5905070304870605,
      "learning_rate": 1.649737853210889e-05,
      "loss": 0.0742,
      "step": 87500
    },
    {
      "epoch": 1.2890089983947217,
      "grad_norm": 1.3501182794570923,
      "learning_rate": 1.649574216014191e-05,
      "loss": 0.0752,
      "step": 87525
    },
    {
      "epoch": 1.2893771814848087,
      "grad_norm": 1.1950966119766235,
      "learning_rate": 1.6494105788174924e-05,
      "loss": 0.0703,
      "step": 87550
    },
    {
      "epoch": 1.2897453645748957,
      "grad_norm": 1.5274617671966553,
      "learning_rate": 1.6492469416207938e-05,
      "loss": 0.0786,
      "step": 87575
    },
    {
      "epoch": 1.2901135476649828,
      "grad_norm": 1.3678492307662964,
      "learning_rate": 1.6490833044240956e-05,
      "loss": 0.0773,
      "step": 87600
    },
    {
      "epoch": 1.29048173075507,
      "grad_norm": 1.5120971202850342,
      "learning_rate": 1.6489196672273967e-05,
      "loss": 0.0864,
      "step": 87625
    },
    {
      "epoch": 1.290849913845157,
      "grad_norm": 0.7234932780265808,
      "learning_rate": 1.6487560300306985e-05,
      "loss": 0.0782,
      "step": 87650
    },
    {
      "epoch": 1.291218096935244,
      "grad_norm": 1.4716856479644775,
      "learning_rate": 1.648592392834e-05,
      "loss": 0.089,
      "step": 87675
    },
    {
      "epoch": 1.291586280025331,
      "grad_norm": 1.5008195638656616,
      "learning_rate": 1.6484287556373014e-05,
      "loss": 0.0723,
      "step": 87700
    },
    {
      "epoch": 1.291954463115418,
      "grad_norm": 1.3393083810806274,
      "learning_rate": 1.6482651184406032e-05,
      "loss": 0.0762,
      "step": 87725
    },
    {
      "epoch": 1.292322646205505,
      "grad_norm": 0.8557813167572021,
      "learning_rate": 1.6481014812439046e-05,
      "loss": 0.0783,
      "step": 87750
    },
    {
      "epoch": 1.292690829295592,
      "grad_norm": 1.3396013975143433,
      "learning_rate": 1.6479378440472064e-05,
      "loss": 0.0793,
      "step": 87775
    },
    {
      "epoch": 1.293059012385679,
      "grad_norm": 1.4153445959091187,
      "learning_rate": 1.647774206850508e-05,
      "loss": 0.078,
      "step": 87800
    },
    {
      "epoch": 1.2934271954757661,
      "grad_norm": 1.2082234621047974,
      "learning_rate": 1.6476105696538093e-05,
      "loss": 0.0772,
      "step": 87825
    },
    {
      "epoch": 1.2937953785658531,
      "grad_norm": 1.2681392431259155,
      "learning_rate": 1.6474469324571107e-05,
      "loss": 0.0861,
      "step": 87850
    },
    {
      "epoch": 1.2941635616559402,
      "grad_norm": 1.1526434421539307,
      "learning_rate": 1.6472832952604122e-05,
      "loss": 0.0637,
      "step": 87875
    },
    {
      "epoch": 1.2945317447460272,
      "grad_norm": 1.3793410062789917,
      "learning_rate": 1.647119658063714e-05,
      "loss": 0.0804,
      "step": 87900
    },
    {
      "epoch": 1.2948999278361144,
      "grad_norm": 1.4132964611053467,
      "learning_rate": 1.6469560208670154e-05,
      "loss": 0.0758,
      "step": 87925
    },
    {
      "epoch": 1.2952681109262014,
      "grad_norm": 1.4679826498031616,
      "learning_rate": 1.646792383670317e-05,
      "loss": 0.0768,
      "step": 87950
    },
    {
      "epoch": 1.2956362940162884,
      "grad_norm": 1.176544189453125,
      "learning_rate": 1.6466287464736186e-05,
      "loss": 0.0775,
      "step": 87975
    },
    {
      "epoch": 1.2960044771063755,
      "grad_norm": 1.8251954317092896,
      "learning_rate": 1.64646510927692e-05,
      "loss": 0.069,
      "step": 88000
    },
    {
      "epoch": 1.2963726601964625,
      "grad_norm": 1.5406668186187744,
      "learning_rate": 1.646301472080222e-05,
      "loss": 0.0695,
      "step": 88025
    },
    {
      "epoch": 1.2967408432865495,
      "grad_norm": 1.1207826137542725,
      "learning_rate": 1.646137834883523e-05,
      "loss": 0.0774,
      "step": 88050
    },
    {
      "epoch": 1.2971090263766365,
      "grad_norm": 1.6414769887924194,
      "learning_rate": 1.6459741976868248e-05,
      "loss": 0.0873,
      "step": 88075
    },
    {
      "epoch": 1.2974772094667237,
      "grad_norm": 1.776997685432434,
      "learning_rate": 1.6458105604901262e-05,
      "loss": 0.0722,
      "step": 88100
    },
    {
      "epoch": 1.2978453925568108,
      "grad_norm": 1.0320223569869995,
      "learning_rate": 1.6456469232934277e-05,
      "loss": 0.0775,
      "step": 88125
    },
    {
      "epoch": 1.2982135756468978,
      "grad_norm": 1.1254173517227173,
      "learning_rate": 1.6454832860967294e-05,
      "loss": 0.0783,
      "step": 88150
    },
    {
      "epoch": 1.2985817587369848,
      "grad_norm": 1.4797089099884033,
      "learning_rate": 1.645319648900031e-05,
      "loss": 0.074,
      "step": 88175
    },
    {
      "epoch": 1.2989499418270718,
      "grad_norm": 1.3233765363693237,
      "learning_rate": 1.6451560117033323e-05,
      "loss": 0.079,
      "step": 88200
    },
    {
      "epoch": 1.2993181249171588,
      "grad_norm": 1.6980239152908325,
      "learning_rate": 1.644992374506634e-05,
      "loss": 0.0821,
      "step": 88225
    },
    {
      "epoch": 1.2996863080072458,
      "grad_norm": 1.635428547859192,
      "learning_rate": 1.6448287373099356e-05,
      "loss": 0.0716,
      "step": 88250
    },
    {
      "epoch": 1.3000544910973328,
      "grad_norm": 0.9146542549133301,
      "learning_rate": 1.644665100113237e-05,
      "loss": 0.0787,
      "step": 88275
    },
    {
      "epoch": 1.3004226741874199,
      "grad_norm": 1.1425437927246094,
      "learning_rate": 1.6445014629165385e-05,
      "loss": 0.0803,
      "step": 88300
    },
    {
      "epoch": 1.3007908572775069,
      "grad_norm": 1.1494662761688232,
      "learning_rate": 1.6443378257198402e-05,
      "loss": 0.0714,
      "step": 88325
    },
    {
      "epoch": 1.3011590403675939,
      "grad_norm": 1.2007160186767578,
      "learning_rate": 1.6441741885231417e-05,
      "loss": 0.072,
      "step": 88350
    },
    {
      "epoch": 1.301527223457681,
      "grad_norm": 1.711274266242981,
      "learning_rate": 1.644010551326443e-05,
      "loss": 0.083,
      "step": 88375
    },
    {
      "epoch": 1.3018954065477681,
      "grad_norm": 1.4959429502487183,
      "learning_rate": 1.643846914129745e-05,
      "loss": 0.0794,
      "step": 88400
    },
    {
      "epoch": 1.3022635896378552,
      "grad_norm": 1.783243179321289,
      "learning_rate": 1.6436832769330464e-05,
      "loss": 0.0719,
      "step": 88425
    },
    {
      "epoch": 1.3026317727279422,
      "grad_norm": 1.0835564136505127,
      "learning_rate": 1.643519639736348e-05,
      "loss": 0.076,
      "step": 88450
    },
    {
      "epoch": 1.3029999558180292,
      "grad_norm": 1.370273470878601,
      "learning_rate": 1.6433560025396493e-05,
      "loss": 0.0679,
      "step": 88475
    },
    {
      "epoch": 1.3033681389081162,
      "grad_norm": 1.517225980758667,
      "learning_rate": 1.643192365342951e-05,
      "loss": 0.0791,
      "step": 88500
    },
    {
      "epoch": 1.3037363219982032,
      "grad_norm": 1.0269920825958252,
      "learning_rate": 1.6430287281462525e-05,
      "loss": 0.0733,
      "step": 88525
    },
    {
      "epoch": 1.3041045050882902,
      "grad_norm": 1.5639657974243164,
      "learning_rate": 1.642865090949554e-05,
      "loss": 0.079,
      "step": 88550
    },
    {
      "epoch": 1.3044726881783775,
      "grad_norm": 1.4998055696487427,
      "learning_rate": 1.6427014537528557e-05,
      "loss": 0.0742,
      "step": 88575
    },
    {
      "epoch": 1.3048408712684645,
      "grad_norm": 1.509335994720459,
      "learning_rate": 1.642537816556157e-05,
      "loss": 0.0777,
      "step": 88600
    },
    {
      "epoch": 1.3052090543585515,
      "grad_norm": 1.7063488960266113,
      "learning_rate": 1.6423741793594586e-05,
      "loss": 0.0712,
      "step": 88625
    },
    {
      "epoch": 1.3055772374486385,
      "grad_norm": 1.6909536123275757,
      "learning_rate": 1.6422105421627604e-05,
      "loss": 0.0822,
      "step": 88650
    },
    {
      "epoch": 1.3059454205387255,
      "grad_norm": 1.8001981973648071,
      "learning_rate": 1.642046904966062e-05,
      "loss": 0.0766,
      "step": 88675
    },
    {
      "epoch": 1.3063136036288125,
      "grad_norm": 1.5624608993530273,
      "learning_rate": 1.6418832677693633e-05,
      "loss": 0.0719,
      "step": 88700
    },
    {
      "epoch": 1.3066817867188996,
      "grad_norm": 1.2220088243484497,
      "learning_rate": 1.6417196305726647e-05,
      "loss": 0.0859,
      "step": 88725
    },
    {
      "epoch": 1.3070499698089866,
      "grad_norm": 1.4403575658798218,
      "learning_rate": 1.6415559933759665e-05,
      "loss": 0.083,
      "step": 88750
    },
    {
      "epoch": 1.3074181528990736,
      "grad_norm": 1.5407556295394897,
      "learning_rate": 1.641392356179268e-05,
      "loss": 0.0781,
      "step": 88775
    },
    {
      "epoch": 1.3077863359891606,
      "grad_norm": 1.4624788761138916,
      "learning_rate": 1.6412287189825694e-05,
      "loss": 0.0682,
      "step": 88800
    },
    {
      "epoch": 1.3081545190792476,
      "grad_norm": 1.3351502418518066,
      "learning_rate": 1.6410650817858712e-05,
      "loss": 0.0803,
      "step": 88825
    },
    {
      "epoch": 1.3085227021693346,
      "grad_norm": 1.2985202074050903,
      "learning_rate": 1.6409014445891726e-05,
      "loss": 0.0748,
      "step": 88850
    },
    {
      "epoch": 1.3088908852594219,
      "grad_norm": 1.2473987340927124,
      "learning_rate": 1.640737807392474e-05,
      "loss": 0.0784,
      "step": 88875
    },
    {
      "epoch": 1.309259068349509,
      "grad_norm": 1.087683916091919,
      "learning_rate": 1.6405741701957755e-05,
      "loss": 0.0746,
      "step": 88900
    },
    {
      "epoch": 1.309627251439596,
      "grad_norm": 1.5098973512649536,
      "learning_rate": 1.6404105329990773e-05,
      "loss": 0.0749,
      "step": 88925
    },
    {
      "epoch": 1.309995434529683,
      "grad_norm": 1.2901890277862549,
      "learning_rate": 1.6402534412902467e-05,
      "loss": 0.0818,
      "step": 88950
    },
    {
      "epoch": 1.31036361761977,
      "grad_norm": 1.1922364234924316,
      "learning_rate": 1.640089804093548e-05,
      "loss": 0.0744,
      "step": 88975
    },
    {
      "epoch": 1.310731800709857,
      "grad_norm": 1.3412866592407227,
      "learning_rate": 1.63992616689685e-05,
      "loss": 0.0691,
      "step": 89000
    },
    {
      "epoch": 1.311099983799944,
      "grad_norm": 1.465410828590393,
      "learning_rate": 1.6397625297001514e-05,
      "loss": 0.0823,
      "step": 89025
    },
    {
      "epoch": 1.3114681668900312,
      "grad_norm": 1.3778318166732788,
      "learning_rate": 1.6395988925034528e-05,
      "loss": 0.0771,
      "step": 89050
    },
    {
      "epoch": 1.3118363499801182,
      "grad_norm": 1.2208822965621948,
      "learning_rate": 1.6394352553067546e-05,
      "loss": 0.0789,
      "step": 89075
    },
    {
      "epoch": 1.3122045330702052,
      "grad_norm": 1.6019970178604126,
      "learning_rate": 1.6392716181100557e-05,
      "loss": 0.0782,
      "step": 89100
    },
    {
      "epoch": 1.3125727161602923,
      "grad_norm": 1.160643458366394,
      "learning_rate": 1.6391079809133575e-05,
      "loss": 0.0796,
      "step": 89125
    },
    {
      "epoch": 1.3129408992503793,
      "grad_norm": 1.4744532108306885,
      "learning_rate": 1.638944343716659e-05,
      "loss": 0.0728,
      "step": 89150
    },
    {
      "epoch": 1.3133090823404663,
      "grad_norm": 1.6466758251190186,
      "learning_rate": 1.6387807065199607e-05,
      "loss": 0.0705,
      "step": 89175
    },
    {
      "epoch": 1.3136772654305533,
      "grad_norm": 1.6497390270233154,
      "learning_rate": 1.6386170693232622e-05,
      "loss": 0.0828,
      "step": 89200
    },
    {
      "epoch": 1.3140454485206403,
      "grad_norm": 1.3599823713302612,
      "learning_rate": 1.6384534321265636e-05,
      "loss": 0.0781,
      "step": 89225
    },
    {
      "epoch": 1.3144136316107273,
      "grad_norm": 1.6341114044189453,
      "learning_rate": 1.6382897949298654e-05,
      "loss": 0.0708,
      "step": 89250
    },
    {
      "epoch": 1.3147818147008143,
      "grad_norm": 1.1234710216522217,
      "learning_rate": 1.638126157733167e-05,
      "loss": 0.0679,
      "step": 89275
    },
    {
      "epoch": 1.3151499977909014,
      "grad_norm": 1.6428710222244263,
      "learning_rate": 1.6379625205364683e-05,
      "loss": 0.0829,
      "step": 89300
    },
    {
      "epoch": 1.3155181808809884,
      "grad_norm": 0.9147220849990845,
      "learning_rate": 1.6377988833397697e-05,
      "loss": 0.0782,
      "step": 89325
    },
    {
      "epoch": 1.3158863639710756,
      "grad_norm": 1.658098578453064,
      "learning_rate": 1.6376352461430712e-05,
      "loss": 0.0709,
      "step": 89350
    },
    {
      "epoch": 1.3162545470611626,
      "grad_norm": 1.2410835027694702,
      "learning_rate": 1.637471608946373e-05,
      "loss": 0.0762,
      "step": 89375
    },
    {
      "epoch": 1.3166227301512496,
      "grad_norm": 1.0997791290283203,
      "learning_rate": 1.6373079717496744e-05,
      "loss": 0.0737,
      "step": 89400
    },
    {
      "epoch": 1.3169909132413367,
      "grad_norm": 1.3778128623962402,
      "learning_rate": 1.6371443345529762e-05,
      "loss": 0.0805,
      "step": 89425
    },
    {
      "epoch": 1.3173590963314237,
      "grad_norm": 1.4913908243179321,
      "learning_rate": 1.6369806973562776e-05,
      "loss": 0.0823,
      "step": 89450
    },
    {
      "epoch": 1.3177272794215107,
      "grad_norm": 1.7200385332107544,
      "learning_rate": 1.636817060159579e-05,
      "loss": 0.0852,
      "step": 89475
    },
    {
      "epoch": 1.3180954625115977,
      "grad_norm": 1.5541396141052246,
      "learning_rate": 1.636653422962881e-05,
      "loss": 0.0824,
      "step": 89500
    },
    {
      "epoch": 1.318463645601685,
      "grad_norm": 1.4841601848602295,
      "learning_rate": 1.636489785766182e-05,
      "loss": 0.0769,
      "step": 89525
    },
    {
      "epoch": 1.318831828691772,
      "grad_norm": 1.5176092386245728,
      "learning_rate": 1.6363261485694838e-05,
      "loss": 0.0707,
      "step": 89550
    },
    {
      "epoch": 1.319200011781859,
      "grad_norm": 1.509604573249817,
      "learning_rate": 1.6361625113727852e-05,
      "loss": 0.0783,
      "step": 89575
    },
    {
      "epoch": 1.319568194871946,
      "grad_norm": 1.4398568868637085,
      "learning_rate": 1.6359988741760867e-05,
      "loss": 0.0708,
      "step": 89600
    },
    {
      "epoch": 1.319936377962033,
      "grad_norm": 1.6234606504440308,
      "learning_rate": 1.6358352369793884e-05,
      "loss": 0.0736,
      "step": 89625
    },
    {
      "epoch": 1.32030456105212,
      "grad_norm": 1.5677322149276733,
      "learning_rate": 1.63567159978269e-05,
      "loss": 0.0757,
      "step": 89650
    },
    {
      "epoch": 1.320672744142207,
      "grad_norm": 1.727790117263794,
      "learning_rate": 1.6355079625859917e-05,
      "loss": 0.0689,
      "step": 89675
    },
    {
      "epoch": 1.321040927232294,
      "grad_norm": 1.2676259279251099,
      "learning_rate": 1.635344325389293e-05,
      "loss": 0.0735,
      "step": 89700
    },
    {
      "epoch": 1.321409110322381,
      "grad_norm": 1.7587922811508179,
      "learning_rate": 1.6351806881925946e-05,
      "loss": 0.0738,
      "step": 89725
    },
    {
      "epoch": 1.321777293412468,
      "grad_norm": 2.1331818103790283,
      "learning_rate": 1.635017050995896e-05,
      "loss": 0.0736,
      "step": 89750
    },
    {
      "epoch": 1.322145476502555,
      "grad_norm": 1.1653778553009033,
      "learning_rate": 1.6348534137991975e-05,
      "loss": 0.0753,
      "step": 89775
    },
    {
      "epoch": 1.322513659592642,
      "grad_norm": 1.4680769443511963,
      "learning_rate": 1.6346897766024992e-05,
      "loss": 0.0764,
      "step": 89800
    },
    {
      "epoch": 1.3228818426827293,
      "grad_norm": 1.3933496475219727,
      "learning_rate": 1.6345261394058007e-05,
      "loss": 0.0657,
      "step": 89825
    },
    {
      "epoch": 1.3232500257728164,
      "grad_norm": 1.3332090377807617,
      "learning_rate": 1.6343625022091025e-05,
      "loss": 0.0761,
      "step": 89850
    },
    {
      "epoch": 1.3236182088629034,
      "grad_norm": 0.8797059059143066,
      "learning_rate": 1.634198865012404e-05,
      "loss": 0.0643,
      "step": 89875
    },
    {
      "epoch": 1.3239863919529904,
      "grad_norm": 1.4298036098480225,
      "learning_rate": 1.6340352278157054e-05,
      "loss": 0.072,
      "step": 89900
    },
    {
      "epoch": 1.3243545750430774,
      "grad_norm": 1.0826364755630493,
      "learning_rate": 1.633871590619007e-05,
      "loss": 0.077,
      "step": 89925
    },
    {
      "epoch": 1.3247227581331644,
      "grad_norm": 1.3877052068710327,
      "learning_rate": 1.6337079534223083e-05,
      "loss": 0.074,
      "step": 89950
    },
    {
      "epoch": 1.3250909412232514,
      "grad_norm": 1.8560642004013062,
      "learning_rate": 1.63354431622561e-05,
      "loss": 0.0769,
      "step": 89975
    },
    {
      "epoch": 1.3254591243133385,
      "grad_norm": 1.0066782236099243,
      "learning_rate": 1.6333806790289115e-05,
      "loss": 0.0772,
      "step": 90000
    },
    {
      "epoch": 1.3258273074034257,
      "grad_norm": 1.5684863328933716,
      "learning_rate": 1.633217041832213e-05,
      "loss": 0.0712,
      "step": 90025
    },
    {
      "epoch": 1.3261954904935127,
      "grad_norm": 1.609997272491455,
      "learning_rate": 1.6330534046355147e-05,
      "loss": 0.0695,
      "step": 90050
    },
    {
      "epoch": 1.3265636735835997,
      "grad_norm": 1.3524547815322876,
      "learning_rate": 1.632889767438816e-05,
      "loss": 0.0913,
      "step": 90075
    },
    {
      "epoch": 1.3269318566736867,
      "grad_norm": 1.690234661102295,
      "learning_rate": 1.632726130242118e-05,
      "loss": 0.0831,
      "step": 90100
    },
    {
      "epoch": 1.3273000397637738,
      "grad_norm": 1.0316253900527954,
      "learning_rate": 1.6325624930454194e-05,
      "loss": 0.0693,
      "step": 90125
    },
    {
      "epoch": 1.3276682228538608,
      "grad_norm": 1.8638652563095093,
      "learning_rate": 1.632398855848721e-05,
      "loss": 0.0868,
      "step": 90150
    },
    {
      "epoch": 1.3280364059439478,
      "grad_norm": 1.6013383865356445,
      "learning_rate": 1.6322352186520223e-05,
      "loss": 0.081,
      "step": 90175
    },
    {
      "epoch": 1.3284045890340348,
      "grad_norm": 1.2709286212921143,
      "learning_rate": 1.6320715814553237e-05,
      "loss": 0.0777,
      "step": 90200
    },
    {
      "epoch": 1.3287727721241218,
      "grad_norm": 1.9263944625854492,
      "learning_rate": 1.6319079442586255e-05,
      "loss": 0.0773,
      "step": 90225
    },
    {
      "epoch": 1.3291409552142088,
      "grad_norm": 1.0371692180633545,
      "learning_rate": 1.631744307061927e-05,
      "loss": 0.077,
      "step": 90250
    },
    {
      "epoch": 1.3295091383042958,
      "grad_norm": 1.599334478378296,
      "learning_rate": 1.6315806698652284e-05,
      "loss": 0.0781,
      "step": 90275
    },
    {
      "epoch": 1.3298773213943829,
      "grad_norm": 1.292439579963684,
      "learning_rate": 1.6314170326685302e-05,
      "loss": 0.0724,
      "step": 90300
    },
    {
      "epoch": 1.33024550448447,
      "grad_norm": 1.5270636081695557,
      "learning_rate": 1.6312533954718316e-05,
      "loss": 0.0755,
      "step": 90325
    },
    {
      "epoch": 1.3306136875745571,
      "grad_norm": 2.00880765914917,
      "learning_rate": 1.6310897582751334e-05,
      "loss": 0.0775,
      "step": 90350
    },
    {
      "epoch": 1.3309818706646441,
      "grad_norm": 1.344334602355957,
      "learning_rate": 1.6309261210784345e-05,
      "loss": 0.0749,
      "step": 90375
    },
    {
      "epoch": 1.3313500537547311,
      "grad_norm": 1.0912142992019653,
      "learning_rate": 1.6307624838817363e-05,
      "loss": 0.0706,
      "step": 90400
    },
    {
      "epoch": 1.3317182368448182,
      "grad_norm": 1.29023277759552,
      "learning_rate": 1.6305988466850378e-05,
      "loss": 0.0729,
      "step": 90425
    },
    {
      "epoch": 1.3320864199349052,
      "grad_norm": 2.004547595977783,
      "learning_rate": 1.6304352094883392e-05,
      "loss": 0.0809,
      "step": 90450
    },
    {
      "epoch": 1.3324546030249922,
      "grad_norm": 1.2141296863555908,
      "learning_rate": 1.630271572291641e-05,
      "loss": 0.0754,
      "step": 90475
    },
    {
      "epoch": 1.3328227861150794,
      "grad_norm": 1.1595202684402466,
      "learning_rate": 1.6301079350949424e-05,
      "loss": 0.0762,
      "step": 90500
    },
    {
      "epoch": 1.3331909692051664,
      "grad_norm": 1.522676944732666,
      "learning_rate": 1.629944297898244e-05,
      "loss": 0.0819,
      "step": 90525
    },
    {
      "epoch": 1.3335591522952535,
      "grad_norm": 1.645128607749939,
      "learning_rate": 1.6297806607015457e-05,
      "loss": 0.0789,
      "step": 90550
    },
    {
      "epoch": 1.3339273353853405,
      "grad_norm": 1.6275830268859863,
      "learning_rate": 1.629617023504847e-05,
      "loss": 0.0777,
      "step": 90575
    },
    {
      "epoch": 1.3342955184754275,
      "grad_norm": 1.7780475616455078,
      "learning_rate": 1.6294533863081486e-05,
      "loss": 0.0751,
      "step": 90600
    },
    {
      "epoch": 1.3346637015655145,
      "grad_norm": 0.8597611784934998,
      "learning_rate": 1.62928974911145e-05,
      "loss": 0.0712,
      "step": 90625
    },
    {
      "epoch": 1.3350318846556015,
      "grad_norm": 1.6526169776916504,
      "learning_rate": 1.6291261119147518e-05,
      "loss": 0.0727,
      "step": 90650
    },
    {
      "epoch": 1.3354000677456885,
      "grad_norm": 1.4190505743026733,
      "learning_rate": 1.6289624747180532e-05,
      "loss": 0.0819,
      "step": 90675
    },
    {
      "epoch": 1.3357682508357756,
      "grad_norm": 1.3069881200790405,
      "learning_rate": 1.6287988375213547e-05,
      "loss": 0.0752,
      "step": 90700
    },
    {
      "epoch": 1.3361364339258626,
      "grad_norm": 1.3888797760009766,
      "learning_rate": 1.6286352003246565e-05,
      "loss": 0.0782,
      "step": 90725
    },
    {
      "epoch": 1.3365046170159496,
      "grad_norm": 1.2229448556900024,
      "learning_rate": 1.628471563127958e-05,
      "loss": 0.0749,
      "step": 90750
    },
    {
      "epoch": 1.3368728001060366,
      "grad_norm": 1.2446955442428589,
      "learning_rate": 1.6283079259312597e-05,
      "loss": 0.072,
      "step": 90775
    },
    {
      "epoch": 1.3372409831961238,
      "grad_norm": 1.2332305908203125,
      "learning_rate": 1.6281442887345608e-05,
      "loss": 0.069,
      "step": 90800
    },
    {
      "epoch": 1.3376091662862108,
      "grad_norm": 1.3013314008712769,
      "learning_rate": 1.6279806515378626e-05,
      "loss": 0.0788,
      "step": 90825
    },
    {
      "epoch": 1.3379773493762979,
      "grad_norm": 1.5836565494537354,
      "learning_rate": 1.627817014341164e-05,
      "loss": 0.0801,
      "step": 90850
    },
    {
      "epoch": 1.3383455324663849,
      "grad_norm": 1.5280481576919556,
      "learning_rate": 1.6276533771444655e-05,
      "loss": 0.0801,
      "step": 90875
    },
    {
      "epoch": 1.338713715556472,
      "grad_norm": 1.3406683206558228,
      "learning_rate": 1.6274897399477673e-05,
      "loss": 0.0767,
      "step": 90900
    },
    {
      "epoch": 1.339081898646559,
      "grad_norm": 0.9091733694076538,
      "learning_rate": 1.6273261027510687e-05,
      "loss": 0.0775,
      "step": 90925
    },
    {
      "epoch": 1.339450081736646,
      "grad_norm": 1.2953158617019653,
      "learning_rate": 1.62716246555437e-05,
      "loss": 0.0733,
      "step": 90950
    },
    {
      "epoch": 1.3398182648267332,
      "grad_norm": 1.3101142644882202,
      "learning_rate": 1.626998828357672e-05,
      "loss": 0.0724,
      "step": 90975
    },
    {
      "epoch": 1.3401864479168202,
      "grad_norm": 1.5352545976638794,
      "learning_rate": 1.626841736648841e-05,
      "loss": 0.0706,
      "step": 91000
    },
    {
      "epoch": 1.3405546310069072,
      "grad_norm": 1.4464690685272217,
      "learning_rate": 1.6266780994521428e-05,
      "loss": 0.0747,
      "step": 91025
    },
    {
      "epoch": 1.3409228140969942,
      "grad_norm": 1.049544334411621,
      "learning_rate": 1.6265144622554442e-05,
      "loss": 0.0731,
      "step": 91050
    },
    {
      "epoch": 1.3412909971870812,
      "grad_norm": 1.006972312927246,
      "learning_rate": 1.626350825058746e-05,
      "loss": 0.0717,
      "step": 91075
    },
    {
      "epoch": 1.3416591802771682,
      "grad_norm": 1.0202698707580566,
      "learning_rate": 1.6261871878620474e-05,
      "loss": 0.0759,
      "step": 91100
    },
    {
      "epoch": 1.3420273633672553,
      "grad_norm": 2.213733434677124,
      "learning_rate": 1.626023550665349e-05,
      "loss": 0.0839,
      "step": 91125
    },
    {
      "epoch": 1.3423955464573423,
      "grad_norm": 1.620505690574646,
      "learning_rate": 1.6258599134686507e-05,
      "loss": 0.0717,
      "step": 91150
    },
    {
      "epoch": 1.3427637295474293,
      "grad_norm": 1.2301806211471558,
      "learning_rate": 1.625696276271952e-05,
      "loss": 0.0726,
      "step": 91175
    },
    {
      "epoch": 1.3431319126375163,
      "grad_norm": 1.664406657218933,
      "learning_rate": 1.6255326390752536e-05,
      "loss": 0.076,
      "step": 91200
    },
    {
      "epoch": 1.3435000957276033,
      "grad_norm": 1.6831588745117188,
      "learning_rate": 1.625369001878555e-05,
      "loss": 0.0744,
      "step": 91225
    },
    {
      "epoch": 1.3438682788176903,
      "grad_norm": 1.3554997444152832,
      "learning_rate": 1.6252053646818568e-05,
      "loss": 0.0735,
      "step": 91250
    },
    {
      "epoch": 1.3442364619077776,
      "grad_norm": 1.3595879077911377,
      "learning_rate": 1.6250417274851582e-05,
      "loss": 0.0796,
      "step": 91275
    },
    {
      "epoch": 1.3446046449978646,
      "grad_norm": 2.183323383331299,
      "learning_rate": 1.6248780902884597e-05,
      "loss": 0.0683,
      "step": 91300
    },
    {
      "epoch": 1.3449728280879516,
      "grad_norm": 1.9501228332519531,
      "learning_rate": 1.6247144530917615e-05,
      "loss": 0.0856,
      "step": 91325
    },
    {
      "epoch": 1.3453410111780386,
      "grad_norm": 1.6886019706726074,
      "learning_rate": 1.624550815895063e-05,
      "loss": 0.0756,
      "step": 91350
    },
    {
      "epoch": 1.3457091942681256,
      "grad_norm": 1.3946233987808228,
      "learning_rate": 1.6243871786983644e-05,
      "loss": 0.0789,
      "step": 91375
    },
    {
      "epoch": 1.3460773773582126,
      "grad_norm": 1.2149133682250977,
      "learning_rate": 1.624223541501666e-05,
      "loss": 0.075,
      "step": 91400
    },
    {
      "epoch": 1.3464455604482997,
      "grad_norm": 1.3756582736968994,
      "learning_rate": 1.6240599043049673e-05,
      "loss": 0.0718,
      "step": 91425
    },
    {
      "epoch": 1.346813743538387,
      "grad_norm": 1.1662389039993286,
      "learning_rate": 1.623896267108269e-05,
      "loss": 0.0749,
      "step": 91450
    },
    {
      "epoch": 1.347181926628474,
      "grad_norm": 1.3211029767990112,
      "learning_rate": 1.6237326299115705e-05,
      "loss": 0.0788,
      "step": 91475
    },
    {
      "epoch": 1.347550109718561,
      "grad_norm": 1.4627496004104614,
      "learning_rate": 1.6235689927148723e-05,
      "loss": 0.0796,
      "step": 91500
    },
    {
      "epoch": 1.347918292808648,
      "grad_norm": 1.4493077993392944,
      "learning_rate": 1.6234053555181737e-05,
      "loss": 0.0742,
      "step": 91525
    },
    {
      "epoch": 1.348286475898735,
      "grad_norm": 1.291533350944519,
      "learning_rate": 1.623241718321475e-05,
      "loss": 0.0705,
      "step": 91550
    },
    {
      "epoch": 1.348654658988822,
      "grad_norm": 1.622793197631836,
      "learning_rate": 1.623078081124777e-05,
      "loss": 0.0708,
      "step": 91575
    },
    {
      "epoch": 1.349022842078909,
      "grad_norm": 1.5103604793548584,
      "learning_rate": 1.6229144439280784e-05,
      "loss": 0.0731,
      "step": 91600
    },
    {
      "epoch": 1.349391025168996,
      "grad_norm": 1.8477883338928223,
      "learning_rate": 1.62275080673138e-05,
      "loss": 0.0825,
      "step": 91625
    },
    {
      "epoch": 1.349759208259083,
      "grad_norm": 1.430349588394165,
      "learning_rate": 1.6225871695346813e-05,
      "loss": 0.0652,
      "step": 91650
    },
    {
      "epoch": 1.35012739134917,
      "grad_norm": 0.872810423374176,
      "learning_rate": 1.6224235323379827e-05,
      "loss": 0.0708,
      "step": 91675
    },
    {
      "epoch": 1.350495574439257,
      "grad_norm": 1.4663913249969482,
      "learning_rate": 1.6222598951412845e-05,
      "loss": 0.0798,
      "step": 91700
    },
    {
      "epoch": 1.350863757529344,
      "grad_norm": 1.6600351333618164,
      "learning_rate": 1.622096257944586e-05,
      "loss": 0.0739,
      "step": 91725
    },
    {
      "epoch": 1.3512319406194313,
      "grad_norm": 1.317004680633545,
      "learning_rate": 1.6219326207478877e-05,
      "loss": 0.0754,
      "step": 91750
    },
    {
      "epoch": 1.3516001237095183,
      "grad_norm": 1.6911221742630005,
      "learning_rate": 1.6217689835511892e-05,
      "loss": 0.0784,
      "step": 91775
    },
    {
      "epoch": 1.3519683067996053,
      "grad_norm": 1.226335048675537,
      "learning_rate": 1.6216053463544906e-05,
      "loss": 0.0776,
      "step": 91800
    },
    {
      "epoch": 1.3523364898896924,
      "grad_norm": 1.7652842998504639,
      "learning_rate": 1.6214417091577924e-05,
      "loss": 0.0767,
      "step": 91825
    },
    {
      "epoch": 1.3527046729797794,
      "grad_norm": 1.5158592462539673,
      "learning_rate": 1.6212780719610935e-05,
      "loss": 0.0793,
      "step": 91850
    },
    {
      "epoch": 1.3530728560698664,
      "grad_norm": 1.1897499561309814,
      "learning_rate": 1.6211144347643953e-05,
      "loss": 0.0741,
      "step": 91875
    },
    {
      "epoch": 1.3534410391599534,
      "grad_norm": 1.5610355138778687,
      "learning_rate": 1.6209507975676968e-05,
      "loss": 0.079,
      "step": 91900
    },
    {
      "epoch": 1.3538092222500406,
      "grad_norm": 1.5645991563796997,
      "learning_rate": 1.6207871603709982e-05,
      "loss": 0.0773,
      "step": 91925
    },
    {
      "epoch": 1.3541774053401276,
      "grad_norm": 1.8131879568099976,
      "learning_rate": 1.6206235231743e-05,
      "loss": 0.0769,
      "step": 91950
    },
    {
      "epoch": 1.3545455884302147,
      "grad_norm": 1.5232982635498047,
      "learning_rate": 1.6204598859776014e-05,
      "loss": 0.0716,
      "step": 91975
    },
    {
      "epoch": 1.3549137715203017,
      "grad_norm": 1.3999429941177368,
      "learning_rate": 1.6202962487809032e-05,
      "loss": 0.0752,
      "step": 92000
    },
    {
      "epoch": 1.3552819546103887,
      "grad_norm": 1.7154897451400757,
      "learning_rate": 1.6201326115842047e-05,
      "loss": 0.0711,
      "step": 92025
    },
    {
      "epoch": 1.3556501377004757,
      "grad_norm": 1.5103822946548462,
      "learning_rate": 1.619968974387506e-05,
      "loss": 0.0788,
      "step": 92050
    },
    {
      "epoch": 1.3560183207905627,
      "grad_norm": 1.2476588487625122,
      "learning_rate": 1.6198053371908076e-05,
      "loss": 0.0808,
      "step": 92075
    },
    {
      "epoch": 1.3563865038806497,
      "grad_norm": 1.5079420804977417,
      "learning_rate": 1.619641699994109e-05,
      "loss": 0.0815,
      "step": 92100
    },
    {
      "epoch": 1.3567546869707368,
      "grad_norm": 1.627220630645752,
      "learning_rate": 1.6194780627974108e-05,
      "loss": 0.0782,
      "step": 92125
    },
    {
      "epoch": 1.3571228700608238,
      "grad_norm": 1.4262968301773071,
      "learning_rate": 1.6193144256007122e-05,
      "loss": 0.0717,
      "step": 92150
    },
    {
      "epoch": 1.3574910531509108,
      "grad_norm": 1.3170403242111206,
      "learning_rate": 1.619150788404014e-05,
      "loss": 0.0777,
      "step": 92175
    },
    {
      "epoch": 1.3578592362409978,
      "grad_norm": 1.3268344402313232,
      "learning_rate": 1.6189871512073155e-05,
      "loss": 0.0709,
      "step": 92200
    },
    {
      "epoch": 1.358227419331085,
      "grad_norm": 1.7530207633972168,
      "learning_rate": 1.618823514010617e-05,
      "loss": 0.0733,
      "step": 92225
    },
    {
      "epoch": 1.358595602421172,
      "grad_norm": 1.6545002460479736,
      "learning_rate": 1.6186598768139187e-05,
      "loss": 0.0778,
      "step": 92250
    },
    {
      "epoch": 1.358963785511259,
      "grad_norm": 1.4598360061645508,
      "learning_rate": 1.6184962396172198e-05,
      "loss": 0.0754,
      "step": 92275
    },
    {
      "epoch": 1.359331968601346,
      "grad_norm": 1.8592778444290161,
      "learning_rate": 1.6183326024205216e-05,
      "loss": 0.0745,
      "step": 92300
    },
    {
      "epoch": 1.359700151691433,
      "grad_norm": 1.261029839515686,
      "learning_rate": 1.618168965223823e-05,
      "loss": 0.0841,
      "step": 92325
    },
    {
      "epoch": 1.3600683347815201,
      "grad_norm": 1.7479112148284912,
      "learning_rate": 1.6180053280271245e-05,
      "loss": 0.0735,
      "step": 92350
    },
    {
      "epoch": 1.3604365178716071,
      "grad_norm": 1.0892970561981201,
      "learning_rate": 1.6178416908304263e-05,
      "loss": 0.0666,
      "step": 92375
    },
    {
      "epoch": 1.3608047009616944,
      "grad_norm": 1.9156078100204468,
      "learning_rate": 1.6176780536337277e-05,
      "loss": 0.0744,
      "step": 92400
    },
    {
      "epoch": 1.3611728840517814,
      "grad_norm": 1.125425100326538,
      "learning_rate": 1.6175144164370295e-05,
      "loss": 0.0782,
      "step": 92425
    },
    {
      "epoch": 1.3615410671418684,
      "grad_norm": 1.3087259531021118,
      "learning_rate": 1.617350779240331e-05,
      "loss": 0.0716,
      "step": 92450
    },
    {
      "epoch": 1.3619092502319554,
      "grad_norm": 1.3223575353622437,
      "learning_rate": 1.6171871420436324e-05,
      "loss": 0.0773,
      "step": 92475
    },
    {
      "epoch": 1.3622774333220424,
      "grad_norm": 1.6559269428253174,
      "learning_rate": 1.6170235048469338e-05,
      "loss": 0.0713,
      "step": 92500
    },
    {
      "epoch": 1.3626456164121294,
      "grad_norm": 1.1485322713851929,
      "learning_rate": 1.6168598676502353e-05,
      "loss": 0.0742,
      "step": 92525
    },
    {
      "epoch": 1.3630137995022165,
      "grad_norm": 1.653432846069336,
      "learning_rate": 1.616696230453537e-05,
      "loss": 0.084,
      "step": 92550
    },
    {
      "epoch": 1.3633819825923035,
      "grad_norm": 1.4452425241470337,
      "learning_rate": 1.6165325932568385e-05,
      "loss": 0.0756,
      "step": 92575
    },
    {
      "epoch": 1.3637501656823905,
      "grad_norm": 1.4553121328353882,
      "learning_rate": 1.61636895606014e-05,
      "loss": 0.0805,
      "step": 92600
    },
    {
      "epoch": 1.3641183487724775,
      "grad_norm": 1.4380722045898438,
      "learning_rate": 1.6162053188634417e-05,
      "loss": 0.0734,
      "step": 92625
    },
    {
      "epoch": 1.3644865318625645,
      "grad_norm": 1.274214506149292,
      "learning_rate": 1.6160416816667432e-05,
      "loss": 0.0701,
      "step": 92650
    },
    {
      "epoch": 1.3648547149526515,
      "grad_norm": 1.813348412513733,
      "learning_rate": 1.615878044470045e-05,
      "loss": 0.0768,
      "step": 92675
    },
    {
      "epoch": 1.3652228980427388,
      "grad_norm": 1.2675639390945435,
      "learning_rate": 1.615714407273346e-05,
      "loss": 0.0788,
      "step": 92700
    },
    {
      "epoch": 1.3655910811328258,
      "grad_norm": 1.216004490852356,
      "learning_rate": 1.615550770076648e-05,
      "loss": 0.0688,
      "step": 92725
    },
    {
      "epoch": 1.3659592642229128,
      "grad_norm": 1.5318211317062378,
      "learning_rate": 1.6153871328799493e-05,
      "loss": 0.0754,
      "step": 92750
    },
    {
      "epoch": 1.3663274473129998,
      "grad_norm": 1.3124147653579712,
      "learning_rate": 1.6152234956832507e-05,
      "loss": 0.0728,
      "step": 92775
    },
    {
      "epoch": 1.3666956304030868,
      "grad_norm": 1.4888821840286255,
      "learning_rate": 1.6150598584865525e-05,
      "loss": 0.0708,
      "step": 92800
    },
    {
      "epoch": 1.3670638134931739,
      "grad_norm": 1.4113868474960327,
      "learning_rate": 1.614896221289854e-05,
      "loss": 0.0704,
      "step": 92825
    },
    {
      "epoch": 1.3674319965832609,
      "grad_norm": 1.3530505895614624,
      "learning_rate": 1.6147325840931554e-05,
      "loss": 0.0741,
      "step": 92850
    },
    {
      "epoch": 1.3678001796733479,
      "grad_norm": 1.3829823732376099,
      "learning_rate": 1.6145689468964572e-05,
      "loss": 0.0709,
      "step": 92875
    },
    {
      "epoch": 1.3681683627634351,
      "grad_norm": 1.209885835647583,
      "learning_rate": 1.6144053096997587e-05,
      "loss": 0.0686,
      "step": 92900
    },
    {
      "epoch": 1.3685365458535221,
      "grad_norm": 1.284749150276184,
      "learning_rate": 1.61424167250306e-05,
      "loss": 0.0749,
      "step": 92925
    },
    {
      "epoch": 1.3689047289436092,
      "grad_norm": 1.4421370029449463,
      "learning_rate": 1.6140780353063615e-05,
      "loss": 0.0768,
      "step": 92950
    },
    {
      "epoch": 1.3692729120336962,
      "grad_norm": 1.3181594610214233,
      "learning_rate": 1.6139143981096633e-05,
      "loss": 0.0777,
      "step": 92975
    },
    {
      "epoch": 1.3696410951237832,
      "grad_norm": 1.1795927286148071,
      "learning_rate": 1.6137507609129648e-05,
      "loss": 0.0684,
      "step": 93000
    },
    {
      "epoch": 1.3700092782138702,
      "grad_norm": 1.5650521516799927,
      "learning_rate": 1.6135871237162662e-05,
      "loss": 0.0792,
      "step": 93025
    },
    {
      "epoch": 1.3703774613039572,
      "grad_norm": 1.4631000757217407,
      "learning_rate": 1.613423486519568e-05,
      "loss": 0.0736,
      "step": 93050
    },
    {
      "epoch": 1.3707456443940442,
      "grad_norm": 1.321958303451538,
      "learning_rate": 1.6132598493228695e-05,
      "loss": 0.0761,
      "step": 93075
    },
    {
      "epoch": 1.3711138274841312,
      "grad_norm": 1.2320996522903442,
      "learning_rate": 1.613096212126171e-05,
      "loss": 0.0715,
      "step": 93100
    },
    {
      "epoch": 1.3714820105742183,
      "grad_norm": 1.4432626962661743,
      "learning_rate": 1.6129325749294723e-05,
      "loss": 0.0749,
      "step": 93125
    },
    {
      "epoch": 1.3718501936643053,
      "grad_norm": 1.2753055095672607,
      "learning_rate": 1.612768937732774e-05,
      "loss": 0.0685,
      "step": 93150
    },
    {
      "epoch": 1.3722183767543923,
      "grad_norm": 1.0910691022872925,
      "learning_rate": 1.6126053005360756e-05,
      "loss": 0.0725,
      "step": 93175
    },
    {
      "epoch": 1.3725865598444795,
      "grad_norm": 1.1770347356796265,
      "learning_rate": 1.612441663339377e-05,
      "loss": 0.0727,
      "step": 93200
    },
    {
      "epoch": 1.3729547429345665,
      "grad_norm": 1.4494625329971313,
      "learning_rate": 1.6122780261426788e-05,
      "loss": 0.0761,
      "step": 93225
    },
    {
      "epoch": 1.3733229260246536,
      "grad_norm": 1.445708990097046,
      "learning_rate": 1.6121143889459802e-05,
      "loss": 0.069,
      "step": 93250
    },
    {
      "epoch": 1.3736911091147406,
      "grad_norm": 1.6922872066497803,
      "learning_rate": 1.6119507517492817e-05,
      "loss": 0.0747,
      "step": 93275
    },
    {
      "epoch": 1.3740592922048276,
      "grad_norm": 1.3769599199295044,
      "learning_rate": 1.6117871145525835e-05,
      "loss": 0.075,
      "step": 93300
    },
    {
      "epoch": 1.3744274752949146,
      "grad_norm": 1.4104198217391968,
      "learning_rate": 1.611623477355885e-05,
      "loss": 0.073,
      "step": 93325
    },
    {
      "epoch": 1.3747956583850016,
      "grad_norm": 1.243233323097229,
      "learning_rate": 1.6114598401591864e-05,
      "loss": 0.0773,
      "step": 93350
    },
    {
      "epoch": 1.3751638414750889,
      "grad_norm": 1.573714017868042,
      "learning_rate": 1.6112962029624878e-05,
      "loss": 0.0723,
      "step": 93375
    },
    {
      "epoch": 1.3755320245651759,
      "grad_norm": 1.5424933433532715,
      "learning_rate": 1.6111325657657896e-05,
      "loss": 0.0785,
      "step": 93400
    },
    {
      "epoch": 1.3759002076552629,
      "grad_norm": 1.051605463027954,
      "learning_rate": 1.610968928569091e-05,
      "loss": 0.0681,
      "step": 93425
    },
    {
      "epoch": 1.37626839074535,
      "grad_norm": 1.186645746231079,
      "learning_rate": 1.6108052913723925e-05,
      "loss": 0.0715,
      "step": 93450
    },
    {
      "epoch": 1.376636573835437,
      "grad_norm": 1.5428658723831177,
      "learning_rate": 1.6106416541756943e-05,
      "loss": 0.0714,
      "step": 93475
    },
    {
      "epoch": 1.377004756925524,
      "grad_norm": 1.181187391281128,
      "learning_rate": 1.6104780169789957e-05,
      "loss": 0.0792,
      "step": 93500
    },
    {
      "epoch": 1.377372940015611,
      "grad_norm": 1.5966691970825195,
      "learning_rate": 1.610314379782297e-05,
      "loss": 0.0817,
      "step": 93525
    },
    {
      "epoch": 1.377741123105698,
      "grad_norm": 1.1199043989181519,
      "learning_rate": 1.6101507425855986e-05,
      "loss": 0.0747,
      "step": 93550
    },
    {
      "epoch": 1.378109306195785,
      "grad_norm": 1.2347215414047241,
      "learning_rate": 1.6099871053889004e-05,
      "loss": 0.0691,
      "step": 93575
    },
    {
      "epoch": 1.378477489285872,
      "grad_norm": 1.4695402383804321,
      "learning_rate": 1.609823468192202e-05,
      "loss": 0.0861,
      "step": 93600
    },
    {
      "epoch": 1.378845672375959,
      "grad_norm": 1.1878154277801514,
      "learning_rate": 1.6096598309955033e-05,
      "loss": 0.0681,
      "step": 93625
    },
    {
      "epoch": 1.379213855466046,
      "grad_norm": 1.849567174911499,
      "learning_rate": 1.609502739286673e-05,
      "loss": 0.0814,
      "step": 93650
    },
    {
      "epoch": 1.3795820385561333,
      "grad_norm": 1.2896862030029297,
      "learning_rate": 1.6093391020899745e-05,
      "loss": 0.0805,
      "step": 93675
    },
    {
      "epoch": 1.3799502216462203,
      "grad_norm": 1.5275253057479858,
      "learning_rate": 1.609175464893276e-05,
      "loss": 0.0767,
      "step": 93700
    },
    {
      "epoch": 1.3803184047363073,
      "grad_norm": 1.5272122621536255,
      "learning_rate": 1.6090118276965777e-05,
      "loss": 0.0747,
      "step": 93725
    },
    {
      "epoch": 1.3806865878263943,
      "grad_norm": 1.2588330507278442,
      "learning_rate": 1.6088481904998788e-05,
      "loss": 0.0708,
      "step": 93750
    },
    {
      "epoch": 1.3810547709164813,
      "grad_norm": 1.479750156402588,
      "learning_rate": 1.6086845533031806e-05,
      "loss": 0.0672,
      "step": 93775
    },
    {
      "epoch": 1.3814229540065683,
      "grad_norm": 1.2768731117248535,
      "learning_rate": 1.608520916106482e-05,
      "loss": 0.0725,
      "step": 93800
    },
    {
      "epoch": 1.3817911370966554,
      "grad_norm": 1.2872675657272339,
      "learning_rate": 1.6083572789097838e-05,
      "loss": 0.0685,
      "step": 93825
    },
    {
      "epoch": 1.3821593201867426,
      "grad_norm": 1.5474956035614014,
      "learning_rate": 1.6081936417130853e-05,
      "loss": 0.0742,
      "step": 93850
    },
    {
      "epoch": 1.3825275032768296,
      "grad_norm": 1.1625415086746216,
      "learning_rate": 1.6080300045163867e-05,
      "loss": 0.0769,
      "step": 93875
    },
    {
      "epoch": 1.3828956863669166,
      "grad_norm": 1.2839393615722656,
      "learning_rate": 1.6078663673196885e-05,
      "loss": 0.0778,
      "step": 93900
    },
    {
      "epoch": 1.3832638694570036,
      "grad_norm": 1.100573182106018,
      "learning_rate": 1.60770273012299e-05,
      "loss": 0.0758,
      "step": 93925
    },
    {
      "epoch": 1.3836320525470907,
      "grad_norm": 1.701767921447754,
      "learning_rate": 1.6075390929262914e-05,
      "loss": 0.0822,
      "step": 93950
    },
    {
      "epoch": 1.3840002356371777,
      "grad_norm": 1.4825780391693115,
      "learning_rate": 1.6073754557295928e-05,
      "loss": 0.0727,
      "step": 93975
    },
    {
      "epoch": 1.3843684187272647,
      "grad_norm": 1.6215708255767822,
      "learning_rate": 1.6072118185328943e-05,
      "loss": 0.0696,
      "step": 94000
    },
    {
      "epoch": 1.3847366018173517,
      "grad_norm": 1.6350444555282593,
      "learning_rate": 1.607048181336196e-05,
      "loss": 0.0714,
      "step": 94025
    },
    {
      "epoch": 1.3851047849074387,
      "grad_norm": 1.4409617185592651,
      "learning_rate": 1.6068845441394975e-05,
      "loss": 0.0735,
      "step": 94050
    },
    {
      "epoch": 1.3854729679975257,
      "grad_norm": 1.2251015901565552,
      "learning_rate": 1.6067209069427993e-05,
      "loss": 0.0704,
      "step": 94075
    },
    {
      "epoch": 1.3858411510876127,
      "grad_norm": 1.1130964756011963,
      "learning_rate": 1.6065572697461007e-05,
      "loss": 0.0837,
      "step": 94100
    },
    {
      "epoch": 1.3862093341776998,
      "grad_norm": 1.407037615776062,
      "learning_rate": 1.6063936325494022e-05,
      "loss": 0.0721,
      "step": 94125
    },
    {
      "epoch": 1.386577517267787,
      "grad_norm": 1.4179117679595947,
      "learning_rate": 1.606229995352704e-05,
      "loss": 0.0737,
      "step": 94150
    },
    {
      "epoch": 1.386945700357874,
      "grad_norm": 1.8100414276123047,
      "learning_rate": 1.606066358156005e-05,
      "loss": 0.0806,
      "step": 94175
    },
    {
      "epoch": 1.387313883447961,
      "grad_norm": 1.518386721611023,
      "learning_rate": 1.605902720959307e-05,
      "loss": 0.068,
      "step": 94200
    },
    {
      "epoch": 1.387682066538048,
      "grad_norm": 1.516484022140503,
      "learning_rate": 1.6057390837626083e-05,
      "loss": 0.07,
      "step": 94225
    },
    {
      "epoch": 1.388050249628135,
      "grad_norm": 1.5263150930404663,
      "learning_rate": 1.6055754465659097e-05,
      "loss": 0.0715,
      "step": 94250
    },
    {
      "epoch": 1.388418432718222,
      "grad_norm": 1.3971295356750488,
      "learning_rate": 1.6054118093692115e-05,
      "loss": 0.0763,
      "step": 94275
    },
    {
      "epoch": 1.388786615808309,
      "grad_norm": 1.7819820642471313,
      "learning_rate": 1.605248172172513e-05,
      "loss": 0.0817,
      "step": 94300
    },
    {
      "epoch": 1.3891547988983963,
      "grad_norm": 1.4265540838241577,
      "learning_rate": 1.6050845349758148e-05,
      "loss": 0.0728,
      "step": 94325
    },
    {
      "epoch": 1.3895229819884833,
      "grad_norm": 1.0876445770263672,
      "learning_rate": 1.6049208977791162e-05,
      "loss": 0.0773,
      "step": 94350
    },
    {
      "epoch": 1.3898911650785704,
      "grad_norm": 1.161629557609558,
      "learning_rate": 1.6047572605824177e-05,
      "loss": 0.0757,
      "step": 94375
    },
    {
      "epoch": 1.3902593481686574,
      "grad_norm": 1.584875226020813,
      "learning_rate": 1.604593623385719e-05,
      "loss": 0.0783,
      "step": 94400
    },
    {
      "epoch": 1.3906275312587444,
      "grad_norm": 0.9503710269927979,
      "learning_rate": 1.6044299861890205e-05,
      "loss": 0.073,
      "step": 94425
    },
    {
      "epoch": 1.3909957143488314,
      "grad_norm": 1.6783994436264038,
      "learning_rate": 1.6042663489923223e-05,
      "loss": 0.0716,
      "step": 94450
    },
    {
      "epoch": 1.3913638974389184,
      "grad_norm": 1.7241624593734741,
      "learning_rate": 1.6041027117956238e-05,
      "loss": 0.0767,
      "step": 94475
    },
    {
      "epoch": 1.3917320805290054,
      "grad_norm": 1.6753015518188477,
      "learning_rate": 1.6039390745989256e-05,
      "loss": 0.0767,
      "step": 94500
    },
    {
      "epoch": 1.3921002636190924,
      "grad_norm": 1.2731801271438599,
      "learning_rate": 1.603775437402227e-05,
      "loss": 0.0759,
      "step": 94525
    },
    {
      "epoch": 1.3924684467091795,
      "grad_norm": 1.4037301540374756,
      "learning_rate": 1.6036118002055285e-05,
      "loss": 0.0731,
      "step": 94550
    },
    {
      "epoch": 1.3928366297992665,
      "grad_norm": 1.5242455005645752,
      "learning_rate": 1.60344816300883e-05,
      "loss": 0.075,
      "step": 94575
    },
    {
      "epoch": 1.3932048128893535,
      "grad_norm": 1.401528000831604,
      "learning_rate": 1.6032845258121313e-05,
      "loss": 0.0748,
      "step": 94600
    },
    {
      "epoch": 1.3935729959794407,
      "grad_norm": 1.6088056564331055,
      "learning_rate": 1.603120888615433e-05,
      "loss": 0.0765,
      "step": 94625
    },
    {
      "epoch": 1.3939411790695277,
      "grad_norm": 1.5187320709228516,
      "learning_rate": 1.6029572514187346e-05,
      "loss": 0.0708,
      "step": 94650
    },
    {
      "epoch": 1.3943093621596148,
      "grad_norm": 1.079283356666565,
      "learning_rate": 1.602793614222036e-05,
      "loss": 0.0707,
      "step": 94675
    },
    {
      "epoch": 1.3946775452497018,
      "grad_norm": 1.1687904596328735,
      "learning_rate": 1.6026299770253378e-05,
      "loss": 0.0775,
      "step": 94700
    },
    {
      "epoch": 1.3950457283397888,
      "grad_norm": 1.5087347030639648,
      "learning_rate": 1.6024663398286392e-05,
      "loss": 0.0727,
      "step": 94725
    },
    {
      "epoch": 1.3954139114298758,
      "grad_norm": 1.6743303537368774,
      "learning_rate": 1.602302702631941e-05,
      "loss": 0.0813,
      "step": 94750
    },
    {
      "epoch": 1.3957820945199628,
      "grad_norm": 1.2112457752227783,
      "learning_rate": 1.6021390654352425e-05,
      "loss": 0.0747,
      "step": 94775
    },
    {
      "epoch": 1.39615027761005,
      "grad_norm": 1.454703450202942,
      "learning_rate": 1.601975428238544e-05,
      "loss": 0.0736,
      "step": 94800
    },
    {
      "epoch": 1.396518460700137,
      "grad_norm": 0.8092736005783081,
      "learning_rate": 1.6018117910418454e-05,
      "loss": 0.067,
      "step": 94825
    },
    {
      "epoch": 1.396886643790224,
      "grad_norm": 1.178327202796936,
      "learning_rate": 1.6016481538451468e-05,
      "loss": 0.0704,
      "step": 94850
    },
    {
      "epoch": 1.397254826880311,
      "grad_norm": 1.6238834857940674,
      "learning_rate": 1.6014845166484486e-05,
      "loss": 0.0777,
      "step": 94875
    },
    {
      "epoch": 1.3976230099703981,
      "grad_norm": 1.8076914548873901,
      "learning_rate": 1.60132087945175e-05,
      "loss": 0.0726,
      "step": 94900
    },
    {
      "epoch": 1.3979911930604851,
      "grad_norm": 1.448445200920105,
      "learning_rate": 1.6011572422550515e-05,
      "loss": 0.0774,
      "step": 94925
    },
    {
      "epoch": 1.3983593761505722,
      "grad_norm": 1.3596991300582886,
      "learning_rate": 1.6009936050583533e-05,
      "loss": 0.0755,
      "step": 94950
    },
    {
      "epoch": 1.3987275592406592,
      "grad_norm": 1.2962175607681274,
      "learning_rate": 1.6008299678616547e-05,
      "loss": 0.0713,
      "step": 94975
    },
    {
      "epoch": 1.3990957423307462,
      "grad_norm": 1.2302311658859253,
      "learning_rate": 1.6006663306649562e-05,
      "loss": 0.0713,
      "step": 95000
    },
    {
      "epoch": 1.3994639254208332,
      "grad_norm": 1.334878921508789,
      "learning_rate": 1.6005026934682576e-05,
      "loss": 0.076,
      "step": 95025
    },
    {
      "epoch": 1.3998321085109202,
      "grad_norm": 1.5879437923431396,
      "learning_rate": 1.6003390562715594e-05,
      "loss": 0.0703,
      "step": 95050
    },
    {
      "epoch": 1.4002002916010072,
      "grad_norm": 1.1994304656982422,
      "learning_rate": 1.600175419074861e-05,
      "loss": 0.0802,
      "step": 95075
    },
    {
      "epoch": 1.4005684746910945,
      "grad_norm": 1.548785924911499,
      "learning_rate": 1.6000117818781623e-05,
      "loss": 0.0779,
      "step": 95100
    },
    {
      "epoch": 1.4009366577811815,
      "grad_norm": 1.2609056234359741,
      "learning_rate": 1.599848144681464e-05,
      "loss": 0.0786,
      "step": 95125
    },
    {
      "epoch": 1.4013048408712685,
      "grad_norm": 1.119028925895691,
      "learning_rate": 1.5996845074847655e-05,
      "loss": 0.0681,
      "step": 95150
    },
    {
      "epoch": 1.4016730239613555,
      "grad_norm": 1.7501965761184692,
      "learning_rate": 1.599520870288067e-05,
      "loss": 0.0805,
      "step": 95175
    },
    {
      "epoch": 1.4020412070514425,
      "grad_norm": 1.458664894104004,
      "learning_rate": 1.5993572330913688e-05,
      "loss": 0.073,
      "step": 95200
    },
    {
      "epoch": 1.4024093901415295,
      "grad_norm": 1.7162827253341675,
      "learning_rate": 1.5991935958946702e-05,
      "loss": 0.0723,
      "step": 95225
    },
    {
      "epoch": 1.4027775732316166,
      "grad_norm": 1.8378931283950806,
      "learning_rate": 1.5990299586979716e-05,
      "loss": 0.0756,
      "step": 95250
    },
    {
      "epoch": 1.4031457563217036,
      "grad_norm": 1.6052461862564087,
      "learning_rate": 1.598866321501273e-05,
      "loss": 0.0681,
      "step": 95275
    },
    {
      "epoch": 1.4035139394117908,
      "grad_norm": 1.5615938901901245,
      "learning_rate": 1.598702684304575e-05,
      "loss": 0.0798,
      "step": 95300
    },
    {
      "epoch": 1.4038821225018778,
      "grad_norm": 1.5934979915618896,
      "learning_rate": 1.5985390471078763e-05,
      "loss": 0.0768,
      "step": 95325
    },
    {
      "epoch": 1.4042503055919648,
      "grad_norm": 1.564051628112793,
      "learning_rate": 1.5983754099111778e-05,
      "loss": 0.0668,
      "step": 95350
    },
    {
      "epoch": 1.4046184886820519,
      "grad_norm": 1.3912962675094604,
      "learning_rate": 1.5982117727144795e-05,
      "loss": 0.0692,
      "step": 95375
    },
    {
      "epoch": 1.4049866717721389,
      "grad_norm": 1.6267430782318115,
      "learning_rate": 1.598048135517781e-05,
      "loss": 0.0828,
      "step": 95400
    },
    {
      "epoch": 1.4053548548622259,
      "grad_norm": 1.5203158855438232,
      "learning_rate": 1.5978844983210824e-05,
      "loss": 0.0864,
      "step": 95425
    },
    {
      "epoch": 1.405723037952313,
      "grad_norm": 1.7149254083633423,
      "learning_rate": 1.597720861124384e-05,
      "loss": 0.0742,
      "step": 95450
    },
    {
      "epoch": 1.4060912210424,
      "grad_norm": 1.3976386785507202,
      "learning_rate": 1.5975572239276857e-05,
      "loss": 0.0671,
      "step": 95475
    },
    {
      "epoch": 1.406459404132487,
      "grad_norm": 1.5091254711151123,
      "learning_rate": 1.597393586730987e-05,
      "loss": 0.0752,
      "step": 95500
    },
    {
      "epoch": 1.406827587222574,
      "grad_norm": 1.513067364692688,
      "learning_rate": 1.5972299495342886e-05,
      "loss": 0.0703,
      "step": 95525
    },
    {
      "epoch": 1.407195770312661,
      "grad_norm": 1.1618019342422485,
      "learning_rate": 1.5970663123375903e-05,
      "loss": 0.0776,
      "step": 95550
    },
    {
      "epoch": 1.407563953402748,
      "grad_norm": 1.1448416709899902,
      "learning_rate": 1.5969026751408918e-05,
      "loss": 0.0786,
      "step": 95575
    },
    {
      "epoch": 1.4079321364928352,
      "grad_norm": 1.1268770694732666,
      "learning_rate": 1.5967390379441932e-05,
      "loss": 0.0666,
      "step": 95600
    },
    {
      "epoch": 1.4083003195829222,
      "grad_norm": 1.377536654472351,
      "learning_rate": 1.596575400747495e-05,
      "loss": 0.078,
      "step": 95625
    },
    {
      "epoch": 1.4086685026730092,
      "grad_norm": 1.4300528764724731,
      "learning_rate": 1.5964117635507965e-05,
      "loss": 0.0754,
      "step": 95650
    },
    {
      "epoch": 1.4090366857630963,
      "grad_norm": 1.8363966941833496,
      "learning_rate": 1.596248126354098e-05,
      "loss": 0.0715,
      "step": 95675
    },
    {
      "epoch": 1.4094048688531833,
      "grad_norm": 1.4692856073379517,
      "learning_rate": 1.5960844891573994e-05,
      "loss": 0.075,
      "step": 95700
    },
    {
      "epoch": 1.4097730519432703,
      "grad_norm": 1.1752272844314575,
      "learning_rate": 1.595927397448569e-05,
      "loss": 0.0714,
      "step": 95725
    },
    {
      "epoch": 1.4101412350333573,
      "grad_norm": 1.2121433019638062,
      "learning_rate": 1.5957637602518705e-05,
      "loss": 0.0797,
      "step": 95750
    },
    {
      "epoch": 1.4105094181234445,
      "grad_norm": 1.3295732736587524,
      "learning_rate": 1.595600123055172e-05,
      "loss": 0.0771,
      "step": 95775
    },
    {
      "epoch": 1.4108776012135316,
      "grad_norm": 1.4161239862442017,
      "learning_rate": 1.5954364858584738e-05,
      "loss": 0.0743,
      "step": 95800
    },
    {
      "epoch": 1.4112457843036186,
      "grad_norm": 1.4361987113952637,
      "learning_rate": 1.5952728486617752e-05,
      "loss": 0.0703,
      "step": 95825
    },
    {
      "epoch": 1.4116139673937056,
      "grad_norm": 1.1932249069213867,
      "learning_rate": 1.5951092114650767e-05,
      "loss": 0.0751,
      "step": 95850
    },
    {
      "epoch": 1.4119821504837926,
      "grad_norm": 1.57851243019104,
      "learning_rate": 1.594945574268378e-05,
      "loss": 0.0781,
      "step": 95875
    },
    {
      "epoch": 1.4123503335738796,
      "grad_norm": 1.181174635887146,
      "learning_rate": 1.59478193707168e-05,
      "loss": 0.0754,
      "step": 95900
    },
    {
      "epoch": 1.4127185166639666,
      "grad_norm": 1.4669899940490723,
      "learning_rate": 1.5946182998749813e-05,
      "loss": 0.0714,
      "step": 95925
    },
    {
      "epoch": 1.4130866997540537,
      "grad_norm": 1.3707056045532227,
      "learning_rate": 1.5944546626782828e-05,
      "loss": 0.0736,
      "step": 95950
    },
    {
      "epoch": 1.4134548828441407,
      "grad_norm": 1.4716074466705322,
      "learning_rate": 1.5942910254815846e-05,
      "loss": 0.076,
      "step": 95975
    },
    {
      "epoch": 1.4138230659342277,
      "grad_norm": 1.4471065998077393,
      "learning_rate": 1.594127388284886e-05,
      "loss": 0.0815,
      "step": 96000
    },
    {
      "epoch": 1.4141912490243147,
      "grad_norm": 1.6333789825439453,
      "learning_rate": 1.5939637510881875e-05,
      "loss": 0.0802,
      "step": 96025
    },
    {
      "epoch": 1.4145594321144017,
      "grad_norm": 1.924338459968567,
      "learning_rate": 1.5938001138914892e-05,
      "loss": 0.0705,
      "step": 96050
    },
    {
      "epoch": 1.414927615204489,
      "grad_norm": 1.0518661737442017,
      "learning_rate": 1.5936364766947903e-05,
      "loss": 0.0748,
      "step": 96075
    },
    {
      "epoch": 1.415295798294576,
      "grad_norm": 1.7915148735046387,
      "learning_rate": 1.593472839498092e-05,
      "loss": 0.0792,
      "step": 96100
    },
    {
      "epoch": 1.415663981384663,
      "grad_norm": 1.7382258176803589,
      "learning_rate": 1.5933092023013936e-05,
      "loss": 0.0756,
      "step": 96125
    },
    {
      "epoch": 1.41603216447475,
      "grad_norm": 1.5208255052566528,
      "learning_rate": 1.5931455651046954e-05,
      "loss": 0.0769,
      "step": 96150
    },
    {
      "epoch": 1.416400347564837,
      "grad_norm": 1.0508660078048706,
      "learning_rate": 1.5929819279079968e-05,
      "loss": 0.0625,
      "step": 96175
    },
    {
      "epoch": 1.416768530654924,
      "grad_norm": 1.800963044166565,
      "learning_rate": 1.5928182907112983e-05,
      "loss": 0.0672,
      "step": 96200
    },
    {
      "epoch": 1.417136713745011,
      "grad_norm": 1.3172813653945923,
      "learning_rate": 1.5926546535146e-05,
      "loss": 0.0733,
      "step": 96225
    },
    {
      "epoch": 1.4175048968350983,
      "grad_norm": 1.6335346698760986,
      "learning_rate": 1.5924910163179015e-05,
      "loss": 0.0747,
      "step": 96250
    },
    {
      "epoch": 1.4178730799251853,
      "grad_norm": 1.1219463348388672,
      "learning_rate": 1.592327379121203e-05,
      "loss": 0.0744,
      "step": 96275
    },
    {
      "epoch": 1.4182412630152723,
      "grad_norm": 1.4757314920425415,
      "learning_rate": 1.5921637419245044e-05,
      "loss": 0.0795,
      "step": 96300
    },
    {
      "epoch": 1.4186094461053593,
      "grad_norm": 1.4103058576583862,
      "learning_rate": 1.5920001047278058e-05,
      "loss": 0.0746,
      "step": 96325
    },
    {
      "epoch": 1.4189776291954463,
      "grad_norm": 1.389687180519104,
      "learning_rate": 1.5918364675311076e-05,
      "loss": 0.0709,
      "step": 96350
    },
    {
      "epoch": 1.4193458122855334,
      "grad_norm": 1.6277658939361572,
      "learning_rate": 1.591672830334409e-05,
      "loss": 0.0777,
      "step": 96375
    },
    {
      "epoch": 1.4197139953756204,
      "grad_norm": 1.535209059715271,
      "learning_rate": 1.591509193137711e-05,
      "loss": 0.0716,
      "step": 96400
    },
    {
      "epoch": 1.4200821784657074,
      "grad_norm": 1.2609444856643677,
      "learning_rate": 1.5913455559410123e-05,
      "loss": 0.0773,
      "step": 96425
    },
    {
      "epoch": 1.4204503615557944,
      "grad_norm": 1.3724538087844849,
      "learning_rate": 1.5911819187443137e-05,
      "loss": 0.0747,
      "step": 96450
    },
    {
      "epoch": 1.4208185446458814,
      "grad_norm": 1.3623946905136108,
      "learning_rate": 1.5910182815476152e-05,
      "loss": 0.0729,
      "step": 96475
    },
    {
      "epoch": 1.4211867277359684,
      "grad_norm": 1.5128545761108398,
      "learning_rate": 1.5908546443509166e-05,
      "loss": 0.0776,
      "step": 96500
    },
    {
      "epoch": 1.4215549108260555,
      "grad_norm": 1.3841090202331543,
      "learning_rate": 1.5906910071542184e-05,
      "loss": 0.0742,
      "step": 96525
    },
    {
      "epoch": 1.4219230939161427,
      "grad_norm": 1.6266148090362549,
      "learning_rate": 1.59052736995752e-05,
      "loss": 0.0719,
      "step": 96550
    },
    {
      "epoch": 1.4222912770062297,
      "grad_norm": 1.0477402210235596,
      "learning_rate": 1.5903637327608213e-05,
      "loss": 0.0693,
      "step": 96575
    },
    {
      "epoch": 1.4226594600963167,
      "grad_norm": 0.8229677081108093,
      "learning_rate": 1.590200095564123e-05,
      "loss": 0.0775,
      "step": 96600
    },
    {
      "epoch": 1.4230276431864037,
      "grad_norm": 1.2981600761413574,
      "learning_rate": 1.5900364583674245e-05,
      "loss": 0.0699,
      "step": 96625
    },
    {
      "epoch": 1.4233958262764907,
      "grad_norm": 1.284982681274414,
      "learning_rate": 1.5898728211707263e-05,
      "loss": 0.08,
      "step": 96650
    },
    {
      "epoch": 1.4237640093665778,
      "grad_norm": 1.4522628784179688,
      "learning_rate": 1.5897091839740278e-05,
      "loss": 0.0689,
      "step": 96675
    },
    {
      "epoch": 1.4241321924566648,
      "grad_norm": 1.519117832183838,
      "learning_rate": 1.5895455467773292e-05,
      "loss": 0.0774,
      "step": 96700
    },
    {
      "epoch": 1.424500375546752,
      "grad_norm": 1.2039984464645386,
      "learning_rate": 1.5893819095806306e-05,
      "loss": 0.0763,
      "step": 96725
    },
    {
      "epoch": 1.424868558636839,
      "grad_norm": 1.3003195524215698,
      "learning_rate": 1.589218272383932e-05,
      "loss": 0.0675,
      "step": 96750
    },
    {
      "epoch": 1.425236741726926,
      "grad_norm": 1.5068836212158203,
      "learning_rate": 1.589054635187234e-05,
      "loss": 0.0725,
      "step": 96775
    },
    {
      "epoch": 1.425604924817013,
      "grad_norm": 1.2453457117080688,
      "learning_rate": 1.5888909979905353e-05,
      "loss": 0.0769,
      "step": 96800
    },
    {
      "epoch": 1.4259731079071,
      "grad_norm": 1.1764041185379028,
      "learning_rate": 1.588727360793837e-05,
      "loss": 0.0784,
      "step": 96825
    },
    {
      "epoch": 1.426341290997187,
      "grad_norm": 1.4045989513397217,
      "learning_rate": 1.5885637235971386e-05,
      "loss": 0.0726,
      "step": 96850
    },
    {
      "epoch": 1.426709474087274,
      "grad_norm": 1.2621376514434814,
      "learning_rate": 1.58840008640044e-05,
      "loss": 0.0711,
      "step": 96875
    },
    {
      "epoch": 1.4270776571773611,
      "grad_norm": 1.2671632766723633,
      "learning_rate": 1.5882364492037414e-05,
      "loss": 0.0748,
      "step": 96900
    },
    {
      "epoch": 1.4274458402674481,
      "grad_norm": 1.3318380117416382,
      "learning_rate": 1.588072812007043e-05,
      "loss": 0.0753,
      "step": 96925
    },
    {
      "epoch": 1.4278140233575352,
      "grad_norm": 1.4506373405456543,
      "learning_rate": 1.5879091748103447e-05,
      "loss": 0.0705,
      "step": 96950
    },
    {
      "epoch": 1.4281822064476222,
      "grad_norm": 1.626297950744629,
      "learning_rate": 1.587745537613646e-05,
      "loss": 0.0747,
      "step": 96975
    },
    {
      "epoch": 1.4285503895377092,
      "grad_norm": 1.2483100891113281,
      "learning_rate": 1.5875819004169476e-05,
      "loss": 0.0774,
      "step": 97000
    },
    {
      "epoch": 1.4289185726277964,
      "grad_norm": 1.2446788549423218,
      "learning_rate": 1.5874182632202493e-05,
      "loss": 0.0716,
      "step": 97025
    },
    {
      "epoch": 1.4292867557178834,
      "grad_norm": 1.527416467666626,
      "learning_rate": 1.5872546260235508e-05,
      "loss": 0.0722,
      "step": 97050
    },
    {
      "epoch": 1.4296549388079705,
      "grad_norm": 1.4439234733581543,
      "learning_rate": 1.5870909888268526e-05,
      "loss": 0.0775,
      "step": 97075
    },
    {
      "epoch": 1.4300231218980575,
      "grad_norm": 1.5147870779037476,
      "learning_rate": 1.586927351630154e-05,
      "loss": 0.0714,
      "step": 97100
    },
    {
      "epoch": 1.4303913049881445,
      "grad_norm": 1.1291157007217407,
      "learning_rate": 1.5867637144334555e-05,
      "loss": 0.0753,
      "step": 97125
    },
    {
      "epoch": 1.4307594880782315,
      "grad_norm": 1.0651323795318604,
      "learning_rate": 1.586600077236757e-05,
      "loss": 0.075,
      "step": 97150
    },
    {
      "epoch": 1.4311276711683185,
      "grad_norm": 1.1143789291381836,
      "learning_rate": 1.5864364400400584e-05,
      "loss": 0.0735,
      "step": 97175
    },
    {
      "epoch": 1.4314958542584058,
      "grad_norm": 1.641231656074524,
      "learning_rate": 1.58627280284336e-05,
      "loss": 0.0772,
      "step": 97200
    },
    {
      "epoch": 1.4318640373484928,
      "grad_norm": 1.8680340051651,
      "learning_rate": 1.5861091656466616e-05,
      "loss": 0.0744,
      "step": 97225
    },
    {
      "epoch": 1.4322322204385798,
      "grad_norm": 1.45437490940094,
      "learning_rate": 1.585945528449963e-05,
      "loss": 0.0748,
      "step": 97250
    },
    {
      "epoch": 1.4326004035286668,
      "grad_norm": 1.2228707075119019,
      "learning_rate": 1.5857818912532648e-05,
      "loss": 0.0804,
      "step": 97275
    },
    {
      "epoch": 1.4329685866187538,
      "grad_norm": 1.3762407302856445,
      "learning_rate": 1.5856182540565663e-05,
      "loss": 0.0732,
      "step": 97300
    },
    {
      "epoch": 1.4333367697088408,
      "grad_norm": 1.4869719743728638,
      "learning_rate": 1.5854546168598677e-05,
      "loss": 0.0689,
      "step": 97325
    },
    {
      "epoch": 1.4337049527989278,
      "grad_norm": 0.9573516249656677,
      "learning_rate": 1.585290979663169e-05,
      "loss": 0.0685,
      "step": 97350
    },
    {
      "epoch": 1.4340731358890149,
      "grad_norm": 1.5761363506317139,
      "learning_rate": 1.585127342466471e-05,
      "loss": 0.0693,
      "step": 97375
    },
    {
      "epoch": 1.4344413189791019,
      "grad_norm": 1.2833172082901,
      "learning_rate": 1.5849637052697724e-05,
      "loss": 0.074,
      "step": 97400
    },
    {
      "epoch": 1.434809502069189,
      "grad_norm": 1.8005805015563965,
      "learning_rate": 1.584800068073074e-05,
      "loss": 0.0776,
      "step": 97425
    },
    {
      "epoch": 1.435177685159276,
      "grad_norm": 1.33426833152771,
      "learning_rate": 1.5846364308763756e-05,
      "loss": 0.0737,
      "step": 97450
    },
    {
      "epoch": 1.435545868249363,
      "grad_norm": 1.4123365879058838,
      "learning_rate": 1.584472793679677e-05,
      "loss": 0.0792,
      "step": 97475
    },
    {
      "epoch": 1.4359140513394502,
      "grad_norm": 1.5246069431304932,
      "learning_rate": 1.5843091564829785e-05,
      "loss": 0.0848,
      "step": 97500
    },
    {
      "epoch": 1.4362822344295372,
      "grad_norm": 1.2983479499816895,
      "learning_rate": 1.5841455192862803e-05,
      "loss": 0.0747,
      "step": 97525
    },
    {
      "epoch": 1.4366504175196242,
      "grad_norm": 0.9336724281311035,
      "learning_rate": 1.5839818820895817e-05,
      "loss": 0.0715,
      "step": 97550
    },
    {
      "epoch": 1.4370186006097112,
      "grad_norm": 1.1193267107009888,
      "learning_rate": 1.5838182448928832e-05,
      "loss": 0.0701,
      "step": 97575
    },
    {
      "epoch": 1.4373867836997982,
      "grad_norm": 1.6512045860290527,
      "learning_rate": 1.5836546076961846e-05,
      "loss": 0.0754,
      "step": 97600
    },
    {
      "epoch": 1.4377549667898852,
      "grad_norm": 1.604440689086914,
      "learning_rate": 1.5834909704994864e-05,
      "loss": 0.0748,
      "step": 97625
    },
    {
      "epoch": 1.4381231498799723,
      "grad_norm": 1.7810752391815186,
      "learning_rate": 1.583327333302788e-05,
      "loss": 0.0746,
      "step": 97650
    },
    {
      "epoch": 1.4384913329700595,
      "grad_norm": 1.3520724773406982,
      "learning_rate": 1.5831636961060893e-05,
      "loss": 0.0713,
      "step": 97675
    },
    {
      "epoch": 1.4388595160601465,
      "grad_norm": 1.6861071586608887,
      "learning_rate": 1.583000058909391e-05,
      "loss": 0.0752,
      "step": 97700
    },
    {
      "epoch": 1.4392276991502335,
      "grad_norm": 1.742470383644104,
      "learning_rate": 1.5828364217126925e-05,
      "loss": 0.0672,
      "step": 97725
    },
    {
      "epoch": 1.4395958822403205,
      "grad_norm": 1.4519072771072388,
      "learning_rate": 1.582672784515994e-05,
      "loss": 0.0748,
      "step": 97750
    },
    {
      "epoch": 1.4399640653304075,
      "grad_norm": 1.3369940519332886,
      "learning_rate": 1.5825091473192954e-05,
      "loss": 0.0781,
      "step": 97775
    },
    {
      "epoch": 1.4403322484204946,
      "grad_norm": 1.3350610733032227,
      "learning_rate": 1.5823455101225972e-05,
      "loss": 0.0748,
      "step": 97800
    },
    {
      "epoch": 1.4407004315105816,
      "grad_norm": 1.5795238018035889,
      "learning_rate": 1.5821818729258987e-05,
      "loss": 0.0776,
      "step": 97825
    },
    {
      "epoch": 1.4410686146006686,
      "grad_norm": 1.3560709953308105,
      "learning_rate": 1.5820182357292e-05,
      "loss": 0.0639,
      "step": 97850
    },
    {
      "epoch": 1.4414367976907556,
      "grad_norm": 1.3758741617202759,
      "learning_rate": 1.581854598532502e-05,
      "loss": 0.0782,
      "step": 97875
    },
    {
      "epoch": 1.4418049807808426,
      "grad_norm": 1.6172165870666504,
      "learning_rate": 1.5816909613358033e-05,
      "loss": 0.0776,
      "step": 97900
    },
    {
      "epoch": 1.4421731638709296,
      "grad_norm": 1.5455402135849,
      "learning_rate": 1.5815273241391048e-05,
      "loss": 0.0779,
      "step": 97925
    },
    {
      "epoch": 1.4425413469610167,
      "grad_norm": 1.3516970872879028,
      "learning_rate": 1.5813636869424062e-05,
      "loss": 0.0714,
      "step": 97950
    },
    {
      "epoch": 1.442909530051104,
      "grad_norm": 1.3554434776306152,
      "learning_rate": 1.5812000497457077e-05,
      "loss": 0.0723,
      "step": 97975
    },
    {
      "epoch": 1.443277713141191,
      "grad_norm": 1.6780847311019897,
      "learning_rate": 1.5810364125490095e-05,
      "loss": 0.0789,
      "step": 98000
    },
    {
      "epoch": 1.443645896231278,
      "grad_norm": 1.2315202951431274,
      "learning_rate": 1.580872775352311e-05,
      "loss": 0.0773,
      "step": 98025
    },
    {
      "epoch": 1.444014079321365,
      "grad_norm": 1.3949530124664307,
      "learning_rate": 1.5807091381556127e-05,
      "loss": 0.071,
      "step": 98050
    },
    {
      "epoch": 1.444382262411452,
      "grad_norm": 1.6291261911392212,
      "learning_rate": 1.580545500958914e-05,
      "loss": 0.0813,
      "step": 98075
    },
    {
      "epoch": 1.444750445501539,
      "grad_norm": 1.2096186876296997,
      "learning_rate": 1.5803818637622156e-05,
      "loss": 0.0726,
      "step": 98100
    },
    {
      "epoch": 1.445118628591626,
      "grad_norm": 1.509542465209961,
      "learning_rate": 1.5802247720533853e-05,
      "loss": 0.0743,
      "step": 98125
    },
    {
      "epoch": 1.445486811681713,
      "grad_norm": 1.62225341796875,
      "learning_rate": 1.5800611348566868e-05,
      "loss": 0.0763,
      "step": 98150
    },
    {
      "epoch": 1.4458549947718002,
      "grad_norm": 1.1660709381103516,
      "learning_rate": 1.5798974976599882e-05,
      "loss": 0.0654,
      "step": 98175
    },
    {
      "epoch": 1.4462231778618873,
      "grad_norm": 1.1793895959854126,
      "learning_rate": 1.5797338604632896e-05,
      "loss": 0.0696,
      "step": 98200
    },
    {
      "epoch": 1.4465913609519743,
      "grad_norm": 1.0981831550598145,
      "learning_rate": 1.5795702232665914e-05,
      "loss": 0.077,
      "step": 98225
    },
    {
      "epoch": 1.4469595440420613,
      "grad_norm": 1.3388949632644653,
      "learning_rate": 1.579406586069893e-05,
      "loss": 0.0803,
      "step": 98250
    },
    {
      "epoch": 1.4473277271321483,
      "grad_norm": 1.572522521018982,
      "learning_rate": 1.5792429488731943e-05,
      "loss": 0.0742,
      "step": 98275
    },
    {
      "epoch": 1.4476959102222353,
      "grad_norm": 1.7953388690948486,
      "learning_rate": 1.579079311676496e-05,
      "loss": 0.0715,
      "step": 98300
    },
    {
      "epoch": 1.4480640933123223,
      "grad_norm": 1.402962327003479,
      "learning_rate": 1.5789156744797976e-05,
      "loss": 0.0797,
      "step": 98325
    },
    {
      "epoch": 1.4484322764024093,
      "grad_norm": 1.1952656507492065,
      "learning_rate": 1.578752037283099e-05,
      "loss": 0.0732,
      "step": 98350
    },
    {
      "epoch": 1.4488004594924964,
      "grad_norm": 1.5005369186401367,
      "learning_rate": 1.5785884000864004e-05,
      "loss": 0.0657,
      "step": 98375
    },
    {
      "epoch": 1.4491686425825834,
      "grad_norm": 1.3722046613693237,
      "learning_rate": 1.578424762889702e-05,
      "loss": 0.0741,
      "step": 98400
    },
    {
      "epoch": 1.4495368256726704,
      "grad_norm": 1.3201954364776611,
      "learning_rate": 1.5782611256930037e-05,
      "loss": 0.0702,
      "step": 98425
    },
    {
      "epoch": 1.4499050087627574,
      "grad_norm": 1.0372276306152344,
      "learning_rate": 1.578097488496305e-05,
      "loss": 0.0694,
      "step": 98450
    },
    {
      "epoch": 1.4502731918528446,
      "grad_norm": 1.4597052335739136,
      "learning_rate": 1.577933851299607e-05,
      "loss": 0.0709,
      "step": 98475
    },
    {
      "epoch": 1.4506413749429317,
      "grad_norm": 1.759572982788086,
      "learning_rate": 1.5777702141029083e-05,
      "loss": 0.0742,
      "step": 98500
    },
    {
      "epoch": 1.4510095580330187,
      "grad_norm": 1.263596773147583,
      "learning_rate": 1.5776065769062098e-05,
      "loss": 0.0752,
      "step": 98525
    },
    {
      "epoch": 1.4513777411231057,
      "grad_norm": 1.363990306854248,
      "learning_rate": 1.5774429397095116e-05,
      "loss": 0.0789,
      "step": 98550
    },
    {
      "epoch": 1.4517459242131927,
      "grad_norm": 1.1553142070770264,
      "learning_rate": 1.577279302512813e-05,
      "loss": 0.0785,
      "step": 98575
    },
    {
      "epoch": 1.4521141073032797,
      "grad_norm": 1.4600054025650024,
      "learning_rate": 1.5771156653161145e-05,
      "loss": 0.0775,
      "step": 98600
    },
    {
      "epoch": 1.4524822903933667,
      "grad_norm": 1.1618602275848389,
      "learning_rate": 1.576952028119416e-05,
      "loss": 0.0708,
      "step": 98625
    },
    {
      "epoch": 1.452850473483454,
      "grad_norm": 1.7354905605316162,
      "learning_rate": 1.5767883909227174e-05,
      "loss": 0.0749,
      "step": 98650
    },
    {
      "epoch": 1.453218656573541,
      "grad_norm": 1.7131768465042114,
      "learning_rate": 1.576624753726019e-05,
      "loss": 0.068,
      "step": 98675
    },
    {
      "epoch": 1.453586839663628,
      "grad_norm": 1.8787506818771362,
      "learning_rate": 1.5764611165293206e-05,
      "loss": 0.0718,
      "step": 98700
    },
    {
      "epoch": 1.453955022753715,
      "grad_norm": 1.067794680595398,
      "learning_rate": 1.5762974793326224e-05,
      "loss": 0.0734,
      "step": 98725
    },
    {
      "epoch": 1.454323205843802,
      "grad_norm": 1.3868683576583862,
      "learning_rate": 1.5761338421359238e-05,
      "loss": 0.0735,
      "step": 98750
    },
    {
      "epoch": 1.454691388933889,
      "grad_norm": 1.3692728281021118,
      "learning_rate": 1.5759702049392253e-05,
      "loss": 0.0748,
      "step": 98775
    },
    {
      "epoch": 1.455059572023976,
      "grad_norm": 1.8034658432006836,
      "learning_rate": 1.5758065677425267e-05,
      "loss": 0.0726,
      "step": 98800
    },
    {
      "epoch": 1.455427755114063,
      "grad_norm": 1.0928919315338135,
      "learning_rate": 1.575642930545828e-05,
      "loss": 0.0727,
      "step": 98825
    },
    {
      "epoch": 1.45579593820415,
      "grad_norm": 1.397918701171875,
      "learning_rate": 1.57547929334913e-05,
      "loss": 0.0785,
      "step": 98850
    },
    {
      "epoch": 1.4561641212942371,
      "grad_norm": 1.60765540599823,
      "learning_rate": 1.5753156561524314e-05,
      "loss": 0.0744,
      "step": 98875
    },
    {
      "epoch": 1.4565323043843241,
      "grad_norm": 1.329447865486145,
      "learning_rate": 1.575152018955733e-05,
      "loss": 0.0697,
      "step": 98900
    },
    {
      "epoch": 1.4569004874744111,
      "grad_norm": 1.276035189628601,
      "learning_rate": 1.5749883817590346e-05,
      "loss": 0.0698,
      "step": 98925
    },
    {
      "epoch": 1.4572686705644984,
      "grad_norm": 1.4128130674362183,
      "learning_rate": 1.574824744562336e-05,
      "loss": 0.0733,
      "step": 98950
    },
    {
      "epoch": 1.4576368536545854,
      "grad_norm": 1.2025258541107178,
      "learning_rate": 1.574661107365638e-05,
      "loss": 0.0679,
      "step": 98975
    },
    {
      "epoch": 1.4580050367446724,
      "grad_norm": 1.434647798538208,
      "learning_rate": 1.5744974701689393e-05,
      "loss": 0.0675,
      "step": 99000
    },
    {
      "epoch": 1.4583732198347594,
      "grad_norm": 1.061908483505249,
      "learning_rate": 1.5743338329722407e-05,
      "loss": 0.0791,
      "step": 99025
    },
    {
      "epoch": 1.4587414029248464,
      "grad_norm": 1.6741228103637695,
      "learning_rate": 1.5741701957755422e-05,
      "loss": 0.0755,
      "step": 99050
    },
    {
      "epoch": 1.4591095860149335,
      "grad_norm": 1.8702377080917358,
      "learning_rate": 1.5740065585788436e-05,
      "loss": 0.0736,
      "step": 99075
    },
    {
      "epoch": 1.4594777691050205,
      "grad_norm": 1.4650659561157227,
      "learning_rate": 1.5738429213821454e-05,
      "loss": 0.0767,
      "step": 99100
    },
    {
      "epoch": 1.4598459521951077,
      "grad_norm": 1.5197646617889404,
      "learning_rate": 1.573679284185447e-05,
      "loss": 0.0723,
      "step": 99125
    },
    {
      "epoch": 1.4602141352851947,
      "grad_norm": 1.5481562614440918,
      "learning_rate": 1.5735156469887486e-05,
      "loss": 0.0769,
      "step": 99150
    },
    {
      "epoch": 1.4605823183752817,
      "grad_norm": 1.1672195196151733,
      "learning_rate": 1.57335200979205e-05,
      "loss": 0.0732,
      "step": 99175
    },
    {
      "epoch": 1.4609505014653688,
      "grad_norm": 1.4054820537567139,
      "learning_rate": 1.5731883725953515e-05,
      "loss": 0.0703,
      "step": 99200
    },
    {
      "epoch": 1.4613186845554558,
      "grad_norm": 1.4027718305587769,
      "learning_rate": 1.573024735398653e-05,
      "loss": 0.0793,
      "step": 99225
    },
    {
      "epoch": 1.4616868676455428,
      "grad_norm": 1.2779954671859741,
      "learning_rate": 1.5728610982019544e-05,
      "loss": 0.0731,
      "step": 99250
    },
    {
      "epoch": 1.4620550507356298,
      "grad_norm": 1.266919732093811,
      "learning_rate": 1.5726974610052562e-05,
      "loss": 0.0654,
      "step": 99275
    },
    {
      "epoch": 1.4624232338257168,
      "grad_norm": 0.884549617767334,
      "learning_rate": 1.5725338238085577e-05,
      "loss": 0.0721,
      "step": 99300
    },
    {
      "epoch": 1.4627914169158038,
      "grad_norm": 1.6032646894454956,
      "learning_rate": 1.572370186611859e-05,
      "loss": 0.0726,
      "step": 99325
    },
    {
      "epoch": 1.4631596000058908,
      "grad_norm": 1.942748785018921,
      "learning_rate": 1.572206549415161e-05,
      "loss": 0.0752,
      "step": 99350
    },
    {
      "epoch": 1.4635277830959779,
      "grad_norm": 1.341539740562439,
      "learning_rate": 1.5720429122184623e-05,
      "loss": 0.0765,
      "step": 99375
    },
    {
      "epoch": 1.4638959661860649,
      "grad_norm": 1.8556504249572754,
      "learning_rate": 1.571879275021764e-05,
      "loss": 0.0688,
      "step": 99400
    },
    {
      "epoch": 1.4642641492761521,
      "grad_norm": 1.3915098905563354,
      "learning_rate": 1.5717156378250652e-05,
      "loss": 0.081,
      "step": 99425
    },
    {
      "epoch": 1.4646323323662391,
      "grad_norm": 1.2533347606658936,
      "learning_rate": 1.571552000628367e-05,
      "loss": 0.067,
      "step": 99450
    },
    {
      "epoch": 1.4650005154563261,
      "grad_norm": 1.229148268699646,
      "learning_rate": 1.5713883634316685e-05,
      "loss": 0.0707,
      "step": 99475
    },
    {
      "epoch": 1.4653686985464132,
      "grad_norm": 1.8005704879760742,
      "learning_rate": 1.57122472623497e-05,
      "loss": 0.0785,
      "step": 99500
    },
    {
      "epoch": 1.4657368816365002,
      "grad_norm": 1.70078706741333,
      "learning_rate": 1.5710610890382717e-05,
      "loss": 0.0715,
      "step": 99525
    },
    {
      "epoch": 1.4661050647265872,
      "grad_norm": 1.2894269227981567,
      "learning_rate": 1.570897451841573e-05,
      "loss": 0.075,
      "step": 99550
    },
    {
      "epoch": 1.4664732478166742,
      "grad_norm": 1.5065540075302124,
      "learning_rate": 1.5707338146448746e-05,
      "loss": 0.0765,
      "step": 99575
    },
    {
      "epoch": 1.4668414309067614,
      "grad_norm": 1.3765555620193481,
      "learning_rate": 1.5705701774481764e-05,
      "loss": 0.0656,
      "step": 99600
    },
    {
      "epoch": 1.4672096139968485,
      "grad_norm": 1.5339314937591553,
      "learning_rate": 1.5704065402514778e-05,
      "loss": 0.0695,
      "step": 99625
    },
    {
      "epoch": 1.4675777970869355,
      "grad_norm": 1.2181205749511719,
      "learning_rate": 1.5702429030547793e-05,
      "loss": 0.0708,
      "step": 99650
    },
    {
      "epoch": 1.4679459801770225,
      "grad_norm": 1.615257740020752,
      "learning_rate": 1.5700792658580807e-05,
      "loss": 0.0829,
      "step": 99675
    },
    {
      "epoch": 1.4683141632671095,
      "grad_norm": 1.094374179840088,
      "learning_rate": 1.5699156286613825e-05,
      "loss": 0.0719,
      "step": 99700
    },
    {
      "epoch": 1.4686823463571965,
      "grad_norm": 1.393290638923645,
      "learning_rate": 1.569751991464684e-05,
      "loss": 0.0781,
      "step": 99725
    },
    {
      "epoch": 1.4690505294472835,
      "grad_norm": 1.2727437019348145,
      "learning_rate": 1.5695883542679854e-05,
      "loss": 0.0736,
      "step": 99750
    },
    {
      "epoch": 1.4694187125373706,
      "grad_norm": 1.0170605182647705,
      "learning_rate": 1.569424717071287e-05,
      "loss": 0.0726,
      "step": 99775
    },
    {
      "epoch": 1.4697868956274576,
      "grad_norm": 1.6831586360931396,
      "learning_rate": 1.5692610798745886e-05,
      "loss": 0.0676,
      "step": 99800
    },
    {
      "epoch": 1.4701550787175446,
      "grad_norm": 1.3987754583358765,
      "learning_rate": 1.56909744267789e-05,
      "loss": 0.0711,
      "step": 99825
    },
    {
      "epoch": 1.4705232618076316,
      "grad_norm": 1.411293387413025,
      "learning_rate": 1.5689338054811915e-05,
      "loss": 0.0754,
      "step": 99850
    },
    {
      "epoch": 1.4708914448977186,
      "grad_norm": 1.3667187690734863,
      "learning_rate": 1.5687701682844933e-05,
      "loss": 0.0687,
      "step": 99875
    },
    {
      "epoch": 1.4712596279878059,
      "grad_norm": 1.6855864524841309,
      "learning_rate": 1.5686065310877947e-05,
      "loss": 0.0752,
      "step": 99900
    },
    {
      "epoch": 1.4716278110778929,
      "grad_norm": 1.7509233951568604,
      "learning_rate": 1.5684428938910962e-05,
      "loss": 0.07,
      "step": 99925
    },
    {
      "epoch": 1.4719959941679799,
      "grad_norm": 1.9707937240600586,
      "learning_rate": 1.568279256694398e-05,
      "loss": 0.074,
      "step": 99950
    },
    {
      "epoch": 1.472364177258067,
      "grad_norm": 1.3641451597213745,
      "learning_rate": 1.5681156194976994e-05,
      "loss": 0.0726,
      "step": 99975
    },
    {
      "epoch": 1.472732360348154,
      "grad_norm": 1.4700021743774414,
      "learning_rate": 1.567951982301001e-05,
      "loss": 0.0637,
      "step": 100000
    },
    {
      "epoch": 1.473100543438241,
      "grad_norm": 1.1824214458465576,
      "learning_rate": 1.5677883451043026e-05,
      "loss": 0.0706,
      "step": 100025
    },
    {
      "epoch": 1.473468726528328,
      "grad_norm": 1.4108469486236572,
      "learning_rate": 1.567624707907604e-05,
      "loss": 0.075,
      "step": 100050
    },
    {
      "epoch": 1.4738369096184152,
      "grad_norm": 1.175336480140686,
      "learning_rate": 1.5674610707109055e-05,
      "loss": 0.0706,
      "step": 100075
    },
    {
      "epoch": 1.4742050927085022,
      "grad_norm": 1.4789600372314453,
      "learning_rate": 1.567297433514207e-05,
      "loss": 0.0667,
      "step": 100100
    },
    {
      "epoch": 1.4745732757985892,
      "grad_norm": 2.167008876800537,
      "learning_rate": 1.5671337963175088e-05,
      "loss": 0.0758,
      "step": 100125
    },
    {
      "epoch": 1.4749414588886762,
      "grad_norm": 1.716241717338562,
      "learning_rate": 1.5669701591208102e-05,
      "loss": 0.0793,
      "step": 100150
    },
    {
      "epoch": 1.4753096419787632,
      "grad_norm": 1.2016184329986572,
      "learning_rate": 1.5668065219241117e-05,
      "loss": 0.0689,
      "step": 100175
    },
    {
      "epoch": 1.4756778250688503,
      "grad_norm": 0.9062038660049438,
      "learning_rate": 1.5666428847274134e-05,
      "loss": 0.0727,
      "step": 100200
    },
    {
      "epoch": 1.4760460081589373,
      "grad_norm": 1.2083696126937866,
      "learning_rate": 1.566479247530715e-05,
      "loss": 0.0664,
      "step": 100225
    },
    {
      "epoch": 1.4764141912490243,
      "grad_norm": 1.3190200328826904,
      "learning_rate": 1.5663156103340163e-05,
      "loss": 0.0693,
      "step": 100250
    },
    {
      "epoch": 1.4767823743391113,
      "grad_norm": 1.0811551809310913,
      "learning_rate": 1.5661519731373178e-05,
      "loss": 0.0656,
      "step": 100275
    },
    {
      "epoch": 1.4771505574291983,
      "grad_norm": 1.033502221107483,
      "learning_rate": 1.5659883359406192e-05,
      "loss": 0.0634,
      "step": 100300
    },
    {
      "epoch": 1.4775187405192853,
      "grad_norm": 0.956745982170105,
      "learning_rate": 1.565824698743921e-05,
      "loss": 0.0708,
      "step": 100325
    },
    {
      "epoch": 1.4778869236093723,
      "grad_norm": 1.2435214519500732,
      "learning_rate": 1.5656610615472224e-05,
      "loss": 0.0681,
      "step": 100350
    },
    {
      "epoch": 1.4782551066994596,
      "grad_norm": 1.8292759656906128,
      "learning_rate": 1.5654974243505242e-05,
      "loss": 0.0781,
      "step": 100375
    },
    {
      "epoch": 1.4786232897895466,
      "grad_norm": 1.631939172744751,
      "learning_rate": 1.5653337871538257e-05,
      "loss": 0.0663,
      "step": 100400
    },
    {
      "epoch": 1.4789914728796336,
      "grad_norm": 0.8636593818664551,
      "learning_rate": 1.565170149957127e-05,
      "loss": 0.0727,
      "step": 100425
    },
    {
      "epoch": 1.4793596559697206,
      "grad_norm": 1.6664172410964966,
      "learning_rate": 1.565006512760429e-05,
      "loss": 0.073,
      "step": 100450
    },
    {
      "epoch": 1.4797278390598076,
      "grad_norm": 1.6664351224899292,
      "learning_rate": 1.5648428755637304e-05,
      "loss": 0.076,
      "step": 100475
    },
    {
      "epoch": 1.4800960221498947,
      "grad_norm": 2.1522881984710693,
      "learning_rate": 1.5646792383670318e-05,
      "loss": 0.0765,
      "step": 100500
    },
    {
      "epoch": 1.4804642052399817,
      "grad_norm": 1.7682160139083862,
      "learning_rate": 1.5645156011703332e-05,
      "loss": 0.0716,
      "step": 100525
    },
    {
      "epoch": 1.4808323883300687,
      "grad_norm": 1.4331692457199097,
      "learning_rate": 1.564351963973635e-05,
      "loss": 0.0725,
      "step": 100550
    },
    {
      "epoch": 1.481200571420156,
      "grad_norm": 1.4662048816680908,
      "learning_rate": 1.5641883267769365e-05,
      "loss": 0.0809,
      "step": 100575
    },
    {
      "epoch": 1.481568754510243,
      "grad_norm": 1.7781012058258057,
      "learning_rate": 1.564024689580238e-05,
      "loss": 0.0739,
      "step": 100600
    },
    {
      "epoch": 1.48193693760033,
      "grad_norm": 1.4978768825531006,
      "learning_rate": 1.5638610523835397e-05,
      "loss": 0.0735,
      "step": 100625
    },
    {
      "epoch": 1.482305120690417,
      "grad_norm": 1.1126130819320679,
      "learning_rate": 1.563697415186841e-05,
      "loss": 0.0756,
      "step": 100650
    },
    {
      "epoch": 1.482673303780504,
      "grad_norm": 1.252961277961731,
      "learning_rate": 1.5635337779901426e-05,
      "loss": 0.0745,
      "step": 100675
    },
    {
      "epoch": 1.483041486870591,
      "grad_norm": 1.9854118824005127,
      "learning_rate": 1.563370140793444e-05,
      "loss": 0.0734,
      "step": 100700
    },
    {
      "epoch": 1.483409669960678,
      "grad_norm": 1.6315757036209106,
      "learning_rate": 1.5632065035967455e-05,
      "loss": 0.0727,
      "step": 100725
    },
    {
      "epoch": 1.483777853050765,
      "grad_norm": 1.4408893585205078,
      "learning_rate": 1.5630428664000473e-05,
      "loss": 0.0801,
      "step": 100750
    },
    {
      "epoch": 1.484146036140852,
      "grad_norm": 1.215943455696106,
      "learning_rate": 1.5628792292033487e-05,
      "loss": 0.0655,
      "step": 100775
    },
    {
      "epoch": 1.484514219230939,
      "grad_norm": 1.3008314371109009,
      "learning_rate": 1.5627155920066505e-05,
      "loss": 0.0714,
      "step": 100800
    },
    {
      "epoch": 1.484882402321026,
      "grad_norm": 1.7536879777908325,
      "learning_rate": 1.562551954809952e-05,
      "loss": 0.0692,
      "step": 100825
    },
    {
      "epoch": 1.485250585411113,
      "grad_norm": 0.9951906204223633,
      "learning_rate": 1.5623883176132534e-05,
      "loss": 0.0805,
      "step": 100850
    },
    {
      "epoch": 1.4856187685012003,
      "grad_norm": 1.1625759601593018,
      "learning_rate": 1.5622246804165552e-05,
      "loss": 0.0697,
      "step": 100875
    },
    {
      "epoch": 1.4859869515912874,
      "grad_norm": 1.129320740699768,
      "learning_rate": 1.5620675887077242e-05,
      "loss": 0.0737,
      "step": 100900
    },
    {
      "epoch": 1.4863551346813744,
      "grad_norm": 1.3437418937683105,
      "learning_rate": 1.561903951511026e-05,
      "loss": 0.0706,
      "step": 100925
    },
    {
      "epoch": 1.4867233177714614,
      "grad_norm": 1.1932883262634277,
      "learning_rate": 1.5617403143143275e-05,
      "loss": 0.0731,
      "step": 100950
    },
    {
      "epoch": 1.4870915008615484,
      "grad_norm": 1.018709421157837,
      "learning_rate": 1.561576677117629e-05,
      "loss": 0.074,
      "step": 100975
    },
    {
      "epoch": 1.4874596839516354,
      "grad_norm": 1.311795949935913,
      "learning_rate": 1.5614130399209307e-05,
      "loss": 0.0732,
      "step": 101000
    },
    {
      "epoch": 1.4878278670417224,
      "grad_norm": 1.0393089056015015,
      "learning_rate": 1.561249402724232e-05,
      "loss": 0.0675,
      "step": 101025
    },
    {
      "epoch": 1.4881960501318097,
      "grad_norm": 1.8284398317337036,
      "learning_rate": 1.561085765527534e-05,
      "loss": 0.0684,
      "step": 101050
    },
    {
      "epoch": 1.4885642332218967,
      "grad_norm": 1.3828203678131104,
      "learning_rate": 1.5609221283308354e-05,
      "loss": 0.0733,
      "step": 101075
    },
    {
      "epoch": 1.4889324163119837,
      "grad_norm": 1.7602605819702148,
      "learning_rate": 1.5607584911341368e-05,
      "loss": 0.0748,
      "step": 101100
    },
    {
      "epoch": 1.4893005994020707,
      "grad_norm": 1.1913679838180542,
      "learning_rate": 1.5605948539374383e-05,
      "loss": 0.0809,
      "step": 101125
    },
    {
      "epoch": 1.4896687824921577,
      "grad_norm": 1.6008341312408447,
      "learning_rate": 1.5604312167407397e-05,
      "loss": 0.076,
      "step": 101150
    },
    {
      "epoch": 1.4900369655822447,
      "grad_norm": 1.4985178709030151,
      "learning_rate": 1.5602675795440415e-05,
      "loss": 0.0812,
      "step": 101175
    },
    {
      "epoch": 1.4904051486723318,
      "grad_norm": 1.3773492574691772,
      "learning_rate": 1.560103942347343e-05,
      "loss": 0.0805,
      "step": 101200
    },
    {
      "epoch": 1.4907733317624188,
      "grad_norm": 1.7686105966567993,
      "learning_rate": 1.5599403051506444e-05,
      "loss": 0.0765,
      "step": 101225
    },
    {
      "epoch": 1.4911415148525058,
      "grad_norm": 1.4000052213668823,
      "learning_rate": 1.559776667953946e-05,
      "loss": 0.0697,
      "step": 101250
    },
    {
      "epoch": 1.4915096979425928,
      "grad_norm": 1.6129021644592285,
      "learning_rate": 1.5596130307572476e-05,
      "loss": 0.0766,
      "step": 101275
    },
    {
      "epoch": 1.4918778810326798,
      "grad_norm": 1.3519954681396484,
      "learning_rate": 1.5594493935605494e-05,
      "loss": 0.0759,
      "step": 101300
    },
    {
      "epoch": 1.4922460641227668,
      "grad_norm": 1.3702207803726196,
      "learning_rate": 1.5592857563638505e-05,
      "loss": 0.0751,
      "step": 101325
    },
    {
      "epoch": 1.492614247212854,
      "grad_norm": 1.0622360706329346,
      "learning_rate": 1.5591221191671523e-05,
      "loss": 0.0696,
      "step": 101350
    },
    {
      "epoch": 1.492982430302941,
      "grad_norm": 1.2110073566436768,
      "learning_rate": 1.5589584819704537e-05,
      "loss": 0.0693,
      "step": 101375
    },
    {
      "epoch": 1.493350613393028,
      "grad_norm": 1.026301383972168,
      "learning_rate": 1.5587948447737552e-05,
      "loss": 0.0751,
      "step": 101400
    },
    {
      "epoch": 1.4937187964831151,
      "grad_norm": 1.6028214693069458,
      "learning_rate": 1.558631207577057e-05,
      "loss": 0.0696,
      "step": 101425
    },
    {
      "epoch": 1.4940869795732021,
      "grad_norm": 1.1455265283584595,
      "learning_rate": 1.5584675703803584e-05,
      "loss": 0.0682,
      "step": 101450
    },
    {
      "epoch": 1.4944551626632891,
      "grad_norm": 1.6425275802612305,
      "learning_rate": 1.55830393318366e-05,
      "loss": 0.0745,
      "step": 101475
    },
    {
      "epoch": 1.4948233457533762,
      "grad_norm": 0.8957460522651672,
      "learning_rate": 1.5581402959869616e-05,
      "loss": 0.0761,
      "step": 101500
    },
    {
      "epoch": 1.4951915288434634,
      "grad_norm": 1.350748062133789,
      "learning_rate": 1.557976658790263e-05,
      "loss": 0.0737,
      "step": 101525
    },
    {
      "epoch": 1.4955597119335504,
      "grad_norm": 1.6226164102554321,
      "learning_rate": 1.5578130215935645e-05,
      "loss": 0.0698,
      "step": 101550
    },
    {
      "epoch": 1.4959278950236374,
      "grad_norm": 1.4175498485565186,
      "learning_rate": 1.557649384396866e-05,
      "loss": 0.0714,
      "step": 101575
    },
    {
      "epoch": 1.4962960781137244,
      "grad_norm": 1.5930166244506836,
      "learning_rate": 1.5574857472001678e-05,
      "loss": 0.0827,
      "step": 101600
    },
    {
      "epoch": 1.4966642612038115,
      "grad_norm": 1.288759469985962,
      "learning_rate": 1.5573221100034692e-05,
      "loss": 0.0778,
      "step": 101625
    },
    {
      "epoch": 1.4970324442938985,
      "grad_norm": 1.360939860343933,
      "learning_rate": 1.5571584728067707e-05,
      "loss": 0.067,
      "step": 101650
    },
    {
      "epoch": 1.4974006273839855,
      "grad_norm": 1.3344268798828125,
      "learning_rate": 1.5569948356100724e-05,
      "loss": 0.0806,
      "step": 101675
    },
    {
      "epoch": 1.4977688104740725,
      "grad_norm": 1.3413747549057007,
      "learning_rate": 1.556831198413374e-05,
      "loss": 0.0731,
      "step": 101700
    },
    {
      "epoch": 1.4981369935641595,
      "grad_norm": 1.6871967315673828,
      "learning_rate": 1.5566675612166757e-05,
      "loss": 0.0802,
      "step": 101725
    },
    {
      "epoch": 1.4985051766542465,
      "grad_norm": 1.1810795068740845,
      "learning_rate": 1.5565039240199768e-05,
      "loss": 0.0695,
      "step": 101750
    },
    {
      "epoch": 1.4988733597443336,
      "grad_norm": 1.5120656490325928,
      "learning_rate": 1.5563402868232786e-05,
      "loss": 0.0692,
      "step": 101775
    },
    {
      "epoch": 1.4992415428344206,
      "grad_norm": 1.4127033948898315,
      "learning_rate": 1.55617664962658e-05,
      "loss": 0.0709,
      "step": 101800
    },
    {
      "epoch": 1.4996097259245078,
      "grad_norm": 1.239804744720459,
      "learning_rate": 1.5560130124298814e-05,
      "loss": 0.0706,
      "step": 101825
    },
    {
      "epoch": 1.4999779090145948,
      "grad_norm": 1.615360140800476,
      "learning_rate": 1.5558493752331832e-05,
      "loss": 0.0744,
      "step": 101850
    },
    {
      "epoch": 1.5003460921046818,
      "grad_norm": 1.5092477798461914,
      "learning_rate": 1.5556857380364847e-05,
      "loss": 0.0734,
      "step": 101875
    },
    {
      "epoch": 1.5007142751947689,
      "grad_norm": 1.6290446519851685,
      "learning_rate": 1.555522100839786e-05,
      "loss": 0.0638,
      "step": 101900
    },
    {
      "epoch": 1.5010824582848559,
      "grad_norm": 1.972309947013855,
      "learning_rate": 1.555358463643088e-05,
      "loss": 0.0671,
      "step": 101925
    },
    {
      "epoch": 1.5014506413749429,
      "grad_norm": 1.2505921125411987,
      "learning_rate": 1.5551948264463894e-05,
      "loss": 0.0706,
      "step": 101950
    },
    {
      "epoch": 1.5018188244650301,
      "grad_norm": 1.4129692316055298,
      "learning_rate": 1.5550311892496908e-05,
      "loss": 0.0817,
      "step": 101975
    },
    {
      "epoch": 1.5021870075551171,
      "grad_norm": 1.4573155641555786,
      "learning_rate": 1.5548675520529922e-05,
      "loss": 0.0755,
      "step": 102000
    },
    {
      "epoch": 1.5025551906452042,
      "grad_norm": 1.9313448667526245,
      "learning_rate": 1.554703914856294e-05,
      "loss": 0.0696,
      "step": 102025
    },
    {
      "epoch": 1.5029233737352912,
      "grad_norm": 1.624696969985962,
      "learning_rate": 1.5545402776595955e-05,
      "loss": 0.0798,
      "step": 102050
    },
    {
      "epoch": 1.5032915568253782,
      "grad_norm": 1.3338876962661743,
      "learning_rate": 1.554376640462897e-05,
      "loss": 0.0748,
      "step": 102075
    },
    {
      "epoch": 1.5036597399154652,
      "grad_norm": 1.4414469003677368,
      "learning_rate": 1.5542130032661987e-05,
      "loss": 0.07,
      "step": 102100
    },
    {
      "epoch": 1.5040279230055522,
      "grad_norm": 1.9996525049209595,
      "learning_rate": 1.5540493660695e-05,
      "loss": 0.0758,
      "step": 102125
    },
    {
      "epoch": 1.5043961060956392,
      "grad_norm": 1.3696807622909546,
      "learning_rate": 1.5538857288728016e-05,
      "loss": 0.071,
      "step": 102150
    },
    {
      "epoch": 1.5047642891857262,
      "grad_norm": 1.4292805194854736,
      "learning_rate": 1.553722091676103e-05,
      "loss": 0.0688,
      "step": 102175
    },
    {
      "epoch": 1.5051324722758133,
      "grad_norm": 1.372391939163208,
      "learning_rate": 1.5535584544794048e-05,
      "loss": 0.0711,
      "step": 102200
    },
    {
      "epoch": 1.5055006553659003,
      "grad_norm": 1.611112356185913,
      "learning_rate": 1.5533948172827063e-05,
      "loss": 0.0711,
      "step": 102225
    },
    {
      "epoch": 1.5058688384559873,
      "grad_norm": 1.2051068544387817,
      "learning_rate": 1.5532311800860077e-05,
      "loss": 0.0723,
      "step": 102250
    },
    {
      "epoch": 1.5062370215460743,
      "grad_norm": 1.5507503747940063,
      "learning_rate": 1.5530675428893095e-05,
      "loss": 0.0714,
      "step": 102275
    },
    {
      "epoch": 1.5066052046361613,
      "grad_norm": 1.495367407798767,
      "learning_rate": 1.552903905692611e-05,
      "loss": 0.0691,
      "step": 102300
    },
    {
      "epoch": 1.5069733877262486,
      "grad_norm": 1.495174765586853,
      "learning_rate": 1.5527402684959124e-05,
      "loss": 0.0744,
      "step": 102325
    },
    {
      "epoch": 1.5073415708163356,
      "grad_norm": 1.657590389251709,
      "learning_rate": 1.5525766312992142e-05,
      "loss": 0.07,
      "step": 102350
    },
    {
      "epoch": 1.5077097539064226,
      "grad_norm": 1.5004724264144897,
      "learning_rate": 1.5524129941025153e-05,
      "loss": 0.0714,
      "step": 102375
    },
    {
      "epoch": 1.5080779369965096,
      "grad_norm": 1.940828800201416,
      "learning_rate": 1.552249356905817e-05,
      "loss": 0.0703,
      "step": 102400
    },
    {
      "epoch": 1.5084461200865966,
      "grad_norm": 1.0247461795806885,
      "learning_rate": 1.5520857197091185e-05,
      "loss": 0.0778,
      "step": 102425
    },
    {
      "epoch": 1.5088143031766839,
      "grad_norm": 1.2785711288452148,
      "learning_rate": 1.5519220825124203e-05,
      "loss": 0.0703,
      "step": 102450
    },
    {
      "epoch": 1.5091824862667709,
      "grad_norm": 1.5319916009902954,
      "learning_rate": 1.5517584453157217e-05,
      "loss": 0.0749,
      "step": 102475
    },
    {
      "epoch": 1.5095506693568579,
      "grad_norm": 1.5376935005187988,
      "learning_rate": 1.5515948081190232e-05,
      "loss": 0.0714,
      "step": 102500
    },
    {
      "epoch": 1.509918852446945,
      "grad_norm": 1.0213650465011597,
      "learning_rate": 1.551431170922325e-05,
      "loss": 0.0616,
      "step": 102525
    },
    {
      "epoch": 1.510287035537032,
      "grad_norm": 1.327736258506775,
      "learning_rate": 1.5512675337256264e-05,
      "loss": 0.0739,
      "step": 102550
    },
    {
      "epoch": 1.510655218627119,
      "grad_norm": 1.6903321743011475,
      "learning_rate": 1.551103896528928e-05,
      "loss": 0.0681,
      "step": 102575
    },
    {
      "epoch": 1.511023401717206,
      "grad_norm": 1.4521526098251343,
      "learning_rate": 1.5509402593322293e-05,
      "loss": 0.0686,
      "step": 102600
    },
    {
      "epoch": 1.511391584807293,
      "grad_norm": 1.2098474502563477,
      "learning_rate": 1.5507766221355308e-05,
      "loss": 0.0688,
      "step": 102625
    },
    {
      "epoch": 1.51175976789738,
      "grad_norm": 1.935627818107605,
      "learning_rate": 1.5506129849388325e-05,
      "loss": 0.0706,
      "step": 102650
    },
    {
      "epoch": 1.512127950987467,
      "grad_norm": 1.023953914642334,
      "learning_rate": 1.550449347742134e-05,
      "loss": 0.0678,
      "step": 102675
    },
    {
      "epoch": 1.512496134077554,
      "grad_norm": 1.8038815259933472,
      "learning_rate": 1.5502857105454358e-05,
      "loss": 0.0838,
      "step": 102700
    },
    {
      "epoch": 1.512864317167641,
      "grad_norm": 1.3764445781707764,
      "learning_rate": 1.5501220733487372e-05,
      "loss": 0.0657,
      "step": 102725
    },
    {
      "epoch": 1.513232500257728,
      "grad_norm": 1.6889528036117554,
      "learning_rate": 1.5499584361520387e-05,
      "loss": 0.0764,
      "step": 102750
    },
    {
      "epoch": 1.513600683347815,
      "grad_norm": 1.3472950458526611,
      "learning_rate": 1.5497947989553405e-05,
      "loss": 0.0717,
      "step": 102775
    },
    {
      "epoch": 1.5139688664379023,
      "grad_norm": 1.6028352975845337,
      "learning_rate": 1.5496311617586416e-05,
      "loss": 0.0812,
      "step": 102800
    },
    {
      "epoch": 1.5143370495279893,
      "grad_norm": 0.8961876630783081,
      "learning_rate": 1.5494675245619433e-05,
      "loss": 0.0797,
      "step": 102825
    },
    {
      "epoch": 1.5147052326180763,
      "grad_norm": 1.5843199491500854,
      "learning_rate": 1.5493038873652448e-05,
      "loss": 0.0757,
      "step": 102850
    },
    {
      "epoch": 1.5150734157081633,
      "grad_norm": 1.6867176294326782,
      "learning_rate": 1.5491402501685466e-05,
      "loss": 0.0752,
      "step": 102875
    },
    {
      "epoch": 1.5154415987982504,
      "grad_norm": 1.3222062587738037,
      "learning_rate": 1.548976612971848e-05,
      "loss": 0.0702,
      "step": 102900
    },
    {
      "epoch": 1.5158097818883376,
      "grad_norm": 1.1368613243103027,
      "learning_rate": 1.5488129757751495e-05,
      "loss": 0.075,
      "step": 102925
    },
    {
      "epoch": 1.5161779649784246,
      "grad_norm": 1.2368009090423584,
      "learning_rate": 1.5486493385784513e-05,
      "loss": 0.0755,
      "step": 102950
    },
    {
      "epoch": 1.5165461480685116,
      "grad_norm": Infinity,
      "learning_rate": 1.5484922468696206e-05,
      "loss": 0.0694,
      "step": 102975
    },
    {
      "epoch": 1.5169143311585986,
      "grad_norm": 1.538528323173523,
      "learning_rate": 1.548328609672922e-05,
      "loss": 0.0783,
      "step": 103000
    },
    {
      "epoch": 1.5172825142486857,
      "grad_norm": 1.6415791511535645,
      "learning_rate": 1.5481649724762235e-05,
      "loss": 0.0702,
      "step": 103025
    },
    {
      "epoch": 1.5176506973387727,
      "grad_norm": 1.1010034084320068,
      "learning_rate": 1.548001335279525e-05,
      "loss": 0.0813,
      "step": 103050
    },
    {
      "epoch": 1.5180188804288597,
      "grad_norm": 1.3240681886672974,
      "learning_rate": 1.5478376980828268e-05,
      "loss": 0.0798,
      "step": 103075
    },
    {
      "epoch": 1.5183870635189467,
      "grad_norm": 1.6091612577438354,
      "learning_rate": 1.5476740608861282e-05,
      "loss": 0.0683,
      "step": 103100
    },
    {
      "epoch": 1.5187552466090337,
      "grad_norm": 1.1695464849472046,
      "learning_rate": 1.54751042368943e-05,
      "loss": 0.0706,
      "step": 103125
    },
    {
      "epoch": 1.5191234296991207,
      "grad_norm": 1.1147621870040894,
      "learning_rate": 1.5473467864927314e-05,
      "loss": 0.0741,
      "step": 103150
    },
    {
      "epoch": 1.5194916127892077,
      "grad_norm": 1.2407925128936768,
      "learning_rate": 1.547183149296033e-05,
      "loss": 0.0728,
      "step": 103175
    },
    {
      "epoch": 1.5198597958792948,
      "grad_norm": 1.8448214530944824,
      "learning_rate": 1.5470195120993347e-05,
      "loss": 0.0774,
      "step": 103200
    },
    {
      "epoch": 1.5202279789693818,
      "grad_norm": 1.5810340642929077,
      "learning_rate": 1.5468558749026358e-05,
      "loss": 0.0701,
      "step": 103225
    },
    {
      "epoch": 1.5205961620594688,
      "grad_norm": 1.7474024295806885,
      "learning_rate": 1.5466922377059376e-05,
      "loss": 0.0703,
      "step": 103250
    },
    {
      "epoch": 1.5209643451495558,
      "grad_norm": 1.7364290952682495,
      "learning_rate": 1.546528600509239e-05,
      "loss": 0.0718,
      "step": 103275
    },
    {
      "epoch": 1.521332528239643,
      "grad_norm": 0.8367420434951782,
      "learning_rate": 1.5463649633125404e-05,
      "loss": 0.0726,
      "step": 103300
    },
    {
      "epoch": 1.52170071132973,
      "grad_norm": 1.2589001655578613,
      "learning_rate": 1.5462013261158422e-05,
      "loss": 0.077,
      "step": 103325
    },
    {
      "epoch": 1.522068894419817,
      "grad_norm": 1.3828003406524658,
      "learning_rate": 1.5460376889191437e-05,
      "loss": 0.0717,
      "step": 103350
    },
    {
      "epoch": 1.522437077509904,
      "grad_norm": 1.029020071029663,
      "learning_rate": 1.5458740517224455e-05,
      "loss": 0.0789,
      "step": 103375
    },
    {
      "epoch": 1.5228052605999913,
      "grad_norm": 1.3277159929275513,
      "learning_rate": 1.545710414525747e-05,
      "loss": 0.0778,
      "step": 103400
    },
    {
      "epoch": 1.5231734436900783,
      "grad_norm": 1.0542221069335938,
      "learning_rate": 1.5455467773290484e-05,
      "loss": 0.0661,
      "step": 103425
    },
    {
      "epoch": 1.5235416267801654,
      "grad_norm": 1.4062111377716064,
      "learning_rate": 1.5453831401323498e-05,
      "loss": 0.0711,
      "step": 103450
    },
    {
      "epoch": 1.5239098098702524,
      "grad_norm": 1.2907397747039795,
      "learning_rate": 1.5452195029356512e-05,
      "loss": 0.0726,
      "step": 103475
    },
    {
      "epoch": 1.5242779929603394,
      "grad_norm": 1.5134557485580444,
      "learning_rate": 1.545055865738953e-05,
      "loss": 0.0652,
      "step": 103500
    },
    {
      "epoch": 1.5246461760504264,
      "grad_norm": 1.8192025423049927,
      "learning_rate": 1.5448922285422545e-05,
      "loss": 0.0725,
      "step": 103525
    },
    {
      "epoch": 1.5250143591405134,
      "grad_norm": 1.3823870420455933,
      "learning_rate": 1.544728591345556e-05,
      "loss": 0.0734,
      "step": 103550
    },
    {
      "epoch": 1.5253825422306004,
      "grad_norm": 1.3769203424453735,
      "learning_rate": 1.5445649541488577e-05,
      "loss": 0.0682,
      "step": 103575
    },
    {
      "epoch": 1.5257507253206875,
      "grad_norm": 1.5179260969161987,
      "learning_rate": 1.544401316952159e-05,
      "loss": 0.0712,
      "step": 103600
    },
    {
      "epoch": 1.5261189084107745,
      "grad_norm": 1.086018443107605,
      "learning_rate": 1.544237679755461e-05,
      "loss": 0.08,
      "step": 103625
    },
    {
      "epoch": 1.5264870915008615,
      "grad_norm": 1.3227345943450928,
      "learning_rate": 1.544074042558762e-05,
      "loss": 0.0731,
      "step": 103650
    },
    {
      "epoch": 1.5268552745909485,
      "grad_norm": 1.4783951044082642,
      "learning_rate": 1.5439104053620638e-05,
      "loss": 0.0678,
      "step": 103675
    },
    {
      "epoch": 1.5272234576810355,
      "grad_norm": 1.495962142944336,
      "learning_rate": 1.5437467681653653e-05,
      "loss": 0.0746,
      "step": 103700
    },
    {
      "epoch": 1.5275916407711225,
      "grad_norm": 1.5806061029434204,
      "learning_rate": 1.5435831309686667e-05,
      "loss": 0.0707,
      "step": 103725
    },
    {
      "epoch": 1.5279598238612095,
      "grad_norm": 1.5442841053009033,
      "learning_rate": 1.5434194937719685e-05,
      "loss": 0.0655,
      "step": 103750
    },
    {
      "epoch": 1.5283280069512968,
      "grad_norm": 1.8976647853851318,
      "learning_rate": 1.54325585657527e-05,
      "loss": 0.0774,
      "step": 103775
    },
    {
      "epoch": 1.5286961900413838,
      "grad_norm": 1.5315282344818115,
      "learning_rate": 1.5430922193785714e-05,
      "loss": 0.0714,
      "step": 103800
    },
    {
      "epoch": 1.5290643731314708,
      "grad_norm": 1.432922124862671,
      "learning_rate": 1.5429285821818732e-05,
      "loss": 0.0708,
      "step": 103825
    },
    {
      "epoch": 1.5294325562215578,
      "grad_norm": 1.3605883121490479,
      "learning_rate": 1.5427649449851746e-05,
      "loss": 0.0687,
      "step": 103850
    },
    {
      "epoch": 1.5298007393116448,
      "grad_norm": 1.4034861326217651,
      "learning_rate": 1.542601307788476e-05,
      "loss": 0.0741,
      "step": 103875
    },
    {
      "epoch": 1.530168922401732,
      "grad_norm": 1.5694468021392822,
      "learning_rate": 1.5424376705917775e-05,
      "loss": 0.0644,
      "step": 103900
    },
    {
      "epoch": 1.530537105491819,
      "grad_norm": 1.4174963235855103,
      "learning_rate": 1.5422740333950793e-05,
      "loss": 0.0774,
      "step": 103925
    },
    {
      "epoch": 1.530905288581906,
      "grad_norm": 1.296543002128601,
      "learning_rate": 1.5421103961983807e-05,
      "loss": 0.0802,
      "step": 103950
    },
    {
      "epoch": 1.5312734716719931,
      "grad_norm": 1.7001346349716187,
      "learning_rate": 1.5419467590016822e-05,
      "loss": 0.0771,
      "step": 103975
    },
    {
      "epoch": 1.5316416547620801,
      "grad_norm": 1.6084377765655518,
      "learning_rate": 1.541783121804984e-05,
      "loss": 0.0656,
      "step": 104000
    },
    {
      "epoch": 1.5320098378521672,
      "grad_norm": 1.2867377996444702,
      "learning_rate": 1.5416194846082854e-05,
      "loss": 0.0644,
      "step": 104025
    },
    {
      "epoch": 1.5323780209422542,
      "grad_norm": 1.7494432926177979,
      "learning_rate": 1.5414558474115872e-05,
      "loss": 0.0746,
      "step": 104050
    },
    {
      "epoch": 1.5327462040323412,
      "grad_norm": 1.155861258506775,
      "learning_rate": 1.5412922102148883e-05,
      "loss": 0.0587,
      "step": 104075
    },
    {
      "epoch": 1.5331143871224282,
      "grad_norm": 1.509307861328125,
      "learning_rate": 1.54112857301819e-05,
      "loss": 0.067,
      "step": 104100
    },
    {
      "epoch": 1.5334825702125152,
      "grad_norm": 1.2659516334533691,
      "learning_rate": 1.5409649358214915e-05,
      "loss": 0.0674,
      "step": 104125
    },
    {
      "epoch": 1.5338507533026022,
      "grad_norm": 1.506854772567749,
      "learning_rate": 1.540801298624793e-05,
      "loss": 0.0728,
      "step": 104150
    },
    {
      "epoch": 1.5342189363926892,
      "grad_norm": 1.6616320610046387,
      "learning_rate": 1.5406376614280948e-05,
      "loss": 0.068,
      "step": 104175
    },
    {
      "epoch": 1.5345871194827763,
      "grad_norm": 1.2628109455108643,
      "learning_rate": 1.5404740242313962e-05,
      "loss": 0.0654,
      "step": 104200
    },
    {
      "epoch": 1.5349553025728633,
      "grad_norm": 1.2735340595245361,
      "learning_rate": 1.5403103870346977e-05,
      "loss": 0.0692,
      "step": 104225
    },
    {
      "epoch": 1.5353234856629505,
      "grad_norm": 1.400356650352478,
      "learning_rate": 1.5401467498379995e-05,
      "loss": 0.068,
      "step": 104250
    },
    {
      "epoch": 1.5356916687530375,
      "grad_norm": 1.6703741550445557,
      "learning_rate": 1.539983112641301e-05,
      "loss": 0.0732,
      "step": 104275
    },
    {
      "epoch": 1.5360598518431245,
      "grad_norm": 1.196559190750122,
      "learning_rate": 1.5398194754446023e-05,
      "loss": 0.0671,
      "step": 104300
    },
    {
      "epoch": 1.5364280349332116,
      "grad_norm": 2.0214624404907227,
      "learning_rate": 1.5396558382479038e-05,
      "loss": 0.0709,
      "step": 104325
    },
    {
      "epoch": 1.5367962180232986,
      "grad_norm": 1.3912500143051147,
      "learning_rate": 1.5394922010512056e-05,
      "loss": 0.071,
      "step": 104350
    },
    {
      "epoch": 1.5371644011133858,
      "grad_norm": 1.0920579433441162,
      "learning_rate": 1.539328563854507e-05,
      "loss": 0.0702,
      "step": 104375
    },
    {
      "epoch": 1.5375325842034728,
      "grad_norm": 1.3705909252166748,
      "learning_rate": 1.5391649266578085e-05,
      "loss": 0.0768,
      "step": 104400
    },
    {
      "epoch": 1.5379007672935598,
      "grad_norm": 1.253305435180664,
      "learning_rate": 1.5390012894611103e-05,
      "loss": 0.0703,
      "step": 104425
    },
    {
      "epoch": 1.5382689503836469,
      "grad_norm": 2.2879838943481445,
      "learning_rate": 1.5388376522644117e-05,
      "loss": 0.0715,
      "step": 104450
    },
    {
      "epoch": 1.5386371334737339,
      "grad_norm": 1.4344496726989746,
      "learning_rate": 1.538674015067713e-05,
      "loss": 0.0652,
      "step": 104475
    },
    {
      "epoch": 1.539005316563821,
      "grad_norm": 0.7615309357643127,
      "learning_rate": 1.5385103778710146e-05,
      "loss": 0.0656,
      "step": 104500
    },
    {
      "epoch": 1.539373499653908,
      "grad_norm": 1.0681302547454834,
      "learning_rate": 1.5383467406743164e-05,
      "loss": 0.0816,
      "step": 104525
    },
    {
      "epoch": 1.539741682743995,
      "grad_norm": 1.8376717567443848,
      "learning_rate": 1.5381831034776178e-05,
      "loss": 0.0732,
      "step": 104550
    },
    {
      "epoch": 1.540109865834082,
      "grad_norm": 1.7124261856079102,
      "learning_rate": 1.5380194662809193e-05,
      "loss": 0.0749,
      "step": 104575
    },
    {
      "epoch": 1.540478048924169,
      "grad_norm": 1.6960835456848145,
      "learning_rate": 1.537855829084221e-05,
      "loss": 0.0794,
      "step": 104600
    },
    {
      "epoch": 1.540846232014256,
      "grad_norm": 1.06011962890625,
      "learning_rate": 1.5376921918875225e-05,
      "loss": 0.066,
      "step": 104625
    },
    {
      "epoch": 1.541214415104343,
      "grad_norm": 1.4902781248092651,
      "learning_rate": 1.537528554690824e-05,
      "loss": 0.0737,
      "step": 104650
    },
    {
      "epoch": 1.54158259819443,
      "grad_norm": 1.2500584125518799,
      "learning_rate": 1.5373649174941257e-05,
      "loss": 0.072,
      "step": 104675
    },
    {
      "epoch": 1.541950781284517,
      "grad_norm": 0.9652636647224426,
      "learning_rate": 1.537201280297427e-05,
      "loss": 0.0735,
      "step": 104700
    },
    {
      "epoch": 1.5423189643746043,
      "grad_norm": 1.2820698022842407,
      "learning_rate": 1.5370376431007286e-05,
      "loss": 0.0658,
      "step": 104725
    },
    {
      "epoch": 1.5426871474646913,
      "grad_norm": 1.5940159559249878,
      "learning_rate": 1.53687400590403e-05,
      "loss": 0.0707,
      "step": 104750
    },
    {
      "epoch": 1.5430553305547783,
      "grad_norm": 0.8580747246742249,
      "learning_rate": 1.536710368707332e-05,
      "loss": 0.0647,
      "step": 104775
    },
    {
      "epoch": 1.5434235136448653,
      "grad_norm": 1.8936817646026611,
      "learning_rate": 1.5365467315106333e-05,
      "loss": 0.0777,
      "step": 104800
    },
    {
      "epoch": 1.5437916967349523,
      "grad_norm": 1.3605011701583862,
      "learning_rate": 1.5363830943139347e-05,
      "loss": 0.064,
      "step": 104825
    },
    {
      "epoch": 1.5441598798250395,
      "grad_norm": 1.314716100692749,
      "learning_rate": 1.5362194571172365e-05,
      "loss": 0.0707,
      "step": 104850
    },
    {
      "epoch": 1.5445280629151266,
      "grad_norm": 1.5477386713027954,
      "learning_rate": 1.536055819920538e-05,
      "loss": 0.0759,
      "step": 104875
    },
    {
      "epoch": 1.5448962460052136,
      "grad_norm": 2.173539876937866,
      "learning_rate": 1.5358921827238394e-05,
      "loss": 0.0776,
      "step": 104900
    },
    {
      "epoch": 1.5452644290953006,
      "grad_norm": 1.4477789402008057,
      "learning_rate": 1.535728545527141e-05,
      "loss": 0.0775,
      "step": 104925
    },
    {
      "epoch": 1.5456326121853876,
      "grad_norm": 1.4179946184158325,
      "learning_rate": 1.5355649083304423e-05,
      "loss": 0.0763,
      "step": 104950
    },
    {
      "epoch": 1.5460007952754746,
      "grad_norm": 1.4506452083587646,
      "learning_rate": 1.535401271133744e-05,
      "loss": 0.0693,
      "step": 104975
    },
    {
      "epoch": 1.5463689783655616,
      "grad_norm": 1.3918483257293701,
      "learning_rate": 1.5352376339370455e-05,
      "loss": 0.0777,
      "step": 105000
    },
    {
      "epoch": 1.5467371614556487,
      "grad_norm": 1.166577696800232,
      "learning_rate": 1.5350739967403473e-05,
      "loss": 0.0692,
      "step": 105025
    },
    {
      "epoch": 1.5471053445457357,
      "grad_norm": 1.01295804977417,
      "learning_rate": 1.5349103595436488e-05,
      "loss": 0.0706,
      "step": 105050
    },
    {
      "epoch": 1.5474735276358227,
      "grad_norm": 1.2988080978393555,
      "learning_rate": 1.5347467223469502e-05,
      "loss": 0.0691,
      "step": 105075
    },
    {
      "epoch": 1.5478417107259097,
      "grad_norm": 1.5271539688110352,
      "learning_rate": 1.534583085150252e-05,
      "loss": 0.0727,
      "step": 105100
    },
    {
      "epoch": 1.5482098938159967,
      "grad_norm": 1.363669753074646,
      "learning_rate": 1.534419447953553e-05,
      "loss": 0.0677,
      "step": 105125
    },
    {
      "epoch": 1.5485780769060837,
      "grad_norm": 1.1423157453536987,
      "learning_rate": 1.534255810756855e-05,
      "loss": 0.0695,
      "step": 105150
    },
    {
      "epoch": 1.5489462599961707,
      "grad_norm": 1.3760061264038086,
      "learning_rate": 1.5340921735601563e-05,
      "loss": 0.0717,
      "step": 105175
    },
    {
      "epoch": 1.549314443086258,
      "grad_norm": 1.4595600366592407,
      "learning_rate": 1.533928536363458e-05,
      "loss": 0.0696,
      "step": 105200
    },
    {
      "epoch": 1.549682626176345,
      "grad_norm": 1.7446681261062622,
      "learning_rate": 1.5337648991667596e-05,
      "loss": 0.0736,
      "step": 105225
    },
    {
      "epoch": 1.550050809266432,
      "grad_norm": 1.1961472034454346,
      "learning_rate": 1.533601261970061e-05,
      "loss": 0.0731,
      "step": 105250
    },
    {
      "epoch": 1.550418992356519,
      "grad_norm": 1.3641124963760376,
      "learning_rate": 1.5334376247733628e-05,
      "loss": 0.0709,
      "step": 105275
    },
    {
      "epoch": 1.550787175446606,
      "grad_norm": 1.3327598571777344,
      "learning_rate": 1.5332739875766642e-05,
      "loss": 0.0733,
      "step": 105300
    },
    {
      "epoch": 1.5511553585366933,
      "grad_norm": 1.0816978216171265,
      "learning_rate": 1.5331103503799657e-05,
      "loss": 0.0714,
      "step": 105325
    },
    {
      "epoch": 1.5515235416267803,
      "grad_norm": 1.0243499279022217,
      "learning_rate": 1.532946713183267e-05,
      "loss": 0.0728,
      "step": 105350
    },
    {
      "epoch": 1.5518917247168673,
      "grad_norm": 1.4931097030639648,
      "learning_rate": 1.5327830759865686e-05,
      "loss": 0.0734,
      "step": 105375
    },
    {
      "epoch": 1.5522599078069543,
      "grad_norm": 0.8803391456604004,
      "learning_rate": 1.5326194387898704e-05,
      "loss": 0.0774,
      "step": 105400
    },
    {
      "epoch": 1.5526280908970413,
      "grad_norm": 1.4650176763534546,
      "learning_rate": 1.5324558015931718e-05,
      "loss": 0.0758,
      "step": 105425
    },
    {
      "epoch": 1.5529962739871284,
      "grad_norm": 1.7435822486877441,
      "learning_rate": 1.5322921643964736e-05,
      "loss": 0.07,
      "step": 105450
    },
    {
      "epoch": 1.5533644570772154,
      "grad_norm": 1.2899059057235718,
      "learning_rate": 1.532128527199775e-05,
      "loss": 0.0662,
      "step": 105475
    },
    {
      "epoch": 1.5537326401673024,
      "grad_norm": 0.8504881858825684,
      "learning_rate": 1.5319648900030765e-05,
      "loss": 0.0703,
      "step": 105500
    },
    {
      "epoch": 1.5541008232573894,
      "grad_norm": 1.1455090045928955,
      "learning_rate": 1.5318012528063783e-05,
      "loss": 0.0745,
      "step": 105525
    },
    {
      "epoch": 1.5544690063474764,
      "grad_norm": 1.7868210077285767,
      "learning_rate": 1.5316376156096794e-05,
      "loss": 0.0792,
      "step": 105550
    },
    {
      "epoch": 1.5548371894375634,
      "grad_norm": 1.8337889909744263,
      "learning_rate": 1.531473978412981e-05,
      "loss": 0.0726,
      "step": 105575
    },
    {
      "epoch": 1.5552053725276505,
      "grad_norm": 1.2802447080612183,
      "learning_rate": 1.5313103412162826e-05,
      "loss": 0.0664,
      "step": 105600
    },
    {
      "epoch": 1.5555735556177375,
      "grad_norm": Infinity,
      "learning_rate": 1.531153249507452e-05,
      "loss": 0.0812,
      "step": 105625
    },
    {
      "epoch": 1.5559417387078245,
      "grad_norm": 1.2857905626296997,
      "learning_rate": 1.5309896123107538e-05,
      "loss": 0.0747,
      "step": 105650
    },
    {
      "epoch": 1.5563099217979115,
      "grad_norm": 1.1183959245681763,
      "learning_rate": 1.5308259751140552e-05,
      "loss": 0.0756,
      "step": 105675
    },
    {
      "epoch": 1.5566781048879987,
      "grad_norm": 1.501734972000122,
      "learning_rate": 1.530662337917357e-05,
      "loss": 0.0772,
      "step": 105700
    },
    {
      "epoch": 1.5570462879780858,
      "grad_norm": 1.4179561138153076,
      "learning_rate": 1.5304987007206585e-05,
      "loss": 0.0697,
      "step": 105725
    },
    {
      "epoch": 1.5574144710681728,
      "grad_norm": 1.4853519201278687,
      "learning_rate": 1.53033506352396e-05,
      "loss": 0.0777,
      "step": 105750
    },
    {
      "epoch": 1.5577826541582598,
      "grad_norm": 1.5254907608032227,
      "learning_rate": 1.5301714263272613e-05,
      "loss": 0.0685,
      "step": 105775
    },
    {
      "epoch": 1.558150837248347,
      "grad_norm": 1.8476866483688354,
      "learning_rate": 1.5300077891305628e-05,
      "loss": 0.0726,
      "step": 105800
    },
    {
      "epoch": 1.558519020338434,
      "grad_norm": 1.456032395362854,
      "learning_rate": 1.5298441519338646e-05,
      "loss": 0.0719,
      "step": 105825
    },
    {
      "epoch": 1.558887203428521,
      "grad_norm": 1.4664596319198608,
      "learning_rate": 1.529680514737166e-05,
      "loss": 0.0735,
      "step": 105850
    },
    {
      "epoch": 1.559255386518608,
      "grad_norm": 1.293497085571289,
      "learning_rate": 1.5295168775404675e-05,
      "loss": 0.0677,
      "step": 105875
    },
    {
      "epoch": 1.559623569608695,
      "grad_norm": 1.593989372253418,
      "learning_rate": 1.5293532403437693e-05,
      "loss": 0.0673,
      "step": 105900
    },
    {
      "epoch": 1.559991752698782,
      "grad_norm": 1.4005393981933594,
      "learning_rate": 1.5291896031470707e-05,
      "loss": 0.0742,
      "step": 105925
    },
    {
      "epoch": 1.5603599357888691,
      "grad_norm": 1.608427882194519,
      "learning_rate": 1.5290259659503725e-05,
      "loss": 0.071,
      "step": 105950
    },
    {
      "epoch": 1.5607281188789561,
      "grad_norm": 1.4881457090377808,
      "learning_rate": 1.5288623287536736e-05,
      "loss": 0.0714,
      "step": 105975
    },
    {
      "epoch": 1.5610963019690431,
      "grad_norm": 1.2351033687591553,
      "learning_rate": 1.5286986915569754e-05,
      "loss": 0.0616,
      "step": 106000
    },
    {
      "epoch": 1.5614644850591302,
      "grad_norm": 1.4147685766220093,
      "learning_rate": 1.5285350543602768e-05,
      "loss": 0.0719,
      "step": 106025
    },
    {
      "epoch": 1.5618326681492172,
      "grad_norm": 1.6583187580108643,
      "learning_rate": 1.5283714171635783e-05,
      "loss": 0.0713,
      "step": 106050
    },
    {
      "epoch": 1.5622008512393042,
      "grad_norm": 1.3932957649230957,
      "learning_rate": 1.52820777996688e-05,
      "loss": 0.0751,
      "step": 106075
    },
    {
      "epoch": 1.5625690343293912,
      "grad_norm": 1.1920204162597656,
      "learning_rate": 1.5280441427701815e-05,
      "loss": 0.0713,
      "step": 106100
    },
    {
      "epoch": 1.5629372174194782,
      "grad_norm": 1.090222716331482,
      "learning_rate": 1.527880505573483e-05,
      "loss": 0.0682,
      "step": 106125
    },
    {
      "epoch": 1.5633054005095652,
      "grad_norm": 1.216540813446045,
      "learning_rate": 1.5277168683767847e-05,
      "loss": 0.071,
      "step": 106150
    },
    {
      "epoch": 1.5636735835996525,
      "grad_norm": 1.424748182296753,
      "learning_rate": 1.5275532311800862e-05,
      "loss": 0.0681,
      "step": 106175
    },
    {
      "epoch": 1.5640417666897395,
      "grad_norm": 1.4271202087402344,
      "learning_rate": 1.5273895939833876e-05,
      "loss": 0.075,
      "step": 106200
    },
    {
      "epoch": 1.5644099497798265,
      "grad_norm": 1.2879153490066528,
      "learning_rate": 1.527225956786689e-05,
      "loss": 0.0643,
      "step": 106225
    },
    {
      "epoch": 1.5647781328699135,
      "grad_norm": 1.4442555904388428,
      "learning_rate": 1.527062319589991e-05,
      "loss": 0.0718,
      "step": 106250
    },
    {
      "epoch": 1.5651463159600005,
      "grad_norm": 0.9468669891357422,
      "learning_rate": 1.5268986823932923e-05,
      "loss": 0.0675,
      "step": 106275
    },
    {
      "epoch": 1.5655144990500878,
      "grad_norm": 1.0760447978973389,
      "learning_rate": 1.5267350451965937e-05,
      "loss": 0.0694,
      "step": 106300
    },
    {
      "epoch": 1.5658826821401748,
      "grad_norm": 1.3462986946105957,
      "learning_rate": 1.5265714079998955e-05,
      "loss": 0.0757,
      "step": 106325
    },
    {
      "epoch": 1.5662508652302618,
      "grad_norm": 1.2779897451400757,
      "learning_rate": 1.526407770803197e-05,
      "loss": 0.0668,
      "step": 106350
    },
    {
      "epoch": 1.5666190483203488,
      "grad_norm": 1.6397979259490967,
      "learning_rate": 1.5262441336064988e-05,
      "loss": 0.0739,
      "step": 106375
    },
    {
      "epoch": 1.5669872314104358,
      "grad_norm": 1.062831163406372,
      "learning_rate": 1.5260804964098e-05,
      "loss": 0.0707,
      "step": 106400
    },
    {
      "epoch": 1.5673554145005228,
      "grad_norm": 1.33720064163208,
      "learning_rate": 1.5259168592131016e-05,
      "loss": 0.0698,
      "step": 106425
    },
    {
      "epoch": 1.5677235975906099,
      "grad_norm": 1.417860984802246,
      "learning_rate": 1.5257532220164031e-05,
      "loss": 0.0711,
      "step": 106450
    },
    {
      "epoch": 1.5680917806806969,
      "grad_norm": 1.456868052482605,
      "learning_rate": 1.5255895848197047e-05,
      "loss": 0.0635,
      "step": 106475
    },
    {
      "epoch": 1.568459963770784,
      "grad_norm": 1.456259846687317,
      "learning_rate": 1.5254259476230062e-05,
      "loss": 0.0686,
      "step": 106500
    },
    {
      "epoch": 1.568828146860871,
      "grad_norm": 1.5034135580062866,
      "learning_rate": 1.5252623104263078e-05,
      "loss": 0.0721,
      "step": 106525
    },
    {
      "epoch": 1.569196329950958,
      "grad_norm": 1.4312362670898438,
      "learning_rate": 1.5250986732296094e-05,
      "loss": 0.0765,
      "step": 106550
    },
    {
      "epoch": 1.569564513041045,
      "grad_norm": 1.6333671808242798,
      "learning_rate": 1.524935036032911e-05,
      "loss": 0.0645,
      "step": 106575
    },
    {
      "epoch": 1.569932696131132,
      "grad_norm": 1.6294881105422974,
      "learning_rate": 1.5247713988362123e-05,
      "loss": 0.0775,
      "step": 106600
    },
    {
      "epoch": 1.570300879221219,
      "grad_norm": 1.624904751777649,
      "learning_rate": 1.5246077616395139e-05,
      "loss": 0.0659,
      "step": 106625
    },
    {
      "epoch": 1.5706690623113062,
      "grad_norm": 1.3214925527572632,
      "learning_rate": 1.5244441244428155e-05,
      "loss": 0.0736,
      "step": 106650
    },
    {
      "epoch": 1.5710372454013932,
      "grad_norm": 1.553481101989746,
      "learning_rate": 1.524280487246117e-05,
      "loss": 0.0785,
      "step": 106675
    },
    {
      "epoch": 1.5714054284914802,
      "grad_norm": 1.1896533966064453,
      "learning_rate": 1.5241168500494186e-05,
      "loss": 0.0732,
      "step": 106700
    },
    {
      "epoch": 1.5717736115815673,
      "grad_norm": 1.3297849893569946,
      "learning_rate": 1.5239532128527202e-05,
      "loss": 0.071,
      "step": 106725
    },
    {
      "epoch": 1.5721417946716543,
      "grad_norm": 1.256745457649231,
      "learning_rate": 1.5237895756560216e-05,
      "loss": 0.0767,
      "step": 106750
    },
    {
      "epoch": 1.5725099777617415,
      "grad_norm": 1.4505422115325928,
      "learning_rate": 1.5236259384593232e-05,
      "loss": 0.0707,
      "step": 106775
    },
    {
      "epoch": 1.5728781608518285,
      "grad_norm": 1.368282437324524,
      "learning_rate": 1.5234623012626249e-05,
      "loss": 0.0762,
      "step": 106800
    },
    {
      "epoch": 1.5732463439419155,
      "grad_norm": 1.8916006088256836,
      "learning_rate": 1.5232986640659261e-05,
      "loss": 0.0712,
      "step": 106825
    },
    {
      "epoch": 1.5736145270320026,
      "grad_norm": 1.4318642616271973,
      "learning_rate": 1.5231350268692277e-05,
      "loss": 0.0712,
      "step": 106850
    },
    {
      "epoch": 1.5739827101220896,
      "grad_norm": 1.4258930683135986,
      "learning_rate": 1.5229713896725294e-05,
      "loss": 0.0779,
      "step": 106875
    },
    {
      "epoch": 1.5743508932121766,
      "grad_norm": 1.4721462726593018,
      "learning_rate": 1.522807752475831e-05,
      "loss": 0.0731,
      "step": 106900
    },
    {
      "epoch": 1.5747190763022636,
      "grad_norm": 1.6253427267074585,
      "learning_rate": 1.5226441152791324e-05,
      "loss": 0.0749,
      "step": 106925
    },
    {
      "epoch": 1.5750872593923506,
      "grad_norm": 1.1349114179611206,
      "learning_rate": 1.522480478082434e-05,
      "loss": 0.0731,
      "step": 106950
    },
    {
      "epoch": 1.5754554424824376,
      "grad_norm": 1.2018711566925049,
      "learning_rate": 1.5223168408857357e-05,
      "loss": 0.0724,
      "step": 106975
    },
    {
      "epoch": 1.5758236255725246,
      "grad_norm": 1.2875616550445557,
      "learning_rate": 1.5221532036890371e-05,
      "loss": 0.0691,
      "step": 107000
    },
    {
      "epoch": 1.5761918086626117,
      "grad_norm": 1.9457770586013794,
      "learning_rate": 1.5219895664923385e-05,
      "loss": 0.0686,
      "step": 107025
    },
    {
      "epoch": 1.5765599917526987,
      "grad_norm": 1.1090826988220215,
      "learning_rate": 1.5218259292956402e-05,
      "loss": 0.075,
      "step": 107050
    },
    {
      "epoch": 1.5769281748427857,
      "grad_norm": 1.6381428241729736,
      "learning_rate": 1.5216622920989416e-05,
      "loss": 0.0748,
      "step": 107075
    },
    {
      "epoch": 1.5772963579328727,
      "grad_norm": 1.4712249040603638,
      "learning_rate": 1.5214986549022432e-05,
      "loss": 0.073,
      "step": 107100
    },
    {
      "epoch": 1.57766454102296,
      "grad_norm": 1.4448765516281128,
      "learning_rate": 1.5213350177055448e-05,
      "loss": 0.0722,
      "step": 107125
    },
    {
      "epoch": 1.578032724113047,
      "grad_norm": 1.2978557348251343,
      "learning_rate": 1.5211713805088465e-05,
      "loss": 0.0692,
      "step": 107150
    },
    {
      "epoch": 1.578400907203134,
      "grad_norm": 1.130519986152649,
      "learning_rate": 1.5210077433121479e-05,
      "loss": 0.0659,
      "step": 107175
    },
    {
      "epoch": 1.578769090293221,
      "grad_norm": 1.5109269618988037,
      "learning_rate": 1.5208441061154495e-05,
      "loss": 0.0761,
      "step": 107200
    },
    {
      "epoch": 1.579137273383308,
      "grad_norm": 1.2745883464813232,
      "learning_rate": 1.5206804689187508e-05,
      "loss": 0.0742,
      "step": 107225
    },
    {
      "epoch": 1.5795054564733952,
      "grad_norm": 1.2200329303741455,
      "learning_rate": 1.5205168317220524e-05,
      "loss": 0.0658,
      "step": 107250
    },
    {
      "epoch": 1.5798736395634823,
      "grad_norm": 1.1669632196426392,
      "learning_rate": 1.520353194525354e-05,
      "loss": 0.0667,
      "step": 107275
    },
    {
      "epoch": 1.5802418226535693,
      "grad_norm": 1.8420226573944092,
      "learning_rate": 1.5201895573286556e-05,
      "loss": 0.0744,
      "step": 107300
    },
    {
      "epoch": 1.5806100057436563,
      "grad_norm": 1.2124704122543335,
      "learning_rate": 1.520025920131957e-05,
      "loss": 0.0745,
      "step": 107325
    },
    {
      "epoch": 1.5809781888337433,
      "grad_norm": 1.3906047344207764,
      "learning_rate": 1.5198622829352587e-05,
      "loss": 0.0681,
      "step": 107350
    },
    {
      "epoch": 1.5813463719238303,
      "grad_norm": 1.08652663230896,
      "learning_rate": 1.5196986457385603e-05,
      "loss": 0.0734,
      "step": 107375
    },
    {
      "epoch": 1.5817145550139173,
      "grad_norm": 1.7926405668258667,
      "learning_rate": 1.519535008541862e-05,
      "loss": 0.073,
      "step": 107400
    },
    {
      "epoch": 1.5820827381040043,
      "grad_norm": 1.416136622428894,
      "learning_rate": 1.5193713713451634e-05,
      "loss": 0.0782,
      "step": 107425
    },
    {
      "epoch": 1.5824509211940914,
      "grad_norm": 1.4756971597671509,
      "learning_rate": 1.5192077341484648e-05,
      "loss": 0.0688,
      "step": 107450
    },
    {
      "epoch": 1.5828191042841784,
      "grad_norm": 1.0753921270370483,
      "learning_rate": 1.5190440969517664e-05,
      "loss": 0.0726,
      "step": 107475
    },
    {
      "epoch": 1.5831872873742654,
      "grad_norm": 1.4241995811462402,
      "learning_rate": 1.5188804597550679e-05,
      "loss": 0.0726,
      "step": 107500
    },
    {
      "epoch": 1.5835554704643524,
      "grad_norm": 1.2430777549743652,
      "learning_rate": 1.5187168225583695e-05,
      "loss": 0.0675,
      "step": 107525
    },
    {
      "epoch": 1.5839236535544394,
      "grad_norm": 1.2054349184036255,
      "learning_rate": 1.5185531853616711e-05,
      "loss": 0.0726,
      "step": 107550
    },
    {
      "epoch": 1.5842918366445264,
      "grad_norm": 1.1671655178070068,
      "learning_rate": 1.5183895481649726e-05,
      "loss": 0.0686,
      "step": 107575
    },
    {
      "epoch": 1.5846600197346137,
      "grad_norm": 1.4217971563339233,
      "learning_rate": 1.5182259109682742e-05,
      "loss": 0.0754,
      "step": 107600
    },
    {
      "epoch": 1.5850282028247007,
      "grad_norm": 1.5682121515274048,
      "learning_rate": 1.5180622737715758e-05,
      "loss": 0.0747,
      "step": 107625
    },
    {
      "epoch": 1.5853963859147877,
      "grad_norm": 2.056432008743286,
      "learning_rate": 1.517898636574877e-05,
      "loss": 0.0741,
      "step": 107650
    },
    {
      "epoch": 1.5857645690048747,
      "grad_norm": 1.61311674118042,
      "learning_rate": 1.5177415448660466e-05,
      "loss": 0.072,
      "step": 107675
    },
    {
      "epoch": 1.5861327520949617,
      "grad_norm": 1.5249961614608765,
      "learning_rate": 1.5175779076693482e-05,
      "loss": 0.0691,
      "step": 107700
    },
    {
      "epoch": 1.586500935185049,
      "grad_norm": 1.4391385316848755,
      "learning_rate": 1.5174142704726498e-05,
      "loss": 0.0726,
      "step": 107725
    },
    {
      "epoch": 1.586869118275136,
      "grad_norm": 1.7815829515457153,
      "learning_rate": 1.5172506332759513e-05,
      "loss": 0.0776,
      "step": 107750
    },
    {
      "epoch": 1.587237301365223,
      "grad_norm": 1.675018548965454,
      "learning_rate": 1.5170869960792529e-05,
      "loss": 0.0692,
      "step": 107775
    },
    {
      "epoch": 1.58760548445531,
      "grad_norm": 1.5182093381881714,
      "learning_rate": 1.5169233588825545e-05,
      "loss": 0.0686,
      "step": 107800
    },
    {
      "epoch": 1.587973667545397,
      "grad_norm": 1.2581740617752075,
      "learning_rate": 1.5167597216858561e-05,
      "loss": 0.0721,
      "step": 107825
    },
    {
      "epoch": 1.588341850635484,
      "grad_norm": 1.1967384815216064,
      "learning_rate": 1.5165960844891576e-05,
      "loss": 0.0658,
      "step": 107850
    },
    {
      "epoch": 1.588710033725571,
      "grad_norm": 1.770078420639038,
      "learning_rate": 1.516432447292459e-05,
      "loss": 0.0715,
      "step": 107875
    },
    {
      "epoch": 1.589078216815658,
      "grad_norm": 1.4590777158737183,
      "learning_rate": 1.5162688100957605e-05,
      "loss": 0.0668,
      "step": 107900
    },
    {
      "epoch": 1.589446399905745,
      "grad_norm": 1.1806178092956543,
      "learning_rate": 1.5161051728990621e-05,
      "loss": 0.0643,
      "step": 107925
    },
    {
      "epoch": 1.5898145829958321,
      "grad_norm": 1.8124110698699951,
      "learning_rate": 1.5159415357023637e-05,
      "loss": 0.0696,
      "step": 107950
    },
    {
      "epoch": 1.5901827660859191,
      "grad_norm": 1.9094833135604858,
      "learning_rate": 1.5157778985056653e-05,
      "loss": 0.0713,
      "step": 107975
    },
    {
      "epoch": 1.5905509491760061,
      "grad_norm": 1.5736355781555176,
      "learning_rate": 1.5156142613089668e-05,
      "loss": 0.0767,
      "step": 108000
    },
    {
      "epoch": 1.5909191322660932,
      "grad_norm": 1.275492548942566,
      "learning_rate": 1.5154506241122684e-05,
      "loss": 0.0723,
      "step": 108025
    },
    {
      "epoch": 1.5912873153561802,
      "grad_norm": 1.7701120376586914,
      "learning_rate": 1.51528698691557e-05,
      "loss": 0.072,
      "step": 108050
    },
    {
      "epoch": 1.5916554984462674,
      "grad_norm": 1.4056119918823242,
      "learning_rate": 1.5151233497188713e-05,
      "loss": 0.0681,
      "step": 108075
    },
    {
      "epoch": 1.5920236815363544,
      "grad_norm": 1.3238954544067383,
      "learning_rate": 1.5149597125221729e-05,
      "loss": 0.0624,
      "step": 108100
    },
    {
      "epoch": 1.5923918646264414,
      "grad_norm": 1.4717309474945068,
      "learning_rate": 1.5147960753254745e-05,
      "loss": 0.0719,
      "step": 108125
    },
    {
      "epoch": 1.5927600477165285,
      "grad_norm": 1.234840989112854,
      "learning_rate": 1.514632438128776e-05,
      "loss": 0.0759,
      "step": 108150
    },
    {
      "epoch": 1.5931282308066155,
      "grad_norm": 1.1387003660202026,
      "learning_rate": 1.5144688009320776e-05,
      "loss": 0.0654,
      "step": 108175
    },
    {
      "epoch": 1.5934964138967027,
      "grad_norm": 0.9490651488304138,
      "learning_rate": 1.5143051637353792e-05,
      "loss": 0.0694,
      "step": 108200
    },
    {
      "epoch": 1.5938645969867897,
      "grad_norm": 1.4842017889022827,
      "learning_rate": 1.5141415265386808e-05,
      "loss": 0.0782,
      "step": 108225
    },
    {
      "epoch": 1.5942327800768767,
      "grad_norm": 1.481844186782837,
      "learning_rate": 1.5139778893419822e-05,
      "loss": 0.071,
      "step": 108250
    },
    {
      "epoch": 1.5946009631669638,
      "grad_norm": 1.2599743604660034,
      "learning_rate": 1.5138142521452839e-05,
      "loss": 0.066,
      "step": 108275
    },
    {
      "epoch": 1.5949691462570508,
      "grad_norm": 1.3949706554412842,
      "learning_rate": 1.5136506149485853e-05,
      "loss": 0.069,
      "step": 108300
    },
    {
      "epoch": 1.5953373293471378,
      "grad_norm": 1.1331928968429565,
      "learning_rate": 1.5134869777518867e-05,
      "loss": 0.0717,
      "step": 108325
    },
    {
      "epoch": 1.5957055124372248,
      "grad_norm": 1.2667357921600342,
      "learning_rate": 1.5133233405551884e-05,
      "loss": 0.0714,
      "step": 108350
    },
    {
      "epoch": 1.5960736955273118,
      "grad_norm": 1.2307201623916626,
      "learning_rate": 1.51315970335849e-05,
      "loss": 0.0699,
      "step": 108375
    },
    {
      "epoch": 1.5964418786173988,
      "grad_norm": 1.3318933248519897,
      "learning_rate": 1.5129960661617916e-05,
      "loss": 0.0753,
      "step": 108400
    },
    {
      "epoch": 1.5968100617074859,
      "grad_norm": 1.3171318769454956,
      "learning_rate": 1.512832428965093e-05,
      "loss": 0.069,
      "step": 108425
    },
    {
      "epoch": 1.5971782447975729,
      "grad_norm": 1.5384852886199951,
      "learning_rate": 1.5126687917683947e-05,
      "loss": 0.0714,
      "step": 108450
    },
    {
      "epoch": 1.5975464278876599,
      "grad_norm": 1.031415581703186,
      "learning_rate": 1.5125051545716963e-05,
      "loss": 0.0671,
      "step": 108475
    },
    {
      "epoch": 1.597914610977747,
      "grad_norm": 1.7239437103271484,
      "learning_rate": 1.5123415173749975e-05,
      "loss": 0.072,
      "step": 108500
    },
    {
      "epoch": 1.598282794067834,
      "grad_norm": 1.6831711530685425,
      "learning_rate": 1.5121778801782992e-05,
      "loss": 0.0729,
      "step": 108525
    },
    {
      "epoch": 1.598650977157921,
      "grad_norm": 1.3334431648254395,
      "learning_rate": 1.5120142429816008e-05,
      "loss": 0.0686,
      "step": 108550
    },
    {
      "epoch": 1.5990191602480082,
      "grad_norm": 1.326115608215332,
      "learning_rate": 1.5118506057849022e-05,
      "loss": 0.0707,
      "step": 108575
    },
    {
      "epoch": 1.5993873433380952,
      "grad_norm": 1.1547571420669556,
      "learning_rate": 1.5116869685882038e-05,
      "loss": 0.0718,
      "step": 108600
    },
    {
      "epoch": 1.5997555264281822,
      "grad_norm": 1.4270943403244019,
      "learning_rate": 1.5115233313915055e-05,
      "loss": 0.0731,
      "step": 108625
    },
    {
      "epoch": 1.6001237095182692,
      "grad_norm": 1.5217137336730957,
      "learning_rate": 1.511359694194807e-05,
      "loss": 0.0717,
      "step": 108650
    },
    {
      "epoch": 1.6004918926083564,
      "grad_norm": 1.583696961402893,
      "learning_rate": 1.5111960569981085e-05,
      "loss": 0.0734,
      "step": 108675
    },
    {
      "epoch": 1.6008600756984435,
      "grad_norm": 1.1852489709854126,
      "learning_rate": 1.51103241980141e-05,
      "loss": 0.0682,
      "step": 108700
    },
    {
      "epoch": 1.6012282587885305,
      "grad_norm": 1.4153369665145874,
      "learning_rate": 1.5108687826047114e-05,
      "loss": 0.0791,
      "step": 108725
    },
    {
      "epoch": 1.6015964418786175,
      "grad_norm": 1.500317096710205,
      "learning_rate": 1.510705145408013e-05,
      "loss": 0.0775,
      "step": 108750
    },
    {
      "epoch": 1.6019646249687045,
      "grad_norm": 1.048874855041504,
      "learning_rate": 1.5105415082113146e-05,
      "loss": 0.0667,
      "step": 108775
    },
    {
      "epoch": 1.6023328080587915,
      "grad_norm": 1.6746413707733154,
      "learning_rate": 1.5103778710146163e-05,
      "loss": 0.0761,
      "step": 108800
    },
    {
      "epoch": 1.6027009911488785,
      "grad_norm": 1.6953356266021729,
      "learning_rate": 1.5102142338179177e-05,
      "loss": 0.0697,
      "step": 108825
    },
    {
      "epoch": 1.6030691742389656,
      "grad_norm": 1.3492467403411865,
      "learning_rate": 1.5100505966212193e-05,
      "loss": 0.0758,
      "step": 108850
    },
    {
      "epoch": 1.6034373573290526,
      "grad_norm": 1.6429636478424072,
      "learning_rate": 1.509886959424521e-05,
      "loss": 0.0749,
      "step": 108875
    },
    {
      "epoch": 1.6038055404191396,
      "grad_norm": 1.4930524826049805,
      "learning_rate": 1.5097233222278225e-05,
      "loss": 0.0759,
      "step": 108900
    },
    {
      "epoch": 1.6041737235092266,
      "grad_norm": 1.3129405975341797,
      "learning_rate": 1.5095596850311238e-05,
      "loss": 0.0707,
      "step": 108925
    },
    {
      "epoch": 1.6045419065993136,
      "grad_norm": 1.0810818672180176,
      "learning_rate": 1.5093960478344254e-05,
      "loss": 0.0775,
      "step": 108950
    },
    {
      "epoch": 1.6049100896894006,
      "grad_norm": 1.2722159624099731,
      "learning_rate": 1.5092324106377269e-05,
      "loss": 0.072,
      "step": 108975
    },
    {
      "epoch": 1.6052782727794876,
      "grad_norm": 1.2757669687271118,
      "learning_rate": 1.5090687734410285e-05,
      "loss": 0.0721,
      "step": 109000
    },
    {
      "epoch": 1.6056464558695747,
      "grad_norm": 1.9006820917129517,
      "learning_rate": 1.5089051362443301e-05,
      "loss": 0.0761,
      "step": 109025
    },
    {
      "epoch": 1.606014638959662,
      "grad_norm": 1.2178272008895874,
      "learning_rate": 1.5087414990476317e-05,
      "loss": 0.0693,
      "step": 109050
    },
    {
      "epoch": 1.606382822049749,
      "grad_norm": 1.3632571697235107,
      "learning_rate": 1.5085778618509332e-05,
      "loss": 0.0716,
      "step": 109075
    },
    {
      "epoch": 1.606751005139836,
      "grad_norm": 2.103853702545166,
      "learning_rate": 1.5084142246542348e-05,
      "loss": 0.081,
      "step": 109100
    },
    {
      "epoch": 1.607119188229923,
      "grad_norm": 1.2864320278167725,
      "learning_rate": 1.5082505874575362e-05,
      "loss": 0.0701,
      "step": 109125
    },
    {
      "epoch": 1.60748737132001,
      "grad_norm": 1.1540031433105469,
      "learning_rate": 1.5080869502608377e-05,
      "loss": 0.0706,
      "step": 109150
    },
    {
      "epoch": 1.6078555544100972,
      "grad_norm": 1.7386056184768677,
      "learning_rate": 1.5079233130641393e-05,
      "loss": 0.0746,
      "step": 109175
    },
    {
      "epoch": 1.6082237375001842,
      "grad_norm": 1.4164113998413086,
      "learning_rate": 1.5077596758674409e-05,
      "loss": 0.0678,
      "step": 109200
    },
    {
      "epoch": 1.6085919205902712,
      "grad_norm": 1.5290884971618652,
      "learning_rate": 1.5075960386707425e-05,
      "loss": 0.0725,
      "step": 109225
    },
    {
      "epoch": 1.6089601036803582,
      "grad_norm": 1.5598176717758179,
      "learning_rate": 1.507432401474044e-05,
      "loss": 0.0781,
      "step": 109250
    },
    {
      "epoch": 1.6093282867704453,
      "grad_norm": 1.4344979524612427,
      "learning_rate": 1.5072687642773456e-05,
      "loss": 0.0709,
      "step": 109275
    },
    {
      "epoch": 1.6096964698605323,
      "grad_norm": 0.9226469397544861,
      "learning_rate": 1.5071051270806472e-05,
      "loss": 0.0766,
      "step": 109300
    },
    {
      "epoch": 1.6100646529506193,
      "grad_norm": 1.2100237607955933,
      "learning_rate": 1.5069414898839486e-05,
      "loss": 0.0685,
      "step": 109325
    },
    {
      "epoch": 1.6104328360407063,
      "grad_norm": 1.4307045936584473,
      "learning_rate": 1.5067778526872501e-05,
      "loss": 0.0669,
      "step": 109350
    },
    {
      "epoch": 1.6108010191307933,
      "grad_norm": 1.2018272876739502,
      "learning_rate": 1.5066142154905517e-05,
      "loss": 0.0692,
      "step": 109375
    },
    {
      "epoch": 1.6111692022208803,
      "grad_norm": 1.6964788436889648,
      "learning_rate": 1.5064505782938532e-05,
      "loss": 0.077,
      "step": 109400
    },
    {
      "epoch": 1.6115373853109674,
      "grad_norm": 1.5950158834457397,
      "learning_rate": 1.5062869410971548e-05,
      "loss": 0.0775,
      "step": 109425
    },
    {
      "epoch": 1.6119055684010544,
      "grad_norm": 1.8836623430252075,
      "learning_rate": 1.5061233039004564e-05,
      "loss": 0.0745,
      "step": 109450
    },
    {
      "epoch": 1.6122737514911414,
      "grad_norm": 1.6316585540771484,
      "learning_rate": 1.505959666703758e-05,
      "loss": 0.0673,
      "step": 109475
    },
    {
      "epoch": 1.6126419345812284,
      "grad_norm": 1.4389818906784058,
      "learning_rate": 1.5057960295070594e-05,
      "loss": 0.0716,
      "step": 109500
    },
    {
      "epoch": 1.6130101176713156,
      "grad_norm": 1.037874698638916,
      "learning_rate": 1.505632392310361e-05,
      "loss": 0.0654,
      "step": 109525
    },
    {
      "epoch": 1.6133783007614027,
      "grad_norm": 1.2963496446609497,
      "learning_rate": 1.5054687551136623e-05,
      "loss": 0.0677,
      "step": 109550
    },
    {
      "epoch": 1.6137464838514897,
      "grad_norm": 1.8723078966140747,
      "learning_rate": 1.505305117916964e-05,
      "loss": 0.0784,
      "step": 109575
    },
    {
      "epoch": 1.6141146669415767,
      "grad_norm": 1.5051474571228027,
      "learning_rate": 1.5051414807202656e-05,
      "loss": 0.072,
      "step": 109600
    },
    {
      "epoch": 1.6144828500316637,
      "grad_norm": 1.5857611894607544,
      "learning_rate": 1.5049778435235672e-05,
      "loss": 0.0703,
      "step": 109625
    },
    {
      "epoch": 1.614851033121751,
      "grad_norm": 1.2172809839248657,
      "learning_rate": 1.5048142063268686e-05,
      "loss": 0.0713,
      "step": 109650
    },
    {
      "epoch": 1.615219216211838,
      "grad_norm": 2.0073728561401367,
      "learning_rate": 1.5046505691301702e-05,
      "loss": 0.0638,
      "step": 109675
    },
    {
      "epoch": 1.615587399301925,
      "grad_norm": 2.3504676818847656,
      "learning_rate": 1.5044869319334719e-05,
      "loss": 0.0751,
      "step": 109700
    },
    {
      "epoch": 1.615955582392012,
      "grad_norm": 1.41877019405365,
      "learning_rate": 1.5043232947367735e-05,
      "loss": 0.0759,
      "step": 109725
    },
    {
      "epoch": 1.616323765482099,
      "grad_norm": 1.5118829011917114,
      "learning_rate": 1.504159657540075e-05,
      "loss": 0.0721,
      "step": 109750
    },
    {
      "epoch": 1.616691948572186,
      "grad_norm": 1.2893067598342896,
      "learning_rate": 1.5039960203433764e-05,
      "loss": 0.0675,
      "step": 109775
    },
    {
      "epoch": 1.617060131662273,
      "grad_norm": 1.3109837770462036,
      "learning_rate": 1.503838928634546e-05,
      "loss": 0.0715,
      "step": 109800
    },
    {
      "epoch": 1.61742831475236,
      "grad_norm": 1.4243651628494263,
      "learning_rate": 1.5036752914378474e-05,
      "loss": 0.0661,
      "step": 109825
    },
    {
      "epoch": 1.617796497842447,
      "grad_norm": 1.4010547399520874,
      "learning_rate": 1.503511654241149e-05,
      "loss": 0.068,
      "step": 109850
    },
    {
      "epoch": 1.618164680932534,
      "grad_norm": 1.4534114599227905,
      "learning_rate": 1.5033480170444506e-05,
      "loss": 0.065,
      "step": 109875
    },
    {
      "epoch": 1.618532864022621,
      "grad_norm": 1.0850321054458618,
      "learning_rate": 1.503184379847752e-05,
      "loss": 0.0695,
      "step": 109900
    },
    {
      "epoch": 1.618901047112708,
      "grad_norm": 1.2090368270874023,
      "learning_rate": 1.5030207426510537e-05,
      "loss": 0.0671,
      "step": 109925
    },
    {
      "epoch": 1.6192692302027951,
      "grad_norm": 1.1374214887619019,
      "learning_rate": 1.5028571054543553e-05,
      "loss": 0.0651,
      "step": 109950
    },
    {
      "epoch": 1.6196374132928821,
      "grad_norm": 1.546884536743164,
      "learning_rate": 1.5026934682576565e-05,
      "loss": 0.0752,
      "step": 109975
    },
    {
      "epoch": 1.6200055963829694,
      "grad_norm": 1.5034337043762207,
      "learning_rate": 1.5025298310609582e-05,
      "loss": 0.0699,
      "step": 110000
    },
    {
      "epoch": 1.6203737794730564,
      "grad_norm": 0.9288470149040222,
      "learning_rate": 1.5023661938642598e-05,
      "loss": 0.0746,
      "step": 110025
    },
    {
      "epoch": 1.6207419625631434,
      "grad_norm": 1.1490124464035034,
      "learning_rate": 1.5022025566675614e-05,
      "loss": 0.0727,
      "step": 110050
    },
    {
      "epoch": 1.6211101456532304,
      "grad_norm": 1.4370003938674927,
      "learning_rate": 1.5020389194708628e-05,
      "loss": 0.0735,
      "step": 110075
    },
    {
      "epoch": 1.6214783287433174,
      "grad_norm": 1.039117693901062,
      "learning_rate": 1.5018752822741645e-05,
      "loss": 0.0667,
      "step": 110100
    },
    {
      "epoch": 1.6218465118334047,
      "grad_norm": 1.1348016262054443,
      "learning_rate": 1.501711645077466e-05,
      "loss": 0.0683,
      "step": 110125
    },
    {
      "epoch": 1.6222146949234917,
      "grad_norm": 1.0824862718582153,
      "learning_rate": 1.5015480078807677e-05,
      "loss": 0.0616,
      "step": 110150
    },
    {
      "epoch": 1.6225828780135787,
      "grad_norm": 0.8653793931007385,
      "learning_rate": 1.5013843706840691e-05,
      "loss": 0.0666,
      "step": 110175
    },
    {
      "epoch": 1.6229510611036657,
      "grad_norm": 1.882455825805664,
      "learning_rate": 1.5012207334873706e-05,
      "loss": 0.0734,
      "step": 110200
    },
    {
      "epoch": 1.6233192441937527,
      "grad_norm": 1.9056713581085205,
      "learning_rate": 1.501057096290672e-05,
      "loss": 0.0734,
      "step": 110225
    },
    {
      "epoch": 1.6236874272838397,
      "grad_norm": 1.5713926553726196,
      "learning_rate": 1.5008934590939736e-05,
      "loss": 0.0824,
      "step": 110250
    },
    {
      "epoch": 1.6240556103739268,
      "grad_norm": 1.3881809711456299,
      "learning_rate": 1.5007298218972753e-05,
      "loss": 0.0658,
      "step": 110275
    },
    {
      "epoch": 1.6244237934640138,
      "grad_norm": 1.973669409751892,
      "learning_rate": 1.5005661847005769e-05,
      "loss": 0.074,
      "step": 110300
    },
    {
      "epoch": 1.6247919765541008,
      "grad_norm": 2.0806632041931152,
      "learning_rate": 1.5004025475038783e-05,
      "loss": 0.0715,
      "step": 110325
    },
    {
      "epoch": 1.6251601596441878,
      "grad_norm": 1.4756126403808594,
      "learning_rate": 1.50023891030718e-05,
      "loss": 0.0722,
      "step": 110350
    },
    {
      "epoch": 1.6255283427342748,
      "grad_norm": 1.3490571975708008,
      "learning_rate": 1.5000752731104815e-05,
      "loss": 0.0707,
      "step": 110375
    },
    {
      "epoch": 1.6258965258243618,
      "grad_norm": 1.4615627527236938,
      "learning_rate": 1.4999116359137828e-05,
      "loss": 0.0719,
      "step": 110400
    },
    {
      "epoch": 1.6262647089144489,
      "grad_norm": 0.8929535150527954,
      "learning_rate": 1.4997479987170844e-05,
      "loss": 0.0717,
      "step": 110425
    },
    {
      "epoch": 1.6266328920045359,
      "grad_norm": 0.678657054901123,
      "learning_rate": 1.499584361520386e-05,
      "loss": 0.0647,
      "step": 110450
    },
    {
      "epoch": 1.627001075094623,
      "grad_norm": 1.6889570951461792,
      "learning_rate": 1.4994207243236875e-05,
      "loss": 0.0728,
      "step": 110475
    },
    {
      "epoch": 1.6273692581847101,
      "grad_norm": 1.0937069654464722,
      "learning_rate": 1.4992570871269891e-05,
      "loss": 0.065,
      "step": 110500
    },
    {
      "epoch": 1.6277374412747971,
      "grad_norm": 1.5447055101394653,
      "learning_rate": 1.4990934499302907e-05,
      "loss": 0.0741,
      "step": 110525
    },
    {
      "epoch": 1.6281056243648842,
      "grad_norm": 1.5269920825958252,
      "learning_rate": 1.4989298127335923e-05,
      "loss": 0.0702,
      "step": 110550
    },
    {
      "epoch": 1.6284738074549712,
      "grad_norm": 1.013918161392212,
      "learning_rate": 1.4987661755368938e-05,
      "loss": 0.0763,
      "step": 110575
    },
    {
      "epoch": 1.6288419905450584,
      "grad_norm": 1.2948002815246582,
      "learning_rate": 1.4986025383401952e-05,
      "loss": 0.0726,
      "step": 110600
    },
    {
      "epoch": 1.6292101736351454,
      "grad_norm": 1.2564746141433716,
      "learning_rate": 1.4984389011434968e-05,
      "loss": 0.0712,
      "step": 110625
    },
    {
      "epoch": 1.6295783567252324,
      "grad_norm": 1.756926417350769,
      "learning_rate": 1.4982752639467983e-05,
      "loss": 0.0679,
      "step": 110650
    },
    {
      "epoch": 1.6299465398153195,
      "grad_norm": 1.5547542572021484,
      "learning_rate": 1.4981116267500999e-05,
      "loss": 0.0697,
      "step": 110675
    },
    {
      "epoch": 1.6303147229054065,
      "grad_norm": 1.1720080375671387,
      "learning_rate": 1.4979479895534015e-05,
      "loss": 0.0677,
      "step": 110700
    },
    {
      "epoch": 1.6306829059954935,
      "grad_norm": 1.7022597789764404,
      "learning_rate": 1.497784352356703e-05,
      "loss": 0.0728,
      "step": 110725
    },
    {
      "epoch": 1.6310510890855805,
      "grad_norm": 1.089576005935669,
      "learning_rate": 1.4976207151600046e-05,
      "loss": 0.0664,
      "step": 110750
    },
    {
      "epoch": 1.6314192721756675,
      "grad_norm": 1.1406174898147583,
      "learning_rate": 1.4974570779633062e-05,
      "loss": 0.0664,
      "step": 110775
    },
    {
      "epoch": 1.6317874552657545,
      "grad_norm": 1.0567905902862549,
      "learning_rate": 1.4972934407666078e-05,
      "loss": 0.071,
      "step": 110800
    },
    {
      "epoch": 1.6321556383558415,
      "grad_norm": 1.025338888168335,
      "learning_rate": 1.4971298035699091e-05,
      "loss": 0.0697,
      "step": 110825
    },
    {
      "epoch": 1.6325238214459286,
      "grad_norm": 1.0947364568710327,
      "learning_rate": 1.4969661663732107e-05,
      "loss": 0.0695,
      "step": 110850
    },
    {
      "epoch": 1.6328920045360156,
      "grad_norm": 1.6678235530853271,
      "learning_rate": 1.4968025291765123e-05,
      "loss": 0.0754,
      "step": 110875
    },
    {
      "epoch": 1.6332601876261026,
      "grad_norm": 1.4429137706756592,
      "learning_rate": 1.4966388919798138e-05,
      "loss": 0.0684,
      "step": 110900
    },
    {
      "epoch": 1.6336283707161896,
      "grad_norm": 1.7626643180847168,
      "learning_rate": 1.4964752547831154e-05,
      "loss": 0.0791,
      "step": 110925
    },
    {
      "epoch": 1.6339965538062766,
      "grad_norm": 1.1349862813949585,
      "learning_rate": 1.496311617586417e-05,
      "loss": 0.0635,
      "step": 110950
    },
    {
      "epoch": 1.6343647368963639,
      "grad_norm": 1.5567644834518433,
      "learning_rate": 1.4961479803897186e-05,
      "loss": 0.0663,
      "step": 110975
    },
    {
      "epoch": 1.6347329199864509,
      "grad_norm": 1.0278818607330322,
      "learning_rate": 1.49598434319302e-05,
      "loss": 0.0702,
      "step": 111000
    },
    {
      "epoch": 1.6351011030765379,
      "grad_norm": 1.507691740989685,
      "learning_rate": 1.4958207059963215e-05,
      "loss": 0.067,
      "step": 111025
    },
    {
      "epoch": 1.635469286166625,
      "grad_norm": 1.0916681289672852,
      "learning_rate": 1.495657068799623e-05,
      "loss": 0.0656,
      "step": 111050
    },
    {
      "epoch": 1.6358374692567121,
      "grad_norm": 1.798743724822998,
      "learning_rate": 1.4954934316029246e-05,
      "loss": 0.0694,
      "step": 111075
    },
    {
      "epoch": 1.6362056523467992,
      "grad_norm": 1.8959791660308838,
      "learning_rate": 1.4953297944062262e-05,
      "loss": 0.071,
      "step": 111100
    },
    {
      "epoch": 1.6365738354368862,
      "grad_norm": 1.5384111404418945,
      "learning_rate": 1.4951661572095278e-05,
      "loss": 0.0675,
      "step": 111125
    },
    {
      "epoch": 1.6369420185269732,
      "grad_norm": 1.4105653762817383,
      "learning_rate": 1.4950025200128292e-05,
      "loss": 0.0722,
      "step": 111150
    },
    {
      "epoch": 1.6373102016170602,
      "grad_norm": 1.0455262660980225,
      "learning_rate": 1.4948388828161309e-05,
      "loss": 0.0748,
      "step": 111175
    },
    {
      "epoch": 1.6376783847071472,
      "grad_norm": 1.0242788791656494,
      "learning_rate": 1.4946752456194325e-05,
      "loss": 0.0683,
      "step": 111200
    },
    {
      "epoch": 1.6380465677972342,
      "grad_norm": 1.522147297859192,
      "learning_rate": 1.4945116084227341e-05,
      "loss": 0.0638,
      "step": 111225
    },
    {
      "epoch": 1.6384147508873212,
      "grad_norm": 1.4644957780838013,
      "learning_rate": 1.4943479712260354e-05,
      "loss": 0.0659,
      "step": 111250
    },
    {
      "epoch": 1.6387829339774083,
      "grad_norm": 1.6227401494979858,
      "learning_rate": 1.494184334029337e-05,
      "loss": 0.0767,
      "step": 111275
    },
    {
      "epoch": 1.6391511170674953,
      "grad_norm": 1.3954418897628784,
      "learning_rate": 1.4940206968326384e-05,
      "loss": 0.0717,
      "step": 111300
    },
    {
      "epoch": 1.6395193001575823,
      "grad_norm": 1.330049753189087,
      "learning_rate": 1.49385705963594e-05,
      "loss": 0.0695,
      "step": 111325
    },
    {
      "epoch": 1.6398874832476693,
      "grad_norm": 1.3499977588653564,
      "learning_rate": 1.4936934224392417e-05,
      "loss": 0.0689,
      "step": 111350
    },
    {
      "epoch": 1.6402556663377563,
      "grad_norm": 1.503440260887146,
      "learning_rate": 1.4935297852425433e-05,
      "loss": 0.0756,
      "step": 111375
    },
    {
      "epoch": 1.6406238494278433,
      "grad_norm": 1.3987048864364624,
      "learning_rate": 1.4933661480458447e-05,
      "loss": 0.0715,
      "step": 111400
    },
    {
      "epoch": 1.6409920325179304,
      "grad_norm": 2.3647842407226562,
      "learning_rate": 1.4932025108491463e-05,
      "loss": 0.0762,
      "step": 111425
    },
    {
      "epoch": 1.6413602156080176,
      "grad_norm": 1.2387139797210693,
      "learning_rate": 1.4930388736524478e-05,
      "loss": 0.0656,
      "step": 111450
    },
    {
      "epoch": 1.6417283986981046,
      "grad_norm": 1.6368547677993774,
      "learning_rate": 1.4928752364557492e-05,
      "loss": 0.0671,
      "step": 111475
    },
    {
      "epoch": 1.6420965817881916,
      "grad_norm": 1.33992600440979,
      "learning_rate": 1.4927115992590508e-05,
      "loss": 0.0756,
      "step": 111500
    },
    {
      "epoch": 1.6424647648782786,
      "grad_norm": 1.510890245437622,
      "learning_rate": 1.4925479620623525e-05,
      "loss": 0.0811,
      "step": 111525
    },
    {
      "epoch": 1.6428329479683659,
      "grad_norm": 1.5336076021194458,
      "learning_rate": 1.492384324865654e-05,
      "loss": 0.0699,
      "step": 111550
    },
    {
      "epoch": 1.643201131058453,
      "grad_norm": 1.6068021059036255,
      "learning_rate": 1.4922206876689555e-05,
      "loss": 0.0748,
      "step": 111575
    },
    {
      "epoch": 1.64356931414854,
      "grad_norm": 1.225089430809021,
      "learning_rate": 1.4920570504722571e-05,
      "loss": 0.0672,
      "step": 111600
    },
    {
      "epoch": 1.643937497238627,
      "grad_norm": 1.656948208808899,
      "learning_rate": 1.4918934132755587e-05,
      "loss": 0.0679,
      "step": 111625
    },
    {
      "epoch": 1.644305680328714,
      "grad_norm": 1.2077441215515137,
      "learning_rate": 1.4917297760788602e-05,
      "loss": 0.0767,
      "step": 111650
    },
    {
      "epoch": 1.644673863418801,
      "grad_norm": 0.8018178939819336,
      "learning_rate": 1.4915661388821616e-05,
      "loss": 0.0682,
      "step": 111675
    },
    {
      "epoch": 1.645042046508888,
      "grad_norm": 1.284868597984314,
      "learning_rate": 1.4914025016854632e-05,
      "loss": 0.0709,
      "step": 111700
    },
    {
      "epoch": 1.645410229598975,
      "grad_norm": 1.558538556098938,
      "learning_rate": 1.4912388644887647e-05,
      "loss": 0.0676,
      "step": 111725
    },
    {
      "epoch": 1.645778412689062,
      "grad_norm": 1.5586585998535156,
      "learning_rate": 1.4910752272920663e-05,
      "loss": 0.0632,
      "step": 111750
    },
    {
      "epoch": 1.646146595779149,
      "grad_norm": 1.346023440361023,
      "learning_rate": 1.490911590095368e-05,
      "loss": 0.0723,
      "step": 111775
    },
    {
      "epoch": 1.646514778869236,
      "grad_norm": 1.0624632835388184,
      "learning_rate": 1.4907479528986695e-05,
      "loss": 0.0725,
      "step": 111800
    },
    {
      "epoch": 1.646882961959323,
      "grad_norm": 1.1744948625564575,
      "learning_rate": 1.490584315701971e-05,
      "loss": 0.0738,
      "step": 111825
    },
    {
      "epoch": 1.64725114504941,
      "grad_norm": 1.0973681211471558,
      "learning_rate": 1.4904206785052726e-05,
      "loss": 0.0686,
      "step": 111850
    },
    {
      "epoch": 1.647619328139497,
      "grad_norm": 1.0810407400131226,
      "learning_rate": 1.4902570413085739e-05,
      "loss": 0.0631,
      "step": 111875
    },
    {
      "epoch": 1.647987511229584,
      "grad_norm": 1.1640533208847046,
      "learning_rate": 1.4900934041118755e-05,
      "loss": 0.0696,
      "step": 111900
    },
    {
      "epoch": 1.6483556943196713,
      "grad_norm": 1.622637391090393,
      "learning_rate": 1.4899297669151771e-05,
      "loss": 0.0743,
      "step": 111925
    },
    {
      "epoch": 1.6487238774097583,
      "grad_norm": 0.9899869561195374,
      "learning_rate": 1.4897661297184787e-05,
      "loss": 0.0665,
      "step": 111950
    },
    {
      "epoch": 1.6490920604998454,
      "grad_norm": 1.7448487281799316,
      "learning_rate": 1.4896024925217802e-05,
      "loss": 0.0711,
      "step": 111975
    },
    {
      "epoch": 1.6494602435899324,
      "grad_norm": 1.130418062210083,
      "learning_rate": 1.4894388553250818e-05,
      "loss": 0.0662,
      "step": 112000
    },
    {
      "epoch": 1.6498284266800194,
      "grad_norm": 1.063306212425232,
      "learning_rate": 1.4892752181283834e-05,
      "loss": 0.0693,
      "step": 112025
    },
    {
      "epoch": 1.6501966097701066,
      "grad_norm": 1.5438015460968018,
      "learning_rate": 1.489111580931685e-05,
      "loss": 0.0643,
      "step": 112050
    },
    {
      "epoch": 1.6505647928601936,
      "grad_norm": 1.6509699821472168,
      "learning_rate": 1.4889479437349863e-05,
      "loss": 0.0723,
      "step": 112075
    },
    {
      "epoch": 1.6509329759502807,
      "grad_norm": 1.7110092639923096,
      "learning_rate": 1.4887843065382879e-05,
      "loss": 0.0761,
      "step": 112100
    },
    {
      "epoch": 1.6513011590403677,
      "grad_norm": 1.7650585174560547,
      "learning_rate": 1.4886272148294575e-05,
      "loss": 0.0758,
      "step": 112125
    },
    {
      "epoch": 1.6516693421304547,
      "grad_norm": 1.121733546257019,
      "learning_rate": 1.4884635776327589e-05,
      "loss": 0.0812,
      "step": 112150
    },
    {
      "epoch": 1.6520375252205417,
      "grad_norm": 1.4370988607406616,
      "learning_rate": 1.4882999404360605e-05,
      "loss": 0.0686,
      "step": 112175
    },
    {
      "epoch": 1.6524057083106287,
      "grad_norm": 1.1857341527938843,
      "learning_rate": 1.4881363032393621e-05,
      "loss": 0.0708,
      "step": 112200
    },
    {
      "epoch": 1.6527738914007157,
      "grad_norm": 1.7822493314743042,
      "learning_rate": 1.4879726660426636e-05,
      "loss": 0.0746,
      "step": 112225
    },
    {
      "epoch": 1.6531420744908027,
      "grad_norm": 1.1039087772369385,
      "learning_rate": 1.4878090288459652e-05,
      "loss": 0.0728,
      "step": 112250
    },
    {
      "epoch": 1.6535102575808898,
      "grad_norm": 1.7532680034637451,
      "learning_rate": 1.4876453916492668e-05,
      "loss": 0.0731,
      "step": 112275
    },
    {
      "epoch": 1.6538784406709768,
      "grad_norm": 1.311147928237915,
      "learning_rate": 1.4874817544525681e-05,
      "loss": 0.0797,
      "step": 112300
    },
    {
      "epoch": 1.6542466237610638,
      "grad_norm": 1.24696683883667,
      "learning_rate": 1.4873181172558697e-05,
      "loss": 0.075,
      "step": 112325
    },
    {
      "epoch": 1.6546148068511508,
      "grad_norm": 1.130251169204712,
      "learning_rate": 1.4871544800591713e-05,
      "loss": 0.066,
      "step": 112350
    },
    {
      "epoch": 1.6549829899412378,
      "grad_norm": 1.2997519969940186,
      "learning_rate": 1.486990842862473e-05,
      "loss": 0.0693,
      "step": 112375
    },
    {
      "epoch": 1.655351173031325,
      "grad_norm": 1.2289658784866333,
      "learning_rate": 1.4868272056657744e-05,
      "loss": 0.0732,
      "step": 112400
    },
    {
      "epoch": 1.655719356121412,
      "grad_norm": 1.9662799835205078,
      "learning_rate": 1.486663568469076e-05,
      "loss": 0.0778,
      "step": 112425
    },
    {
      "epoch": 1.656087539211499,
      "grad_norm": 1.1644535064697266,
      "learning_rate": 1.4864999312723776e-05,
      "loss": 0.0637,
      "step": 112450
    },
    {
      "epoch": 1.656455722301586,
      "grad_norm": 1.3311115503311157,
      "learning_rate": 1.486336294075679e-05,
      "loss": 0.0722,
      "step": 112475
    },
    {
      "epoch": 1.6568239053916731,
      "grad_norm": 1.3293944597244263,
      "learning_rate": 1.4861726568789805e-05,
      "loss": 0.0756,
      "step": 112500
    },
    {
      "epoch": 1.6571920884817604,
      "grad_norm": 1.6116830110549927,
      "learning_rate": 1.4860090196822821e-05,
      "loss": 0.0734,
      "step": 112525
    },
    {
      "epoch": 1.6575602715718474,
      "grad_norm": 1.4583842754364014,
      "learning_rate": 1.4858453824855836e-05,
      "loss": 0.0675,
      "step": 112550
    },
    {
      "epoch": 1.6579284546619344,
      "grad_norm": 1.2287795543670654,
      "learning_rate": 1.4856817452888852e-05,
      "loss": 0.0708,
      "step": 112575
    },
    {
      "epoch": 1.6582966377520214,
      "grad_norm": 2.1830837726593018,
      "learning_rate": 1.4855181080921868e-05,
      "loss": 0.0693,
      "step": 112600
    },
    {
      "epoch": 1.6586648208421084,
      "grad_norm": 1.717875599861145,
      "learning_rate": 1.4853544708954884e-05,
      "loss": 0.0669,
      "step": 112625
    },
    {
      "epoch": 1.6590330039321954,
      "grad_norm": 1.4657039642333984,
      "learning_rate": 1.4851908336987899e-05,
      "loss": 0.0661,
      "step": 112650
    },
    {
      "epoch": 1.6594011870222825,
      "grad_norm": 1.6358797550201416,
      "learning_rate": 1.4850271965020915e-05,
      "loss": 0.0728,
      "step": 112675
    },
    {
      "epoch": 1.6597693701123695,
      "grad_norm": 1.5901389122009277,
      "learning_rate": 1.4848635593053931e-05,
      "loss": 0.0708,
      "step": 112700
    },
    {
      "epoch": 1.6601375532024565,
      "grad_norm": 0.8792926073074341,
      "learning_rate": 1.4846999221086944e-05,
      "loss": 0.072,
      "step": 112725
    },
    {
      "epoch": 1.6605057362925435,
      "grad_norm": 1.4907500743865967,
      "learning_rate": 1.484536284911996e-05,
      "loss": 0.0625,
      "step": 112750
    },
    {
      "epoch": 1.6608739193826305,
      "grad_norm": 1.5523349046707153,
      "learning_rate": 1.4843726477152976e-05,
      "loss": 0.076,
      "step": 112775
    },
    {
      "epoch": 1.6612421024727175,
      "grad_norm": 0.7739635705947876,
      "learning_rate": 1.484209010518599e-05,
      "loss": 0.0704,
      "step": 112800
    },
    {
      "epoch": 1.6616102855628045,
      "grad_norm": 1.430060625076294,
      "learning_rate": 1.4840453733219007e-05,
      "loss": 0.0669,
      "step": 112825
    },
    {
      "epoch": 1.6619784686528916,
      "grad_norm": 1.6013901233673096,
      "learning_rate": 1.4838817361252023e-05,
      "loss": 0.0757,
      "step": 112850
    },
    {
      "epoch": 1.6623466517429788,
      "grad_norm": 1.3758858442306519,
      "learning_rate": 1.4837180989285039e-05,
      "loss": 0.0681,
      "step": 112875
    },
    {
      "epoch": 1.6627148348330658,
      "grad_norm": 1.733469009399414,
      "learning_rate": 1.4835544617318053e-05,
      "loss": 0.0683,
      "step": 112900
    },
    {
      "epoch": 1.6630830179231528,
      "grad_norm": 1.7212387323379517,
      "learning_rate": 1.4833908245351068e-05,
      "loss": 0.0678,
      "step": 112925
    },
    {
      "epoch": 1.6634512010132398,
      "grad_norm": 1.648301362991333,
      "learning_rate": 1.4832271873384084e-05,
      "loss": 0.0709,
      "step": 112950
    },
    {
      "epoch": 1.6638193841033269,
      "grad_norm": 1.7456063032150269,
      "learning_rate": 1.4830635501417098e-05,
      "loss": 0.0702,
      "step": 112975
    },
    {
      "epoch": 1.664187567193414,
      "grad_norm": 0.9833461046218872,
      "learning_rate": 1.4828999129450115e-05,
      "loss": 0.0722,
      "step": 113000
    },
    {
      "epoch": 1.6645557502835011,
      "grad_norm": 1.1886141300201416,
      "learning_rate": 1.482736275748313e-05,
      "loss": 0.0699,
      "step": 113025
    },
    {
      "epoch": 1.6649239333735881,
      "grad_norm": 1.4260448217391968,
      "learning_rate": 1.4825726385516145e-05,
      "loss": 0.0681,
      "step": 113050
    },
    {
      "epoch": 1.6652921164636751,
      "grad_norm": 1.5015608072280884,
      "learning_rate": 1.4824090013549161e-05,
      "loss": 0.0729,
      "step": 113075
    },
    {
      "epoch": 1.6656602995537622,
      "grad_norm": 1.091711163520813,
      "learning_rate": 1.4822453641582177e-05,
      "loss": 0.0641,
      "step": 113100
    },
    {
      "epoch": 1.6660284826438492,
      "grad_norm": 0.8716529607772827,
      "learning_rate": 1.4820817269615194e-05,
      "loss": 0.0645,
      "step": 113125
    },
    {
      "epoch": 1.6663966657339362,
      "grad_norm": 1.4132845401763916,
      "learning_rate": 1.4819180897648206e-05,
      "loss": 0.0755,
      "step": 113150
    },
    {
      "epoch": 1.6667648488240232,
      "grad_norm": 1.70515775680542,
      "learning_rate": 1.4817544525681222e-05,
      "loss": 0.0722,
      "step": 113175
    },
    {
      "epoch": 1.6671330319141102,
      "grad_norm": 1.4492024183273315,
      "learning_rate": 1.4815908153714239e-05,
      "loss": 0.0834,
      "step": 113200
    },
    {
      "epoch": 1.6675012150041972,
      "grad_norm": 1.0968266725540161,
      "learning_rate": 1.4814271781747253e-05,
      "loss": 0.0671,
      "step": 113225
    },
    {
      "epoch": 1.6678693980942843,
      "grad_norm": 1.6026828289031982,
      "learning_rate": 1.481263540978027e-05,
      "loss": 0.0753,
      "step": 113250
    },
    {
      "epoch": 1.6682375811843713,
      "grad_norm": 1.0314505100250244,
      "learning_rate": 1.4810999037813285e-05,
      "loss": 0.0758,
      "step": 113275
    },
    {
      "epoch": 1.6686057642744583,
      "grad_norm": 1.3181610107421875,
      "learning_rate": 1.4809362665846302e-05,
      "loss": 0.0714,
      "step": 113300
    },
    {
      "epoch": 1.6689739473645453,
      "grad_norm": 1.8939374685287476,
      "learning_rate": 1.4807726293879316e-05,
      "loss": 0.0778,
      "step": 113325
    },
    {
      "epoch": 1.6693421304546325,
      "grad_norm": 1.4272085428237915,
      "learning_rate": 1.480608992191233e-05,
      "loss": 0.0719,
      "step": 113350
    },
    {
      "epoch": 1.6697103135447195,
      "grad_norm": 1.146855115890503,
      "learning_rate": 1.4804453549945345e-05,
      "loss": 0.0646,
      "step": 113375
    },
    {
      "epoch": 1.6700784966348066,
      "grad_norm": 1.147344946861267,
      "learning_rate": 1.4802817177978361e-05,
      "loss": 0.0672,
      "step": 113400
    },
    {
      "epoch": 1.6704466797248936,
      "grad_norm": 1.5350712537765503,
      "learning_rate": 1.4801180806011377e-05,
      "loss": 0.0679,
      "step": 113425
    },
    {
      "epoch": 1.6708148628149806,
      "grad_norm": 1.4562945365905762,
      "learning_rate": 1.4799544434044393e-05,
      "loss": 0.0729,
      "step": 113450
    },
    {
      "epoch": 1.6711830459050678,
      "grad_norm": 1.4934054613113403,
      "learning_rate": 1.4797908062077408e-05,
      "loss": 0.0725,
      "step": 113475
    },
    {
      "epoch": 1.6715512289951548,
      "grad_norm": 1.9927408695220947,
      "learning_rate": 1.4796271690110424e-05,
      "loss": 0.068,
      "step": 113500
    },
    {
      "epoch": 1.6719194120852419,
      "grad_norm": 1.384314775466919,
      "learning_rate": 1.479463531814344e-05,
      "loss": 0.0638,
      "step": 113525
    },
    {
      "epoch": 1.6722875951753289,
      "grad_norm": 1.1050392389297485,
      "learning_rate": 1.4792998946176453e-05,
      "loss": 0.0702,
      "step": 113550
    },
    {
      "epoch": 1.672655778265416,
      "grad_norm": 2.284367322921753,
      "learning_rate": 1.4791362574209469e-05,
      "loss": 0.0704,
      "step": 113575
    },
    {
      "epoch": 1.673023961355503,
      "grad_norm": 1.4867076873779297,
      "learning_rate": 1.4789726202242485e-05,
      "loss": 0.0771,
      "step": 113600
    },
    {
      "epoch": 1.67339214444559,
      "grad_norm": 1.328139066696167,
      "learning_rate": 1.47880898302755e-05,
      "loss": 0.0681,
      "step": 113625
    },
    {
      "epoch": 1.673760327535677,
      "grad_norm": 1.6640952825546265,
      "learning_rate": 1.4786453458308516e-05,
      "loss": 0.072,
      "step": 113650
    },
    {
      "epoch": 1.674128510625764,
      "grad_norm": 1.2903510332107544,
      "learning_rate": 1.4784817086341532e-05,
      "loss": 0.0758,
      "step": 113675
    },
    {
      "epoch": 1.674496693715851,
      "grad_norm": 1.137075424194336,
      "learning_rate": 1.4783180714374548e-05,
      "loss": 0.0724,
      "step": 113700
    },
    {
      "epoch": 1.674864876805938,
      "grad_norm": 1.641725778579712,
      "learning_rate": 1.4781544342407563e-05,
      "loss": 0.0606,
      "step": 113725
    },
    {
      "epoch": 1.675233059896025,
      "grad_norm": 1.3679289817810059,
      "learning_rate": 1.4779907970440579e-05,
      "loss": 0.0649,
      "step": 113750
    },
    {
      "epoch": 1.675601242986112,
      "grad_norm": 1.2159751653671265,
      "learning_rate": 1.4778271598473593e-05,
      "loss": 0.0721,
      "step": 113775
    },
    {
      "epoch": 1.675969426076199,
      "grad_norm": 2.194733142852783,
      "learning_rate": 1.4776635226506608e-05,
      "loss": 0.0752,
      "step": 113800
    },
    {
      "epoch": 1.676337609166286,
      "grad_norm": 1.3089299201965332,
      "learning_rate": 1.4774998854539624e-05,
      "loss": 0.0662,
      "step": 113825
    },
    {
      "epoch": 1.6767057922563733,
      "grad_norm": 1.672057867050171,
      "learning_rate": 1.477336248257264e-05,
      "loss": 0.0747,
      "step": 113850
    },
    {
      "epoch": 1.6770739753464603,
      "grad_norm": 1.4456645250320435,
      "learning_rate": 1.4771726110605656e-05,
      "loss": 0.0701,
      "step": 113875
    },
    {
      "epoch": 1.6774421584365473,
      "grad_norm": 1.05272376537323,
      "learning_rate": 1.477008973863867e-05,
      "loss": 0.0726,
      "step": 113900
    },
    {
      "epoch": 1.6778103415266343,
      "grad_norm": 1.6001595258712769,
      "learning_rate": 1.4768453366671687e-05,
      "loss": 0.0696,
      "step": 113925
    },
    {
      "epoch": 1.6781785246167216,
      "grad_norm": 1.5390887260437012,
      "learning_rate": 1.4766816994704703e-05,
      "loss": 0.0618,
      "step": 113950
    },
    {
      "epoch": 1.6785467077068086,
      "grad_norm": 1.4526466131210327,
      "learning_rate": 1.4765180622737716e-05,
      "loss": 0.0627,
      "step": 113975
    },
    {
      "epoch": 1.6789148907968956,
      "grad_norm": 1.9929816722869873,
      "learning_rate": 1.4763544250770732e-05,
      "loss": 0.0741,
      "step": 114000
    },
    {
      "epoch": 1.6792830738869826,
      "grad_norm": 1.51315176486969,
      "learning_rate": 1.4761907878803748e-05,
      "loss": 0.0701,
      "step": 114025
    },
    {
      "epoch": 1.6796512569770696,
      "grad_norm": 1.6183960437774658,
      "learning_rate": 1.4760271506836762e-05,
      "loss": 0.0695,
      "step": 114050
    },
    {
      "epoch": 1.6800194400671566,
      "grad_norm": 1.849167823791504,
      "learning_rate": 1.4758635134869779e-05,
      "loss": 0.0739,
      "step": 114075
    },
    {
      "epoch": 1.6803876231572437,
      "grad_norm": 1.4521594047546387,
      "learning_rate": 1.4756998762902795e-05,
      "loss": 0.0761,
      "step": 114100
    },
    {
      "epoch": 1.6807558062473307,
      "grad_norm": 1.2524521350860596,
      "learning_rate": 1.475536239093581e-05,
      "loss": 0.0721,
      "step": 114125
    },
    {
      "epoch": 1.6811239893374177,
      "grad_norm": 1.1944258213043213,
      "learning_rate": 1.4753726018968825e-05,
      "loss": 0.0669,
      "step": 114150
    },
    {
      "epoch": 1.6814921724275047,
      "grad_norm": 1.4698984622955322,
      "learning_rate": 1.4752089647001841e-05,
      "loss": 0.0701,
      "step": 114175
    },
    {
      "epoch": 1.6818603555175917,
      "grad_norm": 1.1985230445861816,
      "learning_rate": 1.4750453275034854e-05,
      "loss": 0.0714,
      "step": 114200
    },
    {
      "epoch": 1.6822285386076787,
      "grad_norm": 1.5233608484268188,
      "learning_rate": 1.474881690306787e-05,
      "loss": 0.0722,
      "step": 114225
    },
    {
      "epoch": 1.6825967216977658,
      "grad_norm": 1.4962819814682007,
      "learning_rate": 1.4747180531100887e-05,
      "loss": 0.0659,
      "step": 114250
    },
    {
      "epoch": 1.6829649047878528,
      "grad_norm": 1.8392302989959717,
      "learning_rate": 1.4745544159133903e-05,
      "loss": 0.0691,
      "step": 114275
    },
    {
      "epoch": 1.6833330878779398,
      "grad_norm": 1.3911542892456055,
      "learning_rate": 1.4743907787166917e-05,
      "loss": 0.0736,
      "step": 114300
    },
    {
      "epoch": 1.683701270968027,
      "grad_norm": 1.573142409324646,
      "learning_rate": 1.4742271415199933e-05,
      "loss": 0.0755,
      "step": 114325
    },
    {
      "epoch": 1.684069454058114,
      "grad_norm": 1.6133112907409668,
      "learning_rate": 1.474063504323295e-05,
      "loss": 0.0691,
      "step": 114350
    },
    {
      "epoch": 1.684437637148201,
      "grad_norm": 1.4357415437698364,
      "learning_rate": 1.4738998671265966e-05,
      "loss": 0.0675,
      "step": 114375
    },
    {
      "epoch": 1.684805820238288,
      "grad_norm": 1.5966781377792358,
      "learning_rate": 1.4737362299298978e-05,
      "loss": 0.0736,
      "step": 114400
    },
    {
      "epoch": 1.685174003328375,
      "grad_norm": 1.2438653707504272,
      "learning_rate": 1.4735725927331994e-05,
      "loss": 0.0755,
      "step": 114425
    },
    {
      "epoch": 1.6855421864184623,
      "grad_norm": 1.1986197233200073,
      "learning_rate": 1.473408955536501e-05,
      "loss": 0.0697,
      "step": 114450
    },
    {
      "epoch": 1.6859103695085493,
      "grad_norm": 1.7107959985733032,
      "learning_rate": 1.4732453183398025e-05,
      "loss": 0.0787,
      "step": 114475
    },
    {
      "epoch": 1.6862785525986363,
      "grad_norm": 1.837692379951477,
      "learning_rate": 1.4730816811431041e-05,
      "loss": 0.0729,
      "step": 114500
    },
    {
      "epoch": 1.6866467356887234,
      "grad_norm": 1.9337834119796753,
      "learning_rate": 1.4729180439464057e-05,
      "loss": 0.0735,
      "step": 114525
    },
    {
      "epoch": 1.6870149187788104,
      "grad_norm": 1.1843305826187134,
      "learning_rate": 1.4727544067497072e-05,
      "loss": 0.0601,
      "step": 114550
    },
    {
      "epoch": 1.6873831018688974,
      "grad_norm": 1.3090814352035522,
      "learning_rate": 1.4725907695530088e-05,
      "loss": 0.0717,
      "step": 114575
    },
    {
      "epoch": 1.6877512849589844,
      "grad_norm": 1.0865365266799927,
      "learning_rate": 1.4724271323563104e-05,
      "loss": 0.0708,
      "step": 114600
    },
    {
      "epoch": 1.6881194680490714,
      "grad_norm": 1.207167148590088,
      "learning_rate": 1.4722700406474796e-05,
      "loss": 0.0719,
      "step": 114625
    },
    {
      "epoch": 1.6884876511391584,
      "grad_norm": 1.486923336982727,
      "learning_rate": 1.4721064034507813e-05,
      "loss": 0.0691,
      "step": 114650
    },
    {
      "epoch": 1.6888558342292455,
      "grad_norm": 1.125577449798584,
      "learning_rate": 1.4719427662540829e-05,
      "loss": 0.0745,
      "step": 114675
    },
    {
      "epoch": 1.6892240173193325,
      "grad_norm": 1.625648021697998,
      "learning_rate": 1.4717791290573845e-05,
      "loss": 0.0763,
      "step": 114700
    },
    {
      "epoch": 1.6895922004094195,
      "grad_norm": 1.4399183988571167,
      "learning_rate": 1.471615491860686e-05,
      "loss": 0.0666,
      "step": 114725
    },
    {
      "epoch": 1.6899603834995065,
      "grad_norm": 1.0632739067077637,
      "learning_rate": 1.4714518546639875e-05,
      "loss": 0.0692,
      "step": 114750
    },
    {
      "epoch": 1.6903285665895935,
      "grad_norm": 1.558274269104004,
      "learning_rate": 1.4712882174672892e-05,
      "loss": 0.0666,
      "step": 114775
    },
    {
      "epoch": 1.6906967496796808,
      "grad_norm": 1.118222951889038,
      "learning_rate": 1.4711245802705906e-05,
      "loss": 0.0644,
      "step": 114800
    },
    {
      "epoch": 1.6910649327697678,
      "grad_norm": 1.4945584535598755,
      "learning_rate": 1.470960943073892e-05,
      "loss": 0.0667,
      "step": 114825
    },
    {
      "epoch": 1.6914331158598548,
      "grad_norm": 1.81775963306427,
      "learning_rate": 1.4707973058771937e-05,
      "loss": 0.0754,
      "step": 114850
    },
    {
      "epoch": 1.6918012989499418,
      "grad_norm": 1.2940665483474731,
      "learning_rate": 1.4706336686804951e-05,
      "loss": 0.069,
      "step": 114875
    },
    {
      "epoch": 1.6921694820400288,
      "grad_norm": 1.1451246738433838,
      "learning_rate": 1.4704700314837967e-05,
      "loss": 0.07,
      "step": 114900
    },
    {
      "epoch": 1.692537665130116,
      "grad_norm": 1.3590253591537476,
      "learning_rate": 1.4703063942870983e-05,
      "loss": 0.0708,
      "step": 114925
    },
    {
      "epoch": 1.692905848220203,
      "grad_norm": 1.463120460510254,
      "learning_rate": 1.4701427570904e-05,
      "loss": 0.068,
      "step": 114950
    },
    {
      "epoch": 1.69327403131029,
      "grad_norm": 1.0935049057006836,
      "learning_rate": 1.4699791198937014e-05,
      "loss": 0.0754,
      "step": 114975
    },
    {
      "epoch": 1.693642214400377,
      "grad_norm": 1.171722412109375,
      "learning_rate": 1.469815482697003e-05,
      "loss": 0.0719,
      "step": 115000
    },
    {
      "epoch": 1.6940103974904641,
      "grad_norm": 2.0070700645446777,
      "learning_rate": 1.4696518455003043e-05,
      "loss": 0.0775,
      "step": 115025
    },
    {
      "epoch": 1.6943785805805511,
      "grad_norm": 1.0836679935455322,
      "learning_rate": 1.4694882083036059e-05,
      "loss": 0.0631,
      "step": 115050
    },
    {
      "epoch": 1.6947467636706381,
      "grad_norm": 1.061313509941101,
      "learning_rate": 1.4693245711069075e-05,
      "loss": 0.0709,
      "step": 115075
    },
    {
      "epoch": 1.6951149467607252,
      "grad_norm": 1.680694580078125,
      "learning_rate": 1.4691609339102091e-05,
      "loss": 0.0631,
      "step": 115100
    },
    {
      "epoch": 1.6954831298508122,
      "grad_norm": 1.333789348602295,
      "learning_rate": 1.4689972967135106e-05,
      "loss": 0.0697,
      "step": 115125
    },
    {
      "epoch": 1.6958513129408992,
      "grad_norm": 1.414815068244934,
      "learning_rate": 1.4688336595168122e-05,
      "loss": 0.066,
      "step": 115150
    },
    {
      "epoch": 1.6962194960309862,
      "grad_norm": 1.1994245052337646,
      "learning_rate": 1.4686700223201138e-05,
      "loss": 0.0718,
      "step": 115175
    },
    {
      "epoch": 1.6965876791210732,
      "grad_norm": 1.5072047710418701,
      "learning_rate": 1.4685063851234154e-05,
      "loss": 0.0746,
      "step": 115200
    },
    {
      "epoch": 1.6969558622111602,
      "grad_norm": 1.4426829814910889,
      "learning_rate": 1.4683427479267169e-05,
      "loss": 0.0732,
      "step": 115225
    },
    {
      "epoch": 1.6973240453012473,
      "grad_norm": 1.1986182928085327,
      "learning_rate": 1.4681791107300183e-05,
      "loss": 0.0651,
      "step": 115250
    },
    {
      "epoch": 1.6976922283913345,
      "grad_norm": 1.66432523727417,
      "learning_rate": 1.46801547353332e-05,
      "loss": 0.0768,
      "step": 115275
    },
    {
      "epoch": 1.6980604114814215,
      "grad_norm": 1.262149453163147,
      "learning_rate": 1.4678518363366214e-05,
      "loss": 0.0759,
      "step": 115300
    },
    {
      "epoch": 1.6984285945715085,
      "grad_norm": 1.1279561519622803,
      "learning_rate": 1.467688199139923e-05,
      "loss": 0.0751,
      "step": 115325
    },
    {
      "epoch": 1.6987967776615955,
      "grad_norm": 1.1017858982086182,
      "learning_rate": 1.4675245619432246e-05,
      "loss": 0.0631,
      "step": 115350
    },
    {
      "epoch": 1.6991649607516826,
      "grad_norm": 1.5601005554199219,
      "learning_rate": 1.467360924746526e-05,
      "loss": 0.0715,
      "step": 115375
    },
    {
      "epoch": 1.6995331438417698,
      "grad_norm": 1.4502965211868286,
      "learning_rate": 1.4671972875498277e-05,
      "loss": 0.0718,
      "step": 115400
    },
    {
      "epoch": 1.6999013269318568,
      "grad_norm": 1.2339082956314087,
      "learning_rate": 1.4670336503531293e-05,
      "loss": 0.0658,
      "step": 115425
    },
    {
      "epoch": 1.7002695100219438,
      "grad_norm": 1.4210634231567383,
      "learning_rate": 1.4668700131564306e-05,
      "loss": 0.071,
      "step": 115450
    },
    {
      "epoch": 1.7006376931120308,
      "grad_norm": 1.2773572206497192,
      "learning_rate": 1.4667063759597322e-05,
      "loss": 0.0694,
      "step": 115475
    },
    {
      "epoch": 1.7010058762021179,
      "grad_norm": 1.2864840030670166,
      "learning_rate": 1.4665427387630338e-05,
      "loss": 0.0742,
      "step": 115500
    },
    {
      "epoch": 1.7013740592922049,
      "grad_norm": 1.39313542842865,
      "learning_rate": 1.4663791015663354e-05,
      "loss": 0.0715,
      "step": 115525
    },
    {
      "epoch": 1.7017422423822919,
      "grad_norm": 1.2489782571792603,
      "learning_rate": 1.4662154643696369e-05,
      "loss": 0.0743,
      "step": 115550
    },
    {
      "epoch": 1.702110425472379,
      "grad_norm": 1.1297526359558105,
      "learning_rate": 1.4660518271729385e-05,
      "loss": 0.0738,
      "step": 115575
    },
    {
      "epoch": 1.702478608562466,
      "grad_norm": 1.6708920001983643,
      "learning_rate": 1.46588818997624e-05,
      "loss": 0.0774,
      "step": 115600
    },
    {
      "epoch": 1.702846791652553,
      "grad_norm": 1.5281175374984741,
      "learning_rate": 1.4657245527795417e-05,
      "loss": 0.0586,
      "step": 115625
    },
    {
      "epoch": 1.70321497474264,
      "grad_norm": 1.1840823888778687,
      "learning_rate": 1.4655609155828431e-05,
      "loss": 0.059,
      "step": 115650
    },
    {
      "epoch": 1.703583157832727,
      "grad_norm": 1.2361899614334106,
      "learning_rate": 1.4653972783861446e-05,
      "loss": 0.0681,
      "step": 115675
    },
    {
      "epoch": 1.703951340922814,
      "grad_norm": 1.485796570777893,
      "learning_rate": 1.465233641189446e-05,
      "loss": 0.0772,
      "step": 115700
    },
    {
      "epoch": 1.704319524012901,
      "grad_norm": 1.5074681043624878,
      "learning_rate": 1.4650700039927477e-05,
      "loss": 0.0713,
      "step": 115725
    },
    {
      "epoch": 1.7046877071029882,
      "grad_norm": 1.3141595125198364,
      "learning_rate": 1.4649063667960493e-05,
      "loss": 0.0773,
      "step": 115750
    },
    {
      "epoch": 1.7050558901930752,
      "grad_norm": 1.017120361328125,
      "learning_rate": 1.4647427295993509e-05,
      "loss": 0.0653,
      "step": 115775
    },
    {
      "epoch": 1.7054240732831623,
      "grad_norm": 2.0109574794769287,
      "learning_rate": 1.4645790924026523e-05,
      "loss": 0.0718,
      "step": 115800
    },
    {
      "epoch": 1.7057922563732493,
      "grad_norm": 1.1323426961898804,
      "learning_rate": 1.464415455205954e-05,
      "loss": 0.0702,
      "step": 115825
    },
    {
      "epoch": 1.7061604394633363,
      "grad_norm": 1.0215483903884888,
      "learning_rate": 1.4642518180092556e-05,
      "loss": 0.0668,
      "step": 115850
    },
    {
      "epoch": 1.7065286225534235,
      "grad_norm": 1.184708833694458,
      "learning_rate": 1.4640881808125568e-05,
      "loss": 0.0728,
      "step": 115875
    },
    {
      "epoch": 1.7068968056435105,
      "grad_norm": 1.4762532711029053,
      "learning_rate": 1.4639245436158584e-05,
      "loss": 0.0685,
      "step": 115900
    },
    {
      "epoch": 1.7072649887335976,
      "grad_norm": 0.849449098110199,
      "learning_rate": 1.46376090641916e-05,
      "loss": 0.07,
      "step": 115925
    },
    {
      "epoch": 1.7076331718236846,
      "grad_norm": 1.733501672744751,
      "learning_rate": 1.4635972692224615e-05,
      "loss": 0.0747,
      "step": 115950
    },
    {
      "epoch": 1.7080013549137716,
      "grad_norm": 1.7008428573608398,
      "learning_rate": 1.4634336320257631e-05,
      "loss": 0.0746,
      "step": 115975
    },
    {
      "epoch": 1.7083695380038586,
      "grad_norm": 1.203247308731079,
      "learning_rate": 1.4632699948290647e-05,
      "loss": 0.0759,
      "step": 116000
    },
    {
      "epoch": 1.7087377210939456,
      "grad_norm": 1.3310378789901733,
      "learning_rate": 1.4631063576323664e-05,
      "loss": 0.0689,
      "step": 116025
    },
    {
      "epoch": 1.7091059041840326,
      "grad_norm": 1.4498660564422607,
      "learning_rate": 1.4629427204356678e-05,
      "loss": 0.0683,
      "step": 116050
    },
    {
      "epoch": 1.7094740872741196,
      "grad_norm": 1.5515391826629639,
      "learning_rate": 1.4627790832389694e-05,
      "loss": 0.0703,
      "step": 116075
    },
    {
      "epoch": 1.7098422703642067,
      "grad_norm": 1.4486724138259888,
      "learning_rate": 1.4626154460422709e-05,
      "loss": 0.0663,
      "step": 116100
    },
    {
      "epoch": 1.7102104534542937,
      "grad_norm": 1.1945711374282837,
      "learning_rate": 1.4624518088455723e-05,
      "loss": 0.0691,
      "step": 116125
    },
    {
      "epoch": 1.7105786365443807,
      "grad_norm": 1.4841253757476807,
      "learning_rate": 1.462288171648874e-05,
      "loss": 0.0704,
      "step": 116150
    },
    {
      "epoch": 1.7109468196344677,
      "grad_norm": 1.2552186250686646,
      "learning_rate": 1.4621245344521755e-05,
      "loss": 0.0757,
      "step": 116175
    },
    {
      "epoch": 1.7113150027245547,
      "grad_norm": 1.0416038036346436,
      "learning_rate": 1.4619608972554772e-05,
      "loss": 0.0716,
      "step": 116200
    },
    {
      "epoch": 1.7116831858146417,
      "grad_norm": 1.1674039363861084,
      "learning_rate": 1.4617972600587786e-05,
      "loss": 0.0712,
      "step": 116225
    },
    {
      "epoch": 1.712051368904729,
      "grad_norm": 1.4191275835037231,
      "learning_rate": 1.4616336228620802e-05,
      "loss": 0.0711,
      "step": 116250
    },
    {
      "epoch": 1.712419551994816,
      "grad_norm": 1.4573978185653687,
      "learning_rate": 1.4614699856653818e-05,
      "loss": 0.0711,
      "step": 116275
    },
    {
      "epoch": 1.712787735084903,
      "grad_norm": 0.7501809597015381,
      "learning_rate": 1.4613063484686831e-05,
      "loss": 0.065,
      "step": 116300
    },
    {
      "epoch": 1.71315591817499,
      "grad_norm": 1.1468013525009155,
      "learning_rate": 1.4611427112719847e-05,
      "loss": 0.0708,
      "step": 116325
    },
    {
      "epoch": 1.7135241012650773,
      "grad_norm": 1.7565195560455322,
      "learning_rate": 1.4609790740752863e-05,
      "loss": 0.0685,
      "step": 116350
    },
    {
      "epoch": 1.7138922843551643,
      "grad_norm": 1.2514381408691406,
      "learning_rate": 1.4608154368785878e-05,
      "loss": 0.0656,
      "step": 116375
    },
    {
      "epoch": 1.7142604674452513,
      "grad_norm": 1.3440977334976196,
      "learning_rate": 1.4606517996818894e-05,
      "loss": 0.0675,
      "step": 116400
    },
    {
      "epoch": 1.7146286505353383,
      "grad_norm": 1.4262274503707886,
      "learning_rate": 1.460488162485191e-05,
      "loss": 0.0674,
      "step": 116425
    },
    {
      "epoch": 1.7149968336254253,
      "grad_norm": 1.433107614517212,
      "learning_rate": 1.4603245252884926e-05,
      "loss": 0.0706,
      "step": 116450
    },
    {
      "epoch": 1.7153650167155123,
      "grad_norm": 1.055009126663208,
      "learning_rate": 1.460160888091794e-05,
      "loss": 0.0737,
      "step": 116475
    },
    {
      "epoch": 1.7157331998055994,
      "grad_norm": 1.516435980796814,
      "learning_rate": 1.4599972508950955e-05,
      "loss": 0.0684,
      "step": 116500
    },
    {
      "epoch": 1.7161013828956864,
      "grad_norm": 1.2547600269317627,
      "learning_rate": 1.459833613698397e-05,
      "loss": 0.0715,
      "step": 116525
    },
    {
      "epoch": 1.7164695659857734,
      "grad_norm": 1.6059954166412354,
      "learning_rate": 1.4596699765016986e-05,
      "loss": 0.0666,
      "step": 116550
    },
    {
      "epoch": 1.7168377490758604,
      "grad_norm": 1.5357228517532349,
      "learning_rate": 1.4595063393050002e-05,
      "loss": 0.0764,
      "step": 116575
    },
    {
      "epoch": 1.7172059321659474,
      "grad_norm": 1.314332127571106,
      "learning_rate": 1.4593427021083018e-05,
      "loss": 0.0745,
      "step": 116600
    },
    {
      "epoch": 1.7175741152560344,
      "grad_norm": 1.4616749286651611,
      "learning_rate": 1.4591790649116033e-05,
      "loss": 0.0719,
      "step": 116625
    },
    {
      "epoch": 1.7179422983461214,
      "grad_norm": 1.48330557346344,
      "learning_rate": 1.4590154277149049e-05,
      "loss": 0.0741,
      "step": 116650
    },
    {
      "epoch": 1.7183104814362085,
      "grad_norm": 1.3130346536636353,
      "learning_rate": 1.4588517905182065e-05,
      "loss": 0.0674,
      "step": 116675
    },
    {
      "epoch": 1.7186786645262955,
      "grad_norm": 1.320476770401001,
      "learning_rate": 1.4586881533215081e-05,
      "loss": 0.0703,
      "step": 116700
    },
    {
      "epoch": 1.7190468476163827,
      "grad_norm": 1.2791556119918823,
      "learning_rate": 1.4585245161248094e-05,
      "loss": 0.0775,
      "step": 116725
    },
    {
      "epoch": 1.7194150307064697,
      "grad_norm": 1.0319461822509766,
      "learning_rate": 1.458360878928111e-05,
      "loss": 0.0685,
      "step": 116750
    },
    {
      "epoch": 1.7197832137965567,
      "grad_norm": 1.420870065689087,
      "learning_rate": 1.4581972417314126e-05,
      "loss": 0.0643,
      "step": 116775
    },
    {
      "epoch": 1.7201513968866438,
      "grad_norm": 0.9832098484039307,
      "learning_rate": 1.458033604534714e-05,
      "loss": 0.0701,
      "step": 116800
    },
    {
      "epoch": 1.720519579976731,
      "grad_norm": 1.3880716562271118,
      "learning_rate": 1.4578699673380157e-05,
      "loss": 0.0647,
      "step": 116825
    },
    {
      "epoch": 1.720887763066818,
      "grad_norm": 1.3137727975845337,
      "learning_rate": 1.4577063301413173e-05,
      "loss": 0.0664,
      "step": 116850
    },
    {
      "epoch": 1.721255946156905,
      "grad_norm": 1.3756775856018066,
      "learning_rate": 1.4575426929446187e-05,
      "loss": 0.0693,
      "step": 116875
    },
    {
      "epoch": 1.721624129246992,
      "grad_norm": 1.3043729066848755,
      "learning_rate": 1.4573790557479203e-05,
      "loss": 0.068,
      "step": 116900
    },
    {
      "epoch": 1.721992312337079,
      "grad_norm": 0.9949998259544373,
      "learning_rate": 1.4572154185512218e-05,
      "loss": 0.0726,
      "step": 116925
    },
    {
      "epoch": 1.722360495427166,
      "grad_norm": 1.6280750036239624,
      "learning_rate": 1.4570517813545232e-05,
      "loss": 0.0674,
      "step": 116950
    },
    {
      "epoch": 1.722728678517253,
      "grad_norm": 1.2649887800216675,
      "learning_rate": 1.4568881441578249e-05,
      "loss": 0.0676,
      "step": 116975
    },
    {
      "epoch": 1.72309686160734,
      "grad_norm": 1.6157060861587524,
      "learning_rate": 1.4567245069611265e-05,
      "loss": 0.0729,
      "step": 117000
    },
    {
      "epoch": 1.7234650446974271,
      "grad_norm": 1.3674312829971313,
      "learning_rate": 1.456560869764428e-05,
      "loss": 0.0677,
      "step": 117025
    },
    {
      "epoch": 1.7238332277875141,
      "grad_norm": 1.454086184501648,
      "learning_rate": 1.4563972325677295e-05,
      "loss": 0.0672,
      "step": 117050
    },
    {
      "epoch": 1.7242014108776011,
      "grad_norm": 1.434951663017273,
      "learning_rate": 1.4562335953710311e-05,
      "loss": 0.0694,
      "step": 117075
    },
    {
      "epoch": 1.7245695939676882,
      "grad_norm": 1.2908538579940796,
      "learning_rate": 1.4560699581743328e-05,
      "loss": 0.0654,
      "step": 117100
    },
    {
      "epoch": 1.7249377770577752,
      "grad_norm": 1.363983392715454,
      "learning_rate": 1.4559128664655021e-05,
      "loss": 0.0723,
      "step": 117125
    },
    {
      "epoch": 1.7253059601478622,
      "grad_norm": 1.0666635036468506,
      "learning_rate": 1.4557492292688036e-05,
      "loss": 0.0667,
      "step": 117150
    },
    {
      "epoch": 1.7256741432379492,
      "grad_norm": 1.223769187927246,
      "learning_rate": 1.4555855920721052e-05,
      "loss": 0.0632,
      "step": 117175
    },
    {
      "epoch": 1.7260423263280364,
      "grad_norm": 1.1432114839553833,
      "learning_rate": 1.4554219548754067e-05,
      "loss": 0.0716,
      "step": 117200
    },
    {
      "epoch": 1.7264105094181235,
      "grad_norm": 0.8071144819259644,
      "learning_rate": 1.4552583176787083e-05,
      "loss": 0.0751,
      "step": 117225
    },
    {
      "epoch": 1.7267786925082105,
      "grad_norm": 1.6455711126327515,
      "learning_rate": 1.4550946804820099e-05,
      "loss": 0.0653,
      "step": 117250
    },
    {
      "epoch": 1.7271468755982975,
      "grad_norm": 1.1311428546905518,
      "learning_rate": 1.4549310432853115e-05,
      "loss": 0.066,
      "step": 117275
    },
    {
      "epoch": 1.7275150586883845,
      "grad_norm": 1.388342261314392,
      "learning_rate": 1.454767406088613e-05,
      "loss": 0.0651,
      "step": 117300
    },
    {
      "epoch": 1.7278832417784717,
      "grad_norm": 1.2155027389526367,
      "learning_rate": 1.4546037688919146e-05,
      "loss": 0.0728,
      "step": 117325
    },
    {
      "epoch": 1.7282514248685588,
      "grad_norm": 1.494720697402954,
      "learning_rate": 1.4544401316952158e-05,
      "loss": 0.0783,
      "step": 117350
    },
    {
      "epoch": 1.7286196079586458,
      "grad_norm": 1.4385688304901123,
      "learning_rate": 1.4542764944985175e-05,
      "loss": 0.067,
      "step": 117375
    },
    {
      "epoch": 1.7289877910487328,
      "grad_norm": 1.5436705350875854,
      "learning_rate": 1.454112857301819e-05,
      "loss": 0.0753,
      "step": 117400
    },
    {
      "epoch": 1.7293559741388198,
      "grad_norm": 1.8978846073150635,
      "learning_rate": 1.4539492201051207e-05,
      "loss": 0.0686,
      "step": 117425
    },
    {
      "epoch": 1.7297241572289068,
      "grad_norm": 1.692796230316162,
      "learning_rate": 1.4537855829084221e-05,
      "loss": 0.071,
      "step": 117450
    },
    {
      "epoch": 1.7300923403189938,
      "grad_norm": 1.5229357481002808,
      "learning_rate": 1.4536219457117237e-05,
      "loss": 0.0704,
      "step": 117475
    },
    {
      "epoch": 1.7304605234090809,
      "grad_norm": 1.0635727643966675,
      "learning_rate": 1.4534583085150254e-05,
      "loss": 0.0653,
      "step": 117500
    },
    {
      "epoch": 1.7308287064991679,
      "grad_norm": 1.2229571342468262,
      "learning_rate": 1.453294671318327e-05,
      "loss": 0.0674,
      "step": 117525
    },
    {
      "epoch": 1.7311968895892549,
      "grad_norm": 1.3343393802642822,
      "learning_rate": 1.4531310341216284e-05,
      "loss": 0.0666,
      "step": 117550
    },
    {
      "epoch": 1.731565072679342,
      "grad_norm": 1.3549617528915405,
      "learning_rate": 1.4529673969249299e-05,
      "loss": 0.072,
      "step": 117575
    },
    {
      "epoch": 1.731933255769429,
      "grad_norm": 1.5496704578399658,
      "learning_rate": 1.4528037597282315e-05,
      "loss": 0.0718,
      "step": 117600
    },
    {
      "epoch": 1.732301438859516,
      "grad_norm": 1.4900168180465698,
      "learning_rate": 1.452640122531533e-05,
      "loss": 0.0713,
      "step": 117625
    },
    {
      "epoch": 1.732669621949603,
      "grad_norm": 1.6626604795455933,
      "learning_rate": 1.4524764853348345e-05,
      "loss": 0.0647,
      "step": 117650
    },
    {
      "epoch": 1.7330378050396902,
      "grad_norm": 1.471441626548767,
      "learning_rate": 1.4523128481381362e-05,
      "loss": 0.0656,
      "step": 117675
    },
    {
      "epoch": 1.7334059881297772,
      "grad_norm": 1.0678582191467285,
      "learning_rate": 1.4521492109414376e-05,
      "loss": 0.0719,
      "step": 117700
    },
    {
      "epoch": 1.7337741712198642,
      "grad_norm": 1.842948317527771,
      "learning_rate": 1.4519855737447392e-05,
      "loss": 0.0672,
      "step": 117725
    },
    {
      "epoch": 1.7341423543099512,
      "grad_norm": 1.5209269523620605,
      "learning_rate": 1.4518219365480408e-05,
      "loss": 0.0686,
      "step": 117750
    },
    {
      "epoch": 1.7345105374000382,
      "grad_norm": 1.6039365530014038,
      "learning_rate": 1.4516582993513421e-05,
      "loss": 0.0703,
      "step": 117775
    },
    {
      "epoch": 1.7348787204901255,
      "grad_norm": 1.14345383644104,
      "learning_rate": 1.4514946621546437e-05,
      "loss": 0.0722,
      "step": 117800
    },
    {
      "epoch": 1.7352469035802125,
      "grad_norm": 1.0637685060501099,
      "learning_rate": 1.4513310249579453e-05,
      "loss": 0.0661,
      "step": 117825
    },
    {
      "epoch": 1.7356150866702995,
      "grad_norm": 1.4392364025115967,
      "learning_rate": 1.451167387761247e-05,
      "loss": 0.0715,
      "step": 117850
    },
    {
      "epoch": 1.7359832697603865,
      "grad_norm": 1.082470417022705,
      "learning_rate": 1.4510037505645484e-05,
      "loss": 0.0691,
      "step": 117875
    },
    {
      "epoch": 1.7363514528504735,
      "grad_norm": 1.4789104461669922,
      "learning_rate": 1.45084011336785e-05,
      "loss": 0.0735,
      "step": 117900
    },
    {
      "epoch": 1.7367196359405606,
      "grad_norm": 1.1306986808776855,
      "learning_rate": 1.4506764761711516e-05,
      "loss": 0.0677,
      "step": 117925
    },
    {
      "epoch": 1.7370878190306476,
      "grad_norm": 1.4292935132980347,
      "learning_rate": 1.4505128389744532e-05,
      "loss": 0.0735,
      "step": 117950
    },
    {
      "epoch": 1.7374560021207346,
      "grad_norm": 1.753593921661377,
      "learning_rate": 1.4503492017777547e-05,
      "loss": 0.0683,
      "step": 117975
    },
    {
      "epoch": 1.7378241852108216,
      "grad_norm": 1.2248762845993042,
      "learning_rate": 1.4501855645810561e-05,
      "loss": 0.06,
      "step": 118000
    },
    {
      "epoch": 1.7381923683009086,
      "grad_norm": 0.9938359260559082,
      "learning_rate": 1.4500219273843576e-05,
      "loss": 0.0698,
      "step": 118025
    },
    {
      "epoch": 1.7385605513909956,
      "grad_norm": 1.7148356437683105,
      "learning_rate": 1.4498582901876592e-05,
      "loss": 0.077,
      "step": 118050
    },
    {
      "epoch": 1.7389287344810826,
      "grad_norm": 1.6144236326217651,
      "learning_rate": 1.4496946529909608e-05,
      "loss": 0.0653,
      "step": 118075
    },
    {
      "epoch": 1.7392969175711697,
      "grad_norm": 1.386655330657959,
      "learning_rate": 1.4495310157942624e-05,
      "loss": 0.0653,
      "step": 118100
    },
    {
      "epoch": 1.7396651006612567,
      "grad_norm": 1.6300657987594604,
      "learning_rate": 1.4493673785975639e-05,
      "loss": 0.0708,
      "step": 118125
    },
    {
      "epoch": 1.740033283751344,
      "grad_norm": 1.2578023672103882,
      "learning_rate": 1.4492037414008655e-05,
      "loss": 0.0687,
      "step": 118150
    },
    {
      "epoch": 1.740401466841431,
      "grad_norm": 1.3382031917572021,
      "learning_rate": 1.4490401042041671e-05,
      "loss": 0.0671,
      "step": 118175
    },
    {
      "epoch": 1.740769649931518,
      "grad_norm": 1.3886079788208008,
      "learning_rate": 1.4488764670074684e-05,
      "loss": 0.069,
      "step": 118200
    },
    {
      "epoch": 1.741137833021605,
      "grad_norm": 1.4775012731552124,
      "learning_rate": 1.44871282981077e-05,
      "loss": 0.076,
      "step": 118225
    },
    {
      "epoch": 1.741506016111692,
      "grad_norm": 1.1081944704055786,
      "learning_rate": 1.4485491926140716e-05,
      "loss": 0.0627,
      "step": 118250
    },
    {
      "epoch": 1.7418741992017792,
      "grad_norm": 1.001936435699463,
      "learning_rate": 1.448385555417373e-05,
      "loss": 0.0683,
      "step": 118275
    },
    {
      "epoch": 1.7422423822918662,
      "grad_norm": 1.6779181957244873,
      "learning_rate": 1.4482219182206747e-05,
      "loss": 0.0661,
      "step": 118300
    },
    {
      "epoch": 1.7426105653819532,
      "grad_norm": 1.1340508460998535,
      "learning_rate": 1.4480582810239763e-05,
      "loss": 0.0695,
      "step": 118325
    },
    {
      "epoch": 1.7429787484720403,
      "grad_norm": 1.366167426109314,
      "learning_rate": 1.4478946438272779e-05,
      "loss": 0.072,
      "step": 118350
    },
    {
      "epoch": 1.7433469315621273,
      "grad_norm": 1.2603939771652222,
      "learning_rate": 1.4477310066305793e-05,
      "loss": 0.0648,
      "step": 118375
    },
    {
      "epoch": 1.7437151146522143,
      "grad_norm": 1.5625187158584595,
      "learning_rate": 1.4475673694338808e-05,
      "loss": 0.0742,
      "step": 118400
    },
    {
      "epoch": 1.7440832977423013,
      "grad_norm": 2.004777431488037,
      "learning_rate": 1.4474037322371824e-05,
      "loss": 0.079,
      "step": 118425
    },
    {
      "epoch": 1.7444514808323883,
      "grad_norm": 1.3128420114517212,
      "learning_rate": 1.4472400950404839e-05,
      "loss": 0.0697,
      "step": 118450
    },
    {
      "epoch": 1.7448196639224753,
      "grad_norm": 1.628818392753601,
      "learning_rate": 1.4470764578437855e-05,
      "loss": 0.0746,
      "step": 118475
    },
    {
      "epoch": 1.7451878470125624,
      "grad_norm": 1.5265029668807983,
      "learning_rate": 1.446912820647087e-05,
      "loss": 0.0701,
      "step": 118500
    },
    {
      "epoch": 1.7455560301026494,
      "grad_norm": 0.6551544070243835,
      "learning_rate": 1.4467491834503887e-05,
      "loss": 0.0725,
      "step": 118525
    },
    {
      "epoch": 1.7459242131927364,
      "grad_norm": 1.2299360036849976,
      "learning_rate": 1.4465855462536901e-05,
      "loss": 0.0709,
      "step": 118550
    },
    {
      "epoch": 1.7462923962828234,
      "grad_norm": 1.4879930019378662,
      "learning_rate": 1.4464219090569918e-05,
      "loss": 0.0678,
      "step": 118575
    },
    {
      "epoch": 1.7466605793729104,
      "grad_norm": 1.5581705570220947,
      "learning_rate": 1.4462582718602934e-05,
      "loss": 0.065,
      "step": 118600
    },
    {
      "epoch": 1.7470287624629977,
      "grad_norm": 0.8866927623748779,
      "learning_rate": 1.4460946346635947e-05,
      "loss": 0.0649,
      "step": 118625
    },
    {
      "epoch": 1.7473969455530847,
      "grad_norm": 1.5363082885742188,
      "learning_rate": 1.4459309974668963e-05,
      "loss": 0.0638,
      "step": 118650
    },
    {
      "epoch": 1.7477651286431717,
      "grad_norm": 0.9323269128799438,
      "learning_rate": 1.4457673602701979e-05,
      "loss": 0.0738,
      "step": 118675
    },
    {
      "epoch": 1.7481333117332587,
      "grad_norm": 1.2126998901367188,
      "learning_rate": 1.4456037230734993e-05,
      "loss": 0.0699,
      "step": 118700
    },
    {
      "epoch": 1.7485014948233457,
      "grad_norm": 1.1231118440628052,
      "learning_rate": 1.445440085876801e-05,
      "loss": 0.0572,
      "step": 118725
    },
    {
      "epoch": 1.748869677913433,
      "grad_norm": 1.0999683141708374,
      "learning_rate": 1.4452764486801026e-05,
      "loss": 0.0751,
      "step": 118750
    },
    {
      "epoch": 1.74923786100352,
      "grad_norm": 1.5710248947143555,
      "learning_rate": 1.4451128114834042e-05,
      "loss": 0.0712,
      "step": 118775
    },
    {
      "epoch": 1.749606044093607,
      "grad_norm": 1.6222867965698242,
      "learning_rate": 1.4449491742867056e-05,
      "loss": 0.0709,
      "step": 118800
    },
    {
      "epoch": 1.749974227183694,
      "grad_norm": 1.2778717279434204,
      "learning_rate": 1.444785537090007e-05,
      "loss": 0.0717,
      "step": 118825
    },
    {
      "epoch": 1.750342410273781,
      "grad_norm": 0.9944702982902527,
      "learning_rate": 1.4446218998933085e-05,
      "loss": 0.0659,
      "step": 118850
    },
    {
      "epoch": 1.750710593363868,
      "grad_norm": 1.477279543876648,
      "learning_rate": 1.4444582626966101e-05,
      "loss": 0.0701,
      "step": 118875
    },
    {
      "epoch": 1.751078776453955,
      "grad_norm": 1.252055048942566,
      "learning_rate": 1.4442946254999117e-05,
      "loss": 0.0696,
      "step": 118900
    },
    {
      "epoch": 1.751446959544042,
      "grad_norm": 1.0778433084487915,
      "learning_rate": 1.4441309883032134e-05,
      "loss": 0.0674,
      "step": 118925
    },
    {
      "epoch": 1.751815142634129,
      "grad_norm": 0.9636226892471313,
      "learning_rate": 1.4439673511065148e-05,
      "loss": 0.073,
      "step": 118950
    },
    {
      "epoch": 1.752183325724216,
      "grad_norm": 1.8021533489227295,
      "learning_rate": 1.4438037139098164e-05,
      "loss": 0.07,
      "step": 118975
    },
    {
      "epoch": 1.752551508814303,
      "grad_norm": 1.2711902856826782,
      "learning_rate": 1.443640076713118e-05,
      "loss": 0.0678,
      "step": 119000
    },
    {
      "epoch": 1.7529196919043901,
      "grad_norm": 1.6635637283325195,
      "learning_rate": 1.4434764395164196e-05,
      "loss": 0.0715,
      "step": 119025
    },
    {
      "epoch": 1.7532878749944771,
      "grad_norm": 1.0828711986541748,
      "learning_rate": 1.443312802319721e-05,
      "loss": 0.0668,
      "step": 119050
    },
    {
      "epoch": 1.7536560580845642,
      "grad_norm": 0.9765989780426025,
      "learning_rate": 1.4431491651230225e-05,
      "loss": 0.0689,
      "step": 119075
    },
    {
      "epoch": 1.7540242411746512,
      "grad_norm": 1.7505444288253784,
      "learning_rate": 1.4429855279263242e-05,
      "loss": 0.0739,
      "step": 119100
    },
    {
      "epoch": 1.7543924242647384,
      "grad_norm": 1.2258647680282593,
      "learning_rate": 1.4428218907296256e-05,
      "loss": 0.0664,
      "step": 119125
    },
    {
      "epoch": 1.7547606073548254,
      "grad_norm": 1.4681923389434814,
      "learning_rate": 1.4426582535329272e-05,
      "loss": 0.0723,
      "step": 119150
    },
    {
      "epoch": 1.7551287904449124,
      "grad_norm": 1.11790931224823,
      "learning_rate": 1.4425011618240968e-05,
      "loss": 0.0716,
      "step": 119175
    },
    {
      "epoch": 1.7554969735349994,
      "grad_norm": 1.5166959762573242,
      "learning_rate": 1.4423375246273982e-05,
      "loss": 0.072,
      "step": 119200
    },
    {
      "epoch": 1.7558651566250867,
      "grad_norm": 1.3520179986953735,
      "learning_rate": 1.4421738874306998e-05,
      "loss": 0.073,
      "step": 119225
    },
    {
      "epoch": 1.7562333397151737,
      "grad_norm": 1.4081478118896484,
      "learning_rate": 1.4420102502340013e-05,
      "loss": 0.0683,
      "step": 119250
    },
    {
      "epoch": 1.7566015228052607,
      "grad_norm": 1.392702579498291,
      "learning_rate": 1.4418466130373027e-05,
      "loss": 0.0705,
      "step": 119275
    },
    {
      "epoch": 1.7569697058953477,
      "grad_norm": 0.953191339969635,
      "learning_rate": 1.4416829758406043e-05,
      "loss": 0.0694,
      "step": 119300
    },
    {
      "epoch": 1.7573378889854347,
      "grad_norm": 1.3729522228240967,
      "learning_rate": 1.441519338643906e-05,
      "loss": 0.0788,
      "step": 119325
    },
    {
      "epoch": 1.7577060720755218,
      "grad_norm": 2.266904592514038,
      "learning_rate": 1.4413557014472076e-05,
      "loss": 0.0714,
      "step": 119350
    },
    {
      "epoch": 1.7580742551656088,
      "grad_norm": 1.3887821435928345,
      "learning_rate": 1.441192064250509e-05,
      "loss": 0.0694,
      "step": 119375
    },
    {
      "epoch": 1.7584424382556958,
      "grad_norm": 1.1650502681732178,
      "learning_rate": 1.4410284270538106e-05,
      "loss": 0.0687,
      "step": 119400
    },
    {
      "epoch": 1.7588106213457828,
      "grad_norm": 1.1449217796325684,
      "learning_rate": 1.4408647898571122e-05,
      "loss": 0.0652,
      "step": 119425
    },
    {
      "epoch": 1.7591788044358698,
      "grad_norm": 1.301220417022705,
      "learning_rate": 1.4407011526604137e-05,
      "loss": 0.072,
      "step": 119450
    },
    {
      "epoch": 1.7595469875259568,
      "grad_norm": 1.4718124866485596,
      "learning_rate": 1.4405375154637151e-05,
      "loss": 0.0722,
      "step": 119475
    },
    {
      "epoch": 1.7599151706160439,
      "grad_norm": 1.2857835292816162,
      "learning_rate": 1.4403738782670168e-05,
      "loss": 0.0708,
      "step": 119500
    },
    {
      "epoch": 1.7602833537061309,
      "grad_norm": 1.3989564180374146,
      "learning_rate": 1.4402102410703182e-05,
      "loss": 0.0667,
      "step": 119525
    },
    {
      "epoch": 1.7606515367962179,
      "grad_norm": 1.7022169828414917,
      "learning_rate": 1.4400466038736198e-05,
      "loss": 0.07,
      "step": 119550
    },
    {
      "epoch": 1.761019719886305,
      "grad_norm": 1.1868810653686523,
      "learning_rate": 1.4398829666769214e-05,
      "loss": 0.0639,
      "step": 119575
    },
    {
      "epoch": 1.7613879029763921,
      "grad_norm": 1.8378244638442993,
      "learning_rate": 1.439719329480223e-05,
      "loss": 0.0797,
      "step": 119600
    },
    {
      "epoch": 1.7617560860664792,
      "grad_norm": 1.4175103902816772,
      "learning_rate": 1.4395556922835245e-05,
      "loss": 0.0663,
      "step": 119625
    },
    {
      "epoch": 1.7621242691565662,
      "grad_norm": 1.5685625076293945,
      "learning_rate": 1.4393920550868261e-05,
      "loss": 0.0708,
      "step": 119650
    },
    {
      "epoch": 1.7624924522466532,
      "grad_norm": 1.862176537513733,
      "learning_rate": 1.4392284178901274e-05,
      "loss": 0.0652,
      "step": 119675
    },
    {
      "epoch": 1.7628606353367402,
      "grad_norm": 1.4839879274368286,
      "learning_rate": 1.439064780693429e-05,
      "loss": 0.0619,
      "step": 119700
    },
    {
      "epoch": 1.7632288184268274,
      "grad_norm": 1.0859131813049316,
      "learning_rate": 1.4389011434967306e-05,
      "loss": 0.0723,
      "step": 119725
    },
    {
      "epoch": 1.7635970015169145,
      "grad_norm": 1.548510193824768,
      "learning_rate": 1.4387375063000322e-05,
      "loss": 0.0731,
      "step": 119750
    },
    {
      "epoch": 1.7639651846070015,
      "grad_norm": 1.6976302862167358,
      "learning_rate": 1.4385738691033337e-05,
      "loss": 0.0756,
      "step": 119775
    },
    {
      "epoch": 1.7643333676970885,
      "grad_norm": 1.256464958190918,
      "learning_rate": 1.4384102319066353e-05,
      "loss": 0.0765,
      "step": 119800
    },
    {
      "epoch": 1.7647015507871755,
      "grad_norm": 1.97874915599823,
      "learning_rate": 1.4382465947099369e-05,
      "loss": 0.0729,
      "step": 119825
    },
    {
      "epoch": 1.7650697338772625,
      "grad_norm": 1.1645824909210205,
      "learning_rate": 1.4380829575132385e-05,
      "loss": 0.0745,
      "step": 119850
    },
    {
      "epoch": 1.7654379169673495,
      "grad_norm": 1.3096339702606201,
      "learning_rate": 1.4379193203165398e-05,
      "loss": 0.0643,
      "step": 119875
    },
    {
      "epoch": 1.7658061000574365,
      "grad_norm": 1.7846611738204956,
      "learning_rate": 1.4377556831198414e-05,
      "loss": 0.0763,
      "step": 119900
    },
    {
      "epoch": 1.7661742831475236,
      "grad_norm": 1.5958901643753052,
      "learning_rate": 1.437592045923143e-05,
      "loss": 0.0704,
      "step": 119925
    },
    {
      "epoch": 1.7665424662376106,
      "grad_norm": 1.6560559272766113,
      "learning_rate": 1.4374284087264445e-05,
      "loss": 0.075,
      "step": 119950
    },
    {
      "epoch": 1.7669106493276976,
      "grad_norm": 1.3931108713150024,
      "learning_rate": 1.437264771529746e-05,
      "loss": 0.0731,
      "step": 119975
    },
    {
      "epoch": 1.7672788324177846,
      "grad_norm": 1.3844809532165527,
      "learning_rate": 1.4371011343330477e-05,
      "loss": 0.0691,
      "step": 120000
    },
    {
      "epoch": 1.7676470155078716,
      "grad_norm": 1.3142585754394531,
      "learning_rate": 1.4369374971363491e-05,
      "loss": 0.0747,
      "step": 120025
    },
    {
      "epoch": 1.7680151985979586,
      "grad_norm": 1.5986624956130981,
      "learning_rate": 1.4367738599396508e-05,
      "loss": 0.0675,
      "step": 120050
    },
    {
      "epoch": 1.7683833816880459,
      "grad_norm": 1.2626944780349731,
      "learning_rate": 1.4366102227429524e-05,
      "loss": 0.0668,
      "step": 120075
    },
    {
      "epoch": 1.768751564778133,
      "grad_norm": 1.6989572048187256,
      "learning_rate": 1.4364465855462537e-05,
      "loss": 0.0759,
      "step": 120100
    },
    {
      "epoch": 1.76911974786822,
      "grad_norm": 0.8778809905052185,
      "learning_rate": 1.4362829483495553e-05,
      "loss": 0.0671,
      "step": 120125
    },
    {
      "epoch": 1.769487930958307,
      "grad_norm": 1.777380347251892,
      "learning_rate": 1.4361193111528569e-05,
      "loss": 0.0738,
      "step": 120150
    },
    {
      "epoch": 1.769856114048394,
      "grad_norm": 1.3435474634170532,
      "learning_rate": 1.4359556739561585e-05,
      "loss": 0.0719,
      "step": 120175
    },
    {
      "epoch": 1.7702242971384812,
      "grad_norm": 1.1195390224456787,
      "learning_rate": 1.43579203675946e-05,
      "loss": 0.0624,
      "step": 120200
    },
    {
      "epoch": 1.7705924802285682,
      "grad_norm": 1.039779782295227,
      "learning_rate": 1.4356283995627616e-05,
      "loss": 0.0708,
      "step": 120225
    },
    {
      "epoch": 1.7709606633186552,
      "grad_norm": 1.2294856309890747,
      "learning_rate": 1.4354647623660632e-05,
      "loss": 0.0651,
      "step": 120250
    },
    {
      "epoch": 1.7713288464087422,
      "grad_norm": 1.1152254343032837,
      "learning_rate": 1.4353011251693648e-05,
      "loss": 0.0607,
      "step": 120275
    },
    {
      "epoch": 1.7716970294988292,
      "grad_norm": 1.0624229907989502,
      "learning_rate": 1.435137487972666e-05,
      "loss": 0.0706,
      "step": 120300
    },
    {
      "epoch": 1.7720652125889162,
      "grad_norm": 1.59226655960083,
      "learning_rate": 1.4349738507759677e-05,
      "loss": 0.079,
      "step": 120325
    },
    {
      "epoch": 1.7724333956790033,
      "grad_norm": 1.6993451118469238,
      "learning_rate": 1.4348102135792691e-05,
      "loss": 0.0749,
      "step": 120350
    },
    {
      "epoch": 1.7728015787690903,
      "grad_norm": 0.9813623428344727,
      "learning_rate": 1.4346465763825707e-05,
      "loss": 0.0642,
      "step": 120375
    },
    {
      "epoch": 1.7731697618591773,
      "grad_norm": 0.764958918094635,
      "learning_rate": 1.4344829391858724e-05,
      "loss": 0.069,
      "step": 120400
    },
    {
      "epoch": 1.7735379449492643,
      "grad_norm": 1.2145322561264038,
      "learning_rate": 1.434319301989174e-05,
      "loss": 0.0653,
      "step": 120425
    },
    {
      "epoch": 1.7739061280393513,
      "grad_norm": 1.635727047920227,
      "learning_rate": 1.4341556647924754e-05,
      "loss": 0.0751,
      "step": 120450
    },
    {
      "epoch": 1.7742743111294383,
      "grad_norm": 1.4147942066192627,
      "learning_rate": 1.433992027595777e-05,
      "loss": 0.0629,
      "step": 120475
    },
    {
      "epoch": 1.7746424942195254,
      "grad_norm": 1.3779675960540771,
      "learning_rate": 1.4338283903990786e-05,
      "loss": 0.073,
      "step": 120500
    },
    {
      "epoch": 1.7750106773096124,
      "grad_norm": 1.3990569114685059,
      "learning_rate": 1.43366475320238e-05,
      "loss": 0.0698,
      "step": 120525
    },
    {
      "epoch": 1.7753788603996996,
      "grad_norm": 1.5038468837738037,
      "learning_rate": 1.4335011160056815e-05,
      "loss": 0.0607,
      "step": 120550
    },
    {
      "epoch": 1.7757470434897866,
      "grad_norm": 1.332115650177002,
      "learning_rate": 1.4333374788089832e-05,
      "loss": 0.0636,
      "step": 120575
    },
    {
      "epoch": 1.7761152265798736,
      "grad_norm": 1.2773898839950562,
      "learning_rate": 1.4331738416122846e-05,
      "loss": 0.0738,
      "step": 120600
    },
    {
      "epoch": 1.7764834096699607,
      "grad_norm": 1.379075288772583,
      "learning_rate": 1.4330102044155862e-05,
      "loss": 0.068,
      "step": 120625
    },
    {
      "epoch": 1.7768515927600477,
      "grad_norm": 1.3863091468811035,
      "learning_rate": 1.4328465672188878e-05,
      "loss": 0.0692,
      "step": 120650
    },
    {
      "epoch": 1.777219775850135,
      "grad_norm": 1.0513231754302979,
      "learning_rate": 1.4326829300221894e-05,
      "loss": 0.0651,
      "step": 120675
    },
    {
      "epoch": 1.777587958940222,
      "grad_norm": 1.1663014888763428,
      "learning_rate": 1.4325192928254909e-05,
      "loss": 0.0747,
      "step": 120700
    },
    {
      "epoch": 1.777956142030309,
      "grad_norm": 1.2848808765411377,
      "learning_rate": 1.4323556556287923e-05,
      "loss": 0.0664,
      "step": 120725
    },
    {
      "epoch": 1.778324325120396,
      "grad_norm": 1.8034509420394897,
      "learning_rate": 1.432192018432094e-05,
      "loss": 0.0698,
      "step": 120750
    },
    {
      "epoch": 1.778692508210483,
      "grad_norm": 1.3875097036361694,
      "learning_rate": 1.4320283812353954e-05,
      "loss": 0.0751,
      "step": 120775
    },
    {
      "epoch": 1.77906069130057,
      "grad_norm": 1.251055121421814,
      "learning_rate": 1.431864744038697e-05,
      "loss": 0.0647,
      "step": 120800
    },
    {
      "epoch": 1.779428874390657,
      "grad_norm": 1.7127313613891602,
      "learning_rate": 1.4317011068419986e-05,
      "loss": 0.0664,
      "step": 120825
    },
    {
      "epoch": 1.779797057480744,
      "grad_norm": 1.4669241905212402,
      "learning_rate": 1.4315374696453002e-05,
      "loss": 0.0669,
      "step": 120850
    },
    {
      "epoch": 1.780165240570831,
      "grad_norm": 1.3870619535446167,
      "learning_rate": 1.4313738324486017e-05,
      "loss": 0.0677,
      "step": 120875
    },
    {
      "epoch": 1.780533423660918,
      "grad_norm": 1.8698972463607788,
      "learning_rate": 1.4312101952519033e-05,
      "loss": 0.0703,
      "step": 120900
    },
    {
      "epoch": 1.780901606751005,
      "grad_norm": 1.6329165697097778,
      "learning_rate": 1.431046558055205e-05,
      "loss": 0.0694,
      "step": 120925
    },
    {
      "epoch": 1.781269789841092,
      "grad_norm": 1.3736357688903809,
      "learning_rate": 1.4308829208585062e-05,
      "loss": 0.0725,
      "step": 120950
    },
    {
      "epoch": 1.781637972931179,
      "grad_norm": 1.3963435888290405,
      "learning_rate": 1.4307192836618078e-05,
      "loss": 0.0646,
      "step": 120975
    },
    {
      "epoch": 1.782006156021266,
      "grad_norm": 1.7232320308685303,
      "learning_rate": 1.4305556464651094e-05,
      "loss": 0.0795,
      "step": 121000
    },
    {
      "epoch": 1.7823743391113533,
      "grad_norm": 1.8137731552124023,
      "learning_rate": 1.4303920092684109e-05,
      "loss": 0.0737,
      "step": 121025
    },
    {
      "epoch": 1.7827425222014404,
      "grad_norm": 1.3642964363098145,
      "learning_rate": 1.4302283720717125e-05,
      "loss": 0.0682,
      "step": 121050
    },
    {
      "epoch": 1.7831107052915274,
      "grad_norm": 0.9713894724845886,
      "learning_rate": 1.4300647348750141e-05,
      "loss": 0.0714,
      "step": 121075
    },
    {
      "epoch": 1.7834788883816144,
      "grad_norm": 1.5277451276779175,
      "learning_rate": 1.4299010976783157e-05,
      "loss": 0.078,
      "step": 121100
    },
    {
      "epoch": 1.7838470714717014,
      "grad_norm": 1.3738890886306763,
      "learning_rate": 1.4297374604816172e-05,
      "loss": 0.0611,
      "step": 121125
    },
    {
      "epoch": 1.7842152545617886,
      "grad_norm": 1.4161064624786377,
      "learning_rate": 1.4295738232849186e-05,
      "loss": 0.0677,
      "step": 121150
    },
    {
      "epoch": 1.7845834376518757,
      "grad_norm": 1.171356439590454,
      "learning_rate": 1.42941018608822e-05,
      "loss": 0.0709,
      "step": 121175
    },
    {
      "epoch": 1.7849516207419627,
      "grad_norm": 1.4238909482955933,
      "learning_rate": 1.4292465488915217e-05,
      "loss": 0.0718,
      "step": 121200
    },
    {
      "epoch": 1.7853198038320497,
      "grad_norm": 1.3162243366241455,
      "learning_rate": 1.4290829116948233e-05,
      "loss": 0.0708,
      "step": 121225
    },
    {
      "epoch": 1.7856879869221367,
      "grad_norm": 1.1067510843276978,
      "learning_rate": 1.4289192744981249e-05,
      "loss": 0.0678,
      "step": 121250
    },
    {
      "epoch": 1.7860561700122237,
      "grad_norm": 1.2170255184173584,
      "learning_rate": 1.4287556373014263e-05,
      "loss": 0.061,
      "step": 121275
    },
    {
      "epoch": 1.7864243531023107,
      "grad_norm": 1.4518414735794067,
      "learning_rate": 1.4285985455925959e-05,
      "loss": 0.0624,
      "step": 121300
    },
    {
      "epoch": 1.7867925361923978,
      "grad_norm": 0.9676971435546875,
      "learning_rate": 1.4284349083958975e-05,
      "loss": 0.065,
      "step": 121325
    },
    {
      "epoch": 1.7871607192824848,
      "grad_norm": 1.3099478483200073,
      "learning_rate": 1.4282712711991988e-05,
      "loss": 0.0724,
      "step": 121350
    },
    {
      "epoch": 1.7875289023725718,
      "grad_norm": 1.7666794061660767,
      "learning_rate": 1.4281076340025004e-05,
      "loss": 0.0663,
      "step": 121375
    },
    {
      "epoch": 1.7878970854626588,
      "grad_norm": 1.442685842514038,
      "learning_rate": 1.427943996805802e-05,
      "loss": 0.0676,
      "step": 121400
    },
    {
      "epoch": 1.7882652685527458,
      "grad_norm": 1.247192144393921,
      "learning_rate": 1.4277803596091035e-05,
      "loss": 0.0732,
      "step": 121425
    },
    {
      "epoch": 1.7886334516428328,
      "grad_norm": 1.809297800064087,
      "learning_rate": 1.427616722412405e-05,
      "loss": 0.066,
      "step": 121450
    },
    {
      "epoch": 1.7890016347329198,
      "grad_norm": 1.4799872636795044,
      "learning_rate": 1.4274530852157067e-05,
      "loss": 0.0696,
      "step": 121475
    },
    {
      "epoch": 1.7893698178230069,
      "grad_norm": 1.9324668645858765,
      "learning_rate": 1.4272894480190083e-05,
      "loss": 0.0705,
      "step": 121500
    },
    {
      "epoch": 1.789738000913094,
      "grad_norm": 1.2843706607818604,
      "learning_rate": 1.4271258108223098e-05,
      "loss": 0.0634,
      "step": 121525
    },
    {
      "epoch": 1.7901061840031811,
      "grad_norm": 1.338454246520996,
      "learning_rate": 1.4269621736256114e-05,
      "loss": 0.0653,
      "step": 121550
    },
    {
      "epoch": 1.7904743670932681,
      "grad_norm": 1.2111071348190308,
      "learning_rate": 1.4267985364289128e-05,
      "loss": 0.0634,
      "step": 121575
    },
    {
      "epoch": 1.7908425501833551,
      "grad_norm": 1.413698673248291,
      "learning_rate": 1.4266348992322143e-05,
      "loss": 0.0671,
      "step": 121600
    },
    {
      "epoch": 1.7912107332734424,
      "grad_norm": 1.3297215700149536,
      "learning_rate": 1.4264712620355159e-05,
      "loss": 0.0709,
      "step": 121625
    },
    {
      "epoch": 1.7915789163635294,
      "grad_norm": 1.3247640132904053,
      "learning_rate": 1.4263076248388175e-05,
      "loss": 0.0727,
      "step": 121650
    },
    {
      "epoch": 1.7919470994536164,
      "grad_norm": 1.161634922027588,
      "learning_rate": 1.4261439876421191e-05,
      "loss": 0.0749,
      "step": 121675
    },
    {
      "epoch": 1.7923152825437034,
      "grad_norm": 1.5505574941635132,
      "learning_rate": 1.4259803504454206e-05,
      "loss": 0.0656,
      "step": 121700
    },
    {
      "epoch": 1.7926834656337904,
      "grad_norm": 1.5611209869384766,
      "learning_rate": 1.4258167132487222e-05,
      "loss": 0.07,
      "step": 121725
    },
    {
      "epoch": 1.7930516487238775,
      "grad_norm": 2.1446967124938965,
      "learning_rate": 1.4256530760520238e-05,
      "loss": 0.0685,
      "step": 121750
    },
    {
      "epoch": 1.7934198318139645,
      "grad_norm": 1.0909874439239502,
      "learning_rate": 1.425489438855325e-05,
      "loss": 0.0715,
      "step": 121775
    },
    {
      "epoch": 1.7937880149040515,
      "grad_norm": 1.20832097530365,
      "learning_rate": 1.4253258016586267e-05,
      "loss": 0.0648,
      "step": 121800
    },
    {
      "epoch": 1.7941561979941385,
      "grad_norm": 1.3289990425109863,
      "learning_rate": 1.4251621644619283e-05,
      "loss": 0.0722,
      "step": 121825
    },
    {
      "epoch": 1.7945243810842255,
      "grad_norm": 1.5881668329238892,
      "learning_rate": 1.4249985272652297e-05,
      "loss": 0.0658,
      "step": 121850
    },
    {
      "epoch": 1.7948925641743125,
      "grad_norm": 1.1237159967422485,
      "learning_rate": 1.4248348900685314e-05,
      "loss": 0.0608,
      "step": 121875
    },
    {
      "epoch": 1.7952607472643995,
      "grad_norm": 1.7560815811157227,
      "learning_rate": 1.424671252871833e-05,
      "loss": 0.0699,
      "step": 121900
    },
    {
      "epoch": 1.7956289303544866,
      "grad_norm": 1.1331946849822998,
      "learning_rate": 1.4245076156751346e-05,
      "loss": 0.066,
      "step": 121925
    },
    {
      "epoch": 1.7959971134445736,
      "grad_norm": 1.2591310739517212,
      "learning_rate": 1.424343978478436e-05,
      "loss": 0.0627,
      "step": 121950
    },
    {
      "epoch": 1.7963652965346606,
      "grad_norm": 1.442726969718933,
      "learning_rate": 1.4241803412817376e-05,
      "loss": 0.0666,
      "step": 121975
    },
    {
      "epoch": 1.7967334796247478,
      "grad_norm": 1.6420409679412842,
      "learning_rate": 1.424016704085039e-05,
      "loss": 0.0703,
      "step": 122000
    },
    {
      "epoch": 1.7971016627148348,
      "grad_norm": 1.8024204969406128,
      "learning_rate": 1.4238530668883405e-05,
      "loss": 0.0672,
      "step": 122025
    },
    {
      "epoch": 1.7974698458049219,
      "grad_norm": 1.4262099266052246,
      "learning_rate": 1.4236894296916422e-05,
      "loss": 0.0705,
      "step": 122050
    },
    {
      "epoch": 1.7978380288950089,
      "grad_norm": 1.2743639945983887,
      "learning_rate": 1.4235257924949438e-05,
      "loss": 0.0667,
      "step": 122075
    },
    {
      "epoch": 1.7982062119850961,
      "grad_norm": 1.9641082286834717,
      "learning_rate": 1.4233621552982452e-05,
      "loss": 0.074,
      "step": 122100
    },
    {
      "epoch": 1.7985743950751831,
      "grad_norm": 1.8725675344467163,
      "learning_rate": 1.4231985181015468e-05,
      "loss": 0.0637,
      "step": 122125
    },
    {
      "epoch": 1.7989425781652701,
      "grad_norm": 1.1382743120193481,
      "learning_rate": 1.4230348809048484e-05,
      "loss": 0.0645,
      "step": 122150
    },
    {
      "epoch": 1.7993107612553572,
      "grad_norm": 1.1396962404251099,
      "learning_rate": 1.42287124370815e-05,
      "loss": 0.0618,
      "step": 122175
    },
    {
      "epoch": 1.7996789443454442,
      "grad_norm": 1.1252176761627197,
      "learning_rate": 1.4227076065114513e-05,
      "loss": 0.0707,
      "step": 122200
    },
    {
      "epoch": 1.8000471274355312,
      "grad_norm": 1.9477506875991821,
      "learning_rate": 1.422543969314753e-05,
      "loss": 0.0711,
      "step": 122225
    },
    {
      "epoch": 1.8004153105256182,
      "grad_norm": 1.0825227499008179,
      "learning_rate": 1.4223803321180546e-05,
      "loss": 0.0657,
      "step": 122250
    },
    {
      "epoch": 1.8007834936157052,
      "grad_norm": 1.5377202033996582,
      "learning_rate": 1.422216694921356e-05,
      "loss": 0.0706,
      "step": 122275
    },
    {
      "epoch": 1.8011516767057922,
      "grad_norm": 1.994928240776062,
      "learning_rate": 1.4220530577246576e-05,
      "loss": 0.0716,
      "step": 122300
    },
    {
      "epoch": 1.8015198597958793,
      "grad_norm": 1.5081052780151367,
      "learning_rate": 1.4218894205279592e-05,
      "loss": 0.064,
      "step": 122325
    },
    {
      "epoch": 1.8018880428859663,
      "grad_norm": 1.1986535787582397,
      "learning_rate": 1.4217257833312607e-05,
      "loss": 0.0633,
      "step": 122350
    },
    {
      "epoch": 1.8022562259760533,
      "grad_norm": 1.4311184883117676,
      "learning_rate": 1.4215621461345623e-05,
      "loss": 0.0764,
      "step": 122375
    },
    {
      "epoch": 1.8026244090661403,
      "grad_norm": 1.7433348894119263,
      "learning_rate": 1.421398508937864e-05,
      "loss": 0.0659,
      "step": 122400
    },
    {
      "epoch": 1.8029925921562273,
      "grad_norm": 1.5358247756958008,
      "learning_rate": 1.4212348717411652e-05,
      "loss": 0.0674,
      "step": 122425
    },
    {
      "epoch": 1.8033607752463143,
      "grad_norm": 1.254683494567871,
      "learning_rate": 1.4210712345444668e-05,
      "loss": 0.0691,
      "step": 122450
    },
    {
      "epoch": 1.8037289583364016,
      "grad_norm": 1.3888503313064575,
      "learning_rate": 1.4209075973477684e-05,
      "loss": 0.0689,
      "step": 122475
    },
    {
      "epoch": 1.8040971414264886,
      "grad_norm": 1.1659692525863647,
      "learning_rate": 1.42074396015107e-05,
      "loss": 0.0709,
      "step": 122500
    },
    {
      "epoch": 1.8044653245165756,
      "grad_norm": 1.6324865818023682,
      "learning_rate": 1.4205803229543715e-05,
      "loss": 0.0679,
      "step": 122525
    },
    {
      "epoch": 1.8048335076066626,
      "grad_norm": 1.0429387092590332,
      "learning_rate": 1.4204166857576731e-05,
      "loss": 0.0698,
      "step": 122550
    },
    {
      "epoch": 1.8052016906967496,
      "grad_norm": 1.061618447303772,
      "learning_rate": 1.4202530485609747e-05,
      "loss": 0.0685,
      "step": 122575
    },
    {
      "epoch": 1.8055698737868369,
      "grad_norm": 1.6197209358215332,
      "learning_rate": 1.4200894113642763e-05,
      "loss": 0.0708,
      "step": 122600
    },
    {
      "epoch": 1.8059380568769239,
      "grad_norm": 1.4687283039093018,
      "learning_rate": 1.4199257741675776e-05,
      "loss": 0.0686,
      "step": 122625
    },
    {
      "epoch": 1.806306239967011,
      "grad_norm": 1.6074970960617065,
      "learning_rate": 1.4197621369708792e-05,
      "loss": 0.0662,
      "step": 122650
    },
    {
      "epoch": 1.806674423057098,
      "grad_norm": 0.8636961579322815,
      "learning_rate": 1.4195984997741807e-05,
      "loss": 0.0619,
      "step": 122675
    },
    {
      "epoch": 1.807042606147185,
      "grad_norm": 1.3781434297561646,
      "learning_rate": 1.4194348625774823e-05,
      "loss": 0.069,
      "step": 122700
    },
    {
      "epoch": 1.807410789237272,
      "grad_norm": 1.4002094268798828,
      "learning_rate": 1.4192712253807839e-05,
      "loss": 0.0731,
      "step": 122725
    },
    {
      "epoch": 1.807778972327359,
      "grad_norm": 1.388077974319458,
      "learning_rate": 1.4191075881840855e-05,
      "loss": 0.0762,
      "step": 122750
    },
    {
      "epoch": 1.808147155417446,
      "grad_norm": 1.3945006132125854,
      "learning_rate": 1.418943950987387e-05,
      "loss": 0.0688,
      "step": 122775
    },
    {
      "epoch": 1.808515338507533,
      "grad_norm": 1.5331770181655884,
      "learning_rate": 1.4187803137906886e-05,
      "loss": 0.0731,
      "step": 122800
    },
    {
      "epoch": 1.80888352159762,
      "grad_norm": 1.9857841730117798,
      "learning_rate": 1.41861667659399e-05,
      "loss": 0.0738,
      "step": 122825
    },
    {
      "epoch": 1.809251704687707,
      "grad_norm": 1.7090383768081665,
      "learning_rate": 1.4184530393972915e-05,
      "loss": 0.0695,
      "step": 122850
    },
    {
      "epoch": 1.809619887777794,
      "grad_norm": 1.6358251571655273,
      "learning_rate": 1.418289402200593e-05,
      "loss": 0.0678,
      "step": 122875
    },
    {
      "epoch": 1.809988070867881,
      "grad_norm": 1.7000668048858643,
      "learning_rate": 1.4181257650038947e-05,
      "loss": 0.0725,
      "step": 122900
    },
    {
      "epoch": 1.810356253957968,
      "grad_norm": 1.3275409936904907,
      "learning_rate": 1.4179621278071961e-05,
      "loss": 0.0646,
      "step": 122925
    },
    {
      "epoch": 1.8107244370480553,
      "grad_norm": 2.237173080444336,
      "learning_rate": 1.4177984906104978e-05,
      "loss": 0.0697,
      "step": 122950
    },
    {
      "epoch": 1.8110926201381423,
      "grad_norm": 1.398925542831421,
      "learning_rate": 1.4176348534137994e-05,
      "loss": 0.0609,
      "step": 122975
    },
    {
      "epoch": 1.8114608032282293,
      "grad_norm": 1.2925307750701904,
      "learning_rate": 1.417471216217101e-05,
      "loss": 0.071,
      "step": 123000
    },
    {
      "epoch": 1.8118289863183163,
      "grad_norm": 1.352026104927063,
      "learning_rate": 1.4173075790204024e-05,
      "loss": 0.0663,
      "step": 123025
    },
    {
      "epoch": 1.8121971694084034,
      "grad_norm": 1.6469991207122803,
      "learning_rate": 1.4171439418237039e-05,
      "loss": 0.0696,
      "step": 123050
    },
    {
      "epoch": 1.8125653524984906,
      "grad_norm": 1.35920250415802,
      "learning_rate": 1.4169803046270055e-05,
      "loss": 0.0667,
      "step": 123075
    },
    {
      "epoch": 1.8129335355885776,
      "grad_norm": 0.9462726712226868,
      "learning_rate": 1.416816667430307e-05,
      "loss": 0.0729,
      "step": 123100
    },
    {
      "epoch": 1.8133017186786646,
      "grad_norm": 0.9434695839881897,
      "learning_rate": 1.4166530302336086e-05,
      "loss": 0.0644,
      "step": 123125
    },
    {
      "epoch": 1.8136699017687516,
      "grad_norm": 1.3169819116592407,
      "learning_rate": 1.4164893930369102e-05,
      "loss": 0.0684,
      "step": 123150
    },
    {
      "epoch": 1.8140380848588387,
      "grad_norm": 1.188008427619934,
      "learning_rate": 1.4163257558402118e-05,
      "loss": 0.0611,
      "step": 123175
    },
    {
      "epoch": 1.8144062679489257,
      "grad_norm": 1.1293237209320068,
      "learning_rate": 1.4161621186435132e-05,
      "loss": 0.0784,
      "step": 123200
    },
    {
      "epoch": 1.8147744510390127,
      "grad_norm": 1.3106647729873657,
      "learning_rate": 1.4159984814468148e-05,
      "loss": 0.0684,
      "step": 123225
    },
    {
      "epoch": 1.8151426341290997,
      "grad_norm": 1.7924137115478516,
      "learning_rate": 1.4158348442501161e-05,
      "loss": 0.0656,
      "step": 123250
    },
    {
      "epoch": 1.8155108172191867,
      "grad_norm": 1.1363129615783691,
      "learning_rate": 1.4156712070534177e-05,
      "loss": 0.0598,
      "step": 123275
    },
    {
      "epoch": 1.8158790003092737,
      "grad_norm": 1.4987635612487793,
      "learning_rate": 1.4155075698567194e-05,
      "loss": 0.072,
      "step": 123300
    },
    {
      "epoch": 1.8162471833993608,
      "grad_norm": 1.4073323011398315,
      "learning_rate": 1.415343932660021e-05,
      "loss": 0.07,
      "step": 123325
    },
    {
      "epoch": 1.8166153664894478,
      "grad_norm": 1.0494939088821411,
      "learning_rate": 1.4151802954633224e-05,
      "loss": 0.0644,
      "step": 123350
    },
    {
      "epoch": 1.8169835495795348,
      "grad_norm": 1.3889259099960327,
      "learning_rate": 1.415016658266624e-05,
      "loss": 0.0752,
      "step": 123375
    },
    {
      "epoch": 1.8173517326696218,
      "grad_norm": 1.484281301498413,
      "learning_rate": 1.4148530210699256e-05,
      "loss": 0.0692,
      "step": 123400
    },
    {
      "epoch": 1.817719915759709,
      "grad_norm": 2.207677125930786,
      "learning_rate": 1.4146893838732273e-05,
      "loss": 0.0708,
      "step": 123425
    },
    {
      "epoch": 1.818088098849796,
      "grad_norm": 1.457485556602478,
      "learning_rate": 1.4145257466765287e-05,
      "loss": 0.0673,
      "step": 123450
    },
    {
      "epoch": 1.818456281939883,
      "grad_norm": 1.2040473222732544,
      "learning_rate": 1.4143621094798302e-05,
      "loss": 0.0661,
      "step": 123475
    },
    {
      "epoch": 1.81882446502997,
      "grad_norm": 1.5974911451339722,
      "learning_rate": 1.4141984722831316e-05,
      "loss": 0.0651,
      "step": 123500
    },
    {
      "epoch": 1.819192648120057,
      "grad_norm": 1.0894943475723267,
      "learning_rate": 1.4140348350864332e-05,
      "loss": 0.0646,
      "step": 123525
    },
    {
      "epoch": 1.8195608312101443,
      "grad_norm": 1.3080356121063232,
      "learning_rate": 1.4138711978897348e-05,
      "loss": 0.0678,
      "step": 123550
    },
    {
      "epoch": 1.8199290143002314,
      "grad_norm": 0.965703010559082,
      "learning_rate": 1.4137075606930364e-05,
      "loss": 0.0705,
      "step": 123575
    },
    {
      "epoch": 1.8202971973903184,
      "grad_norm": 2.154559373855591,
      "learning_rate": 1.4135439234963379e-05,
      "loss": 0.0693,
      "step": 123600
    },
    {
      "epoch": 1.8206653804804054,
      "grad_norm": 1.5285266637802124,
      "learning_rate": 1.4133802862996395e-05,
      "loss": 0.0654,
      "step": 123625
    },
    {
      "epoch": 1.8210335635704924,
      "grad_norm": 1.1775543689727783,
      "learning_rate": 1.4132166491029411e-05,
      "loss": 0.0665,
      "step": 123650
    },
    {
      "epoch": 1.8214017466605794,
      "grad_norm": 1.6471136808395386,
      "learning_rate": 1.4130530119062424e-05,
      "loss": 0.0704,
      "step": 123675
    },
    {
      "epoch": 1.8217699297506664,
      "grad_norm": 1.290858507156372,
      "learning_rate": 1.412889374709544e-05,
      "loss": 0.0748,
      "step": 123700
    },
    {
      "epoch": 1.8221381128407534,
      "grad_norm": 1.5764347314834595,
      "learning_rate": 1.4127257375128456e-05,
      "loss": 0.0652,
      "step": 123725
    },
    {
      "epoch": 1.8225062959308405,
      "grad_norm": 1.1291478872299194,
      "learning_rate": 1.4125621003161472e-05,
      "loss": 0.0592,
      "step": 123750
    },
    {
      "epoch": 1.8228744790209275,
      "grad_norm": 1.6106886863708496,
      "learning_rate": 1.4123984631194487e-05,
      "loss": 0.0659,
      "step": 123775
    },
    {
      "epoch": 1.8232426621110145,
      "grad_norm": 1.2678749561309814,
      "learning_rate": 1.4122348259227503e-05,
      "loss": 0.0706,
      "step": 123800
    },
    {
      "epoch": 1.8236108452011015,
      "grad_norm": 1.6881310939788818,
      "learning_rate": 1.412071188726052e-05,
      "loss": 0.0679,
      "step": 123825
    },
    {
      "epoch": 1.8239790282911885,
      "grad_norm": 1.330228567123413,
      "learning_rate": 1.4119075515293534e-05,
      "loss": 0.0684,
      "step": 123850
    },
    {
      "epoch": 1.8243472113812755,
      "grad_norm": 1.768144965171814,
      "learning_rate": 1.411743914332655e-05,
      "loss": 0.075,
      "step": 123875
    },
    {
      "epoch": 1.8247153944713628,
      "grad_norm": 1.3235609531402588,
      "learning_rate": 1.4115802771359564e-05,
      "loss": 0.0689,
      "step": 123900
    },
    {
      "epoch": 1.8250835775614498,
      "grad_norm": 1.2848559617996216,
      "learning_rate": 1.4114166399392579e-05,
      "loss": 0.0611,
      "step": 123925
    },
    {
      "epoch": 1.8254517606515368,
      "grad_norm": 1.4224627017974854,
      "learning_rate": 1.4112530027425595e-05,
      "loss": 0.0695,
      "step": 123950
    },
    {
      "epoch": 1.8258199437416238,
      "grad_norm": 1.7034728527069092,
      "learning_rate": 1.4110893655458611e-05,
      "loss": 0.0637,
      "step": 123975
    },
    {
      "epoch": 1.8261881268317108,
      "grad_norm": 1.4163336753845215,
      "learning_rate": 1.4109257283491627e-05,
      "loss": 0.0641,
      "step": 124000
    },
    {
      "epoch": 1.826556309921798,
      "grad_norm": 1.9863816499710083,
      "learning_rate": 1.4107686366403321e-05,
      "loss": 0.0701,
      "step": 124025
    },
    {
      "epoch": 1.826924493011885,
      "grad_norm": 1.6415189504623413,
      "learning_rate": 1.4106049994436337e-05,
      "loss": 0.0602,
      "step": 124050
    },
    {
      "epoch": 1.827292676101972,
      "grad_norm": 1.7405691146850586,
      "learning_rate": 1.4104413622469353e-05,
      "loss": 0.0651,
      "step": 124075
    },
    {
      "epoch": 1.8276608591920591,
      "grad_norm": 1.368970513343811,
      "learning_rate": 1.4102777250502366e-05,
      "loss": 0.0716,
      "step": 124100
    },
    {
      "epoch": 1.8280290422821461,
      "grad_norm": 1.4880112409591675,
      "learning_rate": 1.4101140878535382e-05,
      "loss": 0.0701,
      "step": 124125
    },
    {
      "epoch": 1.8283972253722331,
      "grad_norm": 1.0609428882598877,
      "learning_rate": 1.4099504506568398e-05,
      "loss": 0.0637,
      "step": 124150
    },
    {
      "epoch": 1.8287654084623202,
      "grad_norm": 1.6436680555343628,
      "learning_rate": 1.4097868134601413e-05,
      "loss": 0.0729,
      "step": 124175
    },
    {
      "epoch": 1.8291335915524072,
      "grad_norm": 1.143778681755066,
      "learning_rate": 1.4096231762634429e-05,
      "loss": 0.0672,
      "step": 124200
    },
    {
      "epoch": 1.8295017746424942,
      "grad_norm": 1.2172858715057373,
      "learning_rate": 1.4094595390667445e-05,
      "loss": 0.0722,
      "step": 124225
    },
    {
      "epoch": 1.8298699577325812,
      "grad_norm": 2.0340380668640137,
      "learning_rate": 1.4092959018700461e-05,
      "loss": 0.0754,
      "step": 124250
    },
    {
      "epoch": 1.8302381408226682,
      "grad_norm": 1.838102102279663,
      "learning_rate": 1.4091322646733476e-05,
      "loss": 0.0645,
      "step": 124275
    },
    {
      "epoch": 1.8306063239127552,
      "grad_norm": 1.410643219947815,
      "learning_rate": 1.4089686274766492e-05,
      "loss": 0.0702,
      "step": 124300
    },
    {
      "epoch": 1.8309745070028423,
      "grad_norm": 1.8687243461608887,
      "learning_rate": 1.4088049902799505e-05,
      "loss": 0.0691,
      "step": 124325
    },
    {
      "epoch": 1.8313426900929293,
      "grad_norm": 1.682615876197815,
      "learning_rate": 1.408641353083252e-05,
      "loss": 0.0671,
      "step": 124350
    },
    {
      "epoch": 1.8317108731830163,
      "grad_norm": 1.1685667037963867,
      "learning_rate": 1.4084777158865537e-05,
      "loss": 0.0669,
      "step": 124375
    },
    {
      "epoch": 1.8320790562731035,
      "grad_norm": 1.583894968032837,
      "learning_rate": 1.4083140786898553e-05,
      "loss": 0.0738,
      "step": 124400
    },
    {
      "epoch": 1.8324472393631905,
      "grad_norm": 1.4164410829544067,
      "learning_rate": 1.4081504414931568e-05,
      "loss": 0.0656,
      "step": 124425
    },
    {
      "epoch": 1.8328154224532776,
      "grad_norm": 1.970167636871338,
      "learning_rate": 1.4079868042964584e-05,
      "loss": 0.0705,
      "step": 124450
    },
    {
      "epoch": 1.8331836055433646,
      "grad_norm": 1.6298333406448364,
      "learning_rate": 1.40782316709976e-05,
      "loss": 0.0649,
      "step": 124475
    },
    {
      "epoch": 1.8335517886334518,
      "grad_norm": 1.2371671199798584,
      "learning_rate": 1.4076595299030616e-05,
      "loss": 0.0682,
      "step": 124500
    },
    {
      "epoch": 1.8339199717235388,
      "grad_norm": 1.6401854753494263,
      "learning_rate": 1.4074958927063629e-05,
      "loss": 0.0671,
      "step": 124525
    },
    {
      "epoch": 1.8342881548136258,
      "grad_norm": 1.0946791172027588,
      "learning_rate": 1.4073322555096645e-05,
      "loss": 0.0711,
      "step": 124550
    },
    {
      "epoch": 1.8346563379037129,
      "grad_norm": 1.3830103874206543,
      "learning_rate": 1.4071686183129661e-05,
      "loss": 0.0687,
      "step": 124575
    },
    {
      "epoch": 1.8350245209937999,
      "grad_norm": 1.2596478462219238,
      "learning_rate": 1.4070049811162676e-05,
      "loss": 0.0656,
      "step": 124600
    },
    {
      "epoch": 1.8353927040838869,
      "grad_norm": 1.4882218837738037,
      "learning_rate": 1.4068413439195692e-05,
      "loss": 0.0742,
      "step": 124625
    },
    {
      "epoch": 1.835760887173974,
      "grad_norm": 1.4395406246185303,
      "learning_rate": 1.4066777067228708e-05,
      "loss": 0.0704,
      "step": 124650
    },
    {
      "epoch": 1.836129070264061,
      "grad_norm": 1.1731866598129272,
      "learning_rate": 1.4065140695261722e-05,
      "loss": 0.0609,
      "step": 124675
    },
    {
      "epoch": 1.836497253354148,
      "grad_norm": 1.500097393989563,
      "learning_rate": 1.4063504323294738e-05,
      "loss": 0.0716,
      "step": 124700
    },
    {
      "epoch": 1.836865436444235,
      "grad_norm": 1.510772943496704,
      "learning_rate": 1.4061867951327753e-05,
      "loss": 0.0669,
      "step": 124725
    },
    {
      "epoch": 1.837233619534322,
      "grad_norm": 1.2818152904510498,
      "learning_rate": 1.4060231579360767e-05,
      "loss": 0.0739,
      "step": 124750
    },
    {
      "epoch": 1.837601802624409,
      "grad_norm": 1.9229601621627808,
      "learning_rate": 1.4058595207393784e-05,
      "loss": 0.0651,
      "step": 124775
    },
    {
      "epoch": 1.837969985714496,
      "grad_norm": 1.207069754600525,
      "learning_rate": 1.40569588354268e-05,
      "loss": 0.0704,
      "step": 124800
    },
    {
      "epoch": 1.838338168804583,
      "grad_norm": 1.3579490184783936,
      "learning_rate": 1.4055322463459816e-05,
      "loss": 0.0679,
      "step": 124825
    },
    {
      "epoch": 1.83870635189467,
      "grad_norm": 1.5743180513381958,
      "learning_rate": 1.405368609149283e-05,
      "loss": 0.0741,
      "step": 124850
    },
    {
      "epoch": 1.8390745349847573,
      "grad_norm": 1.741626501083374,
      "learning_rate": 1.4052049719525846e-05,
      "loss": 0.0694,
      "step": 124875
    },
    {
      "epoch": 1.8394427180748443,
      "grad_norm": 1.5830440521240234,
      "learning_rate": 1.4050413347558863e-05,
      "loss": 0.073,
      "step": 124900
    },
    {
      "epoch": 1.8398109011649313,
      "grad_norm": 0.8269080519676208,
      "learning_rate": 1.4048776975591879e-05,
      "loss": 0.0737,
      "step": 124925
    },
    {
      "epoch": 1.8401790842550183,
      "grad_norm": 1.302711844444275,
      "learning_rate": 1.4047140603624892e-05,
      "loss": 0.0662,
      "step": 124950
    },
    {
      "epoch": 1.8405472673451053,
      "grad_norm": 1.016312599182129,
      "learning_rate": 1.4045504231657908e-05,
      "loss": 0.0672,
      "step": 124975
    },
    {
      "epoch": 1.8409154504351926,
      "grad_norm": 1.247109293937683,
      "learning_rate": 1.4043867859690922e-05,
      "loss": 0.0636,
      "step": 125000
    },
    {
      "epoch": 1.8412836335252796,
      "grad_norm": 1.4615647792816162,
      "learning_rate": 1.4042231487723938e-05,
      "loss": 0.0717,
      "step": 125025
    },
    {
      "epoch": 1.8416518166153666,
      "grad_norm": 1.698584794998169,
      "learning_rate": 1.4040595115756954e-05,
      "loss": 0.0705,
      "step": 125050
    },
    {
      "epoch": 1.8420199997054536,
      "grad_norm": 1.2643131017684937,
      "learning_rate": 1.403895874378997e-05,
      "loss": 0.0576,
      "step": 125075
    },
    {
      "epoch": 1.8423881827955406,
      "grad_norm": 1.2368474006652832,
      "learning_rate": 1.4037322371822985e-05,
      "loss": 0.0643,
      "step": 125100
    },
    {
      "epoch": 1.8427563658856276,
      "grad_norm": 1.357359766960144,
      "learning_rate": 1.4035685999856001e-05,
      "loss": 0.0695,
      "step": 125125
    },
    {
      "epoch": 1.8431245489757146,
      "grad_norm": 1.5304204225540161,
      "learning_rate": 1.4034049627889016e-05,
      "loss": 0.071,
      "step": 125150
    },
    {
      "epoch": 1.8434927320658017,
      "grad_norm": 1.029677152633667,
      "learning_rate": 1.403241325592203e-05,
      "loss": 0.0735,
      "step": 125175
    },
    {
      "epoch": 1.8438609151558887,
      "grad_norm": 1.4008866548538208,
      "learning_rate": 1.4030776883955046e-05,
      "loss": 0.0657,
      "step": 125200
    },
    {
      "epoch": 1.8442290982459757,
      "grad_norm": 1.617188811302185,
      "learning_rate": 1.4029140511988062e-05,
      "loss": 0.0643,
      "step": 125225
    },
    {
      "epoch": 1.8445972813360627,
      "grad_norm": 1.5113645792007446,
      "learning_rate": 1.4027504140021077e-05,
      "loss": 0.0752,
      "step": 125250
    },
    {
      "epoch": 1.8449654644261497,
      "grad_norm": 1.2097511291503906,
      "learning_rate": 1.4025867768054093e-05,
      "loss": 0.0636,
      "step": 125275
    },
    {
      "epoch": 1.8453336475162367,
      "grad_norm": 1.6903287172317505,
      "learning_rate": 1.402423139608711e-05,
      "loss": 0.0666,
      "step": 125300
    },
    {
      "epoch": 1.8457018306063238,
      "grad_norm": 0.9059404134750366,
      "learning_rate": 1.4022595024120125e-05,
      "loss": 0.0679,
      "step": 125325
    },
    {
      "epoch": 1.846070013696411,
      "grad_norm": 1.970717430114746,
      "learning_rate": 1.402095865215314e-05,
      "loss": 0.0665,
      "step": 125350
    },
    {
      "epoch": 1.846438196786498,
      "grad_norm": 1.2964277267456055,
      "learning_rate": 1.4019322280186154e-05,
      "loss": 0.0618,
      "step": 125375
    },
    {
      "epoch": 1.846806379876585,
      "grad_norm": 0.9654756188392639,
      "learning_rate": 1.401768590821917e-05,
      "loss": 0.0654,
      "step": 125400
    },
    {
      "epoch": 1.847174562966672,
      "grad_norm": 1.418273687362671,
      "learning_rate": 1.4016049536252185e-05,
      "loss": 0.0694,
      "step": 125425
    },
    {
      "epoch": 1.847542746056759,
      "grad_norm": 1.314102292060852,
      "learning_rate": 1.4014413164285201e-05,
      "loss": 0.0762,
      "step": 125450
    },
    {
      "epoch": 1.8479109291468463,
      "grad_norm": 1.6058028936386108,
      "learning_rate": 1.4012776792318217e-05,
      "loss": 0.0711,
      "step": 125475
    },
    {
      "epoch": 1.8482791122369333,
      "grad_norm": 1.3302124738693237,
      "learning_rate": 1.4011140420351233e-05,
      "loss": 0.063,
      "step": 125500
    },
    {
      "epoch": 1.8486472953270203,
      "grad_norm": 1.6998109817504883,
      "learning_rate": 1.4009504048384248e-05,
      "loss": 0.0647,
      "step": 125525
    },
    {
      "epoch": 1.8490154784171073,
      "grad_norm": 1.5907782316207886,
      "learning_rate": 1.4007867676417264e-05,
      "loss": 0.0713,
      "step": 125550
    },
    {
      "epoch": 1.8493836615071944,
      "grad_norm": 1.1828668117523193,
      "learning_rate": 1.4006231304450277e-05,
      "loss": 0.0769,
      "step": 125575
    },
    {
      "epoch": 1.8497518445972814,
      "grad_norm": 1.1502078771591187,
      "learning_rate": 1.4004594932483293e-05,
      "loss": 0.0664,
      "step": 125600
    },
    {
      "epoch": 1.8501200276873684,
      "grad_norm": 1.2194290161132812,
      "learning_rate": 1.4002958560516309e-05,
      "loss": 0.0619,
      "step": 125625
    },
    {
      "epoch": 1.8504882107774554,
      "grad_norm": 1.4610698223114014,
      "learning_rate": 1.4001322188549325e-05,
      "loss": 0.0669,
      "step": 125650
    },
    {
      "epoch": 1.8508563938675424,
      "grad_norm": 1.318021297454834,
      "learning_rate": 1.399968581658234e-05,
      "loss": 0.0697,
      "step": 125675
    },
    {
      "epoch": 1.8512245769576294,
      "grad_norm": 1.4886027574539185,
      "learning_rate": 1.3998049444615356e-05,
      "loss": 0.0767,
      "step": 125700
    },
    {
      "epoch": 1.8515927600477164,
      "grad_norm": 1.699396014213562,
      "learning_rate": 1.3996413072648372e-05,
      "loss": 0.0692,
      "step": 125725
    },
    {
      "epoch": 1.8519609431378035,
      "grad_norm": 1.5960372686386108,
      "learning_rate": 1.3994776700681388e-05,
      "loss": 0.0653,
      "step": 125750
    },
    {
      "epoch": 1.8523291262278905,
      "grad_norm": 1.6690467596054077,
      "learning_rate": 1.3993140328714402e-05,
      "loss": 0.0748,
      "step": 125775
    },
    {
      "epoch": 1.8526973093179775,
      "grad_norm": 1.5556851625442505,
      "learning_rate": 1.3991503956747417e-05,
      "loss": 0.0666,
      "step": 125800
    },
    {
      "epoch": 1.8530654924080647,
      "grad_norm": 1.1716454029083252,
      "learning_rate": 1.3989867584780431e-05,
      "loss": 0.068,
      "step": 125825
    },
    {
      "epoch": 1.8534336754981517,
      "grad_norm": 1.5424468517303467,
      "learning_rate": 1.3988231212813448e-05,
      "loss": 0.0664,
      "step": 125850
    },
    {
      "epoch": 1.8538018585882388,
      "grad_norm": 1.8190423250198364,
      "learning_rate": 1.3986594840846464e-05,
      "loss": 0.0712,
      "step": 125875
    },
    {
      "epoch": 1.8541700416783258,
      "grad_norm": 1.3615065813064575,
      "learning_rate": 1.398495846887948e-05,
      "loss": 0.0686,
      "step": 125900
    },
    {
      "epoch": 1.8545382247684128,
      "grad_norm": 1.3635457754135132,
      "learning_rate": 1.3983322096912494e-05,
      "loss": 0.075,
      "step": 125925
    },
    {
      "epoch": 1.8549064078585,
      "grad_norm": 1.9268525838851929,
      "learning_rate": 1.398168572494551e-05,
      "loss": 0.0703,
      "step": 125950
    },
    {
      "epoch": 1.855274590948587,
      "grad_norm": 1.5137852430343628,
      "learning_rate": 1.3980049352978527e-05,
      "loss": 0.0651,
      "step": 125975
    },
    {
      "epoch": 1.855642774038674,
      "grad_norm": 1.2876266241073608,
      "learning_rate": 1.397841298101154e-05,
      "loss": 0.0672,
      "step": 126000
    },
    {
      "epoch": 1.856010957128761,
      "grad_norm": 1.370152473449707,
      "learning_rate": 1.3976776609044556e-05,
      "loss": 0.0648,
      "step": 126025
    },
    {
      "epoch": 1.856379140218848,
      "grad_norm": 1.3608373403549194,
      "learning_rate": 1.3975140237077572e-05,
      "loss": 0.0722,
      "step": 126050
    },
    {
      "epoch": 1.856747323308935,
      "grad_norm": 1.3997831344604492,
      "learning_rate": 1.3973503865110586e-05,
      "loss": 0.0683,
      "step": 126075
    },
    {
      "epoch": 1.8571155063990221,
      "grad_norm": 1.488750696182251,
      "learning_rate": 1.3971867493143602e-05,
      "loss": 0.0709,
      "step": 126100
    },
    {
      "epoch": 1.8574836894891091,
      "grad_norm": 1.4350106716156006,
      "learning_rate": 1.3970231121176618e-05,
      "loss": 0.0693,
      "step": 126125
    },
    {
      "epoch": 1.8578518725791962,
      "grad_norm": 1.9889986515045166,
      "learning_rate": 1.3968594749209635e-05,
      "loss": 0.0726,
      "step": 126150
    },
    {
      "epoch": 1.8582200556692832,
      "grad_norm": 0.9648709893226624,
      "learning_rate": 1.3967023832121328e-05,
      "loss": 0.0645,
      "step": 126175
    },
    {
      "epoch": 1.8585882387593702,
      "grad_norm": 1.1292203664779663,
      "learning_rate": 1.3965387460154343e-05,
      "loss": 0.0665,
      "step": 126200
    },
    {
      "epoch": 1.8589564218494572,
      "grad_norm": 1.1202218532562256,
      "learning_rate": 1.3963751088187359e-05,
      "loss": 0.0656,
      "step": 126225
    },
    {
      "epoch": 1.8593246049395442,
      "grad_norm": 1.5572593212127686,
      "learning_rate": 1.3962114716220374e-05,
      "loss": 0.0711,
      "step": 126250
    },
    {
      "epoch": 1.8596927880296312,
      "grad_norm": 1.3085103034973145,
      "learning_rate": 1.396047834425339e-05,
      "loss": 0.0727,
      "step": 126275
    },
    {
      "epoch": 1.8600609711197185,
      "grad_norm": 1.7910100221633911,
      "learning_rate": 1.3958841972286406e-05,
      "loss": 0.0713,
      "step": 126300
    },
    {
      "epoch": 1.8604291542098055,
      "grad_norm": 1.2467713356018066,
      "learning_rate": 1.3957205600319422e-05,
      "loss": 0.0687,
      "step": 126325
    },
    {
      "epoch": 1.8607973372998925,
      "grad_norm": 1.255759596824646,
      "learning_rate": 1.3955569228352436e-05,
      "loss": 0.0737,
      "step": 126350
    },
    {
      "epoch": 1.8611655203899795,
      "grad_norm": 1.3584084510803223,
      "learning_rate": 1.3953932856385453e-05,
      "loss": 0.0657,
      "step": 126375
    },
    {
      "epoch": 1.8615337034800665,
      "grad_norm": 1.4888774156570435,
      "learning_rate": 1.3952296484418469e-05,
      "loss": 0.0722,
      "step": 126400
    },
    {
      "epoch": 1.8619018865701538,
      "grad_norm": 1.4281727075576782,
      "learning_rate": 1.3950660112451482e-05,
      "loss": 0.0702,
      "step": 126425
    },
    {
      "epoch": 1.8622700696602408,
      "grad_norm": 1.7842527627944946,
      "learning_rate": 1.3949023740484498e-05,
      "loss": 0.0714,
      "step": 126450
    },
    {
      "epoch": 1.8626382527503278,
      "grad_norm": 1.6052942276000977,
      "learning_rate": 1.3947387368517514e-05,
      "loss": 0.0695,
      "step": 126475
    },
    {
      "epoch": 1.8630064358404148,
      "grad_norm": 1.844359040260315,
      "learning_rate": 1.3945750996550528e-05,
      "loss": 0.0763,
      "step": 126500
    },
    {
      "epoch": 1.8633746189305018,
      "grad_norm": 1.3661956787109375,
      "learning_rate": 1.3944114624583544e-05,
      "loss": 0.0747,
      "step": 126525
    },
    {
      "epoch": 1.8637428020205888,
      "grad_norm": 1.7540936470031738,
      "learning_rate": 1.394247825261656e-05,
      "loss": 0.0684,
      "step": 126550
    },
    {
      "epoch": 1.8641109851106759,
      "grad_norm": 1.5911238193511963,
      "learning_rate": 1.3940841880649577e-05,
      "loss": 0.0643,
      "step": 126575
    },
    {
      "epoch": 1.8644791682007629,
      "grad_norm": 1.2241568565368652,
      "learning_rate": 1.3939205508682591e-05,
      "loss": 0.0647,
      "step": 126600
    },
    {
      "epoch": 1.8648473512908499,
      "grad_norm": 1.1999590396881104,
      "learning_rate": 1.3937569136715606e-05,
      "loss": 0.0679,
      "step": 126625
    },
    {
      "epoch": 1.865215534380937,
      "grad_norm": 1.4949324131011963,
      "learning_rate": 1.393593276474862e-05,
      "loss": 0.0682,
      "step": 126650
    },
    {
      "epoch": 1.865583717471024,
      "grad_norm": 1.37181556224823,
      "learning_rate": 1.3934296392781636e-05,
      "loss": 0.0688,
      "step": 126675
    },
    {
      "epoch": 1.865951900561111,
      "grad_norm": 2.012047529220581,
      "learning_rate": 1.3932660020814652e-05,
      "loss": 0.0735,
      "step": 126700
    },
    {
      "epoch": 1.866320083651198,
      "grad_norm": 1.7181177139282227,
      "learning_rate": 1.3931023648847669e-05,
      "loss": 0.0653,
      "step": 126725
    },
    {
      "epoch": 1.866688266741285,
      "grad_norm": 1.4143445491790771,
      "learning_rate": 1.3929387276880683e-05,
      "loss": 0.0696,
      "step": 126750
    },
    {
      "epoch": 1.8670564498313722,
      "grad_norm": 1.4846135377883911,
      "learning_rate": 1.39277509049137e-05,
      "loss": 0.0752,
      "step": 126775
    },
    {
      "epoch": 1.8674246329214592,
      "grad_norm": 0.8891171813011169,
      "learning_rate": 1.3926114532946715e-05,
      "loss": 0.0706,
      "step": 126800
    },
    {
      "epoch": 1.8677928160115462,
      "grad_norm": 1.3381410837173462,
      "learning_rate": 1.3924478160979731e-05,
      "loss": 0.0659,
      "step": 126825
    },
    {
      "epoch": 1.8681609991016332,
      "grad_norm": 1.7535873651504517,
      "learning_rate": 1.3922841789012744e-05,
      "loss": 0.0697,
      "step": 126850
    },
    {
      "epoch": 1.8685291821917203,
      "grad_norm": 1.4343935251235962,
      "learning_rate": 1.392120541704576e-05,
      "loss": 0.0653,
      "step": 126875
    },
    {
      "epoch": 1.8688973652818075,
      "grad_norm": 1.4396963119506836,
      "learning_rate": 1.3919569045078777e-05,
      "loss": 0.0711,
      "step": 126900
    },
    {
      "epoch": 1.8692655483718945,
      "grad_norm": 1.6458162069320679,
      "learning_rate": 1.3917932673111791e-05,
      "loss": 0.0623,
      "step": 126925
    },
    {
      "epoch": 1.8696337314619815,
      "grad_norm": 1.4046906232833862,
      "learning_rate": 1.3916296301144807e-05,
      "loss": 0.0607,
      "step": 126950
    },
    {
      "epoch": 1.8700019145520685,
      "grad_norm": 1.6016461849212646,
      "learning_rate": 1.3914659929177823e-05,
      "loss": 0.0682,
      "step": 126975
    },
    {
      "epoch": 1.8703700976421556,
      "grad_norm": 1.6409575939178467,
      "learning_rate": 1.3913023557210838e-05,
      "loss": 0.0684,
      "step": 127000
    },
    {
      "epoch": 1.8707382807322426,
      "grad_norm": 1.0247600078582764,
      "learning_rate": 1.3911387185243854e-05,
      "loss": 0.067,
      "step": 127025
    },
    {
      "epoch": 1.8711064638223296,
      "grad_norm": 0.8390852808952332,
      "learning_rate": 1.3909750813276868e-05,
      "loss": 0.0667,
      "step": 127050
    },
    {
      "epoch": 1.8714746469124166,
      "grad_norm": 1.2949190139770508,
      "learning_rate": 1.3908114441309883e-05,
      "loss": 0.0671,
      "step": 127075
    },
    {
      "epoch": 1.8718428300025036,
      "grad_norm": 1.2676405906677246,
      "learning_rate": 1.3906478069342899e-05,
      "loss": 0.0697,
      "step": 127100
    },
    {
      "epoch": 1.8722110130925906,
      "grad_norm": 1.339078664779663,
      "learning_rate": 1.3904841697375915e-05,
      "loss": 0.0664,
      "step": 127125
    },
    {
      "epoch": 1.8725791961826777,
      "grad_norm": 1.5541309118270874,
      "learning_rate": 1.3903205325408931e-05,
      "loss": 0.0684,
      "step": 127150
    },
    {
      "epoch": 1.8729473792727647,
      "grad_norm": 1.1422303915023804,
      "learning_rate": 1.3901568953441946e-05,
      "loss": 0.0706,
      "step": 127175
    },
    {
      "epoch": 1.8733155623628517,
      "grad_norm": 1.151033639907837,
      "learning_rate": 1.3899932581474962e-05,
      "loss": 0.0639,
      "step": 127200
    },
    {
      "epoch": 1.8736837454529387,
      "grad_norm": 1.6416378021240234,
      "learning_rate": 1.3898296209507978e-05,
      "loss": 0.0704,
      "step": 127225
    },
    {
      "epoch": 1.8740519285430257,
      "grad_norm": 1.3154687881469727,
      "learning_rate": 1.3896659837540993e-05,
      "loss": 0.0624,
      "step": 127250
    },
    {
      "epoch": 1.874420111633113,
      "grad_norm": 1.4494855403900146,
      "learning_rate": 1.3895023465574007e-05,
      "loss": 0.0624,
      "step": 127275
    },
    {
      "epoch": 1.8747882947232,
      "grad_norm": 1.4901341199874878,
      "learning_rate": 1.3893387093607023e-05,
      "loss": 0.0688,
      "step": 127300
    },
    {
      "epoch": 1.875156477813287,
      "grad_norm": 1.6520740985870361,
      "learning_rate": 1.3891750721640038e-05,
      "loss": 0.0657,
      "step": 127325
    },
    {
      "epoch": 1.875524660903374,
      "grad_norm": 1.2405890226364136,
      "learning_rate": 1.3890114349673054e-05,
      "loss": 0.0714,
      "step": 127350
    },
    {
      "epoch": 1.8758928439934612,
      "grad_norm": 1.6109544038772583,
      "learning_rate": 1.388847797770607e-05,
      "loss": 0.0663,
      "step": 127375
    },
    {
      "epoch": 1.8762610270835482,
      "grad_norm": 1.5050334930419922,
      "learning_rate": 1.3886841605739086e-05,
      "loss": 0.0685,
      "step": 127400
    },
    {
      "epoch": 1.8766292101736353,
      "grad_norm": 1.5214776992797852,
      "learning_rate": 1.38852052337721e-05,
      "loss": 0.0731,
      "step": 127425
    },
    {
      "epoch": 1.8769973932637223,
      "grad_norm": 1.0754930973052979,
      "learning_rate": 1.3883568861805117e-05,
      "loss": 0.0618,
      "step": 127450
    },
    {
      "epoch": 1.8773655763538093,
      "grad_norm": 1.4267821311950684,
      "learning_rate": 1.3881932489838131e-05,
      "loss": 0.0662,
      "step": 127475
    },
    {
      "epoch": 1.8777337594438963,
      "grad_norm": 1.4020365476608276,
      "learning_rate": 1.3880296117871146e-05,
      "loss": 0.0767,
      "step": 127500
    },
    {
      "epoch": 1.8781019425339833,
      "grad_norm": 1.3116518259048462,
      "learning_rate": 1.3878659745904162e-05,
      "loss": 0.0662,
      "step": 127525
    },
    {
      "epoch": 1.8784701256240703,
      "grad_norm": 1.164931058883667,
      "learning_rate": 1.3877023373937178e-05,
      "loss": 0.0696,
      "step": 127550
    },
    {
      "epoch": 1.8788383087141574,
      "grad_norm": 1.18403160572052,
      "learning_rate": 1.3875387001970192e-05,
      "loss": 0.0663,
      "step": 127575
    },
    {
      "epoch": 1.8792064918042444,
      "grad_norm": 1.3779178857803345,
      "learning_rate": 1.3873750630003208e-05,
      "loss": 0.0704,
      "step": 127600
    },
    {
      "epoch": 1.8795746748943314,
      "grad_norm": 1.4132893085479736,
      "learning_rate": 1.3872114258036225e-05,
      "loss": 0.0677,
      "step": 127625
    },
    {
      "epoch": 1.8799428579844184,
      "grad_norm": 1.7565144300460815,
      "learning_rate": 1.387047788606924e-05,
      "loss": 0.0707,
      "step": 127650
    },
    {
      "epoch": 1.8803110410745054,
      "grad_norm": 1.2129707336425781,
      "learning_rate": 1.3868841514102254e-05,
      "loss": 0.0618,
      "step": 127675
    },
    {
      "epoch": 1.8806792241645924,
      "grad_norm": 1.627714991569519,
      "learning_rate": 1.386720514213527e-05,
      "loss": 0.0722,
      "step": 127700
    },
    {
      "epoch": 1.8810474072546794,
      "grad_norm": 1.5501409769058228,
      "learning_rate": 1.3865568770168286e-05,
      "loss": 0.0655,
      "step": 127725
    },
    {
      "epoch": 1.8814155903447667,
      "grad_norm": 1.7595138549804688,
      "learning_rate": 1.38639323982013e-05,
      "loss": 0.0745,
      "step": 127750
    },
    {
      "epoch": 1.8817837734348537,
      "grad_norm": 1.1655981540679932,
      "learning_rate": 1.3862296026234316e-05,
      "loss": 0.061,
      "step": 127775
    },
    {
      "epoch": 1.8821519565249407,
      "grad_norm": 1.4186490774154663,
      "learning_rate": 1.3860659654267333e-05,
      "loss": 0.0664,
      "step": 127800
    },
    {
      "epoch": 1.8825201396150277,
      "grad_norm": 1.4063063859939575,
      "learning_rate": 1.3859023282300347e-05,
      "loss": 0.0692,
      "step": 127825
    },
    {
      "epoch": 1.8828883227051147,
      "grad_norm": 2.1492862701416016,
      "learning_rate": 1.3857386910333363e-05,
      "loss": 0.0669,
      "step": 127850
    },
    {
      "epoch": 1.883256505795202,
      "grad_norm": 1.2822747230529785,
      "learning_rate": 1.385575053836638e-05,
      "loss": 0.0655,
      "step": 127875
    },
    {
      "epoch": 1.883624688885289,
      "grad_norm": 1.2677335739135742,
      "learning_rate": 1.3854114166399392e-05,
      "loss": 0.0781,
      "step": 127900
    },
    {
      "epoch": 1.883992871975376,
      "grad_norm": 2.0128173828125,
      "learning_rate": 1.3852477794432408e-05,
      "loss": 0.0698,
      "step": 127925
    },
    {
      "epoch": 1.884361055065463,
      "grad_norm": 1.3676215410232544,
      "learning_rate": 1.3850841422465424e-05,
      "loss": 0.06,
      "step": 127950
    },
    {
      "epoch": 1.88472923815555,
      "grad_norm": 2.009725332260132,
      "learning_rate": 1.384920505049844e-05,
      "loss": 0.0686,
      "step": 127975
    },
    {
      "epoch": 1.885097421245637,
      "grad_norm": 1.3375903367996216,
      "learning_rate": 1.3847568678531455e-05,
      "loss": 0.0631,
      "step": 128000
    },
    {
      "epoch": 1.885465604335724,
      "grad_norm": 1.6937209367752075,
      "learning_rate": 1.3845932306564471e-05,
      "loss": 0.067,
      "step": 128025
    },
    {
      "epoch": 1.885833787425811,
      "grad_norm": 1.276186227798462,
      "learning_rate": 1.3844295934597487e-05,
      "loss": 0.0735,
      "step": 128050
    },
    {
      "epoch": 1.886201970515898,
      "grad_norm": 1.1274360418319702,
      "learning_rate": 1.3842659562630503e-05,
      "loss": 0.0614,
      "step": 128075
    },
    {
      "epoch": 1.8865701536059851,
      "grad_norm": 1.258409023284912,
      "learning_rate": 1.3841023190663516e-05,
      "loss": 0.0692,
      "step": 128100
    },
    {
      "epoch": 1.8869383366960721,
      "grad_norm": 1.585135579109192,
      "learning_rate": 1.3839386818696532e-05,
      "loss": 0.0707,
      "step": 128125
    },
    {
      "epoch": 1.8873065197861592,
      "grad_norm": 1.4370063543319702,
      "learning_rate": 1.3837750446729547e-05,
      "loss": 0.0757,
      "step": 128150
    },
    {
      "epoch": 1.8876747028762462,
      "grad_norm": 1.4753754138946533,
      "learning_rate": 1.3836114074762563e-05,
      "loss": 0.073,
      "step": 128175
    },
    {
      "epoch": 1.8880428859663332,
      "grad_norm": 0.8586186766624451,
      "learning_rate": 1.3834477702795579e-05,
      "loss": 0.0692,
      "step": 128200
    },
    {
      "epoch": 1.8884110690564204,
      "grad_norm": 1.8866338729858398,
      "learning_rate": 1.3832841330828595e-05,
      "loss": 0.0754,
      "step": 128225
    },
    {
      "epoch": 1.8887792521465074,
      "grad_norm": 1.4799463748931885,
      "learning_rate": 1.383127041374029e-05,
      "loss": 0.072,
      "step": 128250
    },
    {
      "epoch": 1.8891474352365945,
      "grad_norm": 1.740657091140747,
      "learning_rate": 1.3829634041773305e-05,
      "loss": 0.0738,
      "step": 128275
    },
    {
      "epoch": 1.8895156183266815,
      "grad_norm": 1.2831006050109863,
      "learning_rate": 1.3827997669806321e-05,
      "loss": 0.0761,
      "step": 128300
    },
    {
      "epoch": 1.8898838014167685,
      "grad_norm": 1.0753891468048096,
      "learning_rate": 1.3826361297839334e-05,
      "loss": 0.0727,
      "step": 128325
    },
    {
      "epoch": 1.8902519845068557,
      "grad_norm": 1.3246605396270752,
      "learning_rate": 1.382472492587235e-05,
      "loss": 0.0738,
      "step": 128350
    },
    {
      "epoch": 1.8906201675969427,
      "grad_norm": 1.0399971008300781,
      "learning_rate": 1.3823088553905367e-05,
      "loss": 0.0668,
      "step": 128375
    },
    {
      "epoch": 1.8909883506870298,
      "grad_norm": 1.3426973819732666,
      "learning_rate": 1.3821452181938381e-05,
      "loss": 0.0675,
      "step": 128400
    },
    {
      "epoch": 1.8913565337771168,
      "grad_norm": 1.9310686588287354,
      "learning_rate": 1.3819815809971397e-05,
      "loss": 0.0693,
      "step": 128425
    },
    {
      "epoch": 1.8917247168672038,
      "grad_norm": 2.069502592086792,
      "learning_rate": 1.3818179438004413e-05,
      "loss": 0.0776,
      "step": 128450
    },
    {
      "epoch": 1.8920928999572908,
      "grad_norm": 1.4830806255340576,
      "learning_rate": 1.381654306603743e-05,
      "loss": 0.0668,
      "step": 128475
    },
    {
      "epoch": 1.8924610830473778,
      "grad_norm": 1.444730281829834,
      "learning_rate": 1.3814906694070444e-05,
      "loss": 0.0769,
      "step": 128500
    },
    {
      "epoch": 1.8928292661374648,
      "grad_norm": 1.0734021663665771,
      "learning_rate": 1.3813270322103458e-05,
      "loss": 0.0658,
      "step": 128525
    },
    {
      "epoch": 1.8931974492275518,
      "grad_norm": 1.1945226192474365,
      "learning_rate": 1.3811633950136475e-05,
      "loss": 0.0667,
      "step": 128550
    },
    {
      "epoch": 1.8935656323176389,
      "grad_norm": 1.3909984827041626,
      "learning_rate": 1.3809997578169489e-05,
      "loss": 0.0644,
      "step": 128575
    },
    {
      "epoch": 1.8939338154077259,
      "grad_norm": 1.0541220903396606,
      "learning_rate": 1.3808361206202505e-05,
      "loss": 0.0673,
      "step": 128600
    },
    {
      "epoch": 1.8943019984978129,
      "grad_norm": 1.4383426904678345,
      "learning_rate": 1.3806724834235521e-05,
      "loss": 0.0693,
      "step": 128625
    },
    {
      "epoch": 1.8946701815879,
      "grad_norm": 1.356213092803955,
      "learning_rate": 1.3805088462268537e-05,
      "loss": 0.067,
      "step": 128650
    },
    {
      "epoch": 1.895038364677987,
      "grad_norm": 0.860144853591919,
      "learning_rate": 1.3803452090301552e-05,
      "loss": 0.0695,
      "step": 128675
    },
    {
      "epoch": 1.8954065477680742,
      "grad_norm": 1.758234977722168,
      "learning_rate": 1.3801815718334568e-05,
      "loss": 0.0663,
      "step": 128700
    },
    {
      "epoch": 1.8957747308581612,
      "grad_norm": 1.6093204021453857,
      "learning_rate": 1.3800179346367584e-05,
      "loss": 0.0693,
      "step": 128725
    },
    {
      "epoch": 1.8961429139482482,
      "grad_norm": 0.980208694934845,
      "learning_rate": 1.3798542974400597e-05,
      "loss": 0.0694,
      "step": 128750
    },
    {
      "epoch": 1.8965110970383352,
      "grad_norm": 1.3171117305755615,
      "learning_rate": 1.3796906602433613e-05,
      "loss": 0.0665,
      "step": 128775
    },
    {
      "epoch": 1.8968792801284222,
      "grad_norm": 1.201322317123413,
      "learning_rate": 1.379527023046663e-05,
      "loss": 0.0659,
      "step": 128800
    },
    {
      "epoch": 1.8972474632185095,
      "grad_norm": 1.3161327838897705,
      "learning_rate": 1.3793633858499644e-05,
      "loss": 0.0615,
      "step": 128825
    },
    {
      "epoch": 1.8976156463085965,
      "grad_norm": 1.1581096649169922,
      "learning_rate": 1.379199748653266e-05,
      "loss": 0.0673,
      "step": 128850
    },
    {
      "epoch": 1.8979838293986835,
      "grad_norm": 1.4302693605422974,
      "learning_rate": 1.3790361114565676e-05,
      "loss": 0.0664,
      "step": 128875
    },
    {
      "epoch": 1.8983520124887705,
      "grad_norm": 0.9522002339363098,
      "learning_rate": 1.3788724742598692e-05,
      "loss": 0.0654,
      "step": 128900
    },
    {
      "epoch": 1.8987201955788575,
      "grad_norm": 1.4306349754333496,
      "learning_rate": 1.3787088370631707e-05,
      "loss": 0.0677,
      "step": 128925
    },
    {
      "epoch": 1.8990883786689445,
      "grad_norm": 1.559780240058899,
      "learning_rate": 1.3785451998664721e-05,
      "loss": 0.072,
      "step": 128950
    },
    {
      "epoch": 1.8994565617590315,
      "grad_norm": 1.4206688404083252,
      "learning_rate": 1.3783815626697736e-05,
      "loss": 0.0714,
      "step": 128975
    },
    {
      "epoch": 1.8998247448491186,
      "grad_norm": 2.223038911819458,
      "learning_rate": 1.3782179254730752e-05,
      "loss": 0.066,
      "step": 129000
    },
    {
      "epoch": 1.9001929279392056,
      "grad_norm": 1.3395779132843018,
      "learning_rate": 1.3780542882763768e-05,
      "loss": 0.0667,
      "step": 129025
    },
    {
      "epoch": 1.9005611110292926,
      "grad_norm": 1.8482961654663086,
      "learning_rate": 1.3778906510796784e-05,
      "loss": 0.0682,
      "step": 129050
    },
    {
      "epoch": 1.9009292941193796,
      "grad_norm": 1.177207589149475,
      "learning_rate": 1.3777270138829798e-05,
      "loss": 0.0626,
      "step": 129075
    },
    {
      "epoch": 1.9012974772094666,
      "grad_norm": 1.647931456565857,
      "learning_rate": 1.3775633766862815e-05,
      "loss": 0.0672,
      "step": 129100
    },
    {
      "epoch": 1.9016656602995536,
      "grad_norm": 1.225214958190918,
      "learning_rate": 1.377399739489583e-05,
      "loss": 0.0709,
      "step": 129125
    },
    {
      "epoch": 1.9020338433896407,
      "grad_norm": 1.8574533462524414,
      "learning_rate": 1.3772361022928844e-05,
      "loss": 0.074,
      "step": 129150
    },
    {
      "epoch": 1.902402026479728,
      "grad_norm": 1.0986663103103638,
      "learning_rate": 1.377072465096186e-05,
      "loss": 0.0735,
      "step": 129175
    },
    {
      "epoch": 1.902770209569815,
      "grad_norm": 1.047900676727295,
      "learning_rate": 1.3769088278994876e-05,
      "loss": 0.0634,
      "step": 129200
    },
    {
      "epoch": 1.903138392659902,
      "grad_norm": 1.5773135423660278,
      "learning_rate": 1.3767451907027892e-05,
      "loss": 0.0704,
      "step": 129225
    },
    {
      "epoch": 1.903506575749989,
      "grad_norm": 1.3682206869125366,
      "learning_rate": 1.3765815535060906e-05,
      "loss": 0.0643,
      "step": 129250
    },
    {
      "epoch": 1.903874758840076,
      "grad_norm": 1.2353607416152954,
      "learning_rate": 1.3764179163093923e-05,
      "loss": 0.0647,
      "step": 129275
    },
    {
      "epoch": 1.9042429419301632,
      "grad_norm": 1.4179573059082031,
      "learning_rate": 1.3762542791126939e-05,
      "loss": 0.0651,
      "step": 129300
    },
    {
      "epoch": 1.9046111250202502,
      "grad_norm": 1.4659348726272583,
      "learning_rate": 1.3760906419159953e-05,
      "loss": 0.0688,
      "step": 129325
    },
    {
      "epoch": 1.9049793081103372,
      "grad_norm": 1.4568201303482056,
      "learning_rate": 1.375927004719297e-05,
      "loss": 0.0729,
      "step": 129350
    },
    {
      "epoch": 1.9053474912004242,
      "grad_norm": 1.249602198600769,
      "learning_rate": 1.3757633675225984e-05,
      "loss": 0.0692,
      "step": 129375
    },
    {
      "epoch": 1.9057156742905113,
      "grad_norm": 0.8342622518539429,
      "learning_rate": 1.3755997303258998e-05,
      "loss": 0.0691,
      "step": 129400
    },
    {
      "epoch": 1.9060838573805983,
      "grad_norm": 1.3442509174346924,
      "learning_rate": 1.3754360931292014e-05,
      "loss": 0.0625,
      "step": 129425
    },
    {
      "epoch": 1.9064520404706853,
      "grad_norm": 1.4729362726211548,
      "learning_rate": 1.375272455932503e-05,
      "loss": 0.0718,
      "step": 129450
    },
    {
      "epoch": 1.9068202235607723,
      "grad_norm": 1.4642016887664795,
      "learning_rate": 1.3751088187358047e-05,
      "loss": 0.068,
      "step": 129475
    },
    {
      "epoch": 1.9071884066508593,
      "grad_norm": 1.5003490447998047,
      "learning_rate": 1.3749451815391061e-05,
      "loss": 0.0622,
      "step": 129500
    },
    {
      "epoch": 1.9075565897409463,
      "grad_norm": 1.138119101524353,
      "learning_rate": 1.3747815443424077e-05,
      "loss": 0.0742,
      "step": 129525
    },
    {
      "epoch": 1.9079247728310333,
      "grad_norm": 1.7713483572006226,
      "learning_rate": 1.3746179071457093e-05,
      "loss": 0.0701,
      "step": 129550
    },
    {
      "epoch": 1.9082929559211204,
      "grad_norm": 1.2612457275390625,
      "learning_rate": 1.3744542699490106e-05,
      "loss": 0.0674,
      "step": 129575
    },
    {
      "epoch": 1.9086611390112074,
      "grad_norm": 1.200155258178711,
      "learning_rate": 1.3742906327523122e-05,
      "loss": 0.0632,
      "step": 129600
    },
    {
      "epoch": 1.9090293221012944,
      "grad_norm": 1.6533445119857788,
      "learning_rate": 1.3741269955556139e-05,
      "loss": 0.0706,
      "step": 129625
    },
    {
      "epoch": 1.9093975051913814,
      "grad_norm": 2.0837175846099854,
      "learning_rate": 1.3739633583589153e-05,
      "loss": 0.0741,
      "step": 129650
    },
    {
      "epoch": 1.9097656882814686,
      "grad_norm": 1.1843743324279785,
      "learning_rate": 1.373799721162217e-05,
      "loss": 0.0627,
      "step": 129675
    },
    {
      "epoch": 1.9101338713715557,
      "grad_norm": 1.6287152767181396,
      "learning_rate": 1.3736360839655185e-05,
      "loss": 0.0688,
      "step": 129700
    },
    {
      "epoch": 1.9105020544616427,
      "grad_norm": 1.1749218702316284,
      "learning_rate": 1.3734724467688201e-05,
      "loss": 0.0699,
      "step": 129725
    },
    {
      "epoch": 1.9108702375517297,
      "grad_norm": 1.2098321914672852,
      "learning_rate": 1.3733088095721216e-05,
      "loss": 0.0638,
      "step": 129750
    },
    {
      "epoch": 1.911238420641817,
      "grad_norm": 1.307051658630371,
      "learning_rate": 1.3731451723754232e-05,
      "loss": 0.0692,
      "step": 129775
    },
    {
      "epoch": 1.911606603731904,
      "grad_norm": 1.4983049631118774,
      "learning_rate": 1.3729815351787245e-05,
      "loss": 0.0715,
      "step": 129800
    },
    {
      "epoch": 1.911974786821991,
      "grad_norm": 1.3979671001434326,
      "learning_rate": 1.3728178979820261e-05,
      "loss": 0.0715,
      "step": 129825
    },
    {
      "epoch": 1.912342969912078,
      "grad_norm": 1.799364686012268,
      "learning_rate": 1.3726542607853277e-05,
      "loss": 0.0704,
      "step": 129850
    },
    {
      "epoch": 1.912711153002165,
      "grad_norm": 1.204068660736084,
      "learning_rate": 1.3724906235886293e-05,
      "loss": 0.0697,
      "step": 129875
    },
    {
      "epoch": 1.913079336092252,
      "grad_norm": 1.6166411638259888,
      "learning_rate": 1.3723269863919308e-05,
      "loss": 0.0705,
      "step": 129900
    },
    {
      "epoch": 1.913447519182339,
      "grad_norm": 1.3805930614471436,
      "learning_rate": 1.3721633491952324e-05,
      "loss": 0.064,
      "step": 129925
    },
    {
      "epoch": 1.913815702272426,
      "grad_norm": 1.4439527988433838,
      "learning_rate": 1.371999711998534e-05,
      "loss": 0.068,
      "step": 129950
    },
    {
      "epoch": 1.914183885362513,
      "grad_norm": 1.1924982070922852,
      "learning_rate": 1.3718360748018356e-05,
      "loss": 0.0655,
      "step": 129975
    },
    {
      "epoch": 1.9145520684526,
      "grad_norm": 1.069240927696228,
      "learning_rate": 1.3716724376051369e-05,
      "loss": 0.073,
      "step": 130000
    },
    {
      "epoch": 1.914920251542687,
      "grad_norm": 1.7875123023986816,
      "learning_rate": 1.3715088004084385e-05,
      "loss": 0.0705,
      "step": 130025
    },
    {
      "epoch": 1.915288434632774,
      "grad_norm": 1.511775255203247,
      "learning_rate": 1.3713451632117401e-05,
      "loss": 0.0648,
      "step": 130050
    },
    {
      "epoch": 1.915656617722861,
      "grad_norm": 1.9179155826568604,
      "learning_rate": 1.3711815260150416e-05,
      "loss": 0.0628,
      "step": 130075
    },
    {
      "epoch": 1.9160248008129481,
      "grad_norm": 1.4427399635314941,
      "learning_rate": 1.3710178888183432e-05,
      "loss": 0.071,
      "step": 130100
    },
    {
      "epoch": 1.9163929839030351,
      "grad_norm": 1.3722892999649048,
      "learning_rate": 1.3708542516216448e-05,
      "loss": 0.069,
      "step": 130125
    },
    {
      "epoch": 1.9167611669931224,
      "grad_norm": 1.374695062637329,
      "learning_rate": 1.3706906144249462e-05,
      "loss": 0.0675,
      "step": 130150
    },
    {
      "epoch": 1.9171293500832094,
      "grad_norm": 1.3429064750671387,
      "learning_rate": 1.3705269772282479e-05,
      "loss": 0.0716,
      "step": 130175
    },
    {
      "epoch": 1.9174975331732964,
      "grad_norm": 1.6231729984283447,
      "learning_rate": 1.3703633400315495e-05,
      "loss": 0.0636,
      "step": 130200
    },
    {
      "epoch": 1.9178657162633834,
      "grad_norm": 1.3547338247299194,
      "learning_rate": 1.3701997028348508e-05,
      "loss": 0.0685,
      "step": 130225
    },
    {
      "epoch": 1.9182338993534704,
      "grad_norm": 0.9752511382102966,
      "learning_rate": 1.3700360656381524e-05,
      "loss": 0.0731,
      "step": 130250
    },
    {
      "epoch": 1.9186020824435577,
      "grad_norm": 1.4045292139053345,
      "learning_rate": 1.369872428441454e-05,
      "loss": 0.0644,
      "step": 130275
    },
    {
      "epoch": 1.9189702655336447,
      "grad_norm": 1.7658567428588867,
      "learning_rate": 1.3697087912447556e-05,
      "loss": 0.0664,
      "step": 130300
    },
    {
      "epoch": 1.9193384486237317,
      "grad_norm": 1.4007624387741089,
      "learning_rate": 1.369545154048057e-05,
      "loss": 0.0765,
      "step": 130325
    },
    {
      "epoch": 1.9197066317138187,
      "grad_norm": 1.3450874090194702,
      "learning_rate": 1.3693815168513587e-05,
      "loss": 0.0684,
      "step": 130350
    },
    {
      "epoch": 1.9200748148039057,
      "grad_norm": 1.3079489469528198,
      "learning_rate": 1.3692178796546603e-05,
      "loss": 0.0669,
      "step": 130375
    },
    {
      "epoch": 1.9204429978939928,
      "grad_norm": 1.468977928161621,
      "learning_rate": 1.3690542424579619e-05,
      "loss": 0.068,
      "step": 130400
    },
    {
      "epoch": 1.9208111809840798,
      "grad_norm": 1.5864746570587158,
      "learning_rate": 1.3688906052612632e-05,
      "loss": 0.0734,
      "step": 130425
    },
    {
      "epoch": 1.9211793640741668,
      "grad_norm": 1.0984007120132446,
      "learning_rate": 1.3687269680645648e-05,
      "loss": 0.0688,
      "step": 130450
    },
    {
      "epoch": 1.9215475471642538,
      "grad_norm": 2.844667911529541,
      "learning_rate": 1.3685633308678662e-05,
      "loss": 0.069,
      "step": 130475
    },
    {
      "epoch": 1.9219157302543408,
      "grad_norm": 1.331337332725525,
      "learning_rate": 1.3683996936711678e-05,
      "loss": 0.0589,
      "step": 130500
    },
    {
      "epoch": 1.9222839133444278,
      "grad_norm": 1.4961379766464233,
      "learning_rate": 1.3682360564744695e-05,
      "loss": 0.0739,
      "step": 130525
    },
    {
      "epoch": 1.9226520964345148,
      "grad_norm": 0.9069755673408508,
      "learning_rate": 1.368072419277771e-05,
      "loss": 0.0714,
      "step": 130550
    },
    {
      "epoch": 1.9230202795246019,
      "grad_norm": 1.4977551698684692,
      "learning_rate": 1.3679087820810725e-05,
      "loss": 0.068,
      "step": 130575
    },
    {
      "epoch": 1.9233884626146889,
      "grad_norm": 1.2419164180755615,
      "learning_rate": 1.3677451448843741e-05,
      "loss": 0.0629,
      "step": 130600
    },
    {
      "epoch": 1.9237566457047761,
      "grad_norm": 1.3516414165496826,
      "learning_rate": 1.3675815076876756e-05,
      "loss": 0.0602,
      "step": 130625
    },
    {
      "epoch": 1.9241248287948631,
      "grad_norm": 0.9260694980621338,
      "learning_rate": 1.367417870490977e-05,
      "loss": 0.072,
      "step": 130650
    },
    {
      "epoch": 1.9244930118849501,
      "grad_norm": 1.580206274986267,
      "learning_rate": 1.3672542332942786e-05,
      "loss": 0.0787,
      "step": 130675
    },
    {
      "epoch": 1.9248611949750372,
      "grad_norm": 1.5083047151565552,
      "learning_rate": 1.3670905960975803e-05,
      "loss": 0.0696,
      "step": 130700
    },
    {
      "epoch": 1.9252293780651242,
      "grad_norm": 1.5534709692001343,
      "learning_rate": 1.3669335043887496e-05,
      "loss": 0.073,
      "step": 130725
    },
    {
      "epoch": 1.9255975611552114,
      "grad_norm": 0.7059693932533264,
      "learning_rate": 1.3667698671920513e-05,
      "loss": 0.0654,
      "step": 130750
    },
    {
      "epoch": 1.9259657442452984,
      "grad_norm": 1.315678358078003,
      "learning_rate": 1.3666062299953529e-05,
      "loss": 0.0696,
      "step": 130775
    },
    {
      "epoch": 1.9263339273353854,
      "grad_norm": 1.7695732116699219,
      "learning_rate": 1.3664425927986545e-05,
      "loss": 0.0677,
      "step": 130800
    },
    {
      "epoch": 1.9267021104254725,
      "grad_norm": 1.2399245500564575,
      "learning_rate": 1.366278955601956e-05,
      "loss": 0.0616,
      "step": 130825
    },
    {
      "epoch": 1.9270702935155595,
      "grad_norm": 1.3680822849273682,
      "learning_rate": 1.3661153184052574e-05,
      "loss": 0.0677,
      "step": 130850
    },
    {
      "epoch": 1.9274384766056465,
      "grad_norm": 1.2249548435211182,
      "learning_rate": 1.365951681208559e-05,
      "loss": 0.0683,
      "step": 130875
    },
    {
      "epoch": 1.9278066596957335,
      "grad_norm": 1.0708106756210327,
      "learning_rate": 1.3657880440118604e-05,
      "loss": 0.0701,
      "step": 130900
    },
    {
      "epoch": 1.9281748427858205,
      "grad_norm": 1.7924996614456177,
      "learning_rate": 1.365624406815162e-05,
      "loss": 0.063,
      "step": 130925
    },
    {
      "epoch": 1.9285430258759075,
      "grad_norm": 1.2274212837219238,
      "learning_rate": 1.3654607696184637e-05,
      "loss": 0.0643,
      "step": 130950
    },
    {
      "epoch": 1.9289112089659946,
      "grad_norm": 1.825490117073059,
      "learning_rate": 1.3652971324217653e-05,
      "loss": 0.0665,
      "step": 130975
    },
    {
      "epoch": 1.9292793920560816,
      "grad_norm": 1.2124346494674683,
      "learning_rate": 1.3651334952250667e-05,
      "loss": 0.0684,
      "step": 131000
    },
    {
      "epoch": 1.9296475751461686,
      "grad_norm": 1.2271982431411743,
      "learning_rate": 1.3649698580283683e-05,
      "loss": 0.068,
      "step": 131025
    },
    {
      "epoch": 1.9300157582362556,
      "grad_norm": 1.4989469051361084,
      "learning_rate": 1.3648062208316696e-05,
      "loss": 0.0696,
      "step": 131050
    },
    {
      "epoch": 1.9303839413263426,
      "grad_norm": 1.6688058376312256,
      "learning_rate": 1.3646425836349712e-05,
      "loss": 0.0631,
      "step": 131075
    },
    {
      "epoch": 1.9307521244164298,
      "grad_norm": 1.2299247980117798,
      "learning_rate": 1.3644789464382729e-05,
      "loss": 0.0704,
      "step": 131100
    },
    {
      "epoch": 1.9311203075065169,
      "grad_norm": 1.480790376663208,
      "learning_rate": 1.3643153092415745e-05,
      "loss": 0.0672,
      "step": 131125
    },
    {
      "epoch": 1.9314884905966039,
      "grad_norm": 1.5254203081130981,
      "learning_rate": 1.364151672044876e-05,
      "loss": 0.0673,
      "step": 131150
    },
    {
      "epoch": 1.931856673686691,
      "grad_norm": 1.2582379579544067,
      "learning_rate": 1.3639880348481775e-05,
      "loss": 0.0711,
      "step": 131175
    },
    {
      "epoch": 1.932224856776778,
      "grad_norm": 1.3952453136444092,
      "learning_rate": 1.3638243976514791e-05,
      "loss": 0.0628,
      "step": 131200
    },
    {
      "epoch": 1.9325930398668651,
      "grad_norm": 1.3503388166427612,
      "learning_rate": 1.3636607604547808e-05,
      "loss": 0.0639,
      "step": 131225
    },
    {
      "epoch": 1.9329612229569522,
      "grad_norm": 1.311848521232605,
      "learning_rate": 1.3634971232580822e-05,
      "loss": 0.0705,
      "step": 131250
    },
    {
      "epoch": 1.9333294060470392,
      "grad_norm": 1.1125034093856812,
      "learning_rate": 1.3633334860613837e-05,
      "loss": 0.0634,
      "step": 131275
    },
    {
      "epoch": 1.9336975891371262,
      "grad_norm": 1.7929441928863525,
      "learning_rate": 1.3631698488646851e-05,
      "loss": 0.0651,
      "step": 131300
    },
    {
      "epoch": 1.9340657722272132,
      "grad_norm": 1.1150362491607666,
      "learning_rate": 1.3630062116679867e-05,
      "loss": 0.0652,
      "step": 131325
    },
    {
      "epoch": 1.9344339553173002,
      "grad_norm": 1.198957085609436,
      "learning_rate": 1.3628425744712883e-05,
      "loss": 0.0604,
      "step": 131350
    },
    {
      "epoch": 1.9348021384073872,
      "grad_norm": 1.4122329950332642,
      "learning_rate": 1.36267893727459e-05,
      "loss": 0.0686,
      "step": 131375
    },
    {
      "epoch": 1.9351703214974743,
      "grad_norm": 1.4253836870193481,
      "learning_rate": 1.3625153000778914e-05,
      "loss": 0.0673,
      "step": 131400
    },
    {
      "epoch": 1.9355385045875613,
      "grad_norm": 1.620025873184204,
      "learning_rate": 1.362351662881193e-05,
      "loss": 0.0697,
      "step": 131425
    },
    {
      "epoch": 1.9359066876776483,
      "grad_norm": 1.8214222192764282,
      "learning_rate": 1.3621880256844946e-05,
      "loss": 0.0668,
      "step": 131450
    },
    {
      "epoch": 1.9362748707677353,
      "grad_norm": 1.322893500328064,
      "learning_rate": 1.3620243884877959e-05,
      "loss": 0.0723,
      "step": 131475
    },
    {
      "epoch": 1.9366430538578223,
      "grad_norm": 0.9099249839782715,
      "learning_rate": 1.3618607512910975e-05,
      "loss": 0.0656,
      "step": 131500
    },
    {
      "epoch": 1.9370112369479093,
      "grad_norm": 1.1030004024505615,
      "learning_rate": 1.3616971140943991e-05,
      "loss": 0.0648,
      "step": 131525
    },
    {
      "epoch": 1.9373794200379963,
      "grad_norm": 1.1647626161575317,
      "learning_rate": 1.3615334768977006e-05,
      "loss": 0.0682,
      "step": 131550
    },
    {
      "epoch": 1.9377476031280836,
      "grad_norm": 1.7726775407791138,
      "learning_rate": 1.3613698397010022e-05,
      "loss": 0.0707,
      "step": 131575
    },
    {
      "epoch": 1.9381157862181706,
      "grad_norm": 1.5560965538024902,
      "learning_rate": 1.3612062025043038e-05,
      "loss": 0.0764,
      "step": 131600
    },
    {
      "epoch": 1.9384839693082576,
      "grad_norm": 1.5568289756774902,
      "learning_rate": 1.3610425653076054e-05,
      "loss": 0.0724,
      "step": 131625
    },
    {
      "epoch": 1.9388521523983446,
      "grad_norm": 1.0233157873153687,
      "learning_rate": 1.3608789281109069e-05,
      "loss": 0.0692,
      "step": 131650
    },
    {
      "epoch": 1.9392203354884316,
      "grad_norm": 1.8779631853103638,
      "learning_rate": 1.3607152909142085e-05,
      "loss": 0.0726,
      "step": 131675
    },
    {
      "epoch": 1.9395885185785189,
      "grad_norm": 1.4789386987686157,
      "learning_rate": 1.36055165371751e-05,
      "loss": 0.0652,
      "step": 131700
    },
    {
      "epoch": 1.939956701668606,
      "grad_norm": 1.684277057647705,
      "learning_rate": 1.3603880165208114e-05,
      "loss": 0.0596,
      "step": 131725
    },
    {
      "epoch": 1.940324884758693,
      "grad_norm": 1.172790288925171,
      "learning_rate": 1.360224379324113e-05,
      "loss": 0.0653,
      "step": 131750
    },
    {
      "epoch": 1.94069306784878,
      "grad_norm": 1.4107803106307983,
      "learning_rate": 1.3600607421274146e-05,
      "loss": 0.0652,
      "step": 131775
    },
    {
      "epoch": 1.941061250938867,
      "grad_norm": 1.1625288724899292,
      "learning_rate": 1.3598971049307162e-05,
      "loss": 0.0644,
      "step": 131800
    },
    {
      "epoch": 1.941429434028954,
      "grad_norm": 1.083372950553894,
      "learning_rate": 1.3597334677340177e-05,
      "loss": 0.0705,
      "step": 131825
    },
    {
      "epoch": 1.941797617119041,
      "grad_norm": 1.3767958879470825,
      "learning_rate": 1.3595698305373193e-05,
      "loss": 0.0673,
      "step": 131850
    },
    {
      "epoch": 1.942165800209128,
      "grad_norm": 1.253279209136963,
      "learning_rate": 1.3594061933406209e-05,
      "loss": 0.0674,
      "step": 131875
    },
    {
      "epoch": 1.942533983299215,
      "grad_norm": 1.4885085821151733,
      "learning_rate": 1.3592425561439222e-05,
      "loss": 0.066,
      "step": 131900
    },
    {
      "epoch": 1.942902166389302,
      "grad_norm": 1.308825969696045,
      "learning_rate": 1.3590789189472238e-05,
      "loss": 0.0617,
      "step": 131925
    },
    {
      "epoch": 1.943270349479389,
      "grad_norm": 1.5395394563674927,
      "learning_rate": 1.3589152817505254e-05,
      "loss": 0.072,
      "step": 131950
    },
    {
      "epoch": 1.943638532569476,
      "grad_norm": 1.3309717178344727,
      "learning_rate": 1.3587516445538268e-05,
      "loss": 0.0634,
      "step": 131975
    },
    {
      "epoch": 1.944006715659563,
      "grad_norm": 1.2955114841461182,
      "learning_rate": 1.3585880073571285e-05,
      "loss": 0.0609,
      "step": 132000
    },
    {
      "epoch": 1.94437489874965,
      "grad_norm": 1.6768534183502197,
      "learning_rate": 1.35842437016043e-05,
      "loss": 0.074,
      "step": 132025
    },
    {
      "epoch": 1.9447430818397373,
      "grad_norm": 1.7316994667053223,
      "learning_rate": 1.3582607329637317e-05,
      "loss": 0.0711,
      "step": 132050
    },
    {
      "epoch": 1.9451112649298243,
      "grad_norm": 1.072589635848999,
      "learning_rate": 1.3580970957670331e-05,
      "loss": 0.0719,
      "step": 132075
    },
    {
      "epoch": 1.9454794480199114,
      "grad_norm": 1.1211479902267456,
      "learning_rate": 1.3579334585703348e-05,
      "loss": 0.0722,
      "step": 132100
    },
    {
      "epoch": 1.9458476311099984,
      "grad_norm": 1.4665725231170654,
      "learning_rate": 1.357769821373636e-05,
      "loss": 0.0613,
      "step": 132125
    },
    {
      "epoch": 1.9462158142000854,
      "grad_norm": 1.207033634185791,
      "learning_rate": 1.3576061841769376e-05,
      "loss": 0.0597,
      "step": 132150
    },
    {
      "epoch": 1.9465839972901726,
      "grad_norm": 1.6699589490890503,
      "learning_rate": 1.3574425469802393e-05,
      "loss": 0.0634,
      "step": 132175
    },
    {
      "epoch": 1.9469521803802596,
      "grad_norm": 0.996124267578125,
      "learning_rate": 1.3572789097835409e-05,
      "loss": 0.0674,
      "step": 132200
    },
    {
      "epoch": 1.9473203634703466,
      "grad_norm": 1.4296915531158447,
      "learning_rate": 1.3571152725868423e-05,
      "loss": 0.0707,
      "step": 132225
    },
    {
      "epoch": 1.9476885465604337,
      "grad_norm": 1.1712855100631714,
      "learning_rate": 1.356951635390144e-05,
      "loss": 0.0721,
      "step": 132250
    },
    {
      "epoch": 1.9480567296505207,
      "grad_norm": 1.4523475170135498,
      "learning_rate": 1.3567879981934455e-05,
      "loss": 0.0717,
      "step": 132275
    },
    {
      "epoch": 1.9484249127406077,
      "grad_norm": 1.652687907218933,
      "learning_rate": 1.3566243609967472e-05,
      "loss": 0.077,
      "step": 132300
    },
    {
      "epoch": 1.9487930958306947,
      "grad_norm": 1.402045726776123,
      "learning_rate": 1.3564607238000484e-05,
      "loss": 0.0647,
      "step": 132325
    },
    {
      "epoch": 1.9491612789207817,
      "grad_norm": 1.2048203945159912,
      "learning_rate": 1.35629708660335e-05,
      "loss": 0.0722,
      "step": 132350
    },
    {
      "epoch": 1.9495294620108687,
      "grad_norm": 1.4486668109893799,
      "learning_rate": 1.3561334494066517e-05,
      "loss": 0.0595,
      "step": 132375
    },
    {
      "epoch": 1.9498976451009558,
      "grad_norm": 0.9311128854751587,
      "learning_rate": 1.3559698122099531e-05,
      "loss": 0.0684,
      "step": 132400
    },
    {
      "epoch": 1.9502658281910428,
      "grad_norm": 1.1343120336532593,
      "learning_rate": 1.3558061750132547e-05,
      "loss": 0.0608,
      "step": 132425
    },
    {
      "epoch": 1.9506340112811298,
      "grad_norm": 1.9259246587753296,
      "learning_rate": 1.3556425378165563e-05,
      "loss": 0.0619,
      "step": 132450
    },
    {
      "epoch": 1.9510021943712168,
      "grad_norm": 1.3511766195297241,
      "learning_rate": 1.3554789006198578e-05,
      "loss": 0.072,
      "step": 132475
    },
    {
      "epoch": 1.9513703774613038,
      "grad_norm": 1.5985580682754517,
      "learning_rate": 1.3553152634231594e-05,
      "loss": 0.0603,
      "step": 132500
    },
    {
      "epoch": 1.9517385605513908,
      "grad_norm": 2.1103451251983643,
      "learning_rate": 1.3551516262264609e-05,
      "loss": 0.0743,
      "step": 132525
    },
    {
      "epoch": 1.952106743641478,
      "grad_norm": 1.3271312713623047,
      "learning_rate": 1.3549879890297623e-05,
      "loss": 0.0611,
      "step": 132550
    },
    {
      "epoch": 1.952474926731565,
      "grad_norm": 1.4392950534820557,
      "learning_rate": 1.3548243518330639e-05,
      "loss": 0.0693,
      "step": 132575
    },
    {
      "epoch": 1.952843109821652,
      "grad_norm": 1.2489006519317627,
      "learning_rate": 1.3546607146363655e-05,
      "loss": 0.0575,
      "step": 132600
    },
    {
      "epoch": 1.9532112929117391,
      "grad_norm": 1.6113096475601196,
      "learning_rate": 1.3544970774396671e-05,
      "loss": 0.0686,
      "step": 132625
    },
    {
      "epoch": 1.9535794760018264,
      "grad_norm": 1.4142565727233887,
      "learning_rate": 1.3543334402429686e-05,
      "loss": 0.067,
      "step": 132650
    },
    {
      "epoch": 1.9539476590919134,
      "grad_norm": 1.5173475742340088,
      "learning_rate": 1.3541698030462702e-05,
      "loss": 0.0629,
      "step": 132675
    },
    {
      "epoch": 1.9543158421820004,
      "grad_norm": 1.070237398147583,
      "learning_rate": 1.3540061658495718e-05,
      "loss": 0.0642,
      "step": 132700
    },
    {
      "epoch": 1.9546840252720874,
      "grad_norm": 1.234835147857666,
      "learning_rate": 1.3538425286528734e-05,
      "loss": 0.0662,
      "step": 132725
    },
    {
      "epoch": 1.9550522083621744,
      "grad_norm": 1.0772147178649902,
      "learning_rate": 1.3536788914561747e-05,
      "loss": 0.0641,
      "step": 132750
    },
    {
      "epoch": 1.9554203914522614,
      "grad_norm": 1.4735125303268433,
      "learning_rate": 1.3535152542594763e-05,
      "loss": 0.0706,
      "step": 132775
    },
    {
      "epoch": 1.9557885745423484,
      "grad_norm": 1.2210239171981812,
      "learning_rate": 1.3533516170627778e-05,
      "loss": 0.0656,
      "step": 132800
    },
    {
      "epoch": 1.9561567576324355,
      "grad_norm": 1.1605870723724365,
      "learning_rate": 1.3531879798660794e-05,
      "loss": 0.0723,
      "step": 132825
    },
    {
      "epoch": 1.9565249407225225,
      "grad_norm": 1.618125319480896,
      "learning_rate": 1.353024342669381e-05,
      "loss": 0.0663,
      "step": 132850
    },
    {
      "epoch": 1.9568931238126095,
      "grad_norm": 1.441646695137024,
      "learning_rate": 1.3528607054726826e-05,
      "loss": 0.0651,
      "step": 132875
    },
    {
      "epoch": 1.9572613069026965,
      "grad_norm": 1.383882761001587,
      "learning_rate": 1.352697068275984e-05,
      "loss": 0.0689,
      "step": 132900
    },
    {
      "epoch": 1.9576294899927835,
      "grad_norm": 1.426594853401184,
      "learning_rate": 1.3525334310792857e-05,
      "loss": 0.0692,
      "step": 132925
    },
    {
      "epoch": 1.9579976730828705,
      "grad_norm": 1.6448640823364258,
      "learning_rate": 1.3523697938825871e-05,
      "loss": 0.0695,
      "step": 132950
    },
    {
      "epoch": 1.9583658561729576,
      "grad_norm": 1.2162638902664185,
      "learning_rate": 1.3522061566858886e-05,
      "loss": 0.0695,
      "step": 132975
    },
    {
      "epoch": 1.9587340392630446,
      "grad_norm": 0.8742014169692993,
      "learning_rate": 1.3520425194891902e-05,
      "loss": 0.0556,
      "step": 133000
    },
    {
      "epoch": 1.9591022223531318,
      "grad_norm": 1.3841526508331299,
      "learning_rate": 1.3518788822924918e-05,
      "loss": 0.0607,
      "step": 133025
    },
    {
      "epoch": 1.9594704054432188,
      "grad_norm": 1.3889743089675903,
      "learning_rate": 1.3517152450957932e-05,
      "loss": 0.0673,
      "step": 133050
    },
    {
      "epoch": 1.9598385885333058,
      "grad_norm": 1.669912576675415,
      "learning_rate": 1.3515516078990949e-05,
      "loss": 0.067,
      "step": 133075
    },
    {
      "epoch": 1.9602067716233929,
      "grad_norm": 1.5863838195800781,
      "learning_rate": 1.3513879707023965e-05,
      "loss": 0.0761,
      "step": 133100
    },
    {
      "epoch": 1.9605749547134799,
      "grad_norm": 1.5423662662506104,
      "learning_rate": 1.3512243335056981e-05,
      "loss": 0.0588,
      "step": 133125
    },
    {
      "epoch": 1.960943137803567,
      "grad_norm": 1.1435093879699707,
      "learning_rate": 1.3510606963089995e-05,
      "loss": 0.0643,
      "step": 133150
    },
    {
      "epoch": 1.9613113208936541,
      "grad_norm": 1.0480999946594238,
      "learning_rate": 1.350897059112301e-05,
      "loss": 0.0637,
      "step": 133175
    },
    {
      "epoch": 1.9616795039837411,
      "grad_norm": 1.140135407447815,
      "learning_rate": 1.3507334219156026e-05,
      "loss": 0.0695,
      "step": 133200
    },
    {
      "epoch": 1.9620476870738282,
      "grad_norm": 1.8027437925338745,
      "learning_rate": 1.350569784718904e-05,
      "loss": 0.068,
      "step": 133225
    },
    {
      "epoch": 1.9624158701639152,
      "grad_norm": 1.075918197631836,
      "learning_rate": 1.3504061475222057e-05,
      "loss": 0.0603,
      "step": 133250
    },
    {
      "epoch": 1.9627840532540022,
      "grad_norm": 1.0406928062438965,
      "learning_rate": 1.3502425103255073e-05,
      "loss": 0.0636,
      "step": 133275
    },
    {
      "epoch": 1.9631522363440892,
      "grad_norm": 1.8164253234863281,
      "learning_rate": 1.3500788731288089e-05,
      "loss": 0.0692,
      "step": 133300
    },
    {
      "epoch": 1.9635204194341762,
      "grad_norm": 2.3341472148895264,
      "learning_rate": 1.3499152359321103e-05,
      "loss": 0.0684,
      "step": 133325
    },
    {
      "epoch": 1.9638886025242632,
      "grad_norm": 1.5393311977386475,
      "learning_rate": 1.349751598735412e-05,
      "loss": 0.0695,
      "step": 133350
    },
    {
      "epoch": 1.9642567856143502,
      "grad_norm": 0.9977759718894958,
      "learning_rate": 1.3495879615387132e-05,
      "loss": 0.0675,
      "step": 133375
    },
    {
      "epoch": 1.9646249687044373,
      "grad_norm": 1.6607754230499268,
      "learning_rate": 1.3494243243420148e-05,
      "loss": 0.072,
      "step": 133400
    },
    {
      "epoch": 1.9649931517945243,
      "grad_norm": 1.4092211723327637,
      "learning_rate": 1.3492606871453165e-05,
      "loss": 0.0699,
      "step": 133425
    },
    {
      "epoch": 1.9653613348846113,
      "grad_norm": 1.3322073221206665,
      "learning_rate": 1.349097049948618e-05,
      "loss": 0.0642,
      "step": 133450
    },
    {
      "epoch": 1.9657295179746983,
      "grad_norm": 1.4318546056747437,
      "learning_rate": 1.3489334127519195e-05,
      "loss": 0.0626,
      "step": 133475
    },
    {
      "epoch": 1.9660977010647855,
      "grad_norm": 0.9655215740203857,
      "learning_rate": 1.3487697755552211e-05,
      "loss": 0.0642,
      "step": 133500
    },
    {
      "epoch": 1.9664658841548726,
      "grad_norm": 1.1807976961135864,
      "learning_rate": 1.3486061383585227e-05,
      "loss": 0.0637,
      "step": 133525
    },
    {
      "epoch": 1.9668340672449596,
      "grad_norm": 1.1131510734558105,
      "learning_rate": 1.3484425011618244e-05,
      "loss": 0.0683,
      "step": 133550
    },
    {
      "epoch": 1.9672022503350466,
      "grad_norm": 1.3223555088043213,
      "learning_rate": 1.3482854094529938e-05,
      "loss": 0.0731,
      "step": 133575
    },
    {
      "epoch": 1.9675704334251336,
      "grad_norm": 2.066756010055542,
      "learning_rate": 1.3481217722562952e-05,
      "loss": 0.0674,
      "step": 133600
    },
    {
      "epoch": 1.9679386165152208,
      "grad_norm": 1.721665382385254,
      "learning_rate": 1.3479581350595966e-05,
      "loss": 0.0683,
      "step": 133625
    },
    {
      "epoch": 1.9683067996053079,
      "grad_norm": 1.6085426807403564,
      "learning_rate": 1.3477944978628983e-05,
      "loss": 0.0637,
      "step": 133650
    },
    {
      "epoch": 1.9686749826953949,
      "grad_norm": 0.8268477916717529,
      "learning_rate": 1.3476308606661999e-05,
      "loss": 0.0673,
      "step": 133675
    },
    {
      "epoch": 1.9690431657854819,
      "grad_norm": 1.0188851356506348,
      "learning_rate": 1.3474672234695015e-05,
      "loss": 0.0703,
      "step": 133700
    },
    {
      "epoch": 1.969411348875569,
      "grad_norm": 1.8331669569015503,
      "learning_rate": 1.347303586272803e-05,
      "loss": 0.0572,
      "step": 133725
    },
    {
      "epoch": 1.969779531965656,
      "grad_norm": 1.1416645050048828,
      "learning_rate": 1.3471399490761045e-05,
      "loss": 0.0684,
      "step": 133750
    },
    {
      "epoch": 1.970147715055743,
      "grad_norm": 1.3836314678192139,
      "learning_rate": 1.3469763118794062e-05,
      "loss": 0.0656,
      "step": 133775
    },
    {
      "epoch": 1.97051589814583,
      "grad_norm": 1.3859492540359497,
      "learning_rate": 1.3468126746827074e-05,
      "loss": 0.0638,
      "step": 133800
    },
    {
      "epoch": 1.970884081235917,
      "grad_norm": 0.8829729557037354,
      "learning_rate": 1.346649037486009e-05,
      "loss": 0.0619,
      "step": 133825
    },
    {
      "epoch": 1.971252264326004,
      "grad_norm": 1.0988966226577759,
      "learning_rate": 1.3464854002893107e-05,
      "loss": 0.0652,
      "step": 133850
    },
    {
      "epoch": 1.971620447416091,
      "grad_norm": 1.6365810632705688,
      "learning_rate": 1.3463217630926121e-05,
      "loss": 0.07,
      "step": 133875
    },
    {
      "epoch": 1.971988630506178,
      "grad_norm": 1.2164117097854614,
      "learning_rate": 1.3461581258959137e-05,
      "loss": 0.0596,
      "step": 133900
    },
    {
      "epoch": 1.972356813596265,
      "grad_norm": 1.175513505935669,
      "learning_rate": 1.3459944886992153e-05,
      "loss": 0.0703,
      "step": 133925
    },
    {
      "epoch": 1.972724996686352,
      "grad_norm": 1.099058747291565,
      "learning_rate": 1.345830851502517e-05,
      "loss": 0.0622,
      "step": 133950
    },
    {
      "epoch": 1.9730931797764393,
      "grad_norm": 1.2795138359069824,
      "learning_rate": 1.3456672143058184e-05,
      "loss": 0.0726,
      "step": 133975
    },
    {
      "epoch": 1.9734613628665263,
      "grad_norm": 1.2164860963821411,
      "learning_rate": 1.3455035771091199e-05,
      "loss": 0.0682,
      "step": 134000
    },
    {
      "epoch": 1.9738295459566133,
      "grad_norm": 1.0763880014419556,
      "learning_rate": 1.3453399399124215e-05,
      "loss": 0.0703,
      "step": 134025
    },
    {
      "epoch": 1.9741977290467003,
      "grad_norm": 1.5583292245864868,
      "learning_rate": 1.3451763027157229e-05,
      "loss": 0.0619,
      "step": 134050
    },
    {
      "epoch": 1.9745659121367873,
      "grad_norm": 1.5683109760284424,
      "learning_rate": 1.3450126655190245e-05,
      "loss": 0.0606,
      "step": 134075
    },
    {
      "epoch": 1.9749340952268746,
      "grad_norm": 1.1754193305969238,
      "learning_rate": 1.3448490283223261e-05,
      "loss": 0.0697,
      "step": 134100
    },
    {
      "epoch": 1.9753022783169616,
      "grad_norm": 1.5604636669158936,
      "learning_rate": 1.3446853911256278e-05,
      "loss": 0.0663,
      "step": 134125
    },
    {
      "epoch": 1.9756704614070486,
      "grad_norm": 1.7430775165557861,
      "learning_rate": 1.3445217539289292e-05,
      "loss": 0.0678,
      "step": 134150
    },
    {
      "epoch": 1.9760386444971356,
      "grad_norm": 1.7419719696044922,
      "learning_rate": 1.3443581167322308e-05,
      "loss": 0.0718,
      "step": 134175
    },
    {
      "epoch": 1.9764068275872226,
      "grad_norm": 1.3962990045547485,
      "learning_rate": 1.3441944795355324e-05,
      "loss": 0.0619,
      "step": 134200
    },
    {
      "epoch": 1.9767750106773097,
      "grad_norm": 1.4117295742034912,
      "learning_rate": 1.3440308423388337e-05,
      "loss": 0.068,
      "step": 134225
    },
    {
      "epoch": 1.9771431937673967,
      "grad_norm": 1.3334444761276245,
      "learning_rate": 1.3438672051421353e-05,
      "loss": 0.0638,
      "step": 134250
    },
    {
      "epoch": 1.9775113768574837,
      "grad_norm": 1.459469199180603,
      "learning_rate": 1.343703567945437e-05,
      "loss": 0.0701,
      "step": 134275
    },
    {
      "epoch": 1.9778795599475707,
      "grad_norm": 1.6322544813156128,
      "learning_rate": 1.3435399307487384e-05,
      "loss": 0.0656,
      "step": 134300
    },
    {
      "epoch": 1.9782477430376577,
      "grad_norm": 1.5134178400039673,
      "learning_rate": 1.34337629355204e-05,
      "loss": 0.0608,
      "step": 134325
    },
    {
      "epoch": 1.9786159261277447,
      "grad_norm": 1.6400991678237915,
      "learning_rate": 1.3432126563553416e-05,
      "loss": 0.0737,
      "step": 134350
    },
    {
      "epoch": 1.9789841092178317,
      "grad_norm": 1.7818231582641602,
      "learning_rate": 1.3430490191586432e-05,
      "loss": 0.067,
      "step": 134375
    },
    {
      "epoch": 1.9793522923079188,
      "grad_norm": 1.3408386707305908,
      "learning_rate": 1.3428853819619447e-05,
      "loss": 0.0607,
      "step": 134400
    },
    {
      "epoch": 1.9797204753980058,
      "grad_norm": 1.1619755029678345,
      "learning_rate": 1.3427217447652461e-05,
      "loss": 0.0652,
      "step": 134425
    },
    {
      "epoch": 1.980088658488093,
      "grad_norm": 1.128059983253479,
      "learning_rate": 1.3425581075685476e-05,
      "loss": 0.0665,
      "step": 134450
    },
    {
      "epoch": 1.98045684157818,
      "grad_norm": 0.9418730735778809,
      "learning_rate": 1.3423944703718492e-05,
      "loss": 0.0591,
      "step": 134475
    },
    {
      "epoch": 1.980825024668267,
      "grad_norm": 1.3298439979553223,
      "learning_rate": 1.3422308331751508e-05,
      "loss": 0.0651,
      "step": 134500
    },
    {
      "epoch": 1.981193207758354,
      "grad_norm": 1.6387840509414673,
      "learning_rate": 1.3420671959784524e-05,
      "loss": 0.0718,
      "step": 134525
    },
    {
      "epoch": 1.981561390848441,
      "grad_norm": 1.352338433265686,
      "learning_rate": 1.3419035587817539e-05,
      "loss": 0.0651,
      "step": 134550
    },
    {
      "epoch": 1.9819295739385283,
      "grad_norm": 1.4517595767974854,
      "learning_rate": 1.3417399215850555e-05,
      "loss": 0.0682,
      "step": 134575
    },
    {
      "epoch": 1.9822977570286153,
      "grad_norm": 1.3293002843856812,
      "learning_rate": 1.3415762843883571e-05,
      "loss": 0.0654,
      "step": 134600
    },
    {
      "epoch": 1.9826659401187023,
      "grad_norm": 1.3806308507919312,
      "learning_rate": 1.3414126471916587e-05,
      "loss": 0.0651,
      "step": 134625
    },
    {
      "epoch": 1.9830341232087894,
      "grad_norm": 1.2611339092254639,
      "learning_rate": 1.34124900999496e-05,
      "loss": 0.0637,
      "step": 134650
    },
    {
      "epoch": 1.9834023062988764,
      "grad_norm": 1.5833430290222168,
      "learning_rate": 1.3410853727982616e-05,
      "loss": 0.0629,
      "step": 134675
    },
    {
      "epoch": 1.9837704893889634,
      "grad_norm": 1.1704708337783813,
      "learning_rate": 1.3409217356015632e-05,
      "loss": 0.0679,
      "step": 134700
    },
    {
      "epoch": 1.9841386724790504,
      "grad_norm": 1.4179396629333496,
      "learning_rate": 1.3407580984048647e-05,
      "loss": 0.0729,
      "step": 134725
    },
    {
      "epoch": 1.9845068555691374,
      "grad_norm": 1.4768248796463013,
      "learning_rate": 1.3405944612081663e-05,
      "loss": 0.0687,
      "step": 134750
    },
    {
      "epoch": 1.9848750386592244,
      "grad_norm": 1.534879446029663,
      "learning_rate": 1.3404308240114679e-05,
      "loss": 0.066,
      "step": 134775
    },
    {
      "epoch": 1.9852432217493114,
      "grad_norm": 1.4434044361114502,
      "learning_rate": 1.3402671868147693e-05,
      "loss": 0.0632,
      "step": 134800
    },
    {
      "epoch": 1.9856114048393985,
      "grad_norm": 1.1986098289489746,
      "learning_rate": 1.340103549618071e-05,
      "loss": 0.066,
      "step": 134825
    },
    {
      "epoch": 1.9859795879294855,
      "grad_norm": 1.2370998859405518,
      "learning_rate": 1.3399399124213724e-05,
      "loss": 0.0646,
      "step": 134850
    },
    {
      "epoch": 1.9863477710195725,
      "grad_norm": 2.0699992179870605,
      "learning_rate": 1.3397762752246738e-05,
      "loss": 0.0718,
      "step": 134875
    },
    {
      "epoch": 1.9867159541096595,
      "grad_norm": 1.4780324697494507,
      "learning_rate": 1.3396126380279755e-05,
      "loss": 0.0644,
      "step": 134900
    },
    {
      "epoch": 1.9870841371997465,
      "grad_norm": 1.4479182958602905,
      "learning_rate": 1.339449000831277e-05,
      "loss": 0.0723,
      "step": 134925
    },
    {
      "epoch": 1.9874523202898338,
      "grad_norm": 1.32070791721344,
      "learning_rate": 1.3392853636345787e-05,
      "loss": 0.0683,
      "step": 134950
    },
    {
      "epoch": 1.9878205033799208,
      "grad_norm": 1.3346669673919678,
      "learning_rate": 1.3391217264378801e-05,
      "loss": 0.0649,
      "step": 134975
    },
    {
      "epoch": 1.9881886864700078,
      "grad_norm": 1.6738038063049316,
      "learning_rate": 1.3389580892411817e-05,
      "loss": 0.0665,
      "step": 135000
    },
    {
      "epoch": 1.9885568695600948,
      "grad_norm": 1.0200092792510986,
      "learning_rate": 1.3387944520444834e-05,
      "loss": 0.0658,
      "step": 135025
    },
    {
      "epoch": 1.988925052650182,
      "grad_norm": 1.3845388889312744,
      "learning_rate": 1.338630814847785e-05,
      "loss": 0.0675,
      "step": 135050
    },
    {
      "epoch": 1.989293235740269,
      "grad_norm": 1.2986143827438354,
      "learning_rate": 1.3384671776510863e-05,
      "loss": 0.0723,
      "step": 135075
    },
    {
      "epoch": 1.989661418830356,
      "grad_norm": 1.4071464538574219,
      "learning_rate": 1.3383035404543879e-05,
      "loss": 0.0649,
      "step": 135100
    },
    {
      "epoch": 1.990029601920443,
      "grad_norm": 1.2547434568405151,
      "learning_rate": 1.3381399032576893e-05,
      "loss": 0.0648,
      "step": 135125
    },
    {
      "epoch": 1.99039778501053,
      "grad_norm": 1.0378402471542358,
      "learning_rate": 1.337976266060991e-05,
      "loss": 0.0652,
      "step": 135150
    },
    {
      "epoch": 1.9907659681006171,
      "grad_norm": 1.137075424194336,
      "learning_rate": 1.3378126288642925e-05,
      "loss": 0.0668,
      "step": 135175
    },
    {
      "epoch": 1.9911341511907041,
      "grad_norm": 1.057377815246582,
      "learning_rate": 1.3376489916675942e-05,
      "loss": 0.0607,
      "step": 135200
    },
    {
      "epoch": 1.9915023342807912,
      "grad_norm": 1.374295711517334,
      "learning_rate": 1.3374853544708956e-05,
      "loss": 0.0587,
      "step": 135225
    },
    {
      "epoch": 1.9918705173708782,
      "grad_norm": 1.5446298122406006,
      "learning_rate": 1.3373217172741972e-05,
      "loss": 0.0707,
      "step": 135250
    },
    {
      "epoch": 1.9922387004609652,
      "grad_norm": 1.2055821418762207,
      "learning_rate": 1.3371580800774987e-05,
      "loss": 0.0622,
      "step": 135275
    },
    {
      "epoch": 1.9926068835510522,
      "grad_norm": 1.558401107788086,
      "learning_rate": 1.3369944428808001e-05,
      "loss": 0.06,
      "step": 135300
    },
    {
      "epoch": 1.9929750666411392,
      "grad_norm": 0.8838926553726196,
      "learning_rate": 1.3368308056841017e-05,
      "loss": 0.0621,
      "step": 135325
    },
    {
      "epoch": 1.9933432497312262,
      "grad_norm": 1.9442501068115234,
      "learning_rate": 1.3366671684874033e-05,
      "loss": 0.064,
      "step": 135350
    },
    {
      "epoch": 1.9937114328213132,
      "grad_norm": 1.5155824422836304,
      "learning_rate": 1.3365035312907048e-05,
      "loss": 0.0697,
      "step": 135375
    },
    {
      "epoch": 1.9940796159114003,
      "grad_norm": 1.2601693868637085,
      "learning_rate": 1.3363398940940064e-05,
      "loss": 0.0696,
      "step": 135400
    },
    {
      "epoch": 1.9944477990014875,
      "grad_norm": 1.1521936655044556,
      "learning_rate": 1.336176256897308e-05,
      "loss": 0.0666,
      "step": 135425
    },
    {
      "epoch": 1.9948159820915745,
      "grad_norm": 1.2650203704833984,
      "learning_rate": 1.3360126197006096e-05,
      "loss": 0.0667,
      "step": 135450
    },
    {
      "epoch": 1.9951841651816615,
      "grad_norm": 1.6142756938934326,
      "learning_rate": 1.3358489825039109e-05,
      "loss": 0.0677,
      "step": 135475
    },
    {
      "epoch": 1.9955523482717485,
      "grad_norm": 1.4373137950897217,
      "learning_rate": 1.3356853453072125e-05,
      "loss": 0.0701,
      "step": 135500
    },
    {
      "epoch": 1.9959205313618356,
      "grad_norm": 1.6161986589431763,
      "learning_rate": 1.3355217081105141e-05,
      "loss": 0.0644,
      "step": 135525
    },
    {
      "epoch": 1.9962887144519228,
      "grad_norm": 1.5426843166351318,
      "learning_rate": 1.3353580709138156e-05,
      "loss": 0.0695,
      "step": 135550
    },
    {
      "epoch": 1.9966568975420098,
      "grad_norm": 1.4742108583450317,
      "learning_rate": 1.3351944337171172e-05,
      "loss": 0.0607,
      "step": 135575
    },
    {
      "epoch": 1.9970250806320968,
      "grad_norm": 1.3358180522918701,
      "learning_rate": 1.3350307965204188e-05,
      "loss": 0.0641,
      "step": 135600
    },
    {
      "epoch": 1.9973932637221838,
      "grad_norm": 0.905835747718811,
      "learning_rate": 1.3348671593237204e-05,
      "loss": 0.0661,
      "step": 135625
    },
    {
      "epoch": 1.9977614468122709,
      "grad_norm": 1.4315563440322876,
      "learning_rate": 1.3347035221270219e-05,
      "loss": 0.0655,
      "step": 135650
    },
    {
      "epoch": 1.9981296299023579,
      "grad_norm": 1.4081978797912598,
      "learning_rate": 1.3345398849303235e-05,
      "loss": 0.0681,
      "step": 135675
    },
    {
      "epoch": 1.9984978129924449,
      "grad_norm": 1.5095083713531494,
      "learning_rate": 1.3343762477336248e-05,
      "loss": 0.0688,
      "step": 135700
    },
    {
      "epoch": 1.998865996082532,
      "grad_norm": 1.2680156230926514,
      "learning_rate": 1.3342126105369264e-05,
      "loss": 0.0671,
      "step": 135725
    },
    {
      "epoch": 1.999234179172619,
      "grad_norm": 1.801788091659546,
      "learning_rate": 1.334048973340228e-05,
      "loss": 0.0649,
      "step": 135750
    },
    {
      "epoch": 1.999602362262706,
      "grad_norm": 1.603861689567566,
      "learning_rate": 1.3338853361435296e-05,
      "loss": 0.0626,
      "step": 135775
    },
    {
      "epoch": 1.999970545352793,
      "grad_norm": 1.4096496105194092,
      "learning_rate": 1.333721698946831e-05,
      "loss": 0.0693,
      "step": 135800
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.06252778321504593,
      "eval_runtime": 115.2179,
      "eval_samples_per_second": 3078.93,
      "eval_steps_per_second": 6.015,
      "step": 135802
    },
    {
      "epoch": 2.00033872844288,
      "grad_norm": 1.7620490789413452,
      "learning_rate": 1.3335580617501327e-05,
      "loss": 0.0664,
      "step": 135825
    },
    {
      "epoch": 2.000706911532967,
      "grad_norm": 1.201253890991211,
      "learning_rate": 1.3333944245534343e-05,
      "loss": 0.0647,
      "step": 135850
    },
    {
      "epoch": 2.001075094623054,
      "grad_norm": 0.907674252986908,
      "learning_rate": 1.3332307873567359e-05,
      "loss": 0.0577,
      "step": 135875
    },
    {
      "epoch": 2.001443277713141,
      "grad_norm": 1.0521012544631958,
      "learning_rate": 1.3330671501600372e-05,
      "loss": 0.0687,
      "step": 135900
    },
    {
      "epoch": 2.001811460803228,
      "grad_norm": 1.1505175828933716,
      "learning_rate": 1.3329035129633388e-05,
      "loss": 0.0696,
      "step": 135925
    },
    {
      "epoch": 2.0021796438933155,
      "grad_norm": 1.0560606718063354,
      "learning_rate": 1.3327398757666402e-05,
      "loss": 0.0571,
      "step": 135950
    },
    {
      "epoch": 2.0025478269834025,
      "grad_norm": 1.2308223247528076,
      "learning_rate": 1.3325762385699419e-05,
      "loss": 0.0582,
      "step": 135975
    },
    {
      "epoch": 2.0029160100734895,
      "grad_norm": 1.277739405632019,
      "learning_rate": 1.3324191468611114e-05,
      "loss": 0.0604,
      "step": 136000
    },
    {
      "epoch": 2.0032841931635765,
      "grad_norm": 1.522047996520996,
      "learning_rate": 1.332255509664413e-05,
      "loss": 0.0607,
      "step": 136025
    },
    {
      "epoch": 2.0036523762536635,
      "grad_norm": 1.4001848697662354,
      "learning_rate": 1.3320918724677145e-05,
      "loss": 0.0602,
      "step": 136050
    },
    {
      "epoch": 2.0040205593437506,
      "grad_norm": 1.267114520072937,
      "learning_rate": 1.3319282352710161e-05,
      "loss": 0.0632,
      "step": 136075
    },
    {
      "epoch": 2.0043887424338376,
      "grad_norm": 1.0386160612106323,
      "learning_rate": 1.3317645980743177e-05,
      "loss": 0.068,
      "step": 136100
    },
    {
      "epoch": 2.0047569255239246,
      "grad_norm": 1.0514968633651733,
      "learning_rate": 1.331600960877619e-05,
      "loss": 0.0633,
      "step": 136125
    },
    {
      "epoch": 2.0051251086140116,
      "grad_norm": 1.0674878358840942,
      "learning_rate": 1.3314373236809206e-05,
      "loss": 0.0601,
      "step": 136150
    },
    {
      "epoch": 2.0054932917040986,
      "grad_norm": 1.092009425163269,
      "learning_rate": 1.3312736864842222e-05,
      "loss": 0.0635,
      "step": 136175
    },
    {
      "epoch": 2.0058614747941856,
      "grad_norm": 1.6346251964569092,
      "learning_rate": 1.3311100492875237e-05,
      "loss": 0.0595,
      "step": 136200
    },
    {
      "epoch": 2.0062296578842727,
      "grad_norm": 1.1024657487869263,
      "learning_rate": 1.3309464120908253e-05,
      "loss": 0.0667,
      "step": 136225
    },
    {
      "epoch": 2.0065978409743597,
      "grad_norm": 1.558005928993225,
      "learning_rate": 1.3307827748941269e-05,
      "loss": 0.0653,
      "step": 136250
    },
    {
      "epoch": 2.0069660240644467,
      "grad_norm": 1.4044561386108398,
      "learning_rate": 1.3306191376974285e-05,
      "loss": 0.0664,
      "step": 136275
    },
    {
      "epoch": 2.0073342071545337,
      "grad_norm": 1.5821421146392822,
      "learning_rate": 1.33045550050073e-05,
      "loss": 0.0521,
      "step": 136300
    },
    {
      "epoch": 2.0077023902446207,
      "grad_norm": 1.2789149284362793,
      "learning_rate": 1.3302918633040314e-05,
      "loss": 0.0565,
      "step": 136325
    },
    {
      "epoch": 2.0080705733347077,
      "grad_norm": 1.0251727104187012,
      "learning_rate": 1.330128226107333e-05,
      "loss": 0.0671,
      "step": 136350
    },
    {
      "epoch": 2.0084387564247947,
      "grad_norm": 1.1342840194702148,
      "learning_rate": 1.3299645889106345e-05,
      "loss": 0.0615,
      "step": 136375
    },
    {
      "epoch": 2.0088069395148818,
      "grad_norm": 1.3609364032745361,
      "learning_rate": 1.329800951713936e-05,
      "loss": 0.0581,
      "step": 136400
    },
    {
      "epoch": 2.0091751226049688,
      "grad_norm": 1.4621198177337646,
      "learning_rate": 1.3296373145172377e-05,
      "loss": 0.061,
      "step": 136425
    },
    {
      "epoch": 2.0095433056950562,
      "grad_norm": 1.2368834018707275,
      "learning_rate": 1.3294736773205393e-05,
      "loss": 0.0609,
      "step": 136450
    },
    {
      "epoch": 2.0099114887851433,
      "grad_norm": 1.1904629468917847,
      "learning_rate": 1.3293100401238407e-05,
      "loss": 0.0648,
      "step": 136475
    },
    {
      "epoch": 2.0102796718752303,
      "grad_norm": 0.9137811660766602,
      "learning_rate": 1.3291464029271424e-05,
      "loss": 0.0653,
      "step": 136500
    },
    {
      "epoch": 2.0106478549653173,
      "grad_norm": 1.5463374853134155,
      "learning_rate": 1.328982765730444e-05,
      "loss": 0.0675,
      "step": 136525
    },
    {
      "epoch": 2.0110160380554043,
      "grad_norm": 1.4156731367111206,
      "learning_rate": 1.3288191285337453e-05,
      "loss": 0.0629,
      "step": 136550
    },
    {
      "epoch": 2.0113842211454913,
      "grad_norm": 1.0640830993652344,
      "learning_rate": 1.3286554913370469e-05,
      "loss": 0.0642,
      "step": 136575
    },
    {
      "epoch": 2.0117524042355783,
      "grad_norm": 0.9086377024650574,
      "learning_rate": 1.3284918541403485e-05,
      "loss": 0.0565,
      "step": 136600
    },
    {
      "epoch": 2.0121205873256653,
      "grad_norm": 1.2068498134613037,
      "learning_rate": 1.32832821694365e-05,
      "loss": 0.055,
      "step": 136625
    },
    {
      "epoch": 2.0124887704157524,
      "grad_norm": 1.6517916917800903,
      "learning_rate": 1.3281645797469515e-05,
      "loss": 0.0588,
      "step": 136650
    },
    {
      "epoch": 2.0128569535058394,
      "grad_norm": 1.911002278327942,
      "learning_rate": 1.3280009425502532e-05,
      "loss": 0.0675,
      "step": 136675
    },
    {
      "epoch": 2.0132251365959264,
      "grad_norm": 1.4228066205978394,
      "learning_rate": 1.3278373053535548e-05,
      "loss": 0.0601,
      "step": 136700
    },
    {
      "epoch": 2.0135933196860134,
      "grad_norm": 1.662040114402771,
      "learning_rate": 1.3276736681568562e-05,
      "loss": 0.0597,
      "step": 136725
    },
    {
      "epoch": 2.0139615027761004,
      "grad_norm": 1.6213383674621582,
      "learning_rate": 1.3275100309601577e-05,
      "loss": 0.0668,
      "step": 136750
    },
    {
      "epoch": 2.0143296858661874,
      "grad_norm": 1.1999540328979492,
      "learning_rate": 1.3273463937634591e-05,
      "loss": 0.0648,
      "step": 136775
    },
    {
      "epoch": 2.0146978689562745,
      "grad_norm": 1.0933380126953125,
      "learning_rate": 1.3271827565667607e-05,
      "loss": 0.0683,
      "step": 136800
    },
    {
      "epoch": 2.0150660520463615,
      "grad_norm": 1.6969233751296997,
      "learning_rate": 1.3270191193700623e-05,
      "loss": 0.0613,
      "step": 136825
    },
    {
      "epoch": 2.0154342351364485,
      "grad_norm": 1.2042580842971802,
      "learning_rate": 1.326855482173364e-05,
      "loss": 0.0668,
      "step": 136850
    },
    {
      "epoch": 2.0158024182265355,
      "grad_norm": 1.8126479387283325,
      "learning_rate": 1.3266918449766654e-05,
      "loss": 0.0611,
      "step": 136875
    },
    {
      "epoch": 2.0161706013166225,
      "grad_norm": 0.9478511214256287,
      "learning_rate": 1.326528207779967e-05,
      "loss": 0.0673,
      "step": 136900
    },
    {
      "epoch": 2.01653878440671,
      "grad_norm": 1.2510234117507935,
      "learning_rate": 1.3263645705832686e-05,
      "loss": 0.0632,
      "step": 136925
    },
    {
      "epoch": 2.016906967496797,
      "grad_norm": 1.2847950458526611,
      "learning_rate": 1.3262009333865699e-05,
      "loss": 0.0567,
      "step": 136950
    },
    {
      "epoch": 2.017275150586884,
      "grad_norm": 1.6596367359161377,
      "learning_rate": 1.3260372961898715e-05,
      "loss": 0.0583,
      "step": 136975
    },
    {
      "epoch": 2.017643333676971,
      "grad_norm": 1.4504663944244385,
      "learning_rate": 1.3258736589931731e-05,
      "loss": 0.0572,
      "step": 137000
    },
    {
      "epoch": 2.018011516767058,
      "grad_norm": 1.1833807229995728,
      "learning_rate": 1.3257100217964748e-05,
      "loss": 0.0664,
      "step": 137025
    },
    {
      "epoch": 2.018379699857145,
      "grad_norm": 0.8727373480796814,
      "learning_rate": 1.3255463845997762e-05,
      "loss": 0.0654,
      "step": 137050
    },
    {
      "epoch": 2.018747882947232,
      "grad_norm": 1.294104814529419,
      "learning_rate": 1.3253827474030778e-05,
      "loss": 0.0648,
      "step": 137075
    },
    {
      "epoch": 2.019116066037319,
      "grad_norm": 1.1471279859542847,
      "learning_rate": 1.3252191102063794e-05,
      "loss": 0.0639,
      "step": 137100
    },
    {
      "epoch": 2.019484249127406,
      "grad_norm": 0.8571379780769348,
      "learning_rate": 1.3250554730096809e-05,
      "loss": 0.0604,
      "step": 137125
    },
    {
      "epoch": 2.019852432217493,
      "grad_norm": 1.138739824295044,
      "learning_rate": 1.3248918358129825e-05,
      "loss": 0.0595,
      "step": 137150
    },
    {
      "epoch": 2.02022061530758,
      "grad_norm": 0.9922412633895874,
      "learning_rate": 1.324728198616284e-05,
      "loss": 0.0587,
      "step": 137175
    },
    {
      "epoch": 2.020588798397667,
      "grad_norm": 1.403673529624939,
      "learning_rate": 1.3245645614195854e-05,
      "loss": 0.0629,
      "step": 137200
    },
    {
      "epoch": 2.020956981487754,
      "grad_norm": 1.358481526374817,
      "learning_rate": 1.324400924222887e-05,
      "loss": 0.0666,
      "step": 137225
    },
    {
      "epoch": 2.021325164577841,
      "grad_norm": 1.0647672414779663,
      "learning_rate": 1.3242372870261886e-05,
      "loss": 0.0615,
      "step": 137250
    },
    {
      "epoch": 2.021693347667928,
      "grad_norm": 1.1101211309432983,
      "learning_rate": 1.3240736498294902e-05,
      "loss": 0.0671,
      "step": 137275
    },
    {
      "epoch": 2.022061530758015,
      "grad_norm": 1.926060438156128,
      "learning_rate": 1.3239100126327917e-05,
      "loss": 0.0638,
      "step": 137300
    },
    {
      "epoch": 2.022429713848102,
      "grad_norm": 1.2956942319869995,
      "learning_rate": 1.3237463754360933e-05,
      "loss": 0.0673,
      "step": 137325
    },
    {
      "epoch": 2.0227978969381892,
      "grad_norm": 0.756001889705658,
      "learning_rate": 1.3235827382393949e-05,
      "loss": 0.0606,
      "step": 137350
    },
    {
      "epoch": 2.0231660800282762,
      "grad_norm": 0.7750321626663208,
      "learning_rate": 1.3234191010426962e-05,
      "loss": 0.061,
      "step": 137375
    },
    {
      "epoch": 2.0235342631183637,
      "grad_norm": 1.663071870803833,
      "learning_rate": 1.3232554638459978e-05,
      "loss": 0.0586,
      "step": 137400
    },
    {
      "epoch": 2.0239024462084507,
      "grad_norm": 1.4343326091766357,
      "learning_rate": 1.3230918266492994e-05,
      "loss": 0.0662,
      "step": 137425
    },
    {
      "epoch": 2.0242706292985377,
      "grad_norm": 1.3194102048873901,
      "learning_rate": 1.3229281894526009e-05,
      "loss": 0.0633,
      "step": 137450
    },
    {
      "epoch": 2.0246388123886248,
      "grad_norm": 1.8688809871673584,
      "learning_rate": 1.3227645522559025e-05,
      "loss": 0.0642,
      "step": 137475
    },
    {
      "epoch": 2.0250069954787118,
      "grad_norm": 1.8178691864013672,
      "learning_rate": 1.3226009150592041e-05,
      "loss": 0.0657,
      "step": 137500
    },
    {
      "epoch": 2.025375178568799,
      "grad_norm": 1.88641357421875,
      "learning_rate": 1.3224372778625057e-05,
      "loss": 0.0606,
      "step": 137525
    },
    {
      "epoch": 2.025743361658886,
      "grad_norm": 1.3782966136932373,
      "learning_rate": 1.3222736406658072e-05,
      "loss": 0.0629,
      "step": 137550
    },
    {
      "epoch": 2.026111544748973,
      "grad_norm": 1.296451210975647,
      "learning_rate": 1.3221100034691088e-05,
      "loss": 0.0609,
      "step": 137575
    },
    {
      "epoch": 2.02647972783906,
      "grad_norm": 1.3301911354064941,
      "learning_rate": 1.3219463662724102e-05,
      "loss": 0.0651,
      "step": 137600
    },
    {
      "epoch": 2.026847910929147,
      "grad_norm": 1.7005969285964966,
      "learning_rate": 1.3217827290757117e-05,
      "loss": 0.0602,
      "step": 137625
    },
    {
      "epoch": 2.027216094019234,
      "grad_norm": 1.0853757858276367,
      "learning_rate": 1.3216190918790133e-05,
      "loss": 0.0587,
      "step": 137650
    },
    {
      "epoch": 2.027584277109321,
      "grad_norm": 1.4156244993209839,
      "learning_rate": 1.3214554546823149e-05,
      "loss": 0.06,
      "step": 137675
    },
    {
      "epoch": 2.027952460199408,
      "grad_norm": 1.3988933563232422,
      "learning_rate": 1.3212918174856163e-05,
      "loss": 0.0617,
      "step": 137700
    },
    {
      "epoch": 2.028320643289495,
      "grad_norm": 1.0410761833190918,
      "learning_rate": 1.321128180288918e-05,
      "loss": 0.0665,
      "step": 137725
    },
    {
      "epoch": 2.028688826379582,
      "grad_norm": 1.1729612350463867,
      "learning_rate": 1.3209645430922196e-05,
      "loss": 0.0627,
      "step": 137750
    },
    {
      "epoch": 2.029057009469669,
      "grad_norm": 0.8793452382087708,
      "learning_rate": 1.3208009058955212e-05,
      "loss": 0.073,
      "step": 137775
    },
    {
      "epoch": 2.029425192559756,
      "grad_norm": 1.0352075099945068,
      "learning_rate": 1.3206372686988225e-05,
      "loss": 0.061,
      "step": 137800
    },
    {
      "epoch": 2.029793375649843,
      "grad_norm": 1.5649809837341309,
      "learning_rate": 1.320473631502124e-05,
      "loss": 0.0608,
      "step": 137825
    },
    {
      "epoch": 2.03016155873993,
      "grad_norm": 1.3111470937728882,
      "learning_rate": 1.3203099943054257e-05,
      "loss": 0.061,
      "step": 137850
    },
    {
      "epoch": 2.0305297418300174,
      "grad_norm": 1.7501057386398315,
      "learning_rate": 1.3201463571087271e-05,
      "loss": 0.0583,
      "step": 137875
    },
    {
      "epoch": 2.0308979249201045,
      "grad_norm": 1.616523027420044,
      "learning_rate": 1.3199827199120287e-05,
      "loss": 0.064,
      "step": 137900
    },
    {
      "epoch": 2.0312661080101915,
      "grad_norm": 1.742025375366211,
      "learning_rate": 1.3198190827153304e-05,
      "loss": 0.063,
      "step": 137925
    },
    {
      "epoch": 2.0316342911002785,
      "grad_norm": 1.0694652795791626,
      "learning_rate": 1.319655445518632e-05,
      "loss": 0.0586,
      "step": 137950
    },
    {
      "epoch": 2.0320024741903655,
      "grad_norm": 1.1424410343170166,
      "learning_rate": 1.3194918083219334e-05,
      "loss": 0.0621,
      "step": 137975
    },
    {
      "epoch": 2.0323706572804525,
      "grad_norm": 1.309401035308838,
      "learning_rate": 1.319328171125235e-05,
      "loss": 0.0643,
      "step": 138000
    },
    {
      "epoch": 2.0327388403705395,
      "grad_norm": 1.3246620893478394,
      "learning_rate": 1.3191645339285363e-05,
      "loss": 0.0622,
      "step": 138025
    },
    {
      "epoch": 2.0331070234606266,
      "grad_norm": 1.3904753923416138,
      "learning_rate": 1.319000896731838e-05,
      "loss": 0.0659,
      "step": 138050
    },
    {
      "epoch": 2.0334752065507136,
      "grad_norm": 1.8712338209152222,
      "learning_rate": 1.3188372595351395e-05,
      "loss": 0.0678,
      "step": 138075
    },
    {
      "epoch": 2.0338433896408006,
      "grad_norm": 1.263144612312317,
      "learning_rate": 1.3186736223384412e-05,
      "loss": 0.0656,
      "step": 138100
    },
    {
      "epoch": 2.0342115727308876,
      "grad_norm": 1.527934193611145,
      "learning_rate": 1.3185099851417426e-05,
      "loss": 0.0684,
      "step": 138125
    },
    {
      "epoch": 2.0345797558209746,
      "grad_norm": 1.329735517501831,
      "learning_rate": 1.3183463479450442e-05,
      "loss": 0.0657,
      "step": 138150
    },
    {
      "epoch": 2.0349479389110616,
      "grad_norm": Infinity,
      "learning_rate": 1.3181892562362138e-05,
      "loss": 0.0623,
      "step": 138175
    },
    {
      "epoch": 2.0353161220011486,
      "grad_norm": 1.3123729228973389,
      "learning_rate": 1.3180256190395154e-05,
      "loss": 0.0575,
      "step": 138200
    },
    {
      "epoch": 2.0356843050912357,
      "grad_norm": 1.5601123571395874,
      "learning_rate": 1.3178619818428167e-05,
      "loss": 0.0674,
      "step": 138225
    },
    {
      "epoch": 2.0360524881813227,
      "grad_norm": 1.5189870595932007,
      "learning_rate": 1.3176983446461183e-05,
      "loss": 0.0626,
      "step": 138250
    },
    {
      "epoch": 2.0364206712714097,
      "grad_norm": 1.7031035423278809,
      "learning_rate": 1.3175347074494197e-05,
      "loss": 0.0675,
      "step": 138275
    },
    {
      "epoch": 2.0367888543614967,
      "grad_norm": 0.8013625741004944,
      "learning_rate": 1.3173710702527213e-05,
      "loss": 0.0617,
      "step": 138300
    },
    {
      "epoch": 2.0371570374515837,
      "grad_norm": 1.5683618783950806,
      "learning_rate": 1.317207433056023e-05,
      "loss": 0.0657,
      "step": 138325
    },
    {
      "epoch": 2.037525220541671,
      "grad_norm": 1.5596259832382202,
      "learning_rate": 1.3170437958593246e-05,
      "loss": 0.0663,
      "step": 138350
    },
    {
      "epoch": 2.037893403631758,
      "grad_norm": 1.3564993143081665,
      "learning_rate": 1.316880158662626e-05,
      "loss": 0.0685,
      "step": 138375
    },
    {
      "epoch": 2.038261586721845,
      "grad_norm": 1.921115517616272,
      "learning_rate": 1.3167165214659276e-05,
      "loss": 0.0621,
      "step": 138400
    },
    {
      "epoch": 2.0386297698119322,
      "grad_norm": 0.9760974049568176,
      "learning_rate": 1.3165528842692293e-05,
      "loss": 0.063,
      "step": 138425
    },
    {
      "epoch": 2.0389979529020192,
      "grad_norm": 0.9111531972885132,
      "learning_rate": 1.3163892470725305e-05,
      "loss": 0.0624,
      "step": 138450
    },
    {
      "epoch": 2.0393661359921063,
      "grad_norm": 1.5316861867904663,
      "learning_rate": 1.3162256098758321e-05,
      "loss": 0.0691,
      "step": 138475
    },
    {
      "epoch": 2.0397343190821933,
      "grad_norm": 1.6244877576828003,
      "learning_rate": 1.3160619726791338e-05,
      "loss": 0.0635,
      "step": 138500
    },
    {
      "epoch": 2.0401025021722803,
      "grad_norm": 1.393915057182312,
      "learning_rate": 1.3158983354824352e-05,
      "loss": 0.0617,
      "step": 138525
    },
    {
      "epoch": 2.0404706852623673,
      "grad_norm": 1.1242659091949463,
      "learning_rate": 1.3157346982857368e-05,
      "loss": 0.0666,
      "step": 138550
    },
    {
      "epoch": 2.0408388683524543,
      "grad_norm": 1.4729751348495483,
      "learning_rate": 1.3155710610890384e-05,
      "loss": 0.069,
      "step": 138575
    },
    {
      "epoch": 2.0412070514425413,
      "grad_norm": 1.425998330116272,
      "learning_rate": 1.31540742389234e-05,
      "loss": 0.062,
      "step": 138600
    },
    {
      "epoch": 2.0415752345326283,
      "grad_norm": 1.3222010135650635,
      "learning_rate": 1.3152437866956415e-05,
      "loss": 0.0671,
      "step": 138625
    },
    {
      "epoch": 2.0419434176227154,
      "grad_norm": 1.410779356956482,
      "learning_rate": 1.315080149498943e-05,
      "loss": 0.0577,
      "step": 138650
    },
    {
      "epoch": 2.0423116007128024,
      "grad_norm": 1.0774645805358887,
      "learning_rate": 1.3149165123022446e-05,
      "loss": 0.0647,
      "step": 138675
    },
    {
      "epoch": 2.0426797838028894,
      "grad_norm": 1.0232930183410645,
      "learning_rate": 1.314752875105546e-05,
      "loss": 0.0578,
      "step": 138700
    },
    {
      "epoch": 2.0430479668929764,
      "grad_norm": 1.0589803457260132,
      "learning_rate": 1.3145892379088476e-05,
      "loss": 0.0574,
      "step": 138725
    },
    {
      "epoch": 2.0434161499830634,
      "grad_norm": 1.8932327032089233,
      "learning_rate": 1.3144256007121492e-05,
      "loss": 0.0644,
      "step": 138750
    },
    {
      "epoch": 2.0437843330731504,
      "grad_norm": 1.52449631690979,
      "learning_rate": 1.3142619635154508e-05,
      "loss": 0.0665,
      "step": 138775
    },
    {
      "epoch": 2.0441525161632375,
      "grad_norm": 1.403548002243042,
      "learning_rate": 1.3140983263187523e-05,
      "loss": 0.0571,
      "step": 138800
    },
    {
      "epoch": 2.044520699253325,
      "grad_norm": 1.053322672843933,
      "learning_rate": 1.3139346891220539e-05,
      "loss": 0.0611,
      "step": 138825
    },
    {
      "epoch": 2.044888882343412,
      "grad_norm": 1.0917490720748901,
      "learning_rate": 1.3137710519253552e-05,
      "loss": 0.0589,
      "step": 138850
    },
    {
      "epoch": 2.045257065433499,
      "grad_norm": 1.9904406070709229,
      "learning_rate": 1.3136074147286568e-05,
      "loss": 0.074,
      "step": 138875
    },
    {
      "epoch": 2.045625248523586,
      "grad_norm": 1.4356249570846558,
      "learning_rate": 1.3134437775319584e-05,
      "loss": 0.0706,
      "step": 138900
    },
    {
      "epoch": 2.045993431613673,
      "grad_norm": 1.5951532125473022,
      "learning_rate": 1.31328014033526e-05,
      "loss": 0.0692,
      "step": 138925
    },
    {
      "epoch": 2.04636161470376,
      "grad_norm": 1.652031421661377,
      "learning_rate": 1.3131165031385615e-05,
      "loss": 0.07,
      "step": 138950
    },
    {
      "epoch": 2.046729797793847,
      "grad_norm": 1.2981295585632324,
      "learning_rate": 1.3129528659418631e-05,
      "loss": 0.0616,
      "step": 138975
    },
    {
      "epoch": 2.047097980883934,
      "grad_norm": 1.3023178577423096,
      "learning_rate": 1.3127892287451647e-05,
      "loss": 0.0641,
      "step": 139000
    },
    {
      "epoch": 2.047466163974021,
      "grad_norm": 1.6468833684921265,
      "learning_rate": 1.3126255915484663e-05,
      "loss": 0.0597,
      "step": 139025
    },
    {
      "epoch": 2.047834347064108,
      "grad_norm": 0.7615518569946289,
      "learning_rate": 1.3124619543517678e-05,
      "loss": 0.0643,
      "step": 139050
    },
    {
      "epoch": 2.048202530154195,
      "grad_norm": 1.8502449989318848,
      "learning_rate": 1.3122983171550692e-05,
      "loss": 0.0587,
      "step": 139075
    },
    {
      "epoch": 2.048570713244282,
      "grad_norm": 1.102052092552185,
      "learning_rate": 1.3121346799583707e-05,
      "loss": 0.0599,
      "step": 139100
    },
    {
      "epoch": 2.048938896334369,
      "grad_norm": 1.6666117906570435,
      "learning_rate": 1.3119710427616723e-05,
      "loss": 0.0627,
      "step": 139125
    },
    {
      "epoch": 2.049307079424456,
      "grad_norm": 0.9726002812385559,
      "learning_rate": 1.3118074055649739e-05,
      "loss": 0.0576,
      "step": 139150
    },
    {
      "epoch": 2.049675262514543,
      "grad_norm": 1.3315876722335815,
      "learning_rate": 1.3116437683682755e-05,
      "loss": 0.0685,
      "step": 139175
    },
    {
      "epoch": 2.05004344560463,
      "grad_norm": 1.5628360509872437,
      "learning_rate": 1.311480131171577e-05,
      "loss": 0.062,
      "step": 139200
    },
    {
      "epoch": 2.050411628694717,
      "grad_norm": 1.05496346950531,
      "learning_rate": 1.3113164939748786e-05,
      "loss": 0.0645,
      "step": 139225
    },
    {
      "epoch": 2.050779811784804,
      "grad_norm": 1.903501272201538,
      "learning_rate": 1.3111528567781802e-05,
      "loss": 0.0617,
      "step": 139250
    },
    {
      "epoch": 2.051147994874891,
      "grad_norm": 1.390907883644104,
      "learning_rate": 1.3109892195814815e-05,
      "loss": 0.0646,
      "step": 139275
    },
    {
      "epoch": 2.051516177964978,
      "grad_norm": 0.9083699584007263,
      "learning_rate": 1.310825582384783e-05,
      "loss": 0.0629,
      "step": 139300
    },
    {
      "epoch": 2.0518843610550657,
      "grad_norm": 0.9529362916946411,
      "learning_rate": 1.3106619451880847e-05,
      "loss": 0.0576,
      "step": 139325
    },
    {
      "epoch": 2.0522525441451527,
      "grad_norm": 1.516104817390442,
      "learning_rate": 1.3104983079913863e-05,
      "loss": 0.0621,
      "step": 139350
    },
    {
      "epoch": 2.0526207272352397,
      "grad_norm": 1.1514497995376587,
      "learning_rate": 1.3103346707946877e-05,
      "loss": 0.0666,
      "step": 139375
    },
    {
      "epoch": 2.0529889103253267,
      "grad_norm": 1.3026103973388672,
      "learning_rate": 1.3101710335979894e-05,
      "loss": 0.0578,
      "step": 139400
    },
    {
      "epoch": 2.0533570934154137,
      "grad_norm": 1.6160250902175903,
      "learning_rate": 1.310007396401291e-05,
      "loss": 0.0665,
      "step": 139425
    },
    {
      "epoch": 2.0537252765055007,
      "grad_norm": 1.8178597688674927,
      "learning_rate": 1.3098437592045924e-05,
      "loss": 0.0696,
      "step": 139450
    },
    {
      "epoch": 2.0540934595955878,
      "grad_norm": 1.6499983072280884,
      "learning_rate": 1.309680122007894e-05,
      "loss": 0.0626,
      "step": 139475
    },
    {
      "epoch": 2.0544616426856748,
      "grad_norm": 1.3892712593078613,
      "learning_rate": 1.3095164848111955e-05,
      "loss": 0.0597,
      "step": 139500
    },
    {
      "epoch": 2.054829825775762,
      "grad_norm": 1.1013695001602173,
      "learning_rate": 1.309352847614497e-05,
      "loss": 0.0659,
      "step": 139525
    },
    {
      "epoch": 2.055198008865849,
      "grad_norm": 1.0680475234985352,
      "learning_rate": 1.3091892104177985e-05,
      "loss": 0.0615,
      "step": 139550
    },
    {
      "epoch": 2.055566191955936,
      "grad_norm": 1.365019679069519,
      "learning_rate": 1.3090255732211002e-05,
      "loss": 0.0607,
      "step": 139575
    },
    {
      "epoch": 2.055934375046023,
      "grad_norm": 1.6709725856781006,
      "learning_rate": 1.3088619360244018e-05,
      "loss": 0.069,
      "step": 139600
    },
    {
      "epoch": 2.05630255813611,
      "grad_norm": 1.7192907333374023,
      "learning_rate": 1.3086982988277032e-05,
      "loss": 0.0613,
      "step": 139625
    },
    {
      "epoch": 2.056670741226197,
      "grad_norm": 1.3610130548477173,
      "learning_rate": 1.3085346616310048e-05,
      "loss": 0.0677,
      "step": 139650
    },
    {
      "epoch": 2.057038924316284,
      "grad_norm": 1.378743052482605,
      "learning_rate": 1.3083710244343065e-05,
      "loss": 0.0727,
      "step": 139675
    },
    {
      "epoch": 2.057407107406371,
      "grad_norm": 0.9532351493835449,
      "learning_rate": 1.3082073872376077e-05,
      "loss": 0.0643,
      "step": 139700
    },
    {
      "epoch": 2.057775290496458,
      "grad_norm": 1.232985496520996,
      "learning_rate": 1.3080437500409093e-05,
      "loss": 0.0585,
      "step": 139725
    },
    {
      "epoch": 2.058143473586545,
      "grad_norm": 1.3010550737380981,
      "learning_rate": 1.307880112844211e-05,
      "loss": 0.0648,
      "step": 139750
    },
    {
      "epoch": 2.058511656676632,
      "grad_norm": 1.1119623184204102,
      "learning_rate": 1.3077164756475124e-05,
      "loss": 0.0562,
      "step": 139775
    },
    {
      "epoch": 2.0588798397667194,
      "grad_norm": 1.2800379991531372,
      "learning_rate": 1.307552838450814e-05,
      "loss": 0.0633,
      "step": 139800
    },
    {
      "epoch": 2.0592480228568064,
      "grad_norm": 1.5540469884872437,
      "learning_rate": 1.3073892012541156e-05,
      "loss": 0.0644,
      "step": 139825
    },
    {
      "epoch": 2.0596162059468934,
      "grad_norm": 1.0997648239135742,
      "learning_rate": 1.3072255640574173e-05,
      "loss": 0.0539,
      "step": 139850
    },
    {
      "epoch": 2.0599843890369804,
      "grad_norm": 1.4437756538391113,
      "learning_rate": 1.3070619268607187e-05,
      "loss": 0.07,
      "step": 139875
    },
    {
      "epoch": 2.0603525721270675,
      "grad_norm": 1.899660587310791,
      "learning_rate": 1.3068982896640203e-05,
      "loss": 0.0638,
      "step": 139900
    },
    {
      "epoch": 2.0607207552171545,
      "grad_norm": 1.098617672920227,
      "learning_rate": 1.3067346524673218e-05,
      "loss": 0.0655,
      "step": 139925
    },
    {
      "epoch": 2.0610889383072415,
      "grad_norm": 0.8406283259391785,
      "learning_rate": 1.3065710152706232e-05,
      "loss": 0.0675,
      "step": 139950
    },
    {
      "epoch": 2.0614571213973285,
      "grad_norm": 1.1740442514419556,
      "learning_rate": 1.3064073780739248e-05,
      "loss": 0.0562,
      "step": 139975
    },
    {
      "epoch": 2.0618253044874155,
      "grad_norm": 1.5432026386260986,
      "learning_rate": 1.3062437408772264e-05,
      "loss": 0.0563,
      "step": 140000
    },
    {
      "epoch": 2.0621934875775025,
      "grad_norm": 1.668741226196289,
      "learning_rate": 1.3060801036805279e-05,
      "loss": 0.0577,
      "step": 140025
    },
    {
      "epoch": 2.0625616706675896,
      "grad_norm": 1.3596270084381104,
      "learning_rate": 1.3059164664838295e-05,
      "loss": 0.0678,
      "step": 140050
    },
    {
      "epoch": 2.0629298537576766,
      "grad_norm": 1.7664932012557983,
      "learning_rate": 1.3057528292871311e-05,
      "loss": 0.0691,
      "step": 140075
    },
    {
      "epoch": 2.0632980368477636,
      "grad_norm": 1.3408604860305786,
      "learning_rate": 1.3055891920904327e-05,
      "loss": 0.0663,
      "step": 140100
    },
    {
      "epoch": 2.0636662199378506,
      "grad_norm": 1.409626841545105,
      "learning_rate": 1.305425554893734e-05,
      "loss": 0.0614,
      "step": 140125
    },
    {
      "epoch": 2.0640344030279376,
      "grad_norm": 1.6057690382003784,
      "learning_rate": 1.3052619176970356e-05,
      "loss": 0.0658,
      "step": 140150
    },
    {
      "epoch": 2.0644025861180246,
      "grad_norm": 1.4680137634277344,
      "learning_rate": 1.3050982805003372e-05,
      "loss": 0.0528,
      "step": 140175
    },
    {
      "epoch": 2.0647707692081116,
      "grad_norm": 1.5226285457611084,
      "learning_rate": 1.3049346433036387e-05,
      "loss": 0.0723,
      "step": 140200
    },
    {
      "epoch": 2.0651389522981987,
      "grad_norm": 1.073876142501831,
      "learning_rate": 1.3047710061069403e-05,
      "loss": 0.0638,
      "step": 140225
    },
    {
      "epoch": 2.0655071353882857,
      "grad_norm": 1.3069032430648804,
      "learning_rate": 1.3046073689102419e-05,
      "loss": 0.0664,
      "step": 140250
    },
    {
      "epoch": 2.065875318478373,
      "grad_norm": 1.4336020946502686,
      "learning_rate": 1.3044502772014113e-05,
      "loss": 0.0688,
      "step": 140275
    },
    {
      "epoch": 2.06624350156846,
      "grad_norm": 1.4683928489685059,
      "learning_rate": 1.3042866400047129e-05,
      "loss": 0.0638,
      "step": 140300
    },
    {
      "epoch": 2.066611684658547,
      "grad_norm": 1.3116852045059204,
      "learning_rate": 1.3041230028080144e-05,
      "loss": 0.063,
      "step": 140325
    },
    {
      "epoch": 2.066979867748634,
      "grad_norm": 1.498652696609497,
      "learning_rate": 1.3039593656113158e-05,
      "loss": 0.0675,
      "step": 140350
    },
    {
      "epoch": 2.067348050838721,
      "grad_norm": 1.148743987083435,
      "learning_rate": 1.3037957284146174e-05,
      "loss": 0.0615,
      "step": 140375
    },
    {
      "epoch": 2.067716233928808,
      "grad_norm": 1.1772137880325317,
      "learning_rate": 1.303632091217919e-05,
      "loss": 0.0593,
      "step": 140400
    },
    {
      "epoch": 2.0680844170188952,
      "grad_norm": 1.1782065629959106,
      "learning_rate": 1.3034684540212206e-05,
      "loss": 0.0593,
      "step": 140425
    },
    {
      "epoch": 2.0684526001089822,
      "grad_norm": 1.458560585975647,
      "learning_rate": 1.3033048168245221e-05,
      "loss": 0.0655,
      "step": 140450
    },
    {
      "epoch": 2.0688207831990693,
      "grad_norm": 0.7723190784454346,
      "learning_rate": 1.3031411796278237e-05,
      "loss": 0.0535,
      "step": 140475
    },
    {
      "epoch": 2.0691889662891563,
      "grad_norm": 1.4423205852508545,
      "learning_rate": 1.3029775424311253e-05,
      "loss": 0.0636,
      "step": 140500
    },
    {
      "epoch": 2.0695571493792433,
      "grad_norm": 0.8405426740646362,
      "learning_rate": 1.302813905234427e-05,
      "loss": 0.0653,
      "step": 140525
    },
    {
      "epoch": 2.0699253324693303,
      "grad_norm": 0.9008001685142517,
      "learning_rate": 1.3026502680377282e-05,
      "loss": 0.0636,
      "step": 140550
    },
    {
      "epoch": 2.0702935155594173,
      "grad_norm": 1.235060691833496,
      "learning_rate": 1.3024866308410298e-05,
      "loss": 0.0654,
      "step": 140575
    },
    {
      "epoch": 2.0706616986495043,
      "grad_norm": 1.0082287788391113,
      "learning_rate": 1.3023229936443313e-05,
      "loss": 0.0668,
      "step": 140600
    },
    {
      "epoch": 2.0710298817395913,
      "grad_norm": 1.3555642366409302,
      "learning_rate": 1.3021593564476329e-05,
      "loss": 0.0602,
      "step": 140625
    },
    {
      "epoch": 2.0713980648296784,
      "grad_norm": 1.2186473608016968,
      "learning_rate": 1.3019957192509345e-05,
      "loss": 0.061,
      "step": 140650
    },
    {
      "epoch": 2.0717662479197654,
      "grad_norm": 1.6986041069030762,
      "learning_rate": 1.3018320820542361e-05,
      "loss": 0.0568,
      "step": 140675
    },
    {
      "epoch": 2.0721344310098524,
      "grad_norm": 1.4862006902694702,
      "learning_rate": 1.3016684448575376e-05,
      "loss": 0.0643,
      "step": 140700
    },
    {
      "epoch": 2.0725026140999394,
      "grad_norm": 1.3662189245224,
      "learning_rate": 1.3015048076608392e-05,
      "loss": 0.064,
      "step": 140725
    },
    {
      "epoch": 2.072870797190027,
      "grad_norm": 1.5313546657562256,
      "learning_rate": 1.3013411704641406e-05,
      "loss": 0.0605,
      "step": 140750
    },
    {
      "epoch": 2.073238980280114,
      "grad_norm": 1.3113452196121216,
      "learning_rate": 1.301177533267442e-05,
      "loss": 0.0595,
      "step": 140775
    },
    {
      "epoch": 2.073607163370201,
      "grad_norm": 1.8002442121505737,
      "learning_rate": 1.3010138960707437e-05,
      "loss": 0.0626,
      "step": 140800
    },
    {
      "epoch": 2.073975346460288,
      "grad_norm": 1.6425055265426636,
      "learning_rate": 1.3008502588740453e-05,
      "loss": 0.0595,
      "step": 140825
    },
    {
      "epoch": 2.074343529550375,
      "grad_norm": 1.4305739402770996,
      "learning_rate": 1.3006866216773467e-05,
      "loss": 0.0609,
      "step": 140850
    },
    {
      "epoch": 2.074711712640462,
      "grad_norm": 1.2356306314468384,
      "learning_rate": 1.3005229844806484e-05,
      "loss": 0.07,
      "step": 140875
    },
    {
      "epoch": 2.075079895730549,
      "grad_norm": 1.367408275604248,
      "learning_rate": 1.30035934728395e-05,
      "loss": 0.0655,
      "step": 140900
    },
    {
      "epoch": 2.075448078820636,
      "grad_norm": 1.344552755355835,
      "learning_rate": 1.3001957100872516e-05,
      "loss": 0.0676,
      "step": 140925
    },
    {
      "epoch": 2.075816261910723,
      "grad_norm": 0.9702365398406982,
      "learning_rate": 1.300032072890553e-05,
      "loss": 0.0657,
      "step": 140950
    },
    {
      "epoch": 2.07618444500081,
      "grad_norm": 1.7759288549423218,
      "learning_rate": 1.2998684356938545e-05,
      "loss": 0.0649,
      "step": 140975
    },
    {
      "epoch": 2.076552628090897,
      "grad_norm": 1.168556571006775,
      "learning_rate": 1.2997047984971561e-05,
      "loss": 0.0662,
      "step": 141000
    },
    {
      "epoch": 2.076920811180984,
      "grad_norm": 0.9602246284484863,
      "learning_rate": 1.2995411613004575e-05,
      "loss": 0.0664,
      "step": 141025
    },
    {
      "epoch": 2.077288994271071,
      "grad_norm": 1.1011443138122559,
      "learning_rate": 1.2993775241037592e-05,
      "loss": 0.0581,
      "step": 141050
    },
    {
      "epoch": 2.077657177361158,
      "grad_norm": 1.2254912853240967,
      "learning_rate": 1.2992138869070608e-05,
      "loss": 0.0648,
      "step": 141075
    },
    {
      "epoch": 2.078025360451245,
      "grad_norm": 1.7616828680038452,
      "learning_rate": 1.2990502497103624e-05,
      "loss": 0.0611,
      "step": 141100
    },
    {
      "epoch": 2.078393543541332,
      "grad_norm": 0.9659159183502197,
      "learning_rate": 1.2988866125136638e-05,
      "loss": 0.0698,
      "step": 141125
    },
    {
      "epoch": 2.078761726631419,
      "grad_norm": 1.216340184211731,
      "learning_rate": 1.2987229753169655e-05,
      "loss": 0.0611,
      "step": 141150
    },
    {
      "epoch": 2.079129909721506,
      "grad_norm": 1.3510624170303345,
      "learning_rate": 1.2985593381202667e-05,
      "loss": 0.0646,
      "step": 141175
    },
    {
      "epoch": 2.079498092811593,
      "grad_norm": 1.135532259941101,
      "learning_rate": 1.2983957009235683e-05,
      "loss": 0.0589,
      "step": 141200
    },
    {
      "epoch": 2.07986627590168,
      "grad_norm": 1.712357521057129,
      "learning_rate": 1.29823206372687e-05,
      "loss": 0.0684,
      "step": 141225
    },
    {
      "epoch": 2.0802344589917676,
      "grad_norm": 1.687016248703003,
      "learning_rate": 1.2980684265301716e-05,
      "loss": 0.0692,
      "step": 141250
    },
    {
      "epoch": 2.0806026420818546,
      "grad_norm": 1.6153839826583862,
      "learning_rate": 1.297904789333473e-05,
      "loss": 0.0674,
      "step": 141275
    },
    {
      "epoch": 2.0809708251719417,
      "grad_norm": 1.5420087575912476,
      "learning_rate": 1.2977411521367746e-05,
      "loss": 0.0684,
      "step": 141300
    },
    {
      "epoch": 2.0813390082620287,
      "grad_norm": 1.437110185623169,
      "learning_rate": 1.2975775149400763e-05,
      "loss": 0.0558,
      "step": 141325
    },
    {
      "epoch": 2.0817071913521157,
      "grad_norm": 1.9535467624664307,
      "learning_rate": 1.2974138777433779e-05,
      "loss": 0.0594,
      "step": 141350
    },
    {
      "epoch": 2.0820753744422027,
      "grad_norm": 1.4665862321853638,
      "learning_rate": 1.2972502405466793e-05,
      "loss": 0.0544,
      "step": 141375
    },
    {
      "epoch": 2.0824435575322897,
      "grad_norm": 1.26799476146698,
      "learning_rate": 1.2970866033499808e-05,
      "loss": 0.0684,
      "step": 141400
    },
    {
      "epoch": 2.0828117406223767,
      "grad_norm": 1.584623098373413,
      "learning_rate": 1.2969229661532822e-05,
      "loss": 0.0646,
      "step": 141425
    },
    {
      "epoch": 2.0831799237124637,
      "grad_norm": 1.1859813928604126,
      "learning_rate": 1.2967593289565838e-05,
      "loss": 0.0679,
      "step": 141450
    },
    {
      "epoch": 2.0835481068025508,
      "grad_norm": 1.1526150703430176,
      "learning_rate": 1.2965956917598854e-05,
      "loss": 0.0648,
      "step": 141475
    },
    {
      "epoch": 2.0839162898926378,
      "grad_norm": 1.5065057277679443,
      "learning_rate": 1.296432054563187e-05,
      "loss": 0.064,
      "step": 141500
    },
    {
      "epoch": 2.084284472982725,
      "grad_norm": 1.4624722003936768,
      "learning_rate": 1.2962684173664885e-05,
      "loss": 0.0673,
      "step": 141525
    },
    {
      "epoch": 2.084652656072812,
      "grad_norm": 1.3338077068328857,
      "learning_rate": 1.2961047801697901e-05,
      "loss": 0.057,
      "step": 141550
    },
    {
      "epoch": 2.085020839162899,
      "grad_norm": 1.2845184803009033,
      "learning_rate": 1.2959411429730917e-05,
      "loss": 0.0663,
      "step": 141575
    },
    {
      "epoch": 2.085389022252986,
      "grad_norm": 1.2876492738723755,
      "learning_rate": 1.295777505776393e-05,
      "loss": 0.0634,
      "step": 141600
    },
    {
      "epoch": 2.085757205343073,
      "grad_norm": 1.4167022705078125,
      "learning_rate": 1.2956138685796946e-05,
      "loss": 0.0629,
      "step": 141625
    },
    {
      "epoch": 2.08612538843316,
      "grad_norm": 1.4296507835388184,
      "learning_rate": 1.2954502313829962e-05,
      "loss": 0.0631,
      "step": 141650
    },
    {
      "epoch": 2.086493571523247,
      "grad_norm": 1.6763315200805664,
      "learning_rate": 1.2952865941862978e-05,
      "loss": 0.0639,
      "step": 141675
    },
    {
      "epoch": 2.0868617546133343,
      "grad_norm": 1.530309796333313,
      "learning_rate": 1.2951229569895993e-05,
      "loss": 0.0592,
      "step": 141700
    },
    {
      "epoch": 2.0872299377034214,
      "grad_norm": 0.849713921546936,
      "learning_rate": 1.2949593197929009e-05,
      "loss": 0.0664,
      "step": 141725
    },
    {
      "epoch": 2.0875981207935084,
      "grad_norm": 1.0692070722579956,
      "learning_rate": 1.2947956825962025e-05,
      "loss": 0.073,
      "step": 141750
    },
    {
      "epoch": 2.0879663038835954,
      "grad_norm": 0.98299241065979,
      "learning_rate": 1.294632045399504e-05,
      "loss": 0.0624,
      "step": 141775
    },
    {
      "epoch": 2.0883344869736824,
      "grad_norm": 1.4880268573760986,
      "learning_rate": 1.2944684082028054e-05,
      "loss": 0.0604,
      "step": 141800
    },
    {
      "epoch": 2.0887026700637694,
      "grad_norm": 1.196270227432251,
      "learning_rate": 1.294304771006107e-05,
      "loss": 0.0681,
      "step": 141825
    },
    {
      "epoch": 2.0890708531538564,
      "grad_norm": 1.146098017692566,
      "learning_rate": 1.2941411338094085e-05,
      "loss": 0.0634,
      "step": 141850
    },
    {
      "epoch": 2.0894390362439434,
      "grad_norm": 1.51388680934906,
      "learning_rate": 1.2939774966127101e-05,
      "loss": 0.0664,
      "step": 141875
    },
    {
      "epoch": 2.0898072193340305,
      "grad_norm": 1.2342647314071655,
      "learning_rate": 1.2938138594160117e-05,
      "loss": 0.0637,
      "step": 141900
    },
    {
      "epoch": 2.0901754024241175,
      "grad_norm": 1.394655704498291,
      "learning_rate": 1.2936502222193133e-05,
      "loss": 0.0635,
      "step": 141925
    },
    {
      "epoch": 2.0905435855142045,
      "grad_norm": 1.622645616531372,
      "learning_rate": 1.2934865850226148e-05,
      "loss": 0.0642,
      "step": 141950
    },
    {
      "epoch": 2.0909117686042915,
      "grad_norm": 1.768418550491333,
      "learning_rate": 1.2933229478259164e-05,
      "loss": 0.0633,
      "step": 141975
    },
    {
      "epoch": 2.0912799516943785,
      "grad_norm": 1.3926706314086914,
      "learning_rate": 1.293159310629218e-05,
      "loss": 0.0604,
      "step": 142000
    },
    {
      "epoch": 2.0916481347844655,
      "grad_norm": 1.6714481115341187,
      "learning_rate": 1.2929956734325193e-05,
      "loss": 0.0634,
      "step": 142025
    },
    {
      "epoch": 2.0920163178745526,
      "grad_norm": 1.5839020013809204,
      "learning_rate": 1.2928320362358209e-05,
      "loss": 0.0706,
      "step": 142050
    },
    {
      "epoch": 2.0923845009646396,
      "grad_norm": 1.0985029935836792,
      "learning_rate": 1.2926683990391225e-05,
      "loss": 0.0571,
      "step": 142075
    },
    {
      "epoch": 2.0927526840547266,
      "grad_norm": 1.868914246559143,
      "learning_rate": 1.292504761842424e-05,
      "loss": 0.067,
      "step": 142100
    },
    {
      "epoch": 2.0931208671448136,
      "grad_norm": 1.129977822303772,
      "learning_rate": 1.2923411246457256e-05,
      "loss": 0.0612,
      "step": 142125
    },
    {
      "epoch": 2.0934890502349006,
      "grad_norm": 1.0147048234939575,
      "learning_rate": 1.2921774874490272e-05,
      "loss": 0.058,
      "step": 142150
    },
    {
      "epoch": 2.0938572333249876,
      "grad_norm": 1.3958303928375244,
      "learning_rate": 1.2920138502523288e-05,
      "loss": 0.0679,
      "step": 142175
    },
    {
      "epoch": 2.094225416415075,
      "grad_norm": 1.1584354639053345,
      "learning_rate": 1.2918502130556302e-05,
      "loss": 0.0635,
      "step": 142200
    },
    {
      "epoch": 2.094593599505162,
      "grad_norm": 1.5718311071395874,
      "learning_rate": 1.2916865758589317e-05,
      "loss": 0.0684,
      "step": 142225
    },
    {
      "epoch": 2.094961782595249,
      "grad_norm": 1.5085594654083252,
      "learning_rate": 1.2915229386622333e-05,
      "loss": 0.0635,
      "step": 142250
    },
    {
      "epoch": 2.095329965685336,
      "grad_norm": 1.1011391878128052,
      "learning_rate": 1.2913593014655347e-05,
      "loss": 0.0657,
      "step": 142275
    },
    {
      "epoch": 2.095698148775423,
      "grad_norm": 1.568986177444458,
      "learning_rate": 1.2911956642688364e-05,
      "loss": 0.0606,
      "step": 142300
    },
    {
      "epoch": 2.09606633186551,
      "grad_norm": 1.4383682012557983,
      "learning_rate": 1.291032027072138e-05,
      "loss": 0.0571,
      "step": 142325
    },
    {
      "epoch": 2.096434514955597,
      "grad_norm": 1.510115146636963,
      "learning_rate": 1.2908683898754394e-05,
      "loss": 0.0629,
      "step": 142350
    },
    {
      "epoch": 2.096802698045684,
      "grad_norm": 1.0108559131622314,
      "learning_rate": 1.290704752678741e-05,
      "loss": 0.0597,
      "step": 142375
    },
    {
      "epoch": 2.097170881135771,
      "grad_norm": 1.4723358154296875,
      "learning_rate": 1.2905411154820427e-05,
      "loss": 0.0604,
      "step": 142400
    },
    {
      "epoch": 2.0975390642258582,
      "grad_norm": 1.4640063047409058,
      "learning_rate": 1.2903774782853443e-05,
      "loss": 0.0629,
      "step": 142425
    },
    {
      "epoch": 2.0979072473159452,
      "grad_norm": 1.620958924293518,
      "learning_rate": 1.2902138410886455e-05,
      "loss": 0.0619,
      "step": 142450
    },
    {
      "epoch": 2.0982754304060323,
      "grad_norm": 1.749944806098938,
      "learning_rate": 1.2900502038919472e-05,
      "loss": 0.0672,
      "step": 142475
    },
    {
      "epoch": 2.0986436134961193,
      "grad_norm": 1.1632822751998901,
      "learning_rate": 1.2898865666952488e-05,
      "loss": 0.0622,
      "step": 142500
    },
    {
      "epoch": 2.0990117965862063,
      "grad_norm": 0.9090245962142944,
      "learning_rate": 1.2897229294985502e-05,
      "loss": 0.0599,
      "step": 142525
    },
    {
      "epoch": 2.0993799796762933,
      "grad_norm": 0.9051256775856018,
      "learning_rate": 1.2895592923018518e-05,
      "loss": 0.062,
      "step": 142550
    },
    {
      "epoch": 2.0997481627663803,
      "grad_norm": 0.896317183971405,
      "learning_rate": 1.2893956551051535e-05,
      "loss": 0.0732,
      "step": 142575
    },
    {
      "epoch": 2.1001163458564673,
      "grad_norm": 1.3453454971313477,
      "learning_rate": 1.2892320179084549e-05,
      "loss": 0.0658,
      "step": 142600
    },
    {
      "epoch": 2.1004845289465544,
      "grad_norm": 1.448994517326355,
      "learning_rate": 1.2890683807117565e-05,
      "loss": 0.0631,
      "step": 142625
    },
    {
      "epoch": 2.1008527120366414,
      "grad_norm": 1.9118424654006958,
      "learning_rate": 1.288904743515058e-05,
      "loss": 0.0581,
      "step": 142650
    },
    {
      "epoch": 2.101220895126729,
      "grad_norm": 1.576361060142517,
      "learning_rate": 1.2887411063183594e-05,
      "loss": 0.0626,
      "step": 142675
    },
    {
      "epoch": 2.101589078216816,
      "grad_norm": 1.3878475427627563,
      "learning_rate": 1.288577469121661e-05,
      "loss": 0.0625,
      "step": 142700
    },
    {
      "epoch": 2.101957261306903,
      "grad_norm": 1.9188108444213867,
      "learning_rate": 1.2884138319249626e-05,
      "loss": 0.059,
      "step": 142725
    },
    {
      "epoch": 2.10232544439699,
      "grad_norm": 1.5311640501022339,
      "learning_rate": 1.2882501947282642e-05,
      "loss": 0.0696,
      "step": 142750
    },
    {
      "epoch": 2.102693627487077,
      "grad_norm": 1.0558075904846191,
      "learning_rate": 1.2880865575315657e-05,
      "loss": 0.065,
      "step": 142775
    },
    {
      "epoch": 2.103061810577164,
      "grad_norm": 1.0887057781219482,
      "learning_rate": 1.2879229203348673e-05,
      "loss": 0.0595,
      "step": 142800
    },
    {
      "epoch": 2.103429993667251,
      "grad_norm": 1.369795322418213,
      "learning_rate": 1.287759283138169e-05,
      "loss": 0.0623,
      "step": 142825
    },
    {
      "epoch": 2.103798176757338,
      "grad_norm": 1.280044674873352,
      "learning_rate": 1.2875956459414705e-05,
      "loss": 0.0615,
      "step": 142850
    },
    {
      "epoch": 2.104166359847425,
      "grad_norm": 1.3107632398605347,
      "learning_rate": 1.2874320087447718e-05,
      "loss": 0.0603,
      "step": 142875
    },
    {
      "epoch": 2.104534542937512,
      "grad_norm": 1.8252980709075928,
      "learning_rate": 1.2872683715480734e-05,
      "loss": 0.0658,
      "step": 142900
    },
    {
      "epoch": 2.104902726027599,
      "grad_norm": 1.4435657262802124,
      "learning_rate": 1.2871047343513749e-05,
      "loss": 0.0643,
      "step": 142925
    },
    {
      "epoch": 2.105270909117686,
      "grad_norm": 1.8467187881469727,
      "learning_rate": 1.2869410971546765e-05,
      "loss": 0.0571,
      "step": 142950
    },
    {
      "epoch": 2.105639092207773,
      "grad_norm": 1.1174285411834717,
      "learning_rate": 1.2867774599579781e-05,
      "loss": 0.0596,
      "step": 142975
    },
    {
      "epoch": 2.10600727529786,
      "grad_norm": 0.9031580090522766,
      "learning_rate": 1.2866138227612797e-05,
      "loss": 0.0627,
      "step": 143000
    },
    {
      "epoch": 2.106375458387947,
      "grad_norm": 1.3084290027618408,
      "learning_rate": 1.2864501855645812e-05,
      "loss": 0.0583,
      "step": 143025
    },
    {
      "epoch": 2.106743641478034,
      "grad_norm": 1.6340535879135132,
      "learning_rate": 1.2862865483678828e-05,
      "loss": 0.0581,
      "step": 143050
    },
    {
      "epoch": 2.107111824568121,
      "grad_norm": 1.4646821022033691,
      "learning_rate": 1.2861229111711842e-05,
      "loss": 0.0684,
      "step": 143075
    },
    {
      "epoch": 2.107480007658208,
      "grad_norm": 1.251671314239502,
      "learning_rate": 1.2859592739744857e-05,
      "loss": 0.0629,
      "step": 143100
    },
    {
      "epoch": 2.107848190748295,
      "grad_norm": 1.1170073747634888,
      "learning_rate": 1.2857956367777873e-05,
      "loss": 0.0582,
      "step": 143125
    },
    {
      "epoch": 2.1082163738383826,
      "grad_norm": 1.2538365125656128,
      "learning_rate": 1.2856319995810889e-05,
      "loss": 0.0631,
      "step": 143150
    },
    {
      "epoch": 2.1085845569284696,
      "grad_norm": 1.4987280368804932,
      "learning_rate": 1.2854683623843904e-05,
      "loss": 0.0649,
      "step": 143175
    },
    {
      "epoch": 2.1089527400185566,
      "grad_norm": 1.2383800745010376,
      "learning_rate": 1.285304725187692e-05,
      "loss": 0.0639,
      "step": 143200
    },
    {
      "epoch": 2.1093209231086436,
      "grad_norm": 0.8765453696250916,
      "learning_rate": 1.2851410879909936e-05,
      "loss": 0.062,
      "step": 143225
    },
    {
      "epoch": 2.1096891061987306,
      "grad_norm": 1.73635733127594,
      "learning_rate": 1.2849774507942952e-05,
      "loss": 0.0696,
      "step": 143250
    },
    {
      "epoch": 2.1100572892888176,
      "grad_norm": 1.0011390447616577,
      "learning_rate": 1.2848138135975965e-05,
      "loss": 0.0668,
      "step": 143275
    },
    {
      "epoch": 2.1104254723789047,
      "grad_norm": 1.3832238912582397,
      "learning_rate": 1.2846501764008981e-05,
      "loss": 0.0562,
      "step": 143300
    },
    {
      "epoch": 2.1107936554689917,
      "grad_norm": 1.3932617902755737,
      "learning_rate": 1.2844865392041997e-05,
      "loss": 0.0607,
      "step": 143325
    },
    {
      "epoch": 2.1111618385590787,
      "grad_norm": 1.249355673789978,
      "learning_rate": 1.2843229020075011e-05,
      "loss": 0.0603,
      "step": 143350
    },
    {
      "epoch": 2.1115300216491657,
      "grad_norm": 1.4410150051116943,
      "learning_rate": 1.2841592648108028e-05,
      "loss": 0.0645,
      "step": 143375
    },
    {
      "epoch": 2.1118982047392527,
      "grad_norm": 1.1042981147766113,
      "learning_rate": 1.2839956276141044e-05,
      "loss": 0.0566,
      "step": 143400
    },
    {
      "epoch": 2.1122663878293397,
      "grad_norm": 1.2523338794708252,
      "learning_rate": 1.283831990417406e-05,
      "loss": 0.0698,
      "step": 143425
    },
    {
      "epoch": 2.1126345709194267,
      "grad_norm": 0.9360388517379761,
      "learning_rate": 1.2836683532207074e-05,
      "loss": 0.061,
      "step": 143450
    },
    {
      "epoch": 2.1130027540095138,
      "grad_norm": 1.7112327814102173,
      "learning_rate": 1.283504716024009e-05,
      "loss": 0.0603,
      "step": 143475
    },
    {
      "epoch": 2.1133709370996008,
      "grad_norm": 1.3567159175872803,
      "learning_rate": 1.2833410788273103e-05,
      "loss": 0.0626,
      "step": 143500
    },
    {
      "epoch": 2.113739120189688,
      "grad_norm": 0.863154411315918,
      "learning_rate": 1.283177441630612e-05,
      "loss": 0.0591,
      "step": 143525
    },
    {
      "epoch": 2.114107303279775,
      "grad_norm": 1.3425259590148926,
      "learning_rate": 1.2830138044339136e-05,
      "loss": 0.0602,
      "step": 143550
    },
    {
      "epoch": 2.114475486369862,
      "grad_norm": Infinity,
      "learning_rate": 1.2828567127250831e-05,
      "loss": 0.0602,
      "step": 143575
    },
    {
      "epoch": 2.114843669459949,
      "grad_norm": 1.4608036279678345,
      "learning_rate": 1.2826930755283846e-05,
      "loss": 0.0579,
      "step": 143600
    },
    {
      "epoch": 2.1152118525500363,
      "grad_norm": 1.1670485734939575,
      "learning_rate": 1.2825294383316862e-05,
      "loss": 0.064,
      "step": 143625
    },
    {
      "epoch": 2.1155800356401233,
      "grad_norm": 1.3197723627090454,
      "learning_rate": 1.2823658011349878e-05,
      "loss": 0.0537,
      "step": 143650
    },
    {
      "epoch": 2.1159482187302103,
      "grad_norm": 0.9913374781608582,
      "learning_rate": 1.2822021639382894e-05,
      "loss": 0.0617,
      "step": 143675
    },
    {
      "epoch": 2.1163164018202973,
      "grad_norm": 1.4832210540771484,
      "learning_rate": 1.2820385267415907e-05,
      "loss": 0.0607,
      "step": 143700
    },
    {
      "epoch": 2.1166845849103844,
      "grad_norm": 1.2942887544631958,
      "learning_rate": 1.2818748895448923e-05,
      "loss": 0.0627,
      "step": 143725
    },
    {
      "epoch": 2.1170527680004714,
      "grad_norm": 1.3849095106124878,
      "learning_rate": 1.2817112523481937e-05,
      "loss": 0.0566,
      "step": 143750
    },
    {
      "epoch": 2.1174209510905584,
      "grad_norm": 1.3855611085891724,
      "learning_rate": 1.2815476151514954e-05,
      "loss": 0.0586,
      "step": 143775
    },
    {
      "epoch": 2.1177891341806454,
      "grad_norm": 1.02754545211792,
      "learning_rate": 1.281383977954797e-05,
      "loss": 0.0603,
      "step": 143800
    },
    {
      "epoch": 2.1181573172707324,
      "grad_norm": 1.3677605390548706,
      "learning_rate": 1.2812203407580986e-05,
      "loss": 0.0622,
      "step": 143825
    },
    {
      "epoch": 2.1185255003608194,
      "grad_norm": 1.3518853187561035,
      "learning_rate": 1.2810567035614e-05,
      "loss": 0.0648,
      "step": 143850
    },
    {
      "epoch": 2.1188936834509065,
      "grad_norm": 1.235385537147522,
      "learning_rate": 1.2808930663647017e-05,
      "loss": 0.0633,
      "step": 143875
    },
    {
      "epoch": 2.1192618665409935,
      "grad_norm": 1.441364049911499,
      "learning_rate": 1.2807294291680033e-05,
      "loss": 0.0611,
      "step": 143900
    },
    {
      "epoch": 2.1196300496310805,
      "grad_norm": 1.1982982158660889,
      "learning_rate": 1.2805657919713045e-05,
      "loss": 0.0605,
      "step": 143925
    },
    {
      "epoch": 2.1199982327211675,
      "grad_norm": 1.425902009010315,
      "learning_rate": 1.2804021547746062e-05,
      "loss": 0.0586,
      "step": 143950
    },
    {
      "epoch": 2.1203664158112545,
      "grad_norm": 1.070468544960022,
      "learning_rate": 1.2802385175779078e-05,
      "loss": 0.0619,
      "step": 143975
    },
    {
      "epoch": 2.1207345989013415,
      "grad_norm": 1.27206289768219,
      "learning_rate": 1.2800748803812094e-05,
      "loss": 0.0612,
      "step": 144000
    },
    {
      "epoch": 2.1211027819914285,
      "grad_norm": 1.3330292701721191,
      "learning_rate": 1.2799112431845108e-05,
      "loss": 0.0646,
      "step": 144025
    },
    {
      "epoch": 2.1214709650815156,
      "grad_norm": 1.4780975580215454,
      "learning_rate": 1.2797476059878125e-05,
      "loss": 0.0615,
      "step": 144050
    },
    {
      "epoch": 2.1218391481716026,
      "grad_norm": 1.175407886505127,
      "learning_rate": 1.279583968791114e-05,
      "loss": 0.0621,
      "step": 144075
    },
    {
      "epoch": 2.1222073312616896,
      "grad_norm": 1.5601645708084106,
      "learning_rate": 1.2794203315944155e-05,
      "loss": 0.0707,
      "step": 144100
    },
    {
      "epoch": 2.122575514351777,
      "grad_norm": 1.0417650938034058,
      "learning_rate": 1.279256694397717e-05,
      "loss": 0.0578,
      "step": 144125
    },
    {
      "epoch": 2.122943697441864,
      "grad_norm": 1.3065272569656372,
      "learning_rate": 1.2790930572010186e-05,
      "loss": 0.0621,
      "step": 144150
    },
    {
      "epoch": 2.123311880531951,
      "grad_norm": 1.6906054019927979,
      "learning_rate": 1.27892942000432e-05,
      "loss": 0.0677,
      "step": 144175
    },
    {
      "epoch": 2.123680063622038,
      "grad_norm": 0.9042579531669617,
      "learning_rate": 1.2787657828076216e-05,
      "loss": 0.0617,
      "step": 144200
    },
    {
      "epoch": 2.124048246712125,
      "grad_norm": 0.9982229471206665,
      "learning_rate": 1.2786021456109232e-05,
      "loss": 0.058,
      "step": 144225
    },
    {
      "epoch": 2.124416429802212,
      "grad_norm": 0.9489299058914185,
      "learning_rate": 1.2784385084142249e-05,
      "loss": 0.0606,
      "step": 144250
    },
    {
      "epoch": 2.124784612892299,
      "grad_norm": 1.6039745807647705,
      "learning_rate": 1.2782748712175263e-05,
      "loss": 0.0661,
      "step": 144275
    },
    {
      "epoch": 2.125152795982386,
      "grad_norm": 1.6360671520233154,
      "learning_rate": 1.278111234020828e-05,
      "loss": 0.0602,
      "step": 144300
    },
    {
      "epoch": 2.125520979072473,
      "grad_norm": 1.4091622829437256,
      "learning_rate": 1.2779475968241295e-05,
      "loss": 0.0568,
      "step": 144325
    },
    {
      "epoch": 2.12588916216256,
      "grad_norm": 1.6434584856033325,
      "learning_rate": 1.2777839596274308e-05,
      "loss": 0.0647,
      "step": 144350
    },
    {
      "epoch": 2.126257345252647,
      "grad_norm": 0.9404569864273071,
      "learning_rate": 1.2776203224307324e-05,
      "loss": 0.0595,
      "step": 144375
    },
    {
      "epoch": 2.126625528342734,
      "grad_norm": 1.516420841217041,
      "learning_rate": 1.277456685234034e-05,
      "loss": 0.0702,
      "step": 144400
    },
    {
      "epoch": 2.1269937114328212,
      "grad_norm": 1.3707910776138306,
      "learning_rate": 1.2772930480373355e-05,
      "loss": 0.0567,
      "step": 144425
    },
    {
      "epoch": 2.1273618945229082,
      "grad_norm": 1.2299379110336304,
      "learning_rate": 1.2771294108406371e-05,
      "loss": 0.0606,
      "step": 144450
    },
    {
      "epoch": 2.1277300776129953,
      "grad_norm": 0.957705557346344,
      "learning_rate": 1.2769657736439387e-05,
      "loss": 0.0609,
      "step": 144475
    },
    {
      "epoch": 2.1280982607030823,
      "grad_norm": 1.8743137121200562,
      "learning_rate": 1.2768021364472403e-05,
      "loss": 0.0581,
      "step": 144500
    },
    {
      "epoch": 2.1284664437931693,
      "grad_norm": 1.2197659015655518,
      "learning_rate": 1.2766384992505418e-05,
      "loss": 0.0678,
      "step": 144525
    },
    {
      "epoch": 2.1288346268832563,
      "grad_norm": 1.4386900663375854,
      "learning_rate": 1.2764748620538432e-05,
      "loss": 0.0629,
      "step": 144550
    },
    {
      "epoch": 2.1292028099733438,
      "grad_norm": 1.5029577016830444,
      "learning_rate": 1.2763112248571447e-05,
      "loss": 0.0631,
      "step": 144575
    },
    {
      "epoch": 2.129570993063431,
      "grad_norm": 1.4606939554214478,
      "learning_rate": 1.2761475876604463e-05,
      "loss": 0.0641,
      "step": 144600
    },
    {
      "epoch": 2.129939176153518,
      "grad_norm": 1.5000444650650024,
      "learning_rate": 1.2759839504637479e-05,
      "loss": 0.0654,
      "step": 144625
    },
    {
      "epoch": 2.130307359243605,
      "grad_norm": 1.001732349395752,
      "learning_rate": 1.2758203132670495e-05,
      "loss": 0.0565,
      "step": 144650
    },
    {
      "epoch": 2.130675542333692,
      "grad_norm": 1.2616219520568848,
      "learning_rate": 1.275656676070351e-05,
      "loss": 0.0604,
      "step": 144675
    },
    {
      "epoch": 2.131043725423779,
      "grad_norm": 1.4114136695861816,
      "learning_rate": 1.2754930388736526e-05,
      "loss": 0.0602,
      "step": 144700
    },
    {
      "epoch": 2.131411908513866,
      "grad_norm": 1.4554424285888672,
      "learning_rate": 1.2753294016769542e-05,
      "loss": 0.0607,
      "step": 144725
    },
    {
      "epoch": 2.131780091603953,
      "grad_norm": 1.5529040098190308,
      "learning_rate": 1.2751657644802555e-05,
      "loss": 0.0559,
      "step": 144750
    },
    {
      "epoch": 2.13214827469404,
      "grad_norm": 1.816643476486206,
      "learning_rate": 1.2750021272835571e-05,
      "loss": 0.0637,
      "step": 144775
    },
    {
      "epoch": 2.132516457784127,
      "grad_norm": 1.499562382698059,
      "learning_rate": 1.2748384900868587e-05,
      "loss": 0.0595,
      "step": 144800
    },
    {
      "epoch": 2.132884640874214,
      "grad_norm": 1.5631037950515747,
      "learning_rate": 1.2746748528901603e-05,
      "loss": 0.0606,
      "step": 144825
    },
    {
      "epoch": 2.133252823964301,
      "grad_norm": 1.1813325881958008,
      "learning_rate": 1.2745112156934618e-05,
      "loss": 0.0619,
      "step": 144850
    },
    {
      "epoch": 2.133621007054388,
      "grad_norm": 1.0146390199661255,
      "learning_rate": 1.2743475784967634e-05,
      "loss": 0.0617,
      "step": 144875
    },
    {
      "epoch": 2.133989190144475,
      "grad_norm": 1.4431804418563843,
      "learning_rate": 1.274183941300065e-05,
      "loss": 0.0678,
      "step": 144900
    },
    {
      "epoch": 2.134357373234562,
      "grad_norm": 1.3941853046417236,
      "learning_rate": 1.2740203041033664e-05,
      "loss": 0.0624,
      "step": 144925
    },
    {
      "epoch": 2.134725556324649,
      "grad_norm": 1.2117805480957031,
      "learning_rate": 1.273856666906668e-05,
      "loss": 0.0615,
      "step": 144950
    },
    {
      "epoch": 2.135093739414736,
      "grad_norm": 1.2068432569503784,
      "learning_rate": 1.2736930297099695e-05,
      "loss": 0.0627,
      "step": 144975
    },
    {
      "epoch": 2.135461922504823,
      "grad_norm": 1.3667653799057007,
      "learning_rate": 1.273529392513271e-05,
      "loss": 0.0674,
      "step": 145000
    },
    {
      "epoch": 2.13583010559491,
      "grad_norm": 1.786485195159912,
      "learning_rate": 1.2733657553165726e-05,
      "loss": 0.055,
      "step": 145025
    },
    {
      "epoch": 2.136198288684997,
      "grad_norm": 1.6920074224472046,
      "learning_rate": 1.2732021181198742e-05,
      "loss": 0.0671,
      "step": 145050
    },
    {
      "epoch": 2.1365664717750845,
      "grad_norm": 1.8676539659500122,
      "learning_rate": 1.2730384809231758e-05,
      "loss": 0.0652,
      "step": 145075
    },
    {
      "epoch": 2.1369346548651715,
      "grad_norm": 1.3670507669448853,
      "learning_rate": 1.2728748437264772e-05,
      "loss": 0.0608,
      "step": 145100
    },
    {
      "epoch": 2.1373028379552585,
      "grad_norm": 1.4896103143692017,
      "learning_rate": 1.2727112065297789e-05,
      "loss": 0.0599,
      "step": 145125
    },
    {
      "epoch": 2.1376710210453456,
      "grad_norm": 1.1572341918945312,
      "learning_rate": 1.2725475693330805e-05,
      "loss": 0.0598,
      "step": 145150
    },
    {
      "epoch": 2.1380392041354326,
      "grad_norm": 1.333055019378662,
      "learning_rate": 1.2723839321363817e-05,
      "loss": 0.0611,
      "step": 145175
    },
    {
      "epoch": 2.1384073872255196,
      "grad_norm": 1.245370864868164,
      "learning_rate": 1.2722202949396834e-05,
      "loss": 0.0633,
      "step": 145200
    },
    {
      "epoch": 2.1387755703156066,
      "grad_norm": 1.2429883480072021,
      "learning_rate": 1.272056657742985e-05,
      "loss": 0.066,
      "step": 145225
    },
    {
      "epoch": 2.1391437534056936,
      "grad_norm": 1.1446257829666138,
      "learning_rate": 1.2718930205462864e-05,
      "loss": 0.0656,
      "step": 145250
    },
    {
      "epoch": 2.1395119364957806,
      "grad_norm": 1.1383090019226074,
      "learning_rate": 1.271729383349588e-05,
      "loss": 0.0674,
      "step": 145275
    },
    {
      "epoch": 2.1398801195858677,
      "grad_norm": 1.2570135593414307,
      "learning_rate": 1.2715657461528897e-05,
      "loss": 0.063,
      "step": 145300
    },
    {
      "epoch": 2.1402483026759547,
      "grad_norm": 1.4377793073654175,
      "learning_rate": 1.2714021089561913e-05,
      "loss": 0.0596,
      "step": 145325
    },
    {
      "epoch": 2.1406164857660417,
      "grad_norm": 1.6421830654144287,
      "learning_rate": 1.2712384717594927e-05,
      "loss": 0.0625,
      "step": 145350
    },
    {
      "epoch": 2.1409846688561287,
      "grad_norm": 1.458400011062622,
      "learning_rate": 1.2710748345627943e-05,
      "loss": 0.0622,
      "step": 145375
    },
    {
      "epoch": 2.1413528519462157,
      "grad_norm": 1.1557886600494385,
      "learning_rate": 1.2709111973660958e-05,
      "loss": 0.0623,
      "step": 145400
    },
    {
      "epoch": 2.1417210350363027,
      "grad_norm": 1.8212825059890747,
      "learning_rate": 1.2707475601693972e-05,
      "loss": 0.0607,
      "step": 145425
    },
    {
      "epoch": 2.1420892181263897,
      "grad_norm": 1.9266533851623535,
      "learning_rate": 1.2705839229726988e-05,
      "loss": 0.0663,
      "step": 145450
    },
    {
      "epoch": 2.1424574012164768,
      "grad_norm": 1.2210057973861694,
      "learning_rate": 1.2704202857760004e-05,
      "loss": 0.0634,
      "step": 145475
    },
    {
      "epoch": 2.142825584306564,
      "grad_norm": 1.2419432401657104,
      "learning_rate": 1.2702566485793019e-05,
      "loss": 0.0637,
      "step": 145500
    },
    {
      "epoch": 2.1431937673966512,
      "grad_norm": 1.3512849807739258,
      "learning_rate": 1.2700930113826035e-05,
      "loss": 0.0619,
      "step": 145525
    },
    {
      "epoch": 2.1435619504867383,
      "grad_norm": 1.3220046758651733,
      "learning_rate": 1.2699293741859051e-05,
      "loss": 0.0598,
      "step": 145550
    },
    {
      "epoch": 2.1439301335768253,
      "grad_norm": 1.4884276390075684,
      "learning_rate": 1.2697657369892067e-05,
      "loss": 0.0633,
      "step": 145575
    },
    {
      "epoch": 2.1442983166669123,
      "grad_norm": 1.504096508026123,
      "learning_rate": 1.269602099792508e-05,
      "loss": 0.0596,
      "step": 145600
    },
    {
      "epoch": 2.1446664997569993,
      "grad_norm": 1.3510218858718872,
      "learning_rate": 1.2694384625958096e-05,
      "loss": 0.0626,
      "step": 145625
    },
    {
      "epoch": 2.1450346828470863,
      "grad_norm": 0.9822535514831543,
      "learning_rate": 1.2692748253991112e-05,
      "loss": 0.0547,
      "step": 145650
    },
    {
      "epoch": 2.1454028659371733,
      "grad_norm": 1.5289455652236938,
      "learning_rate": 1.2691111882024127e-05,
      "loss": 0.0665,
      "step": 145675
    },
    {
      "epoch": 2.1457710490272603,
      "grad_norm": 1.4685269594192505,
      "learning_rate": 1.2689475510057143e-05,
      "loss": 0.056,
      "step": 145700
    },
    {
      "epoch": 2.1461392321173474,
      "grad_norm": 1.6888009309768677,
      "learning_rate": 1.268783913809016e-05,
      "loss": 0.0641,
      "step": 145725
    },
    {
      "epoch": 2.1465074152074344,
      "grad_norm": 1.2333053350448608,
      "learning_rate": 1.2686202766123175e-05,
      "loss": 0.0586,
      "step": 145750
    },
    {
      "epoch": 2.1468755982975214,
      "grad_norm": 1.2496823072433472,
      "learning_rate": 1.268456639415619e-05,
      "loss": 0.058,
      "step": 145775
    },
    {
      "epoch": 2.1472437813876084,
      "grad_norm": 1.2895691394805908,
      "learning_rate": 1.2682930022189206e-05,
      "loss": 0.0575,
      "step": 145800
    },
    {
      "epoch": 2.1476119644776954,
      "grad_norm": 1.2195422649383545,
      "learning_rate": 1.2681293650222219e-05,
      "loss": 0.0603,
      "step": 145825
    },
    {
      "epoch": 2.1479801475677824,
      "grad_norm": 1.4415476322174072,
      "learning_rate": 1.2679657278255235e-05,
      "loss": 0.0507,
      "step": 145850
    },
    {
      "epoch": 2.1483483306578695,
      "grad_norm": 1.3907585144042969,
      "learning_rate": 1.2678020906288251e-05,
      "loss": 0.0594,
      "step": 145875
    },
    {
      "epoch": 2.1487165137479565,
      "grad_norm": 0.7849710583686829,
      "learning_rate": 1.2676384534321267e-05,
      "loss": 0.0668,
      "step": 145900
    },
    {
      "epoch": 2.1490846968380435,
      "grad_norm": 0.9230354428291321,
      "learning_rate": 1.2674748162354282e-05,
      "loss": 0.065,
      "step": 145925
    },
    {
      "epoch": 2.1494528799281305,
      "grad_norm": 1.080701470375061,
      "learning_rate": 1.2673111790387298e-05,
      "loss": 0.0564,
      "step": 145950
    },
    {
      "epoch": 2.1498210630182175,
      "grad_norm": 1.5436818599700928,
      "learning_rate": 1.2671475418420314e-05,
      "loss": 0.0632,
      "step": 145975
    },
    {
      "epoch": 2.1501892461083045,
      "grad_norm": 0.9225010275840759,
      "learning_rate": 1.266983904645333e-05,
      "loss": 0.0647,
      "step": 146000
    },
    {
      "epoch": 2.1505574291983915,
      "grad_norm": 1.6171057224273682,
      "learning_rate": 1.2668202674486343e-05,
      "loss": 0.062,
      "step": 146025
    },
    {
      "epoch": 2.150925612288479,
      "grad_norm": 1.8668445348739624,
      "learning_rate": 1.2666566302519359e-05,
      "loss": 0.061,
      "step": 146050
    },
    {
      "epoch": 2.151293795378566,
      "grad_norm": 1.1698726415634155,
      "learning_rate": 1.2664929930552373e-05,
      "loss": 0.0636,
      "step": 146075
    },
    {
      "epoch": 2.151661978468653,
      "grad_norm": 1.7313923835754395,
      "learning_rate": 1.266329355858539e-05,
      "loss": 0.0632,
      "step": 146100
    },
    {
      "epoch": 2.15203016155874,
      "grad_norm": 1.4423819780349731,
      "learning_rate": 1.2661657186618406e-05,
      "loss": 0.0607,
      "step": 146125
    },
    {
      "epoch": 2.152398344648827,
      "grad_norm": 1.4605655670166016,
      "learning_rate": 1.2660020814651422e-05,
      "loss": 0.0614,
      "step": 146150
    },
    {
      "epoch": 2.152766527738914,
      "grad_norm": 1.4070481061935425,
      "learning_rate": 1.2658384442684436e-05,
      "loss": 0.0625,
      "step": 146175
    },
    {
      "epoch": 2.153134710829001,
      "grad_norm": 1.1662936210632324,
      "learning_rate": 1.2656748070717453e-05,
      "loss": 0.0532,
      "step": 146200
    },
    {
      "epoch": 2.153502893919088,
      "grad_norm": 2.1889052391052246,
      "learning_rate": 1.2655111698750467e-05,
      "loss": 0.0554,
      "step": 146225
    },
    {
      "epoch": 2.153871077009175,
      "grad_norm": 1.0931512117385864,
      "learning_rate": 1.2653475326783481e-05,
      "loss": 0.0544,
      "step": 146250
    },
    {
      "epoch": 2.154239260099262,
      "grad_norm": 1.5517659187316895,
      "learning_rate": 1.2651838954816498e-05,
      "loss": 0.0653,
      "step": 146275
    },
    {
      "epoch": 2.154607443189349,
      "grad_norm": 1.179627776145935,
      "learning_rate": 1.2650202582849514e-05,
      "loss": 0.0655,
      "step": 146300
    },
    {
      "epoch": 2.154975626279436,
      "grad_norm": 1.2843074798583984,
      "learning_rate": 1.264856621088253e-05,
      "loss": 0.0588,
      "step": 146325
    },
    {
      "epoch": 2.155343809369523,
      "grad_norm": 1.0195711851119995,
      "learning_rate": 1.2646929838915544e-05,
      "loss": 0.0586,
      "step": 146350
    },
    {
      "epoch": 2.15571199245961,
      "grad_norm": 0.9936036467552185,
      "learning_rate": 1.264535892182724e-05,
      "loss": 0.0617,
      "step": 146375
    },
    {
      "epoch": 2.156080175549697,
      "grad_norm": 1.2481807470321655,
      "learning_rate": 1.2643722549860256e-05,
      "loss": 0.0586,
      "step": 146400
    },
    {
      "epoch": 2.1564483586397842,
      "grad_norm": 1.269595980644226,
      "learning_rate": 1.264208617789327e-05,
      "loss": 0.0644,
      "step": 146425
    },
    {
      "epoch": 2.1568165417298713,
      "grad_norm": 1.208337426185608,
      "learning_rate": 1.2640449805926285e-05,
      "loss": 0.0647,
      "step": 146450
    },
    {
      "epoch": 2.1571847248199583,
      "grad_norm": 1.532379150390625,
      "learning_rate": 1.2638813433959301e-05,
      "loss": 0.0624,
      "step": 146475
    },
    {
      "epoch": 2.1575529079100457,
      "grad_norm": 1.3638733625411987,
      "learning_rate": 1.2637177061992316e-05,
      "loss": 0.0658,
      "step": 146500
    },
    {
      "epoch": 2.1579210910001327,
      "grad_norm": 1.1447440385818481,
      "learning_rate": 1.2635540690025332e-05,
      "loss": 0.0587,
      "step": 146525
    },
    {
      "epoch": 2.1582892740902198,
      "grad_norm": 1.1884541511535645,
      "learning_rate": 1.2633904318058348e-05,
      "loss": 0.0631,
      "step": 146550
    },
    {
      "epoch": 2.1586574571803068,
      "grad_norm": 1.6917191743850708,
      "learning_rate": 1.2632267946091364e-05,
      "loss": 0.0628,
      "step": 146575
    },
    {
      "epoch": 2.159025640270394,
      "grad_norm": 1.0723440647125244,
      "learning_rate": 1.2630631574124379e-05,
      "loss": 0.0691,
      "step": 146600
    },
    {
      "epoch": 2.159393823360481,
      "grad_norm": 1.411756992340088,
      "learning_rate": 1.2628995202157395e-05,
      "loss": 0.0688,
      "step": 146625
    },
    {
      "epoch": 2.159762006450568,
      "grad_norm": 1.2132647037506104,
      "learning_rate": 1.2627358830190407e-05,
      "loss": 0.0551,
      "step": 146650
    },
    {
      "epoch": 2.160130189540655,
      "grad_norm": 1.6361756324768066,
      "learning_rate": 1.2625722458223424e-05,
      "loss": 0.068,
      "step": 146675
    },
    {
      "epoch": 2.160498372630742,
      "grad_norm": 1.1633211374282837,
      "learning_rate": 1.262408608625644e-05,
      "loss": 0.064,
      "step": 146700
    },
    {
      "epoch": 2.160866555720829,
      "grad_norm": 0.9648963809013367,
      "learning_rate": 1.2622449714289456e-05,
      "loss": 0.0577,
      "step": 146725
    },
    {
      "epoch": 2.161234738810916,
      "grad_norm": 1.452364206314087,
      "learning_rate": 1.262081334232247e-05,
      "loss": 0.06,
      "step": 146750
    },
    {
      "epoch": 2.161602921901003,
      "grad_norm": 1.4971153736114502,
      "learning_rate": 1.2619176970355487e-05,
      "loss": 0.0698,
      "step": 146775
    },
    {
      "epoch": 2.16197110499109,
      "grad_norm": 1.226670265197754,
      "learning_rate": 1.2617540598388503e-05,
      "loss": 0.0654,
      "step": 146800
    },
    {
      "epoch": 2.162339288081177,
      "grad_norm": 1.5489416122436523,
      "learning_rate": 1.2615904226421519e-05,
      "loss": 0.0622,
      "step": 146825
    },
    {
      "epoch": 2.162707471171264,
      "grad_norm": 1.175428867340088,
      "learning_rate": 1.2614267854454533e-05,
      "loss": 0.0605,
      "step": 146850
    },
    {
      "epoch": 2.163075654261351,
      "grad_norm": 1.4683444499969482,
      "learning_rate": 1.2612631482487548e-05,
      "loss": 0.0672,
      "step": 146875
    },
    {
      "epoch": 2.163443837351438,
      "grad_norm": 1.3437857627868652,
      "learning_rate": 1.2610995110520562e-05,
      "loss": 0.0609,
      "step": 146900
    },
    {
      "epoch": 2.163812020441525,
      "grad_norm": 1.406295657157898,
      "learning_rate": 1.2609358738553578e-05,
      "loss": 0.0565,
      "step": 146925
    },
    {
      "epoch": 2.164180203531612,
      "grad_norm": 1.2059030532836914,
      "learning_rate": 1.2607722366586594e-05,
      "loss": 0.0608,
      "step": 146950
    },
    {
      "epoch": 2.164548386621699,
      "grad_norm": 1.5873130559921265,
      "learning_rate": 1.260608599461961e-05,
      "loss": 0.0653,
      "step": 146975
    },
    {
      "epoch": 2.1649165697117865,
      "grad_norm": 1.0940394401550293,
      "learning_rate": 1.2604449622652625e-05,
      "loss": 0.0591,
      "step": 147000
    },
    {
      "epoch": 2.1652847528018735,
      "grad_norm": 1.469257116317749,
      "learning_rate": 1.2602813250685641e-05,
      "loss": 0.0638,
      "step": 147025
    },
    {
      "epoch": 2.1656529358919605,
      "grad_norm": 1.5114831924438477,
      "learning_rate": 1.2601176878718657e-05,
      "loss": 0.0666,
      "step": 147050
    },
    {
      "epoch": 2.1660211189820475,
      "grad_norm": 1.5783467292785645,
      "learning_rate": 1.259954050675167e-05,
      "loss": 0.0589,
      "step": 147075
    },
    {
      "epoch": 2.1663893020721345,
      "grad_norm": 1.2109485864639282,
      "learning_rate": 1.2597904134784686e-05,
      "loss": 0.063,
      "step": 147100
    },
    {
      "epoch": 2.1667574851622216,
      "grad_norm": 1.6290923357009888,
      "learning_rate": 1.2596267762817702e-05,
      "loss": 0.0618,
      "step": 147125
    },
    {
      "epoch": 2.1671256682523086,
      "grad_norm": 1.8701249361038208,
      "learning_rate": 1.2594631390850719e-05,
      "loss": 0.0622,
      "step": 147150
    },
    {
      "epoch": 2.1674938513423956,
      "grad_norm": 1.2110137939453125,
      "learning_rate": 1.2592995018883733e-05,
      "loss": 0.0533,
      "step": 147175
    },
    {
      "epoch": 2.1678620344324826,
      "grad_norm": 1.2196629047393799,
      "learning_rate": 1.259135864691675e-05,
      "loss": 0.0628,
      "step": 147200
    },
    {
      "epoch": 2.1682302175225696,
      "grad_norm": 1.3484305143356323,
      "learning_rate": 1.2589722274949765e-05,
      "loss": 0.0616,
      "step": 147225
    },
    {
      "epoch": 2.1685984006126566,
      "grad_norm": 1.7585560083389282,
      "learning_rate": 1.258808590298278e-05,
      "loss": 0.0661,
      "step": 147250
    },
    {
      "epoch": 2.1689665837027436,
      "grad_norm": 0.9657861590385437,
      "learning_rate": 1.2586449531015796e-05,
      "loss": 0.059,
      "step": 147275
    },
    {
      "epoch": 2.1693347667928307,
      "grad_norm": 1.6587380170822144,
      "learning_rate": 1.258481315904881e-05,
      "loss": 0.0629,
      "step": 147300
    },
    {
      "epoch": 2.1697029498829177,
      "grad_norm": 1.3727116584777832,
      "learning_rate": 1.2583176787081825e-05,
      "loss": 0.0577,
      "step": 147325
    },
    {
      "epoch": 2.1700711329730047,
      "grad_norm": 1.3495436906814575,
      "learning_rate": 1.2581540415114841e-05,
      "loss": 0.0566,
      "step": 147350
    },
    {
      "epoch": 2.1704393160630917,
      "grad_norm": 1.3399096727371216,
      "learning_rate": 1.2579904043147857e-05,
      "loss": 0.0592,
      "step": 147375
    },
    {
      "epoch": 2.1708074991531787,
      "grad_norm": 1.6004917621612549,
      "learning_rate": 1.2578267671180873e-05,
      "loss": 0.067,
      "step": 147400
    },
    {
      "epoch": 2.1711756822432657,
      "grad_norm": 1.8596436977386475,
      "learning_rate": 1.2576631299213888e-05,
      "loss": 0.0564,
      "step": 147425
    },
    {
      "epoch": 2.171543865333353,
      "grad_norm": 1.4039843082427979,
      "learning_rate": 1.2574994927246904e-05,
      "loss": 0.0603,
      "step": 147450
    },
    {
      "epoch": 2.17191204842344,
      "grad_norm": 0.8630525469779968,
      "learning_rate": 1.257335855527992e-05,
      "loss": 0.0621,
      "step": 147475
    },
    {
      "epoch": 2.1722802315135272,
      "grad_norm": 1.2812976837158203,
      "learning_rate": 1.2571722183312933e-05,
      "loss": 0.0636,
      "step": 147500
    },
    {
      "epoch": 2.1726484146036142,
      "grad_norm": 1.1953637599945068,
      "learning_rate": 1.2570085811345949e-05,
      "loss": 0.0627,
      "step": 147525
    },
    {
      "epoch": 2.1730165976937013,
      "grad_norm": 1.6726806163787842,
      "learning_rate": 1.2568449439378965e-05,
      "loss": 0.0616,
      "step": 147550
    },
    {
      "epoch": 2.1733847807837883,
      "grad_norm": 1.4168367385864258,
      "learning_rate": 1.256681306741198e-05,
      "loss": 0.0627,
      "step": 147575
    },
    {
      "epoch": 2.1737529638738753,
      "grad_norm": 1.8928165435791016,
      "learning_rate": 1.2565176695444996e-05,
      "loss": 0.0625,
      "step": 147600
    },
    {
      "epoch": 2.1741211469639623,
      "grad_norm": 1.7232677936553955,
      "learning_rate": 1.2563540323478012e-05,
      "loss": 0.0614,
      "step": 147625
    },
    {
      "epoch": 2.1744893300540493,
      "grad_norm": 0.9866328239440918,
      "learning_rate": 1.2561903951511028e-05,
      "loss": 0.061,
      "step": 147650
    },
    {
      "epoch": 2.1748575131441363,
      "grad_norm": 1.1385637521743774,
      "learning_rate": 1.2560267579544043e-05,
      "loss": 0.0597,
      "step": 147675
    },
    {
      "epoch": 2.1752256962342233,
      "grad_norm": 1.1440985202789307,
      "learning_rate": 1.2558631207577059e-05,
      "loss": 0.0728,
      "step": 147700
    },
    {
      "epoch": 2.1755938793243104,
      "grad_norm": 1.4859899282455444,
      "learning_rate": 1.2556994835610073e-05,
      "loss": 0.0533,
      "step": 147725
    },
    {
      "epoch": 2.1759620624143974,
      "grad_norm": 0.6152855753898621,
      "learning_rate": 1.2555358463643088e-05,
      "loss": 0.0552,
      "step": 147750
    },
    {
      "epoch": 2.1763302455044844,
      "grad_norm": 1.1117513179779053,
      "learning_rate": 1.2553722091676104e-05,
      "loss": 0.0595,
      "step": 147775
    },
    {
      "epoch": 2.1766984285945714,
      "grad_norm": 1.3607580661773682,
      "learning_rate": 1.255208571970912e-05,
      "loss": 0.0591,
      "step": 147800
    },
    {
      "epoch": 2.1770666116846584,
      "grad_norm": 1.2065541744232178,
      "learning_rate": 1.2550449347742134e-05,
      "loss": 0.0529,
      "step": 147825
    },
    {
      "epoch": 2.1774347947747454,
      "grad_norm": 1.242772102355957,
      "learning_rate": 1.254881297577515e-05,
      "loss": 0.067,
      "step": 147850
    },
    {
      "epoch": 2.1778029778648325,
      "grad_norm": 1.523486852645874,
      "learning_rate": 1.2547176603808167e-05,
      "loss": 0.0628,
      "step": 147875
    },
    {
      "epoch": 2.1781711609549195,
      "grad_norm": 1.2696503400802612,
      "learning_rate": 1.2545540231841183e-05,
      "loss": 0.062,
      "step": 147900
    },
    {
      "epoch": 2.1785393440450065,
      "grad_norm": 1.4526379108428955,
      "learning_rate": 1.2543903859874196e-05,
      "loss": 0.0622,
      "step": 147925
    },
    {
      "epoch": 2.178907527135094,
      "grad_norm": 1.4397584199905396,
      "learning_rate": 1.2542267487907212e-05,
      "loss": 0.0716,
      "step": 147950
    },
    {
      "epoch": 2.179275710225181,
      "grad_norm": 1.6296885013580322,
      "learning_rate": 1.2540631115940228e-05,
      "loss": 0.0656,
      "step": 147975
    },
    {
      "epoch": 2.179643893315268,
      "grad_norm": 0.8643805384635925,
      "learning_rate": 1.2538994743973242e-05,
      "loss": 0.0666,
      "step": 148000
    },
    {
      "epoch": 2.180012076405355,
      "grad_norm": 1.0706731081008911,
      "learning_rate": 1.2537358372006259e-05,
      "loss": 0.0616,
      "step": 148025
    },
    {
      "epoch": 2.180380259495442,
      "grad_norm": 1.2822765111923218,
      "learning_rate": 1.2535722000039275e-05,
      "loss": 0.0686,
      "step": 148050
    },
    {
      "epoch": 2.180748442585529,
      "grad_norm": 1.0711289644241333,
      "learning_rate": 1.253408562807229e-05,
      "loss": 0.0665,
      "step": 148075
    },
    {
      "epoch": 2.181116625675616,
      "grad_norm": 1.1549954414367676,
      "learning_rate": 1.2532449256105305e-05,
      "loss": 0.0662,
      "step": 148100
    },
    {
      "epoch": 2.181484808765703,
      "grad_norm": 1.2080084085464478,
      "learning_rate": 1.253081288413832e-05,
      "loss": 0.0662,
      "step": 148125
    },
    {
      "epoch": 2.18185299185579,
      "grad_norm": 1.492305040359497,
      "learning_rate": 1.2529176512171334e-05,
      "loss": 0.0643,
      "step": 148150
    },
    {
      "epoch": 2.182221174945877,
      "grad_norm": 1.7838488817214966,
      "learning_rate": 1.252754014020435e-05,
      "loss": 0.0648,
      "step": 148175
    },
    {
      "epoch": 2.182589358035964,
      "grad_norm": 1.157466173171997,
      "learning_rate": 1.2525903768237366e-05,
      "loss": 0.0649,
      "step": 148200
    },
    {
      "epoch": 2.182957541126051,
      "grad_norm": 0.857884407043457,
      "learning_rate": 1.2524267396270383e-05,
      "loss": 0.0665,
      "step": 148225
    },
    {
      "epoch": 2.183325724216138,
      "grad_norm": 1.0878560543060303,
      "learning_rate": 1.2522631024303397e-05,
      "loss": 0.0607,
      "step": 148250
    },
    {
      "epoch": 2.183693907306225,
      "grad_norm": 1.3036624193191528,
      "learning_rate": 1.2520994652336413e-05,
      "loss": 0.0635,
      "step": 148275
    },
    {
      "epoch": 2.184062090396312,
      "grad_norm": 1.7599238157272339,
      "learning_rate": 1.251935828036943e-05,
      "loss": 0.0621,
      "step": 148300
    },
    {
      "epoch": 2.184430273486399,
      "grad_norm": 1.70572829246521,
      "learning_rate": 1.2517721908402446e-05,
      "loss": 0.0637,
      "step": 148325
    },
    {
      "epoch": 2.184798456576486,
      "grad_norm": 1.495478630065918,
      "learning_rate": 1.2516085536435458e-05,
      "loss": 0.0621,
      "step": 148350
    },
    {
      "epoch": 2.185166639666573,
      "grad_norm": 1.394152045249939,
      "learning_rate": 1.2514449164468474e-05,
      "loss": 0.0671,
      "step": 148375
    },
    {
      "epoch": 2.1855348227566607,
      "grad_norm": 1.3999066352844238,
      "learning_rate": 1.2512812792501489e-05,
      "loss": 0.0645,
      "step": 148400
    },
    {
      "epoch": 2.1859030058467477,
      "grad_norm": 1.5172370672225952,
      "learning_rate": 1.2511176420534505e-05,
      "loss": 0.0662,
      "step": 148425
    },
    {
      "epoch": 2.1862711889368347,
      "grad_norm": 1.1504250764846802,
      "learning_rate": 1.2509540048567521e-05,
      "loss": 0.0614,
      "step": 148450
    },
    {
      "epoch": 2.1866393720269217,
      "grad_norm": 1.031248688697815,
      "learning_rate": 1.2507903676600537e-05,
      "loss": 0.0672,
      "step": 148475
    },
    {
      "epoch": 2.1870075551170087,
      "grad_norm": 1.552599549293518,
      "learning_rate": 1.2506267304633552e-05,
      "loss": 0.0581,
      "step": 148500
    },
    {
      "epoch": 2.1873757382070957,
      "grad_norm": 1.2594009637832642,
      "learning_rate": 1.2504630932666568e-05,
      "loss": 0.0646,
      "step": 148525
    },
    {
      "epoch": 2.1877439212971828,
      "grad_norm": 1.7048428058624268,
      "learning_rate": 1.2502994560699582e-05,
      "loss": 0.0673,
      "step": 148550
    },
    {
      "epoch": 2.1881121043872698,
      "grad_norm": 1.3809187412261963,
      "learning_rate": 1.2501358188732597e-05,
      "loss": 0.0596,
      "step": 148575
    },
    {
      "epoch": 2.188480287477357,
      "grad_norm": 1.5433809757232666,
      "learning_rate": 1.2499721816765613e-05,
      "loss": 0.0628,
      "step": 148600
    },
    {
      "epoch": 2.188848470567444,
      "grad_norm": 1.3568446636199951,
      "learning_rate": 1.249808544479863e-05,
      "loss": 0.0603,
      "step": 148625
    },
    {
      "epoch": 2.189216653657531,
      "grad_norm": 1.197313666343689,
      "learning_rate": 1.2496449072831645e-05,
      "loss": 0.0582,
      "step": 148650
    },
    {
      "epoch": 2.189584836747618,
      "grad_norm": 1.5377166271209717,
      "learning_rate": 1.249481270086466e-05,
      "loss": 0.0634,
      "step": 148675
    },
    {
      "epoch": 2.189953019837705,
      "grad_norm": 0.996576189994812,
      "learning_rate": 1.2493176328897676e-05,
      "loss": 0.0638,
      "step": 148700
    },
    {
      "epoch": 2.190321202927792,
      "grad_norm": 1.5729095935821533,
      "learning_rate": 1.2491539956930692e-05,
      "loss": 0.0635,
      "step": 148725
    },
    {
      "epoch": 2.190689386017879,
      "grad_norm": 1.2386174201965332,
      "learning_rate": 1.2489903584963707e-05,
      "loss": 0.0601,
      "step": 148750
    },
    {
      "epoch": 2.191057569107966,
      "grad_norm": 1.6151951551437378,
      "learning_rate": 1.2488267212996721e-05,
      "loss": 0.0654,
      "step": 148775
    },
    {
      "epoch": 2.191425752198053,
      "grad_norm": 1.192921757698059,
      "learning_rate": 1.2486630841029737e-05,
      "loss": 0.0646,
      "step": 148800
    },
    {
      "epoch": 2.19179393528814,
      "grad_norm": 1.41513192653656,
      "learning_rate": 1.2484994469062752e-05,
      "loss": 0.0665,
      "step": 148825
    },
    {
      "epoch": 2.192162118378227,
      "grad_norm": 1.1037499904632568,
      "learning_rate": 1.2483358097095768e-05,
      "loss": 0.0646,
      "step": 148850
    },
    {
      "epoch": 2.192530301468314,
      "grad_norm": 1.6659947633743286,
      "learning_rate": 1.2481721725128784e-05,
      "loss": 0.061,
      "step": 148875
    },
    {
      "epoch": 2.192898484558401,
      "grad_norm": 1.3756612539291382,
      "learning_rate": 1.24800853531618e-05,
      "loss": 0.0603,
      "step": 148900
    },
    {
      "epoch": 2.1932666676484884,
      "grad_norm": 1.5419251918792725,
      "learning_rate": 1.2478448981194815e-05,
      "loss": 0.0596,
      "step": 148925
    },
    {
      "epoch": 2.1936348507385754,
      "grad_norm": 1.496291995048523,
      "learning_rate": 1.247681260922783e-05,
      "loss": 0.0608,
      "step": 148950
    },
    {
      "epoch": 2.1940030338286625,
      "grad_norm": 1.7220516204833984,
      "learning_rate": 1.2475176237260843e-05,
      "loss": 0.0694,
      "step": 148975
    },
    {
      "epoch": 2.1943712169187495,
      "grad_norm": 1.1940354108810425,
      "learning_rate": 1.247353986529386e-05,
      "loss": 0.0609,
      "step": 149000
    },
    {
      "epoch": 2.1947394000088365,
      "grad_norm": 1.6296696662902832,
      "learning_rate": 1.2471903493326876e-05,
      "loss": 0.0591,
      "step": 149025
    },
    {
      "epoch": 2.1951075830989235,
      "grad_norm": 1.7357245683670044,
      "learning_rate": 1.2470267121359892e-05,
      "loss": 0.0613,
      "step": 149050
    },
    {
      "epoch": 2.1954757661890105,
      "grad_norm": 1.6018069982528687,
      "learning_rate": 1.2468630749392906e-05,
      "loss": 0.0601,
      "step": 149075
    },
    {
      "epoch": 2.1958439492790975,
      "grad_norm": 1.5958993434906006,
      "learning_rate": 1.2466994377425923e-05,
      "loss": 0.065,
      "step": 149100
    },
    {
      "epoch": 2.1962121323691846,
      "grad_norm": 1.0688061714172363,
      "learning_rate": 1.2465358005458939e-05,
      "loss": 0.0582,
      "step": 149125
    },
    {
      "epoch": 2.1965803154592716,
      "grad_norm": 1.1995221376419067,
      "learning_rate": 1.2463721633491955e-05,
      "loss": 0.0645,
      "step": 149150
    },
    {
      "epoch": 2.1969484985493586,
      "grad_norm": 1.3568192720413208,
      "learning_rate": 1.246208526152497e-05,
      "loss": 0.0595,
      "step": 149175
    },
    {
      "epoch": 2.1973166816394456,
      "grad_norm": 1.1569491624832153,
      "learning_rate": 1.2460448889557984e-05,
      "loss": 0.063,
      "step": 149200
    },
    {
      "epoch": 2.1976848647295326,
      "grad_norm": 1.4922245740890503,
      "learning_rate": 1.2458812517591e-05,
      "loss": 0.0671,
      "step": 149225
    },
    {
      "epoch": 2.1980530478196196,
      "grad_norm": 1.4305675029754639,
      "learning_rate": 1.2457176145624014e-05,
      "loss": 0.0633,
      "step": 149250
    },
    {
      "epoch": 2.1984212309097066,
      "grad_norm": 1.384347677230835,
      "learning_rate": 1.245553977365703e-05,
      "loss": 0.0571,
      "step": 149275
    },
    {
      "epoch": 2.1987894139997937,
      "grad_norm": 1.5995211601257324,
      "learning_rate": 1.2453903401690047e-05,
      "loss": 0.0657,
      "step": 149300
    },
    {
      "epoch": 2.1991575970898807,
      "grad_norm": 1.1055463552474976,
      "learning_rate": 1.2452267029723061e-05,
      "loss": 0.0663,
      "step": 149325
    },
    {
      "epoch": 2.1995257801799677,
      "grad_norm": 1.339615821838379,
      "learning_rate": 1.2450630657756077e-05,
      "loss": 0.0598,
      "step": 149350
    },
    {
      "epoch": 2.199893963270055,
      "grad_norm": 0.9209982752799988,
      "learning_rate": 1.2448994285789093e-05,
      "loss": 0.0553,
      "step": 149375
    },
    {
      "epoch": 2.200262146360142,
      "grad_norm": 0.9548072814941406,
      "learning_rate": 1.2447357913822106e-05,
      "loss": 0.0649,
      "step": 149400
    },
    {
      "epoch": 2.200630329450229,
      "grad_norm": 1.3747440576553345,
      "learning_rate": 1.2445721541855122e-05,
      "loss": 0.0544,
      "step": 149425
    },
    {
      "epoch": 2.200998512540316,
      "grad_norm": 1.2757736444473267,
      "learning_rate": 1.2444085169888138e-05,
      "loss": 0.0634,
      "step": 149450
    },
    {
      "epoch": 2.201366695630403,
      "grad_norm": 1.3781328201293945,
      "learning_rate": 1.2442448797921155e-05,
      "loss": 0.0633,
      "step": 149475
    },
    {
      "epoch": 2.2017348787204902,
      "grad_norm": 1.3538663387298584,
      "learning_rate": 1.2440812425954169e-05,
      "loss": 0.065,
      "step": 149500
    },
    {
      "epoch": 2.2021030618105772,
      "grad_norm": 1.1612451076507568,
      "learning_rate": 1.2439176053987185e-05,
      "loss": 0.0641,
      "step": 149525
    },
    {
      "epoch": 2.2024712449006643,
      "grad_norm": 1.6553928852081299,
      "learning_rate": 1.2437539682020201e-05,
      "loss": 0.0643,
      "step": 149550
    },
    {
      "epoch": 2.2028394279907513,
      "grad_norm": 1.4857532978057861,
      "learning_rate": 1.2435903310053218e-05,
      "loss": 0.0601,
      "step": 149575
    },
    {
      "epoch": 2.2032076110808383,
      "grad_norm": 1.23832106590271,
      "learning_rate": 1.243426693808623e-05,
      "loss": 0.0599,
      "step": 149600
    },
    {
      "epoch": 2.2035757941709253,
      "grad_norm": 1.535791277885437,
      "learning_rate": 1.2432630566119246e-05,
      "loss": 0.0617,
      "step": 149625
    },
    {
      "epoch": 2.2039439772610123,
      "grad_norm": 1.5874391794204712,
      "learning_rate": 1.2430994194152261e-05,
      "loss": 0.0652,
      "step": 149650
    },
    {
      "epoch": 2.2043121603510993,
      "grad_norm": 1.652728796005249,
      "learning_rate": 1.2429357822185277e-05,
      "loss": 0.066,
      "step": 149675
    },
    {
      "epoch": 2.2046803434411864,
      "grad_norm": 1.1046514511108398,
      "learning_rate": 1.2427721450218293e-05,
      "loss": 0.0631,
      "step": 149700
    },
    {
      "epoch": 2.2050485265312734,
      "grad_norm": 1.078290343284607,
      "learning_rate": 1.242608507825131e-05,
      "loss": 0.0604,
      "step": 149725
    },
    {
      "epoch": 2.2054167096213604,
      "grad_norm": 1.287468671798706,
      "learning_rate": 1.2424448706284324e-05,
      "loss": 0.0669,
      "step": 149750
    },
    {
      "epoch": 2.2057848927114474,
      "grad_norm": 1.630241870880127,
      "learning_rate": 1.242281233431734e-05,
      "loss": 0.07,
      "step": 149775
    },
    {
      "epoch": 2.2061530758015344,
      "grad_norm": 1.7726587057113647,
      "learning_rate": 1.2421175962350356e-05,
      "loss": 0.061,
      "step": 149800
    },
    {
      "epoch": 2.2065212588916214,
      "grad_norm": 0.9498474597930908,
      "learning_rate": 1.2419539590383369e-05,
      "loss": 0.0532,
      "step": 149825
    },
    {
      "epoch": 2.2068894419817084,
      "grad_norm": 1.8226207494735718,
      "learning_rate": 1.2417903218416385e-05,
      "loss": 0.0644,
      "step": 149850
    },
    {
      "epoch": 2.207257625071796,
      "grad_norm": 1.028624176979065,
      "learning_rate": 1.2416266846449401e-05,
      "loss": 0.0667,
      "step": 149875
    },
    {
      "epoch": 2.207625808161883,
      "grad_norm": 1.1617398262023926,
      "learning_rate": 1.2414630474482416e-05,
      "loss": 0.0567,
      "step": 149900
    },
    {
      "epoch": 2.20799399125197,
      "grad_norm": 0.8688008189201355,
      "learning_rate": 1.2412994102515432e-05,
      "loss": 0.0622,
      "step": 149925
    },
    {
      "epoch": 2.208362174342057,
      "grad_norm": 0.8188631534576416,
      "learning_rate": 1.2411357730548448e-05,
      "loss": 0.0598,
      "step": 149950
    },
    {
      "epoch": 2.208730357432144,
      "grad_norm": 0.9632925391197205,
      "learning_rate": 1.2409721358581464e-05,
      "loss": 0.0625,
      "step": 149975
    },
    {
      "epoch": 2.209098540522231,
      "grad_norm": 1.6538541316986084,
      "learning_rate": 1.2408084986614479e-05,
      "loss": 0.0526,
      "step": 150000
    },
    {
      "epoch": 2.209466723612318,
      "grad_norm": 1.8903040885925293,
      "learning_rate": 1.2406448614647493e-05,
      "loss": 0.0631,
      "step": 150025
    },
    {
      "epoch": 2.209834906702405,
      "grad_norm": 1.2473844289779663,
      "learning_rate": 1.240481224268051e-05,
      "loss": 0.0646,
      "step": 150050
    },
    {
      "epoch": 2.210203089792492,
      "grad_norm": 1.612788200378418,
      "learning_rate": 1.2403175870713524e-05,
      "loss": 0.0646,
      "step": 150075
    },
    {
      "epoch": 2.210571272882579,
      "grad_norm": 1.7383086681365967,
      "learning_rate": 1.240153949874654e-05,
      "loss": 0.0675,
      "step": 150100
    },
    {
      "epoch": 2.210939455972666,
      "grad_norm": 1.6361398696899414,
      "learning_rate": 1.2399903126779556e-05,
      "loss": 0.068,
      "step": 150125
    },
    {
      "epoch": 2.211307639062753,
      "grad_norm": 1.5248494148254395,
      "learning_rate": 1.239826675481257e-05,
      "loss": 0.0663,
      "step": 150150
    },
    {
      "epoch": 2.21167582215284,
      "grad_norm": 0.9361855387687683,
      "learning_rate": 1.2396630382845587e-05,
      "loss": 0.0566,
      "step": 150175
    },
    {
      "epoch": 2.212044005242927,
      "grad_norm": 1.2662063837051392,
      "learning_rate": 1.2394994010878603e-05,
      "loss": 0.0606,
      "step": 150200
    },
    {
      "epoch": 2.212412188333014,
      "grad_norm": 1.3367865085601807,
      "learning_rate": 1.2393357638911619e-05,
      "loss": 0.0647,
      "step": 150225
    },
    {
      "epoch": 2.212780371423101,
      "grad_norm": 0.9757417440414429,
      "learning_rate": 1.2391721266944632e-05,
      "loss": 0.0543,
      "step": 150250
    },
    {
      "epoch": 2.213148554513188,
      "grad_norm": 1.5465177297592163,
      "learning_rate": 1.2390084894977648e-05,
      "loss": 0.0638,
      "step": 150275
    },
    {
      "epoch": 2.213516737603275,
      "grad_norm": 1.637668251991272,
      "learning_rate": 1.2388448523010664e-05,
      "loss": 0.0589,
      "step": 150300
    },
    {
      "epoch": 2.2138849206933626,
      "grad_norm": 1.2209950685501099,
      "learning_rate": 1.2386877605922358e-05,
      "loss": 0.0575,
      "step": 150325
    },
    {
      "epoch": 2.2142531037834496,
      "grad_norm": 1.781244158744812,
      "learning_rate": 1.2385241233955374e-05,
      "loss": 0.0675,
      "step": 150350
    },
    {
      "epoch": 2.2146212868735367,
      "grad_norm": 1.7301344871520996,
      "learning_rate": 1.238360486198839e-05,
      "loss": 0.0661,
      "step": 150375
    },
    {
      "epoch": 2.2149894699636237,
      "grad_norm": 1.3790034055709839,
      "learning_rate": 1.2381968490021406e-05,
      "loss": 0.0611,
      "step": 150400
    },
    {
      "epoch": 2.2153576530537107,
      "grad_norm": 1.5492733716964722,
      "learning_rate": 1.238033211805442e-05,
      "loss": 0.0658,
      "step": 150425
    },
    {
      "epoch": 2.2157258361437977,
      "grad_norm": 1.0602071285247803,
      "learning_rate": 1.2378695746087435e-05,
      "loss": 0.0636,
      "step": 150450
    },
    {
      "epoch": 2.2160940192338847,
      "grad_norm": 1.7389768362045288,
      "learning_rate": 1.237705937412045e-05,
      "loss": 0.0647,
      "step": 150475
    },
    {
      "epoch": 2.2164622023239717,
      "grad_norm": 1.8014004230499268,
      "learning_rate": 1.2375423002153466e-05,
      "loss": 0.0681,
      "step": 150500
    },
    {
      "epoch": 2.2168303854140587,
      "grad_norm": 1.4853734970092773,
      "learning_rate": 1.2373786630186482e-05,
      "loss": 0.0641,
      "step": 150525
    },
    {
      "epoch": 2.2171985685041458,
      "grad_norm": 1.3523290157318115,
      "learning_rate": 1.2372150258219498e-05,
      "loss": 0.0647,
      "step": 150550
    },
    {
      "epoch": 2.2175667515942328,
      "grad_norm": 1.4068225622177124,
      "learning_rate": 1.2370513886252513e-05,
      "loss": 0.0635,
      "step": 150575
    },
    {
      "epoch": 2.21793493468432,
      "grad_norm": 1.6109051704406738,
      "learning_rate": 1.2368877514285529e-05,
      "loss": 0.0644,
      "step": 150600
    },
    {
      "epoch": 2.218303117774407,
      "grad_norm": 1.5518189668655396,
      "learning_rate": 1.2367241142318545e-05,
      "loss": 0.063,
      "step": 150625
    },
    {
      "epoch": 2.218671300864494,
      "grad_norm": 1.8075902462005615,
      "learning_rate": 1.2365604770351561e-05,
      "loss": 0.064,
      "step": 150650
    },
    {
      "epoch": 2.219039483954581,
      "grad_norm": 1.2301523685455322,
      "learning_rate": 1.2363968398384574e-05,
      "loss": 0.0655,
      "step": 150675
    },
    {
      "epoch": 2.219407667044668,
      "grad_norm": 1.6648341417312622,
      "learning_rate": 1.236233202641759e-05,
      "loss": 0.0653,
      "step": 150700
    },
    {
      "epoch": 2.219775850134755,
      "grad_norm": 1.4048888683319092,
      "learning_rate": 1.2360695654450604e-05,
      "loss": 0.0665,
      "step": 150725
    },
    {
      "epoch": 2.220144033224842,
      "grad_norm": 1.1980386972427368,
      "learning_rate": 1.235905928248362e-05,
      "loss": 0.0635,
      "step": 150750
    },
    {
      "epoch": 2.220512216314929,
      "grad_norm": 1.6498727798461914,
      "learning_rate": 1.2357422910516637e-05,
      "loss": 0.0652,
      "step": 150775
    },
    {
      "epoch": 2.220880399405016,
      "grad_norm": 1.7174055576324463,
      "learning_rate": 1.2355786538549653e-05,
      "loss": 0.0659,
      "step": 150800
    },
    {
      "epoch": 2.221248582495103,
      "grad_norm": 1.707761287689209,
      "learning_rate": 1.2354150166582667e-05,
      "loss": 0.0656,
      "step": 150825
    },
    {
      "epoch": 2.2216167655851904,
      "grad_norm": 1.082489252090454,
      "learning_rate": 1.2352513794615683e-05,
      "loss": 0.0508,
      "step": 150850
    },
    {
      "epoch": 2.2219849486752774,
      "grad_norm": 1.2071428298950195,
      "learning_rate": 1.2350877422648698e-05,
      "loss": 0.062,
      "step": 150875
    },
    {
      "epoch": 2.2223531317653644,
      "grad_norm": 1.1847529411315918,
      "learning_rate": 1.2349241050681712e-05,
      "loss": 0.0659,
      "step": 150900
    },
    {
      "epoch": 2.2227213148554514,
      "grad_norm": 1.320918083190918,
      "learning_rate": 1.2347604678714728e-05,
      "loss": 0.0652,
      "step": 150925
    },
    {
      "epoch": 2.2230894979455385,
      "grad_norm": 1.723616361618042,
      "learning_rate": 1.2345968306747745e-05,
      "loss": 0.0621,
      "step": 150950
    },
    {
      "epoch": 2.2234576810356255,
      "grad_norm": 1.40606689453125,
      "learning_rate": 1.234433193478076e-05,
      "loss": 0.0639,
      "step": 150975
    },
    {
      "epoch": 2.2238258641257125,
      "grad_norm": 1.8838179111480713,
      "learning_rate": 1.2342695562813775e-05,
      "loss": 0.0627,
      "step": 151000
    },
    {
      "epoch": 2.2241940472157995,
      "grad_norm": 1.4595658779144287,
      "learning_rate": 1.2341059190846791e-05,
      "loss": 0.0657,
      "step": 151025
    },
    {
      "epoch": 2.2245622303058865,
      "grad_norm": 1.5074530839920044,
      "learning_rate": 1.2339422818879808e-05,
      "loss": 0.064,
      "step": 151050
    },
    {
      "epoch": 2.2249304133959735,
      "grad_norm": 1.2347718477249146,
      "learning_rate": 1.233778644691282e-05,
      "loss": 0.0615,
      "step": 151075
    },
    {
      "epoch": 2.2252985964860605,
      "grad_norm": 1.1741864681243896,
      "learning_rate": 1.2336150074945836e-05,
      "loss": 0.0666,
      "step": 151100
    },
    {
      "epoch": 2.2256667795761476,
      "grad_norm": 1.230065941810608,
      "learning_rate": 1.2334513702978853e-05,
      "loss": 0.0655,
      "step": 151125
    },
    {
      "epoch": 2.2260349626662346,
      "grad_norm": 1.2888094186782837,
      "learning_rate": 1.2332877331011867e-05,
      "loss": 0.0628,
      "step": 151150
    },
    {
      "epoch": 2.2264031457563216,
      "grad_norm": 1.5288974046707153,
      "learning_rate": 1.2331240959044883e-05,
      "loss": 0.0627,
      "step": 151175
    },
    {
      "epoch": 2.2267713288464086,
      "grad_norm": 1.4964913129806519,
      "learning_rate": 1.23296045870779e-05,
      "loss": 0.0656,
      "step": 151200
    },
    {
      "epoch": 2.2271395119364956,
      "grad_norm": 1.3611366748809814,
      "learning_rate": 1.2327968215110916e-05,
      "loss": 0.0623,
      "step": 151225
    },
    {
      "epoch": 2.2275076950265826,
      "grad_norm": 1.4730292558670044,
      "learning_rate": 1.232633184314393e-05,
      "loss": 0.0574,
      "step": 151250
    },
    {
      "epoch": 2.2278758781166697,
      "grad_norm": 1.308506727218628,
      "learning_rate": 1.2324695471176946e-05,
      "loss": 0.057,
      "step": 151275
    },
    {
      "epoch": 2.228244061206757,
      "grad_norm": 1.8183484077453613,
      "learning_rate": 1.2323059099209959e-05,
      "loss": 0.0613,
      "step": 151300
    },
    {
      "epoch": 2.228612244296844,
      "grad_norm": 0.9702703356742859,
      "learning_rate": 1.2321422727242975e-05,
      "loss": 0.0628,
      "step": 151325
    },
    {
      "epoch": 2.228980427386931,
      "grad_norm": 1.2962535619735718,
      "learning_rate": 1.2319786355275991e-05,
      "loss": 0.0573,
      "step": 151350
    },
    {
      "epoch": 2.229348610477018,
      "grad_norm": 0.9257969856262207,
      "learning_rate": 1.2318149983309007e-05,
      "loss": 0.0642,
      "step": 151375
    },
    {
      "epoch": 2.229716793567105,
      "grad_norm": 1.5734418630599976,
      "learning_rate": 1.2316513611342022e-05,
      "loss": 0.0607,
      "step": 151400
    },
    {
      "epoch": 2.230084976657192,
      "grad_norm": 1.3924126625061035,
      "learning_rate": 1.2314877239375038e-05,
      "loss": 0.0624,
      "step": 151425
    },
    {
      "epoch": 2.230453159747279,
      "grad_norm": 1.410024642944336,
      "learning_rate": 1.2313240867408054e-05,
      "loss": 0.065,
      "step": 151450
    },
    {
      "epoch": 2.230821342837366,
      "grad_norm": 1.294050931930542,
      "learning_rate": 1.231160449544107e-05,
      "loss": 0.0573,
      "step": 151475
    },
    {
      "epoch": 2.2311895259274532,
      "grad_norm": 0.7892852425575256,
      "learning_rate": 1.2309968123474083e-05,
      "loss": 0.0581,
      "step": 151500
    },
    {
      "epoch": 2.2315577090175402,
      "grad_norm": 1.2217223644256592,
      "learning_rate": 1.23083317515071e-05,
      "loss": 0.0573,
      "step": 151525
    },
    {
      "epoch": 2.2319258921076273,
      "grad_norm": 1.3074737787246704,
      "learning_rate": 1.2306695379540115e-05,
      "loss": 0.0601,
      "step": 151550
    },
    {
      "epoch": 2.2322940751977143,
      "grad_norm": 1.367594599723816,
      "learning_rate": 1.230505900757313e-05,
      "loss": 0.062,
      "step": 151575
    },
    {
      "epoch": 2.2326622582878013,
      "grad_norm": 0.9579832553863525,
      "learning_rate": 1.2303422635606146e-05,
      "loss": 0.0584,
      "step": 151600
    },
    {
      "epoch": 2.2330304413778883,
      "grad_norm": 1.3861565589904785,
      "learning_rate": 1.2301786263639162e-05,
      "loss": 0.0562,
      "step": 151625
    },
    {
      "epoch": 2.2333986244679753,
      "grad_norm": 1.655278205871582,
      "learning_rate": 1.2300149891672177e-05,
      "loss": 0.0644,
      "step": 151650
    },
    {
      "epoch": 2.2337668075580623,
      "grad_norm": 1.6077373027801514,
      "learning_rate": 1.2298513519705193e-05,
      "loss": 0.0639,
      "step": 151675
    },
    {
      "epoch": 2.2341349906481494,
      "grad_norm": 1.3243197202682495,
      "learning_rate": 1.2296877147738209e-05,
      "loss": 0.0613,
      "step": 151700
    },
    {
      "epoch": 2.2345031737382364,
      "grad_norm": 1.8628376722335815,
      "learning_rate": 1.2295240775771222e-05,
      "loss": 0.0715,
      "step": 151725
    },
    {
      "epoch": 2.2348713568283234,
      "grad_norm": 1.076183557510376,
      "learning_rate": 1.2293604403804238e-05,
      "loss": 0.0642,
      "step": 151750
    },
    {
      "epoch": 2.2352395399184104,
      "grad_norm": 1.0358619689941406,
      "learning_rate": 1.2291968031837254e-05,
      "loss": 0.0638,
      "step": 151775
    },
    {
      "epoch": 2.235607723008498,
      "grad_norm": 1.3663173913955688,
      "learning_rate": 1.229033165987027e-05,
      "loss": 0.0629,
      "step": 151800
    },
    {
      "epoch": 2.235975906098585,
      "grad_norm": 1.3101141452789307,
      "learning_rate": 1.2288695287903285e-05,
      "loss": 0.0538,
      "step": 151825
    },
    {
      "epoch": 2.236344089188672,
      "grad_norm": 1.257340908050537,
      "learning_rate": 1.22870589159363e-05,
      "loss": 0.0585,
      "step": 151850
    },
    {
      "epoch": 2.236712272278759,
      "grad_norm": 1.3124793767929077,
      "learning_rate": 1.2285422543969317e-05,
      "loss": 0.057,
      "step": 151875
    },
    {
      "epoch": 2.237080455368846,
      "grad_norm": 1.228428840637207,
      "learning_rate": 1.2283786172002331e-05,
      "loss": 0.0623,
      "step": 151900
    },
    {
      "epoch": 2.237448638458933,
      "grad_norm": 1.3018091917037964,
      "learning_rate": 1.2282149800035346e-05,
      "loss": 0.0522,
      "step": 151925
    },
    {
      "epoch": 2.23781682154902,
      "grad_norm": 0.9292869567871094,
      "learning_rate": 1.2280513428068362e-05,
      "loss": 0.0632,
      "step": 151950
    },
    {
      "epoch": 2.238185004639107,
      "grad_norm": 1.6583223342895508,
      "learning_rate": 1.2278877056101376e-05,
      "loss": 0.061,
      "step": 151975
    },
    {
      "epoch": 2.238553187729194,
      "grad_norm": 1.4275884628295898,
      "learning_rate": 1.2277240684134393e-05,
      "loss": 0.0609,
      "step": 152000
    },
    {
      "epoch": 2.238921370819281,
      "grad_norm": 1.466074824333191,
      "learning_rate": 1.2275604312167409e-05,
      "loss": 0.0615,
      "step": 152025
    },
    {
      "epoch": 2.239289553909368,
      "grad_norm": 1.985863447189331,
      "learning_rate": 1.2273967940200425e-05,
      "loss": 0.0631,
      "step": 152050
    },
    {
      "epoch": 2.239657736999455,
      "grad_norm": 2.0387206077575684,
      "learning_rate": 1.227233156823344e-05,
      "loss": 0.0668,
      "step": 152075
    },
    {
      "epoch": 2.240025920089542,
      "grad_norm": 1.4524468183517456,
      "learning_rate": 1.2270695196266455e-05,
      "loss": 0.0593,
      "step": 152100
    },
    {
      "epoch": 2.240394103179629,
      "grad_norm": 1.2152190208435059,
      "learning_rate": 1.2269058824299472e-05,
      "loss": 0.0629,
      "step": 152125
    },
    {
      "epoch": 2.240762286269716,
      "grad_norm": 1.3854385614395142,
      "learning_rate": 1.2267422452332484e-05,
      "loss": 0.065,
      "step": 152150
    },
    {
      "epoch": 2.241130469359803,
      "grad_norm": 1.3988215923309326,
      "learning_rate": 1.22657860803655e-05,
      "loss": 0.0657,
      "step": 152175
    },
    {
      "epoch": 2.24149865244989,
      "grad_norm": 1.4388787746429443,
      "learning_rate": 1.2264149708398517e-05,
      "loss": 0.0628,
      "step": 152200
    },
    {
      "epoch": 2.241866835539977,
      "grad_norm": 1.0519180297851562,
      "learning_rate": 1.2262513336431531e-05,
      "loss": 0.0517,
      "step": 152225
    },
    {
      "epoch": 2.2422350186300646,
      "grad_norm": 1.5220091342926025,
      "learning_rate": 1.2260876964464547e-05,
      "loss": 0.0617,
      "step": 152250
    },
    {
      "epoch": 2.2426032017201516,
      "grad_norm": 1.2312278747558594,
      "learning_rate": 1.2259240592497563e-05,
      "loss": 0.0621,
      "step": 152275
    },
    {
      "epoch": 2.2429713848102386,
      "grad_norm": 1.4729104042053223,
      "learning_rate": 1.225760422053058e-05,
      "loss": 0.058,
      "step": 152300
    },
    {
      "epoch": 2.2433395679003256,
      "grad_norm": 1.2740864753723145,
      "learning_rate": 1.2255967848563594e-05,
      "loss": 0.0637,
      "step": 152325
    },
    {
      "epoch": 2.2437077509904126,
      "grad_norm": 1.3165504932403564,
      "learning_rate": 1.2254331476596608e-05,
      "loss": 0.061,
      "step": 152350
    },
    {
      "epoch": 2.2440759340804997,
      "grad_norm": 0.680292010307312,
      "learning_rate": 1.2252695104629625e-05,
      "loss": 0.0606,
      "step": 152375
    },
    {
      "epoch": 2.2444441171705867,
      "grad_norm": 1.5281325578689575,
      "learning_rate": 1.2251058732662639e-05,
      "loss": 0.0655,
      "step": 152400
    },
    {
      "epoch": 2.2448123002606737,
      "grad_norm": 0.857114851474762,
      "learning_rate": 1.2249422360695655e-05,
      "loss": 0.0532,
      "step": 152425
    },
    {
      "epoch": 2.2451804833507607,
      "grad_norm": 1.3292726278305054,
      "learning_rate": 1.224785144360735e-05,
      "loss": 0.0672,
      "step": 152450
    },
    {
      "epoch": 2.2455486664408477,
      "grad_norm": 1.6671534776687622,
      "learning_rate": 1.2246215071640365e-05,
      "loss": 0.0605,
      "step": 152475
    },
    {
      "epoch": 2.2459168495309347,
      "grad_norm": 1.8260165452957153,
      "learning_rate": 1.2244578699673381e-05,
      "loss": 0.0674,
      "step": 152500
    },
    {
      "epoch": 2.2462850326210217,
      "grad_norm": 1.723912000656128,
      "learning_rate": 1.2242942327706398e-05,
      "loss": 0.0534,
      "step": 152525
    },
    {
      "epoch": 2.2466532157111088,
      "grad_norm": 1.1101588010787964,
      "learning_rate": 1.224130595573941e-05,
      "loss": 0.0578,
      "step": 152550
    },
    {
      "epoch": 2.247021398801196,
      "grad_norm": 1.4244182109832764,
      "learning_rate": 1.2239669583772426e-05,
      "loss": 0.0595,
      "step": 152575
    },
    {
      "epoch": 2.247389581891283,
      "grad_norm": 1.4533981084823608,
      "learning_rate": 1.2238033211805443e-05,
      "loss": 0.0614,
      "step": 152600
    },
    {
      "epoch": 2.24775776498137,
      "grad_norm": 1.7205830812454224,
      "learning_rate": 1.2236396839838459e-05,
      "loss": 0.0659,
      "step": 152625
    },
    {
      "epoch": 2.248125948071457,
      "grad_norm": 1.5929219722747803,
      "learning_rate": 1.2234760467871473e-05,
      "loss": 0.0622,
      "step": 152650
    },
    {
      "epoch": 2.248494131161544,
      "grad_norm": 1.1005336046218872,
      "learning_rate": 1.223312409590449e-05,
      "loss": 0.0578,
      "step": 152675
    },
    {
      "epoch": 2.248862314251631,
      "grad_norm": 1.2769103050231934,
      "learning_rate": 1.2231487723937506e-05,
      "loss": 0.0615,
      "step": 152700
    },
    {
      "epoch": 2.249230497341718,
      "grad_norm": 1.293278694152832,
      "learning_rate": 1.2229851351970522e-05,
      "loss": 0.061,
      "step": 152725
    },
    {
      "epoch": 2.2495986804318053,
      "grad_norm": 1.5609077215194702,
      "learning_rate": 1.2228214980003536e-05,
      "loss": 0.0753,
      "step": 152750
    },
    {
      "epoch": 2.2499668635218923,
      "grad_norm": 1.4379901885986328,
      "learning_rate": 1.222657860803655e-05,
      "loss": 0.0593,
      "step": 152775
    },
    {
      "epoch": 2.2503350466119794,
      "grad_norm": 1.272094488143921,
      "learning_rate": 1.2224942236069565e-05,
      "loss": 0.0686,
      "step": 152800
    },
    {
      "epoch": 2.2507032297020664,
      "grad_norm": 1.719151258468628,
      "learning_rate": 1.2223305864102581e-05,
      "loss": 0.0607,
      "step": 152825
    },
    {
      "epoch": 2.2510714127921534,
      "grad_norm": 1.6205966472625732,
      "learning_rate": 1.2221669492135597e-05,
      "loss": 0.0575,
      "step": 152850
    },
    {
      "epoch": 2.2514395958822404,
      "grad_norm": 1.1187539100646973,
      "learning_rate": 1.2220033120168614e-05,
      "loss": 0.0602,
      "step": 152875
    },
    {
      "epoch": 2.2518077789723274,
      "grad_norm": 1.6845979690551758,
      "learning_rate": 1.2218396748201628e-05,
      "loss": 0.0622,
      "step": 152900
    },
    {
      "epoch": 2.2521759620624144,
      "grad_norm": 1.2623274326324463,
      "learning_rate": 1.2216760376234644e-05,
      "loss": 0.0594,
      "step": 152925
    },
    {
      "epoch": 2.2525441451525015,
      "grad_norm": 1.4932770729064941,
      "learning_rate": 1.221512400426766e-05,
      "loss": 0.0556,
      "step": 152950
    },
    {
      "epoch": 2.2529123282425885,
      "grad_norm": 1.5640065670013428,
      "learning_rate": 1.2213487632300673e-05,
      "loss": 0.0653,
      "step": 152975
    },
    {
      "epoch": 2.2532805113326755,
      "grad_norm": 1.2047258615493774,
      "learning_rate": 1.221185126033369e-05,
      "loss": 0.0613,
      "step": 153000
    },
    {
      "epoch": 2.2536486944227625,
      "grad_norm": 1.4226864576339722,
      "learning_rate": 1.2210214888366705e-05,
      "loss": 0.0624,
      "step": 153025
    },
    {
      "epoch": 2.2540168775128495,
      "grad_norm": 1.3719414472579956,
      "learning_rate": 1.220857851639972e-05,
      "loss": 0.0673,
      "step": 153050
    },
    {
      "epoch": 2.2543850606029365,
      "grad_norm": 1.4957154989242554,
      "learning_rate": 1.2206942144432736e-05,
      "loss": 0.0679,
      "step": 153075
    },
    {
      "epoch": 2.2547532436930235,
      "grad_norm": 1.5672136545181274,
      "learning_rate": 1.2205305772465752e-05,
      "loss": 0.0661,
      "step": 153100
    },
    {
      "epoch": 2.2551214267831106,
      "grad_norm": 1.0923973321914673,
      "learning_rate": 1.2203669400498768e-05,
      "loss": 0.0665,
      "step": 153125
    },
    {
      "epoch": 2.2554896098731976,
      "grad_norm": 1.3810386657714844,
      "learning_rate": 1.2202033028531783e-05,
      "loss": 0.0647,
      "step": 153150
    },
    {
      "epoch": 2.2558577929632846,
      "grad_norm": 1.309795618057251,
      "learning_rate": 1.2200396656564799e-05,
      "loss": 0.0588,
      "step": 153175
    },
    {
      "epoch": 2.256225976053372,
      "grad_norm": 1.5182901620864868,
      "learning_rate": 1.2198760284597813e-05,
      "loss": 0.0611,
      "step": 153200
    },
    {
      "epoch": 2.256594159143459,
      "grad_norm": 1.2813936471939087,
      "learning_rate": 1.2197123912630828e-05,
      "loss": 0.0618,
      "step": 153225
    },
    {
      "epoch": 2.256962342233546,
      "grad_norm": 1.1370495557785034,
      "learning_rate": 1.2195487540663844e-05,
      "loss": 0.0632,
      "step": 153250
    },
    {
      "epoch": 2.257330525323633,
      "grad_norm": 1.2709547281265259,
      "learning_rate": 1.219385116869686e-05,
      "loss": 0.064,
      "step": 153275
    },
    {
      "epoch": 2.25769870841372,
      "grad_norm": 1.9685618877410889,
      "learning_rate": 1.2192214796729876e-05,
      "loss": 0.0681,
      "step": 153300
    },
    {
      "epoch": 2.258066891503807,
      "grad_norm": 1.3484185934066772,
      "learning_rate": 1.219057842476289e-05,
      "loss": 0.058,
      "step": 153325
    },
    {
      "epoch": 2.258435074593894,
      "grad_norm": 1.2409491539001465,
      "learning_rate": 1.2188942052795907e-05,
      "loss": 0.0629,
      "step": 153350
    },
    {
      "epoch": 2.258803257683981,
      "grad_norm": 1.8287829160690308,
      "learning_rate": 1.2187305680828923e-05,
      "loss": 0.0575,
      "step": 153375
    },
    {
      "epoch": 2.259171440774068,
      "grad_norm": 1.5454610586166382,
      "learning_rate": 1.2185669308861936e-05,
      "loss": 0.0636,
      "step": 153400
    },
    {
      "epoch": 2.259539623864155,
      "grad_norm": 1.5059220790863037,
      "learning_rate": 1.2184032936894952e-05,
      "loss": 0.0634,
      "step": 153425
    },
    {
      "epoch": 2.259907806954242,
      "grad_norm": 1.4627835750579834,
      "learning_rate": 1.2182396564927968e-05,
      "loss": 0.0597,
      "step": 153450
    },
    {
      "epoch": 2.260275990044329,
      "grad_norm": 1.3997286558151245,
      "learning_rate": 1.2180760192960983e-05,
      "loss": 0.0601,
      "step": 153475
    },
    {
      "epoch": 2.2606441731344162,
      "grad_norm": 1.4016356468200684,
      "learning_rate": 1.2179123820993999e-05,
      "loss": 0.0603,
      "step": 153500
    },
    {
      "epoch": 2.2610123562245033,
      "grad_norm": 1.1766282320022583,
      "learning_rate": 1.2177487449027015e-05,
      "loss": 0.0594,
      "step": 153525
    },
    {
      "epoch": 2.2613805393145903,
      "grad_norm": 1.4122215509414673,
      "learning_rate": 1.2175851077060031e-05,
      "loss": 0.0626,
      "step": 153550
    },
    {
      "epoch": 2.2617487224046773,
      "grad_norm": 1.4782017469406128,
      "learning_rate": 1.2174214705093045e-05,
      "loss": 0.0567,
      "step": 153575
    },
    {
      "epoch": 2.2621169054947643,
      "grad_norm": 1.488305687904358,
      "learning_rate": 1.2172578333126062e-05,
      "loss": 0.0611,
      "step": 153600
    },
    {
      "epoch": 2.2624850885848513,
      "grad_norm": 0.9970173239707947,
      "learning_rate": 1.2170941961159074e-05,
      "loss": 0.064,
      "step": 153625
    },
    {
      "epoch": 2.2628532716749383,
      "grad_norm": 0.7006500363349915,
      "learning_rate": 1.216930558919209e-05,
      "loss": 0.062,
      "step": 153650
    },
    {
      "epoch": 2.2632214547650253,
      "grad_norm": 1.6163110733032227,
      "learning_rate": 1.2167669217225107e-05,
      "loss": 0.062,
      "step": 153675
    },
    {
      "epoch": 2.2635896378551124,
      "grad_norm": 1.6576616764068604,
      "learning_rate": 1.2166032845258123e-05,
      "loss": 0.0632,
      "step": 153700
    },
    {
      "epoch": 2.2639578209452,
      "grad_norm": 1.7387374639511108,
      "learning_rate": 1.2164396473291137e-05,
      "loss": 0.065,
      "step": 153725
    },
    {
      "epoch": 2.264326004035287,
      "grad_norm": 1.4592851400375366,
      "learning_rate": 1.2162760101324153e-05,
      "loss": 0.0617,
      "step": 153750
    },
    {
      "epoch": 2.264694187125374,
      "grad_norm": 1.2785307168960571,
      "learning_rate": 1.216112372935717e-05,
      "loss": 0.0675,
      "step": 153775
    },
    {
      "epoch": 2.265062370215461,
      "grad_norm": 1.5926411151885986,
      "learning_rate": 1.2159487357390186e-05,
      "loss": 0.0581,
      "step": 153800
    },
    {
      "epoch": 2.265430553305548,
      "grad_norm": 1.4977892637252808,
      "learning_rate": 1.2157850985423198e-05,
      "loss": 0.0556,
      "step": 153825
    },
    {
      "epoch": 2.265798736395635,
      "grad_norm": 2.1493022441864014,
      "learning_rate": 1.2156214613456215e-05,
      "loss": 0.065,
      "step": 153850
    },
    {
      "epoch": 2.266166919485722,
      "grad_norm": 1.4621464014053345,
      "learning_rate": 1.2154578241489229e-05,
      "loss": 0.0607,
      "step": 153875
    },
    {
      "epoch": 2.266535102575809,
      "grad_norm": 1.6883494853973389,
      "learning_rate": 1.2152941869522245e-05,
      "loss": 0.0713,
      "step": 153900
    },
    {
      "epoch": 2.266903285665896,
      "grad_norm": 1.9697834253311157,
      "learning_rate": 1.2151305497555261e-05,
      "loss": 0.0613,
      "step": 153925
    },
    {
      "epoch": 2.267271468755983,
      "grad_norm": 1.4346647262573242,
      "learning_rate": 1.2149669125588278e-05,
      "loss": 0.0599,
      "step": 153950
    },
    {
      "epoch": 2.26763965184607,
      "grad_norm": 1.4374207258224487,
      "learning_rate": 1.2148032753621292e-05,
      "loss": 0.0609,
      "step": 153975
    },
    {
      "epoch": 2.268007834936157,
      "grad_norm": 1.854790210723877,
      "learning_rate": 1.2146396381654308e-05,
      "loss": 0.0643,
      "step": 154000
    },
    {
      "epoch": 2.268376018026244,
      "grad_norm": 1.3829092979431152,
      "learning_rate": 1.2144760009687323e-05,
      "loss": 0.0579,
      "step": 154025
    },
    {
      "epoch": 2.268744201116331,
      "grad_norm": 1.3426350355148315,
      "learning_rate": 1.2143123637720337e-05,
      "loss": 0.0678,
      "step": 154050
    },
    {
      "epoch": 2.269112384206418,
      "grad_norm": 1.077064871788025,
      "learning_rate": 1.2141487265753353e-05,
      "loss": 0.0552,
      "step": 154075
    },
    {
      "epoch": 2.269480567296505,
      "grad_norm": 1.560431957244873,
      "learning_rate": 1.213985089378637e-05,
      "loss": 0.0575,
      "step": 154100
    },
    {
      "epoch": 2.269848750386592,
      "grad_norm": 1.2780370712280273,
      "learning_rate": 1.2138214521819386e-05,
      "loss": 0.0619,
      "step": 154125
    },
    {
      "epoch": 2.2702169334766795,
      "grad_norm": 1.0501247644424438,
      "learning_rate": 1.21365781498524e-05,
      "loss": 0.0562,
      "step": 154150
    },
    {
      "epoch": 2.2705851165667665,
      "grad_norm": 1.45809006690979,
      "learning_rate": 1.2134941777885416e-05,
      "loss": 0.062,
      "step": 154175
    },
    {
      "epoch": 2.2709532996568536,
      "grad_norm": 1.7347477674484253,
      "learning_rate": 1.2133305405918432e-05,
      "loss": 0.0632,
      "step": 154200
    },
    {
      "epoch": 2.2713214827469406,
      "grad_norm": 1.1819517612457275,
      "learning_rate": 1.2131669033951447e-05,
      "loss": 0.0628,
      "step": 154225
    },
    {
      "epoch": 2.2716896658370276,
      "grad_norm": 1.4354586601257324,
      "learning_rate": 1.2130032661984461e-05,
      "loss": 0.0612,
      "step": 154250
    },
    {
      "epoch": 2.2720578489271146,
      "grad_norm": 1.2080940008163452,
      "learning_rate": 1.2128396290017477e-05,
      "loss": 0.0647,
      "step": 154275
    },
    {
      "epoch": 2.2724260320172016,
      "grad_norm": 1.4879984855651855,
      "learning_rate": 1.2126759918050492e-05,
      "loss": 0.0531,
      "step": 154300
    },
    {
      "epoch": 2.2727942151072886,
      "grad_norm": 1.7485198974609375,
      "learning_rate": 1.2125123546083508e-05,
      "loss": 0.0605,
      "step": 154325
    },
    {
      "epoch": 2.2731623981973756,
      "grad_norm": 1.7189886569976807,
      "learning_rate": 1.2123487174116524e-05,
      "loss": 0.0605,
      "step": 154350
    },
    {
      "epoch": 2.2735305812874627,
      "grad_norm": 1.657499074935913,
      "learning_rate": 1.212185080214954e-05,
      "loss": 0.0596,
      "step": 154375
    },
    {
      "epoch": 2.2738987643775497,
      "grad_norm": 1.442579984664917,
      "learning_rate": 1.2120214430182555e-05,
      "loss": 0.0698,
      "step": 154400
    },
    {
      "epoch": 2.2742669474676367,
      "grad_norm": 1.1019314527511597,
      "learning_rate": 1.2118578058215571e-05,
      "loss": 0.0531,
      "step": 154425
    },
    {
      "epoch": 2.2746351305577237,
      "grad_norm": 1.1446802616119385,
      "learning_rate": 1.2116941686248584e-05,
      "loss": 0.0541,
      "step": 154450
    },
    {
      "epoch": 2.2750033136478107,
      "grad_norm": 1.2468680143356323,
      "learning_rate": 1.21153053142816e-05,
      "loss": 0.0628,
      "step": 154475
    },
    {
      "epoch": 2.2753714967378977,
      "grad_norm": 1.595442771911621,
      "learning_rate": 1.2113668942314616e-05,
      "loss": 0.0626,
      "step": 154500
    },
    {
      "epoch": 2.2757396798279848,
      "grad_norm": 1.4726616144180298,
      "learning_rate": 1.2112032570347632e-05,
      "loss": 0.0652,
      "step": 154525
    },
    {
      "epoch": 2.2761078629180718,
      "grad_norm": 1.5682780742645264,
      "learning_rate": 1.2110396198380647e-05,
      "loss": 0.055,
      "step": 154550
    },
    {
      "epoch": 2.276476046008159,
      "grad_norm": 1.6534937620162964,
      "learning_rate": 1.2108759826413663e-05,
      "loss": 0.0743,
      "step": 154575
    },
    {
      "epoch": 2.276844229098246,
      "grad_norm": 1.19683039188385,
      "learning_rate": 1.2107123454446679e-05,
      "loss": 0.0602,
      "step": 154600
    },
    {
      "epoch": 2.277212412188333,
      "grad_norm": 1.3283883333206177,
      "learning_rate": 1.2105487082479695e-05,
      "loss": 0.062,
      "step": 154625
    },
    {
      "epoch": 2.27758059527842,
      "grad_norm": 1.3543940782546997,
      "learning_rate": 1.2103916165391389e-05,
      "loss": 0.0675,
      "step": 154650
    },
    {
      "epoch": 2.277948778368507,
      "grad_norm": 1.4500526189804077,
      "learning_rate": 1.2102279793424403e-05,
      "loss": 0.0584,
      "step": 154675
    },
    {
      "epoch": 2.2783169614585943,
      "grad_norm": 1.8910281658172607,
      "learning_rate": 1.210064342145742e-05,
      "loss": 0.0647,
      "step": 154700
    },
    {
      "epoch": 2.2786851445486813,
      "grad_norm": 1.2388887405395508,
      "learning_rate": 1.2099007049490434e-05,
      "loss": 0.0639,
      "step": 154725
    },
    {
      "epoch": 2.2790533276387683,
      "grad_norm": 1.852287769317627,
      "learning_rate": 1.209737067752345e-05,
      "loss": 0.0618,
      "step": 154750
    },
    {
      "epoch": 2.2794215107288553,
      "grad_norm": 1.2877410650253296,
      "learning_rate": 1.2095734305556466e-05,
      "loss": 0.0622,
      "step": 154775
    },
    {
      "epoch": 2.2797896938189424,
      "grad_norm": 1.0189391374588013,
      "learning_rate": 1.209409793358948e-05,
      "loss": 0.0559,
      "step": 154800
    },
    {
      "epoch": 2.2801578769090294,
      "grad_norm": 1.432690143585205,
      "learning_rate": 1.2092461561622497e-05,
      "loss": 0.055,
      "step": 154825
    },
    {
      "epoch": 2.2805260599991164,
      "grad_norm": 1.3184486627578735,
      "learning_rate": 1.2090825189655513e-05,
      "loss": 0.0616,
      "step": 154850
    },
    {
      "epoch": 2.2808942430892034,
      "grad_norm": 1.423694372177124,
      "learning_rate": 1.2089188817688526e-05,
      "loss": 0.0651,
      "step": 154875
    },
    {
      "epoch": 2.2812624261792904,
      "grad_norm": 1.3236284255981445,
      "learning_rate": 1.2087552445721542e-05,
      "loss": 0.0569,
      "step": 154900
    },
    {
      "epoch": 2.2816306092693774,
      "grad_norm": 1.3500868082046509,
      "learning_rate": 1.2085916073754558e-05,
      "loss": 0.0577,
      "step": 154925
    },
    {
      "epoch": 2.2819987923594645,
      "grad_norm": 1.4023374319076538,
      "learning_rate": 1.2084279701787574e-05,
      "loss": 0.0606,
      "step": 154950
    },
    {
      "epoch": 2.2823669754495515,
      "grad_norm": 1.2842373847961426,
      "learning_rate": 1.2082643329820589e-05,
      "loss": 0.0618,
      "step": 154975
    },
    {
      "epoch": 2.2827351585396385,
      "grad_norm": 1.4571912288665771,
      "learning_rate": 1.2081006957853605e-05,
      "loss": 0.0689,
      "step": 155000
    },
    {
      "epoch": 2.2831033416297255,
      "grad_norm": 0.9824802875518799,
      "learning_rate": 1.2079370585886621e-05,
      "loss": 0.0605,
      "step": 155025
    },
    {
      "epoch": 2.2834715247198125,
      "grad_norm": 1.2202214002609253,
      "learning_rate": 1.2077734213919637e-05,
      "loss": 0.0625,
      "step": 155050
    },
    {
      "epoch": 2.2838397078098995,
      "grad_norm": 1.3202885389328003,
      "learning_rate": 1.2076097841952652e-05,
      "loss": 0.0613,
      "step": 155075
    },
    {
      "epoch": 2.284207890899987,
      "grad_norm": 1.7728911638259888,
      "learning_rate": 1.2074461469985666e-05,
      "loss": 0.0651,
      "step": 155100
    },
    {
      "epoch": 2.284576073990074,
      "grad_norm": 1.0166760683059692,
      "learning_rate": 1.207282509801868e-05,
      "loss": 0.0599,
      "step": 155125
    },
    {
      "epoch": 2.284944257080161,
      "grad_norm": 1.1950476169586182,
      "learning_rate": 1.2071188726051697e-05,
      "loss": 0.0578,
      "step": 155150
    },
    {
      "epoch": 2.285312440170248,
      "grad_norm": 1.2729220390319824,
      "learning_rate": 1.2069552354084713e-05,
      "loss": 0.0606,
      "step": 155175
    },
    {
      "epoch": 2.285680623260335,
      "grad_norm": 0.963126540184021,
      "learning_rate": 1.2067915982117729e-05,
      "loss": 0.0615,
      "step": 155200
    },
    {
      "epoch": 2.286048806350422,
      "grad_norm": 1.4809589385986328,
      "learning_rate": 1.2066279610150743e-05,
      "loss": 0.0575,
      "step": 155225
    },
    {
      "epoch": 2.286416989440509,
      "grad_norm": 1.026487946510315,
      "learning_rate": 1.206464323818376e-05,
      "loss": 0.0563,
      "step": 155250
    },
    {
      "epoch": 2.286785172530596,
      "grad_norm": 1.2711668014526367,
      "learning_rate": 1.2063006866216776e-05,
      "loss": 0.0674,
      "step": 155275
    },
    {
      "epoch": 2.287153355620683,
      "grad_norm": 1.0773683786392212,
      "learning_rate": 1.2061370494249788e-05,
      "loss": 0.053,
      "step": 155300
    },
    {
      "epoch": 2.28752153871077,
      "grad_norm": 1.8069260120391846,
      "learning_rate": 1.2059734122282805e-05,
      "loss": 0.0618,
      "step": 155325
    },
    {
      "epoch": 2.287889721800857,
      "grad_norm": 1.054559588432312,
      "learning_rate": 1.205809775031582e-05,
      "loss": 0.0615,
      "step": 155350
    },
    {
      "epoch": 2.288257904890944,
      "grad_norm": 1.3881711959838867,
      "learning_rate": 1.2056461378348835e-05,
      "loss": 0.0621,
      "step": 155375
    },
    {
      "epoch": 2.288626087981031,
      "grad_norm": 1.3065428733825684,
      "learning_rate": 1.2054825006381851e-05,
      "loss": 0.0596,
      "step": 155400
    },
    {
      "epoch": 2.288994271071118,
      "grad_norm": 1.3255311250686646,
      "learning_rate": 1.2053188634414868e-05,
      "loss": 0.0581,
      "step": 155425
    },
    {
      "epoch": 2.289362454161205,
      "grad_norm": 1.5815212726593018,
      "learning_rate": 1.2051552262447884e-05,
      "loss": 0.0615,
      "step": 155450
    },
    {
      "epoch": 2.2897306372512922,
      "grad_norm": 1.2780965566635132,
      "learning_rate": 1.2049915890480898e-05,
      "loss": 0.0615,
      "step": 155475
    },
    {
      "epoch": 2.2900988203413792,
      "grad_norm": 1.1252162456512451,
      "learning_rate": 1.2048279518513914e-05,
      "loss": 0.0592,
      "step": 155500
    },
    {
      "epoch": 2.2904670034314663,
      "grad_norm": 1.3305015563964844,
      "learning_rate": 1.2046643146546929e-05,
      "loss": 0.0592,
      "step": 155525
    },
    {
      "epoch": 2.2908351865215533,
      "grad_norm": 1.5482666492462158,
      "learning_rate": 1.2045006774579943e-05,
      "loss": 0.0645,
      "step": 155550
    },
    {
      "epoch": 2.2912033696116403,
      "grad_norm": 1.7357757091522217,
      "learning_rate": 1.204337040261296e-05,
      "loss": 0.0651,
      "step": 155575
    },
    {
      "epoch": 2.2915715527017273,
      "grad_norm": 1.1715476512908936,
      "learning_rate": 1.2041734030645976e-05,
      "loss": 0.0628,
      "step": 155600
    },
    {
      "epoch": 2.2919397357918143,
      "grad_norm": 1.4161438941955566,
      "learning_rate": 1.204009765867899e-05,
      "loss": 0.0606,
      "step": 155625
    },
    {
      "epoch": 2.2923079188819018,
      "grad_norm": 1.7011719942092896,
      "learning_rate": 1.2038461286712006e-05,
      "loss": 0.0632,
      "step": 155650
    },
    {
      "epoch": 2.292676101971989,
      "grad_norm": 1.2964603900909424,
      "learning_rate": 1.2036824914745022e-05,
      "loss": 0.0648,
      "step": 155675
    },
    {
      "epoch": 2.293044285062076,
      "grad_norm": 1.2411048412322998,
      "learning_rate": 1.2035188542778038e-05,
      "loss": 0.0674,
      "step": 155700
    },
    {
      "epoch": 2.293412468152163,
      "grad_norm": 1.055909514427185,
      "learning_rate": 1.2033552170811051e-05,
      "loss": 0.0589,
      "step": 155725
    },
    {
      "epoch": 2.29378065124225,
      "grad_norm": 1.439857006072998,
      "learning_rate": 1.2031915798844067e-05,
      "loss": 0.075,
      "step": 155750
    },
    {
      "epoch": 2.294148834332337,
      "grad_norm": 2.148745536804199,
      "learning_rate": 1.2030279426877084e-05,
      "loss": 0.0606,
      "step": 155775
    },
    {
      "epoch": 2.294517017422424,
      "grad_norm": 1.3419593572616577,
      "learning_rate": 1.2028643054910098e-05,
      "loss": 0.0685,
      "step": 155800
    },
    {
      "epoch": 2.294885200512511,
      "grad_norm": 1.2838846445083618,
      "learning_rate": 1.2027006682943114e-05,
      "loss": 0.0606,
      "step": 155825
    },
    {
      "epoch": 2.295253383602598,
      "grad_norm": 1.3703469038009644,
      "learning_rate": 1.202537031097613e-05,
      "loss": 0.0548,
      "step": 155850
    },
    {
      "epoch": 2.295621566692685,
      "grad_norm": 1.9246338605880737,
      "learning_rate": 1.2023733939009146e-05,
      "loss": 0.068,
      "step": 155875
    },
    {
      "epoch": 2.295989749782772,
      "grad_norm": 1.618470549583435,
      "learning_rate": 1.2022097567042161e-05,
      "loss": 0.0687,
      "step": 155900
    },
    {
      "epoch": 2.296357932872859,
      "grad_norm": 1.2991777658462524,
      "learning_rate": 1.2020461195075175e-05,
      "loss": 0.0656,
      "step": 155925
    },
    {
      "epoch": 2.296726115962946,
      "grad_norm": 2.071720600128174,
      "learning_rate": 1.201882482310819e-05,
      "loss": 0.0573,
      "step": 155950
    },
    {
      "epoch": 2.297094299053033,
      "grad_norm": 0.9859101176261902,
      "learning_rate": 1.2017188451141206e-05,
      "loss": 0.0568,
      "step": 155975
    },
    {
      "epoch": 2.29746248214312,
      "grad_norm": 1.6727458238601685,
      "learning_rate": 1.2015552079174222e-05,
      "loss": 0.0618,
      "step": 156000
    },
    {
      "epoch": 2.297830665233207,
      "grad_norm": 1.6102430820465088,
      "learning_rate": 1.2013915707207238e-05,
      "loss": 0.0624,
      "step": 156025
    },
    {
      "epoch": 2.298198848323294,
      "grad_norm": 1.8287855386734009,
      "learning_rate": 1.2012279335240253e-05,
      "loss": 0.0679,
      "step": 156050
    },
    {
      "epoch": 2.2985670314133815,
      "grad_norm": 1.5875651836395264,
      "learning_rate": 1.2010642963273269e-05,
      "loss": 0.0614,
      "step": 156075
    },
    {
      "epoch": 2.2989352145034685,
      "grad_norm": 1.2037783861160278,
      "learning_rate": 1.2009006591306285e-05,
      "loss": 0.0595,
      "step": 156100
    },
    {
      "epoch": 2.2993033975935555,
      "grad_norm": 1.473384976387024,
      "learning_rate": 1.2007370219339301e-05,
      "loss": 0.0602,
      "step": 156125
    },
    {
      "epoch": 2.2996715806836425,
      "grad_norm": 1.6098628044128418,
      "learning_rate": 1.2005733847372314e-05,
      "loss": 0.0604,
      "step": 156150
    },
    {
      "epoch": 2.3000397637737295,
      "grad_norm": 1.4224539995193481,
      "learning_rate": 1.200409747540533e-05,
      "loss": 0.059,
      "step": 156175
    },
    {
      "epoch": 2.3004079468638166,
      "grad_norm": 1.2859063148498535,
      "learning_rate": 1.2002461103438345e-05,
      "loss": 0.0588,
      "step": 156200
    },
    {
      "epoch": 2.3007761299539036,
      "grad_norm": 1.760091781616211,
      "learning_rate": 1.200082473147136e-05,
      "loss": 0.0661,
      "step": 156225
    },
    {
      "epoch": 2.3011443130439906,
      "grad_norm": 1.1600080728530884,
      "learning_rate": 1.1999188359504377e-05,
      "loss": 0.0615,
      "step": 156250
    },
    {
      "epoch": 2.3015124961340776,
      "grad_norm": 1.7958961725234985,
      "learning_rate": 1.1997551987537393e-05,
      "loss": 0.0643,
      "step": 156275
    },
    {
      "epoch": 2.3018806792241646,
      "grad_norm": 1.1491156816482544,
      "learning_rate": 1.1995915615570407e-05,
      "loss": 0.0601,
      "step": 156300
    },
    {
      "epoch": 2.3022488623142516,
      "grad_norm": 1.23212730884552,
      "learning_rate": 1.1994279243603424e-05,
      "loss": 0.0581,
      "step": 156325
    },
    {
      "epoch": 2.3026170454043386,
      "grad_norm": 1.3952423334121704,
      "learning_rate": 1.1992642871636438e-05,
      "loss": 0.0641,
      "step": 156350
    },
    {
      "epoch": 2.3029852284944257,
      "grad_norm": 1.6260968446731567,
      "learning_rate": 1.1991006499669453e-05,
      "loss": 0.0646,
      "step": 156375
    },
    {
      "epoch": 2.3033534115845127,
      "grad_norm": 0.9654199481010437,
      "learning_rate": 1.1989370127702469e-05,
      "loss": 0.0738,
      "step": 156400
    },
    {
      "epoch": 2.3037215946745997,
      "grad_norm": 0.9737640619277954,
      "learning_rate": 1.1987733755735485e-05,
      "loss": 0.0594,
      "step": 156425
    },
    {
      "epoch": 2.3040897777646867,
      "grad_norm": 1.3626937866210938,
      "learning_rate": 1.1986097383768501e-05,
      "loss": 0.0613,
      "step": 156450
    },
    {
      "epoch": 2.3044579608547737,
      "grad_norm": 0.9062378406524658,
      "learning_rate": 1.1984461011801515e-05,
      "loss": 0.0644,
      "step": 156475
    },
    {
      "epoch": 2.3048261439448607,
      "grad_norm": 0.6740528345108032,
      "learning_rate": 1.1982824639834532e-05,
      "loss": 0.0589,
      "step": 156500
    },
    {
      "epoch": 2.3051943270349478,
      "grad_norm": 1.082948088645935,
      "learning_rate": 1.1981188267867548e-05,
      "loss": 0.0622,
      "step": 156525
    },
    {
      "epoch": 2.3055625101250348,
      "grad_norm": 1.3589931726455688,
      "learning_rate": 1.1979551895900562e-05,
      "loss": 0.0543,
      "step": 156550
    },
    {
      "epoch": 2.305930693215122,
      "grad_norm": 1.412954330444336,
      "learning_rate": 1.1977915523933577e-05,
      "loss": 0.0517,
      "step": 156575
    },
    {
      "epoch": 2.3062988763052092,
      "grad_norm": 1.567301869392395,
      "learning_rate": 1.1976279151966593e-05,
      "loss": 0.059,
      "step": 156600
    },
    {
      "epoch": 2.3066670593952963,
      "grad_norm": 1.3498588800430298,
      "learning_rate": 1.1974642779999607e-05,
      "loss": 0.0587,
      "step": 156625
    },
    {
      "epoch": 2.3070352424853833,
      "grad_norm": 1.5007323026657104,
      "learning_rate": 1.1973006408032623e-05,
      "loss": 0.062,
      "step": 156650
    },
    {
      "epoch": 2.3074034255754703,
      "grad_norm": 1.3714014291763306,
      "learning_rate": 1.197137003606564e-05,
      "loss": 0.0606,
      "step": 156675
    },
    {
      "epoch": 2.3077716086655573,
      "grad_norm": 1.85383939743042,
      "learning_rate": 1.1969733664098656e-05,
      "loss": 0.066,
      "step": 156700
    },
    {
      "epoch": 2.3081397917556443,
      "grad_norm": 1.467696189880371,
      "learning_rate": 1.196809729213167e-05,
      "loss": 0.0655,
      "step": 156725
    },
    {
      "epoch": 2.3085079748457313,
      "grad_norm": 1.5151034593582153,
      "learning_rate": 1.1966460920164686e-05,
      "loss": 0.0667,
      "step": 156750
    },
    {
      "epoch": 2.3088761579358184,
      "grad_norm": 1.274662733078003,
      "learning_rate": 1.1964824548197699e-05,
      "loss": 0.06,
      "step": 156775
    },
    {
      "epoch": 2.3092443410259054,
      "grad_norm": 1.2610223293304443,
      "learning_rate": 1.1963188176230715e-05,
      "loss": 0.0609,
      "step": 156800
    },
    {
      "epoch": 2.3096125241159924,
      "grad_norm": 1.4022438526153564,
      "learning_rate": 1.1961551804263731e-05,
      "loss": 0.0571,
      "step": 156825
    },
    {
      "epoch": 2.3099807072060794,
      "grad_norm": 1.190034031867981,
      "learning_rate": 1.1959915432296748e-05,
      "loss": 0.0598,
      "step": 156850
    },
    {
      "epoch": 2.3103488902961664,
      "grad_norm": 1.7318341732025146,
      "learning_rate": 1.1958279060329762e-05,
      "loss": 0.0634,
      "step": 156875
    },
    {
      "epoch": 2.3107170733862534,
      "grad_norm": 1.3399244546890259,
      "learning_rate": 1.1956642688362778e-05,
      "loss": 0.0589,
      "step": 156900
    },
    {
      "epoch": 2.3110852564763404,
      "grad_norm": 1.327311396598816,
      "learning_rate": 1.1955006316395794e-05,
      "loss": 0.0652,
      "step": 156925
    },
    {
      "epoch": 2.3114534395664275,
      "grad_norm": 1.235003113746643,
      "learning_rate": 1.195343539930749e-05,
      "loss": 0.0675,
      "step": 156950
    },
    {
      "epoch": 2.3118216226565145,
      "grad_norm": 1.0948799848556519,
      "learning_rate": 1.1951799027340504e-05,
      "loss": 0.0621,
      "step": 156975
    },
    {
      "epoch": 2.3121898057466015,
      "grad_norm": 1.446100115776062,
      "learning_rate": 1.1950162655373519e-05,
      "loss": 0.0587,
      "step": 157000
    },
    {
      "epoch": 2.312557988836689,
      "grad_norm": 1.5128823518753052,
      "learning_rate": 1.1948526283406535e-05,
      "loss": 0.0602,
      "step": 157025
    },
    {
      "epoch": 2.312926171926776,
      "grad_norm": 1.706076741218567,
      "learning_rate": 1.194688991143955e-05,
      "loss": 0.0578,
      "step": 157050
    },
    {
      "epoch": 2.313294355016863,
      "grad_norm": 1.1051557064056396,
      "learning_rate": 1.1945253539472566e-05,
      "loss": 0.0544,
      "step": 157075
    },
    {
      "epoch": 2.31366253810695,
      "grad_norm": 1.1977298259735107,
      "learning_rate": 1.1943617167505582e-05,
      "loss": 0.0622,
      "step": 157100
    },
    {
      "epoch": 2.314030721197037,
      "grad_norm": 1.512346625328064,
      "learning_rate": 1.1941980795538596e-05,
      "loss": 0.0595,
      "step": 157125
    },
    {
      "epoch": 2.314398904287124,
      "grad_norm": 1.0016157627105713,
      "learning_rate": 1.1940344423571612e-05,
      "loss": 0.0646,
      "step": 157150
    },
    {
      "epoch": 2.314767087377211,
      "grad_norm": 1.4108219146728516,
      "learning_rate": 1.1938708051604628e-05,
      "loss": 0.0576,
      "step": 157175
    },
    {
      "epoch": 2.315135270467298,
      "grad_norm": 1.4841899871826172,
      "learning_rate": 1.1937071679637641e-05,
      "loss": 0.0578,
      "step": 157200
    },
    {
      "epoch": 2.315503453557385,
      "grad_norm": 1.4161380529403687,
      "learning_rate": 1.1935435307670657e-05,
      "loss": 0.063,
      "step": 157225
    },
    {
      "epoch": 2.315871636647472,
      "grad_norm": 1.1011296510696411,
      "learning_rate": 1.1933798935703674e-05,
      "loss": 0.0533,
      "step": 157250
    },
    {
      "epoch": 2.316239819737559,
      "grad_norm": 1.3405083417892456,
      "learning_rate": 1.193216256373669e-05,
      "loss": 0.0645,
      "step": 157275
    },
    {
      "epoch": 2.316608002827646,
      "grad_norm": 1.6028244495391846,
      "learning_rate": 1.1930526191769704e-05,
      "loss": 0.0645,
      "step": 157300
    },
    {
      "epoch": 2.316976185917733,
      "grad_norm": 1.0115848779678345,
      "learning_rate": 1.192888981980272e-05,
      "loss": 0.0536,
      "step": 157325
    },
    {
      "epoch": 2.31734436900782,
      "grad_norm": 1.0993667840957642,
      "learning_rate": 1.1927253447835736e-05,
      "loss": 0.06,
      "step": 157350
    },
    {
      "epoch": 2.317712552097907,
      "grad_norm": 1.1778128147125244,
      "learning_rate": 1.1925617075868751e-05,
      "loss": 0.0643,
      "step": 157375
    },
    {
      "epoch": 2.318080735187994,
      "grad_norm": 1.7951834201812744,
      "learning_rate": 1.1923980703901765e-05,
      "loss": 0.0676,
      "step": 157400
    },
    {
      "epoch": 2.318448918278081,
      "grad_norm": 1.0738455057144165,
      "learning_rate": 1.1922344331934781e-05,
      "loss": 0.0577,
      "step": 157425
    },
    {
      "epoch": 2.318817101368168,
      "grad_norm": 1.2450037002563477,
      "learning_rate": 1.1920707959967796e-05,
      "loss": 0.0617,
      "step": 157450
    },
    {
      "epoch": 2.3191852844582552,
      "grad_norm": 0.9938094615936279,
      "learning_rate": 1.1919071588000812e-05,
      "loss": 0.0565,
      "step": 157475
    },
    {
      "epoch": 2.3195534675483422,
      "grad_norm": 1.3486922979354858,
      "learning_rate": 1.1917435216033828e-05,
      "loss": 0.0554,
      "step": 157500
    },
    {
      "epoch": 2.3199216506384293,
      "grad_norm": 1.1815550327301025,
      "learning_rate": 1.1915798844066844e-05,
      "loss": 0.0592,
      "step": 157525
    },
    {
      "epoch": 2.3202898337285163,
      "grad_norm": 0.7472695112228394,
      "learning_rate": 1.1914162472099859e-05,
      "loss": 0.0625,
      "step": 157550
    },
    {
      "epoch": 2.3206580168186037,
      "grad_norm": 1.0809193849563599,
      "learning_rate": 1.1912526100132875e-05,
      "loss": 0.0566,
      "step": 157575
    },
    {
      "epoch": 2.3210261999086907,
      "grad_norm": 0.8146666884422302,
      "learning_rate": 1.1910889728165891e-05,
      "loss": 0.0531,
      "step": 157600
    },
    {
      "epoch": 2.3213943829987778,
      "grad_norm": 1.1890565156936646,
      "learning_rate": 1.1909253356198904e-05,
      "loss": 0.0612,
      "step": 157625
    },
    {
      "epoch": 2.3217625660888648,
      "grad_norm": 1.3931154012680054,
      "learning_rate": 1.190761698423192e-05,
      "loss": 0.0609,
      "step": 157650
    },
    {
      "epoch": 2.322130749178952,
      "grad_norm": 1.0479291677474976,
      "learning_rate": 1.1905980612264936e-05,
      "loss": 0.0605,
      "step": 157675
    },
    {
      "epoch": 2.322498932269039,
      "grad_norm": 1.5202665328979492,
      "learning_rate": 1.190434424029795e-05,
      "loss": 0.0629,
      "step": 157700
    },
    {
      "epoch": 2.322867115359126,
      "grad_norm": 1.2968895435333252,
      "learning_rate": 1.1902707868330967e-05,
      "loss": 0.0572,
      "step": 157725
    },
    {
      "epoch": 2.323235298449213,
      "grad_norm": 1.9595389366149902,
      "learning_rate": 1.1901071496363983e-05,
      "loss": 0.0663,
      "step": 157750
    },
    {
      "epoch": 2.3236034815393,
      "grad_norm": 1.2240461111068726,
      "learning_rate": 1.1899435124396999e-05,
      "loss": 0.0659,
      "step": 157775
    },
    {
      "epoch": 2.323971664629387,
      "grad_norm": 1.2727786302566528,
      "learning_rate": 1.1897798752430014e-05,
      "loss": 0.0623,
      "step": 157800
    },
    {
      "epoch": 2.324339847719474,
      "grad_norm": 1.2986425161361694,
      "learning_rate": 1.1896162380463028e-05,
      "loss": 0.0651,
      "step": 157825
    },
    {
      "epoch": 2.324708030809561,
      "grad_norm": 1.1853927373886108,
      "learning_rate": 1.1894526008496044e-05,
      "loss": 0.0558,
      "step": 157850
    },
    {
      "epoch": 2.325076213899648,
      "grad_norm": 1.7799198627471924,
      "learning_rate": 1.1892889636529059e-05,
      "loss": 0.0598,
      "step": 157875
    },
    {
      "epoch": 2.325444396989735,
      "grad_norm": 1.1868650913238525,
      "learning_rate": 1.1891253264562075e-05,
      "loss": 0.06,
      "step": 157900
    },
    {
      "epoch": 2.325812580079822,
      "grad_norm": 1.1879189014434814,
      "learning_rate": 1.1889616892595091e-05,
      "loss": 0.0605,
      "step": 157925
    },
    {
      "epoch": 2.326180763169909,
      "grad_norm": 1.43667733669281,
      "learning_rate": 1.1887980520628105e-05,
      "loss": 0.0625,
      "step": 157950
    },
    {
      "epoch": 2.3265489462599964,
      "grad_norm": 1.117746114730835,
      "learning_rate": 1.1886344148661122e-05,
      "loss": 0.0587,
      "step": 157975
    },
    {
      "epoch": 2.3269171293500834,
      "grad_norm": 1.7764755487442017,
      "learning_rate": 1.1884707776694138e-05,
      "loss": 0.0647,
      "step": 158000
    },
    {
      "epoch": 2.3272853124401705,
      "grad_norm": 1.1765077114105225,
      "learning_rate": 1.1883071404727154e-05,
      "loss": 0.0546,
      "step": 158025
    },
    {
      "epoch": 2.3276534955302575,
      "grad_norm": 1.383188247680664,
      "learning_rate": 1.1881435032760167e-05,
      "loss": 0.0584,
      "step": 158050
    },
    {
      "epoch": 2.3280216786203445,
      "grad_norm": 1.0768449306488037,
      "learning_rate": 1.1879798660793183e-05,
      "loss": 0.0577,
      "step": 158075
    },
    {
      "epoch": 2.3283898617104315,
      "grad_norm": 1.5985080003738403,
      "learning_rate": 1.1878162288826199e-05,
      "loss": 0.068,
      "step": 158100
    },
    {
      "epoch": 2.3287580448005185,
      "grad_norm": 1.1434884071350098,
      "learning_rate": 1.1876525916859213e-05,
      "loss": 0.06,
      "step": 158125
    },
    {
      "epoch": 2.3291262278906055,
      "grad_norm": 1.3976846933364868,
      "learning_rate": 1.187488954489223e-05,
      "loss": 0.0613,
      "step": 158150
    },
    {
      "epoch": 2.3294944109806925,
      "grad_norm": 0.9468764066696167,
      "learning_rate": 1.1873253172925246e-05,
      "loss": 0.0592,
      "step": 158175
    },
    {
      "epoch": 2.3298625940707796,
      "grad_norm": 1.5256427526474,
      "learning_rate": 1.1871616800958262e-05,
      "loss": 0.0641,
      "step": 158200
    },
    {
      "epoch": 2.3302307771608666,
      "grad_norm": 1.3651533126831055,
      "learning_rate": 1.1869980428991276e-05,
      "loss": 0.0662,
      "step": 158225
    },
    {
      "epoch": 2.3305989602509536,
      "grad_norm": 1.3118497133255005,
      "learning_rate": 1.186834405702429e-05,
      "loss": 0.0662,
      "step": 158250
    },
    {
      "epoch": 2.3309671433410406,
      "grad_norm": 1.3593465089797974,
      "learning_rate": 1.1866707685057305e-05,
      "loss": 0.0617,
      "step": 158275
    },
    {
      "epoch": 2.3313353264311276,
      "grad_norm": 1.5234158039093018,
      "learning_rate": 1.1865071313090321e-05,
      "loss": 0.0627,
      "step": 158300
    },
    {
      "epoch": 2.3317035095212146,
      "grad_norm": 1.2156411409378052,
      "learning_rate": 1.1863434941123338e-05,
      "loss": 0.0621,
      "step": 158325
    },
    {
      "epoch": 2.3320716926113016,
      "grad_norm": 1.3550735712051392,
      "learning_rate": 1.1861798569156354e-05,
      "loss": 0.0585,
      "step": 158350
    },
    {
      "epoch": 2.3324398757013887,
      "grad_norm": 1.3472539186477661,
      "learning_rate": 1.1860162197189368e-05,
      "loss": 0.0586,
      "step": 158375
    },
    {
      "epoch": 2.3328080587914757,
      "grad_norm": 1.8611524105072021,
      "learning_rate": 1.1858525825222384e-05,
      "loss": 0.0604,
      "step": 158400
    },
    {
      "epoch": 2.3331762418815627,
      "grad_norm": 1.456943392753601,
      "learning_rate": 1.18568894532554e-05,
      "loss": 0.0693,
      "step": 158425
    },
    {
      "epoch": 2.3335444249716497,
      "grad_norm": 1.9718719720840454,
      "learning_rate": 1.1855253081288417e-05,
      "loss": 0.0622,
      "step": 158450
    },
    {
      "epoch": 2.3339126080617367,
      "grad_norm": 1.6115795373916626,
      "learning_rate": 1.185361670932143e-05,
      "loss": 0.0623,
      "step": 158475
    },
    {
      "epoch": 2.3342807911518237,
      "grad_norm": 1.6292591094970703,
      "learning_rate": 1.1851980337354446e-05,
      "loss": 0.0582,
      "step": 158500
    },
    {
      "epoch": 2.334648974241911,
      "grad_norm": 1.5077576637268066,
      "learning_rate": 1.185034396538746e-05,
      "loss": 0.0562,
      "step": 158525
    },
    {
      "epoch": 2.335017157331998,
      "grad_norm": 1.3961241245269775,
      "learning_rate": 1.1848707593420476e-05,
      "loss": 0.0666,
      "step": 158550
    },
    {
      "epoch": 2.3353853404220852,
      "grad_norm": 1.9080363512039185,
      "learning_rate": 1.1847071221453492e-05,
      "loss": 0.0641,
      "step": 158575
    },
    {
      "epoch": 2.3357535235121722,
      "grad_norm": 1.4271209239959717,
      "learning_rate": 1.1845434849486508e-05,
      "loss": 0.0591,
      "step": 158600
    },
    {
      "epoch": 2.3361217066022593,
      "grad_norm": 1.1924054622650146,
      "learning_rate": 1.1843798477519523e-05,
      "loss": 0.0626,
      "step": 158625
    },
    {
      "epoch": 2.3364898896923463,
      "grad_norm": 1.6412979364395142,
      "learning_rate": 1.1842162105552539e-05,
      "loss": 0.0736,
      "step": 158650
    },
    {
      "epoch": 2.3368580727824333,
      "grad_norm": 1.3793342113494873,
      "learning_rate": 1.1840525733585553e-05,
      "loss": 0.0608,
      "step": 158675
    },
    {
      "epoch": 2.3372262558725203,
      "grad_norm": 1.7255667448043823,
      "learning_rate": 1.1838889361618568e-05,
      "loss": 0.0653,
      "step": 158700
    },
    {
      "epoch": 2.3375944389626073,
      "grad_norm": 1.3862090110778809,
      "learning_rate": 1.1837252989651584e-05,
      "loss": 0.0594,
      "step": 158725
    },
    {
      "epoch": 2.3379626220526943,
      "grad_norm": 1.1461907625198364,
      "learning_rate": 1.18356166176846e-05,
      "loss": 0.0604,
      "step": 158750
    },
    {
      "epoch": 2.3383308051427814,
      "grad_norm": 1.108074426651001,
      "learning_rate": 1.1833980245717616e-05,
      "loss": 0.0636,
      "step": 158775
    },
    {
      "epoch": 2.3386989882328684,
      "grad_norm": 1.3630497455596924,
      "learning_rate": 1.1832343873750631e-05,
      "loss": 0.0607,
      "step": 158800
    },
    {
      "epoch": 2.3390671713229554,
      "grad_norm": 1.5388332605361938,
      "learning_rate": 1.1830707501783647e-05,
      "loss": 0.06,
      "step": 158825
    },
    {
      "epoch": 2.3394353544130424,
      "grad_norm": 0.6975927948951721,
      "learning_rate": 1.1829071129816663e-05,
      "loss": 0.0649,
      "step": 158850
    },
    {
      "epoch": 2.3398035375031294,
      "grad_norm": 1.3464326858520508,
      "learning_rate": 1.1827434757849676e-05,
      "loss": 0.0565,
      "step": 158875
    },
    {
      "epoch": 2.3401717205932164,
      "grad_norm": 1.3845411539077759,
      "learning_rate": 1.1825798385882692e-05,
      "loss": 0.0579,
      "step": 158900
    },
    {
      "epoch": 2.3405399036833034,
      "grad_norm": 1.7086868286132812,
      "learning_rate": 1.1824162013915708e-05,
      "loss": 0.0588,
      "step": 158925
    },
    {
      "epoch": 2.340908086773391,
      "grad_norm": 1.5713566541671753,
      "learning_rate": 1.1822525641948723e-05,
      "loss": 0.0571,
      "step": 158950
    },
    {
      "epoch": 2.341276269863478,
      "grad_norm": 1.0964891910552979,
      "learning_rate": 1.1820889269981739e-05,
      "loss": 0.0618,
      "step": 158975
    },
    {
      "epoch": 2.341644452953565,
      "grad_norm": 1.443109393119812,
      "learning_rate": 1.1819252898014755e-05,
      "loss": 0.0568,
      "step": 159000
    },
    {
      "epoch": 2.342012636043652,
      "grad_norm": 1.610060453414917,
      "learning_rate": 1.1817616526047771e-05,
      "loss": 0.0608,
      "step": 159025
    },
    {
      "epoch": 2.342380819133739,
      "grad_norm": 2.101233959197998,
      "learning_rate": 1.1815980154080786e-05,
      "loss": 0.063,
      "step": 159050
    },
    {
      "epoch": 2.342749002223826,
      "grad_norm": 1.2336227893829346,
      "learning_rate": 1.1814343782113802e-05,
      "loss": 0.0589,
      "step": 159075
    },
    {
      "epoch": 2.343117185313913,
      "grad_norm": 1.50912344455719,
      "learning_rate": 1.1812707410146815e-05,
      "loss": 0.0705,
      "step": 159100
    },
    {
      "epoch": 2.343485368404,
      "grad_norm": 1.507298231124878,
      "learning_rate": 1.181107103817983e-05,
      "loss": 0.0624,
      "step": 159125
    },
    {
      "epoch": 2.343853551494087,
      "grad_norm": 0.8697091341018677,
      "learning_rate": 1.1809434666212847e-05,
      "loss": 0.0592,
      "step": 159150
    },
    {
      "epoch": 2.344221734584174,
      "grad_norm": 0.6300997138023376,
      "learning_rate": 1.1807798294245863e-05,
      "loss": 0.0596,
      "step": 159175
    },
    {
      "epoch": 2.344589917674261,
      "grad_norm": 0.9704003930091858,
      "learning_rate": 1.1806161922278877e-05,
      "loss": 0.0517,
      "step": 159200
    },
    {
      "epoch": 2.344958100764348,
      "grad_norm": 1.3032547235488892,
      "learning_rate": 1.1804591005190573e-05,
      "loss": 0.0634,
      "step": 159225
    },
    {
      "epoch": 2.345326283854435,
      "grad_norm": 1.7722431421279907,
      "learning_rate": 1.1802954633223589e-05,
      "loss": 0.0586,
      "step": 159250
    },
    {
      "epoch": 2.345694466944522,
      "grad_norm": 1.3605221509933472,
      "learning_rate": 1.1801318261256605e-05,
      "loss": 0.0634,
      "step": 159275
    },
    {
      "epoch": 2.346062650034609,
      "grad_norm": 1.5096231698989868,
      "learning_rate": 1.1799681889289618e-05,
      "loss": 0.0621,
      "step": 159300
    },
    {
      "epoch": 2.346430833124696,
      "grad_norm": 1.2020301818847656,
      "learning_rate": 1.1798045517322634e-05,
      "loss": 0.06,
      "step": 159325
    },
    {
      "epoch": 2.346799016214783,
      "grad_norm": 1.5449987649917603,
      "learning_rate": 1.1796409145355649e-05,
      "loss": 0.0608,
      "step": 159350
    },
    {
      "epoch": 2.34716719930487,
      "grad_norm": 1.3683671951293945,
      "learning_rate": 1.1794772773388665e-05,
      "loss": 0.0639,
      "step": 159375
    },
    {
      "epoch": 2.347535382394957,
      "grad_norm": 1.0007554292678833,
      "learning_rate": 1.1793136401421681e-05,
      "loss": 0.0586,
      "step": 159400
    },
    {
      "epoch": 2.347903565485044,
      "grad_norm": 1.7813198566436768,
      "learning_rate": 1.1791500029454697e-05,
      "loss": 0.0613,
      "step": 159425
    },
    {
      "epoch": 2.348271748575131,
      "grad_norm": 1.5094960927963257,
      "learning_rate": 1.1789863657487712e-05,
      "loss": 0.059,
      "step": 159450
    },
    {
      "epoch": 2.3486399316652187,
      "grad_norm": 1.5805097818374634,
      "learning_rate": 1.1788227285520728e-05,
      "loss": 0.0659,
      "step": 159475
    },
    {
      "epoch": 2.3490081147553057,
      "grad_norm": 1.2663806676864624,
      "learning_rate": 1.1786590913553744e-05,
      "loss": 0.0547,
      "step": 159500
    },
    {
      "epoch": 2.3493762978453927,
      "grad_norm": 1.15206778049469,
      "learning_rate": 1.1784954541586757e-05,
      "loss": 0.059,
      "step": 159525
    },
    {
      "epoch": 2.3497444809354797,
      "grad_norm": 1.3665211200714111,
      "learning_rate": 1.1783318169619773e-05,
      "loss": 0.0583,
      "step": 159550
    },
    {
      "epoch": 2.3501126640255667,
      "grad_norm": 0.8548767566680908,
      "learning_rate": 1.1781681797652789e-05,
      "loss": 0.0592,
      "step": 159575
    },
    {
      "epoch": 2.3504808471156537,
      "grad_norm": 1.1451300382614136,
      "learning_rate": 1.1780045425685805e-05,
      "loss": 0.0665,
      "step": 159600
    },
    {
      "epoch": 2.3508490302057408,
      "grad_norm": 1.5223888158798218,
      "learning_rate": 1.177840905371882e-05,
      "loss": 0.0622,
      "step": 159625
    },
    {
      "epoch": 2.351217213295828,
      "grad_norm": 1.9416120052337646,
      "learning_rate": 1.1776772681751836e-05,
      "loss": 0.0625,
      "step": 159650
    },
    {
      "epoch": 2.351585396385915,
      "grad_norm": 1.0877821445465088,
      "learning_rate": 1.1775136309784852e-05,
      "loss": 0.0626,
      "step": 159675
    },
    {
      "epoch": 2.351953579476002,
      "grad_norm": 1.207059383392334,
      "learning_rate": 1.1773499937817866e-05,
      "loss": 0.0612,
      "step": 159700
    },
    {
      "epoch": 2.352321762566089,
      "grad_norm": 1.0630658864974976,
      "learning_rate": 1.177186356585088e-05,
      "loss": 0.0602,
      "step": 159725
    },
    {
      "epoch": 2.352689945656176,
      "grad_norm": 1.6132158041000366,
      "learning_rate": 1.1770227193883897e-05,
      "loss": 0.0611,
      "step": 159750
    },
    {
      "epoch": 2.353058128746263,
      "grad_norm": 0.8465059399604797,
      "learning_rate": 1.1768590821916911e-05,
      "loss": 0.0634,
      "step": 159775
    },
    {
      "epoch": 2.35342631183635,
      "grad_norm": 1.100665807723999,
      "learning_rate": 1.1766954449949928e-05,
      "loss": 0.0551,
      "step": 159800
    },
    {
      "epoch": 2.353794494926437,
      "grad_norm": 1.456581473350525,
      "learning_rate": 1.1765318077982944e-05,
      "loss": 0.0585,
      "step": 159825
    },
    {
      "epoch": 2.354162678016524,
      "grad_norm": 1.5542956590652466,
      "learning_rate": 1.176368170601596e-05,
      "loss": 0.0635,
      "step": 159850
    },
    {
      "epoch": 2.354530861106611,
      "grad_norm": 1.1105351448059082,
      "learning_rate": 1.1762045334048974e-05,
      "loss": 0.0601,
      "step": 159875
    },
    {
      "epoch": 2.3548990441966984,
      "grad_norm": 1.045846939086914,
      "learning_rate": 1.176040896208199e-05,
      "loss": 0.0564,
      "step": 159900
    },
    {
      "epoch": 2.3552672272867854,
      "grad_norm": 1.3951945304870605,
      "learning_rate": 1.1758772590115007e-05,
      "loss": 0.0596,
      "step": 159925
    },
    {
      "epoch": 2.3556354103768724,
      "grad_norm": 0.9000104069709778,
      "learning_rate": 1.175713621814802e-05,
      "loss": 0.0494,
      "step": 159950
    },
    {
      "epoch": 2.3560035934669594,
      "grad_norm": 1.3923025131225586,
      "learning_rate": 1.1755499846181036e-05,
      "loss": 0.0725,
      "step": 159975
    },
    {
      "epoch": 2.3563717765570464,
      "grad_norm": 1.5339983701705933,
      "learning_rate": 1.1753863474214052e-05,
      "loss": 0.0619,
      "step": 160000
    },
    {
      "epoch": 2.3567399596471335,
      "grad_norm": 1.526166558265686,
      "learning_rate": 1.1752227102247066e-05,
      "loss": 0.0671,
      "step": 160025
    },
    {
      "epoch": 2.3571081427372205,
      "grad_norm": 1.3462245464324951,
      "learning_rate": 1.1750590730280082e-05,
      "loss": 0.0609,
      "step": 160050
    },
    {
      "epoch": 2.3574763258273075,
      "grad_norm": 1.4161617755889893,
      "learning_rate": 1.1748954358313098e-05,
      "loss": 0.0627,
      "step": 160075
    },
    {
      "epoch": 2.3578445089173945,
      "grad_norm": 1.2137469053268433,
      "learning_rate": 1.1747317986346115e-05,
      "loss": 0.0588,
      "step": 160100
    },
    {
      "epoch": 2.3582126920074815,
      "grad_norm": 1.8838942050933838,
      "learning_rate": 1.1745681614379129e-05,
      "loss": 0.0626,
      "step": 160125
    },
    {
      "epoch": 2.3585808750975685,
      "grad_norm": 1.1949462890625,
      "learning_rate": 1.1744045242412143e-05,
      "loss": 0.0582,
      "step": 160150
    },
    {
      "epoch": 2.3589490581876555,
      "grad_norm": 1.2934921979904175,
      "learning_rate": 1.174240887044516e-05,
      "loss": 0.0571,
      "step": 160175
    },
    {
      "epoch": 2.3593172412777426,
      "grad_norm": 0.8720579147338867,
      "learning_rate": 1.1740772498478174e-05,
      "loss": 0.0563,
      "step": 160200
    },
    {
      "epoch": 2.3596854243678296,
      "grad_norm": 1.112901210784912,
      "learning_rate": 1.173913612651119e-05,
      "loss": 0.0649,
      "step": 160225
    },
    {
      "epoch": 2.3600536074579166,
      "grad_norm": 1.310280442237854,
      "learning_rate": 1.1737499754544206e-05,
      "loss": 0.0492,
      "step": 160250
    },
    {
      "epoch": 2.3604217905480036,
      "grad_norm": 1.4563299417495728,
      "learning_rate": 1.1735863382577221e-05,
      "loss": 0.0647,
      "step": 160275
    },
    {
      "epoch": 2.3607899736380906,
      "grad_norm": 1.1201913356781006,
      "learning_rate": 1.1734227010610237e-05,
      "loss": 0.0592,
      "step": 160300
    },
    {
      "epoch": 2.3611581567281776,
      "grad_norm": 1.393536925315857,
      "learning_rate": 1.1732590638643253e-05,
      "loss": 0.0669,
      "step": 160325
    },
    {
      "epoch": 2.3615263398182647,
      "grad_norm": 1.3970112800598145,
      "learning_rate": 1.1730954266676266e-05,
      "loss": 0.0627,
      "step": 160350
    },
    {
      "epoch": 2.3618945229083517,
      "grad_norm": 1.3412425518035889,
      "learning_rate": 1.1729317894709282e-05,
      "loss": 0.0596,
      "step": 160375
    },
    {
      "epoch": 2.3622627059984387,
      "grad_norm": 1.4249790906906128,
      "learning_rate": 1.1727681522742298e-05,
      "loss": 0.0685,
      "step": 160400
    },
    {
      "epoch": 2.3626308890885257,
      "grad_norm": 1.6485384702682495,
      "learning_rate": 1.1726045150775314e-05,
      "loss": 0.0557,
      "step": 160425
    },
    {
      "epoch": 2.362999072178613,
      "grad_norm": 1.3390308618545532,
      "learning_rate": 1.1724408778808329e-05,
      "loss": 0.0627,
      "step": 160450
    },
    {
      "epoch": 2.3633672552687,
      "grad_norm": 1.3587732315063477,
      "learning_rate": 1.1722772406841345e-05,
      "loss": 0.0667,
      "step": 160475
    },
    {
      "epoch": 2.363735438358787,
      "grad_norm": 1.5958024263381958,
      "learning_rate": 1.1721136034874361e-05,
      "loss": 0.0626,
      "step": 160500
    },
    {
      "epoch": 2.364103621448874,
      "grad_norm": 1.5098737478256226,
      "learning_rate": 1.1719499662907377e-05,
      "loss": 0.0643,
      "step": 160525
    },
    {
      "epoch": 2.364471804538961,
      "grad_norm": 0.945949375629425,
      "learning_rate": 1.1717863290940392e-05,
      "loss": 0.0651,
      "step": 160550
    },
    {
      "epoch": 2.3648399876290482,
      "grad_norm": 1.5439256429672241,
      "learning_rate": 1.1716226918973406e-05,
      "loss": 0.0645,
      "step": 160575
    },
    {
      "epoch": 2.3652081707191352,
      "grad_norm": 1.4468693733215332,
      "learning_rate": 1.171459054700642e-05,
      "loss": 0.0613,
      "step": 160600
    },
    {
      "epoch": 2.3655763538092223,
      "grad_norm": 1.000960350036621,
      "learning_rate": 1.1712954175039437e-05,
      "loss": 0.0565,
      "step": 160625
    },
    {
      "epoch": 2.3659445368993093,
      "grad_norm": 1.3444818258285522,
      "learning_rate": 1.1711317803072453e-05,
      "loss": 0.0558,
      "step": 160650
    },
    {
      "epoch": 2.3663127199893963,
      "grad_norm": 1.2307623624801636,
      "learning_rate": 1.1709681431105469e-05,
      "loss": 0.0613,
      "step": 160675
    },
    {
      "epoch": 2.3666809030794833,
      "grad_norm": 1.2238882780075073,
      "learning_rate": 1.1708045059138484e-05,
      "loss": 0.0602,
      "step": 160700
    },
    {
      "epoch": 2.3670490861695703,
      "grad_norm": 1.5728293657302856,
      "learning_rate": 1.17064086871715e-05,
      "loss": 0.0584,
      "step": 160725
    },
    {
      "epoch": 2.3674172692596573,
      "grad_norm": 1.3549801111221313,
      "learning_rate": 1.1704772315204516e-05,
      "loss": 0.0595,
      "step": 160750
    },
    {
      "epoch": 2.3677854523497444,
      "grad_norm": 1.713382363319397,
      "learning_rate": 1.1703135943237529e-05,
      "loss": 0.0636,
      "step": 160775
    },
    {
      "epoch": 2.3681536354398314,
      "grad_norm": 1.4275914430618286,
      "learning_rate": 1.1701499571270545e-05,
      "loss": 0.0639,
      "step": 160800
    },
    {
      "epoch": 2.3685218185299184,
      "grad_norm": 1.5068387985229492,
      "learning_rate": 1.1699863199303561e-05,
      "loss": 0.0628,
      "step": 160825
    },
    {
      "epoch": 2.3688900016200054,
      "grad_norm": 1.2079943418502808,
      "learning_rate": 1.1698226827336575e-05,
      "loss": 0.0617,
      "step": 160850
    },
    {
      "epoch": 2.369258184710093,
      "grad_norm": 1.4133532047271729,
      "learning_rate": 1.1696590455369592e-05,
      "loss": 0.0692,
      "step": 160875
    },
    {
      "epoch": 2.36962636780018,
      "grad_norm": 1.2821435928344727,
      "learning_rate": 1.1694954083402608e-05,
      "loss": 0.0564,
      "step": 160900
    },
    {
      "epoch": 2.369994550890267,
      "grad_norm": 1.1887179613113403,
      "learning_rate": 1.1693317711435624e-05,
      "loss": 0.0603,
      "step": 160925
    },
    {
      "epoch": 2.370362733980354,
      "grad_norm": 1.1365894079208374,
      "learning_rate": 1.1691681339468638e-05,
      "loss": 0.0576,
      "step": 160950
    },
    {
      "epoch": 2.370730917070441,
      "grad_norm": 1.0952905416488647,
      "learning_rate": 1.1690044967501654e-05,
      "loss": 0.0641,
      "step": 160975
    },
    {
      "epoch": 2.371099100160528,
      "grad_norm": 1.5661420822143555,
      "learning_rate": 1.1688408595534669e-05,
      "loss": 0.0652,
      "step": 161000
    },
    {
      "epoch": 2.371467283250615,
      "grad_norm": 1.1456741094589233,
      "learning_rate": 1.1686772223567683e-05,
      "loss": 0.0643,
      "step": 161025
    },
    {
      "epoch": 2.371835466340702,
      "grad_norm": 1.2231078147888184,
      "learning_rate": 1.16851358516007e-05,
      "loss": 0.0606,
      "step": 161050
    },
    {
      "epoch": 2.372203649430789,
      "grad_norm": 1.5011032819747925,
      "learning_rate": 1.1683499479633716e-05,
      "loss": 0.0574,
      "step": 161075
    },
    {
      "epoch": 2.372571832520876,
      "grad_norm": 1.4581764936447144,
      "learning_rate": 1.1681863107666732e-05,
      "loss": 0.0565,
      "step": 161100
    },
    {
      "epoch": 2.372940015610963,
      "grad_norm": 1.2929635047912598,
      "learning_rate": 1.1680226735699746e-05,
      "loss": 0.0612,
      "step": 161125
    },
    {
      "epoch": 2.37330819870105,
      "grad_norm": 1.8550294637680054,
      "learning_rate": 1.1678590363732762e-05,
      "loss": 0.0621,
      "step": 161150
    },
    {
      "epoch": 2.373676381791137,
      "grad_norm": 1.3191032409667969,
      "learning_rate": 1.1676953991765779e-05,
      "loss": 0.0609,
      "step": 161175
    },
    {
      "epoch": 2.374044564881224,
      "grad_norm": 1.435896635055542,
      "learning_rate": 1.1675317619798791e-05,
      "loss": 0.0619,
      "step": 161200
    },
    {
      "epoch": 2.374412747971311,
      "grad_norm": 1.0836715698242188,
      "learning_rate": 1.1673681247831808e-05,
      "loss": 0.0559,
      "step": 161225
    },
    {
      "epoch": 2.374780931061398,
      "grad_norm": 1.4972012042999268,
      "learning_rate": 1.1672044875864824e-05,
      "loss": 0.0602,
      "step": 161250
    },
    {
      "epoch": 2.375149114151485,
      "grad_norm": 1.6567645072937012,
      "learning_rate": 1.1670408503897838e-05,
      "loss": 0.0634,
      "step": 161275
    },
    {
      "epoch": 2.375517297241572,
      "grad_norm": 1.0782804489135742,
      "learning_rate": 1.1668772131930854e-05,
      "loss": 0.0598,
      "step": 161300
    },
    {
      "epoch": 2.375885480331659,
      "grad_norm": 1.5296897888183594,
      "learning_rate": 1.166713575996387e-05,
      "loss": 0.0632,
      "step": 161325
    },
    {
      "epoch": 2.376253663421746,
      "grad_norm": 1.4100537300109863,
      "learning_rate": 1.1665499387996887e-05,
      "loss": 0.0629,
      "step": 161350
    },
    {
      "epoch": 2.376621846511833,
      "grad_norm": 1.6575449705123901,
      "learning_rate": 1.1663863016029901e-05,
      "loss": 0.058,
      "step": 161375
    },
    {
      "epoch": 2.3769900296019206,
      "grad_norm": 1.2804405689239502,
      "learning_rate": 1.1662226644062917e-05,
      "loss": 0.0611,
      "step": 161400
    },
    {
      "epoch": 2.3773582126920076,
      "grad_norm": 1.5220341682434082,
      "learning_rate": 1.166065572697461e-05,
      "loss": 0.0617,
      "step": 161425
    },
    {
      "epoch": 2.3777263957820947,
      "grad_norm": 1.1972874402999878,
      "learning_rate": 1.1659019355007626e-05,
      "loss": 0.0541,
      "step": 161450
    },
    {
      "epoch": 2.3780945788721817,
      "grad_norm": 1.238508701324463,
      "learning_rate": 1.1657382983040642e-05,
      "loss": 0.0598,
      "step": 161475
    },
    {
      "epoch": 2.3784627619622687,
      "grad_norm": 1.4524447917938232,
      "learning_rate": 1.1655746611073658e-05,
      "loss": 0.0625,
      "step": 161500
    },
    {
      "epoch": 2.3788309450523557,
      "grad_norm": 1.0858079195022583,
      "learning_rate": 1.1654110239106672e-05,
      "loss": 0.0596,
      "step": 161525
    },
    {
      "epoch": 2.3791991281424427,
      "grad_norm": 0.8202728033065796,
      "learning_rate": 1.1652473867139688e-05,
      "loss": 0.0607,
      "step": 161550
    },
    {
      "epoch": 2.3795673112325297,
      "grad_norm": 1.1476420164108276,
      "learning_rate": 1.1650837495172705e-05,
      "loss": 0.06,
      "step": 161575
    },
    {
      "epoch": 2.3799354943226168,
      "grad_norm": 1.1841585636138916,
      "learning_rate": 1.164920112320572e-05,
      "loss": 0.0588,
      "step": 161600
    },
    {
      "epoch": 2.3803036774127038,
      "grad_norm": 0.8197386264801025,
      "learning_rate": 1.1647564751238734e-05,
      "loss": 0.0653,
      "step": 161625
    },
    {
      "epoch": 2.380671860502791,
      "grad_norm": 1.3758143186569214,
      "learning_rate": 1.164592837927175e-05,
      "loss": 0.0572,
      "step": 161650
    },
    {
      "epoch": 2.381040043592878,
      "grad_norm": 1.315423846244812,
      "learning_rate": 1.1644292007304764e-05,
      "loss": 0.0581,
      "step": 161675
    },
    {
      "epoch": 2.381408226682965,
      "grad_norm": 1.294613003730774,
      "learning_rate": 1.164265563533778e-05,
      "loss": 0.0611,
      "step": 161700
    },
    {
      "epoch": 2.381776409773052,
      "grad_norm": 1.762940526008606,
      "learning_rate": 1.1641019263370796e-05,
      "loss": 0.0663,
      "step": 161725
    },
    {
      "epoch": 2.382144592863139,
      "grad_norm": 0.876607358455658,
      "learning_rate": 1.1639382891403813e-05,
      "loss": 0.0638,
      "step": 161750
    },
    {
      "epoch": 2.382512775953226,
      "grad_norm": 1.163330316543579,
      "learning_rate": 1.1637746519436827e-05,
      "loss": 0.0664,
      "step": 161775
    },
    {
      "epoch": 2.382880959043313,
      "grad_norm": 1.4517558813095093,
      "learning_rate": 1.1636110147469843e-05,
      "loss": 0.0638,
      "step": 161800
    },
    {
      "epoch": 2.3832491421334003,
      "grad_norm": 1.4793587923049927,
      "learning_rate": 1.1634473775502858e-05,
      "loss": 0.0574,
      "step": 161825
    },
    {
      "epoch": 2.3836173252234873,
      "grad_norm": 1.6949927806854248,
      "learning_rate": 1.1632837403535872e-05,
      "loss": 0.0626,
      "step": 161850
    },
    {
      "epoch": 2.3839855083135744,
      "grad_norm": 1.3969281911849976,
      "learning_rate": 1.1631201031568888e-05,
      "loss": 0.0549,
      "step": 161875
    },
    {
      "epoch": 2.3843536914036614,
      "grad_norm": 1.4277383089065552,
      "learning_rate": 1.1629564659601904e-05,
      "loss": 0.0642,
      "step": 161900
    },
    {
      "epoch": 2.3847218744937484,
      "grad_norm": 1.18282949924469,
      "learning_rate": 1.162792828763492e-05,
      "loss": 0.0611,
      "step": 161925
    },
    {
      "epoch": 2.3850900575838354,
      "grad_norm": 1.4680016040802002,
      "learning_rate": 1.1626291915667935e-05,
      "loss": 0.059,
      "step": 161950
    },
    {
      "epoch": 2.3854582406739224,
      "grad_norm": 2.0155391693115234,
      "learning_rate": 1.1624655543700951e-05,
      "loss": 0.0674,
      "step": 161975
    },
    {
      "epoch": 2.3858264237640094,
      "grad_norm": 0.9847386479377747,
      "learning_rate": 1.1623019171733967e-05,
      "loss": 0.0558,
      "step": 162000
    },
    {
      "epoch": 2.3861946068540965,
      "grad_norm": 0.9667565822601318,
      "learning_rate": 1.1621382799766982e-05,
      "loss": 0.0646,
      "step": 162025
    },
    {
      "epoch": 2.3865627899441835,
      "grad_norm": 1.3795044422149658,
      "learning_rate": 1.1619746427799996e-05,
      "loss": 0.0615,
      "step": 162050
    },
    {
      "epoch": 2.3869309730342705,
      "grad_norm": 1.7952836751937866,
      "learning_rate": 1.1618110055833012e-05,
      "loss": 0.0587,
      "step": 162075
    },
    {
      "epoch": 2.3872991561243575,
      "grad_norm": 1.4881402254104614,
      "learning_rate": 1.1616473683866027e-05,
      "loss": 0.0648,
      "step": 162100
    },
    {
      "epoch": 2.3876673392144445,
      "grad_norm": 1.7149633169174194,
      "learning_rate": 1.1614837311899043e-05,
      "loss": 0.0653,
      "step": 162125
    },
    {
      "epoch": 2.3880355223045315,
      "grad_norm": 1.3947250843048096,
      "learning_rate": 1.1613200939932059e-05,
      "loss": 0.0529,
      "step": 162150
    },
    {
      "epoch": 2.3884037053946185,
      "grad_norm": 1.2231005430221558,
      "learning_rate": 1.1611564567965075e-05,
      "loss": 0.0626,
      "step": 162175
    },
    {
      "epoch": 2.3887718884847056,
      "grad_norm": 1.3092209100723267,
      "learning_rate": 1.160992819599809e-05,
      "loss": 0.0622,
      "step": 162200
    },
    {
      "epoch": 2.3891400715747926,
      "grad_norm": 1.6007519960403442,
      "learning_rate": 1.1608291824031106e-05,
      "loss": 0.0597,
      "step": 162225
    },
    {
      "epoch": 2.3895082546648796,
      "grad_norm": 1.2384833097457886,
      "learning_rate": 1.1606655452064119e-05,
      "loss": 0.065,
      "step": 162250
    },
    {
      "epoch": 2.3898764377549666,
      "grad_norm": 1.6072911024093628,
      "learning_rate": 1.1605019080097135e-05,
      "loss": 0.0629,
      "step": 162275
    },
    {
      "epoch": 2.3902446208450536,
      "grad_norm": 1.4751777648925781,
      "learning_rate": 1.1603382708130151e-05,
      "loss": 0.0641,
      "step": 162300
    },
    {
      "epoch": 2.3906128039351406,
      "grad_norm": 1.5758445262908936,
      "learning_rate": 1.1601746336163167e-05,
      "loss": 0.0636,
      "step": 162325
    },
    {
      "epoch": 2.390980987025228,
      "grad_norm": 1.7630226612091064,
      "learning_rate": 1.1600109964196182e-05,
      "loss": 0.0639,
      "step": 162350
    },
    {
      "epoch": 2.391349170115315,
      "grad_norm": 1.7897053956985474,
      "learning_rate": 1.1598473592229198e-05,
      "loss": 0.0603,
      "step": 162375
    },
    {
      "epoch": 2.391717353205402,
      "grad_norm": 1.2426737546920776,
      "learning_rate": 1.1596837220262214e-05,
      "loss": 0.0614,
      "step": 162400
    },
    {
      "epoch": 2.392085536295489,
      "grad_norm": 1.243597388267517,
      "learning_rate": 1.159520084829523e-05,
      "loss": 0.0608,
      "step": 162425
    },
    {
      "epoch": 2.392453719385576,
      "grad_norm": 0.9385548233985901,
      "learning_rate": 1.1593564476328244e-05,
      "loss": 0.0522,
      "step": 162450
    },
    {
      "epoch": 2.392821902475663,
      "grad_norm": 1.703717589378357,
      "learning_rate": 1.1591928104361259e-05,
      "loss": 0.0596,
      "step": 162475
    },
    {
      "epoch": 2.39319008556575,
      "grad_norm": 0.9092941880226135,
      "learning_rate": 1.1590291732394275e-05,
      "loss": 0.0618,
      "step": 162500
    },
    {
      "epoch": 2.393558268655837,
      "grad_norm": 1.6146618127822876,
      "learning_rate": 1.158865536042729e-05,
      "loss": 0.0588,
      "step": 162525
    },
    {
      "epoch": 2.3939264517459242,
      "grad_norm": 1.4100440740585327,
      "learning_rate": 1.1587018988460306e-05,
      "loss": 0.0645,
      "step": 162550
    },
    {
      "epoch": 2.3942946348360112,
      "grad_norm": 1.964572787284851,
      "learning_rate": 1.1585382616493322e-05,
      "loss": 0.0603,
      "step": 162575
    },
    {
      "epoch": 2.3946628179260983,
      "grad_norm": 1.041839599609375,
      "learning_rate": 1.1583746244526336e-05,
      "loss": 0.0624,
      "step": 162600
    },
    {
      "epoch": 2.3950310010161853,
      "grad_norm": 1.6516464948654175,
      "learning_rate": 1.1582109872559352e-05,
      "loss": 0.0584,
      "step": 162625
    },
    {
      "epoch": 2.3953991841062723,
      "grad_norm": 0.9532708525657654,
      "learning_rate": 1.1580473500592369e-05,
      "loss": 0.0611,
      "step": 162650
    },
    {
      "epoch": 2.3957673671963593,
      "grad_norm": 1.1092602014541626,
      "learning_rate": 1.1578837128625381e-05,
      "loss": 0.0614,
      "step": 162675
    },
    {
      "epoch": 2.3961355502864463,
      "grad_norm": 1.2556697130203247,
      "learning_rate": 1.1577200756658398e-05,
      "loss": 0.0647,
      "step": 162700
    },
    {
      "epoch": 2.3965037333765333,
      "grad_norm": 1.751400351524353,
      "learning_rate": 1.1575564384691414e-05,
      "loss": 0.0594,
      "step": 162725
    },
    {
      "epoch": 2.3968719164666203,
      "grad_norm": 1.47752046585083,
      "learning_rate": 1.157392801272443e-05,
      "loss": 0.0592,
      "step": 162750
    },
    {
      "epoch": 2.397240099556708,
      "grad_norm": 1.470221996307373,
      "learning_rate": 1.1572291640757444e-05,
      "loss": 0.0617,
      "step": 162775
    },
    {
      "epoch": 2.397608282646795,
      "grad_norm": 1.528430700302124,
      "learning_rate": 1.157065526879046e-05,
      "loss": 0.0579,
      "step": 162800
    },
    {
      "epoch": 2.397976465736882,
      "grad_norm": 1.5887783765792847,
      "learning_rate": 1.1569018896823477e-05,
      "loss": 0.0688,
      "step": 162825
    },
    {
      "epoch": 2.398344648826969,
      "grad_norm": 0.9407671093940735,
      "learning_rate": 1.1567382524856493e-05,
      "loss": 0.0595,
      "step": 162850
    },
    {
      "epoch": 2.398712831917056,
      "grad_norm": 1.6214011907577515,
      "learning_rate": 1.1565746152889507e-05,
      "loss": 0.0619,
      "step": 162875
    },
    {
      "epoch": 2.399081015007143,
      "grad_norm": 1.2414709329605103,
      "learning_rate": 1.1564109780922522e-05,
      "loss": 0.0543,
      "step": 162900
    },
    {
      "epoch": 2.39944919809723,
      "grad_norm": 1.1898075342178345,
      "learning_rate": 1.1562473408955536e-05,
      "loss": 0.0671,
      "step": 162925
    },
    {
      "epoch": 2.399817381187317,
      "grad_norm": 1.2103432416915894,
      "learning_rate": 1.1560837036988552e-05,
      "loss": 0.0552,
      "step": 162950
    },
    {
      "epoch": 2.400185564277404,
      "grad_norm": 1.0854477882385254,
      "learning_rate": 1.1559200665021568e-05,
      "loss": 0.0619,
      "step": 162975
    },
    {
      "epoch": 2.400553747367491,
      "grad_norm": 1.0311617851257324,
      "learning_rate": 1.1557564293054585e-05,
      "loss": 0.0562,
      "step": 163000
    },
    {
      "epoch": 2.400921930457578,
      "grad_norm": 1.5999698638916016,
      "learning_rate": 1.1555927921087599e-05,
      "loss": 0.0589,
      "step": 163025
    },
    {
      "epoch": 2.401290113547665,
      "grad_norm": 1.5684162378311157,
      "learning_rate": 1.1554291549120615e-05,
      "loss": 0.0607,
      "step": 163050
    },
    {
      "epoch": 2.401658296637752,
      "grad_norm": 1.2197459936141968,
      "learning_rate": 1.1552655177153631e-05,
      "loss": 0.0571,
      "step": 163075
    },
    {
      "epoch": 2.402026479727839,
      "grad_norm": 1.5436311960220337,
      "learning_rate": 1.1551018805186644e-05,
      "loss": 0.0681,
      "step": 163100
    },
    {
      "epoch": 2.402394662817926,
      "grad_norm": 0.9872592687606812,
      "learning_rate": 1.154938243321966e-05,
      "loss": 0.0588,
      "step": 163125
    },
    {
      "epoch": 2.402762845908013,
      "grad_norm": 1.3021868467330933,
      "learning_rate": 1.1547746061252676e-05,
      "loss": 0.0567,
      "step": 163150
    },
    {
      "epoch": 2.4031310289981,
      "grad_norm": 1.247228980064392,
      "learning_rate": 1.154610968928569e-05,
      "loss": 0.0538,
      "step": 163175
    },
    {
      "epoch": 2.403499212088187,
      "grad_norm": 1.711039662361145,
      "learning_rate": 1.1544473317318707e-05,
      "loss": 0.0625,
      "step": 163200
    },
    {
      "epoch": 2.403867395178274,
      "grad_norm": 1.3409584760665894,
      "learning_rate": 1.1542836945351723e-05,
      "loss": 0.0625,
      "step": 163225
    },
    {
      "epoch": 2.404235578268361,
      "grad_norm": 1.0156556367874146,
      "learning_rate": 1.154120057338474e-05,
      "loss": 0.0654,
      "step": 163250
    },
    {
      "epoch": 2.404603761358448,
      "grad_norm": 0.9794853925704956,
      "learning_rate": 1.1539564201417754e-05,
      "loss": 0.0541,
      "step": 163275
    },
    {
      "epoch": 2.404971944448535,
      "grad_norm": 2.182584285736084,
      "learning_rate": 1.1537927829450768e-05,
      "loss": 0.0598,
      "step": 163300
    },
    {
      "epoch": 2.4053401275386226,
      "grad_norm": 1.5068104267120361,
      "learning_rate": 1.1536291457483784e-05,
      "loss": 0.0599,
      "step": 163325
    },
    {
      "epoch": 2.4057083106287096,
      "grad_norm": 1.6922208070755005,
      "learning_rate": 1.1534655085516799e-05,
      "loss": 0.0581,
      "step": 163350
    },
    {
      "epoch": 2.4060764937187966,
      "grad_norm": 1.176310420036316,
      "learning_rate": 1.1533018713549815e-05,
      "loss": 0.057,
      "step": 163375
    },
    {
      "epoch": 2.4064446768088836,
      "grad_norm": 1.4202582836151123,
      "learning_rate": 1.1531382341582831e-05,
      "loss": 0.0626,
      "step": 163400
    },
    {
      "epoch": 2.4068128598989706,
      "grad_norm": 1.572019338607788,
      "learning_rate": 1.1529745969615847e-05,
      "loss": 0.0627,
      "step": 163425
    },
    {
      "epoch": 2.4071810429890577,
      "grad_norm": 1.3447861671447754,
      "learning_rate": 1.1528109597648862e-05,
      "loss": 0.066,
      "step": 163450
    },
    {
      "epoch": 2.4075492260791447,
      "grad_norm": 1.3755595684051514,
      "learning_rate": 1.1526473225681878e-05,
      "loss": 0.058,
      "step": 163475
    },
    {
      "epoch": 2.4079174091692317,
      "grad_norm": 1.036713719367981,
      "learning_rate": 1.1524836853714894e-05,
      "loss": 0.0597,
      "step": 163500
    },
    {
      "epoch": 2.4082855922593187,
      "grad_norm": 1.1989190578460693,
      "learning_rate": 1.1523200481747907e-05,
      "loss": 0.0539,
      "step": 163525
    },
    {
      "epoch": 2.4086537753494057,
      "grad_norm": 1.349012017250061,
      "learning_rate": 1.1521629564659602e-05,
      "loss": 0.0603,
      "step": 163550
    },
    {
      "epoch": 2.4090219584394927,
      "grad_norm": 1.689965844154358,
      "learning_rate": 1.1519993192692619e-05,
      "loss": 0.0656,
      "step": 163575
    },
    {
      "epoch": 2.4093901415295798,
      "grad_norm": 1.2762181758880615,
      "learning_rate": 1.1518356820725633e-05,
      "loss": 0.056,
      "step": 163600
    },
    {
      "epoch": 2.4097583246196668,
      "grad_norm": 1.458579659461975,
      "learning_rate": 1.1516720448758649e-05,
      "loss": 0.0668,
      "step": 163625
    },
    {
      "epoch": 2.410126507709754,
      "grad_norm": 1.2224823236465454,
      "learning_rate": 1.1515084076791665e-05,
      "loss": 0.0622,
      "step": 163650
    },
    {
      "epoch": 2.410494690799841,
      "grad_norm": 1.6604403257369995,
      "learning_rate": 1.1513447704824681e-05,
      "loss": 0.0627,
      "step": 163675
    },
    {
      "epoch": 2.410862873889928,
      "grad_norm": 1.5779081583023071,
      "learning_rate": 1.1511811332857696e-05,
      "loss": 0.0592,
      "step": 163700
    },
    {
      "epoch": 2.411231056980015,
      "grad_norm": 1.486918330192566,
      "learning_rate": 1.151017496089071e-05,
      "loss": 0.0611,
      "step": 163725
    },
    {
      "epoch": 2.4115992400701023,
      "grad_norm": 1.4067714214324951,
      "learning_rate": 1.1508538588923725e-05,
      "loss": 0.0594,
      "step": 163750
    },
    {
      "epoch": 2.4119674231601893,
      "grad_norm": 1.2232433557510376,
      "learning_rate": 1.1506902216956741e-05,
      "loss": 0.06,
      "step": 163775
    },
    {
      "epoch": 2.4123356062502763,
      "grad_norm": 1.5029042959213257,
      "learning_rate": 1.1505265844989757e-05,
      "loss": 0.0691,
      "step": 163800
    },
    {
      "epoch": 2.4127037893403633,
      "grad_norm": 1.155443787574768,
      "learning_rate": 1.1503629473022773e-05,
      "loss": 0.0591,
      "step": 163825
    },
    {
      "epoch": 2.4130719724304504,
      "grad_norm": 1.2783972024917603,
      "learning_rate": 1.1501993101055788e-05,
      "loss": 0.0525,
      "step": 163850
    },
    {
      "epoch": 2.4134401555205374,
      "grad_norm": 1.4131832122802734,
      "learning_rate": 1.1500356729088804e-05,
      "loss": 0.0634,
      "step": 163875
    },
    {
      "epoch": 2.4138083386106244,
      "grad_norm": 0.925175130367279,
      "learning_rate": 1.149872035712182e-05,
      "loss": 0.0598,
      "step": 163900
    },
    {
      "epoch": 2.4141765217007114,
      "grad_norm": 1.3498001098632812,
      "learning_rate": 1.1497083985154836e-05,
      "loss": 0.0558,
      "step": 163925
    },
    {
      "epoch": 2.4145447047907984,
      "grad_norm": 1.1745197772979736,
      "learning_rate": 1.1495447613187849e-05,
      "loss": 0.0579,
      "step": 163950
    },
    {
      "epoch": 2.4149128878808854,
      "grad_norm": 1.5623067617416382,
      "learning_rate": 1.1493811241220865e-05,
      "loss": 0.0643,
      "step": 163975
    },
    {
      "epoch": 2.4152810709709724,
      "grad_norm": 1.6745076179504395,
      "learning_rate": 1.149217486925388e-05,
      "loss": 0.0644,
      "step": 164000
    },
    {
      "epoch": 2.4156492540610595,
      "grad_norm": 1.1848735809326172,
      "learning_rate": 1.1490538497286896e-05,
      "loss": 0.0606,
      "step": 164025
    },
    {
      "epoch": 2.4160174371511465,
      "grad_norm": 1.2545878887176514,
      "learning_rate": 1.1488902125319912e-05,
      "loss": 0.0592,
      "step": 164050
    },
    {
      "epoch": 2.4163856202412335,
      "grad_norm": 0.9535441994667053,
      "learning_rate": 1.1487265753352928e-05,
      "loss": 0.0623,
      "step": 164075
    },
    {
      "epoch": 2.4167538033313205,
      "grad_norm": 2.1380114555358887,
      "learning_rate": 1.1485629381385942e-05,
      "loss": 0.0609,
      "step": 164100
    },
    {
      "epoch": 2.4171219864214075,
      "grad_norm": 1.0810357332229614,
      "learning_rate": 1.1483993009418959e-05,
      "loss": 0.0632,
      "step": 164125
    },
    {
      "epoch": 2.4174901695114945,
      "grad_norm": 1.2533912658691406,
      "learning_rate": 1.1482356637451973e-05,
      "loss": 0.0606,
      "step": 164150
    },
    {
      "epoch": 2.4178583526015816,
      "grad_norm": 0.8738951086997986,
      "learning_rate": 1.1480720265484988e-05,
      "loss": 0.0615,
      "step": 164175
    },
    {
      "epoch": 2.4182265356916686,
      "grad_norm": 1.5970969200134277,
      "learning_rate": 1.1479083893518004e-05,
      "loss": 0.0663,
      "step": 164200
    },
    {
      "epoch": 2.4185947187817556,
      "grad_norm": 1.8518763780593872,
      "learning_rate": 1.147744752155102e-05,
      "loss": 0.0599,
      "step": 164225
    },
    {
      "epoch": 2.4189629018718426,
      "grad_norm": 1.0967353582382202,
      "learning_rate": 1.1475811149584036e-05,
      "loss": 0.0679,
      "step": 164250
    },
    {
      "epoch": 2.41933108496193,
      "grad_norm": 1.0748199224472046,
      "learning_rate": 1.147417477761705e-05,
      "loss": 0.0602,
      "step": 164275
    },
    {
      "epoch": 2.419699268052017,
      "grad_norm": 1.453350305557251,
      "learning_rate": 1.1472538405650067e-05,
      "loss": 0.0577,
      "step": 164300
    },
    {
      "epoch": 2.420067451142104,
      "grad_norm": 1.313693881034851,
      "learning_rate": 1.1470902033683083e-05,
      "loss": 0.0561,
      "step": 164325
    },
    {
      "epoch": 2.420435634232191,
      "grad_norm": 1.944935917854309,
      "learning_rate": 1.1469265661716097e-05,
      "loss": 0.0668,
      "step": 164350
    },
    {
      "epoch": 2.420803817322278,
      "grad_norm": 1.3409377336502075,
      "learning_rate": 1.1467629289749112e-05,
      "loss": 0.0585,
      "step": 164375
    },
    {
      "epoch": 2.421172000412365,
      "grad_norm": 1.2738739252090454,
      "learning_rate": 1.1465992917782128e-05,
      "loss": 0.0526,
      "step": 164400
    },
    {
      "epoch": 2.421540183502452,
      "grad_norm": 1.3180350065231323,
      "learning_rate": 1.1464356545815142e-05,
      "loss": 0.0562,
      "step": 164425
    },
    {
      "epoch": 2.421908366592539,
      "grad_norm": 1.4948314428329468,
      "learning_rate": 1.1462720173848158e-05,
      "loss": 0.0628,
      "step": 164450
    },
    {
      "epoch": 2.422276549682626,
      "grad_norm": 2.0097382068634033,
      "learning_rate": 1.1461083801881175e-05,
      "loss": 0.0683,
      "step": 164475
    },
    {
      "epoch": 2.422644732772713,
      "grad_norm": 1.7877976894378662,
      "learning_rate": 1.145944742991419e-05,
      "loss": 0.0582,
      "step": 164500
    },
    {
      "epoch": 2.4230129158628,
      "grad_norm": 1.1123054027557373,
      "learning_rate": 1.1457811057947205e-05,
      "loss": 0.0641,
      "step": 164525
    },
    {
      "epoch": 2.4233810989528872,
      "grad_norm": 1.3095401525497437,
      "learning_rate": 1.1456174685980221e-05,
      "loss": 0.0645,
      "step": 164550
    },
    {
      "epoch": 2.4237492820429742,
      "grad_norm": 1.6199350357055664,
      "learning_rate": 1.1454538314013234e-05,
      "loss": 0.0572,
      "step": 164575
    },
    {
      "epoch": 2.4241174651330613,
      "grad_norm": 1.5714999437332153,
      "learning_rate": 1.145290194204625e-05,
      "loss": 0.0604,
      "step": 164600
    },
    {
      "epoch": 2.4244856482231483,
      "grad_norm": 0.8159253597259521,
      "learning_rate": 1.1451265570079266e-05,
      "loss": 0.0546,
      "step": 164625
    },
    {
      "epoch": 2.4248538313132353,
      "grad_norm": 1.240932583808899,
      "learning_rate": 1.1449629198112283e-05,
      "loss": 0.0572,
      "step": 164650
    },
    {
      "epoch": 2.4252220144033223,
      "grad_norm": 1.8340493440628052,
      "learning_rate": 1.1447992826145297e-05,
      "loss": 0.0666,
      "step": 164675
    },
    {
      "epoch": 2.4255901974934098,
      "grad_norm": 1.129704236984253,
      "learning_rate": 1.1446356454178313e-05,
      "loss": 0.0636,
      "step": 164700
    },
    {
      "epoch": 2.4259583805834968,
      "grad_norm": 1.0799140930175781,
      "learning_rate": 1.144472008221133e-05,
      "loss": 0.0583,
      "step": 164725
    },
    {
      "epoch": 2.426326563673584,
      "grad_norm": 1.3294018507003784,
      "learning_rate": 1.1443083710244345e-05,
      "loss": 0.056,
      "step": 164750
    },
    {
      "epoch": 2.426694746763671,
      "grad_norm": 1.4548373222351074,
      "learning_rate": 1.144144733827736e-05,
      "loss": 0.0635,
      "step": 164775
    },
    {
      "epoch": 2.427062929853758,
      "grad_norm": 1.2410050630569458,
      "learning_rate": 1.1439810966310374e-05,
      "loss": 0.0563,
      "step": 164800
    },
    {
      "epoch": 2.427431112943845,
      "grad_norm": 1.2411009073257446,
      "learning_rate": 1.143817459434339e-05,
      "loss": 0.0623,
      "step": 164825
    },
    {
      "epoch": 2.427799296033932,
      "grad_norm": 1.343772530555725,
      "learning_rate": 1.1436538222376405e-05,
      "loss": 0.0569,
      "step": 164850
    },
    {
      "epoch": 2.428167479124019,
      "grad_norm": 1.093827724456787,
      "learning_rate": 1.1434901850409421e-05,
      "loss": 0.0626,
      "step": 164875
    },
    {
      "epoch": 2.428535662214106,
      "grad_norm": 1.6780364513397217,
      "learning_rate": 1.1433265478442437e-05,
      "loss": 0.0639,
      "step": 164900
    },
    {
      "epoch": 2.428903845304193,
      "grad_norm": 1.2943000793457031,
      "learning_rate": 1.1431629106475452e-05,
      "loss": 0.0705,
      "step": 164925
    },
    {
      "epoch": 2.42927202839428,
      "grad_norm": 1.2445746660232544,
      "learning_rate": 1.1429992734508468e-05,
      "loss": 0.0588,
      "step": 164950
    },
    {
      "epoch": 2.429640211484367,
      "grad_norm": 1.9415868520736694,
      "learning_rate": 1.1428356362541484e-05,
      "loss": 0.057,
      "step": 164975
    },
    {
      "epoch": 2.430008394574454,
      "grad_norm": 1.4405913352966309,
      "learning_rate": 1.1426719990574497e-05,
      "loss": 0.0604,
      "step": 165000
    },
    {
      "epoch": 2.430376577664541,
      "grad_norm": 1.8014392852783203,
      "learning_rate": 1.1425083618607513e-05,
      "loss": 0.0629,
      "step": 165025
    },
    {
      "epoch": 2.430744760754628,
      "grad_norm": 1.5329076051712036,
      "learning_rate": 1.1423447246640529e-05,
      "loss": 0.0619,
      "step": 165050
    },
    {
      "epoch": 2.431112943844715,
      "grad_norm": 1.0596028566360474,
      "learning_rate": 1.1421810874673545e-05,
      "loss": 0.056,
      "step": 165075
    },
    {
      "epoch": 2.431481126934802,
      "grad_norm": 1.4429690837860107,
      "learning_rate": 1.142017450270656e-05,
      "loss": 0.0613,
      "step": 165100
    },
    {
      "epoch": 2.431849310024889,
      "grad_norm": 1.3741044998168945,
      "learning_rate": 1.1418538130739576e-05,
      "loss": 0.0546,
      "step": 165125
    },
    {
      "epoch": 2.432217493114976,
      "grad_norm": 1.1319230794906616,
      "learning_rate": 1.1416901758772592e-05,
      "loss": 0.0618,
      "step": 165150
    },
    {
      "epoch": 2.432585676205063,
      "grad_norm": 0.992778480052948,
      "learning_rate": 1.1415265386805608e-05,
      "loss": 0.0543,
      "step": 165175
    },
    {
      "epoch": 2.43295385929515,
      "grad_norm": 1.1014723777770996,
      "learning_rate": 1.1413629014838621e-05,
      "loss": 0.0591,
      "step": 165200
    },
    {
      "epoch": 2.433322042385237,
      "grad_norm": 1.273636817932129,
      "learning_rate": 1.1411992642871637e-05,
      "loss": 0.0554,
      "step": 165225
    },
    {
      "epoch": 2.4336902254753245,
      "grad_norm": 1.494178295135498,
      "learning_rate": 1.1410356270904652e-05,
      "loss": 0.0603,
      "step": 165250
    },
    {
      "epoch": 2.4340584085654116,
      "grad_norm": 1.1151559352874756,
      "learning_rate": 1.1408719898937668e-05,
      "loss": 0.062,
      "step": 165275
    },
    {
      "epoch": 2.4344265916554986,
      "grad_norm": 1.3149724006652832,
      "learning_rate": 1.1407083526970684e-05,
      "loss": 0.0622,
      "step": 165300
    },
    {
      "epoch": 2.4347947747455856,
      "grad_norm": 1.5353071689605713,
      "learning_rate": 1.14054471550037e-05,
      "loss": 0.0586,
      "step": 165325
    },
    {
      "epoch": 2.4351629578356726,
      "grad_norm": 1.3900264501571655,
      "learning_rate": 1.1403810783036714e-05,
      "loss": 0.0565,
      "step": 165350
    },
    {
      "epoch": 2.4355311409257596,
      "grad_norm": 1.4981166124343872,
      "learning_rate": 1.140217441106973e-05,
      "loss": 0.0581,
      "step": 165375
    },
    {
      "epoch": 2.4358993240158466,
      "grad_norm": 1.3365064859390259,
      "learning_rate": 1.1400538039102747e-05,
      "loss": 0.0585,
      "step": 165400
    },
    {
      "epoch": 2.4362675071059336,
      "grad_norm": 1.0220973491668701,
      "learning_rate": 1.139890166713576e-05,
      "loss": 0.0643,
      "step": 165425
    },
    {
      "epoch": 2.4366356901960207,
      "grad_norm": 1.3133913278579712,
      "learning_rate": 1.1397265295168776e-05,
      "loss": 0.0573,
      "step": 165450
    },
    {
      "epoch": 2.4370038732861077,
      "grad_norm": 2.3548390865325928,
      "learning_rate": 1.1395628923201792e-05,
      "loss": 0.0658,
      "step": 165475
    },
    {
      "epoch": 2.4373720563761947,
      "grad_norm": 1.3287529945373535,
      "learning_rate": 1.1393992551234806e-05,
      "loss": 0.0611,
      "step": 165500
    },
    {
      "epoch": 2.4377402394662817,
      "grad_norm": 1.0778312683105469,
      "learning_rate": 1.1392356179267822e-05,
      "loss": 0.0604,
      "step": 165525
    },
    {
      "epoch": 2.4381084225563687,
      "grad_norm": 1.2553143501281738,
      "learning_rate": 1.1390719807300839e-05,
      "loss": 0.0574,
      "step": 165550
    },
    {
      "epoch": 2.4384766056464557,
      "grad_norm": 1.6024491786956787,
      "learning_rate": 1.1389083435333855e-05,
      "loss": 0.0612,
      "step": 165575
    },
    {
      "epoch": 2.4388447887365428,
      "grad_norm": 1.249430775642395,
      "learning_rate": 1.138744706336687e-05,
      "loss": 0.0572,
      "step": 165600
    },
    {
      "epoch": 2.4392129718266298,
      "grad_norm": 1.5314658880233765,
      "learning_rate": 1.1385810691399884e-05,
      "loss": 0.0632,
      "step": 165625
    },
    {
      "epoch": 2.4395811549167172,
      "grad_norm": 1.287026286125183,
      "learning_rate": 1.13841743194329e-05,
      "loss": 0.0534,
      "step": 165650
    },
    {
      "epoch": 2.4399493380068042,
      "grad_norm": 1.6620677709579468,
      "learning_rate": 1.1382537947465914e-05,
      "loss": 0.0626,
      "step": 165675
    },
    {
      "epoch": 2.4403175210968913,
      "grad_norm": 1.3048840761184692,
      "learning_rate": 1.138090157549893e-05,
      "loss": 0.0637,
      "step": 165700
    },
    {
      "epoch": 2.4406857041869783,
      "grad_norm": 1.2793484926223755,
      "learning_rate": 1.1379265203531947e-05,
      "loss": 0.0563,
      "step": 165725
    },
    {
      "epoch": 2.4410538872770653,
      "grad_norm": 1.4037551879882812,
      "learning_rate": 1.1377628831564963e-05,
      "loss": 0.0549,
      "step": 165750
    },
    {
      "epoch": 2.4414220703671523,
      "grad_norm": 1.3582558631896973,
      "learning_rate": 1.1375992459597977e-05,
      "loss": 0.056,
      "step": 165775
    },
    {
      "epoch": 2.4417902534572393,
      "grad_norm": 1.780892014503479,
      "learning_rate": 1.1374356087630993e-05,
      "loss": 0.0642,
      "step": 165800
    },
    {
      "epoch": 2.4421584365473263,
      "grad_norm": 1.6464052200317383,
      "learning_rate": 1.137271971566401e-05,
      "loss": 0.0579,
      "step": 165825
    },
    {
      "epoch": 2.4425266196374134,
      "grad_norm": 1.0196897983551025,
      "learning_rate": 1.1371083343697022e-05,
      "loss": 0.0546,
      "step": 165850
    },
    {
      "epoch": 2.4428948027275004,
      "grad_norm": 1.2940099239349365,
      "learning_rate": 1.1369446971730038e-05,
      "loss": 0.0604,
      "step": 165875
    },
    {
      "epoch": 2.4432629858175874,
      "grad_norm": 1.2686991691589355,
      "learning_rate": 1.1367810599763055e-05,
      "loss": 0.0661,
      "step": 165900
    },
    {
      "epoch": 2.4436311689076744,
      "grad_norm": 0.5272985696792603,
      "learning_rate": 1.1366174227796069e-05,
      "loss": 0.0554,
      "step": 165925
    },
    {
      "epoch": 2.4439993519977614,
      "grad_norm": 1.1308406591415405,
      "learning_rate": 1.1364537855829085e-05,
      "loss": 0.0534,
      "step": 165950
    },
    {
      "epoch": 2.4443675350878484,
      "grad_norm": 1.2716814279556274,
      "learning_rate": 1.1362901483862101e-05,
      "loss": 0.0632,
      "step": 165975
    },
    {
      "epoch": 2.4447357181779354,
      "grad_norm": 1.2231541872024536,
      "learning_rate": 1.1361265111895117e-05,
      "loss": 0.0507,
      "step": 166000
    },
    {
      "epoch": 2.4451039012680225,
      "grad_norm": 1.2079716920852661,
      "learning_rate": 1.1359628739928132e-05,
      "loss": 0.0593,
      "step": 166025
    },
    {
      "epoch": 2.4454720843581095,
      "grad_norm": 1.2885600328445435,
      "learning_rate": 1.1357992367961146e-05,
      "loss": 0.0641,
      "step": 166050
    },
    {
      "epoch": 2.4458402674481965,
      "grad_norm": 1.2220584154129028,
      "learning_rate": 1.135635599599416e-05,
      "loss": 0.0544,
      "step": 166075
    },
    {
      "epoch": 2.4462084505382835,
      "grad_norm": 1.1342827081680298,
      "learning_rate": 1.1354719624027177e-05,
      "loss": 0.06,
      "step": 166100
    },
    {
      "epoch": 2.4465766336283705,
      "grad_norm": 1.4868804216384888,
      "learning_rate": 1.1353083252060193e-05,
      "loss": 0.0654,
      "step": 166125
    },
    {
      "epoch": 2.4469448167184575,
      "grad_norm": 1.4469658136367798,
      "learning_rate": 1.135144688009321e-05,
      "loss": 0.0593,
      "step": 166150
    },
    {
      "epoch": 2.4473129998085446,
      "grad_norm": 1.7173411846160889,
      "learning_rate": 1.1349810508126224e-05,
      "loss": 0.0649,
      "step": 166175
    },
    {
      "epoch": 2.447681182898632,
      "grad_norm": 1.4442107677459717,
      "learning_rate": 1.134817413615924e-05,
      "loss": 0.0578,
      "step": 166200
    },
    {
      "epoch": 2.448049365988719,
      "grad_norm": 1.5394749641418457,
      "learning_rate": 1.1346537764192256e-05,
      "loss": 0.0544,
      "step": 166225
    },
    {
      "epoch": 2.448417549078806,
      "grad_norm": 1.4512736797332764,
      "learning_rate": 1.1344901392225272e-05,
      "loss": 0.0584,
      "step": 166250
    },
    {
      "epoch": 2.448785732168893,
      "grad_norm": 1.2789298295974731,
      "learning_rate": 1.1343265020258285e-05,
      "loss": 0.0526,
      "step": 166275
    },
    {
      "epoch": 2.44915391525898,
      "grad_norm": 0.9591279625892639,
      "learning_rate": 1.1341628648291301e-05,
      "loss": 0.0586,
      "step": 166300
    },
    {
      "epoch": 2.449522098349067,
      "grad_norm": 1.3781404495239258,
      "learning_rate": 1.1339992276324317e-05,
      "loss": 0.0604,
      "step": 166325
    },
    {
      "epoch": 2.449890281439154,
      "grad_norm": 1.503045678138733,
      "learning_rate": 1.1338355904357332e-05,
      "loss": 0.0593,
      "step": 166350
    },
    {
      "epoch": 2.450258464529241,
      "grad_norm": 1.400524616241455,
      "learning_rate": 1.1336719532390348e-05,
      "loss": 0.065,
      "step": 166375
    },
    {
      "epoch": 2.450626647619328,
      "grad_norm": 1.530967354774475,
      "learning_rate": 1.1335083160423364e-05,
      "loss": 0.0679,
      "step": 166400
    },
    {
      "epoch": 2.450994830709415,
      "grad_norm": 0.9463401436805725,
      "learning_rate": 1.1333446788456378e-05,
      "loss": 0.0574,
      "step": 166425
    },
    {
      "epoch": 2.451363013799502,
      "grad_norm": 1.41310715675354,
      "learning_rate": 1.1331810416489395e-05,
      "loss": 0.057,
      "step": 166450
    },
    {
      "epoch": 2.451731196889589,
      "grad_norm": 1.444229006767273,
      "learning_rate": 1.1330174044522409e-05,
      "loss": 0.061,
      "step": 166475
    },
    {
      "epoch": 2.452099379979676,
      "grad_norm": 2.0020642280578613,
      "learning_rate": 1.1328537672555424e-05,
      "loss": 0.065,
      "step": 166500
    },
    {
      "epoch": 2.452467563069763,
      "grad_norm": 1.286094069480896,
      "learning_rate": 1.132690130058844e-05,
      "loss": 0.0589,
      "step": 166525
    },
    {
      "epoch": 2.4528357461598502,
      "grad_norm": 1.7219420671463013,
      "learning_rate": 1.1325330383500135e-05,
      "loss": 0.0585,
      "step": 166550
    },
    {
      "epoch": 2.4532039292499372,
      "grad_norm": 1.1732163429260254,
      "learning_rate": 1.1323694011533151e-05,
      "loss": 0.0568,
      "step": 166575
    },
    {
      "epoch": 2.4535721123400243,
      "grad_norm": 1.4670276641845703,
      "learning_rate": 1.1322057639566166e-05,
      "loss": 0.063,
      "step": 166600
    },
    {
      "epoch": 2.4539402954301117,
      "grad_norm": 1.4043080806732178,
      "learning_rate": 1.1320421267599182e-05,
      "loss": 0.0589,
      "step": 166625
    },
    {
      "epoch": 2.4543084785201987,
      "grad_norm": 0.8559965491294861,
      "learning_rate": 1.1318784895632198e-05,
      "loss": 0.0612,
      "step": 166650
    },
    {
      "epoch": 2.4546766616102857,
      "grad_norm": 1.137347936630249,
      "learning_rate": 1.1317148523665211e-05,
      "loss": 0.0626,
      "step": 166675
    },
    {
      "epoch": 2.4550448447003728,
      "grad_norm": 0.715144157409668,
      "learning_rate": 1.1315512151698227e-05,
      "loss": 0.0508,
      "step": 166700
    },
    {
      "epoch": 2.45541302779046,
      "grad_norm": 1.409767746925354,
      "learning_rate": 1.1313875779731243e-05,
      "loss": 0.0688,
      "step": 166725
    },
    {
      "epoch": 2.455781210880547,
      "grad_norm": 1.1211822032928467,
      "learning_rate": 1.1312239407764258e-05,
      "loss": 0.0679,
      "step": 166750
    },
    {
      "epoch": 2.456149393970634,
      "grad_norm": 1.1836835145950317,
      "learning_rate": 1.1310603035797274e-05,
      "loss": 0.0549,
      "step": 166775
    },
    {
      "epoch": 2.456517577060721,
      "grad_norm": 0.9444347023963928,
      "learning_rate": 1.130896666383029e-05,
      "loss": 0.0564,
      "step": 166800
    },
    {
      "epoch": 2.456885760150808,
      "grad_norm": 1.400810956954956,
      "learning_rate": 1.1307330291863306e-05,
      "loss": 0.0552,
      "step": 166825
    },
    {
      "epoch": 2.457253943240895,
      "grad_norm": 1.3926701545715332,
      "learning_rate": 1.130569391989632e-05,
      "loss": 0.0646,
      "step": 166850
    },
    {
      "epoch": 2.457622126330982,
      "grad_norm": 1.5172147750854492,
      "learning_rate": 1.1304057547929337e-05,
      "loss": 0.0562,
      "step": 166875
    },
    {
      "epoch": 2.457990309421069,
      "grad_norm": 1.601259708404541,
      "learning_rate": 1.130242117596235e-05,
      "loss": 0.0536,
      "step": 166900
    },
    {
      "epoch": 2.458358492511156,
      "grad_norm": 1.2095301151275635,
      "learning_rate": 1.1300784803995366e-05,
      "loss": 0.0563,
      "step": 166925
    },
    {
      "epoch": 2.458726675601243,
      "grad_norm": 1.4847033023834229,
      "learning_rate": 1.1299148432028382e-05,
      "loss": 0.0651,
      "step": 166950
    },
    {
      "epoch": 2.45909485869133,
      "grad_norm": 1.2221888303756714,
      "learning_rate": 1.1297512060061398e-05,
      "loss": 0.0614,
      "step": 166975
    },
    {
      "epoch": 2.459463041781417,
      "grad_norm": 1.0640538930892944,
      "learning_rate": 1.1295875688094412e-05,
      "loss": 0.0684,
      "step": 167000
    },
    {
      "epoch": 2.459831224871504,
      "grad_norm": 1.0552587509155273,
      "learning_rate": 1.1294239316127429e-05,
      "loss": 0.065,
      "step": 167025
    },
    {
      "epoch": 2.460199407961591,
      "grad_norm": 1.2368358373641968,
      "learning_rate": 1.1292602944160445e-05,
      "loss": 0.063,
      "step": 167050
    },
    {
      "epoch": 2.460567591051678,
      "grad_norm": 1.5190926790237427,
      "learning_rate": 1.1290966572193461e-05,
      "loss": 0.0612,
      "step": 167075
    },
    {
      "epoch": 2.460935774141765,
      "grad_norm": 1.2478116750717163,
      "learning_rate": 1.1289330200226474e-05,
      "loss": 0.0551,
      "step": 167100
    },
    {
      "epoch": 2.461303957231852,
      "grad_norm": 1.3048303127288818,
      "learning_rate": 1.128769382825949e-05,
      "loss": 0.0613,
      "step": 167125
    },
    {
      "epoch": 2.4616721403219395,
      "grad_norm": 2.1593852043151855,
      "learning_rate": 1.1286057456292506e-05,
      "loss": 0.0627,
      "step": 167150
    },
    {
      "epoch": 2.4620403234120265,
      "grad_norm": 1.3679306507110596,
      "learning_rate": 1.128442108432552e-05,
      "loss": 0.0598,
      "step": 167175
    },
    {
      "epoch": 2.4624085065021135,
      "grad_norm": 1.7907465696334839,
      "learning_rate": 1.1282784712358537e-05,
      "loss": 0.0606,
      "step": 167200
    },
    {
      "epoch": 2.4627766895922005,
      "grad_norm": 1.0296827554702759,
      "learning_rate": 1.1281148340391553e-05,
      "loss": 0.0631,
      "step": 167225
    },
    {
      "epoch": 2.4631448726822875,
      "grad_norm": 1.2811992168426514,
      "learning_rate": 1.1279511968424567e-05,
      "loss": 0.059,
      "step": 167250
    },
    {
      "epoch": 2.4635130557723746,
      "grad_norm": 1.2746224403381348,
      "learning_rate": 1.1277875596457583e-05,
      "loss": 0.0709,
      "step": 167275
    },
    {
      "epoch": 2.4638812388624616,
      "grad_norm": 1.7382864952087402,
      "learning_rate": 1.12762392244906e-05,
      "loss": 0.06,
      "step": 167300
    },
    {
      "epoch": 2.4642494219525486,
      "grad_norm": 1.1114996671676636,
      "learning_rate": 1.1274602852523612e-05,
      "loss": 0.0586,
      "step": 167325
    },
    {
      "epoch": 2.4646176050426356,
      "grad_norm": 1.0237200260162354,
      "learning_rate": 1.1272966480556628e-05,
      "loss": 0.0596,
      "step": 167350
    },
    {
      "epoch": 2.4649857881327226,
      "grad_norm": 1.5585252046585083,
      "learning_rate": 1.1271330108589645e-05,
      "loss": 0.0589,
      "step": 167375
    },
    {
      "epoch": 2.4653539712228096,
      "grad_norm": 1.7459185123443604,
      "learning_rate": 1.126969373662266e-05,
      "loss": 0.0593,
      "step": 167400
    },
    {
      "epoch": 2.4657221543128967,
      "grad_norm": 1.231163501739502,
      "learning_rate": 1.1268057364655675e-05,
      "loss": 0.0489,
      "step": 167425
    },
    {
      "epoch": 2.4660903374029837,
      "grad_norm": 1.2625170946121216,
      "learning_rate": 1.1266420992688691e-05,
      "loss": 0.0606,
      "step": 167450
    },
    {
      "epoch": 2.4664585204930707,
      "grad_norm": 1.569287657737732,
      "learning_rate": 1.1264784620721707e-05,
      "loss": 0.0607,
      "step": 167475
    },
    {
      "epoch": 2.4668267035831577,
      "grad_norm": 1.0937808752059937,
      "learning_rate": 1.1263148248754724e-05,
      "loss": 0.0587,
      "step": 167500
    },
    {
      "epoch": 2.4671948866732447,
      "grad_norm": 1.7669326066970825,
      "learning_rate": 1.1261511876787736e-05,
      "loss": 0.0674,
      "step": 167525
    },
    {
      "epoch": 2.4675630697633317,
      "grad_norm": 1.7553832530975342,
      "learning_rate": 1.1259875504820753e-05,
      "loss": 0.0643,
      "step": 167550
    },
    {
      "epoch": 2.467931252853419,
      "grad_norm": 1.7348791360855103,
      "learning_rate": 1.1258239132853767e-05,
      "loss": 0.0598,
      "step": 167575
    },
    {
      "epoch": 2.468299435943506,
      "grad_norm": 1.446185827255249,
      "learning_rate": 1.1256602760886783e-05,
      "loss": 0.0569,
      "step": 167600
    },
    {
      "epoch": 2.468667619033593,
      "grad_norm": 1.5019747018814087,
      "learning_rate": 1.12549663889198e-05,
      "loss": 0.0595,
      "step": 167625
    },
    {
      "epoch": 2.4690358021236802,
      "grad_norm": 1.207907795906067,
      "learning_rate": 1.1253330016952815e-05,
      "loss": 0.0568,
      "step": 167650
    },
    {
      "epoch": 2.4694039852137672,
      "grad_norm": 1.4958751201629639,
      "learning_rate": 1.125169364498583e-05,
      "loss": 0.0669,
      "step": 167675
    },
    {
      "epoch": 2.4697721683038543,
      "grad_norm": 1.3696891069412231,
      "learning_rate": 1.1250057273018846e-05,
      "loss": 0.0614,
      "step": 167700
    },
    {
      "epoch": 2.4701403513939413,
      "grad_norm": 1.3949445486068726,
      "learning_rate": 1.1248420901051862e-05,
      "loss": 0.0557,
      "step": 167725
    },
    {
      "epoch": 2.4705085344840283,
      "grad_norm": 1.7682507038116455,
      "learning_rate": 1.1246784529084875e-05,
      "loss": 0.0569,
      "step": 167750
    },
    {
      "epoch": 2.4708767175741153,
      "grad_norm": 1.3156243562698364,
      "learning_rate": 1.1245148157117891e-05,
      "loss": 0.0528,
      "step": 167775
    },
    {
      "epoch": 2.4712449006642023,
      "grad_norm": 1.1397193670272827,
      "learning_rate": 1.1243511785150907e-05,
      "loss": 0.0695,
      "step": 167800
    },
    {
      "epoch": 2.4716130837542893,
      "grad_norm": 1.5308188199996948,
      "learning_rate": 1.1241875413183922e-05,
      "loss": 0.0604,
      "step": 167825
    },
    {
      "epoch": 2.4719812668443764,
      "grad_norm": 1.0728814601898193,
      "learning_rate": 1.1240239041216938e-05,
      "loss": 0.0652,
      "step": 167850
    },
    {
      "epoch": 2.4723494499344634,
      "grad_norm": 1.5396006107330322,
      "learning_rate": 1.1238602669249954e-05,
      "loss": 0.0592,
      "step": 167875
    },
    {
      "epoch": 2.4727176330245504,
      "grad_norm": 1.8049131631851196,
      "learning_rate": 1.123696629728297e-05,
      "loss": 0.0626,
      "step": 167900
    },
    {
      "epoch": 2.4730858161146374,
      "grad_norm": 1.6638224124908447,
      "learning_rate": 1.1235329925315985e-05,
      "loss": 0.0599,
      "step": 167925
    },
    {
      "epoch": 2.4734539992047244,
      "grad_norm": 1.2251381874084473,
      "learning_rate": 1.1233693553348999e-05,
      "loss": 0.0556,
      "step": 167950
    },
    {
      "epoch": 2.4738221822948114,
      "grad_norm": 1.7455753087997437,
      "learning_rate": 1.1232057181382015e-05,
      "loss": 0.0598,
      "step": 167975
    },
    {
      "epoch": 2.4741903653848984,
      "grad_norm": 1.4482413530349731,
      "learning_rate": 1.123042080941503e-05,
      "loss": 0.059,
      "step": 168000
    },
    {
      "epoch": 2.4745585484749855,
      "grad_norm": 1.4696309566497803,
      "learning_rate": 1.1228784437448046e-05,
      "loss": 0.0654,
      "step": 168025
    },
    {
      "epoch": 2.4749267315650725,
      "grad_norm": 1.1336841583251953,
      "learning_rate": 1.1227148065481062e-05,
      "loss": 0.0552,
      "step": 168050
    },
    {
      "epoch": 2.4752949146551595,
      "grad_norm": 1.566132664680481,
      "learning_rate": 1.1225511693514078e-05,
      "loss": 0.0559,
      "step": 168075
    },
    {
      "epoch": 2.4756630977452465,
      "grad_norm": 1.8225576877593994,
      "learning_rate": 1.1223875321547093e-05,
      "loss": 0.058,
      "step": 168100
    },
    {
      "epoch": 2.476031280835334,
      "grad_norm": 1.896125078201294,
      "learning_rate": 1.1222238949580109e-05,
      "loss": 0.0632,
      "step": 168125
    },
    {
      "epoch": 2.476399463925421,
      "grad_norm": 1.2454347610473633,
      "learning_rate": 1.1220602577613122e-05,
      "loss": 0.0575,
      "step": 168150
    },
    {
      "epoch": 2.476767647015508,
      "grad_norm": 1.1391030550003052,
      "learning_rate": 1.1218966205646138e-05,
      "loss": 0.0587,
      "step": 168175
    },
    {
      "epoch": 2.477135830105595,
      "grad_norm": 1.3579012155532837,
      "learning_rate": 1.1217329833679154e-05,
      "loss": 0.0595,
      "step": 168200
    },
    {
      "epoch": 2.477504013195682,
      "grad_norm": 1.5360424518585205,
      "learning_rate": 1.121569346171217e-05,
      "loss": 0.0605,
      "step": 168225
    },
    {
      "epoch": 2.477872196285769,
      "grad_norm": 1.5696964263916016,
      "learning_rate": 1.1214057089745184e-05,
      "loss": 0.0562,
      "step": 168250
    },
    {
      "epoch": 2.478240379375856,
      "grad_norm": 0.9135528802871704,
      "learning_rate": 1.12124207177782e-05,
      "loss": 0.056,
      "step": 168275
    },
    {
      "epoch": 2.478608562465943,
      "grad_norm": 1.3621424436569214,
      "learning_rate": 1.1210784345811217e-05,
      "loss": 0.0658,
      "step": 168300
    },
    {
      "epoch": 2.47897674555603,
      "grad_norm": 1.819261908531189,
      "learning_rate": 1.1209147973844233e-05,
      "loss": 0.0595,
      "step": 168325
    },
    {
      "epoch": 2.479344928646117,
      "grad_norm": 1.3147543668746948,
      "learning_rate": 1.1207511601877247e-05,
      "loss": 0.0565,
      "step": 168350
    },
    {
      "epoch": 2.479713111736204,
      "grad_norm": 1.3996645212173462,
      "learning_rate": 1.1205875229910262e-05,
      "loss": 0.0694,
      "step": 168375
    },
    {
      "epoch": 2.480081294826291,
      "grad_norm": 1.4199289083480835,
      "learning_rate": 1.1204238857943276e-05,
      "loss": 0.0585,
      "step": 168400
    },
    {
      "epoch": 2.480449477916378,
      "grad_norm": 1.6981226205825806,
      "learning_rate": 1.1202602485976292e-05,
      "loss": 0.0599,
      "step": 168425
    },
    {
      "epoch": 2.480817661006465,
      "grad_norm": 1.2532683610916138,
      "learning_rate": 1.1200966114009309e-05,
      "loss": 0.0592,
      "step": 168450
    },
    {
      "epoch": 2.481185844096552,
      "grad_norm": 1.2565698623657227,
      "learning_rate": 1.1199329742042325e-05,
      "loss": 0.0608,
      "step": 168475
    },
    {
      "epoch": 2.481554027186639,
      "grad_norm": 1.2574912309646606,
      "learning_rate": 1.119769337007534e-05,
      "loss": 0.0611,
      "step": 168500
    },
    {
      "epoch": 2.4819222102767267,
      "grad_norm": 1.2927576303482056,
      "learning_rate": 1.1196056998108355e-05,
      "loss": 0.0591,
      "step": 168525
    },
    {
      "epoch": 2.4822903933668137,
      "grad_norm": 1.5040113925933838,
      "learning_rate": 1.1194420626141371e-05,
      "loss": 0.0655,
      "step": 168550
    },
    {
      "epoch": 2.4826585764569007,
      "grad_norm": 1.3379175662994385,
      "learning_rate": 1.1192784254174384e-05,
      "loss": 0.0668,
      "step": 168575
    },
    {
      "epoch": 2.4830267595469877,
      "grad_norm": 1.7368855476379395,
      "learning_rate": 1.11911478822074e-05,
      "loss": 0.0651,
      "step": 168600
    },
    {
      "epoch": 2.4833949426370747,
      "grad_norm": 0.9847592711448669,
      "learning_rate": 1.1189511510240417e-05,
      "loss": 0.0595,
      "step": 168625
    },
    {
      "epoch": 2.4837631257271617,
      "grad_norm": 1.8552279472351074,
      "learning_rate": 1.1187875138273433e-05,
      "loss": 0.0628,
      "step": 168650
    },
    {
      "epoch": 2.4841313088172488,
      "grad_norm": 0.9592670798301697,
      "learning_rate": 1.1186238766306447e-05,
      "loss": 0.0563,
      "step": 168675
    },
    {
      "epoch": 2.4844994919073358,
      "grad_norm": 1.369267463684082,
      "learning_rate": 1.1184602394339463e-05,
      "loss": 0.0614,
      "step": 168700
    },
    {
      "epoch": 2.484867674997423,
      "grad_norm": 1.1809782981872559,
      "learning_rate": 1.118296602237248e-05,
      "loss": 0.0536,
      "step": 168725
    },
    {
      "epoch": 2.48523585808751,
      "grad_norm": 1.6489936113357544,
      "learning_rate": 1.1181329650405494e-05,
      "loss": 0.0556,
      "step": 168750
    },
    {
      "epoch": 2.485604041177597,
      "grad_norm": 1.207336664199829,
      "learning_rate": 1.117969327843851e-05,
      "loss": 0.0591,
      "step": 168775
    },
    {
      "epoch": 2.485972224267684,
      "grad_norm": 1.1333446502685547,
      "learning_rate": 1.1178056906471525e-05,
      "loss": 0.0529,
      "step": 168800
    },
    {
      "epoch": 2.486340407357771,
      "grad_norm": 0.8984009027481079,
      "learning_rate": 1.1176420534504539e-05,
      "loss": 0.0616,
      "step": 168825
    },
    {
      "epoch": 2.486708590447858,
      "grad_norm": 1.1152271032333374,
      "learning_rate": 1.1174784162537555e-05,
      "loss": 0.0648,
      "step": 168850
    },
    {
      "epoch": 2.487076773537945,
      "grad_norm": 1.551584243774414,
      "learning_rate": 1.1173147790570571e-05,
      "loss": 0.0607,
      "step": 168875
    },
    {
      "epoch": 2.487444956628032,
      "grad_norm": 1.3065776824951172,
      "learning_rate": 1.1171511418603587e-05,
      "loss": 0.0533,
      "step": 168900
    },
    {
      "epoch": 2.487813139718119,
      "grad_norm": 1.4792369604110718,
      "learning_rate": 1.1169875046636602e-05,
      "loss": 0.066,
      "step": 168925
    },
    {
      "epoch": 2.488181322808206,
      "grad_norm": 1.4735231399536133,
      "learning_rate": 1.1168238674669618e-05,
      "loss": 0.06,
      "step": 168950
    },
    {
      "epoch": 2.488549505898293,
      "grad_norm": 0.8678383827209473,
      "learning_rate": 1.1166602302702634e-05,
      "loss": 0.0623,
      "step": 168975
    },
    {
      "epoch": 2.48891768898838,
      "grad_norm": 1.2267992496490479,
      "learning_rate": 1.1164965930735647e-05,
      "loss": 0.0543,
      "step": 169000
    },
    {
      "epoch": 2.489285872078467,
      "grad_norm": 1.070836067199707,
      "learning_rate": 1.1163329558768663e-05,
      "loss": 0.0582,
      "step": 169025
    },
    {
      "epoch": 2.489654055168554,
      "grad_norm": 1.0685932636260986,
      "learning_rate": 1.116169318680168e-05,
      "loss": 0.0554,
      "step": 169050
    },
    {
      "epoch": 2.4900222382586414,
      "grad_norm": 1.4122798442840576,
      "learning_rate": 1.1160056814834694e-05,
      "loss": 0.0612,
      "step": 169075
    },
    {
      "epoch": 2.4903904213487285,
      "grad_norm": 1.4628891944885254,
      "learning_rate": 1.115842044286771e-05,
      "loss": 0.0583,
      "step": 169100
    },
    {
      "epoch": 2.4907586044388155,
      "grad_norm": 1.6943978071212769,
      "learning_rate": 1.1156784070900726e-05,
      "loss": 0.0501,
      "step": 169125
    },
    {
      "epoch": 2.4911267875289025,
      "grad_norm": 1.1437104940414429,
      "learning_rate": 1.1155147698933742e-05,
      "loss": 0.0673,
      "step": 169150
    },
    {
      "epoch": 2.4914949706189895,
      "grad_norm": 1.5388097763061523,
      "learning_rate": 1.1153511326966757e-05,
      "loss": 0.0578,
      "step": 169175
    },
    {
      "epoch": 2.4918631537090765,
      "grad_norm": 1.300085186958313,
      "learning_rate": 1.1151874954999773e-05,
      "loss": 0.0583,
      "step": 169200
    },
    {
      "epoch": 2.4922313367991635,
      "grad_norm": 1.4053467512130737,
      "learning_rate": 1.1150238583032786e-05,
      "loss": 0.0626,
      "step": 169225
    },
    {
      "epoch": 2.4925995198892505,
      "grad_norm": 1.4101793766021729,
      "learning_rate": 1.1148602211065802e-05,
      "loss": 0.0663,
      "step": 169250
    },
    {
      "epoch": 2.4929677029793376,
      "grad_norm": 1.3437551259994507,
      "learning_rate": 1.1146965839098818e-05,
      "loss": 0.06,
      "step": 169275
    },
    {
      "epoch": 2.4933358860694246,
      "grad_norm": 1.3879871368408203,
      "learning_rate": 1.1145329467131834e-05,
      "loss": 0.0576,
      "step": 169300
    },
    {
      "epoch": 2.4937040691595116,
      "grad_norm": 1.4387868642807007,
      "learning_rate": 1.1143693095164848e-05,
      "loss": 0.0584,
      "step": 169325
    },
    {
      "epoch": 2.4940722522495986,
      "grad_norm": 1.3832757472991943,
      "learning_rate": 1.1142056723197865e-05,
      "loss": 0.0656,
      "step": 169350
    },
    {
      "epoch": 2.4944404353396856,
      "grad_norm": 1.2323750257492065,
      "learning_rate": 1.114042035123088e-05,
      "loss": 0.0608,
      "step": 169375
    },
    {
      "epoch": 2.4948086184297726,
      "grad_norm": 1.284777045249939,
      "learning_rate": 1.1138783979263897e-05,
      "loss": 0.0528,
      "step": 169400
    },
    {
      "epoch": 2.4951768015198597,
      "grad_norm": 1.9294309616088867,
      "learning_rate": 1.113714760729691e-05,
      "loss": 0.0682,
      "step": 169425
    },
    {
      "epoch": 2.4955449846099467,
      "grad_norm": 1.6158751249313354,
      "learning_rate": 1.1135511235329926e-05,
      "loss": 0.0629,
      "step": 169450
    },
    {
      "epoch": 2.4959131677000337,
      "grad_norm": 1.494902491569519,
      "learning_rate": 1.1133874863362942e-05,
      "loss": 0.0576,
      "step": 169475
    },
    {
      "epoch": 2.496281350790121,
      "grad_norm": 1.114361047744751,
      "learning_rate": 1.1132238491395956e-05,
      "loss": 0.0642,
      "step": 169500
    },
    {
      "epoch": 2.496649533880208,
      "grad_norm": 1.5320008993148804,
      "learning_rate": 1.1130602119428973e-05,
      "loss": 0.0625,
      "step": 169525
    },
    {
      "epoch": 2.497017716970295,
      "grad_norm": 1.7157726287841797,
      "learning_rate": 1.1128965747461989e-05,
      "loss": 0.0605,
      "step": 169550
    },
    {
      "epoch": 2.497385900060382,
      "grad_norm": 1.2340632677078247,
      "learning_rate": 1.1127329375495003e-05,
      "loss": 0.0582,
      "step": 169575
    },
    {
      "epoch": 2.497754083150469,
      "grad_norm": 0.5398813486099243,
      "learning_rate": 1.112569300352802e-05,
      "loss": 0.0602,
      "step": 169600
    },
    {
      "epoch": 2.498122266240556,
      "grad_norm": 1.3092273473739624,
      "learning_rate": 1.1124056631561034e-05,
      "loss": 0.0641,
      "step": 169625
    },
    {
      "epoch": 2.4984904493306432,
      "grad_norm": 1.3375014066696167,
      "learning_rate": 1.1122420259594048e-05,
      "loss": 0.0639,
      "step": 169650
    },
    {
      "epoch": 2.4988586324207303,
      "grad_norm": 1.0961357355117798,
      "learning_rate": 1.1120783887627064e-05,
      "loss": 0.0557,
      "step": 169675
    },
    {
      "epoch": 2.4992268155108173,
      "grad_norm": 1.2570444345474243,
      "learning_rate": 1.111914751566008e-05,
      "loss": 0.0522,
      "step": 169700
    },
    {
      "epoch": 2.4995949986009043,
      "grad_norm": 0.7900687456130981,
      "learning_rate": 1.1117511143693097e-05,
      "loss": 0.0534,
      "step": 169725
    },
    {
      "epoch": 2.4999631816909913,
      "grad_norm": 1.4797840118408203,
      "learning_rate": 1.1115874771726111e-05,
      "loss": 0.0582,
      "step": 169750
    },
    {
      "epoch": 2.5003313647810783,
      "grad_norm": 1.2982577085494995,
      "learning_rate": 1.1114238399759127e-05,
      "loss": 0.0588,
      "step": 169775
    },
    {
      "epoch": 2.5006995478711653,
      "grad_norm": 1.4716572761535645,
      "learning_rate": 1.1112602027792143e-05,
      "loss": 0.0634,
      "step": 169800
    },
    {
      "epoch": 2.5010677309612523,
      "grad_norm": 1.3682631254196167,
      "learning_rate": 1.111096565582516e-05,
      "loss": 0.0577,
      "step": 169825
    },
    {
      "epoch": 2.5014359140513394,
      "grad_norm": 1.2293293476104736,
      "learning_rate": 1.1109329283858172e-05,
      "loss": 0.0544,
      "step": 169850
    },
    {
      "epoch": 2.5018040971414264,
      "grad_norm": 0.9996431469917297,
      "learning_rate": 1.1107692911891189e-05,
      "loss": 0.0644,
      "step": 169875
    },
    {
      "epoch": 2.5021722802315134,
      "grad_norm": 1.39128839969635,
      "learning_rate": 1.1106056539924203e-05,
      "loss": 0.0578,
      "step": 169900
    },
    {
      "epoch": 2.5025404633216004,
      "grad_norm": 1.3627259731292725,
      "learning_rate": 1.1104485622835899e-05,
      "loss": 0.0621,
      "step": 169925
    },
    {
      "epoch": 2.5029086464116874,
      "grad_norm": 1.1350576877593994,
      "learning_rate": 1.1102849250868915e-05,
      "loss": 0.0619,
      "step": 169950
    },
    {
      "epoch": 2.5032768295017744,
      "grad_norm": 1.7639429569244385,
      "learning_rate": 1.1101212878901931e-05,
      "loss": 0.0592,
      "step": 169975
    },
    {
      "epoch": 2.5036450125918615,
      "grad_norm": 1.236167311668396,
      "learning_rate": 1.1099576506934945e-05,
      "loss": 0.0545,
      "step": 170000
    },
    {
      "epoch": 2.5040131956819485,
      "grad_norm": 0.9044005274772644,
      "learning_rate": 1.1097940134967961e-05,
      "loss": 0.0596,
      "step": 170025
    },
    {
      "epoch": 2.5043813787720355,
      "grad_norm": 1.596166729927063,
      "learning_rate": 1.1096303763000976e-05,
      "loss": 0.0638,
      "step": 170050
    },
    {
      "epoch": 2.504749561862123,
      "grad_norm": 1.2062618732452393,
      "learning_rate": 1.109466739103399e-05,
      "loss": 0.0594,
      "step": 170075
    },
    {
      "epoch": 2.50511774495221,
      "grad_norm": 1.2923988103866577,
      "learning_rate": 1.1093031019067007e-05,
      "loss": 0.0563,
      "step": 170100
    },
    {
      "epoch": 2.505485928042297,
      "grad_norm": 0.9151227474212646,
      "learning_rate": 1.1091394647100023e-05,
      "loss": 0.0602,
      "step": 170125
    },
    {
      "epoch": 2.505854111132384,
      "grad_norm": 1.2792390584945679,
      "learning_rate": 1.1089758275133037e-05,
      "loss": 0.0499,
      "step": 170150
    },
    {
      "epoch": 2.506222294222471,
      "grad_norm": 0.8770138025283813,
      "learning_rate": 1.1088121903166053e-05,
      "loss": 0.0576,
      "step": 170175
    },
    {
      "epoch": 2.506590477312558,
      "grad_norm": 1.378714680671692,
      "learning_rate": 1.108648553119907e-05,
      "loss": 0.0598,
      "step": 170200
    },
    {
      "epoch": 2.506958660402645,
      "grad_norm": 1.3941881656646729,
      "learning_rate": 1.1084849159232086e-05,
      "loss": 0.0584,
      "step": 170225
    },
    {
      "epoch": 2.507326843492732,
      "grad_norm": 1.348970651626587,
      "learning_rate": 1.10832127872651e-05,
      "loss": 0.0593,
      "step": 170250
    },
    {
      "epoch": 2.507695026582819,
      "grad_norm": 1.1845784187316895,
      "learning_rate": 1.1081576415298115e-05,
      "loss": 0.0588,
      "step": 170275
    },
    {
      "epoch": 2.508063209672906,
      "grad_norm": 1.4029450416564941,
      "learning_rate": 1.107994004333113e-05,
      "loss": 0.0561,
      "step": 170300
    },
    {
      "epoch": 2.508431392762993,
      "grad_norm": 0.9983207583427429,
      "learning_rate": 1.1078303671364145e-05,
      "loss": 0.0593,
      "step": 170325
    },
    {
      "epoch": 2.50879957585308,
      "grad_norm": 1.1089082956314087,
      "learning_rate": 1.1076667299397161e-05,
      "loss": 0.0662,
      "step": 170350
    },
    {
      "epoch": 2.509167758943167,
      "grad_norm": 1.2326053380966187,
      "learning_rate": 1.1075030927430177e-05,
      "loss": 0.0642,
      "step": 170375
    },
    {
      "epoch": 2.509535942033254,
      "grad_norm": 1.4079872369766235,
      "learning_rate": 1.1073394555463194e-05,
      "loss": 0.0551,
      "step": 170400
    },
    {
      "epoch": 2.5099041251233416,
      "grad_norm": 1.554484486579895,
      "learning_rate": 1.1071758183496208e-05,
      "loss": 0.0622,
      "step": 170425
    },
    {
      "epoch": 2.5102723082134286,
      "grad_norm": 1.2747653722763062,
      "learning_rate": 1.1070121811529224e-05,
      "loss": 0.0631,
      "step": 170450
    },
    {
      "epoch": 2.5106404913035156,
      "grad_norm": 2.409205913543701,
      "learning_rate": 1.1068485439562237e-05,
      "loss": 0.0681,
      "step": 170475
    },
    {
      "epoch": 2.5110086743936026,
      "grad_norm": 0.8650424480438232,
      "learning_rate": 1.1066849067595253e-05,
      "loss": 0.0619,
      "step": 170500
    },
    {
      "epoch": 2.5113768574836897,
      "grad_norm": 1.2267897129058838,
      "learning_rate": 1.106521269562827e-05,
      "loss": 0.0557,
      "step": 170525
    },
    {
      "epoch": 2.5117450405737767,
      "grad_norm": 1.2166086435317993,
      "learning_rate": 1.1063576323661285e-05,
      "loss": 0.0664,
      "step": 170550
    },
    {
      "epoch": 2.5121132236638637,
      "grad_norm": 0.9450547099113464,
      "learning_rate": 1.10619399516943e-05,
      "loss": 0.0574,
      "step": 170575
    },
    {
      "epoch": 2.5124814067539507,
      "grad_norm": 1.2446368932724,
      "learning_rate": 1.1060303579727316e-05,
      "loss": 0.0671,
      "step": 170600
    },
    {
      "epoch": 2.5128495898440377,
      "grad_norm": 1.5243786573410034,
      "learning_rate": 1.1058667207760332e-05,
      "loss": 0.0647,
      "step": 170625
    },
    {
      "epoch": 2.5132177729341247,
      "grad_norm": 1.6719969511032104,
      "learning_rate": 1.1057030835793348e-05,
      "loss": 0.0602,
      "step": 170650
    },
    {
      "epoch": 2.5135859560242118,
      "grad_norm": 1.5910475254058838,
      "learning_rate": 1.1055394463826363e-05,
      "loss": 0.0579,
      "step": 170675
    },
    {
      "epoch": 2.5139541391142988,
      "grad_norm": 1.9802005290985107,
      "learning_rate": 1.1053758091859377e-05,
      "loss": 0.0683,
      "step": 170700
    },
    {
      "epoch": 2.514322322204386,
      "grad_norm": 1.0029993057250977,
      "learning_rate": 1.1052121719892392e-05,
      "loss": 0.0549,
      "step": 170725
    },
    {
      "epoch": 2.514690505294473,
      "grad_norm": 1.3782914876937866,
      "learning_rate": 1.1050485347925408e-05,
      "loss": 0.0572,
      "step": 170750
    },
    {
      "epoch": 2.51505868838456,
      "grad_norm": 1.233562707901001,
      "learning_rate": 1.1048848975958424e-05,
      "loss": 0.0572,
      "step": 170775
    },
    {
      "epoch": 2.515426871474647,
      "grad_norm": 1.3055835962295532,
      "learning_rate": 1.104721260399144e-05,
      "loss": 0.0587,
      "step": 170800
    },
    {
      "epoch": 2.515795054564734,
      "grad_norm": 0.9142024517059326,
      "learning_rate": 1.1045576232024455e-05,
      "loss": 0.0573,
      "step": 170825
    },
    {
      "epoch": 2.516163237654821,
      "grad_norm": 1.1494581699371338,
      "learning_rate": 1.104393986005747e-05,
      "loss": 0.0607,
      "step": 170850
    },
    {
      "epoch": 2.516531420744908,
      "grad_norm": 1.1684163808822632,
      "learning_rate": 1.1042303488090487e-05,
      "loss": 0.0575,
      "step": 170875
    },
    {
      "epoch": 2.516899603834995,
      "grad_norm": 1.2969334125518799,
      "learning_rate": 1.10406671161235e-05,
      "loss": 0.0613,
      "step": 170900
    },
    {
      "epoch": 2.517267786925082,
      "grad_norm": 1.63365638256073,
      "learning_rate": 1.1039030744156516e-05,
      "loss": 0.0608,
      "step": 170925
    },
    {
      "epoch": 2.517635970015169,
      "grad_norm": 1.5004760026931763,
      "learning_rate": 1.1037394372189532e-05,
      "loss": 0.0576,
      "step": 170950
    },
    {
      "epoch": 2.518004153105256,
      "grad_norm": 1.3111398220062256,
      "learning_rate": 1.1035758000222546e-05,
      "loss": 0.0582,
      "step": 170975
    },
    {
      "epoch": 2.518372336195343,
      "grad_norm": 1.0577459335327148,
      "learning_rate": 1.1034121628255563e-05,
      "loss": 0.0601,
      "step": 171000
    },
    {
      "epoch": 2.5187405192854304,
      "grad_norm": 1.2442892789840698,
      "learning_rate": 1.1032485256288579e-05,
      "loss": 0.0623,
      "step": 171025
    },
    {
      "epoch": 2.5191087023755174,
      "grad_norm": 0.9899790287017822,
      "learning_rate": 1.1030848884321595e-05,
      "loss": 0.0555,
      "step": 171050
    },
    {
      "epoch": 2.5194768854656044,
      "grad_norm": 1.1004408597946167,
      "learning_rate": 1.102921251235461e-05,
      "loss": 0.0639,
      "step": 171075
    },
    {
      "epoch": 2.5198450685556915,
      "grad_norm": 1.7649691104888916,
      "learning_rate": 1.1027576140387624e-05,
      "loss": 0.0641,
      "step": 171100
    },
    {
      "epoch": 2.5202132516457785,
      "grad_norm": 1.2538145780563354,
      "learning_rate": 1.102593976842064e-05,
      "loss": 0.0651,
      "step": 171125
    },
    {
      "epoch": 2.5205814347358655,
      "grad_norm": 1.5799659490585327,
      "learning_rate": 1.1024303396453654e-05,
      "loss": 0.0566,
      "step": 171150
    },
    {
      "epoch": 2.5209496178259525,
      "grad_norm": 1.5274546146392822,
      "learning_rate": 1.102266702448667e-05,
      "loss": 0.0529,
      "step": 171175
    },
    {
      "epoch": 2.5213178009160395,
      "grad_norm": 1.1489367485046387,
      "learning_rate": 1.1021030652519687e-05,
      "loss": 0.0608,
      "step": 171200
    },
    {
      "epoch": 2.5216859840061265,
      "grad_norm": 1.1045472621917725,
      "learning_rate": 1.1019394280552703e-05,
      "loss": 0.0709,
      "step": 171225
    },
    {
      "epoch": 2.5220541670962136,
      "grad_norm": 0.944532573223114,
      "learning_rate": 1.1017757908585717e-05,
      "loss": 0.0585,
      "step": 171250
    },
    {
      "epoch": 2.5224223501863006,
      "grad_norm": 1.4226242303848267,
      "learning_rate": 1.1016121536618733e-05,
      "loss": 0.0618,
      "step": 171275
    },
    {
      "epoch": 2.5227905332763876,
      "grad_norm": 1.2734806537628174,
      "learning_rate": 1.101448516465175e-05,
      "loss": 0.0634,
      "step": 171300
    },
    {
      "epoch": 2.5231587163664746,
      "grad_norm": 1.162075161933899,
      "learning_rate": 1.1012848792684762e-05,
      "loss": 0.0565,
      "step": 171325
    },
    {
      "epoch": 2.5235268994565616,
      "grad_norm": 1.5107145309448242,
      "learning_rate": 1.1011212420717779e-05,
      "loss": 0.0586,
      "step": 171350
    },
    {
      "epoch": 2.523895082546649,
      "grad_norm": 1.5202807188034058,
      "learning_rate": 1.1009576048750795e-05,
      "loss": 0.0656,
      "step": 171375
    },
    {
      "epoch": 2.524263265636736,
      "grad_norm": 1.1547309160232544,
      "learning_rate": 1.100793967678381e-05,
      "loss": 0.0579,
      "step": 171400
    },
    {
      "epoch": 2.524631448726823,
      "grad_norm": 1.107877254486084,
      "learning_rate": 1.1006303304816825e-05,
      "loss": 0.0572,
      "step": 171425
    },
    {
      "epoch": 2.52499963181691,
      "grad_norm": 1.3621753454208374,
      "learning_rate": 1.1004666932849841e-05,
      "loss": 0.0583,
      "step": 171450
    },
    {
      "epoch": 2.525367814906997,
      "grad_norm": 1.2165898084640503,
      "learning_rate": 1.1003030560882858e-05,
      "loss": 0.0681,
      "step": 171475
    },
    {
      "epoch": 2.525735997997084,
      "grad_norm": 0.8010907173156738,
      "learning_rate": 1.1001394188915872e-05,
      "loss": 0.0596,
      "step": 171500
    },
    {
      "epoch": 2.526104181087171,
      "grad_norm": 1.4865715503692627,
      "learning_rate": 1.0999757816948887e-05,
      "loss": 0.0603,
      "step": 171525
    },
    {
      "epoch": 2.526472364177258,
      "grad_norm": 1.533263921737671,
      "learning_rate": 1.0998121444981901e-05,
      "loss": 0.061,
      "step": 171550
    },
    {
      "epoch": 2.526840547267345,
      "grad_norm": 1.4903484582901,
      "learning_rate": 1.0996485073014917e-05,
      "loss": 0.0564,
      "step": 171575
    },
    {
      "epoch": 2.527208730357432,
      "grad_norm": 0.9554063081741333,
      "learning_rate": 1.0994848701047933e-05,
      "loss": 0.0644,
      "step": 171600
    },
    {
      "epoch": 2.5275769134475192,
      "grad_norm": 1.4139825105667114,
      "learning_rate": 1.099321232908095e-05,
      "loss": 0.0728,
      "step": 171625
    },
    {
      "epoch": 2.5279450965376062,
      "grad_norm": 1.3340976238250732,
      "learning_rate": 1.0991575957113964e-05,
      "loss": 0.0621,
      "step": 171650
    },
    {
      "epoch": 2.5283132796276933,
      "grad_norm": 1.2663319110870361,
      "learning_rate": 1.098993958514698e-05,
      "loss": 0.0575,
      "step": 171675
    },
    {
      "epoch": 2.5286814627177803,
      "grad_norm": 1.643074631690979,
      "learning_rate": 1.0988303213179996e-05,
      "loss": 0.0551,
      "step": 171700
    },
    {
      "epoch": 2.5290496458078673,
      "grad_norm": 1.7283785343170166,
      "learning_rate": 1.0986666841213012e-05,
      "loss": 0.0674,
      "step": 171725
    },
    {
      "epoch": 2.5294178288979543,
      "grad_norm": 1.2985864877700806,
      "learning_rate": 1.0985030469246025e-05,
      "loss": 0.0588,
      "step": 171750
    },
    {
      "epoch": 2.5297860119880413,
      "grad_norm": 1.1789016723632812,
      "learning_rate": 1.0983394097279041e-05,
      "loss": 0.0598,
      "step": 171775
    },
    {
      "epoch": 2.5301541950781283,
      "grad_norm": 0.9327289462089539,
      "learning_rate": 1.0981757725312057e-05,
      "loss": 0.0624,
      "step": 171800
    },
    {
      "epoch": 2.5305223781682153,
      "grad_norm": 1.516343355178833,
      "learning_rate": 1.0980121353345072e-05,
      "loss": 0.0547,
      "step": 171825
    },
    {
      "epoch": 2.5308905612583024,
      "grad_norm": 1.5967307090759277,
      "learning_rate": 1.0978484981378088e-05,
      "loss": 0.0579,
      "step": 171850
    },
    {
      "epoch": 2.5312587443483894,
      "grad_norm": 1.1730389595031738,
      "learning_rate": 1.0976848609411104e-05,
      "loss": 0.0571,
      "step": 171875
    },
    {
      "epoch": 2.5316269274384764,
      "grad_norm": 1.4589629173278809,
      "learning_rate": 1.0975212237444119e-05,
      "loss": 0.0624,
      "step": 171900
    },
    {
      "epoch": 2.5319951105285634,
      "grad_norm": 1.6210570335388184,
      "learning_rate": 1.0973575865477135e-05,
      "loss": 0.0543,
      "step": 171925
    },
    {
      "epoch": 2.5323632936186504,
      "grad_norm": 1.1741224527359009,
      "learning_rate": 1.097193949351015e-05,
      "loss": 0.0584,
      "step": 171950
    },
    {
      "epoch": 2.532731476708738,
      "grad_norm": 1.2996466159820557,
      "learning_rate": 1.0970303121543164e-05,
      "loss": 0.0611,
      "step": 171975
    },
    {
      "epoch": 2.533099659798825,
      "grad_norm": 1.371118426322937,
      "learning_rate": 1.096866674957618e-05,
      "loss": 0.0574,
      "step": 172000
    },
    {
      "epoch": 2.533467842888912,
      "grad_norm": 1.527109980583191,
      "learning_rate": 1.0967030377609196e-05,
      "loss": 0.0606,
      "step": 172025
    },
    {
      "epoch": 2.533836025978999,
      "grad_norm": 0.7290975451469421,
      "learning_rate": 1.0965394005642212e-05,
      "loss": 0.0625,
      "step": 172050
    },
    {
      "epoch": 2.534204209069086,
      "grad_norm": 1.2977904081344604,
      "learning_rate": 1.0963757633675227e-05,
      "loss": 0.0614,
      "step": 172075
    },
    {
      "epoch": 2.534572392159173,
      "grad_norm": 1.3101640939712524,
      "learning_rate": 1.0962121261708243e-05,
      "loss": 0.0562,
      "step": 172100
    },
    {
      "epoch": 2.53494057524926,
      "grad_norm": 1.228004813194275,
      "learning_rate": 1.0960484889741259e-05,
      "loss": 0.0626,
      "step": 172125
    },
    {
      "epoch": 2.535308758339347,
      "grad_norm": 1.4276083707809448,
      "learning_rate": 1.0958848517774275e-05,
      "loss": 0.0571,
      "step": 172150
    },
    {
      "epoch": 2.535676941429434,
      "grad_norm": 1.3443721532821655,
      "learning_rate": 1.0957212145807288e-05,
      "loss": 0.062,
      "step": 172175
    },
    {
      "epoch": 2.536045124519521,
      "grad_norm": 0.9576521515846252,
      "learning_rate": 1.0955575773840304e-05,
      "loss": 0.0548,
      "step": 172200
    },
    {
      "epoch": 2.536413307609608,
      "grad_norm": 1.0474224090576172,
      "learning_rate": 1.0953939401873318e-05,
      "loss": 0.0593,
      "step": 172225
    },
    {
      "epoch": 2.536781490699695,
      "grad_norm": 1.3131436109542847,
      "learning_rate": 1.0952303029906335e-05,
      "loss": 0.0601,
      "step": 172250
    },
    {
      "epoch": 2.537149673789782,
      "grad_norm": 1.1439170837402344,
      "learning_rate": 1.095066665793935e-05,
      "loss": 0.0612,
      "step": 172275
    },
    {
      "epoch": 2.537517856879869,
      "grad_norm": 1.5104656219482422,
      "learning_rate": 1.0949030285972367e-05,
      "loss": 0.0612,
      "step": 172300
    },
    {
      "epoch": 2.537886039969956,
      "grad_norm": 0.8834033012390137,
      "learning_rate": 1.0947393914005381e-05,
      "loss": 0.0595,
      "step": 172325
    },
    {
      "epoch": 2.5382542230600436,
      "grad_norm": 1.4126570224761963,
      "learning_rate": 1.0945757542038398e-05,
      "loss": 0.0645,
      "step": 172350
    },
    {
      "epoch": 2.5386224061501306,
      "grad_norm": 1.0129644870758057,
      "learning_rate": 1.0944121170071412e-05,
      "loss": 0.0574,
      "step": 172375
    },
    {
      "epoch": 2.5389905892402176,
      "grad_norm": 1.1276706457138062,
      "learning_rate": 1.0942484798104426e-05,
      "loss": 0.0569,
      "step": 172400
    },
    {
      "epoch": 2.5393587723303046,
      "grad_norm": 0.7559683918952942,
      "learning_rate": 1.0940848426137443e-05,
      "loss": 0.0558,
      "step": 172425
    },
    {
      "epoch": 2.5397269554203916,
      "grad_norm": 0.7431871891021729,
      "learning_rate": 1.0939212054170459e-05,
      "loss": 0.0549,
      "step": 172450
    },
    {
      "epoch": 2.5400951385104786,
      "grad_norm": 1.4502934217453003,
      "learning_rate": 1.0937575682203473e-05,
      "loss": 0.0563,
      "step": 172475
    },
    {
      "epoch": 2.5404633216005656,
      "grad_norm": 1.8606306314468384,
      "learning_rate": 1.093593931023649e-05,
      "loss": 0.0565,
      "step": 172500
    },
    {
      "epoch": 2.5408315046906527,
      "grad_norm": 1.6751878261566162,
      "learning_rate": 1.0934302938269505e-05,
      "loss": 0.062,
      "step": 172525
    },
    {
      "epoch": 2.5411996877807397,
      "grad_norm": 1.163288950920105,
      "learning_rate": 1.0932666566302522e-05,
      "loss": 0.0578,
      "step": 172550
    },
    {
      "epoch": 2.5415678708708267,
      "grad_norm": 1.3800410032272339,
      "learning_rate": 1.0931030194335534e-05,
      "loss": 0.0639,
      "step": 172575
    },
    {
      "epoch": 2.5419360539609137,
      "grad_norm": 1.4605334997177124,
      "learning_rate": 1.092939382236855e-05,
      "loss": 0.0645,
      "step": 172600
    },
    {
      "epoch": 2.5423042370510007,
      "grad_norm": 1.5603034496307373,
      "learning_rate": 1.0927757450401567e-05,
      "loss": 0.0629,
      "step": 172625
    },
    {
      "epoch": 2.5426724201410877,
      "grad_norm": 1.1956393718719482,
      "learning_rate": 1.0926121078434581e-05,
      "loss": 0.0638,
      "step": 172650
    },
    {
      "epoch": 2.5430406032311748,
      "grad_norm": 1.3152358531951904,
      "learning_rate": 1.0924484706467597e-05,
      "loss": 0.0564,
      "step": 172675
    },
    {
      "epoch": 2.5434087863212618,
      "grad_norm": 1.3094853162765503,
      "learning_rate": 1.0922848334500613e-05,
      "loss": 0.0609,
      "step": 172700
    },
    {
      "epoch": 2.543776969411349,
      "grad_norm": 1.2942171096801758,
      "learning_rate": 1.092121196253363e-05,
      "loss": 0.0546,
      "step": 172725
    },
    {
      "epoch": 2.544145152501436,
      "grad_norm": 1.4587411880493164,
      "learning_rate": 1.0919575590566644e-05,
      "loss": 0.0564,
      "step": 172750
    },
    {
      "epoch": 2.544513335591523,
      "grad_norm": 1.1731085777282715,
      "learning_rate": 1.091793921859966e-05,
      "loss": 0.0624,
      "step": 172775
    },
    {
      "epoch": 2.54488151868161,
      "grad_norm": 1.484205722808838,
      "learning_rate": 1.0916302846632673e-05,
      "loss": 0.0599,
      "step": 172800
    },
    {
      "epoch": 2.545249701771697,
      "grad_norm": 1.0528786182403564,
      "learning_rate": 1.0914666474665689e-05,
      "loss": 0.0589,
      "step": 172825
    },
    {
      "epoch": 2.545617884861784,
      "grad_norm": 1.2374013662338257,
      "learning_rate": 1.0913030102698705e-05,
      "loss": 0.0566,
      "step": 172850
    },
    {
      "epoch": 2.545986067951871,
      "grad_norm": 1.175444483757019,
      "learning_rate": 1.0911393730731721e-05,
      "loss": 0.0592,
      "step": 172875
    },
    {
      "epoch": 2.546354251041958,
      "grad_norm": 1.2255836725234985,
      "learning_rate": 1.0909757358764736e-05,
      "loss": 0.0564,
      "step": 172900
    },
    {
      "epoch": 2.546722434132045,
      "grad_norm": 2.067763328552246,
      "learning_rate": 1.0908120986797752e-05,
      "loss": 0.0662,
      "step": 172925
    },
    {
      "epoch": 2.5470906172221324,
      "grad_norm": 1.4129890203475952,
      "learning_rate": 1.0906484614830768e-05,
      "loss": 0.0515,
      "step": 172950
    },
    {
      "epoch": 2.5474588003122194,
      "grad_norm": 1.695087194442749,
      "learning_rate": 1.0904848242863784e-05,
      "loss": 0.0631,
      "step": 172975
    },
    {
      "epoch": 2.5478269834023064,
      "grad_norm": 1.375633716583252,
      "learning_rate": 1.0903211870896797e-05,
      "loss": 0.0611,
      "step": 173000
    },
    {
      "epoch": 2.5481951664923934,
      "grad_norm": 1.4690459966659546,
      "learning_rate": 1.0901575498929813e-05,
      "loss": 0.0593,
      "step": 173025
    },
    {
      "epoch": 2.5485633495824804,
      "grad_norm": 2.098743438720703,
      "learning_rate": 1.0899939126962828e-05,
      "loss": 0.0638,
      "step": 173050
    },
    {
      "epoch": 2.5489315326725674,
      "grad_norm": 1.653788685798645,
      "learning_rate": 1.0898302754995844e-05,
      "loss": 0.0625,
      "step": 173075
    },
    {
      "epoch": 2.5492997157626545,
      "grad_norm": 1.2757081985473633,
      "learning_rate": 1.089666638302886e-05,
      "loss": 0.0621,
      "step": 173100
    },
    {
      "epoch": 2.5496678988527415,
      "grad_norm": 1.3836207389831543,
      "learning_rate": 1.0895030011061876e-05,
      "loss": 0.0575,
      "step": 173125
    },
    {
      "epoch": 2.5500360819428285,
      "grad_norm": 0.8787174224853516,
      "learning_rate": 1.089339363909489e-05,
      "loss": 0.0509,
      "step": 173150
    },
    {
      "epoch": 2.5504042650329155,
      "grad_norm": 1.0540931224822998,
      "learning_rate": 1.0891757267127907e-05,
      "loss": 0.0606,
      "step": 173175
    },
    {
      "epoch": 2.5507724481230025,
      "grad_norm": 0.9889383912086487,
      "learning_rate": 1.0890120895160923e-05,
      "loss": 0.055,
      "step": 173200
    },
    {
      "epoch": 2.5511406312130895,
      "grad_norm": 1.5400831699371338,
      "learning_rate": 1.0888484523193936e-05,
      "loss": 0.0574,
      "step": 173225
    },
    {
      "epoch": 2.5515088143031766,
      "grad_norm": 1.1952531337738037,
      "learning_rate": 1.0886848151226952e-05,
      "loss": 0.0573,
      "step": 173250
    },
    {
      "epoch": 2.5518769973932636,
      "grad_norm": 1.44627845287323,
      "learning_rate": 1.0885211779259968e-05,
      "loss": 0.0586,
      "step": 173275
    },
    {
      "epoch": 2.552245180483351,
      "grad_norm": 1.1690912246704102,
      "learning_rate": 1.0883575407292984e-05,
      "loss": 0.0588,
      "step": 173300
    },
    {
      "epoch": 2.552613363573438,
      "grad_norm": 1.0434222221374512,
      "learning_rate": 1.0882004490204678e-05,
      "loss": 0.059,
      "step": 173325
    },
    {
      "epoch": 2.552981546663525,
      "grad_norm": 1.2067543268203735,
      "learning_rate": 1.0880368118237694e-05,
      "loss": 0.0609,
      "step": 173350
    },
    {
      "epoch": 2.553349729753612,
      "grad_norm": 1.1828426122665405,
      "learning_rate": 1.087873174627071e-05,
      "loss": 0.0577,
      "step": 173375
    },
    {
      "epoch": 2.553717912843699,
      "grad_norm": 1.1234699487686157,
      "learning_rate": 1.0877095374303725e-05,
      "loss": 0.0528,
      "step": 173400
    },
    {
      "epoch": 2.554086095933786,
      "grad_norm": 1.2688320875167847,
      "learning_rate": 1.087545900233674e-05,
      "loss": 0.0558,
      "step": 173425
    },
    {
      "epoch": 2.554454279023873,
      "grad_norm": 1.5327723026275635,
      "learning_rate": 1.0873822630369755e-05,
      "loss": 0.06,
      "step": 173450
    },
    {
      "epoch": 2.55482246211396,
      "grad_norm": 1.2880806922912598,
      "learning_rate": 1.087218625840277e-05,
      "loss": 0.0557,
      "step": 173475
    },
    {
      "epoch": 2.555190645204047,
      "grad_norm": 1.1509093046188354,
      "learning_rate": 1.0870549886435786e-05,
      "loss": 0.0611,
      "step": 173500
    },
    {
      "epoch": 2.555558828294134,
      "grad_norm": 1.282781958580017,
      "learning_rate": 1.0868913514468802e-05,
      "loss": 0.0599,
      "step": 173525
    },
    {
      "epoch": 2.555927011384221,
      "grad_norm": 1.4733446836471558,
      "learning_rate": 1.0867277142501818e-05,
      "loss": 0.0553,
      "step": 173550
    },
    {
      "epoch": 2.556295194474308,
      "grad_norm": 1.279699444770813,
      "learning_rate": 1.0865640770534833e-05,
      "loss": 0.0565,
      "step": 173575
    },
    {
      "epoch": 2.556663377564395,
      "grad_norm": 1.5490727424621582,
      "learning_rate": 1.0864004398567849e-05,
      "loss": 0.063,
      "step": 173600
    },
    {
      "epoch": 2.5570315606544822,
      "grad_norm": 1.3272525072097778,
      "learning_rate": 1.0862368026600865e-05,
      "loss": 0.0592,
      "step": 173625
    },
    {
      "epoch": 2.5573997437445692,
      "grad_norm": 1.7039073705673218,
      "learning_rate": 1.0860731654633878e-05,
      "loss": 0.0627,
      "step": 173650
    },
    {
      "epoch": 2.5577679268346563,
      "grad_norm": 1.239800214767456,
      "learning_rate": 1.0859095282666894e-05,
      "loss": 0.0652,
      "step": 173675
    },
    {
      "epoch": 2.5581361099247433,
      "grad_norm": 1.62569260597229,
      "learning_rate": 1.085745891069991e-05,
      "loss": 0.065,
      "step": 173700
    },
    {
      "epoch": 2.5585042930148303,
      "grad_norm": 1.069297194480896,
      "learning_rate": 1.0855822538732925e-05,
      "loss": 0.058,
      "step": 173725
    },
    {
      "epoch": 2.5588724761049173,
      "grad_norm": 1.8260818719863892,
      "learning_rate": 1.085418616676594e-05,
      "loss": 0.0539,
      "step": 173750
    },
    {
      "epoch": 2.5592406591950043,
      "grad_norm": 1.5096160173416138,
      "learning_rate": 1.0852549794798957e-05,
      "loss": 0.0625,
      "step": 173775
    },
    {
      "epoch": 2.5596088422850913,
      "grad_norm": 1.12891685962677,
      "learning_rate": 1.0850913422831973e-05,
      "loss": 0.0599,
      "step": 173800
    },
    {
      "epoch": 2.5599770253751783,
      "grad_norm": 1.8938446044921875,
      "learning_rate": 1.0849277050864988e-05,
      "loss": 0.0677,
      "step": 173825
    },
    {
      "epoch": 2.5603452084652654,
      "grad_norm": 0.8113965392112732,
      "learning_rate": 1.0847640678898002e-05,
      "loss": 0.0572,
      "step": 173850
    },
    {
      "epoch": 2.5607133915553524,
      "grad_norm": 1.1232788562774658,
      "learning_rate": 1.0846004306931016e-05,
      "loss": 0.061,
      "step": 173875
    },
    {
      "epoch": 2.56108157464544,
      "grad_norm": 1.623877763748169,
      "learning_rate": 1.0844367934964033e-05,
      "loss": 0.0707,
      "step": 173900
    },
    {
      "epoch": 2.561449757735527,
      "grad_norm": 1.5341931581497192,
      "learning_rate": 1.0842731562997049e-05,
      "loss": 0.0569,
      "step": 173925
    },
    {
      "epoch": 2.561817940825614,
      "grad_norm": 1.5959268808364868,
      "learning_rate": 1.0841095191030065e-05,
      "loss": 0.0597,
      "step": 173950
    },
    {
      "epoch": 2.562186123915701,
      "grad_norm": 1.1079158782958984,
      "learning_rate": 1.083945881906308e-05,
      "loss": 0.0651,
      "step": 173975
    },
    {
      "epoch": 2.562554307005788,
      "grad_norm": 1.3437291383743286,
      "learning_rate": 1.0837822447096095e-05,
      "loss": 0.0605,
      "step": 174000
    },
    {
      "epoch": 2.562922490095875,
      "grad_norm": 1.8788952827453613,
      "learning_rate": 1.0836186075129112e-05,
      "loss": 0.0646,
      "step": 174025
    },
    {
      "epoch": 2.563290673185962,
      "grad_norm": 1.4978293180465698,
      "learning_rate": 1.0834549703162128e-05,
      "loss": 0.0648,
      "step": 174050
    },
    {
      "epoch": 2.563658856276049,
      "grad_norm": 1.351623773574829,
      "learning_rate": 1.083291333119514e-05,
      "loss": 0.0655,
      "step": 174075
    },
    {
      "epoch": 2.564027039366136,
      "grad_norm": 1.4173803329467773,
      "learning_rate": 1.0831276959228157e-05,
      "loss": 0.0547,
      "step": 174100
    },
    {
      "epoch": 2.564395222456223,
      "grad_norm": 0.7470574975013733,
      "learning_rate": 1.0829640587261173e-05,
      "loss": 0.0613,
      "step": 174125
    },
    {
      "epoch": 2.56476340554631,
      "grad_norm": 1.2393810749053955,
      "learning_rate": 1.0828004215294187e-05,
      "loss": 0.0655,
      "step": 174150
    },
    {
      "epoch": 2.565131588636397,
      "grad_norm": 1.1187108755111694,
      "learning_rate": 1.0826367843327203e-05,
      "loss": 0.058,
      "step": 174175
    },
    {
      "epoch": 2.565499771726484,
      "grad_norm": 1.1925551891326904,
      "learning_rate": 1.082473147136022e-05,
      "loss": 0.0612,
      "step": 174200
    },
    {
      "epoch": 2.565867954816571,
      "grad_norm": 1.777707815170288,
      "learning_rate": 1.0823095099393234e-05,
      "loss": 0.0609,
      "step": 174225
    },
    {
      "epoch": 2.5662361379066585,
      "grad_norm": 1.914480209350586,
      "learning_rate": 1.082145872742625e-05,
      "loss": 0.0545,
      "step": 174250
    },
    {
      "epoch": 2.5666043209967455,
      "grad_norm": 1.142600655555725,
      "learning_rate": 1.0819822355459265e-05,
      "loss": 0.0564,
      "step": 174275
    },
    {
      "epoch": 2.5669725040868325,
      "grad_norm": 1.641832947731018,
      "learning_rate": 1.0818185983492279e-05,
      "loss": 0.0694,
      "step": 174300
    },
    {
      "epoch": 2.5673406871769195,
      "grad_norm": 1.7670449018478394,
      "learning_rate": 1.0816549611525295e-05,
      "loss": 0.0577,
      "step": 174325
    },
    {
      "epoch": 2.5677088702670066,
      "grad_norm": 1.3882704973220825,
      "learning_rate": 1.0814913239558311e-05,
      "loss": 0.0603,
      "step": 174350
    },
    {
      "epoch": 2.5680770533570936,
      "grad_norm": 1.5545358657836914,
      "learning_rate": 1.0813276867591328e-05,
      "loss": 0.0563,
      "step": 174375
    },
    {
      "epoch": 2.5684452364471806,
      "grad_norm": 1.2787203788757324,
      "learning_rate": 1.0811640495624342e-05,
      "loss": 0.0528,
      "step": 174400
    },
    {
      "epoch": 2.5688134195372676,
      "grad_norm": 1.0915101766586304,
      "learning_rate": 1.0810004123657358e-05,
      "loss": 0.0554,
      "step": 174425
    },
    {
      "epoch": 2.5691816026273546,
      "grad_norm": 1.2762171030044556,
      "learning_rate": 1.0808367751690374e-05,
      "loss": 0.0593,
      "step": 174450
    },
    {
      "epoch": 2.5695497857174416,
      "grad_norm": 1.1823515892028809,
      "learning_rate": 1.0806731379723387e-05,
      "loss": 0.0623,
      "step": 174475
    },
    {
      "epoch": 2.5699179688075287,
      "grad_norm": 1.8453294038772583,
      "learning_rate": 1.0805095007756403e-05,
      "loss": 0.0611,
      "step": 174500
    },
    {
      "epoch": 2.5702861518976157,
      "grad_norm": 1.1715410947799683,
      "learning_rate": 1.080345863578942e-05,
      "loss": 0.0582,
      "step": 174525
    },
    {
      "epoch": 2.5706543349877027,
      "grad_norm": 1.1181731224060059,
      "learning_rate": 1.0801822263822434e-05,
      "loss": 0.0557,
      "step": 174550
    },
    {
      "epoch": 2.5710225180777897,
      "grad_norm": 1.0414645671844482,
      "learning_rate": 1.080018589185545e-05,
      "loss": 0.0575,
      "step": 174575
    },
    {
      "epoch": 2.5713907011678767,
      "grad_norm": 1.3872711658477783,
      "learning_rate": 1.0798549519888466e-05,
      "loss": 0.0664,
      "step": 174600
    },
    {
      "epoch": 2.5717588842579637,
      "grad_norm": 1.711892008781433,
      "learning_rate": 1.0796913147921482e-05,
      "loss": 0.0582,
      "step": 174625
    },
    {
      "epoch": 2.5721270673480507,
      "grad_norm": 1.81852388381958,
      "learning_rate": 1.0795276775954497e-05,
      "loss": 0.0632,
      "step": 174650
    },
    {
      "epoch": 2.5724952504381378,
      "grad_norm": 1.6783087253570557,
      "learning_rate": 1.0793640403987513e-05,
      "loss": 0.0642,
      "step": 174675
    },
    {
      "epoch": 2.5728634335282248,
      "grad_norm": 1.4971789121627808,
      "learning_rate": 1.0792004032020527e-05,
      "loss": 0.0624,
      "step": 174700
    },
    {
      "epoch": 2.573231616618312,
      "grad_norm": 1.3275350332260132,
      "learning_rate": 1.0790367660053542e-05,
      "loss": 0.0637,
      "step": 174725
    },
    {
      "epoch": 2.573599799708399,
      "grad_norm": 1.269001841545105,
      "learning_rate": 1.0788731288086558e-05,
      "loss": 0.0545,
      "step": 174750
    },
    {
      "epoch": 2.573967982798486,
      "grad_norm": 1.5252752304077148,
      "learning_rate": 1.0787094916119574e-05,
      "loss": 0.0596,
      "step": 174775
    },
    {
      "epoch": 2.574336165888573,
      "grad_norm": 1.285728096961975,
      "learning_rate": 1.0785458544152589e-05,
      "loss": 0.0595,
      "step": 174800
    },
    {
      "epoch": 2.57470434897866,
      "grad_norm": 1.3012200593948364,
      "learning_rate": 1.0783822172185605e-05,
      "loss": 0.0563,
      "step": 174825
    },
    {
      "epoch": 2.5750725320687473,
      "grad_norm": 1.4969555139541626,
      "learning_rate": 1.0782185800218621e-05,
      "loss": 0.0619,
      "step": 174850
    },
    {
      "epoch": 2.5754407151588343,
      "grad_norm": 1.3967996835708618,
      "learning_rate": 1.0780549428251637e-05,
      "loss": 0.0596,
      "step": 174875
    },
    {
      "epoch": 2.5758088982489213,
      "grad_norm": 1.6889322996139526,
      "learning_rate": 1.077891305628465e-05,
      "loss": 0.0588,
      "step": 174900
    },
    {
      "epoch": 2.5761770813390084,
      "grad_norm": 1.378534197807312,
      "learning_rate": 1.0777276684317666e-05,
      "loss": 0.0602,
      "step": 174925
    },
    {
      "epoch": 2.5765452644290954,
      "grad_norm": 1.597387433052063,
      "learning_rate": 1.0775640312350682e-05,
      "loss": 0.0559,
      "step": 174950
    },
    {
      "epoch": 2.5769134475191824,
      "grad_norm": 1.8913077116012573,
      "learning_rate": 1.0774003940383697e-05,
      "loss": 0.068,
      "step": 174975
    },
    {
      "epoch": 2.5772816306092694,
      "grad_norm": 1.259865403175354,
      "learning_rate": 1.0772367568416713e-05,
      "loss": 0.0538,
      "step": 175000
    },
    {
      "epoch": 2.5776498136993564,
      "grad_norm": 0.9669091105461121,
      "learning_rate": 1.0770731196449729e-05,
      "loss": 0.0602,
      "step": 175025
    },
    {
      "epoch": 2.5780179967894434,
      "grad_norm": 1.1503703594207764,
      "learning_rate": 1.0769094824482745e-05,
      "loss": 0.0486,
      "step": 175050
    },
    {
      "epoch": 2.5783861798795304,
      "grad_norm": 1.5549144744873047,
      "learning_rate": 1.076745845251576e-05,
      "loss": 0.0566,
      "step": 175075
    },
    {
      "epoch": 2.5787543629696175,
      "grad_norm": 0.9050664901733398,
      "learning_rate": 1.0765822080548776e-05,
      "loss": 0.0601,
      "step": 175100
    },
    {
      "epoch": 2.5791225460597045,
      "grad_norm": 1.4891036748886108,
      "learning_rate": 1.0764185708581788e-05,
      "loss": 0.0568,
      "step": 175125
    },
    {
      "epoch": 2.5794907291497915,
      "grad_norm": 1.5478847026824951,
      "learning_rate": 1.0762549336614805e-05,
      "loss": 0.0596,
      "step": 175150
    },
    {
      "epoch": 2.5798589122398785,
      "grad_norm": 1.3112157583236694,
      "learning_rate": 1.076091296464782e-05,
      "loss": 0.0574,
      "step": 175175
    },
    {
      "epoch": 2.5802270953299655,
      "grad_norm": 1.233964443206787,
      "learning_rate": 1.0759276592680837e-05,
      "loss": 0.053,
      "step": 175200
    },
    {
      "epoch": 2.580595278420053,
      "grad_norm": 0.8812665343284607,
      "learning_rate": 1.0757640220713851e-05,
      "loss": 0.0581,
      "step": 175225
    },
    {
      "epoch": 2.58096346151014,
      "grad_norm": 1.3952767848968506,
      "learning_rate": 1.0756003848746867e-05,
      "loss": 0.0504,
      "step": 175250
    },
    {
      "epoch": 2.581331644600227,
      "grad_norm": 1.034899115562439,
      "learning_rate": 1.0754367476779884e-05,
      "loss": 0.0546,
      "step": 175275
    },
    {
      "epoch": 2.581699827690314,
      "grad_norm": 2.012953996658325,
      "learning_rate": 1.07527311048129e-05,
      "loss": 0.0577,
      "step": 175300
    },
    {
      "epoch": 2.582068010780401,
      "grad_norm": 1.2987000942230225,
      "learning_rate": 1.0751094732845913e-05,
      "loss": 0.0645,
      "step": 175325
    },
    {
      "epoch": 2.582436193870488,
      "grad_norm": 1.3182604312896729,
      "learning_rate": 1.0749458360878929e-05,
      "loss": 0.057,
      "step": 175350
    },
    {
      "epoch": 2.582804376960575,
      "grad_norm": 1.5478579998016357,
      "learning_rate": 1.0747821988911943e-05,
      "loss": 0.0605,
      "step": 175375
    },
    {
      "epoch": 2.583172560050662,
      "grad_norm": 1.0221810340881348,
      "learning_rate": 1.074618561694496e-05,
      "loss": 0.0603,
      "step": 175400
    },
    {
      "epoch": 2.583540743140749,
      "grad_norm": 1.7871437072753906,
      "learning_rate": 1.0744549244977975e-05,
      "loss": 0.0553,
      "step": 175425
    },
    {
      "epoch": 2.583908926230836,
      "grad_norm": 1.3152027130126953,
      "learning_rate": 1.0742912873010992e-05,
      "loss": 0.0568,
      "step": 175450
    },
    {
      "epoch": 2.584277109320923,
      "grad_norm": 1.2218319177627563,
      "learning_rate": 1.0741276501044006e-05,
      "loss": 0.0606,
      "step": 175475
    },
    {
      "epoch": 2.58464529241101,
      "grad_norm": 1.649019479751587,
      "learning_rate": 1.0739640129077022e-05,
      "loss": 0.0582,
      "step": 175500
    },
    {
      "epoch": 2.585013475501097,
      "grad_norm": 1.5051950216293335,
      "learning_rate": 1.0738003757110038e-05,
      "loss": 0.0602,
      "step": 175525
    },
    {
      "epoch": 2.585381658591184,
      "grad_norm": 1.6265712976455688,
      "learning_rate": 1.0736367385143051e-05,
      "loss": 0.0691,
      "step": 175550
    },
    {
      "epoch": 2.585749841681271,
      "grad_norm": 1.2473348379135132,
      "learning_rate": 1.0734731013176067e-05,
      "loss": 0.0581,
      "step": 175575
    },
    {
      "epoch": 2.586118024771358,
      "grad_norm": 1.1284838914871216,
      "learning_rate": 1.0733094641209083e-05,
      "loss": 0.058,
      "step": 175600
    },
    {
      "epoch": 2.5864862078614452,
      "grad_norm": 1.1544748544692993,
      "learning_rate": 1.07314582692421e-05,
      "loss": 0.0625,
      "step": 175625
    },
    {
      "epoch": 2.5868543909515322,
      "grad_norm": 1.856658935546875,
      "learning_rate": 1.0729821897275114e-05,
      "loss": 0.0602,
      "step": 175650
    },
    {
      "epoch": 2.5872225740416193,
      "grad_norm": 1.3996042013168335,
      "learning_rate": 1.072818552530813e-05,
      "loss": 0.0626,
      "step": 175675
    },
    {
      "epoch": 2.5875907571317063,
      "grad_norm": 1.4461251497268677,
      "learning_rate": 1.0726549153341146e-05,
      "loss": 0.0661,
      "step": 175700
    },
    {
      "epoch": 2.5879589402217933,
      "grad_norm": 1.5906156301498413,
      "learning_rate": 1.072491278137416e-05,
      "loss": 0.0666,
      "step": 175725
    },
    {
      "epoch": 2.5883271233118803,
      "grad_norm": 1.1601279973983765,
      "learning_rate": 1.0723276409407175e-05,
      "loss": 0.0535,
      "step": 175750
    },
    {
      "epoch": 2.5886953064019673,
      "grad_norm": 1.4376877546310425,
      "learning_rate": 1.0721640037440191e-05,
      "loss": 0.0573,
      "step": 175775
    },
    {
      "epoch": 2.5890634894920543,
      "grad_norm": 1.4215503931045532,
      "learning_rate": 1.0720003665473206e-05,
      "loss": 0.0592,
      "step": 175800
    },
    {
      "epoch": 2.589431672582142,
      "grad_norm": 1.3166005611419678,
      "learning_rate": 1.0718367293506222e-05,
      "loss": 0.0643,
      "step": 175825
    },
    {
      "epoch": 2.589799855672229,
      "grad_norm": 1.6120201349258423,
      "learning_rate": 1.0716730921539238e-05,
      "loss": 0.0646,
      "step": 175850
    },
    {
      "epoch": 2.590168038762316,
      "grad_norm": 1.91206693649292,
      "learning_rate": 1.0715094549572254e-05,
      "loss": 0.0594,
      "step": 175875
    },
    {
      "epoch": 2.590536221852403,
      "grad_norm": 1.233803629875183,
      "learning_rate": 1.0713458177605269e-05,
      "loss": 0.0539,
      "step": 175900
    },
    {
      "epoch": 2.59090440494249,
      "grad_norm": 1.8472914695739746,
      "learning_rate": 1.0711821805638285e-05,
      "loss": 0.0652,
      "step": 175925
    },
    {
      "epoch": 2.591272588032577,
      "grad_norm": 1.537962555885315,
      "learning_rate": 1.0710185433671298e-05,
      "loss": 0.0577,
      "step": 175950
    },
    {
      "epoch": 2.591640771122664,
      "grad_norm": 0.8386574387550354,
      "learning_rate": 1.0708549061704314e-05,
      "loss": 0.0536,
      "step": 175975
    },
    {
      "epoch": 2.592008954212751,
      "grad_norm": 1.2876434326171875,
      "learning_rate": 1.070691268973733e-05,
      "loss": 0.061,
      "step": 176000
    },
    {
      "epoch": 2.592377137302838,
      "grad_norm": 1.6931270360946655,
      "learning_rate": 1.0705276317770346e-05,
      "loss": 0.0539,
      "step": 176025
    },
    {
      "epoch": 2.592745320392925,
      "grad_norm": 1.1909981966018677,
      "learning_rate": 1.070363994580336e-05,
      "loss": 0.0619,
      "step": 176050
    },
    {
      "epoch": 2.593113503483012,
      "grad_norm": 1.2086323499679565,
      "learning_rate": 1.0702003573836377e-05,
      "loss": 0.0532,
      "step": 176075
    },
    {
      "epoch": 2.593481686573099,
      "grad_norm": 1.4572724103927612,
      "learning_rate": 1.0700367201869393e-05,
      "loss": 0.056,
      "step": 176100
    },
    {
      "epoch": 2.593849869663186,
      "grad_norm": 1.4585648775100708,
      "learning_rate": 1.0698730829902409e-05,
      "loss": 0.0677,
      "step": 176125
    },
    {
      "epoch": 2.594218052753273,
      "grad_norm": 1.7802493572235107,
      "learning_rate": 1.0697094457935424e-05,
      "loss": 0.0601,
      "step": 176150
    },
    {
      "epoch": 2.5945862358433605,
      "grad_norm": 1.6509461402893066,
      "learning_rate": 1.0695458085968438e-05,
      "loss": 0.0589,
      "step": 176175
    },
    {
      "epoch": 2.5949544189334475,
      "grad_norm": 1.4764903783798218,
      "learning_rate": 1.0693821714001454e-05,
      "loss": 0.0637,
      "step": 176200
    },
    {
      "epoch": 2.5953226020235345,
      "grad_norm": 1.2520935535430908,
      "learning_rate": 1.0692185342034469e-05,
      "loss": 0.0604,
      "step": 176225
    },
    {
      "epoch": 2.5956907851136215,
      "grad_norm": 1.5063178539276123,
      "learning_rate": 1.0690548970067485e-05,
      "loss": 0.0556,
      "step": 176250
    },
    {
      "epoch": 2.5960589682037085,
      "grad_norm": 1.5612396001815796,
      "learning_rate": 1.0688912598100501e-05,
      "loss": 0.0626,
      "step": 176275
    },
    {
      "epoch": 2.5964271512937955,
      "grad_norm": 1.3243896961212158,
      "learning_rate": 1.0687276226133515e-05,
      "loss": 0.065,
      "step": 176300
    },
    {
      "epoch": 2.5967953343838825,
      "grad_norm": 0.7383653521537781,
      "learning_rate": 1.0685639854166532e-05,
      "loss": 0.0543,
      "step": 176325
    },
    {
      "epoch": 2.5971635174739696,
      "grad_norm": 1.701586365699768,
      "learning_rate": 1.0684003482199548e-05,
      "loss": 0.0538,
      "step": 176350
    },
    {
      "epoch": 2.5975317005640566,
      "grad_norm": 1.2080687284469604,
      "learning_rate": 1.068236711023256e-05,
      "loss": 0.054,
      "step": 176375
    },
    {
      "epoch": 2.5978998836541436,
      "grad_norm": 0.8733327984809875,
      "learning_rate": 1.0680730738265577e-05,
      "loss": 0.0581,
      "step": 176400
    },
    {
      "epoch": 2.5982680667442306,
      "grad_norm": 1.3134326934814453,
      "learning_rate": 1.0679094366298593e-05,
      "loss": 0.0651,
      "step": 176425
    },
    {
      "epoch": 2.5986362498343176,
      "grad_norm": 1.738545298576355,
      "learning_rate": 1.0677457994331609e-05,
      "loss": 0.0623,
      "step": 176450
    },
    {
      "epoch": 2.5990044329244046,
      "grad_norm": 1.4502187967300415,
      "learning_rate": 1.0675821622364623e-05,
      "loss": 0.0648,
      "step": 176475
    },
    {
      "epoch": 2.5993726160144917,
      "grad_norm": 1.1931694746017456,
      "learning_rate": 1.067418525039764e-05,
      "loss": 0.0623,
      "step": 176500
    },
    {
      "epoch": 2.5997407991045787,
      "grad_norm": 2.380855083465576,
      "learning_rate": 1.0672548878430656e-05,
      "loss": 0.0677,
      "step": 176525
    },
    {
      "epoch": 2.6001089821946657,
      "grad_norm": 1.5158504247665405,
      "learning_rate": 1.067091250646367e-05,
      "loss": 0.058,
      "step": 176550
    },
    {
      "epoch": 2.6004771652847527,
      "grad_norm": 1.609561800956726,
      "learning_rate": 1.0669276134496686e-05,
      "loss": 0.0576,
      "step": 176575
    },
    {
      "epoch": 2.6008453483748397,
      "grad_norm": 1.1952263116836548,
      "learning_rate": 1.06676397625297e-05,
      "loss": 0.0568,
      "step": 176600
    },
    {
      "epoch": 2.6012135314649267,
      "grad_norm": 0.928909182548523,
      "learning_rate": 1.0666003390562715e-05,
      "loss": 0.0591,
      "step": 176625
    },
    {
      "epoch": 2.6015817145550137,
      "grad_norm": 1.8413697481155396,
      "learning_rate": 1.0664367018595731e-05,
      "loss": 0.0657,
      "step": 176650
    },
    {
      "epoch": 2.6019498976451008,
      "grad_norm": 1.7064995765686035,
      "learning_rate": 1.0662730646628747e-05,
      "loss": 0.0651,
      "step": 176675
    },
    {
      "epoch": 2.6023180807351878,
      "grad_norm": 1.1123034954071045,
      "learning_rate": 1.0661094274661764e-05,
      "loss": 0.056,
      "step": 176700
    },
    {
      "epoch": 2.602686263825275,
      "grad_norm": 1.1000996828079224,
      "learning_rate": 1.0659457902694778e-05,
      "loss": 0.0579,
      "step": 176725
    },
    {
      "epoch": 2.603054446915362,
      "grad_norm": 1.422594428062439,
      "learning_rate": 1.0657821530727794e-05,
      "loss": 0.0564,
      "step": 176750
    },
    {
      "epoch": 2.6034226300054493,
      "grad_norm": 1.125012755393982,
      "learning_rate": 1.065618515876081e-05,
      "loss": 0.0528,
      "step": 176775
    },
    {
      "epoch": 2.6037908130955363,
      "grad_norm": 1.297001838684082,
      "learning_rate": 1.0654548786793823e-05,
      "loss": 0.0614,
      "step": 176800
    },
    {
      "epoch": 2.6041589961856233,
      "grad_norm": 1.121140480041504,
      "learning_rate": 1.065291241482684e-05,
      "loss": 0.0554,
      "step": 176825
    },
    {
      "epoch": 2.6045271792757103,
      "grad_norm": 1.47142493724823,
      "learning_rate": 1.0651276042859855e-05,
      "loss": 0.0611,
      "step": 176850
    },
    {
      "epoch": 2.6048953623657973,
      "grad_norm": 1.4502065181732178,
      "learning_rate": 1.064963967089287e-05,
      "loss": 0.0663,
      "step": 176875
    },
    {
      "epoch": 2.6052635454558843,
      "grad_norm": 1.3370851278305054,
      "learning_rate": 1.0648003298925886e-05,
      "loss": 0.0592,
      "step": 176900
    },
    {
      "epoch": 2.6056317285459714,
      "grad_norm": 0.9253253936767578,
      "learning_rate": 1.0646366926958902e-05,
      "loss": 0.0599,
      "step": 176925
    },
    {
      "epoch": 2.6059999116360584,
      "grad_norm": 0.8026204705238342,
      "learning_rate": 1.0644730554991918e-05,
      "loss": 0.0581,
      "step": 176950
    },
    {
      "epoch": 2.6063680947261454,
      "grad_norm": 1.0097384452819824,
      "learning_rate": 1.0643094183024933e-05,
      "loss": 0.0579,
      "step": 176975
    },
    {
      "epoch": 2.6067362778162324,
      "grad_norm": 1.2711149454116821,
      "learning_rate": 1.0641457811057949e-05,
      "loss": 0.0634,
      "step": 177000
    },
    {
      "epoch": 2.6071044609063194,
      "grad_norm": 1.2087106704711914,
      "learning_rate": 1.0639821439090963e-05,
      "loss": 0.0588,
      "step": 177025
    },
    {
      "epoch": 2.6074726439964064,
      "grad_norm": 1.7671455144882202,
      "learning_rate": 1.0638185067123978e-05,
      "loss": 0.0627,
      "step": 177050
    },
    {
      "epoch": 2.6078408270864935,
      "grad_norm": 1.6097408533096313,
      "learning_rate": 1.0636548695156994e-05,
      "loss": 0.0637,
      "step": 177075
    },
    {
      "epoch": 2.6082090101765805,
      "grad_norm": 1.0626698732376099,
      "learning_rate": 1.063491232319001e-05,
      "loss": 0.0553,
      "step": 177100
    },
    {
      "epoch": 2.6085771932666675,
      "grad_norm": 1.4588572978973389,
      "learning_rate": 1.0633275951223025e-05,
      "loss": 0.065,
      "step": 177125
    },
    {
      "epoch": 2.608945376356755,
      "grad_norm": 0.9147449731826782,
      "learning_rate": 1.063163957925604e-05,
      "loss": 0.056,
      "step": 177150
    },
    {
      "epoch": 2.609313559446842,
      "grad_norm": 1.1432408094406128,
      "learning_rate": 1.0630003207289057e-05,
      "loss": 0.0652,
      "step": 177175
    },
    {
      "epoch": 2.609681742536929,
      "grad_norm": 1.362931489944458,
      "learning_rate": 1.0628366835322073e-05,
      "loss": 0.0697,
      "step": 177200
    },
    {
      "epoch": 2.610049925627016,
      "grad_norm": 0.7644778490066528,
      "learning_rate": 1.0626730463355086e-05,
      "loss": 0.055,
      "step": 177225
    },
    {
      "epoch": 2.610418108717103,
      "grad_norm": 1.5069628953933716,
      "learning_rate": 1.0625094091388102e-05,
      "loss": 0.0588,
      "step": 177250
    },
    {
      "epoch": 2.61078629180719,
      "grad_norm": 1.3445383310317993,
      "learning_rate": 1.0623523174299798e-05,
      "loss": 0.0705,
      "step": 177275
    },
    {
      "epoch": 2.611154474897277,
      "grad_norm": 1.5314680337905884,
      "learning_rate": 1.0621886802332812e-05,
      "loss": 0.0605,
      "step": 177300
    },
    {
      "epoch": 2.611522657987364,
      "grad_norm": 1.4227596521377563,
      "learning_rate": 1.0620250430365828e-05,
      "loss": 0.0615,
      "step": 177325
    },
    {
      "epoch": 2.611890841077451,
      "grad_norm": 1.6466411352157593,
      "learning_rate": 1.0618614058398844e-05,
      "loss": 0.0557,
      "step": 177350
    },
    {
      "epoch": 2.612259024167538,
      "grad_norm": 1.2078698873519897,
      "learning_rate": 1.061697768643186e-05,
      "loss": 0.0673,
      "step": 177375
    },
    {
      "epoch": 2.612627207257625,
      "grad_norm": 1.6566219329833984,
      "learning_rate": 1.0615341314464875e-05,
      "loss": 0.0542,
      "step": 177400
    },
    {
      "epoch": 2.612995390347712,
      "grad_norm": 1.2760077714920044,
      "learning_rate": 1.061370494249789e-05,
      "loss": 0.0585,
      "step": 177425
    },
    {
      "epoch": 2.613363573437799,
      "grad_norm": 1.257825493812561,
      "learning_rate": 1.0612068570530904e-05,
      "loss": 0.061,
      "step": 177450
    },
    {
      "epoch": 2.613731756527886,
      "grad_norm": 0.9882104396820068,
      "learning_rate": 1.061043219856392e-05,
      "loss": 0.0563,
      "step": 177475
    },
    {
      "epoch": 2.614099939617973,
      "grad_norm": 1.5817950963974,
      "learning_rate": 1.0608795826596936e-05,
      "loss": 0.0602,
      "step": 177500
    },
    {
      "epoch": 2.61446812270806,
      "grad_norm": 1.0192517042160034,
      "learning_rate": 1.0607159454629952e-05,
      "loss": 0.0618,
      "step": 177525
    },
    {
      "epoch": 2.614836305798147,
      "grad_norm": 1.6604156494140625,
      "learning_rate": 1.0605523082662967e-05,
      "loss": 0.069,
      "step": 177550
    },
    {
      "epoch": 2.615204488888234,
      "grad_norm": 1.4611319303512573,
      "learning_rate": 1.0603886710695983e-05,
      "loss": 0.0656,
      "step": 177575
    },
    {
      "epoch": 2.615572671978321,
      "grad_norm": 1.1192989349365234,
      "learning_rate": 1.0602250338728999e-05,
      "loss": 0.0592,
      "step": 177600
    },
    {
      "epoch": 2.6159408550684082,
      "grad_norm": 1.1363897323608398,
      "learning_rate": 1.0600613966762015e-05,
      "loss": 0.0633,
      "step": 177625
    },
    {
      "epoch": 2.6163090381584952,
      "grad_norm": 1.5756871700286865,
      "learning_rate": 1.0598977594795028e-05,
      "loss": 0.0561,
      "step": 177650
    },
    {
      "epoch": 2.6166772212485823,
      "grad_norm": 1.8618441820144653,
      "learning_rate": 1.0597341222828044e-05,
      "loss": 0.0657,
      "step": 177675
    },
    {
      "epoch": 2.6170454043386693,
      "grad_norm": 0.9189126491546631,
      "learning_rate": 1.0595704850861059e-05,
      "loss": 0.0553,
      "step": 177700
    },
    {
      "epoch": 2.6174135874287567,
      "grad_norm": 1.7426941394805908,
      "learning_rate": 1.0594068478894075e-05,
      "loss": 0.063,
      "step": 177725
    },
    {
      "epoch": 2.6177817705188438,
      "grad_norm": 1.4612746238708496,
      "learning_rate": 1.0592432106927091e-05,
      "loss": 0.0561,
      "step": 177750
    },
    {
      "epoch": 2.6181499536089308,
      "grad_norm": 1.1249966621398926,
      "learning_rate": 1.0590795734960107e-05,
      "loss": 0.0691,
      "step": 177775
    },
    {
      "epoch": 2.618518136699018,
      "grad_norm": 1.174182653427124,
      "learning_rate": 1.0589159362993122e-05,
      "loss": 0.0618,
      "step": 177800
    },
    {
      "epoch": 2.618886319789105,
      "grad_norm": 0.9717144966125488,
      "learning_rate": 1.0587522991026138e-05,
      "loss": 0.0547,
      "step": 177825
    },
    {
      "epoch": 2.619254502879192,
      "grad_norm": 1.2711896896362305,
      "learning_rate": 1.0585886619059152e-05,
      "loss": 0.0626,
      "step": 177850
    },
    {
      "epoch": 2.619622685969279,
      "grad_norm": 1.4679142236709595,
      "learning_rate": 1.0584250247092167e-05,
      "loss": 0.0562,
      "step": 177875
    },
    {
      "epoch": 2.619990869059366,
      "grad_norm": 1.3813141584396362,
      "learning_rate": 1.0582613875125183e-05,
      "loss": 0.0577,
      "step": 177900
    },
    {
      "epoch": 2.620359052149453,
      "grad_norm": 1.1439592838287354,
      "learning_rate": 1.0580977503158199e-05,
      "loss": 0.0544,
      "step": 177925
    },
    {
      "epoch": 2.62072723523954,
      "grad_norm": 1.5633246898651123,
      "learning_rate": 1.0579341131191215e-05,
      "loss": 0.0523,
      "step": 177950
    },
    {
      "epoch": 2.621095418329627,
      "grad_norm": 1.2544136047363281,
      "learning_rate": 1.057770475922423e-05,
      "loss": 0.0643,
      "step": 177975
    },
    {
      "epoch": 2.621463601419714,
      "grad_norm": 1.5179665088653564,
      "learning_rate": 1.0576068387257246e-05,
      "loss": 0.0619,
      "step": 178000
    },
    {
      "epoch": 2.621831784509801,
      "grad_norm": 1.1079788208007812,
      "learning_rate": 1.0574432015290262e-05,
      "loss": 0.06,
      "step": 178025
    },
    {
      "epoch": 2.622199967599888,
      "grad_norm": 1.4913833141326904,
      "learning_rate": 1.0572795643323276e-05,
      "loss": 0.0645,
      "step": 178050
    },
    {
      "epoch": 2.622568150689975,
      "grad_norm": 1.2368390560150146,
      "learning_rate": 1.057115927135629e-05,
      "loss": 0.0581,
      "step": 178075
    },
    {
      "epoch": 2.6229363337800624,
      "grad_norm": 1.7132432460784912,
      "learning_rate": 1.0569522899389307e-05,
      "loss": 0.0593,
      "step": 178100
    },
    {
      "epoch": 2.6233045168701494,
      "grad_norm": 1.515574336051941,
      "learning_rate": 1.0567886527422321e-05,
      "loss": 0.0579,
      "step": 178125
    },
    {
      "epoch": 2.6236726999602364,
      "grad_norm": 1.7901968955993652,
      "learning_rate": 1.0566250155455337e-05,
      "loss": 0.0569,
      "step": 178150
    },
    {
      "epoch": 2.6240408830503235,
      "grad_norm": 1.4704879522323608,
      "learning_rate": 1.0564613783488354e-05,
      "loss": 0.05,
      "step": 178175
    },
    {
      "epoch": 2.6244090661404105,
      "grad_norm": 0.9033634066581726,
      "learning_rate": 1.056297741152137e-05,
      "loss": 0.0617,
      "step": 178200
    },
    {
      "epoch": 2.6247772492304975,
      "grad_norm": 1.3718925714492798,
      "learning_rate": 1.0561341039554384e-05,
      "loss": 0.0597,
      "step": 178225
    },
    {
      "epoch": 2.6251454323205845,
      "grad_norm": 1.5202617645263672,
      "learning_rate": 1.05597046675874e-05,
      "loss": 0.0643,
      "step": 178250
    },
    {
      "epoch": 2.6255136154106715,
      "grad_norm": 1.2252256870269775,
      "learning_rate": 1.0558068295620413e-05,
      "loss": 0.0608,
      "step": 178275
    },
    {
      "epoch": 2.6258817985007585,
      "grad_norm": 1.1312342882156372,
      "learning_rate": 1.055643192365343e-05,
      "loss": 0.0628,
      "step": 178300
    },
    {
      "epoch": 2.6262499815908456,
      "grad_norm": 1.2931057214736938,
      "learning_rate": 1.0554795551686445e-05,
      "loss": 0.0628,
      "step": 178325
    },
    {
      "epoch": 2.6266181646809326,
      "grad_norm": 1.1888556480407715,
      "learning_rate": 1.0553159179719462e-05,
      "loss": 0.0635,
      "step": 178350
    },
    {
      "epoch": 2.6269863477710196,
      "grad_norm": 1.7914178371429443,
      "learning_rate": 1.0551522807752476e-05,
      "loss": 0.0554,
      "step": 178375
    },
    {
      "epoch": 2.6273545308611066,
      "grad_norm": 1.2818809747695923,
      "learning_rate": 1.0549886435785492e-05,
      "loss": 0.062,
      "step": 178400
    },
    {
      "epoch": 2.6277227139511936,
      "grad_norm": 1.2769396305084229,
      "learning_rate": 1.0548250063818508e-05,
      "loss": 0.0591,
      "step": 178425
    },
    {
      "epoch": 2.6280908970412806,
      "grad_norm": 1.2644082307815552,
      "learning_rate": 1.0546613691851525e-05,
      "loss": 0.0544,
      "step": 178450
    },
    {
      "epoch": 2.6284590801313676,
      "grad_norm": 1.585900068283081,
      "learning_rate": 1.0544977319884539e-05,
      "loss": 0.0576,
      "step": 178475
    },
    {
      "epoch": 2.6288272632214547,
      "grad_norm": 1.2027477025985718,
      "learning_rate": 1.0543340947917553e-05,
      "loss": 0.0575,
      "step": 178500
    },
    {
      "epoch": 2.6291954463115417,
      "grad_norm": 1.3439468145370483,
      "learning_rate": 1.0541704575950568e-05,
      "loss": 0.0623,
      "step": 178525
    },
    {
      "epoch": 2.6295636294016287,
      "grad_norm": 1.4071685075759888,
      "learning_rate": 1.0540068203983584e-05,
      "loss": 0.0554,
      "step": 178550
    },
    {
      "epoch": 2.6299318124917157,
      "grad_norm": 1.3623868227005005,
      "learning_rate": 1.05384318320166e-05,
      "loss": 0.0625,
      "step": 178575
    },
    {
      "epoch": 2.6302999955818027,
      "grad_norm": 1.6273113489151,
      "learning_rate": 1.0536795460049616e-05,
      "loss": 0.0586,
      "step": 178600
    },
    {
      "epoch": 2.6306681786718897,
      "grad_norm": 1.7515143156051636,
      "learning_rate": 1.053515908808263e-05,
      "loss": 0.0587,
      "step": 178625
    },
    {
      "epoch": 2.6310363617619767,
      "grad_norm": 1.0457935333251953,
      "learning_rate": 1.0533522716115647e-05,
      "loss": 0.0591,
      "step": 178650
    },
    {
      "epoch": 2.6314045448520638,
      "grad_norm": 0.7781147360801697,
      "learning_rate": 1.0531886344148663e-05,
      "loss": 0.056,
      "step": 178675
    },
    {
      "epoch": 2.6317727279421512,
      "grad_norm": 1.553816318511963,
      "learning_rate": 1.0530249972181676e-05,
      "loss": 0.0545,
      "step": 178700
    },
    {
      "epoch": 2.6321409110322382,
      "grad_norm": 1.633156657218933,
      "learning_rate": 1.0528613600214692e-05,
      "loss": 0.0559,
      "step": 178725
    },
    {
      "epoch": 2.6325090941223253,
      "grad_norm": 1.4244391918182373,
      "learning_rate": 1.0526977228247708e-05,
      "loss": 0.0552,
      "step": 178750
    },
    {
      "epoch": 2.6328772772124123,
      "grad_norm": 0.9048569202423096,
      "learning_rate": 1.0525340856280724e-05,
      "loss": 0.0548,
      "step": 178775
    },
    {
      "epoch": 2.6332454603024993,
      "grad_norm": 1.279308795928955,
      "learning_rate": 1.0523704484313739e-05,
      "loss": 0.0544,
      "step": 178800
    },
    {
      "epoch": 2.6336136433925863,
      "grad_norm": 1.0054556131362915,
      "learning_rate": 1.0522068112346755e-05,
      "loss": 0.0565,
      "step": 178825
    },
    {
      "epoch": 2.6339818264826733,
      "grad_norm": 1.6993483304977417,
      "learning_rate": 1.0520431740379771e-05,
      "loss": 0.0599,
      "step": 178850
    },
    {
      "epoch": 2.6343500095727603,
      "grad_norm": 1.3486400842666626,
      "learning_rate": 1.0518795368412786e-05,
      "loss": 0.0606,
      "step": 178875
    },
    {
      "epoch": 2.6347181926628473,
      "grad_norm": 1.2938765287399292,
      "learning_rate": 1.05171589964458e-05,
      "loss": 0.0579,
      "step": 178900
    },
    {
      "epoch": 2.6350863757529344,
      "grad_norm": 1.1513739824295044,
      "learning_rate": 1.0515522624478816e-05,
      "loss": 0.0558,
      "step": 178925
    },
    {
      "epoch": 2.6354545588430214,
      "grad_norm": 1.2724298238754272,
      "learning_rate": 1.051388625251183e-05,
      "loss": 0.0531,
      "step": 178950
    },
    {
      "epoch": 2.6358227419331084,
      "grad_norm": 1.5667345523834229,
      "learning_rate": 1.0512249880544847e-05,
      "loss": 0.0578,
      "step": 178975
    },
    {
      "epoch": 2.6361909250231954,
      "grad_norm": 1.5931990146636963,
      "learning_rate": 1.0510613508577863e-05,
      "loss": 0.0597,
      "step": 179000
    },
    {
      "epoch": 2.6365591081132824,
      "grad_norm": 1.4145405292510986,
      "learning_rate": 1.0508977136610879e-05,
      "loss": 0.0558,
      "step": 179025
    },
    {
      "epoch": 2.63692729120337,
      "grad_norm": 1.402555227279663,
      "learning_rate": 1.0507340764643894e-05,
      "loss": 0.0583,
      "step": 179050
    },
    {
      "epoch": 2.637295474293457,
      "grad_norm": 1.4087753295898438,
      "learning_rate": 1.050570439267691e-05,
      "loss": 0.0596,
      "step": 179075
    },
    {
      "epoch": 2.637663657383544,
      "grad_norm": 1.6323907375335693,
      "learning_rate": 1.0504068020709926e-05,
      "loss": 0.0617,
      "step": 179100
    },
    {
      "epoch": 2.638031840473631,
      "grad_norm": 1.080856442451477,
      "learning_rate": 1.0502431648742939e-05,
      "loss": 0.0622,
      "step": 179125
    },
    {
      "epoch": 2.638400023563718,
      "grad_norm": 1.7374264001846313,
      "learning_rate": 1.0500795276775955e-05,
      "loss": 0.0564,
      "step": 179150
    },
    {
      "epoch": 2.638768206653805,
      "grad_norm": 1.0557068586349487,
      "learning_rate": 1.0499158904808971e-05,
      "loss": 0.0551,
      "step": 179175
    },
    {
      "epoch": 2.639136389743892,
      "grad_norm": 0.7824804782867432,
      "learning_rate": 1.0497522532841985e-05,
      "loss": 0.05,
      "step": 179200
    },
    {
      "epoch": 2.639504572833979,
      "grad_norm": 1.5242996215820312,
      "learning_rate": 1.0495886160875001e-05,
      "loss": 0.0595,
      "step": 179225
    },
    {
      "epoch": 2.639872755924066,
      "grad_norm": 1.93537437915802,
      "learning_rate": 1.0494249788908018e-05,
      "loss": 0.0596,
      "step": 179250
    },
    {
      "epoch": 2.640240939014153,
      "grad_norm": 0.8486847281455994,
      "learning_rate": 1.0492613416941034e-05,
      "loss": 0.0664,
      "step": 179275
    },
    {
      "epoch": 2.64060912210424,
      "grad_norm": 1.0348448753356934,
      "learning_rate": 1.0490977044974048e-05,
      "loss": 0.0648,
      "step": 179300
    },
    {
      "epoch": 2.640977305194327,
      "grad_norm": 1.3246134519577026,
      "learning_rate": 1.0489340673007063e-05,
      "loss": 0.0591,
      "step": 179325
    },
    {
      "epoch": 2.641345488284414,
      "grad_norm": 1.2293293476104736,
      "learning_rate": 1.0487704301040079e-05,
      "loss": 0.0618,
      "step": 179350
    },
    {
      "epoch": 2.641713671374501,
      "grad_norm": 1.6534684896469116,
      "learning_rate": 1.0486067929073093e-05,
      "loss": 0.0616,
      "step": 179375
    },
    {
      "epoch": 2.642081854464588,
      "grad_norm": 1.2842737436294556,
      "learning_rate": 1.048443155710611e-05,
      "loss": 0.0509,
      "step": 179400
    },
    {
      "epoch": 2.642450037554675,
      "grad_norm": 1.3370314836502075,
      "learning_rate": 1.0482795185139126e-05,
      "loss": 0.0632,
      "step": 179425
    },
    {
      "epoch": 2.642818220644762,
      "grad_norm": 1.1738921403884888,
      "learning_rate": 1.048115881317214e-05,
      "loss": 0.0631,
      "step": 179450
    },
    {
      "epoch": 2.643186403734849,
      "grad_norm": 1.4158613681793213,
      "learning_rate": 1.0479522441205156e-05,
      "loss": 0.0536,
      "step": 179475
    },
    {
      "epoch": 2.643554586824936,
      "grad_norm": 1.729436993598938,
      "learning_rate": 1.0477886069238172e-05,
      "loss": 0.06,
      "step": 179500
    },
    {
      "epoch": 2.643922769915023,
      "grad_norm": 1.5001217126846313,
      "learning_rate": 1.0476249697271189e-05,
      "loss": 0.0559,
      "step": 179525
    },
    {
      "epoch": 2.64429095300511,
      "grad_norm": 1.3951103687286377,
      "learning_rate": 1.0474613325304201e-05,
      "loss": 0.0672,
      "step": 179550
    },
    {
      "epoch": 2.644659136095197,
      "grad_norm": 1.2371768951416016,
      "learning_rate": 1.0472976953337217e-05,
      "loss": 0.0588,
      "step": 179575
    },
    {
      "epoch": 2.645027319185284,
      "grad_norm": 1.9210978746414185,
      "learning_rate": 1.0471340581370234e-05,
      "loss": 0.0629,
      "step": 179600
    },
    {
      "epoch": 2.6453955022753712,
      "grad_norm": 1.996050238609314,
      "learning_rate": 1.0469704209403248e-05,
      "loss": 0.0604,
      "step": 179625
    },
    {
      "epoch": 2.6457636853654587,
      "grad_norm": 1.2609702348709106,
      "learning_rate": 1.0468067837436264e-05,
      "loss": 0.0575,
      "step": 179650
    },
    {
      "epoch": 2.6461318684555457,
      "grad_norm": 1.7031487226486206,
      "learning_rate": 1.046643146546928e-05,
      "loss": 0.0625,
      "step": 179675
    },
    {
      "epoch": 2.6465000515456327,
      "grad_norm": 1.2939832210540771,
      "learning_rate": 1.0464795093502297e-05,
      "loss": 0.058,
      "step": 179700
    },
    {
      "epoch": 2.6468682346357197,
      "grad_norm": 1.6751508712768555,
      "learning_rate": 1.0463158721535311e-05,
      "loss": 0.0618,
      "step": 179725
    },
    {
      "epoch": 2.6472364177258068,
      "grad_norm": 1.25525963306427,
      "learning_rate": 1.0461522349568325e-05,
      "loss": 0.0661,
      "step": 179750
    },
    {
      "epoch": 2.6476046008158938,
      "grad_norm": 0.7105354070663452,
      "learning_rate": 1.045988597760134e-05,
      "loss": 0.0616,
      "step": 179775
    },
    {
      "epoch": 2.647972783905981,
      "grad_norm": 1.251990556716919,
      "learning_rate": 1.0458249605634356e-05,
      "loss": 0.0592,
      "step": 179800
    },
    {
      "epoch": 2.648340966996068,
      "grad_norm": 0.9045290946960449,
      "learning_rate": 1.0456613233667372e-05,
      "loss": 0.0612,
      "step": 179825
    },
    {
      "epoch": 2.648709150086155,
      "grad_norm": 1.2470132112503052,
      "learning_rate": 1.0454976861700388e-05,
      "loss": 0.0642,
      "step": 179850
    },
    {
      "epoch": 2.649077333176242,
      "grad_norm": 1.1372416019439697,
      "learning_rate": 1.0453340489733403e-05,
      "loss": 0.0578,
      "step": 179875
    },
    {
      "epoch": 2.649445516266329,
      "grad_norm": 1.4001555442810059,
      "learning_rate": 1.0451704117766419e-05,
      "loss": 0.0601,
      "step": 179900
    },
    {
      "epoch": 2.649813699356416,
      "grad_norm": 1.0540432929992676,
      "learning_rate": 1.0450067745799435e-05,
      "loss": 0.0537,
      "step": 179925
    },
    {
      "epoch": 2.650181882446503,
      "grad_norm": 0.8042251467704773,
      "learning_rate": 1.0448431373832451e-05,
      "loss": 0.0598,
      "step": 179950
    },
    {
      "epoch": 2.65055006553659,
      "grad_norm": 1.184349775314331,
      "learning_rate": 1.0446795001865464e-05,
      "loss": 0.0679,
      "step": 179975
    },
    {
      "epoch": 2.650918248626677,
      "grad_norm": 1.085429072380066,
      "learning_rate": 1.044515862989848e-05,
      "loss": 0.0613,
      "step": 180000
    },
    {
      "epoch": 2.6512864317167644,
      "grad_norm": 1.4456406831741333,
      "learning_rate": 1.0443522257931495e-05,
      "loss": 0.0649,
      "step": 180025
    },
    {
      "epoch": 2.6516546148068514,
      "grad_norm": 1.3142204284667969,
      "learning_rate": 1.044188588596451e-05,
      "loss": 0.0624,
      "step": 180050
    },
    {
      "epoch": 2.6520227978969384,
      "grad_norm": 1.2352559566497803,
      "learning_rate": 1.0440249513997527e-05,
      "loss": 0.0614,
      "step": 180075
    },
    {
      "epoch": 2.6523909809870254,
      "grad_norm": 1.8774738311767578,
      "learning_rate": 1.0438613142030543e-05,
      "loss": 0.0564,
      "step": 180100
    },
    {
      "epoch": 2.6527591640771124,
      "grad_norm": 1.3439046144485474,
      "learning_rate": 1.0436976770063558e-05,
      "loss": 0.0552,
      "step": 180125
    },
    {
      "epoch": 2.6531273471671994,
      "grad_norm": 1.3229684829711914,
      "learning_rate": 1.0435340398096574e-05,
      "loss": 0.0612,
      "step": 180150
    },
    {
      "epoch": 2.6534955302572865,
      "grad_norm": 1.5237793922424316,
      "learning_rate": 1.0433704026129588e-05,
      "loss": 0.0587,
      "step": 180175
    },
    {
      "epoch": 2.6538637133473735,
      "grad_norm": 1.4181277751922607,
      "learning_rate": 1.0432067654162603e-05,
      "loss": 0.0579,
      "step": 180200
    },
    {
      "epoch": 2.6542318964374605,
      "grad_norm": 0.8959699869155884,
      "learning_rate": 1.0430431282195619e-05,
      "loss": 0.0641,
      "step": 180225
    },
    {
      "epoch": 2.6546000795275475,
      "grad_norm": 1.299402117729187,
      "learning_rate": 1.0428794910228635e-05,
      "loss": 0.0669,
      "step": 180250
    },
    {
      "epoch": 2.6549682626176345,
      "grad_norm": 1.4478684663772583,
      "learning_rate": 1.0427158538261651e-05,
      "loss": 0.0588,
      "step": 180275
    },
    {
      "epoch": 2.6553364457077215,
      "grad_norm": 1.7965478897094727,
      "learning_rate": 1.0425522166294666e-05,
      "loss": 0.0587,
      "step": 180300
    },
    {
      "epoch": 2.6557046287978086,
      "grad_norm": 1.7480050325393677,
      "learning_rate": 1.0423885794327682e-05,
      "loss": 0.06,
      "step": 180325
    },
    {
      "epoch": 2.6560728118878956,
      "grad_norm": 1.9673129320144653,
      "learning_rate": 1.0422249422360698e-05,
      "loss": 0.0626,
      "step": 180350
    },
    {
      "epoch": 2.6564409949779826,
      "grad_norm": 1.3611135482788086,
      "learning_rate": 1.042061305039371e-05,
      "loss": 0.059,
      "step": 180375
    },
    {
      "epoch": 2.6568091780680696,
      "grad_norm": 1.392541766166687,
      "learning_rate": 1.0418976678426727e-05,
      "loss": 0.0619,
      "step": 180400
    },
    {
      "epoch": 2.6571773611581566,
      "grad_norm": 1.6021356582641602,
      "learning_rate": 1.0417340306459743e-05,
      "loss": 0.0605,
      "step": 180425
    },
    {
      "epoch": 2.6575455442482436,
      "grad_norm": 1.6446949243545532,
      "learning_rate": 1.0415703934492757e-05,
      "loss": 0.0616,
      "step": 180450
    },
    {
      "epoch": 2.6579137273383306,
      "grad_norm": 1.6089822053909302,
      "learning_rate": 1.0414067562525773e-05,
      "loss": 0.0617,
      "step": 180475
    },
    {
      "epoch": 2.6582819104284177,
      "grad_norm": 1.604913592338562,
      "learning_rate": 1.041243119055879e-05,
      "loss": 0.0607,
      "step": 180500
    },
    {
      "epoch": 2.6586500935185047,
      "grad_norm": 1.532718539237976,
      "learning_rate": 1.0410794818591806e-05,
      "loss": 0.0671,
      "step": 180525
    },
    {
      "epoch": 2.6590182766085917,
      "grad_norm": 1.3487820625305176,
      "learning_rate": 1.040915844662482e-05,
      "loss": 0.0561,
      "step": 180550
    },
    {
      "epoch": 2.6593864596986787,
      "grad_norm": 1.6589494943618774,
      "learning_rate": 1.0407522074657836e-05,
      "loss": 0.0679,
      "step": 180575
    },
    {
      "epoch": 2.6597546427887657,
      "grad_norm": 1.7940930128097534,
      "learning_rate": 1.040588570269085e-05,
      "loss": 0.0604,
      "step": 180600
    },
    {
      "epoch": 2.660122825878853,
      "grad_norm": 1.5940213203430176,
      "learning_rate": 1.0404249330723865e-05,
      "loss": 0.0629,
      "step": 180625
    },
    {
      "epoch": 2.66049100896894,
      "grad_norm": 1.478488564491272,
      "learning_rate": 1.0402612958756881e-05,
      "loss": 0.0647,
      "step": 180650
    },
    {
      "epoch": 2.660859192059027,
      "grad_norm": 1.3178554773330688,
      "learning_rate": 1.0400976586789898e-05,
      "loss": 0.058,
      "step": 180675
    },
    {
      "epoch": 2.6612273751491142,
      "grad_norm": 1.3025761842727661,
      "learning_rate": 1.0399340214822912e-05,
      "loss": 0.0574,
      "step": 180700
    },
    {
      "epoch": 2.6615955582392012,
      "grad_norm": 1.010762095451355,
      "learning_rate": 1.0397703842855928e-05,
      "loss": 0.0566,
      "step": 180725
    },
    {
      "epoch": 2.6619637413292883,
      "grad_norm": 1.4706687927246094,
      "learning_rate": 1.0396067470888944e-05,
      "loss": 0.0647,
      "step": 180750
    },
    {
      "epoch": 2.6623319244193753,
      "grad_norm": 1.3642903566360474,
      "learning_rate": 1.039443109892196e-05,
      "loss": 0.0571,
      "step": 180775
    },
    {
      "epoch": 2.6627001075094623,
      "grad_norm": 1.0870933532714844,
      "learning_rate": 1.0392794726954973e-05,
      "loss": 0.0578,
      "step": 180800
    },
    {
      "epoch": 2.6630682905995493,
      "grad_norm": 0.9486274123191833,
      "learning_rate": 1.039115835498799e-05,
      "loss": 0.0534,
      "step": 180825
    },
    {
      "epoch": 2.6634364736896363,
      "grad_norm": 1.5033544301986694,
      "learning_rate": 1.0389521983021006e-05,
      "loss": 0.0597,
      "step": 180850
    },
    {
      "epoch": 2.6638046567797233,
      "grad_norm": 1.4026621580123901,
      "learning_rate": 1.038788561105402e-05,
      "loss": 0.0551,
      "step": 180875
    },
    {
      "epoch": 2.6641728398698103,
      "grad_norm": 1.653326392173767,
      "learning_rate": 1.0386249239087036e-05,
      "loss": 0.0607,
      "step": 180900
    },
    {
      "epoch": 2.6645410229598974,
      "grad_norm": 1.20106840133667,
      "learning_rate": 1.0384612867120052e-05,
      "loss": 0.0609,
      "step": 180925
    },
    {
      "epoch": 2.6649092060499844,
      "grad_norm": 1.3902958631515503,
      "learning_rate": 1.0382976495153067e-05,
      "loss": 0.0563,
      "step": 180950
    },
    {
      "epoch": 2.665277389140072,
      "grad_norm": 1.240664005279541,
      "learning_rate": 1.0381340123186083e-05,
      "loss": 0.0548,
      "step": 180975
    },
    {
      "epoch": 2.665645572230159,
      "grad_norm": 1.4866517782211304,
      "learning_rate": 1.0379703751219099e-05,
      "loss": 0.0525,
      "step": 181000
    },
    {
      "epoch": 2.666013755320246,
      "grad_norm": 1.3387285470962524,
      "learning_rate": 1.0378132834130791e-05,
      "loss": 0.0574,
      "step": 181025
    },
    {
      "epoch": 2.666381938410333,
      "grad_norm": 1.2695602178573608,
      "learning_rate": 1.0376496462163807e-05,
      "loss": 0.0538,
      "step": 181050
    },
    {
      "epoch": 2.66675012150042,
      "grad_norm": 0.8715903162956238,
      "learning_rate": 1.0374860090196824e-05,
      "loss": 0.0577,
      "step": 181075
    },
    {
      "epoch": 2.667118304590507,
      "grad_norm": 1.5755671262741089,
      "learning_rate": 1.037322371822984e-05,
      "loss": 0.0568,
      "step": 181100
    },
    {
      "epoch": 2.667486487680594,
      "grad_norm": 1.0684443712234497,
      "learning_rate": 1.0371587346262854e-05,
      "loss": 0.0573,
      "step": 181125
    },
    {
      "epoch": 2.667854670770681,
      "grad_norm": 1.036328911781311,
      "learning_rate": 1.036995097429587e-05,
      "loss": 0.0599,
      "step": 181150
    },
    {
      "epoch": 2.668222853860768,
      "grad_norm": 1.6287823915481567,
      "learning_rate": 1.0368314602328887e-05,
      "loss": 0.0638,
      "step": 181175
    },
    {
      "epoch": 2.668591036950855,
      "grad_norm": 1.5044418573379517,
      "learning_rate": 1.0366678230361901e-05,
      "loss": 0.0543,
      "step": 181200
    },
    {
      "epoch": 2.668959220040942,
      "grad_norm": 1.2453765869140625,
      "learning_rate": 1.0365041858394915e-05,
      "loss": 0.0606,
      "step": 181225
    },
    {
      "epoch": 2.669327403131029,
      "grad_norm": 1.1694879531860352,
      "learning_rate": 1.0363405486427932e-05,
      "loss": 0.0556,
      "step": 181250
    },
    {
      "epoch": 2.669695586221116,
      "grad_norm": 1.2286295890808105,
      "learning_rate": 1.0361769114460946e-05,
      "loss": 0.0574,
      "step": 181275
    },
    {
      "epoch": 2.670063769311203,
      "grad_norm": 1.068738579750061,
      "learning_rate": 1.0360132742493962e-05,
      "loss": 0.0645,
      "step": 181300
    },
    {
      "epoch": 2.67043195240129,
      "grad_norm": 1.381239891052246,
      "learning_rate": 1.0358496370526978e-05,
      "loss": 0.0567,
      "step": 181325
    },
    {
      "epoch": 2.670800135491377,
      "grad_norm": 1.7819762229919434,
      "learning_rate": 1.0356859998559994e-05,
      "loss": 0.0624,
      "step": 181350
    },
    {
      "epoch": 2.671168318581464,
      "grad_norm": 1.2299953699111938,
      "learning_rate": 1.0355223626593009e-05,
      "loss": 0.0551,
      "step": 181375
    },
    {
      "epoch": 2.671536501671551,
      "grad_norm": 1.6271673440933228,
      "learning_rate": 1.0353587254626025e-05,
      "loss": 0.0583,
      "step": 181400
    },
    {
      "epoch": 2.671904684761638,
      "grad_norm": 1.5756698846817017,
      "learning_rate": 1.0351950882659041e-05,
      "loss": 0.0567,
      "step": 181425
    },
    {
      "epoch": 2.672272867851725,
      "grad_norm": 1.4155449867248535,
      "learning_rate": 1.0350314510692054e-05,
      "loss": 0.0588,
      "step": 181450
    },
    {
      "epoch": 2.672641050941812,
      "grad_norm": 1.3141456842422485,
      "learning_rate": 1.034867813872507e-05,
      "loss": 0.065,
      "step": 181475
    },
    {
      "epoch": 2.673009234031899,
      "grad_norm": 1.2345174551010132,
      "learning_rate": 1.0347041766758086e-05,
      "loss": 0.0513,
      "step": 181500
    },
    {
      "epoch": 2.673377417121986,
      "grad_norm": 1.3157939910888672,
      "learning_rate": 1.03454053947911e-05,
      "loss": 0.0706,
      "step": 181525
    },
    {
      "epoch": 2.673745600212073,
      "grad_norm": 1.5634199380874634,
      "learning_rate": 1.0343769022824117e-05,
      "loss": 0.0549,
      "step": 181550
    },
    {
      "epoch": 2.6741137833021607,
      "grad_norm": 1.149841070175171,
      "learning_rate": 1.0342132650857133e-05,
      "loss": 0.0657,
      "step": 181575
    },
    {
      "epoch": 2.6744819663922477,
      "grad_norm": 1.2981230020523071,
      "learning_rate": 1.034049627889015e-05,
      "loss": 0.0591,
      "step": 181600
    },
    {
      "epoch": 2.6748501494823347,
      "grad_norm": 0.9457674026489258,
      "learning_rate": 1.0338859906923164e-05,
      "loss": 0.0559,
      "step": 181625
    },
    {
      "epoch": 2.6752183325724217,
      "grad_norm": 1.519731044769287,
      "learning_rate": 1.0337223534956178e-05,
      "loss": 0.0629,
      "step": 181650
    },
    {
      "epoch": 2.6755865156625087,
      "grad_norm": 0.9381436109542847,
      "learning_rate": 1.0335587162989194e-05,
      "loss": 0.0591,
      "step": 181675
    },
    {
      "epoch": 2.6759546987525957,
      "grad_norm": 1.9244869947433472,
      "learning_rate": 1.0333950791022209e-05,
      "loss": 0.0602,
      "step": 181700
    },
    {
      "epoch": 2.6763228818426827,
      "grad_norm": 1.1269614696502686,
      "learning_rate": 1.0332314419055225e-05,
      "loss": 0.0569,
      "step": 181725
    },
    {
      "epoch": 2.6766910649327698,
      "grad_norm": 1.834871530532837,
      "learning_rate": 1.0330678047088241e-05,
      "loss": 0.0593,
      "step": 181750
    },
    {
      "epoch": 2.6770592480228568,
      "grad_norm": 2.0192348957061768,
      "learning_rate": 1.0329041675121256e-05,
      "loss": 0.065,
      "step": 181775
    },
    {
      "epoch": 2.677427431112944,
      "grad_norm": 1.3955726623535156,
      "learning_rate": 1.0327405303154272e-05,
      "loss": 0.0611,
      "step": 181800
    },
    {
      "epoch": 2.677795614203031,
      "grad_norm": 0.8697972297668457,
      "learning_rate": 1.0325768931187288e-05,
      "loss": 0.0708,
      "step": 181825
    },
    {
      "epoch": 2.678163797293118,
      "grad_norm": 1.6199911832809448,
      "learning_rate": 1.03241325592203e-05,
      "loss": 0.0612,
      "step": 181850
    },
    {
      "epoch": 2.678531980383205,
      "grad_norm": 1.4861690998077393,
      "learning_rate": 1.0322496187253317e-05,
      "loss": 0.0614,
      "step": 181875
    },
    {
      "epoch": 2.678900163473292,
      "grad_norm": 1.5810816287994385,
      "learning_rate": 1.0320859815286333e-05,
      "loss": 0.061,
      "step": 181900
    },
    {
      "epoch": 2.6792683465633793,
      "grad_norm": 0.9216818809509277,
      "learning_rate": 1.0319223443319349e-05,
      "loss": 0.0607,
      "step": 181925
    },
    {
      "epoch": 2.6796365296534663,
      "grad_norm": 1.6507874727249146,
      "learning_rate": 1.0317587071352363e-05,
      "loss": 0.0598,
      "step": 181950
    },
    {
      "epoch": 2.6800047127435533,
      "grad_norm": 0.9667916297912598,
      "learning_rate": 1.031595069938538e-05,
      "loss": 0.0605,
      "step": 181975
    },
    {
      "epoch": 2.6803728958336404,
      "grad_norm": 1.180269718170166,
      "learning_rate": 1.0314314327418396e-05,
      "loss": 0.0568,
      "step": 182000
    },
    {
      "epoch": 2.6807410789237274,
      "grad_norm": 1.100769281387329,
      "learning_rate": 1.0312677955451412e-05,
      "loss": 0.0626,
      "step": 182025
    },
    {
      "epoch": 2.6811092620138144,
      "grad_norm": 1.3334780931472778,
      "learning_rate": 1.0311041583484426e-05,
      "loss": 0.0689,
      "step": 182050
    },
    {
      "epoch": 2.6814774451039014,
      "grad_norm": 1.484243631362915,
      "learning_rate": 1.0309405211517441e-05,
      "loss": 0.0606,
      "step": 182075
    },
    {
      "epoch": 2.6818456281939884,
      "grad_norm": 1.574515700340271,
      "learning_rate": 1.0307768839550455e-05,
      "loss": 0.0495,
      "step": 182100
    },
    {
      "epoch": 2.6822138112840754,
      "grad_norm": 1.226787805557251,
      "learning_rate": 1.0306132467583471e-05,
      "loss": 0.0526,
      "step": 182125
    },
    {
      "epoch": 2.6825819943741624,
      "grad_norm": 1.495360255241394,
      "learning_rate": 1.0304496095616488e-05,
      "loss": 0.0601,
      "step": 182150
    },
    {
      "epoch": 2.6829501774642495,
      "grad_norm": 1.9710766077041626,
      "learning_rate": 1.0302859723649504e-05,
      "loss": 0.062,
      "step": 182175
    },
    {
      "epoch": 2.6833183605543365,
      "grad_norm": 1.479754090309143,
      "learning_rate": 1.0301223351682518e-05,
      "loss": 0.0616,
      "step": 182200
    },
    {
      "epoch": 2.6836865436444235,
      "grad_norm": 1.2248257398605347,
      "learning_rate": 1.0299586979715534e-05,
      "loss": 0.0529,
      "step": 182225
    },
    {
      "epoch": 2.6840547267345105,
      "grad_norm": 1.264366626739502,
      "learning_rate": 1.029795060774855e-05,
      "loss": 0.0598,
      "step": 182250
    },
    {
      "epoch": 2.6844229098245975,
      "grad_norm": 1.2543889284133911,
      "learning_rate": 1.0296314235781563e-05,
      "loss": 0.0615,
      "step": 182275
    },
    {
      "epoch": 2.6847910929146845,
      "grad_norm": 1.1692509651184082,
      "learning_rate": 1.029467786381458e-05,
      "loss": 0.0611,
      "step": 182300
    },
    {
      "epoch": 2.6851592760047716,
      "grad_norm": 2.0396153926849365,
      "learning_rate": 1.0293041491847596e-05,
      "loss": 0.0589,
      "step": 182325
    },
    {
      "epoch": 2.6855274590948586,
      "grad_norm": 1.1993850469589233,
      "learning_rate": 1.029140511988061e-05,
      "loss": 0.0634,
      "step": 182350
    },
    {
      "epoch": 2.6858956421849456,
      "grad_norm": 1.1579844951629639,
      "learning_rate": 1.0289768747913626e-05,
      "loss": 0.0566,
      "step": 182375
    },
    {
      "epoch": 2.6862638252750326,
      "grad_norm": 0.9792632460594177,
      "learning_rate": 1.0288132375946642e-05,
      "loss": 0.0551,
      "step": 182400
    },
    {
      "epoch": 2.6866320083651196,
      "grad_norm": 1.3327845335006714,
      "learning_rate": 1.0286496003979659e-05,
      "loss": 0.0551,
      "step": 182425
    },
    {
      "epoch": 2.6870001914552066,
      "grad_norm": 1.2504613399505615,
      "learning_rate": 1.0284859632012673e-05,
      "loss": 0.0575,
      "step": 182450
    },
    {
      "epoch": 2.6873683745452936,
      "grad_norm": 1.36167311668396,
      "learning_rate": 1.0283223260045689e-05,
      "loss": 0.0582,
      "step": 182475
    },
    {
      "epoch": 2.6877365576353807,
      "grad_norm": 1.4674620628356934,
      "learning_rate": 1.0281586888078704e-05,
      "loss": 0.0553,
      "step": 182500
    },
    {
      "epoch": 2.688104740725468,
      "grad_norm": 1.6915929317474365,
      "learning_rate": 1.0279950516111718e-05,
      "loss": 0.0558,
      "step": 182525
    },
    {
      "epoch": 2.688472923815555,
      "grad_norm": 1.7129229307174683,
      "learning_rate": 1.0278314144144734e-05,
      "loss": 0.058,
      "step": 182550
    },
    {
      "epoch": 2.688841106905642,
      "grad_norm": 1.9740345478057861,
      "learning_rate": 1.027667777217775e-05,
      "loss": 0.0554,
      "step": 182575
    },
    {
      "epoch": 2.689209289995729,
      "grad_norm": 1.830592393875122,
      "learning_rate": 1.0275041400210766e-05,
      "loss": 0.0591,
      "step": 182600
    },
    {
      "epoch": 2.689577473085816,
      "grad_norm": 1.1833752393722534,
      "learning_rate": 1.0273405028243781e-05,
      "loss": 0.0594,
      "step": 182625
    },
    {
      "epoch": 2.689945656175903,
      "grad_norm": 1.2470561265945435,
      "learning_rate": 1.0271768656276797e-05,
      "loss": 0.0619,
      "step": 182650
    },
    {
      "epoch": 2.69031383926599,
      "grad_norm": 1.4791810512542725,
      "learning_rate": 1.0270132284309813e-05,
      "loss": 0.062,
      "step": 182675
    },
    {
      "epoch": 2.6906820223560772,
      "grad_norm": 1.0999714136123657,
      "learning_rate": 1.0268495912342826e-05,
      "loss": 0.0613,
      "step": 182700
    },
    {
      "epoch": 2.6910502054461642,
      "grad_norm": 1.560540795326233,
      "learning_rate": 1.0266859540375842e-05,
      "loss": 0.0626,
      "step": 182725
    },
    {
      "epoch": 2.6914183885362513,
      "grad_norm": 1.7665736675262451,
      "learning_rate": 1.0265223168408858e-05,
      "loss": 0.0655,
      "step": 182750
    },
    {
      "epoch": 2.6917865716263383,
      "grad_norm": 1.2932590246200562,
      "learning_rate": 1.0263586796441873e-05,
      "loss": 0.0583,
      "step": 182775
    },
    {
      "epoch": 2.6921547547164253,
      "grad_norm": 0.8844801187515259,
      "learning_rate": 1.0261950424474889e-05,
      "loss": 0.0517,
      "step": 182800
    },
    {
      "epoch": 2.6925229378065123,
      "grad_norm": 1.4140905141830444,
      "learning_rate": 1.0260314052507905e-05,
      "loss": 0.0613,
      "step": 182825
    },
    {
      "epoch": 2.6928911208965993,
      "grad_norm": 1.438362717628479,
      "learning_rate": 1.0258677680540921e-05,
      "loss": 0.0495,
      "step": 182850
    },
    {
      "epoch": 2.6932593039866863,
      "grad_norm": 1.283368706703186,
      "learning_rate": 1.0257041308573936e-05,
      "loss": 0.0623,
      "step": 182875
    },
    {
      "epoch": 2.693627487076774,
      "grad_norm": 1.3483699560165405,
      "learning_rate": 1.0255404936606952e-05,
      "loss": 0.0664,
      "step": 182900
    },
    {
      "epoch": 2.693995670166861,
      "grad_norm": 1.1190013885498047,
      "learning_rate": 1.0253768564639965e-05,
      "loss": 0.0613,
      "step": 182925
    },
    {
      "epoch": 2.694363853256948,
      "grad_norm": 1.2656058073043823,
      "learning_rate": 1.025213219267298e-05,
      "loss": 0.0557,
      "step": 182950
    },
    {
      "epoch": 2.694732036347035,
      "grad_norm": 1.3406060934066772,
      "learning_rate": 1.0250495820705997e-05,
      "loss": 0.0638,
      "step": 182975
    },
    {
      "epoch": 2.695100219437122,
      "grad_norm": 1.6723088026046753,
      "learning_rate": 1.0248859448739013e-05,
      "loss": 0.0623,
      "step": 183000
    },
    {
      "epoch": 2.695468402527209,
      "grad_norm": 1.4660754203796387,
      "learning_rate": 1.0247223076772028e-05,
      "loss": 0.0602,
      "step": 183025
    },
    {
      "epoch": 2.695836585617296,
      "grad_norm": 1.482211947441101,
      "learning_rate": 1.0245586704805044e-05,
      "loss": 0.0583,
      "step": 183050
    },
    {
      "epoch": 2.696204768707383,
      "grad_norm": 1.2991197109222412,
      "learning_rate": 1.024395033283806e-05,
      "loss": 0.0672,
      "step": 183075
    },
    {
      "epoch": 2.69657295179747,
      "grad_norm": 1.8289599418640137,
      "learning_rate": 1.0242313960871076e-05,
      "loss": 0.0647,
      "step": 183100
    },
    {
      "epoch": 2.696941134887557,
      "grad_norm": 1.2484925985336304,
      "learning_rate": 1.0240743043782768e-05,
      "loss": 0.0571,
      "step": 183125
    },
    {
      "epoch": 2.697309317977644,
      "grad_norm": 1.3957743644714355,
      "learning_rate": 1.0239106671815784e-05,
      "loss": 0.0583,
      "step": 183150
    },
    {
      "epoch": 2.697677501067731,
      "grad_norm": 0.8832376599311829,
      "learning_rate": 1.0237470299848799e-05,
      "loss": 0.0627,
      "step": 183175
    },
    {
      "epoch": 2.698045684157818,
      "grad_norm": 1.1454840898513794,
      "learning_rate": 1.0235833927881815e-05,
      "loss": 0.0547,
      "step": 183200
    },
    {
      "epoch": 2.698413867247905,
      "grad_norm": 1.428775668144226,
      "learning_rate": 1.0234197555914831e-05,
      "loss": 0.0609,
      "step": 183225
    },
    {
      "epoch": 2.698782050337992,
      "grad_norm": 1.228224754333496,
      "learning_rate": 1.0232561183947847e-05,
      "loss": 0.0587,
      "step": 183250
    },
    {
      "epoch": 2.699150233428079,
      "grad_norm": 1.730570673942566,
      "learning_rate": 1.0230924811980862e-05,
      "loss": 0.0708,
      "step": 183275
    },
    {
      "epoch": 2.699518416518166,
      "grad_norm": 1.202194094657898,
      "learning_rate": 1.0229288440013878e-05,
      "loss": 0.0561,
      "step": 183300
    },
    {
      "epoch": 2.699886599608253,
      "grad_norm": 1.2887914180755615,
      "learning_rate": 1.0227652068046894e-05,
      "loss": 0.0612,
      "step": 183325
    },
    {
      "epoch": 2.70025478269834,
      "grad_norm": 1.7454698085784912,
      "learning_rate": 1.0226015696079907e-05,
      "loss": 0.0716,
      "step": 183350
    },
    {
      "epoch": 2.700622965788427,
      "grad_norm": 1.3642715215682983,
      "learning_rate": 1.0224379324112923e-05,
      "loss": 0.0597,
      "step": 183375
    },
    {
      "epoch": 2.700991148878514,
      "grad_norm": 1.5880589485168457,
      "learning_rate": 1.0222742952145939e-05,
      "loss": 0.0587,
      "step": 183400
    },
    {
      "epoch": 2.701359331968601,
      "grad_norm": 1.3546175956726074,
      "learning_rate": 1.0221106580178955e-05,
      "loss": 0.0556,
      "step": 183425
    },
    {
      "epoch": 2.701727515058688,
      "grad_norm": 1.6443510055541992,
      "learning_rate": 1.021947020821197e-05,
      "loss": 0.0564,
      "step": 183450
    },
    {
      "epoch": 2.702095698148775,
      "grad_norm": 1.2800062894821167,
      "learning_rate": 1.0217833836244986e-05,
      "loss": 0.05,
      "step": 183475
    },
    {
      "epoch": 2.7024638812388626,
      "grad_norm": 0.8485159873962402,
      "learning_rate": 1.0216197464278002e-05,
      "loss": 0.0629,
      "step": 183500
    },
    {
      "epoch": 2.7028320643289496,
      "grad_norm": 1.1817010641098022,
      "learning_rate": 1.0214561092311016e-05,
      "loss": 0.057,
      "step": 183525
    },
    {
      "epoch": 2.7032002474190366,
      "grad_norm": 0.9812033772468567,
      "learning_rate": 1.0212924720344031e-05,
      "loss": 0.0588,
      "step": 183550
    },
    {
      "epoch": 2.7035684305091237,
      "grad_norm": 1.533164381980896,
      "learning_rate": 1.0211288348377047e-05,
      "loss": 0.0617,
      "step": 183575
    },
    {
      "epoch": 2.7039366135992107,
      "grad_norm": 1.741667628288269,
      "learning_rate": 1.0209651976410061e-05,
      "loss": 0.0591,
      "step": 183600
    },
    {
      "epoch": 2.7043047966892977,
      "grad_norm": 1.0580602884292603,
      "learning_rate": 1.0208015604443078e-05,
      "loss": 0.0608,
      "step": 183625
    },
    {
      "epoch": 2.7046729797793847,
      "grad_norm": 1.3728079795837402,
      "learning_rate": 1.0206379232476094e-05,
      "loss": 0.0539,
      "step": 183650
    },
    {
      "epoch": 2.7050411628694717,
      "grad_norm": 1.7485744953155518,
      "learning_rate": 1.020474286050911e-05,
      "loss": 0.058,
      "step": 183675
    },
    {
      "epoch": 2.7054093459595587,
      "grad_norm": 1.710504412651062,
      "learning_rate": 1.0203106488542124e-05,
      "loss": 0.0619,
      "step": 183700
    },
    {
      "epoch": 2.7057775290496457,
      "grad_norm": 1.5417767763137817,
      "learning_rate": 1.020147011657514e-05,
      "loss": 0.0622,
      "step": 183725
    },
    {
      "epoch": 2.7061457121397328,
      "grad_norm": 1.6605935096740723,
      "learning_rate": 1.0199833744608153e-05,
      "loss": 0.0631,
      "step": 183750
    },
    {
      "epoch": 2.7065138952298198,
      "grad_norm": 1.4925915002822876,
      "learning_rate": 1.019819737264117e-05,
      "loss": 0.0607,
      "step": 183775
    },
    {
      "epoch": 2.706882078319907,
      "grad_norm": 0.6819592118263245,
      "learning_rate": 1.0196561000674186e-05,
      "loss": 0.0568,
      "step": 183800
    },
    {
      "epoch": 2.707250261409994,
      "grad_norm": 1.4788029193878174,
      "learning_rate": 1.0194924628707202e-05,
      "loss": 0.064,
      "step": 183825
    },
    {
      "epoch": 2.7076184445000813,
      "grad_norm": 1.4906667470932007,
      "learning_rate": 1.0193288256740216e-05,
      "loss": 0.0507,
      "step": 183850
    },
    {
      "epoch": 2.7079866275901683,
      "grad_norm": 1.268228530883789,
      "learning_rate": 1.0191651884773232e-05,
      "loss": 0.0611,
      "step": 183875
    },
    {
      "epoch": 2.7083548106802553,
      "grad_norm": 1.1930062770843506,
      "learning_rate": 1.0190015512806249e-05,
      "loss": 0.0586,
      "step": 183900
    },
    {
      "epoch": 2.7087229937703423,
      "grad_norm": 1.4462981224060059,
      "learning_rate": 1.0188379140839265e-05,
      "loss": 0.0587,
      "step": 183925
    },
    {
      "epoch": 2.7090911768604293,
      "grad_norm": 1.7341102361679077,
      "learning_rate": 1.0186742768872279e-05,
      "loss": 0.0656,
      "step": 183950
    },
    {
      "epoch": 2.7094593599505163,
      "grad_norm": 1.2396494150161743,
      "learning_rate": 1.0185106396905294e-05,
      "loss": 0.0585,
      "step": 183975
    },
    {
      "epoch": 2.7098275430406034,
      "grad_norm": 1.207575798034668,
      "learning_rate": 1.018347002493831e-05,
      "loss": 0.0662,
      "step": 184000
    },
    {
      "epoch": 2.7101957261306904,
      "grad_norm": 1.2569626569747925,
      "learning_rate": 1.0181833652971324e-05,
      "loss": 0.0669,
      "step": 184025
    },
    {
      "epoch": 2.7105639092207774,
      "grad_norm": 1.3071489334106445,
      "learning_rate": 1.018019728100434e-05,
      "loss": 0.0563,
      "step": 184050
    },
    {
      "epoch": 2.7109320923108644,
      "grad_norm": 1.3904650211334229,
      "learning_rate": 1.0178560909037357e-05,
      "loss": 0.0715,
      "step": 184075
    },
    {
      "epoch": 2.7113002754009514,
      "grad_norm": 1.1116451025009155,
      "learning_rate": 1.0176924537070371e-05,
      "loss": 0.0616,
      "step": 184100
    },
    {
      "epoch": 2.7116684584910384,
      "grad_norm": 1.0635896921157837,
      "learning_rate": 1.0175288165103387e-05,
      "loss": 0.0577,
      "step": 184125
    },
    {
      "epoch": 2.7120366415811255,
      "grad_norm": 1.2569341659545898,
      "learning_rate": 1.0173651793136403e-05,
      "loss": 0.053,
      "step": 184150
    },
    {
      "epoch": 2.7124048246712125,
      "grad_norm": 1.6762291193008423,
      "learning_rate": 1.0172015421169416e-05,
      "loss": 0.0583,
      "step": 184175
    },
    {
      "epoch": 2.7127730077612995,
      "grad_norm": 1.5419416427612305,
      "learning_rate": 1.0170379049202432e-05,
      "loss": 0.0592,
      "step": 184200
    },
    {
      "epoch": 2.7131411908513865,
      "grad_norm": 1.745168924331665,
      "learning_rate": 1.0168742677235448e-05,
      "loss": 0.0552,
      "step": 184225
    },
    {
      "epoch": 2.7135093739414735,
      "grad_norm": 0.9518269300460815,
      "learning_rate": 1.0167106305268464e-05,
      "loss": 0.0631,
      "step": 184250
    },
    {
      "epoch": 2.7138775570315605,
      "grad_norm": 1.358918309211731,
      "learning_rate": 1.0165469933301479e-05,
      "loss": 0.0698,
      "step": 184275
    },
    {
      "epoch": 2.7142457401216475,
      "grad_norm": 1.5168896913528442,
      "learning_rate": 1.0163833561334495e-05,
      "loss": 0.0604,
      "step": 184300
    },
    {
      "epoch": 2.7146139232117346,
      "grad_norm": 1.1325515508651733,
      "learning_rate": 1.0162197189367511e-05,
      "loss": 0.0613,
      "step": 184325
    },
    {
      "epoch": 2.7149821063018216,
      "grad_norm": 1.3978608846664429,
      "learning_rate": 1.0160560817400527e-05,
      "loss": 0.0605,
      "step": 184350
    },
    {
      "epoch": 2.7153502893919086,
      "grad_norm": 2.056851863861084,
      "learning_rate": 1.0158924445433542e-05,
      "loss": 0.0594,
      "step": 184375
    },
    {
      "epoch": 2.7157184724819956,
      "grad_norm": 0.8224292993545532,
      "learning_rate": 1.0157288073466556e-05,
      "loss": 0.0596,
      "step": 184400
    },
    {
      "epoch": 2.7160866555720826,
      "grad_norm": 1.7935566902160645,
      "learning_rate": 1.015565170149957e-05,
      "loss": 0.0646,
      "step": 184425
    },
    {
      "epoch": 2.71645483866217,
      "grad_norm": 1.458459734916687,
      "learning_rate": 1.0154015329532587e-05,
      "loss": 0.0604,
      "step": 184450
    },
    {
      "epoch": 2.716823021752257,
      "grad_norm": 1.405131220817566,
      "learning_rate": 1.0152378957565603e-05,
      "loss": 0.0624,
      "step": 184475
    },
    {
      "epoch": 2.717191204842344,
      "grad_norm": 1.4670943021774292,
      "learning_rate": 1.015074258559862e-05,
      "loss": 0.062,
      "step": 184500
    },
    {
      "epoch": 2.717559387932431,
      "grad_norm": 1.455169439315796,
      "learning_rate": 1.0149106213631634e-05,
      "loss": 0.0629,
      "step": 184525
    },
    {
      "epoch": 2.717927571022518,
      "grad_norm": 1.6445952653884888,
      "learning_rate": 1.014746984166465e-05,
      "loss": 0.0568,
      "step": 184550
    },
    {
      "epoch": 2.718295754112605,
      "grad_norm": 1.038856863975525,
      "learning_rate": 1.0145833469697666e-05,
      "loss": 0.0519,
      "step": 184575
    },
    {
      "epoch": 2.718663937202692,
      "grad_norm": 1.7846399545669556,
      "learning_rate": 1.0144197097730679e-05,
      "loss": 0.0638,
      "step": 184600
    },
    {
      "epoch": 2.719032120292779,
      "grad_norm": 1.5178264379501343,
      "learning_rate": 1.0142560725763695e-05,
      "loss": 0.0596,
      "step": 184625
    },
    {
      "epoch": 2.719400303382866,
      "grad_norm": 1.2321747541427612,
      "learning_rate": 1.0140924353796711e-05,
      "loss": 0.0601,
      "step": 184650
    },
    {
      "epoch": 2.719768486472953,
      "grad_norm": 1.2565826177597046,
      "learning_rate": 1.0139287981829726e-05,
      "loss": 0.056,
      "step": 184675
    },
    {
      "epoch": 2.7201366695630402,
      "grad_norm": 1.3941514492034912,
      "learning_rate": 1.0137651609862742e-05,
      "loss": 0.0562,
      "step": 184700
    },
    {
      "epoch": 2.7205048526531272,
      "grad_norm": 1.3635790348052979,
      "learning_rate": 1.0136015237895758e-05,
      "loss": 0.062,
      "step": 184725
    },
    {
      "epoch": 2.7208730357432143,
      "grad_norm": 1.7387267351150513,
      "learning_rate": 1.0134378865928774e-05,
      "loss": 0.0541,
      "step": 184750
    },
    {
      "epoch": 2.7212412188333013,
      "grad_norm": 1.214390516281128,
      "learning_rate": 1.0132742493961788e-05,
      "loss": 0.0603,
      "step": 184775
    },
    {
      "epoch": 2.7216094019233887,
      "grad_norm": 1.81710946559906,
      "learning_rate": 1.0131106121994805e-05,
      "loss": 0.0648,
      "step": 184800
    },
    {
      "epoch": 2.7219775850134758,
      "grad_norm": 1.5733720064163208,
      "learning_rate": 1.0129469750027819e-05,
      "loss": 0.0607,
      "step": 184825
    },
    {
      "epoch": 2.7223457681035628,
      "grad_norm": 1.319262146949768,
      "learning_rate": 1.0127833378060833e-05,
      "loss": 0.062,
      "step": 184850
    },
    {
      "epoch": 2.72271395119365,
      "grad_norm": 1.2286608219146729,
      "learning_rate": 1.012619700609385e-05,
      "loss": 0.0655,
      "step": 184875
    },
    {
      "epoch": 2.723082134283737,
      "grad_norm": 1.5763959884643555,
      "learning_rate": 1.0124560634126866e-05,
      "loss": 0.0624,
      "step": 184900
    },
    {
      "epoch": 2.723450317373824,
      "grad_norm": 0.7446858882904053,
      "learning_rate": 1.0122924262159882e-05,
      "loss": 0.057,
      "step": 184925
    },
    {
      "epoch": 2.723818500463911,
      "grad_norm": 1.4298956394195557,
      "learning_rate": 1.0121287890192896e-05,
      "loss": 0.0563,
      "step": 184950
    },
    {
      "epoch": 2.724186683553998,
      "grad_norm": 1.0701384544372559,
      "learning_rate": 1.0119651518225913e-05,
      "loss": 0.0582,
      "step": 184975
    },
    {
      "epoch": 2.724554866644085,
      "grad_norm": 1.5140502452850342,
      "learning_rate": 1.0118015146258929e-05,
      "loss": 0.056,
      "step": 185000
    },
    {
      "epoch": 2.724923049734172,
      "grad_norm": 1.266882300376892,
      "learning_rate": 1.0116378774291941e-05,
      "loss": 0.0578,
      "step": 185025
    },
    {
      "epoch": 2.725291232824259,
      "grad_norm": 1.3724349737167358,
      "learning_rate": 1.0114742402324958e-05,
      "loss": 0.0575,
      "step": 185050
    },
    {
      "epoch": 2.725659415914346,
      "grad_norm": 1.0935099124908447,
      "learning_rate": 1.0113106030357974e-05,
      "loss": 0.0615,
      "step": 185075
    },
    {
      "epoch": 2.726027599004433,
      "grad_norm": 1.2079156637191772,
      "learning_rate": 1.0111469658390988e-05,
      "loss": 0.0582,
      "step": 185100
    },
    {
      "epoch": 2.72639578209452,
      "grad_norm": 1.390535831451416,
      "learning_rate": 1.0109833286424004e-05,
      "loss": 0.0651,
      "step": 185125
    },
    {
      "epoch": 2.726763965184607,
      "grad_norm": 1.5886473655700684,
      "learning_rate": 1.010819691445702e-05,
      "loss": 0.0611,
      "step": 185150
    },
    {
      "epoch": 2.727132148274694,
      "grad_norm": 1.565333366394043,
      "learning_rate": 1.0106560542490037e-05,
      "loss": 0.055,
      "step": 185175
    },
    {
      "epoch": 2.727500331364781,
      "grad_norm": 0.9570712447166443,
      "learning_rate": 1.0104924170523051e-05,
      "loss": 0.0584,
      "step": 185200
    },
    {
      "epoch": 2.727868514454868,
      "grad_norm": 1.2584855556488037,
      "learning_rate": 1.0103287798556066e-05,
      "loss": 0.0547,
      "step": 185225
    },
    {
      "epoch": 2.728236697544955,
      "grad_norm": 2.0586068630218506,
      "learning_rate": 1.010165142658908e-05,
      "loss": 0.0679,
      "step": 185250
    },
    {
      "epoch": 2.728604880635042,
      "grad_norm": 1.3739484548568726,
      "learning_rate": 1.0100015054622096e-05,
      "loss": 0.0623,
      "step": 185275
    },
    {
      "epoch": 2.728973063725129,
      "grad_norm": 1.7004926204681396,
      "learning_rate": 1.0098378682655112e-05,
      "loss": 0.0607,
      "step": 185300
    },
    {
      "epoch": 2.729341246815216,
      "grad_norm": 1.7843974828720093,
      "learning_rate": 1.0096742310688128e-05,
      "loss": 0.0642,
      "step": 185325
    },
    {
      "epoch": 2.729709429905303,
      "grad_norm": 1.2455390691757202,
      "learning_rate": 1.0095105938721143e-05,
      "loss": 0.0569,
      "step": 185350
    },
    {
      "epoch": 2.73007761299539,
      "grad_norm": 1.5210189819335938,
      "learning_rate": 1.0093469566754159e-05,
      "loss": 0.0582,
      "step": 185375
    },
    {
      "epoch": 2.7304457960854776,
      "grad_norm": 1.5243985652923584,
      "learning_rate": 1.0091833194787175e-05,
      "loss": 0.058,
      "step": 185400
    },
    {
      "epoch": 2.7308139791755646,
      "grad_norm": 1.044806718826294,
      "learning_rate": 1.0090196822820191e-05,
      "loss": 0.0558,
      "step": 185425
    },
    {
      "epoch": 2.7311821622656516,
      "grad_norm": 1.198866367340088,
      "learning_rate": 1.0088560450853204e-05,
      "loss": 0.0577,
      "step": 185450
    },
    {
      "epoch": 2.7315503453557386,
      "grad_norm": 1.286551833152771,
      "learning_rate": 1.008692407888622e-05,
      "loss": 0.0575,
      "step": 185475
    },
    {
      "epoch": 2.7319185284458256,
      "grad_norm": 1.1551345586776733,
      "learning_rate": 1.0085287706919236e-05,
      "loss": 0.0534,
      "step": 185500
    },
    {
      "epoch": 2.7322867115359126,
      "grad_norm": 1.4485493898391724,
      "learning_rate": 1.0083651334952251e-05,
      "loss": 0.0637,
      "step": 185525
    },
    {
      "epoch": 2.7326548946259996,
      "grad_norm": 1.0969319343566895,
      "learning_rate": 1.0082014962985267e-05,
      "loss": 0.0602,
      "step": 185550
    },
    {
      "epoch": 2.7330230777160867,
      "grad_norm": 0.8755432963371277,
      "learning_rate": 1.0080378591018283e-05,
      "loss": 0.0587,
      "step": 185575
    },
    {
      "epoch": 2.7333912608061737,
      "grad_norm": 0.9794855713844299,
      "learning_rate": 1.0078742219051298e-05,
      "loss": 0.0532,
      "step": 185600
    },
    {
      "epoch": 2.7337594438962607,
      "grad_norm": 1.1577626466751099,
      "learning_rate": 1.0077105847084314e-05,
      "loss": 0.0601,
      "step": 185625
    },
    {
      "epoch": 2.7341276269863477,
      "grad_norm": 0.9449526071548462,
      "learning_rate": 1.0075469475117328e-05,
      "loss": 0.0572,
      "step": 185650
    },
    {
      "epoch": 2.7344958100764347,
      "grad_norm": 1.190841794013977,
      "learning_rate": 1.0073833103150343e-05,
      "loss": 0.059,
      "step": 185675
    },
    {
      "epoch": 2.7348639931665217,
      "grad_norm": 1.2565597295761108,
      "learning_rate": 1.0072196731183359e-05,
      "loss": 0.0605,
      "step": 185700
    },
    {
      "epoch": 2.7352321762566087,
      "grad_norm": 1.2178332805633545,
      "learning_rate": 1.0070560359216375e-05,
      "loss": 0.0591,
      "step": 185725
    },
    {
      "epoch": 2.7356003593466958,
      "grad_norm": 0.8831363916397095,
      "learning_rate": 1.0068923987249391e-05,
      "loss": 0.0542,
      "step": 185750
    },
    {
      "epoch": 2.7359685424367832,
      "grad_norm": 2.107327938079834,
      "learning_rate": 1.0067287615282406e-05,
      "loss": 0.0552,
      "step": 185775
    },
    {
      "epoch": 2.7363367255268702,
      "grad_norm": 1.45307195186615,
      "learning_rate": 1.0065651243315422e-05,
      "loss": 0.062,
      "step": 185800
    },
    {
      "epoch": 2.7367049086169573,
      "grad_norm": 1.8094033002853394,
      "learning_rate": 1.0064014871348438e-05,
      "loss": 0.0566,
      "step": 185825
    },
    {
      "epoch": 2.7370730917070443,
      "grad_norm": 1.331187129020691,
      "learning_rate": 1.0062378499381454e-05,
      "loss": 0.064,
      "step": 185850
    },
    {
      "epoch": 2.7374412747971313,
      "grad_norm": 1.1713544130325317,
      "learning_rate": 1.0060742127414467e-05,
      "loss": 0.0624,
      "step": 185875
    },
    {
      "epoch": 2.7378094578872183,
      "grad_norm": 1.4612226486206055,
      "learning_rate": 1.0059105755447483e-05,
      "loss": 0.0611,
      "step": 185900
    },
    {
      "epoch": 2.7381776409773053,
      "grad_norm": 1.2365119457244873,
      "learning_rate": 1.0057469383480497e-05,
      "loss": 0.0594,
      "step": 185925
    },
    {
      "epoch": 2.7385458240673923,
      "grad_norm": 1.6546189785003662,
      "learning_rate": 1.0055833011513514e-05,
      "loss": 0.0648,
      "step": 185950
    },
    {
      "epoch": 2.7389140071574793,
      "grad_norm": 1.514780879020691,
      "learning_rate": 1.005419663954653e-05,
      "loss": 0.0567,
      "step": 185975
    },
    {
      "epoch": 2.7392821902475664,
      "grad_norm": 0.8987876772880554,
      "learning_rate": 1.0052560267579546e-05,
      "loss": 0.063,
      "step": 186000
    },
    {
      "epoch": 2.7396503733376534,
      "grad_norm": 1.7933945655822754,
      "learning_rate": 1.005092389561256e-05,
      "loss": 0.0563,
      "step": 186025
    },
    {
      "epoch": 2.7400185564277404,
      "grad_norm": 1.4574695825576782,
      "learning_rate": 1.0049287523645577e-05,
      "loss": 0.0541,
      "step": 186050
    },
    {
      "epoch": 2.7403867395178274,
      "grad_norm": 1.4761006832122803,
      "learning_rate": 1.0047651151678591e-05,
      "loss": 0.0591,
      "step": 186075
    },
    {
      "epoch": 2.7407549226079144,
      "grad_norm": 1.0592039823532104,
      "learning_rate": 1.0046014779711605e-05,
      "loss": 0.0572,
      "step": 186100
    },
    {
      "epoch": 2.7411231056980014,
      "grad_norm": 1.2496352195739746,
      "learning_rate": 1.0044378407744622e-05,
      "loss": 0.0622,
      "step": 186125
    },
    {
      "epoch": 2.7414912887880885,
      "grad_norm": 1.2604317665100098,
      "learning_rate": 1.0042742035777638e-05,
      "loss": 0.059,
      "step": 186150
    },
    {
      "epoch": 2.7418594718781755,
      "grad_norm": 1.5644952058792114,
      "learning_rate": 1.0041105663810652e-05,
      "loss": 0.057,
      "step": 186175
    },
    {
      "epoch": 2.7422276549682625,
      "grad_norm": 1.5964096784591675,
      "learning_rate": 1.0039469291843668e-05,
      "loss": 0.0601,
      "step": 186200
    },
    {
      "epoch": 2.7425958380583495,
      "grad_norm": 1.1815171241760254,
      "learning_rate": 1.0037832919876685e-05,
      "loss": 0.057,
      "step": 186225
    },
    {
      "epoch": 2.7429640211484365,
      "grad_norm": 1.429641842842102,
      "learning_rate": 1.00361965479097e-05,
      "loss": 0.0606,
      "step": 186250
    },
    {
      "epoch": 2.7433322042385235,
      "grad_norm": 1.0084843635559082,
      "learning_rate": 1.0034560175942715e-05,
      "loss": 0.0547,
      "step": 186275
    },
    {
      "epoch": 2.7437003873286105,
      "grad_norm": 1.1361294984817505,
      "learning_rate": 1.003292380397573e-05,
      "loss": 0.0585,
      "step": 186300
    },
    {
      "epoch": 2.7440685704186976,
      "grad_norm": 1.2982580661773682,
      "learning_rate": 1.0031287432008746e-05,
      "loss": 0.0613,
      "step": 186325
    },
    {
      "epoch": 2.7444367535087846,
      "grad_norm": 1.3378002643585205,
      "learning_rate": 1.002965106004176e-05,
      "loss": 0.0548,
      "step": 186350
    },
    {
      "epoch": 2.744804936598872,
      "grad_norm": 1.4267215728759766,
      "learning_rate": 1.0028014688074776e-05,
      "loss": 0.0565,
      "step": 186375
    },
    {
      "epoch": 2.745173119688959,
      "grad_norm": 1.2975184917449951,
      "learning_rate": 1.0026378316107793e-05,
      "loss": 0.0607,
      "step": 186400
    },
    {
      "epoch": 2.745541302779046,
      "grad_norm": 1.2684420347213745,
      "learning_rate": 1.0024741944140807e-05,
      "loss": 0.0543,
      "step": 186425
    },
    {
      "epoch": 2.745909485869133,
      "grad_norm": 1.9877132177352905,
      "learning_rate": 1.0023105572173823e-05,
      "loss": 0.0593,
      "step": 186450
    },
    {
      "epoch": 2.74627766895922,
      "grad_norm": 1.5988712310791016,
      "learning_rate": 1.002146920020684e-05,
      "loss": 0.0593,
      "step": 186475
    },
    {
      "epoch": 2.746645852049307,
      "grad_norm": 1.7441291809082031,
      "learning_rate": 1.0019832828239852e-05,
      "loss": 0.0604,
      "step": 186500
    },
    {
      "epoch": 2.747014035139394,
      "grad_norm": 1.3226075172424316,
      "learning_rate": 1.0018196456272868e-05,
      "loss": 0.0599,
      "step": 186525
    },
    {
      "epoch": 2.747382218229481,
      "grad_norm": 1.805211067199707,
      "learning_rate": 1.0016560084305884e-05,
      "loss": 0.0656,
      "step": 186550
    },
    {
      "epoch": 2.747750401319568,
      "grad_norm": 1.4552751779556274,
      "learning_rate": 1.00149237123389e-05,
      "loss": 0.0653,
      "step": 186575
    },
    {
      "epoch": 2.748118584409655,
      "grad_norm": 1.0965235233306885,
      "learning_rate": 1.0013352795250594e-05,
      "loss": 0.0565,
      "step": 186600
    },
    {
      "epoch": 2.748486767499742,
      "grad_norm": 1.0309356451034546,
      "learning_rate": 1.001171642328361e-05,
      "loss": 0.0553,
      "step": 186625
    },
    {
      "epoch": 2.748854950589829,
      "grad_norm": 1.480932593345642,
      "learning_rate": 1.0010080051316627e-05,
      "loss": 0.0655,
      "step": 186650
    },
    {
      "epoch": 2.749223133679916,
      "grad_norm": 2.2531635761260986,
      "learning_rate": 1.0008443679349643e-05,
      "loss": 0.0674,
      "step": 186675
    },
    {
      "epoch": 2.7495913167700032,
      "grad_norm": 0.9868398308753967,
      "learning_rate": 1.0006807307382656e-05,
      "loss": 0.0566,
      "step": 186700
    },
    {
      "epoch": 2.7499594998600907,
      "grad_norm": 1.5144098997116089,
      "learning_rate": 1.0005170935415672e-05,
      "loss": 0.0554,
      "step": 186725
    },
    {
      "epoch": 2.7503276829501777,
      "grad_norm": 0.9801357388496399,
      "learning_rate": 1.0003534563448686e-05,
      "loss": 0.061,
      "step": 186750
    },
    {
      "epoch": 2.7506958660402647,
      "grad_norm": 1.4268430471420288,
      "learning_rate": 1.0001898191481702e-05,
      "loss": 0.0629,
      "step": 186775
    },
    {
      "epoch": 2.7510640491303517,
      "grad_norm": 1.3382500410079956,
      "learning_rate": 1.0000261819514719e-05,
      "loss": 0.0605,
      "step": 186800
    },
    {
      "epoch": 2.7514322322204388,
      "grad_norm": 1.1464941501617432,
      "learning_rate": 9.998625447547735e-06,
      "loss": 0.0562,
      "step": 186825
    },
    {
      "epoch": 2.7518004153105258,
      "grad_norm": 1.2671126127243042,
      "learning_rate": 9.996989075580749e-06,
      "loss": 0.0576,
      "step": 186850
    },
    {
      "epoch": 2.752168598400613,
      "grad_norm": 1.2742894887924194,
      "learning_rate": 9.995352703613764e-06,
      "loss": 0.0568,
      "step": 186875
    },
    {
      "epoch": 2.7525367814907,
      "grad_norm": 0.9946377873420715,
      "learning_rate": 9.99371633164678e-06,
      "loss": 0.0608,
      "step": 186900
    },
    {
      "epoch": 2.752904964580787,
      "grad_norm": 1.217646837234497,
      "learning_rate": 9.992079959679796e-06,
      "loss": 0.0678,
      "step": 186925
    },
    {
      "epoch": 2.753273147670874,
      "grad_norm": 1.749219536781311,
      "learning_rate": 9.990443587712812e-06,
      "loss": 0.064,
      "step": 186950
    },
    {
      "epoch": 2.753641330760961,
      "grad_norm": 1.1158767938613892,
      "learning_rate": 9.988807215745826e-06,
      "loss": 0.0521,
      "step": 186975
    },
    {
      "epoch": 2.754009513851048,
      "grad_norm": 1.01271390914917,
      "learning_rate": 9.987170843778841e-06,
      "loss": 0.0547,
      "step": 187000
    },
    {
      "epoch": 2.754377696941135,
      "grad_norm": 1.6766948699951172,
      "learning_rate": 9.985534471811857e-06,
      "loss": 0.0592,
      "step": 187025
    },
    {
      "epoch": 2.754745880031222,
      "grad_norm": 1.7003371715545654,
      "learning_rate": 9.983898099844873e-06,
      "loss": 0.0603,
      "step": 187050
    },
    {
      "epoch": 2.755114063121309,
      "grad_norm": 1.6806586980819702,
      "learning_rate": 9.982261727877888e-06,
      "loss": 0.056,
      "step": 187075
    },
    {
      "epoch": 2.755482246211396,
      "grad_norm": 1.0180401802062988,
      "learning_rate": 9.980625355910904e-06,
      "loss": 0.0636,
      "step": 187100
    },
    {
      "epoch": 2.755850429301483,
      "grad_norm": 1.5644352436065674,
      "learning_rate": 9.978988983943918e-06,
      "loss": 0.0576,
      "step": 187125
    },
    {
      "epoch": 2.75621861239157,
      "grad_norm": 1.1711918115615845,
      "learning_rate": 9.977352611976934e-06,
      "loss": 0.0517,
      "step": 187150
    },
    {
      "epoch": 2.756586795481657,
      "grad_norm": 1.4307109117507935,
      "learning_rate": 9.975716240009949e-06,
      "loss": 0.0627,
      "step": 187175
    },
    {
      "epoch": 2.756954978571744,
      "grad_norm": 1.8698443174362183,
      "learning_rate": 9.974079868042965e-06,
      "loss": 0.0569,
      "step": 187200
    },
    {
      "epoch": 2.757323161661831,
      "grad_norm": 1.1340711116790771,
      "learning_rate": 9.972443496075981e-06,
      "loss": 0.0543,
      "step": 187225
    },
    {
      "epoch": 2.757691344751918,
      "grad_norm": 1.3797444105148315,
      "learning_rate": 9.970807124108997e-06,
      "loss": 0.0633,
      "step": 187250
    },
    {
      "epoch": 2.758059527842005,
      "grad_norm": 1.2021266222000122,
      "learning_rate": 9.969170752142012e-06,
      "loss": 0.0642,
      "step": 187275
    },
    {
      "epoch": 2.758427710932092,
      "grad_norm": 1.1204200983047485,
      "learning_rate": 9.967534380175026e-06,
      "loss": 0.0626,
      "step": 187300
    },
    {
      "epoch": 2.7587958940221795,
      "grad_norm": 1.6233242750167847,
      "learning_rate": 9.965898008208042e-06,
      "loss": 0.0634,
      "step": 187325
    },
    {
      "epoch": 2.7591640771122665,
      "grad_norm": 1.3069599866867065,
      "learning_rate": 9.964261636241059e-06,
      "loss": 0.0581,
      "step": 187350
    },
    {
      "epoch": 2.7595322602023535,
      "grad_norm": 1.0715394020080566,
      "learning_rate": 9.962625264274075e-06,
      "loss": 0.0556,
      "step": 187375
    },
    {
      "epoch": 2.7599004432924406,
      "grad_norm": 1.428015112876892,
      "learning_rate": 9.96098889230709e-06,
      "loss": 0.0596,
      "step": 187400
    },
    {
      "epoch": 2.7602686263825276,
      "grad_norm": 1.0343254804611206,
      "learning_rate": 9.959352520340104e-06,
      "loss": 0.0498,
      "step": 187425
    },
    {
      "epoch": 2.7606368094726146,
      "grad_norm": 1.025355339050293,
      "learning_rate": 9.95771614837312e-06,
      "loss": 0.0572,
      "step": 187450
    },
    {
      "epoch": 2.7610049925627016,
      "grad_norm": 1.2642844915390015,
      "learning_rate": 9.956079776406136e-06,
      "loss": 0.0647,
      "step": 187475
    },
    {
      "epoch": 2.7613731756527886,
      "grad_norm": 1.3289872407913208,
      "learning_rate": 9.95444340443915e-06,
      "loss": 0.0674,
      "step": 187500
    },
    {
      "epoch": 2.7617413587428756,
      "grad_norm": 1.2489908933639526,
      "learning_rate": 9.952807032472167e-06,
      "loss": 0.0621,
      "step": 187525
    },
    {
      "epoch": 2.7621095418329626,
      "grad_norm": 1.3466240167617798,
      "learning_rate": 9.951170660505181e-06,
      "loss": 0.0587,
      "step": 187550
    },
    {
      "epoch": 2.7624777249230497,
      "grad_norm": 1.1342620849609375,
      "learning_rate": 9.949534288538197e-06,
      "loss": 0.0596,
      "step": 187575
    },
    {
      "epoch": 2.7628459080131367,
      "grad_norm": 1.3637194633483887,
      "learning_rate": 9.947897916571212e-06,
      "loss": 0.0626,
      "step": 187600
    },
    {
      "epoch": 2.7632140911032237,
      "grad_norm": 1.1403675079345703,
      "learning_rate": 9.946261544604228e-06,
      "loss": 0.0602,
      "step": 187625
    },
    {
      "epoch": 2.7635822741933107,
      "grad_norm": 2.7129833698272705,
      "learning_rate": 9.944625172637244e-06,
      "loss": 0.0615,
      "step": 187650
    },
    {
      "epoch": 2.763950457283398,
      "grad_norm": 1.598002314567566,
      "learning_rate": 9.942988800670258e-06,
      "loss": 0.0648,
      "step": 187675
    },
    {
      "epoch": 2.764318640373485,
      "grad_norm": 1.5435110330581665,
      "learning_rate": 9.941352428703273e-06,
      "loss": 0.0631,
      "step": 187700
    },
    {
      "epoch": 2.764686823463572,
      "grad_norm": 1.4801486730575562,
      "learning_rate": 9.939716056736289e-06,
      "loss": 0.0577,
      "step": 187725
    },
    {
      "epoch": 2.765055006553659,
      "grad_norm": 1.0425965785980225,
      "learning_rate": 9.938079684769305e-06,
      "loss": 0.063,
      "step": 187750
    },
    {
      "epoch": 2.7654231896437462,
      "grad_norm": 0.7088472843170166,
      "learning_rate": 9.936443312802321e-06,
      "loss": 0.0616,
      "step": 187775
    },
    {
      "epoch": 2.7657913727338332,
      "grad_norm": 1.5590195655822754,
      "learning_rate": 9.934806940835336e-06,
      "loss": 0.0561,
      "step": 187800
    },
    {
      "epoch": 2.7661595558239203,
      "grad_norm": 1.3413126468658447,
      "learning_rate": 9.933170568868352e-06,
      "loss": 0.0524,
      "step": 187825
    },
    {
      "epoch": 2.7665277389140073,
      "grad_norm": 1.69755220413208,
      "learning_rate": 9.931534196901366e-06,
      "loss": 0.0671,
      "step": 187850
    },
    {
      "epoch": 2.7668959220040943,
      "grad_norm": 1.50053870677948,
      "learning_rate": 9.929897824934383e-06,
      "loss": 0.0546,
      "step": 187875
    },
    {
      "epoch": 2.7672641050941813,
      "grad_norm": 1.535345435142517,
      "learning_rate": 9.928261452967399e-06,
      "loss": 0.0575,
      "step": 187900
    },
    {
      "epoch": 2.7676322881842683,
      "grad_norm": 1.266869068145752,
      "learning_rate": 9.926625081000413e-06,
      "loss": 0.059,
      "step": 187925
    },
    {
      "epoch": 2.7680004712743553,
      "grad_norm": 1.2413359880447388,
      "learning_rate": 9.92498870903343e-06,
      "loss": 0.0599,
      "step": 187950
    },
    {
      "epoch": 2.7683686543644423,
      "grad_norm": 0.9187526702880859,
      "learning_rate": 9.923352337066444e-06,
      "loss": 0.0549,
      "step": 187975
    },
    {
      "epoch": 2.7687368374545294,
      "grad_norm": 1.7610441446304321,
      "learning_rate": 9.92171596509946e-06,
      "loss": 0.0609,
      "step": 188000
    },
    {
      "epoch": 2.7691050205446164,
      "grad_norm": 1.9958604574203491,
      "learning_rate": 9.920079593132474e-06,
      "loss": 0.0594,
      "step": 188025
    },
    {
      "epoch": 2.7694732036347034,
      "grad_norm": 1.1109797954559326,
      "learning_rate": 9.91844322116549e-06,
      "loss": 0.0619,
      "step": 188050
    },
    {
      "epoch": 2.7698413867247904,
      "grad_norm": 1.0043673515319824,
      "learning_rate": 9.916806849198507e-06,
      "loss": 0.0604,
      "step": 188075
    },
    {
      "epoch": 2.7702095698148774,
      "grad_norm": 1.6157029867172241,
      "learning_rate": 9.915170477231521e-06,
      "loss": 0.0533,
      "step": 188100
    },
    {
      "epoch": 2.7705777529049644,
      "grad_norm": 1.8857016563415527,
      "learning_rate": 9.913534105264536e-06,
      "loss": 0.0631,
      "step": 188125
    },
    {
      "epoch": 2.7709459359950515,
      "grad_norm": 1.3368146419525146,
      "learning_rate": 9.911897733297552e-06,
      "loss": 0.0583,
      "step": 188150
    },
    {
      "epoch": 2.7713141190851385,
      "grad_norm": 1.1812440156936646,
      "learning_rate": 9.910261361330568e-06,
      "loss": 0.0519,
      "step": 188175
    },
    {
      "epoch": 2.7716823021752255,
      "grad_norm": 1.7864316701889038,
      "learning_rate": 9.908624989363584e-06,
      "loss": 0.0555,
      "step": 188200
    },
    {
      "epoch": 2.7720504852653125,
      "grad_norm": 1.478463888168335,
      "learning_rate": 9.906988617396598e-06,
      "loss": 0.0604,
      "step": 188225
    },
    {
      "epoch": 2.7724186683553995,
      "grad_norm": 1.3287866115570068,
      "learning_rate": 9.905352245429613e-06,
      "loss": 0.0594,
      "step": 188250
    },
    {
      "epoch": 2.772786851445487,
      "grad_norm": 1.3550938367843628,
      "learning_rate": 9.903715873462629e-06,
      "loss": 0.0692,
      "step": 188275
    },
    {
      "epoch": 2.773155034535574,
      "grad_norm": 1.0496293306350708,
      "learning_rate": 9.902079501495645e-06,
      "loss": 0.0567,
      "step": 188300
    },
    {
      "epoch": 2.773523217625661,
      "grad_norm": 1.3580968379974365,
      "learning_rate": 9.900443129528661e-06,
      "loss": 0.0606,
      "step": 188325
    },
    {
      "epoch": 2.773891400715748,
      "grad_norm": 1.5281096696853638,
      "learning_rate": 9.898806757561676e-06,
      "loss": 0.0646,
      "step": 188350
    },
    {
      "epoch": 2.774259583805835,
      "grad_norm": 1.658506155014038,
      "learning_rate": 9.89717038559469e-06,
      "loss": 0.0572,
      "step": 188375
    },
    {
      "epoch": 2.774627766895922,
      "grad_norm": 1.673966884613037,
      "learning_rate": 9.895534013627706e-06,
      "loss": 0.0615,
      "step": 188400
    },
    {
      "epoch": 2.774995949986009,
      "grad_norm": 1.0819004774093628,
      "learning_rate": 9.893897641660723e-06,
      "loss": 0.0494,
      "step": 188425
    },
    {
      "epoch": 2.775364133076096,
      "grad_norm": 1.4841569662094116,
      "learning_rate": 9.892261269693737e-06,
      "loss": 0.0585,
      "step": 188450
    },
    {
      "epoch": 2.775732316166183,
      "grad_norm": 1.232907772064209,
      "learning_rate": 9.890624897726753e-06,
      "loss": 0.0551,
      "step": 188475
    },
    {
      "epoch": 2.77610049925627,
      "grad_norm": 1.1234492063522339,
      "learning_rate": 9.888988525759768e-06,
      "loss": 0.053,
      "step": 188500
    },
    {
      "epoch": 2.776468682346357,
      "grad_norm": 1.416610836982727,
      "learning_rate": 9.887352153792784e-06,
      "loss": 0.0585,
      "step": 188525
    },
    {
      "epoch": 2.776836865436444,
      "grad_norm": 1.4943337440490723,
      "learning_rate": 9.885715781825798e-06,
      "loss": 0.0635,
      "step": 188550
    },
    {
      "epoch": 2.777205048526531,
      "grad_norm": 1.4836252927780151,
      "learning_rate": 9.884079409858814e-06,
      "loss": 0.0642,
      "step": 188575
    },
    {
      "epoch": 2.777573231616618,
      "grad_norm": 2.0967702865600586,
      "learning_rate": 9.88244303789183e-06,
      "loss": 0.0603,
      "step": 188600
    },
    {
      "epoch": 2.777941414706705,
      "grad_norm": 0.9067227244377136,
      "learning_rate": 9.880806665924845e-06,
      "loss": 0.0653,
      "step": 188625
    },
    {
      "epoch": 2.7783095977967927,
      "grad_norm": 1.601743459701538,
      "learning_rate": 9.879170293957861e-06,
      "loss": 0.0596,
      "step": 188650
    },
    {
      "epoch": 2.7786777808868797,
      "grad_norm": 1.953127145767212,
      "learning_rate": 9.877533921990876e-06,
      "loss": 0.067,
      "step": 188675
    },
    {
      "epoch": 2.7790459639769667,
      "grad_norm": 0.987812340259552,
      "learning_rate": 9.875897550023892e-06,
      "loss": 0.0592,
      "step": 188700
    },
    {
      "epoch": 2.7794141470670537,
      "grad_norm": 1.4707144498825073,
      "learning_rate": 9.874261178056908e-06,
      "loss": 0.0603,
      "step": 188725
    },
    {
      "epoch": 2.7797823301571407,
      "grad_norm": 1.6171125173568726,
      "learning_rate": 9.872624806089922e-06,
      "loss": 0.0611,
      "step": 188750
    },
    {
      "epoch": 2.7801505132472277,
      "grad_norm": 1.6431132555007935,
      "learning_rate": 9.870988434122939e-06,
      "loss": 0.0632,
      "step": 188775
    },
    {
      "epoch": 2.7805186963373147,
      "grad_norm": 1.7907843589782715,
      "learning_rate": 9.869352062155953e-06,
      "loss": 0.0607,
      "step": 188800
    },
    {
      "epoch": 2.7808868794274018,
      "grad_norm": 1.4770323038101196,
      "learning_rate": 9.86771569018897e-06,
      "loss": 0.0611,
      "step": 188825
    },
    {
      "epoch": 2.7812550625174888,
      "grad_norm": 1.2739169597625732,
      "learning_rate": 9.866079318221985e-06,
      "loss": 0.0581,
      "step": 188850
    },
    {
      "epoch": 2.781623245607576,
      "grad_norm": 1.1502888202667236,
      "learning_rate": 9.864442946255e-06,
      "loss": 0.0657,
      "step": 188875
    },
    {
      "epoch": 2.781991428697663,
      "grad_norm": 1.880157709121704,
      "learning_rate": 9.862806574288016e-06,
      "loss": 0.0618,
      "step": 188900
    },
    {
      "epoch": 2.78235961178775,
      "grad_norm": 1.2659341096878052,
      "learning_rate": 9.86117020232103e-06,
      "loss": 0.0601,
      "step": 188925
    },
    {
      "epoch": 2.782727794877837,
      "grad_norm": 0.9823083281517029,
      "learning_rate": 9.859533830354047e-06,
      "loss": 0.0592,
      "step": 188950
    },
    {
      "epoch": 2.783095977967924,
      "grad_norm": 1.0225110054016113,
      "learning_rate": 9.857897458387061e-06,
      "loss": 0.0605,
      "step": 188975
    },
    {
      "epoch": 2.783464161058011,
      "grad_norm": 1.818889856338501,
      "learning_rate": 9.856261086420077e-06,
      "loss": 0.0557,
      "step": 189000
    },
    {
      "epoch": 2.783832344148098,
      "grad_norm": 1.8652188777923584,
      "learning_rate": 9.854624714453093e-06,
      "loss": 0.0629,
      "step": 189025
    },
    {
      "epoch": 2.784200527238185,
      "grad_norm": 1.6811672449111938,
      "learning_rate": 9.852988342486108e-06,
      "loss": 0.0602,
      "step": 189050
    },
    {
      "epoch": 2.784568710328272,
      "grad_norm": 1.3678489923477173,
      "learning_rate": 9.851351970519122e-06,
      "loss": 0.0664,
      "step": 189075
    },
    {
      "epoch": 2.784936893418359,
      "grad_norm": 1.0080233812332153,
      "learning_rate": 9.849715598552138e-06,
      "loss": 0.0593,
      "step": 189100
    },
    {
      "epoch": 2.785305076508446,
      "grad_norm": 1.2449897527694702,
      "learning_rate": 9.848144681463834e-06,
      "loss": 0.054,
      "step": 189125
    },
    {
      "epoch": 2.785673259598533,
      "grad_norm": 0.9559805393218994,
      "learning_rate": 9.84650830949685e-06,
      "loss": 0.0592,
      "step": 189150
    },
    {
      "epoch": 2.78604144268862,
      "grad_norm": 1.4270541667938232,
      "learning_rate": 9.844871937529865e-06,
      "loss": 0.0537,
      "step": 189175
    },
    {
      "epoch": 2.786409625778707,
      "grad_norm": 1.5277680158615112,
      "learning_rate": 9.843235565562879e-06,
      "loss": 0.0583,
      "step": 189200
    },
    {
      "epoch": 2.786777808868794,
      "grad_norm": 1.2945669889450073,
      "learning_rate": 9.841599193595895e-06,
      "loss": 0.0563,
      "step": 189225
    },
    {
      "epoch": 2.7871459919588815,
      "grad_norm": 0.9818471670150757,
      "learning_rate": 9.839962821628911e-06,
      "loss": 0.0534,
      "step": 189250
    },
    {
      "epoch": 2.7875141750489685,
      "grad_norm": 2.0116989612579346,
      "learning_rate": 9.838326449661927e-06,
      "loss": 0.0615,
      "step": 189275
    },
    {
      "epoch": 2.7878823581390555,
      "grad_norm": 1.2373155355453491,
      "learning_rate": 9.836690077694942e-06,
      "loss": 0.0562,
      "step": 189300
    },
    {
      "epoch": 2.7882505412291425,
      "grad_norm": 1.25845468044281,
      "learning_rate": 9.835053705727956e-06,
      "loss": 0.057,
      "step": 189325
    },
    {
      "epoch": 2.7886187243192295,
      "grad_norm": 1.2892045974731445,
      "learning_rate": 9.833417333760973e-06,
      "loss": 0.0591,
      "step": 189350
    },
    {
      "epoch": 2.7889869074093165,
      "grad_norm": 1.18462073802948,
      "learning_rate": 9.831780961793989e-06,
      "loss": 0.062,
      "step": 189375
    },
    {
      "epoch": 2.7893550904994036,
      "grad_norm": 1.44797682762146,
      "learning_rate": 9.830144589827003e-06,
      "loss": 0.0601,
      "step": 189400
    },
    {
      "epoch": 2.7897232735894906,
      "grad_norm": 0.7727309465408325,
      "learning_rate": 9.82850821786002e-06,
      "loss": 0.0568,
      "step": 189425
    },
    {
      "epoch": 2.7900914566795776,
      "grad_norm": 1.4913910627365112,
      "learning_rate": 9.826871845893034e-06,
      "loss": 0.0618,
      "step": 189450
    },
    {
      "epoch": 2.7904596397696646,
      "grad_norm": 1.3875844478607178,
      "learning_rate": 9.82523547392605e-06,
      "loss": 0.0559,
      "step": 189475
    },
    {
      "epoch": 2.7908278228597516,
      "grad_norm": 1.2802366018295288,
      "learning_rate": 9.823599101959064e-06,
      "loss": 0.0603,
      "step": 189500
    },
    {
      "epoch": 2.7911960059498386,
      "grad_norm": 1.418169617652893,
      "learning_rate": 9.82196272999208e-06,
      "loss": 0.0662,
      "step": 189525
    },
    {
      "epoch": 2.7915641890399256,
      "grad_norm": 1.1409392356872559,
      "learning_rate": 9.820326358025097e-06,
      "loss": 0.0594,
      "step": 189550
    },
    {
      "epoch": 2.7919323721300127,
      "grad_norm": 1.0275858640670776,
      "learning_rate": 9.818689986058113e-06,
      "loss": 0.0587,
      "step": 189575
    },
    {
      "epoch": 2.7923005552201,
      "grad_norm": 1.3041986227035522,
      "learning_rate": 9.817053614091127e-06,
      "loss": 0.0517,
      "step": 189600
    },
    {
      "epoch": 2.792668738310187,
      "grad_norm": 1.662649154663086,
      "learning_rate": 9.815417242124142e-06,
      "loss": 0.0593,
      "step": 189625
    },
    {
      "epoch": 2.793036921400274,
      "grad_norm": 1.0946348905563354,
      "learning_rate": 9.813780870157158e-06,
      "loss": 0.0533,
      "step": 189650
    },
    {
      "epoch": 2.793405104490361,
      "grad_norm": 1.1640589237213135,
      "learning_rate": 9.812144498190174e-06,
      "loss": 0.0634,
      "step": 189675
    },
    {
      "epoch": 2.793773287580448,
      "grad_norm": 1.4606651067733765,
      "learning_rate": 9.81050812622319e-06,
      "loss": 0.0626,
      "step": 189700
    },
    {
      "epoch": 2.794141470670535,
      "grad_norm": 1.110967755317688,
      "learning_rate": 9.808871754256205e-06,
      "loss": 0.0572,
      "step": 189725
    },
    {
      "epoch": 2.794509653760622,
      "grad_norm": 0.9042086601257324,
      "learning_rate": 9.807235382289219e-06,
      "loss": 0.0502,
      "step": 189750
    },
    {
      "epoch": 2.7948778368507092,
      "grad_norm": 1.7488985061645508,
      "learning_rate": 9.805599010322235e-06,
      "loss": 0.0617,
      "step": 189775
    },
    {
      "epoch": 2.7952460199407962,
      "grad_norm": 1.443360447883606,
      "learning_rate": 9.803962638355251e-06,
      "loss": 0.0574,
      "step": 189800
    },
    {
      "epoch": 2.7956142030308833,
      "grad_norm": 1.2800196409225464,
      "learning_rate": 9.802326266388266e-06,
      "loss": 0.0626,
      "step": 189825
    },
    {
      "epoch": 2.7959823861209703,
      "grad_norm": 1.0426499843597412,
      "learning_rate": 9.800689894421282e-06,
      "loss": 0.059,
      "step": 189850
    },
    {
      "epoch": 2.7963505692110573,
      "grad_norm": 0.7933362126350403,
      "learning_rate": 9.799053522454296e-06,
      "loss": 0.0581,
      "step": 189875
    },
    {
      "epoch": 2.7967187523011443,
      "grad_norm": 1.4540982246398926,
      "learning_rate": 9.797417150487313e-06,
      "loss": 0.0576,
      "step": 189900
    },
    {
      "epoch": 2.7970869353912313,
      "grad_norm": 1.2096306085586548,
      "learning_rate": 9.795780778520327e-06,
      "loss": 0.059,
      "step": 189925
    },
    {
      "epoch": 2.7974551184813183,
      "grad_norm": 1.3250669240951538,
      "learning_rate": 9.794144406553343e-06,
      "loss": 0.0562,
      "step": 189950
    },
    {
      "epoch": 2.7978233015714054,
      "grad_norm": 2.0905303955078125,
      "learning_rate": 9.79250803458636e-06,
      "loss": 0.063,
      "step": 189975
    },
    {
      "epoch": 2.7981914846614924,
      "grad_norm": 1.696402668952942,
      "learning_rate": 9.790871662619374e-06,
      "loss": 0.061,
      "step": 190000
    },
    {
      "epoch": 2.7985596677515794,
      "grad_norm": 1.4953043460845947,
      "learning_rate": 9.789235290652388e-06,
      "loss": 0.0581,
      "step": 190025
    },
    {
      "epoch": 2.7989278508416664,
      "grad_norm": 1.6427489519119263,
      "learning_rate": 9.787598918685404e-06,
      "loss": 0.0556,
      "step": 190050
    },
    {
      "epoch": 2.7992960339317534,
      "grad_norm": 1.1342544555664062,
      "learning_rate": 9.78596254671842e-06,
      "loss": 0.0566,
      "step": 190075
    },
    {
      "epoch": 2.7996642170218404,
      "grad_norm": 1.2183283567428589,
      "learning_rate": 9.784326174751437e-06,
      "loss": 0.0587,
      "step": 190100
    },
    {
      "epoch": 2.8000324001119274,
      "grad_norm": 1.477626085281372,
      "learning_rate": 9.782689802784451e-06,
      "loss": 0.0547,
      "step": 190125
    },
    {
      "epoch": 2.8004005832020145,
      "grad_norm": 1.677478551864624,
      "learning_rate": 9.781053430817466e-06,
      "loss": 0.0615,
      "step": 190150
    },
    {
      "epoch": 2.8007687662921015,
      "grad_norm": 1.1885257959365845,
      "learning_rate": 9.779417058850482e-06,
      "loss": 0.0563,
      "step": 190175
    },
    {
      "epoch": 2.801136949382189,
      "grad_norm": 1.6312246322631836,
      "learning_rate": 9.777780686883498e-06,
      "loss": 0.0563,
      "step": 190200
    },
    {
      "epoch": 2.801505132472276,
      "grad_norm": 1.4285744428634644,
      "learning_rate": 9.776144314916514e-06,
      "loss": 0.0634,
      "step": 190225
    },
    {
      "epoch": 2.801873315562363,
      "grad_norm": 1.203006625175476,
      "learning_rate": 9.774507942949529e-06,
      "loss": 0.0681,
      "step": 190250
    },
    {
      "epoch": 2.80224149865245,
      "grad_norm": 1.2275274991989136,
      "learning_rate": 9.772871570982545e-06,
      "loss": 0.067,
      "step": 190275
    },
    {
      "epoch": 2.802609681742537,
      "grad_norm": 0.7865505814552307,
      "learning_rate": 9.77123519901556e-06,
      "loss": 0.0616,
      "step": 190300
    },
    {
      "epoch": 2.802977864832624,
      "grad_norm": 1.6754999160766602,
      "learning_rate": 9.769598827048575e-06,
      "loss": 0.0605,
      "step": 190325
    },
    {
      "epoch": 2.803346047922711,
      "grad_norm": 1.0922694206237793,
      "learning_rate": 9.76796245508159e-06,
      "loss": 0.0603,
      "step": 190350
    },
    {
      "epoch": 2.803714231012798,
      "grad_norm": 1.393845558166504,
      "learning_rate": 9.766326083114606e-06,
      "loss": 0.058,
      "step": 190375
    },
    {
      "epoch": 2.804082414102885,
      "grad_norm": 1.4487552642822266,
      "learning_rate": 9.764689711147622e-06,
      "loss": 0.0651,
      "step": 190400
    },
    {
      "epoch": 2.804450597192972,
      "grad_norm": 1.6551408767700195,
      "learning_rate": 9.763053339180637e-06,
      "loss": 0.0579,
      "step": 190425
    },
    {
      "epoch": 2.804818780283059,
      "grad_norm": 1.6418046951293945,
      "learning_rate": 9.761416967213651e-06,
      "loss": 0.0545,
      "step": 190450
    },
    {
      "epoch": 2.805186963373146,
      "grad_norm": 1.3582541942596436,
      "learning_rate": 9.759780595246667e-06,
      "loss": 0.0539,
      "step": 190475
    },
    {
      "epoch": 2.805555146463233,
      "grad_norm": 1.7742160558700562,
      "learning_rate": 9.758144223279683e-06,
      "loss": 0.0687,
      "step": 190500
    },
    {
      "epoch": 2.80592332955332,
      "grad_norm": 1.258548378944397,
      "learning_rate": 9.7565078513127e-06,
      "loss": 0.0592,
      "step": 190525
    },
    {
      "epoch": 2.806291512643407,
      "grad_norm": 1.7473278045654297,
      "learning_rate": 9.754871479345714e-06,
      "loss": 0.0584,
      "step": 190550
    },
    {
      "epoch": 2.8066596957334946,
      "grad_norm": 1.3154525756835938,
      "learning_rate": 9.753235107378728e-06,
      "loss": 0.0556,
      "step": 190575
    },
    {
      "epoch": 2.8070278788235816,
      "grad_norm": 1.2229403257369995,
      "learning_rate": 9.751598735411745e-06,
      "loss": 0.0636,
      "step": 190600
    },
    {
      "epoch": 2.8073960619136686,
      "grad_norm": 1.0417224168777466,
      "learning_rate": 9.74996236344476e-06,
      "loss": 0.0575,
      "step": 190625
    },
    {
      "epoch": 2.8077642450037557,
      "grad_norm": 1.4170892238616943,
      "learning_rate": 9.748325991477777e-06,
      "loss": 0.062,
      "step": 190650
    },
    {
      "epoch": 2.8081324280938427,
      "grad_norm": 1.2770042419433594,
      "learning_rate": 9.746689619510791e-06,
      "loss": 0.0586,
      "step": 190675
    },
    {
      "epoch": 2.8085006111839297,
      "grad_norm": 1.2343826293945312,
      "learning_rate": 9.745053247543806e-06,
      "loss": 0.0518,
      "step": 190700
    },
    {
      "epoch": 2.8088687942740167,
      "grad_norm": 1.0419585704803467,
      "learning_rate": 9.743416875576822e-06,
      "loss": 0.0569,
      "step": 190725
    },
    {
      "epoch": 2.8092369773641037,
      "grad_norm": 1.4943057298660278,
      "learning_rate": 9.741780503609838e-06,
      "loss": 0.0646,
      "step": 190750
    },
    {
      "epoch": 2.8096051604541907,
      "grad_norm": 1.158626675605774,
      "learning_rate": 9.740144131642853e-06,
      "loss": 0.0563,
      "step": 190775
    },
    {
      "epoch": 2.8099733435442777,
      "grad_norm": 1.5623724460601807,
      "learning_rate": 9.738507759675869e-06,
      "loss": 0.0594,
      "step": 190800
    },
    {
      "epoch": 2.8103415266343648,
      "grad_norm": 1.8177307844161987,
      "learning_rate": 9.736871387708883e-06,
      "loss": 0.0676,
      "step": 190825
    },
    {
      "epoch": 2.8107097097244518,
      "grad_norm": 1.16116201877594,
      "learning_rate": 9.7352350157419e-06,
      "loss": 0.0592,
      "step": 190850
    },
    {
      "epoch": 2.811077892814539,
      "grad_norm": 1.3950963020324707,
      "learning_rate": 9.733598643774914e-06,
      "loss": 0.0591,
      "step": 190875
    },
    {
      "epoch": 2.811446075904626,
      "grad_norm": 1.4604241847991943,
      "learning_rate": 9.73196227180793e-06,
      "loss": 0.0623,
      "step": 190900
    },
    {
      "epoch": 2.811814258994713,
      "grad_norm": 1.6212618350982666,
      "learning_rate": 9.730325899840946e-06,
      "loss": 0.065,
      "step": 190925
    },
    {
      "epoch": 2.8121824420848,
      "grad_norm": 1.211201548576355,
      "learning_rate": 9.72868952787396e-06,
      "loss": 0.0585,
      "step": 190950
    },
    {
      "epoch": 2.812550625174887,
      "grad_norm": 1.1423178911209106,
      "learning_rate": 9.727053155906977e-06,
      "loss": 0.0554,
      "step": 190975
    },
    {
      "epoch": 2.812918808264974,
      "grad_norm": 0.9317471385002136,
      "learning_rate": 9.725416783939991e-06,
      "loss": 0.0626,
      "step": 191000
    },
    {
      "epoch": 2.813286991355061,
      "grad_norm": 1.9102895259857178,
      "learning_rate": 9.723780411973007e-06,
      "loss": 0.0574,
      "step": 191025
    },
    {
      "epoch": 2.813655174445148,
      "grad_norm": 1.346814751625061,
      "learning_rate": 9.722144040006023e-06,
      "loss": 0.0589,
      "step": 191050
    },
    {
      "epoch": 2.814023357535235,
      "grad_norm": 1.6709024906158447,
      "learning_rate": 9.720507668039038e-06,
      "loss": 0.0641,
      "step": 191075
    },
    {
      "epoch": 2.814391540625322,
      "grad_norm": 1.28024423122406,
      "learning_rate": 9.718871296072054e-06,
      "loss": 0.0578,
      "step": 191100
    },
    {
      "epoch": 2.814759723715409,
      "grad_norm": 1.3274121284484863,
      "learning_rate": 9.717234924105068e-06,
      "loss": 0.0573,
      "step": 191125
    },
    {
      "epoch": 2.815127906805496,
      "grad_norm": 1.5620265007019043,
      "learning_rate": 9.715598552138085e-06,
      "loss": 0.0617,
      "step": 191150
    },
    {
      "epoch": 2.8154960898955834,
      "grad_norm": 0.5231131911277771,
      "learning_rate": 9.7139621801711e-06,
      "loss": 0.051,
      "step": 191175
    },
    {
      "epoch": 2.8158642729856704,
      "grad_norm": 1.5454624891281128,
      "learning_rate": 9.712325808204115e-06,
      "loss": 0.0581,
      "step": 191200
    },
    {
      "epoch": 2.8162324560757575,
      "grad_norm": 0.9377264380455017,
      "learning_rate": 9.710689436237131e-06,
      "loss": 0.0558,
      "step": 191225
    },
    {
      "epoch": 2.8166006391658445,
      "grad_norm": 1.467632532119751,
      "learning_rate": 9.709053064270146e-06,
      "loss": 0.0579,
      "step": 191250
    },
    {
      "epoch": 2.8169688222559315,
      "grad_norm": 1.655173659324646,
      "learning_rate": 9.707416692303162e-06,
      "loss": 0.0616,
      "step": 191275
    },
    {
      "epoch": 2.8173370053460185,
      "grad_norm": 1.3735733032226562,
      "learning_rate": 9.705780320336176e-06,
      "loss": 0.0591,
      "step": 191300
    },
    {
      "epoch": 2.8177051884361055,
      "grad_norm": 1.1340259313583374,
      "learning_rate": 9.704143948369193e-06,
      "loss": 0.0587,
      "step": 191325
    },
    {
      "epoch": 2.8180733715261925,
      "grad_norm": 1.3795756101608276,
      "learning_rate": 9.702507576402209e-06,
      "loss": 0.0538,
      "step": 191350
    },
    {
      "epoch": 2.8184415546162795,
      "grad_norm": 0.8095846772193909,
      "learning_rate": 9.700871204435223e-06,
      "loss": 0.0554,
      "step": 191375
    },
    {
      "epoch": 2.8188097377063666,
      "grad_norm": 1.9934107065200806,
      "learning_rate": 9.699234832468238e-06,
      "loss": 0.0655,
      "step": 191400
    },
    {
      "epoch": 2.8191779207964536,
      "grad_norm": 1.474034309387207,
      "learning_rate": 9.697598460501254e-06,
      "loss": 0.0621,
      "step": 191425
    },
    {
      "epoch": 2.8195461038865406,
      "grad_norm": 0.9095655679702759,
      "learning_rate": 9.69596208853427e-06,
      "loss": 0.053,
      "step": 191450
    },
    {
      "epoch": 2.8199142869766276,
      "grad_norm": 0.9106353521347046,
      "learning_rate": 9.694325716567286e-06,
      "loss": 0.0551,
      "step": 191475
    },
    {
      "epoch": 2.8202824700667146,
      "grad_norm": 1.4420366287231445,
      "learning_rate": 9.6926893446003e-06,
      "loss": 0.0645,
      "step": 191500
    },
    {
      "epoch": 2.820650653156802,
      "grad_norm": 1.4797770977020264,
      "learning_rate": 9.691052972633315e-06,
      "loss": 0.0609,
      "step": 191525
    },
    {
      "epoch": 2.821018836246889,
      "grad_norm": 1.2653862237930298,
      "learning_rate": 9.689416600666331e-06,
      "loss": 0.0594,
      "step": 191550
    },
    {
      "epoch": 2.821387019336976,
      "grad_norm": 1.404115915298462,
      "learning_rate": 9.687780228699347e-06,
      "loss": 0.062,
      "step": 191575
    },
    {
      "epoch": 2.821755202427063,
      "grad_norm": 1.2101653814315796,
      "learning_rate": 9.686143856732362e-06,
      "loss": 0.0537,
      "step": 191600
    },
    {
      "epoch": 2.82212338551715,
      "grad_norm": 1.290461540222168,
      "learning_rate": 9.684507484765378e-06,
      "loss": 0.0535,
      "step": 191625
    },
    {
      "epoch": 2.822491568607237,
      "grad_norm": 1.5852519273757935,
      "learning_rate": 9.682936567677072e-06,
      "loss": 0.0689,
      "step": 191650
    },
    {
      "epoch": 2.822859751697324,
      "grad_norm": 1.3701965808868408,
      "learning_rate": 9.681300195710088e-06,
      "loss": 0.0636,
      "step": 191675
    },
    {
      "epoch": 2.823227934787411,
      "grad_norm": 1.1893665790557861,
      "learning_rate": 9.679663823743104e-06,
      "loss": 0.0517,
      "step": 191700
    },
    {
      "epoch": 2.823596117877498,
      "grad_norm": 1.3377516269683838,
      "learning_rate": 9.678027451776119e-06,
      "loss": 0.0552,
      "step": 191725
    },
    {
      "epoch": 2.823964300967585,
      "grad_norm": 1.5781387090682983,
      "learning_rate": 9.676391079809135e-06,
      "loss": 0.0565,
      "step": 191750
    },
    {
      "epoch": 2.8243324840576722,
      "grad_norm": 1.5812073945999146,
      "learning_rate": 9.67475470784215e-06,
      "loss": 0.0554,
      "step": 191775
    },
    {
      "epoch": 2.8247006671477592,
      "grad_norm": 1.8488749265670776,
      "learning_rate": 9.673118335875165e-06,
      "loss": 0.0596,
      "step": 191800
    },
    {
      "epoch": 2.8250688502378463,
      "grad_norm": 1.237788200378418,
      "learning_rate": 9.67148196390818e-06,
      "loss": 0.0626,
      "step": 191825
    },
    {
      "epoch": 2.8254370333279333,
      "grad_norm": 1.0844690799713135,
      "learning_rate": 9.669845591941196e-06,
      "loss": 0.0583,
      "step": 191850
    },
    {
      "epoch": 2.8258052164180203,
      "grad_norm": 1.0248757600784302,
      "learning_rate": 9.668209219974212e-06,
      "loss": 0.0657,
      "step": 191875
    },
    {
      "epoch": 2.8261733995081073,
      "grad_norm": 1.0006017684936523,
      "learning_rate": 9.666572848007227e-06,
      "loss": 0.0588,
      "step": 191900
    },
    {
      "epoch": 2.8265415825981943,
      "grad_norm": 1.8060781955718994,
      "learning_rate": 9.664936476040243e-06,
      "loss": 0.0618,
      "step": 191925
    },
    {
      "epoch": 2.8269097656882813,
      "grad_norm": 0.8489823341369629,
      "learning_rate": 9.663300104073257e-06,
      "loss": 0.0511,
      "step": 191950
    },
    {
      "epoch": 2.8272779487783684,
      "grad_norm": 1.4254478216171265,
      "learning_rate": 9.661663732106273e-06,
      "loss": 0.0586,
      "step": 191975
    },
    {
      "epoch": 2.8276461318684554,
      "grad_norm": 1.8723089694976807,
      "learning_rate": 9.66002736013929e-06,
      "loss": 0.0651,
      "step": 192000
    },
    {
      "epoch": 2.8280143149585424,
      "grad_norm": 1.5793136358261108,
      "learning_rate": 9.658390988172304e-06,
      "loss": 0.0566,
      "step": 192025
    },
    {
      "epoch": 2.8283824980486294,
      "grad_norm": 0.5374088883399963,
      "learning_rate": 9.65675461620532e-06,
      "loss": 0.0612,
      "step": 192050
    },
    {
      "epoch": 2.8287506811387164,
      "grad_norm": 0.8083250522613525,
      "learning_rate": 9.655118244238335e-06,
      "loss": 0.056,
      "step": 192075
    },
    {
      "epoch": 2.8291188642288034,
      "grad_norm": 2.0485692024230957,
      "learning_rate": 9.65348187227135e-06,
      "loss": 0.0553,
      "step": 192100
    },
    {
      "epoch": 2.829487047318891,
      "grad_norm": 1.5060235261917114,
      "learning_rate": 9.651845500304367e-06,
      "loss": 0.0591,
      "step": 192125
    },
    {
      "epoch": 2.829855230408978,
      "grad_norm": 1.2350915670394897,
      "learning_rate": 9.650209128337381e-06,
      "loss": 0.0597,
      "step": 192150
    },
    {
      "epoch": 2.830223413499065,
      "grad_norm": 1.228594422340393,
      "learning_rate": 9.648572756370397e-06,
      "loss": 0.0626,
      "step": 192175
    },
    {
      "epoch": 2.830591596589152,
      "grad_norm": 1.490804672241211,
      "learning_rate": 9.646936384403412e-06,
      "loss": 0.065,
      "step": 192200
    },
    {
      "epoch": 2.830959779679239,
      "grad_norm": 1.1721911430358887,
      "learning_rate": 9.645300012436428e-06,
      "loss": 0.0636,
      "step": 192225
    },
    {
      "epoch": 2.831327962769326,
      "grad_norm": 1.5919784307479858,
      "learning_rate": 9.643663640469443e-06,
      "loss": 0.0646,
      "step": 192250
    },
    {
      "epoch": 2.831696145859413,
      "grad_norm": 1.3846238851547241,
      "learning_rate": 9.642027268502459e-06,
      "loss": 0.0604,
      "step": 192275
    },
    {
      "epoch": 2.8320643289495,
      "grad_norm": 1.2938507795333862,
      "learning_rate": 9.640390896535475e-06,
      "loss": 0.0586,
      "step": 192300
    },
    {
      "epoch": 2.832432512039587,
      "grad_norm": 1.9460163116455078,
      "learning_rate": 9.63875452456849e-06,
      "loss": 0.0597,
      "step": 192325
    },
    {
      "epoch": 2.832800695129674,
      "grad_norm": 1.302515983581543,
      "learning_rate": 9.637118152601504e-06,
      "loss": 0.0601,
      "step": 192350
    },
    {
      "epoch": 2.833168878219761,
      "grad_norm": 1.4734759330749512,
      "learning_rate": 9.63548178063452e-06,
      "loss": 0.0626,
      "step": 192375
    },
    {
      "epoch": 2.833537061309848,
      "grad_norm": 1.3492851257324219,
      "learning_rate": 9.633845408667536e-06,
      "loss": 0.0581,
      "step": 192400
    },
    {
      "epoch": 2.833905244399935,
      "grad_norm": 1.4389796257019043,
      "learning_rate": 9.632209036700552e-06,
      "loss": 0.0604,
      "step": 192425
    },
    {
      "epoch": 2.834273427490022,
      "grad_norm": 1.309415578842163,
      "learning_rate": 9.630572664733567e-06,
      "loss": 0.0644,
      "step": 192450
    },
    {
      "epoch": 2.8346416105801095,
      "grad_norm": 1.527631402015686,
      "learning_rate": 9.628936292766581e-06,
      "loss": 0.0616,
      "step": 192475
    },
    {
      "epoch": 2.8350097936701966,
      "grad_norm": 1.1195049285888672,
      "learning_rate": 9.627299920799597e-06,
      "loss": 0.0597,
      "step": 192500
    },
    {
      "epoch": 2.8353779767602836,
      "grad_norm": 1.2835516929626465,
      "learning_rate": 9.625663548832613e-06,
      "loss": 0.0567,
      "step": 192525
    },
    {
      "epoch": 2.8357461598503706,
      "grad_norm": 1.5211371183395386,
      "learning_rate": 9.624027176865628e-06,
      "loss": 0.0539,
      "step": 192550
    },
    {
      "epoch": 2.8361143429404576,
      "grad_norm": 1.4089497327804565,
      "learning_rate": 9.622390804898644e-06,
      "loss": 0.06,
      "step": 192575
    },
    {
      "epoch": 2.8364825260305446,
      "grad_norm": 1.1427394151687622,
      "learning_rate": 9.62075443293166e-06,
      "loss": 0.0577,
      "step": 192600
    },
    {
      "epoch": 2.8368507091206316,
      "grad_norm": 1.4656684398651123,
      "learning_rate": 9.619118060964675e-06,
      "loss": 0.0575,
      "step": 192625
    },
    {
      "epoch": 2.8372188922107187,
      "grad_norm": 1.2018636465072632,
      "learning_rate": 9.61748168899769e-06,
      "loss": 0.0594,
      "step": 192650
    },
    {
      "epoch": 2.8375870753008057,
      "grad_norm": 1.888146162033081,
      "learning_rate": 9.615845317030705e-06,
      "loss": 0.0596,
      "step": 192675
    },
    {
      "epoch": 2.8379552583908927,
      "grad_norm": 1.2280315160751343,
      "learning_rate": 9.614208945063721e-06,
      "loss": 0.0596,
      "step": 192700
    },
    {
      "epoch": 2.8383234414809797,
      "grad_norm": 1.18323814868927,
      "learning_rate": 9.612572573096738e-06,
      "loss": 0.053,
      "step": 192725
    },
    {
      "epoch": 2.8386916245710667,
      "grad_norm": 1.4740184545516968,
      "learning_rate": 9.610936201129752e-06,
      "loss": 0.0569,
      "step": 192750
    },
    {
      "epoch": 2.8390598076611537,
      "grad_norm": 1.4008374214172363,
      "learning_rate": 9.609299829162766e-06,
      "loss": 0.0659,
      "step": 192775
    },
    {
      "epoch": 2.8394279907512407,
      "grad_norm": 1.2718358039855957,
      "learning_rate": 9.607663457195783e-06,
      "loss": 0.0608,
      "step": 192800
    },
    {
      "epoch": 2.8397961738413278,
      "grad_norm": 1.1847437620162964,
      "learning_rate": 9.606027085228799e-06,
      "loss": 0.0551,
      "step": 192825
    },
    {
      "epoch": 2.840164356931415,
      "grad_norm": 1.2949888706207275,
      "learning_rate": 9.604390713261815e-06,
      "loss": 0.0625,
      "step": 192850
    },
    {
      "epoch": 2.840532540021502,
      "grad_norm": 1.5082757472991943,
      "learning_rate": 9.60275434129483e-06,
      "loss": 0.0596,
      "step": 192875
    },
    {
      "epoch": 2.840900723111589,
      "grad_norm": 1.6205650568008423,
      "learning_rate": 9.601117969327844e-06,
      "loss": 0.0605,
      "step": 192900
    },
    {
      "epoch": 2.841268906201676,
      "grad_norm": 1.9189810752868652,
      "learning_rate": 9.59948159736086e-06,
      "loss": 0.0649,
      "step": 192925
    },
    {
      "epoch": 2.841637089291763,
      "grad_norm": 1.4988007545471191,
      "learning_rate": 9.597845225393876e-06,
      "loss": 0.0558,
      "step": 192950
    },
    {
      "epoch": 2.84200527238185,
      "grad_norm": 1.3511971235275269,
      "learning_rate": 9.59620885342689e-06,
      "loss": 0.0567,
      "step": 192975
    },
    {
      "epoch": 2.842373455471937,
      "grad_norm": 1.4849319458007812,
      "learning_rate": 9.594572481459907e-06,
      "loss": 0.0648,
      "step": 193000
    },
    {
      "epoch": 2.842741638562024,
      "grad_norm": 1.265085220336914,
      "learning_rate": 9.592936109492921e-06,
      "loss": 0.0598,
      "step": 193025
    },
    {
      "epoch": 2.843109821652111,
      "grad_norm": 1.3015947341918945,
      "learning_rate": 9.591299737525937e-06,
      "loss": 0.059,
      "step": 193050
    },
    {
      "epoch": 2.8434780047421984,
      "grad_norm": 1.3761097192764282,
      "learning_rate": 9.589663365558953e-06,
      "loss": 0.0505,
      "step": 193075
    },
    {
      "epoch": 2.8438461878322854,
      "grad_norm": 1.2499710321426392,
      "learning_rate": 9.588026993591968e-06,
      "loss": 0.0639,
      "step": 193100
    },
    {
      "epoch": 2.8442143709223724,
      "grad_norm": 0.9448404908180237,
      "learning_rate": 9.586390621624984e-06,
      "loss": 0.0584,
      "step": 193125
    },
    {
      "epoch": 2.8445825540124594,
      "grad_norm": 0.8087537884712219,
      "learning_rate": 9.584754249657999e-06,
      "loss": 0.0629,
      "step": 193150
    },
    {
      "epoch": 2.8449507371025464,
      "grad_norm": 1.224900245666504,
      "learning_rate": 9.583117877691015e-06,
      "loss": 0.0561,
      "step": 193175
    },
    {
      "epoch": 2.8453189201926334,
      "grad_norm": 1.2548829317092896,
      "learning_rate": 9.58148150572403e-06,
      "loss": 0.0615,
      "step": 193200
    },
    {
      "epoch": 2.8456871032827205,
      "grad_norm": 1.1911438703536987,
      "learning_rate": 9.579845133757045e-06,
      "loss": 0.0556,
      "step": 193225
    },
    {
      "epoch": 2.8460552863728075,
      "grad_norm": 1.4781550168991089,
      "learning_rate": 9.578208761790061e-06,
      "loss": 0.0599,
      "step": 193250
    },
    {
      "epoch": 2.8464234694628945,
      "grad_norm": 1.4878883361816406,
      "learning_rate": 9.576572389823076e-06,
      "loss": 0.0603,
      "step": 193275
    },
    {
      "epoch": 2.8467916525529815,
      "grad_norm": 1.0348962545394897,
      "learning_rate": 9.574936017856092e-06,
      "loss": 0.0594,
      "step": 193300
    },
    {
      "epoch": 2.8471598356430685,
      "grad_norm": 1.458094835281372,
      "learning_rate": 9.573299645889107e-06,
      "loss": 0.0619,
      "step": 193325
    },
    {
      "epoch": 2.8475280187331555,
      "grad_norm": 0.9041836857795715,
      "learning_rate": 9.571663273922123e-06,
      "loss": 0.0569,
      "step": 193350
    },
    {
      "epoch": 2.8478962018232425,
      "grad_norm": 1.3955724239349365,
      "learning_rate": 9.570026901955139e-06,
      "loss": 0.0633,
      "step": 193375
    },
    {
      "epoch": 2.8482643849133296,
      "grad_norm": 1.040228009223938,
      "learning_rate": 9.568390529988153e-06,
      "loss": 0.0605,
      "step": 193400
    },
    {
      "epoch": 2.8486325680034166,
      "grad_norm": 0.4016520082950592,
      "learning_rate": 9.56675415802117e-06,
      "loss": 0.061,
      "step": 193425
    },
    {
      "epoch": 2.849000751093504,
      "grad_norm": 1.5045928955078125,
      "learning_rate": 9.565117786054184e-06,
      "loss": 0.0589,
      "step": 193450
    },
    {
      "epoch": 2.849368934183591,
      "grad_norm": 1.043886661529541,
      "learning_rate": 9.5634814140872e-06,
      "loss": 0.0527,
      "step": 193475
    },
    {
      "epoch": 2.849737117273678,
      "grad_norm": 1.453575849533081,
      "learning_rate": 9.561845042120215e-06,
      "loss": 0.0613,
      "step": 193500
    },
    {
      "epoch": 2.850105300363765,
      "grad_norm": 1.272387146949768,
      "learning_rate": 9.56020867015323e-06,
      "loss": 0.0514,
      "step": 193525
    },
    {
      "epoch": 2.850473483453852,
      "grad_norm": 1.2844825983047485,
      "learning_rate": 9.558572298186247e-06,
      "loss": 0.0662,
      "step": 193550
    },
    {
      "epoch": 2.850841666543939,
      "grad_norm": 2.290719509124756,
      "learning_rate": 9.556935926219261e-06,
      "loss": 0.06,
      "step": 193575
    },
    {
      "epoch": 2.851209849634026,
      "grad_norm": 1.6012428998947144,
      "learning_rate": 9.555299554252277e-06,
      "loss": 0.0598,
      "step": 193600
    },
    {
      "epoch": 2.851578032724113,
      "grad_norm": 1.329693078994751,
      "learning_rate": 9.553663182285292e-06,
      "loss": 0.0666,
      "step": 193625
    },
    {
      "epoch": 2.8519462158142,
      "grad_norm": 1.206016182899475,
      "learning_rate": 9.552026810318308e-06,
      "loss": 0.0569,
      "step": 193650
    },
    {
      "epoch": 2.852314398904287,
      "grad_norm": 1.5623241662979126,
      "learning_rate": 9.550390438351324e-06,
      "loss": 0.0614,
      "step": 193675
    },
    {
      "epoch": 2.852682581994374,
      "grad_norm": 1.1449590921401978,
      "learning_rate": 9.548754066384339e-06,
      "loss": 0.0513,
      "step": 193700
    },
    {
      "epoch": 2.853050765084461,
      "grad_norm": 1.553481101989746,
      "learning_rate": 9.547117694417353e-06,
      "loss": 0.0553,
      "step": 193725
    },
    {
      "epoch": 2.853418948174548,
      "grad_norm": 1.6526141166687012,
      "learning_rate": 9.54548132245037e-06,
      "loss": 0.0574,
      "step": 193750
    },
    {
      "epoch": 2.8537871312646352,
      "grad_norm": 0.970827579498291,
      "learning_rate": 9.543844950483385e-06,
      "loss": 0.0504,
      "step": 193775
    },
    {
      "epoch": 2.8541553143547223,
      "grad_norm": 0.8908568620681763,
      "learning_rate": 9.542208578516402e-06,
      "loss": 0.0592,
      "step": 193800
    },
    {
      "epoch": 2.8545234974448093,
      "grad_norm": 1.5711696147918701,
      "learning_rate": 9.540572206549416e-06,
      "loss": 0.0597,
      "step": 193825
    },
    {
      "epoch": 2.8548916805348963,
      "grad_norm": 1.2680684328079224,
      "learning_rate": 9.53893583458243e-06,
      "loss": 0.0555,
      "step": 193850
    },
    {
      "epoch": 2.8552598636249833,
      "grad_norm": 1.4148808717727661,
      "learning_rate": 9.537299462615447e-06,
      "loss": 0.0617,
      "step": 193875
    },
    {
      "epoch": 2.8556280467150703,
      "grad_norm": 1.4733003377914429,
      "learning_rate": 9.535663090648463e-06,
      "loss": 0.0561,
      "step": 193900
    },
    {
      "epoch": 2.8559962298051573,
      "grad_norm": 1.8704921007156372,
      "learning_rate": 9.534026718681477e-06,
      "loss": 0.0557,
      "step": 193925
    },
    {
      "epoch": 2.8563644128952443,
      "grad_norm": 1.3796284198760986,
      "learning_rate": 9.532390346714493e-06,
      "loss": 0.0627,
      "step": 193950
    },
    {
      "epoch": 2.8567325959853314,
      "grad_norm": 1.5480225086212158,
      "learning_rate": 9.530753974747508e-06,
      "loss": 0.0536,
      "step": 193975
    },
    {
      "epoch": 2.8571007790754184,
      "grad_norm": 1.2552282810211182,
      "learning_rate": 9.529117602780524e-06,
      "loss": 0.0569,
      "step": 194000
    },
    {
      "epoch": 2.8574689621655054,
      "grad_norm": 1.0808730125427246,
      "learning_rate": 9.527481230813538e-06,
      "loss": 0.0578,
      "step": 194025
    },
    {
      "epoch": 2.857837145255593,
      "grad_norm": 1.4201253652572632,
      "learning_rate": 9.525844858846555e-06,
      "loss": 0.0571,
      "step": 194050
    },
    {
      "epoch": 2.85820532834568,
      "grad_norm": 1.670086145401001,
      "learning_rate": 9.52420848687957e-06,
      "loss": 0.0618,
      "step": 194075
    },
    {
      "epoch": 2.858573511435767,
      "grad_norm": 1.5384279489517212,
      "learning_rate": 9.522572114912585e-06,
      "loss": 0.0603,
      "step": 194100
    },
    {
      "epoch": 2.858941694525854,
      "grad_norm": 2.0621936321258545,
      "learning_rate": 9.520935742945601e-06,
      "loss": 0.0569,
      "step": 194125
    },
    {
      "epoch": 2.859309877615941,
      "grad_norm": 1.7698636054992676,
      "learning_rate": 9.519299370978616e-06,
      "loss": 0.0597,
      "step": 194150
    },
    {
      "epoch": 2.859678060706028,
      "grad_norm": 1.7682251930236816,
      "learning_rate": 9.517662999011632e-06,
      "loss": 0.0596,
      "step": 194175
    },
    {
      "epoch": 2.860046243796115,
      "grad_norm": 1.6107062101364136,
      "learning_rate": 9.516026627044648e-06,
      "loss": 0.0608,
      "step": 194200
    },
    {
      "epoch": 2.860414426886202,
      "grad_norm": 1.0938622951507568,
      "learning_rate": 9.514390255077664e-06,
      "loss": 0.0572,
      "step": 194225
    },
    {
      "epoch": 2.860782609976289,
      "grad_norm": 1.2379047870635986,
      "learning_rate": 9.512753883110679e-06,
      "loss": 0.0575,
      "step": 194250
    },
    {
      "epoch": 2.861150793066376,
      "grad_norm": 1.2361403703689575,
      "learning_rate": 9.511117511143693e-06,
      "loss": 0.0634,
      "step": 194275
    },
    {
      "epoch": 2.861518976156463,
      "grad_norm": 0.9031369686126709,
      "learning_rate": 9.50948113917671e-06,
      "loss": 0.0578,
      "step": 194300
    },
    {
      "epoch": 2.86188715924655,
      "grad_norm": 1.6158697605133057,
      "learning_rate": 9.507844767209725e-06,
      "loss": 0.0577,
      "step": 194325
    },
    {
      "epoch": 2.862255342336637,
      "grad_norm": 1.3848204612731934,
      "learning_rate": 9.50620839524274e-06,
      "loss": 0.0631,
      "step": 194350
    },
    {
      "epoch": 2.862623525426724,
      "grad_norm": 1.1341049671173096,
      "learning_rate": 9.504572023275756e-06,
      "loss": 0.0628,
      "step": 194375
    },
    {
      "epoch": 2.8629917085168115,
      "grad_norm": 1.663681983947754,
      "learning_rate": 9.50293565130877e-06,
      "loss": 0.0583,
      "step": 194400
    },
    {
      "epoch": 2.8633598916068985,
      "grad_norm": 1.6502104997634888,
      "learning_rate": 9.501299279341787e-06,
      "loss": 0.0617,
      "step": 194425
    },
    {
      "epoch": 2.8637280746969855,
      "grad_norm": 1.509641170501709,
      "learning_rate": 9.499662907374801e-06,
      "loss": 0.0594,
      "step": 194450
    },
    {
      "epoch": 2.8640962577870726,
      "grad_norm": 1.3160899877548218,
      "learning_rate": 9.498091990286497e-06,
      "loss": 0.0549,
      "step": 194475
    },
    {
      "epoch": 2.8644644408771596,
      "grad_norm": 1.005088448524475,
      "learning_rate": 9.496455618319513e-06,
      "loss": 0.0578,
      "step": 194500
    },
    {
      "epoch": 2.8648326239672466,
      "grad_norm": 1.698104977607727,
      "learning_rate": 9.494819246352527e-06,
      "loss": 0.0589,
      "step": 194525
    },
    {
      "epoch": 2.8652008070573336,
      "grad_norm": 0.8770014643669128,
      "learning_rate": 9.493182874385543e-06,
      "loss": 0.0618,
      "step": 194550
    },
    {
      "epoch": 2.8655689901474206,
      "grad_norm": 1.386398196220398,
      "learning_rate": 9.491546502418558e-06,
      "loss": 0.0615,
      "step": 194575
    },
    {
      "epoch": 2.8659371732375076,
      "grad_norm": 1.7904797792434692,
      "learning_rate": 9.489910130451574e-06,
      "loss": 0.0643,
      "step": 194600
    },
    {
      "epoch": 2.8663053563275946,
      "grad_norm": 1.0641735792160034,
      "learning_rate": 9.48827375848459e-06,
      "loss": 0.0583,
      "step": 194625
    },
    {
      "epoch": 2.8666735394176817,
      "grad_norm": 1.3294438123703003,
      "learning_rate": 9.486637386517605e-06,
      "loss": 0.0579,
      "step": 194650
    },
    {
      "epoch": 2.8670417225077687,
      "grad_norm": 1.261055827140808,
      "learning_rate": 9.48500101455062e-06,
      "loss": 0.0602,
      "step": 194675
    },
    {
      "epoch": 2.8674099055978557,
      "grad_norm": 1.0894662141799927,
      "learning_rate": 9.483364642583635e-06,
      "loss": 0.0589,
      "step": 194700
    },
    {
      "epoch": 2.8677780886879427,
      "grad_norm": 1.9460142850875854,
      "learning_rate": 9.481728270616651e-06,
      "loss": 0.0592,
      "step": 194725
    },
    {
      "epoch": 2.8681462717780297,
      "grad_norm": 1.5306462049484253,
      "learning_rate": 9.480091898649668e-06,
      "loss": 0.0643,
      "step": 194750
    },
    {
      "epoch": 2.8685144548681167,
      "grad_norm": 1.7471768856048584,
      "learning_rate": 9.478455526682682e-06,
      "loss": 0.0592,
      "step": 194775
    },
    {
      "epoch": 2.8688826379582038,
      "grad_norm": 1.1541129350662231,
      "learning_rate": 9.476819154715697e-06,
      "loss": 0.0547,
      "step": 194800
    },
    {
      "epoch": 2.8692508210482908,
      "grad_norm": 1.23127281665802,
      "learning_rate": 9.475182782748713e-06,
      "loss": 0.0625,
      "step": 194825
    },
    {
      "epoch": 2.869619004138378,
      "grad_norm": 1.8355498313903809,
      "learning_rate": 9.473546410781729e-06,
      "loss": 0.0646,
      "step": 194850
    },
    {
      "epoch": 2.869987187228465,
      "grad_norm": 1.1252093315124512,
      "learning_rate": 9.471910038814743e-06,
      "loss": 0.0601,
      "step": 194875
    },
    {
      "epoch": 2.870355370318552,
      "grad_norm": 1.7468353509902954,
      "learning_rate": 9.47027366684776e-06,
      "loss": 0.0624,
      "step": 194900
    },
    {
      "epoch": 2.870723553408639,
      "grad_norm": 1.70834219455719,
      "learning_rate": 9.468637294880776e-06,
      "loss": 0.0636,
      "step": 194925
    },
    {
      "epoch": 2.871091736498726,
      "grad_norm": 1.8979442119598389,
      "learning_rate": 9.46700092291379e-06,
      "loss": 0.0595,
      "step": 194950
    },
    {
      "epoch": 2.871459919588813,
      "grad_norm": 1.704552412033081,
      "learning_rate": 9.465364550946805e-06,
      "loss": 0.0559,
      "step": 194975
    },
    {
      "epoch": 2.8718281026789003,
      "grad_norm": 1.685447335243225,
      "learning_rate": 9.46372817897982e-06,
      "loss": 0.0633,
      "step": 195000
    },
    {
      "epoch": 2.8721962857689873,
      "grad_norm": 1.3126622438430786,
      "learning_rate": 9.462091807012837e-06,
      "loss": 0.0629,
      "step": 195025
    },
    {
      "epoch": 2.8725644688590743,
      "grad_norm": 1.6387419700622559,
      "learning_rate": 9.460455435045853e-06,
      "loss": 0.0617,
      "step": 195050
    },
    {
      "epoch": 2.8729326519491614,
      "grad_norm": 1.2038549184799194,
      "learning_rate": 9.458819063078867e-06,
      "loss": 0.0581,
      "step": 195075
    },
    {
      "epoch": 2.8733008350392484,
      "grad_norm": 1.4955627918243408,
      "learning_rate": 9.457182691111882e-06,
      "loss": 0.0631,
      "step": 195100
    },
    {
      "epoch": 2.8736690181293354,
      "grad_norm": 2.1704089641571045,
      "learning_rate": 9.455546319144898e-06,
      "loss": 0.0614,
      "step": 195125
    },
    {
      "epoch": 2.8740372012194224,
      "grad_norm": 0.9554424285888672,
      "learning_rate": 9.453909947177914e-06,
      "loss": 0.0629,
      "step": 195150
    },
    {
      "epoch": 2.8744053843095094,
      "grad_norm": 1.3388690948486328,
      "learning_rate": 9.45227357521093e-06,
      "loss": 0.059,
      "step": 195175
    },
    {
      "epoch": 2.8747735673995964,
      "grad_norm": 1.060782551765442,
      "learning_rate": 9.450637203243945e-06,
      "loss": 0.0617,
      "step": 195200
    },
    {
      "epoch": 2.8751417504896835,
      "grad_norm": 1.4692620038986206,
      "learning_rate": 9.44900083127696e-06,
      "loss": 0.0555,
      "step": 195225
    },
    {
      "epoch": 2.8755099335797705,
      "grad_norm": 1.1032123565673828,
      "learning_rate": 9.447364459309975e-06,
      "loss": 0.0612,
      "step": 195250
    },
    {
      "epoch": 2.8758781166698575,
      "grad_norm": 1.2734709978103638,
      "learning_rate": 9.445728087342992e-06,
      "loss": 0.0573,
      "step": 195275
    },
    {
      "epoch": 2.8762462997599445,
      "grad_norm": 1.4377365112304688,
      "learning_rate": 9.444091715376006e-06,
      "loss": 0.0535,
      "step": 195300
    },
    {
      "epoch": 2.8766144828500315,
      "grad_norm": 1.3288897275924683,
      "learning_rate": 9.442455343409022e-06,
      "loss": 0.0546,
      "step": 195325
    },
    {
      "epoch": 2.876982665940119,
      "grad_norm": 1.4181761741638184,
      "learning_rate": 9.440818971442037e-06,
      "loss": 0.0606,
      "step": 195350
    },
    {
      "epoch": 2.877350849030206,
      "grad_norm": 1.2221509218215942,
      "learning_rate": 9.439182599475053e-06,
      "loss": 0.0523,
      "step": 195375
    },
    {
      "epoch": 2.877719032120293,
      "grad_norm": 0.9942285418510437,
      "learning_rate": 9.437546227508067e-06,
      "loss": 0.0583,
      "step": 195400
    },
    {
      "epoch": 2.87808721521038,
      "grad_norm": 1.4243788719177246,
      "learning_rate": 9.435909855541083e-06,
      "loss": 0.0582,
      "step": 195425
    },
    {
      "epoch": 2.878455398300467,
      "grad_norm": 1.2714022397994995,
      "learning_rate": 9.4342734835741e-06,
      "loss": 0.0633,
      "step": 195450
    },
    {
      "epoch": 2.878823581390554,
      "grad_norm": 1.3662664890289307,
      "learning_rate": 9.432637111607114e-06,
      "loss": 0.0602,
      "step": 195475
    },
    {
      "epoch": 2.879191764480641,
      "grad_norm": 1.4568315744400024,
      "learning_rate": 9.431000739640128e-06,
      "loss": 0.0698,
      "step": 195500
    },
    {
      "epoch": 2.879559947570728,
      "grad_norm": 1.3788787126541138,
      "learning_rate": 9.429364367673145e-06,
      "loss": 0.0584,
      "step": 195525
    },
    {
      "epoch": 2.879928130660815,
      "grad_norm": 1.4465004205703735,
      "learning_rate": 9.42772799570616e-06,
      "loss": 0.0563,
      "step": 195550
    },
    {
      "epoch": 2.880296313750902,
      "grad_norm": 1.3699222803115845,
      "learning_rate": 9.426091623739177e-06,
      "loss": 0.0591,
      "step": 195575
    },
    {
      "epoch": 2.880664496840989,
      "grad_norm": 1.3015691041946411,
      "learning_rate": 9.424455251772191e-06,
      "loss": 0.06,
      "step": 195600
    },
    {
      "epoch": 2.881032679931076,
      "grad_norm": 1.9338303804397583,
      "learning_rate": 9.422818879805208e-06,
      "loss": 0.0577,
      "step": 195625
    },
    {
      "epoch": 2.881400863021163,
      "grad_norm": 1.7295316457748413,
      "learning_rate": 9.421182507838222e-06,
      "loss": 0.0693,
      "step": 195650
    },
    {
      "epoch": 2.88176904611125,
      "grad_norm": 1.2147678136825562,
      "learning_rate": 9.419546135871238e-06,
      "loss": 0.0608,
      "step": 195675
    },
    {
      "epoch": 2.882137229201337,
      "grad_norm": 0.9938950538635254,
      "learning_rate": 9.417909763904254e-06,
      "loss": 0.0562,
      "step": 195700
    },
    {
      "epoch": 2.882505412291424,
      "grad_norm": 1.5220997333526611,
      "learning_rate": 9.416273391937269e-06,
      "loss": 0.0556,
      "step": 195725
    },
    {
      "epoch": 2.8828735953815112,
      "grad_norm": 1.557158350944519,
      "learning_rate": 9.414637019970285e-06,
      "loss": 0.0599,
      "step": 195750
    },
    {
      "epoch": 2.8832417784715982,
      "grad_norm": 1.4891746044158936,
      "learning_rate": 9.4130006480033e-06,
      "loss": 0.0526,
      "step": 195775
    },
    {
      "epoch": 2.8836099615616853,
      "grad_norm": 1.334468126296997,
      "learning_rate": 9.411364276036315e-06,
      "loss": 0.0585,
      "step": 195800
    },
    {
      "epoch": 2.8839781446517723,
      "grad_norm": 0.9822672009468079,
      "learning_rate": 9.40972790406933e-06,
      "loss": 0.0563,
      "step": 195825
    },
    {
      "epoch": 2.8843463277418593,
      "grad_norm": 1.0776010751724243,
      "learning_rate": 9.408091532102346e-06,
      "loss": 0.053,
      "step": 195850
    },
    {
      "epoch": 2.8847145108319463,
      "grad_norm": 1.5215586423873901,
      "learning_rate": 9.406455160135362e-06,
      "loss": 0.059,
      "step": 195875
    },
    {
      "epoch": 2.8850826939220333,
      "grad_norm": 1.5357248783111572,
      "learning_rate": 9.404818788168377e-06,
      "loss": 0.0637,
      "step": 195900
    },
    {
      "epoch": 2.8854508770121203,
      "grad_norm": 1.2864866256713867,
      "learning_rate": 9.403182416201391e-06,
      "loss": 0.056,
      "step": 195925
    },
    {
      "epoch": 2.885819060102208,
      "grad_norm": 1.7680096626281738,
      "learning_rate": 9.401546044234407e-06,
      "loss": 0.0611,
      "step": 195950
    },
    {
      "epoch": 2.886187243192295,
      "grad_norm": 1.2400763034820557,
      "learning_rate": 9.399909672267423e-06,
      "loss": 0.0527,
      "step": 195975
    },
    {
      "epoch": 2.886555426282382,
      "grad_norm": 1.0655015707015991,
      "learning_rate": 9.39827330030044e-06,
      "loss": 0.0686,
      "step": 196000
    },
    {
      "epoch": 2.886923609372469,
      "grad_norm": 1.801745057106018,
      "learning_rate": 9.396636928333454e-06,
      "loss": 0.065,
      "step": 196025
    },
    {
      "epoch": 2.887291792462556,
      "grad_norm": 1.1378285884857178,
      "learning_rate": 9.395000556366469e-06,
      "loss": 0.0592,
      "step": 196050
    },
    {
      "epoch": 2.887659975552643,
      "grad_norm": 1.8716405630111694,
      "learning_rate": 9.393364184399485e-06,
      "loss": 0.0656,
      "step": 196075
    },
    {
      "epoch": 2.88802815864273,
      "grad_norm": 1.603352427482605,
      "learning_rate": 9.3917278124325e-06,
      "loss": 0.0562,
      "step": 196100
    },
    {
      "epoch": 2.888396341732817,
      "grad_norm": 1.1756269931793213,
      "learning_rate": 9.390091440465517e-06,
      "loss": 0.0566,
      "step": 196125
    },
    {
      "epoch": 2.888764524822904,
      "grad_norm": 1.252945899963379,
      "learning_rate": 9.388455068498531e-06,
      "loss": 0.0642,
      "step": 196150
    },
    {
      "epoch": 2.889132707912991,
      "grad_norm": 1.6491494178771973,
      "learning_rate": 9.386818696531546e-06,
      "loss": 0.061,
      "step": 196175
    },
    {
      "epoch": 2.889500891003078,
      "grad_norm": 1.4682763814926147,
      "learning_rate": 9.385182324564562e-06,
      "loss": 0.0612,
      "step": 196200
    },
    {
      "epoch": 2.889869074093165,
      "grad_norm": 1.8137708902359009,
      "learning_rate": 9.383545952597578e-06,
      "loss": 0.0563,
      "step": 196225
    },
    {
      "epoch": 2.890237257183252,
      "grad_norm": 1.2542399168014526,
      "learning_rate": 9.381909580630593e-06,
      "loss": 0.0631,
      "step": 196250
    },
    {
      "epoch": 2.890605440273339,
      "grad_norm": 1.7905925512313843,
      "learning_rate": 9.380273208663609e-06,
      "loss": 0.0564,
      "step": 196275
    },
    {
      "epoch": 2.890973623363426,
      "grad_norm": 1.4301177263259888,
      "learning_rate": 9.378636836696623e-06,
      "loss": 0.0633,
      "step": 196300
    },
    {
      "epoch": 2.8913418064535135,
      "grad_norm": 1.5495744943618774,
      "learning_rate": 9.37700046472964e-06,
      "loss": 0.0541,
      "step": 196325
    },
    {
      "epoch": 2.8917099895436005,
      "grad_norm": 1.355925440788269,
      "learning_rate": 9.375364092762654e-06,
      "loss": 0.0589,
      "step": 196350
    },
    {
      "epoch": 2.8920781726336875,
      "grad_norm": 1.479114055633545,
      "learning_rate": 9.37372772079567e-06,
      "loss": 0.0607,
      "step": 196375
    },
    {
      "epoch": 2.8924463557237745,
      "grad_norm": 1.381609320640564,
      "learning_rate": 9.372091348828686e-06,
      "loss": 0.0608,
      "step": 196400
    },
    {
      "epoch": 2.8928145388138615,
      "grad_norm": 0.8497568368911743,
      "learning_rate": 9.3704549768617e-06,
      "loss": 0.0556,
      "step": 196425
    },
    {
      "epoch": 2.8931827219039485,
      "grad_norm": 1.6425498723983765,
      "learning_rate": 9.368818604894717e-06,
      "loss": 0.0596,
      "step": 196450
    },
    {
      "epoch": 2.8935509049940356,
      "grad_norm": 1.2890487909317017,
      "learning_rate": 9.367182232927731e-06,
      "loss": 0.0525,
      "step": 196475
    },
    {
      "epoch": 2.8939190880841226,
      "grad_norm": 1.6077224016189575,
      "learning_rate": 9.365545860960747e-06,
      "loss": 0.0558,
      "step": 196500
    },
    {
      "epoch": 2.8942872711742096,
      "grad_norm": 1.4123090505599976,
      "learning_rate": 9.363909488993764e-06,
      "loss": 0.0582,
      "step": 196525
    },
    {
      "epoch": 2.8946554542642966,
      "grad_norm": 1.333767056465149,
      "learning_rate": 9.36227311702678e-06,
      "loss": 0.0609,
      "step": 196550
    },
    {
      "epoch": 2.8950236373543836,
      "grad_norm": 0.9862918257713318,
      "learning_rate": 9.360636745059794e-06,
      "loss": 0.0604,
      "step": 196575
    },
    {
      "epoch": 2.8953918204444706,
      "grad_norm": 1.1639968156814575,
      "learning_rate": 9.359000373092809e-06,
      "loss": 0.0567,
      "step": 196600
    },
    {
      "epoch": 2.8957600035345576,
      "grad_norm": 1.0329935550689697,
      "learning_rate": 9.357364001125825e-06,
      "loss": 0.0574,
      "step": 196625
    },
    {
      "epoch": 2.8961281866246447,
      "grad_norm": 1.156352162361145,
      "learning_rate": 9.355727629158841e-06,
      "loss": 0.0612,
      "step": 196650
    },
    {
      "epoch": 2.8964963697147317,
      "grad_norm": 1.3425453901290894,
      "learning_rate": 9.354091257191855e-06,
      "loss": 0.0593,
      "step": 196675
    },
    {
      "epoch": 2.8968645528048187,
      "grad_norm": 1.839534878730774,
      "learning_rate": 9.352454885224872e-06,
      "loss": 0.0631,
      "step": 196700
    },
    {
      "epoch": 2.8972327358949057,
      "grad_norm": 1.8900396823883057,
      "learning_rate": 9.350818513257886e-06,
      "loss": 0.056,
      "step": 196725
    },
    {
      "epoch": 2.8976009189849927,
      "grad_norm": 0.9381417036056519,
      "learning_rate": 9.349182141290902e-06,
      "loss": 0.052,
      "step": 196750
    },
    {
      "epoch": 2.8979691020750797,
      "grad_norm": 1.9540637731552124,
      "learning_rate": 9.347545769323917e-06,
      "loss": 0.0603,
      "step": 196775
    },
    {
      "epoch": 2.8983372851651668,
      "grad_norm": 1.345794677734375,
      "learning_rate": 9.345909397356933e-06,
      "loss": 0.056,
      "step": 196800
    },
    {
      "epoch": 2.8987054682552538,
      "grad_norm": 1.263490915298462,
      "learning_rate": 9.344273025389949e-06,
      "loss": 0.054,
      "step": 196825
    },
    {
      "epoch": 2.899073651345341,
      "grad_norm": 1.1459684371948242,
      "learning_rate": 9.342636653422963e-06,
      "loss": 0.0543,
      "step": 196850
    },
    {
      "epoch": 2.899441834435428,
      "grad_norm": 1.0339220762252808,
      "learning_rate": 9.341000281455978e-06,
      "loss": 0.0596,
      "step": 196875
    },
    {
      "epoch": 2.899810017525515,
      "grad_norm": 1.0365582704544067,
      "learning_rate": 9.339363909488994e-06,
      "loss": 0.0649,
      "step": 196900
    },
    {
      "epoch": 2.9001782006156023,
      "grad_norm": 1.3709920644760132,
      "learning_rate": 9.33772753752201e-06,
      "loss": 0.0697,
      "step": 196925
    },
    {
      "epoch": 2.9005463837056893,
      "grad_norm": 0.9024575352668762,
      "learning_rate": 9.336091165555026e-06,
      "loss": 0.0477,
      "step": 196950
    },
    {
      "epoch": 2.9009145667957763,
      "grad_norm": 1.5772453546524048,
      "learning_rate": 9.33445479358804e-06,
      "loss": 0.0617,
      "step": 196975
    },
    {
      "epoch": 2.9012827498858633,
      "grad_norm": 1.1659525632858276,
      "learning_rate": 9.332818421621055e-06,
      "loss": 0.056,
      "step": 197000
    },
    {
      "epoch": 2.9016509329759503,
      "grad_norm": 1.7544668912887573,
      "learning_rate": 9.331182049654071e-06,
      "loss": 0.0618,
      "step": 197025
    },
    {
      "epoch": 2.9020191160660374,
      "grad_norm": 1.6566730737686157,
      "learning_rate": 9.329545677687087e-06,
      "loss": 0.058,
      "step": 197050
    },
    {
      "epoch": 2.9023872991561244,
      "grad_norm": 1.152753472328186,
      "learning_rate": 9.327909305720104e-06,
      "loss": 0.063,
      "step": 197075
    },
    {
      "epoch": 2.9027554822462114,
      "grad_norm": 1.2455028295516968,
      "learning_rate": 9.326272933753118e-06,
      "loss": 0.0614,
      "step": 197100
    },
    {
      "epoch": 2.9031236653362984,
      "grad_norm": 1.3942725658416748,
      "learning_rate": 9.324636561786134e-06,
      "loss": 0.055,
      "step": 197125
    },
    {
      "epoch": 2.9034918484263854,
      "grad_norm": 1.179827094078064,
      "learning_rate": 9.323000189819149e-06,
      "loss": 0.0592,
      "step": 197150
    },
    {
      "epoch": 2.9038600315164724,
      "grad_norm": 1.185843825340271,
      "learning_rate": 9.321363817852165e-06,
      "loss": 0.0595,
      "step": 197175
    },
    {
      "epoch": 2.9042282146065594,
      "grad_norm": 1.4101228713989258,
      "learning_rate": 9.31972744588518e-06,
      "loss": 0.0604,
      "step": 197200
    },
    {
      "epoch": 2.9045963976966465,
      "grad_norm": 1.3172492980957031,
      "learning_rate": 9.318091073918195e-06,
      "loss": 0.0562,
      "step": 197225
    },
    {
      "epoch": 2.9049645807867335,
      "grad_norm": 1.3225313425064087,
      "learning_rate": 9.316454701951212e-06,
      "loss": 0.0604,
      "step": 197250
    },
    {
      "epoch": 2.905332763876821,
      "grad_norm": 1.6840494871139526,
      "learning_rate": 9.314818329984226e-06,
      "loss": 0.0543,
      "step": 197275
    },
    {
      "epoch": 2.905700946966908,
      "grad_norm": 1.3955174684524536,
      "learning_rate": 9.31318195801724e-06,
      "loss": 0.0597,
      "step": 197300
    },
    {
      "epoch": 2.906069130056995,
      "grad_norm": 1.1126434803009033,
      "learning_rate": 9.311545586050257e-06,
      "loss": 0.0571,
      "step": 197325
    },
    {
      "epoch": 2.906437313147082,
      "grad_norm": 1.7837954759597778,
      "learning_rate": 9.309909214083273e-06,
      "loss": 0.0552,
      "step": 197350
    },
    {
      "epoch": 2.906805496237169,
      "grad_norm": 1.4580146074295044,
      "learning_rate": 9.308272842116289e-06,
      "loss": 0.0573,
      "step": 197375
    },
    {
      "epoch": 2.907173679327256,
      "grad_norm": 1.662578821182251,
      "learning_rate": 9.306636470149303e-06,
      "loss": 0.0571,
      "step": 197400
    },
    {
      "epoch": 2.907541862417343,
      "grad_norm": 1.2120906114578247,
      "learning_rate": 9.305000098182318e-06,
      "loss": 0.0608,
      "step": 197425
    },
    {
      "epoch": 2.90791004550743,
      "grad_norm": 1.644189715385437,
      "learning_rate": 9.303363726215334e-06,
      "loss": 0.0601,
      "step": 197450
    },
    {
      "epoch": 2.908278228597517,
      "grad_norm": 1.4400080442428589,
      "learning_rate": 9.30172735424835e-06,
      "loss": 0.0661,
      "step": 197475
    },
    {
      "epoch": 2.908646411687604,
      "grad_norm": 1.6892335414886475,
      "learning_rate": 9.300090982281366e-06,
      "loss": 0.0634,
      "step": 197500
    },
    {
      "epoch": 2.909014594777691,
      "grad_norm": 1.0236717462539673,
      "learning_rate": 9.29845461031438e-06,
      "loss": 0.0546,
      "step": 197525
    },
    {
      "epoch": 2.909382777867778,
      "grad_norm": 1.3156410455703735,
      "learning_rate": 9.296818238347395e-06,
      "loss": 0.0505,
      "step": 197550
    },
    {
      "epoch": 2.909750960957865,
      "grad_norm": 0.8439908027648926,
      "learning_rate": 9.295181866380411e-06,
      "loss": 0.0625,
      "step": 197575
    },
    {
      "epoch": 2.910119144047952,
      "grad_norm": 1.2165777683258057,
      "learning_rate": 9.293545494413428e-06,
      "loss": 0.052,
      "step": 197600
    },
    {
      "epoch": 2.910487327138039,
      "grad_norm": 1.3894983530044556,
      "learning_rate": 9.291909122446442e-06,
      "loss": 0.0577,
      "step": 197625
    },
    {
      "epoch": 2.910855510228126,
      "grad_norm": 1.21957528591156,
      "learning_rate": 9.290338205358138e-06,
      "loss": 0.0574,
      "step": 197650
    },
    {
      "epoch": 2.911223693318213,
      "grad_norm": 1.7782771587371826,
      "learning_rate": 9.288701833391152e-06,
      "loss": 0.06,
      "step": 197675
    },
    {
      "epoch": 2.9115918764083,
      "grad_norm": 1.6327035427093506,
      "learning_rate": 9.287065461424168e-06,
      "loss": 0.0569,
      "step": 197700
    },
    {
      "epoch": 2.911960059498387,
      "grad_norm": 1.442806363105774,
      "learning_rate": 9.285429089457183e-06,
      "loss": 0.0612,
      "step": 197725
    },
    {
      "epoch": 2.9123282425884742,
      "grad_norm": 1.6058634519577026,
      "learning_rate": 9.283792717490199e-06,
      "loss": 0.0602,
      "step": 197750
    },
    {
      "epoch": 2.9126964256785612,
      "grad_norm": 1.7758101224899292,
      "learning_rate": 9.282156345523215e-06,
      "loss": 0.0549,
      "step": 197775
    },
    {
      "epoch": 2.9130646087686483,
      "grad_norm": 1.0256634950637817,
      "learning_rate": 9.28051997355623e-06,
      "loss": 0.054,
      "step": 197800
    },
    {
      "epoch": 2.9134327918587353,
      "grad_norm": 1.1333004236221313,
      "learning_rate": 9.278883601589244e-06,
      "loss": 0.0582,
      "step": 197825
    },
    {
      "epoch": 2.9138009749488223,
      "grad_norm": 1.649069905281067,
      "learning_rate": 9.27724722962226e-06,
      "loss": 0.0603,
      "step": 197850
    },
    {
      "epoch": 2.9141691580389097,
      "grad_norm": 1.7726221084594727,
      "learning_rate": 9.275610857655276e-06,
      "loss": 0.0603,
      "step": 197875
    },
    {
      "epoch": 2.9145373411289968,
      "grad_norm": 0.7078713178634644,
      "learning_rate": 9.273974485688292e-06,
      "loss": 0.058,
      "step": 197900
    },
    {
      "epoch": 2.9149055242190838,
      "grad_norm": 1.1463229656219482,
      "learning_rate": 9.272338113721307e-06,
      "loss": 0.0632,
      "step": 197925
    },
    {
      "epoch": 2.915273707309171,
      "grad_norm": 1.6430611610412598,
      "learning_rate": 9.270701741754323e-06,
      "loss": 0.0608,
      "step": 197950
    },
    {
      "epoch": 2.915641890399258,
      "grad_norm": 1.5461478233337402,
      "learning_rate": 9.269065369787337e-06,
      "loss": 0.0612,
      "step": 197975
    },
    {
      "epoch": 2.916010073489345,
      "grad_norm": 1.0525239706039429,
      "learning_rate": 9.267428997820354e-06,
      "loss": 0.055,
      "step": 198000
    },
    {
      "epoch": 2.916378256579432,
      "grad_norm": 1.4291107654571533,
      "learning_rate": 9.26579262585337e-06,
      "loss": 0.054,
      "step": 198025
    },
    {
      "epoch": 2.916746439669519,
      "grad_norm": 0.988843560218811,
      "learning_rate": 9.264156253886384e-06,
      "loss": 0.0608,
      "step": 198050
    },
    {
      "epoch": 2.917114622759606,
      "grad_norm": 1.3206547498703003,
      "learning_rate": 9.2625198819194e-06,
      "loss": 0.0537,
      "step": 198075
    },
    {
      "epoch": 2.917482805849693,
      "grad_norm": 1.5068535804748535,
      "learning_rate": 9.260883509952415e-06,
      "loss": 0.0538,
      "step": 198100
    },
    {
      "epoch": 2.91785098893978,
      "grad_norm": 1.1686986684799194,
      "learning_rate": 9.259247137985431e-06,
      "loss": 0.063,
      "step": 198125
    },
    {
      "epoch": 2.918219172029867,
      "grad_norm": 1.5856209993362427,
      "learning_rate": 9.257610766018445e-06,
      "loss": 0.0515,
      "step": 198150
    },
    {
      "epoch": 2.918587355119954,
      "grad_norm": 1.4969663619995117,
      "learning_rate": 9.255974394051462e-06,
      "loss": 0.0587,
      "step": 198175
    },
    {
      "epoch": 2.918955538210041,
      "grad_norm": 1.4149047136306763,
      "learning_rate": 9.254338022084478e-06,
      "loss": 0.0609,
      "step": 198200
    },
    {
      "epoch": 2.9193237213001284,
      "grad_norm": 1.1060090065002441,
      "learning_rate": 9.252701650117492e-06,
      "loss": 0.0565,
      "step": 198225
    },
    {
      "epoch": 2.9196919043902154,
      "grad_norm": 1.0932490825653076,
      "learning_rate": 9.251065278150507e-06,
      "loss": 0.0558,
      "step": 198250
    },
    {
      "epoch": 2.9200600874803024,
      "grad_norm": 1.1115100383758545,
      "learning_rate": 9.249428906183523e-06,
      "loss": 0.0587,
      "step": 198275
    },
    {
      "epoch": 2.9204282705703895,
      "grad_norm": 1.6504942178726196,
      "learning_rate": 9.247792534216539e-06,
      "loss": 0.0586,
      "step": 198300
    },
    {
      "epoch": 2.9207964536604765,
      "grad_norm": 1.6405014991760254,
      "learning_rate": 9.246156162249555e-06,
      "loss": 0.0609,
      "step": 198325
    },
    {
      "epoch": 2.9211646367505635,
      "grad_norm": 1.4614890813827515,
      "learning_rate": 9.24451979028257e-06,
      "loss": 0.0601,
      "step": 198350
    },
    {
      "epoch": 2.9215328198406505,
      "grad_norm": 1.1469447612762451,
      "learning_rate": 9.242883418315584e-06,
      "loss": 0.0569,
      "step": 198375
    },
    {
      "epoch": 2.9219010029307375,
      "grad_norm": 1.199053406715393,
      "learning_rate": 9.2412470463486e-06,
      "loss": 0.057,
      "step": 198400
    },
    {
      "epoch": 2.9222691860208245,
      "grad_norm": 1.3672250509262085,
      "learning_rate": 9.239610674381616e-06,
      "loss": 0.0545,
      "step": 198425
    },
    {
      "epoch": 2.9226373691109115,
      "grad_norm": 1.0338282585144043,
      "learning_rate": 9.237974302414632e-06,
      "loss": 0.0598,
      "step": 198450
    },
    {
      "epoch": 2.9230055522009986,
      "grad_norm": 1.2343991994857788,
      "learning_rate": 9.236337930447647e-06,
      "loss": 0.058,
      "step": 198475
    },
    {
      "epoch": 2.9233737352910856,
      "grad_norm": 1.6773756742477417,
      "learning_rate": 9.234701558480661e-06,
      "loss": 0.0596,
      "step": 198500
    },
    {
      "epoch": 2.9237419183811726,
      "grad_norm": 1.4384047985076904,
      "learning_rate": 9.233065186513677e-06,
      "loss": 0.061,
      "step": 198525
    },
    {
      "epoch": 2.9241101014712596,
      "grad_norm": 1.8510615825653076,
      "learning_rate": 9.231428814546694e-06,
      "loss": 0.0583,
      "step": 198550
    },
    {
      "epoch": 2.9244782845613466,
      "grad_norm": 1.3557552099227905,
      "learning_rate": 9.229792442579708e-06,
      "loss": 0.0562,
      "step": 198575
    },
    {
      "epoch": 2.9248464676514336,
      "grad_norm": 1.180547833442688,
      "learning_rate": 9.228156070612724e-06,
      "loss": 0.0526,
      "step": 198600
    },
    {
      "epoch": 2.9252146507415207,
      "grad_norm": 1.7722705602645874,
      "learning_rate": 9.226519698645739e-06,
      "loss": 0.0636,
      "step": 198625
    },
    {
      "epoch": 2.9255828338316077,
      "grad_norm": 0.9078608751296997,
      "learning_rate": 9.224883326678755e-06,
      "loss": 0.0607,
      "step": 198650
    },
    {
      "epoch": 2.9259510169216947,
      "grad_norm": 1.3928896188735962,
      "learning_rate": 9.22324695471177e-06,
      "loss": 0.0588,
      "step": 198675
    },
    {
      "epoch": 2.9263192000117817,
      "grad_norm": 1.5424017906188965,
      "learning_rate": 9.221610582744785e-06,
      "loss": 0.0616,
      "step": 198700
    },
    {
      "epoch": 2.9266873831018687,
      "grad_norm": 1.3789658546447754,
      "learning_rate": 9.219974210777802e-06,
      "loss": 0.0584,
      "step": 198725
    },
    {
      "epoch": 2.9270555661919557,
      "grad_norm": 0.9868493676185608,
      "learning_rate": 9.218337838810816e-06,
      "loss": 0.0649,
      "step": 198750
    },
    {
      "epoch": 2.9274237492820427,
      "grad_norm": 0.9839689135551453,
      "learning_rate": 9.216701466843832e-06,
      "loss": 0.0616,
      "step": 198775
    },
    {
      "epoch": 2.9277919323721298,
      "grad_norm": 1.1599619388580322,
      "learning_rate": 9.215065094876847e-06,
      "loss": 0.0574,
      "step": 198800
    },
    {
      "epoch": 2.928160115462217,
      "grad_norm": 1.0458452701568604,
      "learning_rate": 9.213428722909863e-06,
      "loss": 0.0501,
      "step": 198825
    },
    {
      "epoch": 2.9285282985523042,
      "grad_norm": 1.0514401197433472,
      "learning_rate": 9.211792350942879e-06,
      "loss": 0.0658,
      "step": 198850
    },
    {
      "epoch": 2.9288964816423912,
      "grad_norm": 1.8742988109588623,
      "learning_rate": 9.210155978975893e-06,
      "loss": 0.0627,
      "step": 198875
    },
    {
      "epoch": 2.9292646647324783,
      "grad_norm": 1.4387950897216797,
      "learning_rate": 9.20851960700891e-06,
      "loss": 0.0605,
      "step": 198900
    },
    {
      "epoch": 2.9296328478225653,
      "grad_norm": 0.8504129648208618,
      "learning_rate": 9.206883235041924e-06,
      "loss": 0.0612,
      "step": 198925
    },
    {
      "epoch": 2.9300010309126523,
      "grad_norm": 0.9518362283706665,
      "learning_rate": 9.20524686307494e-06,
      "loss": 0.0659,
      "step": 198950
    },
    {
      "epoch": 2.9303692140027393,
      "grad_norm": 1.277635931968689,
      "learning_rate": 9.203610491107956e-06,
      "loss": 0.0616,
      "step": 198975
    },
    {
      "epoch": 2.9307373970928263,
      "grad_norm": 1.2637767791748047,
      "learning_rate": 9.20197411914097e-06,
      "loss": 0.0636,
      "step": 199000
    },
    {
      "epoch": 2.9311055801829133,
      "grad_norm": 1.6286725997924805,
      "learning_rate": 9.200337747173987e-06,
      "loss": 0.0535,
      "step": 199025
    },
    {
      "epoch": 2.9314737632730004,
      "grad_norm": 1.6246912479400635,
      "learning_rate": 9.198701375207001e-06,
      "loss": 0.0609,
      "step": 199050
    },
    {
      "epoch": 2.9318419463630874,
      "grad_norm": 1.269919514656067,
      "learning_rate": 9.197065003240018e-06,
      "loss": 0.0646,
      "step": 199075
    },
    {
      "epoch": 2.9322101294531744,
      "grad_norm": 1.518126368522644,
      "learning_rate": 9.195428631273032e-06,
      "loss": 0.0554,
      "step": 199100
    },
    {
      "epoch": 2.9325783125432614,
      "grad_norm": 1.5499554872512817,
      "learning_rate": 9.193792259306048e-06,
      "loss": 0.064,
      "step": 199125
    },
    {
      "epoch": 2.9329464956333484,
      "grad_norm": 1.7440624237060547,
      "learning_rate": 9.192155887339064e-06,
      "loss": 0.0578,
      "step": 199150
    },
    {
      "epoch": 2.9333146787234354,
      "grad_norm": 1.2438544034957886,
      "learning_rate": 9.190519515372079e-06,
      "loss": 0.0614,
      "step": 199175
    },
    {
      "epoch": 2.933682861813523,
      "grad_norm": 1.824089527130127,
      "learning_rate": 9.188883143405093e-06,
      "loss": 0.0548,
      "step": 199200
    },
    {
      "epoch": 2.93405104490361,
      "grad_norm": 1.448590636253357,
      "learning_rate": 9.18724677143811e-06,
      "loss": 0.0616,
      "step": 199225
    },
    {
      "epoch": 2.934419227993697,
      "grad_norm": 1.3385372161865234,
      "learning_rate": 9.185610399471126e-06,
      "loss": 0.0615,
      "step": 199250
    },
    {
      "epoch": 2.934787411083784,
      "grad_norm": 1.1906473636627197,
      "learning_rate": 9.183974027504142e-06,
      "loss": 0.0554,
      "step": 199275
    },
    {
      "epoch": 2.935155594173871,
      "grad_norm": 1.8147873878479004,
      "learning_rate": 9.182337655537156e-06,
      "loss": 0.0562,
      "step": 199300
    },
    {
      "epoch": 2.935523777263958,
      "grad_norm": 1.5189435482025146,
      "learning_rate": 9.18070128357017e-06,
      "loss": 0.0559,
      "step": 199325
    },
    {
      "epoch": 2.935891960354045,
      "grad_norm": 1.525354027748108,
      "learning_rate": 9.179064911603187e-06,
      "loss": 0.0564,
      "step": 199350
    },
    {
      "epoch": 2.936260143444132,
      "grad_norm": 1.9442694187164307,
      "learning_rate": 9.177428539636203e-06,
      "loss": 0.0565,
      "step": 199375
    },
    {
      "epoch": 2.936628326534219,
      "grad_norm": 1.1037707328796387,
      "learning_rate": 9.175792167669217e-06,
      "loss": 0.0606,
      "step": 199400
    },
    {
      "epoch": 2.936996509624306,
      "grad_norm": 1.0446170568466187,
      "learning_rate": 9.174155795702234e-06,
      "loss": 0.0557,
      "step": 199425
    },
    {
      "epoch": 2.937364692714393,
      "grad_norm": 1.5533865690231323,
      "learning_rate": 9.172519423735248e-06,
      "loss": 0.0559,
      "step": 199450
    },
    {
      "epoch": 2.93773287580448,
      "grad_norm": 1.2778167724609375,
      "learning_rate": 9.170883051768264e-06,
      "loss": 0.0583,
      "step": 199475
    },
    {
      "epoch": 2.938101058894567,
      "grad_norm": 1.3502825498580933,
      "learning_rate": 9.16924667980128e-06,
      "loss": 0.0588,
      "step": 199500
    },
    {
      "epoch": 2.938469241984654,
      "grad_norm": 1.5013251304626465,
      "learning_rate": 9.167610307834295e-06,
      "loss": 0.0586,
      "step": 199525
    },
    {
      "epoch": 2.938837425074741,
      "grad_norm": 1.59925377368927,
      "learning_rate": 9.165973935867311e-06,
      "loss": 0.0609,
      "step": 199550
    },
    {
      "epoch": 2.939205608164828,
      "grad_norm": 1.480793833732605,
      "learning_rate": 9.164337563900327e-06,
      "loss": 0.0593,
      "step": 199575
    },
    {
      "epoch": 2.939573791254915,
      "grad_norm": 1.4268132448196411,
      "learning_rate": 9.162701191933342e-06,
      "loss": 0.0571,
      "step": 199600
    },
    {
      "epoch": 2.939941974345002,
      "grad_norm": 1.4799308776855469,
      "learning_rate": 9.161064819966356e-06,
      "loss": 0.0535,
      "step": 199625
    },
    {
      "epoch": 2.940310157435089,
      "grad_norm": 1.0943411588668823,
      "learning_rate": 9.159428447999372e-06,
      "loss": 0.0608,
      "step": 199650
    },
    {
      "epoch": 2.940678340525176,
      "grad_norm": 1.4410858154296875,
      "learning_rate": 9.157792076032388e-06,
      "loss": 0.0656,
      "step": 199675
    },
    {
      "epoch": 2.941046523615263,
      "grad_norm": 1.3466460704803467,
      "learning_rate": 9.156155704065404e-06,
      "loss": 0.06,
      "step": 199700
    },
    {
      "epoch": 2.94141470670535,
      "grad_norm": 0.9451752305030823,
      "learning_rate": 9.154519332098419e-06,
      "loss": 0.0545,
      "step": 199725
    },
    {
      "epoch": 2.9417828897954372,
      "grad_norm": 0.7051096558570862,
      "learning_rate": 9.152882960131433e-06,
      "loss": 0.0615,
      "step": 199750
    },
    {
      "epoch": 2.9421510728855242,
      "grad_norm": 1.1519935131072998,
      "learning_rate": 9.15124658816445e-06,
      "loss": 0.0599,
      "step": 199775
    },
    {
      "epoch": 2.9425192559756117,
      "grad_norm": 1.8065868616104126,
      "learning_rate": 9.149610216197466e-06,
      "loss": 0.064,
      "step": 199800
    },
    {
      "epoch": 2.9428874390656987,
      "grad_norm": 1.9942238330841064,
      "learning_rate": 9.14797384423048e-06,
      "loss": 0.0599,
      "step": 199825
    },
    {
      "epoch": 2.9432556221557857,
      "grad_norm": 1.9204283952713013,
      "learning_rate": 9.146337472263496e-06,
      "loss": 0.0658,
      "step": 199850
    },
    {
      "epoch": 2.9436238052458727,
      "grad_norm": 1.3510510921478271,
      "learning_rate": 9.14470110029651e-06,
      "loss": 0.0569,
      "step": 199875
    },
    {
      "epoch": 2.9439919883359598,
      "grad_norm": 1.260505199432373,
      "learning_rate": 9.143064728329527e-06,
      "loss": 0.0515,
      "step": 199900
    },
    {
      "epoch": 2.944360171426047,
      "grad_norm": 1.3015497922897339,
      "learning_rate": 9.141428356362543e-06,
      "loss": 0.0483,
      "step": 199925
    },
    {
      "epoch": 2.944728354516134,
      "grad_norm": 1.0609599351882935,
      "learning_rate": 9.139791984395557e-06,
      "loss": 0.0621,
      "step": 199950
    },
    {
      "epoch": 2.945096537606221,
      "grad_norm": 1.3109073638916016,
      "learning_rate": 9.138155612428574e-06,
      "loss": 0.0612,
      "step": 199975
    },
    {
      "epoch": 2.945464720696308,
      "grad_norm": 1.788352608680725,
      "learning_rate": 9.136519240461588e-06,
      "loss": 0.0572,
      "step": 200000
    },
    {
      "epoch": 2.945832903786395,
      "grad_norm": 1.0755196809768677,
      "learning_rate": 9.134882868494604e-06,
      "loss": 0.0545,
      "step": 200025
    },
    {
      "epoch": 2.946201086876482,
      "grad_norm": 1.3027923107147217,
      "learning_rate": 9.133246496527619e-06,
      "loss": 0.0629,
      "step": 200050
    },
    {
      "epoch": 2.946569269966569,
      "grad_norm": 1.7882213592529297,
      "learning_rate": 9.131610124560635e-06,
      "loss": 0.0601,
      "step": 200075
    },
    {
      "epoch": 2.946937453056656,
      "grad_norm": 1.8756657838821411,
      "learning_rate": 9.129973752593651e-06,
      "loss": 0.0638,
      "step": 200100
    },
    {
      "epoch": 2.947305636146743,
      "grad_norm": 1.380398154258728,
      "learning_rate": 9.128337380626665e-06,
      "loss": 0.0557,
      "step": 200125
    },
    {
      "epoch": 2.9476738192368304,
      "grad_norm": 1.7494428157806396,
      "learning_rate": 9.126701008659682e-06,
      "loss": 0.0545,
      "step": 200150
    },
    {
      "epoch": 2.9480420023269174,
      "grad_norm": 1.886902928352356,
      "learning_rate": 9.125064636692696e-06,
      "loss": 0.0556,
      "step": 200175
    },
    {
      "epoch": 2.9484101854170044,
      "grad_norm": 0.9188262224197388,
      "learning_rate": 9.123428264725712e-06,
      "loss": 0.0586,
      "step": 200200
    },
    {
      "epoch": 2.9487783685070914,
      "grad_norm": 1.4986294507980347,
      "learning_rate": 9.121791892758728e-06,
      "loss": 0.0536,
      "step": 200225
    },
    {
      "epoch": 2.9491465515971784,
      "grad_norm": 1.2192198038101196,
      "learning_rate": 9.120155520791743e-06,
      "loss": 0.0552,
      "step": 200250
    },
    {
      "epoch": 2.9495147346872654,
      "grad_norm": 1.0087519884109497,
      "learning_rate": 9.118519148824759e-06,
      "loss": 0.05,
      "step": 200275
    },
    {
      "epoch": 2.9498829177773525,
      "grad_norm": 1.7876970767974854,
      "learning_rate": 9.116882776857773e-06,
      "loss": 0.0607,
      "step": 200300
    },
    {
      "epoch": 2.9502511008674395,
      "grad_norm": 0.9782829880714417,
      "learning_rate": 9.11524640489079e-06,
      "loss": 0.053,
      "step": 200325
    },
    {
      "epoch": 2.9506192839575265,
      "grad_norm": 1.240944743156433,
      "learning_rate": 9.113610032923804e-06,
      "loss": 0.0592,
      "step": 200350
    },
    {
      "epoch": 2.9509874670476135,
      "grad_norm": 1.4407756328582764,
      "learning_rate": 9.11197366095682e-06,
      "loss": 0.0614,
      "step": 200375
    },
    {
      "epoch": 2.9513556501377005,
      "grad_norm": 2.1824145317077637,
      "learning_rate": 9.110337288989836e-06,
      "loss": 0.0605,
      "step": 200400
    },
    {
      "epoch": 2.9517238332277875,
      "grad_norm": 1.4299499988555908,
      "learning_rate": 9.10870091702285e-06,
      "loss": 0.0575,
      "step": 200425
    },
    {
      "epoch": 2.9520920163178745,
      "grad_norm": 1.2828664779663086,
      "learning_rate": 9.107064545055867e-06,
      "loss": 0.0557,
      "step": 200450
    },
    {
      "epoch": 2.9524601994079616,
      "grad_norm": 1.1413097381591797,
      "learning_rate": 9.105428173088881e-06,
      "loss": 0.0514,
      "step": 200475
    },
    {
      "epoch": 2.9528283824980486,
      "grad_norm": 2.0417468547821045,
      "learning_rate": 9.103791801121898e-06,
      "loss": 0.0568,
      "step": 200500
    },
    {
      "epoch": 2.9531965655881356,
      "grad_norm": 1.0465189218521118,
      "learning_rate": 9.102155429154914e-06,
      "loss": 0.0556,
      "step": 200525
    },
    {
      "epoch": 2.9535647486782226,
      "grad_norm": 1.49721360206604,
      "learning_rate": 9.100519057187928e-06,
      "loss": 0.0662,
      "step": 200550
    },
    {
      "epoch": 2.9539329317683096,
      "grad_norm": 1.5440685749053955,
      "learning_rate": 9.098882685220943e-06,
      "loss": 0.0557,
      "step": 200575
    },
    {
      "epoch": 2.9543011148583966,
      "grad_norm": 1.5729954242706299,
      "learning_rate": 9.097246313253959e-06,
      "loss": 0.0699,
      "step": 200600
    },
    {
      "epoch": 2.9546692979484837,
      "grad_norm": 1.790576457977295,
      "learning_rate": 9.095609941286975e-06,
      "loss": 0.062,
      "step": 200625
    },
    {
      "epoch": 2.9550374810385707,
      "grad_norm": 0.7784260511398315,
      "learning_rate": 9.093973569319991e-06,
      "loss": 0.0564,
      "step": 200650
    },
    {
      "epoch": 2.9554056641286577,
      "grad_norm": 1.3388988971710205,
      "learning_rate": 9.092337197353006e-06,
      "loss": 0.0562,
      "step": 200675
    },
    {
      "epoch": 2.9557738472187447,
      "grad_norm": 1.1975830793380737,
      "learning_rate": 9.09070082538602e-06,
      "loss": 0.0563,
      "step": 200700
    },
    {
      "epoch": 2.9561420303088317,
      "grad_norm": 1.378483772277832,
      "learning_rate": 9.089064453419036e-06,
      "loss": 0.0588,
      "step": 200725
    },
    {
      "epoch": 2.956510213398919,
      "grad_norm": 1.3845373392105103,
      "learning_rate": 9.087428081452052e-06,
      "loss": 0.0582,
      "step": 200750
    },
    {
      "epoch": 2.956878396489006,
      "grad_norm": 1.903778314590454,
      "learning_rate": 9.085791709485067e-06,
      "loss": 0.0548,
      "step": 200775
    },
    {
      "epoch": 2.957246579579093,
      "grad_norm": 1.680147647857666,
      "learning_rate": 9.084155337518083e-06,
      "loss": 0.0606,
      "step": 200800
    },
    {
      "epoch": 2.95761476266918,
      "grad_norm": 1.376102328300476,
      "learning_rate": 9.082518965551097e-06,
      "loss": 0.0615,
      "step": 200825
    },
    {
      "epoch": 2.9579829457592672,
      "grad_norm": 1.7232086658477783,
      "learning_rate": 9.080882593584114e-06,
      "loss": 0.0573,
      "step": 200850
    },
    {
      "epoch": 2.9583511288493543,
      "grad_norm": 1.1853758096694946,
      "learning_rate": 9.079246221617128e-06,
      "loss": 0.0613,
      "step": 200875
    },
    {
      "epoch": 2.9587193119394413,
      "grad_norm": 1.0940239429473877,
      "learning_rate": 9.077609849650144e-06,
      "loss": 0.0551,
      "step": 200900
    },
    {
      "epoch": 2.9590874950295283,
      "grad_norm": 1.3854516744613647,
      "learning_rate": 9.07597347768316e-06,
      "loss": 0.0622,
      "step": 200925
    },
    {
      "epoch": 2.9594556781196153,
      "grad_norm": 0.9698584675788879,
      "learning_rate": 9.074337105716175e-06,
      "loss": 0.054,
      "step": 200950
    },
    {
      "epoch": 2.9598238612097023,
      "grad_norm": 1.4902364015579224,
      "learning_rate": 9.072700733749191e-06,
      "loss": 0.0583,
      "step": 200975
    },
    {
      "epoch": 2.9601920442997893,
      "grad_norm": 1.326014518737793,
      "learning_rate": 9.071064361782205e-06,
      "loss": 0.0629,
      "step": 201000
    },
    {
      "epoch": 2.9605602273898763,
      "grad_norm": 1.4434059858322144,
      "learning_rate": 9.069427989815221e-06,
      "loss": 0.057,
      "step": 201025
    },
    {
      "epoch": 2.9609284104799634,
      "grad_norm": 1.6042250394821167,
      "learning_rate": 9.067791617848238e-06,
      "loss": 0.0554,
      "step": 201050
    },
    {
      "epoch": 2.9612965935700504,
      "grad_norm": 1.7414075136184692,
      "learning_rate": 9.066155245881254e-06,
      "loss": 0.0574,
      "step": 201075
    },
    {
      "epoch": 2.9616647766601374,
      "grad_norm": 1.4316617250442505,
      "learning_rate": 9.064518873914268e-06,
      "loss": 0.0581,
      "step": 201100
    },
    {
      "epoch": 2.962032959750225,
      "grad_norm": 1.2498750686645508,
      "learning_rate": 9.062882501947283e-06,
      "loss": 0.0556,
      "step": 201125
    },
    {
      "epoch": 2.962401142840312,
      "grad_norm": 1.5134295225143433,
      "learning_rate": 9.061246129980299e-06,
      "loss": 0.0615,
      "step": 201150
    },
    {
      "epoch": 2.962769325930399,
      "grad_norm": 1.3428409099578857,
      "learning_rate": 9.059609758013315e-06,
      "loss": 0.0595,
      "step": 201175
    },
    {
      "epoch": 2.963137509020486,
      "grad_norm": 1.099374771118164,
      "learning_rate": 9.05797338604633e-06,
      "loss": 0.062,
      "step": 201200
    },
    {
      "epoch": 2.963505692110573,
      "grad_norm": 1.5205073356628418,
      "learning_rate": 9.056337014079346e-06,
      "loss": 0.0608,
      "step": 201225
    },
    {
      "epoch": 2.96387387520066,
      "grad_norm": 0.7118973731994629,
      "learning_rate": 9.05470064211236e-06,
      "loss": 0.0672,
      "step": 201250
    },
    {
      "epoch": 2.964242058290747,
      "grad_norm": 1.3384051322937012,
      "learning_rate": 9.053064270145376e-06,
      "loss": 0.0537,
      "step": 201275
    },
    {
      "epoch": 2.964610241380834,
      "grad_norm": 1.144583821296692,
      "learning_rate": 9.05142789817839e-06,
      "loss": 0.064,
      "step": 201300
    },
    {
      "epoch": 2.964978424470921,
      "grad_norm": 1.328416347503662,
      "learning_rate": 9.049791526211407e-06,
      "loss": 0.056,
      "step": 201325
    },
    {
      "epoch": 2.965346607561008,
      "grad_norm": 1.483860969543457,
      "learning_rate": 9.048155154244423e-06,
      "loss": 0.0551,
      "step": 201350
    },
    {
      "epoch": 2.965714790651095,
      "grad_norm": 1.4993629455566406,
      "learning_rate": 9.046518782277437e-06,
      "loss": 0.0649,
      "step": 201375
    },
    {
      "epoch": 2.966082973741182,
      "grad_norm": 1.240976095199585,
      "learning_rate": 9.044882410310454e-06,
      "loss": 0.0623,
      "step": 201400
    },
    {
      "epoch": 2.966451156831269,
      "grad_norm": 0.7770968079566956,
      "learning_rate": 9.043246038343468e-06,
      "loss": 0.0542,
      "step": 201425
    },
    {
      "epoch": 2.966819339921356,
      "grad_norm": 0.8711233735084534,
      "learning_rate": 9.041609666376484e-06,
      "loss": 0.0562,
      "step": 201450
    },
    {
      "epoch": 2.967187523011443,
      "grad_norm": 1.3974469900131226,
      "learning_rate": 9.0399732944095e-06,
      "loss": 0.0617,
      "step": 201475
    },
    {
      "epoch": 2.96755570610153,
      "grad_norm": 1.3322957754135132,
      "learning_rate": 9.038336922442515e-06,
      "loss": 0.0611,
      "step": 201500
    },
    {
      "epoch": 2.967923889191617,
      "grad_norm": 1.739691972732544,
      "learning_rate": 9.03670055047553e-06,
      "loss": 0.0596,
      "step": 201525
    },
    {
      "epoch": 2.968292072281704,
      "grad_norm": 1.4545892477035522,
      "learning_rate": 9.035129633387225e-06,
      "loss": 0.0627,
      "step": 201550
    },
    {
      "epoch": 2.968660255371791,
      "grad_norm": 1.595164179801941,
      "learning_rate": 9.033493261420241e-06,
      "loss": 0.0578,
      "step": 201575
    },
    {
      "epoch": 2.969028438461878,
      "grad_norm": 1.5550832748413086,
      "learning_rate": 9.031856889453257e-06,
      "loss": 0.0572,
      "step": 201600
    },
    {
      "epoch": 2.969396621551965,
      "grad_norm": 0.9695870876312256,
      "learning_rate": 9.030220517486272e-06,
      "loss": 0.0557,
      "step": 201625
    },
    {
      "epoch": 2.969764804642052,
      "grad_norm": 1.5909755229949951,
      "learning_rate": 9.028584145519286e-06,
      "loss": 0.0558,
      "step": 201650
    },
    {
      "epoch": 2.970132987732139,
      "grad_norm": 1.476220965385437,
      "learning_rate": 9.026947773552302e-06,
      "loss": 0.0612,
      "step": 201675
    },
    {
      "epoch": 2.970501170822226,
      "grad_norm": 1.614236831665039,
      "learning_rate": 9.025311401585318e-06,
      "loss": 0.0528,
      "step": 201700
    },
    {
      "epoch": 2.9708693539123137,
      "grad_norm": 1.523722529411316,
      "learning_rate": 9.023675029618333e-06,
      "loss": 0.054,
      "step": 201725
    },
    {
      "epoch": 2.9712375370024007,
      "grad_norm": 0.7770445942878723,
      "learning_rate": 9.022038657651349e-06,
      "loss": 0.0628,
      "step": 201750
    },
    {
      "epoch": 2.9716057200924877,
      "grad_norm": 1.3149023056030273,
      "learning_rate": 9.020402285684363e-06,
      "loss": 0.0623,
      "step": 201775
    },
    {
      "epoch": 2.9719739031825747,
      "grad_norm": 1.6708271503448486,
      "learning_rate": 9.01876591371738e-06,
      "loss": 0.0618,
      "step": 201800
    },
    {
      "epoch": 2.9723420862726617,
      "grad_norm": 1.0455543994903564,
      "learning_rate": 9.017129541750394e-06,
      "loss": 0.0539,
      "step": 201825
    },
    {
      "epoch": 2.9727102693627487,
      "grad_norm": 1.2296518087387085,
      "learning_rate": 9.01549316978341e-06,
      "loss": 0.0632,
      "step": 201850
    },
    {
      "epoch": 2.9730784524528358,
      "grad_norm": 1.0646535158157349,
      "learning_rate": 9.013856797816426e-06,
      "loss": 0.0502,
      "step": 201875
    },
    {
      "epoch": 2.9734466355429228,
      "grad_norm": 1.8293626308441162,
      "learning_rate": 9.012220425849443e-06,
      "loss": 0.0555,
      "step": 201900
    },
    {
      "epoch": 2.97381481863301,
      "grad_norm": 1.4057419300079346,
      "learning_rate": 9.010584053882457e-06,
      "loss": 0.0578,
      "step": 201925
    },
    {
      "epoch": 2.974183001723097,
      "grad_norm": 1.6182752847671509,
      "learning_rate": 9.008947681915471e-06,
      "loss": 0.0578,
      "step": 201950
    },
    {
      "epoch": 2.974551184813184,
      "grad_norm": 1.1909171342849731,
      "learning_rate": 9.007311309948488e-06,
      "loss": 0.0606,
      "step": 201975
    },
    {
      "epoch": 2.974919367903271,
      "grad_norm": 1.3816593885421753,
      "learning_rate": 9.005674937981504e-06,
      "loss": 0.0545,
      "step": 202000
    },
    {
      "epoch": 2.975287550993358,
      "grad_norm": 1.3536043167114258,
      "learning_rate": 9.00403856601452e-06,
      "loss": 0.0511,
      "step": 202025
    },
    {
      "epoch": 2.975655734083445,
      "grad_norm": 1.563293218612671,
      "learning_rate": 9.002402194047534e-06,
      "loss": 0.0567,
      "step": 202050
    },
    {
      "epoch": 2.9760239171735323,
      "grad_norm": 1.500271201133728,
      "learning_rate": 9.000765822080549e-06,
      "loss": 0.0553,
      "step": 202075
    },
    {
      "epoch": 2.9763921002636193,
      "grad_norm": 1.253664493560791,
      "learning_rate": 8.999129450113565e-06,
      "loss": 0.0568,
      "step": 202100
    },
    {
      "epoch": 2.9767602833537063,
      "grad_norm": 1.2963294982910156,
      "learning_rate": 8.997493078146581e-06,
      "loss": 0.0566,
      "step": 202125
    },
    {
      "epoch": 2.9771284664437934,
      "grad_norm": 0.7844330072402954,
      "learning_rate": 8.995856706179596e-06,
      "loss": 0.0574,
      "step": 202150
    },
    {
      "epoch": 2.9774966495338804,
      "grad_norm": 1.3363014459609985,
      "learning_rate": 8.994220334212612e-06,
      "loss": 0.0627,
      "step": 202175
    },
    {
      "epoch": 2.9778648326239674,
      "grad_norm": 0.9135760068893433,
      "learning_rate": 8.992583962245626e-06,
      "loss": 0.0576,
      "step": 202200
    },
    {
      "epoch": 2.9782330157140544,
      "grad_norm": 0.970763623714447,
      "learning_rate": 8.990947590278642e-06,
      "loss": 0.0569,
      "step": 202225
    },
    {
      "epoch": 2.9786011988041414,
      "grad_norm": 1.6973836421966553,
      "learning_rate": 8.989311218311657e-06,
      "loss": 0.0597,
      "step": 202250
    },
    {
      "epoch": 2.9789693818942284,
      "grad_norm": 1.1714738607406616,
      "learning_rate": 8.987674846344673e-06,
      "loss": 0.055,
      "step": 202275
    },
    {
      "epoch": 2.9793375649843155,
      "grad_norm": 0.9725155830383301,
      "learning_rate": 8.986038474377689e-06,
      "loss": 0.0542,
      "step": 202300
    },
    {
      "epoch": 2.9797057480744025,
      "grad_norm": 1.2944258451461792,
      "learning_rate": 8.984402102410704e-06,
      "loss": 0.0546,
      "step": 202325
    },
    {
      "epoch": 2.9800739311644895,
      "grad_norm": 1.6862541437149048,
      "learning_rate": 8.98276573044372e-06,
      "loss": 0.0538,
      "step": 202350
    },
    {
      "epoch": 2.9804421142545765,
      "grad_norm": 0.9478617906570435,
      "learning_rate": 8.981129358476734e-06,
      "loss": 0.0555,
      "step": 202375
    },
    {
      "epoch": 2.9808102973446635,
      "grad_norm": 1.4293984174728394,
      "learning_rate": 8.97949298650975e-06,
      "loss": 0.0619,
      "step": 202400
    },
    {
      "epoch": 2.9811784804347505,
      "grad_norm": 1.564691185951233,
      "learning_rate": 8.977856614542766e-06,
      "loss": 0.0564,
      "step": 202425
    },
    {
      "epoch": 2.9815466635248375,
      "grad_norm": 1.2600538730621338,
      "learning_rate": 8.976220242575781e-06,
      "loss": 0.0525,
      "step": 202450
    },
    {
      "epoch": 2.9819148466149246,
      "grad_norm": 0.8747884035110474,
      "learning_rate": 8.974583870608797e-06,
      "loss": 0.0524,
      "step": 202475
    },
    {
      "epoch": 2.9822830297050116,
      "grad_norm": 1.313508152961731,
      "learning_rate": 8.972947498641812e-06,
      "loss": 0.0638,
      "step": 202500
    },
    {
      "epoch": 2.9826512127950986,
      "grad_norm": 1.1806327104568481,
      "learning_rate": 8.971311126674828e-06,
      "loss": 0.0597,
      "step": 202525
    },
    {
      "epoch": 2.9830193958851856,
      "grad_norm": 1.6251575946807861,
      "learning_rate": 8.969674754707844e-06,
      "loss": 0.0625,
      "step": 202550
    },
    {
      "epoch": 2.9833875789752726,
      "grad_norm": 1.1241422891616821,
      "learning_rate": 8.968038382740858e-06,
      "loss": 0.0569,
      "step": 202575
    },
    {
      "epoch": 2.9837557620653596,
      "grad_norm": 1.5648202896118164,
      "learning_rate": 8.966402010773874e-06,
      "loss": 0.0571,
      "step": 202600
    },
    {
      "epoch": 2.9841239451554467,
      "grad_norm": 1.427259922027588,
      "learning_rate": 8.964765638806889e-06,
      "loss": 0.0574,
      "step": 202625
    },
    {
      "epoch": 2.9844921282455337,
      "grad_norm": 1.5021920204162598,
      "learning_rate": 8.963129266839905e-06,
      "loss": 0.0554,
      "step": 202650
    },
    {
      "epoch": 2.984860311335621,
      "grad_norm": 1.5881065130233765,
      "learning_rate": 8.96149289487292e-06,
      "loss": 0.0637,
      "step": 202675
    },
    {
      "epoch": 2.985228494425708,
      "grad_norm": 1.0749396085739136,
      "learning_rate": 8.959856522905936e-06,
      "loss": 0.0513,
      "step": 202700
    },
    {
      "epoch": 2.985596677515795,
      "grad_norm": 1.4305154085159302,
      "learning_rate": 8.958220150938952e-06,
      "loss": 0.061,
      "step": 202725
    },
    {
      "epoch": 2.985964860605882,
      "grad_norm": 1.6493737697601318,
      "learning_rate": 8.956583778971966e-06,
      "loss": 0.054,
      "step": 202750
    },
    {
      "epoch": 2.986333043695969,
      "grad_norm": 1.2181254625320435,
      "learning_rate": 8.95494740700498e-06,
      "loss": 0.0627,
      "step": 202775
    },
    {
      "epoch": 2.986701226786056,
      "grad_norm": 1.3569931983947754,
      "learning_rate": 8.953311035037997e-06,
      "loss": 0.0571,
      "step": 202800
    },
    {
      "epoch": 2.9870694098761432,
      "grad_norm": 1.7755765914916992,
      "learning_rate": 8.951674663071013e-06,
      "loss": 0.0591,
      "step": 202825
    },
    {
      "epoch": 2.9874375929662302,
      "grad_norm": 1.5601924657821655,
      "learning_rate": 8.950038291104029e-06,
      "loss": 0.0562,
      "step": 202850
    },
    {
      "epoch": 2.9878057760563173,
      "grad_norm": 1.135627269744873,
      "learning_rate": 8.948401919137044e-06,
      "loss": 0.068,
      "step": 202875
    },
    {
      "epoch": 2.9881739591464043,
      "grad_norm": 1.6177241802215576,
      "learning_rate": 8.946765547170058e-06,
      "loss": 0.0546,
      "step": 202900
    },
    {
      "epoch": 2.9885421422364913,
      "grad_norm": 1.8751848936080933,
      "learning_rate": 8.945129175203074e-06,
      "loss": 0.0577,
      "step": 202925
    },
    {
      "epoch": 2.9889103253265783,
      "grad_norm": 1.1617170572280884,
      "learning_rate": 8.94349280323609e-06,
      "loss": 0.0544,
      "step": 202950
    },
    {
      "epoch": 2.9892785084166653,
      "grad_norm": 1.0930885076522827,
      "learning_rate": 8.941856431269107e-06,
      "loss": 0.0616,
      "step": 202975
    },
    {
      "epoch": 2.9896466915067523,
      "grad_norm": 1.3557119369506836,
      "learning_rate": 8.940220059302121e-06,
      "loss": 0.0615,
      "step": 203000
    },
    {
      "epoch": 2.99001487459684,
      "grad_norm": 1.8400906324386597,
      "learning_rate": 8.938583687335135e-06,
      "loss": 0.0619,
      "step": 203025
    },
    {
      "epoch": 2.990383057686927,
      "grad_norm": 1.0624231100082397,
      "learning_rate": 8.936947315368152e-06,
      "loss": 0.0588,
      "step": 203050
    },
    {
      "epoch": 2.990751240777014,
      "grad_norm": 1.1801598072052002,
      "learning_rate": 8.935310943401168e-06,
      "loss": 0.0612,
      "step": 203075
    },
    {
      "epoch": 2.991119423867101,
      "grad_norm": 1.715049386024475,
      "learning_rate": 8.933674571434182e-06,
      "loss": 0.061,
      "step": 203100
    },
    {
      "epoch": 2.991487606957188,
      "grad_norm": 1.2938154935836792,
      "learning_rate": 8.932038199467198e-06,
      "loss": 0.051,
      "step": 203125
    },
    {
      "epoch": 2.991855790047275,
      "grad_norm": 1.0665907859802246,
      "learning_rate": 8.930401827500213e-06,
      "loss": 0.0647,
      "step": 203150
    },
    {
      "epoch": 2.992223973137362,
      "grad_norm": 1.6673771142959595,
      "learning_rate": 8.928765455533229e-06,
      "loss": 0.0506,
      "step": 203175
    },
    {
      "epoch": 2.992592156227449,
      "grad_norm": 1.6066478490829468,
      "learning_rate": 8.927129083566243e-06,
      "loss": 0.0584,
      "step": 203200
    },
    {
      "epoch": 2.992960339317536,
      "grad_norm": 1.2847256660461426,
      "learning_rate": 8.92549271159926e-06,
      "loss": 0.0567,
      "step": 203225
    },
    {
      "epoch": 2.993328522407623,
      "grad_norm": 1.3385009765625,
      "learning_rate": 8.923856339632276e-06,
      "loss": 0.0508,
      "step": 203250
    },
    {
      "epoch": 2.99369670549771,
      "grad_norm": 1.4735643863677979,
      "learning_rate": 8.92221996766529e-06,
      "loss": 0.0552,
      "step": 203275
    },
    {
      "epoch": 2.994064888587797,
      "grad_norm": 1.4854005575180054,
      "learning_rate": 8.920583595698306e-06,
      "loss": 0.062,
      "step": 203300
    },
    {
      "epoch": 2.994433071677884,
      "grad_norm": 2.120142698287964,
      "learning_rate": 8.91894722373132e-06,
      "loss": 0.0566,
      "step": 203325
    },
    {
      "epoch": 2.994801254767971,
      "grad_norm": 1.3660907745361328,
      "learning_rate": 8.917310851764337e-06,
      "loss": 0.0588,
      "step": 203350
    },
    {
      "epoch": 2.995169437858058,
      "grad_norm": 0.9961246848106384,
      "learning_rate": 8.915674479797353e-06,
      "loss": 0.0563,
      "step": 203375
    },
    {
      "epoch": 2.995537620948145,
      "grad_norm": 1.458559513092041,
      "learning_rate": 8.914038107830368e-06,
      "loss": 0.0606,
      "step": 203400
    },
    {
      "epoch": 2.995905804038232,
      "grad_norm": 1.472598671913147,
      "learning_rate": 8.912401735863384e-06,
      "loss": 0.063,
      "step": 203425
    },
    {
      "epoch": 2.996273987128319,
      "grad_norm": 1.3394176959991455,
      "learning_rate": 8.910765363896398e-06,
      "loss": 0.0547,
      "step": 203450
    },
    {
      "epoch": 2.996642170218406,
      "grad_norm": 1.1453043222427368,
      "learning_rate": 8.909128991929414e-06,
      "loss": 0.0596,
      "step": 203475
    },
    {
      "epoch": 2.997010353308493,
      "grad_norm": 1.615200400352478,
      "learning_rate": 8.90749261996243e-06,
      "loss": 0.0607,
      "step": 203500
    },
    {
      "epoch": 2.99737853639858,
      "grad_norm": 1.1500487327575684,
      "learning_rate": 8.905856247995445e-06,
      "loss": 0.0564,
      "step": 203525
    },
    {
      "epoch": 2.997746719488667,
      "grad_norm": 1.5393701791763306,
      "learning_rate": 8.904219876028461e-06,
      "loss": 0.0578,
      "step": 203550
    },
    {
      "epoch": 2.998114902578754,
      "grad_norm": 0.8329442143440247,
      "learning_rate": 8.902583504061476e-06,
      "loss": 0.063,
      "step": 203575
    },
    {
      "epoch": 2.998483085668841,
      "grad_norm": 1.5081946849822998,
      "learning_rate": 8.900947132094492e-06,
      "loss": 0.0597,
      "step": 203600
    },
    {
      "epoch": 2.9988512687589286,
      "grad_norm": 1.3962911367416382,
      "learning_rate": 8.899310760127506e-06,
      "loss": 0.0704,
      "step": 203625
    },
    {
      "epoch": 2.9992194518490156,
      "grad_norm": 1.4709864854812622,
      "learning_rate": 8.897674388160522e-06,
      "loss": 0.0514,
      "step": 203650
    },
    {
      "epoch": 2.9995876349391026,
      "grad_norm": 1.0525881052017212,
      "learning_rate": 8.896038016193538e-06,
      "loss": 0.0569,
      "step": 203675
    },
    {
      "epoch": 2.9999558180291896,
      "grad_norm": 1.207714557647705,
      "learning_rate": 8.894401644226553e-06,
      "loss": 0.0535,
      "step": 203700
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.06028754264116287,
      "eval_runtime": 112.9082,
      "eval_samples_per_second": 3141.916,
      "eval_steps_per_second": 6.138,
      "step": 203703
    },
    {
      "epoch": 3.0003240011192767,
      "grad_norm": 1.2547531127929688,
      "learning_rate": 8.892765272259567e-06,
      "loss": 0.0586,
      "step": 203725
    },
    {
      "epoch": 3.0006921842093637,
      "grad_norm": 0.9733414053916931,
      "learning_rate": 8.891128900292583e-06,
      "loss": 0.0563,
      "step": 203750
    },
    {
      "epoch": 3.0010603672994507,
      "grad_norm": 1.3888561725616455,
      "learning_rate": 8.8894925283256e-06,
      "loss": 0.0544,
      "step": 203775
    },
    {
      "epoch": 3.0014285503895377,
      "grad_norm": 1.5861791372299194,
      "learning_rate": 8.887856156358616e-06,
      "loss": 0.0526,
      "step": 203800
    },
    {
      "epoch": 3.0017967334796247,
      "grad_norm": 1.382179856300354,
      "learning_rate": 8.88621978439163e-06,
      "loss": 0.059,
      "step": 203825
    },
    {
      "epoch": 3.0021649165697117,
      "grad_norm": 1.2970032691955566,
      "learning_rate": 8.884583412424645e-06,
      "loss": 0.0511,
      "step": 203850
    },
    {
      "epoch": 3.0025330996597988,
      "grad_norm": 1.096897840499878,
      "learning_rate": 8.882947040457661e-06,
      "loss": 0.0616,
      "step": 203875
    },
    {
      "epoch": 3.0029012827498858,
      "grad_norm": 1.4717472791671753,
      "learning_rate": 8.881310668490677e-06,
      "loss": 0.0544,
      "step": 203900
    },
    {
      "epoch": 3.003269465839973,
      "grad_norm": 1.314205527305603,
      "learning_rate": 8.879674296523693e-06,
      "loss": 0.0562,
      "step": 203925
    },
    {
      "epoch": 3.00363764893006,
      "grad_norm": 1.1430346965789795,
      "learning_rate": 8.878037924556708e-06,
      "loss": 0.0515,
      "step": 203950
    },
    {
      "epoch": 3.004005832020147,
      "grad_norm": 1.457344651222229,
      "learning_rate": 8.876401552589722e-06,
      "loss": 0.0498,
      "step": 203975
    },
    {
      "epoch": 3.004374015110234,
      "grad_norm": 1.669184684753418,
      "learning_rate": 8.874765180622738e-06,
      "loss": 0.0593,
      "step": 204000
    },
    {
      "epoch": 3.004742198200321,
      "grad_norm": 1.449445128440857,
      "learning_rate": 8.873128808655754e-06,
      "loss": 0.0548,
      "step": 204025
    },
    {
      "epoch": 3.0051103812904083,
      "grad_norm": 1.431996464729309,
      "learning_rate": 8.871492436688769e-06,
      "loss": 0.0537,
      "step": 204050
    },
    {
      "epoch": 3.0054785643804953,
      "grad_norm": 0.9228318333625793,
      "learning_rate": 8.869856064721785e-06,
      "loss": 0.0592,
      "step": 204075
    },
    {
      "epoch": 3.0058467474705823,
      "grad_norm": 1.6553899049758911,
      "learning_rate": 8.868219692754801e-06,
      "loss": 0.0557,
      "step": 204100
    },
    {
      "epoch": 3.0062149305606694,
      "grad_norm": 1.8629385232925415,
      "learning_rate": 8.866583320787816e-06,
      "loss": 0.0577,
      "step": 204125
    },
    {
      "epoch": 3.0065831136507564,
      "grad_norm": 1.2063720226287842,
      "learning_rate": 8.86494694882083e-06,
      "loss": 0.0594,
      "step": 204150
    },
    {
      "epoch": 3.0069512967408434,
      "grad_norm": 0.875553548336029,
      "learning_rate": 8.863310576853846e-06,
      "loss": 0.0564,
      "step": 204175
    },
    {
      "epoch": 3.0073194798309304,
      "grad_norm": 1.0950829982757568,
      "learning_rate": 8.861674204886862e-06,
      "loss": 0.0533,
      "step": 204200
    },
    {
      "epoch": 3.0076876629210174,
      "grad_norm": 1.1200106143951416,
      "learning_rate": 8.860037832919879e-06,
      "loss": 0.0531,
      "step": 204225
    },
    {
      "epoch": 3.0080558460111044,
      "grad_norm": 0.9940056204795837,
      "learning_rate": 8.858401460952893e-06,
      "loss": 0.0574,
      "step": 204250
    },
    {
      "epoch": 3.0084240291011914,
      "grad_norm": 1.985337257385254,
      "learning_rate": 8.856765088985907e-06,
      "loss": 0.0631,
      "step": 204275
    },
    {
      "epoch": 3.0087922121912785,
      "grad_norm": 1.3617161512374878,
      "learning_rate": 8.855128717018924e-06,
      "loss": 0.0542,
      "step": 204300
    },
    {
      "epoch": 3.0091603952813655,
      "grad_norm": 2.3319380283355713,
      "learning_rate": 8.85349234505194e-06,
      "loss": 0.0554,
      "step": 204325
    },
    {
      "epoch": 3.0095285783714525,
      "grad_norm": 1.4009677171707153,
      "learning_rate": 8.851855973084956e-06,
      "loss": 0.0568,
      "step": 204350
    },
    {
      "epoch": 3.0098967614615395,
      "grad_norm": 0.9551072716712952,
      "learning_rate": 8.85021960111797e-06,
      "loss": 0.0503,
      "step": 204375
    },
    {
      "epoch": 3.0102649445516265,
      "grad_norm": 1.7699204683303833,
      "learning_rate": 8.848583229150985e-06,
      "loss": 0.0598,
      "step": 204400
    },
    {
      "epoch": 3.0106331276417135,
      "grad_norm": 1.314049482345581,
      "learning_rate": 8.846946857184001e-06,
      "loss": 0.0571,
      "step": 204425
    },
    {
      "epoch": 3.0110013107318006,
      "grad_norm": 1.1528035402297974,
      "learning_rate": 8.845310485217017e-06,
      "loss": 0.0577,
      "step": 204450
    },
    {
      "epoch": 3.0113694938218876,
      "grad_norm": 1.3904610872268677,
      "learning_rate": 8.843674113250032e-06,
      "loss": 0.0514,
      "step": 204475
    },
    {
      "epoch": 3.0117376769119746,
      "grad_norm": 1.1227728128433228,
      "learning_rate": 8.842037741283048e-06,
      "loss": 0.0568,
      "step": 204500
    },
    {
      "epoch": 3.012105860002062,
      "grad_norm": 1.5878812074661255,
      "learning_rate": 8.840401369316062e-06,
      "loss": 0.0521,
      "step": 204525
    },
    {
      "epoch": 3.012474043092149,
      "grad_norm": 1.4791150093078613,
      "learning_rate": 8.838764997349078e-06,
      "loss": 0.0559,
      "step": 204550
    },
    {
      "epoch": 3.012842226182236,
      "grad_norm": 1.7063350677490234,
      "learning_rate": 8.837128625382093e-06,
      "loss": 0.0521,
      "step": 204575
    },
    {
      "epoch": 3.013210409272323,
      "grad_norm": 1.2506731748580933,
      "learning_rate": 8.835492253415109e-06,
      "loss": 0.0524,
      "step": 204600
    },
    {
      "epoch": 3.01357859236241,
      "grad_norm": 1.1786755323410034,
      "learning_rate": 8.833855881448125e-06,
      "loss": 0.054,
      "step": 204625
    },
    {
      "epoch": 3.013946775452497,
      "grad_norm": 1.3485809564590454,
      "learning_rate": 8.83221950948114e-06,
      "loss": 0.0499,
      "step": 204650
    },
    {
      "epoch": 3.014314958542584,
      "grad_norm": 1.3246203660964966,
      "learning_rate": 8.830583137514156e-06,
      "loss": 0.0598,
      "step": 204675
    },
    {
      "epoch": 3.014683141632671,
      "grad_norm": 0.915708065032959,
      "learning_rate": 8.82894676554717e-06,
      "loss": 0.0589,
      "step": 204700
    },
    {
      "epoch": 3.015051324722758,
      "grad_norm": 1.1194193363189697,
      "learning_rate": 8.827310393580186e-06,
      "loss": 0.0566,
      "step": 204725
    },
    {
      "epoch": 3.015419507812845,
      "grad_norm": 0.7385013699531555,
      "learning_rate": 8.825674021613202e-06,
      "loss": 0.0546,
      "step": 204750
    },
    {
      "epoch": 3.015787690902932,
      "grad_norm": 1.127184271812439,
      "learning_rate": 8.824037649646217e-06,
      "loss": 0.0537,
      "step": 204775
    },
    {
      "epoch": 3.016155873993019,
      "grad_norm": 1.0766645669937134,
      "learning_rate": 8.822401277679233e-06,
      "loss": 0.0615,
      "step": 204800
    },
    {
      "epoch": 3.0165240570831062,
      "grad_norm": 1.3340160846710205,
      "learning_rate": 8.820764905712248e-06,
      "loss": 0.056,
      "step": 204825
    },
    {
      "epoch": 3.0168922401731932,
      "grad_norm": 1.3009932041168213,
      "learning_rate": 8.819128533745264e-06,
      "loss": 0.0511,
      "step": 204850
    },
    {
      "epoch": 3.0172604232632803,
      "grad_norm": 1.339657187461853,
      "learning_rate": 8.81749216177828e-06,
      "loss": 0.0588,
      "step": 204875
    },
    {
      "epoch": 3.0176286063533673,
      "grad_norm": 1.4364664554595947,
      "learning_rate": 8.815855789811294e-06,
      "loss": 0.0541,
      "step": 204900
    },
    {
      "epoch": 3.0179967894434543,
      "grad_norm": 1.4257307052612305,
      "learning_rate": 8.81421941784431e-06,
      "loss": 0.0583,
      "step": 204925
    },
    {
      "epoch": 3.0183649725335413,
      "grad_norm": 1.0394444465637207,
      "learning_rate": 8.812583045877325e-06,
      "loss": 0.0503,
      "step": 204950
    },
    {
      "epoch": 3.0187331556236283,
      "grad_norm": 1.498558759689331,
      "learning_rate": 8.810946673910341e-06,
      "loss": 0.0567,
      "step": 204975
    },
    {
      "epoch": 3.0191013387137158,
      "grad_norm": 1.1389732360839844,
      "learning_rate": 8.809310301943355e-06,
      "loss": 0.0566,
      "step": 205000
    },
    {
      "epoch": 3.019469521803803,
      "grad_norm": 1.5723432302474976,
      "learning_rate": 8.807673929976372e-06,
      "loss": 0.0536,
      "step": 205025
    },
    {
      "epoch": 3.01983770489389,
      "grad_norm": 1.2834051847457886,
      "learning_rate": 8.806037558009388e-06,
      "loss": 0.0591,
      "step": 205050
    },
    {
      "epoch": 3.020205887983977,
      "grad_norm": 1.2777409553527832,
      "learning_rate": 8.804401186042402e-06,
      "loss": 0.0575,
      "step": 205075
    },
    {
      "epoch": 3.020574071074064,
      "grad_norm": 1.115095853805542,
      "learning_rate": 8.802764814075417e-06,
      "loss": 0.0553,
      "step": 205100
    },
    {
      "epoch": 3.020942254164151,
      "grad_norm": 1.5836453437805176,
      "learning_rate": 8.801128442108433e-06,
      "loss": 0.0568,
      "step": 205125
    },
    {
      "epoch": 3.021310437254238,
      "grad_norm": 1.372407078742981,
      "learning_rate": 8.799492070141449e-06,
      "loss": 0.0552,
      "step": 205150
    },
    {
      "epoch": 3.021678620344325,
      "grad_norm": 1.5378984212875366,
      "learning_rate": 8.797855698174465e-06,
      "loss": 0.0615,
      "step": 205175
    },
    {
      "epoch": 3.022046803434412,
      "grad_norm": 1.590171217918396,
      "learning_rate": 8.79621932620748e-06,
      "loss": 0.0553,
      "step": 205200
    },
    {
      "epoch": 3.022414986524499,
      "grad_norm": 1.4625080823898315,
      "learning_rate": 8.794582954240494e-06,
      "loss": 0.0505,
      "step": 205225
    },
    {
      "epoch": 3.022783169614586,
      "grad_norm": 1.2913883924484253,
      "learning_rate": 8.79294658227351e-06,
      "loss": 0.0584,
      "step": 205250
    },
    {
      "epoch": 3.023151352704673,
      "grad_norm": 1.7851440906524658,
      "learning_rate": 8.791310210306526e-06,
      "loss": 0.0608,
      "step": 205275
    },
    {
      "epoch": 3.02351953579476,
      "grad_norm": 1.3867429494857788,
      "learning_rate": 8.789673838339543e-06,
      "loss": 0.0579,
      "step": 205300
    },
    {
      "epoch": 3.023887718884847,
      "grad_norm": 1.428532600402832,
      "learning_rate": 8.788037466372557e-06,
      "loss": 0.0502,
      "step": 205325
    },
    {
      "epoch": 3.024255901974934,
      "grad_norm": 1.3660624027252197,
      "learning_rate": 8.786401094405571e-06,
      "loss": 0.0576,
      "step": 205350
    },
    {
      "epoch": 3.024624085065021,
      "grad_norm": 1.6445670127868652,
      "learning_rate": 8.784764722438588e-06,
      "loss": 0.0601,
      "step": 205375
    },
    {
      "epoch": 3.024992268155108,
      "grad_norm": 1.4324208498001099,
      "learning_rate": 8.783128350471604e-06,
      "loss": 0.0666,
      "step": 205400
    },
    {
      "epoch": 3.025360451245195,
      "grad_norm": 0.9703266620635986,
      "learning_rate": 8.781491978504618e-06,
      "loss": 0.0522,
      "step": 205425
    },
    {
      "epoch": 3.025728634335282,
      "grad_norm": 1.2462025880813599,
      "learning_rate": 8.779855606537634e-06,
      "loss": 0.0556,
      "step": 205450
    },
    {
      "epoch": 3.026096817425369,
      "grad_norm": 0.980064332485199,
      "learning_rate": 8.778219234570649e-06,
      "loss": 0.0499,
      "step": 205475
    },
    {
      "epoch": 3.0264650005154565,
      "grad_norm": 1.148585319519043,
      "learning_rate": 8.776582862603665e-06,
      "loss": 0.0566,
      "step": 205500
    },
    {
      "epoch": 3.0268331836055435,
      "grad_norm": 1.5142290592193604,
      "learning_rate": 8.77494649063668e-06,
      "loss": 0.0591,
      "step": 205525
    },
    {
      "epoch": 3.0272013666956306,
      "grad_norm": 1.345026969909668,
      "learning_rate": 8.773375573548375e-06,
      "loss": 0.0563,
      "step": 205550
    },
    {
      "epoch": 3.0275695497857176,
      "grad_norm": 1.6352711915969849,
      "learning_rate": 8.771739201581391e-06,
      "loss": 0.0598,
      "step": 205575
    },
    {
      "epoch": 3.0279377328758046,
      "grad_norm": 1.354448676109314,
      "learning_rate": 8.770102829614406e-06,
      "loss": 0.0531,
      "step": 205600
    },
    {
      "epoch": 3.0283059159658916,
      "grad_norm": 1.3734674453735352,
      "learning_rate": 8.768466457647422e-06,
      "loss": 0.0595,
      "step": 205625
    },
    {
      "epoch": 3.0286740990559786,
      "grad_norm": 1.3386586904525757,
      "learning_rate": 8.766830085680436e-06,
      "loss": 0.0573,
      "step": 205650
    },
    {
      "epoch": 3.0290422821460656,
      "grad_norm": 1.2852375507354736,
      "learning_rate": 8.765193713713452e-06,
      "loss": 0.0601,
      "step": 205675
    },
    {
      "epoch": 3.0294104652361526,
      "grad_norm": 1.6926796436309814,
      "learning_rate": 8.763557341746469e-06,
      "loss": 0.0571,
      "step": 205700
    },
    {
      "epoch": 3.0297786483262397,
      "grad_norm": 1.0144689083099365,
      "learning_rate": 8.761920969779483e-06,
      "loss": 0.0509,
      "step": 205725
    },
    {
      "epoch": 3.0301468314163267,
      "grad_norm": 1.588392734527588,
      "learning_rate": 8.760284597812499e-06,
      "loss": 0.0545,
      "step": 205750
    },
    {
      "epoch": 3.0305150145064137,
      "grad_norm": 1.201865792274475,
      "learning_rate": 8.758648225845514e-06,
      "loss": 0.0504,
      "step": 205775
    },
    {
      "epoch": 3.0308831975965007,
      "grad_norm": 1.4972093105316162,
      "learning_rate": 8.75701185387853e-06,
      "loss": 0.0568,
      "step": 205800
    },
    {
      "epoch": 3.0312513806865877,
      "grad_norm": 1.420656442642212,
      "learning_rate": 8.755375481911546e-06,
      "loss": 0.0516,
      "step": 205825
    },
    {
      "epoch": 3.0316195637766747,
      "grad_norm": 1.4574306011199951,
      "learning_rate": 8.75373910994456e-06,
      "loss": 0.0535,
      "step": 205850
    },
    {
      "epoch": 3.0319877468667618,
      "grad_norm": 1.0388892889022827,
      "learning_rate": 8.752102737977577e-06,
      "loss": 0.0557,
      "step": 205875
    },
    {
      "epoch": 3.0323559299568488,
      "grad_norm": 1.3123277425765991,
      "learning_rate": 8.750466366010591e-06,
      "loss": 0.0609,
      "step": 205900
    },
    {
      "epoch": 3.032724113046936,
      "grad_norm": 1.2200952768325806,
      "learning_rate": 8.748829994043607e-06,
      "loss": 0.0484,
      "step": 205925
    },
    {
      "epoch": 3.033092296137023,
      "grad_norm": 1.0080150365829468,
      "learning_rate": 8.747193622076622e-06,
      "loss": 0.0581,
      "step": 205950
    },
    {
      "epoch": 3.0334604792271103,
      "grad_norm": 1.3287502527236938,
      "learning_rate": 8.745557250109638e-06,
      "loss": 0.0568,
      "step": 205975
    },
    {
      "epoch": 3.0338286623171973,
      "grad_norm": 1.698480248451233,
      "learning_rate": 8.743920878142654e-06,
      "loss": 0.0586,
      "step": 206000
    },
    {
      "epoch": 3.0341968454072843,
      "grad_norm": 1.0519176721572876,
      "learning_rate": 8.742284506175668e-06,
      "loss": 0.0511,
      "step": 206025
    },
    {
      "epoch": 3.0345650284973713,
      "grad_norm": 1.6990594863891602,
      "learning_rate": 8.740648134208683e-06,
      "loss": 0.0595,
      "step": 206050
    },
    {
      "epoch": 3.0349332115874583,
      "grad_norm": 1.2284382581710815,
      "learning_rate": 8.739011762241699e-06,
      "loss": 0.0609,
      "step": 206075
    },
    {
      "epoch": 3.0353013946775453,
      "grad_norm": 1.9676870107650757,
      "learning_rate": 8.737375390274715e-06,
      "loss": 0.059,
      "step": 206100
    },
    {
      "epoch": 3.0356695777676324,
      "grad_norm": 1.1042401790618896,
      "learning_rate": 8.735739018307731e-06,
      "loss": 0.054,
      "step": 206125
    },
    {
      "epoch": 3.0360377608577194,
      "grad_norm": 1.369608759880066,
      "learning_rate": 8.734102646340746e-06,
      "loss": 0.0533,
      "step": 206150
    },
    {
      "epoch": 3.0364059439478064,
      "grad_norm": 1.3927401304244995,
      "learning_rate": 8.73246627437376e-06,
      "loss": 0.0541,
      "step": 206175
    },
    {
      "epoch": 3.0367741270378934,
      "grad_norm": 1.536010980606079,
      "learning_rate": 8.730829902406776e-06,
      "loss": 0.0533,
      "step": 206200
    },
    {
      "epoch": 3.0371423101279804,
      "grad_norm": 1.2623169422149658,
      "learning_rate": 8.729193530439792e-06,
      "loss": 0.0581,
      "step": 206225
    },
    {
      "epoch": 3.0375104932180674,
      "grad_norm": 1.2068862915039062,
      "learning_rate": 8.727557158472807e-06,
      "loss": 0.0572,
      "step": 206250
    },
    {
      "epoch": 3.0378786763081544,
      "grad_norm": 1.3528281450271606,
      "learning_rate": 8.725920786505823e-06,
      "loss": 0.0546,
      "step": 206275
    },
    {
      "epoch": 3.0382468593982415,
      "grad_norm": 1.5392383337020874,
      "learning_rate": 8.724284414538838e-06,
      "loss": 0.0603,
      "step": 206300
    },
    {
      "epoch": 3.0386150424883285,
      "grad_norm": 1.4895553588867188,
      "learning_rate": 8.722648042571854e-06,
      "loss": 0.0505,
      "step": 206325
    },
    {
      "epoch": 3.0389832255784155,
      "grad_norm": 1.4076825380325317,
      "learning_rate": 8.72101167060487e-06,
      "loss": 0.0524,
      "step": 206350
    },
    {
      "epoch": 3.0393514086685025,
      "grad_norm": 1.4680297374725342,
      "learning_rate": 8.719375298637884e-06,
      "loss": 0.0511,
      "step": 206375
    },
    {
      "epoch": 3.0397195917585895,
      "grad_norm": 1.3740334510803223,
      "learning_rate": 8.7177389266709e-06,
      "loss": 0.0602,
      "step": 206400
    },
    {
      "epoch": 3.0400877748486765,
      "grad_norm": 1.6759331226348877,
      "learning_rate": 8.716102554703917e-06,
      "loss": 0.0566,
      "step": 206425
    },
    {
      "epoch": 3.040455957938764,
      "grad_norm": 1.0811715126037598,
      "learning_rate": 8.714466182736931e-06,
      "loss": 0.0514,
      "step": 206450
    },
    {
      "epoch": 3.040824141028851,
      "grad_norm": 1.4177398681640625,
      "learning_rate": 8.712829810769946e-06,
      "loss": 0.0546,
      "step": 206475
    },
    {
      "epoch": 3.041192324118938,
      "grad_norm": 1.4018912315368652,
      "learning_rate": 8.711193438802962e-06,
      "loss": 0.0549,
      "step": 206500
    },
    {
      "epoch": 3.041560507209025,
      "grad_norm": 1.1432408094406128,
      "learning_rate": 8.709557066835978e-06,
      "loss": 0.0579,
      "step": 206525
    },
    {
      "epoch": 3.041928690299112,
      "grad_norm": 0.9459879398345947,
      "learning_rate": 8.707920694868994e-06,
      "loss": 0.0518,
      "step": 206550
    },
    {
      "epoch": 3.042296873389199,
      "grad_norm": 1.3017204999923706,
      "learning_rate": 8.706284322902008e-06,
      "loss": 0.0532,
      "step": 206575
    },
    {
      "epoch": 3.042665056479286,
      "grad_norm": 1.4303687810897827,
      "learning_rate": 8.704647950935023e-06,
      "loss": 0.0592,
      "step": 206600
    },
    {
      "epoch": 3.043033239569373,
      "grad_norm": 1.5987792015075684,
      "learning_rate": 8.703011578968039e-06,
      "loss": 0.0548,
      "step": 206625
    },
    {
      "epoch": 3.04340142265946,
      "grad_norm": 1.5667405128479004,
      "learning_rate": 8.701375207001055e-06,
      "loss": 0.0602,
      "step": 206650
    },
    {
      "epoch": 3.043769605749547,
      "grad_norm": 1.3785927295684814,
      "learning_rate": 8.69973883503407e-06,
      "loss": 0.0552,
      "step": 206675
    },
    {
      "epoch": 3.044137788839634,
      "grad_norm": 1.4220634698867798,
      "learning_rate": 8.698102463067086e-06,
      "loss": 0.0648,
      "step": 206700
    },
    {
      "epoch": 3.044505971929721,
      "grad_norm": 1.6044307947158813,
      "learning_rate": 8.6964660911001e-06,
      "loss": 0.0559,
      "step": 206725
    },
    {
      "epoch": 3.044874155019808,
      "grad_norm": 1.194157600402832,
      "learning_rate": 8.694829719133116e-06,
      "loss": 0.0547,
      "step": 206750
    },
    {
      "epoch": 3.045242338109895,
      "grad_norm": 0.9766545295715332,
      "learning_rate": 8.693258802044812e-06,
      "loss": 0.0572,
      "step": 206775
    },
    {
      "epoch": 3.045610521199982,
      "grad_norm": 1.2415086030960083,
      "learning_rate": 8.691622430077826e-06,
      "loss": 0.0521,
      "step": 206800
    },
    {
      "epoch": 3.0459787042900692,
      "grad_norm": 0.9433152675628662,
      "learning_rate": 8.689986058110843e-06,
      "loss": 0.0596,
      "step": 206825
    },
    {
      "epoch": 3.0463468873801562,
      "grad_norm": 1.1773619651794434,
      "learning_rate": 8.688349686143857e-06,
      "loss": 0.0606,
      "step": 206850
    },
    {
      "epoch": 3.0467150704702433,
      "grad_norm": 1.4022718667984009,
      "learning_rate": 8.686713314176873e-06,
      "loss": 0.0459,
      "step": 206875
    },
    {
      "epoch": 3.0470832535603303,
      "grad_norm": 1.40048348903656,
      "learning_rate": 8.685076942209888e-06,
      "loss": 0.0577,
      "step": 206900
    },
    {
      "epoch": 3.0474514366504177,
      "grad_norm": 1.2992240190505981,
      "learning_rate": 8.683440570242904e-06,
      "loss": 0.0548,
      "step": 206925
    },
    {
      "epoch": 3.0478196197405047,
      "grad_norm": 1.4067257642745972,
      "learning_rate": 8.68180419827592e-06,
      "loss": 0.0571,
      "step": 206950
    },
    {
      "epoch": 3.0481878028305918,
      "grad_norm": 1.6077123880386353,
      "learning_rate": 8.680167826308934e-06,
      "loss": 0.059,
      "step": 206975
    },
    {
      "epoch": 3.048555985920679,
      "grad_norm": 1.639776587486267,
      "learning_rate": 8.678531454341949e-06,
      "loss": 0.0514,
      "step": 207000
    },
    {
      "epoch": 3.048924169010766,
      "grad_norm": 1.1378449201583862,
      "learning_rate": 8.676895082374965e-06,
      "loss": 0.056,
      "step": 207025
    },
    {
      "epoch": 3.049292352100853,
      "grad_norm": 1.3766497373580933,
      "learning_rate": 8.675258710407981e-06,
      "loss": 0.0566,
      "step": 207050
    },
    {
      "epoch": 3.04966053519094,
      "grad_norm": 1.2629650831222534,
      "learning_rate": 8.673622338440997e-06,
      "loss": 0.0508,
      "step": 207075
    },
    {
      "epoch": 3.050028718281027,
      "grad_norm": 1.3769396543502808,
      "learning_rate": 8.671985966474012e-06,
      "loss": 0.0563,
      "step": 207100
    },
    {
      "epoch": 3.050396901371114,
      "grad_norm": 1.697031021118164,
      "learning_rate": 8.670349594507026e-06,
      "loss": 0.0537,
      "step": 207125
    },
    {
      "epoch": 3.050765084461201,
      "grad_norm": 1.6207547187805176,
      "learning_rate": 8.668713222540042e-06,
      "loss": 0.0566,
      "step": 207150
    },
    {
      "epoch": 3.051133267551288,
      "grad_norm": 1.6054742336273193,
      "learning_rate": 8.667076850573059e-06,
      "loss": 0.0588,
      "step": 207175
    },
    {
      "epoch": 3.051501450641375,
      "grad_norm": 1.6720421314239502,
      "learning_rate": 8.665440478606073e-06,
      "loss": 0.0484,
      "step": 207200
    },
    {
      "epoch": 3.051869633731462,
      "grad_norm": 1.7105470895767212,
      "learning_rate": 8.663804106639089e-06,
      "loss": 0.0646,
      "step": 207225
    },
    {
      "epoch": 3.052237816821549,
      "grad_norm": 1.3694008588790894,
      "learning_rate": 8.662167734672105e-06,
      "loss": 0.0544,
      "step": 207250
    },
    {
      "epoch": 3.052605999911636,
      "grad_norm": 1.3467543125152588,
      "learning_rate": 8.66053136270512e-06,
      "loss": 0.0512,
      "step": 207275
    },
    {
      "epoch": 3.052974183001723,
      "grad_norm": 1.4946752786636353,
      "learning_rate": 8.658894990738136e-06,
      "loss": 0.0589,
      "step": 207300
    },
    {
      "epoch": 3.05334236609181,
      "grad_norm": 1.2056653499603271,
      "learning_rate": 8.65725861877115e-06,
      "loss": 0.0494,
      "step": 207325
    },
    {
      "epoch": 3.053710549181897,
      "grad_norm": 1.4029792547225952,
      "learning_rate": 8.655622246804167e-06,
      "loss": 0.0587,
      "step": 207350
    },
    {
      "epoch": 3.054078732271984,
      "grad_norm": 1.0971531867980957,
      "learning_rate": 8.653985874837183e-06,
      "loss": 0.0527,
      "step": 207375
    },
    {
      "epoch": 3.0544469153620715,
      "grad_norm": 1.40229332447052,
      "learning_rate": 8.652349502870197e-06,
      "loss": 0.0592,
      "step": 207400
    },
    {
      "epoch": 3.0548150984521585,
      "grad_norm": 1.3901351690292358,
      "learning_rate": 8.650713130903212e-06,
      "loss": 0.055,
      "step": 207425
    },
    {
      "epoch": 3.0551832815422455,
      "grad_norm": 1.6501914262771606,
      "learning_rate": 8.649076758936228e-06,
      "loss": 0.0602,
      "step": 207450
    },
    {
      "epoch": 3.0555514646323325,
      "grad_norm": 1.0003635883331299,
      "learning_rate": 8.647440386969244e-06,
      "loss": 0.0536,
      "step": 207475
    },
    {
      "epoch": 3.0559196477224195,
      "grad_norm": 1.4143460988998413,
      "learning_rate": 8.64580401500226e-06,
      "loss": 0.0499,
      "step": 207500
    },
    {
      "epoch": 3.0562878308125065,
      "grad_norm": 1.3231967687606812,
      "learning_rate": 8.644167643035274e-06,
      "loss": 0.0593,
      "step": 207525
    },
    {
      "epoch": 3.0566560139025936,
      "grad_norm": 1.5237478017807007,
      "learning_rate": 8.642531271068289e-06,
      "loss": 0.0532,
      "step": 207550
    },
    {
      "epoch": 3.0570241969926806,
      "grad_norm": 0.9174051880836487,
      "learning_rate": 8.640894899101305e-06,
      "loss": 0.0554,
      "step": 207575
    },
    {
      "epoch": 3.0573923800827676,
      "grad_norm": 1.524187684059143,
      "learning_rate": 8.639258527134321e-06,
      "loss": 0.052,
      "step": 207600
    },
    {
      "epoch": 3.0577605631728546,
      "grad_norm": 0.962013304233551,
      "learning_rate": 8.637622155167336e-06,
      "loss": 0.0465,
      "step": 207625
    },
    {
      "epoch": 3.0581287462629416,
      "grad_norm": 1.402522325515747,
      "learning_rate": 8.635985783200352e-06,
      "loss": 0.0551,
      "step": 207650
    },
    {
      "epoch": 3.0584969293530286,
      "grad_norm": 1.3614004850387573,
      "learning_rate": 8.634349411233366e-06,
      "loss": 0.0565,
      "step": 207675
    },
    {
      "epoch": 3.0588651124431157,
      "grad_norm": 1.1091094017028809,
      "learning_rate": 8.632713039266382e-06,
      "loss": 0.0597,
      "step": 207700
    },
    {
      "epoch": 3.0592332955332027,
      "grad_norm": 0.9271122217178345,
      "learning_rate": 8.631076667299399e-06,
      "loss": 0.0537,
      "step": 207725
    },
    {
      "epoch": 3.0596014786232897,
      "grad_norm": 0.9111393690109253,
      "learning_rate": 8.629440295332413e-06,
      "loss": 0.054,
      "step": 207750
    },
    {
      "epoch": 3.0599696617133767,
      "grad_norm": 1.7573012113571167,
      "learning_rate": 8.62780392336543e-06,
      "loss": 0.054,
      "step": 207775
    },
    {
      "epoch": 3.0603378448034637,
      "grad_norm": 1.514750599861145,
      "learning_rate": 8.626167551398444e-06,
      "loss": 0.057,
      "step": 207800
    },
    {
      "epoch": 3.0607060278935507,
      "grad_norm": 1.525256872177124,
      "learning_rate": 8.62453117943146e-06,
      "loss": 0.0501,
      "step": 207825
    },
    {
      "epoch": 3.0610742109836377,
      "grad_norm": 1.2433980703353882,
      "learning_rate": 8.622894807464474e-06,
      "loss": 0.0526,
      "step": 207850
    },
    {
      "epoch": 3.061442394073725,
      "grad_norm": 0.9487237930297852,
      "learning_rate": 8.62125843549749e-06,
      "loss": 0.0523,
      "step": 207875
    },
    {
      "epoch": 3.061810577163812,
      "grad_norm": 1.2259085178375244,
      "learning_rate": 8.619622063530507e-06,
      "loss": 0.0578,
      "step": 207900
    },
    {
      "epoch": 3.0621787602538992,
      "grad_norm": 1.2869065999984741,
      "learning_rate": 8.617985691563521e-06,
      "loss": 0.0536,
      "step": 207925
    },
    {
      "epoch": 3.0625469433439862,
      "grad_norm": 1.5098904371261597,
      "learning_rate": 8.616349319596537e-06,
      "loss": 0.047,
      "step": 207950
    },
    {
      "epoch": 3.0629151264340733,
      "grad_norm": 1.5204015970230103,
      "learning_rate": 8.614712947629552e-06,
      "loss": 0.056,
      "step": 207975
    },
    {
      "epoch": 3.0632833095241603,
      "grad_norm": 1.3810440301895142,
      "learning_rate": 8.613076575662568e-06,
      "loss": 0.0656,
      "step": 208000
    },
    {
      "epoch": 3.0636514926142473,
      "grad_norm": 1.6976685523986816,
      "learning_rate": 8.611440203695584e-06,
      "loss": 0.0576,
      "step": 208025
    },
    {
      "epoch": 3.0640196757043343,
      "grad_norm": 1.728892207145691,
      "learning_rate": 8.609803831728598e-06,
      "loss": 0.0496,
      "step": 208050
    },
    {
      "epoch": 3.0643878587944213,
      "grad_norm": 1.0475125312805176,
      "learning_rate": 8.608167459761615e-06,
      "loss": 0.0546,
      "step": 208075
    },
    {
      "epoch": 3.0647560418845083,
      "grad_norm": 1.3587110042572021,
      "learning_rate": 8.606531087794629e-06,
      "loss": 0.0632,
      "step": 208100
    },
    {
      "epoch": 3.0651242249745954,
      "grad_norm": 1.8400919437408447,
      "learning_rate": 8.604894715827645e-06,
      "loss": 0.0621,
      "step": 208125
    },
    {
      "epoch": 3.0654924080646824,
      "grad_norm": 1.7645920515060425,
      "learning_rate": 8.60325834386066e-06,
      "loss": 0.0601,
      "step": 208150
    },
    {
      "epoch": 3.0658605911547694,
      "grad_norm": 1.384613037109375,
      "learning_rate": 8.601621971893676e-06,
      "loss": 0.054,
      "step": 208175
    },
    {
      "epoch": 3.0662287742448564,
      "grad_norm": 1.311444878578186,
      "learning_rate": 8.599985599926692e-06,
      "loss": 0.0647,
      "step": 208200
    },
    {
      "epoch": 3.0665969573349434,
      "grad_norm": 1.5473036766052246,
      "learning_rate": 8.598349227959706e-06,
      "loss": 0.061,
      "step": 208225
    },
    {
      "epoch": 3.0669651404250304,
      "grad_norm": 1.3374066352844238,
      "learning_rate": 8.596712855992723e-06,
      "loss": 0.0517,
      "step": 208250
    },
    {
      "epoch": 3.0673333235151174,
      "grad_norm": 1.1354910135269165,
      "learning_rate": 8.595076484025737e-06,
      "loss": 0.0548,
      "step": 208275
    },
    {
      "epoch": 3.0677015066052045,
      "grad_norm": 1.0074881315231323,
      "learning_rate": 8.593440112058753e-06,
      "loss": 0.0545,
      "step": 208300
    },
    {
      "epoch": 3.0680696896952915,
      "grad_norm": 1.469630241394043,
      "learning_rate": 8.59180374009177e-06,
      "loss": 0.0557,
      "step": 208325
    },
    {
      "epoch": 3.0684378727853785,
      "grad_norm": 1.3384222984313965,
      "learning_rate": 8.590167368124784e-06,
      "loss": 0.0515,
      "step": 208350
    },
    {
      "epoch": 3.068806055875466,
      "grad_norm": 1.2757782936096191,
      "learning_rate": 8.588530996157798e-06,
      "loss": 0.0561,
      "step": 208375
    },
    {
      "epoch": 3.069174238965553,
      "grad_norm": 1.3328989744186401,
      "learning_rate": 8.586894624190814e-06,
      "loss": 0.0518,
      "step": 208400
    },
    {
      "epoch": 3.06954242205564,
      "grad_norm": 0.8638681173324585,
      "learning_rate": 8.58525825222383e-06,
      "loss": 0.052,
      "step": 208425
    },
    {
      "epoch": 3.069910605145727,
      "grad_norm": 1.485156536102295,
      "learning_rate": 8.583621880256847e-06,
      "loss": 0.0546,
      "step": 208450
    },
    {
      "epoch": 3.070278788235814,
      "grad_norm": 1.0470079183578491,
      "learning_rate": 8.581985508289861e-06,
      "loss": 0.0529,
      "step": 208475
    },
    {
      "epoch": 3.070646971325901,
      "grad_norm": 1.503570795059204,
      "learning_rate": 8.580349136322876e-06,
      "loss": 0.0565,
      "step": 208500
    },
    {
      "epoch": 3.071015154415988,
      "grad_norm": 1.1559982299804688,
      "learning_rate": 8.578712764355892e-06,
      "loss": 0.057,
      "step": 208525
    },
    {
      "epoch": 3.071383337506075,
      "grad_norm": 1.3871711492538452,
      "learning_rate": 8.577076392388908e-06,
      "loss": 0.0565,
      "step": 208550
    },
    {
      "epoch": 3.071751520596162,
      "grad_norm": 1.815867304801941,
      "learning_rate": 8.575440020421922e-06,
      "loss": 0.0525,
      "step": 208575
    },
    {
      "epoch": 3.072119703686249,
      "grad_norm": 1.232560396194458,
      "learning_rate": 8.573803648454939e-06,
      "loss": 0.054,
      "step": 208600
    },
    {
      "epoch": 3.072487886776336,
      "grad_norm": 1.6233042478561401,
      "learning_rate": 8.572167276487953e-06,
      "loss": 0.0485,
      "step": 208625
    },
    {
      "epoch": 3.072856069866423,
      "grad_norm": 1.4989091157913208,
      "learning_rate": 8.570530904520969e-06,
      "loss": 0.0633,
      "step": 208650
    },
    {
      "epoch": 3.07322425295651,
      "grad_norm": 1.7640851736068726,
      "learning_rate": 8.568894532553984e-06,
      "loss": 0.0558,
      "step": 208675
    },
    {
      "epoch": 3.073592436046597,
      "grad_norm": 1.2119972705841064,
      "learning_rate": 8.567258160587e-06,
      "loss": 0.0493,
      "step": 208700
    },
    {
      "epoch": 3.073960619136684,
      "grad_norm": 1.429002046585083,
      "learning_rate": 8.565621788620016e-06,
      "loss": 0.0583,
      "step": 208725
    },
    {
      "epoch": 3.074328802226771,
      "grad_norm": 1.5290645360946655,
      "learning_rate": 8.56398541665303e-06,
      "loss": 0.0542,
      "step": 208750
    },
    {
      "epoch": 3.074696985316858,
      "grad_norm": 1.0177441835403442,
      "learning_rate": 8.562349044686046e-06,
      "loss": 0.0495,
      "step": 208775
    },
    {
      "epoch": 3.075065168406945,
      "grad_norm": 1.3806568384170532,
      "learning_rate": 8.560712672719061e-06,
      "loss": 0.0545,
      "step": 208800
    },
    {
      "epoch": 3.0754333514970322,
      "grad_norm": 0.9060839414596558,
      "learning_rate": 8.559076300752077e-06,
      "loss": 0.0508,
      "step": 208825
    },
    {
      "epoch": 3.0758015345871197,
      "grad_norm": 1.34939444065094,
      "learning_rate": 8.557439928785093e-06,
      "loss": 0.0624,
      "step": 208850
    },
    {
      "epoch": 3.0761697176772067,
      "grad_norm": 0.9613877534866333,
      "learning_rate": 8.55580355681811e-06,
      "loss": 0.0539,
      "step": 208875
    },
    {
      "epoch": 3.0765379007672937,
      "grad_norm": 1.4004034996032715,
      "learning_rate": 8.554167184851124e-06,
      "loss": 0.0499,
      "step": 208900
    },
    {
      "epoch": 3.0769060838573807,
      "grad_norm": 1.7539432048797607,
      "learning_rate": 8.552530812884138e-06,
      "loss": 0.0573,
      "step": 208925
    },
    {
      "epoch": 3.0772742669474678,
      "grad_norm": 1.5943663120269775,
      "learning_rate": 8.550894440917154e-06,
      "loss": 0.0537,
      "step": 208950
    },
    {
      "epoch": 3.0776424500375548,
      "grad_norm": 1.2340283393859863,
      "learning_rate": 8.54925806895017e-06,
      "loss": 0.0517,
      "step": 208975
    },
    {
      "epoch": 3.078010633127642,
      "grad_norm": 1.14974045753479,
      "learning_rate": 8.547621696983185e-06,
      "loss": 0.0515,
      "step": 209000
    },
    {
      "epoch": 3.078378816217729,
      "grad_norm": 1.496490478515625,
      "learning_rate": 8.545985325016201e-06,
      "loss": 0.0548,
      "step": 209025
    },
    {
      "epoch": 3.078746999307816,
      "grad_norm": 1.1287086009979248,
      "learning_rate": 8.544348953049216e-06,
      "loss": 0.0586,
      "step": 209050
    },
    {
      "epoch": 3.079115182397903,
      "grad_norm": 1.3333474397659302,
      "learning_rate": 8.542712581082232e-06,
      "loss": 0.0548,
      "step": 209075
    },
    {
      "epoch": 3.07948336548799,
      "grad_norm": 1.4309800863265991,
      "learning_rate": 8.541076209115246e-06,
      "loss": 0.0556,
      "step": 209100
    },
    {
      "epoch": 3.079851548578077,
      "grad_norm": 1.2619292736053467,
      "learning_rate": 8.539439837148262e-06,
      "loss": 0.0574,
      "step": 209125
    },
    {
      "epoch": 3.080219731668164,
      "grad_norm": 1.5033754110336304,
      "learning_rate": 8.537803465181279e-06,
      "loss": 0.0534,
      "step": 209150
    },
    {
      "epoch": 3.080587914758251,
      "grad_norm": 1.121462345123291,
      "learning_rate": 8.536167093214293e-06,
      "loss": 0.0618,
      "step": 209175
    },
    {
      "epoch": 3.080956097848338,
      "grad_norm": 1.0048859119415283,
      "learning_rate": 8.53453072124731e-06,
      "loss": 0.0501,
      "step": 209200
    },
    {
      "epoch": 3.081324280938425,
      "grad_norm": 1.148051381111145,
      "learning_rate": 8.532894349280324e-06,
      "loss": 0.048,
      "step": 209225
    },
    {
      "epoch": 3.081692464028512,
      "grad_norm": 1.5723342895507812,
      "learning_rate": 8.53125797731334e-06,
      "loss": 0.0574,
      "step": 209250
    },
    {
      "epoch": 3.082060647118599,
      "grad_norm": 0.7662875056266785,
      "learning_rate": 8.529621605346356e-06,
      "loss": 0.0475,
      "step": 209275
    },
    {
      "epoch": 3.082428830208686,
      "grad_norm": 0.8803878426551819,
      "learning_rate": 8.52798523337937e-06,
      "loss": 0.0483,
      "step": 209300
    },
    {
      "epoch": 3.0827970132987734,
      "grad_norm": 0.956765353679657,
      "learning_rate": 8.526348861412385e-06,
      "loss": 0.0566,
      "step": 209325
    },
    {
      "epoch": 3.0831651963888604,
      "grad_norm": 0.9687982797622681,
      "learning_rate": 8.524712489445401e-06,
      "loss": 0.0548,
      "step": 209350
    },
    {
      "epoch": 3.0835333794789475,
      "grad_norm": 1.46809720993042,
      "learning_rate": 8.523076117478417e-06,
      "loss": 0.0574,
      "step": 209375
    },
    {
      "epoch": 3.0839015625690345,
      "grad_norm": 1.3305907249450684,
      "learning_rate": 8.521439745511433e-06,
      "loss": 0.0555,
      "step": 209400
    },
    {
      "epoch": 3.0842697456591215,
      "grad_norm": 1.184473991394043,
      "learning_rate": 8.519803373544448e-06,
      "loss": 0.0628,
      "step": 209425
    },
    {
      "epoch": 3.0846379287492085,
      "grad_norm": 1.128836750984192,
      "learning_rate": 8.518167001577464e-06,
      "loss": 0.055,
      "step": 209450
    },
    {
      "epoch": 3.0850061118392955,
      "grad_norm": 0.7413156032562256,
      "learning_rate": 8.516530629610478e-06,
      "loss": 0.0486,
      "step": 209475
    },
    {
      "epoch": 3.0853742949293825,
      "grad_norm": 1.2718126773834229,
      "learning_rate": 8.514894257643495e-06,
      "loss": 0.0625,
      "step": 209500
    },
    {
      "epoch": 3.0857424780194695,
      "grad_norm": 1.653568148612976,
      "learning_rate": 8.513257885676509e-06,
      "loss": 0.0579,
      "step": 209525
    },
    {
      "epoch": 3.0861106611095566,
      "grad_norm": 1.485002875328064,
      "learning_rate": 8.511621513709525e-06,
      "loss": 0.0569,
      "step": 209550
    },
    {
      "epoch": 3.0864788441996436,
      "grad_norm": 1.282753586769104,
      "learning_rate": 8.509985141742541e-06,
      "loss": 0.0599,
      "step": 209575
    },
    {
      "epoch": 3.0868470272897306,
      "grad_norm": 1.3891710042953491,
      "learning_rate": 8.508348769775556e-06,
      "loss": 0.0532,
      "step": 209600
    },
    {
      "epoch": 3.0872152103798176,
      "grad_norm": 1.182939052581787,
      "learning_rate": 8.50671239780857e-06,
      "loss": 0.0592,
      "step": 209625
    },
    {
      "epoch": 3.0875833934699046,
      "grad_norm": 1.310529351234436,
      "learning_rate": 8.505076025841586e-06,
      "loss": 0.057,
      "step": 209650
    },
    {
      "epoch": 3.0879515765599916,
      "grad_norm": 1.4837828874588013,
      "learning_rate": 8.503439653874603e-06,
      "loss": 0.0526,
      "step": 209675
    },
    {
      "epoch": 3.0883197596500787,
      "grad_norm": 1.3674213886260986,
      "learning_rate": 8.501803281907619e-06,
      "loss": 0.0565,
      "step": 209700
    },
    {
      "epoch": 3.0886879427401657,
      "grad_norm": 2.05519962310791,
      "learning_rate": 8.500166909940633e-06,
      "loss": 0.0582,
      "step": 209725
    },
    {
      "epoch": 3.0890561258302527,
      "grad_norm": 1.9169014692306519,
      "learning_rate": 8.498530537973648e-06,
      "loss": 0.0563,
      "step": 209750
    },
    {
      "epoch": 3.0894243089203397,
      "grad_norm": 1.029412031173706,
      "learning_rate": 8.496894166006664e-06,
      "loss": 0.0565,
      "step": 209775
    },
    {
      "epoch": 3.089792492010427,
      "grad_norm": 1.0069315433502197,
      "learning_rate": 8.49525779403968e-06,
      "loss": 0.062,
      "step": 209800
    },
    {
      "epoch": 3.090160675100514,
      "grad_norm": 1.0237654447555542,
      "learning_rate": 8.493621422072696e-06,
      "loss": 0.0544,
      "step": 209825
    },
    {
      "epoch": 3.090528858190601,
      "grad_norm": 1.139911413192749,
      "learning_rate": 8.49198505010571e-06,
      "loss": 0.0542,
      "step": 209850
    },
    {
      "epoch": 3.090897041280688,
      "grad_norm": 1.4588353633880615,
      "learning_rate": 8.490348678138725e-06,
      "loss": 0.0542,
      "step": 209875
    },
    {
      "epoch": 3.091265224370775,
      "grad_norm": 1.0450915098190308,
      "learning_rate": 8.488712306171741e-06,
      "loss": 0.0559,
      "step": 209900
    },
    {
      "epoch": 3.0916334074608622,
      "grad_norm": 1.5998412370681763,
      "learning_rate": 8.487075934204757e-06,
      "loss": 0.0517,
      "step": 209925
    },
    {
      "epoch": 3.0920015905509493,
      "grad_norm": 1.2449195384979248,
      "learning_rate": 8.485439562237772e-06,
      "loss": 0.0509,
      "step": 209950
    },
    {
      "epoch": 3.0923697736410363,
      "grad_norm": 2.1324307918548584,
      "learning_rate": 8.483803190270788e-06,
      "loss": 0.0577,
      "step": 209975
    },
    {
      "epoch": 3.0927379567311233,
      "grad_norm": 1.532668113708496,
      "learning_rate": 8.482166818303802e-06,
      "loss": 0.0591,
      "step": 210000
    },
    {
      "epoch": 3.0931061398212103,
      "grad_norm": 1.3548530340194702,
      "learning_rate": 8.480530446336818e-06,
      "loss": 0.0496,
      "step": 210025
    },
    {
      "epoch": 3.0934743229112973,
      "grad_norm": 0.7626944780349731,
      "learning_rate": 8.478894074369833e-06,
      "loss": 0.0537,
      "step": 210050
    },
    {
      "epoch": 3.0938425060013843,
      "grad_norm": 1.320664644241333,
      "learning_rate": 8.477257702402849e-06,
      "loss": 0.0529,
      "step": 210075
    },
    {
      "epoch": 3.0942106890914713,
      "grad_norm": 2.0523860454559326,
      "learning_rate": 8.475621330435865e-06,
      "loss": 0.0531,
      "step": 210100
    },
    {
      "epoch": 3.0945788721815584,
      "grad_norm": 1.1885137557983398,
      "learning_rate": 8.47398495846888e-06,
      "loss": 0.0482,
      "step": 210125
    },
    {
      "epoch": 3.0949470552716454,
      "grad_norm": 1.2213784456253052,
      "learning_rate": 8.472414041380575e-06,
      "loss": 0.0639,
      "step": 210150
    },
    {
      "epoch": 3.0953152383617324,
      "grad_norm": 1.1088509559631348,
      "learning_rate": 8.47077766941359e-06,
      "loss": 0.0543,
      "step": 210175
    },
    {
      "epoch": 3.0956834214518194,
      "grad_norm": 1.1907039880752563,
      "learning_rate": 8.469141297446606e-06,
      "loss": 0.0512,
      "step": 210200
    },
    {
      "epoch": 3.0960516045419064,
      "grad_norm": 1.47086501121521,
      "learning_rate": 8.467504925479622e-06,
      "loss": 0.0511,
      "step": 210225
    },
    {
      "epoch": 3.0964197876319934,
      "grad_norm": 1.4203814268112183,
      "learning_rate": 8.465868553512636e-06,
      "loss": 0.0519,
      "step": 210250
    },
    {
      "epoch": 3.0967879707220805,
      "grad_norm": 1.4843871593475342,
      "learning_rate": 8.464232181545653e-06,
      "loss": 0.0471,
      "step": 210275
    },
    {
      "epoch": 3.097156153812168,
      "grad_norm": 1.4037944078445435,
      "learning_rate": 8.462595809578667e-06,
      "loss": 0.0571,
      "step": 210300
    },
    {
      "epoch": 3.097524336902255,
      "grad_norm": 1.4150410890579224,
      "learning_rate": 8.460959437611683e-06,
      "loss": 0.0548,
      "step": 210325
    },
    {
      "epoch": 3.097892519992342,
      "grad_norm": 1.319433569908142,
      "learning_rate": 8.4593230656447e-06,
      "loss": 0.0577,
      "step": 210350
    },
    {
      "epoch": 3.098260703082429,
      "grad_norm": 1.613512396812439,
      "learning_rate": 8.457686693677714e-06,
      "loss": 0.0595,
      "step": 210375
    },
    {
      "epoch": 3.098628886172516,
      "grad_norm": 0.9161540269851685,
      "learning_rate": 8.45605032171073e-06,
      "loss": 0.0516,
      "step": 210400
    },
    {
      "epoch": 3.098997069262603,
      "grad_norm": 1.211728811264038,
      "learning_rate": 8.454413949743744e-06,
      "loss": 0.0539,
      "step": 210425
    },
    {
      "epoch": 3.09936525235269,
      "grad_norm": 1.1488088369369507,
      "learning_rate": 8.45277757777676e-06,
      "loss": 0.0641,
      "step": 210450
    },
    {
      "epoch": 3.099733435442777,
      "grad_norm": 1.1961686611175537,
      "learning_rate": 8.451141205809775e-06,
      "loss": 0.0552,
      "step": 210475
    },
    {
      "epoch": 3.100101618532864,
      "grad_norm": 0.6367381811141968,
      "learning_rate": 8.449504833842791e-06,
      "loss": 0.0547,
      "step": 210500
    },
    {
      "epoch": 3.100469801622951,
      "grad_norm": 1.1959617137908936,
      "learning_rate": 8.447868461875807e-06,
      "loss": 0.0534,
      "step": 210525
    },
    {
      "epoch": 3.100837984713038,
      "grad_norm": 1.5913487672805786,
      "learning_rate": 8.446232089908822e-06,
      "loss": 0.0577,
      "step": 210550
    },
    {
      "epoch": 3.101206167803125,
      "grad_norm": 1.4227579832077026,
      "learning_rate": 8.444595717941836e-06,
      "loss": 0.0524,
      "step": 210575
    },
    {
      "epoch": 3.101574350893212,
      "grad_norm": 1.454413652420044,
      "learning_rate": 8.442959345974852e-06,
      "loss": 0.0584,
      "step": 210600
    },
    {
      "epoch": 3.101942533983299,
      "grad_norm": 0.9828397631645203,
      "learning_rate": 8.441322974007869e-06,
      "loss": 0.0547,
      "step": 210625
    },
    {
      "epoch": 3.102310717073386,
      "grad_norm": 1.0633606910705566,
      "learning_rate": 8.439686602040885e-06,
      "loss": 0.0526,
      "step": 210650
    },
    {
      "epoch": 3.102678900163473,
      "grad_norm": 1.8901829719543457,
      "learning_rate": 8.4380502300739e-06,
      "loss": 0.0582,
      "step": 210675
    },
    {
      "epoch": 3.10304708325356,
      "grad_norm": 1.520228385925293,
      "learning_rate": 8.436413858106914e-06,
      "loss": 0.0488,
      "step": 210700
    },
    {
      "epoch": 3.103415266343647,
      "grad_norm": 1.1276758909225464,
      "learning_rate": 8.43477748613993e-06,
      "loss": 0.0621,
      "step": 210725
    },
    {
      "epoch": 3.1037834494337346,
      "grad_norm": 0.8954167366027832,
      "learning_rate": 8.433141114172946e-06,
      "loss": 0.0591,
      "step": 210750
    },
    {
      "epoch": 3.1041516325238216,
      "grad_norm": 1.574902057647705,
      "learning_rate": 8.431504742205962e-06,
      "loss": 0.0574,
      "step": 210775
    },
    {
      "epoch": 3.1045198156139087,
      "grad_norm": 1.3450320959091187,
      "learning_rate": 8.429868370238977e-06,
      "loss": 0.0593,
      "step": 210800
    },
    {
      "epoch": 3.1048879987039957,
      "grad_norm": 1.4459121227264404,
      "learning_rate": 8.428231998271991e-06,
      "loss": 0.0454,
      "step": 210825
    },
    {
      "epoch": 3.1052561817940827,
      "grad_norm": 1.0224316120147705,
      "learning_rate": 8.426595626305007e-06,
      "loss": 0.0522,
      "step": 210850
    },
    {
      "epoch": 3.1056243648841697,
      "grad_norm": 1.315860629081726,
      "learning_rate": 8.424959254338023e-06,
      "loss": 0.0568,
      "step": 210875
    },
    {
      "epoch": 3.1059925479742567,
      "grad_norm": 0.7082265019416809,
      "learning_rate": 8.423322882371038e-06,
      "loss": 0.0608,
      "step": 210900
    },
    {
      "epoch": 3.1063607310643437,
      "grad_norm": 1.3857876062393188,
      "learning_rate": 8.421686510404054e-06,
      "loss": 0.0495,
      "step": 210925
    },
    {
      "epoch": 3.1067289141544308,
      "grad_norm": 1.9409323930740356,
      "learning_rate": 8.420050138437068e-06,
      "loss": 0.0554,
      "step": 210950
    },
    {
      "epoch": 3.1070970972445178,
      "grad_norm": 1.4113516807556152,
      "learning_rate": 8.418413766470085e-06,
      "loss": 0.052,
      "step": 210975
    },
    {
      "epoch": 3.107465280334605,
      "grad_norm": 1.1748491525650024,
      "learning_rate": 8.416777394503099e-06,
      "loss": 0.0519,
      "step": 211000
    },
    {
      "epoch": 3.107833463424692,
      "grad_norm": 1.3287670612335205,
      "learning_rate": 8.415141022536115e-06,
      "loss": 0.0539,
      "step": 211025
    },
    {
      "epoch": 3.108201646514779,
      "grad_norm": 1.2790247201919556,
      "learning_rate": 8.413504650569131e-06,
      "loss": 0.052,
      "step": 211050
    },
    {
      "epoch": 3.108569829604866,
      "grad_norm": 1.527503252029419,
      "learning_rate": 8.411868278602146e-06,
      "loss": 0.0527,
      "step": 211075
    },
    {
      "epoch": 3.108938012694953,
      "grad_norm": 1.5122714042663574,
      "learning_rate": 8.410231906635162e-06,
      "loss": 0.0568,
      "step": 211100
    },
    {
      "epoch": 3.10930619578504,
      "grad_norm": 1.2688982486724854,
      "learning_rate": 8.408595534668176e-06,
      "loss": 0.0517,
      "step": 211125
    },
    {
      "epoch": 3.109674378875127,
      "grad_norm": 0.7769589424133301,
      "learning_rate": 8.406959162701193e-06,
      "loss": 0.0525,
      "step": 211150
    },
    {
      "epoch": 3.110042561965214,
      "grad_norm": 1.918821096420288,
      "learning_rate": 8.405322790734209e-06,
      "loss": 0.0649,
      "step": 211175
    },
    {
      "epoch": 3.110410745055301,
      "grad_norm": 1.1520462036132812,
      "learning_rate": 8.403686418767225e-06,
      "loss": 0.0498,
      "step": 211200
    },
    {
      "epoch": 3.110778928145388,
      "grad_norm": 1.44691002368927,
      "learning_rate": 8.40205004680024e-06,
      "loss": 0.0519,
      "step": 211225
    },
    {
      "epoch": 3.1111471112354754,
      "grad_norm": 1.6217703819274902,
      "learning_rate": 8.400413674833254e-06,
      "loss": 0.053,
      "step": 211250
    },
    {
      "epoch": 3.1115152943255624,
      "grad_norm": 1.853597640991211,
      "learning_rate": 8.39877730286627e-06,
      "loss": 0.0582,
      "step": 211275
    },
    {
      "epoch": 3.1118834774156494,
      "grad_norm": 0.60493004322052,
      "learning_rate": 8.397140930899286e-06,
      "loss": 0.0543,
      "step": 211300
    },
    {
      "epoch": 3.1122516605057364,
      "grad_norm": 1.6704847812652588,
      "learning_rate": 8.3955045589323e-06,
      "loss": 0.0631,
      "step": 211325
    },
    {
      "epoch": 3.1126198435958234,
      "grad_norm": 1.2225840091705322,
      "learning_rate": 8.393868186965317e-06,
      "loss": 0.0611,
      "step": 211350
    },
    {
      "epoch": 3.1129880266859105,
      "grad_norm": 1.5563147068023682,
      "learning_rate": 8.392231814998331e-06,
      "loss": 0.0493,
      "step": 211375
    },
    {
      "epoch": 3.1133562097759975,
      "grad_norm": 1.3426176309585571,
      "learning_rate": 8.390595443031347e-06,
      "loss": 0.0558,
      "step": 211400
    },
    {
      "epoch": 3.1137243928660845,
      "grad_norm": 1.3655571937561035,
      "learning_rate": 8.388959071064362e-06,
      "loss": 0.0581,
      "step": 211425
    },
    {
      "epoch": 3.1140925759561715,
      "grad_norm": 1.1556378602981567,
      "learning_rate": 8.387322699097378e-06,
      "loss": 0.0551,
      "step": 211450
    },
    {
      "epoch": 3.1144607590462585,
      "grad_norm": 0.9666590094566345,
      "learning_rate": 8.385686327130394e-06,
      "loss": 0.0511,
      "step": 211475
    },
    {
      "epoch": 3.1148289421363455,
      "grad_norm": 1.3704558610916138,
      "learning_rate": 8.384049955163408e-06,
      "loss": 0.0543,
      "step": 211500
    },
    {
      "epoch": 3.1151971252264326,
      "grad_norm": 1.1888535022735596,
      "learning_rate": 8.382413583196423e-06,
      "loss": 0.0578,
      "step": 211525
    },
    {
      "epoch": 3.1155653083165196,
      "grad_norm": 1.8262602090835571,
      "learning_rate": 8.380777211229439e-06,
      "loss": 0.0535,
      "step": 211550
    },
    {
      "epoch": 3.1159334914066066,
      "grad_norm": 1.3509559631347656,
      "learning_rate": 8.379140839262455e-06,
      "loss": 0.0556,
      "step": 211575
    },
    {
      "epoch": 3.1163016744966936,
      "grad_norm": 1.738990068435669,
      "learning_rate": 8.377504467295471e-06,
      "loss": 0.0557,
      "step": 211600
    },
    {
      "epoch": 3.1166698575867806,
      "grad_norm": 1.7641172409057617,
      "learning_rate": 8.375868095328486e-06,
      "loss": 0.0554,
      "step": 211625
    },
    {
      "epoch": 3.1170380406768676,
      "grad_norm": 1.572106122970581,
      "learning_rate": 8.3742317233615e-06,
      "loss": 0.0591,
      "step": 211650
    },
    {
      "epoch": 3.1174062237669546,
      "grad_norm": 1.3219088315963745,
      "learning_rate": 8.372595351394516e-06,
      "loss": 0.0557,
      "step": 211675
    },
    {
      "epoch": 3.1177744068570417,
      "grad_norm": 1.7581079006195068,
      "learning_rate": 8.370958979427533e-06,
      "loss": 0.0602,
      "step": 211700
    },
    {
      "epoch": 3.118142589947129,
      "grad_norm": 1.3362586498260498,
      "learning_rate": 8.369322607460549e-06,
      "loss": 0.0508,
      "step": 211725
    },
    {
      "epoch": 3.118510773037216,
      "grad_norm": 1.5656627416610718,
      "learning_rate": 8.367686235493563e-06,
      "loss": 0.059,
      "step": 211750
    },
    {
      "epoch": 3.118878956127303,
      "grad_norm": 1.342461109161377,
      "learning_rate": 8.36604986352658e-06,
      "loss": 0.058,
      "step": 211775
    },
    {
      "epoch": 3.11924713921739,
      "grad_norm": 1.5516109466552734,
      "learning_rate": 8.364413491559594e-06,
      "loss": 0.0573,
      "step": 211800
    },
    {
      "epoch": 3.119615322307477,
      "grad_norm": 1.3761303424835205,
      "learning_rate": 8.36277711959261e-06,
      "loss": 0.0581,
      "step": 211825
    },
    {
      "epoch": 3.119983505397564,
      "grad_norm": 1.4315694570541382,
      "learning_rate": 8.361140747625624e-06,
      "loss": 0.0587,
      "step": 211850
    },
    {
      "epoch": 3.120351688487651,
      "grad_norm": 1.0159449577331543,
      "learning_rate": 8.35950437565864e-06,
      "loss": 0.0522,
      "step": 211875
    },
    {
      "epoch": 3.1207198715777382,
      "grad_norm": 1.3247125148773193,
      "learning_rate": 8.357868003691657e-06,
      "loss": 0.0552,
      "step": 211900
    },
    {
      "epoch": 3.1210880546678252,
      "grad_norm": 0.9327214360237122,
      "learning_rate": 8.356231631724671e-06,
      "loss": 0.0531,
      "step": 211925
    },
    {
      "epoch": 3.1214562377579123,
      "grad_norm": 1.2175869941711426,
      "learning_rate": 8.354595259757686e-06,
      "loss": 0.0562,
      "step": 211950
    },
    {
      "epoch": 3.1218244208479993,
      "grad_norm": 1.0258172750473022,
      "learning_rate": 8.352958887790702e-06,
      "loss": 0.0558,
      "step": 211975
    },
    {
      "epoch": 3.1221926039380863,
      "grad_norm": 1.298048973083496,
      "learning_rate": 8.351322515823718e-06,
      "loss": 0.047,
      "step": 212000
    },
    {
      "epoch": 3.1225607870281733,
      "grad_norm": 1.1345291137695312,
      "learning_rate": 8.349686143856734e-06,
      "loss": 0.0568,
      "step": 212025
    },
    {
      "epoch": 3.1229289701182603,
      "grad_norm": 1.2308984994888306,
      "learning_rate": 8.348049771889749e-06,
      "loss": 0.0558,
      "step": 212050
    },
    {
      "epoch": 3.1232971532083473,
      "grad_norm": 1.4277386665344238,
      "learning_rate": 8.346413399922763e-06,
      "loss": 0.0511,
      "step": 212075
    },
    {
      "epoch": 3.1236653362984343,
      "grad_norm": 1.3885278701782227,
      "learning_rate": 8.34477702795578e-06,
      "loss": 0.0613,
      "step": 212100
    },
    {
      "epoch": 3.1240335193885214,
      "grad_norm": 1.5365197658538818,
      "learning_rate": 8.343140655988795e-06,
      "loss": 0.0627,
      "step": 212125
    },
    {
      "epoch": 3.1244017024786084,
      "grad_norm": 1.2687935829162598,
      "learning_rate": 8.341504284021811e-06,
      "loss": 0.0484,
      "step": 212150
    },
    {
      "epoch": 3.1247698855686954,
      "grad_norm": 1.5429891347885132,
      "learning_rate": 8.339867912054826e-06,
      "loss": 0.0558,
      "step": 212175
    },
    {
      "epoch": 3.1251380686587824,
      "grad_norm": 1.4517898559570312,
      "learning_rate": 8.33823154008784e-06,
      "loss": 0.0509,
      "step": 212200
    },
    {
      "epoch": 3.12550625174887,
      "grad_norm": 1.2259372472763062,
      "learning_rate": 8.336595168120857e-06,
      "loss": 0.0584,
      "step": 212225
    },
    {
      "epoch": 3.125874434838957,
      "grad_norm": 1.4298137426376343,
      "learning_rate": 8.334958796153873e-06,
      "loss": 0.0597,
      "step": 212250
    },
    {
      "epoch": 3.126242617929044,
      "grad_norm": 0.9515334963798523,
      "learning_rate": 8.333322424186887e-06,
      "loss": 0.0475,
      "step": 212275
    },
    {
      "epoch": 3.126610801019131,
      "grad_norm": 1.2227177619934082,
      "learning_rate": 8.331686052219903e-06,
      "loss": 0.0557,
      "step": 212300
    },
    {
      "epoch": 3.126978984109218,
      "grad_norm": 1.1603729724884033,
      "learning_rate": 8.330049680252918e-06,
      "loss": 0.0525,
      "step": 212325
    },
    {
      "epoch": 3.127347167199305,
      "grad_norm": 1.310678243637085,
      "learning_rate": 8.328413308285934e-06,
      "loss": 0.0497,
      "step": 212350
    },
    {
      "epoch": 3.127715350289392,
      "grad_norm": 1.4992417097091675,
      "learning_rate": 8.326776936318948e-06,
      "loss": 0.0495,
      "step": 212375
    },
    {
      "epoch": 3.128083533379479,
      "grad_norm": 1.0512217283248901,
      "learning_rate": 8.325140564351965e-06,
      "loss": 0.06,
      "step": 212400
    },
    {
      "epoch": 3.128451716469566,
      "grad_norm": 1.5721431970596313,
      "learning_rate": 8.32350419238498e-06,
      "loss": 0.056,
      "step": 212425
    },
    {
      "epoch": 3.128819899559653,
      "grad_norm": 0.7191261649131775,
      "learning_rate": 8.321867820417995e-06,
      "loss": 0.049,
      "step": 212450
    },
    {
      "epoch": 3.12918808264974,
      "grad_norm": 1.365514874458313,
      "learning_rate": 8.320231448451011e-06,
      "loss": 0.057,
      "step": 212475
    },
    {
      "epoch": 3.129556265739827,
      "grad_norm": 1.22919499874115,
      "learning_rate": 8.318595076484026e-06,
      "loss": 0.0594,
      "step": 212500
    },
    {
      "epoch": 3.129924448829914,
      "grad_norm": 1.4038081169128418,
      "learning_rate": 8.316958704517042e-06,
      "loss": 0.0547,
      "step": 212525
    },
    {
      "epoch": 3.130292631920001,
      "grad_norm": 1.0304944515228271,
      "learning_rate": 8.315322332550058e-06,
      "loss": 0.056,
      "step": 212550
    },
    {
      "epoch": 3.130660815010088,
      "grad_norm": 1.1844600439071655,
      "learning_rate": 8.313685960583073e-06,
      "loss": 0.0554,
      "step": 212575
    },
    {
      "epoch": 3.131028998100175,
      "grad_norm": 1.4965578317642212,
      "learning_rate": 8.312115043494768e-06,
      "loss": 0.0565,
      "step": 212600
    },
    {
      "epoch": 3.131397181190262,
      "grad_norm": 1.496036410331726,
      "learning_rate": 8.310478671527783e-06,
      "loss": 0.0581,
      "step": 212625
    },
    {
      "epoch": 3.131765364280349,
      "grad_norm": 1.601948857307434,
      "learning_rate": 8.308842299560799e-06,
      "loss": 0.0638,
      "step": 212650
    },
    {
      "epoch": 3.1321335473704366,
      "grad_norm": 0.8362765908241272,
      "learning_rate": 8.307205927593815e-06,
      "loss": 0.046,
      "step": 212675
    },
    {
      "epoch": 3.1325017304605236,
      "grad_norm": 1.1488044261932373,
      "learning_rate": 8.30556955562683e-06,
      "loss": 0.0548,
      "step": 212700
    },
    {
      "epoch": 3.1328699135506106,
      "grad_norm": 1.212314486503601,
      "learning_rate": 8.303933183659845e-06,
      "loss": 0.0532,
      "step": 212725
    },
    {
      "epoch": 3.1332380966406976,
      "grad_norm": 1.6238683462142944,
      "learning_rate": 8.30229681169286e-06,
      "loss": 0.0572,
      "step": 212750
    },
    {
      "epoch": 3.1336062797307846,
      "grad_norm": 1.1864604949951172,
      "learning_rate": 8.300660439725876e-06,
      "loss": 0.0611,
      "step": 212775
    },
    {
      "epoch": 3.1339744628208717,
      "grad_norm": 1.8363687992095947,
      "learning_rate": 8.29902406775889e-06,
      "loss": 0.0588,
      "step": 212800
    },
    {
      "epoch": 3.1343426459109587,
      "grad_norm": 1.286948561668396,
      "learning_rate": 8.297387695791907e-06,
      "loss": 0.0578,
      "step": 212825
    },
    {
      "epoch": 3.1347108290010457,
      "grad_norm": 1.3574045896530151,
      "learning_rate": 8.295751323824923e-06,
      "loss": 0.05,
      "step": 212850
    },
    {
      "epoch": 3.1350790120911327,
      "grad_norm": 1.433182716369629,
      "learning_rate": 8.294114951857937e-06,
      "loss": 0.0588,
      "step": 212875
    },
    {
      "epoch": 3.1354471951812197,
      "grad_norm": 1.83694589138031,
      "learning_rate": 8.292478579890952e-06,
      "loss": 0.0622,
      "step": 212900
    },
    {
      "epoch": 3.1358153782713067,
      "grad_norm": 1.7200795412063599,
      "learning_rate": 8.290842207923968e-06,
      "loss": 0.0576,
      "step": 212925
    },
    {
      "epoch": 3.1361835613613938,
      "grad_norm": 1.1987969875335693,
      "learning_rate": 8.289205835956984e-06,
      "loss": 0.0565,
      "step": 212950
    },
    {
      "epoch": 3.1365517444514808,
      "grad_norm": 1.3952667713165283,
      "learning_rate": 8.28756946399e-06,
      "loss": 0.0581,
      "step": 212975
    },
    {
      "epoch": 3.136919927541568,
      "grad_norm": 1.2205032110214233,
      "learning_rate": 8.285933092023015e-06,
      "loss": 0.0637,
      "step": 213000
    },
    {
      "epoch": 3.137288110631655,
      "grad_norm": 1.66941499710083,
      "learning_rate": 8.284296720056029e-06,
      "loss": 0.0553,
      "step": 213025
    },
    {
      "epoch": 3.137656293721742,
      "grad_norm": 1.975106954574585,
      "learning_rate": 8.282660348089045e-06,
      "loss": 0.0505,
      "step": 213050
    },
    {
      "epoch": 3.138024476811829,
      "grad_norm": 1.55607271194458,
      "learning_rate": 8.281023976122061e-06,
      "loss": 0.0541,
      "step": 213075
    },
    {
      "epoch": 3.138392659901916,
      "grad_norm": 1.6888313293457031,
      "learning_rate": 8.279387604155078e-06,
      "loss": 0.0577,
      "step": 213100
    },
    {
      "epoch": 3.138760842992003,
      "grad_norm": 1.2630295753479004,
      "learning_rate": 8.277751232188092e-06,
      "loss": 0.0628,
      "step": 213125
    },
    {
      "epoch": 3.13912902608209,
      "grad_norm": 1.3446288108825684,
      "learning_rate": 8.276114860221106e-06,
      "loss": 0.0536,
      "step": 213150
    },
    {
      "epoch": 3.1394972091721773,
      "grad_norm": 0.8002458810806274,
      "learning_rate": 8.274478488254123e-06,
      "loss": 0.0549,
      "step": 213175
    },
    {
      "epoch": 3.1398653922622644,
      "grad_norm": 1.483613133430481,
      "learning_rate": 8.272842116287139e-06,
      "loss": 0.0573,
      "step": 213200
    },
    {
      "epoch": 3.1402335753523514,
      "grad_norm": 1.3966134786605835,
      "learning_rate": 8.271205744320153e-06,
      "loss": 0.0576,
      "step": 213225
    },
    {
      "epoch": 3.1406017584424384,
      "grad_norm": 1.3493943214416504,
      "learning_rate": 8.26956937235317e-06,
      "loss": 0.0559,
      "step": 213250
    },
    {
      "epoch": 3.1409699415325254,
      "grad_norm": 1.8125332593917847,
      "learning_rate": 8.267933000386184e-06,
      "loss": 0.0561,
      "step": 213275
    },
    {
      "epoch": 3.1413381246226124,
      "grad_norm": 1.1806167364120483,
      "learning_rate": 8.2662966284192e-06,
      "loss": 0.0573,
      "step": 213300
    },
    {
      "epoch": 3.1417063077126994,
      "grad_norm": 1.107303500175476,
      "learning_rate": 8.264660256452214e-06,
      "loss": 0.057,
      "step": 213325
    },
    {
      "epoch": 3.1420744908027864,
      "grad_norm": 1.1827845573425293,
      "learning_rate": 8.26302388448523e-06,
      "loss": 0.0553,
      "step": 213350
    },
    {
      "epoch": 3.1424426738928735,
      "grad_norm": 1.6559511423110962,
      "learning_rate": 8.261387512518247e-06,
      "loss": 0.0516,
      "step": 213375
    },
    {
      "epoch": 3.1428108569829605,
      "grad_norm": 0.8393713235855103,
      "learning_rate": 8.259751140551261e-06,
      "loss": 0.0527,
      "step": 213400
    },
    {
      "epoch": 3.1431790400730475,
      "grad_norm": 1.2095073461532593,
      "learning_rate": 8.258114768584277e-06,
      "loss": 0.052,
      "step": 213425
    },
    {
      "epoch": 3.1435472231631345,
      "grad_norm": 0.9588323831558228,
      "learning_rate": 8.256478396617292e-06,
      "loss": 0.0555,
      "step": 213450
    },
    {
      "epoch": 3.1439154062532215,
      "grad_norm": 1.6538848876953125,
      "learning_rate": 8.254842024650308e-06,
      "loss": 0.0639,
      "step": 213475
    },
    {
      "epoch": 3.1442835893433085,
      "grad_norm": 0.5957412123680115,
      "learning_rate": 8.253205652683324e-06,
      "loss": 0.0582,
      "step": 213500
    },
    {
      "epoch": 3.1446517724333956,
      "grad_norm": 1.141301155090332,
      "learning_rate": 8.251569280716339e-06,
      "loss": 0.0514,
      "step": 213525
    },
    {
      "epoch": 3.1450199555234826,
      "grad_norm": 1.6773204803466797,
      "learning_rate": 8.249932908749355e-06,
      "loss": 0.0594,
      "step": 213550
    },
    {
      "epoch": 3.1453881386135696,
      "grad_norm": 1.4257261753082275,
      "learning_rate": 8.24829653678237e-06,
      "loss": 0.0525,
      "step": 213575
    },
    {
      "epoch": 3.1457563217036566,
      "grad_norm": 1.009519338607788,
      "learning_rate": 8.246660164815385e-06,
      "loss": 0.0544,
      "step": 213600
    },
    {
      "epoch": 3.146124504793744,
      "grad_norm": 1.6525260210037231,
      "learning_rate": 8.245023792848401e-06,
      "loss": 0.061,
      "step": 213625
    },
    {
      "epoch": 3.146492687883831,
      "grad_norm": 1.2808359861373901,
      "learning_rate": 8.243387420881416e-06,
      "loss": 0.0521,
      "step": 213650
    },
    {
      "epoch": 3.146860870973918,
      "grad_norm": 1.2175997495651245,
      "learning_rate": 8.241751048914432e-06,
      "loss": 0.0582,
      "step": 213675
    },
    {
      "epoch": 3.147229054064005,
      "grad_norm": 1.5359950065612793,
      "learning_rate": 8.240114676947447e-06,
      "loss": 0.0647,
      "step": 213700
    },
    {
      "epoch": 3.147597237154092,
      "grad_norm": 1.3045196533203125,
      "learning_rate": 8.238478304980463e-06,
      "loss": 0.0558,
      "step": 213725
    },
    {
      "epoch": 3.147965420244179,
      "grad_norm": 1.7997868061065674,
      "learning_rate": 8.236841933013477e-06,
      "loss": 0.0585,
      "step": 213750
    },
    {
      "epoch": 3.148333603334266,
      "grad_norm": 1.3765270709991455,
      "learning_rate": 8.235205561046493e-06,
      "loss": 0.063,
      "step": 213775
    },
    {
      "epoch": 3.148701786424353,
      "grad_norm": 1.4333709478378296,
      "learning_rate": 8.23356918907951e-06,
      "loss": 0.0503,
      "step": 213800
    },
    {
      "epoch": 3.14906996951444,
      "grad_norm": 1.555262804031372,
      "learning_rate": 8.231932817112524e-06,
      "loss": 0.0584,
      "step": 213825
    },
    {
      "epoch": 3.149438152604527,
      "grad_norm": 0.9358018636703491,
      "learning_rate": 8.230296445145538e-06,
      "loss": 0.0581,
      "step": 213850
    },
    {
      "epoch": 3.149806335694614,
      "grad_norm": 0.9457089900970459,
      "learning_rate": 8.228660073178555e-06,
      "loss": 0.0563,
      "step": 213875
    },
    {
      "epoch": 3.1501745187847012,
      "grad_norm": 1.402043104171753,
      "learning_rate": 8.22702370121157e-06,
      "loss": 0.053,
      "step": 213900
    },
    {
      "epoch": 3.1505427018747882,
      "grad_norm": 1.2711849212646484,
      "learning_rate": 8.225387329244587e-06,
      "loss": 0.0588,
      "step": 213925
    },
    {
      "epoch": 3.1509108849648753,
      "grad_norm": 1.3364461660385132,
      "learning_rate": 8.223750957277601e-06,
      "loss": 0.059,
      "step": 213950
    },
    {
      "epoch": 3.1512790680549623,
      "grad_norm": 1.0449579954147339,
      "learning_rate": 8.222114585310616e-06,
      "loss": 0.0504,
      "step": 213975
    },
    {
      "epoch": 3.1516472511450493,
      "grad_norm": 1.3899604082107544,
      "learning_rate": 8.220478213343632e-06,
      "loss": 0.0499,
      "step": 214000
    },
    {
      "epoch": 3.1520154342351363,
      "grad_norm": 1.4784462451934814,
      "learning_rate": 8.218841841376648e-06,
      "loss": 0.0611,
      "step": 214025
    },
    {
      "epoch": 3.1523836173252233,
      "grad_norm": 0.9336026906967163,
      "learning_rate": 8.217205469409663e-06,
      "loss": 0.054,
      "step": 214050
    },
    {
      "epoch": 3.1527518004153103,
      "grad_norm": 1.2687602043151855,
      "learning_rate": 8.215569097442679e-06,
      "loss": 0.0548,
      "step": 214075
    },
    {
      "epoch": 3.1531199835053974,
      "grad_norm": 1.118599534034729,
      "learning_rate": 8.213932725475695e-06,
      "loss": 0.0546,
      "step": 214100
    },
    {
      "epoch": 3.153488166595485,
      "grad_norm": 1.5339264869689941,
      "learning_rate": 8.21229635350871e-06,
      "loss": 0.0502,
      "step": 214125
    },
    {
      "epoch": 3.153856349685572,
      "grad_norm": 0.9971723556518555,
      "learning_rate": 8.210659981541725e-06,
      "loss": 0.0533,
      "step": 214150
    },
    {
      "epoch": 3.154224532775659,
      "grad_norm": 1.572286605834961,
      "learning_rate": 8.20902360957474e-06,
      "loss": 0.0546,
      "step": 214175
    },
    {
      "epoch": 3.154592715865746,
      "grad_norm": 1.5081692934036255,
      "learning_rate": 8.207387237607756e-06,
      "loss": 0.055,
      "step": 214200
    },
    {
      "epoch": 3.154960898955833,
      "grad_norm": 1.7470306158065796,
      "learning_rate": 8.205750865640772e-06,
      "loss": 0.0585,
      "step": 214225
    },
    {
      "epoch": 3.15532908204592,
      "grad_norm": 1.0209217071533203,
      "learning_rate": 8.204114493673787e-06,
      "loss": 0.0559,
      "step": 214250
    },
    {
      "epoch": 3.155697265136007,
      "grad_norm": 0.9646037220954895,
      "learning_rate": 8.202478121706801e-06,
      "loss": 0.054,
      "step": 214275
    },
    {
      "epoch": 3.156065448226094,
      "grad_norm": 1.0945680141448975,
      "learning_rate": 8.200841749739817e-06,
      "loss": 0.054,
      "step": 214300
    },
    {
      "epoch": 3.156433631316181,
      "grad_norm": 1.498083472251892,
      "learning_rate": 8.199205377772833e-06,
      "loss": 0.055,
      "step": 214325
    },
    {
      "epoch": 3.156801814406268,
      "grad_norm": 1.2775437831878662,
      "learning_rate": 8.19756900580585e-06,
      "loss": 0.0579,
      "step": 214350
    },
    {
      "epoch": 3.157169997496355,
      "grad_norm": 1.5905368328094482,
      "learning_rate": 8.195932633838864e-06,
      "loss": 0.0576,
      "step": 214375
    },
    {
      "epoch": 3.157538180586442,
      "grad_norm": 1.3423746824264526,
      "learning_rate": 8.194296261871878e-06,
      "loss": 0.0574,
      "step": 214400
    },
    {
      "epoch": 3.157906363676529,
      "grad_norm": 1.3613702058792114,
      "learning_rate": 8.192659889904895e-06,
      "loss": 0.0546,
      "step": 214425
    },
    {
      "epoch": 3.158274546766616,
      "grad_norm": 1.3552147150039673,
      "learning_rate": 8.19102351793791e-06,
      "loss": 0.0554,
      "step": 214450
    },
    {
      "epoch": 3.158642729856703,
      "grad_norm": 1.9894176721572876,
      "learning_rate": 8.189387145970925e-06,
      "loss": 0.0618,
      "step": 214475
    },
    {
      "epoch": 3.15901091294679,
      "grad_norm": 1.0688800811767578,
      "learning_rate": 8.187750774003941e-06,
      "loss": 0.064,
      "step": 214500
    },
    {
      "epoch": 3.159379096036877,
      "grad_norm": 1.731858730316162,
      "learning_rate": 8.186114402036956e-06,
      "loss": 0.0568,
      "step": 214525
    },
    {
      "epoch": 3.159747279126964,
      "grad_norm": 1.0963916778564453,
      "learning_rate": 8.184478030069972e-06,
      "loss": 0.0572,
      "step": 214550
    },
    {
      "epoch": 3.1601154622170515,
      "grad_norm": 0.7976915240287781,
      "learning_rate": 8.182841658102988e-06,
      "loss": 0.0546,
      "step": 214575
    },
    {
      "epoch": 3.1604836453071385,
      "grad_norm": 1.5268316268920898,
      "learning_rate": 8.181205286136003e-06,
      "loss": 0.0531,
      "step": 214600
    },
    {
      "epoch": 3.1608518283972256,
      "grad_norm": 0.8235132694244385,
      "learning_rate": 8.179568914169019e-06,
      "loss": 0.0597,
      "step": 214625
    },
    {
      "epoch": 3.1612200114873126,
      "grad_norm": 1.3213927745819092,
      "learning_rate": 8.177932542202033e-06,
      "loss": 0.0587,
      "step": 214650
    },
    {
      "epoch": 3.1615881945773996,
      "grad_norm": 1.0174158811569214,
      "learning_rate": 8.17629617023505e-06,
      "loss": 0.0541,
      "step": 214675
    },
    {
      "epoch": 3.1619563776674866,
      "grad_norm": 1.4569728374481201,
      "learning_rate": 8.174659798268064e-06,
      "loss": 0.0572,
      "step": 214700
    },
    {
      "epoch": 3.1623245607575736,
      "grad_norm": 1.13290536403656,
      "learning_rate": 8.17302342630108e-06,
      "loss": 0.0521,
      "step": 214725
    },
    {
      "epoch": 3.1626927438476606,
      "grad_norm": 1.3021537065505981,
      "learning_rate": 8.171387054334096e-06,
      "loss": 0.0518,
      "step": 214750
    },
    {
      "epoch": 3.1630609269377477,
      "grad_norm": 1.4855173826217651,
      "learning_rate": 8.16975068236711e-06,
      "loss": 0.0572,
      "step": 214775
    },
    {
      "epoch": 3.1634291100278347,
      "grad_norm": 1.2575362920761108,
      "learning_rate": 8.168114310400127e-06,
      "loss": 0.0526,
      "step": 214800
    },
    {
      "epoch": 3.1637972931179217,
      "grad_norm": 1.4837589263916016,
      "learning_rate": 8.166477938433141e-06,
      "loss": 0.0501,
      "step": 214825
    },
    {
      "epoch": 3.1641654762080087,
      "grad_norm": 1.3201878070831299,
      "learning_rate": 8.164841566466157e-06,
      "loss": 0.0529,
      "step": 214850
    },
    {
      "epoch": 3.1645336592980957,
      "grad_norm": 0.935027539730072,
      "learning_rate": 8.163205194499173e-06,
      "loss": 0.0548,
      "step": 214875
    },
    {
      "epoch": 3.1649018423881827,
      "grad_norm": 1.7025067806243896,
      "learning_rate": 8.161568822532188e-06,
      "loss": 0.0552,
      "step": 214900
    },
    {
      "epoch": 3.1652700254782697,
      "grad_norm": 1.2447409629821777,
      "learning_rate": 8.159932450565204e-06,
      "loss": 0.0571,
      "step": 214925
    },
    {
      "epoch": 3.1656382085683568,
      "grad_norm": 1.6836856603622437,
      "learning_rate": 8.158296078598219e-06,
      "loss": 0.0515,
      "step": 214950
    },
    {
      "epoch": 3.1660063916584438,
      "grad_norm": 1.638223648071289,
      "learning_rate": 8.156659706631235e-06,
      "loss": 0.0545,
      "step": 214975
    },
    {
      "epoch": 3.166374574748531,
      "grad_norm": 1.0520964860916138,
      "learning_rate": 8.15502333466425e-06,
      "loss": 0.0565,
      "step": 215000
    },
    {
      "epoch": 3.166742757838618,
      "grad_norm": 1.174172043800354,
      "learning_rate": 8.153452417575945e-06,
      "loss": 0.0568,
      "step": 215025
    },
    {
      "epoch": 3.167110940928705,
      "grad_norm": 1.3834646940231323,
      "learning_rate": 8.151816045608961e-06,
      "loss": 0.0588,
      "step": 215050
    },
    {
      "epoch": 3.167479124018792,
      "grad_norm": 0.9561067223548889,
      "learning_rate": 8.150179673641975e-06,
      "loss": 0.058,
      "step": 215075
    },
    {
      "epoch": 3.1678473071088793,
      "grad_norm": 1.3745793104171753,
      "learning_rate": 8.148543301674992e-06,
      "loss": 0.0536,
      "step": 215100
    },
    {
      "epoch": 3.1682154901989663,
      "grad_norm": 1.4004675149917603,
      "learning_rate": 8.146906929708006e-06,
      "loss": 0.0561,
      "step": 215125
    },
    {
      "epoch": 3.1685836732890533,
      "grad_norm": 1.1246006488800049,
      "learning_rate": 8.145270557741022e-06,
      "loss": 0.0568,
      "step": 215150
    },
    {
      "epoch": 3.1689518563791403,
      "grad_norm": 1.1674959659576416,
      "learning_rate": 8.143634185774038e-06,
      "loss": 0.0515,
      "step": 215175
    },
    {
      "epoch": 3.1693200394692274,
      "grad_norm": 1.3476957082748413,
      "learning_rate": 8.141997813807053e-06,
      "loss": 0.0513,
      "step": 215200
    },
    {
      "epoch": 3.1696882225593144,
      "grad_norm": 1.0233887434005737,
      "learning_rate": 8.140361441840067e-06,
      "loss": 0.0562,
      "step": 215225
    },
    {
      "epoch": 3.1700564056494014,
      "grad_norm": 1.5655750036239624,
      "learning_rate": 8.138725069873083e-06,
      "loss": 0.0531,
      "step": 215250
    },
    {
      "epoch": 3.1704245887394884,
      "grad_norm": 1.1323137283325195,
      "learning_rate": 8.1370886979061e-06,
      "loss": 0.0572,
      "step": 215275
    },
    {
      "epoch": 3.1707927718295754,
      "grad_norm": 1.1090760231018066,
      "learning_rate": 8.135452325939116e-06,
      "loss": 0.0596,
      "step": 215300
    },
    {
      "epoch": 3.1711609549196624,
      "grad_norm": 1.6740914583206177,
      "learning_rate": 8.13381595397213e-06,
      "loss": 0.0502,
      "step": 215325
    },
    {
      "epoch": 3.1715291380097494,
      "grad_norm": 1.2162855863571167,
      "learning_rate": 8.132179582005145e-06,
      "loss": 0.0483,
      "step": 215350
    },
    {
      "epoch": 3.1718973210998365,
      "grad_norm": 1.3601698875427246,
      "learning_rate": 8.13054321003816e-06,
      "loss": 0.0549,
      "step": 215375
    },
    {
      "epoch": 3.1722655041899235,
      "grad_norm": 1.5169645547866821,
      "learning_rate": 8.128906838071177e-06,
      "loss": 0.048,
      "step": 215400
    },
    {
      "epoch": 3.1726336872800105,
      "grad_norm": 1.4372507333755493,
      "learning_rate": 8.127270466104191e-06,
      "loss": 0.0533,
      "step": 215425
    },
    {
      "epoch": 3.1730018703700975,
      "grad_norm": 1.2877602577209473,
      "learning_rate": 8.125634094137207e-06,
      "loss": 0.0597,
      "step": 215450
    },
    {
      "epoch": 3.1733700534601845,
      "grad_norm": 0.972962498664856,
      "learning_rate": 8.123997722170222e-06,
      "loss": 0.0514,
      "step": 215475
    },
    {
      "epoch": 3.1737382365502715,
      "grad_norm": 1.4570029973983765,
      "learning_rate": 8.122361350203238e-06,
      "loss": 0.053,
      "step": 215500
    },
    {
      "epoch": 3.1741064196403586,
      "grad_norm": 1.2261537313461304,
      "learning_rate": 8.120724978236254e-06,
      "loss": 0.055,
      "step": 215525
    },
    {
      "epoch": 3.174474602730446,
      "grad_norm": 1.1197057962417603,
      "learning_rate": 8.119088606269269e-06,
      "loss": 0.0616,
      "step": 215550
    },
    {
      "epoch": 3.174842785820533,
      "grad_norm": 1.6672215461730957,
      "learning_rate": 8.117452234302285e-06,
      "loss": 0.0549,
      "step": 215575
    },
    {
      "epoch": 3.17521096891062,
      "grad_norm": 1.498630404472351,
      "learning_rate": 8.1158158623353e-06,
      "loss": 0.0488,
      "step": 215600
    },
    {
      "epoch": 3.175579152000707,
      "grad_norm": 1.8891905546188354,
      "learning_rate": 8.114179490368315e-06,
      "loss": 0.056,
      "step": 215625
    },
    {
      "epoch": 3.175947335090794,
      "grad_norm": 1.3721314668655396,
      "learning_rate": 8.11254311840133e-06,
      "loss": 0.0583,
      "step": 215650
    },
    {
      "epoch": 3.176315518180881,
      "grad_norm": 1.6834851503372192,
      "learning_rate": 8.110906746434346e-06,
      "loss": 0.0589,
      "step": 215675
    },
    {
      "epoch": 3.176683701270968,
      "grad_norm": 1.2526143789291382,
      "learning_rate": 8.109270374467362e-06,
      "loss": 0.0529,
      "step": 215700
    },
    {
      "epoch": 3.177051884361055,
      "grad_norm": 1.2112129926681519,
      "learning_rate": 8.107634002500377e-06,
      "loss": 0.0486,
      "step": 215725
    },
    {
      "epoch": 3.177420067451142,
      "grad_norm": 1.319957971572876,
      "learning_rate": 8.105997630533393e-06,
      "loss": 0.0524,
      "step": 215750
    },
    {
      "epoch": 3.177788250541229,
      "grad_norm": 1.3872143030166626,
      "learning_rate": 8.104361258566407e-06,
      "loss": 0.0578,
      "step": 215775
    },
    {
      "epoch": 3.178156433631316,
      "grad_norm": 1.1528382301330566,
      "learning_rate": 8.102724886599423e-06,
      "loss": 0.0534,
      "step": 215800
    },
    {
      "epoch": 3.178524616721403,
      "grad_norm": 1.0561034679412842,
      "learning_rate": 8.10108851463244e-06,
      "loss": 0.0567,
      "step": 215825
    },
    {
      "epoch": 3.17889279981149,
      "grad_norm": 1.3813204765319824,
      "learning_rate": 8.099452142665454e-06,
      "loss": 0.0506,
      "step": 215850
    },
    {
      "epoch": 3.179260982901577,
      "grad_norm": 1.1070430278778076,
      "learning_rate": 8.09781577069847e-06,
      "loss": 0.0543,
      "step": 215875
    },
    {
      "epoch": 3.1796291659916642,
      "grad_norm": 1.2274774312973022,
      "learning_rate": 8.096179398731485e-06,
      "loss": 0.0532,
      "step": 215900
    },
    {
      "epoch": 3.1799973490817512,
      "grad_norm": 1.2431498765945435,
      "learning_rate": 8.0945430267645e-06,
      "loss": 0.0594,
      "step": 215925
    },
    {
      "epoch": 3.1803655321718383,
      "grad_norm": 1.5270816087722778,
      "learning_rate": 8.092906654797515e-06,
      "loss": 0.0535,
      "step": 215950
    },
    {
      "epoch": 3.1807337152619253,
      "grad_norm": 1.6544066667556763,
      "learning_rate": 8.091270282830531e-06,
      "loss": 0.0556,
      "step": 215975
    },
    {
      "epoch": 3.1811018983520123,
      "grad_norm": 1.0528546571731567,
      "learning_rate": 8.089633910863548e-06,
      "loss": 0.0546,
      "step": 216000
    },
    {
      "epoch": 3.1814700814420993,
      "grad_norm": 1.2223880290985107,
      "learning_rate": 8.087997538896562e-06,
      "loss": 0.0572,
      "step": 216025
    },
    {
      "epoch": 3.1818382645321868,
      "grad_norm": 1.477240800857544,
      "learning_rate": 8.086361166929578e-06,
      "loss": 0.0539,
      "step": 216050
    },
    {
      "epoch": 3.182206447622274,
      "grad_norm": 1.1323323249816895,
      "learning_rate": 8.084724794962593e-06,
      "loss": 0.056,
      "step": 216075
    },
    {
      "epoch": 3.182574630712361,
      "grad_norm": 1.3468780517578125,
      "learning_rate": 8.083088422995609e-06,
      "loss": 0.0596,
      "step": 216100
    },
    {
      "epoch": 3.182942813802448,
      "grad_norm": 1.3269598484039307,
      "learning_rate": 8.081452051028625e-06,
      "loss": 0.0544,
      "step": 216125
    },
    {
      "epoch": 3.183310996892535,
      "grad_norm": 1.4006699323654175,
      "learning_rate": 8.07981567906164e-06,
      "loss": 0.0544,
      "step": 216150
    },
    {
      "epoch": 3.183679179982622,
      "grad_norm": 1.5041946172714233,
      "learning_rate": 8.078179307094654e-06,
      "loss": 0.0598,
      "step": 216175
    },
    {
      "epoch": 3.184047363072709,
      "grad_norm": 1.2448440790176392,
      "learning_rate": 8.07654293512767e-06,
      "loss": 0.0508,
      "step": 216200
    },
    {
      "epoch": 3.184415546162796,
      "grad_norm": 1.682237148284912,
      "learning_rate": 8.074906563160686e-06,
      "loss": 0.054,
      "step": 216225
    },
    {
      "epoch": 3.184783729252883,
      "grad_norm": 1.3867461681365967,
      "learning_rate": 8.073270191193702e-06,
      "loss": 0.0534,
      "step": 216250
    },
    {
      "epoch": 3.18515191234297,
      "grad_norm": 1.4437001943588257,
      "learning_rate": 8.071633819226717e-06,
      "loss": 0.0593,
      "step": 216275
    },
    {
      "epoch": 3.185520095433057,
      "grad_norm": 1.1554878950119019,
      "learning_rate": 8.069997447259731e-06,
      "loss": 0.0556,
      "step": 216300
    },
    {
      "epoch": 3.185888278523144,
      "grad_norm": 1.4752306938171387,
      "learning_rate": 8.068361075292747e-06,
      "loss": 0.0628,
      "step": 216325
    },
    {
      "epoch": 3.186256461613231,
      "grad_norm": 1.122360110282898,
      "learning_rate": 8.066724703325763e-06,
      "loss": 0.054,
      "step": 216350
    },
    {
      "epoch": 3.186624644703318,
      "grad_norm": 1.4905937910079956,
      "learning_rate": 8.065088331358778e-06,
      "loss": 0.0588,
      "step": 216375
    },
    {
      "epoch": 3.186992827793405,
      "grad_norm": 1.376253366470337,
      "learning_rate": 8.063451959391794e-06,
      "loss": 0.052,
      "step": 216400
    },
    {
      "epoch": 3.187361010883492,
      "grad_norm": 1.725206732749939,
      "learning_rate": 8.061815587424809e-06,
      "loss": 0.0672,
      "step": 216425
    },
    {
      "epoch": 3.187729193973579,
      "grad_norm": 1.6516075134277344,
      "learning_rate": 8.060179215457825e-06,
      "loss": 0.0637,
      "step": 216450
    },
    {
      "epoch": 3.188097377063666,
      "grad_norm": 1.9658507108688354,
      "learning_rate": 8.05854284349084e-06,
      "loss": 0.0559,
      "step": 216475
    },
    {
      "epoch": 3.1884655601537535,
      "grad_norm": 1.446641206741333,
      "learning_rate": 8.056906471523855e-06,
      "loss": 0.0554,
      "step": 216500
    },
    {
      "epoch": 3.1888337432438405,
      "grad_norm": 1.4606941938400269,
      "learning_rate": 8.055270099556871e-06,
      "loss": 0.0561,
      "step": 216525
    },
    {
      "epoch": 3.1892019263339275,
      "grad_norm": 1.0025992393493652,
      "learning_rate": 8.053633727589888e-06,
      "loss": 0.0522,
      "step": 216550
    },
    {
      "epoch": 3.1895701094240145,
      "grad_norm": 1.394100308418274,
      "learning_rate": 8.051997355622902e-06,
      "loss": 0.0536,
      "step": 216575
    },
    {
      "epoch": 3.1899382925141015,
      "grad_norm": 1.3884831666946411,
      "learning_rate": 8.050360983655917e-06,
      "loss": 0.0563,
      "step": 216600
    },
    {
      "epoch": 3.1903064756041886,
      "grad_norm": 1.4022445678710938,
      "learning_rate": 8.048724611688933e-06,
      "loss": 0.056,
      "step": 216625
    },
    {
      "epoch": 3.1906746586942756,
      "grad_norm": 1.6145238876342773,
      "learning_rate": 8.047088239721949e-06,
      "loss": 0.058,
      "step": 216650
    },
    {
      "epoch": 3.1910428417843626,
      "grad_norm": 1.4306340217590332,
      "learning_rate": 8.045451867754965e-06,
      "loss": 0.0552,
      "step": 216675
    },
    {
      "epoch": 3.1914110248744496,
      "grad_norm": 0.9680589437484741,
      "learning_rate": 8.04381549578798e-06,
      "loss": 0.0551,
      "step": 216700
    },
    {
      "epoch": 3.1917792079645366,
      "grad_norm": 1.6016921997070312,
      "learning_rate": 8.042179123820994e-06,
      "loss": 0.0593,
      "step": 216725
    },
    {
      "epoch": 3.1921473910546236,
      "grad_norm": 1.6941205263137817,
      "learning_rate": 8.04054275185401e-06,
      "loss": 0.0565,
      "step": 216750
    },
    {
      "epoch": 3.1925155741447107,
      "grad_norm": 1.2751888036727905,
      "learning_rate": 8.038906379887026e-06,
      "loss": 0.0561,
      "step": 216775
    },
    {
      "epoch": 3.1928837572347977,
      "grad_norm": 1.2650781869888306,
      "learning_rate": 8.03727000792004e-06,
      "loss": 0.0521,
      "step": 216800
    },
    {
      "epoch": 3.1932519403248847,
      "grad_norm": 1.5958646535873413,
      "learning_rate": 8.035633635953057e-06,
      "loss": 0.0631,
      "step": 216825
    },
    {
      "epoch": 3.1936201234149717,
      "grad_norm": 1.2072230577468872,
      "learning_rate": 8.033997263986071e-06,
      "loss": 0.0508,
      "step": 216850
    },
    {
      "epoch": 3.1939883065050587,
      "grad_norm": 1.642998456954956,
      "learning_rate": 8.032360892019087e-06,
      "loss": 0.0588,
      "step": 216875
    },
    {
      "epoch": 3.1943564895951457,
      "grad_norm": 1.6794160604476929,
      "learning_rate": 8.030724520052102e-06,
      "loss": 0.053,
      "step": 216900
    },
    {
      "epoch": 3.1947246726852327,
      "grad_norm": 1.5724201202392578,
      "learning_rate": 8.029088148085118e-06,
      "loss": 0.0476,
      "step": 216925
    },
    {
      "epoch": 3.1950928557753198,
      "grad_norm": 1.1608883142471313,
      "learning_rate": 8.027451776118134e-06,
      "loss": 0.0615,
      "step": 216950
    },
    {
      "epoch": 3.1954610388654068,
      "grad_norm": 0.874894380569458,
      "learning_rate": 8.025815404151149e-06,
      "loss": 0.0574,
      "step": 216975
    },
    {
      "epoch": 3.195829221955494,
      "grad_norm": 1.4091254472732544,
      "learning_rate": 8.024179032184165e-06,
      "loss": 0.0551,
      "step": 217000
    },
    {
      "epoch": 3.1961974050455813,
      "grad_norm": 1.1484310626983643,
      "learning_rate": 8.02254266021718e-06,
      "loss": 0.0565,
      "step": 217025
    },
    {
      "epoch": 3.1965655881356683,
      "grad_norm": 1.597987174987793,
      "learning_rate": 8.020906288250195e-06,
      "loss": 0.0535,
      "step": 217050
    },
    {
      "epoch": 3.1969337712257553,
      "grad_norm": 1.5632587671279907,
      "learning_rate": 8.019335371161891e-06,
      "loss": 0.0541,
      "step": 217075
    },
    {
      "epoch": 3.1973019543158423,
      "grad_norm": 1.4962395429611206,
      "learning_rate": 8.017698999194905e-06,
      "loss": 0.0621,
      "step": 217100
    },
    {
      "epoch": 3.1976701374059293,
      "grad_norm": 1.2469425201416016,
      "learning_rate": 8.01606262722792e-06,
      "loss": 0.0533,
      "step": 217125
    },
    {
      "epoch": 3.1980383204960163,
      "grad_norm": 1.5091272592544556,
      "learning_rate": 8.014426255260936e-06,
      "loss": 0.057,
      "step": 217150
    },
    {
      "epoch": 3.1984065035861033,
      "grad_norm": 1.6714004278182983,
      "learning_rate": 8.012789883293952e-06,
      "loss": 0.0569,
      "step": 217175
    },
    {
      "epoch": 3.1987746866761904,
      "grad_norm": 1.188004493713379,
      "learning_rate": 8.011153511326968e-06,
      "loss": 0.0571,
      "step": 217200
    },
    {
      "epoch": 3.1991428697662774,
      "grad_norm": 1.5033217668533325,
      "learning_rate": 8.009517139359983e-06,
      "loss": 0.0617,
      "step": 217225
    },
    {
      "epoch": 3.1995110528563644,
      "grad_norm": 1.7540327310562134,
      "learning_rate": 8.007880767392999e-06,
      "loss": 0.0521,
      "step": 217250
    },
    {
      "epoch": 3.1998792359464514,
      "grad_norm": 1.2581640481948853,
      "learning_rate": 8.006244395426013e-06,
      "loss": 0.0459,
      "step": 217275
    },
    {
      "epoch": 3.2002474190365384,
      "grad_norm": 1.2538645267486572,
      "learning_rate": 8.00460802345903e-06,
      "loss": 0.0527,
      "step": 217300
    },
    {
      "epoch": 3.2006156021266254,
      "grad_norm": 1.5519791841506958,
      "learning_rate": 8.002971651492044e-06,
      "loss": 0.0586,
      "step": 217325
    },
    {
      "epoch": 3.2009837852167125,
      "grad_norm": 1.1620047092437744,
      "learning_rate": 8.00133527952506e-06,
      "loss": 0.0588,
      "step": 217350
    },
    {
      "epoch": 3.2013519683067995,
      "grad_norm": 1.399783730506897,
      "learning_rate": 7.999698907558076e-06,
      "loss": 0.0553,
      "step": 217375
    },
    {
      "epoch": 3.2017201513968865,
      "grad_norm": 1.1367632150650024,
      "learning_rate": 7.99806253559109e-06,
      "loss": 0.0571,
      "step": 217400
    },
    {
      "epoch": 3.2020883344869735,
      "grad_norm": 1.2106285095214844,
      "learning_rate": 7.996426163624105e-06,
      "loss": 0.0518,
      "step": 217425
    },
    {
      "epoch": 3.2024565175770605,
      "grad_norm": 1.589905858039856,
      "learning_rate": 7.994789791657121e-06,
      "loss": 0.0645,
      "step": 217450
    },
    {
      "epoch": 3.202824700667148,
      "grad_norm": 1.630958080291748,
      "learning_rate": 7.993153419690138e-06,
      "loss": 0.0537,
      "step": 217475
    },
    {
      "epoch": 3.203192883757235,
      "grad_norm": 1.2025760412216187,
      "learning_rate": 7.991517047723154e-06,
      "loss": 0.0534,
      "step": 217500
    },
    {
      "epoch": 3.203561066847322,
      "grad_norm": 1.398712158203125,
      "learning_rate": 7.989880675756168e-06,
      "loss": 0.0541,
      "step": 217525
    },
    {
      "epoch": 3.203929249937409,
      "grad_norm": 0.9454752802848816,
      "learning_rate": 7.988244303789183e-06,
      "loss": 0.0537,
      "step": 217550
    },
    {
      "epoch": 3.204297433027496,
      "grad_norm": 1.665662407875061,
      "learning_rate": 7.986607931822199e-06,
      "loss": 0.0614,
      "step": 217575
    },
    {
      "epoch": 3.204665616117583,
      "grad_norm": 1.098403811454773,
      "learning_rate": 7.984971559855215e-06,
      "loss": 0.0609,
      "step": 217600
    },
    {
      "epoch": 3.20503379920767,
      "grad_norm": 1.2247380018234253,
      "learning_rate": 7.983335187888231e-06,
      "loss": 0.0591,
      "step": 217625
    },
    {
      "epoch": 3.205401982297757,
      "grad_norm": 1.1882036924362183,
      "learning_rate": 7.981698815921246e-06,
      "loss": 0.0536,
      "step": 217650
    },
    {
      "epoch": 3.205770165387844,
      "grad_norm": 1.586959719657898,
      "learning_rate": 7.98006244395426e-06,
      "loss": 0.0545,
      "step": 217675
    },
    {
      "epoch": 3.206138348477931,
      "grad_norm": 1.1314295530319214,
      "learning_rate": 7.978426071987276e-06,
      "loss": 0.0591,
      "step": 217700
    },
    {
      "epoch": 3.206506531568018,
      "grad_norm": 1.2921476364135742,
      "learning_rate": 7.976789700020292e-06,
      "loss": 0.0577,
      "step": 217725
    },
    {
      "epoch": 3.206874714658105,
      "grad_norm": 0.9713476896286011,
      "learning_rate": 7.975153328053307e-06,
      "loss": 0.0491,
      "step": 217750
    },
    {
      "epoch": 3.207242897748192,
      "grad_norm": 1.641406536102295,
      "learning_rate": 7.973516956086323e-06,
      "loss": 0.059,
      "step": 217775
    },
    {
      "epoch": 3.207611080838279,
      "grad_norm": 1.4800699949264526,
      "learning_rate": 7.971880584119337e-06,
      "loss": 0.05,
      "step": 217800
    },
    {
      "epoch": 3.207979263928366,
      "grad_norm": 0.901652455329895,
      "learning_rate": 7.970244212152354e-06,
      "loss": 0.0564,
      "step": 217825
    },
    {
      "epoch": 3.208347447018453,
      "grad_norm": 1.2968181371688843,
      "learning_rate": 7.968607840185368e-06,
      "loss": 0.0571,
      "step": 217850
    },
    {
      "epoch": 3.20871563010854,
      "grad_norm": 1.3947423696517944,
      "learning_rate": 7.966971468218384e-06,
      "loss": 0.0585,
      "step": 217875
    },
    {
      "epoch": 3.2090838131986272,
      "grad_norm": 0.698746919631958,
      "learning_rate": 7.9653350962514e-06,
      "loss": 0.0577,
      "step": 217900
    },
    {
      "epoch": 3.2094519962887142,
      "grad_norm": 1.2787644863128662,
      "learning_rate": 7.963698724284415e-06,
      "loss": 0.0599,
      "step": 217925
    },
    {
      "epoch": 3.2098201793788013,
      "grad_norm": 1.4996448755264282,
      "learning_rate": 7.962062352317431e-06,
      "loss": 0.0571,
      "step": 217950
    },
    {
      "epoch": 3.2101883624688887,
      "grad_norm": 1.2406779527664185,
      "learning_rate": 7.960425980350445e-06,
      "loss": 0.0509,
      "step": 217975
    },
    {
      "epoch": 3.2105565455589757,
      "grad_norm": 2.0091066360473633,
      "learning_rate": 7.958789608383461e-06,
      "loss": 0.056,
      "step": 218000
    },
    {
      "epoch": 3.2109247286490628,
      "grad_norm": 1.932629942893982,
      "learning_rate": 7.957153236416478e-06,
      "loss": 0.0506,
      "step": 218025
    },
    {
      "epoch": 3.2112929117391498,
      "grad_norm": 1.0832786560058594,
      "learning_rate": 7.955516864449492e-06,
      "loss": 0.0602,
      "step": 218050
    },
    {
      "epoch": 3.211661094829237,
      "grad_norm": 1.1417902708053589,
      "learning_rate": 7.953880492482508e-06,
      "loss": 0.0529,
      "step": 218075
    },
    {
      "epoch": 3.212029277919324,
      "grad_norm": 1.7789398431777954,
      "learning_rate": 7.952244120515523e-06,
      "loss": 0.0589,
      "step": 218100
    },
    {
      "epoch": 3.212397461009411,
      "grad_norm": 1.2323665618896484,
      "learning_rate": 7.950607748548539e-06,
      "loss": 0.0561,
      "step": 218125
    },
    {
      "epoch": 3.212765644099498,
      "grad_norm": 1.3466113805770874,
      "learning_rate": 7.948971376581555e-06,
      "loss": 0.0542,
      "step": 218150
    },
    {
      "epoch": 3.213133827189585,
      "grad_norm": 1.055102825164795,
      "learning_rate": 7.94733500461457e-06,
      "loss": 0.054,
      "step": 218175
    },
    {
      "epoch": 3.213502010279672,
      "grad_norm": 1.2706629037857056,
      "learning_rate": 7.945698632647586e-06,
      "loss": 0.0529,
      "step": 218200
    },
    {
      "epoch": 3.213870193369759,
      "grad_norm": 1.5244171619415283,
      "learning_rate": 7.9440622606806e-06,
      "loss": 0.0519,
      "step": 218225
    },
    {
      "epoch": 3.214238376459846,
      "grad_norm": 1.0767817497253418,
      "learning_rate": 7.942425888713616e-06,
      "loss": 0.057,
      "step": 218250
    },
    {
      "epoch": 3.214606559549933,
      "grad_norm": 1.303810954093933,
      "learning_rate": 7.94078951674663e-06,
      "loss": 0.057,
      "step": 218275
    },
    {
      "epoch": 3.21497474264002,
      "grad_norm": 1.4361202716827393,
      "learning_rate": 7.939153144779647e-06,
      "loss": 0.0512,
      "step": 218300
    },
    {
      "epoch": 3.215342925730107,
      "grad_norm": 1.2506037950515747,
      "learning_rate": 7.937516772812663e-06,
      "loss": 0.0606,
      "step": 218325
    },
    {
      "epoch": 3.215711108820194,
      "grad_norm": 1.189992070198059,
      "learning_rate": 7.935880400845677e-06,
      "loss": 0.0563,
      "step": 218350
    },
    {
      "epoch": 3.216079291910281,
      "grad_norm": 1.3108410835266113,
      "learning_rate": 7.934244028878692e-06,
      "loss": 0.0478,
      "step": 218375
    },
    {
      "epoch": 3.216447475000368,
      "grad_norm": 1.2580293416976929,
      "learning_rate": 7.932607656911708e-06,
      "loss": 0.0583,
      "step": 218400
    },
    {
      "epoch": 3.2168156580904554,
      "grad_norm": 1.4991685152053833,
      "learning_rate": 7.930971284944724e-06,
      "loss": 0.0591,
      "step": 218425
    },
    {
      "epoch": 3.2171838411805425,
      "grad_norm": 1.218895435333252,
      "learning_rate": 7.92933491297774e-06,
      "loss": 0.0555,
      "step": 218450
    },
    {
      "epoch": 3.2175520242706295,
      "grad_norm": 1.5605872869491577,
      "learning_rate": 7.927698541010755e-06,
      "loss": 0.0492,
      "step": 218475
    },
    {
      "epoch": 3.2179202073607165,
      "grad_norm": 1.4264897108078003,
      "learning_rate": 7.92606216904377e-06,
      "loss": 0.0561,
      "step": 218500
    },
    {
      "epoch": 3.2182883904508035,
      "grad_norm": 1.3707656860351562,
      "learning_rate": 7.924425797076785e-06,
      "loss": 0.0539,
      "step": 218525
    },
    {
      "epoch": 3.2186565735408905,
      "grad_norm": 0.6610348224639893,
      "learning_rate": 7.922789425109802e-06,
      "loss": 0.0507,
      "step": 218550
    },
    {
      "epoch": 3.2190247566309775,
      "grad_norm": 1.1268141269683838,
      "learning_rate": 7.921153053142818e-06,
      "loss": 0.0576,
      "step": 218575
    },
    {
      "epoch": 3.2193929397210646,
      "grad_norm": 1.5163272619247437,
      "learning_rate": 7.919516681175832e-06,
      "loss": 0.0616,
      "step": 218600
    },
    {
      "epoch": 3.2197611228111516,
      "grad_norm": 1.032633900642395,
      "learning_rate": 7.917880309208847e-06,
      "loss": 0.0504,
      "step": 218625
    },
    {
      "epoch": 3.2201293059012386,
      "grad_norm": 1.1997606754302979,
      "learning_rate": 7.916243937241863e-06,
      "loss": 0.0589,
      "step": 218650
    },
    {
      "epoch": 3.2204974889913256,
      "grad_norm": 0.787285327911377,
      "learning_rate": 7.914607565274879e-06,
      "loss": 0.0579,
      "step": 218675
    },
    {
      "epoch": 3.2208656720814126,
      "grad_norm": 1.094734787940979,
      "learning_rate": 7.912971193307893e-06,
      "loss": 0.0512,
      "step": 218700
    },
    {
      "epoch": 3.2212338551714996,
      "grad_norm": 1.3936247825622559,
      "learning_rate": 7.91133482134091e-06,
      "loss": 0.0556,
      "step": 218725
    },
    {
      "epoch": 3.2216020382615866,
      "grad_norm": 0.9421206116676331,
      "learning_rate": 7.909698449373924e-06,
      "loss": 0.0539,
      "step": 218750
    },
    {
      "epoch": 3.2219702213516737,
      "grad_norm": 1.6949080228805542,
      "learning_rate": 7.90806207740694e-06,
      "loss": 0.0549,
      "step": 218775
    },
    {
      "epoch": 3.2223384044417607,
      "grad_norm": 1.5208719968795776,
      "learning_rate": 7.906425705439955e-06,
      "loss": 0.0615,
      "step": 218800
    },
    {
      "epoch": 3.2227065875318477,
      "grad_norm": 1.4702141284942627,
      "learning_rate": 7.90478933347297e-06,
      "loss": 0.0506,
      "step": 218825
    },
    {
      "epoch": 3.2230747706219347,
      "grad_norm": 1.4829256534576416,
      "learning_rate": 7.903152961505987e-06,
      "loss": 0.0545,
      "step": 218850
    },
    {
      "epoch": 3.2234429537120217,
      "grad_norm": 1.4444411993026733,
      "learning_rate": 7.901516589539003e-06,
      "loss": 0.0598,
      "step": 218875
    },
    {
      "epoch": 3.2238111368021087,
      "grad_norm": 1.365617275238037,
      "learning_rate": 7.899880217572018e-06,
      "loss": 0.0511,
      "step": 218900
    },
    {
      "epoch": 3.224179319892196,
      "grad_norm": 1.714347243309021,
      "learning_rate": 7.898243845605032e-06,
      "loss": 0.0561,
      "step": 218925
    },
    {
      "epoch": 3.224547502982283,
      "grad_norm": 0.9410088658332825,
      "learning_rate": 7.896607473638048e-06,
      "loss": 0.0564,
      "step": 218950
    },
    {
      "epoch": 3.2249156860723702,
      "grad_norm": 1.8370044231414795,
      "learning_rate": 7.894971101671064e-06,
      "loss": 0.0526,
      "step": 218975
    },
    {
      "epoch": 3.2252838691624572,
      "grad_norm": 1.2382044792175293,
      "learning_rate": 7.89333472970408e-06,
      "loss": 0.0573,
      "step": 219000
    },
    {
      "epoch": 3.2256520522525443,
      "grad_norm": 1.8614258766174316,
      "learning_rate": 7.891698357737095e-06,
      "loss": 0.0631,
      "step": 219025
    },
    {
      "epoch": 3.2260202353426313,
      "grad_norm": 1.6997545957565308,
      "learning_rate": 7.89006198577011e-06,
      "loss": 0.0569,
      "step": 219050
    },
    {
      "epoch": 3.2263884184327183,
      "grad_norm": 2.166200876235962,
      "learning_rate": 7.888425613803126e-06,
      "loss": 0.0611,
      "step": 219075
    },
    {
      "epoch": 3.2267566015228053,
      "grad_norm": 1.0739834308624268,
      "learning_rate": 7.886789241836142e-06,
      "loss": 0.0496,
      "step": 219100
    },
    {
      "epoch": 3.2271247846128923,
      "grad_norm": 1.9614357948303223,
      "learning_rate": 7.885152869869156e-06,
      "loss": 0.0512,
      "step": 219125
    },
    {
      "epoch": 3.2274929677029793,
      "grad_norm": 1.7242183685302734,
      "learning_rate": 7.883516497902172e-06,
      "loss": 0.0567,
      "step": 219150
    },
    {
      "epoch": 3.2278611507930663,
      "grad_norm": 1.8307344913482666,
      "learning_rate": 7.881880125935187e-06,
      "loss": 0.0564,
      "step": 219175
    },
    {
      "epoch": 3.2282293338831534,
      "grad_norm": 1.1590774059295654,
      "learning_rate": 7.880243753968203e-06,
      "loss": 0.051,
      "step": 219200
    },
    {
      "epoch": 3.2285975169732404,
      "grad_norm": 1.5442832708358765,
      "learning_rate": 7.878607382001217e-06,
      "loss": 0.0506,
      "step": 219225
    },
    {
      "epoch": 3.2289657000633274,
      "grad_norm": 0.9634265303611755,
      "learning_rate": 7.876971010034233e-06,
      "loss": 0.0601,
      "step": 219250
    },
    {
      "epoch": 3.2293338831534144,
      "grad_norm": 1.2625250816345215,
      "learning_rate": 7.87533463806725e-06,
      "loss": 0.0562,
      "step": 219275
    },
    {
      "epoch": 3.2297020662435014,
      "grad_norm": 1.1419134140014648,
      "learning_rate": 7.873698266100264e-06,
      "loss": 0.0611,
      "step": 219300
    },
    {
      "epoch": 3.2300702493335884,
      "grad_norm": 1.245415210723877,
      "learning_rate": 7.872061894133279e-06,
      "loss": 0.061,
      "step": 219325
    },
    {
      "epoch": 3.2304384324236755,
      "grad_norm": 1.7258957624435425,
      "learning_rate": 7.870425522166295e-06,
      "loss": 0.0547,
      "step": 219350
    },
    {
      "epoch": 3.230806615513763,
      "grad_norm": 1.766653299331665,
      "learning_rate": 7.868789150199311e-06,
      "loss": 0.0571,
      "step": 219375
    },
    {
      "epoch": 3.23117479860385,
      "grad_norm": 0.9274715781211853,
      "learning_rate": 7.867152778232327e-06,
      "loss": 0.0531,
      "step": 219400
    },
    {
      "epoch": 3.231542981693937,
      "grad_norm": 1.7551229000091553,
      "learning_rate": 7.865581861144021e-06,
      "loss": 0.053,
      "step": 219425
    },
    {
      "epoch": 3.231911164784024,
      "grad_norm": 1.257689118385315,
      "learning_rate": 7.863945489177035e-06,
      "loss": 0.0591,
      "step": 219450
    },
    {
      "epoch": 3.232279347874111,
      "grad_norm": 1.8327311277389526,
      "learning_rate": 7.862309117210051e-06,
      "loss": 0.0539,
      "step": 219475
    },
    {
      "epoch": 3.232647530964198,
      "grad_norm": 1.2686773538589478,
      "learning_rate": 7.860672745243068e-06,
      "loss": 0.0525,
      "step": 219500
    },
    {
      "epoch": 3.233015714054285,
      "grad_norm": 1.14503812789917,
      "learning_rate": 7.859036373276084e-06,
      "loss": 0.0586,
      "step": 219525
    },
    {
      "epoch": 3.233383897144372,
      "grad_norm": 1.4078103303909302,
      "learning_rate": 7.857400001309098e-06,
      "loss": 0.0493,
      "step": 219550
    },
    {
      "epoch": 3.233752080234459,
      "grad_norm": 0.8492733240127563,
      "learning_rate": 7.855763629342114e-06,
      "loss": 0.0533,
      "step": 219575
    },
    {
      "epoch": 3.234120263324546,
      "grad_norm": 1.3695659637451172,
      "learning_rate": 7.854127257375129e-06,
      "loss": 0.061,
      "step": 219600
    },
    {
      "epoch": 3.234488446414633,
      "grad_norm": 1.4010695219039917,
      "learning_rate": 7.852490885408145e-06,
      "loss": 0.0503,
      "step": 219625
    },
    {
      "epoch": 3.23485662950472,
      "grad_norm": 1.7307606935501099,
      "learning_rate": 7.85085451344116e-06,
      "loss": 0.0576,
      "step": 219650
    },
    {
      "epoch": 3.235224812594807,
      "grad_norm": 1.6170222759246826,
      "learning_rate": 7.849218141474176e-06,
      "loss": 0.0533,
      "step": 219675
    },
    {
      "epoch": 3.235592995684894,
      "grad_norm": 1.5843403339385986,
      "learning_rate": 7.847581769507192e-06,
      "loss": 0.053,
      "step": 219700
    },
    {
      "epoch": 3.235961178774981,
      "grad_norm": 1.438511848449707,
      "learning_rate": 7.845945397540206e-06,
      "loss": 0.0525,
      "step": 219725
    },
    {
      "epoch": 3.236329361865068,
      "grad_norm": 1.6156588792800903,
      "learning_rate": 7.84430902557322e-06,
      "loss": 0.0607,
      "step": 219750
    },
    {
      "epoch": 3.236697544955155,
      "grad_norm": 1.6585315465927124,
      "learning_rate": 7.842672653606237e-06,
      "loss": 0.0512,
      "step": 219775
    },
    {
      "epoch": 3.237065728045242,
      "grad_norm": 0.9308322072029114,
      "learning_rate": 7.841036281639253e-06,
      "loss": 0.051,
      "step": 219800
    },
    {
      "epoch": 3.237433911135329,
      "grad_norm": 0.7710166573524475,
      "learning_rate": 7.839399909672269e-06,
      "loss": 0.0578,
      "step": 219825
    },
    {
      "epoch": 3.237802094225416,
      "grad_norm": 1.2792218923568726,
      "learning_rate": 7.837763537705284e-06,
      "loss": 0.0548,
      "step": 219850
    },
    {
      "epoch": 3.238170277315503,
      "grad_norm": 0.8353171348571777,
      "learning_rate": 7.836127165738298e-06,
      "loss": 0.0533,
      "step": 219875
    },
    {
      "epoch": 3.2385384604055907,
      "grad_norm": 1.236276388168335,
      "learning_rate": 7.834490793771314e-06,
      "loss": 0.0612,
      "step": 219900
    },
    {
      "epoch": 3.2389066434956777,
      "grad_norm": 1.0535063743591309,
      "learning_rate": 7.83285442180433e-06,
      "loss": 0.0608,
      "step": 219925
    },
    {
      "epoch": 3.2392748265857647,
      "grad_norm": 1.2724214792251587,
      "learning_rate": 7.831218049837347e-06,
      "loss": 0.0556,
      "step": 219950
    },
    {
      "epoch": 3.2396430096758517,
      "grad_norm": 1.3391023874282837,
      "learning_rate": 7.829581677870361e-06,
      "loss": 0.0543,
      "step": 219975
    },
    {
      "epoch": 3.2400111927659387,
      "grad_norm": 1.8974146842956543,
      "learning_rate": 7.827945305903375e-06,
      "loss": 0.0621,
      "step": 220000
    },
    {
      "epoch": 3.2403793758560258,
      "grad_norm": 1.2320915460586548,
      "learning_rate": 7.826308933936392e-06,
      "loss": 0.0557,
      "step": 220025
    },
    {
      "epoch": 3.2407475589461128,
      "grad_norm": 1.5294811725616455,
      "learning_rate": 7.824672561969408e-06,
      "loss": 0.0521,
      "step": 220050
    },
    {
      "epoch": 3.2411157420362,
      "grad_norm": 1.2936004400253296,
      "learning_rate": 7.823036190002422e-06,
      "loss": 0.0531,
      "step": 220075
    },
    {
      "epoch": 3.241483925126287,
      "grad_norm": 1.2569042444229126,
      "learning_rate": 7.821399818035438e-06,
      "loss": 0.0474,
      "step": 220100
    },
    {
      "epoch": 3.241852108216374,
      "grad_norm": 1.6288968324661255,
      "learning_rate": 7.819763446068453e-06,
      "loss": 0.0529,
      "step": 220125
    },
    {
      "epoch": 3.242220291306461,
      "grad_norm": 1.7834948301315308,
      "learning_rate": 7.818127074101469e-06,
      "loss": 0.0606,
      "step": 220150
    },
    {
      "epoch": 3.242588474396548,
      "grad_norm": 1.0031009912490845,
      "learning_rate": 7.816490702134483e-06,
      "loss": 0.0532,
      "step": 220175
    },
    {
      "epoch": 3.242956657486635,
      "grad_norm": 1.7676444053649902,
      "learning_rate": 7.8148543301675e-06,
      "loss": 0.0574,
      "step": 220200
    },
    {
      "epoch": 3.243324840576722,
      "grad_norm": 1.1878416538238525,
      "learning_rate": 7.813217958200516e-06,
      "loss": 0.0529,
      "step": 220225
    },
    {
      "epoch": 3.243693023666809,
      "grad_norm": 1.1438179016113281,
      "learning_rate": 7.81158158623353e-06,
      "loss": 0.0555,
      "step": 220250
    },
    {
      "epoch": 3.244061206756896,
      "grad_norm": 1.3004798889160156,
      "learning_rate": 7.809945214266546e-06,
      "loss": 0.0545,
      "step": 220275
    },
    {
      "epoch": 3.244429389846983,
      "grad_norm": 1.0291725397109985,
      "learning_rate": 7.80830884229956e-06,
      "loss": 0.049,
      "step": 220300
    },
    {
      "epoch": 3.24479757293707,
      "grad_norm": 1.079509973526001,
      "learning_rate": 7.806672470332577e-06,
      "loss": 0.055,
      "step": 220325
    },
    {
      "epoch": 3.2451657560271574,
      "grad_norm": 1.3070451021194458,
      "learning_rate": 7.805036098365593e-06,
      "loss": 0.0477,
      "step": 220350
    },
    {
      "epoch": 3.2455339391172444,
      "grad_norm": 1.6348049640655518,
      "learning_rate": 7.803399726398608e-06,
      "loss": 0.0557,
      "step": 220375
    },
    {
      "epoch": 3.2459021222073314,
      "grad_norm": 1.5824440717697144,
      "learning_rate": 7.801763354431624e-06,
      "loss": 0.0544,
      "step": 220400
    },
    {
      "epoch": 3.2462703052974184,
      "grad_norm": 1.23793625831604,
      "learning_rate": 7.800126982464638e-06,
      "loss": 0.0522,
      "step": 220425
    },
    {
      "epoch": 3.2466384883875055,
      "grad_norm": 1.3498425483703613,
      "learning_rate": 7.798490610497654e-06,
      "loss": 0.0551,
      "step": 220450
    },
    {
      "epoch": 3.2470066714775925,
      "grad_norm": 1.681096076965332,
      "learning_rate": 7.79685423853067e-06,
      "loss": 0.0566,
      "step": 220475
    },
    {
      "epoch": 3.2473748545676795,
      "grad_norm": 1.1958085298538208,
      "learning_rate": 7.795217866563685e-06,
      "loss": 0.0551,
      "step": 220500
    },
    {
      "epoch": 3.2477430376577665,
      "grad_norm": 1.3471393585205078,
      "learning_rate": 7.793581494596701e-06,
      "loss": 0.059,
      "step": 220525
    },
    {
      "epoch": 3.2481112207478535,
      "grad_norm": 0.9900381565093994,
      "learning_rate": 7.791945122629716e-06,
      "loss": 0.0549,
      "step": 220550
    },
    {
      "epoch": 3.2484794038379405,
      "grad_norm": 1.659070372581482,
      "learning_rate": 7.790308750662732e-06,
      "loss": 0.0581,
      "step": 220575
    },
    {
      "epoch": 3.2488475869280276,
      "grad_norm": 1.5467759370803833,
      "learning_rate": 7.788672378695746e-06,
      "loss": 0.0526,
      "step": 220600
    },
    {
      "epoch": 3.2492157700181146,
      "grad_norm": 1.1430433988571167,
      "learning_rate": 7.787036006728762e-06,
      "loss": 0.0567,
      "step": 220625
    },
    {
      "epoch": 3.2495839531082016,
      "grad_norm": 1.223158597946167,
      "learning_rate": 7.785399634761778e-06,
      "loss": 0.0551,
      "step": 220650
    },
    {
      "epoch": 3.2499521361982886,
      "grad_norm": 1.2660014629364014,
      "learning_rate": 7.783763262794793e-06,
      "loss": 0.0553,
      "step": 220675
    },
    {
      "epoch": 3.2503203192883756,
      "grad_norm": 1.2932569980621338,
      "learning_rate": 7.782126890827807e-06,
      "loss": 0.0536,
      "step": 220700
    },
    {
      "epoch": 3.2506885023784626,
      "grad_norm": 1.11372971534729,
      "learning_rate": 7.780490518860823e-06,
      "loss": 0.0511,
      "step": 220725
    },
    {
      "epoch": 3.2510566854685496,
      "grad_norm": 1.1959940195083618,
      "learning_rate": 7.77885414689384e-06,
      "loss": 0.0541,
      "step": 220750
    },
    {
      "epoch": 3.2514248685586367,
      "grad_norm": 1.5139520168304443,
      "learning_rate": 7.777217774926856e-06,
      "loss": 0.0592,
      "step": 220775
    },
    {
      "epoch": 3.2517930516487237,
      "grad_norm": 1.145010232925415,
      "learning_rate": 7.77558140295987e-06,
      "loss": 0.0588,
      "step": 220800
    },
    {
      "epoch": 3.2521612347388107,
      "grad_norm": 0.9240149855613708,
      "learning_rate": 7.773945030992885e-06,
      "loss": 0.0485,
      "step": 220825
    },
    {
      "epoch": 3.2525294178288977,
      "grad_norm": 1.4335905313491821,
      "learning_rate": 7.772308659025901e-06,
      "loss": 0.0573,
      "step": 220850
    },
    {
      "epoch": 3.252897600918985,
      "grad_norm": 1.374603509902954,
      "learning_rate": 7.770672287058917e-06,
      "loss": 0.0491,
      "step": 220875
    },
    {
      "epoch": 3.253265784009072,
      "grad_norm": 1.4705373048782349,
      "learning_rate": 7.769035915091933e-06,
      "loss": 0.0537,
      "step": 220900
    },
    {
      "epoch": 3.253633967099159,
      "grad_norm": 1.4508769512176514,
      "learning_rate": 7.767399543124948e-06,
      "loss": 0.061,
      "step": 220925
    },
    {
      "epoch": 3.254002150189246,
      "grad_norm": 1.0141245126724243,
      "learning_rate": 7.765763171157962e-06,
      "loss": 0.0487,
      "step": 220950
    },
    {
      "epoch": 3.2543703332793332,
      "grad_norm": 1.3924833536148071,
      "learning_rate": 7.764126799190978e-06,
      "loss": 0.052,
      "step": 220975
    },
    {
      "epoch": 3.2547385163694202,
      "grad_norm": 1.394154667854309,
      "learning_rate": 7.762490427223994e-06,
      "loss": 0.0537,
      "step": 221000
    },
    {
      "epoch": 3.2551066994595073,
      "grad_norm": 1.1420148611068726,
      "learning_rate": 7.760854055257009e-06,
      "loss": 0.0542,
      "step": 221025
    },
    {
      "epoch": 3.2554748825495943,
      "grad_norm": 1.1677132844924927,
      "learning_rate": 7.759217683290025e-06,
      "loss": 0.0553,
      "step": 221050
    },
    {
      "epoch": 3.2558430656396813,
      "grad_norm": 1.5021580457687378,
      "learning_rate": 7.75758131132304e-06,
      "loss": 0.0505,
      "step": 221075
    },
    {
      "epoch": 3.2562112487297683,
      "grad_norm": 1.3404242992401123,
      "learning_rate": 7.755944939356056e-06,
      "loss": 0.0535,
      "step": 221100
    },
    {
      "epoch": 3.2565794318198553,
      "grad_norm": 1.3072950839996338,
      "learning_rate": 7.75430856738907e-06,
      "loss": 0.0597,
      "step": 221125
    },
    {
      "epoch": 3.2569476149099423,
      "grad_norm": 1.358812689781189,
      "learning_rate": 7.752672195422086e-06,
      "loss": 0.0521,
      "step": 221150
    },
    {
      "epoch": 3.2573157980000293,
      "grad_norm": 0.9844956398010254,
      "learning_rate": 7.751035823455102e-06,
      "loss": 0.0548,
      "step": 221175
    },
    {
      "epoch": 3.2576839810901164,
      "grad_norm": 0.9220600128173828,
      "learning_rate": 7.749399451488119e-06,
      "loss": 0.0549,
      "step": 221200
    },
    {
      "epoch": 3.2580521641802034,
      "grad_norm": 0.9293469190597534,
      "learning_rate": 7.747763079521133e-06,
      "loss": 0.0521,
      "step": 221225
    },
    {
      "epoch": 3.2584203472702904,
      "grad_norm": 2.073119878768921,
      "learning_rate": 7.746126707554147e-06,
      "loss": 0.0565,
      "step": 221250
    },
    {
      "epoch": 3.258788530360378,
      "grad_norm": 1.2151141166687012,
      "learning_rate": 7.744490335587164e-06,
      "loss": 0.0498,
      "step": 221275
    },
    {
      "epoch": 3.259156713450465,
      "grad_norm": 1.2879892587661743,
      "learning_rate": 7.74285396362018e-06,
      "loss": 0.0567,
      "step": 221300
    },
    {
      "epoch": 3.259524896540552,
      "grad_norm": 0.9492918252944946,
      "learning_rate": 7.741217591653194e-06,
      "loss": 0.0536,
      "step": 221325
    },
    {
      "epoch": 3.259893079630639,
      "grad_norm": 0.952765941619873,
      "learning_rate": 7.73958121968621e-06,
      "loss": 0.0495,
      "step": 221350
    },
    {
      "epoch": 3.260261262720726,
      "grad_norm": 1.2155275344848633,
      "learning_rate": 7.737944847719225e-06,
      "loss": 0.06,
      "step": 221375
    },
    {
      "epoch": 3.260629445810813,
      "grad_norm": 0.9193699359893799,
      "learning_rate": 7.736308475752241e-06,
      "loss": 0.0503,
      "step": 221400
    },
    {
      "epoch": 3.2609976289009,
      "grad_norm": 1.2600046396255493,
      "learning_rate": 7.734672103785257e-06,
      "loss": 0.053,
      "step": 221425
    },
    {
      "epoch": 3.261365811990987,
      "grad_norm": 1.485835075378418,
      "learning_rate": 7.733035731818272e-06,
      "loss": 0.0533,
      "step": 221450
    },
    {
      "epoch": 3.261733995081074,
      "grad_norm": 1.4178886413574219,
      "learning_rate": 7.731399359851288e-06,
      "loss": 0.0627,
      "step": 221475
    },
    {
      "epoch": 3.262102178171161,
      "grad_norm": 1.5273215770721436,
      "learning_rate": 7.729762987884302e-06,
      "loss": 0.053,
      "step": 221500
    },
    {
      "epoch": 3.262470361261248,
      "grad_norm": 1.0496892929077148,
      "learning_rate": 7.728126615917318e-06,
      "loss": 0.0589,
      "step": 221525
    },
    {
      "epoch": 3.262838544351335,
      "grad_norm": 1.0900356769561768,
      "learning_rate": 7.726555698829012e-06,
      "loss": 0.0566,
      "step": 221550
    },
    {
      "epoch": 3.263206727441422,
      "grad_norm": 1.4395400285720825,
      "learning_rate": 7.724919326862028e-06,
      "loss": 0.0497,
      "step": 221575
    },
    {
      "epoch": 3.263574910531509,
      "grad_norm": 1.1687188148498535,
      "learning_rate": 7.723282954895044e-06,
      "loss": 0.0484,
      "step": 221600
    },
    {
      "epoch": 3.263943093621596,
      "grad_norm": 1.5220705270767212,
      "learning_rate": 7.721646582928059e-06,
      "loss": 0.0531,
      "step": 221625
    },
    {
      "epoch": 3.264311276711683,
      "grad_norm": 1.4611964225769043,
      "learning_rate": 7.720010210961073e-06,
      "loss": 0.0574,
      "step": 221650
    },
    {
      "epoch": 3.26467945980177,
      "grad_norm": 1.3516196012496948,
      "learning_rate": 7.71837383899409e-06,
      "loss": 0.0591,
      "step": 221675
    },
    {
      "epoch": 3.265047642891857,
      "grad_norm": 1.595139503479004,
      "learning_rate": 7.716737467027106e-06,
      "loss": 0.0592,
      "step": 221700
    },
    {
      "epoch": 3.265415825981944,
      "grad_norm": 1.6821563243865967,
      "learning_rate": 7.715101095060122e-06,
      "loss": 0.0538,
      "step": 221725
    },
    {
      "epoch": 3.265784009072031,
      "grad_norm": 1.530360221862793,
      "learning_rate": 7.713464723093136e-06,
      "loss": 0.0529,
      "step": 221750
    },
    {
      "epoch": 3.266152192162118,
      "grad_norm": 1.4588305950164795,
      "learning_rate": 7.71182835112615e-06,
      "loss": 0.0556,
      "step": 221775
    },
    {
      "epoch": 3.266520375252205,
      "grad_norm": 1.5564113855361938,
      "learning_rate": 7.710191979159167e-06,
      "loss": 0.0526,
      "step": 221800
    },
    {
      "epoch": 3.2668885583422926,
      "grad_norm": 1.1926299333572388,
      "learning_rate": 7.708555607192183e-06,
      "loss": 0.0548,
      "step": 221825
    },
    {
      "epoch": 3.2672567414323797,
      "grad_norm": 0.8720672726631165,
      "learning_rate": 7.7069192352252e-06,
      "loss": 0.0579,
      "step": 221850
    },
    {
      "epoch": 3.2676249245224667,
      "grad_norm": 1.4807361364364624,
      "learning_rate": 7.705282863258214e-06,
      "loss": 0.0551,
      "step": 221875
    },
    {
      "epoch": 3.2679931076125537,
      "grad_norm": 0.8349344730377197,
      "learning_rate": 7.703646491291228e-06,
      "loss": 0.0611,
      "step": 221900
    },
    {
      "epoch": 3.2683612907026407,
      "grad_norm": 1.3450521230697632,
      "learning_rate": 7.702010119324244e-06,
      "loss": 0.0596,
      "step": 221925
    },
    {
      "epoch": 3.2687294737927277,
      "grad_norm": 1.3241890668869019,
      "learning_rate": 7.70037374735726e-06,
      "loss": 0.0576,
      "step": 221950
    },
    {
      "epoch": 3.2690976568828147,
      "grad_norm": 1.565602421760559,
      "learning_rate": 7.698737375390275e-06,
      "loss": 0.0546,
      "step": 221975
    },
    {
      "epoch": 3.2694658399729017,
      "grad_norm": 0.6321175694465637,
      "learning_rate": 7.697101003423291e-06,
      "loss": 0.0529,
      "step": 222000
    },
    {
      "epoch": 3.2698340230629888,
      "grad_norm": 0.8589783906936646,
      "learning_rate": 7.695464631456307e-06,
      "loss": 0.0505,
      "step": 222025
    },
    {
      "epoch": 3.2702022061530758,
      "grad_norm": 1.4351216554641724,
      "learning_rate": 7.693828259489322e-06,
      "loss": 0.0538,
      "step": 222050
    },
    {
      "epoch": 3.270570389243163,
      "grad_norm": 1.1147154569625854,
      "learning_rate": 7.692191887522336e-06,
      "loss": 0.0557,
      "step": 222075
    },
    {
      "epoch": 3.27093857233325,
      "grad_norm": 1.3109734058380127,
      "learning_rate": 7.690555515555352e-06,
      "loss": 0.0634,
      "step": 222100
    },
    {
      "epoch": 3.271306755423337,
      "grad_norm": 1.5449657440185547,
      "learning_rate": 7.688919143588368e-06,
      "loss": 0.05,
      "step": 222125
    },
    {
      "epoch": 3.271674938513424,
      "grad_norm": 1.3285760879516602,
      "learning_rate": 7.687282771621385e-06,
      "loss": 0.0539,
      "step": 222150
    },
    {
      "epoch": 3.272043121603511,
      "grad_norm": 1.5189287662506104,
      "learning_rate": 7.685646399654399e-06,
      "loss": 0.0515,
      "step": 222175
    },
    {
      "epoch": 3.272411304693598,
      "grad_norm": 0.8532837629318237,
      "learning_rate": 7.684010027687413e-06,
      "loss": 0.0499,
      "step": 222200
    },
    {
      "epoch": 3.272779487783685,
      "grad_norm": 1.6627578735351562,
      "learning_rate": 7.68237365572043e-06,
      "loss": 0.0526,
      "step": 222225
    },
    {
      "epoch": 3.2731476708737723,
      "grad_norm": 1.2204079627990723,
      "learning_rate": 7.680737283753446e-06,
      "loss": 0.0464,
      "step": 222250
    },
    {
      "epoch": 3.2735158539638594,
      "grad_norm": 1.2260034084320068,
      "learning_rate": 7.67910091178646e-06,
      "loss": 0.0608,
      "step": 222275
    },
    {
      "epoch": 3.2738840370539464,
      "grad_norm": 1.2964593172073364,
      "learning_rate": 7.677464539819476e-06,
      "loss": 0.051,
      "step": 222300
    },
    {
      "epoch": 3.2742522201440334,
      "grad_norm": 1.4139735698699951,
      "learning_rate": 7.675828167852491e-06,
      "loss": 0.0528,
      "step": 222325
    },
    {
      "epoch": 3.2746204032341204,
      "grad_norm": 0.9862179160118103,
      "learning_rate": 7.674191795885507e-06,
      "loss": 0.0529,
      "step": 222350
    },
    {
      "epoch": 3.2749885863242074,
      "grad_norm": 1.6559228897094727,
      "learning_rate": 7.672555423918523e-06,
      "loss": 0.0565,
      "step": 222375
    },
    {
      "epoch": 3.2753567694142944,
      "grad_norm": 1.298337697982788,
      "learning_rate": 7.670919051951538e-06,
      "loss": 0.0555,
      "step": 222400
    },
    {
      "epoch": 3.2757249525043814,
      "grad_norm": 1.1173251867294312,
      "learning_rate": 7.669282679984554e-06,
      "loss": 0.052,
      "step": 222425
    },
    {
      "epoch": 3.2760931355944685,
      "grad_norm": 1.8743599653244019,
      "learning_rate": 7.667646308017568e-06,
      "loss": 0.0544,
      "step": 222450
    },
    {
      "epoch": 3.2764613186845555,
      "grad_norm": 1.701632022857666,
      "learning_rate": 7.666009936050584e-06,
      "loss": 0.0546,
      "step": 222475
    },
    {
      "epoch": 3.2768295017746425,
      "grad_norm": 1.3790738582611084,
      "learning_rate": 7.664373564083599e-06,
      "loss": 0.0542,
      "step": 222500
    },
    {
      "epoch": 3.2771976848647295,
      "grad_norm": 1.1943485736846924,
      "learning_rate": 7.662737192116615e-06,
      "loss": 0.0577,
      "step": 222525
    },
    {
      "epoch": 3.2775658679548165,
      "grad_norm": 1.142665147781372,
      "learning_rate": 7.661100820149631e-06,
      "loss": 0.0598,
      "step": 222550
    },
    {
      "epoch": 3.2779340510449035,
      "grad_norm": 1.531950831413269,
      "learning_rate": 7.659464448182646e-06,
      "loss": 0.0549,
      "step": 222575
    },
    {
      "epoch": 3.2783022341349906,
      "grad_norm": 1.0448013544082642,
      "learning_rate": 7.657828076215662e-06,
      "loss": 0.0558,
      "step": 222600
    },
    {
      "epoch": 3.2786704172250776,
      "grad_norm": 1.3565757274627686,
      "learning_rate": 7.656191704248676e-06,
      "loss": 0.0508,
      "step": 222625
    },
    {
      "epoch": 3.2790386003151646,
      "grad_norm": 0.9218998551368713,
      "learning_rate": 7.654555332281692e-06,
      "loss": 0.054,
      "step": 222650
    },
    {
      "epoch": 3.2794067834052516,
      "grad_norm": 1.3896846771240234,
      "learning_rate": 7.652918960314709e-06,
      "loss": 0.0561,
      "step": 222675
    },
    {
      "epoch": 3.2797749664953386,
      "grad_norm": 1.2601679563522339,
      "learning_rate": 7.651282588347723e-06,
      "loss": 0.0566,
      "step": 222700
    },
    {
      "epoch": 3.2801431495854256,
      "grad_norm": 1.8283239603042603,
      "learning_rate": 7.649646216380739e-06,
      "loss": 0.0537,
      "step": 222725
    },
    {
      "epoch": 3.2805113326755126,
      "grad_norm": 1.262842059135437,
      "learning_rate": 7.648009844413754e-06,
      "loss": 0.0514,
      "step": 222750
    },
    {
      "epoch": 3.2808795157656,
      "grad_norm": 1.830797791481018,
      "learning_rate": 7.64637347244677e-06,
      "loss": 0.0597,
      "step": 222775
    },
    {
      "epoch": 3.281247698855687,
      "grad_norm": 1.9591783285140991,
      "learning_rate": 7.644737100479784e-06,
      "loss": 0.047,
      "step": 222800
    },
    {
      "epoch": 3.281615881945774,
      "grad_norm": 1.268332600593567,
      "learning_rate": 7.6431007285128e-06,
      "loss": 0.0482,
      "step": 222825
    },
    {
      "epoch": 3.281984065035861,
      "grad_norm": 1.3524540662765503,
      "learning_rate": 7.641464356545816e-06,
      "loss": 0.05,
      "step": 222850
    },
    {
      "epoch": 3.282352248125948,
      "grad_norm": 0.8394463062286377,
      "learning_rate": 7.639827984578831e-06,
      "loss": 0.05,
      "step": 222875
    },
    {
      "epoch": 3.282720431216035,
      "grad_norm": 0.9698912501335144,
      "learning_rate": 7.638191612611847e-06,
      "loss": 0.0587,
      "step": 222900
    },
    {
      "epoch": 3.283088614306122,
      "grad_norm": 1.1424485445022583,
      "learning_rate": 7.636555240644862e-06,
      "loss": 0.0528,
      "step": 222925
    },
    {
      "epoch": 3.283456797396209,
      "grad_norm": 1.3660345077514648,
      "learning_rate": 7.634918868677878e-06,
      "loss": 0.0519,
      "step": 222950
    },
    {
      "epoch": 3.2838249804862962,
      "grad_norm": 1.464294195175171,
      "learning_rate": 7.633282496710894e-06,
      "loss": 0.0526,
      "step": 222975
    },
    {
      "epoch": 3.2841931635763832,
      "grad_norm": 1.4236983060836792,
      "learning_rate": 7.631646124743908e-06,
      "loss": 0.0545,
      "step": 223000
    },
    {
      "epoch": 3.2845613466664703,
      "grad_norm": 1.4798903465270996,
      "learning_rate": 7.630009752776923e-06,
      "loss": 0.0556,
      "step": 223025
    },
    {
      "epoch": 3.2849295297565573,
      "grad_norm": 1.7093205451965332,
      "learning_rate": 7.628373380809939e-06,
      "loss": 0.0548,
      "step": 223050
    },
    {
      "epoch": 3.2852977128466443,
      "grad_norm": 1.6315637826919556,
      "learning_rate": 7.626737008842955e-06,
      "loss": 0.0522,
      "step": 223075
    },
    {
      "epoch": 3.2856658959367313,
      "grad_norm": 1.3786839246749878,
      "learning_rate": 7.62510063687597e-06,
      "loss": 0.0552,
      "step": 223100
    },
    {
      "epoch": 3.2860340790268183,
      "grad_norm": 1.337576985359192,
      "learning_rate": 7.623464264908985e-06,
      "loss": 0.0552,
      "step": 223125
    },
    {
      "epoch": 3.2864022621169053,
      "grad_norm": 1.6407849788665771,
      "learning_rate": 7.621827892942001e-06,
      "loss": 0.0538,
      "step": 223150
    },
    {
      "epoch": 3.2867704452069924,
      "grad_norm": 1.7331223487854004,
      "learning_rate": 7.620191520975016e-06,
      "loss": 0.0566,
      "step": 223175
    },
    {
      "epoch": 3.28713862829708,
      "grad_norm": 1.4217900037765503,
      "learning_rate": 7.6185551490080324e-06,
      "loss": 0.0609,
      "step": 223200
    },
    {
      "epoch": 3.287506811387167,
      "grad_norm": 1.0790061950683594,
      "learning_rate": 7.616918777041047e-06,
      "loss": 0.0574,
      "step": 223225
    },
    {
      "epoch": 3.287874994477254,
      "grad_norm": 1.3927891254425049,
      "learning_rate": 7.615282405074062e-06,
      "loss": 0.0535,
      "step": 223250
    },
    {
      "epoch": 3.288243177567341,
      "grad_norm": 1.0607072114944458,
      "learning_rate": 7.613646033107078e-06,
      "loss": 0.0543,
      "step": 223275
    },
    {
      "epoch": 3.288611360657428,
      "grad_norm": 1.1236480474472046,
      "learning_rate": 7.612009661140094e-06,
      "loss": 0.0567,
      "step": 223300
    },
    {
      "epoch": 3.288979543747515,
      "grad_norm": 1.216027855873108,
      "learning_rate": 7.61037328917311e-06,
      "loss": 0.0479,
      "step": 223325
    },
    {
      "epoch": 3.289347726837602,
      "grad_norm": 0.8837573528289795,
      "learning_rate": 7.608736917206124e-06,
      "loss": 0.0547,
      "step": 223350
    },
    {
      "epoch": 3.289715909927689,
      "grad_norm": 1.6418942213058472,
      "learning_rate": 7.60710054523914e-06,
      "loss": 0.0575,
      "step": 223375
    },
    {
      "epoch": 3.290084093017776,
      "grad_norm": 1.8690115213394165,
      "learning_rate": 7.605464173272156e-06,
      "loss": 0.0563,
      "step": 223400
    },
    {
      "epoch": 3.290452276107863,
      "grad_norm": 1.184334635734558,
      "learning_rate": 7.603827801305171e-06,
      "loss": 0.0504,
      "step": 223425
    },
    {
      "epoch": 3.29082045919795,
      "grad_norm": 1.2403236627578735,
      "learning_rate": 7.602191429338186e-06,
      "loss": 0.0543,
      "step": 223450
    },
    {
      "epoch": 3.291188642288037,
      "grad_norm": 1.2870265245437622,
      "learning_rate": 7.600555057371202e-06,
      "loss": 0.0514,
      "step": 223475
    },
    {
      "epoch": 3.291556825378124,
      "grad_norm": 1.1583762168884277,
      "learning_rate": 7.598918685404218e-06,
      "loss": 0.0509,
      "step": 223500
    },
    {
      "epoch": 3.291925008468211,
      "grad_norm": 1.5692108869552612,
      "learning_rate": 7.597282313437233e-06,
      "loss": 0.0639,
      "step": 223525
    },
    {
      "epoch": 3.292293191558298,
      "grad_norm": 1.4680817127227783,
      "learning_rate": 7.5956459414702476e-06,
      "loss": 0.0563,
      "step": 223550
    },
    {
      "epoch": 3.292661374648385,
      "grad_norm": 0.9752763509750366,
      "learning_rate": 7.594009569503264e-06,
      "loss": 0.0619,
      "step": 223575
    },
    {
      "epoch": 3.293029557738472,
      "grad_norm": 0.8564492464065552,
      "learning_rate": 7.592373197536279e-06,
      "loss": 0.0571,
      "step": 223600
    },
    {
      "epoch": 3.293397740828559,
      "grad_norm": 1.0177819728851318,
      "learning_rate": 7.590736825569295e-06,
      "loss": 0.0602,
      "step": 223625
    },
    {
      "epoch": 3.293765923918646,
      "grad_norm": 1.4920368194580078,
      "learning_rate": 7.58910045360231e-06,
      "loss": 0.0609,
      "step": 223650
    },
    {
      "epoch": 3.294134107008733,
      "grad_norm": 1.2452385425567627,
      "learning_rate": 7.587464081635325e-06,
      "loss": 0.0541,
      "step": 223675
    },
    {
      "epoch": 3.29450229009882,
      "grad_norm": 1.4523124694824219,
      "learning_rate": 7.585827709668341e-06,
      "loss": 0.0534,
      "step": 223700
    },
    {
      "epoch": 3.294870473188907,
      "grad_norm": 1.3547228574752808,
      "learning_rate": 7.584191337701356e-06,
      "loss": 0.053,
      "step": 223725
    },
    {
      "epoch": 3.2952386562789946,
      "grad_norm": 1.2725343704223633,
      "learning_rate": 7.582554965734371e-06,
      "loss": 0.048,
      "step": 223750
    },
    {
      "epoch": 3.2956068393690816,
      "grad_norm": 1.9595698118209839,
      "learning_rate": 7.580918593767387e-06,
      "loss": 0.0587,
      "step": 223775
    },
    {
      "epoch": 3.2959750224591686,
      "grad_norm": 1.6414002180099487,
      "learning_rate": 7.579282221800402e-06,
      "loss": 0.0557,
      "step": 223800
    },
    {
      "epoch": 3.2963432055492556,
      "grad_norm": 1.3708178997039795,
      "learning_rate": 7.5776458498334184e-06,
      "loss": 0.0546,
      "step": 223825
    },
    {
      "epoch": 3.2967113886393427,
      "grad_norm": 1.808488130569458,
      "learning_rate": 7.576009477866434e-06,
      "loss": 0.0584,
      "step": 223850
    },
    {
      "epoch": 3.2970795717294297,
      "grad_norm": 1.3050129413604736,
      "learning_rate": 7.574373105899448e-06,
      "loss": 0.0456,
      "step": 223875
    },
    {
      "epoch": 3.2974477548195167,
      "grad_norm": 0.8597230315208435,
      "learning_rate": 7.572736733932464e-06,
      "loss": 0.0521,
      "step": 223900
    },
    {
      "epoch": 3.2978159379096037,
      "grad_norm": 1.233884334564209,
      "learning_rate": 7.57110036196548e-06,
      "loss": 0.055,
      "step": 223925
    },
    {
      "epoch": 3.2981841209996907,
      "grad_norm": 1.0324082374572754,
      "learning_rate": 7.569463989998496e-06,
      "loss": 0.0531,
      "step": 223950
    },
    {
      "epoch": 3.2985523040897777,
      "grad_norm": 1.320130705833435,
      "learning_rate": 7.56782761803151e-06,
      "loss": 0.0556,
      "step": 223975
    },
    {
      "epoch": 3.2989204871798647,
      "grad_norm": 1.1132978200912476,
      "learning_rate": 7.566191246064526e-06,
      "loss": 0.0532,
      "step": 224000
    },
    {
      "epoch": 3.2992886702699518,
      "grad_norm": 0.8644164800643921,
      "learning_rate": 7.564554874097542e-06,
      "loss": 0.0527,
      "step": 224025
    },
    {
      "epoch": 3.2996568533600388,
      "grad_norm": 1.2989580631256104,
      "learning_rate": 7.562918502130557e-06,
      "loss": 0.0531,
      "step": 224050
    },
    {
      "epoch": 3.300025036450126,
      "grad_norm": 1.2543586492538452,
      "learning_rate": 7.561282130163572e-06,
      "loss": 0.0528,
      "step": 224075
    },
    {
      "epoch": 3.300393219540213,
      "grad_norm": 0.7723637819290161,
      "learning_rate": 7.559645758196588e-06,
      "loss": 0.0503,
      "step": 224100
    },
    {
      "epoch": 3.3007614026303,
      "grad_norm": 1.3822739124298096,
      "learning_rate": 7.558009386229603e-06,
      "loss": 0.0564,
      "step": 224125
    },
    {
      "epoch": 3.3011295857203873,
      "grad_norm": 1.8552175760269165,
      "learning_rate": 7.556373014262619e-06,
      "loss": 0.0587,
      "step": 224150
    },
    {
      "epoch": 3.3014977688104743,
      "grad_norm": 1.2169073820114136,
      "learning_rate": 7.5547366422956336e-06,
      "loss": 0.0583,
      "step": 224175
    },
    {
      "epoch": 3.3018659519005613,
      "grad_norm": 1.2105604410171509,
      "learning_rate": 7.55310027032865e-06,
      "loss": 0.0534,
      "step": 224200
    },
    {
      "epoch": 3.3022341349906483,
      "grad_norm": 0.9494795203208923,
      "learning_rate": 7.551463898361665e-06,
      "loss": 0.0528,
      "step": 224225
    },
    {
      "epoch": 3.3026023180807353,
      "grad_norm": 1.4743846654891968,
      "learning_rate": 7.549827526394681e-06,
      "loss": 0.0533,
      "step": 224250
    },
    {
      "epoch": 3.3029705011708224,
      "grad_norm": 1.7627805471420288,
      "learning_rate": 7.548191154427696e-06,
      "loss": 0.0564,
      "step": 224275
    },
    {
      "epoch": 3.3033386842609094,
      "grad_norm": 1.4553340673446655,
      "learning_rate": 7.546554782460711e-06,
      "loss": 0.0478,
      "step": 224300
    },
    {
      "epoch": 3.3037068673509964,
      "grad_norm": 1.1054805517196655,
      "learning_rate": 7.544918410493727e-06,
      "loss": 0.0603,
      "step": 224325
    },
    {
      "epoch": 3.3040750504410834,
      "grad_norm": 1.101176142692566,
      "learning_rate": 7.543282038526742e-06,
      "loss": 0.0599,
      "step": 224350
    },
    {
      "epoch": 3.3044432335311704,
      "grad_norm": 1.6016474962234497,
      "learning_rate": 7.5416456665597585e-06,
      "loss": 0.0506,
      "step": 224375
    },
    {
      "epoch": 3.3048114166212574,
      "grad_norm": 1.2569962739944458,
      "learning_rate": 7.540009294592773e-06,
      "loss": 0.052,
      "step": 224400
    },
    {
      "epoch": 3.3051795997113445,
      "grad_norm": 1.3738640546798706,
      "learning_rate": 7.538372922625788e-06,
      "loss": 0.0489,
      "step": 224425
    },
    {
      "epoch": 3.3055477828014315,
      "grad_norm": 0.9245179295539856,
      "learning_rate": 7.5367365506588044e-06,
      "loss": 0.0504,
      "step": 224450
    },
    {
      "epoch": 3.3059159658915185,
      "grad_norm": 1.9862191677093506,
      "learning_rate": 7.53510017869182e-06,
      "loss": 0.0522,
      "step": 224475
    },
    {
      "epoch": 3.3062841489816055,
      "grad_norm": 1.3871967792510986,
      "learning_rate": 7.533463806724834e-06,
      "loss": 0.0539,
      "step": 224500
    },
    {
      "epoch": 3.3066523320716925,
      "grad_norm": 1.0422518253326416,
      "learning_rate": 7.53182743475785e-06,
      "loss": 0.0535,
      "step": 224525
    },
    {
      "epoch": 3.3070205151617795,
      "grad_norm": 0.7369499802589417,
      "learning_rate": 7.530191062790866e-06,
      "loss": 0.0526,
      "step": 224550
    },
    {
      "epoch": 3.3073886982518665,
      "grad_norm": 1.549147605895996,
      "learning_rate": 7.528554690823882e-06,
      "loss": 0.0554,
      "step": 224575
    },
    {
      "epoch": 3.3077568813419536,
      "grad_norm": 1.5977694988250732,
      "learning_rate": 7.526918318856896e-06,
      "loss": 0.0555,
      "step": 224600
    },
    {
      "epoch": 3.3081250644320406,
      "grad_norm": 1.2408370971679688,
      "learning_rate": 7.525281946889912e-06,
      "loss": 0.0538,
      "step": 224625
    },
    {
      "epoch": 3.3084932475221276,
      "grad_norm": 1.279453992843628,
      "learning_rate": 7.523645574922928e-06,
      "loss": 0.051,
      "step": 224650
    },
    {
      "epoch": 3.3088614306122146,
      "grad_norm": 1.4527429342269897,
      "learning_rate": 7.522009202955943e-06,
      "loss": 0.0575,
      "step": 224675
    },
    {
      "epoch": 3.309229613702302,
      "grad_norm": 1.3381928205490112,
      "learning_rate": 7.5203728309889575e-06,
      "loss": 0.0541,
      "step": 224700
    },
    {
      "epoch": 3.309597796792389,
      "grad_norm": 0.5282140374183655,
      "learning_rate": 7.518736459021974e-06,
      "loss": 0.0549,
      "step": 224725
    },
    {
      "epoch": 3.309965979882476,
      "grad_norm": 1.9055311679840088,
      "learning_rate": 7.517100087054989e-06,
      "loss": 0.0529,
      "step": 224750
    },
    {
      "epoch": 3.310334162972563,
      "grad_norm": 0.9043813347816467,
      "learning_rate": 7.515463715088005e-06,
      "loss": 0.0584,
      "step": 224775
    },
    {
      "epoch": 3.31070234606265,
      "grad_norm": 1.080385446548462,
      "learning_rate": 7.51382734312102e-06,
      "loss": 0.0663,
      "step": 224800
    },
    {
      "epoch": 3.311070529152737,
      "grad_norm": 1.3524212837219238,
      "learning_rate": 7.512190971154036e-06,
      "loss": 0.0558,
      "step": 224825
    },
    {
      "epoch": 3.311438712242824,
      "grad_norm": 1.1148847341537476,
      "learning_rate": 7.510554599187051e-06,
      "loss": 0.0569,
      "step": 224850
    },
    {
      "epoch": 3.311806895332911,
      "grad_norm": 1.5092246532440186,
      "learning_rate": 7.508918227220066e-06,
      "loss": 0.0605,
      "step": 224875
    },
    {
      "epoch": 3.312175078422998,
      "grad_norm": 1.1484147310256958,
      "learning_rate": 7.5072818552530825e-06,
      "loss": 0.0458,
      "step": 224900
    },
    {
      "epoch": 3.312543261513085,
      "grad_norm": 1.0924971103668213,
      "learning_rate": 7.505645483286097e-06,
      "loss": 0.0524,
      "step": 224925
    },
    {
      "epoch": 3.312911444603172,
      "grad_norm": 1.2831860780715942,
      "learning_rate": 7.504009111319113e-06,
      "loss": 0.056,
      "step": 224950
    },
    {
      "epoch": 3.3132796276932592,
      "grad_norm": 1.0059807300567627,
      "learning_rate": 7.502372739352128e-06,
      "loss": 0.0485,
      "step": 224975
    },
    {
      "epoch": 3.3136478107833462,
      "grad_norm": 1.245866060256958,
      "learning_rate": 7.500736367385144e-06,
      "loss": 0.0537,
      "step": 225000
    },
    {
      "epoch": 3.3140159938734333,
      "grad_norm": 0.3531804084777832,
      "learning_rate": 7.499099995418159e-06,
      "loss": 0.0534,
      "step": 225025
    },
    {
      "epoch": 3.3143841769635203,
      "grad_norm": 1.1652743816375732,
      "learning_rate": 7.497463623451174e-06,
      "loss": 0.0563,
      "step": 225050
    },
    {
      "epoch": 3.3147523600536073,
      "grad_norm": 0.6103206276893616,
      "learning_rate": 7.4958272514841904e-06,
      "loss": 0.0512,
      "step": 225075
    },
    {
      "epoch": 3.3151205431436943,
      "grad_norm": 0.5627954006195068,
      "learning_rate": 7.494190879517206e-06,
      "loss": 0.0495,
      "step": 225100
    },
    {
      "epoch": 3.3154887262337818,
      "grad_norm": 1.6367796659469604,
      "learning_rate": 7.49255450755022e-06,
      "loss": 0.0554,
      "step": 225125
    },
    {
      "epoch": 3.315856909323869,
      "grad_norm": 1.0813084840774536,
      "learning_rate": 7.490918135583236e-06,
      "loss": 0.0508,
      "step": 225150
    },
    {
      "epoch": 3.316225092413956,
      "grad_norm": 0.8711093068122864,
      "learning_rate": 7.489281763616252e-06,
      "loss": 0.0558,
      "step": 225175
    },
    {
      "epoch": 3.316593275504043,
      "grad_norm": 1.6254273653030396,
      "learning_rate": 7.487645391649268e-06,
      "loss": 0.0511,
      "step": 225200
    },
    {
      "epoch": 3.31696145859413,
      "grad_norm": 1.4052600860595703,
      "learning_rate": 7.486009019682282e-06,
      "loss": 0.0543,
      "step": 225225
    },
    {
      "epoch": 3.317329641684217,
      "grad_norm": 1.354114294052124,
      "learning_rate": 7.4843726477152976e-06,
      "loss": 0.0582,
      "step": 225250
    },
    {
      "epoch": 3.317697824774304,
      "grad_norm": 1.7040506601333618,
      "learning_rate": 7.482736275748314e-06,
      "loss": 0.0528,
      "step": 225275
    },
    {
      "epoch": 3.318066007864391,
      "grad_norm": 1.3592195510864258,
      "learning_rate": 7.481099903781329e-06,
      "loss": 0.0515,
      "step": 225300
    },
    {
      "epoch": 3.318434190954478,
      "grad_norm": 1.500755786895752,
      "learning_rate": 7.479463531814345e-06,
      "loss": 0.0506,
      "step": 225325
    },
    {
      "epoch": 3.318802374044565,
      "grad_norm": 1.3757877349853516,
      "learning_rate": 7.47782715984736e-06,
      "loss": 0.0528,
      "step": 225350
    },
    {
      "epoch": 3.319170557134652,
      "grad_norm": 1.2323228120803833,
      "learning_rate": 7.476190787880375e-06,
      "loss": 0.0517,
      "step": 225375
    },
    {
      "epoch": 3.319538740224739,
      "grad_norm": 1.5338979959487915,
      "learning_rate": 7.474554415913391e-06,
      "loss": 0.0597,
      "step": 225400
    },
    {
      "epoch": 3.319906923314826,
      "grad_norm": 1.4789302349090576,
      "learning_rate": 7.472918043946406e-06,
      "loss": 0.0587,
      "step": 225425
    },
    {
      "epoch": 3.320275106404913,
      "grad_norm": 0.9365647435188293,
      "learning_rate": 7.471281671979421e-06,
      "loss": 0.0507,
      "step": 225450
    },
    {
      "epoch": 3.320643289495,
      "grad_norm": 1.2973297834396362,
      "learning_rate": 7.469645300012437e-06,
      "loss": 0.0533,
      "step": 225475
    },
    {
      "epoch": 3.321011472585087,
      "grad_norm": 1.6962288618087769,
      "learning_rate": 7.468008928045452e-06,
      "loss": 0.0508,
      "step": 225500
    },
    {
      "epoch": 3.321379655675174,
      "grad_norm": 1.545454978942871,
      "learning_rate": 7.4663725560784685e-06,
      "loss": 0.0553,
      "step": 225525
    },
    {
      "epoch": 3.321747838765261,
      "grad_norm": 1.3650883436203003,
      "learning_rate": 7.464801638990162e-06,
      "loss": 0.0533,
      "step": 225550
    },
    {
      "epoch": 3.322116021855348,
      "grad_norm": 1.0064499378204346,
      "learning_rate": 7.463165267023178e-06,
      "loss": 0.0446,
      "step": 225575
    },
    {
      "epoch": 3.322484204945435,
      "grad_norm": 1.7375867366790771,
      "learning_rate": 7.461528895056194e-06,
      "loss": 0.0571,
      "step": 225600
    },
    {
      "epoch": 3.322852388035522,
      "grad_norm": 1.4480394124984741,
      "learning_rate": 7.459892523089209e-06,
      "loss": 0.0604,
      "step": 225625
    },
    {
      "epoch": 3.3232205711256095,
      "grad_norm": 1.0826350450515747,
      "learning_rate": 7.458256151122224e-06,
      "loss": 0.0596,
      "step": 225650
    },
    {
      "epoch": 3.3235887542156966,
      "grad_norm": 1.8368470668792725,
      "learning_rate": 7.45661977915524e-06,
      "loss": 0.0588,
      "step": 225675
    },
    {
      "epoch": 3.3239569373057836,
      "grad_norm": 1.5385063886642456,
      "learning_rate": 7.454983407188255e-06,
      "loss": 0.052,
      "step": 225700
    },
    {
      "epoch": 3.3243251203958706,
      "grad_norm": 1.1620228290557861,
      "learning_rate": 7.453347035221271e-06,
      "loss": 0.0578,
      "step": 225725
    },
    {
      "epoch": 3.3246933034859576,
      "grad_norm": 0.9635811448097229,
      "learning_rate": 7.451710663254286e-06,
      "loss": 0.0536,
      "step": 225750
    },
    {
      "epoch": 3.3250614865760446,
      "grad_norm": 1.5155373811721802,
      "learning_rate": 7.450074291287302e-06,
      "loss": 0.0568,
      "step": 225775
    },
    {
      "epoch": 3.3254296696661316,
      "grad_norm": 1.4668232202529907,
      "learning_rate": 7.448437919320317e-06,
      "loss": 0.0574,
      "step": 225800
    },
    {
      "epoch": 3.3257978527562186,
      "grad_norm": 1.0898630619049072,
      "learning_rate": 7.446801547353333e-06,
      "loss": 0.0559,
      "step": 225825
    },
    {
      "epoch": 3.3261660358463057,
      "grad_norm": 1.5009483098983765,
      "learning_rate": 7.4451651753863485e-06,
      "loss": 0.0578,
      "step": 225850
    },
    {
      "epoch": 3.3265342189363927,
      "grad_norm": 1.4300683736801147,
      "learning_rate": 7.443528803419363e-06,
      "loss": 0.0567,
      "step": 225875
    },
    {
      "epoch": 3.3269024020264797,
      "grad_norm": 1.7828330993652344,
      "learning_rate": 7.441892431452379e-06,
      "loss": 0.0556,
      "step": 225900
    },
    {
      "epoch": 3.3272705851165667,
      "grad_norm": 1.6853739023208618,
      "learning_rate": 7.4402560594853945e-06,
      "loss": 0.0485,
      "step": 225925
    },
    {
      "epoch": 3.3276387682066537,
      "grad_norm": 1.880748987197876,
      "learning_rate": 7.438619687518411e-06,
      "loss": 0.0519,
      "step": 225950
    },
    {
      "epoch": 3.3280069512967407,
      "grad_norm": 1.1433602571487427,
      "learning_rate": 7.436983315551425e-06,
      "loss": 0.0534,
      "step": 225975
    },
    {
      "epoch": 3.3283751343868277,
      "grad_norm": 1.6150449514389038,
      "learning_rate": 7.43534694358444e-06,
      "loss": 0.0561,
      "step": 226000
    },
    {
      "epoch": 3.3287433174769148,
      "grad_norm": 1.256623387336731,
      "learning_rate": 7.4337105716174565e-06,
      "loss": 0.0546,
      "step": 226025
    },
    {
      "epoch": 3.329111500567002,
      "grad_norm": 1.558573603630066,
      "learning_rate": 7.432074199650472e-06,
      "loss": 0.0546,
      "step": 226050
    },
    {
      "epoch": 3.3294796836570892,
      "grad_norm": 1.0619813203811646,
      "learning_rate": 7.430437827683486e-06,
      "loss": 0.0486,
      "step": 226075
    },
    {
      "epoch": 3.3298478667471763,
      "grad_norm": 1.3027280569076538,
      "learning_rate": 7.4288014557165024e-06,
      "loss": 0.0578,
      "step": 226100
    },
    {
      "epoch": 3.3302160498372633,
      "grad_norm": 1.3097243309020996,
      "learning_rate": 7.427165083749518e-06,
      "loss": 0.0589,
      "step": 226125
    },
    {
      "epoch": 3.3305842329273503,
      "grad_norm": 1.410872220993042,
      "learning_rate": 7.425528711782534e-06,
      "loss": 0.0492,
      "step": 226150
    },
    {
      "epoch": 3.3309524160174373,
      "grad_norm": 1.0632275342941284,
      "learning_rate": 7.423892339815548e-06,
      "loss": 0.0482,
      "step": 226175
    },
    {
      "epoch": 3.3313205991075243,
      "grad_norm": 1.3405839204788208,
      "learning_rate": 7.422255967848564e-06,
      "loss": 0.0573,
      "step": 226200
    },
    {
      "epoch": 3.3316887821976113,
      "grad_norm": 1.5575217008590698,
      "learning_rate": 7.42061959588158e-06,
      "loss": 0.0601,
      "step": 226225
    },
    {
      "epoch": 3.3320569652876983,
      "grad_norm": 1.3362696170806885,
      "learning_rate": 7.418983223914595e-06,
      "loss": 0.0546,
      "step": 226250
    },
    {
      "epoch": 3.3324251483777854,
      "grad_norm": 1.7967804670333862,
      "learning_rate": 7.417346851947611e-06,
      "loss": 0.0626,
      "step": 226275
    },
    {
      "epoch": 3.3327933314678724,
      "grad_norm": 1.905088186264038,
      "learning_rate": 7.415710479980626e-06,
      "loss": 0.0691,
      "step": 226300
    },
    {
      "epoch": 3.3331615145579594,
      "grad_norm": 1.680367112159729,
      "learning_rate": 7.414074108013641e-06,
      "loss": 0.0545,
      "step": 226325
    },
    {
      "epoch": 3.3335296976480464,
      "grad_norm": 1.2259553670883179,
      "learning_rate": 7.412437736046657e-06,
      "loss": 0.0588,
      "step": 226350
    },
    {
      "epoch": 3.3338978807381334,
      "grad_norm": 1.1680890321731567,
      "learning_rate": 7.4108013640796725e-06,
      "loss": 0.054,
      "step": 226375
    },
    {
      "epoch": 3.3342660638282204,
      "grad_norm": 1.0716922283172607,
      "learning_rate": 7.409164992112688e-06,
      "loss": 0.0569,
      "step": 226400
    },
    {
      "epoch": 3.3346342469183075,
      "grad_norm": 1.0347439050674438,
      "learning_rate": 7.407528620145703e-06,
      "loss": 0.0546,
      "step": 226425
    },
    {
      "epoch": 3.3350024300083945,
      "grad_norm": 1.017482876777649,
      "learning_rate": 7.405892248178718e-06,
      "loss": 0.0581,
      "step": 226450
    },
    {
      "epoch": 3.3353706130984815,
      "grad_norm": 1.238437533378601,
      "learning_rate": 7.4042558762117345e-06,
      "loss": 0.0472,
      "step": 226475
    },
    {
      "epoch": 3.3357387961885685,
      "grad_norm": 1.1666393280029297,
      "learning_rate": 7.402619504244749e-06,
      "loss": 0.0502,
      "step": 226500
    },
    {
      "epoch": 3.3361069792786555,
      "grad_norm": 1.5748651027679443,
      "learning_rate": 7.400983132277765e-06,
      "loss": 0.0515,
      "step": 226525
    },
    {
      "epoch": 3.3364751623687425,
      "grad_norm": 1.6308422088623047,
      "learning_rate": 7.3993467603107805e-06,
      "loss": 0.056,
      "step": 226550
    },
    {
      "epoch": 3.3368433454588295,
      "grad_norm": 1.0964595079421997,
      "learning_rate": 7.397710388343796e-06,
      "loss": 0.0559,
      "step": 226575
    },
    {
      "epoch": 3.3372115285489166,
      "grad_norm": 1.0749626159667969,
      "learning_rate": 7.396074016376811e-06,
      "loss": 0.0528,
      "step": 226600
    },
    {
      "epoch": 3.337579711639004,
      "grad_norm": 1.2637847661972046,
      "learning_rate": 7.394437644409826e-06,
      "loss": 0.0524,
      "step": 226625
    },
    {
      "epoch": 3.337947894729091,
      "grad_norm": 1.138476014137268,
      "learning_rate": 7.3928012724428425e-06,
      "loss": 0.0512,
      "step": 226650
    },
    {
      "epoch": 3.338316077819178,
      "grad_norm": 1.6251795291900635,
      "learning_rate": 7.391164900475858e-06,
      "loss": 0.0581,
      "step": 226675
    },
    {
      "epoch": 3.338684260909265,
      "grad_norm": 1.6297962665557861,
      "learning_rate": 7.389528528508872e-06,
      "loss": 0.0478,
      "step": 226700
    },
    {
      "epoch": 3.339052443999352,
      "grad_norm": 1.0838072299957275,
      "learning_rate": 7.387957611420568e-06,
      "loss": 0.0564,
      "step": 226725
    },
    {
      "epoch": 3.339420627089439,
      "grad_norm": 1.4168065786361694,
      "learning_rate": 7.386321239453583e-06,
      "loss": 0.0536,
      "step": 226750
    },
    {
      "epoch": 3.339788810179526,
      "grad_norm": 1.5672047138214111,
      "learning_rate": 7.384684867486599e-06,
      "loss": 0.052,
      "step": 226775
    },
    {
      "epoch": 3.340156993269613,
      "grad_norm": 1.6075565814971924,
      "learning_rate": 7.383048495519615e-06,
      "loss": 0.0557,
      "step": 226800
    },
    {
      "epoch": 3.3405251763597,
      "grad_norm": 1.1818798780441284,
      "learning_rate": 7.381412123552629e-06,
      "loss": 0.061,
      "step": 226825
    },
    {
      "epoch": 3.340893359449787,
      "grad_norm": 1.5154752731323242,
      "learning_rate": 7.379775751585645e-06,
      "loss": 0.0622,
      "step": 226850
    },
    {
      "epoch": 3.341261542539874,
      "grad_norm": 1.3670189380645752,
      "learning_rate": 7.3781393796186605e-06,
      "loss": 0.0515,
      "step": 226875
    },
    {
      "epoch": 3.341629725629961,
      "grad_norm": 1.474528193473816,
      "learning_rate": 7.376503007651677e-06,
      "loss": 0.0562,
      "step": 226900
    },
    {
      "epoch": 3.341997908720048,
      "grad_norm": 1.8051512241363525,
      "learning_rate": 7.374866635684691e-06,
      "loss": 0.0547,
      "step": 226925
    },
    {
      "epoch": 3.342366091810135,
      "grad_norm": 1.8945709466934204,
      "learning_rate": 7.3732302637177064e-06,
      "loss": 0.058,
      "step": 226950
    },
    {
      "epoch": 3.3427342749002222,
      "grad_norm": 1.313443660736084,
      "learning_rate": 7.371593891750723e-06,
      "loss": 0.0502,
      "step": 226975
    },
    {
      "epoch": 3.3431024579903093,
      "grad_norm": 1.4861992597579956,
      "learning_rate": 7.369957519783738e-06,
      "loss": 0.0553,
      "step": 227000
    },
    {
      "epoch": 3.3434706410803963,
      "grad_norm": 1.0212277173995972,
      "learning_rate": 7.368321147816752e-06,
      "loss": 0.0526,
      "step": 227025
    },
    {
      "epoch": 3.3438388241704837,
      "grad_norm": 1.5201823711395264,
      "learning_rate": 7.3666847758497685e-06,
      "loss": 0.0617,
      "step": 227050
    },
    {
      "epoch": 3.3442070072605707,
      "grad_norm": 0.9945889711380005,
      "learning_rate": 7.365048403882784e-06,
      "loss": 0.0525,
      "step": 227075
    },
    {
      "epoch": 3.3445751903506578,
      "grad_norm": 0.9937597513198853,
      "learning_rate": 7.3634120319158e-06,
      "loss": 0.0508,
      "step": 227100
    },
    {
      "epoch": 3.3449433734407448,
      "grad_norm": 1.2471506595611572,
      "learning_rate": 7.361775659948814e-06,
      "loss": 0.0567,
      "step": 227125
    },
    {
      "epoch": 3.345311556530832,
      "grad_norm": 1.2965233325958252,
      "learning_rate": 7.36013928798183e-06,
      "loss": 0.0561,
      "step": 227150
    },
    {
      "epoch": 3.345679739620919,
      "grad_norm": 1.4884933233261108,
      "learning_rate": 7.358502916014846e-06,
      "loss": 0.0543,
      "step": 227175
    },
    {
      "epoch": 3.346047922711006,
      "grad_norm": 0.7720845937728882,
      "learning_rate": 7.356866544047861e-06,
      "loss": 0.0554,
      "step": 227200
    },
    {
      "epoch": 3.346416105801093,
      "grad_norm": 1.334080457687378,
      "learning_rate": 7.355230172080877e-06,
      "loss": 0.0561,
      "step": 227225
    },
    {
      "epoch": 3.34678428889118,
      "grad_norm": 1.4637643098831177,
      "learning_rate": 7.353593800113892e-06,
      "loss": 0.0572,
      "step": 227250
    },
    {
      "epoch": 3.347152471981267,
      "grad_norm": 1.6052371263504028,
      "learning_rate": 7.351957428146907e-06,
      "loss": 0.0499,
      "step": 227275
    },
    {
      "epoch": 3.347520655071354,
      "grad_norm": 1.3566983938217163,
      "learning_rate": 7.350321056179923e-06,
      "loss": 0.0542,
      "step": 227300
    },
    {
      "epoch": 3.347888838161441,
      "grad_norm": 1.2417707443237305,
      "learning_rate": 7.3486846842129385e-06,
      "loss": 0.0595,
      "step": 227325
    },
    {
      "epoch": 3.348257021251528,
      "grad_norm": 1.0273051261901855,
      "learning_rate": 7.347048312245954e-06,
      "loss": 0.0545,
      "step": 227350
    },
    {
      "epoch": 3.348625204341615,
      "grad_norm": 1.5369634628295898,
      "learning_rate": 7.345411940278969e-06,
      "loss": 0.049,
      "step": 227375
    },
    {
      "epoch": 3.348993387431702,
      "grad_norm": 0.9035630822181702,
      "learning_rate": 7.343775568311985e-06,
      "loss": 0.0541,
      "step": 227400
    },
    {
      "epoch": 3.349361570521789,
      "grad_norm": 1.1546244621276855,
      "learning_rate": 7.342139196345001e-06,
      "loss": 0.0559,
      "step": 227425
    },
    {
      "epoch": 3.349729753611876,
      "grad_norm": 1.3052910566329956,
      "learning_rate": 7.340502824378015e-06,
      "loss": 0.0568,
      "step": 227450
    },
    {
      "epoch": 3.350097936701963,
      "grad_norm": 1.5565319061279297,
      "learning_rate": 7.338866452411031e-06,
      "loss": 0.0546,
      "step": 227475
    },
    {
      "epoch": 3.35046611979205,
      "grad_norm": 1.9359947443008423,
      "learning_rate": 7.3372300804440465e-06,
      "loss": 0.0556,
      "step": 227500
    },
    {
      "epoch": 3.350834302882137,
      "grad_norm": 1.249371886253357,
      "learning_rate": 7.335593708477063e-06,
      "loss": 0.0524,
      "step": 227525
    },
    {
      "epoch": 3.351202485972224,
      "grad_norm": 1.4207818508148193,
      "learning_rate": 7.333957336510077e-06,
      "loss": 0.0494,
      "step": 227550
    },
    {
      "epoch": 3.3515706690623115,
      "grad_norm": 1.0526865720748901,
      "learning_rate": 7.3323209645430924e-06,
      "loss": 0.057,
      "step": 227575
    },
    {
      "epoch": 3.3519388521523985,
      "grad_norm": 1.489443063735962,
      "learning_rate": 7.330684592576109e-06,
      "loss": 0.0591,
      "step": 227600
    },
    {
      "epoch": 3.3523070352424855,
      "grad_norm": 0.8755283355712891,
      "learning_rate": 7.329048220609124e-06,
      "loss": 0.0584,
      "step": 227625
    },
    {
      "epoch": 3.3526752183325725,
      "grad_norm": 1.4304816722869873,
      "learning_rate": 7.327411848642138e-06,
      "loss": 0.0529,
      "step": 227650
    },
    {
      "epoch": 3.3530434014226596,
      "grad_norm": 0.8949004411697388,
      "learning_rate": 7.3257754766751545e-06,
      "loss": 0.0477,
      "step": 227675
    },
    {
      "epoch": 3.3534115845127466,
      "grad_norm": 1.4937458038330078,
      "learning_rate": 7.32413910470817e-06,
      "loss": 0.0555,
      "step": 227700
    },
    {
      "epoch": 3.3537797676028336,
      "grad_norm": 1.59857976436615,
      "learning_rate": 7.322502732741186e-06,
      "loss": 0.0613,
      "step": 227725
    },
    {
      "epoch": 3.3541479506929206,
      "grad_norm": 1.2917006015777588,
      "learning_rate": 7.320866360774201e-06,
      "loss": 0.0516,
      "step": 227750
    },
    {
      "epoch": 3.3545161337830076,
      "grad_norm": 1.10425865650177,
      "learning_rate": 7.319229988807216e-06,
      "loss": 0.0515,
      "step": 227775
    },
    {
      "epoch": 3.3548843168730946,
      "grad_norm": 1.6088265180587769,
      "learning_rate": 7.317593616840232e-06,
      "loss": 0.0585,
      "step": 227800
    },
    {
      "epoch": 3.3552524999631816,
      "grad_norm": 1.1360076665878296,
      "learning_rate": 7.315957244873247e-06,
      "loss": 0.056,
      "step": 227825
    },
    {
      "epoch": 3.3556206830532687,
      "grad_norm": 1.2189058065414429,
      "learning_rate": 7.314320872906263e-06,
      "loss": 0.0539,
      "step": 227850
    },
    {
      "epoch": 3.3559888661433557,
      "grad_norm": 0.9098104238510132,
      "learning_rate": 7.312684500939278e-06,
      "loss": 0.0617,
      "step": 227875
    },
    {
      "epoch": 3.3563570492334427,
      "grad_norm": 1.2700965404510498,
      "learning_rate": 7.311048128972293e-06,
      "loss": 0.0516,
      "step": 227900
    },
    {
      "epoch": 3.3567252323235297,
      "grad_norm": 0.8122754693031311,
      "learning_rate": 7.309411757005309e-06,
      "loss": 0.0552,
      "step": 227925
    },
    {
      "epoch": 3.3570934154136167,
      "grad_norm": 1.3640238046646118,
      "learning_rate": 7.3077753850383245e-06,
      "loss": 0.054,
      "step": 227950
    },
    {
      "epoch": 3.3574615985037037,
      "grad_norm": 1.0851056575775146,
      "learning_rate": 7.30613901307134e-06,
      "loss": 0.05,
      "step": 227975
    },
    {
      "epoch": 3.357829781593791,
      "grad_norm": 1.0979098081588745,
      "learning_rate": 7.304502641104355e-06,
      "loss": 0.0485,
      "step": 228000
    },
    {
      "epoch": 3.358197964683878,
      "grad_norm": 1.2680208683013916,
      "learning_rate": 7.3028662691373705e-06,
      "loss": 0.0552,
      "step": 228025
    },
    {
      "epoch": 3.3585661477739652,
      "grad_norm": 1.0618665218353271,
      "learning_rate": 7.301229897170387e-06,
      "loss": 0.0533,
      "step": 228050
    },
    {
      "epoch": 3.3589343308640522,
      "grad_norm": 1.4491623640060425,
      "learning_rate": 7.299593525203401e-06,
      "loss": 0.0573,
      "step": 228075
    },
    {
      "epoch": 3.3593025139541393,
      "grad_norm": 1.700790524482727,
      "learning_rate": 7.297957153236417e-06,
      "loss": 0.0519,
      "step": 228100
    },
    {
      "epoch": 3.3596706970442263,
      "grad_norm": 1.0543335676193237,
      "learning_rate": 7.2963207812694325e-06,
      "loss": 0.062,
      "step": 228125
    },
    {
      "epoch": 3.3600388801343133,
      "grad_norm": 1.0651147365570068,
      "learning_rate": 7.294684409302449e-06,
      "loss": 0.0483,
      "step": 228150
    },
    {
      "epoch": 3.3604070632244003,
      "grad_norm": 1.6682809591293335,
      "learning_rate": 7.293048037335463e-06,
      "loss": 0.0528,
      "step": 228175
    },
    {
      "epoch": 3.3607752463144873,
      "grad_norm": 1.1392656564712524,
      "learning_rate": 7.2914116653684784e-06,
      "loss": 0.0465,
      "step": 228200
    },
    {
      "epoch": 3.3611434294045743,
      "grad_norm": 1.4422379732131958,
      "learning_rate": 7.289775293401495e-06,
      "loss": 0.0562,
      "step": 228225
    },
    {
      "epoch": 3.3615116124946613,
      "grad_norm": 1.3540560007095337,
      "learning_rate": 7.28813892143451e-06,
      "loss": 0.0502,
      "step": 228250
    },
    {
      "epoch": 3.3618797955847484,
      "grad_norm": 1.4447195529937744,
      "learning_rate": 7.286502549467526e-06,
      "loss": 0.0609,
      "step": 228275
    },
    {
      "epoch": 3.3622479786748354,
      "grad_norm": 1.2681094408035278,
      "learning_rate": 7.2848661775005405e-06,
      "loss": 0.0583,
      "step": 228300
    },
    {
      "epoch": 3.3626161617649224,
      "grad_norm": 1.6062184572219849,
      "learning_rate": 7.283229805533556e-06,
      "loss": 0.0527,
      "step": 228325
    },
    {
      "epoch": 3.3629843448550094,
      "grad_norm": 1.0012041330337524,
      "learning_rate": 7.281593433566572e-06,
      "loss": 0.06,
      "step": 228350
    },
    {
      "epoch": 3.3633525279450964,
      "grad_norm": 1.3650450706481934,
      "learning_rate": 7.279957061599587e-06,
      "loss": 0.06,
      "step": 228375
    },
    {
      "epoch": 3.3637207110351834,
      "grad_norm": 1.7142605781555176,
      "learning_rate": 7.278320689632602e-06,
      "loss": 0.0503,
      "step": 228400
    },
    {
      "epoch": 3.3640888941252705,
      "grad_norm": 1.285312294960022,
      "learning_rate": 7.276684317665618e-06,
      "loss": 0.0504,
      "step": 228425
    },
    {
      "epoch": 3.3644570772153575,
      "grad_norm": 1.5487838983535767,
      "learning_rate": 7.275047945698633e-06,
      "loss": 0.0513,
      "step": 228450
    },
    {
      "epoch": 3.3648252603054445,
      "grad_norm": 0.7486552596092224,
      "learning_rate": 7.273411573731649e-06,
      "loss": 0.0554,
      "step": 228475
    },
    {
      "epoch": 3.3651934433955315,
      "grad_norm": 0.9599723219871521,
      "learning_rate": 7.271775201764664e-06,
      "loss": 0.0551,
      "step": 228500
    },
    {
      "epoch": 3.365561626485619,
      "grad_norm": 1.0872753858566284,
      "learning_rate": 7.270138829797679e-06,
      "loss": 0.0523,
      "step": 228525
    },
    {
      "epoch": 3.365929809575706,
      "grad_norm": 1.714653491973877,
      "learning_rate": 7.268502457830695e-06,
      "loss": 0.0519,
      "step": 228550
    },
    {
      "epoch": 3.366297992665793,
      "grad_norm": 0.8374559879302979,
      "learning_rate": 7.2668660858637105e-06,
      "loss": 0.0543,
      "step": 228575
    },
    {
      "epoch": 3.36666617575588,
      "grad_norm": 1.2753009796142578,
      "learning_rate": 7.265229713896725e-06,
      "loss": 0.0565,
      "step": 228600
    },
    {
      "epoch": 3.367034358845967,
      "grad_norm": 1.3866500854492188,
      "learning_rate": 7.263593341929741e-06,
      "loss": 0.0528,
      "step": 228625
    },
    {
      "epoch": 3.367402541936054,
      "grad_norm": 1.2218362092971802,
      "learning_rate": 7.2619569699627565e-06,
      "loss": 0.0536,
      "step": 228650
    },
    {
      "epoch": 3.367770725026141,
      "grad_norm": 1.0911751985549927,
      "learning_rate": 7.260320597995773e-06,
      "loss": 0.0455,
      "step": 228675
    },
    {
      "epoch": 3.368138908116228,
      "grad_norm": 1.4038351774215698,
      "learning_rate": 7.258684226028788e-06,
      "loss": 0.049,
      "step": 228700
    },
    {
      "epoch": 3.368507091206315,
      "grad_norm": 1.325892448425293,
      "learning_rate": 7.257047854061802e-06,
      "loss": 0.0555,
      "step": 228725
    },
    {
      "epoch": 3.368875274296402,
      "grad_norm": 1.778721809387207,
      "learning_rate": 7.2554114820948185e-06,
      "loss": 0.0554,
      "step": 228750
    },
    {
      "epoch": 3.369243457386489,
      "grad_norm": 1.1900590658187866,
      "learning_rate": 7.253775110127834e-06,
      "loss": 0.0571,
      "step": 228775
    },
    {
      "epoch": 3.369611640476576,
      "grad_norm": 1.177790880203247,
      "learning_rate": 7.25213873816085e-06,
      "loss": 0.055,
      "step": 228800
    },
    {
      "epoch": 3.369979823566663,
      "grad_norm": 1.2539998292922974,
      "learning_rate": 7.2505023661938644e-06,
      "loss": 0.0593,
      "step": 228825
    },
    {
      "epoch": 3.37034800665675,
      "grad_norm": 1.387547492980957,
      "learning_rate": 7.248865994226881e-06,
      "loss": 0.0464,
      "step": 228850
    },
    {
      "epoch": 3.370716189746837,
      "grad_norm": 1.865710973739624,
      "learning_rate": 7.247229622259896e-06,
      "loss": 0.0544,
      "step": 228875
    },
    {
      "epoch": 3.371084372836924,
      "grad_norm": 1.3285530805587769,
      "learning_rate": 7.245593250292911e-06,
      "loss": 0.056,
      "step": 228900
    },
    {
      "epoch": 3.371452555927011,
      "grad_norm": 0.9275400638580322,
      "learning_rate": 7.2439568783259265e-06,
      "loss": 0.0556,
      "step": 228925
    },
    {
      "epoch": 3.3718207390170987,
      "grad_norm": 1.1684253215789795,
      "learning_rate": 7.242320506358942e-06,
      "loss": 0.051,
      "step": 228950
    },
    {
      "epoch": 3.3721889221071857,
      "grad_norm": 1.2132936716079712,
      "learning_rate": 7.240684134391958e-06,
      "loss": 0.0516,
      "step": 228975
    },
    {
      "epoch": 3.3725571051972727,
      "grad_norm": 1.4304368495941162,
      "learning_rate": 7.239047762424973e-06,
      "loss": 0.0638,
      "step": 229000
    },
    {
      "epoch": 3.3729252882873597,
      "grad_norm": 1.1957331895828247,
      "learning_rate": 7.237411390457988e-06,
      "loss": 0.0504,
      "step": 229025
    },
    {
      "epoch": 3.3732934713774467,
      "grad_norm": 1.043702244758606,
      "learning_rate": 7.235775018491004e-06,
      "loss": 0.0557,
      "step": 229050
    },
    {
      "epoch": 3.3736616544675337,
      "grad_norm": 1.3535196781158447,
      "learning_rate": 7.234138646524019e-06,
      "loss": 0.054,
      "step": 229075
    },
    {
      "epoch": 3.3740298375576208,
      "grad_norm": 1.148145318031311,
      "learning_rate": 7.232502274557035e-06,
      "loss": 0.0523,
      "step": 229100
    },
    {
      "epoch": 3.3743980206477078,
      "grad_norm": 0.9355340003967285,
      "learning_rate": 7.23086590259005e-06,
      "loss": 0.0522,
      "step": 229125
    },
    {
      "epoch": 3.374766203737795,
      "grad_norm": 1.1868191957473755,
      "learning_rate": 7.229229530623065e-06,
      "loss": 0.0605,
      "step": 229150
    },
    {
      "epoch": 3.375134386827882,
      "grad_norm": 1.4411094188690186,
      "learning_rate": 7.227593158656081e-06,
      "loss": 0.0529,
      "step": 229175
    },
    {
      "epoch": 3.375502569917969,
      "grad_norm": 1.2888022661209106,
      "learning_rate": 7.2259567866890965e-06,
      "loss": 0.0558,
      "step": 229200
    },
    {
      "epoch": 3.375870753008056,
      "grad_norm": 1.340494155883789,
      "learning_rate": 7.224320414722113e-06,
      "loss": 0.0562,
      "step": 229225
    },
    {
      "epoch": 3.376238936098143,
      "grad_norm": 1.0565993785858154,
      "learning_rate": 7.222684042755127e-06,
      "loss": 0.0558,
      "step": 229250
    },
    {
      "epoch": 3.37660711918823,
      "grad_norm": 1.1885859966278076,
      "learning_rate": 7.2210476707881425e-06,
      "loss": 0.0568,
      "step": 229275
    },
    {
      "epoch": 3.376975302278317,
      "grad_norm": 1.6759604215621948,
      "learning_rate": 7.219411298821159e-06,
      "loss": 0.0489,
      "step": 229300
    },
    {
      "epoch": 3.377343485368404,
      "grad_norm": 1.3945995569229126,
      "learning_rate": 7.217774926854174e-06,
      "loss": 0.0625,
      "step": 229325
    },
    {
      "epoch": 3.377711668458491,
      "grad_norm": 1.4647125005722046,
      "learning_rate": 7.216138554887188e-06,
      "loss": 0.0607,
      "step": 229350
    },
    {
      "epoch": 3.378079851548578,
      "grad_norm": 1.204837679862976,
      "learning_rate": 7.2145021829202045e-06,
      "loss": 0.0607,
      "step": 229375
    },
    {
      "epoch": 3.378448034638665,
      "grad_norm": 0.9089778661727905,
      "learning_rate": 7.21286581095322e-06,
      "loss": 0.051,
      "step": 229400
    },
    {
      "epoch": 3.378816217728752,
      "grad_norm": 1.097982406616211,
      "learning_rate": 7.211229438986236e-06,
      "loss": 0.0529,
      "step": 229425
    },
    {
      "epoch": 3.379184400818839,
      "grad_norm": 0.9045083522796631,
      "learning_rate": 7.2095930670192504e-06,
      "loss": 0.0499,
      "step": 229450
    },
    {
      "epoch": 3.379552583908926,
      "grad_norm": 1.2514936923980713,
      "learning_rate": 7.207956695052266e-06,
      "loss": 0.0509,
      "step": 229475
    },
    {
      "epoch": 3.3799207669990134,
      "grad_norm": 1.1990993022918701,
      "learning_rate": 7.206320323085282e-06,
      "loss": 0.0557,
      "step": 229500
    },
    {
      "epoch": 3.3802889500891005,
      "grad_norm": 1.3038480281829834,
      "learning_rate": 7.204683951118297e-06,
      "loss": 0.0618,
      "step": 229525
    },
    {
      "epoch": 3.3806571331791875,
      "grad_norm": 0.908562421798706,
      "learning_rate": 7.2030475791513125e-06,
      "loss": 0.0572,
      "step": 229550
    },
    {
      "epoch": 3.3810253162692745,
      "grad_norm": 1.503128170967102,
      "learning_rate": 7.201411207184328e-06,
      "loss": 0.0548,
      "step": 229575
    },
    {
      "epoch": 3.3813934993593615,
      "grad_norm": 1.3939077854156494,
      "learning_rate": 7.199774835217344e-06,
      "loss": 0.0553,
      "step": 229600
    },
    {
      "epoch": 3.3817616824494485,
      "grad_norm": 1.074191927909851,
      "learning_rate": 7.198138463250359e-06,
      "loss": 0.0559,
      "step": 229625
    },
    {
      "epoch": 3.3821298655395355,
      "grad_norm": 1.2822694778442383,
      "learning_rate": 7.196502091283374e-06,
      "loss": 0.0501,
      "step": 229650
    },
    {
      "epoch": 3.3824980486296226,
      "grad_norm": 1.416955590248108,
      "learning_rate": 7.19486571931639e-06,
      "loss": 0.0548,
      "step": 229675
    },
    {
      "epoch": 3.3828662317197096,
      "grad_norm": 1.4085408449172974,
      "learning_rate": 7.193229347349405e-06,
      "loss": 0.0484,
      "step": 229700
    },
    {
      "epoch": 3.3832344148097966,
      "grad_norm": 1.1416951417922974,
      "learning_rate": 7.191592975382421e-06,
      "loss": 0.0538,
      "step": 229725
    },
    {
      "epoch": 3.3836025978998836,
      "grad_norm": 1.8507288694381714,
      "learning_rate": 7.189956603415437e-06,
      "loss": 0.0531,
      "step": 229750
    },
    {
      "epoch": 3.3839707809899706,
      "grad_norm": 1.6278001070022583,
      "learning_rate": 7.188320231448451e-06,
      "loss": 0.063,
      "step": 229775
    },
    {
      "epoch": 3.3843389640800576,
      "grad_norm": 1.0278292894363403,
      "learning_rate": 7.186683859481467e-06,
      "loss": 0.0528,
      "step": 229800
    },
    {
      "epoch": 3.3847071471701446,
      "grad_norm": 0.9253020286560059,
      "learning_rate": 7.1850474875144825e-06,
      "loss": 0.0529,
      "step": 229825
    },
    {
      "epoch": 3.3850753302602317,
      "grad_norm": 1.118381142616272,
      "learning_rate": 7.183411115547499e-06,
      "loss": 0.0489,
      "step": 229850
    },
    {
      "epoch": 3.3854435133503187,
      "grad_norm": 1.3826417922973633,
      "learning_rate": 7.181774743580513e-06,
      "loss": 0.0564,
      "step": 229875
    },
    {
      "epoch": 3.3858116964404057,
      "grad_norm": 1.6267495155334473,
      "learning_rate": 7.1801383716135285e-06,
      "loss": 0.0496,
      "step": 229900
    },
    {
      "epoch": 3.386179879530493,
      "grad_norm": 1.306330919265747,
      "learning_rate": 7.178501999646545e-06,
      "loss": 0.0495,
      "step": 229925
    },
    {
      "epoch": 3.38654806262058,
      "grad_norm": 1.3307965993881226,
      "learning_rate": 7.17686562767956e-06,
      "loss": 0.0487,
      "step": 229950
    },
    {
      "epoch": 3.386916245710667,
      "grad_norm": 0.4965097904205322,
      "learning_rate": 7.175229255712574e-06,
      "loss": 0.0506,
      "step": 229975
    },
    {
      "epoch": 3.387284428800754,
      "grad_norm": 1.696635127067566,
      "learning_rate": 7.1735928837455905e-06,
      "loss": 0.0596,
      "step": 230000
    },
    {
      "epoch": 3.387652611890841,
      "grad_norm": 1.322624683380127,
      "learning_rate": 7.171956511778606e-06,
      "loss": 0.0526,
      "step": 230025
    },
    {
      "epoch": 3.3880207949809282,
      "grad_norm": 1.0035916566848755,
      "learning_rate": 7.170320139811622e-06,
      "loss": 0.0582,
      "step": 230050
    },
    {
      "epoch": 3.3883889780710152,
      "grad_norm": 0.8434239625930786,
      "learning_rate": 7.1686837678446364e-06,
      "loss": 0.0513,
      "step": 230075
    },
    {
      "epoch": 3.3887571611611023,
      "grad_norm": 1.328847050666809,
      "learning_rate": 7.167047395877652e-06,
      "loss": 0.0535,
      "step": 230100
    },
    {
      "epoch": 3.3891253442511893,
      "grad_norm": 1.5471599102020264,
      "learning_rate": 7.165411023910668e-06,
      "loss": 0.0571,
      "step": 230125
    },
    {
      "epoch": 3.3894935273412763,
      "grad_norm": 1.2296767234802246,
      "learning_rate": 7.163774651943683e-06,
      "loss": 0.0613,
      "step": 230150
    },
    {
      "epoch": 3.3898617104313633,
      "grad_norm": 1.2912365198135376,
      "learning_rate": 7.162138279976699e-06,
      "loss": 0.063,
      "step": 230175
    },
    {
      "epoch": 3.3902298935214503,
      "grad_norm": 1.0140345096588135,
      "learning_rate": 7.160501908009714e-06,
      "loss": 0.0514,
      "step": 230200
    },
    {
      "epoch": 3.3905980766115373,
      "grad_norm": 1.5158782005310059,
      "learning_rate": 7.158865536042729e-06,
      "loss": 0.0495,
      "step": 230225
    },
    {
      "epoch": 3.3909662597016244,
      "grad_norm": 1.2484372854232788,
      "learning_rate": 7.157229164075745e-06,
      "loss": 0.0536,
      "step": 230250
    },
    {
      "epoch": 3.3913344427917114,
      "grad_norm": 1.3410732746124268,
      "learning_rate": 7.1555927921087606e-06,
      "loss": 0.0509,
      "step": 230275
    },
    {
      "epoch": 3.3917026258817984,
      "grad_norm": 1.0487955808639526,
      "learning_rate": 7.153956420141776e-06,
      "loss": 0.0489,
      "step": 230300
    },
    {
      "epoch": 3.3920708089718854,
      "grad_norm": 1.2039780616760254,
      "learning_rate": 7.152320048174791e-06,
      "loss": 0.0549,
      "step": 230325
    },
    {
      "epoch": 3.3924389920619724,
      "grad_norm": 1.4711581468582153,
      "learning_rate": 7.150683676207807e-06,
      "loss": 0.0601,
      "step": 230350
    },
    {
      "epoch": 3.3928071751520594,
      "grad_norm": 1.4990562200546265,
      "learning_rate": 7.149047304240823e-06,
      "loss": 0.058,
      "step": 230375
    },
    {
      "epoch": 3.3931753582421464,
      "grad_norm": 1.0785878896713257,
      "learning_rate": 7.147410932273837e-06,
      "loss": 0.0601,
      "step": 230400
    },
    {
      "epoch": 3.3935435413322335,
      "grad_norm": 1.5416322946548462,
      "learning_rate": 7.145774560306853e-06,
      "loss": 0.054,
      "step": 230425
    },
    {
      "epoch": 3.393911724422321,
      "grad_norm": 1.7337473630905151,
      "learning_rate": 7.1441381883398685e-06,
      "loss": 0.0535,
      "step": 230450
    },
    {
      "epoch": 3.394279907512408,
      "grad_norm": 1.2901270389556885,
      "learning_rate": 7.142501816372885e-06,
      "loss": 0.0589,
      "step": 230475
    },
    {
      "epoch": 3.394648090602495,
      "grad_norm": 1.308258056640625,
      "learning_rate": 7.140865444405899e-06,
      "loss": 0.0627,
      "step": 230500
    },
    {
      "epoch": 3.395016273692582,
      "grad_norm": 0.9888496398925781,
      "learning_rate": 7.1392290724389145e-06,
      "loss": 0.0524,
      "step": 230525
    },
    {
      "epoch": 3.395384456782669,
      "grad_norm": 1.0918967723846436,
      "learning_rate": 7.137592700471931e-06,
      "loss": 0.0499,
      "step": 230550
    },
    {
      "epoch": 3.395752639872756,
      "grad_norm": 1.1751959323883057,
      "learning_rate": 7.135956328504946e-06,
      "loss": 0.0532,
      "step": 230575
    },
    {
      "epoch": 3.396120822962843,
      "grad_norm": 1.5641647577285767,
      "learning_rate": 7.13431995653796e-06,
      "loss": 0.0499,
      "step": 230600
    },
    {
      "epoch": 3.39648900605293,
      "grad_norm": 1.6345034837722778,
      "learning_rate": 7.1326835845709765e-06,
      "loss": 0.0551,
      "step": 230625
    },
    {
      "epoch": 3.396857189143017,
      "grad_norm": 1.2135564088821411,
      "learning_rate": 7.131047212603992e-06,
      "loss": 0.0487,
      "step": 230650
    },
    {
      "epoch": 3.397225372233104,
      "grad_norm": 1.112851619720459,
      "learning_rate": 7.129410840637008e-06,
      "loss": 0.0578,
      "step": 230675
    },
    {
      "epoch": 3.397593555323191,
      "grad_norm": 1.4080071449279785,
      "learning_rate": 7.127774468670023e-06,
      "loss": 0.0576,
      "step": 230700
    },
    {
      "epoch": 3.397961738413278,
      "grad_norm": 1.3997987508773804,
      "learning_rate": 7.126203551581717e-06,
      "loss": 0.0562,
      "step": 230725
    },
    {
      "epoch": 3.398329921503365,
      "grad_norm": 1.2018193006515503,
      "learning_rate": 7.124567179614733e-06,
      "loss": 0.0567,
      "step": 230750
    },
    {
      "epoch": 3.398698104593452,
      "grad_norm": 1.1015087366104126,
      "learning_rate": 7.122930807647749e-06,
      "loss": 0.0507,
      "step": 230775
    },
    {
      "epoch": 3.399066287683539,
      "grad_norm": 1.5534456968307495,
      "learning_rate": 7.121294435680765e-06,
      "loss": 0.0549,
      "step": 230800
    },
    {
      "epoch": 3.399434470773626,
      "grad_norm": 1.714211106300354,
      "learning_rate": 7.119658063713779e-06,
      "loss": 0.0549,
      "step": 230825
    },
    {
      "epoch": 3.399802653863713,
      "grad_norm": 0.5852401256561279,
      "learning_rate": 7.1180216917467945e-06,
      "loss": 0.0458,
      "step": 230850
    },
    {
      "epoch": 3.4001708369538006,
      "grad_norm": 1.0412797927856445,
      "learning_rate": 7.116385319779811e-06,
      "loss": 0.0514,
      "step": 230875
    },
    {
      "epoch": 3.4005390200438876,
      "grad_norm": 1.4167077541351318,
      "learning_rate": 7.114748947812826e-06,
      "loss": 0.0541,
      "step": 230900
    },
    {
      "epoch": 3.4009072031339747,
      "grad_norm": 1.3110685348510742,
      "learning_rate": 7.1131125758458404e-06,
      "loss": 0.0557,
      "step": 230925
    },
    {
      "epoch": 3.4012753862240617,
      "grad_norm": 0.5840967893600464,
      "learning_rate": 7.111476203878857e-06,
      "loss": 0.054,
      "step": 230950
    },
    {
      "epoch": 3.4016435693141487,
      "grad_norm": 1.4448232650756836,
      "learning_rate": 7.109839831911872e-06,
      "loss": 0.0494,
      "step": 230975
    },
    {
      "epoch": 3.4020117524042357,
      "grad_norm": 1.3682793378829956,
      "learning_rate": 7.108203459944888e-06,
      "loss": 0.0531,
      "step": 231000
    },
    {
      "epoch": 3.4023799354943227,
      "grad_norm": 1.5229613780975342,
      "learning_rate": 7.1065670879779025e-06,
      "loss": 0.059,
      "step": 231025
    },
    {
      "epoch": 3.4027481185844097,
      "grad_norm": 1.8227739334106445,
      "learning_rate": 7.104930716010918e-06,
      "loss": 0.0553,
      "step": 231050
    },
    {
      "epoch": 3.4031163016744967,
      "grad_norm": 1.4641097784042358,
      "learning_rate": 7.103294344043934e-06,
      "loss": 0.0518,
      "step": 231075
    },
    {
      "epoch": 3.4034844847645838,
      "grad_norm": 1.3100905418395996,
      "learning_rate": 7.101657972076949e-06,
      "loss": 0.0601,
      "step": 231100
    },
    {
      "epoch": 3.4038526678546708,
      "grad_norm": 1.0178974866867065,
      "learning_rate": 7.100021600109965e-06,
      "loss": 0.0556,
      "step": 231125
    },
    {
      "epoch": 3.404220850944758,
      "grad_norm": 1.049385666847229,
      "learning_rate": 7.09838522814298e-06,
      "loss": 0.0534,
      "step": 231150
    },
    {
      "epoch": 3.404589034034845,
      "grad_norm": 1.3061366081237793,
      "learning_rate": 7.096748856175996e-06,
      "loss": 0.055,
      "step": 231175
    },
    {
      "epoch": 3.404957217124932,
      "grad_norm": 1.6441543102264404,
      "learning_rate": 7.095112484209011e-06,
      "loss": 0.0492,
      "step": 231200
    },
    {
      "epoch": 3.405325400215019,
      "grad_norm": 1.2504527568817139,
      "learning_rate": 7.093476112242027e-06,
      "loss": 0.0524,
      "step": 231225
    },
    {
      "epoch": 3.405693583305106,
      "grad_norm": 1.9197078943252563,
      "learning_rate": 7.091839740275042e-06,
      "loss": 0.055,
      "step": 231250
    },
    {
      "epoch": 3.406061766395193,
      "grad_norm": 1.5676006078720093,
      "learning_rate": 7.090203368308057e-06,
      "loss": 0.0501,
      "step": 231275
    },
    {
      "epoch": 3.40642994948528,
      "grad_norm": 1.5482673645019531,
      "learning_rate": 7.088566996341073e-06,
      "loss": 0.0593,
      "step": 231300
    },
    {
      "epoch": 3.406798132575367,
      "grad_norm": 1.7814412117004395,
      "learning_rate": 7.086930624374089e-06,
      "loss": 0.053,
      "step": 231325
    },
    {
      "epoch": 3.407166315665454,
      "grad_norm": 1.2105423212051392,
      "learning_rate": 7.085294252407103e-06,
      "loss": 0.0518,
      "step": 231350
    },
    {
      "epoch": 3.407534498755541,
      "grad_norm": 1.210797667503357,
      "learning_rate": 7.083657880440119e-06,
      "loss": 0.0525,
      "step": 231375
    },
    {
      "epoch": 3.407902681845628,
      "grad_norm": 1.19452702999115,
      "learning_rate": 7.082021508473135e-06,
      "loss": 0.0474,
      "step": 231400
    },
    {
      "epoch": 3.4082708649357154,
      "grad_norm": 1.1709809303283691,
      "learning_rate": 7.080385136506151e-06,
      "loss": 0.0544,
      "step": 231425
    },
    {
      "epoch": 3.4086390480258024,
      "grad_norm": 0.8450188040733337,
      "learning_rate": 7.078748764539165e-06,
      "loss": 0.0562,
      "step": 231450
    },
    {
      "epoch": 3.4090072311158894,
      "grad_norm": 1.330223560333252,
      "learning_rate": 7.0771123925721805e-06,
      "loss": 0.0541,
      "step": 231475
    },
    {
      "epoch": 3.4093754142059765,
      "grad_norm": 1.1931023597717285,
      "learning_rate": 7.075476020605197e-06,
      "loss": 0.0479,
      "step": 231500
    },
    {
      "epoch": 3.4097435972960635,
      "grad_norm": 1.5765554904937744,
      "learning_rate": 7.073839648638212e-06,
      "loss": 0.0518,
      "step": 231525
    },
    {
      "epoch": 3.4101117803861505,
      "grad_norm": 1.6435822248458862,
      "learning_rate": 7.0722032766712264e-06,
      "loss": 0.0589,
      "step": 231550
    },
    {
      "epoch": 3.4104799634762375,
      "grad_norm": 0.9774888753890991,
      "learning_rate": 7.070566904704243e-06,
      "loss": 0.0564,
      "step": 231575
    },
    {
      "epoch": 3.4108481465663245,
      "grad_norm": 1.090767502784729,
      "learning_rate": 7.068930532737258e-06,
      "loss": 0.0587,
      "step": 231600
    },
    {
      "epoch": 3.4112163296564115,
      "grad_norm": 1.4417084455490112,
      "learning_rate": 7.067294160770274e-06,
      "loss": 0.059,
      "step": 231625
    },
    {
      "epoch": 3.4115845127464985,
      "grad_norm": 1.5252876281738281,
      "learning_rate": 7.065657788803289e-06,
      "loss": 0.0558,
      "step": 231650
    },
    {
      "epoch": 3.4119526958365856,
      "grad_norm": 1.1169487237930298,
      "learning_rate": 7.064021416836304e-06,
      "loss": 0.0562,
      "step": 231675
    },
    {
      "epoch": 3.4123208789266726,
      "grad_norm": 1.637233853340149,
      "learning_rate": 7.06238504486932e-06,
      "loss": 0.0585,
      "step": 231700
    },
    {
      "epoch": 3.4126890620167596,
      "grad_norm": 1.1862943172454834,
      "learning_rate": 7.060748672902335e-06,
      "loss": 0.0576,
      "step": 231725
    },
    {
      "epoch": 3.4130572451068466,
      "grad_norm": 1.5288313627243042,
      "learning_rate": 7.059112300935351e-06,
      "loss": 0.057,
      "step": 231750
    },
    {
      "epoch": 3.4134254281969336,
      "grad_norm": 1.928776502609253,
      "learning_rate": 7.057475928968366e-06,
      "loss": 0.059,
      "step": 231775
    },
    {
      "epoch": 3.4137936112870206,
      "grad_norm": 1.1547839641571045,
      "learning_rate": 7.055839557001381e-06,
      "loss": 0.0514,
      "step": 231800
    },
    {
      "epoch": 3.414161794377108,
      "grad_norm": 1.022402048110962,
      "learning_rate": 7.054203185034397e-06,
      "loss": 0.0555,
      "step": 231825
    },
    {
      "epoch": 3.414529977467195,
      "grad_norm": 0.7116663455963135,
      "learning_rate": 7.052566813067413e-06,
      "loss": 0.0595,
      "step": 231850
    },
    {
      "epoch": 3.414898160557282,
      "grad_norm": 1.7710940837860107,
      "learning_rate": 7.050930441100428e-06,
      "loss": 0.0579,
      "step": 231875
    },
    {
      "epoch": 3.415266343647369,
      "grad_norm": 1.4213794469833374,
      "learning_rate": 7.049294069133443e-06,
      "loss": 0.0519,
      "step": 231900
    },
    {
      "epoch": 3.415634526737456,
      "grad_norm": 1.1554690599441528,
      "learning_rate": 7.047657697166459e-06,
      "loss": 0.0537,
      "step": 231925
    },
    {
      "epoch": 3.416002709827543,
      "grad_norm": 1.626205325126648,
      "learning_rate": 7.046021325199475e-06,
      "loss": 0.0541,
      "step": 231950
    },
    {
      "epoch": 3.41637089291763,
      "grad_norm": 1.1741199493408203,
      "learning_rate": 7.044384953232489e-06,
      "loss": 0.0551,
      "step": 231975
    },
    {
      "epoch": 3.416739076007717,
      "grad_norm": 1.0156229734420776,
      "learning_rate": 7.042748581265505e-06,
      "loss": 0.0501,
      "step": 232000
    },
    {
      "epoch": 3.417107259097804,
      "grad_norm": 1.2444950342178345,
      "learning_rate": 7.041112209298521e-06,
      "loss": 0.0503,
      "step": 232025
    },
    {
      "epoch": 3.4174754421878912,
      "grad_norm": 1.3393328189849854,
      "learning_rate": 7.039475837331537e-06,
      "loss": 0.0603,
      "step": 232050
    },
    {
      "epoch": 3.4178436252779782,
      "grad_norm": 0.9660595655441284,
      "learning_rate": 7.037839465364551e-06,
      "loss": 0.058,
      "step": 232075
    },
    {
      "epoch": 3.4182118083680653,
      "grad_norm": 1.1569231748580933,
      "learning_rate": 7.0362030933975665e-06,
      "loss": 0.0473,
      "step": 232100
    },
    {
      "epoch": 3.4185799914581523,
      "grad_norm": 1.8087416887283325,
      "learning_rate": 7.034566721430583e-06,
      "loss": 0.0591,
      "step": 232125
    },
    {
      "epoch": 3.4189481745482393,
      "grad_norm": 1.3858063220977783,
      "learning_rate": 7.032930349463598e-06,
      "loss": 0.0535,
      "step": 232150
    },
    {
      "epoch": 3.4193163576383263,
      "grad_norm": 1.208710789680481,
      "learning_rate": 7.031293977496614e-06,
      "loss": 0.0552,
      "step": 232175
    },
    {
      "epoch": 3.4196845407284133,
      "grad_norm": 1.5608861446380615,
      "learning_rate": 7.029657605529629e-06,
      "loss": 0.0545,
      "step": 232200
    },
    {
      "epoch": 3.4200527238185003,
      "grad_norm": 1.5005319118499756,
      "learning_rate": 7.028021233562644e-06,
      "loss": 0.057,
      "step": 232225
    },
    {
      "epoch": 3.4204209069085874,
      "grad_norm": 1.1500242948532104,
      "learning_rate": 7.02638486159566e-06,
      "loss": 0.0565,
      "step": 232250
    },
    {
      "epoch": 3.4207890899986744,
      "grad_norm": 1.6404985189437866,
      "learning_rate": 7.024748489628675e-06,
      "loss": 0.0511,
      "step": 232275
    },
    {
      "epoch": 3.4211572730887614,
      "grad_norm": 1.4335654973983765,
      "learning_rate": 7.02311211766169e-06,
      "loss": 0.0556,
      "step": 232300
    },
    {
      "epoch": 3.4215254561788484,
      "grad_norm": 1.354269027709961,
      "learning_rate": 7.021475745694706e-06,
      "loss": 0.0563,
      "step": 232325
    },
    {
      "epoch": 3.4218936392689354,
      "grad_norm": 1.8647336959838867,
      "learning_rate": 7.019839373727721e-06,
      "loss": 0.0564,
      "step": 232350
    },
    {
      "epoch": 3.422261822359023,
      "grad_norm": 1.2038220167160034,
      "learning_rate": 7.018203001760737e-06,
      "loss": 0.0572,
      "step": 232375
    },
    {
      "epoch": 3.42263000544911,
      "grad_norm": 1.2991139888763428,
      "learning_rate": 7.016566629793752e-06,
      "loss": 0.0508,
      "step": 232400
    },
    {
      "epoch": 3.422998188539197,
      "grad_norm": 1.2577146291732788,
      "learning_rate": 7.014930257826767e-06,
      "loss": 0.0518,
      "step": 232425
    },
    {
      "epoch": 3.423366371629284,
      "grad_norm": 1.1498643159866333,
      "learning_rate": 7.013293885859783e-06,
      "loss": 0.0535,
      "step": 232450
    },
    {
      "epoch": 3.423734554719371,
      "grad_norm": 0.9440375566482544,
      "learning_rate": 7.011657513892799e-06,
      "loss": 0.0528,
      "step": 232475
    },
    {
      "epoch": 3.424102737809458,
      "grad_norm": 1.3459793329238892,
      "learning_rate": 7.010021141925813e-06,
      "loss": 0.0563,
      "step": 232500
    },
    {
      "epoch": 3.424470920899545,
      "grad_norm": 1.218077301979065,
      "learning_rate": 7.008384769958829e-06,
      "loss": 0.0562,
      "step": 232525
    },
    {
      "epoch": 3.424839103989632,
      "grad_norm": 1.4415664672851562,
      "learning_rate": 7.0067483979918445e-06,
      "loss": 0.0511,
      "step": 232550
    },
    {
      "epoch": 3.425207287079719,
      "grad_norm": 1.043233871459961,
      "learning_rate": 7.005112026024861e-06,
      "loss": 0.0574,
      "step": 232575
    },
    {
      "epoch": 3.425575470169806,
      "grad_norm": 1.5398774147033691,
      "learning_rate": 7.003475654057876e-06,
      "loss": 0.0539,
      "step": 232600
    },
    {
      "epoch": 3.425943653259893,
      "grad_norm": 1.1564363241195679,
      "learning_rate": 7.001839282090891e-06,
      "loss": 0.0582,
      "step": 232625
    },
    {
      "epoch": 3.42631183634998,
      "grad_norm": 1.0484740734100342,
      "learning_rate": 7.000202910123907e-06,
      "loss": 0.0545,
      "step": 232650
    },
    {
      "epoch": 3.426680019440067,
      "grad_norm": 0.7289891839027405,
      "learning_rate": 6.998566538156922e-06,
      "loss": 0.0551,
      "step": 232675
    },
    {
      "epoch": 3.427048202530154,
      "grad_norm": 1.0609796047210693,
      "learning_rate": 6.996930166189938e-06,
      "loss": 0.0612,
      "step": 232700
    },
    {
      "epoch": 3.427416385620241,
      "grad_norm": 1.637547254562378,
      "learning_rate": 6.9952937942229525e-06,
      "loss": 0.0547,
      "step": 232725
    },
    {
      "epoch": 3.427784568710328,
      "grad_norm": 1.0543664693832397,
      "learning_rate": 6.993722877134648e-06,
      "loss": 0.0573,
      "step": 232750
    },
    {
      "epoch": 3.428152751800415,
      "grad_norm": 0.9847107529640198,
      "learning_rate": 6.992086505167663e-06,
      "loss": 0.0516,
      "step": 232775
    },
    {
      "epoch": 3.4285209348905026,
      "grad_norm": 1.437808632850647,
      "learning_rate": 6.990450133200679e-06,
      "loss": 0.0529,
      "step": 232800
    },
    {
      "epoch": 3.4288891179805896,
      "grad_norm": 0.8354343175888062,
      "learning_rate": 6.988813761233694e-06,
      "loss": 0.0568,
      "step": 232825
    },
    {
      "epoch": 3.4292573010706766,
      "grad_norm": 1.4230772256851196,
      "learning_rate": 6.987177389266709e-06,
      "loss": 0.0568,
      "step": 232850
    },
    {
      "epoch": 3.4296254841607636,
      "grad_norm": 1.260440707206726,
      "learning_rate": 6.9855410172997255e-06,
      "loss": 0.0534,
      "step": 232875
    },
    {
      "epoch": 3.4299936672508506,
      "grad_norm": 1.4289093017578125,
      "learning_rate": 6.983904645332741e-06,
      "loss": 0.0605,
      "step": 232900
    },
    {
      "epoch": 3.4303618503409377,
      "grad_norm": 1.2366105318069458,
      "learning_rate": 6.982268273365755e-06,
      "loss": 0.0523,
      "step": 232925
    },
    {
      "epoch": 3.4307300334310247,
      "grad_norm": 1.1550147533416748,
      "learning_rate": 6.980631901398771e-06,
      "loss": 0.0574,
      "step": 232950
    },
    {
      "epoch": 3.4310982165211117,
      "grad_norm": 1.513540506362915,
      "learning_rate": 6.978995529431787e-06,
      "loss": 0.0592,
      "step": 232975
    },
    {
      "epoch": 3.4314663996111987,
      "grad_norm": 1.570636510848999,
      "learning_rate": 6.977359157464803e-06,
      "loss": 0.0549,
      "step": 233000
    },
    {
      "epoch": 3.4318345827012857,
      "grad_norm": 1.3403128385543823,
      "learning_rate": 6.975722785497817e-06,
      "loss": 0.0487,
      "step": 233025
    },
    {
      "epoch": 3.4322027657913727,
      "grad_norm": 1.6959725618362427,
      "learning_rate": 6.974086413530833e-06,
      "loss": 0.0494,
      "step": 233050
    },
    {
      "epoch": 3.4325709488814597,
      "grad_norm": 1.3732138872146606,
      "learning_rate": 6.972450041563849e-06,
      "loss": 0.0578,
      "step": 233075
    },
    {
      "epoch": 3.4329391319715468,
      "grad_norm": 1.0029683113098145,
      "learning_rate": 6.970813669596864e-06,
      "loss": 0.0648,
      "step": 233100
    },
    {
      "epoch": 3.433307315061634,
      "grad_norm": 1.0218791961669922,
      "learning_rate": 6.96917729762988e-06,
      "loss": 0.0514,
      "step": 233125
    },
    {
      "epoch": 3.433675498151721,
      "grad_norm": 1.437760829925537,
      "learning_rate": 6.967540925662895e-06,
      "loss": 0.057,
      "step": 233150
    },
    {
      "epoch": 3.434043681241808,
      "grad_norm": 1.7047245502471924,
      "learning_rate": 6.96590455369591e-06,
      "loss": 0.0621,
      "step": 233175
    },
    {
      "epoch": 3.434411864331895,
      "grad_norm": 1.3571045398712158,
      "learning_rate": 6.964268181728926e-06,
      "loss": 0.0571,
      "step": 233200
    },
    {
      "epoch": 3.434780047421982,
      "grad_norm": 1.0178911685943604,
      "learning_rate": 6.962631809761941e-06,
      "loss": 0.0593,
      "step": 233225
    },
    {
      "epoch": 3.435148230512069,
      "grad_norm": 1.3114897012710571,
      "learning_rate": 6.960995437794956e-06,
      "loss": 0.0526,
      "step": 233250
    },
    {
      "epoch": 3.435516413602156,
      "grad_norm": 1.376524567604065,
      "learning_rate": 6.959359065827972e-06,
      "loss": 0.0567,
      "step": 233275
    },
    {
      "epoch": 3.435884596692243,
      "grad_norm": 0.8352702260017395,
      "learning_rate": 6.957722693860987e-06,
      "loss": 0.0568,
      "step": 233300
    },
    {
      "epoch": 3.4362527797823303,
      "grad_norm": 1.4313902854919434,
      "learning_rate": 6.9560863218940035e-06,
      "loss": 0.0513,
      "step": 233325
    },
    {
      "epoch": 3.4366209628724174,
      "grad_norm": 1.2757450342178345,
      "learning_rate": 6.954449949927018e-06,
      "loss": 0.0535,
      "step": 233350
    },
    {
      "epoch": 3.4369891459625044,
      "grad_norm": 1.646310567855835,
      "learning_rate": 6.952813577960033e-06,
      "loss": 0.0617,
      "step": 233375
    },
    {
      "epoch": 3.4373573290525914,
      "grad_norm": 1.7936944961547852,
      "learning_rate": 6.951177205993049e-06,
      "loss": 0.0546,
      "step": 233400
    },
    {
      "epoch": 3.4377255121426784,
      "grad_norm": 0.933220624923706,
      "learning_rate": 6.949540834026065e-06,
      "loss": 0.0603,
      "step": 233425
    },
    {
      "epoch": 3.4380936952327654,
      "grad_norm": 1.5044418573379517,
      "learning_rate": 6.94790446205908e-06,
      "loss": 0.0494,
      "step": 233450
    },
    {
      "epoch": 3.4384618783228524,
      "grad_norm": 1.3826124668121338,
      "learning_rate": 6.946268090092095e-06,
      "loss": 0.0528,
      "step": 233475
    },
    {
      "epoch": 3.4388300614129395,
      "grad_norm": 1.0711140632629395,
      "learning_rate": 6.9446317181251115e-06,
      "loss": 0.0508,
      "step": 233500
    },
    {
      "epoch": 3.4391982445030265,
      "grad_norm": 1.4220081567764282,
      "learning_rate": 6.942995346158127e-06,
      "loss": 0.0602,
      "step": 233525
    },
    {
      "epoch": 3.4395664275931135,
      "grad_norm": 1.2843989133834839,
      "learning_rate": 6.941358974191141e-06,
      "loss": 0.0485,
      "step": 233550
    },
    {
      "epoch": 3.4399346106832005,
      "grad_norm": 1.3126015663146973,
      "learning_rate": 6.939722602224157e-06,
      "loss": 0.0592,
      "step": 233575
    },
    {
      "epoch": 3.4403027937732875,
      "grad_norm": 1.6128236055374146,
      "learning_rate": 6.938086230257173e-06,
      "loss": 0.0537,
      "step": 233600
    },
    {
      "epoch": 3.4406709768633745,
      "grad_norm": 0.8044310808181763,
      "learning_rate": 6.936449858290189e-06,
      "loss": 0.049,
      "step": 233625
    },
    {
      "epoch": 3.4410391599534615,
      "grad_norm": 1.4766666889190674,
      "learning_rate": 6.9348789412018836e-06,
      "loss": 0.0547,
      "step": 233650
    },
    {
      "epoch": 3.4414073430435486,
      "grad_norm": 2.1881563663482666,
      "learning_rate": 6.933242569234898e-06,
      "loss": 0.0575,
      "step": 233675
    },
    {
      "epoch": 3.4417755261336356,
      "grad_norm": 1.0944713354110718,
      "learning_rate": 6.931606197267914e-06,
      "loss": 0.0571,
      "step": 233700
    },
    {
      "epoch": 3.4421437092237226,
      "grad_norm": 1.5938423871994019,
      "learning_rate": 6.9299698253009295e-06,
      "loss": 0.0666,
      "step": 233725
    },
    {
      "epoch": 3.44251189231381,
      "grad_norm": 0.8704099655151367,
      "learning_rate": 6.928333453333946e-06,
      "loss": 0.0514,
      "step": 233750
    },
    {
      "epoch": 3.442880075403897,
      "grad_norm": 1.607376217842102,
      "learning_rate": 6.92669708136696e-06,
      "loss": 0.0601,
      "step": 233775
    },
    {
      "epoch": 3.443248258493984,
      "grad_norm": 1.2371896505355835,
      "learning_rate": 6.925060709399975e-06,
      "loss": 0.0601,
      "step": 233800
    },
    {
      "epoch": 3.443616441584071,
      "grad_norm": 1.1890604496002197,
      "learning_rate": 6.9234243374329915e-06,
      "loss": 0.0523,
      "step": 233825
    },
    {
      "epoch": 3.443984624674158,
      "grad_norm": 0.8377152681350708,
      "learning_rate": 6.921787965466007e-06,
      "loss": 0.055,
      "step": 233850
    },
    {
      "epoch": 3.444352807764245,
      "grad_norm": 1.3999760150909424,
      "learning_rate": 6.920151593499021e-06,
      "loss": 0.0543,
      "step": 233875
    },
    {
      "epoch": 3.444720990854332,
      "grad_norm": 1.1638010740280151,
      "learning_rate": 6.9185152215320375e-06,
      "loss": 0.0593,
      "step": 233900
    },
    {
      "epoch": 3.445089173944419,
      "grad_norm": 1.4614380598068237,
      "learning_rate": 6.916878849565053e-06,
      "loss": 0.0526,
      "step": 233925
    },
    {
      "epoch": 3.445457357034506,
      "grad_norm": 1.6576213836669922,
      "learning_rate": 6.915242477598069e-06,
      "loss": 0.0557,
      "step": 233950
    },
    {
      "epoch": 3.445825540124593,
      "grad_norm": 1.156779170036316,
      "learning_rate": 6.913606105631083e-06,
      "loss": 0.0528,
      "step": 233975
    },
    {
      "epoch": 3.44619372321468,
      "grad_norm": 1.4911894798278809,
      "learning_rate": 6.911969733664099e-06,
      "loss": 0.0565,
      "step": 234000
    },
    {
      "epoch": 3.446561906304767,
      "grad_norm": 1.1513633728027344,
      "learning_rate": 6.910333361697115e-06,
      "loss": 0.0526,
      "step": 234025
    },
    {
      "epoch": 3.4469300893948542,
      "grad_norm": 1.9557138681411743,
      "learning_rate": 6.90869698973013e-06,
      "loss": 0.0601,
      "step": 234050
    },
    {
      "epoch": 3.4472982724849413,
      "grad_norm": 1.3454968929290771,
      "learning_rate": 6.907060617763146e-06,
      "loss": 0.0518,
      "step": 234075
    },
    {
      "epoch": 3.4476664555750283,
      "grad_norm": 1.2555756568908691,
      "learning_rate": 6.905424245796161e-06,
      "loss": 0.0492,
      "step": 234100
    },
    {
      "epoch": 3.4480346386651153,
      "grad_norm": 1.12812340259552,
      "learning_rate": 6.903787873829176e-06,
      "loss": 0.055,
      "step": 234125
    },
    {
      "epoch": 3.4484028217552023,
      "grad_norm": 1.457519292831421,
      "learning_rate": 6.902151501862192e-06,
      "loss": 0.0505,
      "step": 234150
    },
    {
      "epoch": 3.4487710048452893,
      "grad_norm": 1.1737881898880005,
      "learning_rate": 6.9005151298952075e-06,
      "loss": 0.0526,
      "step": 234175
    },
    {
      "epoch": 3.4491391879353763,
      "grad_norm": 1.009140133857727,
      "learning_rate": 6.898878757928222e-06,
      "loss": 0.059,
      "step": 234200
    },
    {
      "epoch": 3.4495073710254633,
      "grad_norm": 1.4208605289459229,
      "learning_rate": 6.897242385961238e-06,
      "loss": 0.0505,
      "step": 234225
    },
    {
      "epoch": 3.4498755541155504,
      "grad_norm": 1.17922043800354,
      "learning_rate": 6.895606013994253e-06,
      "loss": 0.0506,
      "step": 234250
    },
    {
      "epoch": 3.4502437372056374,
      "grad_norm": 1.5503113269805908,
      "learning_rate": 6.8939696420272696e-06,
      "loss": 0.053,
      "step": 234275
    },
    {
      "epoch": 3.450611920295725,
      "grad_norm": 1.4906063079833984,
      "learning_rate": 6.892333270060284e-06,
      "loss": 0.0575,
      "step": 234300
    },
    {
      "epoch": 3.450980103385812,
      "grad_norm": 2.207207441329956,
      "learning_rate": 6.8906968980933e-06,
      "loss": 0.0526,
      "step": 234325
    },
    {
      "epoch": 3.451348286475899,
      "grad_norm": 1.1196998357772827,
      "learning_rate": 6.8890605261263155e-06,
      "loss": 0.0545,
      "step": 234350
    },
    {
      "epoch": 3.451716469565986,
      "grad_norm": 1.8272604942321777,
      "learning_rate": 6.887424154159331e-06,
      "loss": 0.0558,
      "step": 234375
    },
    {
      "epoch": 3.452084652656073,
      "grad_norm": 0.8513809442520142,
      "learning_rate": 6.885787782192346e-06,
      "loss": 0.0489,
      "step": 234400
    },
    {
      "epoch": 3.45245283574616,
      "grad_norm": 1.5188515186309814,
      "learning_rate": 6.884151410225361e-06,
      "loss": 0.0515,
      "step": 234425
    },
    {
      "epoch": 3.452821018836247,
      "grad_norm": 1.2470561265945435,
      "learning_rate": 6.8825150382583775e-06,
      "loss": 0.052,
      "step": 234450
    },
    {
      "epoch": 3.453189201926334,
      "grad_norm": 1.6135742664337158,
      "learning_rate": 6.880878666291393e-06,
      "loss": 0.0545,
      "step": 234475
    },
    {
      "epoch": 3.453557385016421,
      "grad_norm": 1.383779764175415,
      "learning_rate": 6.879242294324407e-06,
      "loss": 0.0564,
      "step": 234500
    },
    {
      "epoch": 3.453925568106508,
      "grad_norm": 1.1124207973480225,
      "learning_rate": 6.8776059223574235e-06,
      "loss": 0.05,
      "step": 234525
    },
    {
      "epoch": 3.454293751196595,
      "grad_norm": 1.3962866067886353,
      "learning_rate": 6.875969550390439e-06,
      "loss": 0.0515,
      "step": 234550
    },
    {
      "epoch": 3.454661934286682,
      "grad_norm": 1.374168038368225,
      "learning_rate": 6.874333178423455e-06,
      "loss": 0.0505,
      "step": 234575
    },
    {
      "epoch": 3.455030117376769,
      "grad_norm": 1.6096205711364746,
      "learning_rate": 6.87269680645647e-06,
      "loss": 0.0594,
      "step": 234600
    },
    {
      "epoch": 3.455398300466856,
      "grad_norm": 1.3983277082443237,
      "learning_rate": 6.871060434489485e-06,
      "loss": 0.0562,
      "step": 234625
    },
    {
      "epoch": 3.455766483556943,
      "grad_norm": 1.3306199312210083,
      "learning_rate": 6.869424062522501e-06,
      "loss": 0.0512,
      "step": 234650
    },
    {
      "epoch": 3.45613466664703,
      "grad_norm": 1.9823153018951416,
      "learning_rate": 6.867787690555516e-06,
      "loss": 0.0684,
      "step": 234675
    },
    {
      "epoch": 3.4565028497371175,
      "grad_norm": 1.5583148002624512,
      "learning_rate": 6.866151318588532e-06,
      "loss": 0.0573,
      "step": 234700
    },
    {
      "epoch": 3.4568710328272045,
      "grad_norm": 1.6550252437591553,
      "learning_rate": 6.864514946621547e-06,
      "loss": 0.0523,
      "step": 234725
    },
    {
      "epoch": 3.4572392159172916,
      "grad_norm": 1.2345621585845947,
      "learning_rate": 6.862878574654562e-06,
      "loss": 0.0519,
      "step": 234750
    },
    {
      "epoch": 3.4576073990073786,
      "grad_norm": 1.5852617025375366,
      "learning_rate": 6.861242202687578e-06,
      "loss": 0.0532,
      "step": 234775
    },
    {
      "epoch": 3.4579755820974656,
      "grad_norm": 1.7967497110366821,
      "learning_rate": 6.8596058307205935e-06,
      "loss": 0.0597,
      "step": 234800
    },
    {
      "epoch": 3.4583437651875526,
      "grad_norm": 1.4641871452331543,
      "learning_rate": 6.857969458753608e-06,
      "loss": 0.0578,
      "step": 234825
    },
    {
      "epoch": 3.4587119482776396,
      "grad_norm": 1.144559383392334,
      "learning_rate": 6.856333086786624e-06,
      "loss": 0.0468,
      "step": 234850
    },
    {
      "epoch": 3.4590801313677266,
      "grad_norm": 1.3692418336868286,
      "learning_rate": 6.854696714819639e-06,
      "loss": 0.0619,
      "step": 234875
    },
    {
      "epoch": 3.4594483144578136,
      "grad_norm": 1.4208827018737793,
      "learning_rate": 6.8530603428526556e-06,
      "loss": 0.0591,
      "step": 234900
    },
    {
      "epoch": 3.4598164975479007,
      "grad_norm": 1.0685988664627075,
      "learning_rate": 6.85142397088567e-06,
      "loss": 0.0564,
      "step": 234925
    },
    {
      "epoch": 3.4601846806379877,
      "grad_norm": 1.3136134147644043,
      "learning_rate": 6.849787598918685e-06,
      "loss": 0.0557,
      "step": 234950
    },
    {
      "epoch": 3.4605528637280747,
      "grad_norm": 1.1276713609695435,
      "learning_rate": 6.8481512269517015e-06,
      "loss": 0.0462,
      "step": 234975
    },
    {
      "epoch": 3.4609210468181617,
      "grad_norm": 1.357824444770813,
      "learning_rate": 6.846514854984717e-06,
      "loss": 0.0562,
      "step": 235000
    },
    {
      "epoch": 3.4612892299082487,
      "grad_norm": 1.4743969440460205,
      "learning_rate": 6.844878483017733e-06,
      "loss": 0.0569,
      "step": 235025
    },
    {
      "epoch": 3.4616574129983357,
      "grad_norm": 1.5803524255752563,
      "learning_rate": 6.843242111050747e-06,
      "loss": 0.0541,
      "step": 235050
    },
    {
      "epoch": 3.4620255960884228,
      "grad_norm": 1.5557721853256226,
      "learning_rate": 6.8416057390837635e-06,
      "loss": 0.0562,
      "step": 235075
    },
    {
      "epoch": 3.4623937791785098,
      "grad_norm": 1.3310322761535645,
      "learning_rate": 6.839969367116779e-06,
      "loss": 0.0584,
      "step": 235100
    },
    {
      "epoch": 3.462761962268597,
      "grad_norm": 1.1937814950942993,
      "learning_rate": 6.838332995149794e-06,
      "loss": 0.0591,
      "step": 235125
    },
    {
      "epoch": 3.463130145358684,
      "grad_norm": 0.9994338750839233,
      "learning_rate": 6.8366966231828094e-06,
      "loss": 0.0531,
      "step": 235150
    },
    {
      "epoch": 3.463498328448771,
      "grad_norm": 1.452646017074585,
      "learning_rate": 6.835060251215825e-06,
      "loss": 0.051,
      "step": 235175
    },
    {
      "epoch": 3.463866511538858,
      "grad_norm": 1.3941121101379395,
      "learning_rate": 6.833423879248841e-06,
      "loss": 0.0566,
      "step": 235200
    },
    {
      "epoch": 3.464234694628945,
      "grad_norm": 0.8465602993965149,
      "learning_rate": 6.831787507281856e-06,
      "loss": 0.0569,
      "step": 235225
    },
    {
      "epoch": 3.4646028777190323,
      "grad_norm": 1.2280335426330566,
      "learning_rate": 6.830151135314871e-06,
      "loss": 0.0599,
      "step": 235250
    },
    {
      "epoch": 3.4649710608091193,
      "grad_norm": 1.3824517726898193,
      "learning_rate": 6.828514763347887e-06,
      "loss": 0.0534,
      "step": 235275
    },
    {
      "epoch": 3.4653392438992063,
      "grad_norm": 1.3251019716262817,
      "learning_rate": 6.826878391380902e-06,
      "loss": 0.0491,
      "step": 235300
    },
    {
      "epoch": 3.4657074269892933,
      "grad_norm": 0.8468957543373108,
      "learning_rate": 6.825242019413918e-06,
      "loss": 0.0526,
      "step": 235325
    },
    {
      "epoch": 3.4660756100793804,
      "grad_norm": 0.9797238111495972,
      "learning_rate": 6.823605647446933e-06,
      "loss": 0.0514,
      "step": 235350
    },
    {
      "epoch": 3.4664437931694674,
      "grad_norm": 1.248177170753479,
      "learning_rate": 6.821969275479948e-06,
      "loss": 0.0521,
      "step": 235375
    },
    {
      "epoch": 3.4668119762595544,
      "grad_norm": 1.6597363948822021,
      "learning_rate": 6.820332903512964e-06,
      "loss": 0.0578,
      "step": 235400
    },
    {
      "epoch": 3.4671801593496414,
      "grad_norm": 1.5750609636306763,
      "learning_rate": 6.8186965315459795e-06,
      "loss": 0.0555,
      "step": 235425
    },
    {
      "epoch": 3.4675483424397284,
      "grad_norm": 2.0586562156677246,
      "learning_rate": 6.817060159578994e-06,
      "loss": 0.0505,
      "step": 235450
    },
    {
      "epoch": 3.4679165255298154,
      "grad_norm": 1.3287327289581299,
      "learning_rate": 6.81542378761201e-06,
      "loss": 0.0532,
      "step": 235475
    },
    {
      "epoch": 3.4682847086199025,
      "grad_norm": 1.8020696640014648,
      "learning_rate": 6.813787415645025e-06,
      "loss": 0.0554,
      "step": 235500
    },
    {
      "epoch": 3.4686528917099895,
      "grad_norm": 1.4357680082321167,
      "learning_rate": 6.8121510436780416e-06,
      "loss": 0.0547,
      "step": 235525
    },
    {
      "epoch": 3.4690210748000765,
      "grad_norm": 1.7434290647506714,
      "learning_rate": 6.810514671711057e-06,
      "loss": 0.0505,
      "step": 235550
    },
    {
      "epoch": 3.4693892578901635,
      "grad_norm": 1.6416116952896118,
      "learning_rate": 6.808878299744071e-06,
      "loss": 0.0586,
      "step": 235575
    },
    {
      "epoch": 3.4697574409802505,
      "grad_norm": 1.632177472114563,
      "learning_rate": 6.8072419277770875e-06,
      "loss": 0.0547,
      "step": 235600
    },
    {
      "epoch": 3.4701256240703375,
      "grad_norm": 1.4830220937728882,
      "learning_rate": 6.805605555810103e-06,
      "loss": 0.0541,
      "step": 235625
    },
    {
      "epoch": 3.4704938071604245,
      "grad_norm": 1.9248520135879517,
      "learning_rate": 6.803969183843119e-06,
      "loss": 0.0521,
      "step": 235650
    },
    {
      "epoch": 3.470861990250512,
      "grad_norm": 1.7081308364868164,
      "learning_rate": 6.802332811876133e-06,
      "loss": 0.0511,
      "step": 235675
    },
    {
      "epoch": 3.471230173340599,
      "grad_norm": 0.9971753358840942,
      "learning_rate": 6.800696439909149e-06,
      "loss": 0.0474,
      "step": 235700
    },
    {
      "epoch": 3.471598356430686,
      "grad_norm": 1.6527999639511108,
      "learning_rate": 6.799060067942165e-06,
      "loss": 0.0574,
      "step": 235725
    },
    {
      "epoch": 3.471966539520773,
      "grad_norm": 1.3607879877090454,
      "learning_rate": 6.79742369597518e-06,
      "loss": 0.055,
      "step": 235750
    },
    {
      "epoch": 3.47233472261086,
      "grad_norm": 1.4134190082550049,
      "learning_rate": 6.7957873240081954e-06,
      "loss": 0.0532,
      "step": 235775
    },
    {
      "epoch": 3.472702905700947,
      "grad_norm": 0.9669596552848816,
      "learning_rate": 6.794150952041211e-06,
      "loss": 0.062,
      "step": 235800
    },
    {
      "epoch": 3.473071088791034,
      "grad_norm": 0.6366238594055176,
      "learning_rate": 6.792514580074227e-06,
      "loss": 0.054,
      "step": 235825
    },
    {
      "epoch": 3.473439271881121,
      "grad_norm": 1.2645734548568726,
      "learning_rate": 6.790878208107242e-06,
      "loss": 0.0629,
      "step": 235850
    },
    {
      "epoch": 3.473807454971208,
      "grad_norm": 1.1953133344650269,
      "learning_rate": 6.789241836140257e-06,
      "loss": 0.053,
      "step": 235875
    },
    {
      "epoch": 3.474175638061295,
      "grad_norm": 1.2158385515213013,
      "learning_rate": 6.787605464173273e-06,
      "loss": 0.0557,
      "step": 235900
    },
    {
      "epoch": 3.474543821151382,
      "grad_norm": 0.9192224740982056,
      "learning_rate": 6.785969092206288e-06,
      "loss": 0.0551,
      "step": 235925
    },
    {
      "epoch": 3.474912004241469,
      "grad_norm": 1.5386022329330444,
      "learning_rate": 6.784332720239304e-06,
      "loss": 0.0591,
      "step": 235950
    },
    {
      "epoch": 3.475280187331556,
      "grad_norm": 1.353676438331604,
      "learning_rate": 6.782696348272319e-06,
      "loss": 0.0535,
      "step": 235975
    },
    {
      "epoch": 3.475648370421643,
      "grad_norm": 1.5959298610687256,
      "learning_rate": 6.781059976305334e-06,
      "loss": 0.052,
      "step": 236000
    },
    {
      "epoch": 3.4760165535117302,
      "grad_norm": 1.2088018655776978,
      "learning_rate": 6.77942360433835e-06,
      "loss": 0.0528,
      "step": 236025
    },
    {
      "epoch": 3.4763847366018172,
      "grad_norm": 1.6774694919586182,
      "learning_rate": 6.7777872323713655e-06,
      "loss": 0.0543,
      "step": 236050
    },
    {
      "epoch": 3.4767529196919043,
      "grad_norm": 1.322643756866455,
      "learning_rate": 6.776150860404382e-06,
      "loss": 0.061,
      "step": 236075
    },
    {
      "epoch": 3.4771211027819913,
      "grad_norm": 1.6292881965637207,
      "learning_rate": 6.774514488437396e-06,
      "loss": 0.0585,
      "step": 236100
    },
    {
      "epoch": 3.4774892858720783,
      "grad_norm": 1.2966444492340088,
      "learning_rate": 6.772878116470411e-06,
      "loss": 0.0526,
      "step": 236125
    },
    {
      "epoch": 3.4778574689621653,
      "grad_norm": 1.436832308769226,
      "learning_rate": 6.7712417445034276e-06,
      "loss": 0.0562,
      "step": 236150
    },
    {
      "epoch": 3.4782256520522523,
      "grad_norm": 1.6515053510665894,
      "learning_rate": 6.769605372536443e-06,
      "loss": 0.0509,
      "step": 236175
    },
    {
      "epoch": 3.4785938351423398,
      "grad_norm": 1.1039272546768188,
      "learning_rate": 6.767969000569457e-06,
      "loss": 0.0535,
      "step": 236200
    },
    {
      "epoch": 3.478962018232427,
      "grad_norm": 1.2447068691253662,
      "learning_rate": 6.7663326286024735e-06,
      "loss": 0.0556,
      "step": 236225
    },
    {
      "epoch": 3.479330201322514,
      "grad_norm": 1.0118263959884644,
      "learning_rate": 6.764696256635489e-06,
      "loss": 0.0563,
      "step": 236250
    },
    {
      "epoch": 3.479698384412601,
      "grad_norm": 1.262508511543274,
      "learning_rate": 6.763059884668505e-06,
      "loss": 0.0511,
      "step": 236275
    },
    {
      "epoch": 3.480066567502688,
      "grad_norm": 1.5445690155029297,
      "learning_rate": 6.761423512701519e-06,
      "loss": 0.0547,
      "step": 236300
    },
    {
      "epoch": 3.480434750592775,
      "grad_norm": 0.6799076199531555,
      "learning_rate": 6.759787140734535e-06,
      "loss": 0.0556,
      "step": 236325
    },
    {
      "epoch": 3.480802933682862,
      "grad_norm": 0.9605331420898438,
      "learning_rate": 6.758150768767551e-06,
      "loss": 0.0499,
      "step": 236350
    },
    {
      "epoch": 3.481171116772949,
      "grad_norm": 0.9172486662864685,
      "learning_rate": 6.756514396800566e-06,
      "loss": 0.0504,
      "step": 236375
    },
    {
      "epoch": 3.481539299863036,
      "grad_norm": 1.4085545539855957,
      "learning_rate": 6.754878024833581e-06,
      "loss": 0.0539,
      "step": 236400
    },
    {
      "epoch": 3.481907482953123,
      "grad_norm": 1.6286646127700806,
      "learning_rate": 6.753241652866597e-06,
      "loss": 0.0559,
      "step": 236425
    },
    {
      "epoch": 3.48227566604321,
      "grad_norm": 1.5250663757324219,
      "learning_rate": 6.751605280899612e-06,
      "loss": 0.0565,
      "step": 236450
    },
    {
      "epoch": 3.482643849133297,
      "grad_norm": 1.1588718891143799,
      "learning_rate": 6.749968908932628e-06,
      "loss": 0.0576,
      "step": 236475
    },
    {
      "epoch": 3.483012032223384,
      "grad_norm": 1.209121584892273,
      "learning_rate": 6.7483325369656435e-06,
      "loss": 0.0554,
      "step": 236500
    },
    {
      "epoch": 3.483380215313471,
      "grad_norm": 1.5117765665054321,
      "learning_rate": 6.746696164998659e-06,
      "loss": 0.0532,
      "step": 236525
    },
    {
      "epoch": 3.483748398403558,
      "grad_norm": 1.2792174816131592,
      "learning_rate": 6.745059793031674e-06,
      "loss": 0.0592,
      "step": 236550
    },
    {
      "epoch": 3.484116581493645,
      "grad_norm": 0.8825824856758118,
      "learning_rate": 6.7434234210646894e-06,
      "loss": 0.0546,
      "step": 236575
    },
    {
      "epoch": 3.484484764583732,
      "grad_norm": 1.230028510093689,
      "learning_rate": 6.741787049097706e-06,
      "loss": 0.054,
      "step": 236600
    },
    {
      "epoch": 3.4848529476738195,
      "grad_norm": 1.2112821340560913,
      "learning_rate": 6.74015067713072e-06,
      "loss": 0.0623,
      "step": 236625
    },
    {
      "epoch": 3.4852211307639065,
      "grad_norm": 1.7815861701965332,
      "learning_rate": 6.738514305163736e-06,
      "loss": 0.0515,
      "step": 236650
    },
    {
      "epoch": 3.4855893138539935,
      "grad_norm": 1.379918098449707,
      "learning_rate": 6.7368779331967515e-06,
      "loss": 0.0598,
      "step": 236675
    },
    {
      "epoch": 3.4859574969440805,
      "grad_norm": 1.303359031677246,
      "learning_rate": 6.735241561229768e-06,
      "loss": 0.0574,
      "step": 236700
    },
    {
      "epoch": 3.4863256800341675,
      "grad_norm": 1.2450717687606812,
      "learning_rate": 6.733605189262782e-06,
      "loss": 0.0552,
      "step": 236725
    },
    {
      "epoch": 3.4866938631242546,
      "grad_norm": 1.3029987812042236,
      "learning_rate": 6.731968817295797e-06,
      "loss": 0.0528,
      "step": 236750
    },
    {
      "epoch": 3.4870620462143416,
      "grad_norm": 1.0911694765090942,
      "learning_rate": 6.7303324453288136e-06,
      "loss": 0.0572,
      "step": 236775
    },
    {
      "epoch": 3.4874302293044286,
      "grad_norm": 1.6581034660339355,
      "learning_rate": 6.728696073361829e-06,
      "loss": 0.0532,
      "step": 236800
    },
    {
      "epoch": 3.4877984123945156,
      "grad_norm": 1.311486005783081,
      "learning_rate": 6.727059701394843e-06,
      "loss": 0.054,
      "step": 236825
    },
    {
      "epoch": 3.4881665954846026,
      "grad_norm": 1.219870924949646,
      "learning_rate": 6.7254233294278595e-06,
      "loss": 0.0646,
      "step": 236850
    },
    {
      "epoch": 3.4885347785746896,
      "grad_norm": 1.3772090673446655,
      "learning_rate": 6.723786957460875e-06,
      "loss": 0.0509,
      "step": 236875
    },
    {
      "epoch": 3.4889029616647766,
      "grad_norm": 1.1852127313613892,
      "learning_rate": 6.722150585493891e-06,
      "loss": 0.0579,
      "step": 236900
    },
    {
      "epoch": 3.4892711447548637,
      "grad_norm": 1.2634761333465576,
      "learning_rate": 6.720514213526905e-06,
      "loss": 0.0578,
      "step": 236925
    },
    {
      "epoch": 3.4896393278449507,
      "grad_norm": 1.434631586074829,
      "learning_rate": 6.718877841559921e-06,
      "loss": 0.0497,
      "step": 236950
    },
    {
      "epoch": 3.4900075109350377,
      "grad_norm": 1.5747640132904053,
      "learning_rate": 6.717241469592937e-06,
      "loss": 0.0523,
      "step": 236975
    },
    {
      "epoch": 3.4903756940251247,
      "grad_norm": 1.2704558372497559,
      "learning_rate": 6.715605097625952e-06,
      "loss": 0.0645,
      "step": 237000
    },
    {
      "epoch": 3.4907438771152117,
      "grad_norm": 1.4632608890533447,
      "learning_rate": 6.713968725658968e-06,
      "loss": 0.0531,
      "step": 237025
    },
    {
      "epoch": 3.4911120602052987,
      "grad_norm": 1.9516022205352783,
      "learning_rate": 6.712332353691983e-06,
      "loss": 0.0571,
      "step": 237050
    },
    {
      "epoch": 3.4914802432953858,
      "grad_norm": 1.6178995370864868,
      "learning_rate": 6.710695981724998e-06,
      "loss": 0.0499,
      "step": 237075
    },
    {
      "epoch": 3.4918484263854728,
      "grad_norm": 1.7712401151657104,
      "learning_rate": 6.709059609758014e-06,
      "loss": 0.0619,
      "step": 237100
    },
    {
      "epoch": 3.49221660947556,
      "grad_norm": 1.120568871498108,
      "learning_rate": 6.707488692669709e-06,
      "loss": 0.0592,
      "step": 237125
    },
    {
      "epoch": 3.492584792565647,
      "grad_norm": 1.1097540855407715,
      "learning_rate": 6.705852320702723e-06,
      "loss": 0.0571,
      "step": 237150
    },
    {
      "epoch": 3.4929529756557343,
      "grad_norm": 1.0537636280059814,
      "learning_rate": 6.7042159487357395e-06,
      "loss": 0.0554,
      "step": 237175
    },
    {
      "epoch": 3.4933211587458213,
      "grad_norm": 1.3439337015151978,
      "learning_rate": 6.702579576768755e-06,
      "loss": 0.0588,
      "step": 237200
    },
    {
      "epoch": 3.4936893418359083,
      "grad_norm": 1.8948816061019897,
      "learning_rate": 6.700943204801771e-06,
      "loss": 0.0539,
      "step": 237225
    },
    {
      "epoch": 3.4940575249259953,
      "grad_norm": 1.3520216941833496,
      "learning_rate": 6.6993068328347855e-06,
      "loss": 0.0602,
      "step": 237250
    },
    {
      "epoch": 3.4944257080160823,
      "grad_norm": 1.1100796461105347,
      "learning_rate": 6.697670460867801e-06,
      "loss": 0.0537,
      "step": 237275
    },
    {
      "epoch": 3.4947938911061693,
      "grad_norm": 1.159147024154663,
      "learning_rate": 6.696034088900817e-06,
      "loss": 0.0618,
      "step": 237300
    },
    {
      "epoch": 3.4951620741962564,
      "grad_norm": 1.400046467781067,
      "learning_rate": 6.694397716933832e-06,
      "loss": 0.0524,
      "step": 237325
    },
    {
      "epoch": 3.4955302572863434,
      "grad_norm": 1.0062354803085327,
      "learning_rate": 6.6927613449668475e-06,
      "loss": 0.0512,
      "step": 237350
    },
    {
      "epoch": 3.4958984403764304,
      "grad_norm": 1.3053812980651855,
      "learning_rate": 6.691124972999863e-06,
      "loss": 0.0503,
      "step": 237375
    },
    {
      "epoch": 3.4962666234665174,
      "grad_norm": 1.707844853401184,
      "learning_rate": 6.689488601032879e-06,
      "loss": 0.0567,
      "step": 237400
    },
    {
      "epoch": 3.4966348065566044,
      "grad_norm": 1.378909707069397,
      "learning_rate": 6.687852229065894e-06,
      "loss": 0.0561,
      "step": 237425
    },
    {
      "epoch": 3.4970029896466914,
      "grad_norm": 1.0125243663787842,
      "learning_rate": 6.68621585709891e-06,
      "loss": 0.0493,
      "step": 237450
    },
    {
      "epoch": 3.4973711727367784,
      "grad_norm": 1.4544315338134766,
      "learning_rate": 6.684579485131925e-06,
      "loss": 0.0628,
      "step": 237475
    },
    {
      "epoch": 3.4977393558268655,
      "grad_norm": 1.692452073097229,
      "learning_rate": 6.68294311316494e-06,
      "loss": 0.0542,
      "step": 237500
    },
    {
      "epoch": 3.4981075389169525,
      "grad_norm": 1.0853323936462402,
      "learning_rate": 6.681306741197956e-06,
      "loss": 0.0563,
      "step": 237525
    },
    {
      "epoch": 3.4984757220070395,
      "grad_norm": 1.0275168418884277,
      "learning_rate": 6.679670369230972e-06,
      "loss": 0.0589,
      "step": 237550
    },
    {
      "epoch": 3.4988439050971265,
      "grad_norm": 2.008805274963379,
      "learning_rate": 6.678033997263986e-06,
      "loss": 0.0589,
      "step": 237575
    },
    {
      "epoch": 3.499212088187214,
      "grad_norm": 1.8501317501068115,
      "learning_rate": 6.676397625297002e-06,
      "loss": 0.0563,
      "step": 237600
    },
    {
      "epoch": 3.499580271277301,
      "grad_norm": 1.7143025398254395,
      "learning_rate": 6.6747612533300176e-06,
      "loss": 0.0553,
      "step": 237625
    },
    {
      "epoch": 3.499948454367388,
      "grad_norm": 1.7455675601959229,
      "learning_rate": 6.673124881363034e-06,
      "loss": 0.0596,
      "step": 237650
    },
    {
      "epoch": 3.500316637457475,
      "grad_norm": 1.641186237335205,
      "learning_rate": 6.671488509396048e-06,
      "loss": 0.0578,
      "step": 237675
    },
    {
      "epoch": 3.500684820547562,
      "grad_norm": 1.3982226848602295,
      "learning_rate": 6.6698521374290635e-06,
      "loss": 0.0524,
      "step": 237700
    },
    {
      "epoch": 3.501053003637649,
      "grad_norm": 1.5200787782669067,
      "learning_rate": 6.66821576546208e-06,
      "loss": 0.0542,
      "step": 237725
    },
    {
      "epoch": 3.501421186727736,
      "grad_norm": 1.4995650053024292,
      "learning_rate": 6.666579393495095e-06,
      "loss": 0.0483,
      "step": 237750
    },
    {
      "epoch": 3.501789369817823,
      "grad_norm": 1.4795305728912354,
      "learning_rate": 6.664943021528109e-06,
      "loss": 0.0564,
      "step": 237775
    },
    {
      "epoch": 3.50215755290791,
      "grad_norm": 1.0468206405639648,
      "learning_rate": 6.6633066495611255e-06,
      "loss": 0.0516,
      "step": 237800
    },
    {
      "epoch": 3.502525735997997,
      "grad_norm": 1.4110145568847656,
      "learning_rate": 6.661670277594141e-06,
      "loss": 0.0537,
      "step": 237825
    },
    {
      "epoch": 3.502893919088084,
      "grad_norm": 1.5681344270706177,
      "learning_rate": 6.660033905627157e-06,
      "loss": 0.0568,
      "step": 237850
    },
    {
      "epoch": 3.503262102178171,
      "grad_norm": 1.2710176706314087,
      "learning_rate": 6.6583975336601715e-06,
      "loss": 0.0527,
      "step": 237875
    },
    {
      "epoch": 3.503630285268258,
      "grad_norm": 1.0221668481826782,
      "learning_rate": 6.656761161693187e-06,
      "loss": 0.0497,
      "step": 237900
    },
    {
      "epoch": 3.503998468358345,
      "grad_norm": 1.3735289573669434,
      "learning_rate": 6.655124789726203e-06,
      "loss": 0.0543,
      "step": 237925
    },
    {
      "epoch": 3.504366651448432,
      "grad_norm": 1.8345123529434204,
      "learning_rate": 6.653488417759218e-06,
      "loss": 0.0564,
      "step": 237950
    },
    {
      "epoch": 3.504734834538519,
      "grad_norm": 1.3793649673461914,
      "learning_rate": 6.651852045792234e-06,
      "loss": 0.0598,
      "step": 237975
    },
    {
      "epoch": 3.505103017628606,
      "grad_norm": 0.896697998046875,
      "learning_rate": 6.650215673825249e-06,
      "loss": 0.0555,
      "step": 238000
    },
    {
      "epoch": 3.5054712007186932,
      "grad_norm": 1.7305998802185059,
      "learning_rate": 6.648579301858264e-06,
      "loss": 0.0489,
      "step": 238025
    },
    {
      "epoch": 3.5058393838087802,
      "grad_norm": 1.2747411727905273,
      "learning_rate": 6.64694292989128e-06,
      "loss": 0.0592,
      "step": 238050
    },
    {
      "epoch": 3.5062075668988673,
      "grad_norm": 1.5657886266708374,
      "learning_rate": 6.645306557924296e-06,
      "loss": 0.0609,
      "step": 238075
    },
    {
      "epoch": 3.5065757499889543,
      "grad_norm": 1.0818257331848145,
      "learning_rate": 6.643670185957311e-06,
      "loss": 0.0587,
      "step": 238100
    },
    {
      "epoch": 3.5069439330790413,
      "grad_norm": 1.3494561910629272,
      "learning_rate": 6.642033813990326e-06,
      "loss": 0.0607,
      "step": 238125
    },
    {
      "epoch": 3.5073121161691287,
      "grad_norm": 1.1361825466156006,
      "learning_rate": 6.6403974420233415e-06,
      "loss": 0.0543,
      "step": 238150
    },
    {
      "epoch": 3.5076802992592158,
      "grad_norm": 1.1191574335098267,
      "learning_rate": 6.638761070056358e-06,
      "loss": 0.0476,
      "step": 238175
    },
    {
      "epoch": 3.5080484823493028,
      "grad_norm": 1.2232915163040161,
      "learning_rate": 6.637124698089372e-06,
      "loss": 0.0529,
      "step": 238200
    },
    {
      "epoch": 3.50841666543939,
      "grad_norm": 0.8323943614959717,
      "learning_rate": 6.635488326122388e-06,
      "loss": 0.0472,
      "step": 238225
    },
    {
      "epoch": 3.508784848529477,
      "grad_norm": 1.4062432050704956,
      "learning_rate": 6.6338519541554036e-06,
      "loss": 0.0613,
      "step": 238250
    },
    {
      "epoch": 3.509153031619564,
      "grad_norm": 1.2244796752929688,
      "learning_rate": 6.63221558218842e-06,
      "loss": 0.0579,
      "step": 238275
    },
    {
      "epoch": 3.509521214709651,
      "grad_norm": 1.6239651441574097,
      "learning_rate": 6.630579210221434e-06,
      "loss": 0.0549,
      "step": 238300
    },
    {
      "epoch": 3.509889397799738,
      "grad_norm": 1.101367473602295,
      "learning_rate": 6.6289428382544495e-06,
      "loss": 0.0517,
      "step": 238325
    },
    {
      "epoch": 3.510257580889825,
      "grad_norm": 1.7837995290756226,
      "learning_rate": 6.627306466287466e-06,
      "loss": 0.0539,
      "step": 238350
    },
    {
      "epoch": 3.510625763979912,
      "grad_norm": 1.1036487817764282,
      "learning_rate": 6.625670094320481e-06,
      "loss": 0.0531,
      "step": 238375
    },
    {
      "epoch": 3.510993947069999,
      "grad_norm": 1.0700322389602661,
      "learning_rate": 6.624033722353495e-06,
      "loss": 0.0537,
      "step": 238400
    },
    {
      "epoch": 3.511362130160086,
      "grad_norm": 1.2576203346252441,
      "learning_rate": 6.6223973503865115e-06,
      "loss": 0.0534,
      "step": 238425
    },
    {
      "epoch": 3.511730313250173,
      "grad_norm": 0.8141782879829407,
      "learning_rate": 6.620760978419527e-06,
      "loss": 0.0484,
      "step": 238450
    },
    {
      "epoch": 3.51209849634026,
      "grad_norm": 1.3504191637039185,
      "learning_rate": 6.619124606452543e-06,
      "loss": 0.0566,
      "step": 238475
    },
    {
      "epoch": 3.512466679430347,
      "grad_norm": 1.2083675861358643,
      "learning_rate": 6.617488234485558e-06,
      "loss": 0.0545,
      "step": 238500
    },
    {
      "epoch": 3.5128348625204344,
      "grad_norm": 1.0957146883010864,
      "learning_rate": 6.615851862518573e-06,
      "loss": 0.055,
      "step": 238525
    },
    {
      "epoch": 3.5132030456105214,
      "grad_norm": 1.055159568786621,
      "learning_rate": 6.614215490551589e-06,
      "loss": 0.0535,
      "step": 238550
    },
    {
      "epoch": 3.5135712287006085,
      "grad_norm": 1.4225109815597534,
      "learning_rate": 6.612579118584604e-06,
      "loss": 0.0529,
      "step": 238575
    },
    {
      "epoch": 3.5139394117906955,
      "grad_norm": 1.0083485841751099,
      "learning_rate": 6.61094274661762e-06,
      "loss": 0.0585,
      "step": 238600
    },
    {
      "epoch": 3.5143075948807825,
      "grad_norm": 0.8519147038459778,
      "learning_rate": 6.609306374650635e-06,
      "loss": 0.0564,
      "step": 238625
    },
    {
      "epoch": 3.5146757779708695,
      "grad_norm": 1.3270843029022217,
      "learning_rate": 6.60767000268365e-06,
      "loss": 0.0536,
      "step": 238650
    },
    {
      "epoch": 3.5150439610609565,
      "grad_norm": 1.1831995248794556,
      "learning_rate": 6.606033630716666e-06,
      "loss": 0.0545,
      "step": 238675
    },
    {
      "epoch": 3.5154121441510435,
      "grad_norm": 2.0273499488830566,
      "learning_rate": 6.604397258749682e-06,
      "loss": 0.0581,
      "step": 238700
    },
    {
      "epoch": 3.5157803272411305,
      "grad_norm": 1.365114688873291,
      "learning_rate": 6.602760886782696e-06,
      "loss": 0.0555,
      "step": 238725
    },
    {
      "epoch": 3.5161485103312176,
      "grad_norm": 1.4722472429275513,
      "learning_rate": 6.601124514815712e-06,
      "loss": 0.0558,
      "step": 238750
    },
    {
      "epoch": 3.5165166934213046,
      "grad_norm": 1.4480513334274292,
      "learning_rate": 6.5994881428487275e-06,
      "loss": 0.0493,
      "step": 238775
    },
    {
      "epoch": 3.5168848765113916,
      "grad_norm": 1.4381823539733887,
      "learning_rate": 6.597851770881744e-06,
      "loss": 0.0626,
      "step": 238800
    },
    {
      "epoch": 3.5172530596014786,
      "grad_norm": 0.95949786901474,
      "learning_rate": 6.596215398914758e-06,
      "loss": 0.0631,
      "step": 238825
    },
    {
      "epoch": 3.5176212426915656,
      "grad_norm": 1.0725449323654175,
      "learning_rate": 6.594579026947774e-06,
      "loss": 0.052,
      "step": 238850
    },
    {
      "epoch": 3.5179894257816526,
      "grad_norm": 1.4058222770690918,
      "learning_rate": 6.5929426549807896e-06,
      "loss": 0.0502,
      "step": 238875
    },
    {
      "epoch": 3.5183576088717397,
      "grad_norm": 1.219465732574463,
      "learning_rate": 6.591306283013805e-06,
      "loss": 0.0506,
      "step": 238900
    },
    {
      "epoch": 3.5187257919618267,
      "grad_norm": 1.4697891473770142,
      "learning_rate": 6.589669911046821e-06,
      "loss": 0.0483,
      "step": 238925
    },
    {
      "epoch": 3.5190939750519137,
      "grad_norm": 1.4222756624221802,
      "learning_rate": 6.5880335390798355e-06,
      "loss": 0.0543,
      "step": 238950
    },
    {
      "epoch": 3.5194621581420007,
      "grad_norm": 1.1545531749725342,
      "learning_rate": 6.586397167112852e-06,
      "loss": 0.0513,
      "step": 238975
    },
    {
      "epoch": 3.5198303412320877,
      "grad_norm": 1.1878083944320679,
      "learning_rate": 6.584760795145867e-06,
      "loss": 0.0504,
      "step": 239000
    },
    {
      "epoch": 3.5201985243221747,
      "grad_norm": 1.4583110809326172,
      "learning_rate": 6.583124423178883e-06,
      "loss": 0.0561,
      "step": 239025
    },
    {
      "epoch": 3.5205667074122617,
      "grad_norm": 1.1590267419815063,
      "learning_rate": 6.5814880512118975e-06,
      "loss": 0.0593,
      "step": 239050
    },
    {
      "epoch": 3.5209348905023488,
      "grad_norm": 1.353603482246399,
      "learning_rate": 6.579851679244913e-06,
      "loss": 0.0522,
      "step": 239075
    },
    {
      "epoch": 3.5213030735924358,
      "grad_norm": 1.6820868253707886,
      "learning_rate": 6.578215307277929e-06,
      "loss": 0.058,
      "step": 239100
    },
    {
      "epoch": 3.5216712566825232,
      "grad_norm": 1.0427802801132202,
      "learning_rate": 6.576578935310944e-06,
      "loss": 0.0495,
      "step": 239125
    },
    {
      "epoch": 3.5220394397726102,
      "grad_norm": 1.4165109395980835,
      "learning_rate": 6.574942563343959e-06,
      "loss": 0.0517,
      "step": 239150
    },
    {
      "epoch": 3.5224076228626973,
      "grad_norm": 1.5403512716293335,
      "learning_rate": 6.573306191376975e-06,
      "loss": 0.0607,
      "step": 239175
    },
    {
      "epoch": 3.5227758059527843,
      "grad_norm": 1.482560634613037,
      "learning_rate": 6.57166981940999e-06,
      "loss": 0.058,
      "step": 239200
    },
    {
      "epoch": 3.5231439890428713,
      "grad_norm": 1.7138524055480957,
      "learning_rate": 6.570033447443006e-06,
      "loss": 0.0581,
      "step": 239225
    },
    {
      "epoch": 3.5235121721329583,
      "grad_norm": 1.4876329898834229,
      "learning_rate": 6.568397075476021e-06,
      "loss": 0.0556,
      "step": 239250
    },
    {
      "epoch": 3.5238803552230453,
      "grad_norm": 1.2482026815414429,
      "learning_rate": 6.566760703509036e-06,
      "loss": 0.0603,
      "step": 239275
    },
    {
      "epoch": 3.5242485383131323,
      "grad_norm": 1.320499062538147,
      "learning_rate": 6.565124331542052e-06,
      "loss": 0.0525,
      "step": 239300
    },
    {
      "epoch": 3.5246167214032194,
      "grad_norm": 1.2832379341125488,
      "learning_rate": 6.563553414453747e-06,
      "loss": 0.0659,
      "step": 239325
    },
    {
      "epoch": 3.5249849044933064,
      "grad_norm": 1.1847057342529297,
      "learning_rate": 6.5619170424867615e-06,
      "loss": 0.0519,
      "step": 239350
    },
    {
      "epoch": 3.5253530875833934,
      "grad_norm": 1.2214913368225098,
      "learning_rate": 6.560280670519778e-06,
      "loss": 0.06,
      "step": 239375
    },
    {
      "epoch": 3.5257212706734804,
      "grad_norm": 1.207951307296753,
      "learning_rate": 6.558644298552793e-06,
      "loss": 0.0469,
      "step": 239400
    },
    {
      "epoch": 3.5260894537635674,
      "grad_norm": 1.6714341640472412,
      "learning_rate": 6.557007926585809e-06,
      "loss": 0.0599,
      "step": 239425
    },
    {
      "epoch": 3.5264576368536544,
      "grad_norm": 1.8564172983169556,
      "learning_rate": 6.555371554618824e-06,
      "loss": 0.0528,
      "step": 239450
    },
    {
      "epoch": 3.526825819943742,
      "grad_norm": 1.068618655204773,
      "learning_rate": 6.553735182651839e-06,
      "loss": 0.051,
      "step": 239475
    },
    {
      "epoch": 3.527194003033829,
      "grad_norm": 1.001126766204834,
      "learning_rate": 6.552098810684855e-06,
      "loss": 0.0622,
      "step": 239500
    },
    {
      "epoch": 3.527562186123916,
      "grad_norm": 1.6345113515853882,
      "learning_rate": 6.55046243871787e-06,
      "loss": 0.0506,
      "step": 239525
    },
    {
      "epoch": 3.527930369214003,
      "grad_norm": 1.3190492391586304,
      "learning_rate": 6.5488260667508864e-06,
      "loss": 0.0554,
      "step": 239550
    },
    {
      "epoch": 3.52829855230409,
      "grad_norm": 1.4491008520126343,
      "learning_rate": 6.547189694783901e-06,
      "loss": 0.0598,
      "step": 239575
    },
    {
      "epoch": 3.528666735394177,
      "grad_norm": 1.3110566139221191,
      "learning_rate": 6.545553322816916e-06,
      "loss": 0.0538,
      "step": 239600
    },
    {
      "epoch": 3.529034918484264,
      "grad_norm": 1.1355332136154175,
      "learning_rate": 6.543916950849932e-06,
      "loss": 0.0545,
      "step": 239625
    },
    {
      "epoch": 3.529403101574351,
      "grad_norm": 0.5813958644866943,
      "learning_rate": 6.542280578882948e-06,
      "loss": 0.0528,
      "step": 239650
    },
    {
      "epoch": 3.529771284664438,
      "grad_norm": 1.1517421007156372,
      "learning_rate": 6.540644206915963e-06,
      "loss": 0.0488,
      "step": 239675
    },
    {
      "epoch": 3.530139467754525,
      "grad_norm": 1.0873079299926758,
      "learning_rate": 6.539007834948978e-06,
      "loss": 0.0519,
      "step": 239700
    },
    {
      "epoch": 3.530507650844612,
      "grad_norm": 1.235602855682373,
      "learning_rate": 6.5373714629819936e-06,
      "loss": 0.0562,
      "step": 239725
    },
    {
      "epoch": 3.530875833934699,
      "grad_norm": 1.028826117515564,
      "learning_rate": 6.53573509101501e-06,
      "loss": 0.0544,
      "step": 239750
    },
    {
      "epoch": 3.531244017024786,
      "grad_norm": 1.5451363325119019,
      "learning_rate": 6.534098719048024e-06,
      "loss": 0.0528,
      "step": 239775
    },
    {
      "epoch": 3.531612200114873,
      "grad_norm": 1.3391128778457642,
      "learning_rate": 6.53246234708104e-06,
      "loss": 0.0519,
      "step": 239800
    },
    {
      "epoch": 3.53198038320496,
      "grad_norm": 1.2500059604644775,
      "learning_rate": 6.530825975114056e-06,
      "loss": 0.0533,
      "step": 239825
    },
    {
      "epoch": 3.532348566295047,
      "grad_norm": 1.6858341693878174,
      "learning_rate": 6.529189603147072e-06,
      "loss": 0.0555,
      "step": 239850
    },
    {
      "epoch": 3.532716749385134,
      "grad_norm": 1.4265520572662354,
      "learning_rate": 6.527553231180086e-06,
      "loss": 0.0515,
      "step": 239875
    },
    {
      "epoch": 3.533084932475221,
      "grad_norm": 1.5402430295944214,
      "learning_rate": 6.5259168592131015e-06,
      "loss": 0.0594,
      "step": 239900
    },
    {
      "epoch": 3.533453115565308,
      "grad_norm": 1.5277717113494873,
      "learning_rate": 6.524280487246118e-06,
      "loss": 0.05,
      "step": 239925
    },
    {
      "epoch": 3.533821298655395,
      "grad_norm": 1.59786057472229,
      "learning_rate": 6.522644115279133e-06,
      "loss": 0.0512,
      "step": 239950
    },
    {
      "epoch": 3.534189481745482,
      "grad_norm": 1.531436562538147,
      "learning_rate": 6.521007743312149e-06,
      "loss": 0.0573,
      "step": 239975
    },
    {
      "epoch": 3.534557664835569,
      "grad_norm": 1.2132772207260132,
      "learning_rate": 6.519371371345164e-06,
      "loss": 0.0581,
      "step": 240000
    },
    {
      "epoch": 3.5349258479256562,
      "grad_norm": 0.9268856048583984,
      "learning_rate": 6.517734999378179e-06,
      "loss": 0.0573,
      "step": 240025
    },
    {
      "epoch": 3.5352940310157432,
      "grad_norm": 1.2429592609405518,
      "learning_rate": 6.516098627411195e-06,
      "loss": 0.0555,
      "step": 240050
    },
    {
      "epoch": 3.5356622141058307,
      "grad_norm": 1.0812588930130005,
      "learning_rate": 6.51446225544421e-06,
      "loss": 0.0512,
      "step": 240075
    },
    {
      "epoch": 3.5360303971959177,
      "grad_norm": 1.0637778043746948,
      "learning_rate": 6.512825883477225e-06,
      "loss": 0.0523,
      "step": 240100
    },
    {
      "epoch": 3.5363985802860047,
      "grad_norm": 1.4924452304840088,
      "learning_rate": 6.511189511510241e-06,
      "loss": 0.057,
      "step": 240125
    },
    {
      "epoch": 3.5367667633760917,
      "grad_norm": 1.6891506910324097,
      "learning_rate": 6.509553139543256e-06,
      "loss": 0.0567,
      "step": 240150
    },
    {
      "epoch": 3.5371349464661788,
      "grad_norm": 1.0667810440063477,
      "learning_rate": 6.5079167675762724e-06,
      "loss": 0.0528,
      "step": 240175
    },
    {
      "epoch": 3.537503129556266,
      "grad_norm": 1.1576346158981323,
      "learning_rate": 6.506280395609287e-06,
      "loss": 0.0541,
      "step": 240200
    },
    {
      "epoch": 3.537871312646353,
      "grad_norm": 1.3905154466629028,
      "learning_rate": 6.504644023642302e-06,
      "loss": 0.0487,
      "step": 240225
    },
    {
      "epoch": 3.53823949573644,
      "grad_norm": 1.4382922649383545,
      "learning_rate": 6.503007651675318e-06,
      "loss": 0.0548,
      "step": 240250
    },
    {
      "epoch": 3.538607678826527,
      "grad_norm": 1.3392270803451538,
      "learning_rate": 6.501371279708334e-06,
      "loss": 0.0572,
      "step": 240275
    },
    {
      "epoch": 3.538975861916614,
      "grad_norm": 1.4199312925338745,
      "learning_rate": 6.499734907741348e-06,
      "loss": 0.0554,
      "step": 240300
    },
    {
      "epoch": 3.539344045006701,
      "grad_norm": 1.397796630859375,
      "learning_rate": 6.498098535774364e-06,
      "loss": 0.0556,
      "step": 240325
    },
    {
      "epoch": 3.539712228096788,
      "grad_norm": 1.3609561920166016,
      "learning_rate": 6.4964621638073796e-06,
      "loss": 0.0484,
      "step": 240350
    },
    {
      "epoch": 3.540080411186875,
      "grad_norm": 1.6534154415130615,
      "learning_rate": 6.494825791840396e-06,
      "loss": 0.0479,
      "step": 240375
    },
    {
      "epoch": 3.540448594276962,
      "grad_norm": 1.191688895225525,
      "learning_rate": 6.493189419873411e-06,
      "loss": 0.0503,
      "step": 240400
    },
    {
      "epoch": 3.5408167773670494,
      "grad_norm": 1.6409965753555298,
      "learning_rate": 6.491553047906426e-06,
      "loss": 0.0526,
      "step": 240425
    },
    {
      "epoch": 3.5411849604571364,
      "grad_norm": 1.658474087715149,
      "learning_rate": 6.489916675939442e-06,
      "loss": 0.0573,
      "step": 240450
    },
    {
      "epoch": 3.5415531435472234,
      "grad_norm": 1.2004069089889526,
      "learning_rate": 6.488280303972457e-06,
      "loss": 0.055,
      "step": 240475
    },
    {
      "epoch": 3.5419213266373104,
      "grad_norm": 1.3615010976791382,
      "learning_rate": 6.486643932005473e-06,
      "loss": 0.0617,
      "step": 240500
    },
    {
      "epoch": 3.5422895097273974,
      "grad_norm": 1.157260775566101,
      "learning_rate": 6.4850075600384875e-06,
      "loss": 0.0551,
      "step": 240525
    },
    {
      "epoch": 3.5426576928174844,
      "grad_norm": 1.0880311727523804,
      "learning_rate": 6.483371188071504e-06,
      "loss": 0.0488,
      "step": 240550
    },
    {
      "epoch": 3.5430258759075715,
      "grad_norm": 1.7445656061172485,
      "learning_rate": 6.481734816104519e-06,
      "loss": 0.0542,
      "step": 240575
    },
    {
      "epoch": 3.5433940589976585,
      "grad_norm": 1.377503514289856,
      "learning_rate": 6.480098444137535e-06,
      "loss": 0.0509,
      "step": 240600
    },
    {
      "epoch": 3.5437622420877455,
      "grad_norm": 1.4644700288772583,
      "learning_rate": 6.47846207217055e-06,
      "loss": 0.0538,
      "step": 240625
    },
    {
      "epoch": 3.5441304251778325,
      "grad_norm": 1.2659029960632324,
      "learning_rate": 6.476825700203565e-06,
      "loss": 0.0516,
      "step": 240650
    },
    {
      "epoch": 3.5444986082679195,
      "grad_norm": 1.73847234249115,
      "learning_rate": 6.475189328236581e-06,
      "loss": 0.0577,
      "step": 240675
    },
    {
      "epoch": 3.5448667913580065,
      "grad_norm": 1.5496197938919067,
      "learning_rate": 6.473552956269596e-06,
      "loss": 0.055,
      "step": 240700
    },
    {
      "epoch": 3.5452349744480935,
      "grad_norm": 1.3950140476226807,
      "learning_rate": 6.471916584302611e-06,
      "loss": 0.0492,
      "step": 240725
    },
    {
      "epoch": 3.5456031575381806,
      "grad_norm": 1.6770752668380737,
      "learning_rate": 6.470280212335627e-06,
      "loss": 0.0542,
      "step": 240750
    },
    {
      "epoch": 3.5459713406282676,
      "grad_norm": 1.4783209562301636,
      "learning_rate": 6.468643840368642e-06,
      "loss": 0.0499,
      "step": 240775
    },
    {
      "epoch": 3.5463395237183546,
      "grad_norm": 1.030889630317688,
      "learning_rate": 6.4670074684016584e-06,
      "loss": 0.055,
      "step": 240800
    },
    {
      "epoch": 3.5467077068084416,
      "grad_norm": 1.4858734607696533,
      "learning_rate": 6.465371096434673e-06,
      "loss": 0.0559,
      "step": 240825
    },
    {
      "epoch": 3.5470758898985286,
      "grad_norm": 1.286137342453003,
      "learning_rate": 6.463734724467688e-06,
      "loss": 0.0516,
      "step": 240850
    },
    {
      "epoch": 3.5474440729886156,
      "grad_norm": 1.4824665784835815,
      "learning_rate": 6.462098352500704e-06,
      "loss": 0.0542,
      "step": 240875
    },
    {
      "epoch": 3.5478122560787027,
      "grad_norm": 1.2878632545471191,
      "learning_rate": 6.46046198053372e-06,
      "loss": 0.0573,
      "step": 240900
    },
    {
      "epoch": 3.5481804391687897,
      "grad_norm": 1.1780720949172974,
      "learning_rate": 6.458825608566736e-06,
      "loss": 0.0547,
      "step": 240925
    },
    {
      "epoch": 3.5485486222588767,
      "grad_norm": 0.6847298741340637,
      "learning_rate": 6.45718923659975e-06,
      "loss": 0.0529,
      "step": 240950
    },
    {
      "epoch": 3.5489168053489637,
      "grad_norm": 1.3146413564682007,
      "learning_rate": 6.4555528646327656e-06,
      "loss": 0.0492,
      "step": 240975
    },
    {
      "epoch": 3.5492849884390507,
      "grad_norm": 1.2977025508880615,
      "learning_rate": 6.453916492665782e-06,
      "loss": 0.0533,
      "step": 241000
    },
    {
      "epoch": 3.549653171529138,
      "grad_norm": 1.5004416704177856,
      "learning_rate": 6.452280120698797e-06,
      "loss": 0.0489,
      "step": 241025
    },
    {
      "epoch": 3.550021354619225,
      "grad_norm": 1.2740706205368042,
      "learning_rate": 6.4506437487318115e-06,
      "loss": 0.0594,
      "step": 241050
    },
    {
      "epoch": 3.550389537709312,
      "grad_norm": 1.189389944076538,
      "learning_rate": 6.449007376764828e-06,
      "loss": 0.055,
      "step": 241075
    },
    {
      "epoch": 3.550757720799399,
      "grad_norm": 1.220145344734192,
      "learning_rate": 6.447371004797843e-06,
      "loss": 0.0525,
      "step": 241100
    },
    {
      "epoch": 3.5511259038894862,
      "grad_norm": 1.157399296760559,
      "learning_rate": 6.445734632830859e-06,
      "loss": 0.0537,
      "step": 241125
    },
    {
      "epoch": 3.5514940869795733,
      "grad_norm": 0.7158670425415039,
      "learning_rate": 6.4440982608638735e-06,
      "loss": 0.0544,
      "step": 241150
    },
    {
      "epoch": 3.5518622700696603,
      "grad_norm": 0.9459164142608643,
      "learning_rate": 6.44246188889689e-06,
      "loss": 0.0542,
      "step": 241175
    },
    {
      "epoch": 3.5522304531597473,
      "grad_norm": 1.2373384237289429,
      "learning_rate": 6.440825516929905e-06,
      "loss": 0.0583,
      "step": 241200
    },
    {
      "epoch": 3.5525986362498343,
      "grad_norm": 1.261459231376648,
      "learning_rate": 6.43918914496292e-06,
      "loss": 0.051,
      "step": 241225
    },
    {
      "epoch": 3.5529668193399213,
      "grad_norm": 1.8835111856460571,
      "learning_rate": 6.437552772995936e-06,
      "loss": 0.061,
      "step": 241250
    },
    {
      "epoch": 3.5533350024300083,
      "grad_norm": 0.9285681843757629,
      "learning_rate": 6.435916401028951e-06,
      "loss": 0.0508,
      "step": 241275
    },
    {
      "epoch": 3.5537031855200953,
      "grad_norm": 1.9005039930343628,
      "learning_rate": 6.434280029061967e-06,
      "loss": 0.0538,
      "step": 241300
    },
    {
      "epoch": 3.5540713686101824,
      "grad_norm": 0.9641976952552795,
      "learning_rate": 6.432643657094982e-06,
      "loss": 0.0504,
      "step": 241325
    },
    {
      "epoch": 3.5544395517002694,
      "grad_norm": 1.4524327516555786,
      "learning_rate": 6.431007285127997e-06,
      "loss": 0.0564,
      "step": 241350
    },
    {
      "epoch": 3.5548077347903564,
      "grad_norm": 1.343019723892212,
      "learning_rate": 6.429370913161013e-06,
      "loss": 0.052,
      "step": 241375
    },
    {
      "epoch": 3.555175917880444,
      "grad_norm": 0.9352220892906189,
      "learning_rate": 6.427734541194028e-06,
      "loss": 0.0563,
      "step": 241400
    },
    {
      "epoch": 3.555544100970531,
      "grad_norm": 1.2362533807754517,
      "learning_rate": 6.4260981692270444e-06,
      "loss": 0.0523,
      "step": 241425
    },
    {
      "epoch": 3.555912284060618,
      "grad_norm": 1.329370141029358,
      "learning_rate": 6.42446179726006e-06,
      "loss": 0.0556,
      "step": 241450
    },
    {
      "epoch": 3.556280467150705,
      "grad_norm": 1.212063193321228,
      "learning_rate": 6.422825425293074e-06,
      "loss": 0.0592,
      "step": 241475
    },
    {
      "epoch": 3.556648650240792,
      "grad_norm": 1.6169648170471191,
      "learning_rate": 6.42118905332609e-06,
      "loss": 0.0511,
      "step": 241500
    },
    {
      "epoch": 3.557016833330879,
      "grad_norm": 1.2143300771713257,
      "learning_rate": 6.419552681359106e-06,
      "loss": 0.0499,
      "step": 241525
    },
    {
      "epoch": 3.557385016420966,
      "grad_norm": 1.3234690427780151,
      "learning_rate": 6.417916309392122e-06,
      "loss": 0.0579,
      "step": 241550
    },
    {
      "epoch": 3.557753199511053,
      "grad_norm": 1.5765708684921265,
      "learning_rate": 6.416279937425136e-06,
      "loss": 0.0621,
      "step": 241575
    },
    {
      "epoch": 3.55812138260114,
      "grad_norm": 1.1030410528182983,
      "learning_rate": 6.4146435654581516e-06,
      "loss": 0.0521,
      "step": 241600
    },
    {
      "epoch": 3.558489565691227,
      "grad_norm": 1.5761111974716187,
      "learning_rate": 6.413072648369847e-06,
      "loss": 0.0609,
      "step": 241625
    },
    {
      "epoch": 3.558857748781314,
      "grad_norm": 1.3453253507614136,
      "learning_rate": 6.4114362764028624e-06,
      "loss": 0.0545,
      "step": 241650
    },
    {
      "epoch": 3.559225931871401,
      "grad_norm": 1.520203948020935,
      "learning_rate": 6.409799904435877e-06,
      "loss": 0.0537,
      "step": 241675
    },
    {
      "epoch": 3.559594114961488,
      "grad_norm": 0.9888783693313599,
      "learning_rate": 6.408163532468893e-06,
      "loss": 0.0559,
      "step": 241700
    },
    {
      "epoch": 3.559962298051575,
      "grad_norm": 1.7509855031967163,
      "learning_rate": 6.406527160501908e-06,
      "loss": 0.0464,
      "step": 241725
    },
    {
      "epoch": 3.560330481141662,
      "grad_norm": 1.3090428113937378,
      "learning_rate": 6.4048907885349245e-06,
      "loss": 0.0536,
      "step": 241750
    },
    {
      "epoch": 3.560698664231749,
      "grad_norm": 1.4758968353271484,
      "learning_rate": 6.403254416567939e-06,
      "loss": 0.0507,
      "step": 241775
    },
    {
      "epoch": 3.561066847321836,
      "grad_norm": 1.620352029800415,
      "learning_rate": 6.401618044600954e-06,
      "loss": 0.0555,
      "step": 241800
    },
    {
      "epoch": 3.561435030411923,
      "grad_norm": 1.500105381011963,
      "learning_rate": 6.39998167263397e-06,
      "loss": 0.0584,
      "step": 241825
    },
    {
      "epoch": 3.56180321350201,
      "grad_norm": 1.3465381860733032,
      "learning_rate": 6.398345300666986e-06,
      "loss": 0.0458,
      "step": 241850
    },
    {
      "epoch": 3.562171396592097,
      "grad_norm": 1.4730098247528076,
      "learning_rate": 6.396708928700002e-06,
      "loss": 0.0496,
      "step": 241875
    },
    {
      "epoch": 3.562539579682184,
      "grad_norm": 1.381256103515625,
      "learning_rate": 6.395072556733016e-06,
      "loss": 0.0484,
      "step": 241900
    },
    {
      "epoch": 3.562907762772271,
      "grad_norm": 1.3158462047576904,
      "learning_rate": 6.393436184766032e-06,
      "loss": 0.0548,
      "step": 241925
    },
    {
      "epoch": 3.563275945862358,
      "grad_norm": 0.9338003993034363,
      "learning_rate": 6.391799812799048e-06,
      "loss": 0.0539,
      "step": 241950
    },
    {
      "epoch": 3.563644128952445,
      "grad_norm": 1.442128300666809,
      "learning_rate": 6.390163440832063e-06,
      "loss": 0.0545,
      "step": 241975
    },
    {
      "epoch": 3.5640123120425327,
      "grad_norm": 1.8567023277282715,
      "learning_rate": 6.388527068865078e-06,
      "loss": 0.0612,
      "step": 242000
    },
    {
      "epoch": 3.5643804951326197,
      "grad_norm": 1.1950697898864746,
      "learning_rate": 6.386890696898094e-06,
      "loss": 0.0516,
      "step": 242025
    },
    {
      "epoch": 3.5647486782227067,
      "grad_norm": 0.8833608627319336,
      "learning_rate": 6.385254324931109e-06,
      "loss": 0.0533,
      "step": 242050
    },
    {
      "epoch": 3.5651168613127937,
      "grad_norm": 0.7630489468574524,
      "learning_rate": 6.383617952964125e-06,
      "loss": 0.0542,
      "step": 242075
    },
    {
      "epoch": 3.5654850444028807,
      "grad_norm": 1.2349568605422974,
      "learning_rate": 6.38198158099714e-06,
      "loss": 0.0558,
      "step": 242100
    },
    {
      "epoch": 3.5658532274929677,
      "grad_norm": 1.4984163045883179,
      "learning_rate": 6.380345209030156e-06,
      "loss": 0.0527,
      "step": 242125
    },
    {
      "epoch": 3.5662214105830548,
      "grad_norm": 1.307085633277893,
      "learning_rate": 6.378708837063171e-06,
      "loss": 0.0475,
      "step": 242150
    },
    {
      "epoch": 3.5665895936731418,
      "grad_norm": 1.2279508113861084,
      "learning_rate": 6.377072465096187e-06,
      "loss": 0.0536,
      "step": 242175
    },
    {
      "epoch": 3.566957776763229,
      "grad_norm": 0.9844478368759155,
      "learning_rate": 6.375436093129202e-06,
      "loss": 0.0559,
      "step": 242200
    },
    {
      "epoch": 3.567325959853316,
      "grad_norm": 1.0601154565811157,
      "learning_rate": 6.373799721162217e-06,
      "loss": 0.0533,
      "step": 242225
    },
    {
      "epoch": 3.567694142943403,
      "grad_norm": 0.9835296273231506,
      "learning_rate": 6.372163349195233e-06,
      "loss": 0.048,
      "step": 242250
    },
    {
      "epoch": 3.56806232603349,
      "grad_norm": 1.6406023502349854,
      "learning_rate": 6.3705269772282484e-06,
      "loss": 0.0538,
      "step": 242275
    },
    {
      "epoch": 3.568430509123577,
      "grad_norm": 1.235166072845459,
      "learning_rate": 6.368890605261263e-06,
      "loss": 0.0505,
      "step": 242300
    },
    {
      "epoch": 3.568798692213664,
      "grad_norm": 1.4901325702667236,
      "learning_rate": 6.367254233294279e-06,
      "loss": 0.0518,
      "step": 242325
    },
    {
      "epoch": 3.5691668753037513,
      "grad_norm": 1.6954319477081299,
      "learning_rate": 6.365617861327294e-06,
      "loss": 0.0569,
      "step": 242350
    },
    {
      "epoch": 3.5695350583938383,
      "grad_norm": 1.4798673391342163,
      "learning_rate": 6.3639814893603105e-06,
      "loss": 0.0536,
      "step": 242375
    },
    {
      "epoch": 3.5699032414839253,
      "grad_norm": 1.34373939037323,
      "learning_rate": 6.362345117393326e-06,
      "loss": 0.0473,
      "step": 242400
    },
    {
      "epoch": 3.5702714245740124,
      "grad_norm": 1.538964033126831,
      "learning_rate": 6.36070874542634e-06,
      "loss": 0.0544,
      "step": 242425
    },
    {
      "epoch": 3.5706396076640994,
      "grad_norm": 1.8510009050369263,
      "learning_rate": 6.359072373459356e-06,
      "loss": 0.0558,
      "step": 242450
    },
    {
      "epoch": 3.5710077907541864,
      "grad_norm": 1.198325514793396,
      "learning_rate": 6.357436001492372e-06,
      "loss": 0.0568,
      "step": 242475
    },
    {
      "epoch": 3.5713759738442734,
      "grad_norm": 0.910335123538971,
      "learning_rate": 6.355799629525388e-06,
      "loss": 0.0642,
      "step": 242500
    },
    {
      "epoch": 3.5717441569343604,
      "grad_norm": 1.613919734954834,
      "learning_rate": 6.354163257558402e-06,
      "loss": 0.0513,
      "step": 242525
    },
    {
      "epoch": 3.5721123400244474,
      "grad_norm": 1.2130235433578491,
      "learning_rate": 6.352526885591418e-06,
      "loss": 0.0568,
      "step": 242550
    },
    {
      "epoch": 3.5724805231145345,
      "grad_norm": 1.123764157295227,
      "learning_rate": 6.350890513624434e-06,
      "loss": 0.0543,
      "step": 242575
    },
    {
      "epoch": 3.5728487062046215,
      "grad_norm": 1.3343775272369385,
      "learning_rate": 6.349254141657449e-06,
      "loss": 0.0518,
      "step": 242600
    },
    {
      "epoch": 3.5732168892947085,
      "grad_norm": 1.30911386013031,
      "learning_rate": 6.3476177696904636e-06,
      "loss": 0.0515,
      "step": 242625
    },
    {
      "epoch": 3.5735850723847955,
      "grad_norm": 1.0751773118972778,
      "learning_rate": 6.34598139772348e-06,
      "loss": 0.0546,
      "step": 242650
    },
    {
      "epoch": 3.5739532554748825,
      "grad_norm": 0.7018923163414001,
      "learning_rate": 6.344345025756495e-06,
      "loss": 0.0563,
      "step": 242675
    },
    {
      "epoch": 3.5743214385649695,
      "grad_norm": 1.2422815561294556,
      "learning_rate": 6.342708653789511e-06,
      "loss": 0.0601,
      "step": 242700
    },
    {
      "epoch": 3.5746896216550565,
      "grad_norm": 1.1047272682189941,
      "learning_rate": 6.341072281822526e-06,
      "loss": 0.058,
      "step": 242725
    },
    {
      "epoch": 3.5750578047451436,
      "grad_norm": 1.252564787864685,
      "learning_rate": 6.339435909855542e-06,
      "loss": 0.0471,
      "step": 242750
    },
    {
      "epoch": 3.5754259878352306,
      "grad_norm": 1.2016949653625488,
      "learning_rate": 6.337799537888557e-06,
      "loss": 0.0506,
      "step": 242775
    },
    {
      "epoch": 3.5757941709253176,
      "grad_norm": 1.1374620199203491,
      "learning_rate": 6.336163165921572e-06,
      "loss": 0.0541,
      "step": 242800
    },
    {
      "epoch": 3.5761623540154046,
      "grad_norm": 1.4337965250015259,
      "learning_rate": 6.3345267939545885e-06,
      "loss": 0.0584,
      "step": 242825
    },
    {
      "epoch": 3.5765305371054916,
      "grad_norm": 1.2424113750457764,
      "learning_rate": 6.332890421987603e-06,
      "loss": 0.0495,
      "step": 242850
    },
    {
      "epoch": 3.5768987201955786,
      "grad_norm": 1.4434897899627686,
      "learning_rate": 6.331254050020619e-06,
      "loss": 0.0499,
      "step": 242875
    },
    {
      "epoch": 3.5772669032856657,
      "grad_norm": 1.5763555765151978,
      "learning_rate": 6.3296176780536344e-06,
      "loss": 0.0548,
      "step": 242900
    },
    {
      "epoch": 3.5776350863757527,
      "grad_norm": 1.4369443655014038,
      "learning_rate": 6.327981306086651e-06,
      "loss": 0.0551,
      "step": 242925
    },
    {
      "epoch": 3.57800326946584,
      "grad_norm": 1.4109026193618774,
      "learning_rate": 6.326344934119665e-06,
      "loss": 0.0537,
      "step": 242950
    },
    {
      "epoch": 3.578371452555927,
      "grad_norm": 1.2654056549072266,
      "learning_rate": 6.32470856215268e-06,
      "loss": 0.053,
      "step": 242975
    },
    {
      "epoch": 3.578739635646014,
      "grad_norm": 2.049483060836792,
      "learning_rate": 6.3230721901856965e-06,
      "loss": 0.0615,
      "step": 243000
    },
    {
      "epoch": 3.579107818736101,
      "grad_norm": 1.703001618385315,
      "learning_rate": 6.321435818218712e-06,
      "loss": 0.0514,
      "step": 243025
    },
    {
      "epoch": 3.579476001826188,
      "grad_norm": 1.5279680490493774,
      "learning_rate": 6.319799446251726e-06,
      "loss": 0.0509,
      "step": 243050
    },
    {
      "epoch": 3.579844184916275,
      "grad_norm": 1.4448721408843994,
      "learning_rate": 6.318163074284742e-06,
      "loss": 0.0556,
      "step": 243075
    },
    {
      "epoch": 3.5802123680063622,
      "grad_norm": 1.5780704021453857,
      "learning_rate": 6.316526702317758e-06,
      "loss": 0.0513,
      "step": 243100
    },
    {
      "epoch": 3.5805805510964492,
      "grad_norm": 1.4930514097213745,
      "learning_rate": 6.314890330350774e-06,
      "loss": 0.0562,
      "step": 243125
    },
    {
      "epoch": 3.5809487341865363,
      "grad_norm": 1.1155985593795776,
      "learning_rate": 6.313253958383788e-06,
      "loss": 0.0503,
      "step": 243150
    },
    {
      "epoch": 3.5813169172766233,
      "grad_norm": 1.4989086389541626,
      "learning_rate": 6.311617586416804e-06,
      "loss": 0.0572,
      "step": 243175
    },
    {
      "epoch": 3.5816851003667103,
      "grad_norm": 1.3143268823623657,
      "learning_rate": 6.30998121444982e-06,
      "loss": 0.058,
      "step": 243200
    },
    {
      "epoch": 3.5820532834567973,
      "grad_norm": 1.2218605279922485,
      "learning_rate": 6.308344842482835e-06,
      "loss": 0.0512,
      "step": 243225
    },
    {
      "epoch": 3.5824214665468843,
      "grad_norm": 1.347113847732544,
      "learning_rate": 6.3067084705158495e-06,
      "loss": 0.0541,
      "step": 243250
    },
    {
      "epoch": 3.5827896496369713,
      "grad_norm": 1.5165753364562988,
      "learning_rate": 6.305072098548866e-06,
      "loss": 0.0612,
      "step": 243275
    },
    {
      "epoch": 3.583157832727059,
      "grad_norm": 0.5842728614807129,
      "learning_rate": 6.303435726581881e-06,
      "loss": 0.0507,
      "step": 243300
    },
    {
      "epoch": 3.583526015817146,
      "grad_norm": 1.269214153289795,
      "learning_rate": 6.301799354614897e-06,
      "loss": 0.0624,
      "step": 243325
    },
    {
      "epoch": 3.583894198907233,
      "grad_norm": 1.4616782665252686,
      "learning_rate": 6.3001629826479125e-06,
      "loss": 0.0595,
      "step": 243350
    },
    {
      "epoch": 3.58426238199732,
      "grad_norm": 1.0306979417800903,
      "learning_rate": 6.298526610680927e-06,
      "loss": 0.0515,
      "step": 243375
    },
    {
      "epoch": 3.584630565087407,
      "grad_norm": 1.4185216426849365,
      "learning_rate": 6.296890238713943e-06,
      "loss": 0.054,
      "step": 243400
    },
    {
      "epoch": 3.584998748177494,
      "grad_norm": 1.3975141048431396,
      "learning_rate": 6.295253866746958e-06,
      "loss": 0.0582,
      "step": 243425
    },
    {
      "epoch": 3.585366931267581,
      "grad_norm": 1.7203360795974731,
      "learning_rate": 6.2936174947799745e-06,
      "loss": 0.0595,
      "step": 243450
    },
    {
      "epoch": 3.585735114357668,
      "grad_norm": 1.608855128288269,
      "learning_rate": 6.291981122812989e-06,
      "loss": 0.0557,
      "step": 243475
    },
    {
      "epoch": 3.586103297447755,
      "grad_norm": 1.4659583568572998,
      "learning_rate": 6.290344750846004e-06,
      "loss": 0.0546,
      "step": 243500
    },
    {
      "epoch": 3.586471480537842,
      "grad_norm": 1.4480702877044678,
      "learning_rate": 6.2887083788790204e-06,
      "loss": 0.0498,
      "step": 243525
    },
    {
      "epoch": 3.586839663627929,
      "grad_norm": 1.438968300819397,
      "learning_rate": 6.287072006912036e-06,
      "loss": 0.0577,
      "step": 243550
    },
    {
      "epoch": 3.587207846718016,
      "grad_norm": 0.9120373725891113,
      "learning_rate": 6.285435634945051e-06,
      "loss": 0.0511,
      "step": 243575
    },
    {
      "epoch": 3.587576029808103,
      "grad_norm": 1.5426757335662842,
      "learning_rate": 6.283799262978066e-06,
      "loss": 0.0555,
      "step": 243600
    },
    {
      "epoch": 3.58794421289819,
      "grad_norm": 1.5347541570663452,
      "learning_rate": 6.2821628910110825e-06,
      "loss": 0.0557,
      "step": 243625
    },
    {
      "epoch": 3.588312395988277,
      "grad_norm": 1.6872762441635132,
      "learning_rate": 6.280526519044098e-06,
      "loss": 0.0566,
      "step": 243650
    },
    {
      "epoch": 3.588680579078364,
      "grad_norm": 1.684535264968872,
      "learning_rate": 6.278890147077112e-06,
      "loss": 0.0583,
      "step": 243675
    },
    {
      "epoch": 3.589048762168451,
      "grad_norm": 1.1899725198745728,
      "learning_rate": 6.277253775110128e-06,
      "loss": 0.0578,
      "step": 243700
    },
    {
      "epoch": 3.589416945258538,
      "grad_norm": 1.234108328819275,
      "learning_rate": 6.275617403143144e-06,
      "loss": 0.0563,
      "step": 243725
    },
    {
      "epoch": 3.589785128348625,
      "grad_norm": 0.7505812644958496,
      "learning_rate": 6.27398103117616e-06,
      "loss": 0.053,
      "step": 243750
    },
    {
      "epoch": 3.590153311438712,
      "grad_norm": 1.1787515878677368,
      "learning_rate": 6.272344659209174e-06,
      "loss": 0.0551,
      "step": 243775
    },
    {
      "epoch": 3.590521494528799,
      "grad_norm": 1.2294164896011353,
      "learning_rate": 6.27070828724219e-06,
      "loss": 0.0651,
      "step": 243800
    },
    {
      "epoch": 3.590889677618886,
      "grad_norm": 1.1803584098815918,
      "learning_rate": 6.269071915275206e-06,
      "loss": 0.0573,
      "step": 243825
    },
    {
      "epoch": 3.591257860708973,
      "grad_norm": 0.7361862659454346,
      "learning_rate": 6.267435543308221e-06,
      "loss": 0.0509,
      "step": 243850
    },
    {
      "epoch": 3.59162604379906,
      "grad_norm": 1.3564823865890503,
      "learning_rate": 6.265799171341237e-06,
      "loss": 0.0492,
      "step": 243875
    },
    {
      "epoch": 3.5919942268891476,
      "grad_norm": 1.2647794485092163,
      "learning_rate": 6.264162799374252e-06,
      "loss": 0.0555,
      "step": 243900
    },
    {
      "epoch": 3.5923624099792346,
      "grad_norm": 1.1084637641906738,
      "learning_rate": 6.262526427407267e-06,
      "loss": 0.0526,
      "step": 243925
    },
    {
      "epoch": 3.5927305930693216,
      "grad_norm": 0.8604913949966431,
      "learning_rate": 6.260890055440283e-06,
      "loss": 0.0536,
      "step": 243950
    },
    {
      "epoch": 3.5930987761594086,
      "grad_norm": 1.1622568368911743,
      "learning_rate": 6.2592536834732985e-06,
      "loss": 0.0588,
      "step": 243975
    },
    {
      "epoch": 3.5934669592494957,
      "grad_norm": 1.5048962831497192,
      "learning_rate": 6.257617311506313e-06,
      "loss": 0.0527,
      "step": 244000
    },
    {
      "epoch": 3.5938351423395827,
      "grad_norm": 1.2005444765090942,
      "learning_rate": 6.255980939539329e-06,
      "loss": 0.0582,
      "step": 244025
    },
    {
      "epoch": 3.5942033254296697,
      "grad_norm": 1.516257643699646,
      "learning_rate": 6.254344567572344e-06,
      "loss": 0.055,
      "step": 244050
    },
    {
      "epoch": 3.5945715085197567,
      "grad_norm": 1.3147369623184204,
      "learning_rate": 6.2527081956053605e-06,
      "loss": 0.0648,
      "step": 244075
    },
    {
      "epoch": 3.5949396916098437,
      "grad_norm": 1.078008770942688,
      "learning_rate": 6.251071823638375e-06,
      "loss": 0.0515,
      "step": 244100
    },
    {
      "epoch": 3.5953078746999307,
      "grad_norm": 1.3717511892318726,
      "learning_rate": 6.24943545167139e-06,
      "loss": 0.0538,
      "step": 244125
    },
    {
      "epoch": 3.5956760577900178,
      "grad_norm": 0.8229482173919678,
      "learning_rate": 6.2477990797044064e-06,
      "loss": 0.052,
      "step": 244150
    },
    {
      "epoch": 3.5960442408801048,
      "grad_norm": 0.9627171158790588,
      "learning_rate": 6.246162707737422e-06,
      "loss": 0.0505,
      "step": 244175
    },
    {
      "epoch": 3.596412423970192,
      "grad_norm": 1.3391168117523193,
      "learning_rate": 6.244526335770437e-06,
      "loss": 0.0507,
      "step": 244200
    },
    {
      "epoch": 3.596780607060279,
      "grad_norm": 1.6255720853805542,
      "learning_rate": 6.242889963803452e-06,
      "loss": 0.0504,
      "step": 244225
    },
    {
      "epoch": 3.597148790150366,
      "grad_norm": 1.5499217510223389,
      "learning_rate": 6.241253591836468e-06,
      "loss": 0.0518,
      "step": 244250
    },
    {
      "epoch": 3.5975169732404533,
      "grad_norm": 1.371516466140747,
      "learning_rate": 6.239617219869484e-06,
      "loss": 0.0507,
      "step": 244275
    },
    {
      "epoch": 3.5978851563305403,
      "grad_norm": 1.1114004850387573,
      "learning_rate": 6.237980847902499e-06,
      "loss": 0.0507,
      "step": 244300
    },
    {
      "epoch": 3.5982533394206273,
      "grad_norm": 1.0851997137069702,
      "learning_rate": 6.236344475935514e-06,
      "loss": 0.0547,
      "step": 244325
    },
    {
      "epoch": 3.5986215225107143,
      "grad_norm": 1.604949951171875,
      "learning_rate": 6.23470810396853e-06,
      "loss": 0.0476,
      "step": 244350
    },
    {
      "epoch": 3.5989897056008013,
      "grad_norm": 0.6757989525794983,
      "learning_rate": 6.233071732001546e-06,
      "loss": 0.0556,
      "step": 244375
    },
    {
      "epoch": 3.5993578886908884,
      "grad_norm": 1.1998815536499023,
      "learning_rate": 6.231435360034561e-06,
      "loss": 0.062,
      "step": 244400
    },
    {
      "epoch": 3.5997260717809754,
      "grad_norm": 1.7735071182250977,
      "learning_rate": 6.229798988067576e-06,
      "loss": 0.0549,
      "step": 244425
    },
    {
      "epoch": 3.6000942548710624,
      "grad_norm": 1.3361785411834717,
      "learning_rate": 6.228162616100592e-06,
      "loss": 0.0549,
      "step": 244450
    },
    {
      "epoch": 3.6004624379611494,
      "grad_norm": 1.4690197706222534,
      "learning_rate": 6.226526244133607e-06,
      "loss": 0.0552,
      "step": 244475
    },
    {
      "epoch": 3.6008306210512364,
      "grad_norm": 2.2308056354522705,
      "learning_rate": 6.224889872166623e-06,
      "loss": 0.0618,
      "step": 244500
    },
    {
      "epoch": 3.6011988041413234,
      "grad_norm": 1.2314001321792603,
      "learning_rate": 6.223253500199638e-06,
      "loss": 0.0474,
      "step": 244525
    },
    {
      "epoch": 3.6015669872314104,
      "grad_norm": 1.5631963014602661,
      "learning_rate": 6.221617128232653e-06,
      "loss": 0.051,
      "step": 244550
    },
    {
      "epoch": 3.6019351703214975,
      "grad_norm": 1.363520622253418,
      "learning_rate": 6.219980756265669e-06,
      "loss": 0.0515,
      "step": 244575
    },
    {
      "epoch": 3.6023033534115845,
      "grad_norm": 1.4217227697372437,
      "learning_rate": 6.2183443842986845e-06,
      "loss": 0.0512,
      "step": 244600
    },
    {
      "epoch": 3.6026715365016715,
      "grad_norm": 1.4224117994308472,
      "learning_rate": 6.216708012331699e-06,
      "loss": 0.0554,
      "step": 244625
    },
    {
      "epoch": 3.6030397195917585,
      "grad_norm": 1.8205037117004395,
      "learning_rate": 6.215071640364715e-06,
      "loss": 0.0527,
      "step": 244650
    },
    {
      "epoch": 3.6034079026818455,
      "grad_norm": 1.1793156862258911,
      "learning_rate": 6.21343526839773e-06,
      "loss": 0.0509,
      "step": 244675
    },
    {
      "epoch": 3.6037760857719325,
      "grad_norm": 1.0544114112854004,
      "learning_rate": 6.2117988964307465e-06,
      "loss": 0.0465,
      "step": 244700
    },
    {
      "epoch": 3.6041442688620196,
      "grad_norm": 0.9881449937820435,
      "learning_rate": 6.21022797934244e-06,
      "loss": 0.0504,
      "step": 244725
    },
    {
      "epoch": 3.6045124519521066,
      "grad_norm": 0.9702735543251038,
      "learning_rate": 6.208591607375456e-06,
      "loss": 0.0525,
      "step": 244750
    },
    {
      "epoch": 3.6048806350421936,
      "grad_norm": 1.267856478691101,
      "learning_rate": 6.206955235408472e-06,
      "loss": 0.053,
      "step": 244775
    },
    {
      "epoch": 3.6052488181322806,
      "grad_norm": 1.4718348979949951,
      "learning_rate": 6.205318863441487e-06,
      "loss": 0.0584,
      "step": 244800
    },
    {
      "epoch": 3.6056170012223676,
      "grad_norm": 1.0116479396820068,
      "learning_rate": 6.203682491474503e-06,
      "loss": 0.0499,
      "step": 244825
    },
    {
      "epoch": 3.6059851843124546,
      "grad_norm": 1.7999377250671387,
      "learning_rate": 6.202046119507518e-06,
      "loss": 0.0543,
      "step": 244850
    },
    {
      "epoch": 3.606353367402542,
      "grad_norm": 1.6597347259521484,
      "learning_rate": 6.200409747540533e-06,
      "loss": 0.0597,
      "step": 244875
    },
    {
      "epoch": 3.606721550492629,
      "grad_norm": 0.996076762676239,
      "learning_rate": 6.198773375573549e-06,
      "loss": 0.06,
      "step": 244900
    },
    {
      "epoch": 3.607089733582716,
      "grad_norm": 0.8949547410011292,
      "learning_rate": 6.1971370036065645e-06,
      "loss": 0.048,
      "step": 244925
    },
    {
      "epoch": 3.607457916672803,
      "grad_norm": 1.2150899171829224,
      "learning_rate": 6.195500631639579e-06,
      "loss": 0.0494,
      "step": 244950
    },
    {
      "epoch": 3.60782609976289,
      "grad_norm": 1.2176042795181274,
      "learning_rate": 6.193864259672595e-06,
      "loss": 0.052,
      "step": 244975
    },
    {
      "epoch": 3.608194282852977,
      "grad_norm": 1.6427092552185059,
      "learning_rate": 6.1922278877056104e-06,
      "loss": 0.0637,
      "step": 245000
    },
    {
      "epoch": 3.608562465943064,
      "grad_norm": 1.6079386472702026,
      "learning_rate": 6.190591515738627e-06,
      "loss": 0.0515,
      "step": 245025
    },
    {
      "epoch": 3.608930649033151,
      "grad_norm": 1.4702075719833374,
      "learning_rate": 6.188955143771641e-06,
      "loss": 0.0504,
      "step": 245050
    },
    {
      "epoch": 3.609298832123238,
      "grad_norm": 1.0616623163223267,
      "learning_rate": 6.187318771804657e-06,
      "loss": 0.0564,
      "step": 245075
    },
    {
      "epoch": 3.6096670152133252,
      "grad_norm": 1.1923943758010864,
      "learning_rate": 6.1856823998376725e-06,
      "loss": 0.0555,
      "step": 245100
    },
    {
      "epoch": 3.6100351983034122,
      "grad_norm": 1.1287178993225098,
      "learning_rate": 6.184046027870688e-06,
      "loss": 0.0524,
      "step": 245125
    },
    {
      "epoch": 3.6104033813934993,
      "grad_norm": 1.0990760326385498,
      "learning_rate": 6.182409655903703e-06,
      "loss": 0.0599,
      "step": 245150
    },
    {
      "epoch": 3.6107715644835863,
      "grad_norm": 1.4486708641052246,
      "learning_rate": 6.1807732839367184e-06,
      "loss": 0.0547,
      "step": 245175
    },
    {
      "epoch": 3.6111397475736733,
      "grad_norm": 1.107300877571106,
      "learning_rate": 6.1791369119697346e-06,
      "loss": 0.053,
      "step": 245200
    },
    {
      "epoch": 3.6115079306637607,
      "grad_norm": 1.3102949857711792,
      "learning_rate": 6.17750054000275e-06,
      "loss": 0.0583,
      "step": 245225
    },
    {
      "epoch": 3.6118761137538478,
      "grad_norm": 1.5146225690841675,
      "learning_rate": 6.175864168035765e-06,
      "loss": 0.0561,
      "step": 245250
    },
    {
      "epoch": 3.6122442968439348,
      "grad_norm": 1.256121039390564,
      "learning_rate": 6.1742277960687805e-06,
      "loss": 0.0608,
      "step": 245275
    },
    {
      "epoch": 3.612612479934022,
      "grad_norm": 1.4065275192260742,
      "learning_rate": 6.172591424101796e-06,
      "loss": 0.0639,
      "step": 245300
    },
    {
      "epoch": 3.612980663024109,
      "grad_norm": 1.9923678636550903,
      "learning_rate": 6.170955052134812e-06,
      "loss": 0.045,
      "step": 245325
    },
    {
      "epoch": 3.613348846114196,
      "grad_norm": 1.399849534034729,
      "learning_rate": 6.169318680167827e-06,
      "loss": 0.0507,
      "step": 245350
    },
    {
      "epoch": 3.613717029204283,
      "grad_norm": 1.5250312089920044,
      "learning_rate": 6.167682308200842e-06,
      "loss": 0.0517,
      "step": 245375
    },
    {
      "epoch": 3.61408521229437,
      "grad_norm": 1.3271722793579102,
      "learning_rate": 6.166045936233858e-06,
      "loss": 0.0508,
      "step": 245400
    },
    {
      "epoch": 3.614453395384457,
      "grad_norm": 1.3831508159637451,
      "learning_rate": 6.164409564266873e-06,
      "loss": 0.0506,
      "step": 245425
    },
    {
      "epoch": 3.614821578474544,
      "grad_norm": 1.142835021018982,
      "learning_rate": 6.162773192299889e-06,
      "loss": 0.0519,
      "step": 245450
    },
    {
      "epoch": 3.615189761564631,
      "grad_norm": 1.3392393589019775,
      "learning_rate": 6.161136820332904e-06,
      "loss": 0.0499,
      "step": 245475
    },
    {
      "epoch": 3.615557944654718,
      "grad_norm": 1.1853067874908447,
      "learning_rate": 6.159500448365919e-06,
      "loss": 0.0528,
      "step": 245500
    },
    {
      "epoch": 3.615926127744805,
      "grad_norm": 1.3341588973999023,
      "learning_rate": 6.157864076398935e-06,
      "loss": 0.0523,
      "step": 245525
    },
    {
      "epoch": 3.616294310834892,
      "grad_norm": 1.495862603187561,
      "learning_rate": 6.1562277044319505e-06,
      "loss": 0.0533,
      "step": 245550
    },
    {
      "epoch": 3.616662493924979,
      "grad_norm": 1.5366014242172241,
      "learning_rate": 6.154591332464965e-06,
      "loss": 0.0498,
      "step": 245575
    },
    {
      "epoch": 3.617030677015066,
      "grad_norm": 1.1389579772949219,
      "learning_rate": 6.152954960497981e-06,
      "loss": 0.0542,
      "step": 245600
    },
    {
      "epoch": 3.617398860105153,
      "grad_norm": 0.932179868221283,
      "learning_rate": 6.1513185885309964e-06,
      "loss": 0.0535,
      "step": 245625
    },
    {
      "epoch": 3.61776704319524,
      "grad_norm": 1.5072063207626343,
      "learning_rate": 6.149682216564013e-06,
      "loss": 0.0537,
      "step": 245650
    },
    {
      "epoch": 3.618135226285327,
      "grad_norm": 1.2622642517089844,
      "learning_rate": 6.148045844597027e-06,
      "loss": 0.049,
      "step": 245675
    },
    {
      "epoch": 3.618503409375414,
      "grad_norm": 1.2886232137680054,
      "learning_rate": 6.146409472630042e-06,
      "loss": 0.06,
      "step": 245700
    },
    {
      "epoch": 3.618871592465501,
      "grad_norm": 2.2847864627838135,
      "learning_rate": 6.1447731006630585e-06,
      "loss": 0.0578,
      "step": 245725
    },
    {
      "epoch": 3.619239775555588,
      "grad_norm": 0.6749784350395203,
      "learning_rate": 6.143136728696074e-06,
      "loss": 0.0515,
      "step": 245750
    },
    {
      "epoch": 3.619607958645675,
      "grad_norm": 1.4920670986175537,
      "learning_rate": 6.14150035672909e-06,
      "loss": 0.0611,
      "step": 245775
    },
    {
      "epoch": 3.619976141735762,
      "grad_norm": 1.5810109376907349,
      "learning_rate": 6.139863984762104e-06,
      "loss": 0.0558,
      "step": 245800
    },
    {
      "epoch": 3.6203443248258496,
      "grad_norm": 1.413640022277832,
      "learning_rate": 6.13822761279512e-06,
      "loss": 0.0578,
      "step": 245825
    },
    {
      "epoch": 3.6207125079159366,
      "grad_norm": 0.9894798398017883,
      "learning_rate": 6.136591240828136e-06,
      "loss": 0.0493,
      "step": 245850
    },
    {
      "epoch": 3.6210806910060236,
      "grad_norm": 1.2460707426071167,
      "learning_rate": 6.134954868861151e-06,
      "loss": 0.0585,
      "step": 245875
    },
    {
      "epoch": 3.6214488740961106,
      "grad_norm": 1.248345136642456,
      "learning_rate": 6.1333184968941665e-06,
      "loss": 0.0483,
      "step": 245900
    },
    {
      "epoch": 3.6218170571861976,
      "grad_norm": 1.5993595123291016,
      "learning_rate": 6.131682124927182e-06,
      "loss": 0.0524,
      "step": 245925
    },
    {
      "epoch": 3.6221852402762846,
      "grad_norm": 1.0702323913574219,
      "learning_rate": 6.130045752960198e-06,
      "loss": 0.0457,
      "step": 245950
    },
    {
      "epoch": 3.6225534233663716,
      "grad_norm": 1.7296591997146606,
      "learning_rate": 6.128409380993213e-06,
      "loss": 0.0596,
      "step": 245975
    },
    {
      "epoch": 3.6229216064564587,
      "grad_norm": 1.5586003065109253,
      "learning_rate": 6.126773009026228e-06,
      "loss": 0.05,
      "step": 246000
    },
    {
      "epoch": 3.6232897895465457,
      "grad_norm": 1.4900046586990356,
      "learning_rate": 6.125136637059244e-06,
      "loss": 0.0494,
      "step": 246025
    },
    {
      "epoch": 3.6236579726366327,
      "grad_norm": 1.5979666709899902,
      "learning_rate": 6.123500265092259e-06,
      "loss": 0.0533,
      "step": 246050
    },
    {
      "epoch": 3.6240261557267197,
      "grad_norm": 1.6462072134017944,
      "learning_rate": 6.121863893125275e-06,
      "loss": 0.0519,
      "step": 246075
    },
    {
      "epoch": 3.6243943388168067,
      "grad_norm": 1.8660732507705688,
      "learning_rate": 6.12022752115829e-06,
      "loss": 0.0535,
      "step": 246100
    },
    {
      "epoch": 3.6247625219068937,
      "grad_norm": 1.1489410400390625,
      "learning_rate": 6.118591149191305e-06,
      "loss": 0.0509,
      "step": 246125
    },
    {
      "epoch": 3.6251307049969808,
      "grad_norm": 1.7616811990737915,
      "learning_rate": 6.116954777224321e-06,
      "loss": 0.0587,
      "step": 246150
    },
    {
      "epoch": 3.6254988880870678,
      "grad_norm": 1.6229524612426758,
      "learning_rate": 6.1153184052573365e-06,
      "loss": 0.0493,
      "step": 246175
    },
    {
      "epoch": 3.6258670711771552,
      "grad_norm": 1.388395071029663,
      "learning_rate": 6.113682033290351e-06,
      "loss": 0.0532,
      "step": 246200
    },
    {
      "epoch": 3.6262352542672422,
      "grad_norm": 1.6239497661590576,
      "learning_rate": 6.112045661323367e-06,
      "loss": 0.0581,
      "step": 246225
    },
    {
      "epoch": 3.6266034373573293,
      "grad_norm": 0.9362977147102356,
      "learning_rate": 6.1104092893563824e-06,
      "loss": 0.0495,
      "step": 246250
    },
    {
      "epoch": 3.6269716204474163,
      "grad_norm": 1.5556639432907104,
      "learning_rate": 6.108772917389399e-06,
      "loss": 0.0554,
      "step": 246275
    },
    {
      "epoch": 3.6273398035375033,
      "grad_norm": 1.2726385593414307,
      "learning_rate": 6.107136545422414e-06,
      "loss": 0.0502,
      "step": 246300
    },
    {
      "epoch": 3.6277079866275903,
      "grad_norm": 1.9834768772125244,
      "learning_rate": 6.105500173455428e-06,
      "loss": 0.0531,
      "step": 246325
    },
    {
      "epoch": 3.6280761697176773,
      "grad_norm": 1.2573577165603638,
      "learning_rate": 6.1038638014884445e-06,
      "loss": 0.0545,
      "step": 246350
    },
    {
      "epoch": 3.6284443528077643,
      "grad_norm": 1.6854448318481445,
      "learning_rate": 6.10222742952146e-06,
      "loss": 0.057,
      "step": 246375
    },
    {
      "epoch": 3.6288125358978514,
      "grad_norm": 1.40595281124115,
      "learning_rate": 6.100591057554476e-06,
      "loss": 0.0517,
      "step": 246400
    },
    {
      "epoch": 3.6291807189879384,
      "grad_norm": 1.1282232999801636,
      "learning_rate": 6.09895468558749e-06,
      "loss": 0.0577,
      "step": 246425
    },
    {
      "epoch": 3.6295489020780254,
      "grad_norm": 0.7684329152107239,
      "learning_rate": 6.097318313620506e-06,
      "loss": 0.0479,
      "step": 246450
    },
    {
      "epoch": 3.6299170851681124,
      "grad_norm": 0.6485980153083801,
      "learning_rate": 6.095681941653522e-06,
      "loss": 0.0499,
      "step": 246475
    },
    {
      "epoch": 3.6302852682581994,
      "grad_norm": 1.4307173490524292,
      "learning_rate": 6.094045569686537e-06,
      "loss": 0.0562,
      "step": 246500
    },
    {
      "epoch": 3.6306534513482864,
      "grad_norm": 1.3257659673690796,
      "learning_rate": 6.0924091977195525e-06,
      "loss": 0.0591,
      "step": 246525
    },
    {
      "epoch": 3.6310216344383734,
      "grad_norm": 1.5286110639572144,
      "learning_rate": 6.090772825752568e-06,
      "loss": 0.0469,
      "step": 246550
    },
    {
      "epoch": 3.6313898175284605,
      "grad_norm": 0.8919461369514465,
      "learning_rate": 6.089136453785583e-06,
      "loss": 0.055,
      "step": 246575
    },
    {
      "epoch": 3.6317580006185475,
      "grad_norm": 1.9157183170318604,
      "learning_rate": 6.087500081818599e-06,
      "loss": 0.0557,
      "step": 246600
    },
    {
      "epoch": 3.6321261837086345,
      "grad_norm": 1.1851670742034912,
      "learning_rate": 6.085863709851614e-06,
      "loss": 0.0606,
      "step": 246625
    },
    {
      "epoch": 3.6324943667987215,
      "grad_norm": 1.8410229682922363,
      "learning_rate": 6.08422733788463e-06,
      "loss": 0.0544,
      "step": 246650
    },
    {
      "epoch": 3.6328625498888085,
      "grad_norm": 1.2160470485687256,
      "learning_rate": 6.082590965917645e-06,
      "loss": 0.0518,
      "step": 246675
    },
    {
      "epoch": 3.6332307329788955,
      "grad_norm": 1.572349190711975,
      "learning_rate": 6.080954593950661e-06,
      "loss": 0.0565,
      "step": 246700
    },
    {
      "epoch": 3.6335989160689826,
      "grad_norm": 1.3269578218460083,
      "learning_rate": 6.079318221983677e-06,
      "loss": 0.0541,
      "step": 246725
    },
    {
      "epoch": 3.6339670991590696,
      "grad_norm": 0.7197899222373962,
      "learning_rate": 6.077681850016691e-06,
      "loss": 0.0494,
      "step": 246750
    },
    {
      "epoch": 3.6343352822491566,
      "grad_norm": 1.1615010499954224,
      "learning_rate": 6.076045478049707e-06,
      "loss": 0.0591,
      "step": 246775
    },
    {
      "epoch": 3.634703465339244,
      "grad_norm": 0.9506219625473022,
      "learning_rate": 6.0744091060827225e-06,
      "loss": 0.0603,
      "step": 246800
    },
    {
      "epoch": 3.635071648429331,
      "grad_norm": 1.126420021057129,
      "learning_rate": 6.072772734115739e-06,
      "loss": 0.0579,
      "step": 246825
    },
    {
      "epoch": 3.635439831519418,
      "grad_norm": 1.2277495861053467,
      "learning_rate": 6.071136362148753e-06,
      "loss": 0.0623,
      "step": 246850
    },
    {
      "epoch": 3.635808014609505,
      "grad_norm": 1.1536575555801392,
      "learning_rate": 6.0694999901817684e-06,
      "loss": 0.0529,
      "step": 246875
    },
    {
      "epoch": 3.636176197699592,
      "grad_norm": 1.2662118673324585,
      "learning_rate": 6.067863618214785e-06,
      "loss": 0.0538,
      "step": 246900
    },
    {
      "epoch": 3.636544380789679,
      "grad_norm": 1.493308186531067,
      "learning_rate": 6.0662272462478e-06,
      "loss": 0.047,
      "step": 246925
    },
    {
      "epoch": 3.636912563879766,
      "grad_norm": 2.351929187774658,
      "learning_rate": 6.064590874280814e-06,
      "loss": 0.057,
      "step": 246950
    },
    {
      "epoch": 3.637280746969853,
      "grad_norm": 1.4156646728515625,
      "learning_rate": 6.0629545023138305e-06,
      "loss": 0.0573,
      "step": 246975
    },
    {
      "epoch": 3.63764893005994,
      "grad_norm": 1.434942364692688,
      "learning_rate": 6.061318130346846e-06,
      "loss": 0.0522,
      "step": 247000
    },
    {
      "epoch": 3.638017113150027,
      "grad_norm": 1.510393738746643,
      "learning_rate": 6.059681758379862e-06,
      "loss": 0.0662,
      "step": 247025
    },
    {
      "epoch": 3.638385296240114,
      "grad_norm": 1.399326205253601,
      "learning_rate": 6.058045386412876e-06,
      "loss": 0.0522,
      "step": 247050
    },
    {
      "epoch": 3.638753479330201,
      "grad_norm": 1.0100810527801514,
      "learning_rate": 6.056409014445892e-06,
      "loss": 0.0481,
      "step": 247075
    },
    {
      "epoch": 3.6391216624202882,
      "grad_norm": 1.1176669597625732,
      "learning_rate": 6.054772642478908e-06,
      "loss": 0.054,
      "step": 247100
    },
    {
      "epoch": 3.6394898455103752,
      "grad_norm": 1.309601902961731,
      "learning_rate": 6.053136270511923e-06,
      "loss": 0.0521,
      "step": 247125
    },
    {
      "epoch": 3.6398580286004627,
      "grad_norm": 1.0684806108474731,
      "learning_rate": 6.051499898544938e-06,
      "loss": 0.0553,
      "step": 247150
    },
    {
      "epoch": 3.6402262116905497,
      "grad_norm": 1.735277771949768,
      "learning_rate": 6.049863526577954e-06,
      "loss": 0.0543,
      "step": 247175
    },
    {
      "epoch": 3.6405943947806367,
      "grad_norm": 1.1124500036239624,
      "learning_rate": 6.048227154610969e-06,
      "loss": 0.058,
      "step": 247200
    },
    {
      "epoch": 3.6409625778707237,
      "grad_norm": 1.3430871963500977,
      "learning_rate": 6.046590782643985e-06,
      "loss": 0.0536,
      "step": 247225
    },
    {
      "epoch": 3.6413307609608108,
      "grad_norm": 2.3065345287323,
      "learning_rate": 6.0449544106770006e-06,
      "loss": 0.0618,
      "step": 247250
    },
    {
      "epoch": 3.641698944050898,
      "grad_norm": 0.9468328952789307,
      "learning_rate": 6.043318038710016e-06,
      "loss": 0.0491,
      "step": 247275
    },
    {
      "epoch": 3.642067127140985,
      "grad_norm": 1.0123200416564941,
      "learning_rate": 6.041681666743031e-06,
      "loss": 0.0552,
      "step": 247300
    },
    {
      "epoch": 3.642435310231072,
      "grad_norm": 1.094185471534729,
      "learning_rate": 6.0400452947760465e-06,
      "loss": 0.0512,
      "step": 247325
    },
    {
      "epoch": 3.642803493321159,
      "grad_norm": 1.2349109649658203,
      "learning_rate": 6.038408922809063e-06,
      "loss": 0.0472,
      "step": 247350
    },
    {
      "epoch": 3.643171676411246,
      "grad_norm": 1.3068591356277466,
      "learning_rate": 6.036772550842077e-06,
      "loss": 0.0525,
      "step": 247375
    },
    {
      "epoch": 3.643539859501333,
      "grad_norm": 1.3664003610610962,
      "learning_rate": 6.035136178875093e-06,
      "loss": 0.0537,
      "step": 247400
    },
    {
      "epoch": 3.64390804259142,
      "grad_norm": 1.492756962776184,
      "learning_rate": 6.0334998069081085e-06,
      "loss": 0.0526,
      "step": 247425
    },
    {
      "epoch": 3.644276225681507,
      "grad_norm": 1.2986410856246948,
      "learning_rate": 6.031863434941124e-06,
      "loss": 0.0499,
      "step": 247450
    },
    {
      "epoch": 3.644644408771594,
      "grad_norm": 1.2511528730392456,
      "learning_rate": 6.030227062974139e-06,
      "loss": 0.0577,
      "step": 247475
    },
    {
      "epoch": 3.645012591861681,
      "grad_norm": 1.599492073059082,
      "learning_rate": 6.0285906910071544e-06,
      "loss": 0.0557,
      "step": 247500
    },
    {
      "epoch": 3.645380774951768,
      "grad_norm": 1.4056516885757446,
      "learning_rate": 6.02701977391885e-06,
      "loss": 0.061,
      "step": 247525
    },
    {
      "epoch": 3.645748958041855,
      "grad_norm": 1.0515146255493164,
      "learning_rate": 6.025383401951865e-06,
      "loss": 0.0507,
      "step": 247550
    },
    {
      "epoch": 3.646117141131942,
      "grad_norm": 1.6306889057159424,
      "learning_rate": 6.02374702998488e-06,
      "loss": 0.0574,
      "step": 247575
    },
    {
      "epoch": 3.646485324222029,
      "grad_norm": 1.2514461278915405,
      "learning_rate": 6.022110658017896e-06,
      "loss": 0.054,
      "step": 247600
    },
    {
      "epoch": 3.646853507312116,
      "grad_norm": 0.9605504870414734,
      "learning_rate": 6.020474286050911e-06,
      "loss": 0.0476,
      "step": 247625
    },
    {
      "epoch": 3.647221690402203,
      "grad_norm": 1.020743489265442,
      "learning_rate": 6.018837914083927e-06,
      "loss": 0.0555,
      "step": 247650
    },
    {
      "epoch": 3.64758987349229,
      "grad_norm": 1.1369078159332275,
      "learning_rate": 6.017201542116942e-06,
      "loss": 0.0569,
      "step": 247675
    },
    {
      "epoch": 3.647958056582377,
      "grad_norm": 1.3633928298950195,
      "learning_rate": 6.015565170149957e-06,
      "loss": 0.0516,
      "step": 247700
    },
    {
      "epoch": 3.648326239672464,
      "grad_norm": 1.1503260135650635,
      "learning_rate": 6.013928798182973e-06,
      "loss": 0.0499,
      "step": 247725
    },
    {
      "epoch": 3.6486944227625515,
      "grad_norm": 0.8007208108901978,
      "learning_rate": 6.012292426215989e-06,
      "loss": 0.048,
      "step": 247750
    },
    {
      "epoch": 3.6490626058526385,
      "grad_norm": 1.6318391561508179,
      "learning_rate": 6.010656054249005e-06,
      "loss": 0.0495,
      "step": 247775
    },
    {
      "epoch": 3.6494307889427255,
      "grad_norm": 1.3860453367233276,
      "learning_rate": 6.009019682282019e-06,
      "loss": 0.0501,
      "step": 247800
    },
    {
      "epoch": 3.6497989720328126,
      "grad_norm": 1.5322635173797607,
      "learning_rate": 6.0073833103150345e-06,
      "loss": 0.0552,
      "step": 247825
    },
    {
      "epoch": 3.6501671551228996,
      "grad_norm": 1.2735567092895508,
      "learning_rate": 6.005746938348051e-06,
      "loss": 0.0492,
      "step": 247850
    },
    {
      "epoch": 3.6505353382129866,
      "grad_norm": 1.451747179031372,
      "learning_rate": 6.004110566381066e-06,
      "loss": 0.0505,
      "step": 247875
    },
    {
      "epoch": 3.6509035213030736,
      "grad_norm": 0.9211105108261108,
      "learning_rate": 6.0024741944140804e-06,
      "loss": 0.0555,
      "step": 247900
    },
    {
      "epoch": 3.6512717043931606,
      "grad_norm": 1.5851130485534668,
      "learning_rate": 6.000837822447097e-06,
      "loss": 0.0574,
      "step": 247925
    },
    {
      "epoch": 3.6516398874832476,
      "grad_norm": 1.8523852825164795,
      "learning_rate": 5.999201450480112e-06,
      "loss": 0.0568,
      "step": 247950
    },
    {
      "epoch": 3.6520080705733347,
      "grad_norm": 0.8993156552314758,
      "learning_rate": 5.997565078513128e-06,
      "loss": 0.0514,
      "step": 247975
    },
    {
      "epoch": 3.6523762536634217,
      "grad_norm": 1.0645922422409058,
      "learning_rate": 5.9959287065461425e-06,
      "loss": 0.052,
      "step": 248000
    },
    {
      "epoch": 3.6527444367535087,
      "grad_norm": 1.5108942985534668,
      "learning_rate": 5.994292334579158e-06,
      "loss": 0.0585,
      "step": 248025
    },
    {
      "epoch": 3.6531126198435957,
      "grad_norm": 1.2397645711898804,
      "learning_rate": 5.992655962612174e-06,
      "loss": 0.0609,
      "step": 248050
    },
    {
      "epoch": 3.6534808029336827,
      "grad_norm": 2.0706255435943604,
      "learning_rate": 5.991019590645189e-06,
      "loss": 0.0505,
      "step": 248075
    },
    {
      "epoch": 3.65384898602377,
      "grad_norm": 1.3555341958999634,
      "learning_rate": 5.9893832186782046e-06,
      "loss": 0.0529,
      "step": 248100
    },
    {
      "epoch": 3.654217169113857,
      "grad_norm": 1.3032127618789673,
      "learning_rate": 5.98774684671122e-06,
      "loss": 0.0532,
      "step": 248125
    },
    {
      "epoch": 3.654585352203944,
      "grad_norm": 1.8432704210281372,
      "learning_rate": 5.986110474744235e-06,
      "loss": 0.0518,
      "step": 248150
    },
    {
      "epoch": 3.654953535294031,
      "grad_norm": 1.8248027563095093,
      "learning_rate": 5.984474102777251e-06,
      "loss": 0.0521,
      "step": 248175
    },
    {
      "epoch": 3.6553217183841182,
      "grad_norm": 1.4147013425827026,
      "learning_rate": 5.982837730810267e-06,
      "loss": 0.0546,
      "step": 248200
    },
    {
      "epoch": 3.6556899014742053,
      "grad_norm": 1.702972650527954,
      "learning_rate": 5.981201358843282e-06,
      "loss": 0.0616,
      "step": 248225
    },
    {
      "epoch": 3.6560580845642923,
      "grad_norm": 1.338383674621582,
      "learning_rate": 5.979564986876297e-06,
      "loss": 0.0544,
      "step": 248250
    },
    {
      "epoch": 3.6564262676543793,
      "grad_norm": 1.0849289894104004,
      "learning_rate": 5.977928614909313e-06,
      "loss": 0.0489,
      "step": 248275
    },
    {
      "epoch": 3.6567944507444663,
      "grad_norm": 1.5459071397781372,
      "learning_rate": 5.976292242942329e-06,
      "loss": 0.0556,
      "step": 248300
    },
    {
      "epoch": 3.6571626338345533,
      "grad_norm": 0.8210369348526001,
      "learning_rate": 5.974655870975343e-06,
      "loss": 0.044,
      "step": 248325
    },
    {
      "epoch": 3.6575308169246403,
      "grad_norm": 1.5008289813995361,
      "learning_rate": 5.973019499008359e-06,
      "loss": 0.0513,
      "step": 248350
    },
    {
      "epoch": 3.6578990000147273,
      "grad_norm": 1.6575672626495361,
      "learning_rate": 5.971383127041375e-06,
      "loss": 0.0537,
      "step": 248375
    },
    {
      "epoch": 3.6582671831048144,
      "grad_norm": 1.339687705039978,
      "learning_rate": 5.969746755074391e-06,
      "loss": 0.0507,
      "step": 248400
    },
    {
      "epoch": 3.6586353661949014,
      "grad_norm": 2.0025038719177246,
      "learning_rate": 5.968110383107405e-06,
      "loss": 0.0536,
      "step": 248425
    },
    {
      "epoch": 3.6590035492849884,
      "grad_norm": 1.4699890613555908,
      "learning_rate": 5.9664740111404205e-06,
      "loss": 0.0566,
      "step": 248450
    },
    {
      "epoch": 3.6593717323750754,
      "grad_norm": 1.4215010404586792,
      "learning_rate": 5.964837639173437e-06,
      "loss": 0.0518,
      "step": 248475
    },
    {
      "epoch": 3.6597399154651624,
      "grad_norm": 1.5836677551269531,
      "learning_rate": 5.963201267206452e-06,
      "loss": 0.0515,
      "step": 248500
    },
    {
      "epoch": 3.6601080985552494,
      "grad_norm": 1.1467339992523193,
      "learning_rate": 5.9615648952394664e-06,
      "loss": 0.0558,
      "step": 248525
    },
    {
      "epoch": 3.6604762816453364,
      "grad_norm": 0.9520508050918579,
      "learning_rate": 5.959928523272483e-06,
      "loss": 0.0579,
      "step": 248550
    },
    {
      "epoch": 3.6608444647354235,
      "grad_norm": 1.471479058265686,
      "learning_rate": 5.958292151305498e-06,
      "loss": 0.0529,
      "step": 248575
    },
    {
      "epoch": 3.6612126478255105,
      "grad_norm": 1.4006043672561646,
      "learning_rate": 5.956655779338514e-06,
      "loss": 0.0476,
      "step": 248600
    },
    {
      "epoch": 3.6615808309155975,
      "grad_norm": 1.6726654767990112,
      "learning_rate": 5.9550194073715285e-06,
      "loss": 0.0577,
      "step": 248625
    },
    {
      "epoch": 3.6619490140056845,
      "grad_norm": 1.358842372894287,
      "learning_rate": 5.953383035404544e-06,
      "loss": 0.0548,
      "step": 248650
    },
    {
      "epoch": 3.6623171970957715,
      "grad_norm": 1.4992647171020508,
      "learning_rate": 5.95174666343756e-06,
      "loss": 0.0512,
      "step": 248675
    },
    {
      "epoch": 3.662685380185859,
      "grad_norm": 1.303995966911316,
      "learning_rate": 5.950110291470575e-06,
      "loss": 0.0519,
      "step": 248700
    },
    {
      "epoch": 3.663053563275946,
      "grad_norm": 1.8302656412124634,
      "learning_rate": 5.948473919503591e-06,
      "loss": 0.0555,
      "step": 248725
    },
    {
      "epoch": 3.663421746366033,
      "grad_norm": 1.6934587955474854,
      "learning_rate": 5.946837547536606e-06,
      "loss": 0.0552,
      "step": 248750
    },
    {
      "epoch": 3.66378992945612,
      "grad_norm": 0.9948172569274902,
      "learning_rate": 5.945201175569621e-06,
      "loss": 0.0513,
      "step": 248775
    },
    {
      "epoch": 3.664158112546207,
      "grad_norm": 1.5662486553192139,
      "learning_rate": 5.943564803602637e-06,
      "loss": 0.0522,
      "step": 248800
    },
    {
      "epoch": 3.664526295636294,
      "grad_norm": 0.9182396531105042,
      "learning_rate": 5.941928431635653e-06,
      "loss": 0.0523,
      "step": 248825
    },
    {
      "epoch": 3.664894478726381,
      "grad_norm": 1.5740448236465454,
      "learning_rate": 5.940292059668668e-06,
      "loss": 0.0566,
      "step": 248850
    },
    {
      "epoch": 3.665262661816468,
      "grad_norm": 1.4985628128051758,
      "learning_rate": 5.938655687701683e-06,
      "loss": 0.0611,
      "step": 248875
    },
    {
      "epoch": 3.665630844906555,
      "grad_norm": 0.7719475030899048,
      "learning_rate": 5.9370193157346985e-06,
      "loss": 0.0585,
      "step": 248900
    },
    {
      "epoch": 3.665999027996642,
      "grad_norm": 1.188104510307312,
      "learning_rate": 5.935382943767715e-06,
      "loss": 0.0587,
      "step": 248925
    },
    {
      "epoch": 3.666367211086729,
      "grad_norm": 1.3426990509033203,
      "learning_rate": 5.933746571800729e-06,
      "loss": 0.0576,
      "step": 248950
    },
    {
      "epoch": 3.666735394176816,
      "grad_norm": 1.6435117721557617,
      "learning_rate": 5.932110199833745e-06,
      "loss": 0.0541,
      "step": 248975
    },
    {
      "epoch": 3.667103577266903,
      "grad_norm": 1.4483036994934082,
      "learning_rate": 5.930473827866761e-06,
      "loss": 0.0494,
      "step": 249000
    },
    {
      "epoch": 3.66747176035699,
      "grad_norm": 1.3431450128555298,
      "learning_rate": 5.928837455899777e-06,
      "loss": 0.0562,
      "step": 249025
    },
    {
      "epoch": 3.667839943447077,
      "grad_norm": 1.640661597251892,
      "learning_rate": 5.927201083932791e-06,
      "loss": 0.0557,
      "step": 249050
    },
    {
      "epoch": 3.6682081265371647,
      "grad_norm": 1.4969080686569214,
      "learning_rate": 5.9255647119658065e-06,
      "loss": 0.0553,
      "step": 249075
    },
    {
      "epoch": 3.6685763096272517,
      "grad_norm": 1.5314726829528809,
      "learning_rate": 5.923928339998823e-06,
      "loss": 0.053,
      "step": 249100
    },
    {
      "epoch": 3.6689444927173387,
      "grad_norm": 1.4686625003814697,
      "learning_rate": 5.922291968031838e-06,
      "loss": 0.057,
      "step": 249125
    },
    {
      "epoch": 3.6693126758074257,
      "grad_norm": 1.0762619972229004,
      "learning_rate": 5.9206555960648524e-06,
      "loss": 0.0504,
      "step": 249150
    },
    {
      "epoch": 3.6696808588975127,
      "grad_norm": 1.2639774084091187,
      "learning_rate": 5.919019224097869e-06,
      "loss": 0.0556,
      "step": 249175
    },
    {
      "epoch": 3.6700490419875997,
      "grad_norm": 1.439120888710022,
      "learning_rate": 5.917382852130884e-06,
      "loss": 0.0555,
      "step": 249200
    },
    {
      "epoch": 3.6704172250776868,
      "grad_norm": 1.29892897605896,
      "learning_rate": 5.9157464801639e-06,
      "loss": 0.0537,
      "step": 249225
    },
    {
      "epoch": 3.6707854081677738,
      "grad_norm": 1.665852427482605,
      "learning_rate": 5.914110108196915e-06,
      "loss": 0.0597,
      "step": 249250
    },
    {
      "epoch": 3.671153591257861,
      "grad_norm": 1.8004980087280273,
      "learning_rate": 5.91247373622993e-06,
      "loss": 0.0568,
      "step": 249275
    },
    {
      "epoch": 3.671521774347948,
      "grad_norm": 1.6189017295837402,
      "learning_rate": 5.910837364262946e-06,
      "loss": 0.0556,
      "step": 249300
    },
    {
      "epoch": 3.671889957438035,
      "grad_norm": 1.3257699012756348,
      "learning_rate": 5.909200992295961e-06,
      "loss": 0.0511,
      "step": 249325
    },
    {
      "epoch": 3.672258140528122,
      "grad_norm": 1.8579190969467163,
      "learning_rate": 5.907564620328977e-06,
      "loss": 0.0547,
      "step": 249350
    },
    {
      "epoch": 3.672626323618209,
      "grad_norm": 1.4295527935028076,
      "learning_rate": 5.905928248361992e-06,
      "loss": 0.0476,
      "step": 249375
    },
    {
      "epoch": 3.672994506708296,
      "grad_norm": 0.9922029376029968,
      "learning_rate": 5.904291876395007e-06,
      "loss": 0.052,
      "step": 249400
    },
    {
      "epoch": 3.673362689798383,
      "grad_norm": 1.6572712659835815,
      "learning_rate": 5.902655504428023e-06,
      "loss": 0.0593,
      "step": 249425
    },
    {
      "epoch": 3.67373087288847,
      "grad_norm": 1.2601135969161987,
      "learning_rate": 5.901019132461039e-06,
      "loss": 0.0494,
      "step": 249450
    },
    {
      "epoch": 3.674099055978557,
      "grad_norm": 0.8301722407341003,
      "learning_rate": 5.899382760494053e-06,
      "loss": 0.0488,
      "step": 249475
    },
    {
      "epoch": 3.674467239068644,
      "grad_norm": 1.1777797937393188,
      "learning_rate": 5.897746388527069e-06,
      "loss": 0.0546,
      "step": 249500
    },
    {
      "epoch": 3.674835422158731,
      "grad_norm": 1.8586320877075195,
      "learning_rate": 5.8961100165600845e-06,
      "loss": 0.0567,
      "step": 249525
    },
    {
      "epoch": 3.675203605248818,
      "grad_norm": 1.4060636758804321,
      "learning_rate": 5.894473644593101e-06,
      "loss": 0.055,
      "step": 249550
    },
    {
      "epoch": 3.675571788338905,
      "grad_norm": 1.2396047115325928,
      "learning_rate": 5.892837272626115e-06,
      "loss": 0.0553,
      "step": 249575
    },
    {
      "epoch": 3.675939971428992,
      "grad_norm": 1.1717571020126343,
      "learning_rate": 5.8912009006591304e-06,
      "loss": 0.0481,
      "step": 249600
    },
    {
      "epoch": 3.676308154519079,
      "grad_norm": 1.4184919595718384,
      "learning_rate": 5.889564528692147e-06,
      "loss": 0.0537,
      "step": 249625
    },
    {
      "epoch": 3.676676337609166,
      "grad_norm": 0.9783734083175659,
      "learning_rate": 5.887928156725162e-06,
      "loss": 0.0481,
      "step": 249650
    },
    {
      "epoch": 3.6770445206992535,
      "grad_norm": 0.8995241522789001,
      "learning_rate": 5.886291784758178e-06,
      "loss": 0.0527,
      "step": 249675
    },
    {
      "epoch": 3.6774127037893405,
      "grad_norm": 0.9122073650360107,
      "learning_rate": 5.8846554127911925e-06,
      "loss": 0.0528,
      "step": 249700
    },
    {
      "epoch": 3.6777808868794275,
      "grad_norm": 1.0846768617630005,
      "learning_rate": 5.883019040824209e-06,
      "loss": 0.0472,
      "step": 249725
    },
    {
      "epoch": 3.6781490699695145,
      "grad_norm": 1.857362985610962,
      "learning_rate": 5.881382668857224e-06,
      "loss": 0.0554,
      "step": 249750
    },
    {
      "epoch": 3.6785172530596015,
      "grad_norm": 1.790833592414856,
      "learning_rate": 5.879746296890239e-06,
      "loss": 0.0556,
      "step": 249775
    },
    {
      "epoch": 3.6788854361496885,
      "grad_norm": 1.3882207870483398,
      "learning_rate": 5.878109924923255e-06,
      "loss": 0.0505,
      "step": 249800
    },
    {
      "epoch": 3.6792536192397756,
      "grad_norm": 1.0891906023025513,
      "learning_rate": 5.87647355295627e-06,
      "loss": 0.0473,
      "step": 249825
    },
    {
      "epoch": 3.6796218023298626,
      "grad_norm": 1.1623611450195312,
      "learning_rate": 5.874837180989286e-06,
      "loss": 0.0608,
      "step": 249850
    },
    {
      "epoch": 3.6799899854199496,
      "grad_norm": 0.8595478534698486,
      "learning_rate": 5.873200809022301e-06,
      "loss": 0.0541,
      "step": 249875
    },
    {
      "epoch": 3.6803581685100366,
      "grad_norm": 1.7485387325286865,
      "learning_rate": 5.871564437055316e-06,
      "loss": 0.0546,
      "step": 249900
    },
    {
      "epoch": 3.6807263516001236,
      "grad_norm": 0.9385265707969666,
      "learning_rate": 5.869928065088332e-06,
      "loss": 0.0498,
      "step": 249925
    },
    {
      "epoch": 3.6810945346902106,
      "grad_norm": 1.4399908781051636,
      "learning_rate": 5.868291693121347e-06,
      "loss": 0.0463,
      "step": 249950
    },
    {
      "epoch": 3.6814627177802977,
      "grad_norm": 1.282204031944275,
      "learning_rate": 5.866655321154363e-06,
      "loss": 0.0519,
      "step": 249975
    },
    {
      "epoch": 3.6818309008703847,
      "grad_norm": 1.3285976648330688,
      "learning_rate": 5.865018949187378e-06,
      "loss": 0.0486,
      "step": 250000
    },
    {
      "epoch": 3.682199083960472,
      "grad_norm": 1.1242409944534302,
      "learning_rate": 5.863382577220393e-06,
      "loss": 0.0625,
      "step": 250025
    },
    {
      "epoch": 3.682567267050559,
      "grad_norm": 1.1674872636795044,
      "learning_rate": 5.861746205253409e-06,
      "loss": 0.0594,
      "step": 250050
    },
    {
      "epoch": 3.682935450140646,
      "grad_norm": 1.760764241218567,
      "learning_rate": 5.860109833286425e-06,
      "loss": 0.0592,
      "step": 250075
    },
    {
      "epoch": 3.683303633230733,
      "grad_norm": 1.2220913171768188,
      "learning_rate": 5.858473461319439e-06,
      "loss": 0.0539,
      "step": 250100
    },
    {
      "epoch": 3.68367181632082,
      "grad_norm": 0.8641085028648376,
      "learning_rate": 5.856837089352455e-06,
      "loss": 0.0482,
      "step": 250125
    },
    {
      "epoch": 3.684039999410907,
      "grad_norm": 1.2287153005599976,
      "learning_rate": 5.8552007173854705e-06,
      "loss": 0.0487,
      "step": 250150
    },
    {
      "epoch": 3.6844081825009942,
      "grad_norm": 1.0244650840759277,
      "learning_rate": 5.853564345418487e-06,
      "loss": 0.0571,
      "step": 250175
    },
    {
      "epoch": 3.6847763655910812,
      "grad_norm": 1.1917394399642944,
      "learning_rate": 5.851927973451502e-06,
      "loss": 0.0567,
      "step": 250200
    },
    {
      "epoch": 3.6851445486811683,
      "grad_norm": 1.7676109075546265,
      "learning_rate": 5.8502916014845164e-06,
      "loss": 0.0509,
      "step": 250225
    },
    {
      "epoch": 3.6855127317712553,
      "grad_norm": 1.4125123023986816,
      "learning_rate": 5.848655229517533e-06,
      "loss": 0.057,
      "step": 250250
    },
    {
      "epoch": 3.6858809148613423,
      "grad_norm": 1.3902137279510498,
      "learning_rate": 5.847018857550548e-06,
      "loss": 0.0571,
      "step": 250275
    },
    {
      "epoch": 3.6862490979514293,
      "grad_norm": 1.337487816810608,
      "learning_rate": 5.845382485583564e-06,
      "loss": 0.0578,
      "step": 250300
    },
    {
      "epoch": 3.6866172810415163,
      "grad_norm": 1.1797250509262085,
      "learning_rate": 5.8437461136165785e-06,
      "loss": 0.0547,
      "step": 250325
    },
    {
      "epoch": 3.6869854641316033,
      "grad_norm": 1.3180617094039917,
      "learning_rate": 5.842109741649594e-06,
      "loss": 0.0558,
      "step": 250350
    },
    {
      "epoch": 3.6873536472216903,
      "grad_norm": 1.0192644596099854,
      "learning_rate": 5.84047336968261e-06,
      "loss": 0.0516,
      "step": 250375
    },
    {
      "epoch": 3.6877218303117774,
      "grad_norm": 1.0768439769744873,
      "learning_rate": 5.838836997715625e-06,
      "loss": 0.0627,
      "step": 250400
    },
    {
      "epoch": 3.6880900134018644,
      "grad_norm": 1.3138222694396973,
      "learning_rate": 5.8372006257486406e-06,
      "loss": 0.051,
      "step": 250425
    },
    {
      "epoch": 3.6884581964919514,
      "grad_norm": 1.0106867551803589,
      "learning_rate": 5.835564253781656e-06,
      "loss": 0.0534,
      "step": 250450
    },
    {
      "epoch": 3.6888263795820384,
      "grad_norm": 1.5678738355636597,
      "learning_rate": 5.833927881814672e-06,
      "loss": 0.0561,
      "step": 250475
    },
    {
      "epoch": 3.6891945626721254,
      "grad_norm": 1.904113531112671,
      "learning_rate": 5.832291509847687e-06,
      "loss": 0.0569,
      "step": 250500
    },
    {
      "epoch": 3.6895627457622124,
      "grad_norm": 0.9363990426063538,
      "learning_rate": 5.830655137880702e-06,
      "loss": 0.0522,
      "step": 250525
    },
    {
      "epoch": 3.6899309288522995,
      "grad_norm": 1.0286802053451538,
      "learning_rate": 5.829018765913718e-06,
      "loss": 0.0524,
      "step": 250550
    },
    {
      "epoch": 3.6902991119423865,
      "grad_norm": 0.9147318005561829,
      "learning_rate": 5.827382393946733e-06,
      "loss": 0.0498,
      "step": 250575
    },
    {
      "epoch": 3.6906672950324735,
      "grad_norm": 1.3984861373901367,
      "learning_rate": 5.825746021979749e-06,
      "loss": 0.0511,
      "step": 250600
    },
    {
      "epoch": 3.691035478122561,
      "grad_norm": 2.1696903705596924,
      "learning_rate": 5.824109650012764e-06,
      "loss": 0.0578,
      "step": 250625
    },
    {
      "epoch": 3.691403661212648,
      "grad_norm": 1.66084885597229,
      "learning_rate": 5.822473278045779e-06,
      "loss": 0.0522,
      "step": 250650
    },
    {
      "epoch": 3.691771844302735,
      "grad_norm": 1.3171850442886353,
      "learning_rate": 5.820836906078795e-06,
      "loss": 0.0497,
      "step": 250675
    },
    {
      "epoch": 3.692140027392822,
      "grad_norm": 0.9930122494697571,
      "learning_rate": 5.819200534111811e-06,
      "loss": 0.0504,
      "step": 250700
    },
    {
      "epoch": 3.692508210482909,
      "grad_norm": 1.7931667566299438,
      "learning_rate": 5.817564162144827e-06,
      "loss": 0.0501,
      "step": 250725
    },
    {
      "epoch": 3.692876393572996,
      "grad_norm": 1.1997019052505493,
      "learning_rate": 5.815927790177841e-06,
      "loss": 0.0492,
      "step": 250750
    },
    {
      "epoch": 3.693244576663083,
      "grad_norm": 1.5087249279022217,
      "learning_rate": 5.8142914182108565e-06,
      "loss": 0.0576,
      "step": 250775
    },
    {
      "epoch": 3.69361275975317,
      "grad_norm": 1.2138445377349854,
      "learning_rate": 5.812655046243873e-06,
      "loss": 0.0531,
      "step": 250800
    },
    {
      "epoch": 3.693980942843257,
      "grad_norm": 0.6003586053848267,
      "learning_rate": 5.811018674276888e-06,
      "loss": 0.0526,
      "step": 250825
    },
    {
      "epoch": 3.694349125933344,
      "grad_norm": 1.2139582633972168,
      "learning_rate": 5.8093823023099024e-06,
      "loss": 0.0496,
      "step": 250850
    },
    {
      "epoch": 3.694717309023431,
      "grad_norm": 1.5219069719314575,
      "learning_rate": 5.807745930342919e-06,
      "loss": 0.0548,
      "step": 250875
    },
    {
      "epoch": 3.695085492113518,
      "grad_norm": 1.328459620475769,
      "learning_rate": 5.806109558375934e-06,
      "loss": 0.0574,
      "step": 250900
    },
    {
      "epoch": 3.695453675203605,
      "grad_norm": 1.2728067636489868,
      "learning_rate": 5.80447318640895e-06,
      "loss": 0.0572,
      "step": 250925
    },
    {
      "epoch": 3.695821858293692,
      "grad_norm": 1.318516731262207,
      "learning_rate": 5.8028368144419645e-06,
      "loss": 0.0514,
      "step": 250950
    },
    {
      "epoch": 3.6961900413837796,
      "grad_norm": 1.19840407371521,
      "learning_rate": 5.80120044247498e-06,
      "loss": 0.0629,
      "step": 250975
    },
    {
      "epoch": 3.6965582244738666,
      "grad_norm": 0.6499559283256531,
      "learning_rate": 5.799564070507996e-06,
      "loss": 0.0482,
      "step": 251000
    },
    {
      "epoch": 3.6969264075639536,
      "grad_norm": 1.533064842224121,
      "learning_rate": 5.797927698541011e-06,
      "loss": 0.0587,
      "step": 251025
    },
    {
      "epoch": 3.6972945906540406,
      "grad_norm": 1.0622810125350952,
      "learning_rate": 5.7962913265740266e-06,
      "loss": 0.052,
      "step": 251050
    },
    {
      "epoch": 3.6976627737441277,
      "grad_norm": 1.281508207321167,
      "learning_rate": 5.794654954607042e-06,
      "loss": 0.0539,
      "step": 251075
    },
    {
      "epoch": 3.6980309568342147,
      "grad_norm": 1.468364953994751,
      "learning_rate": 5.793018582640057e-06,
      "loss": 0.0548,
      "step": 251100
    },
    {
      "epoch": 3.6983991399243017,
      "grad_norm": 1.375420331954956,
      "learning_rate": 5.791382210673073e-06,
      "loss": 0.0569,
      "step": 251125
    },
    {
      "epoch": 3.6987673230143887,
      "grad_norm": 1.0663232803344727,
      "learning_rate": 5.789745838706089e-06,
      "loss": 0.0545,
      "step": 251150
    },
    {
      "epoch": 3.6991355061044757,
      "grad_norm": 1.4968196153640747,
      "learning_rate": 5.788109466739104e-06,
      "loss": 0.0494,
      "step": 251175
    },
    {
      "epoch": 3.6995036891945627,
      "grad_norm": 1.2448912858963013,
      "learning_rate": 5.786473094772119e-06,
      "loss": 0.0535,
      "step": 251200
    },
    {
      "epoch": 3.6998718722846498,
      "grad_norm": 1.282294511795044,
      "learning_rate": 5.7848367228051346e-06,
      "loss": 0.0531,
      "step": 251225
    },
    {
      "epoch": 3.7002400553747368,
      "grad_norm": 1.6939245462417603,
      "learning_rate": 5.783200350838151e-06,
      "loss": 0.0562,
      "step": 251250
    },
    {
      "epoch": 3.700608238464824,
      "grad_norm": 1.7599190473556519,
      "learning_rate": 5.781563978871165e-06,
      "loss": 0.062,
      "step": 251275
    },
    {
      "epoch": 3.700976421554911,
      "grad_norm": 1.3975493907928467,
      "learning_rate": 5.779927606904181e-06,
      "loss": 0.048,
      "step": 251300
    },
    {
      "epoch": 3.701344604644998,
      "grad_norm": 1.0987061262130737,
      "learning_rate": 5.778291234937197e-06,
      "loss": 0.0583,
      "step": 251325
    },
    {
      "epoch": 3.701712787735085,
      "grad_norm": 1.5857268571853638,
      "learning_rate": 5.776654862970213e-06,
      "loss": 0.055,
      "step": 251350
    },
    {
      "epoch": 3.702080970825172,
      "grad_norm": 1.0517363548278809,
      "learning_rate": 5.775018491003227e-06,
      "loss": 0.0558,
      "step": 251375
    },
    {
      "epoch": 3.702449153915259,
      "grad_norm": 1.3491718769073486,
      "learning_rate": 5.7733821190362425e-06,
      "loss": 0.0519,
      "step": 251400
    },
    {
      "epoch": 3.702817337005346,
      "grad_norm": 1.9140602350234985,
      "learning_rate": 5.771745747069259e-06,
      "loss": 0.0527,
      "step": 251425
    },
    {
      "epoch": 3.703185520095433,
      "grad_norm": 1.5180044174194336,
      "learning_rate": 5.770109375102274e-06,
      "loss": 0.0459,
      "step": 251450
    },
    {
      "epoch": 3.70355370318552,
      "grad_norm": 1.3003180027008057,
      "learning_rate": 5.7684730031352884e-06,
      "loss": 0.0477,
      "step": 251475
    },
    {
      "epoch": 3.703921886275607,
      "grad_norm": 1.350250005722046,
      "learning_rate": 5.766836631168305e-06,
      "loss": 0.0612,
      "step": 251500
    },
    {
      "epoch": 3.704290069365694,
      "grad_norm": 1.6882920265197754,
      "learning_rate": 5.765265714079999e-06,
      "loss": 0.0461,
      "step": 251525
    },
    {
      "epoch": 3.704658252455781,
      "grad_norm": 1.538088083267212,
      "learning_rate": 5.7636293421130155e-06,
      "loss": 0.0487,
      "step": 251550
    },
    {
      "epoch": 3.7050264355458684,
      "grad_norm": 1.2595710754394531,
      "learning_rate": 5.76199297014603e-06,
      "loss": 0.0485,
      "step": 251575
    },
    {
      "epoch": 3.7053946186359554,
      "grad_norm": 1.2149754762649536,
      "learning_rate": 5.760356598179045e-06,
      "loss": 0.055,
      "step": 251600
    },
    {
      "epoch": 3.7057628017260424,
      "grad_norm": 0.9896544218063354,
      "learning_rate": 5.758720226212061e-06,
      "loss": 0.0564,
      "step": 251625
    },
    {
      "epoch": 3.7061309848161295,
      "grad_norm": 1.3770445585250854,
      "learning_rate": 5.757083854245077e-06,
      "loss": 0.052,
      "step": 251650
    },
    {
      "epoch": 3.7064991679062165,
      "grad_norm": 1.1469231843948364,
      "learning_rate": 5.755447482278093e-06,
      "loss": 0.045,
      "step": 251675
    },
    {
      "epoch": 3.7068673509963035,
      "grad_norm": 1.002680778503418,
      "learning_rate": 5.753811110311107e-06,
      "loss": 0.0502,
      "step": 251700
    },
    {
      "epoch": 3.7072355340863905,
      "grad_norm": 1.060140609741211,
      "learning_rate": 5.752174738344123e-06,
      "loss": 0.0603,
      "step": 251725
    },
    {
      "epoch": 3.7076037171764775,
      "grad_norm": 1.5683294534683228,
      "learning_rate": 5.750538366377139e-06,
      "loss": 0.0599,
      "step": 251750
    },
    {
      "epoch": 3.7079719002665645,
      "grad_norm": 0.8978787064552307,
      "learning_rate": 5.748901994410154e-06,
      "loss": 0.0538,
      "step": 251775
    },
    {
      "epoch": 3.7083400833566516,
      "grad_norm": 1.6562714576721191,
      "learning_rate": 5.7472656224431685e-06,
      "loss": 0.0569,
      "step": 251800
    },
    {
      "epoch": 3.7087082664467386,
      "grad_norm": 1.2582255601882935,
      "learning_rate": 5.745629250476185e-06,
      "loss": 0.0552,
      "step": 251825
    },
    {
      "epoch": 3.7090764495368256,
      "grad_norm": 1.4364217519760132,
      "learning_rate": 5.7439928785092e-06,
      "loss": 0.0544,
      "step": 251850
    },
    {
      "epoch": 3.7094446326269126,
      "grad_norm": 1.3657604455947876,
      "learning_rate": 5.742356506542216e-06,
      "loss": 0.0607,
      "step": 251875
    },
    {
      "epoch": 3.7098128157169996,
      "grad_norm": 1.0781053304672241,
      "learning_rate": 5.740720134575231e-06,
      "loss": 0.0524,
      "step": 251900
    },
    {
      "epoch": 3.7101809988070866,
      "grad_norm": 1.0399731397628784,
      "learning_rate": 5.739083762608246e-06,
      "loss": 0.0516,
      "step": 251925
    },
    {
      "epoch": 3.710549181897174,
      "grad_norm": 0.8589385151863098,
      "learning_rate": 5.7375128455199415e-06,
      "loss": 0.0544,
      "step": 251950
    },
    {
      "epoch": 3.710917364987261,
      "grad_norm": 0.9743944406509399,
      "learning_rate": 5.735876473552957e-06,
      "loss": 0.0545,
      "step": 251975
    },
    {
      "epoch": 3.711285548077348,
      "grad_norm": 1.2415448427200317,
      "learning_rate": 5.734240101585972e-06,
      "loss": 0.0526,
      "step": 252000
    },
    {
      "epoch": 3.711653731167435,
      "grad_norm": 1.7873594760894775,
      "learning_rate": 5.732603729618987e-06,
      "loss": 0.052,
      "step": 252025
    },
    {
      "epoch": 3.712021914257522,
      "grad_norm": 1.1923871040344238,
      "learning_rate": 5.730967357652003e-06,
      "loss": 0.057,
      "step": 252050
    },
    {
      "epoch": 3.712390097347609,
      "grad_norm": 1.191372275352478,
      "learning_rate": 5.729330985685019e-06,
      "loss": 0.0481,
      "step": 252075
    },
    {
      "epoch": 3.712758280437696,
      "grad_norm": 1.3412585258483887,
      "learning_rate": 5.727694613718034e-06,
      "loss": 0.0631,
      "step": 252100
    },
    {
      "epoch": 3.713126463527783,
      "grad_norm": 1.0661895275115967,
      "learning_rate": 5.7260582417510494e-06,
      "loss": 0.0543,
      "step": 252125
    },
    {
      "epoch": 3.71349464661787,
      "grad_norm": 1.5244656801223755,
      "learning_rate": 5.724421869784065e-06,
      "loss": 0.0526,
      "step": 252150
    },
    {
      "epoch": 3.7138628297079572,
      "grad_norm": 1.5156444311141968,
      "learning_rate": 5.722785497817081e-06,
      "loss": 0.0557,
      "step": 252175
    },
    {
      "epoch": 3.7142310127980442,
      "grad_norm": 2.343926429748535,
      "learning_rate": 5.721149125850096e-06,
      "loss": 0.0552,
      "step": 252200
    },
    {
      "epoch": 3.7145991958881313,
      "grad_norm": 1.3030555248260498,
      "learning_rate": 5.719512753883111e-06,
      "loss": 0.0514,
      "step": 252225
    },
    {
      "epoch": 3.7149673789782183,
      "grad_norm": 1.3795397281646729,
      "learning_rate": 5.717876381916127e-06,
      "loss": 0.051,
      "step": 252250
    },
    {
      "epoch": 3.7153355620683053,
      "grad_norm": 1.3526875972747803,
      "learning_rate": 5.716240009949142e-06,
      "loss": 0.0551,
      "step": 252275
    },
    {
      "epoch": 3.7157037451583923,
      "grad_norm": 0.9389641284942627,
      "learning_rate": 5.714603637982158e-06,
      "loss": 0.0528,
      "step": 252300
    },
    {
      "epoch": 3.7160719282484793,
      "grad_norm": 1.8082996606826782,
      "learning_rate": 5.712967266015173e-06,
      "loss": 0.0549,
      "step": 252325
    },
    {
      "epoch": 3.7164401113385663,
      "grad_norm": 2.0994391441345215,
      "learning_rate": 5.711330894048188e-06,
      "loss": 0.0546,
      "step": 252350
    },
    {
      "epoch": 3.7168082944286533,
      "grad_norm": 1.2304348945617676,
      "learning_rate": 5.709694522081204e-06,
      "loss": 0.0579,
      "step": 252375
    },
    {
      "epoch": 3.7171764775187404,
      "grad_norm": 1.218194603919983,
      "learning_rate": 5.7080581501142195e-06,
      "loss": 0.051,
      "step": 252400
    },
    {
      "epoch": 3.7175446606088274,
      "grad_norm": 0.8290098309516907,
      "learning_rate": 5.706421778147234e-06,
      "loss": 0.0501,
      "step": 252425
    },
    {
      "epoch": 3.7179128436989144,
      "grad_norm": 1.408225417137146,
      "learning_rate": 5.70478540618025e-06,
      "loss": 0.0575,
      "step": 252450
    },
    {
      "epoch": 3.7182810267890014,
      "grad_norm": 1.3667598962783813,
      "learning_rate": 5.703149034213265e-06,
      "loss": 0.047,
      "step": 252475
    },
    {
      "epoch": 3.7186492098790884,
      "grad_norm": 1.0042376518249512,
      "learning_rate": 5.7015126622462815e-06,
      "loss": 0.0482,
      "step": 252500
    },
    {
      "epoch": 3.7190173929691754,
      "grad_norm": 1.4553728103637695,
      "learning_rate": 5.699876290279296e-06,
      "loss": 0.056,
      "step": 252525
    },
    {
      "epoch": 3.719385576059263,
      "grad_norm": 1.4743157625198364,
      "learning_rate": 5.698239918312311e-06,
      "loss": 0.0497,
      "step": 252550
    },
    {
      "epoch": 3.71975375914935,
      "grad_norm": 1.0868481397628784,
      "learning_rate": 5.6966035463453275e-06,
      "loss": 0.0534,
      "step": 252575
    },
    {
      "epoch": 3.720121942239437,
      "grad_norm": 0.7166820168495178,
      "learning_rate": 5.694967174378343e-06,
      "loss": 0.0467,
      "step": 252600
    },
    {
      "epoch": 3.720490125329524,
      "grad_norm": 1.4140657186508179,
      "learning_rate": 5.693330802411359e-06,
      "loss": 0.0567,
      "step": 252625
    },
    {
      "epoch": 3.720858308419611,
      "grad_norm": 1.2269924879074097,
      "learning_rate": 5.691694430444373e-06,
      "loss": 0.0523,
      "step": 252650
    },
    {
      "epoch": 3.721226491509698,
      "grad_norm": 2.159184455871582,
      "learning_rate": 5.690058058477389e-06,
      "loss": 0.0526,
      "step": 252675
    },
    {
      "epoch": 3.721594674599785,
      "grad_norm": 1.8485740423202515,
      "learning_rate": 5.688421686510405e-06,
      "loss": 0.059,
      "step": 252700
    },
    {
      "epoch": 3.721962857689872,
      "grad_norm": 2.0930328369140625,
      "learning_rate": 5.68678531454342e-06,
      "loss": 0.0494,
      "step": 252725
    },
    {
      "epoch": 3.722331040779959,
      "grad_norm": 1.4516451358795166,
      "learning_rate": 5.6851489425764354e-06,
      "loss": 0.0565,
      "step": 252750
    },
    {
      "epoch": 3.722699223870046,
      "grad_norm": 1.2605059146881104,
      "learning_rate": 5.683512570609451e-06,
      "loss": 0.0529,
      "step": 252775
    },
    {
      "epoch": 3.723067406960133,
      "grad_norm": 0.9492141604423523,
      "learning_rate": 5.681876198642466e-06,
      "loss": 0.0521,
      "step": 252800
    },
    {
      "epoch": 3.72343559005022,
      "grad_norm": 1.4231098890304565,
      "learning_rate": 5.680239826675482e-06,
      "loss": 0.0567,
      "step": 252825
    },
    {
      "epoch": 3.723803773140307,
      "grad_norm": 1.3626117706298828,
      "learning_rate": 5.678603454708497e-06,
      "loss": 0.0498,
      "step": 252850
    },
    {
      "epoch": 3.724171956230394,
      "grad_norm": 1.5522030591964722,
      "learning_rate": 5.676967082741513e-06,
      "loss": 0.0495,
      "step": 252875
    },
    {
      "epoch": 3.7245401393204816,
      "grad_norm": 0.9307212829589844,
      "learning_rate": 5.675330710774528e-06,
      "loss": 0.0603,
      "step": 252900
    },
    {
      "epoch": 3.7249083224105686,
      "grad_norm": 1.206346869468689,
      "learning_rate": 5.673694338807543e-06,
      "loss": 0.0558,
      "step": 252925
    },
    {
      "epoch": 3.7252765055006556,
      "grad_norm": 1.2777684926986694,
      "learning_rate": 5.672057966840559e-06,
      "loss": 0.0517,
      "step": 252950
    },
    {
      "epoch": 3.7256446885907426,
      "grad_norm": 1.1456856727600098,
      "learning_rate": 5.670421594873574e-06,
      "loss": 0.0513,
      "step": 252975
    },
    {
      "epoch": 3.7260128716808296,
      "grad_norm": 1.420722246170044,
      "learning_rate": 5.66878522290659e-06,
      "loss": 0.0494,
      "step": 253000
    },
    {
      "epoch": 3.7263810547709166,
      "grad_norm": 1.6824190616607666,
      "learning_rate": 5.6671488509396055e-06,
      "loss": 0.0534,
      "step": 253025
    },
    {
      "epoch": 3.7267492378610036,
      "grad_norm": 1.2085250616073608,
      "learning_rate": 5.66551247897262e-06,
      "loss": 0.0528,
      "step": 253050
    },
    {
      "epoch": 3.7271174209510907,
      "grad_norm": 1.712480902671814,
      "learning_rate": 5.663876107005636e-06,
      "loss": 0.0572,
      "step": 253075
    },
    {
      "epoch": 3.7274856040411777,
      "grad_norm": 1.1883583068847656,
      "learning_rate": 5.662239735038651e-06,
      "loss": 0.0539,
      "step": 253100
    },
    {
      "epoch": 3.7278537871312647,
      "grad_norm": 1.7856664657592773,
      "learning_rate": 5.6606033630716675e-06,
      "loss": 0.0526,
      "step": 253125
    },
    {
      "epoch": 3.7282219702213517,
      "grad_norm": 1.5447059869766235,
      "learning_rate": 5.658966991104683e-06,
      "loss": 0.0526,
      "step": 253150
    },
    {
      "epoch": 3.7285901533114387,
      "grad_norm": 1.5127798318862915,
      "learning_rate": 5.657330619137697e-06,
      "loss": 0.0597,
      "step": 253175
    },
    {
      "epoch": 3.7289583364015257,
      "grad_norm": 1.3642851114273071,
      "learning_rate": 5.6556942471707135e-06,
      "loss": 0.0452,
      "step": 253200
    },
    {
      "epoch": 3.7293265194916128,
      "grad_norm": 1.419844150543213,
      "learning_rate": 5.654057875203729e-06,
      "loss": 0.0544,
      "step": 253225
    },
    {
      "epoch": 3.7296947025816998,
      "grad_norm": 0.8943883180618286,
      "learning_rate": 5.652421503236745e-06,
      "loss": 0.0484,
      "step": 253250
    },
    {
      "epoch": 3.730062885671787,
      "grad_norm": 1.8702971935272217,
      "learning_rate": 5.650785131269759e-06,
      "loss": 0.0543,
      "step": 253275
    },
    {
      "epoch": 3.730431068761874,
      "grad_norm": 1.5334993600845337,
      "learning_rate": 5.649148759302775e-06,
      "loss": 0.0561,
      "step": 253300
    },
    {
      "epoch": 3.730799251851961,
      "grad_norm": 1.957545518875122,
      "learning_rate": 5.647512387335791e-06,
      "loss": 0.0592,
      "step": 253325
    },
    {
      "epoch": 3.731167434942048,
      "grad_norm": 1.1380528211593628,
      "learning_rate": 5.645876015368806e-06,
      "loss": 0.0587,
      "step": 253350
    },
    {
      "epoch": 3.731535618032135,
      "grad_norm": 1.2202318906784058,
      "learning_rate": 5.644239643401821e-06,
      "loss": 0.0531,
      "step": 253375
    },
    {
      "epoch": 3.731903801122222,
      "grad_norm": 1.397199034690857,
      "learning_rate": 5.642603271434837e-06,
      "loss": 0.0544,
      "step": 253400
    },
    {
      "epoch": 3.732271984212309,
      "grad_norm": 1.5993881225585938,
      "learning_rate": 5.640966899467852e-06,
      "loss": 0.0527,
      "step": 253425
    },
    {
      "epoch": 3.732640167302396,
      "grad_norm": 1.0414845943450928,
      "learning_rate": 5.639330527500868e-06,
      "loss": 0.0472,
      "step": 253450
    },
    {
      "epoch": 3.733008350392483,
      "grad_norm": 1.3304674625396729,
      "learning_rate": 5.637694155533883e-06,
      "loss": 0.0538,
      "step": 253475
    },
    {
      "epoch": 3.7333765334825704,
      "grad_norm": 0.9491090774536133,
      "learning_rate": 5.636057783566898e-06,
      "loss": 0.0534,
      "step": 253500
    },
    {
      "epoch": 3.7337447165726574,
      "grad_norm": 1.0399004220962524,
      "learning_rate": 5.634421411599914e-06,
      "loss": 0.0565,
      "step": 253525
    },
    {
      "epoch": 3.7341128996627444,
      "grad_norm": 1.2867871522903442,
      "learning_rate": 5.632785039632929e-06,
      "loss": 0.0564,
      "step": 253550
    },
    {
      "epoch": 3.7344810827528314,
      "grad_norm": 1.3402962684631348,
      "learning_rate": 5.6311486676659456e-06,
      "loss": 0.0501,
      "step": 253575
    },
    {
      "epoch": 3.7348492658429184,
      "grad_norm": 0.974835216999054,
      "learning_rate": 5.62951229569896e-06,
      "loss": 0.0551,
      "step": 253600
    },
    {
      "epoch": 3.7352174489330054,
      "grad_norm": 1.0583165884017944,
      "learning_rate": 5.627875923731976e-06,
      "loss": 0.0535,
      "step": 253625
    },
    {
      "epoch": 3.7355856320230925,
      "grad_norm": 2.0062127113342285,
      "learning_rate": 5.6262395517649915e-06,
      "loss": 0.0567,
      "step": 253650
    },
    {
      "epoch": 3.7359538151131795,
      "grad_norm": 1.4993880987167358,
      "learning_rate": 5.624603179798007e-06,
      "loss": 0.0569,
      "step": 253675
    },
    {
      "epoch": 3.7363219982032665,
      "grad_norm": 1.6953601837158203,
      "learning_rate": 5.622966807831022e-06,
      "loss": 0.0545,
      "step": 253700
    },
    {
      "epoch": 3.7366901812933535,
      "grad_norm": 1.5041608810424805,
      "learning_rate": 5.621330435864037e-06,
      "loss": 0.0463,
      "step": 253725
    },
    {
      "epoch": 3.7370583643834405,
      "grad_norm": 1.3922077417373657,
      "learning_rate": 5.6196940638970535e-06,
      "loss": 0.0555,
      "step": 253750
    },
    {
      "epoch": 3.7374265474735275,
      "grad_norm": 1.3642810583114624,
      "learning_rate": 5.618057691930069e-06,
      "loss": 0.0523,
      "step": 253775
    },
    {
      "epoch": 3.7377947305636146,
      "grad_norm": 1.7539656162261963,
      "learning_rate": 5.616421319963083e-06,
      "loss": 0.0549,
      "step": 253800
    },
    {
      "epoch": 3.7381629136537016,
      "grad_norm": 1.55299973487854,
      "learning_rate": 5.6147849479960995e-06,
      "loss": 0.0561,
      "step": 253825
    },
    {
      "epoch": 3.738531096743789,
      "grad_norm": 1.2092570066452026,
      "learning_rate": 5.613148576029115e-06,
      "loss": 0.057,
      "step": 253850
    },
    {
      "epoch": 3.738899279833876,
      "grad_norm": 1.6553634405136108,
      "learning_rate": 5.611512204062131e-06,
      "loss": 0.0541,
      "step": 253875
    },
    {
      "epoch": 3.739267462923963,
      "grad_norm": 1.1082477569580078,
      "learning_rate": 5.609875832095145e-06,
      "loss": 0.0537,
      "step": 253900
    },
    {
      "epoch": 3.73963564601405,
      "grad_norm": 0.8774492144584656,
      "learning_rate": 5.608239460128161e-06,
      "loss": 0.0575,
      "step": 253925
    },
    {
      "epoch": 3.740003829104137,
      "grad_norm": 1.1553659439086914,
      "learning_rate": 5.606603088161177e-06,
      "loss": 0.0563,
      "step": 253950
    },
    {
      "epoch": 3.740372012194224,
      "grad_norm": 1.5415480136871338,
      "learning_rate": 5.604966716194192e-06,
      "loss": 0.0562,
      "step": 253975
    },
    {
      "epoch": 3.740740195284311,
      "grad_norm": 0.9468506574630737,
      "learning_rate": 5.603330344227207e-06,
      "loss": 0.0574,
      "step": 254000
    },
    {
      "epoch": 3.741108378374398,
      "grad_norm": 1.3352893590927124,
      "learning_rate": 5.601693972260223e-06,
      "loss": 0.0531,
      "step": 254025
    },
    {
      "epoch": 3.741476561464485,
      "grad_norm": 1.564494252204895,
      "learning_rate": 5.600057600293238e-06,
      "loss": 0.0516,
      "step": 254050
    },
    {
      "epoch": 3.741844744554572,
      "grad_norm": 1.5080046653747559,
      "learning_rate": 5.598421228326254e-06,
      "loss": 0.0501,
      "step": 254075
    },
    {
      "epoch": 3.742212927644659,
      "grad_norm": 1.3194935321807861,
      "learning_rate": 5.5967848563592695e-06,
      "loss": 0.0597,
      "step": 254100
    },
    {
      "epoch": 3.742581110734746,
      "grad_norm": 1.4618470668792725,
      "learning_rate": 5.595148484392284e-06,
      "loss": 0.0552,
      "step": 254125
    },
    {
      "epoch": 3.742949293824833,
      "grad_norm": 1.3278696537017822,
      "learning_rate": 5.5935775673039795e-06,
      "loss": 0.0577,
      "step": 254150
    },
    {
      "epoch": 3.7433174769149202,
      "grad_norm": 2.116835832595825,
      "learning_rate": 5.591941195336995e-06,
      "loss": 0.0624,
      "step": 254175
    },
    {
      "epoch": 3.7436856600050072,
      "grad_norm": 1.7616889476776123,
      "learning_rate": 5.590304823370011e-06,
      "loss": 0.0499,
      "step": 254200
    },
    {
      "epoch": 3.7440538430950943,
      "grad_norm": 2.1772472858428955,
      "learning_rate": 5.5886684514030254e-06,
      "loss": 0.0627,
      "step": 254225
    },
    {
      "epoch": 3.7444220261851813,
      "grad_norm": 1.0219078063964844,
      "learning_rate": 5.587032079436041e-06,
      "loss": 0.0526,
      "step": 254250
    },
    {
      "epoch": 3.7447902092752683,
      "grad_norm": 1.9629639387130737,
      "learning_rate": 5.585395707469057e-06,
      "loss": 0.0552,
      "step": 254275
    },
    {
      "epoch": 3.7451583923653553,
      "grad_norm": 1.9039746522903442,
      "learning_rate": 5.583759335502072e-06,
      "loss": 0.0523,
      "step": 254300
    },
    {
      "epoch": 3.7455265754554423,
      "grad_norm": 1.5746433734893799,
      "learning_rate": 5.5821229635350875e-06,
      "loss": 0.0534,
      "step": 254325
    },
    {
      "epoch": 3.7458947585455293,
      "grad_norm": 1.8572666645050049,
      "learning_rate": 5.580486591568103e-06,
      "loss": 0.0589,
      "step": 254350
    },
    {
      "epoch": 3.7462629416356164,
      "grad_norm": 0.8775358200073242,
      "learning_rate": 5.578850219601118e-06,
      "loss": 0.0491,
      "step": 254375
    },
    {
      "epoch": 3.7466311247257034,
      "grad_norm": 1.6323455572128296,
      "learning_rate": 5.577213847634134e-06,
      "loss": 0.0582,
      "step": 254400
    },
    {
      "epoch": 3.7469993078157904,
      "grad_norm": 1.3335814476013184,
      "learning_rate": 5.575577475667149e-06,
      "loss": 0.0557,
      "step": 254425
    },
    {
      "epoch": 3.747367490905878,
      "grad_norm": 0.894173264503479,
      "learning_rate": 5.573941103700165e-06,
      "loss": 0.0588,
      "step": 254450
    },
    {
      "epoch": 3.747735673995965,
      "grad_norm": 1.5173102617263794,
      "learning_rate": 5.57230473173318e-06,
      "loss": 0.0527,
      "step": 254475
    },
    {
      "epoch": 3.748103857086052,
      "grad_norm": 1.433074951171875,
      "learning_rate": 5.5706683597661955e-06,
      "loss": 0.0522,
      "step": 254500
    },
    {
      "epoch": 3.748472040176139,
      "grad_norm": 1.2709681987762451,
      "learning_rate": 5.569031987799212e-06,
      "loss": 0.0518,
      "step": 254525
    },
    {
      "epoch": 3.748840223266226,
      "grad_norm": 1.0563149452209473,
      "learning_rate": 5.567395615832226e-06,
      "loss": 0.0534,
      "step": 254550
    },
    {
      "epoch": 3.749208406356313,
      "grad_norm": 1.5839401483535767,
      "learning_rate": 5.565759243865242e-06,
      "loss": 0.0558,
      "step": 254575
    },
    {
      "epoch": 3.7495765894464,
      "grad_norm": 1.1206369400024414,
      "learning_rate": 5.5641228718982576e-06,
      "loss": 0.0594,
      "step": 254600
    },
    {
      "epoch": 3.749944772536487,
      "grad_norm": 1.8041942119598389,
      "learning_rate": 5.562486499931274e-06,
      "loss": 0.0549,
      "step": 254625
    },
    {
      "epoch": 3.750312955626574,
      "grad_norm": 1.2459949254989624,
      "learning_rate": 5.560850127964288e-06,
      "loss": 0.0529,
      "step": 254650
    },
    {
      "epoch": 3.750681138716661,
      "grad_norm": 1.857454776763916,
      "learning_rate": 5.5592137559973035e-06,
      "loss": 0.0643,
      "step": 254675
    },
    {
      "epoch": 3.751049321806748,
      "grad_norm": 0.8499749898910522,
      "learning_rate": 5.55757738403032e-06,
      "loss": 0.0506,
      "step": 254700
    },
    {
      "epoch": 3.751417504896835,
      "grad_norm": 1.7271147966384888,
      "learning_rate": 5.555941012063335e-06,
      "loss": 0.0575,
      "step": 254725
    },
    {
      "epoch": 3.751785687986922,
      "grad_norm": 1.2960542440414429,
      "learning_rate": 5.554304640096349e-06,
      "loss": 0.0592,
      "step": 254750
    },
    {
      "epoch": 3.752153871077009,
      "grad_norm": 1.0242388248443604,
      "learning_rate": 5.5526682681293655e-06,
      "loss": 0.0557,
      "step": 254775
    },
    {
      "epoch": 3.752522054167096,
      "grad_norm": 1.5435205698013306,
      "learning_rate": 5.551031896162381e-06,
      "loss": 0.0631,
      "step": 254800
    },
    {
      "epoch": 3.7528902372571835,
      "grad_norm": 1.2656018733978271,
      "learning_rate": 5.549395524195397e-06,
      "loss": 0.0463,
      "step": 254825
    },
    {
      "epoch": 3.7532584203472705,
      "grad_norm": 1.3564785718917847,
      "learning_rate": 5.5477591522284114e-06,
      "loss": 0.0563,
      "step": 254850
    },
    {
      "epoch": 3.7536266034373575,
      "grad_norm": 1.6872167587280273,
      "learning_rate": 5.546122780261427e-06,
      "loss": 0.0564,
      "step": 254875
    },
    {
      "epoch": 3.7539947865274446,
      "grad_norm": 1.6043498516082764,
      "learning_rate": 5.544486408294443e-06,
      "loss": 0.0574,
      "step": 254900
    },
    {
      "epoch": 3.7543629696175316,
      "grad_norm": 1.449638843536377,
      "learning_rate": 5.542850036327458e-06,
      "loss": 0.0598,
      "step": 254925
    },
    {
      "epoch": 3.7547311527076186,
      "grad_norm": 1.5360407829284668,
      "learning_rate": 5.541213664360473e-06,
      "loss": 0.0526,
      "step": 254950
    },
    {
      "epoch": 3.7550993357977056,
      "grad_norm": 1.4203702211380005,
      "learning_rate": 5.539577292393489e-06,
      "loss": 0.0487,
      "step": 254975
    },
    {
      "epoch": 3.7554675188877926,
      "grad_norm": 1.0612748861312866,
      "learning_rate": 5.537940920426504e-06,
      "loss": 0.0511,
      "step": 255000
    },
    {
      "epoch": 3.7558357019778796,
      "grad_norm": 1.6567435264587402,
      "learning_rate": 5.53630454845952e-06,
      "loss": 0.0613,
      "step": 255025
    },
    {
      "epoch": 3.7562038850679667,
      "grad_norm": 0.856922447681427,
      "learning_rate": 5.5346681764925356e-06,
      "loss": 0.0532,
      "step": 255050
    },
    {
      "epoch": 3.7565720681580537,
      "grad_norm": 1.659920573234558,
      "learning_rate": 5.53303180452555e-06,
      "loss": 0.0536,
      "step": 255075
    },
    {
      "epoch": 3.7569402512481407,
      "grad_norm": 1.4628435373306274,
      "learning_rate": 5.531395432558566e-06,
      "loss": 0.0571,
      "step": 255100
    },
    {
      "epoch": 3.7573084343382277,
      "grad_norm": 1.719956398010254,
      "learning_rate": 5.5297590605915815e-06,
      "loss": 0.0608,
      "step": 255125
    },
    {
      "epoch": 3.7576766174283147,
      "grad_norm": 1.5746419429779053,
      "learning_rate": 5.528122688624598e-06,
      "loss": 0.0561,
      "step": 255150
    },
    {
      "epoch": 3.7580448005184017,
      "grad_norm": 1.9329231977462769,
      "learning_rate": 5.526486316657612e-06,
      "loss": 0.054,
      "step": 255175
    },
    {
      "epoch": 3.7584129836084887,
      "grad_norm": 1.1097121238708496,
      "learning_rate": 5.524849944690628e-06,
      "loss": 0.0516,
      "step": 255200
    },
    {
      "epoch": 3.7587811666985758,
      "grad_norm": 1.089319109916687,
      "learning_rate": 5.5232135727236436e-06,
      "loss": 0.0494,
      "step": 255225
    },
    {
      "epoch": 3.7591493497886628,
      "grad_norm": 1.215819239616394,
      "learning_rate": 5.521577200756659e-06,
      "loss": 0.0556,
      "step": 255250
    },
    {
      "epoch": 3.75951753287875,
      "grad_norm": 1.770950436592102,
      "learning_rate": 5.519940828789674e-06,
      "loss": 0.0563,
      "step": 255275
    },
    {
      "epoch": 3.759885715968837,
      "grad_norm": 1.3925715684890747,
      "learning_rate": 5.5183044568226895e-06,
      "loss": 0.0575,
      "step": 255300
    },
    {
      "epoch": 3.760253899058924,
      "grad_norm": 1.2943941354751587,
      "learning_rate": 5.516668084855706e-06,
      "loss": 0.0629,
      "step": 255325
    },
    {
      "epoch": 3.760622082149011,
      "grad_norm": 1.6082313060760498,
      "learning_rate": 5.515031712888721e-06,
      "loss": 0.0563,
      "step": 255350
    },
    {
      "epoch": 3.760990265239098,
      "grad_norm": 1.6059378385543823,
      "learning_rate": 5.513395340921735e-06,
      "loss": 0.0478,
      "step": 255375
    },
    {
      "epoch": 3.761358448329185,
      "grad_norm": 1.0294955968856812,
      "learning_rate": 5.5117589689547515e-06,
      "loss": 0.0518,
      "step": 255400
    },
    {
      "epoch": 3.7617266314192723,
      "grad_norm": 1.270851492881775,
      "learning_rate": 5.510122596987767e-06,
      "loss": 0.0547,
      "step": 255425
    },
    {
      "epoch": 3.7620948145093593,
      "grad_norm": 1.1567546129226685,
      "learning_rate": 5.508486225020783e-06,
      "loss": 0.0548,
      "step": 255450
    },
    {
      "epoch": 3.7624629975994464,
      "grad_norm": 0.7632060647010803,
      "learning_rate": 5.5068498530537974e-06,
      "loss": 0.0521,
      "step": 255475
    },
    {
      "epoch": 3.7628311806895334,
      "grad_norm": 1.4948337078094482,
      "learning_rate": 5.505213481086813e-06,
      "loss": 0.0553,
      "step": 255500
    },
    {
      "epoch": 3.7631993637796204,
      "grad_norm": 1.7801198959350586,
      "learning_rate": 5.503577109119829e-06,
      "loss": 0.0581,
      "step": 255525
    },
    {
      "epoch": 3.7635675468697074,
      "grad_norm": 1.4474761486053467,
      "learning_rate": 5.501940737152844e-06,
      "loss": 0.0568,
      "step": 255550
    },
    {
      "epoch": 3.7639357299597944,
      "grad_norm": 1.2973910570144653,
      "learning_rate": 5.50030436518586e-06,
      "loss": 0.0514,
      "step": 255575
    },
    {
      "epoch": 3.7643039130498814,
      "grad_norm": 1.047560214996338,
      "learning_rate": 5.498667993218875e-06,
      "loss": 0.0508,
      "step": 255600
    },
    {
      "epoch": 3.7646720961399684,
      "grad_norm": 1.7053852081298828,
      "learning_rate": 5.49703162125189e-06,
      "loss": 0.0549,
      "step": 255625
    },
    {
      "epoch": 3.7650402792300555,
      "grad_norm": 1.4982037544250488,
      "learning_rate": 5.495395249284906e-06,
      "loss": 0.054,
      "step": 255650
    },
    {
      "epoch": 3.7654084623201425,
      "grad_norm": 1.4650391340255737,
      "learning_rate": 5.4937588773179216e-06,
      "loss": 0.0546,
      "step": 255675
    },
    {
      "epoch": 3.7657766454102295,
      "grad_norm": 0.7948215007781982,
      "learning_rate": 5.492122505350936e-06,
      "loss": 0.061,
      "step": 255700
    },
    {
      "epoch": 3.7661448285003165,
      "grad_norm": 1.7028142213821411,
      "learning_rate": 5.490486133383952e-06,
      "loss": 0.0559,
      "step": 255725
    },
    {
      "epoch": 3.7665130115904035,
      "grad_norm": 1.1448397636413574,
      "learning_rate": 5.4888497614169675e-06,
      "loss": 0.0504,
      "step": 255750
    },
    {
      "epoch": 3.766881194680491,
      "grad_norm": 1.411699891090393,
      "learning_rate": 5.487213389449984e-06,
      "loss": 0.0575,
      "step": 255775
    },
    {
      "epoch": 3.767249377770578,
      "grad_norm": 1.5449858903884888,
      "learning_rate": 5.485577017482998e-06,
      "loss": 0.052,
      "step": 255800
    },
    {
      "epoch": 3.767617560860665,
      "grad_norm": 1.1064664125442505,
      "learning_rate": 5.483940645516013e-06,
      "loss": 0.0474,
      "step": 255825
    },
    {
      "epoch": 3.767985743950752,
      "grad_norm": 1.473809838294983,
      "learning_rate": 5.4823042735490295e-06,
      "loss": 0.0528,
      "step": 255850
    },
    {
      "epoch": 3.768353927040839,
      "grad_norm": 1.6745480298995972,
      "learning_rate": 5.480667901582045e-06,
      "loss": 0.0589,
      "step": 255875
    },
    {
      "epoch": 3.768722110130926,
      "grad_norm": 1.6598762273788452,
      "learning_rate": 5.47903152961506e-06,
      "loss": 0.0554,
      "step": 255900
    },
    {
      "epoch": 3.769090293221013,
      "grad_norm": 1.3508578538894653,
      "learning_rate": 5.4773951576480755e-06,
      "loss": 0.0585,
      "step": 255925
    },
    {
      "epoch": 3.7694584763111,
      "grad_norm": 1.256160855293274,
      "learning_rate": 5.475758785681092e-06,
      "loss": 0.0509,
      "step": 255950
    },
    {
      "epoch": 3.769826659401187,
      "grad_norm": 1.4047116041183472,
      "learning_rate": 5.474122413714107e-06,
      "loss": 0.0552,
      "step": 255975
    },
    {
      "epoch": 3.770194842491274,
      "grad_norm": 0.8932319283485413,
      "learning_rate": 5.472486041747122e-06,
      "loss": 0.0521,
      "step": 256000
    },
    {
      "epoch": 3.770563025581361,
      "grad_norm": 1.0128567218780518,
      "learning_rate": 5.4708496697801375e-06,
      "loss": 0.053,
      "step": 256025
    },
    {
      "epoch": 3.770931208671448,
      "grad_norm": 1.258533000946045,
      "learning_rate": 5.469213297813153e-06,
      "loss": 0.0471,
      "step": 256050
    },
    {
      "epoch": 3.771299391761535,
      "grad_norm": 1.0562033653259277,
      "learning_rate": 5.467576925846169e-06,
      "loss": 0.052,
      "step": 256075
    },
    {
      "epoch": 3.771667574851622,
      "grad_norm": 1.2327269315719604,
      "learning_rate": 5.465940553879184e-06,
      "loss": 0.0549,
      "step": 256100
    },
    {
      "epoch": 3.772035757941709,
      "grad_norm": 1.4154163599014282,
      "learning_rate": 5.464304181912199e-06,
      "loss": 0.0561,
      "step": 256125
    },
    {
      "epoch": 3.772403941031796,
      "grad_norm": 1.3224729299545288,
      "learning_rate": 5.462667809945215e-06,
      "loss": 0.0526,
      "step": 256150
    },
    {
      "epoch": 3.7727721241218832,
      "grad_norm": 1.5392817258834839,
      "learning_rate": 5.46103143797823e-06,
      "loss": 0.0576,
      "step": 256175
    },
    {
      "epoch": 3.7731403072119702,
      "grad_norm": 1.4855889081954956,
      "learning_rate": 5.459395066011246e-06,
      "loss": 0.057,
      "step": 256200
    },
    {
      "epoch": 3.7735084903020573,
      "grad_norm": 1.3495925664901733,
      "learning_rate": 5.457758694044261e-06,
      "loss": 0.0538,
      "step": 256225
    },
    {
      "epoch": 3.7738766733921443,
      "grad_norm": 1.6467121839523315,
      "learning_rate": 5.456122322077276e-06,
      "loss": 0.0534,
      "step": 256250
    },
    {
      "epoch": 3.7742448564822313,
      "grad_norm": 1.9972038269042969,
      "learning_rate": 5.454485950110292e-06,
      "loss": 0.0595,
      "step": 256275
    },
    {
      "epoch": 3.7746130395723183,
      "grad_norm": 1.1029690504074097,
      "learning_rate": 5.4528495781433076e-06,
      "loss": 0.0505,
      "step": 256300
    },
    {
      "epoch": 3.7749812226624053,
      "grad_norm": 1.0059043169021606,
      "learning_rate": 5.451213206176322e-06,
      "loss": 0.0509,
      "step": 256325
    },
    {
      "epoch": 3.7753494057524923,
      "grad_norm": 0.9450321197509766,
      "learning_rate": 5.449576834209338e-06,
      "loss": 0.0504,
      "step": 256350
    },
    {
      "epoch": 3.77571758884258,
      "grad_norm": 0.9107796549797058,
      "learning_rate": 5.4479404622423535e-06,
      "loss": 0.0505,
      "step": 256375
    },
    {
      "epoch": 3.776085771932667,
      "grad_norm": 1.2592682838439941,
      "learning_rate": 5.44630409027537e-06,
      "loss": 0.0511,
      "step": 256400
    },
    {
      "epoch": 3.776453955022754,
      "grad_norm": 1.16099214553833,
      "learning_rate": 5.444667718308384e-06,
      "loss": 0.0488,
      "step": 256425
    },
    {
      "epoch": 3.776822138112841,
      "grad_norm": 0.9327324628829956,
      "learning_rate": 5.443031346341399e-06,
      "loss": 0.049,
      "step": 256450
    },
    {
      "epoch": 3.777190321202928,
      "grad_norm": 0.9738727807998657,
      "learning_rate": 5.4413949743744155e-06,
      "loss": 0.0494,
      "step": 256475
    },
    {
      "epoch": 3.777558504293015,
      "grad_norm": 1.8228766918182373,
      "learning_rate": 5.439758602407431e-06,
      "loss": 0.0548,
      "step": 256500
    },
    {
      "epoch": 3.777926687383102,
      "grad_norm": 1.2142353057861328,
      "learning_rate": 5.438122230440447e-06,
      "loss": 0.0512,
      "step": 256525
    },
    {
      "epoch": 3.778294870473189,
      "grad_norm": 0.9079768657684326,
      "learning_rate": 5.4364858584734615e-06,
      "loss": 0.0514,
      "step": 256550
    },
    {
      "epoch": 3.778663053563276,
      "grad_norm": 0.9658983945846558,
      "learning_rate": 5.434849486506477e-06,
      "loss": 0.0577,
      "step": 256575
    },
    {
      "epoch": 3.779031236653363,
      "grad_norm": 0.7127007246017456,
      "learning_rate": 5.433213114539493e-06,
      "loss": 0.0494,
      "step": 256600
    },
    {
      "epoch": 3.77939941974345,
      "grad_norm": 1.2438246011734009,
      "learning_rate": 5.431576742572508e-06,
      "loss": 0.0602,
      "step": 256625
    },
    {
      "epoch": 3.779767602833537,
      "grad_norm": 1.516676664352417,
      "learning_rate": 5.4299403706055235e-06,
      "loss": 0.057,
      "step": 256650
    },
    {
      "epoch": 3.780135785923624,
      "grad_norm": 1.3705201148986816,
      "learning_rate": 5.428303998638539e-06,
      "loss": 0.0524,
      "step": 256675
    },
    {
      "epoch": 3.780503969013711,
      "grad_norm": 0.9629042148590088,
      "learning_rate": 5.426667626671554e-06,
      "loss": 0.0547,
      "step": 256700
    },
    {
      "epoch": 3.780872152103798,
      "grad_norm": 1.5228092670440674,
      "learning_rate": 5.42503125470457e-06,
      "loss": 0.0534,
      "step": 256725
    },
    {
      "epoch": 3.7812403351938855,
      "grad_norm": 0.9180556535720825,
      "learning_rate": 5.423460337616264e-06,
      "loss": 0.0454,
      "step": 256750
    },
    {
      "epoch": 3.7816085182839725,
      "grad_norm": 1.1951689720153809,
      "learning_rate": 5.42182396564928e-06,
      "loss": 0.054,
      "step": 256775
    },
    {
      "epoch": 3.7819767013740595,
      "grad_norm": 1.231819748878479,
      "learning_rate": 5.420187593682296e-06,
      "loss": 0.0459,
      "step": 256800
    },
    {
      "epoch": 3.7823448844641465,
      "grad_norm": 1.658528447151184,
      "learning_rate": 5.418551221715311e-06,
      "loss": 0.0583,
      "step": 256825
    },
    {
      "epoch": 3.7827130675542335,
      "grad_norm": 1.0634377002716064,
      "learning_rate": 5.416914849748326e-06,
      "loss": 0.0495,
      "step": 256850
    },
    {
      "epoch": 3.7830812506443205,
      "grad_norm": 1.3042906522750854,
      "learning_rate": 5.4152784777813415e-06,
      "loss": 0.0478,
      "step": 256875
    },
    {
      "epoch": 3.7834494337344076,
      "grad_norm": 1.640803575515747,
      "learning_rate": 5.413642105814358e-06,
      "loss": 0.0506,
      "step": 256900
    },
    {
      "epoch": 3.7838176168244946,
      "grad_norm": 1.1699762344360352,
      "learning_rate": 5.412005733847373e-06,
      "loss": 0.0562,
      "step": 256925
    },
    {
      "epoch": 3.7841857999145816,
      "grad_norm": 1.746739387512207,
      "learning_rate": 5.410369361880389e-06,
      "loss": 0.0497,
      "step": 256950
    },
    {
      "epoch": 3.7845539830046686,
      "grad_norm": 1.1424062252044678,
      "learning_rate": 5.408732989913404e-06,
      "loss": 0.0565,
      "step": 256975
    },
    {
      "epoch": 3.7849221660947556,
      "grad_norm": 1.3724812269210815,
      "learning_rate": 5.407096617946419e-06,
      "loss": 0.0548,
      "step": 257000
    },
    {
      "epoch": 3.7852903491848426,
      "grad_norm": 1.2119040489196777,
      "learning_rate": 5.405460245979435e-06,
      "loss": 0.05,
      "step": 257025
    },
    {
      "epoch": 3.7856585322749297,
      "grad_norm": 1.6405060291290283,
      "learning_rate": 5.40382387401245e-06,
      "loss": 0.0516,
      "step": 257050
    },
    {
      "epoch": 3.7860267153650167,
      "grad_norm": 1.511218547821045,
      "learning_rate": 5.402187502045465e-06,
      "loss": 0.0593,
      "step": 257075
    },
    {
      "epoch": 3.7863948984551037,
      "grad_norm": 1.2260838747024536,
      "learning_rate": 5.400551130078481e-06,
      "loss": 0.0492,
      "step": 257100
    },
    {
      "epoch": 3.7867630815451907,
      "grad_norm": 0.8578277826309204,
      "learning_rate": 5.398914758111496e-06,
      "loss": 0.0527,
      "step": 257125
    },
    {
      "epoch": 3.7871312646352777,
      "grad_norm": 1.2542473077774048,
      "learning_rate": 5.3972783861445124e-06,
      "loss": 0.0541,
      "step": 257150
    },
    {
      "epoch": 3.7874994477253647,
      "grad_norm": 1.1171574592590332,
      "learning_rate": 5.395642014177527e-06,
      "loss": 0.0549,
      "step": 257175
    },
    {
      "epoch": 3.7878676308154517,
      "grad_norm": 1.5778684616088867,
      "learning_rate": 5.394005642210542e-06,
      "loss": 0.0609,
      "step": 257200
    },
    {
      "epoch": 3.7882358139055388,
      "grad_norm": 1.9366440773010254,
      "learning_rate": 5.392369270243558e-06,
      "loss": 0.052,
      "step": 257225
    },
    {
      "epoch": 3.7886039969956258,
      "grad_norm": 1.5419654846191406,
      "learning_rate": 5.390732898276574e-06,
      "loss": 0.0547,
      "step": 257250
    },
    {
      "epoch": 3.788972180085713,
      "grad_norm": 1.533220648765564,
      "learning_rate": 5.389096526309588e-06,
      "loss": 0.0539,
      "step": 257275
    },
    {
      "epoch": 3.7893403631758,
      "grad_norm": 1.63277268409729,
      "learning_rate": 5.387460154342604e-06,
      "loss": 0.0561,
      "step": 257300
    },
    {
      "epoch": 3.789708546265887,
      "grad_norm": 1.1307544708251953,
      "learning_rate": 5.3858237823756196e-06,
      "loss": 0.0546,
      "step": 257325
    },
    {
      "epoch": 3.7900767293559743,
      "grad_norm": 1.6596407890319824,
      "learning_rate": 5.384187410408636e-06,
      "loss": 0.0522,
      "step": 257350
    },
    {
      "epoch": 3.7904449124460613,
      "grad_norm": 1.5582321882247925,
      "learning_rate": 5.38255103844165e-06,
      "loss": 0.0566,
      "step": 257375
    },
    {
      "epoch": 3.7908130955361483,
      "grad_norm": 1.3519046306610107,
      "learning_rate": 5.3809146664746655e-06,
      "loss": 0.0588,
      "step": 257400
    },
    {
      "epoch": 3.7911812786262353,
      "grad_norm": 1.0576746463775635,
      "learning_rate": 5.379278294507682e-06,
      "loss": 0.0539,
      "step": 257425
    },
    {
      "epoch": 3.7915494617163223,
      "grad_norm": 0.9776924848556519,
      "learning_rate": 5.377641922540697e-06,
      "loss": 0.0547,
      "step": 257450
    },
    {
      "epoch": 3.7919176448064094,
      "grad_norm": 1.6325174570083618,
      "learning_rate": 5.376005550573713e-06,
      "loss": 0.0503,
      "step": 257475
    },
    {
      "epoch": 3.7922858278964964,
      "grad_norm": 1.3017586469650269,
      "learning_rate": 5.3743691786067275e-06,
      "loss": 0.0612,
      "step": 257500
    },
    {
      "epoch": 3.7926540109865834,
      "grad_norm": 1.3604521751403809,
      "learning_rate": 5.372732806639744e-06,
      "loss": 0.0495,
      "step": 257525
    },
    {
      "epoch": 3.7930221940766704,
      "grad_norm": 0.9956885576248169,
      "learning_rate": 5.371096434672759e-06,
      "loss": 0.0534,
      "step": 257550
    },
    {
      "epoch": 3.7933903771667574,
      "grad_norm": 1.5488227605819702,
      "learning_rate": 5.369460062705774e-06,
      "loss": 0.0565,
      "step": 257575
    },
    {
      "epoch": 3.7937585602568444,
      "grad_norm": 1.2222161293029785,
      "learning_rate": 5.36782369073879e-06,
      "loss": 0.0543,
      "step": 257600
    },
    {
      "epoch": 3.7941267433469315,
      "grad_norm": 1.633481740951538,
      "learning_rate": 5.366187318771805e-06,
      "loss": 0.0516,
      "step": 257625
    },
    {
      "epoch": 3.7944949264370185,
      "grad_norm": 1.9253498315811157,
      "learning_rate": 5.364550946804821e-06,
      "loss": 0.055,
      "step": 257650
    },
    {
      "epoch": 3.7948631095271055,
      "grad_norm": 1.1269255876541138,
      "learning_rate": 5.362914574837836e-06,
      "loss": 0.0598,
      "step": 257675
    },
    {
      "epoch": 3.795231292617193,
      "grad_norm": 1.2639343738555908,
      "learning_rate": 5.361278202870851e-06,
      "loss": 0.0523,
      "step": 257700
    },
    {
      "epoch": 3.79559947570728,
      "grad_norm": 0.8839380145072937,
      "learning_rate": 5.359641830903867e-06,
      "loss": 0.0578,
      "step": 257725
    },
    {
      "epoch": 3.795967658797367,
      "grad_norm": 1.2928338050842285,
      "learning_rate": 5.358005458936882e-06,
      "loss": 0.0519,
      "step": 257750
    },
    {
      "epoch": 3.796335841887454,
      "grad_norm": 0.8707910776138306,
      "learning_rate": 5.3563690869698984e-06,
      "loss": 0.0553,
      "step": 257775
    },
    {
      "epoch": 3.796704024977541,
      "grad_norm": 1.1992079019546509,
      "learning_rate": 5.354732715002913e-06,
      "loss": 0.054,
      "step": 257800
    },
    {
      "epoch": 3.797072208067628,
      "grad_norm": 2.014716148376465,
      "learning_rate": 5.353096343035928e-06,
      "loss": 0.0556,
      "step": 257825
    },
    {
      "epoch": 3.797440391157715,
      "grad_norm": 1.2445027828216553,
      "learning_rate": 5.351459971068944e-06,
      "loss": 0.0549,
      "step": 257850
    },
    {
      "epoch": 3.797808574247802,
      "grad_norm": 1.348606824874878,
      "learning_rate": 5.34982359910196e-06,
      "loss": 0.0552,
      "step": 257875
    },
    {
      "epoch": 3.798176757337889,
      "grad_norm": 1.4702867269515991,
      "learning_rate": 5.348187227134974e-06,
      "loss": 0.0526,
      "step": 257900
    },
    {
      "epoch": 3.798544940427976,
      "grad_norm": 2.106762409210205,
      "learning_rate": 5.34655085516799e-06,
      "loss": 0.057,
      "step": 257925
    },
    {
      "epoch": 3.798913123518063,
      "grad_norm": 1.7450799942016602,
      "learning_rate": 5.3449144832010056e-06,
      "loss": 0.0539,
      "step": 257950
    },
    {
      "epoch": 3.79928130660815,
      "grad_norm": 1.56754469871521,
      "learning_rate": 5.343278111234022e-06,
      "loss": 0.0484,
      "step": 257975
    },
    {
      "epoch": 3.799649489698237,
      "grad_norm": 1.1302671432495117,
      "learning_rate": 5.341641739267037e-06,
      "loss": 0.0552,
      "step": 258000
    },
    {
      "epoch": 3.800017672788324,
      "grad_norm": 1.7833666801452637,
      "learning_rate": 5.3400053673000515e-06,
      "loss": 0.0575,
      "step": 258025
    },
    {
      "epoch": 3.800385855878411,
      "grad_norm": 1.1021578311920166,
      "learning_rate": 5.338368995333068e-06,
      "loss": 0.059,
      "step": 258050
    },
    {
      "epoch": 3.800754038968498,
      "grad_norm": 1.235292673110962,
      "learning_rate": 5.336732623366083e-06,
      "loss": 0.0503,
      "step": 258075
    },
    {
      "epoch": 3.801122222058585,
      "grad_norm": 1.4408515691757202,
      "learning_rate": 5.335096251399099e-06,
      "loss": 0.0605,
      "step": 258100
    },
    {
      "epoch": 3.801490405148672,
      "grad_norm": 1.5457375049591064,
      "learning_rate": 5.3334598794321135e-06,
      "loss": 0.0512,
      "step": 258125
    },
    {
      "epoch": 3.801858588238759,
      "grad_norm": 1.3230416774749756,
      "learning_rate": 5.331823507465129e-06,
      "loss": 0.0516,
      "step": 258150
    },
    {
      "epoch": 3.8022267713288462,
      "grad_norm": 0.7777018547058105,
      "learning_rate": 5.330187135498145e-06,
      "loss": 0.0459,
      "step": 258175
    },
    {
      "epoch": 3.8025949544189332,
      "grad_norm": 1.3024753332138062,
      "learning_rate": 5.32855076353116e-06,
      "loss": 0.0559,
      "step": 258200
    },
    {
      "epoch": 3.8029631375090203,
      "grad_norm": 1.4986072778701782,
      "learning_rate": 5.326914391564176e-06,
      "loss": 0.058,
      "step": 258225
    },
    {
      "epoch": 3.8033313205991073,
      "grad_norm": 1.0987248420715332,
      "learning_rate": 5.325278019597191e-06,
      "loss": 0.052,
      "step": 258250
    },
    {
      "epoch": 3.8036995036891943,
      "grad_norm": 1.2068313360214233,
      "learning_rate": 5.323641647630207e-06,
      "loss": 0.0524,
      "step": 258275
    },
    {
      "epoch": 3.8040676867792818,
      "grad_norm": 1.3169461488723755,
      "learning_rate": 5.322005275663222e-06,
      "loss": 0.0561,
      "step": 258300
    },
    {
      "epoch": 3.8044358698693688,
      "grad_norm": 0.6551405191421509,
      "learning_rate": 5.320368903696237e-06,
      "loss": 0.0533,
      "step": 258325
    },
    {
      "epoch": 3.804804052959456,
      "grad_norm": 1.0101947784423828,
      "learning_rate": 5.318732531729253e-06,
      "loss": 0.0532,
      "step": 258350
    },
    {
      "epoch": 3.805172236049543,
      "grad_norm": 0.8900593519210815,
      "learning_rate": 5.317096159762268e-06,
      "loss": 0.0472,
      "step": 258375
    },
    {
      "epoch": 3.80554041913963,
      "grad_norm": 1.1752017736434937,
      "learning_rate": 5.315459787795284e-06,
      "loss": 0.0559,
      "step": 258400
    },
    {
      "epoch": 3.805908602229717,
      "grad_norm": 1.813315987586975,
      "learning_rate": 5.3138234158283e-06,
      "loss": 0.0616,
      "step": 258425
    },
    {
      "epoch": 3.806276785319804,
      "grad_norm": 1.255013108253479,
      "learning_rate": 5.312187043861314e-06,
      "loss": 0.0555,
      "step": 258450
    },
    {
      "epoch": 3.806644968409891,
      "grad_norm": 1.2944269180297852,
      "learning_rate": 5.31055067189433e-06,
      "loss": 0.0603,
      "step": 258475
    },
    {
      "epoch": 3.807013151499978,
      "grad_norm": 1.3191167116165161,
      "learning_rate": 5.308914299927346e-06,
      "loss": 0.0504,
      "step": 258500
    },
    {
      "epoch": 3.807381334590065,
      "grad_norm": 1.7232218980789185,
      "learning_rate": 5.307277927960362e-06,
      "loss": 0.0567,
      "step": 258525
    },
    {
      "epoch": 3.807749517680152,
      "grad_norm": 1.3582383394241333,
      "learning_rate": 5.305641555993376e-06,
      "loss": 0.0576,
      "step": 258550
    },
    {
      "epoch": 3.808117700770239,
      "grad_norm": 1.328294038772583,
      "learning_rate": 5.3040051840263916e-06,
      "loss": 0.0601,
      "step": 258575
    },
    {
      "epoch": 3.808485883860326,
      "grad_norm": 1.5954055786132812,
      "learning_rate": 5.302368812059408e-06,
      "loss": 0.0512,
      "step": 258600
    },
    {
      "epoch": 3.808854066950413,
      "grad_norm": 1.3190314769744873,
      "learning_rate": 5.300732440092423e-06,
      "loss": 0.0492,
      "step": 258625
    },
    {
      "epoch": 3.8092222500405004,
      "grad_norm": 1.1978700160980225,
      "learning_rate": 5.2990960681254375e-06,
      "loss": 0.0588,
      "step": 258650
    },
    {
      "epoch": 3.8095904331305874,
      "grad_norm": 1.455289363861084,
      "learning_rate": 5.297459696158454e-06,
      "loss": 0.0479,
      "step": 258675
    },
    {
      "epoch": 3.8099586162206744,
      "grad_norm": 1.4614241123199463,
      "learning_rate": 5.295823324191469e-06,
      "loss": 0.0562,
      "step": 258700
    },
    {
      "epoch": 3.8103267993107615,
      "grad_norm": 1.2461235523223877,
      "learning_rate": 5.294186952224485e-06,
      "loss": 0.0497,
      "step": 258725
    },
    {
      "epoch": 3.8106949824008485,
      "grad_norm": 1.2991175651550293,
      "learning_rate": 5.2925505802574995e-06,
      "loss": 0.0614,
      "step": 258750
    },
    {
      "epoch": 3.8110631654909355,
      "grad_norm": 1.3581284284591675,
      "learning_rate": 5.290914208290515e-06,
      "loss": 0.0553,
      "step": 258775
    },
    {
      "epoch": 3.8114313485810225,
      "grad_norm": 1.3191865682601929,
      "learning_rate": 5.289277836323531e-06,
      "loss": 0.0487,
      "step": 258800
    },
    {
      "epoch": 3.8117995316711095,
      "grad_norm": 1.309084415435791,
      "learning_rate": 5.287641464356546e-06,
      "loss": 0.0535,
      "step": 258825
    },
    {
      "epoch": 3.8121677147611965,
      "grad_norm": 1.2470643520355225,
      "learning_rate": 5.286005092389561e-06,
      "loss": 0.0565,
      "step": 258850
    },
    {
      "epoch": 3.8125358978512836,
      "grad_norm": 1.290117859840393,
      "learning_rate": 5.284368720422577e-06,
      "loss": 0.0492,
      "step": 258875
    },
    {
      "epoch": 3.8129040809413706,
      "grad_norm": 1.2575085163116455,
      "learning_rate": 5.282732348455592e-06,
      "loss": 0.0467,
      "step": 258900
    },
    {
      "epoch": 3.8132722640314576,
      "grad_norm": 1.0561531782150269,
      "learning_rate": 5.281095976488608e-06,
      "loss": 0.0465,
      "step": 258925
    },
    {
      "epoch": 3.8136404471215446,
      "grad_norm": 1.3675233125686646,
      "learning_rate": 5.279459604521624e-06,
      "loss": 0.0576,
      "step": 258950
    },
    {
      "epoch": 3.8140086302116316,
      "grad_norm": 1.7130180597305298,
      "learning_rate": 5.277823232554639e-06,
      "loss": 0.0512,
      "step": 258975
    },
    {
      "epoch": 3.8143768133017186,
      "grad_norm": 1.4887468814849854,
      "learning_rate": 5.276186860587654e-06,
      "loss": 0.0585,
      "step": 259000
    },
    {
      "epoch": 3.8147449963918056,
      "grad_norm": 1.6468786001205444,
      "learning_rate": 5.2745504886206696e-06,
      "loss": 0.0577,
      "step": 259025
    },
    {
      "epoch": 3.8151131794818927,
      "grad_norm": 1.610073447227478,
      "learning_rate": 5.272914116653686e-06,
      "loss": 0.0531,
      "step": 259050
    },
    {
      "epoch": 3.8154813625719797,
      "grad_norm": 1.2798658609390259,
      "learning_rate": 5.2712777446867e-06,
      "loss": 0.0537,
      "step": 259075
    },
    {
      "epoch": 3.8158495456620667,
      "grad_norm": 1.083975911140442,
      "learning_rate": 5.269706827598396e-06,
      "loss": 0.0554,
      "step": 259100
    },
    {
      "epoch": 3.8162177287521537,
      "grad_norm": 1.5547819137573242,
      "learning_rate": 5.268070455631411e-06,
      "loss": 0.0537,
      "step": 259125
    },
    {
      "epoch": 3.8165859118422407,
      "grad_norm": 1.2542880773544312,
      "learning_rate": 5.266434083664426e-06,
      "loss": 0.0545,
      "step": 259150
    },
    {
      "epoch": 3.8169540949323277,
      "grad_norm": 1.0080958604812622,
      "learning_rate": 5.264797711697442e-06,
      "loss": 0.0548,
      "step": 259175
    },
    {
      "epoch": 3.8173222780224147,
      "grad_norm": 1.3396576642990112,
      "learning_rate": 5.263161339730457e-06,
      "loss": 0.0487,
      "step": 259200
    },
    {
      "epoch": 3.8176904611125018,
      "grad_norm": 1.5910086631774902,
      "learning_rate": 5.261524967763473e-06,
      "loss": 0.0472,
      "step": 259225
    },
    {
      "epoch": 3.8180586442025892,
      "grad_norm": 1.5560002326965332,
      "learning_rate": 5.2598885957964884e-06,
      "loss": 0.0481,
      "step": 259250
    },
    {
      "epoch": 3.8184268272926762,
      "grad_norm": 1.1967967748641968,
      "learning_rate": 5.258252223829503e-06,
      "loss": 0.0675,
      "step": 259275
    },
    {
      "epoch": 3.8187950103827633,
      "grad_norm": 1.1806341409683228,
      "learning_rate": 5.256615851862519e-06,
      "loss": 0.0596,
      "step": 259300
    },
    {
      "epoch": 3.8191631934728503,
      "grad_norm": 1.0007059574127197,
      "learning_rate": 5.254979479895534e-06,
      "loss": 0.0547,
      "step": 259325
    },
    {
      "epoch": 3.8195313765629373,
      "grad_norm": 1.5026322603225708,
      "learning_rate": 5.2533431079285505e-06,
      "loss": 0.053,
      "step": 259350
    },
    {
      "epoch": 3.8198995596530243,
      "grad_norm": 2.126260280609131,
      "learning_rate": 5.251706735961565e-06,
      "loss": 0.0604,
      "step": 259375
    },
    {
      "epoch": 3.8202677427431113,
      "grad_norm": 1.415503740310669,
      "learning_rate": 5.25007036399458e-06,
      "loss": 0.0551,
      "step": 259400
    },
    {
      "epoch": 3.8206359258331983,
      "grad_norm": 1.8415940999984741,
      "learning_rate": 5.248433992027596e-06,
      "loss": 0.0535,
      "step": 259425
    },
    {
      "epoch": 3.8210041089232853,
      "grad_norm": 1.6008800268173218,
      "learning_rate": 5.246797620060612e-06,
      "loss": 0.0547,
      "step": 259450
    },
    {
      "epoch": 3.8213722920133724,
      "grad_norm": 1.4544116258621216,
      "learning_rate": 5.245161248093628e-06,
      "loss": 0.0492,
      "step": 259475
    },
    {
      "epoch": 3.8217404751034594,
      "grad_norm": 0.896043062210083,
      "learning_rate": 5.243524876126642e-06,
      "loss": 0.0548,
      "step": 259500
    },
    {
      "epoch": 3.8221086581935464,
      "grad_norm": 1.6785802841186523,
      "learning_rate": 5.241888504159658e-06,
      "loss": 0.0565,
      "step": 259525
    },
    {
      "epoch": 3.8224768412836334,
      "grad_norm": 0.8544201254844666,
      "learning_rate": 5.240252132192674e-06,
      "loss": 0.0536,
      "step": 259550
    },
    {
      "epoch": 3.8228450243737204,
      "grad_norm": 1.707797884941101,
      "learning_rate": 5.238615760225689e-06,
      "loss": 0.0525,
      "step": 259575
    },
    {
      "epoch": 3.8232132074638074,
      "grad_norm": 1.360618233680725,
      "learning_rate": 5.2369793882587035e-06,
      "loss": 0.0506,
      "step": 259600
    },
    {
      "epoch": 3.823581390553895,
      "grad_norm": 1.5496617555618286,
      "learning_rate": 5.23534301629172e-06,
      "loss": 0.0517,
      "step": 259625
    },
    {
      "epoch": 3.823949573643982,
      "grad_norm": 1.1844253540039062,
      "learning_rate": 5.233706644324735e-06,
      "loss": 0.0532,
      "step": 259650
    },
    {
      "epoch": 3.824317756734069,
      "grad_norm": 1.2134591341018677,
      "learning_rate": 5.232070272357751e-06,
      "loss": 0.0538,
      "step": 259675
    },
    {
      "epoch": 3.824685939824156,
      "grad_norm": 1.4161320924758911,
      "learning_rate": 5.230433900390766e-06,
      "loss": 0.0616,
      "step": 259700
    },
    {
      "epoch": 3.825054122914243,
      "grad_norm": 1.357408881187439,
      "learning_rate": 5.228797528423781e-06,
      "loss": 0.055,
      "step": 259725
    },
    {
      "epoch": 3.82542230600433,
      "grad_norm": 1.6354411840438843,
      "learning_rate": 5.227161156456797e-06,
      "loss": 0.0621,
      "step": 259750
    },
    {
      "epoch": 3.825790489094417,
      "grad_norm": 1.4002183675765991,
      "learning_rate": 5.225524784489812e-06,
      "loss": 0.0521,
      "step": 259775
    },
    {
      "epoch": 3.826158672184504,
      "grad_norm": 1.368042230606079,
      "learning_rate": 5.223888412522828e-06,
      "loss": 0.0456,
      "step": 259800
    },
    {
      "epoch": 3.826526855274591,
      "grad_norm": 1.519740104675293,
      "learning_rate": 5.222252040555843e-06,
      "loss": 0.0555,
      "step": 259825
    },
    {
      "epoch": 3.826895038364678,
      "grad_norm": 1.3023755550384521,
      "learning_rate": 5.220615668588859e-06,
      "loss": 0.0566,
      "step": 259850
    },
    {
      "epoch": 3.827263221454765,
      "grad_norm": 1.1263644695281982,
      "learning_rate": 5.2189792966218744e-06,
      "loss": 0.0544,
      "step": 259875
    },
    {
      "epoch": 3.827631404544852,
      "grad_norm": 1.0691365003585815,
      "learning_rate": 5.21734292465489e-06,
      "loss": 0.0468,
      "step": 259900
    },
    {
      "epoch": 3.827999587634939,
      "grad_norm": 1.4326441287994385,
      "learning_rate": 5.215706552687905e-06,
      "loss": 0.0489,
      "step": 259925
    },
    {
      "epoch": 3.828367770725026,
      "grad_norm": 1.323684573173523,
      "learning_rate": 5.21407018072092e-06,
      "loss": 0.0589,
      "step": 259950
    },
    {
      "epoch": 3.828735953815113,
      "grad_norm": 1.7488555908203125,
      "learning_rate": 5.2124338087539365e-06,
      "loss": 0.0552,
      "step": 259975
    },
    {
      "epoch": 3.8291041369052,
      "grad_norm": 1.8526763916015625,
      "learning_rate": 5.210797436786952e-06,
      "loss": 0.0521,
      "step": 260000
    },
    {
      "epoch": 3.829472319995287,
      "grad_norm": 1.5918937921524048,
      "learning_rate": 5.209161064819966e-06,
      "loss": 0.0545,
      "step": 260025
    },
    {
      "epoch": 3.829840503085374,
      "grad_norm": 1.4427952766418457,
      "learning_rate": 5.207524692852982e-06,
      "loss": 0.05,
      "step": 260050
    },
    {
      "epoch": 3.830208686175461,
      "grad_norm": 1.1258643865585327,
      "learning_rate": 5.205888320885998e-06,
      "loss": 0.0542,
      "step": 260075
    },
    {
      "epoch": 3.830576869265548,
      "grad_norm": 1.606545329093933,
      "learning_rate": 5.204251948919014e-06,
      "loss": 0.0619,
      "step": 260100
    },
    {
      "epoch": 3.830945052355635,
      "grad_norm": 1.4542641639709473,
      "learning_rate": 5.202615576952028e-06,
      "loss": 0.0513,
      "step": 260125
    },
    {
      "epoch": 3.831313235445722,
      "grad_norm": 1.8232176303863525,
      "learning_rate": 5.200979204985044e-06,
      "loss": 0.0529,
      "step": 260150
    },
    {
      "epoch": 3.8316814185358092,
      "grad_norm": 1.6592656373977661,
      "learning_rate": 5.19934283301806e-06,
      "loss": 0.0504,
      "step": 260175
    },
    {
      "epoch": 3.8320496016258963,
      "grad_norm": 1.9837111234664917,
      "learning_rate": 5.197706461051075e-06,
      "loss": 0.0633,
      "step": 260200
    },
    {
      "epoch": 3.8324177847159837,
      "grad_norm": 1.306334376335144,
      "learning_rate": 5.1960700890840895e-06,
      "loss": 0.0496,
      "step": 260225
    },
    {
      "epoch": 3.8327859678060707,
      "grad_norm": 1.4030455350875854,
      "learning_rate": 5.194433717117106e-06,
      "loss": 0.0469,
      "step": 260250
    },
    {
      "epoch": 3.8331541508961577,
      "grad_norm": 1.6958765983581543,
      "learning_rate": 5.192797345150121e-06,
      "loss": 0.0579,
      "step": 260275
    },
    {
      "epoch": 3.8335223339862448,
      "grad_norm": 1.386186957359314,
      "learning_rate": 5.191160973183137e-06,
      "loss": 0.0532,
      "step": 260300
    },
    {
      "epoch": 3.8338905170763318,
      "grad_norm": 1.8941980600357056,
      "learning_rate": 5.189524601216152e-06,
      "loss": 0.0579,
      "step": 260325
    },
    {
      "epoch": 3.834258700166419,
      "grad_norm": 1.5502009391784668,
      "learning_rate": 5.187888229249167e-06,
      "loss": 0.0561,
      "step": 260350
    },
    {
      "epoch": 3.834626883256506,
      "grad_norm": 1.8309978246688843,
      "learning_rate": 5.186251857282183e-06,
      "loss": 0.0554,
      "step": 260375
    },
    {
      "epoch": 3.834995066346593,
      "grad_norm": 1.4434939622879028,
      "learning_rate": 5.184615485315198e-06,
      "loss": 0.0545,
      "step": 260400
    },
    {
      "epoch": 3.83536324943668,
      "grad_norm": 0.6179027557373047,
      "learning_rate": 5.1829791133482145e-06,
      "loss": 0.0498,
      "step": 260425
    },
    {
      "epoch": 3.835731432526767,
      "grad_norm": 1.7048661708831787,
      "learning_rate": 5.181342741381229e-06,
      "loss": 0.0601,
      "step": 260450
    },
    {
      "epoch": 3.836099615616854,
      "grad_norm": 1.4048082828521729,
      "learning_rate": 5.179706369414244e-06,
      "loss": 0.0586,
      "step": 260475
    },
    {
      "epoch": 3.836467798706941,
      "grad_norm": 1.4202415943145752,
      "learning_rate": 5.1780699974472604e-06,
      "loss": 0.0634,
      "step": 260500
    },
    {
      "epoch": 3.836835981797028,
      "grad_norm": 1.163485050201416,
      "learning_rate": 5.176433625480276e-06,
      "loss": 0.0556,
      "step": 260525
    },
    {
      "epoch": 3.837204164887115,
      "grad_norm": 1.1400727033615112,
      "learning_rate": 5.174797253513291e-06,
      "loss": 0.0541,
      "step": 260550
    },
    {
      "epoch": 3.8375723479772024,
      "grad_norm": 1.8711127042770386,
      "learning_rate": 5.173160881546306e-06,
      "loss": 0.0542,
      "step": 260575
    },
    {
      "epoch": 3.8379405310672894,
      "grad_norm": 1.0933290719985962,
      "learning_rate": 5.171524509579322e-06,
      "loss": 0.0541,
      "step": 260600
    },
    {
      "epoch": 3.8383087141573764,
      "grad_norm": 1.738000512123108,
      "learning_rate": 5.169888137612338e-06,
      "loss": 0.0534,
      "step": 260625
    },
    {
      "epoch": 3.8386768972474634,
      "grad_norm": 1.521758794784546,
      "learning_rate": 5.168251765645352e-06,
      "loss": 0.0523,
      "step": 260650
    },
    {
      "epoch": 3.8390450803375504,
      "grad_norm": 1.4863930940628052,
      "learning_rate": 5.166615393678368e-06,
      "loss": 0.0525,
      "step": 260675
    },
    {
      "epoch": 3.8394132634276374,
      "grad_norm": 1.4816769361495972,
      "learning_rate": 5.164979021711384e-06,
      "loss": 0.0519,
      "step": 260700
    },
    {
      "epoch": 3.8397814465177245,
      "grad_norm": 1.8269590139389038,
      "learning_rate": 5.1633426497444e-06,
      "loss": 0.0539,
      "step": 260725
    },
    {
      "epoch": 3.8401496296078115,
      "grad_norm": 1.7181164026260376,
      "learning_rate": 5.161706277777414e-06,
      "loss": 0.0513,
      "step": 260750
    },
    {
      "epoch": 3.8405178126978985,
      "grad_norm": 1.3441529273986816,
      "learning_rate": 5.16006990581043e-06,
      "loss": 0.0532,
      "step": 260775
    },
    {
      "epoch": 3.8408859957879855,
      "grad_norm": 1.526574730873108,
      "learning_rate": 5.158433533843446e-06,
      "loss": 0.053,
      "step": 260800
    },
    {
      "epoch": 3.8412541788780725,
      "grad_norm": 0.9288049936294556,
      "learning_rate": 5.156797161876461e-06,
      "loss": 0.0479,
      "step": 260825
    },
    {
      "epoch": 3.8416223619681595,
      "grad_norm": 0.9120572209358215,
      "learning_rate": 5.1551607899094755e-06,
      "loss": 0.0524,
      "step": 260850
    },
    {
      "epoch": 3.8419905450582466,
      "grad_norm": 1.4088650941848755,
      "learning_rate": 5.153524417942492e-06,
      "loss": 0.0588,
      "step": 260875
    },
    {
      "epoch": 3.8423587281483336,
      "grad_norm": 1.0027408599853516,
      "learning_rate": 5.151888045975507e-06,
      "loss": 0.0625,
      "step": 260900
    },
    {
      "epoch": 3.8427269112384206,
      "grad_norm": 1.4070595502853394,
      "learning_rate": 5.150251674008523e-06,
      "loss": 0.0522,
      "step": 260925
    },
    {
      "epoch": 3.8430950943285076,
      "grad_norm": 0.9803788661956787,
      "learning_rate": 5.1486153020415384e-06,
      "loss": 0.0584,
      "step": 260950
    },
    {
      "epoch": 3.8434632774185946,
      "grad_norm": 1.607533574104309,
      "learning_rate": 5.146978930074553e-06,
      "loss": 0.0547,
      "step": 260975
    },
    {
      "epoch": 3.8438314605086816,
      "grad_norm": 1.5357824563980103,
      "learning_rate": 5.145342558107569e-06,
      "loss": 0.0584,
      "step": 261000
    },
    {
      "epoch": 3.8441996435987686,
      "grad_norm": 1.1385902166366577,
      "learning_rate": 5.143706186140584e-06,
      "loss": 0.0518,
      "step": 261025
    },
    {
      "epoch": 3.8445678266888557,
      "grad_norm": 2.078275680541992,
      "learning_rate": 5.1420698141736005e-06,
      "loss": 0.0593,
      "step": 261050
    },
    {
      "epoch": 3.8449360097789427,
      "grad_norm": 1.5946872234344482,
      "learning_rate": 5.140433442206615e-06,
      "loss": 0.0513,
      "step": 261075
    },
    {
      "epoch": 3.8453041928690297,
      "grad_norm": 1.8258835077285767,
      "learning_rate": 5.13879707023963e-06,
      "loss": 0.0475,
      "step": 261100
    },
    {
      "epoch": 3.8456723759591167,
      "grad_norm": 0.797872006893158,
      "learning_rate": 5.1371606982726464e-06,
      "loss": 0.0485,
      "step": 261125
    },
    {
      "epoch": 3.8460405590492037,
      "grad_norm": 1.5468593835830688,
      "learning_rate": 5.135524326305662e-06,
      "loss": 0.0551,
      "step": 261150
    },
    {
      "epoch": 3.846408742139291,
      "grad_norm": 0.8762492537498474,
      "learning_rate": 5.133887954338676e-06,
      "loss": 0.0548,
      "step": 261175
    },
    {
      "epoch": 3.846776925229378,
      "grad_norm": 1.2396718263626099,
      "learning_rate": 5.132251582371692e-06,
      "loss": 0.0543,
      "step": 261200
    },
    {
      "epoch": 3.847145108319465,
      "grad_norm": 1.6593016386032104,
      "learning_rate": 5.130615210404708e-06,
      "loss": 0.0505,
      "step": 261225
    },
    {
      "epoch": 3.8475132914095522,
      "grad_norm": 1.3239442110061646,
      "learning_rate": 5.128978838437724e-06,
      "loss": 0.0545,
      "step": 261250
    },
    {
      "epoch": 3.8478814744996392,
      "grad_norm": 1.420785903930664,
      "learning_rate": 5.127342466470738e-06,
      "loss": 0.0545,
      "step": 261275
    },
    {
      "epoch": 3.8482496575897263,
      "grad_norm": 1.5579971075057983,
      "learning_rate": 5.125706094503754e-06,
      "loss": 0.047,
      "step": 261300
    },
    {
      "epoch": 3.8486178406798133,
      "grad_norm": 1.2454949617385864,
      "learning_rate": 5.12406972253677e-06,
      "loss": 0.0555,
      "step": 261325
    },
    {
      "epoch": 3.8489860237699003,
      "grad_norm": 1.3548647165298462,
      "learning_rate": 5.122433350569785e-06,
      "loss": 0.0551,
      "step": 261350
    },
    {
      "epoch": 3.8493542068599873,
      "grad_norm": 1.1681543588638306,
      "learning_rate": 5.120796978602801e-06,
      "loss": 0.0557,
      "step": 261375
    },
    {
      "epoch": 3.8497223899500743,
      "grad_norm": 1.1274433135986328,
      "learning_rate": 5.119160606635816e-06,
      "loss": 0.0525,
      "step": 261400
    },
    {
      "epoch": 3.8500905730401613,
      "grad_norm": 1.9322034120559692,
      "learning_rate": 5.117524234668832e-06,
      "loss": 0.0577,
      "step": 261425
    },
    {
      "epoch": 3.8504587561302484,
      "grad_norm": 1.1695538759231567,
      "learning_rate": 5.115887862701847e-06,
      "loss": 0.0544,
      "step": 261450
    },
    {
      "epoch": 3.8508269392203354,
      "grad_norm": 1.565733790397644,
      "learning_rate": 5.114251490734863e-06,
      "loss": 0.0507,
      "step": 261475
    },
    {
      "epoch": 3.8511951223104224,
      "grad_norm": 1.5190887451171875,
      "learning_rate": 5.112615118767878e-06,
      "loss": 0.0532,
      "step": 261500
    },
    {
      "epoch": 3.85156330540051,
      "grad_norm": 1.5840818881988525,
      "learning_rate": 5.110978746800893e-06,
      "loss": 0.0562,
      "step": 261525
    },
    {
      "epoch": 3.851931488490597,
      "grad_norm": 2.0657613277435303,
      "learning_rate": 5.109342374833909e-06,
      "loss": 0.0574,
      "step": 261550
    },
    {
      "epoch": 3.852299671580684,
      "grad_norm": 1.2047373056411743,
      "learning_rate": 5.1077060028669244e-06,
      "loss": 0.0537,
      "step": 261575
    },
    {
      "epoch": 3.852667854670771,
      "grad_norm": 1.5429242849349976,
      "learning_rate": 5.106069630899939e-06,
      "loss": 0.0486,
      "step": 261600
    },
    {
      "epoch": 3.853036037760858,
      "grad_norm": 1.3733842372894287,
      "learning_rate": 5.104433258932955e-06,
      "loss": 0.0576,
      "step": 261625
    },
    {
      "epoch": 3.853404220850945,
      "grad_norm": 0.98934406042099,
      "learning_rate": 5.10279688696597e-06,
      "loss": 0.0474,
      "step": 261650
    },
    {
      "epoch": 3.853772403941032,
      "grad_norm": 1.2279473543167114,
      "learning_rate": 5.1011605149989865e-06,
      "loss": 0.0564,
      "step": 261675
    },
    {
      "epoch": 3.854140587031119,
      "grad_norm": 1.2108162641525269,
      "learning_rate": 5.099524143032001e-06,
      "loss": 0.0472,
      "step": 261700
    },
    {
      "epoch": 3.854508770121206,
      "grad_norm": 1.415026307106018,
      "learning_rate": 5.097887771065016e-06,
      "loss": 0.0558,
      "step": 261725
    },
    {
      "epoch": 3.854876953211293,
      "grad_norm": 1.457869291305542,
      "learning_rate": 5.0962513990980324e-06,
      "loss": 0.0556,
      "step": 261750
    },
    {
      "epoch": 3.85524513630138,
      "grad_norm": 1.6702696084976196,
      "learning_rate": 5.094615027131048e-06,
      "loss": 0.0513,
      "step": 261775
    },
    {
      "epoch": 3.855613319391467,
      "grad_norm": 1.2122410535812378,
      "learning_rate": 5.092978655164062e-06,
      "loss": 0.0528,
      "step": 261800
    },
    {
      "epoch": 3.855981502481554,
      "grad_norm": 1.269836664199829,
      "learning_rate": 5.091342283197078e-06,
      "loss": 0.0487,
      "step": 261825
    },
    {
      "epoch": 3.856349685571641,
      "grad_norm": 1.2878093719482422,
      "learning_rate": 5.089705911230094e-06,
      "loss": 0.046,
      "step": 261850
    },
    {
      "epoch": 3.856717868661728,
      "grad_norm": 1.4793931245803833,
      "learning_rate": 5.08806953926311e-06,
      "loss": 0.0509,
      "step": 261875
    },
    {
      "epoch": 3.857086051751815,
      "grad_norm": 1.5541657209396362,
      "learning_rate": 5.086433167296125e-06,
      "loss": 0.057,
      "step": 261900
    },
    {
      "epoch": 3.857454234841902,
      "grad_norm": 1.2834160327911377,
      "learning_rate": 5.0847967953291396e-06,
      "loss": 0.0547,
      "step": 261925
    },
    {
      "epoch": 3.857822417931989,
      "grad_norm": 1.3856425285339355,
      "learning_rate": 5.083160423362156e-06,
      "loss": 0.0525,
      "step": 261950
    },
    {
      "epoch": 3.858190601022076,
      "grad_norm": 1.2109434604644775,
      "learning_rate": 5.081524051395171e-06,
      "loss": 0.0509,
      "step": 261975
    },
    {
      "epoch": 3.858558784112163,
      "grad_norm": 1.5357753038406372,
      "learning_rate": 5.079887679428187e-06,
      "loss": 0.0554,
      "step": 262000
    },
    {
      "epoch": 3.85892696720225,
      "grad_norm": 1.0928494930267334,
      "learning_rate": 5.078251307461202e-06,
      "loss": 0.0502,
      "step": 262025
    },
    {
      "epoch": 3.859295150292337,
      "grad_norm": 1.6509815454483032,
      "learning_rate": 5.076680390372896e-06,
      "loss": 0.0555,
      "step": 262050
    },
    {
      "epoch": 3.859663333382424,
      "grad_norm": 1.2722944021224976,
      "learning_rate": 5.0750440184059125e-06,
      "loss": 0.0559,
      "step": 262075
    },
    {
      "epoch": 3.860031516472511,
      "grad_norm": 1.725926399230957,
      "learning_rate": 5.073407646438928e-06,
      "loss": 0.0572,
      "step": 262100
    },
    {
      "epoch": 3.8603996995625987,
      "grad_norm": 1.0288758277893066,
      "learning_rate": 5.071771274471943e-06,
      "loss": 0.0526,
      "step": 262125
    },
    {
      "epoch": 3.8607678826526857,
      "grad_norm": 1.2656705379486084,
      "learning_rate": 5.070134902504958e-06,
      "loss": 0.0496,
      "step": 262150
    },
    {
      "epoch": 3.8611360657427727,
      "grad_norm": 1.0348396301269531,
      "learning_rate": 5.068498530537974e-06,
      "loss": 0.0495,
      "step": 262175
    },
    {
      "epoch": 3.8615042488328597,
      "grad_norm": 1.073555827140808,
      "learning_rate": 5.06686215857099e-06,
      "loss": 0.0506,
      "step": 262200
    },
    {
      "epoch": 3.8618724319229467,
      "grad_norm": 1.2936421632766724,
      "learning_rate": 5.065225786604004e-06,
      "loss": 0.0522,
      "step": 262225
    },
    {
      "epoch": 3.8622406150130337,
      "grad_norm": 1.5264995098114014,
      "learning_rate": 5.0635894146370205e-06,
      "loss": 0.0502,
      "step": 262250
    },
    {
      "epoch": 3.8626087981031207,
      "grad_norm": 1.0455107688903809,
      "learning_rate": 5.061953042670036e-06,
      "loss": 0.0585,
      "step": 262275
    },
    {
      "epoch": 3.8629769811932078,
      "grad_norm": 1.1604970693588257,
      "learning_rate": 5.060316670703052e-06,
      "loss": 0.0518,
      "step": 262300
    },
    {
      "epoch": 3.8633451642832948,
      "grad_norm": 1.308096170425415,
      "learning_rate": 5.058680298736067e-06,
      "loss": 0.0488,
      "step": 262325
    },
    {
      "epoch": 3.863713347373382,
      "grad_norm": 0.7083810567855835,
      "learning_rate": 5.057043926769082e-06,
      "loss": 0.0527,
      "step": 262350
    },
    {
      "epoch": 3.864081530463469,
      "grad_norm": 1.0349743366241455,
      "learning_rate": 5.055407554802098e-06,
      "loss": 0.0579,
      "step": 262375
    },
    {
      "epoch": 3.864449713553556,
      "grad_norm": 1.4834082126617432,
      "learning_rate": 5.053771182835113e-06,
      "loss": 0.0609,
      "step": 262400
    },
    {
      "epoch": 3.864817896643643,
      "grad_norm": 1.3020261526107788,
      "learning_rate": 5.052134810868129e-06,
      "loss": 0.0538,
      "step": 262425
    },
    {
      "epoch": 3.86518607973373,
      "grad_norm": 1.718116283416748,
      "learning_rate": 5.050498438901144e-06,
      "loss": 0.0554,
      "step": 262450
    },
    {
      "epoch": 3.865554262823817,
      "grad_norm": 1.4437057971954346,
      "learning_rate": 5.048862066934159e-06,
      "loss": 0.0605,
      "step": 262475
    },
    {
      "epoch": 3.8659224459139043,
      "grad_norm": 1.3646372556686401,
      "learning_rate": 5.047225694967175e-06,
      "loss": 0.0439,
      "step": 262500
    },
    {
      "epoch": 3.8662906290039913,
      "grad_norm": 1.2433618307113647,
      "learning_rate": 5.0455893230001905e-06,
      "loss": 0.051,
      "step": 262525
    },
    {
      "epoch": 3.8666588120940784,
      "grad_norm": 1.3432698249816895,
      "learning_rate": 5.043952951033205e-06,
      "loss": 0.0494,
      "step": 262550
    },
    {
      "epoch": 3.8670269951841654,
      "grad_norm": 1.52839994430542,
      "learning_rate": 5.042316579066221e-06,
      "loss": 0.057,
      "step": 262575
    },
    {
      "epoch": 3.8673951782742524,
      "grad_norm": 0.7989602088928223,
      "learning_rate": 5.0406802070992364e-06,
      "loss": 0.0539,
      "step": 262600
    },
    {
      "epoch": 3.8677633613643394,
      "grad_norm": 1.0238368511199951,
      "learning_rate": 5.039043835132253e-06,
      "loss": 0.0535,
      "step": 262625
    },
    {
      "epoch": 3.8681315444544264,
      "grad_norm": 1.4565939903259277,
      "learning_rate": 5.037407463165267e-06,
      "loss": 0.0566,
      "step": 262650
    },
    {
      "epoch": 3.8684997275445134,
      "grad_norm": 1.0134023427963257,
      "learning_rate": 5.035771091198282e-06,
      "loss": 0.0469,
      "step": 262675
    },
    {
      "epoch": 3.8688679106346004,
      "grad_norm": 0.8739012479782104,
      "learning_rate": 5.0341347192312985e-06,
      "loss": 0.0476,
      "step": 262700
    },
    {
      "epoch": 3.8692360937246875,
      "grad_norm": 1.611704707145691,
      "learning_rate": 5.032498347264314e-06,
      "loss": 0.0558,
      "step": 262725
    },
    {
      "epoch": 3.8696042768147745,
      "grad_norm": 1.6190578937530518,
      "learning_rate": 5.030861975297328e-06,
      "loss": 0.0551,
      "step": 262750
    },
    {
      "epoch": 3.8699724599048615,
      "grad_norm": 1.473643183708191,
      "learning_rate": 5.029225603330344e-06,
      "loss": 0.0559,
      "step": 262775
    },
    {
      "epoch": 3.8703406429949485,
      "grad_norm": 1.2018488645553589,
      "learning_rate": 5.02758923136336e-06,
      "loss": 0.052,
      "step": 262800
    },
    {
      "epoch": 3.8707088260850355,
      "grad_norm": 0.9915162324905396,
      "learning_rate": 5.025952859396376e-06,
      "loss": 0.051,
      "step": 262825
    },
    {
      "epoch": 3.8710770091751225,
      "grad_norm": 1.7892978191375732,
      "learning_rate": 5.024316487429391e-06,
      "loss": 0.0601,
      "step": 262850
    },
    {
      "epoch": 3.8714451922652096,
      "grad_norm": 1.0372949838638306,
      "learning_rate": 5.0226801154624065e-06,
      "loss": 0.0492,
      "step": 262875
    },
    {
      "epoch": 3.8718133753552966,
      "grad_norm": 1.0803349018096924,
      "learning_rate": 5.021043743495422e-06,
      "loss": 0.05,
      "step": 262900
    },
    {
      "epoch": 3.8721815584453836,
      "grad_norm": 1.1820727586746216,
      "learning_rate": 5.019407371528437e-06,
      "loss": 0.0535,
      "step": 262925
    },
    {
      "epoch": 3.8725497415354706,
      "grad_norm": 1.3209728002548218,
      "learning_rate": 5.017770999561453e-06,
      "loss": 0.0577,
      "step": 262950
    },
    {
      "epoch": 3.8729179246255576,
      "grad_norm": 1.413087010383606,
      "learning_rate": 5.016134627594468e-06,
      "loss": 0.0524,
      "step": 262975
    },
    {
      "epoch": 3.8732861077156446,
      "grad_norm": 1.6288573741912842,
      "learning_rate": 5.014498255627484e-06,
      "loss": 0.054,
      "step": 263000
    },
    {
      "epoch": 3.8736542908057316,
      "grad_norm": 1.2065969705581665,
      "learning_rate": 5.012861883660499e-06,
      "loss": 0.0538,
      "step": 263025
    },
    {
      "epoch": 3.8740224738958187,
      "grad_norm": 1.3036879301071167,
      "learning_rate": 5.011225511693515e-06,
      "loss": 0.0498,
      "step": 263050
    },
    {
      "epoch": 3.8743906569859057,
      "grad_norm": 0.9634690284729004,
      "learning_rate": 5.00958913972653e-06,
      "loss": 0.054,
      "step": 263075
    },
    {
      "epoch": 3.874758840075993,
      "grad_norm": 1.1450031995773315,
      "learning_rate": 5.007952767759545e-06,
      "loss": 0.0643,
      "step": 263100
    },
    {
      "epoch": 3.87512702316608,
      "grad_norm": 1.444369912147522,
      "learning_rate": 5.006316395792561e-06,
      "loss": 0.0589,
      "step": 263125
    },
    {
      "epoch": 3.875495206256167,
      "grad_norm": 1.1926497220993042,
      "learning_rate": 5.0046800238255765e-06,
      "loss": 0.0553,
      "step": 263150
    },
    {
      "epoch": 3.875863389346254,
      "grad_norm": 1.1588252782821655,
      "learning_rate": 5.003043651858591e-06,
      "loss": 0.0512,
      "step": 263175
    },
    {
      "epoch": 3.876231572436341,
      "grad_norm": 1.4155786037445068,
      "learning_rate": 5.001407279891607e-06,
      "loss": 0.0596,
      "step": 263200
    },
    {
      "epoch": 3.876599755526428,
      "grad_norm": 1.4117778539657593,
      "learning_rate": 4.9997709079246224e-06,
      "loss": 0.0534,
      "step": 263225
    },
    {
      "epoch": 3.8769679386165152,
      "grad_norm": 1.7850371599197388,
      "learning_rate": 4.998134535957638e-06,
      "loss": 0.0533,
      "step": 263250
    },
    {
      "epoch": 3.8773361217066022,
      "grad_norm": 1.54677152633667,
      "learning_rate": 4.996498163990653e-06,
      "loss": 0.0532,
      "step": 263275
    },
    {
      "epoch": 3.8777043047966893,
      "grad_norm": 1.1184360980987549,
      "learning_rate": 4.994861792023669e-06,
      "loss": 0.0563,
      "step": 263300
    },
    {
      "epoch": 3.8780724878867763,
      "grad_norm": 1.3255177736282349,
      "learning_rate": 4.9932254200566845e-06,
      "loss": 0.0518,
      "step": 263325
    },
    {
      "epoch": 3.8784406709768633,
      "grad_norm": 1.4861990213394165,
      "learning_rate": 4.9915890480897e-06,
      "loss": 0.0508,
      "step": 263350
    },
    {
      "epoch": 3.8788088540669503,
      "grad_norm": 1.1583616733551025,
      "learning_rate": 4.989952676122715e-06,
      "loss": 0.0527,
      "step": 263375
    },
    {
      "epoch": 3.8791770371570373,
      "grad_norm": 1.5475693941116333,
      "learning_rate": 4.988316304155731e-06,
      "loss": 0.0523,
      "step": 263400
    },
    {
      "epoch": 3.8795452202471243,
      "grad_norm": 1.5123289823532104,
      "learning_rate": 4.986679932188746e-06,
      "loss": 0.0545,
      "step": 263425
    },
    {
      "epoch": 3.879913403337212,
      "grad_norm": 1.5087281465530396,
      "learning_rate": 4.985043560221762e-06,
      "loss": 0.0503,
      "step": 263450
    },
    {
      "epoch": 3.880281586427299,
      "grad_norm": 1.328429937362671,
      "learning_rate": 4.983407188254777e-06,
      "loss": 0.0508,
      "step": 263475
    },
    {
      "epoch": 3.880649769517386,
      "grad_norm": 1.2900867462158203,
      "learning_rate": 4.9817708162877925e-06,
      "loss": 0.0566,
      "step": 263500
    },
    {
      "epoch": 3.881017952607473,
      "grad_norm": 1.924965500831604,
      "learning_rate": 4.980134444320808e-06,
      "loss": 0.0527,
      "step": 263525
    },
    {
      "epoch": 3.88138613569756,
      "grad_norm": 1.3238420486450195,
      "learning_rate": 4.978498072353823e-06,
      "loss": 0.0523,
      "step": 263550
    },
    {
      "epoch": 3.881754318787647,
      "grad_norm": 1.123356819152832,
      "learning_rate": 4.976861700386838e-06,
      "loss": 0.0523,
      "step": 263575
    },
    {
      "epoch": 3.882122501877734,
      "grad_norm": 1.0134892463684082,
      "learning_rate": 4.9752253284198545e-06,
      "loss": 0.0435,
      "step": 263600
    },
    {
      "epoch": 3.882490684967821,
      "grad_norm": 1.3513299226760864,
      "learning_rate": 4.97358895645287e-06,
      "loss": 0.0528,
      "step": 263625
    },
    {
      "epoch": 3.882858868057908,
      "grad_norm": 1.4515777826309204,
      "learning_rate": 4.971952584485885e-06,
      "loss": 0.0521,
      "step": 263650
    },
    {
      "epoch": 3.883227051147995,
      "grad_norm": 1.0249426364898682,
      "learning_rate": 4.9703162125189005e-06,
      "loss": 0.0549,
      "step": 263675
    },
    {
      "epoch": 3.883595234238082,
      "grad_norm": 1.5847457647323608,
      "learning_rate": 4.968679840551916e-06,
      "loss": 0.0542,
      "step": 263700
    },
    {
      "epoch": 3.883963417328169,
      "grad_norm": 1.0688776969909668,
      "learning_rate": 4.967043468584931e-06,
      "loss": 0.0513,
      "step": 263725
    },
    {
      "epoch": 3.884331600418256,
      "grad_norm": 1.574480414390564,
      "learning_rate": 4.965407096617947e-06,
      "loss": 0.0675,
      "step": 263750
    },
    {
      "epoch": 3.884699783508343,
      "grad_norm": 1.8858754634857178,
      "learning_rate": 4.963770724650962e-06,
      "loss": 0.0537,
      "step": 263775
    },
    {
      "epoch": 3.88506796659843,
      "grad_norm": 1.6930290460586548,
      "learning_rate": 4.962134352683978e-06,
      "loss": 0.0499,
      "step": 263800
    },
    {
      "epoch": 3.885436149688517,
      "grad_norm": 1.8441869020462036,
      "learning_rate": 4.960497980716993e-06,
      "loss": 0.0534,
      "step": 263825
    },
    {
      "epoch": 3.885804332778604,
      "grad_norm": 1.0883420705795288,
      "learning_rate": 4.9588616087500084e-06,
      "loss": 0.0518,
      "step": 263850
    },
    {
      "epoch": 3.886172515868691,
      "grad_norm": 0.9594785571098328,
      "learning_rate": 4.957225236783025e-06,
      "loss": 0.0551,
      "step": 263875
    },
    {
      "epoch": 3.886540698958778,
      "grad_norm": 1.289876937866211,
      "learning_rate": 4.955588864816039e-06,
      "loss": 0.0551,
      "step": 263900
    },
    {
      "epoch": 3.886908882048865,
      "grad_norm": 0.6620969772338867,
      "learning_rate": 4.953952492849055e-06,
      "loss": 0.0458,
      "step": 263925
    },
    {
      "epoch": 3.887277065138952,
      "grad_norm": 1.5180296897888184,
      "learning_rate": 4.9523161208820705e-06,
      "loss": 0.0603,
      "step": 263950
    },
    {
      "epoch": 3.887645248229039,
      "grad_norm": 1.2539643049240112,
      "learning_rate": 4.950679748915086e-06,
      "loss": 0.049,
      "step": 263975
    },
    {
      "epoch": 3.888013431319126,
      "grad_norm": 1.4199472665786743,
      "learning_rate": 4.949043376948101e-06,
      "loss": 0.0582,
      "step": 264000
    },
    {
      "epoch": 3.888381614409213,
      "grad_norm": 1.2492717504501343,
      "learning_rate": 4.947407004981116e-06,
      "loss": 0.0476,
      "step": 264025
    },
    {
      "epoch": 3.8887497974993006,
      "grad_norm": 1.4540746212005615,
      "learning_rate": 4.945770633014132e-06,
      "loss": 0.0511,
      "step": 264050
    },
    {
      "epoch": 3.8891179805893876,
      "grad_norm": 1.660137414932251,
      "learning_rate": 4.944134261047148e-06,
      "loss": 0.0598,
      "step": 264075
    },
    {
      "epoch": 3.8894861636794746,
      "grad_norm": 1.6908013820648193,
      "learning_rate": 4.942497889080163e-06,
      "loss": 0.0497,
      "step": 264100
    },
    {
      "epoch": 3.8898543467695617,
      "grad_norm": 1.3535715341567993,
      "learning_rate": 4.9408615171131785e-06,
      "loss": 0.0515,
      "step": 264125
    },
    {
      "epoch": 3.8902225298596487,
      "grad_norm": 1.08555006980896,
      "learning_rate": 4.939225145146194e-06,
      "loss": 0.0538,
      "step": 264150
    },
    {
      "epoch": 3.8905907129497357,
      "grad_norm": 1.4909448623657227,
      "learning_rate": 4.937588773179209e-06,
      "loss": 0.0535,
      "step": 264175
    },
    {
      "epoch": 3.8909588960398227,
      "grad_norm": 1.6206080913543701,
      "learning_rate": 4.935952401212224e-06,
      "loss": 0.0493,
      "step": 264200
    },
    {
      "epoch": 3.8913270791299097,
      "grad_norm": 1.244004726409912,
      "learning_rate": 4.9343160292452405e-06,
      "loss": 0.057,
      "step": 264225
    },
    {
      "epoch": 3.8916952622199967,
      "grad_norm": 1.2292659282684326,
      "learning_rate": 4.932679657278255e-06,
      "loss": 0.0506,
      "step": 264250
    },
    {
      "epoch": 3.8920634453100837,
      "grad_norm": 0.9641752243041992,
      "learning_rate": 4.931043285311271e-06,
      "loss": 0.055,
      "step": 264275
    },
    {
      "epoch": 3.8924316284001708,
      "grad_norm": 1.9379572868347168,
      "learning_rate": 4.9294069133442865e-06,
      "loss": 0.0576,
      "step": 264300
    },
    {
      "epoch": 3.8927998114902578,
      "grad_norm": 1.2098896503448486,
      "learning_rate": 4.927770541377302e-06,
      "loss": 0.0567,
      "step": 264325
    },
    {
      "epoch": 3.893167994580345,
      "grad_norm": 1.4082597494125366,
      "learning_rate": 4.926134169410318e-06,
      "loss": 0.0521,
      "step": 264350
    },
    {
      "epoch": 3.893536177670432,
      "grad_norm": 1.9065814018249512,
      "learning_rate": 4.924497797443332e-06,
      "loss": 0.063,
      "step": 264375
    },
    {
      "epoch": 3.8939043607605193,
      "grad_norm": 1.0186904668807983,
      "learning_rate": 4.9228614254763485e-06,
      "loss": 0.0491,
      "step": 264400
    },
    {
      "epoch": 3.8942725438506063,
      "grad_norm": 1.4114601612091064,
      "learning_rate": 4.921225053509364e-06,
      "loss": 0.0519,
      "step": 264425
    },
    {
      "epoch": 3.8946407269406933,
      "grad_norm": 1.9180808067321777,
      "learning_rate": 4.919588681542379e-06,
      "loss": 0.0509,
      "step": 264450
    },
    {
      "epoch": 3.8950089100307803,
      "grad_norm": 1.1067109107971191,
      "learning_rate": 4.9179523095753944e-06,
      "loss": 0.0531,
      "step": 264475
    },
    {
      "epoch": 3.8953770931208673,
      "grad_norm": 1.7426888942718506,
      "learning_rate": 4.916315937608411e-06,
      "loss": 0.0531,
      "step": 264500
    },
    {
      "epoch": 3.8957452762109543,
      "grad_norm": 1.4335108995437622,
      "learning_rate": 4.914679565641425e-06,
      "loss": 0.0556,
      "step": 264525
    },
    {
      "epoch": 3.8961134593010414,
      "grad_norm": 1.5560462474822998,
      "learning_rate": 4.913043193674441e-06,
      "loss": 0.0539,
      "step": 264550
    },
    {
      "epoch": 3.8964816423911284,
      "grad_norm": 1.5154043436050415,
      "learning_rate": 4.9114068217074565e-06,
      "loss": 0.0518,
      "step": 264575
    },
    {
      "epoch": 3.8968498254812154,
      "grad_norm": 1.4359437227249146,
      "learning_rate": 4.909770449740472e-06,
      "loss": 0.0522,
      "step": 264600
    },
    {
      "epoch": 3.8972180085713024,
      "grad_norm": 1.1610114574432373,
      "learning_rate": 4.908134077773487e-06,
      "loss": 0.0522,
      "step": 264625
    },
    {
      "epoch": 3.8975861916613894,
      "grad_norm": 1.2974045276641846,
      "learning_rate": 4.906497705806502e-06,
      "loss": 0.055,
      "step": 264650
    },
    {
      "epoch": 3.8979543747514764,
      "grad_norm": 0.9947437644004822,
      "learning_rate": 4.904861333839518e-06,
      "loss": 0.0524,
      "step": 264675
    },
    {
      "epoch": 3.8983225578415635,
      "grad_norm": 1.2135608196258545,
      "learning_rate": 4.903224961872534e-06,
      "loss": 0.0524,
      "step": 264700
    },
    {
      "epoch": 3.8986907409316505,
      "grad_norm": 1.270745873451233,
      "learning_rate": 4.901588589905549e-06,
      "loss": 0.052,
      "step": 264725
    },
    {
      "epoch": 3.8990589240217375,
      "grad_norm": 1.5922800302505493,
      "learning_rate": 4.8999522179385645e-06,
      "loss": 0.0521,
      "step": 264750
    },
    {
      "epoch": 3.8994271071118245,
      "grad_norm": 1.6764752864837646,
      "learning_rate": 4.89831584597158e-06,
      "loss": 0.0508,
      "step": 264775
    },
    {
      "epoch": 3.8997952902019115,
      "grad_norm": 1.5361818075180054,
      "learning_rate": 4.896679474004595e-06,
      "loss": 0.0517,
      "step": 264800
    },
    {
      "epoch": 3.9001634732919985,
      "grad_norm": 1.6716116666793823,
      "learning_rate": 4.895043102037611e-06,
      "loss": 0.0558,
      "step": 264825
    },
    {
      "epoch": 3.9005316563820855,
      "grad_norm": 0.8519721031188965,
      "learning_rate": 4.8934067300706265e-06,
      "loss": 0.0507,
      "step": 264850
    },
    {
      "epoch": 3.9008998394721726,
      "grad_norm": 1.0493004322052002,
      "learning_rate": 4.891770358103642e-06,
      "loss": 0.0485,
      "step": 264875
    },
    {
      "epoch": 3.9012680225622596,
      "grad_norm": 1.2470858097076416,
      "learning_rate": 4.890133986136657e-06,
      "loss": 0.0549,
      "step": 264900
    },
    {
      "epoch": 3.9016362056523466,
      "grad_norm": 1.2583844661712646,
      "learning_rate": 4.8884976141696725e-06,
      "loss": 0.0578,
      "step": 264925
    },
    {
      "epoch": 3.9020043887424336,
      "grad_norm": 1.489508032798767,
      "learning_rate": 4.886861242202688e-06,
      "loss": 0.0513,
      "step": 264950
    },
    {
      "epoch": 3.9023725718325206,
      "grad_norm": 1.6750906705856323,
      "learning_rate": 4.885224870235704e-06,
      "loss": 0.0547,
      "step": 264975
    },
    {
      "epoch": 3.902740754922608,
      "grad_norm": 0.6309359669685364,
      "learning_rate": 4.883588498268718e-06,
      "loss": 0.0565,
      "step": 265000
    },
    {
      "epoch": 3.903108938012695,
      "grad_norm": 0.8855849504470825,
      "learning_rate": 4.8819521263017345e-06,
      "loss": 0.0504,
      "step": 265025
    },
    {
      "epoch": 3.903477121102782,
      "grad_norm": 1.6380351781845093,
      "learning_rate": 4.88031575433475e-06,
      "loss": 0.0582,
      "step": 265050
    },
    {
      "epoch": 3.903845304192869,
      "grad_norm": 1.115140676498413,
      "learning_rate": 4.878679382367765e-06,
      "loss": 0.0549,
      "step": 265075
    },
    {
      "epoch": 3.904213487282956,
      "grad_norm": 0.9697654843330383,
      "learning_rate": 4.8770430104007804e-06,
      "loss": 0.0525,
      "step": 265100
    },
    {
      "epoch": 3.904581670373043,
      "grad_norm": 1.5834414958953857,
      "learning_rate": 4.875406638433796e-06,
      "loss": 0.06,
      "step": 265125
    },
    {
      "epoch": 3.90494985346313,
      "grad_norm": 1.3147239685058594,
      "learning_rate": 4.873770266466811e-06,
      "loss": 0.0602,
      "step": 265150
    },
    {
      "epoch": 3.905318036553217,
      "grad_norm": 1.2951992750167847,
      "learning_rate": 4.872133894499827e-06,
      "loss": 0.0535,
      "step": 265175
    },
    {
      "epoch": 3.905686219643304,
      "grad_norm": 1.5094690322875977,
      "learning_rate": 4.8704975225328425e-06,
      "loss": 0.0498,
      "step": 265200
    },
    {
      "epoch": 3.906054402733391,
      "grad_norm": 1.495611548423767,
      "learning_rate": 4.868861150565858e-06,
      "loss": 0.0577,
      "step": 265225
    },
    {
      "epoch": 3.9064225858234782,
      "grad_norm": 1.5796841382980347,
      "learning_rate": 4.867224778598873e-06,
      "loss": 0.0531,
      "step": 265250
    },
    {
      "epoch": 3.9067907689135652,
      "grad_norm": 1.49833083152771,
      "learning_rate": 4.865588406631888e-06,
      "loss": 0.0563,
      "step": 265275
    },
    {
      "epoch": 3.9071589520036523,
      "grad_norm": 1.3084923028945923,
      "learning_rate": 4.8639520346649046e-06,
      "loss": 0.0498,
      "step": 265300
    },
    {
      "epoch": 3.9075271350937393,
      "grad_norm": 1.0368095636367798,
      "learning_rate": 4.86231566269792e-06,
      "loss": 0.0488,
      "step": 265325
    },
    {
      "epoch": 3.9078953181838263,
      "grad_norm": 1.35903799533844,
      "learning_rate": 4.860679290730935e-06,
      "loss": 0.0602,
      "step": 265350
    },
    {
      "epoch": 3.9082635012739138,
      "grad_norm": 1.058145523071289,
      "learning_rate": 4.8590429187639505e-06,
      "loss": 0.0462,
      "step": 265375
    },
    {
      "epoch": 3.9086316843640008,
      "grad_norm": 1.3978030681610107,
      "learning_rate": 4.857406546796966e-06,
      "loss": 0.0501,
      "step": 265400
    },
    {
      "epoch": 3.908999867454088,
      "grad_norm": 1.2879433631896973,
      "learning_rate": 4.855770174829981e-06,
      "loss": 0.0524,
      "step": 265425
    },
    {
      "epoch": 3.909368050544175,
      "grad_norm": 1.6850285530090332,
      "learning_rate": 4.854133802862997e-06,
      "loss": 0.0505,
      "step": 265450
    },
    {
      "epoch": 3.909736233634262,
      "grad_norm": 1.6365737915039062,
      "learning_rate": 4.852497430896012e-06,
      "loss": 0.0553,
      "step": 265475
    },
    {
      "epoch": 3.910104416724349,
      "grad_norm": 1.3211332559585571,
      "learning_rate": 4.850861058929028e-06,
      "loss": 0.061,
      "step": 265500
    },
    {
      "epoch": 3.910472599814436,
      "grad_norm": 1.489636778831482,
      "learning_rate": 4.849224686962043e-06,
      "loss": 0.055,
      "step": 265525
    },
    {
      "epoch": 3.910840782904523,
      "grad_norm": 1.7707267999649048,
      "learning_rate": 4.8475883149950585e-06,
      "loss": 0.0501,
      "step": 265550
    },
    {
      "epoch": 3.91120896599461,
      "grad_norm": 1.2592132091522217,
      "learning_rate": 4.845951943028074e-06,
      "loss": 0.0469,
      "step": 265575
    },
    {
      "epoch": 3.911577149084697,
      "grad_norm": 1.3604294061660767,
      "learning_rate": 4.84431557106109e-06,
      "loss": 0.0499,
      "step": 265600
    },
    {
      "epoch": 3.911945332174784,
      "grad_norm": 1.6320096254348755,
      "learning_rate": 4.842679199094104e-06,
      "loss": 0.0579,
      "step": 265625
    },
    {
      "epoch": 3.912313515264871,
      "grad_norm": 1.7693284749984741,
      "learning_rate": 4.8410428271271205e-06,
      "loss": 0.0587,
      "step": 265650
    },
    {
      "epoch": 3.912681698354958,
      "grad_norm": 1.6865720748901367,
      "learning_rate": 4.839406455160136e-06,
      "loss": 0.0528,
      "step": 265675
    },
    {
      "epoch": 3.913049881445045,
      "grad_norm": 1.4077274799346924,
      "learning_rate": 4.837770083193151e-06,
      "loss": 0.0535,
      "step": 265700
    },
    {
      "epoch": 3.913418064535132,
      "grad_norm": 1.789424180984497,
      "learning_rate": 4.8361337112261664e-06,
      "loss": 0.0508,
      "step": 265725
    },
    {
      "epoch": 3.913786247625219,
      "grad_norm": 1.5183805227279663,
      "learning_rate": 4.834562794137861e-06,
      "loss": 0.0581,
      "step": 265750
    },
    {
      "epoch": 3.914154430715306,
      "grad_norm": 1.3152365684509277,
      "learning_rate": 4.832926422170877e-06,
      "loss": 0.0572,
      "step": 265775
    },
    {
      "epoch": 3.914522613805393,
      "grad_norm": 1.2793784141540527,
      "learning_rate": 4.831290050203893e-06,
      "loss": 0.0532,
      "step": 265800
    },
    {
      "epoch": 3.91489079689548,
      "grad_norm": 0.9389621615409851,
      "learning_rate": 4.829653678236908e-06,
      "loss": 0.0491,
      "step": 265825
    },
    {
      "epoch": 3.915258979985567,
      "grad_norm": 1.8882869482040405,
      "learning_rate": 4.828017306269923e-06,
      "loss": 0.056,
      "step": 265850
    },
    {
      "epoch": 3.915627163075654,
      "grad_norm": 1.0818278789520264,
      "learning_rate": 4.8263809343029385e-06,
      "loss": 0.0528,
      "step": 265875
    },
    {
      "epoch": 3.915995346165741,
      "grad_norm": 1.7939482927322388,
      "learning_rate": 4.824744562335954e-06,
      "loss": 0.0538,
      "step": 265900
    },
    {
      "epoch": 3.916363529255828,
      "grad_norm": 0.9074087142944336,
      "learning_rate": 4.82310819036897e-06,
      "loss": 0.0457,
      "step": 265925
    },
    {
      "epoch": 3.916731712345915,
      "grad_norm": 1.7172048091888428,
      "learning_rate": 4.8214718184019844e-06,
      "loss": 0.0566,
      "step": 265950
    },
    {
      "epoch": 3.9170998954360026,
      "grad_norm": 0.9454116225242615,
      "learning_rate": 4.819835446435001e-06,
      "loss": 0.0521,
      "step": 265975
    },
    {
      "epoch": 3.9174680785260896,
      "grad_norm": 1.4153268337249756,
      "learning_rate": 4.818199074468016e-06,
      "loss": 0.0572,
      "step": 266000
    },
    {
      "epoch": 3.9178362616161766,
      "grad_norm": 1.7052981853485107,
      "learning_rate": 4.816562702501031e-06,
      "loss": 0.0532,
      "step": 266025
    },
    {
      "epoch": 3.9182044447062636,
      "grad_norm": 1.160554051399231,
      "learning_rate": 4.8149263305340465e-06,
      "loss": 0.0527,
      "step": 266050
    },
    {
      "epoch": 3.9185726277963506,
      "grad_norm": 1.0327438116073608,
      "learning_rate": 4.813289958567063e-06,
      "loss": 0.0535,
      "step": 266075
    },
    {
      "epoch": 3.9189408108864376,
      "grad_norm": 1.128814935684204,
      "learning_rate": 4.811653586600077e-06,
      "loss": 0.0545,
      "step": 266100
    },
    {
      "epoch": 3.9193089939765247,
      "grad_norm": 1.0633081197738647,
      "learning_rate": 4.810017214633093e-06,
      "loss": 0.054,
      "step": 266125
    },
    {
      "epoch": 3.9196771770666117,
      "grad_norm": 1.7413182258605957,
      "learning_rate": 4.8083808426661086e-06,
      "loss": 0.0497,
      "step": 266150
    },
    {
      "epoch": 3.9200453601566987,
      "grad_norm": 1.3477416038513184,
      "learning_rate": 4.806744470699124e-06,
      "loss": 0.0495,
      "step": 266175
    },
    {
      "epoch": 3.9204135432467857,
      "grad_norm": 1.4275422096252441,
      "learning_rate": 4.805108098732139e-06,
      "loss": 0.0514,
      "step": 266200
    },
    {
      "epoch": 3.9207817263368727,
      "grad_norm": 1.2836915254592896,
      "learning_rate": 4.8034717267651545e-06,
      "loss": 0.0534,
      "step": 266225
    },
    {
      "epoch": 3.9211499094269597,
      "grad_norm": 1.096981167793274,
      "learning_rate": 4.801835354798171e-06,
      "loss": 0.0511,
      "step": 266250
    },
    {
      "epoch": 3.9215180925170467,
      "grad_norm": 1.4388827085494995,
      "learning_rate": 4.800198982831186e-06,
      "loss": 0.0544,
      "step": 266275
    },
    {
      "epoch": 3.9218862756071338,
      "grad_norm": 1.7638380527496338,
      "learning_rate": 4.798562610864201e-06,
      "loss": 0.0554,
      "step": 266300
    },
    {
      "epoch": 3.9222544586972212,
      "grad_norm": 1.970281958580017,
      "learning_rate": 4.7969262388972165e-06,
      "loss": 0.0504,
      "step": 266325
    },
    {
      "epoch": 3.9226226417873082,
      "grad_norm": 1.3509507179260254,
      "learning_rate": 4.795289866930232e-06,
      "loss": 0.0558,
      "step": 266350
    },
    {
      "epoch": 3.9229908248773953,
      "grad_norm": 1.155869722366333,
      "learning_rate": 4.793653494963247e-06,
      "loss": 0.056,
      "step": 266375
    },
    {
      "epoch": 3.9233590079674823,
      "grad_norm": 1.2204054594039917,
      "learning_rate": 4.792017122996263e-06,
      "loss": 0.0491,
      "step": 266400
    },
    {
      "epoch": 3.9237271910575693,
      "grad_norm": 1.7223869562149048,
      "learning_rate": 4.790380751029279e-06,
      "loss": 0.0561,
      "step": 266425
    },
    {
      "epoch": 3.9240953741476563,
      "grad_norm": 1.4869788885116577,
      "learning_rate": 4.788744379062294e-06,
      "loss": 0.0595,
      "step": 266450
    },
    {
      "epoch": 3.9244635572377433,
      "grad_norm": 1.2139626741409302,
      "learning_rate": 4.787108007095309e-06,
      "loss": 0.0548,
      "step": 266475
    },
    {
      "epoch": 3.9248317403278303,
      "grad_norm": 1.4303646087646484,
      "learning_rate": 4.7854716351283245e-06,
      "loss": 0.0503,
      "step": 266500
    },
    {
      "epoch": 3.9251999234179173,
      "grad_norm": 1.713517665863037,
      "learning_rate": 4.78383526316134e-06,
      "loss": 0.0549,
      "step": 266525
    },
    {
      "epoch": 3.9255681065080044,
      "grad_norm": 2.059519052505493,
      "learning_rate": 4.782198891194356e-06,
      "loss": 0.0585,
      "step": 266550
    },
    {
      "epoch": 3.9259362895980914,
      "grad_norm": 1.407524585723877,
      "learning_rate": 4.7805625192273704e-06,
      "loss": 0.0612,
      "step": 266575
    },
    {
      "epoch": 3.9263044726881784,
      "grad_norm": 1.0501809120178223,
      "learning_rate": 4.778926147260387e-06,
      "loss": 0.0493,
      "step": 266600
    },
    {
      "epoch": 3.9266726557782654,
      "grad_norm": 1.40870201587677,
      "learning_rate": 4.777289775293402e-06,
      "loss": 0.0537,
      "step": 266625
    },
    {
      "epoch": 3.9270408388683524,
      "grad_norm": 1.3485110998153687,
      "learning_rate": 4.775653403326417e-06,
      "loss": 0.0518,
      "step": 266650
    },
    {
      "epoch": 3.9274090219584394,
      "grad_norm": 1.403550624847412,
      "learning_rate": 4.7740170313594325e-06,
      "loss": 0.0598,
      "step": 266675
    },
    {
      "epoch": 3.9277772050485265,
      "grad_norm": 1.2981377840042114,
      "learning_rate": 4.772380659392448e-06,
      "loss": 0.0493,
      "step": 266700
    },
    {
      "epoch": 3.9281453881386135,
      "grad_norm": 1.3137564659118652,
      "learning_rate": 4.770744287425464e-06,
      "loss": 0.0533,
      "step": 266725
    },
    {
      "epoch": 3.9285135712287005,
      "grad_norm": 0.9053286910057068,
      "learning_rate": 4.769107915458479e-06,
      "loss": 0.0497,
      "step": 266750
    },
    {
      "epoch": 3.9288817543187875,
      "grad_norm": 1.2881439924240112,
      "learning_rate": 4.7674715434914946e-06,
      "loss": 0.0564,
      "step": 266775
    },
    {
      "epoch": 3.9292499374088745,
      "grad_norm": 1.4748735427856445,
      "learning_rate": 4.76583517152451e-06,
      "loss": 0.0599,
      "step": 266800
    },
    {
      "epoch": 3.9296181204989615,
      "grad_norm": 1.5210356712341309,
      "learning_rate": 4.764198799557526e-06,
      "loss": 0.0589,
      "step": 266825
    },
    {
      "epoch": 3.9299863035890485,
      "grad_norm": 1.4726735353469849,
      "learning_rate": 4.7625624275905405e-06,
      "loss": 0.0527,
      "step": 266850
    },
    {
      "epoch": 3.9303544866791356,
      "grad_norm": 1.6104947328567505,
      "learning_rate": 4.760926055623557e-06,
      "loss": 0.0535,
      "step": 266875
    },
    {
      "epoch": 3.9307226697692226,
      "grad_norm": 1.428181767463684,
      "learning_rate": 4.759289683656572e-06,
      "loss": 0.0501,
      "step": 266900
    },
    {
      "epoch": 3.93109085285931,
      "grad_norm": 1.2220829725265503,
      "learning_rate": 4.757653311689587e-06,
      "loss": 0.0562,
      "step": 266925
    },
    {
      "epoch": 3.931459035949397,
      "grad_norm": 1.426183819770813,
      "learning_rate": 4.7560169397226025e-06,
      "loss": 0.0576,
      "step": 266950
    },
    {
      "epoch": 3.931827219039484,
      "grad_norm": 1.261764645576477,
      "learning_rate": 4.754380567755618e-06,
      "loss": 0.0523,
      "step": 266975
    },
    {
      "epoch": 3.932195402129571,
      "grad_norm": 1.4767704010009766,
      "learning_rate": 4.752744195788633e-06,
      "loss": 0.0509,
      "step": 267000
    },
    {
      "epoch": 3.932563585219658,
      "grad_norm": 1.2506531476974487,
      "learning_rate": 4.751107823821649e-06,
      "loss": 0.0472,
      "step": 267025
    },
    {
      "epoch": 3.932931768309745,
      "grad_norm": 1.3366501331329346,
      "learning_rate": 4.749471451854664e-06,
      "loss": 0.053,
      "step": 267050
    },
    {
      "epoch": 3.933299951399832,
      "grad_norm": 1.314798355102539,
      "learning_rate": 4.74783507988768e-06,
      "loss": 0.0533,
      "step": 267075
    },
    {
      "epoch": 3.933668134489919,
      "grad_norm": 1.0986489057540894,
      "learning_rate": 4.746198707920695e-06,
      "loss": 0.0539,
      "step": 267100
    },
    {
      "epoch": 3.934036317580006,
      "grad_norm": 1.0679348707199097,
      "learning_rate": 4.7445623359537105e-06,
      "loss": 0.0498,
      "step": 267125
    },
    {
      "epoch": 3.934404500670093,
      "grad_norm": 1.3667182922363281,
      "learning_rate": 4.742925963986726e-06,
      "loss": 0.0575,
      "step": 267150
    },
    {
      "epoch": 3.93477268376018,
      "grad_norm": 0.7756701707839966,
      "learning_rate": 4.741289592019742e-06,
      "loss": 0.0502,
      "step": 267175
    },
    {
      "epoch": 3.935140866850267,
      "grad_norm": 1.2835673093795776,
      "learning_rate": 4.7396532200527564e-06,
      "loss": 0.0503,
      "step": 267200
    },
    {
      "epoch": 3.935509049940354,
      "grad_norm": 1.3682050704956055,
      "learning_rate": 4.738016848085773e-06,
      "loss": 0.0517,
      "step": 267225
    },
    {
      "epoch": 3.9358772330304412,
      "grad_norm": 1.4204814434051514,
      "learning_rate": 4.736380476118788e-06,
      "loss": 0.0551,
      "step": 267250
    },
    {
      "epoch": 3.9362454161205283,
      "grad_norm": 1.366399884223938,
      "learning_rate": 4.734744104151803e-06,
      "loss": 0.0561,
      "step": 267275
    },
    {
      "epoch": 3.9366135992106157,
      "grad_norm": 1.3777496814727783,
      "learning_rate": 4.733107732184819e-06,
      "loss": 0.059,
      "step": 267300
    },
    {
      "epoch": 3.9369817823007027,
      "grad_norm": 0.5394716858863831,
      "learning_rate": 4.731471360217834e-06,
      "loss": 0.0454,
      "step": 267325
    },
    {
      "epoch": 3.9373499653907897,
      "grad_norm": 1.2963823080062866,
      "learning_rate": 4.72983498825085e-06,
      "loss": 0.0542,
      "step": 267350
    },
    {
      "epoch": 3.9377181484808768,
      "grad_norm": 1.249991536140442,
      "learning_rate": 4.728198616283865e-06,
      "loss": 0.0518,
      "step": 267375
    },
    {
      "epoch": 3.9380863315709638,
      "grad_norm": 1.4893606901168823,
      "learning_rate": 4.7265622443168806e-06,
      "loss": 0.0628,
      "step": 267400
    },
    {
      "epoch": 3.938454514661051,
      "grad_norm": 1.330801248550415,
      "learning_rate": 4.724925872349896e-06,
      "loss": 0.0523,
      "step": 267425
    },
    {
      "epoch": 3.938822697751138,
      "grad_norm": 2.1741268634796143,
      "learning_rate": 4.723289500382911e-06,
      "loss": 0.0617,
      "step": 267450
    },
    {
      "epoch": 3.939190880841225,
      "grad_norm": 0.8185716867446899,
      "learning_rate": 4.7216531284159265e-06,
      "loss": 0.0502,
      "step": 267475
    },
    {
      "epoch": 3.939559063931312,
      "grad_norm": 1.2055408954620361,
      "learning_rate": 4.720016756448943e-06,
      "loss": 0.0575,
      "step": 267500
    },
    {
      "epoch": 3.939927247021399,
      "grad_norm": 1.2894607782363892,
      "learning_rate": 4.718380384481958e-06,
      "loss": 0.0557,
      "step": 267525
    },
    {
      "epoch": 3.940295430111486,
      "grad_norm": 1.1453173160552979,
      "learning_rate": 4.716744012514973e-06,
      "loss": 0.0499,
      "step": 267550
    },
    {
      "epoch": 3.940663613201573,
      "grad_norm": 1.7617239952087402,
      "learning_rate": 4.7151076405479885e-06,
      "loss": 0.0576,
      "step": 267575
    },
    {
      "epoch": 3.94103179629166,
      "grad_norm": 0.9825162887573242,
      "learning_rate": 4.713471268581004e-06,
      "loss": 0.0532,
      "step": 267600
    },
    {
      "epoch": 3.941399979381747,
      "grad_norm": 1.490334153175354,
      "learning_rate": 4.711834896614019e-06,
      "loss": 0.0552,
      "step": 267625
    },
    {
      "epoch": 3.941768162471834,
      "grad_norm": 1.5881626605987549,
      "learning_rate": 4.710198524647035e-06,
      "loss": 0.0549,
      "step": 267650
    },
    {
      "epoch": 3.942136345561921,
      "grad_norm": 0.9396759867668152,
      "learning_rate": 4.70856215268005e-06,
      "loss": 0.0602,
      "step": 267675
    },
    {
      "epoch": 3.942504528652008,
      "grad_norm": 1.4920693635940552,
      "learning_rate": 4.706925780713066e-06,
      "loss": 0.0531,
      "step": 267700
    },
    {
      "epoch": 3.942872711742095,
      "grad_norm": 1.5342576503753662,
      "learning_rate": 4.705289408746081e-06,
      "loss": 0.0526,
      "step": 267725
    },
    {
      "epoch": 3.943240894832182,
      "grad_norm": 1.1870781183242798,
      "learning_rate": 4.7036530367790965e-06,
      "loss": 0.0532,
      "step": 267750
    },
    {
      "epoch": 3.943609077922269,
      "grad_norm": 1.4740474224090576,
      "learning_rate": 4.702016664812113e-06,
      "loss": 0.0501,
      "step": 267775
    },
    {
      "epoch": 3.943977261012356,
      "grad_norm": 1.4664939641952515,
      "learning_rate": 4.700380292845127e-06,
      "loss": 0.0484,
      "step": 267800
    },
    {
      "epoch": 3.944345444102443,
      "grad_norm": 1.5314321517944336,
      "learning_rate": 4.698743920878143e-06,
      "loss": 0.0599,
      "step": 267825
    },
    {
      "epoch": 3.94471362719253,
      "grad_norm": 1.668631672859192,
      "learning_rate": 4.697107548911159e-06,
      "loss": 0.0535,
      "step": 267850
    },
    {
      "epoch": 3.945081810282617,
      "grad_norm": 1.4098851680755615,
      "learning_rate": 4.695471176944174e-06,
      "loss": 0.0612,
      "step": 267875
    },
    {
      "epoch": 3.9454499933727045,
      "grad_norm": 1.2308943271636963,
      "learning_rate": 4.693834804977189e-06,
      "loss": 0.0485,
      "step": 267900
    },
    {
      "epoch": 3.9458181764627915,
      "grad_norm": 1.3214237689971924,
      "learning_rate": 4.692198433010205e-06,
      "loss": 0.0514,
      "step": 267925
    },
    {
      "epoch": 3.9461863595528786,
      "grad_norm": 0.8661231398582458,
      "learning_rate": 4.69056206104322e-06,
      "loss": 0.048,
      "step": 267950
    },
    {
      "epoch": 3.9465545426429656,
      "grad_norm": 1.5837327241897583,
      "learning_rate": 4.688925689076236e-06,
      "loss": 0.0534,
      "step": 267975
    },
    {
      "epoch": 3.9469227257330526,
      "grad_norm": 1.7167258262634277,
      "learning_rate": 4.687289317109251e-06,
      "loss": 0.0558,
      "step": 268000
    },
    {
      "epoch": 3.9472909088231396,
      "grad_norm": 2.0640058517456055,
      "learning_rate": 4.6856529451422666e-06,
      "loss": 0.0567,
      "step": 268025
    },
    {
      "epoch": 3.9476590919132266,
      "grad_norm": 0.9550992846488953,
      "learning_rate": 4.684016573175282e-06,
      "loss": 0.0557,
      "step": 268050
    },
    {
      "epoch": 3.9480272750033136,
      "grad_norm": 1.574422001838684,
      "learning_rate": 4.682380201208297e-06,
      "loss": 0.0576,
      "step": 268075
    },
    {
      "epoch": 3.9483954580934006,
      "grad_norm": 1.7121474742889404,
      "learning_rate": 4.6807438292413125e-06,
      "loss": 0.0572,
      "step": 268100
    },
    {
      "epoch": 3.9487636411834877,
      "grad_norm": 1.6270625591278076,
      "learning_rate": 4.679107457274329e-06,
      "loss": 0.0515,
      "step": 268125
    },
    {
      "epoch": 3.9491318242735747,
      "grad_norm": 1.489768147468567,
      "learning_rate": 4.677471085307343e-06,
      "loss": 0.0488,
      "step": 268150
    },
    {
      "epoch": 3.9495000073636617,
      "grad_norm": 1.512560486793518,
      "learning_rate": 4.675834713340359e-06,
      "loss": 0.048,
      "step": 268175
    },
    {
      "epoch": 3.9498681904537487,
      "grad_norm": 0.9951837062835693,
      "learning_rate": 4.6741983413733745e-06,
      "loss": 0.0512,
      "step": 268200
    },
    {
      "epoch": 3.9502363735438357,
      "grad_norm": 1.2083882093429565,
      "learning_rate": 4.67256196940639e-06,
      "loss": 0.0599,
      "step": 268225
    },
    {
      "epoch": 3.950604556633923,
      "grad_norm": 1.7658652067184448,
      "learning_rate": 4.670925597439406e-06,
      "loss": 0.0566,
      "step": 268250
    },
    {
      "epoch": 3.95097273972401,
      "grad_norm": 1.414762258529663,
      "learning_rate": 4.669289225472421e-06,
      "loss": 0.046,
      "step": 268275
    },
    {
      "epoch": 3.951340922814097,
      "grad_norm": 1.7980754375457764,
      "learning_rate": 4.667652853505437e-06,
      "loss": 0.053,
      "step": 268300
    },
    {
      "epoch": 3.9517091059041842,
      "grad_norm": 1.034510850906372,
      "learning_rate": 4.666016481538452e-06,
      "loss": 0.0503,
      "step": 268325
    },
    {
      "epoch": 3.9520772889942712,
      "grad_norm": 1.8255705833435059,
      "learning_rate": 4.664380109571467e-06,
      "loss": 0.0476,
      "step": 268350
    },
    {
      "epoch": 3.9524454720843583,
      "grad_norm": 2.084826946258545,
      "learning_rate": 4.6627437376044825e-06,
      "loss": 0.0561,
      "step": 268375
    },
    {
      "epoch": 3.9528136551744453,
      "grad_norm": 1.2996611595153809,
      "learning_rate": 4.661107365637499e-06,
      "loss": 0.0519,
      "step": 268400
    },
    {
      "epoch": 3.9531818382645323,
      "grad_norm": 1.4135969877243042,
      "learning_rate": 4.659470993670513e-06,
      "loss": 0.0502,
      "step": 268425
    },
    {
      "epoch": 3.9535500213546193,
      "grad_norm": Infinity,
      "learning_rate": 4.657900076582209e-06,
      "loss": 0.0513,
      "step": 268450
    },
    {
      "epoch": 3.9539182044447063,
      "grad_norm": 1.0560312271118164,
      "learning_rate": 4.656263704615224e-06,
      "loss": 0.0484,
      "step": 268475
    },
    {
      "epoch": 3.9542863875347933,
      "grad_norm": 1.3681700229644775,
      "learning_rate": 4.654627332648239e-06,
      "loss": 0.0521,
      "step": 268500
    },
    {
      "epoch": 3.9546545706248803,
      "grad_norm": 1.2107115983963013,
      "learning_rate": 4.652990960681255e-06,
      "loss": 0.053,
      "step": 268525
    },
    {
      "epoch": 3.9550227537149674,
      "grad_norm": 1.4557937383651733,
      "learning_rate": 4.65135458871427e-06,
      "loss": 0.0573,
      "step": 268550
    },
    {
      "epoch": 3.9553909368050544,
      "grad_norm": 0.8258102536201477,
      "learning_rate": 4.649718216747285e-06,
      "loss": 0.0542,
      "step": 268575
    },
    {
      "epoch": 3.9557591198951414,
      "grad_norm": 1.3532434701919556,
      "learning_rate": 4.648081844780301e-06,
      "loss": 0.0568,
      "step": 268600
    },
    {
      "epoch": 3.9561273029852284,
      "grad_norm": 1.3047311305999756,
      "learning_rate": 4.646445472813317e-06,
      "loss": 0.0566,
      "step": 268625
    },
    {
      "epoch": 3.9564954860753154,
      "grad_norm": 1.172406792640686,
      "learning_rate": 4.644809100846332e-06,
      "loss": 0.0603,
      "step": 268650
    },
    {
      "epoch": 3.9568636691654024,
      "grad_norm": 1.6092957258224487,
      "learning_rate": 4.643172728879347e-06,
      "loss": 0.0595,
      "step": 268675
    },
    {
      "epoch": 3.9572318522554895,
      "grad_norm": 0.9011870622634888,
      "learning_rate": 4.641536356912363e-06,
      "loss": 0.0537,
      "step": 268700
    },
    {
      "epoch": 3.9576000353455765,
      "grad_norm": 1.2932192087173462,
      "learning_rate": 4.639899984945379e-06,
      "loss": 0.05,
      "step": 268725
    },
    {
      "epoch": 3.9579682184356635,
      "grad_norm": 1.4305644035339355,
      "learning_rate": 4.638263612978394e-06,
      "loss": 0.0585,
      "step": 268750
    },
    {
      "epoch": 3.9583364015257505,
      "grad_norm": 1.1784656047821045,
      "learning_rate": 4.636627241011409e-06,
      "loss": 0.0566,
      "step": 268775
    },
    {
      "epoch": 3.9587045846158375,
      "grad_norm": 1.1213839054107666,
      "learning_rate": 4.634990869044425e-06,
      "loss": 0.0515,
      "step": 268800
    },
    {
      "epoch": 3.9590727677059245,
      "grad_norm": 1.3900045156478882,
      "learning_rate": 4.63335449707744e-06,
      "loss": 0.0472,
      "step": 268825
    },
    {
      "epoch": 3.959440950796012,
      "grad_norm": 1.5762546062469482,
      "learning_rate": 4.631718125110455e-06,
      "loss": 0.0637,
      "step": 268850
    },
    {
      "epoch": 3.959809133886099,
      "grad_norm": 1.1227755546569824,
      "learning_rate": 4.630081753143471e-06,
      "loss": 0.0502,
      "step": 268875
    },
    {
      "epoch": 3.960177316976186,
      "grad_norm": 1.465013861656189,
      "learning_rate": 4.628445381176486e-06,
      "loss": 0.0566,
      "step": 268900
    },
    {
      "epoch": 3.960545500066273,
      "grad_norm": 1.5864336490631104,
      "learning_rate": 4.626809009209502e-06,
      "loss": 0.0597,
      "step": 268925
    },
    {
      "epoch": 3.96091368315636,
      "grad_norm": 1.635076642036438,
      "learning_rate": 4.625172637242517e-06,
      "loss": 0.049,
      "step": 268950
    },
    {
      "epoch": 3.961281866246447,
      "grad_norm": 1.3956019878387451,
      "learning_rate": 4.623536265275533e-06,
      "loss": 0.0589,
      "step": 268975
    },
    {
      "epoch": 3.961650049336534,
      "grad_norm": 1.9567335844039917,
      "learning_rate": 4.621899893308548e-06,
      "loss": 0.0588,
      "step": 269000
    },
    {
      "epoch": 3.962018232426621,
      "grad_norm": 1.4911231994628906,
      "learning_rate": 4.620263521341563e-06,
      "loss": 0.0563,
      "step": 269025
    },
    {
      "epoch": 3.962386415516708,
      "grad_norm": 1.437920093536377,
      "learning_rate": 4.6186271493745785e-06,
      "loss": 0.0533,
      "step": 269050
    },
    {
      "epoch": 3.962754598606795,
      "grad_norm": 1.3426780700683594,
      "learning_rate": 4.616990777407595e-06,
      "loss": 0.0586,
      "step": 269075
    },
    {
      "epoch": 3.963122781696882,
      "grad_norm": 1.3258249759674072,
      "learning_rate": 4.61535440544061e-06,
      "loss": 0.0557,
      "step": 269100
    },
    {
      "epoch": 3.963490964786969,
      "grad_norm": 1.2463361024856567,
      "learning_rate": 4.613718033473625e-06,
      "loss": 0.0537,
      "step": 269125
    },
    {
      "epoch": 3.963859147877056,
      "grad_norm": 1.011029601097107,
      "learning_rate": 4.6120816615066415e-06,
      "loss": 0.0528,
      "step": 269150
    },
    {
      "epoch": 3.964227330967143,
      "grad_norm": 1.9055616855621338,
      "learning_rate": 4.610445289539656e-06,
      "loss": 0.0605,
      "step": 269175
    },
    {
      "epoch": 3.9645955140572307,
      "grad_norm": 1.4764559268951416,
      "learning_rate": 4.608808917572672e-06,
      "loss": 0.0507,
      "step": 269200
    },
    {
      "epoch": 3.9649636971473177,
      "grad_norm": 1.0713492631912231,
      "learning_rate": 4.607172545605687e-06,
      "loss": 0.0538,
      "step": 269225
    },
    {
      "epoch": 3.9653318802374047,
      "grad_norm": 1.5301762819290161,
      "learning_rate": 4.605536173638703e-06,
      "loss": 0.0552,
      "step": 269250
    },
    {
      "epoch": 3.9657000633274917,
      "grad_norm": 1.4716534614562988,
      "learning_rate": 4.603899801671718e-06,
      "loss": 0.0492,
      "step": 269275
    },
    {
      "epoch": 3.9660682464175787,
      "grad_norm": 1.1124612092971802,
      "learning_rate": 4.602263429704733e-06,
      "loss": 0.052,
      "step": 269300
    },
    {
      "epoch": 3.9664364295076657,
      "grad_norm": 1.2334586381912231,
      "learning_rate": 4.600627057737749e-06,
      "loss": 0.0635,
      "step": 269325
    },
    {
      "epoch": 3.9668046125977527,
      "grad_norm": 1.4479761123657227,
      "learning_rate": 4.598990685770765e-06,
      "loss": 0.0533,
      "step": 269350
    },
    {
      "epoch": 3.9671727956878398,
      "grad_norm": 1.0653736591339111,
      "learning_rate": 4.597354313803779e-06,
      "loss": 0.0498,
      "step": 269375
    },
    {
      "epoch": 3.9675409787779268,
      "grad_norm": 1.6402326822280884,
      "learning_rate": 4.595717941836795e-06,
      "loss": 0.0543,
      "step": 269400
    },
    {
      "epoch": 3.967909161868014,
      "grad_norm": 1.0338371992111206,
      "learning_rate": 4.594081569869811e-06,
      "loss": 0.0569,
      "step": 269425
    },
    {
      "epoch": 3.968277344958101,
      "grad_norm": 1.7422449588775635,
      "learning_rate": 4.592445197902826e-06,
      "loss": 0.0586,
      "step": 269450
    },
    {
      "epoch": 3.968645528048188,
      "grad_norm": 1.5586791038513184,
      "learning_rate": 4.590808825935841e-06,
      "loss": 0.0544,
      "step": 269475
    },
    {
      "epoch": 3.969013711138275,
      "grad_norm": 1.1228259801864624,
      "learning_rate": 4.589172453968857e-06,
      "loss": 0.055,
      "step": 269500
    },
    {
      "epoch": 3.969381894228362,
      "grad_norm": 1.6707308292388916,
      "learning_rate": 4.587536082001872e-06,
      "loss": 0.0485,
      "step": 269525
    },
    {
      "epoch": 3.969750077318449,
      "grad_norm": 1.7484828233718872,
      "learning_rate": 4.585899710034888e-06,
      "loss": 0.0548,
      "step": 269550
    },
    {
      "epoch": 3.970118260408536,
      "grad_norm": 1.5427132844924927,
      "learning_rate": 4.584263338067903e-06,
      "loss": 0.0548,
      "step": 269575
    },
    {
      "epoch": 3.970486443498623,
      "grad_norm": 1.3491164445877075,
      "learning_rate": 4.582626966100919e-06,
      "loss": 0.0542,
      "step": 269600
    },
    {
      "epoch": 3.97085462658871,
      "grad_norm": 0.9970679879188538,
      "learning_rate": 4.580990594133934e-06,
      "loss": 0.0508,
      "step": 269625
    },
    {
      "epoch": 3.971222809678797,
      "grad_norm": 1.3284558057785034,
      "learning_rate": 4.579354222166949e-06,
      "loss": 0.055,
      "step": 269650
    },
    {
      "epoch": 3.971590992768884,
      "grad_norm": 1.1876049041748047,
      "learning_rate": 4.577717850199965e-06,
      "loss": 0.054,
      "step": 269675
    },
    {
      "epoch": 3.971959175858971,
      "grad_norm": 1.3055858612060547,
      "learning_rate": 4.576081478232981e-06,
      "loss": 0.049,
      "step": 269700
    },
    {
      "epoch": 3.972327358949058,
      "grad_norm": 1.3097436428070068,
      "learning_rate": 4.574445106265996e-06,
      "loss": 0.0503,
      "step": 269725
    },
    {
      "epoch": 3.972695542039145,
      "grad_norm": 0.7987446188926697,
      "learning_rate": 4.572808734299011e-06,
      "loss": 0.0566,
      "step": 269750
    },
    {
      "epoch": 3.973063725129232,
      "grad_norm": 0.9984710216522217,
      "learning_rate": 4.571172362332027e-06,
      "loss": 0.0494,
      "step": 269775
    },
    {
      "epoch": 3.9734319082193195,
      "grad_norm": 1.938915491104126,
      "learning_rate": 4.569535990365042e-06,
      "loss": 0.0545,
      "step": 269800
    },
    {
      "epoch": 3.9738000913094065,
      "grad_norm": 1.6826567649841309,
      "learning_rate": 4.567899618398058e-06,
      "loss": 0.0516,
      "step": 269825
    },
    {
      "epoch": 3.9741682743994935,
      "grad_norm": 1.4236141443252563,
      "learning_rate": 4.566263246431073e-06,
      "loss": 0.0538,
      "step": 269850
    },
    {
      "epoch": 3.9745364574895805,
      "grad_norm": 1.7639985084533691,
      "learning_rate": 4.564626874464089e-06,
      "loss": 0.0567,
      "step": 269875
    },
    {
      "epoch": 3.9749046405796675,
      "grad_norm": 1.604337215423584,
      "learning_rate": 4.562990502497104e-06,
      "loss": 0.0555,
      "step": 269900
    },
    {
      "epoch": 3.9752728236697545,
      "grad_norm": 1.2788645029067993,
      "learning_rate": 4.561354130530119e-06,
      "loss": 0.0568,
      "step": 269925
    },
    {
      "epoch": 3.9756410067598416,
      "grad_norm": 1.5617719888687134,
      "learning_rate": 4.559717758563135e-06,
      "loss": 0.0547,
      "step": 269950
    },
    {
      "epoch": 3.9760091898499286,
      "grad_norm": 1.111019492149353,
      "learning_rate": 4.558081386596151e-06,
      "loss": 0.0519,
      "step": 269975
    },
    {
      "epoch": 3.9763773729400156,
      "grad_norm": 1.796332597732544,
      "learning_rate": 4.556445014629165e-06,
      "loss": 0.0556,
      "step": 270000
    },
    {
      "epoch": 3.9767455560301026,
      "grad_norm": 1.623569130897522,
      "learning_rate": 4.554808642662181e-06,
      "loss": 0.0517,
      "step": 270025
    },
    {
      "epoch": 3.9771137391201896,
      "grad_norm": 0.7278119921684265,
      "learning_rate": 4.553172270695197e-06,
      "loss": 0.0524,
      "step": 270050
    },
    {
      "epoch": 3.9774819222102766,
      "grad_norm": 1.6087461709976196,
      "learning_rate": 4.551535898728212e-06,
      "loss": 0.0581,
      "step": 270075
    },
    {
      "epoch": 3.9778501053003636,
      "grad_norm": 1.0999680757522583,
      "learning_rate": 4.549899526761227e-06,
      "loss": 0.0504,
      "step": 270100
    },
    {
      "epoch": 3.9782182883904507,
      "grad_norm": 1.5412969589233398,
      "learning_rate": 4.5482631547942426e-06,
      "loss": 0.0545,
      "step": 270125
    },
    {
      "epoch": 3.9785864714805377,
      "grad_norm": 1.3886959552764893,
      "learning_rate": 4.546626782827259e-06,
      "loss": 0.0625,
      "step": 270150
    },
    {
      "epoch": 3.978954654570625,
      "grad_norm": 1.2410922050476074,
      "learning_rate": 4.544990410860274e-06,
      "loss": 0.0581,
      "step": 270175
    },
    {
      "epoch": 3.979322837660712,
      "grad_norm": 1.2803741693496704,
      "learning_rate": 4.543354038893289e-06,
      "loss": 0.0449,
      "step": 270200
    },
    {
      "epoch": 3.979691020750799,
      "grad_norm": 0.9015976786613464,
      "learning_rate": 4.541717666926305e-06,
      "loss": 0.0501,
      "step": 270225
    },
    {
      "epoch": 3.980059203840886,
      "grad_norm": 1.4364522695541382,
      "learning_rate": 4.540081294959321e-06,
      "loss": 0.0576,
      "step": 270250
    },
    {
      "epoch": 3.980427386930973,
      "grad_norm": 1.476516842842102,
      "learning_rate": 4.538444922992335e-06,
      "loss": 0.0577,
      "step": 270275
    },
    {
      "epoch": 3.98079557002106,
      "grad_norm": 1.1962724924087524,
      "learning_rate": 4.536808551025351e-06,
      "loss": 0.0454,
      "step": 270300
    },
    {
      "epoch": 3.9811637531111472,
      "grad_norm": 1.4422894716262817,
      "learning_rate": 4.535172179058367e-06,
      "loss": 0.0553,
      "step": 270325
    },
    {
      "epoch": 3.9815319362012342,
      "grad_norm": 1.0398858785629272,
      "learning_rate": 4.533535807091382e-06,
      "loss": 0.0487,
      "step": 270350
    },
    {
      "epoch": 3.9819001192913213,
      "grad_norm": 0.9367135763168335,
      "learning_rate": 4.531899435124397e-06,
      "loss": 0.0509,
      "step": 270375
    },
    {
      "epoch": 3.9822683023814083,
      "grad_norm": 1.9063533544540405,
      "learning_rate": 4.530263063157413e-06,
      "loss": 0.0587,
      "step": 270400
    },
    {
      "epoch": 3.9826364854714953,
      "grad_norm": 1.2517178058624268,
      "learning_rate": 4.528626691190428e-06,
      "loss": 0.0542,
      "step": 270425
    },
    {
      "epoch": 3.9830046685615823,
      "grad_norm": 1.330877423286438,
      "learning_rate": 4.526990319223444e-06,
      "loss": 0.0501,
      "step": 270450
    },
    {
      "epoch": 3.9833728516516693,
      "grad_norm": 1.5600701570510864,
      "learning_rate": 4.5253539472564585e-06,
      "loss": 0.0572,
      "step": 270475
    },
    {
      "epoch": 3.9837410347417563,
      "grad_norm": 1.7591938972473145,
      "learning_rate": 4.523717575289475e-06,
      "loss": 0.0562,
      "step": 270500
    },
    {
      "epoch": 3.9841092178318434,
      "grad_norm": 1.2087832689285278,
      "learning_rate": 4.52208120332249e-06,
      "loss": 0.0587,
      "step": 270525
    },
    {
      "epoch": 3.9844774009219304,
      "grad_norm": 1.0783907175064087,
      "learning_rate": 4.520444831355505e-06,
      "loss": 0.0538,
      "step": 270550
    },
    {
      "epoch": 3.9848455840120174,
      "grad_norm": 1.8448988199234009,
      "learning_rate": 4.518808459388521e-06,
      "loss": 0.0566,
      "step": 270575
    },
    {
      "epoch": 3.9852137671021044,
      "grad_norm": 1.0080883502960205,
      "learning_rate": 4.517172087421537e-06,
      "loss": 0.0567,
      "step": 270600
    },
    {
      "epoch": 3.9855819501921914,
      "grad_norm": 1.3988831043243408,
      "learning_rate": 4.515535715454552e-06,
      "loss": 0.0508,
      "step": 270625
    },
    {
      "epoch": 3.9859501332822784,
      "grad_norm": 1.3583345413208008,
      "learning_rate": 4.513899343487567e-06,
      "loss": 0.0602,
      "step": 270650
    },
    {
      "epoch": 3.9863183163723654,
      "grad_norm": 1.6458958387374878,
      "learning_rate": 4.512262971520583e-06,
      "loss": 0.0534,
      "step": 270675
    },
    {
      "epoch": 3.9866864994624525,
      "grad_norm": 1.4257688522338867,
      "learning_rate": 4.510626599553598e-06,
      "loss": 0.0498,
      "step": 270700
    },
    {
      "epoch": 3.9870546825525395,
      "grad_norm": 0.8840227127075195,
      "learning_rate": 4.508990227586614e-06,
      "loss": 0.0555,
      "step": 270725
    },
    {
      "epoch": 3.9874228656426265,
      "grad_norm": 1.1897773742675781,
      "learning_rate": 4.5073538556196286e-06,
      "loss": 0.0562,
      "step": 270750
    },
    {
      "epoch": 3.987791048732714,
      "grad_norm": 1.2229746580123901,
      "learning_rate": 4.505717483652645e-06,
      "loss": 0.0553,
      "step": 270775
    },
    {
      "epoch": 3.988159231822801,
      "grad_norm": 0.9923685193061829,
      "learning_rate": 4.50408111168566e-06,
      "loss": 0.0513,
      "step": 270800
    },
    {
      "epoch": 3.988527414912888,
      "grad_norm": 0.8561668992042542,
      "learning_rate": 4.502444739718675e-06,
      "loss": 0.0501,
      "step": 270825
    },
    {
      "epoch": 3.988895598002975,
      "grad_norm": 1.9197190999984741,
      "learning_rate": 4.500808367751691e-06,
      "loss": 0.057,
      "step": 270850
    },
    {
      "epoch": 3.989263781093062,
      "grad_norm": 1.3146735429763794,
      "learning_rate": 4.499171995784706e-06,
      "loss": 0.0609,
      "step": 270875
    },
    {
      "epoch": 3.989631964183149,
      "grad_norm": 1.598151683807373,
      "learning_rate": 4.497535623817721e-06,
      "loss": 0.0557,
      "step": 270900
    },
    {
      "epoch": 3.990000147273236,
      "grad_norm": 1.4592117071151733,
      "learning_rate": 4.495899251850737e-06,
      "loss": 0.0607,
      "step": 270925
    },
    {
      "epoch": 3.990368330363323,
      "grad_norm": 1.409018874168396,
      "learning_rate": 4.494262879883753e-06,
      "loss": 0.0529,
      "step": 270950
    },
    {
      "epoch": 3.99073651345341,
      "grad_norm": 1.162003517150879,
      "learning_rate": 4.492626507916768e-06,
      "loss": 0.0535,
      "step": 270975
    },
    {
      "epoch": 3.991104696543497,
      "grad_norm": 1.278775691986084,
      "learning_rate": 4.490990135949783e-06,
      "loss": 0.0572,
      "step": 271000
    },
    {
      "epoch": 3.991472879633584,
      "grad_norm": 0.8224690556526184,
      "learning_rate": 4.489353763982799e-06,
      "loss": 0.0527,
      "step": 271025
    },
    {
      "epoch": 3.991841062723671,
      "grad_norm": 1.4066331386566162,
      "learning_rate": 4.487717392015814e-06,
      "loss": 0.0553,
      "step": 271050
    },
    {
      "epoch": 3.992209245813758,
      "grad_norm": 1.0099693536758423,
      "learning_rate": 4.48608102004883e-06,
      "loss": 0.0586,
      "step": 271075
    },
    {
      "epoch": 3.992577428903845,
      "grad_norm": 1.4977132081985474,
      "learning_rate": 4.4844446480818445e-06,
      "loss": 0.0474,
      "step": 271100
    },
    {
      "epoch": 3.9929456119939326,
      "grad_norm": 1.8170990943908691,
      "learning_rate": 4.482808276114861e-06,
      "loss": 0.0636,
      "step": 271125
    },
    {
      "epoch": 3.9933137950840196,
      "grad_norm": 1.537147879600525,
      "learning_rate": 4.481171904147876e-06,
      "loss": 0.0509,
      "step": 271150
    },
    {
      "epoch": 3.9936819781741066,
      "grad_norm": 1.4248909950256348,
      "learning_rate": 4.479535532180891e-06,
      "loss": 0.0543,
      "step": 271175
    },
    {
      "epoch": 3.9940501612641937,
      "grad_norm": 0.7828039526939392,
      "learning_rate": 4.4778991602139074e-06,
      "loss": 0.0486,
      "step": 271200
    },
    {
      "epoch": 3.9944183443542807,
      "grad_norm": 1.4615932703018188,
      "learning_rate": 4.476262788246922e-06,
      "loss": 0.0573,
      "step": 271225
    },
    {
      "epoch": 3.9947865274443677,
      "grad_norm": 0.7762133479118347,
      "learning_rate": 4.474626416279938e-06,
      "loss": 0.0567,
      "step": 271250
    },
    {
      "epoch": 3.9951547105344547,
      "grad_norm": 1.386361002922058,
      "learning_rate": 4.472990044312953e-06,
      "loss": 0.0522,
      "step": 271275
    },
    {
      "epoch": 3.9955228936245417,
      "grad_norm": 1.1249874830245972,
      "learning_rate": 4.471353672345969e-06,
      "loss": 0.0607,
      "step": 271300
    },
    {
      "epoch": 3.9958910767146287,
      "grad_norm": 0.9048838019371033,
      "learning_rate": 4.469717300378984e-06,
      "loss": 0.0491,
      "step": 271325
    },
    {
      "epoch": 3.9962592598047157,
      "grad_norm": 1.614412546157837,
      "learning_rate": 4.468080928412e-06,
      "loss": 0.0507,
      "step": 271350
    },
    {
      "epoch": 3.9966274428948028,
      "grad_norm": 1.4011964797973633,
      "learning_rate": 4.4664445564450146e-06,
      "loss": 0.052,
      "step": 271375
    },
    {
      "epoch": 3.9969956259848898,
      "grad_norm": 1.6174417734146118,
      "learning_rate": 4.464808184478031e-06,
      "loss": 0.0498,
      "step": 271400
    },
    {
      "epoch": 3.997363809074977,
      "grad_norm": 1.1240805387496948,
      "learning_rate": 4.463171812511046e-06,
      "loss": 0.0514,
      "step": 271425
    },
    {
      "epoch": 3.997731992165064,
      "grad_norm": 0.9346145987510681,
      "learning_rate": 4.461535440544061e-06,
      "loss": 0.0478,
      "step": 271450
    },
    {
      "epoch": 3.998100175255151,
      "grad_norm": 1.1280893087387085,
      "learning_rate": 4.459899068577077e-06,
      "loss": 0.05,
      "step": 271475
    },
    {
      "epoch": 3.998468358345238,
      "grad_norm": 1.5235371589660645,
      "learning_rate": 4.458262696610092e-06,
      "loss": 0.0571,
      "step": 271500
    },
    {
      "epoch": 3.998836541435325,
      "grad_norm": 1.5510966777801514,
      "learning_rate": 4.456626324643107e-06,
      "loss": 0.0491,
      "step": 271525
    },
    {
      "epoch": 3.999204724525412,
      "grad_norm": 1.0778614282608032,
      "learning_rate": 4.454989952676123e-06,
      "loss": 0.0483,
      "step": 271550
    },
    {
      "epoch": 3.999572907615499,
      "grad_norm": 1.274242639541626,
      "learning_rate": 4.453353580709138e-06,
      "loss": 0.0556,
      "step": 271575
    },
    {
      "epoch": 3.999941090705586,
      "grad_norm": 1.295026183128357,
      "learning_rate": 4.451717208742154e-06,
      "loss": 0.0519,
      "step": 271600
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.058289363980293274,
      "eval_runtime": 112.6241,
      "eval_samples_per_second": 3149.84,
      "eval_steps_per_second": 6.153,
      "step": 271604
    },
    {
      "epoch": 4.000309273795673,
      "grad_norm": 1.0060703754425049,
      "learning_rate": 4.450080836775169e-06,
      "loss": 0.0538,
      "step": 271625
    },
    {
      "epoch": 4.00067745688576,
      "grad_norm": 1.6256768703460693,
      "learning_rate": 4.448444464808185e-06,
      "loss": 0.0523,
      "step": 271650
    },
    {
      "epoch": 4.001045639975847,
      "grad_norm": 1.5601869821548462,
      "learning_rate": 4.446808092841201e-06,
      "loss": 0.0566,
      "step": 271675
    },
    {
      "epoch": 4.001413823065934,
      "grad_norm": 1.1750060319900513,
      "learning_rate": 4.445171720874216e-06,
      "loss": 0.0642,
      "step": 271700
    },
    {
      "epoch": 4.001782006156021,
      "grad_norm": 1.1978039741516113,
      "learning_rate": 4.443535348907231e-06,
      "loss": 0.0529,
      "step": 271725
    },
    {
      "epoch": 4.002150189246108,
      "grad_norm": 1.0152662992477417,
      "learning_rate": 4.441898976940247e-06,
      "loss": 0.0509,
      "step": 271750
    },
    {
      "epoch": 4.002518372336195,
      "grad_norm": 1.405200481414795,
      "learning_rate": 4.440262604973262e-06,
      "loss": 0.0546,
      "step": 271775
    },
    {
      "epoch": 4.002886555426282,
      "grad_norm": 1.8479878902435303,
      "learning_rate": 4.438626233006277e-06,
      "loss": 0.0545,
      "step": 271800
    },
    {
      "epoch": 4.003254738516369,
      "grad_norm": 1.265377163887024,
      "learning_rate": 4.4369898610392934e-06,
      "loss": 0.0633,
      "step": 271825
    },
    {
      "epoch": 4.003622921606456,
      "grad_norm": 1.418918251991272,
      "learning_rate": 4.435353489072308e-06,
      "loss": 0.0524,
      "step": 271850
    },
    {
      "epoch": 4.003991104696543,
      "grad_norm": 1.582596778869629,
      "learning_rate": 4.433717117105324e-06,
      "loss": 0.054,
      "step": 271875
    },
    {
      "epoch": 4.004359287786631,
      "grad_norm": 1.4304360151290894,
      "learning_rate": 4.432080745138339e-06,
      "loss": 0.0525,
      "step": 271900
    },
    {
      "epoch": 4.004727470876718,
      "grad_norm": 1.1909329891204834,
      "learning_rate": 4.430444373171355e-06,
      "loss": 0.0545,
      "step": 271925
    },
    {
      "epoch": 4.005095653966805,
      "grad_norm": 1.5431170463562012,
      "learning_rate": 4.42880800120437e-06,
      "loss": 0.046,
      "step": 271950
    },
    {
      "epoch": 4.005463837056892,
      "grad_norm": 1.464337706565857,
      "learning_rate": 4.427171629237385e-06,
      "loss": 0.048,
      "step": 271975
    },
    {
      "epoch": 4.005832020146979,
      "grad_norm": 0.8372604250907898,
      "learning_rate": 4.4255352572704006e-06,
      "loss": 0.0545,
      "step": 272000
    },
    {
      "epoch": 4.006200203237066,
      "grad_norm": 0.969363808631897,
      "learning_rate": 4.423898885303417e-06,
      "loss": 0.0471,
      "step": 272025
    },
    {
      "epoch": 4.006568386327153,
      "grad_norm": 1.3070334196090698,
      "learning_rate": 4.422262513336432e-06,
      "loss": 0.0523,
      "step": 272050
    },
    {
      "epoch": 4.00693656941724,
      "grad_norm": 1.105718731880188,
      "learning_rate": 4.420626141369447e-06,
      "loss": 0.0478,
      "step": 272075
    },
    {
      "epoch": 4.007304752507327,
      "grad_norm": 1.1749552488327026,
      "learning_rate": 4.418989769402463e-06,
      "loss": 0.049,
      "step": 272100
    },
    {
      "epoch": 4.007672935597414,
      "grad_norm": 1.798275351524353,
      "learning_rate": 4.417353397435478e-06,
      "loss": 0.0511,
      "step": 272125
    },
    {
      "epoch": 4.008041118687501,
      "grad_norm": 1.2194712162017822,
      "learning_rate": 4.415717025468494e-06,
      "loss": 0.0469,
      "step": 272150
    },
    {
      "epoch": 4.008409301777588,
      "grad_norm": 1.7627757787704468,
      "learning_rate": 4.414080653501509e-06,
      "loss": 0.0523,
      "step": 272175
    },
    {
      "epoch": 4.008777484867675,
      "grad_norm": 1.3913440704345703,
      "learning_rate": 4.412444281534525e-06,
      "loss": 0.0508,
      "step": 272200
    },
    {
      "epoch": 4.009145667957762,
      "grad_norm": 1.026561975479126,
      "learning_rate": 4.41080790956754e-06,
      "loss": 0.055,
      "step": 272225
    },
    {
      "epoch": 4.009513851047849,
      "grad_norm": 1.353627324104309,
      "learning_rate": 4.409171537600555e-06,
      "loss": 0.0545,
      "step": 272250
    },
    {
      "epoch": 4.009882034137936,
      "grad_norm": 1.5493100881576538,
      "learning_rate": 4.407535165633571e-06,
      "loss": 0.0486,
      "step": 272275
    },
    {
      "epoch": 4.010250217228023,
      "grad_norm": 1.1753915548324585,
      "learning_rate": 4.405898793666587e-06,
      "loss": 0.053,
      "step": 272300
    },
    {
      "epoch": 4.01061840031811,
      "grad_norm": 1.3746318817138672,
      "learning_rate": 4.404262421699601e-06,
      "loss": 0.0486,
      "step": 272325
    },
    {
      "epoch": 4.010986583408197,
      "grad_norm": 1.4474351406097412,
      "learning_rate": 4.402626049732617e-06,
      "loss": 0.053,
      "step": 272350
    },
    {
      "epoch": 4.011354766498284,
      "grad_norm": 1.0470190048217773,
      "learning_rate": 4.400989677765633e-06,
      "loss": 0.0548,
      "step": 272375
    },
    {
      "epoch": 4.011722949588371,
      "grad_norm": 1.1962851285934448,
      "learning_rate": 4.399353305798648e-06,
      "loss": 0.0563,
      "step": 272400
    },
    {
      "epoch": 4.012091132678458,
      "grad_norm": 1.5802463293075562,
      "learning_rate": 4.397716933831663e-06,
      "loss": 0.0536,
      "step": 272425
    },
    {
      "epoch": 4.012459315768545,
      "grad_norm": 1.2477504014968872,
      "learning_rate": 4.3960805618646794e-06,
      "loss": 0.067,
      "step": 272450
    },
    {
      "epoch": 4.012827498858632,
      "grad_norm": 1.3278979063034058,
      "learning_rate": 4.394509644776373e-06,
      "loss": 0.0536,
      "step": 272475
    },
    {
      "epoch": 4.013195681948719,
      "grad_norm": 1.1894779205322266,
      "learning_rate": 4.3928732728093895e-06,
      "loss": 0.0478,
      "step": 272500
    },
    {
      "epoch": 4.013563865038806,
      "grad_norm": 1.1184048652648926,
      "learning_rate": 4.391236900842405e-06,
      "loss": 0.0542,
      "step": 272525
    },
    {
      "epoch": 4.013932048128893,
      "grad_norm": 1.0860342979431152,
      "learning_rate": 4.38960052887542e-06,
      "loss": 0.0477,
      "step": 272550
    },
    {
      "epoch": 4.01430023121898,
      "grad_norm": 0.7316974997520447,
      "learning_rate": 4.387964156908435e-06,
      "loss": 0.0534,
      "step": 272575
    },
    {
      "epoch": 4.014668414309067,
      "grad_norm": 1.1899856328964233,
      "learning_rate": 4.386327784941451e-06,
      "loss": 0.0506,
      "step": 272600
    },
    {
      "epoch": 4.015036597399154,
      "grad_norm": 1.1814501285552979,
      "learning_rate": 4.384691412974467e-06,
      "loss": 0.0557,
      "step": 272625
    },
    {
      "epoch": 4.015404780489241,
      "grad_norm": 1.3784784078598022,
      "learning_rate": 4.383055041007482e-06,
      "loss": 0.0521,
      "step": 272650
    },
    {
      "epoch": 4.0157729635793284,
      "grad_norm": 0.9269449710845947,
      "learning_rate": 4.3814186690404974e-06,
      "loss": 0.0513,
      "step": 272675
    },
    {
      "epoch": 4.0161411466694155,
      "grad_norm": 1.6109111309051514,
      "learning_rate": 4.379782297073513e-06,
      "loss": 0.0481,
      "step": 272700
    },
    {
      "epoch": 4.0165093297595025,
      "grad_norm": 1.2411295175552368,
      "learning_rate": 4.378145925106528e-06,
      "loss": 0.053,
      "step": 272725
    },
    {
      "epoch": 4.0168775128495895,
      "grad_norm": 1.662144660949707,
      "learning_rate": 4.376509553139543e-06,
      "loss": 0.056,
      "step": 272750
    },
    {
      "epoch": 4.0172456959396765,
      "grad_norm": 1.5705420970916748,
      "learning_rate": 4.3748731811725595e-06,
      "loss": 0.055,
      "step": 272775
    },
    {
      "epoch": 4.0176138790297635,
      "grad_norm": 1.402494192123413,
      "learning_rate": 4.373236809205574e-06,
      "loss": 0.0438,
      "step": 272800
    },
    {
      "epoch": 4.0179820621198505,
      "grad_norm": 1.5950024127960205,
      "learning_rate": 4.37160043723859e-06,
      "loss": 0.0467,
      "step": 272825
    },
    {
      "epoch": 4.0183502452099376,
      "grad_norm": 1.5156582593917847,
      "learning_rate": 4.369964065271605e-06,
      "loss": 0.0469,
      "step": 272850
    },
    {
      "epoch": 4.0187184283000255,
      "grad_norm": 1.172438621520996,
      "learning_rate": 4.368327693304621e-06,
      "loss": 0.0486,
      "step": 272875
    },
    {
      "epoch": 4.0190866113901125,
      "grad_norm": 1.0187807083129883,
      "learning_rate": 4.366691321337636e-06,
      "loss": 0.0586,
      "step": 272900
    },
    {
      "epoch": 4.0194547944801995,
      "grad_norm": 1.3460967540740967,
      "learning_rate": 4.365054949370652e-06,
      "loss": 0.054,
      "step": 272925
    },
    {
      "epoch": 4.0198229775702865,
      "grad_norm": 1.8421919345855713,
      "learning_rate": 4.363418577403667e-06,
      "loss": 0.0517,
      "step": 272950
    },
    {
      "epoch": 4.0201911606603735,
      "grad_norm": 1.0277374982833862,
      "learning_rate": 4.361782205436683e-06,
      "loss": 0.0455,
      "step": 272975
    },
    {
      "epoch": 4.0205593437504605,
      "grad_norm": 0.9868427515029907,
      "learning_rate": 4.360145833469698e-06,
      "loss": 0.0535,
      "step": 273000
    },
    {
      "epoch": 4.0209275268405476,
      "grad_norm": 1.530604600906372,
      "learning_rate": 4.358509461502713e-06,
      "loss": 0.0516,
      "step": 273025
    },
    {
      "epoch": 4.021295709930635,
      "grad_norm": 0.7439272403717041,
      "learning_rate": 4.356873089535729e-06,
      "loss": 0.0455,
      "step": 273050
    },
    {
      "epoch": 4.021663893020722,
      "grad_norm": 1.4210286140441895,
      "learning_rate": 4.355236717568744e-06,
      "loss": 0.053,
      "step": 273075
    },
    {
      "epoch": 4.022032076110809,
      "grad_norm": 1.286408543586731,
      "learning_rate": 4.35360034560176e-06,
      "loss": 0.0546,
      "step": 273100
    },
    {
      "epoch": 4.022400259200896,
      "grad_norm": 1.2621315717697144,
      "learning_rate": 4.3519639736347755e-06,
      "loss": 0.05,
      "step": 273125
    },
    {
      "epoch": 4.022768442290983,
      "grad_norm": 1.1417889595031738,
      "learning_rate": 4.350327601667791e-06,
      "loss": 0.0485,
      "step": 273150
    },
    {
      "epoch": 4.02313662538107,
      "grad_norm": 2.103334665298462,
      "learning_rate": 4.348691229700806e-06,
      "loss": 0.0502,
      "step": 273175
    },
    {
      "epoch": 4.023504808471157,
      "grad_norm": 1.42632257938385,
      "learning_rate": 4.347054857733821e-06,
      "loss": 0.0516,
      "step": 273200
    },
    {
      "epoch": 4.023872991561244,
      "grad_norm": 1.1343106031417847,
      "learning_rate": 4.345418485766837e-06,
      "loss": 0.0503,
      "step": 273225
    },
    {
      "epoch": 4.024241174651331,
      "grad_norm": 1.399665117263794,
      "learning_rate": 4.343782113799853e-06,
      "loss": 0.0543,
      "step": 273250
    },
    {
      "epoch": 4.024609357741418,
      "grad_norm": 1.361912488937378,
      "learning_rate": 4.342145741832868e-06,
      "loss": 0.0522,
      "step": 273275
    },
    {
      "epoch": 4.024977540831505,
      "grad_norm": 1.5529487133026123,
      "learning_rate": 4.3405093698658834e-06,
      "loss": 0.055,
      "step": 273300
    },
    {
      "epoch": 4.025345723921592,
      "grad_norm": 1.268615961074829,
      "learning_rate": 4.338872997898899e-06,
      "loss": 0.0511,
      "step": 273325
    },
    {
      "epoch": 4.025713907011679,
      "grad_norm": 1.5254231691360474,
      "learning_rate": 4.337236625931914e-06,
      "loss": 0.0533,
      "step": 273350
    },
    {
      "epoch": 4.026082090101766,
      "grad_norm": 1.4463733434677124,
      "learning_rate": 4.335600253964929e-06,
      "loss": 0.0575,
      "step": 273375
    },
    {
      "epoch": 4.026450273191853,
      "grad_norm": 1.6330395936965942,
      "learning_rate": 4.3339638819979455e-06,
      "loss": 0.057,
      "step": 273400
    },
    {
      "epoch": 4.02681845628194,
      "grad_norm": 1.4489814043045044,
      "learning_rate": 4.33232751003096e-06,
      "loss": 0.0537,
      "step": 273425
    },
    {
      "epoch": 4.027186639372027,
      "grad_norm": 1.350322961807251,
      "learning_rate": 4.330691138063976e-06,
      "loss": 0.0499,
      "step": 273450
    },
    {
      "epoch": 4.027554822462114,
      "grad_norm": 1.1626771688461304,
      "learning_rate": 4.329054766096991e-06,
      "loss": 0.048,
      "step": 273475
    },
    {
      "epoch": 4.027923005552201,
      "grad_norm": 1.0108155012130737,
      "learning_rate": 4.327418394130007e-06,
      "loss": 0.0527,
      "step": 273500
    },
    {
      "epoch": 4.028291188642288,
      "grad_norm": 1.025947093963623,
      "learning_rate": 4.325782022163022e-06,
      "loss": 0.0497,
      "step": 273525
    },
    {
      "epoch": 4.028659371732375,
      "grad_norm": 0.9710474014282227,
      "learning_rate": 4.324211105074717e-06,
      "loss": 0.0526,
      "step": 273550
    },
    {
      "epoch": 4.029027554822462,
      "grad_norm": 0.8908804655075073,
      "learning_rate": 4.322574733107733e-06,
      "loss": 0.0558,
      "step": 273575
    },
    {
      "epoch": 4.029395737912549,
      "grad_norm": 1.1707205772399902,
      "learning_rate": 4.320938361140748e-06,
      "loss": 0.0425,
      "step": 273600
    },
    {
      "epoch": 4.029763921002636,
      "grad_norm": 1.9893933534622192,
      "learning_rate": 4.3193019891737635e-06,
      "loss": 0.0541,
      "step": 273625
    },
    {
      "epoch": 4.030132104092723,
      "grad_norm": 1.2278904914855957,
      "learning_rate": 4.317665617206779e-06,
      "loss": 0.0488,
      "step": 273650
    },
    {
      "epoch": 4.03050028718281,
      "grad_norm": 1.466922640800476,
      "learning_rate": 4.316029245239794e-06,
      "loss": 0.0507,
      "step": 273675
    },
    {
      "epoch": 4.030868470272897,
      "grad_norm": 1.2266409397125244,
      "learning_rate": 4.3143928732728094e-06,
      "loss": 0.0505,
      "step": 273700
    },
    {
      "epoch": 4.031236653362984,
      "grad_norm": 1.2560009956359863,
      "learning_rate": 4.312756501305826e-06,
      "loss": 0.0495,
      "step": 273725
    },
    {
      "epoch": 4.031604836453071,
      "grad_norm": 1.5691561698913574,
      "learning_rate": 4.311120129338841e-06,
      "loss": 0.058,
      "step": 273750
    },
    {
      "epoch": 4.031973019543158,
      "grad_norm": 1.390404224395752,
      "learning_rate": 4.309483757371856e-06,
      "loss": 0.0472,
      "step": 273775
    },
    {
      "epoch": 4.032341202633245,
      "grad_norm": 1.7149642705917358,
      "learning_rate": 4.3078473854048715e-06,
      "loss": 0.0545,
      "step": 273800
    },
    {
      "epoch": 4.032709385723333,
      "grad_norm": 1.1567271947860718,
      "learning_rate": 4.306211013437887e-06,
      "loss": 0.0432,
      "step": 273825
    },
    {
      "epoch": 4.03307756881342,
      "grad_norm": 1.3016043901443481,
      "learning_rate": 4.304574641470902e-06,
      "loss": 0.0502,
      "step": 273850
    },
    {
      "epoch": 4.033445751903507,
      "grad_norm": 1.5264921188354492,
      "learning_rate": 4.302938269503918e-06,
      "loss": 0.0527,
      "step": 273875
    },
    {
      "epoch": 4.033813934993594,
      "grad_norm": 1.398117184638977,
      "learning_rate": 4.301301897536933e-06,
      "loss": 0.0474,
      "step": 273900
    },
    {
      "epoch": 4.034182118083681,
      "grad_norm": 1.3414297103881836,
      "learning_rate": 4.299665525569949e-06,
      "loss": 0.0436,
      "step": 273925
    },
    {
      "epoch": 4.034550301173768,
      "grad_norm": 1.3082327842712402,
      "learning_rate": 4.298029153602964e-06,
      "loss": 0.0564,
      "step": 273950
    },
    {
      "epoch": 4.034918484263855,
      "grad_norm": 1.8074843883514404,
      "learning_rate": 4.2963927816359795e-06,
      "loss": 0.0561,
      "step": 273975
    },
    {
      "epoch": 4.035286667353942,
      "grad_norm": 1.2525572776794434,
      "learning_rate": 4.294756409668995e-06,
      "loss": 0.0516,
      "step": 274000
    },
    {
      "epoch": 4.035654850444029,
      "grad_norm": 0.9967262744903564,
      "learning_rate": 4.29312003770201e-06,
      "loss": 0.0515,
      "step": 274025
    },
    {
      "epoch": 4.036023033534116,
      "grad_norm": 0.7842008471488953,
      "learning_rate": 4.291483665735026e-06,
      "loss": 0.0538,
      "step": 274050
    },
    {
      "epoch": 4.036391216624203,
      "grad_norm": 1.54563307762146,
      "learning_rate": 4.2898472937680415e-06,
      "loss": 0.049,
      "step": 274075
    },
    {
      "epoch": 4.03675939971429,
      "grad_norm": 1.1523215770721436,
      "learning_rate": 4.288210921801057e-06,
      "loss": 0.053,
      "step": 274100
    },
    {
      "epoch": 4.037127582804377,
      "grad_norm": 1.4274777173995972,
      "learning_rate": 4.286574549834072e-06,
      "loss": 0.0516,
      "step": 274125
    },
    {
      "epoch": 4.037495765894464,
      "grad_norm": 1.737199306488037,
      "learning_rate": 4.284938177867088e-06,
      "loss": 0.0494,
      "step": 274150
    },
    {
      "epoch": 4.037863948984551,
      "grad_norm": 0.4694286584854126,
      "learning_rate": 4.283301805900103e-06,
      "loss": 0.0536,
      "step": 274175
    },
    {
      "epoch": 4.038232132074638,
      "grad_norm": 1.7536449432373047,
      "learning_rate": 4.281665433933119e-06,
      "loss": 0.0569,
      "step": 274200
    },
    {
      "epoch": 4.038600315164725,
      "grad_norm": 1.1667371988296509,
      "learning_rate": 4.280029061966134e-06,
      "loss": 0.0511,
      "step": 274225
    },
    {
      "epoch": 4.038968498254812,
      "grad_norm": 1.3396191596984863,
      "learning_rate": 4.2783926899991495e-06,
      "loss": 0.0561,
      "step": 274250
    },
    {
      "epoch": 4.039336681344899,
      "grad_norm": 1.516628384590149,
      "learning_rate": 4.276756318032165e-06,
      "loss": 0.0535,
      "step": 274275
    },
    {
      "epoch": 4.039704864434986,
      "grad_norm": 1.5235519409179688,
      "learning_rate": 4.27511994606518e-06,
      "loss": 0.0545,
      "step": 274300
    },
    {
      "epoch": 4.040073047525073,
      "grad_norm": 1.9207894802093506,
      "learning_rate": 4.2734835740981954e-06,
      "loss": 0.0494,
      "step": 274325
    },
    {
      "epoch": 4.04044123061516,
      "grad_norm": 1.4713939428329468,
      "learning_rate": 4.271847202131212e-06,
      "loss": 0.05,
      "step": 274350
    },
    {
      "epoch": 4.040809413705247,
      "grad_norm": 1.1645876169204712,
      "learning_rate": 4.270210830164226e-06,
      "loss": 0.0486,
      "step": 274375
    },
    {
      "epoch": 4.041177596795334,
      "grad_norm": 0.8513479232788086,
      "learning_rate": 4.268574458197242e-06,
      "loss": 0.0488,
      "step": 274400
    },
    {
      "epoch": 4.041545779885421,
      "grad_norm": 1.943671464920044,
      "learning_rate": 4.2669380862302575e-06,
      "loss": 0.0507,
      "step": 274425
    },
    {
      "epoch": 4.041913962975508,
      "grad_norm": 1.4613126516342163,
      "learning_rate": 4.265301714263273e-06,
      "loss": 0.0495,
      "step": 274450
    },
    {
      "epoch": 4.042282146065595,
      "grad_norm": 1.0549331903457642,
      "learning_rate": 4.263665342296288e-06,
      "loss": 0.0459,
      "step": 274475
    },
    {
      "epoch": 4.042650329155682,
      "grad_norm": 1.629569411277771,
      "learning_rate": 4.262028970329304e-06,
      "loss": 0.0459,
      "step": 274500
    },
    {
      "epoch": 4.043018512245769,
      "grad_norm": 1.6735188961029053,
      "learning_rate": 4.2603925983623196e-06,
      "loss": 0.0529,
      "step": 274525
    },
    {
      "epoch": 4.043386695335856,
      "grad_norm": 1.1502186059951782,
      "learning_rate": 4.258756226395335e-06,
      "loss": 0.0474,
      "step": 274550
    },
    {
      "epoch": 4.043754878425943,
      "grad_norm": 1.0216065645217896,
      "learning_rate": 4.25711985442835e-06,
      "loss": 0.0554,
      "step": 274575
    },
    {
      "epoch": 4.04412306151603,
      "grad_norm": 1.4073421955108643,
      "learning_rate": 4.2554834824613655e-06,
      "loss": 0.0561,
      "step": 274600
    },
    {
      "epoch": 4.044491244606117,
      "grad_norm": 1.106372594833374,
      "learning_rate": 4.253847110494382e-06,
      "loss": 0.0584,
      "step": 274625
    },
    {
      "epoch": 4.044859427696204,
      "grad_norm": 1.419223427772522,
      "learning_rate": 4.252210738527396e-06,
      "loss": 0.0471,
      "step": 274650
    },
    {
      "epoch": 4.0452276107862915,
      "grad_norm": 1.1996541023254395,
      "learning_rate": 4.250574366560412e-06,
      "loss": 0.0497,
      "step": 274675
    },
    {
      "epoch": 4.0455957938763785,
      "grad_norm": 1.3995702266693115,
      "learning_rate": 4.2489379945934275e-06,
      "loss": 0.0493,
      "step": 274700
    },
    {
      "epoch": 4.0459639769664655,
      "grad_norm": 1.0698034763336182,
      "learning_rate": 4.247301622626443e-06,
      "loss": 0.047,
      "step": 274725
    },
    {
      "epoch": 4.0463321600565525,
      "grad_norm": 1.38143789768219,
      "learning_rate": 4.245665250659458e-06,
      "loss": 0.0501,
      "step": 274750
    },
    {
      "epoch": 4.04670034314664,
      "grad_norm": 1.5964407920837402,
      "learning_rate": 4.2440288786924734e-06,
      "loss": 0.0504,
      "step": 274775
    },
    {
      "epoch": 4.047068526236727,
      "grad_norm": 1.6865134239196777,
      "learning_rate": 4.242392506725489e-06,
      "loss": 0.0448,
      "step": 274800
    },
    {
      "epoch": 4.047436709326814,
      "grad_norm": 1.5240774154663086,
      "learning_rate": 4.240756134758505e-06,
      "loss": 0.0509,
      "step": 274825
    },
    {
      "epoch": 4.0478048924169014,
      "grad_norm": 1.4935787916183472,
      "learning_rate": 4.23911976279152e-06,
      "loss": 0.0498,
      "step": 274850
    },
    {
      "epoch": 4.0481730755069885,
      "grad_norm": 1.2989176511764526,
      "learning_rate": 4.2374833908245355e-06,
      "loss": 0.0534,
      "step": 274875
    },
    {
      "epoch": 4.0485412585970755,
      "grad_norm": 1.2426633834838867,
      "learning_rate": 4.235847018857551e-06,
      "loss": 0.055,
      "step": 274900
    },
    {
      "epoch": 4.0489094416871625,
      "grad_norm": 1.389223575592041,
      "learning_rate": 4.234210646890566e-06,
      "loss": 0.054,
      "step": 274925
    },
    {
      "epoch": 4.0492776247772495,
      "grad_norm": 1.1781851053237915,
      "learning_rate": 4.2325742749235814e-06,
      "loss": 0.053,
      "step": 274950
    },
    {
      "epoch": 4.0496458078673365,
      "grad_norm": 1.1376879215240479,
      "learning_rate": 4.230937902956598e-06,
      "loss": 0.0493,
      "step": 274975
    },
    {
      "epoch": 4.0500139909574235,
      "grad_norm": 1.385263442993164,
      "learning_rate": 4.229301530989612e-06,
      "loss": 0.0512,
      "step": 275000
    },
    {
      "epoch": 4.0503821740475106,
      "grad_norm": 1.1493520736694336,
      "learning_rate": 4.227665159022628e-06,
      "loss": 0.0588,
      "step": 275025
    },
    {
      "epoch": 4.050750357137598,
      "grad_norm": 0.9774383306503296,
      "learning_rate": 4.2260287870556435e-06,
      "loss": 0.0561,
      "step": 275050
    },
    {
      "epoch": 4.051118540227685,
      "grad_norm": 1.7937761545181274,
      "learning_rate": 4.224392415088659e-06,
      "loss": 0.0505,
      "step": 275075
    },
    {
      "epoch": 4.051486723317772,
      "grad_norm": 0.9946050643920898,
      "learning_rate": 4.222756043121675e-06,
      "loss": 0.0524,
      "step": 275100
    },
    {
      "epoch": 4.051854906407859,
      "grad_norm": 1.591372013092041,
      "learning_rate": 4.221119671154689e-06,
      "loss": 0.0568,
      "step": 275125
    },
    {
      "epoch": 4.052223089497946,
      "grad_norm": 1.4732050895690918,
      "learning_rate": 4.2194832991877056e-06,
      "loss": 0.0497,
      "step": 275150
    },
    {
      "epoch": 4.052591272588033,
      "grad_norm": 2.2683539390563965,
      "learning_rate": 4.217846927220721e-06,
      "loss": 0.0538,
      "step": 275175
    },
    {
      "epoch": 4.05295945567812,
      "grad_norm": 1.082350254058838,
      "learning_rate": 4.216210555253736e-06,
      "loss": 0.0492,
      "step": 275200
    },
    {
      "epoch": 4.053327638768207,
      "grad_norm": 1.438417673110962,
      "learning_rate": 4.2145741832867515e-06,
      "loss": 0.0474,
      "step": 275225
    },
    {
      "epoch": 4.053695821858294,
      "grad_norm": 1.1476970911026,
      "learning_rate": 4.212937811319768e-06,
      "loss": 0.0513,
      "step": 275250
    },
    {
      "epoch": 4.054064004948381,
      "grad_norm": 1.1711137294769287,
      "learning_rate": 4.211301439352782e-06,
      "loss": 0.0496,
      "step": 275275
    },
    {
      "epoch": 4.054432188038468,
      "grad_norm": 1.4307100772857666,
      "learning_rate": 4.209665067385798e-06,
      "loss": 0.0533,
      "step": 275300
    },
    {
      "epoch": 4.054800371128555,
      "grad_norm": 1.5378831624984741,
      "learning_rate": 4.2080286954188135e-06,
      "loss": 0.055,
      "step": 275325
    },
    {
      "epoch": 4.055168554218642,
      "grad_norm": 1.559235692024231,
      "learning_rate": 4.206392323451829e-06,
      "loss": 0.054,
      "step": 275350
    },
    {
      "epoch": 4.055536737308729,
      "grad_norm": 1.107239842414856,
      "learning_rate": 4.204755951484844e-06,
      "loss": 0.0499,
      "step": 275375
    },
    {
      "epoch": 4.055904920398816,
      "grad_norm": 1.0033743381500244,
      "learning_rate": 4.2031195795178594e-06,
      "loss": 0.0475,
      "step": 275400
    },
    {
      "epoch": 4.056273103488903,
      "grad_norm": 1.441225290298462,
      "learning_rate": 4.201483207550875e-06,
      "loss": 0.0493,
      "step": 275425
    },
    {
      "epoch": 4.05664128657899,
      "grad_norm": 1.8936614990234375,
      "learning_rate": 4.199846835583891e-06,
      "loss": 0.0439,
      "step": 275450
    },
    {
      "epoch": 4.057009469669077,
      "grad_norm": 1.6256004571914673,
      "learning_rate": 4.198210463616905e-06,
      "loss": 0.0531,
      "step": 275475
    },
    {
      "epoch": 4.057377652759164,
      "grad_norm": 1.1434792280197144,
      "learning_rate": 4.1965740916499215e-06,
      "loss": 0.0534,
      "step": 275500
    },
    {
      "epoch": 4.057745835849251,
      "grad_norm": 1.2260769605636597,
      "learning_rate": 4.194937719682937e-06,
      "loss": 0.0438,
      "step": 275525
    },
    {
      "epoch": 4.058114018939338,
      "grad_norm": 1.2241178750991821,
      "learning_rate": 4.193301347715952e-06,
      "loss": 0.051,
      "step": 275550
    },
    {
      "epoch": 4.058482202029425,
      "grad_norm": 1.7049682140350342,
      "learning_rate": 4.191664975748968e-06,
      "loss": 0.0481,
      "step": 275575
    },
    {
      "epoch": 4.058850385119512,
      "grad_norm": 1.180618166923523,
      "learning_rate": 4.190028603781984e-06,
      "loss": 0.0471,
      "step": 275600
    },
    {
      "epoch": 4.059218568209599,
      "grad_norm": 1.0967459678649902,
      "learning_rate": 4.188392231814999e-06,
      "loss": 0.0566,
      "step": 275625
    },
    {
      "epoch": 4.059586751299686,
      "grad_norm": 1.3460229635238647,
      "learning_rate": 4.186755859848014e-06,
      "loss": 0.0489,
      "step": 275650
    },
    {
      "epoch": 4.059954934389773,
      "grad_norm": 1.5702486038208008,
      "learning_rate": 4.185184942759709e-06,
      "loss": 0.0602,
      "step": 275675
    },
    {
      "epoch": 4.06032311747986,
      "grad_norm": 1.0367343425750732,
      "learning_rate": 4.183548570792724e-06,
      "loss": 0.0506,
      "step": 275700
    },
    {
      "epoch": 4.060691300569948,
      "grad_norm": 0.8412879109382629,
      "learning_rate": 4.18191219882574e-06,
      "loss": 0.0564,
      "step": 275725
    },
    {
      "epoch": 4.061059483660035,
      "grad_norm": 1.5939195156097412,
      "learning_rate": 4.180275826858755e-06,
      "loss": 0.0486,
      "step": 275750
    },
    {
      "epoch": 4.061427666750122,
      "grad_norm": 1.1993752717971802,
      "learning_rate": 4.178639454891771e-06,
      "loss": 0.0484,
      "step": 275775
    },
    {
      "epoch": 4.061795849840209,
      "grad_norm": 1.3193093538284302,
      "learning_rate": 4.177003082924786e-06,
      "loss": 0.0497,
      "step": 275800
    },
    {
      "epoch": 4.062164032930296,
      "grad_norm": 1.7645987272262573,
      "learning_rate": 4.175366710957802e-06,
      "loss": 0.0512,
      "step": 275825
    },
    {
      "epoch": 4.062532216020383,
      "grad_norm": 1.0492877960205078,
      "learning_rate": 4.173730338990817e-06,
      "loss": 0.0487,
      "step": 275850
    },
    {
      "epoch": 4.06290039911047,
      "grad_norm": 1.2446444034576416,
      "learning_rate": 4.172093967023832e-06,
      "loss": 0.053,
      "step": 275875
    },
    {
      "epoch": 4.063268582200557,
      "grad_norm": 0.8729466795921326,
      "learning_rate": 4.1704575950568475e-06,
      "loss": 0.0557,
      "step": 275900
    },
    {
      "epoch": 4.063636765290644,
      "grad_norm": 1.142698884010315,
      "learning_rate": 4.168821223089864e-06,
      "loss": 0.0529,
      "step": 275925
    },
    {
      "epoch": 4.064004948380731,
      "grad_norm": 1.6176023483276367,
      "learning_rate": 4.167184851122878e-06,
      "loss": 0.045,
      "step": 275950
    },
    {
      "epoch": 4.064373131470818,
      "grad_norm": 1.1004976034164429,
      "learning_rate": 4.165548479155894e-06,
      "loss": 0.0483,
      "step": 275975
    },
    {
      "epoch": 4.064741314560905,
      "grad_norm": 0.9486537575721741,
      "learning_rate": 4.1639121071889096e-06,
      "loss": 0.0537,
      "step": 276000
    },
    {
      "epoch": 4.065109497650992,
      "grad_norm": 1.4191198348999023,
      "learning_rate": 4.162275735221925e-06,
      "loss": 0.0428,
      "step": 276025
    },
    {
      "epoch": 4.065477680741079,
      "grad_norm": 1.368203043937683,
      "learning_rate": 4.160639363254941e-06,
      "loss": 0.0496,
      "step": 276050
    },
    {
      "epoch": 4.065845863831166,
      "grad_norm": 1.5861294269561768,
      "learning_rate": 4.159002991287956e-06,
      "loss": 0.0529,
      "step": 276075
    },
    {
      "epoch": 4.066214046921253,
      "grad_norm": 1.3480197191238403,
      "learning_rate": 4.157366619320972e-06,
      "loss": 0.058,
      "step": 276100
    },
    {
      "epoch": 4.06658223001134,
      "grad_norm": 1.3167120218276978,
      "learning_rate": 4.155730247353987e-06,
      "loss": 0.0487,
      "step": 276125
    },
    {
      "epoch": 4.066950413101427,
      "grad_norm": 1.538070797920227,
      "learning_rate": 4.154093875387002e-06,
      "loss": 0.0564,
      "step": 276150
    },
    {
      "epoch": 4.067318596191514,
      "grad_norm": 1.1279892921447754,
      "learning_rate": 4.1524575034200175e-06,
      "loss": 0.0475,
      "step": 276175
    },
    {
      "epoch": 4.067686779281601,
      "grad_norm": 1.28226637840271,
      "learning_rate": 4.150821131453034e-06,
      "loss": 0.0518,
      "step": 276200
    },
    {
      "epoch": 4.068054962371688,
      "grad_norm": 1.1002174615859985,
      "learning_rate": 4.149184759486048e-06,
      "loss": 0.0457,
      "step": 276225
    },
    {
      "epoch": 4.068423145461775,
      "grad_norm": 1.401401162147522,
      "learning_rate": 4.147548387519064e-06,
      "loss": 0.0533,
      "step": 276250
    },
    {
      "epoch": 4.068791328551862,
      "grad_norm": 1.037559151649475,
      "learning_rate": 4.14591201555208e-06,
      "loss": 0.0458,
      "step": 276275
    },
    {
      "epoch": 4.069159511641949,
      "grad_norm": 1.0772476196289062,
      "learning_rate": 4.144275643585095e-06,
      "loss": 0.0499,
      "step": 276300
    },
    {
      "epoch": 4.069527694732036,
      "grad_norm": 0.9725455641746521,
      "learning_rate": 4.14263927161811e-06,
      "loss": 0.0546,
      "step": 276325
    },
    {
      "epoch": 4.069895877822123,
      "grad_norm": 1.4394406080245972,
      "learning_rate": 4.1410028996511255e-06,
      "loss": 0.0556,
      "step": 276350
    },
    {
      "epoch": 4.07026406091221,
      "grad_norm": 1.7116791009902954,
      "learning_rate": 4.139366527684141e-06,
      "loss": 0.0476,
      "step": 276375
    },
    {
      "epoch": 4.070632244002297,
      "grad_norm": 1.3169617652893066,
      "learning_rate": 4.137730155717157e-06,
      "loss": 0.05,
      "step": 276400
    },
    {
      "epoch": 4.071000427092384,
      "grad_norm": 1.8277477025985718,
      "learning_rate": 4.136093783750172e-06,
      "loss": 0.0529,
      "step": 276425
    },
    {
      "epoch": 4.071368610182471,
      "grad_norm": 1.1039022207260132,
      "learning_rate": 4.134457411783188e-06,
      "loss": 0.053,
      "step": 276450
    },
    {
      "epoch": 4.071736793272558,
      "grad_norm": 0.8616538047790527,
      "learning_rate": 4.132821039816203e-06,
      "loss": 0.0515,
      "step": 276475
    },
    {
      "epoch": 4.072104976362645,
      "grad_norm": 0.7401161789894104,
      "learning_rate": 4.131184667849218e-06,
      "loss": 0.0474,
      "step": 276500
    },
    {
      "epoch": 4.072473159452732,
      "grad_norm": 0.8509481549263,
      "learning_rate": 4.129548295882234e-06,
      "loss": 0.0539,
      "step": 276525
    },
    {
      "epoch": 4.072841342542819,
      "grad_norm": 0.9534338712692261,
      "learning_rate": 4.12791192391525e-06,
      "loss": 0.053,
      "step": 276550
    },
    {
      "epoch": 4.073209525632906,
      "grad_norm": 1.070122480392456,
      "learning_rate": 4.126275551948265e-06,
      "loss": 0.0465,
      "step": 276575
    },
    {
      "epoch": 4.073577708722993,
      "grad_norm": 1.021049976348877,
      "learning_rate": 4.12463917998128e-06,
      "loss": 0.0511,
      "step": 276600
    },
    {
      "epoch": 4.07394589181308,
      "grad_norm": 1.3801721334457397,
      "learning_rate": 4.1230028080142956e-06,
      "loss": 0.053,
      "step": 276625
    },
    {
      "epoch": 4.074314074903167,
      "grad_norm": 1.4900864362716675,
      "learning_rate": 4.121366436047311e-06,
      "loss": 0.0502,
      "step": 276650
    },
    {
      "epoch": 4.0746822579932545,
      "grad_norm": 1.3341633081436157,
      "learning_rate": 4.119730064080327e-06,
      "loss": 0.0515,
      "step": 276675
    },
    {
      "epoch": 4.075050441083342,
      "grad_norm": 1.6892269849777222,
      "learning_rate": 4.1180936921133415e-06,
      "loss": 0.0586,
      "step": 276700
    },
    {
      "epoch": 4.075418624173429,
      "grad_norm": 1.7524826526641846,
      "learning_rate": 4.116457320146358e-06,
      "loss": 0.0524,
      "step": 276725
    },
    {
      "epoch": 4.075786807263516,
      "grad_norm": 1.2068977355957031,
      "learning_rate": 4.114820948179373e-06,
      "loss": 0.0529,
      "step": 276750
    },
    {
      "epoch": 4.076154990353603,
      "grad_norm": 1.2106729745864868,
      "learning_rate": 4.113184576212388e-06,
      "loss": 0.0515,
      "step": 276775
    },
    {
      "epoch": 4.07652317344369,
      "grad_norm": 1.0615928173065186,
      "learning_rate": 4.1115482042454035e-06,
      "loss": 0.0488,
      "step": 276800
    },
    {
      "epoch": 4.076891356533777,
      "grad_norm": 0.9959233403205872,
      "learning_rate": 4.10991183227842e-06,
      "loss": 0.0593,
      "step": 276825
    },
    {
      "epoch": 4.0772595396238644,
      "grad_norm": 1.3831136226654053,
      "learning_rate": 4.108275460311434e-06,
      "loss": 0.0532,
      "step": 276850
    },
    {
      "epoch": 4.0776277227139515,
      "grad_norm": 1.4975295066833496,
      "learning_rate": 4.10663908834445e-06,
      "loss": 0.0491,
      "step": 276875
    },
    {
      "epoch": 4.0779959058040385,
      "grad_norm": 0.7060441374778748,
      "learning_rate": 4.105002716377466e-06,
      "loss": 0.0475,
      "step": 276900
    },
    {
      "epoch": 4.0783640888941255,
      "grad_norm": 1.2254325151443481,
      "learning_rate": 4.103366344410481e-06,
      "loss": 0.0554,
      "step": 276925
    },
    {
      "epoch": 4.0787322719842125,
      "grad_norm": 1.3210432529449463,
      "learning_rate": 4.101729972443497e-06,
      "loss": 0.0507,
      "step": 276950
    },
    {
      "epoch": 4.0791004550742995,
      "grad_norm": 1.3877944946289062,
      "learning_rate": 4.1000936004765115e-06,
      "loss": 0.0502,
      "step": 276975
    },
    {
      "epoch": 4.0794686381643865,
      "grad_norm": 1.5770561695098877,
      "learning_rate": 4.098457228509528e-06,
      "loss": 0.0502,
      "step": 277000
    },
    {
      "epoch": 4.0798368212544736,
      "grad_norm": 1.1314924955368042,
      "learning_rate": 4.096820856542543e-06,
      "loss": 0.0499,
      "step": 277025
    },
    {
      "epoch": 4.080205004344561,
      "grad_norm": 1.0255426168441772,
      "learning_rate": 4.095184484575558e-06,
      "loss": 0.0513,
      "step": 277050
    },
    {
      "epoch": 4.080573187434648,
      "grad_norm": 1.539833664894104,
      "learning_rate": 4.093548112608574e-06,
      "loss": 0.0566,
      "step": 277075
    },
    {
      "epoch": 4.080941370524735,
      "grad_norm": 0.6666017770767212,
      "learning_rate": 4.091911740641589e-06,
      "loss": 0.0452,
      "step": 277100
    },
    {
      "epoch": 4.081309553614822,
      "grad_norm": 0.8971357941627502,
      "learning_rate": 4.090275368674604e-06,
      "loss": 0.0467,
      "step": 277125
    },
    {
      "epoch": 4.081677736704909,
      "grad_norm": 1.290385127067566,
      "learning_rate": 4.08863899670762e-06,
      "loss": 0.0473,
      "step": 277150
    },
    {
      "epoch": 4.082045919794996,
      "grad_norm": 1.6305418014526367,
      "learning_rate": 4.087002624740636e-06,
      "loss": 0.0545,
      "step": 277175
    },
    {
      "epoch": 4.082414102885083,
      "grad_norm": 1.1996963024139404,
      "learning_rate": 4.085366252773651e-06,
      "loss": 0.0521,
      "step": 277200
    },
    {
      "epoch": 4.08278228597517,
      "grad_norm": 0.871673583984375,
      "learning_rate": 4.083729880806666e-06,
      "loss": 0.0485,
      "step": 277225
    },
    {
      "epoch": 4.083150469065257,
      "grad_norm": 0.8779425024986267,
      "learning_rate": 4.0820935088396816e-06,
      "loss": 0.0519,
      "step": 277250
    },
    {
      "epoch": 4.083518652155344,
      "grad_norm": 1.308681607246399,
      "learning_rate": 4.080457136872697e-06,
      "loss": 0.048,
      "step": 277275
    },
    {
      "epoch": 4.083886835245431,
      "grad_norm": 1.2734334468841553,
      "learning_rate": 4.078820764905713e-06,
      "loss": 0.0516,
      "step": 277300
    },
    {
      "epoch": 4.084255018335518,
      "grad_norm": 1.5363825559616089,
      "learning_rate": 4.0771843929387275e-06,
      "loss": 0.0474,
      "step": 277325
    },
    {
      "epoch": 4.084623201425605,
      "grad_norm": 1.5124701261520386,
      "learning_rate": 4.075548020971744e-06,
      "loss": 0.0508,
      "step": 277350
    },
    {
      "epoch": 4.084991384515692,
      "grad_norm": 1.2399643659591675,
      "learning_rate": 4.073911649004759e-06,
      "loss": 0.0547,
      "step": 277375
    },
    {
      "epoch": 4.085359567605779,
      "grad_norm": 1.130418062210083,
      "learning_rate": 4.072275277037774e-06,
      "loss": 0.0534,
      "step": 277400
    },
    {
      "epoch": 4.085727750695866,
      "grad_norm": 1.3833637237548828,
      "learning_rate": 4.0706389050707895e-06,
      "loss": 0.0524,
      "step": 277425
    },
    {
      "epoch": 4.086095933785953,
      "grad_norm": 0.9419562220573425,
      "learning_rate": 4.069002533103805e-06,
      "loss": 0.0584,
      "step": 277450
    },
    {
      "epoch": 4.08646411687604,
      "grad_norm": 1.365135669708252,
      "learning_rate": 4.067366161136821e-06,
      "loss": 0.0566,
      "step": 277475
    },
    {
      "epoch": 4.086832299966127,
      "grad_norm": 0.8531058430671692,
      "learning_rate": 4.065729789169836e-06,
      "loss": 0.051,
      "step": 277500
    },
    {
      "epoch": 4.087200483056214,
      "grad_norm": 1.4153810739517212,
      "learning_rate": 4.064093417202852e-06,
      "loss": 0.0543,
      "step": 277525
    },
    {
      "epoch": 4.087568666146301,
      "grad_norm": 0.8011882305145264,
      "learning_rate": 4.062457045235867e-06,
      "loss": 0.0449,
      "step": 277550
    },
    {
      "epoch": 4.087936849236388,
      "grad_norm": 1.3054267168045044,
      "learning_rate": 4.060820673268882e-06,
      "loss": 0.0596,
      "step": 277575
    },
    {
      "epoch": 4.088305032326475,
      "grad_norm": 1.223548173904419,
      "learning_rate": 4.0591843013018975e-06,
      "loss": 0.0469,
      "step": 277600
    },
    {
      "epoch": 4.088673215416562,
      "grad_norm": 1.245463252067566,
      "learning_rate": 4.057547929334914e-06,
      "loss": 0.0501,
      "step": 277625
    },
    {
      "epoch": 4.08904139850665,
      "grad_norm": 1.0151854753494263,
      "learning_rate": 4.055911557367929e-06,
      "loss": 0.0552,
      "step": 277650
    },
    {
      "epoch": 4.089409581596737,
      "grad_norm": 1.369176983833313,
      "learning_rate": 4.054275185400944e-06,
      "loss": 0.0495,
      "step": 277675
    },
    {
      "epoch": 4.089777764686824,
      "grad_norm": 0.9294801950454712,
      "learning_rate": 4.05263881343396e-06,
      "loss": 0.0513,
      "step": 277700
    },
    {
      "epoch": 4.090145947776911,
      "grad_norm": 1.0617151260375977,
      "learning_rate": 4.051002441466975e-06,
      "loss": 0.0512,
      "step": 277725
    },
    {
      "epoch": 4.090514130866998,
      "grad_norm": 1.2690285444259644,
      "learning_rate": 4.04936606949999e-06,
      "loss": 0.0534,
      "step": 277750
    },
    {
      "epoch": 4.090882313957085,
      "grad_norm": 1.0979222059249878,
      "learning_rate": 4.047729697533006e-06,
      "loss": 0.0505,
      "step": 277775
    },
    {
      "epoch": 4.091250497047172,
      "grad_norm": 1.1480523347854614,
      "learning_rate": 4.046093325566021e-06,
      "loss": 0.0534,
      "step": 277800
    },
    {
      "epoch": 4.091618680137259,
      "grad_norm": 1.834741473197937,
      "learning_rate": 4.044456953599037e-06,
      "loss": 0.0596,
      "step": 277825
    },
    {
      "epoch": 4.091986863227346,
      "grad_norm": 1.4808859825134277,
      "learning_rate": 4.042820581632052e-06,
      "loss": 0.0496,
      "step": 277850
    },
    {
      "epoch": 4.092355046317433,
      "grad_norm": 1.0610440969467163,
      "learning_rate": 4.0411842096650676e-06,
      "loss": 0.0529,
      "step": 277875
    },
    {
      "epoch": 4.09272322940752,
      "grad_norm": 1.6741849184036255,
      "learning_rate": 4.039547837698083e-06,
      "loss": 0.0515,
      "step": 277900
    },
    {
      "epoch": 4.093091412497607,
      "grad_norm": 1.0648272037506104,
      "learning_rate": 4.037911465731099e-06,
      "loss": 0.0534,
      "step": 277925
    },
    {
      "epoch": 4.093459595587694,
      "grad_norm": 0.7898022532463074,
      "learning_rate": 4.036275093764114e-06,
      "loss": 0.0449,
      "step": 277950
    },
    {
      "epoch": 4.093827778677781,
      "grad_norm": 1.3693745136260986,
      "learning_rate": 4.03463872179713e-06,
      "loss": 0.049,
      "step": 277975
    },
    {
      "epoch": 4.094195961767868,
      "grad_norm": 0.8926895260810852,
      "learning_rate": 4.033002349830145e-06,
      "loss": 0.0468,
      "step": 278000
    },
    {
      "epoch": 4.094564144857955,
      "grad_norm": 1.4297376871109009,
      "learning_rate": 4.03136597786316e-06,
      "loss": 0.0568,
      "step": 278025
    },
    {
      "epoch": 4.094932327948042,
      "grad_norm": 1.0375431776046753,
      "learning_rate": 4.029729605896176e-06,
      "loss": 0.0521,
      "step": 278050
    },
    {
      "epoch": 4.095300511038129,
      "grad_norm": 1.9625438451766968,
      "learning_rate": 4.028093233929191e-06,
      "loss": 0.0533,
      "step": 278075
    },
    {
      "epoch": 4.095668694128216,
      "grad_norm": 1.1361236572265625,
      "learning_rate": 4.026456861962207e-06,
      "loss": 0.048,
      "step": 278100
    },
    {
      "epoch": 4.096036877218303,
      "grad_norm": 1.4360212087631226,
      "learning_rate": 4.024820489995222e-06,
      "loss": 0.056,
      "step": 278125
    },
    {
      "epoch": 4.09640506030839,
      "grad_norm": 1.3297650814056396,
      "learning_rate": 4.023184118028238e-06,
      "loss": 0.0484,
      "step": 278150
    },
    {
      "epoch": 4.096773243398477,
      "grad_norm": 0.8162236213684082,
      "learning_rate": 4.021547746061253e-06,
      "loss": 0.0562,
      "step": 278175
    },
    {
      "epoch": 4.097141426488564,
      "grad_norm": 1.244158387184143,
      "learning_rate": 4.019911374094268e-06,
      "loss": 0.0492,
      "step": 278200
    },
    {
      "epoch": 4.097509609578651,
      "grad_norm": 1.918629765510559,
      "learning_rate": 4.0182750021272835e-06,
      "loss": 0.0518,
      "step": 278225
    },
    {
      "epoch": 4.097877792668738,
      "grad_norm": 1.3182944059371948,
      "learning_rate": 4.0166386301603e-06,
      "loss": 0.0562,
      "step": 278250
    },
    {
      "epoch": 4.098245975758825,
      "grad_norm": 1.0243265628814697,
      "learning_rate": 4.015002258193315e-06,
      "loss": 0.0483,
      "step": 278275
    },
    {
      "epoch": 4.098614158848912,
      "grad_norm": 1.120998740196228,
      "learning_rate": 4.01336588622633e-06,
      "loss": 0.0537,
      "step": 278300
    },
    {
      "epoch": 4.098982341938999,
      "grad_norm": 0.9357917308807373,
      "learning_rate": 4.011729514259346e-06,
      "loss": 0.0553,
      "step": 278325
    },
    {
      "epoch": 4.099350525029086,
      "grad_norm": 1.3658519983291626,
      "learning_rate": 4.010093142292361e-06,
      "loss": 0.053,
      "step": 278350
    },
    {
      "epoch": 4.099718708119173,
      "grad_norm": 1.928463339805603,
      "learning_rate": 4.008456770325376e-06,
      "loss": 0.0541,
      "step": 278375
    },
    {
      "epoch": 4.10008689120926,
      "grad_norm": 1.170521855354309,
      "learning_rate": 4.006820398358392e-06,
      "loss": 0.0487,
      "step": 278400
    },
    {
      "epoch": 4.100455074299347,
      "grad_norm": 1.3885964155197144,
      "learning_rate": 4.005184026391408e-06,
      "loss": 0.0551,
      "step": 278425
    },
    {
      "epoch": 4.100823257389434,
      "grad_norm": 1.4824920892715454,
      "learning_rate": 4.003547654424423e-06,
      "loss": 0.053,
      "step": 278450
    },
    {
      "epoch": 4.101191440479521,
      "grad_norm": 1.451856017112732,
      "learning_rate": 4.001911282457438e-06,
      "loss": 0.0503,
      "step": 278475
    },
    {
      "epoch": 4.101559623569608,
      "grad_norm": 1.8395278453826904,
      "learning_rate": 4.0002749104904536e-06,
      "loss": 0.0504,
      "step": 278500
    },
    {
      "epoch": 4.101927806659695,
      "grad_norm": 1.5039464235305786,
      "learning_rate": 3.99863853852347e-06,
      "loss": 0.0535,
      "step": 278525
    },
    {
      "epoch": 4.102295989749782,
      "grad_norm": 0.820640504360199,
      "learning_rate": 3.997002166556484e-06,
      "loss": 0.0454,
      "step": 278550
    },
    {
      "epoch": 4.102664172839869,
      "grad_norm": 1.6552753448486328,
      "learning_rate": 3.9953657945895e-06,
      "loss": 0.0519,
      "step": 278575
    },
    {
      "epoch": 4.103032355929956,
      "grad_norm": 1.631272792816162,
      "learning_rate": 3.993729422622516e-06,
      "loss": 0.0524,
      "step": 278600
    },
    {
      "epoch": 4.103400539020044,
      "grad_norm": 1.3876011371612549,
      "learning_rate": 3.992093050655531e-06,
      "loss": 0.0533,
      "step": 278625
    },
    {
      "epoch": 4.103768722110131,
      "grad_norm": 1.0428255796432495,
      "learning_rate": 3.990456678688546e-06,
      "loss": 0.0561,
      "step": 278650
    },
    {
      "epoch": 4.104136905200218,
      "grad_norm": 1.0484966039657593,
      "learning_rate": 3.9888203067215615e-06,
      "loss": 0.0495,
      "step": 278675
    },
    {
      "epoch": 4.104505088290305,
      "grad_norm": 1.3320986032485962,
      "learning_rate": 3.987183934754577e-06,
      "loss": 0.047,
      "step": 278700
    },
    {
      "epoch": 4.104873271380392,
      "grad_norm": 0.9794262647628784,
      "learning_rate": 3.985547562787593e-06,
      "loss": 0.0517,
      "step": 278725
    },
    {
      "epoch": 4.105241454470479,
      "grad_norm": 1.261979103088379,
      "learning_rate": 3.983911190820608e-06,
      "loss": 0.0511,
      "step": 278750
    },
    {
      "epoch": 4.105609637560566,
      "grad_norm": 1.1944228410720825,
      "learning_rate": 3.982274818853624e-06,
      "loss": 0.051,
      "step": 278775
    },
    {
      "epoch": 4.105977820650653,
      "grad_norm": 1.2041842937469482,
      "learning_rate": 3.980638446886639e-06,
      "loss": 0.0544,
      "step": 278800
    },
    {
      "epoch": 4.10634600374074,
      "grad_norm": 1.5027638673782349,
      "learning_rate": 3.979002074919654e-06,
      "loss": 0.053,
      "step": 278825
    },
    {
      "epoch": 4.1067141868308275,
      "grad_norm": 1.4179985523223877,
      "learning_rate": 3.9773657029526695e-06,
      "loss": 0.0505,
      "step": 278850
    },
    {
      "epoch": 4.1070823699209145,
      "grad_norm": 1.3135807514190674,
      "learning_rate": 3.975729330985686e-06,
      "loss": 0.0547,
      "step": 278875
    },
    {
      "epoch": 4.1074505530110015,
      "grad_norm": 1.3814418315887451,
      "learning_rate": 3.9740929590187e-06,
      "loss": 0.0517,
      "step": 278900
    },
    {
      "epoch": 4.1078187361010885,
      "grad_norm": 1.207383394241333,
      "learning_rate": 3.972456587051716e-06,
      "loss": 0.0519,
      "step": 278925
    },
    {
      "epoch": 4.1081869191911755,
      "grad_norm": 1.3203582763671875,
      "learning_rate": 3.970820215084732e-06,
      "loss": 0.0626,
      "step": 278950
    },
    {
      "epoch": 4.1085551022812625,
      "grad_norm": 1.9190857410430908,
      "learning_rate": 3.969183843117747e-06,
      "loss": 0.0546,
      "step": 278975
    },
    {
      "epoch": 4.1089232853713495,
      "grad_norm": 1.5795048475265503,
      "learning_rate": 3.967547471150763e-06,
      "loss": 0.0543,
      "step": 279000
    },
    {
      "epoch": 4.109291468461437,
      "grad_norm": 1.1442686319351196,
      "learning_rate": 3.965911099183778e-06,
      "loss": 0.0572,
      "step": 279025
    },
    {
      "epoch": 4.109659651551524,
      "grad_norm": 1.3957563638687134,
      "learning_rate": 3.964274727216794e-06,
      "loss": 0.0546,
      "step": 279050
    },
    {
      "epoch": 4.110027834641611,
      "grad_norm": 1.3441755771636963,
      "learning_rate": 3.962638355249809e-06,
      "loss": 0.0479,
      "step": 279075
    },
    {
      "epoch": 4.110396017731698,
      "grad_norm": 1.243120551109314,
      "learning_rate": 3.961001983282824e-06,
      "loss": 0.0532,
      "step": 279100
    },
    {
      "epoch": 4.110764200821785,
      "grad_norm": 1.1591179370880127,
      "learning_rate": 3.9593656113158396e-06,
      "loss": 0.051,
      "step": 279125
    },
    {
      "epoch": 4.111132383911872,
      "grad_norm": 1.3041765689849854,
      "learning_rate": 3.957729239348856e-06,
      "loss": 0.0517,
      "step": 279150
    },
    {
      "epoch": 4.111500567001959,
      "grad_norm": 1.2058486938476562,
      "learning_rate": 3.95609286738187e-06,
      "loss": 0.0461,
      "step": 279175
    },
    {
      "epoch": 4.111868750092046,
      "grad_norm": 1.1057459115982056,
      "learning_rate": 3.954456495414886e-06,
      "loss": 0.0529,
      "step": 279200
    },
    {
      "epoch": 4.112236933182133,
      "grad_norm": 1.5318320989608765,
      "learning_rate": 3.952820123447902e-06,
      "loss": 0.0525,
      "step": 279225
    },
    {
      "epoch": 4.11260511627222,
      "grad_norm": 1.34402334690094,
      "learning_rate": 3.951183751480917e-06,
      "loss": 0.0534,
      "step": 279250
    },
    {
      "epoch": 4.112973299362307,
      "grad_norm": 0.8894164562225342,
      "learning_rate": 3.949547379513932e-06,
      "loss": 0.052,
      "step": 279275
    },
    {
      "epoch": 4.113341482452394,
      "grad_norm": 1.1744776964187622,
      "learning_rate": 3.9479110075469475e-06,
      "loss": 0.0469,
      "step": 279300
    },
    {
      "epoch": 4.113709665542481,
      "grad_norm": 0.9439015984535217,
      "learning_rate": 3.946274635579963e-06,
      "loss": 0.0522,
      "step": 279325
    },
    {
      "epoch": 4.114077848632568,
      "grad_norm": 1.3068386316299438,
      "learning_rate": 3.944638263612979e-06,
      "loss": 0.0508,
      "step": 279350
    },
    {
      "epoch": 4.114446031722655,
      "grad_norm": 0.9515126347541809,
      "learning_rate": 3.943001891645994e-06,
      "loss": 0.0563,
      "step": 279375
    },
    {
      "epoch": 4.114814214812742,
      "grad_norm": 1.7591304779052734,
      "learning_rate": 3.94136551967901e-06,
      "loss": 0.0522,
      "step": 279400
    },
    {
      "epoch": 4.115182397902829,
      "grad_norm": 1.4973087310791016,
      "learning_rate": 3.939729147712025e-06,
      "loss": 0.0506,
      "step": 279425
    },
    {
      "epoch": 4.115550580992916,
      "grad_norm": 1.8153070211410522,
      "learning_rate": 3.93809277574504e-06,
      "loss": 0.0517,
      "step": 279450
    },
    {
      "epoch": 4.115918764083003,
      "grad_norm": 1.1951584815979004,
      "learning_rate": 3.936456403778056e-06,
      "loss": 0.0539,
      "step": 279475
    },
    {
      "epoch": 4.11628694717309,
      "grad_norm": 1.7845375537872314,
      "learning_rate": 3.934820031811072e-06,
      "loss": 0.0538,
      "step": 279500
    },
    {
      "epoch": 4.116655130263177,
      "grad_norm": 0.9844855070114136,
      "learning_rate": 3.933183659844087e-06,
      "loss": 0.0434,
      "step": 279525
    },
    {
      "epoch": 4.117023313353264,
      "grad_norm": 1.21844482421875,
      "learning_rate": 3.931547287877102e-06,
      "loss": 0.0499,
      "step": 279550
    },
    {
      "epoch": 4.117391496443352,
      "grad_norm": 1.4345084428787231,
      "learning_rate": 3.929910915910118e-06,
      "loss": 0.0498,
      "step": 279575
    },
    {
      "epoch": 4.117759679533439,
      "grad_norm": 1.4880101680755615,
      "learning_rate": 3.928274543943133e-06,
      "loss": 0.0472,
      "step": 279600
    },
    {
      "epoch": 4.118127862623526,
      "grad_norm": 1.296320915222168,
      "learning_rate": 3.926638171976149e-06,
      "loss": 0.0511,
      "step": 279625
    },
    {
      "epoch": 4.118496045713613,
      "grad_norm": 1.3764183521270752,
      "learning_rate": 3.9250018000091635e-06,
      "loss": 0.0473,
      "step": 279650
    },
    {
      "epoch": 4.1188642288037,
      "grad_norm": 1.530476450920105,
      "learning_rate": 3.923430882920859e-06,
      "loss": 0.0589,
      "step": 279675
    },
    {
      "epoch": 4.119232411893787,
      "grad_norm": 1.2778176069259644,
      "learning_rate": 3.921794510953874e-06,
      "loss": 0.05,
      "step": 279700
    },
    {
      "epoch": 4.119600594983874,
      "grad_norm": 1.380972146987915,
      "learning_rate": 3.92015813898689e-06,
      "loss": 0.0483,
      "step": 279725
    },
    {
      "epoch": 4.119968778073961,
      "grad_norm": 1.468732476234436,
      "learning_rate": 3.918521767019905e-06,
      "loss": 0.0538,
      "step": 279750
    },
    {
      "epoch": 4.120336961164048,
      "grad_norm": 1.6027483940124512,
      "learning_rate": 3.91688539505292e-06,
      "loss": 0.0546,
      "step": 279775
    },
    {
      "epoch": 4.120705144254135,
      "grad_norm": 0.9671803712844849,
      "learning_rate": 3.915249023085936e-06,
      "loss": 0.0496,
      "step": 279800
    },
    {
      "epoch": 4.121073327344222,
      "grad_norm": 1.381050705909729,
      "learning_rate": 3.913612651118952e-06,
      "loss": 0.0454,
      "step": 279825
    },
    {
      "epoch": 4.121441510434309,
      "grad_norm": 0.8141024708747864,
      "learning_rate": 3.911976279151967e-06,
      "loss": 0.0513,
      "step": 279850
    },
    {
      "epoch": 4.121809693524396,
      "grad_norm": 1.382088303565979,
      "learning_rate": 3.910339907184982e-06,
      "loss": 0.0521,
      "step": 279875
    },
    {
      "epoch": 4.122177876614483,
      "grad_norm": 1.3073772192001343,
      "learning_rate": 3.908703535217998e-06,
      "loss": 0.0542,
      "step": 279900
    },
    {
      "epoch": 4.12254605970457,
      "grad_norm": 1.2595528364181519,
      "learning_rate": 3.907067163251013e-06,
      "loss": 0.0564,
      "step": 279925
    },
    {
      "epoch": 4.122914242794657,
      "grad_norm": 0.7806668877601624,
      "learning_rate": 3.905430791284029e-06,
      "loss": 0.0516,
      "step": 279950
    },
    {
      "epoch": 4.123282425884744,
      "grad_norm": 1.7917717695236206,
      "learning_rate": 3.903794419317044e-06,
      "loss": 0.0454,
      "step": 279975
    },
    {
      "epoch": 4.123650608974831,
      "grad_norm": 0.6870917081832886,
      "learning_rate": 3.90215804735006e-06,
      "loss": 0.058,
      "step": 280000
    },
    {
      "epoch": 4.124018792064918,
      "grad_norm": 1.9068914651870728,
      "learning_rate": 3.900521675383075e-06,
      "loss": 0.0508,
      "step": 280025
    },
    {
      "epoch": 4.124386975155005,
      "grad_norm": 1.4279999732971191,
      "learning_rate": 3.89888530341609e-06,
      "loss": 0.051,
      "step": 280050
    },
    {
      "epoch": 4.124755158245092,
      "grad_norm": 1.2259025573730469,
      "learning_rate": 3.897248931449106e-06,
      "loss": 0.0484,
      "step": 280075
    },
    {
      "epoch": 4.125123341335179,
      "grad_norm": 1.1658406257629395,
      "learning_rate": 3.895612559482122e-06,
      "loss": 0.0487,
      "step": 280100
    },
    {
      "epoch": 4.125491524425266,
      "grad_norm": 1.4479068517684937,
      "learning_rate": 3.893976187515136e-06,
      "loss": 0.0539,
      "step": 280125
    },
    {
      "epoch": 4.125859707515353,
      "grad_norm": 1.5897122621536255,
      "learning_rate": 3.892339815548152e-06,
      "loss": 0.0564,
      "step": 280150
    },
    {
      "epoch": 4.12622789060544,
      "grad_norm": 1.1869397163391113,
      "learning_rate": 3.890703443581168e-06,
      "loss": 0.052,
      "step": 280175
    },
    {
      "epoch": 4.126596073695527,
      "grad_norm": 1.2326768636703491,
      "learning_rate": 3.889067071614183e-06,
      "loss": 0.0488,
      "step": 280200
    },
    {
      "epoch": 4.126964256785614,
      "grad_norm": 1.4007612466812134,
      "learning_rate": 3.887430699647198e-06,
      "loss": 0.0512,
      "step": 280225
    },
    {
      "epoch": 4.127332439875701,
      "grad_norm": 1.311048150062561,
      "learning_rate": 3.885794327680214e-06,
      "loss": 0.0575,
      "step": 280250
    },
    {
      "epoch": 4.127700622965788,
      "grad_norm": 1.2580080032348633,
      "learning_rate": 3.884157955713229e-06,
      "loss": 0.0552,
      "step": 280275
    },
    {
      "epoch": 4.128068806055875,
      "grad_norm": 1.7356899976730347,
      "learning_rate": 3.882521583746245e-06,
      "loss": 0.0478,
      "step": 280300
    },
    {
      "epoch": 4.128436989145962,
      "grad_norm": 0.8031937479972839,
      "learning_rate": 3.88088521177926e-06,
      "loss": 0.0539,
      "step": 280325
    },
    {
      "epoch": 4.128805172236049,
      "grad_norm": 1.1134426593780518,
      "learning_rate": 3.879248839812276e-06,
      "loss": 0.0445,
      "step": 280350
    },
    {
      "epoch": 4.129173355326136,
      "grad_norm": 1.4369693994522095,
      "learning_rate": 3.877612467845292e-06,
      "loss": 0.0482,
      "step": 280375
    },
    {
      "epoch": 4.129541538416223,
      "grad_norm": 1.549274206161499,
      "learning_rate": 3.875976095878306e-06,
      "loss": 0.047,
      "step": 280400
    },
    {
      "epoch": 4.12990972150631,
      "grad_norm": 1.5910199880599976,
      "learning_rate": 3.8743397239113224e-06,
      "loss": 0.0522,
      "step": 280425
    },
    {
      "epoch": 4.130277904596397,
      "grad_norm": 1.6414375305175781,
      "learning_rate": 3.872703351944338e-06,
      "loss": 0.0538,
      "step": 280450
    },
    {
      "epoch": 4.130646087686484,
      "grad_norm": 1.5115987062454224,
      "learning_rate": 3.871066979977353e-06,
      "loss": 0.0464,
      "step": 280475
    },
    {
      "epoch": 4.131014270776571,
      "grad_norm": 1.0115933418273926,
      "learning_rate": 3.869430608010368e-06,
      "loss": 0.0526,
      "step": 280500
    },
    {
      "epoch": 4.131382453866658,
      "grad_norm": 1.1249902248382568,
      "learning_rate": 3.867794236043384e-06,
      "loss": 0.0546,
      "step": 280525
    },
    {
      "epoch": 4.131750636956746,
      "grad_norm": 0.9938558340072632,
      "learning_rate": 3.866157864076399e-06,
      "loss": 0.0603,
      "step": 280550
    },
    {
      "epoch": 4.132118820046833,
      "grad_norm": 1.1485445499420166,
      "learning_rate": 3.864521492109415e-06,
      "loss": 0.0466,
      "step": 280575
    },
    {
      "epoch": 4.13248700313692,
      "grad_norm": 1.3947577476501465,
      "learning_rate": 3.86288512014243e-06,
      "loss": 0.0562,
      "step": 280600
    },
    {
      "epoch": 4.132855186227007,
      "grad_norm": 1.539109468460083,
      "learning_rate": 3.861248748175446e-06,
      "loss": 0.0477,
      "step": 280625
    },
    {
      "epoch": 4.133223369317094,
      "grad_norm": 1.8791552782058716,
      "learning_rate": 3.859612376208461e-06,
      "loss": 0.0525,
      "step": 280650
    },
    {
      "epoch": 4.133591552407181,
      "grad_norm": 1.3559869527816772,
      "learning_rate": 3.857976004241476e-06,
      "loss": 0.0536,
      "step": 280675
    },
    {
      "epoch": 4.133959735497268,
      "grad_norm": 1.8116252422332764,
      "learning_rate": 3.856339632274492e-06,
      "loss": 0.0521,
      "step": 280700
    },
    {
      "epoch": 4.134327918587355,
      "grad_norm": 1.504961371421814,
      "learning_rate": 3.854703260307508e-06,
      "loss": 0.0488,
      "step": 280725
    },
    {
      "epoch": 4.134696101677442,
      "grad_norm": 1.4178041219711304,
      "learning_rate": 3.853066888340522e-06,
      "loss": 0.0557,
      "step": 280750
    },
    {
      "epoch": 4.135064284767529,
      "grad_norm": 1.3786544799804688,
      "learning_rate": 3.851430516373538e-06,
      "loss": 0.0477,
      "step": 280775
    },
    {
      "epoch": 4.135432467857616,
      "grad_norm": 0.9783445000648499,
      "learning_rate": 3.849794144406554e-06,
      "loss": 0.0527,
      "step": 280800
    },
    {
      "epoch": 4.135800650947703,
      "grad_norm": 1.117689609527588,
      "learning_rate": 3.848157772439569e-06,
      "loss": 0.0536,
      "step": 280825
    },
    {
      "epoch": 4.1361688340377905,
      "grad_norm": 1.316820740699768,
      "learning_rate": 3.846521400472584e-06,
      "loss": 0.0487,
      "step": 280850
    },
    {
      "epoch": 4.1365370171278775,
      "grad_norm": 1.0695276260375977,
      "learning_rate": 3.8448850285056e-06,
      "loss": 0.0513,
      "step": 280875
    },
    {
      "epoch": 4.1369052002179645,
      "grad_norm": 0.7669237852096558,
      "learning_rate": 3.843248656538616e-06,
      "loss": 0.0465,
      "step": 280900
    },
    {
      "epoch": 4.1372733833080515,
      "grad_norm": 1.5514988899230957,
      "learning_rate": 3.841612284571631e-06,
      "loss": 0.0475,
      "step": 280925
    },
    {
      "epoch": 4.1376415663981385,
      "grad_norm": 1.8899503946304321,
      "learning_rate": 3.839975912604646e-06,
      "loss": 0.0518,
      "step": 280950
    },
    {
      "epoch": 4.1380097494882255,
      "grad_norm": 1.3596923351287842,
      "learning_rate": 3.838339540637662e-06,
      "loss": 0.0535,
      "step": 280975
    },
    {
      "epoch": 4.1383779325783125,
      "grad_norm": 1.0282924175262451,
      "learning_rate": 3.836703168670677e-06,
      "loss": 0.0504,
      "step": 281000
    },
    {
      "epoch": 4.1387461156684,
      "grad_norm": 0.9863817095756531,
      "learning_rate": 3.835066796703692e-06,
      "loss": 0.0546,
      "step": 281025
    },
    {
      "epoch": 4.139114298758487,
      "grad_norm": 1.4705183506011963,
      "learning_rate": 3.8334304247367084e-06,
      "loss": 0.0484,
      "step": 281050
    },
    {
      "epoch": 4.139482481848574,
      "grad_norm": 1.3739672899246216,
      "learning_rate": 3.831794052769724e-06,
      "loss": 0.0481,
      "step": 281075
    },
    {
      "epoch": 4.139850664938661,
      "grad_norm": 1.5501339435577393,
      "learning_rate": 3.830157680802739e-06,
      "loss": 0.056,
      "step": 281100
    },
    {
      "epoch": 4.140218848028748,
      "grad_norm": 1.1224405765533447,
      "learning_rate": 3.828521308835754e-06,
      "loss": 0.056,
      "step": 281125
    },
    {
      "epoch": 4.140587031118835,
      "grad_norm": 1.4552491903305054,
      "learning_rate": 3.82688493686877e-06,
      "loss": 0.0528,
      "step": 281150
    },
    {
      "epoch": 4.140955214208922,
      "grad_norm": 1.668328881263733,
      "learning_rate": 3.825248564901785e-06,
      "loss": 0.0459,
      "step": 281175
    },
    {
      "epoch": 4.141323397299009,
      "grad_norm": 1.903546690940857,
      "learning_rate": 3.823612192934801e-06,
      "loss": 0.0577,
      "step": 281200
    },
    {
      "epoch": 4.141691580389096,
      "grad_norm": 1.1953068971633911,
      "learning_rate": 3.8219758209678156e-06,
      "loss": 0.048,
      "step": 281225
    },
    {
      "epoch": 4.142059763479183,
      "grad_norm": 1.5615509748458862,
      "learning_rate": 3.820339449000832e-06,
      "loss": 0.0507,
      "step": 281250
    },
    {
      "epoch": 4.14242794656927,
      "grad_norm": 1.4437997341156006,
      "learning_rate": 3.818703077033847e-06,
      "loss": 0.0469,
      "step": 281275
    },
    {
      "epoch": 4.142796129659357,
      "grad_norm": 1.070771336555481,
      "learning_rate": 3.817066705066862e-06,
      "loss": 0.0509,
      "step": 281300
    },
    {
      "epoch": 4.143164312749444,
      "grad_norm": 1.5310577154159546,
      "learning_rate": 3.815430333099878e-06,
      "loss": 0.0474,
      "step": 281325
    },
    {
      "epoch": 4.143532495839531,
      "grad_norm": 0.6705294847488403,
      "learning_rate": 3.8137939611328934e-06,
      "loss": 0.0467,
      "step": 281350
    },
    {
      "epoch": 4.143900678929618,
      "grad_norm": 1.2362991571426392,
      "learning_rate": 3.812157589165909e-06,
      "loss": 0.0566,
      "step": 281375
    },
    {
      "epoch": 4.144268862019705,
      "grad_norm": 0.9386043548583984,
      "learning_rate": 3.8105212171989244e-06,
      "loss": 0.0529,
      "step": 281400
    },
    {
      "epoch": 4.144637045109792,
      "grad_norm": 1.6510556936264038,
      "learning_rate": 3.80888484523194e-06,
      "loss": 0.0514,
      "step": 281425
    },
    {
      "epoch": 4.145005228199879,
      "grad_norm": 1.139722466468811,
      "learning_rate": 3.807248473264955e-06,
      "loss": 0.054,
      "step": 281450
    },
    {
      "epoch": 4.145373411289967,
      "grad_norm": 1.3769398927688599,
      "learning_rate": 3.8056121012979707e-06,
      "loss": 0.0584,
      "step": 281475
    },
    {
      "epoch": 4.145741594380054,
      "grad_norm": 1.5605382919311523,
      "learning_rate": 3.803975729330986e-06,
      "loss": 0.0593,
      "step": 281500
    },
    {
      "epoch": 4.146109777470141,
      "grad_norm": 0.9440943002700806,
      "learning_rate": 3.8023393573640018e-06,
      "loss": 0.0458,
      "step": 281525
    },
    {
      "epoch": 4.146477960560228,
      "grad_norm": 0.9544996619224548,
      "learning_rate": 3.8007029853970166e-06,
      "loss": 0.0498,
      "step": 281550
    },
    {
      "epoch": 4.146846143650315,
      "grad_norm": 1.5641615390777588,
      "learning_rate": 3.7990666134300324e-06,
      "loss": 0.0515,
      "step": 281575
    },
    {
      "epoch": 4.147214326740402,
      "grad_norm": 1.2624694108963013,
      "learning_rate": 3.7974302414630477e-06,
      "loss": 0.0474,
      "step": 281600
    },
    {
      "epoch": 4.147582509830489,
      "grad_norm": 1.328643798828125,
      "learning_rate": 3.7957938694960634e-06,
      "loss": 0.0571,
      "step": 281625
    },
    {
      "epoch": 4.147950692920576,
      "grad_norm": 1.238291621208191,
      "learning_rate": 3.7941574975290783e-06,
      "loss": 0.0466,
      "step": 281650
    },
    {
      "epoch": 4.148318876010663,
      "grad_norm": 1.3468068838119507,
      "learning_rate": 3.792521125562094e-06,
      "loss": 0.05,
      "step": 281675
    },
    {
      "epoch": 4.14868705910075,
      "grad_norm": 1.4807982444763184,
      "learning_rate": 3.7909502084737887e-06,
      "loss": 0.048,
      "step": 281700
    },
    {
      "epoch": 4.149055242190837,
      "grad_norm": 1.6809862852096558,
      "learning_rate": 3.7893138365068045e-06,
      "loss": 0.0453,
      "step": 281725
    },
    {
      "epoch": 4.149423425280924,
      "grad_norm": 0.9660292267799377,
      "learning_rate": 3.7876774645398193e-06,
      "loss": 0.0534,
      "step": 281750
    },
    {
      "epoch": 4.149791608371011,
      "grad_norm": 1.4939463138580322,
      "learning_rate": 3.786041092572835e-06,
      "loss": 0.0493,
      "step": 281775
    },
    {
      "epoch": 4.150159791461098,
      "grad_norm": 1.056085467338562,
      "learning_rate": 3.7844047206058504e-06,
      "loss": 0.0547,
      "step": 281800
    },
    {
      "epoch": 4.150527974551185,
      "grad_norm": 1.0270249843597412,
      "learning_rate": 3.782768348638866e-06,
      "loss": 0.0548,
      "step": 281825
    },
    {
      "epoch": 4.150896157641272,
      "grad_norm": 0.9781486988067627,
      "learning_rate": 3.781131976671882e-06,
      "loss": 0.051,
      "step": 281850
    },
    {
      "epoch": 4.151264340731359,
      "grad_norm": 1.783305287361145,
      "learning_rate": 3.779495604704897e-06,
      "loss": 0.0494,
      "step": 281875
    },
    {
      "epoch": 4.151632523821446,
      "grad_norm": 1.3520704507827759,
      "learning_rate": 3.777859232737913e-06,
      "loss": 0.0534,
      "step": 281900
    },
    {
      "epoch": 4.152000706911533,
      "grad_norm": 1.6302826404571533,
      "learning_rate": 3.7762228607709277e-06,
      "loss": 0.0593,
      "step": 281925
    },
    {
      "epoch": 4.15236889000162,
      "grad_norm": 1.2334368228912354,
      "learning_rate": 3.7745864888039435e-06,
      "loss": 0.0515,
      "step": 281950
    },
    {
      "epoch": 4.152737073091707,
      "grad_norm": 1.8050910234451294,
      "learning_rate": 3.7729501168369588e-06,
      "loss": 0.0606,
      "step": 281975
    },
    {
      "epoch": 4.153105256181794,
      "grad_norm": 1.1785519123077393,
      "learning_rate": 3.7713137448699745e-06,
      "loss": 0.0569,
      "step": 282000
    },
    {
      "epoch": 4.153473439271881,
      "grad_norm": 1.6257681846618652,
      "learning_rate": 3.7696773729029894e-06,
      "loss": 0.053,
      "step": 282025
    },
    {
      "epoch": 4.153841622361968,
      "grad_norm": 1.4788684844970703,
      "learning_rate": 3.768041000936005e-06,
      "loss": 0.053,
      "step": 282050
    },
    {
      "epoch": 4.154209805452055,
      "grad_norm": 1.4266949892044067,
      "learning_rate": 3.7664046289690204e-06,
      "loss": 0.0541,
      "step": 282075
    },
    {
      "epoch": 4.154577988542142,
      "grad_norm": 1.146416425704956,
      "learning_rate": 3.764768257002036e-06,
      "loss": 0.0556,
      "step": 282100
    },
    {
      "epoch": 4.154946171632229,
      "grad_norm": 1.081526756286621,
      "learning_rate": 3.763131885035051e-06,
      "loss": 0.044,
      "step": 282125
    },
    {
      "epoch": 4.155314354722316,
      "grad_norm": 1.3343247175216675,
      "learning_rate": 3.7614955130680668e-06,
      "loss": 0.0517,
      "step": 282150
    },
    {
      "epoch": 4.155682537812403,
      "grad_norm": 0.8478697538375854,
      "learning_rate": 3.7599245959797615e-06,
      "loss": 0.0556,
      "step": 282175
    },
    {
      "epoch": 4.15605072090249,
      "grad_norm": 1.4555424451828003,
      "learning_rate": 3.758288224012777e-06,
      "loss": 0.0588,
      "step": 282200
    },
    {
      "epoch": 4.156418903992577,
      "grad_norm": 1.3658803701400757,
      "learning_rate": 3.7566518520457925e-06,
      "loss": 0.0528,
      "step": 282225
    },
    {
      "epoch": 4.156787087082664,
      "grad_norm": 1.5523141622543335,
      "learning_rate": 3.755015480078808e-06,
      "loss": 0.05,
      "step": 282250
    },
    {
      "epoch": 4.157155270172751,
      "grad_norm": 1.4585939645767212,
      "learning_rate": 3.753379108111823e-06,
      "loss": 0.0588,
      "step": 282275
    },
    {
      "epoch": 4.157523453262838,
      "grad_norm": 1.540112018585205,
      "learning_rate": 3.751742736144839e-06,
      "loss": 0.0543,
      "step": 282300
    },
    {
      "epoch": 4.157891636352925,
      "grad_norm": 1.4808543920516968,
      "learning_rate": 3.7501063641778546e-06,
      "loss": 0.0517,
      "step": 282325
    },
    {
      "epoch": 4.158259819443012,
      "grad_norm": 1.9939229488372803,
      "learning_rate": 3.74846999221087e-06,
      "loss": 0.0483,
      "step": 282350
    },
    {
      "epoch": 4.158628002533099,
      "grad_norm": 1.9306612014770508,
      "learning_rate": 3.7468336202438856e-06,
      "loss": 0.0544,
      "step": 282375
    },
    {
      "epoch": 4.158996185623186,
      "grad_norm": 0.961968183517456,
      "learning_rate": 3.7451972482769005e-06,
      "loss": 0.0463,
      "step": 282400
    },
    {
      "epoch": 4.159364368713273,
      "grad_norm": 1.0841031074523926,
      "learning_rate": 3.7435608763099162e-06,
      "loss": 0.055,
      "step": 282425
    },
    {
      "epoch": 4.15973255180336,
      "grad_norm": 1.4258852005004883,
      "learning_rate": 3.7419245043429315e-06,
      "loss": 0.0507,
      "step": 282450
    },
    {
      "epoch": 4.160100734893448,
      "grad_norm": 0.4652763307094574,
      "learning_rate": 3.7402881323759473e-06,
      "loss": 0.0538,
      "step": 282475
    },
    {
      "epoch": 4.160468917983535,
      "grad_norm": 1.4698705673217773,
      "learning_rate": 3.738651760408962e-06,
      "loss": 0.0444,
      "step": 282500
    },
    {
      "epoch": 4.160837101073622,
      "grad_norm": 0.6853455901145935,
      "learning_rate": 3.737015388441978e-06,
      "loss": 0.047,
      "step": 282525
    },
    {
      "epoch": 4.161205284163709,
      "grad_norm": 1.4518296718597412,
      "learning_rate": 3.735379016474993e-06,
      "loss": 0.0554,
      "step": 282550
    },
    {
      "epoch": 4.161573467253796,
      "grad_norm": 1.57321035861969,
      "learning_rate": 3.733742644508009e-06,
      "loss": 0.0568,
      "step": 282575
    },
    {
      "epoch": 4.161941650343883,
      "grad_norm": 1.0352312326431274,
      "learning_rate": 3.7321062725410238e-06,
      "loss": 0.0507,
      "step": 282600
    },
    {
      "epoch": 4.16230983343397,
      "grad_norm": 1.6174521446228027,
      "learning_rate": 3.7304699005740395e-06,
      "loss": 0.0541,
      "step": 282625
    },
    {
      "epoch": 4.162678016524057,
      "grad_norm": 1.185634732246399,
      "learning_rate": 3.728833528607055e-06,
      "loss": 0.0472,
      "step": 282650
    },
    {
      "epoch": 4.163046199614144,
      "grad_norm": 1.0461548566818237,
      "learning_rate": 3.7271971566400705e-06,
      "loss": 0.0491,
      "step": 282675
    },
    {
      "epoch": 4.163414382704231,
      "grad_norm": 2.0412161350250244,
      "learning_rate": 3.725560784673086e-06,
      "loss": 0.0483,
      "step": 282700
    },
    {
      "epoch": 4.163782565794318,
      "grad_norm": 1.139068603515625,
      "learning_rate": 3.7239244127061016e-06,
      "loss": 0.0546,
      "step": 282725
    },
    {
      "epoch": 4.164150748884405,
      "grad_norm": 1.1163042783737183,
      "learning_rate": 3.7222880407391164e-06,
      "loss": 0.0493,
      "step": 282750
    },
    {
      "epoch": 4.164518931974492,
      "grad_norm": 1.081230640411377,
      "learning_rate": 3.720651668772132e-06,
      "loss": 0.0481,
      "step": 282775
    },
    {
      "epoch": 4.164887115064579,
      "grad_norm": 1.8118054866790771,
      "learning_rate": 3.719015296805148e-06,
      "loss": 0.0491,
      "step": 282800
    },
    {
      "epoch": 4.165255298154666,
      "grad_norm": 1.4853826761245728,
      "learning_rate": 3.717378924838163e-06,
      "loss": 0.0468,
      "step": 282825
    },
    {
      "epoch": 4.1656234812447535,
      "grad_norm": 1.6419068574905396,
      "learning_rate": 3.715742552871179e-06,
      "loss": 0.0491,
      "step": 282850
    },
    {
      "epoch": 4.1659916643348405,
      "grad_norm": 1.3722608089447021,
      "learning_rate": 3.714106180904194e-06,
      "loss": 0.0527,
      "step": 282875
    },
    {
      "epoch": 4.1663598474249275,
      "grad_norm": 0.9031792879104614,
      "learning_rate": 3.7124698089372095e-06,
      "loss": 0.0489,
      "step": 282900
    },
    {
      "epoch": 4.1667280305150145,
      "grad_norm": 1.5830729007720947,
      "learning_rate": 3.710833436970225e-06,
      "loss": 0.0519,
      "step": 282925
    },
    {
      "epoch": 4.1670962136051015,
      "grad_norm": 1.047120451927185,
      "learning_rate": 3.7091970650032406e-06,
      "loss": 0.0441,
      "step": 282950
    },
    {
      "epoch": 4.1674643966951885,
      "grad_norm": 0.8555548787117004,
      "learning_rate": 3.7075606930362555e-06,
      "loss": 0.0478,
      "step": 282975
    },
    {
      "epoch": 4.1678325797852755,
      "grad_norm": 1.1296738386154175,
      "learning_rate": 3.705924321069271e-06,
      "loss": 0.0483,
      "step": 283000
    },
    {
      "epoch": 4.168200762875363,
      "grad_norm": 1.3878998756408691,
      "learning_rate": 3.7042879491022865e-06,
      "loss": 0.0553,
      "step": 283025
    },
    {
      "epoch": 4.16856894596545,
      "grad_norm": 1.4197288751602173,
      "learning_rate": 3.7026515771353022e-06,
      "loss": 0.0536,
      "step": 283050
    },
    {
      "epoch": 4.168937129055537,
      "grad_norm": 1.2826813459396362,
      "learning_rate": 3.7010152051683175e-06,
      "loss": 0.0496,
      "step": 283075
    },
    {
      "epoch": 4.169305312145624,
      "grad_norm": 1.5160043239593506,
      "learning_rate": 3.6993788332013333e-06,
      "loss": 0.0534,
      "step": 283100
    },
    {
      "epoch": 4.169673495235711,
      "grad_norm": 1.5410345792770386,
      "learning_rate": 3.697742461234348e-06,
      "loss": 0.0488,
      "step": 283125
    },
    {
      "epoch": 4.170041678325798,
      "grad_norm": 1.1372733116149902,
      "learning_rate": 3.696106089267364e-06,
      "loss": 0.0434,
      "step": 283150
    },
    {
      "epoch": 4.170409861415885,
      "grad_norm": 1.4371058940887451,
      "learning_rate": 3.694469717300379e-06,
      "loss": 0.0554,
      "step": 283175
    },
    {
      "epoch": 4.170778044505972,
      "grad_norm": 1.4943219423294067,
      "learning_rate": 3.692833345333395e-06,
      "loss": 0.0533,
      "step": 283200
    },
    {
      "epoch": 4.171146227596059,
      "grad_norm": 1.2485432624816895,
      "learning_rate": 3.6911969733664098e-06,
      "loss": 0.0548,
      "step": 283225
    },
    {
      "epoch": 4.171514410686146,
      "grad_norm": 1.5465879440307617,
      "learning_rate": 3.6895606013994255e-06,
      "loss": 0.0497,
      "step": 283250
    },
    {
      "epoch": 4.171882593776233,
      "grad_norm": 0.9757363200187683,
      "learning_rate": 3.687924229432441e-06,
      "loss": 0.0489,
      "step": 283275
    },
    {
      "epoch": 4.17225077686632,
      "grad_norm": 1.5635976791381836,
      "learning_rate": 3.6862878574654565e-06,
      "loss": 0.0562,
      "step": 283300
    },
    {
      "epoch": 4.172618959956407,
      "grad_norm": 1.2802245616912842,
      "learning_rate": 3.6846514854984723e-06,
      "loss": 0.0537,
      "step": 283325
    },
    {
      "epoch": 4.172987143046494,
      "grad_norm": 1.107739806175232,
      "learning_rate": 3.683015113531487e-06,
      "loss": 0.0481,
      "step": 283350
    },
    {
      "epoch": 4.173355326136581,
      "grad_norm": 1.1620582342147827,
      "learning_rate": 3.681378741564503e-06,
      "loss": 0.0512,
      "step": 283375
    },
    {
      "epoch": 4.173723509226669,
      "grad_norm": 1.1617927551269531,
      "learning_rate": 3.679742369597518e-06,
      "loss": 0.0483,
      "step": 283400
    },
    {
      "epoch": 4.174091692316756,
      "grad_norm": 1.037222981452942,
      "learning_rate": 3.678105997630534e-06,
      "loss": 0.0521,
      "step": 283425
    },
    {
      "epoch": 4.174459875406843,
      "grad_norm": 1.8881289958953857,
      "learning_rate": 3.676469625663549e-06,
      "loss": 0.0548,
      "step": 283450
    },
    {
      "epoch": 4.17482805849693,
      "grad_norm": 0.6663896441459656,
      "learning_rate": 3.674833253696565e-06,
      "loss": 0.0434,
      "step": 283475
    },
    {
      "epoch": 4.175196241587017,
      "grad_norm": 2.1065988540649414,
      "learning_rate": 3.67319688172958e-06,
      "loss": 0.0535,
      "step": 283500
    },
    {
      "epoch": 4.175564424677104,
      "grad_norm": 1.6611994504928589,
      "learning_rate": 3.6715605097625955e-06,
      "loss": 0.0504,
      "step": 283525
    },
    {
      "epoch": 4.175932607767191,
      "grad_norm": 1.2438486814498901,
      "learning_rate": 3.669924137795611e-06,
      "loss": 0.0539,
      "step": 283550
    },
    {
      "epoch": 4.176300790857278,
      "grad_norm": 1.3907822370529175,
      "learning_rate": 3.6682877658286266e-06,
      "loss": 0.0517,
      "step": 283575
    },
    {
      "epoch": 4.176668973947365,
      "grad_norm": 1.1284421682357788,
      "learning_rate": 3.6666513938616415e-06,
      "loss": 0.0522,
      "step": 283600
    },
    {
      "epoch": 4.177037157037452,
      "grad_norm": 1.4884239435195923,
      "learning_rate": 3.665015021894657e-06,
      "loss": 0.0595,
      "step": 283625
    },
    {
      "epoch": 4.177405340127539,
      "grad_norm": 1.5920871496200562,
      "learning_rate": 3.6633786499276725e-06,
      "loss": 0.0522,
      "step": 283650
    },
    {
      "epoch": 4.177773523217626,
      "grad_norm": 1.177476406097412,
      "learning_rate": 3.6617422779606882e-06,
      "loss": 0.0556,
      "step": 283675
    },
    {
      "epoch": 4.178141706307713,
      "grad_norm": 0.9358722567558289,
      "learning_rate": 3.660105905993703e-06,
      "loss": 0.0521,
      "step": 283700
    },
    {
      "epoch": 4.1785098893978,
      "grad_norm": 1.1437467336654663,
      "learning_rate": 3.658469534026719e-06,
      "loss": 0.0509,
      "step": 283725
    },
    {
      "epoch": 4.178878072487887,
      "grad_norm": 0.9306302666664124,
      "learning_rate": 3.656833162059734e-06,
      "loss": 0.0546,
      "step": 283750
    },
    {
      "epoch": 4.179246255577974,
      "grad_norm": 1.3671047687530518,
      "learning_rate": 3.65519679009275e-06,
      "loss": 0.0467,
      "step": 283775
    },
    {
      "epoch": 4.179614438668061,
      "grad_norm": 1.409212350845337,
      "learning_rate": 3.6535604181257656e-06,
      "loss": 0.0466,
      "step": 283800
    },
    {
      "epoch": 4.179982621758148,
      "grad_norm": 1.377893090248108,
      "learning_rate": 3.651924046158781e-06,
      "loss": 0.0567,
      "step": 283825
    },
    {
      "epoch": 4.180350804848235,
      "grad_norm": 1.5031005144119263,
      "learning_rate": 3.6502876741917966e-06,
      "loss": 0.0531,
      "step": 283850
    },
    {
      "epoch": 4.180718987938322,
      "grad_norm": 1.449777603149414,
      "learning_rate": 3.6486513022248115e-06,
      "loss": 0.0527,
      "step": 283875
    },
    {
      "epoch": 4.181087171028409,
      "grad_norm": 1.289481282234192,
      "learning_rate": 3.6470149302578272e-06,
      "loss": 0.0471,
      "step": 283900
    },
    {
      "epoch": 4.181455354118496,
      "grad_norm": 0.9766039252281189,
      "learning_rate": 3.6453785582908425e-06,
      "loss": 0.0474,
      "step": 283925
    },
    {
      "epoch": 4.181823537208583,
      "grad_norm": 1.397099494934082,
      "learning_rate": 3.6437421863238583e-06,
      "loss": 0.0487,
      "step": 283950
    },
    {
      "epoch": 4.18219172029867,
      "grad_norm": 1.3325046300888062,
      "learning_rate": 3.642105814356873e-06,
      "loss": 0.0547,
      "step": 283975
    },
    {
      "epoch": 4.182559903388757,
      "grad_norm": 0.7550286650657654,
      "learning_rate": 3.640469442389889e-06,
      "loss": 0.0505,
      "step": 284000
    },
    {
      "epoch": 4.182928086478844,
      "grad_norm": 1.2550404071807861,
      "learning_rate": 3.638833070422904e-06,
      "loss": 0.0571,
      "step": 284025
    },
    {
      "epoch": 4.183296269568931,
      "grad_norm": 1.4222699403762817,
      "learning_rate": 3.63719669845592e-06,
      "loss": 0.0497,
      "step": 284050
    },
    {
      "epoch": 4.183664452659018,
      "grad_norm": 1.6280872821807861,
      "learning_rate": 3.6355603264889348e-06,
      "loss": 0.054,
      "step": 284075
    },
    {
      "epoch": 4.184032635749105,
      "grad_norm": 0.9154566526412964,
      "learning_rate": 3.6339239545219505e-06,
      "loss": 0.0521,
      "step": 284100
    },
    {
      "epoch": 4.184400818839192,
      "grad_norm": 1.3835906982421875,
      "learning_rate": 3.632287582554966e-06,
      "loss": 0.055,
      "step": 284125
    },
    {
      "epoch": 4.184769001929279,
      "grad_norm": 1.424846887588501,
      "learning_rate": 3.6306512105879815e-06,
      "loss": 0.0504,
      "step": 284150
    },
    {
      "epoch": 4.185137185019366,
      "grad_norm": 1.2080756425857544,
      "learning_rate": 3.629014838620997e-06,
      "loss": 0.0547,
      "step": 284175
    },
    {
      "epoch": 4.185505368109453,
      "grad_norm": 1.2325810194015503,
      "learning_rate": 3.6273784666540126e-06,
      "loss": 0.0459,
      "step": 284200
    },
    {
      "epoch": 4.18587355119954,
      "grad_norm": 1.4504190683364868,
      "learning_rate": 3.6257420946870275e-06,
      "loss": 0.0446,
      "step": 284225
    },
    {
      "epoch": 4.186241734289627,
      "grad_norm": 0.8496175408363342,
      "learning_rate": 3.624105722720043e-06,
      "loss": 0.051,
      "step": 284250
    },
    {
      "epoch": 4.186609917379714,
      "grad_norm": 1.655601978302002,
      "learning_rate": 3.622469350753059e-06,
      "loss": 0.0564,
      "step": 284275
    },
    {
      "epoch": 4.186978100469801,
      "grad_norm": 1.0685817003250122,
      "learning_rate": 3.6208329787860742e-06,
      "loss": 0.0496,
      "step": 284300
    },
    {
      "epoch": 4.187346283559888,
      "grad_norm": 1.3044464588165283,
      "learning_rate": 3.61919660681909e-06,
      "loss": 0.0498,
      "step": 284325
    },
    {
      "epoch": 4.187714466649975,
      "grad_norm": 1.1293905973434448,
      "learning_rate": 3.617560234852105e-06,
      "loss": 0.0518,
      "step": 284350
    },
    {
      "epoch": 4.188082649740063,
      "grad_norm": 1.2364485263824463,
      "learning_rate": 3.6159893177638e-06,
      "loss": 0.049,
      "step": 284375
    },
    {
      "epoch": 4.18845083283015,
      "grad_norm": 1.1875444650650024,
      "learning_rate": 3.6143529457968153e-06,
      "loss": 0.0555,
      "step": 284400
    },
    {
      "epoch": 4.188819015920237,
      "grad_norm": 1.6219980716705322,
      "learning_rate": 3.612716573829831e-06,
      "loss": 0.0468,
      "step": 284425
    },
    {
      "epoch": 4.189187199010324,
      "grad_norm": 1.4333478212356567,
      "learning_rate": 3.611080201862846e-06,
      "loss": 0.0527,
      "step": 284450
    },
    {
      "epoch": 4.189555382100411,
      "grad_norm": 1.323794960975647,
      "learning_rate": 3.6094438298958616e-06,
      "loss": 0.0519,
      "step": 284475
    },
    {
      "epoch": 4.189923565190498,
      "grad_norm": 1.1227856874465942,
      "learning_rate": 3.607807457928877e-06,
      "loss": 0.0518,
      "step": 284500
    },
    {
      "epoch": 4.190291748280585,
      "grad_norm": 1.4547703266143799,
      "learning_rate": 3.6061710859618926e-06,
      "loss": 0.052,
      "step": 284525
    },
    {
      "epoch": 4.190659931370672,
      "grad_norm": 2.2386605739593506,
      "learning_rate": 3.6045347139949075e-06,
      "loss": 0.0551,
      "step": 284550
    },
    {
      "epoch": 4.191028114460759,
      "grad_norm": 1.3500272035598755,
      "learning_rate": 3.6028983420279233e-06,
      "loss": 0.0535,
      "step": 284575
    },
    {
      "epoch": 4.191396297550846,
      "grad_norm": 1.15129816532135,
      "learning_rate": 3.6012619700609386e-06,
      "loss": 0.0556,
      "step": 284600
    },
    {
      "epoch": 4.191764480640933,
      "grad_norm": 1.3669862747192383,
      "learning_rate": 3.5996255980939543e-06,
      "loss": 0.0541,
      "step": 284625
    },
    {
      "epoch": 4.19213266373102,
      "grad_norm": 1.4275031089782715,
      "learning_rate": 3.5979892261269696e-06,
      "loss": 0.0509,
      "step": 284650
    },
    {
      "epoch": 4.192500846821107,
      "grad_norm": 1.125692367553711,
      "learning_rate": 3.5963528541599853e-06,
      "loss": 0.0494,
      "step": 284675
    },
    {
      "epoch": 4.192869029911194,
      "grad_norm": 1.472734808921814,
      "learning_rate": 3.594716482193e-06,
      "loss": 0.0452,
      "step": 284700
    },
    {
      "epoch": 4.193237213001281,
      "grad_norm": 1.5592925548553467,
      "learning_rate": 3.593080110226016e-06,
      "loss": 0.0498,
      "step": 284725
    },
    {
      "epoch": 4.193605396091368,
      "grad_norm": 1.3757771253585815,
      "learning_rate": 3.5914437382590317e-06,
      "loss": 0.054,
      "step": 284750
    },
    {
      "epoch": 4.193973579181455,
      "grad_norm": 1.4402618408203125,
      "learning_rate": 3.589807366292047e-06,
      "loss": 0.0445,
      "step": 284775
    },
    {
      "epoch": 4.194341762271542,
      "grad_norm": 1.5237786769866943,
      "learning_rate": 3.5881709943250627e-06,
      "loss": 0.0544,
      "step": 284800
    },
    {
      "epoch": 4.194709945361629,
      "grad_norm": 1.1614034175872803,
      "learning_rate": 3.5865346223580776e-06,
      "loss": 0.0479,
      "step": 284825
    },
    {
      "epoch": 4.1950781284517165,
      "grad_norm": 1.3960483074188232,
      "learning_rate": 3.5848982503910933e-06,
      "loss": 0.0549,
      "step": 284850
    },
    {
      "epoch": 4.1954463115418035,
      "grad_norm": 1.4039913415908813,
      "learning_rate": 3.5832618784241086e-06,
      "loss": 0.05,
      "step": 284875
    },
    {
      "epoch": 4.1958144946318905,
      "grad_norm": 0.9151574969291687,
      "learning_rate": 3.5816255064571243e-06,
      "loss": 0.05,
      "step": 284900
    },
    {
      "epoch": 4.1961826777219775,
      "grad_norm": 1.6494172811508179,
      "learning_rate": 3.5799891344901392e-06,
      "loss": 0.0517,
      "step": 284925
    },
    {
      "epoch": 4.1965508608120645,
      "grad_norm": 1.4494985342025757,
      "learning_rate": 3.578352762523155e-06,
      "loss": 0.0533,
      "step": 284950
    },
    {
      "epoch": 4.1969190439021515,
      "grad_norm": 0.9448962211608887,
      "learning_rate": 3.5767163905561702e-06,
      "loss": 0.0572,
      "step": 284975
    },
    {
      "epoch": 4.1972872269922386,
      "grad_norm": 1.1785589456558228,
      "learning_rate": 3.575080018589186e-06,
      "loss": 0.0555,
      "step": 285000
    },
    {
      "epoch": 4.197655410082326,
      "grad_norm": 1.3010982275009155,
      "learning_rate": 3.5734436466222013e-06,
      "loss": 0.0525,
      "step": 285025
    },
    {
      "epoch": 4.198023593172413,
      "grad_norm": 1.6826128959655762,
      "learning_rate": 3.571807274655217e-06,
      "loss": 0.0543,
      "step": 285050
    },
    {
      "epoch": 4.1983917762625,
      "grad_norm": 0.929847002029419,
      "learning_rate": 3.570170902688232e-06,
      "loss": 0.0522,
      "step": 285075
    },
    {
      "epoch": 4.198759959352587,
      "grad_norm": 1.49109947681427,
      "learning_rate": 3.5685345307212476e-06,
      "loss": 0.0489,
      "step": 285100
    },
    {
      "epoch": 4.199128142442674,
      "grad_norm": 0.9287909865379333,
      "learning_rate": 3.566898158754263e-06,
      "loss": 0.0509,
      "step": 285125
    },
    {
      "epoch": 4.199496325532761,
      "grad_norm": 1.263285517692566,
      "learning_rate": 3.5652617867872786e-06,
      "loss": 0.0587,
      "step": 285150
    },
    {
      "epoch": 4.199864508622848,
      "grad_norm": 0.84065181016922,
      "learning_rate": 3.5636254148202935e-06,
      "loss": 0.0575,
      "step": 285175
    },
    {
      "epoch": 4.200232691712935,
      "grad_norm": 1.635359525680542,
      "learning_rate": 3.5619890428533093e-06,
      "loss": 0.0481,
      "step": 285200
    },
    {
      "epoch": 4.200600874803022,
      "grad_norm": 1.077462077140808,
      "learning_rate": 3.560352670886325e-06,
      "loss": 0.0508,
      "step": 285225
    },
    {
      "epoch": 4.200969057893109,
      "grad_norm": 1.0449047088623047,
      "learning_rate": 3.5587162989193403e-06,
      "loss": 0.0507,
      "step": 285250
    },
    {
      "epoch": 4.201337240983196,
      "grad_norm": 1.7464070320129395,
      "learning_rate": 3.557079926952356e-06,
      "loss": 0.0468,
      "step": 285275
    },
    {
      "epoch": 4.201705424073283,
      "grad_norm": 1.5482630729675293,
      "learning_rate": 3.555443554985371e-06,
      "loss": 0.0554,
      "step": 285300
    },
    {
      "epoch": 4.202073607163371,
      "grad_norm": 1.3082911968231201,
      "learning_rate": 3.5538071830183866e-06,
      "loss": 0.0518,
      "step": 285325
    },
    {
      "epoch": 4.202441790253458,
      "grad_norm": 1.0178221464157104,
      "learning_rate": 3.552170811051402e-06,
      "loss": 0.0479,
      "step": 285350
    },
    {
      "epoch": 4.202809973343545,
      "grad_norm": 1.0560585260391235,
      "learning_rate": 3.5505344390844177e-06,
      "loss": 0.0512,
      "step": 285375
    },
    {
      "epoch": 4.203178156433632,
      "grad_norm": 1.4004467725753784,
      "learning_rate": 3.548898067117433e-06,
      "loss": 0.0557,
      "step": 285400
    },
    {
      "epoch": 4.203546339523719,
      "grad_norm": 1.5529333353042603,
      "learning_rate": 3.5472616951504487e-06,
      "loss": 0.0498,
      "step": 285425
    },
    {
      "epoch": 4.203914522613806,
      "grad_norm": 1.1342477798461914,
      "learning_rate": 3.5456253231834636e-06,
      "loss": 0.0541,
      "step": 285450
    },
    {
      "epoch": 4.204282705703893,
      "grad_norm": 1.4639195203781128,
      "learning_rate": 3.5439889512164793e-06,
      "loss": 0.0545,
      "step": 285475
    },
    {
      "epoch": 4.20465088879398,
      "grad_norm": 1.3467246294021606,
      "learning_rate": 3.5423525792494946e-06,
      "loss": 0.0494,
      "step": 285500
    },
    {
      "epoch": 4.205019071884067,
      "grad_norm": 1.3431875705718994,
      "learning_rate": 3.5407162072825103e-06,
      "loss": 0.0549,
      "step": 285525
    },
    {
      "epoch": 4.205387254974154,
      "grad_norm": 0.9523720741271973,
      "learning_rate": 3.5390798353155252e-06,
      "loss": 0.0509,
      "step": 285550
    },
    {
      "epoch": 4.205755438064241,
      "grad_norm": 1.1625728607177734,
      "learning_rate": 3.537443463348541e-06,
      "loss": 0.0514,
      "step": 285575
    },
    {
      "epoch": 4.206123621154328,
      "grad_norm": 1.586590051651001,
      "learning_rate": 3.5358070913815562e-06,
      "loss": 0.0492,
      "step": 285600
    },
    {
      "epoch": 4.206491804244415,
      "grad_norm": 1.7192764282226562,
      "learning_rate": 3.534170719414572e-06,
      "loss": 0.0527,
      "step": 285625
    },
    {
      "epoch": 4.206859987334502,
      "grad_norm": 0.8004681468009949,
      "learning_rate": 3.532534347447587e-06,
      "loss": 0.0478,
      "step": 285650
    },
    {
      "epoch": 4.207228170424589,
      "grad_norm": 0.976959228515625,
      "learning_rate": 3.5308979754806026e-06,
      "loss": 0.0536,
      "step": 285675
    },
    {
      "epoch": 4.207596353514676,
      "grad_norm": 1.2287458181381226,
      "learning_rate": 3.529261603513618e-06,
      "loss": 0.0519,
      "step": 285700
    },
    {
      "epoch": 4.207964536604763,
      "grad_norm": 1.6525061130523682,
      "learning_rate": 3.5276252315466336e-06,
      "loss": 0.0564,
      "step": 285725
    },
    {
      "epoch": 4.20833271969485,
      "grad_norm": 1.0059435367584229,
      "learning_rate": 3.5259888595796493e-06,
      "loss": 0.053,
      "step": 285750
    },
    {
      "epoch": 4.208700902784937,
      "grad_norm": 1.0858955383300781,
      "learning_rate": 3.5243524876126646e-06,
      "loss": 0.0458,
      "step": 285775
    },
    {
      "epoch": 4.209069085875024,
      "grad_norm": 0.9993805289268494,
      "learning_rate": 3.5227161156456804e-06,
      "loss": 0.0493,
      "step": 285800
    },
    {
      "epoch": 4.209437268965111,
      "grad_norm": 1.0551505088806152,
      "learning_rate": 3.5210797436786953e-06,
      "loss": 0.0514,
      "step": 285825
    },
    {
      "epoch": 4.209805452055198,
      "grad_norm": 0.9285268187522888,
      "learning_rate": 3.519443371711711e-06,
      "loss": 0.0492,
      "step": 285850
    },
    {
      "epoch": 4.210173635145285,
      "grad_norm": 0.6509037017822266,
      "learning_rate": 3.5178069997447263e-06,
      "loss": 0.0492,
      "step": 285875
    },
    {
      "epoch": 4.210541818235372,
      "grad_norm": 1.442220687866211,
      "learning_rate": 3.516170627777742e-06,
      "loss": 0.051,
      "step": 285900
    },
    {
      "epoch": 4.210910001325459,
      "grad_norm": 1.7208672761917114,
      "learning_rate": 3.514534255810757e-06,
      "loss": 0.0512,
      "step": 285925
    },
    {
      "epoch": 4.211278184415546,
      "grad_norm": 1.4775081872940063,
      "learning_rate": 3.5128978838437726e-06,
      "loss": 0.0509,
      "step": 285950
    },
    {
      "epoch": 4.211646367505633,
      "grad_norm": 1.3374617099761963,
      "learning_rate": 3.511261511876788e-06,
      "loss": 0.0586,
      "step": 285975
    },
    {
      "epoch": 4.21201455059572,
      "grad_norm": 1.54880690574646,
      "learning_rate": 3.5096251399098037e-06,
      "loss": 0.0491,
      "step": 286000
    },
    {
      "epoch": 4.212382733685807,
      "grad_norm": 1.2430872917175293,
      "learning_rate": 3.5079887679428185e-06,
      "loss": 0.0569,
      "step": 286025
    },
    {
      "epoch": 4.212750916775894,
      "grad_norm": 1.08426833152771,
      "learning_rate": 3.5063523959758343e-06,
      "loss": 0.0508,
      "step": 286050
    },
    {
      "epoch": 4.213119099865981,
      "grad_norm": 1.4624437093734741,
      "learning_rate": 3.5047160240088496e-06,
      "loss": 0.0487,
      "step": 286075
    },
    {
      "epoch": 4.213487282956068,
      "grad_norm": 1.0623691082000732,
      "learning_rate": 3.5030796520418653e-06,
      "loss": 0.053,
      "step": 286100
    },
    {
      "epoch": 4.213855466046155,
      "grad_norm": 1.3077738285064697,
      "learning_rate": 3.5014432800748806e-06,
      "loss": 0.0495,
      "step": 286125
    },
    {
      "epoch": 4.214223649136242,
      "grad_norm": 1.6256091594696045,
      "learning_rate": 3.4998069081078963e-06,
      "loss": 0.0486,
      "step": 286150
    },
    {
      "epoch": 4.214591832226329,
      "grad_norm": 1.4712008237838745,
      "learning_rate": 3.4981705361409112e-06,
      "loss": 0.047,
      "step": 286175
    },
    {
      "epoch": 4.214960015316416,
      "grad_norm": 1.390589952468872,
      "learning_rate": 3.496534164173927e-06,
      "loss": 0.0507,
      "step": 286200
    },
    {
      "epoch": 4.215328198406503,
      "grad_norm": 1.317811131477356,
      "learning_rate": 3.4948977922069427e-06,
      "loss": 0.0506,
      "step": 286225
    },
    {
      "epoch": 4.21569638149659,
      "grad_norm": 0.9666112065315247,
      "learning_rate": 3.493261420239958e-06,
      "loss": 0.0484,
      "step": 286250
    },
    {
      "epoch": 4.216064564586677,
      "grad_norm": 1.3020298480987549,
      "learning_rate": 3.4916250482729737e-06,
      "loss": 0.0521,
      "step": 286275
    },
    {
      "epoch": 4.216432747676765,
      "grad_norm": 1.4063552618026733,
      "learning_rate": 3.4899886763059886e-06,
      "loss": 0.0509,
      "step": 286300
    },
    {
      "epoch": 4.216800930766852,
      "grad_norm": 1.3392831087112427,
      "learning_rate": 3.4883523043390043e-06,
      "loss": 0.0502,
      "step": 286325
    },
    {
      "epoch": 4.217169113856939,
      "grad_norm": 1.3368010520935059,
      "learning_rate": 3.4867159323720196e-06,
      "loss": 0.0532,
      "step": 286350
    },
    {
      "epoch": 4.217537296947026,
      "grad_norm": 1.1430208683013916,
      "learning_rate": 3.4850795604050353e-06,
      "loss": 0.059,
      "step": 286375
    },
    {
      "epoch": 4.217905480037113,
      "grad_norm": 1.1475892066955566,
      "learning_rate": 3.4834431884380502e-06,
      "loss": 0.0501,
      "step": 286400
    },
    {
      "epoch": 4.2182736631272,
      "grad_norm": 1.3156911134719849,
      "learning_rate": 3.481806816471066e-06,
      "loss": 0.0544,
      "step": 286425
    },
    {
      "epoch": 4.218641846217287,
      "grad_norm": 1.2806178331375122,
      "learning_rate": 3.4801704445040813e-06,
      "loss": 0.054,
      "step": 286450
    },
    {
      "epoch": 4.219010029307374,
      "grad_norm": 1.2796999216079712,
      "learning_rate": 3.478534072537097e-06,
      "loss": 0.0483,
      "step": 286475
    },
    {
      "epoch": 4.219378212397461,
      "grad_norm": 0.8412948250770569,
      "learning_rate": 3.4768977005701123e-06,
      "loss": 0.05,
      "step": 286500
    },
    {
      "epoch": 4.219746395487548,
      "grad_norm": 1.1998164653778076,
      "learning_rate": 3.475261328603128e-06,
      "loss": 0.0564,
      "step": 286525
    },
    {
      "epoch": 4.220114578577635,
      "grad_norm": 1.5432463884353638,
      "learning_rate": 3.473624956636143e-06,
      "loss": 0.0571,
      "step": 286550
    },
    {
      "epoch": 4.220482761667722,
      "grad_norm": 1.593366265296936,
      "learning_rate": 3.4719885846691586e-06,
      "loss": 0.0594,
      "step": 286575
    },
    {
      "epoch": 4.220850944757809,
      "grad_norm": 1.1081656217575073,
      "learning_rate": 3.470352212702174e-06,
      "loss": 0.0482,
      "step": 286600
    },
    {
      "epoch": 4.221219127847896,
      "grad_norm": 1.377360463142395,
      "learning_rate": 3.4687158407351897e-06,
      "loss": 0.0547,
      "step": 286625
    },
    {
      "epoch": 4.221587310937983,
      "grad_norm": 1.5743542909622192,
      "learning_rate": 3.4670794687682045e-06,
      "loss": 0.0519,
      "step": 286650
    },
    {
      "epoch": 4.22195549402807,
      "grad_norm": 1.342956304550171,
      "learning_rate": 3.4654430968012203e-06,
      "loss": 0.0505,
      "step": 286675
    },
    {
      "epoch": 4.222323677118157,
      "grad_norm": 1.1712003946304321,
      "learning_rate": 3.463806724834236e-06,
      "loss": 0.0556,
      "step": 286700
    },
    {
      "epoch": 4.222691860208244,
      "grad_norm": 1.3126769065856934,
      "learning_rate": 3.4621703528672513e-06,
      "loss": 0.0515,
      "step": 286725
    },
    {
      "epoch": 4.223060043298331,
      "grad_norm": 1.6280109882354736,
      "learning_rate": 3.460533980900267e-06,
      "loss": 0.0562,
      "step": 286750
    },
    {
      "epoch": 4.223428226388418,
      "grad_norm": 1.3670576810836792,
      "learning_rate": 3.458897608933282e-06,
      "loss": 0.0525,
      "step": 286775
    },
    {
      "epoch": 4.223796409478505,
      "grad_norm": 1.4677999019622803,
      "learning_rate": 3.4572612369662976e-06,
      "loss": 0.0455,
      "step": 286800
    },
    {
      "epoch": 4.2241645925685924,
      "grad_norm": 1.2276616096496582,
      "learning_rate": 3.455624864999313e-06,
      "loss": 0.0477,
      "step": 286825
    },
    {
      "epoch": 4.2245327756586795,
      "grad_norm": 1.565685510635376,
      "learning_rate": 3.4539884930323287e-06,
      "loss": 0.0573,
      "step": 286850
    },
    {
      "epoch": 4.2249009587487665,
      "grad_norm": 1.4173812866210938,
      "learning_rate": 3.452352121065344e-06,
      "loss": 0.0501,
      "step": 286875
    },
    {
      "epoch": 4.2252691418388535,
      "grad_norm": 1.1952711343765259,
      "learning_rate": 3.4507157490983597e-06,
      "loss": 0.0485,
      "step": 286900
    },
    {
      "epoch": 4.2256373249289405,
      "grad_norm": 1.399548053741455,
      "learning_rate": 3.4490793771313746e-06,
      "loss": 0.0563,
      "step": 286925
    },
    {
      "epoch": 4.2260055080190275,
      "grad_norm": 1.5876163244247437,
      "learning_rate": 3.4474430051643903e-06,
      "loss": 0.0471,
      "step": 286950
    },
    {
      "epoch": 4.2263736911091145,
      "grad_norm": 1.351813793182373,
      "learning_rate": 3.4458066331974056e-06,
      "loss": 0.0533,
      "step": 286975
    },
    {
      "epoch": 4.2267418741992016,
      "grad_norm": 1.292729139328003,
      "learning_rate": 3.4441702612304213e-06,
      "loss": 0.0451,
      "step": 287000
    },
    {
      "epoch": 4.227110057289289,
      "grad_norm": 1.1002082824707031,
      "learning_rate": 3.4425338892634362e-06,
      "loss": 0.0495,
      "step": 287025
    },
    {
      "epoch": 4.227478240379376,
      "grad_norm": 1.2477836608886719,
      "learning_rate": 3.440897517296452e-06,
      "loss": 0.0536,
      "step": 287050
    },
    {
      "epoch": 4.227846423469463,
      "grad_norm": 0.9137217998504639,
      "learning_rate": 3.4392611453294673e-06,
      "loss": 0.0458,
      "step": 287075
    },
    {
      "epoch": 4.22821460655955,
      "grad_norm": 0.747760534286499,
      "learning_rate": 3.437624773362483e-06,
      "loss": 0.0497,
      "step": 287100
    },
    {
      "epoch": 4.228582789649637,
      "grad_norm": 1.180064082145691,
      "learning_rate": 3.435988401395498e-06,
      "loss": 0.0536,
      "step": 287125
    },
    {
      "epoch": 4.228950972739724,
      "grad_norm": 1.5068246126174927,
      "learning_rate": 3.4343520294285136e-06,
      "loss": 0.0498,
      "step": 287150
    },
    {
      "epoch": 4.229319155829811,
      "grad_norm": 1.101651668548584,
      "learning_rate": 3.432715657461529e-06,
      "loss": 0.0536,
      "step": 287175
    },
    {
      "epoch": 4.229687338919898,
      "grad_norm": 1.7168573141098022,
      "learning_rate": 3.431144740373224e-06,
      "loss": 0.0529,
      "step": 287200
    },
    {
      "epoch": 4.230055522009986,
      "grad_norm": 1.4192029237747192,
      "learning_rate": 3.4295083684062398e-06,
      "loss": 0.0465,
      "step": 287225
    },
    {
      "epoch": 4.230423705100073,
      "grad_norm": 1.5027098655700684,
      "learning_rate": 3.4278719964392547e-06,
      "loss": 0.0549,
      "step": 287250
    },
    {
      "epoch": 4.23079188819016,
      "grad_norm": 1.5143821239471436,
      "learning_rate": 3.4262356244722704e-06,
      "loss": 0.0535,
      "step": 287275
    },
    {
      "epoch": 4.231160071280247,
      "grad_norm": 1.3449275493621826,
      "learning_rate": 3.4245992525052857e-06,
      "loss": 0.047,
      "step": 287300
    },
    {
      "epoch": 4.231528254370334,
      "grad_norm": 1.2649414539337158,
      "learning_rate": 3.4229628805383014e-06,
      "loss": 0.0526,
      "step": 287325
    },
    {
      "epoch": 4.231896437460421,
      "grad_norm": 1.6541857719421387,
      "learning_rate": 3.4213265085713167e-06,
      "loss": 0.0535,
      "step": 287350
    },
    {
      "epoch": 4.232264620550508,
      "grad_norm": 1.4935722351074219,
      "learning_rate": 3.4196901366043324e-06,
      "loss": 0.0615,
      "step": 287375
    },
    {
      "epoch": 4.232632803640595,
      "grad_norm": 1.4454498291015625,
      "learning_rate": 3.4180537646373473e-06,
      "loss": 0.0546,
      "step": 287400
    },
    {
      "epoch": 4.233000986730682,
      "grad_norm": 0.9084637761116028,
      "learning_rate": 3.416417392670363e-06,
      "loss": 0.0535,
      "step": 287425
    },
    {
      "epoch": 4.233369169820769,
      "grad_norm": 1.5626475811004639,
      "learning_rate": 3.4147810207033784e-06,
      "loss": 0.0495,
      "step": 287450
    },
    {
      "epoch": 4.233737352910856,
      "grad_norm": 1.3954142332077026,
      "learning_rate": 3.413144648736394e-06,
      "loss": 0.0537,
      "step": 287475
    },
    {
      "epoch": 4.234105536000943,
      "grad_norm": 1.2264585494995117,
      "learning_rate": 3.411508276769409e-06,
      "loss": 0.0523,
      "step": 287500
    },
    {
      "epoch": 4.23447371909103,
      "grad_norm": 1.78122878074646,
      "learning_rate": 3.4098719048024247e-06,
      "loss": 0.0535,
      "step": 287525
    },
    {
      "epoch": 4.234841902181117,
      "grad_norm": 1.4941461086273193,
      "learning_rate": 3.40823553283544e-06,
      "loss": 0.0474,
      "step": 287550
    },
    {
      "epoch": 4.235210085271204,
      "grad_norm": 1.4036809206008911,
      "learning_rate": 3.4065991608684557e-06,
      "loss": 0.053,
      "step": 287575
    },
    {
      "epoch": 4.235578268361291,
      "grad_norm": 1.3087482452392578,
      "learning_rate": 3.4049627889014706e-06,
      "loss": 0.0598,
      "step": 287600
    },
    {
      "epoch": 4.235946451451378,
      "grad_norm": 1.2929563522338867,
      "learning_rate": 3.4033264169344863e-06,
      "loss": 0.0502,
      "step": 287625
    },
    {
      "epoch": 4.236314634541465,
      "grad_norm": 1.1535840034484863,
      "learning_rate": 3.4016900449675016e-06,
      "loss": 0.0509,
      "step": 287650
    },
    {
      "epoch": 4.236682817631552,
      "grad_norm": 1.151566743850708,
      "learning_rate": 3.4000536730005174e-06,
      "loss": 0.0499,
      "step": 287675
    },
    {
      "epoch": 4.237051000721639,
      "grad_norm": 1.2608799934387207,
      "learning_rate": 3.398417301033533e-06,
      "loss": 0.054,
      "step": 287700
    },
    {
      "epoch": 4.237419183811726,
      "grad_norm": 1.074532389640808,
      "learning_rate": 3.3967809290665484e-06,
      "loss": 0.0517,
      "step": 287725
    },
    {
      "epoch": 4.237787366901813,
      "grad_norm": 1.3132317066192627,
      "learning_rate": 3.395144557099564e-06,
      "loss": 0.0536,
      "step": 287750
    },
    {
      "epoch": 4.2381555499919,
      "grad_norm": 1.4164564609527588,
      "learning_rate": 3.393508185132579e-06,
      "loss": 0.0569,
      "step": 287775
    },
    {
      "epoch": 4.238523733081987,
      "grad_norm": 0.8150655031204224,
      "learning_rate": 3.3918718131655947e-06,
      "loss": 0.0475,
      "step": 287800
    },
    {
      "epoch": 4.238891916172074,
      "grad_norm": 1.7023588418960571,
      "learning_rate": 3.39023544119861e-06,
      "loss": 0.0533,
      "step": 287825
    },
    {
      "epoch": 4.239260099262161,
      "grad_norm": 1.1218279600143433,
      "learning_rate": 3.3885990692316258e-06,
      "loss": 0.0591,
      "step": 287850
    },
    {
      "epoch": 4.239628282352248,
      "grad_norm": 1.4748696088790894,
      "learning_rate": 3.3869626972646407e-06,
      "loss": 0.0513,
      "step": 287875
    },
    {
      "epoch": 4.239996465442335,
      "grad_norm": 1.154908299446106,
      "learning_rate": 3.3853263252976564e-06,
      "loss": 0.052,
      "step": 287900
    },
    {
      "epoch": 4.240364648532422,
      "grad_norm": 1.170432448387146,
      "learning_rate": 3.3836899533306717e-06,
      "loss": 0.0453,
      "step": 287925
    },
    {
      "epoch": 4.240732831622509,
      "grad_norm": 1.052100658416748,
      "learning_rate": 3.3820535813636874e-06,
      "loss": 0.0595,
      "step": 287950
    },
    {
      "epoch": 4.241101014712596,
      "grad_norm": 1.3566515445709229,
      "learning_rate": 3.3804172093967023e-06,
      "loss": 0.0505,
      "step": 287975
    },
    {
      "epoch": 4.241469197802683,
      "grad_norm": 2.012906551361084,
      "learning_rate": 3.378780837429718e-06,
      "loss": 0.0473,
      "step": 288000
    },
    {
      "epoch": 4.24183738089277,
      "grad_norm": 1.811879277229309,
      "learning_rate": 3.3771444654627333e-06,
      "loss": 0.0533,
      "step": 288025
    },
    {
      "epoch": 4.242205563982857,
      "grad_norm": 1.1512092351913452,
      "learning_rate": 3.375508093495749e-06,
      "loss": 0.0489,
      "step": 288050
    },
    {
      "epoch": 4.242573747072944,
      "grad_norm": 1.3555552959442139,
      "learning_rate": 3.3738717215287644e-06,
      "loss": 0.0543,
      "step": 288075
    },
    {
      "epoch": 4.242941930163031,
      "grad_norm": 1.0794495344161987,
      "learning_rate": 3.37223534956178e-06,
      "loss": 0.0523,
      "step": 288100
    },
    {
      "epoch": 4.243310113253118,
      "grad_norm": 1.7101385593414307,
      "learning_rate": 3.370598977594795e-06,
      "loss": 0.0592,
      "step": 288125
    },
    {
      "epoch": 4.243678296343205,
      "grad_norm": 1.3256570100784302,
      "learning_rate": 3.3689626056278107e-06,
      "loss": 0.0571,
      "step": 288150
    },
    {
      "epoch": 4.244046479433292,
      "grad_norm": 1.3084527254104614,
      "learning_rate": 3.3673262336608264e-06,
      "loss": 0.057,
      "step": 288175
    },
    {
      "epoch": 4.244414662523379,
      "grad_norm": 0.7994515299797058,
      "learning_rate": 3.3656898616938417e-06,
      "loss": 0.052,
      "step": 288200
    },
    {
      "epoch": 4.244782845613467,
      "grad_norm": 1.5635790824890137,
      "learning_rate": 3.3640534897268575e-06,
      "loss": 0.0573,
      "step": 288225
    },
    {
      "epoch": 4.245151028703554,
      "grad_norm": 1.1247919797897339,
      "learning_rate": 3.3624171177598723e-06,
      "loss": 0.055,
      "step": 288250
    },
    {
      "epoch": 4.245519211793641,
      "grad_norm": 1.2731386423110962,
      "learning_rate": 3.360780745792888e-06,
      "loss": 0.0541,
      "step": 288275
    },
    {
      "epoch": 4.245887394883728,
      "grad_norm": 1.2959508895874023,
      "learning_rate": 3.3591443738259034e-06,
      "loss": 0.0426,
      "step": 288300
    },
    {
      "epoch": 4.246255577973815,
      "grad_norm": 1.116158127784729,
      "learning_rate": 3.357508001858919e-06,
      "loss": 0.0561,
      "step": 288325
    },
    {
      "epoch": 4.246623761063902,
      "grad_norm": 0.6781604290008545,
      "learning_rate": 3.355871629891934e-06,
      "loss": 0.0398,
      "step": 288350
    },
    {
      "epoch": 4.246991944153989,
      "grad_norm": 1.5362576246261597,
      "learning_rate": 3.3542352579249497e-06,
      "loss": 0.0574,
      "step": 288375
    },
    {
      "epoch": 4.247360127244076,
      "grad_norm": 1.4149580001831055,
      "learning_rate": 3.352598885957965e-06,
      "loss": 0.0517,
      "step": 288400
    },
    {
      "epoch": 4.247728310334163,
      "grad_norm": 1.4112199544906616,
      "learning_rate": 3.3509625139909807e-06,
      "loss": 0.0545,
      "step": 288425
    },
    {
      "epoch": 4.24809649342425,
      "grad_norm": 0.7767438888549805,
      "learning_rate": 3.349326142023996e-06,
      "loss": 0.0446,
      "step": 288450
    },
    {
      "epoch": 4.248464676514337,
      "grad_norm": 1.3089207410812378,
      "learning_rate": 3.3476897700570118e-06,
      "loss": 0.0492,
      "step": 288475
    },
    {
      "epoch": 4.248832859604424,
      "grad_norm": 1.3279422521591187,
      "learning_rate": 3.3460533980900267e-06,
      "loss": 0.054,
      "step": 288500
    },
    {
      "epoch": 4.249201042694511,
      "grad_norm": 1.2852131128311157,
      "learning_rate": 3.3444170261230424e-06,
      "loss": 0.0488,
      "step": 288525
    },
    {
      "epoch": 4.249569225784598,
      "grad_norm": 1.088971495628357,
      "learning_rate": 3.3427806541560577e-06,
      "loss": 0.0527,
      "step": 288550
    },
    {
      "epoch": 4.249937408874685,
      "grad_norm": 1.6746782064437866,
      "learning_rate": 3.3411442821890734e-06,
      "loss": 0.0581,
      "step": 288575
    },
    {
      "epoch": 4.250305591964772,
      "grad_norm": 1.1501498222351074,
      "learning_rate": 3.3395079102220883e-06,
      "loss": 0.0525,
      "step": 288600
    },
    {
      "epoch": 4.250673775054859,
      "grad_norm": 1.5668296813964844,
      "learning_rate": 3.337871538255104e-06,
      "loss": 0.048,
      "step": 288625
    },
    {
      "epoch": 4.251041958144946,
      "grad_norm": 1.3431293964385986,
      "learning_rate": 3.3362351662881197e-06,
      "loss": 0.0458,
      "step": 288650
    },
    {
      "epoch": 4.251410141235033,
      "grad_norm": 1.4084354639053345,
      "learning_rate": 3.334598794321135e-06,
      "loss": 0.0539,
      "step": 288675
    },
    {
      "epoch": 4.25177832432512,
      "grad_norm": 1.9039839506149292,
      "learning_rate": 3.3329624223541508e-06,
      "loss": 0.0532,
      "step": 288700
    },
    {
      "epoch": 4.252146507415207,
      "grad_norm": 1.0529685020446777,
      "learning_rate": 3.3313260503871657e-06,
      "loss": 0.0555,
      "step": 288725
    },
    {
      "epoch": 4.252514690505294,
      "grad_norm": 1.086586356163025,
      "learning_rate": 3.3296896784201814e-06,
      "loss": 0.0485,
      "step": 288750
    },
    {
      "epoch": 4.252882873595381,
      "grad_norm": 0.8603990077972412,
      "learning_rate": 3.3280533064531967e-06,
      "loss": 0.0555,
      "step": 288775
    },
    {
      "epoch": 4.253251056685468,
      "grad_norm": 1.4172135591506958,
      "learning_rate": 3.3264169344862124e-06,
      "loss": 0.0494,
      "step": 288800
    },
    {
      "epoch": 4.2536192397755554,
      "grad_norm": 1.1183111667633057,
      "learning_rate": 3.3247805625192277e-06,
      "loss": 0.0562,
      "step": 288825
    },
    {
      "epoch": 4.2539874228656425,
      "grad_norm": 1.5140810012817383,
      "learning_rate": 3.3231441905522435e-06,
      "loss": 0.0507,
      "step": 288850
    },
    {
      "epoch": 4.2543556059557295,
      "grad_norm": 1.7189756631851196,
      "learning_rate": 3.3215078185852583e-06,
      "loss": 0.0505,
      "step": 288875
    },
    {
      "epoch": 4.2547237890458165,
      "grad_norm": 0.8740991353988647,
      "learning_rate": 3.319871446618274e-06,
      "loss": 0.0513,
      "step": 288900
    },
    {
      "epoch": 4.2550919721359035,
      "grad_norm": 1.5767619609832764,
      "learning_rate": 3.3182350746512894e-06,
      "loss": 0.0513,
      "step": 288925
    },
    {
      "epoch": 4.2554601552259905,
      "grad_norm": 1.383811593055725,
      "learning_rate": 3.316598702684305e-06,
      "loss": 0.0604,
      "step": 288950
    },
    {
      "epoch": 4.2558283383160775,
      "grad_norm": 1.1832149028778076,
      "learning_rate": 3.31496233071732e-06,
      "loss": 0.0531,
      "step": 288975
    },
    {
      "epoch": 4.256196521406165,
      "grad_norm": 1.482468605041504,
      "learning_rate": 3.3133259587503357e-06,
      "loss": 0.0528,
      "step": 289000
    },
    {
      "epoch": 4.256564704496252,
      "grad_norm": 1.53411865234375,
      "learning_rate": 3.311689586783351e-06,
      "loss": 0.0499,
      "step": 289025
    },
    {
      "epoch": 4.256932887586339,
      "grad_norm": 1.00519597530365,
      "learning_rate": 3.3100532148163667e-06,
      "loss": 0.0533,
      "step": 289050
    },
    {
      "epoch": 4.257301070676426,
      "grad_norm": 1.3609029054641724,
      "learning_rate": 3.3084168428493816e-06,
      "loss": 0.0557,
      "step": 289075
    },
    {
      "epoch": 4.257669253766513,
      "grad_norm": 1.0265083312988281,
      "learning_rate": 3.3067804708823973e-06,
      "loss": 0.0508,
      "step": 289100
    },
    {
      "epoch": 4.2580374368566005,
      "grad_norm": 1.1739802360534668,
      "learning_rate": 3.3051440989154127e-06,
      "loss": 0.0551,
      "step": 289125
    },
    {
      "epoch": 4.2584056199466875,
      "grad_norm": 0.864486038684845,
      "learning_rate": 3.3035077269484284e-06,
      "loss": 0.0504,
      "step": 289150
    },
    {
      "epoch": 4.2587738030367746,
      "grad_norm": 1.1211373805999756,
      "learning_rate": 3.301871354981444e-06,
      "loss": 0.0467,
      "step": 289175
    },
    {
      "epoch": 4.259141986126862,
      "grad_norm": 1.3712124824523926,
      "learning_rate": 3.3002349830144594e-06,
      "loss": 0.0519,
      "step": 289200
    },
    {
      "epoch": 4.259510169216949,
      "grad_norm": 1.7577154636383057,
      "learning_rate": 3.2985986110474747e-06,
      "loss": 0.0527,
      "step": 289225
    },
    {
      "epoch": 4.259878352307036,
      "grad_norm": 1.3306009769439697,
      "learning_rate": 3.29696223908049e-06,
      "loss": 0.0524,
      "step": 289250
    },
    {
      "epoch": 4.260246535397123,
      "grad_norm": 1.6227326393127441,
      "learning_rate": 3.2953258671135057e-06,
      "loss": 0.0522,
      "step": 289275
    },
    {
      "epoch": 4.26061471848721,
      "grad_norm": 0.8738011717796326,
      "learning_rate": 3.293689495146521e-06,
      "loss": 0.0472,
      "step": 289300
    },
    {
      "epoch": 4.260982901577297,
      "grad_norm": 1.1673873662948608,
      "learning_rate": 3.2920531231795368e-06,
      "loss": 0.0488,
      "step": 289325
    },
    {
      "epoch": 4.261351084667384,
      "grad_norm": 2.038783073425293,
      "learning_rate": 3.2904167512125517e-06,
      "loss": 0.0544,
      "step": 289350
    },
    {
      "epoch": 4.261719267757471,
      "grad_norm": 1.1564655303955078,
      "learning_rate": 3.2887803792455674e-06,
      "loss": 0.0568,
      "step": 289375
    },
    {
      "epoch": 4.262087450847558,
      "grad_norm": 1.0990837812423706,
      "learning_rate": 3.2871440072785827e-06,
      "loss": 0.05,
      "step": 289400
    },
    {
      "epoch": 4.262455633937645,
      "grad_norm": 0.9731162786483765,
      "learning_rate": 3.2855076353115984e-06,
      "loss": 0.0587,
      "step": 289425
    },
    {
      "epoch": 4.262823817027732,
      "grad_norm": 1.4380793571472168,
      "learning_rate": 3.2838712633446133e-06,
      "loss": 0.0554,
      "step": 289450
    },
    {
      "epoch": 4.263192000117819,
      "grad_norm": 1.3496243953704834,
      "learning_rate": 3.282234891377629e-06,
      "loss": 0.0509,
      "step": 289475
    },
    {
      "epoch": 4.263560183207906,
      "grad_norm": 0.9886029958724976,
      "learning_rate": 3.2805985194106443e-06,
      "loss": 0.0513,
      "step": 289500
    },
    {
      "epoch": 4.263928366297993,
      "grad_norm": 1.2706990242004395,
      "learning_rate": 3.27896214744366e-06,
      "loss": 0.056,
      "step": 289525
    },
    {
      "epoch": 4.26429654938808,
      "grad_norm": 1.5971804857254028,
      "learning_rate": 3.2773257754766754e-06,
      "loss": 0.0473,
      "step": 289550
    },
    {
      "epoch": 4.264664732478167,
      "grad_norm": 1.1203545331954956,
      "learning_rate": 3.275689403509691e-06,
      "loss": 0.0518,
      "step": 289575
    },
    {
      "epoch": 4.265032915568254,
      "grad_norm": 1.5526471138000488,
      "learning_rate": 3.274053031542706e-06,
      "loss": 0.0499,
      "step": 289600
    },
    {
      "epoch": 4.265401098658341,
      "grad_norm": 1.7908813953399658,
      "learning_rate": 3.2724166595757217e-06,
      "loss": 0.0546,
      "step": 289625
    },
    {
      "epoch": 4.265769281748428,
      "grad_norm": 0.9146240949630737,
      "learning_rate": 3.2707802876087374e-06,
      "loss": 0.0497,
      "step": 289650
    },
    {
      "epoch": 4.266137464838515,
      "grad_norm": 1.3962552547454834,
      "learning_rate": 3.2691439156417527e-06,
      "loss": 0.0518,
      "step": 289675
    },
    {
      "epoch": 4.266505647928602,
      "grad_norm": 1.0758328437805176,
      "learning_rate": 3.2675075436747685e-06,
      "loss": 0.0508,
      "step": 289700
    },
    {
      "epoch": 4.266873831018689,
      "grad_norm": 1.6994529962539673,
      "learning_rate": 3.2658711717077833e-06,
      "loss": 0.052,
      "step": 289725
    },
    {
      "epoch": 4.267242014108776,
      "grad_norm": 1.4449158906936646,
      "learning_rate": 3.264234799740799e-06,
      "loss": 0.046,
      "step": 289750
    },
    {
      "epoch": 4.267610197198863,
      "grad_norm": 1.1649595499038696,
      "learning_rate": 3.2625984277738144e-06,
      "loss": 0.05,
      "step": 289775
    },
    {
      "epoch": 4.26797838028895,
      "grad_norm": 1.8054165840148926,
      "learning_rate": 3.26096205580683e-06,
      "loss": 0.0518,
      "step": 289800
    },
    {
      "epoch": 4.268346563379037,
      "grad_norm": 1.7467937469482422,
      "learning_rate": 3.259325683839845e-06,
      "loss": 0.0448,
      "step": 289825
    },
    {
      "epoch": 4.268714746469124,
      "grad_norm": 1.802024006843567,
      "learning_rate": 3.2576893118728607e-06,
      "loss": 0.0512,
      "step": 289850
    },
    {
      "epoch": 4.269082929559211,
      "grad_norm": 1.099867582321167,
      "learning_rate": 3.256052939905876e-06,
      "loss": 0.0514,
      "step": 289875
    },
    {
      "epoch": 4.269451112649298,
      "grad_norm": 1.5718309879302979,
      "learning_rate": 3.2544165679388917e-06,
      "loss": 0.0531,
      "step": 289900
    },
    {
      "epoch": 4.269819295739385,
      "grad_norm": 1.1007776260375977,
      "learning_rate": 3.252780195971907e-06,
      "loss": 0.0473,
      "step": 289925
    },
    {
      "epoch": 4.270187478829472,
      "grad_norm": 1.678215503692627,
      "learning_rate": 3.2511438240049224e-06,
      "loss": 0.0522,
      "step": 289950
    },
    {
      "epoch": 4.270555661919559,
      "grad_norm": 1.1936955451965332,
      "learning_rate": 3.2495074520379377e-06,
      "loss": 0.048,
      "step": 289975
    },
    {
      "epoch": 4.270923845009646,
      "grad_norm": 1.3668286800384521,
      "learning_rate": 3.2478710800709534e-06,
      "loss": 0.0544,
      "step": 290000
    },
    {
      "epoch": 4.271292028099733,
      "grad_norm": 1.1381443738937378,
      "learning_rate": 3.2462347081039687e-06,
      "loss": 0.0463,
      "step": 290025
    },
    {
      "epoch": 4.27166021118982,
      "grad_norm": 1.5934035778045654,
      "learning_rate": 3.2445983361369844e-06,
      "loss": 0.0562,
      "step": 290050
    },
    {
      "epoch": 4.272028394279907,
      "grad_norm": 1.540291666984558,
      "learning_rate": 3.2429619641699993e-06,
      "loss": 0.047,
      "step": 290075
    },
    {
      "epoch": 4.272396577369994,
      "grad_norm": 1.237172245979309,
      "learning_rate": 3.241325592203015e-06,
      "loss": 0.0585,
      "step": 290100
    },
    {
      "epoch": 4.272764760460081,
      "grad_norm": 2.0267882347106934,
      "learning_rate": 3.2396892202360308e-06,
      "loss": 0.051,
      "step": 290125
    },
    {
      "epoch": 4.273132943550169,
      "grad_norm": 1.0281363725662231,
      "learning_rate": 3.238052848269046e-06,
      "loss": 0.0505,
      "step": 290150
    },
    {
      "epoch": 4.273501126640256,
      "grad_norm": 1.1336170434951782,
      "learning_rate": 3.236416476302062e-06,
      "loss": 0.0536,
      "step": 290175
    },
    {
      "epoch": 4.273869309730343,
      "grad_norm": 1.7467669248580933,
      "learning_rate": 3.2347801043350767e-06,
      "loss": 0.0541,
      "step": 290200
    },
    {
      "epoch": 4.27423749282043,
      "grad_norm": 0.6383174061775208,
      "learning_rate": 3.233209187246772e-06,
      "loss": 0.0587,
      "step": 290225
    },
    {
      "epoch": 4.274605675910517,
      "grad_norm": 1.2602483034133911,
      "learning_rate": 3.231572815279787e-06,
      "loss": 0.056,
      "step": 290250
    },
    {
      "epoch": 4.274973859000604,
      "grad_norm": 1.03874933719635,
      "learning_rate": 3.229936443312803e-06,
      "loss": 0.0537,
      "step": 290275
    },
    {
      "epoch": 4.275342042090691,
      "grad_norm": 1.1729580163955688,
      "learning_rate": 3.2283000713458177e-06,
      "loss": 0.0477,
      "step": 290300
    },
    {
      "epoch": 4.275710225180778,
      "grad_norm": 0.864612877368927,
      "learning_rate": 3.2266636993788335e-06,
      "loss": 0.0489,
      "step": 290325
    },
    {
      "epoch": 4.276078408270865,
      "grad_norm": 1.0861895084381104,
      "learning_rate": 3.2250273274118488e-06,
      "loss": 0.0451,
      "step": 290350
    },
    {
      "epoch": 4.276446591360952,
      "grad_norm": 1.7863030433654785,
      "learning_rate": 3.2233909554448645e-06,
      "loss": 0.0501,
      "step": 290375
    },
    {
      "epoch": 4.276814774451039,
      "grad_norm": 1.3945504426956177,
      "learning_rate": 3.22175458347788e-06,
      "loss": 0.0533,
      "step": 290400
    },
    {
      "epoch": 4.277182957541126,
      "grad_norm": 1.0132365226745605,
      "learning_rate": 3.2201182115108955e-06,
      "loss": 0.0545,
      "step": 290425
    },
    {
      "epoch": 4.277551140631213,
      "grad_norm": 1.4781955480575562,
      "learning_rate": 3.2184818395439104e-06,
      "loss": 0.0472,
      "step": 290450
    },
    {
      "epoch": 4.2779193237213,
      "grad_norm": 1.2693023681640625,
      "learning_rate": 3.216845467576926e-06,
      "loss": 0.0581,
      "step": 290475
    },
    {
      "epoch": 4.278287506811387,
      "grad_norm": 1.305580496788025,
      "learning_rate": 3.2152090956099414e-06,
      "loss": 0.0518,
      "step": 290500
    },
    {
      "epoch": 4.278655689901474,
      "grad_norm": 1.3671510219573975,
      "learning_rate": 3.213572723642957e-06,
      "loss": 0.0466,
      "step": 290525
    },
    {
      "epoch": 4.279023872991561,
      "grad_norm": 1.381828784942627,
      "learning_rate": 3.211936351675972e-06,
      "loss": 0.0572,
      "step": 290550
    },
    {
      "epoch": 4.279392056081648,
      "grad_norm": 1.5436588525772095,
      "learning_rate": 3.2102999797089878e-06,
      "loss": 0.0461,
      "step": 290575
    },
    {
      "epoch": 4.279760239171735,
      "grad_norm": 1.1214696168899536,
      "learning_rate": 3.2086636077420035e-06,
      "loss": 0.0508,
      "step": 290600
    },
    {
      "epoch": 4.280128422261822,
      "grad_norm": 1.500473141670227,
      "learning_rate": 3.207027235775019e-06,
      "loss": 0.0493,
      "step": 290625
    },
    {
      "epoch": 4.280496605351909,
      "grad_norm": 1.0295681953430176,
      "learning_rate": 3.2053908638080345e-06,
      "loss": 0.044,
      "step": 290650
    },
    {
      "epoch": 4.280864788441996,
      "grad_norm": 1.176110863685608,
      "learning_rate": 3.2037544918410494e-06,
      "loss": 0.0536,
      "step": 290675
    },
    {
      "epoch": 4.281232971532083,
      "grad_norm": 1.3825201988220215,
      "learning_rate": 3.202118119874065e-06,
      "loss": 0.0532,
      "step": 290700
    },
    {
      "epoch": 4.28160115462217,
      "grad_norm": 1.329816460609436,
      "learning_rate": 3.2004817479070804e-06,
      "loss": 0.051,
      "step": 290725
    },
    {
      "epoch": 4.281969337712257,
      "grad_norm": 1.1210793256759644,
      "learning_rate": 3.198845375940096e-06,
      "loss": 0.054,
      "step": 290750
    },
    {
      "epoch": 4.282337520802344,
      "grad_norm": 0.8068587779998779,
      "learning_rate": 3.1972090039731115e-06,
      "loss": 0.0564,
      "step": 290775
    },
    {
      "epoch": 4.282705703892431,
      "grad_norm": 1.3938854932785034,
      "learning_rate": 3.1955726320061268e-06,
      "loss": 0.0519,
      "step": 290800
    },
    {
      "epoch": 4.2830738869825185,
      "grad_norm": 1.214111328125,
      "learning_rate": 3.193936260039142e-06,
      "loss": 0.0508,
      "step": 290825
    },
    {
      "epoch": 4.2834420700726055,
      "grad_norm": 1.3451392650604248,
      "learning_rate": 3.192299888072158e-06,
      "loss": 0.0543,
      "step": 290850
    },
    {
      "epoch": 4.2838102531626925,
      "grad_norm": 1.3039880990982056,
      "learning_rate": 3.190663516105173e-06,
      "loss": 0.0494,
      "step": 290875
    },
    {
      "epoch": 4.2841784362527795,
      "grad_norm": 1.424365520477295,
      "learning_rate": 3.189027144138189e-06,
      "loss": 0.0526,
      "step": 290900
    },
    {
      "epoch": 4.2845466193428665,
      "grad_norm": 1.1638636589050293,
      "learning_rate": 3.1873907721712037e-06,
      "loss": 0.0499,
      "step": 290925
    },
    {
      "epoch": 4.2849148024329535,
      "grad_norm": 1.5044561624526978,
      "learning_rate": 3.1857544002042195e-06,
      "loss": 0.0549,
      "step": 290950
    },
    {
      "epoch": 4.2852829855230405,
      "grad_norm": 1.4784969091415405,
      "learning_rate": 3.1841180282372348e-06,
      "loss": 0.0525,
      "step": 290975
    },
    {
      "epoch": 4.285651168613128,
      "grad_norm": 1.465497612953186,
      "learning_rate": 3.1824816562702505e-06,
      "loss": 0.0513,
      "step": 291000
    },
    {
      "epoch": 4.286019351703215,
      "grad_norm": 1.3716065883636475,
      "learning_rate": 3.1808452843032654e-06,
      "loss": 0.0503,
      "step": 291025
    },
    {
      "epoch": 4.2863875347933025,
      "grad_norm": 1.2589154243469238,
      "learning_rate": 3.179208912336281e-06,
      "loss": 0.0562,
      "step": 291050
    },
    {
      "epoch": 4.2867557178833895,
      "grad_norm": 1.7774900197982788,
      "learning_rate": 3.1775725403692964e-06,
      "loss": 0.0551,
      "step": 291075
    },
    {
      "epoch": 4.2871239009734765,
      "grad_norm": 1.2855439186096191,
      "learning_rate": 3.175936168402312e-06,
      "loss": 0.0465,
      "step": 291100
    },
    {
      "epoch": 4.2874920840635635,
      "grad_norm": 1.6866042613983154,
      "learning_rate": 3.174299796435328e-06,
      "loss": 0.0497,
      "step": 291125
    },
    {
      "epoch": 4.2878602671536505,
      "grad_norm": 1.8169424533843994,
      "learning_rate": 3.172663424468343e-06,
      "loss": 0.0517,
      "step": 291150
    },
    {
      "epoch": 4.2882284502437376,
      "grad_norm": 1.103255033493042,
      "learning_rate": 3.1710270525013585e-06,
      "loss": 0.0538,
      "step": 291175
    },
    {
      "epoch": 4.288596633333825,
      "grad_norm": 1.1007641553878784,
      "learning_rate": 3.1693906805343738e-06,
      "loss": 0.0483,
      "step": 291200
    },
    {
      "epoch": 4.288964816423912,
      "grad_norm": 1.074811339378357,
      "learning_rate": 3.1677543085673895e-06,
      "loss": 0.0486,
      "step": 291225
    },
    {
      "epoch": 4.289332999513999,
      "grad_norm": 1.32887864112854,
      "learning_rate": 3.166117936600405e-06,
      "loss": 0.0455,
      "step": 291250
    },
    {
      "epoch": 4.289701182604086,
      "grad_norm": 0.8534607291221619,
      "learning_rate": 3.1644815646334205e-06,
      "loss": 0.0539,
      "step": 291275
    },
    {
      "epoch": 4.290069365694173,
      "grad_norm": 1.475752353668213,
      "learning_rate": 3.1628451926664354e-06,
      "loss": 0.0547,
      "step": 291300
    },
    {
      "epoch": 4.29043754878426,
      "grad_norm": 0.9513261318206787,
      "learning_rate": 3.161208820699451e-06,
      "loss": 0.05,
      "step": 291325
    },
    {
      "epoch": 4.290805731874347,
      "grad_norm": 1.2565200328826904,
      "learning_rate": 3.1595724487324664e-06,
      "loss": 0.0515,
      "step": 291350
    },
    {
      "epoch": 4.291173914964434,
      "grad_norm": 1.3913260698318481,
      "learning_rate": 3.157936076765482e-06,
      "loss": 0.0509,
      "step": 291375
    },
    {
      "epoch": 4.291542098054521,
      "grad_norm": 0.9858401417732239,
      "learning_rate": 3.156299704798497e-06,
      "loss": 0.0537,
      "step": 291400
    },
    {
      "epoch": 4.291910281144608,
      "grad_norm": 1.2199232578277588,
      "learning_rate": 3.1546633328315128e-06,
      "loss": 0.0462,
      "step": 291425
    },
    {
      "epoch": 4.292278464234695,
      "grad_norm": 0.9362841248512268,
      "learning_rate": 3.153026960864528e-06,
      "loss": 0.0518,
      "step": 291450
    },
    {
      "epoch": 4.292646647324782,
      "grad_norm": 1.0767419338226318,
      "learning_rate": 3.151390588897544e-06,
      "loss": 0.0493,
      "step": 291475
    },
    {
      "epoch": 4.293014830414869,
      "grad_norm": 1.5425621271133423,
      "learning_rate": 3.149754216930559e-06,
      "loss": 0.052,
      "step": 291500
    },
    {
      "epoch": 4.293383013504956,
      "grad_norm": 1.0434218645095825,
      "learning_rate": 3.148117844963575e-06,
      "loss": 0.0555,
      "step": 291525
    },
    {
      "epoch": 4.293751196595043,
      "grad_norm": 1.1621745824813843,
      "learning_rate": 3.1464814729965897e-06,
      "loss": 0.048,
      "step": 291550
    },
    {
      "epoch": 4.29411937968513,
      "grad_norm": 0.9913632273674011,
      "learning_rate": 3.1448451010296055e-06,
      "loss": 0.0553,
      "step": 291575
    },
    {
      "epoch": 4.294487562775217,
      "grad_norm": 1.3171082735061646,
      "learning_rate": 3.143208729062621e-06,
      "loss": 0.0506,
      "step": 291600
    },
    {
      "epoch": 4.294855745865304,
      "grad_norm": 1.4461736679077148,
      "learning_rate": 3.1415723570956365e-06,
      "loss": 0.0535,
      "step": 291625
    },
    {
      "epoch": 4.295223928955391,
      "grad_norm": 1.1955137252807617,
      "learning_rate": 3.1399359851286522e-06,
      "loss": 0.0508,
      "step": 291650
    },
    {
      "epoch": 4.295592112045478,
      "grad_norm": 1.2526497840881348,
      "learning_rate": 3.138299613161667e-06,
      "loss": 0.0507,
      "step": 291675
    },
    {
      "epoch": 4.295960295135565,
      "grad_norm": 1.2451164722442627,
      "learning_rate": 3.136663241194683e-06,
      "loss": 0.0563,
      "step": 291700
    },
    {
      "epoch": 4.296328478225652,
      "grad_norm": 1.0359917879104614,
      "learning_rate": 3.135026869227698e-06,
      "loss": 0.0457,
      "step": 291725
    },
    {
      "epoch": 4.296696661315739,
      "grad_norm": 1.4772802591323853,
      "learning_rate": 3.133390497260714e-06,
      "loss": 0.0493,
      "step": 291750
    },
    {
      "epoch": 4.297064844405826,
      "grad_norm": 1.2262344360351562,
      "learning_rate": 3.1317541252937287e-06,
      "loss": 0.0498,
      "step": 291775
    },
    {
      "epoch": 4.297433027495913,
      "grad_norm": 0.5714721083641052,
      "learning_rate": 3.1301177533267445e-06,
      "loss": 0.0545,
      "step": 291800
    },
    {
      "epoch": 4.297801210586,
      "grad_norm": 1.4304720163345337,
      "learning_rate": 3.1284813813597598e-06,
      "loss": 0.0518,
      "step": 291825
    },
    {
      "epoch": 4.298169393676087,
      "grad_norm": 0.8661518096923828,
      "learning_rate": 3.1268450093927755e-06,
      "loss": 0.051,
      "step": 291850
    },
    {
      "epoch": 4.298537576766174,
      "grad_norm": 1.01231050491333,
      "learning_rate": 3.125208637425791e-06,
      "loss": 0.0448,
      "step": 291875
    },
    {
      "epoch": 4.298905759856261,
      "grad_norm": 1.9943183660507202,
      "learning_rate": 3.123572265458806e-06,
      "loss": 0.064,
      "step": 291900
    },
    {
      "epoch": 4.299273942946348,
      "grad_norm": 1.2635256052017212,
      "learning_rate": 3.1219358934918214e-06,
      "loss": 0.052,
      "step": 291925
    },
    {
      "epoch": 4.299642126036435,
      "grad_norm": 1.0197324752807617,
      "learning_rate": 3.120299521524837e-06,
      "loss": 0.0449,
      "step": 291950
    },
    {
      "epoch": 4.300010309126522,
      "grad_norm": 1.2457128763198853,
      "learning_rate": 3.1186631495578524e-06,
      "loss": 0.0415,
      "step": 291975
    },
    {
      "epoch": 4.300378492216609,
      "grad_norm": 1.4862420558929443,
      "learning_rate": 3.117026777590868e-06,
      "loss": 0.054,
      "step": 292000
    },
    {
      "epoch": 4.300746675306696,
      "grad_norm": 1.4623264074325562,
      "learning_rate": 3.115390405623883e-06,
      "loss": 0.0482,
      "step": 292025
    },
    {
      "epoch": 4.301114858396783,
      "grad_norm": 1.565063714981079,
      "learning_rate": 3.1137540336568988e-06,
      "loss": 0.0478,
      "step": 292050
    },
    {
      "epoch": 4.301483041486871,
      "grad_norm": 1.4657037258148193,
      "learning_rate": 3.1121176616899145e-06,
      "loss": 0.0474,
      "step": 292075
    },
    {
      "epoch": 4.301851224576958,
      "grad_norm": 1.2396297454833984,
      "learning_rate": 3.11048128972293e-06,
      "loss": 0.0496,
      "step": 292100
    },
    {
      "epoch": 4.302219407667045,
      "grad_norm": 1.2692071199417114,
      "learning_rate": 3.1088449177559455e-06,
      "loss": 0.0449,
      "step": 292125
    },
    {
      "epoch": 4.302587590757132,
      "grad_norm": 1.3308403491973877,
      "learning_rate": 3.1072085457889604e-06,
      "loss": 0.0601,
      "step": 292150
    },
    {
      "epoch": 4.302955773847219,
      "grad_norm": 1.3547325134277344,
      "learning_rate": 3.105572173821976e-06,
      "loss": 0.0604,
      "step": 292175
    },
    {
      "epoch": 4.303323956937306,
      "grad_norm": 1.7643423080444336,
      "learning_rate": 3.1039358018549915e-06,
      "loss": 0.0498,
      "step": 292200
    },
    {
      "epoch": 4.303692140027393,
      "grad_norm": 1.1142398118972778,
      "learning_rate": 3.102299429888007e-06,
      "loss": 0.0549,
      "step": 292225
    },
    {
      "epoch": 4.30406032311748,
      "grad_norm": 1.2440831661224365,
      "learning_rate": 3.1006630579210225e-06,
      "loss": 0.0506,
      "step": 292250
    },
    {
      "epoch": 4.304428506207567,
      "grad_norm": 1.3940349817276,
      "learning_rate": 3.099026685954038e-06,
      "loss": 0.0546,
      "step": 292275
    },
    {
      "epoch": 4.304796689297654,
      "grad_norm": 1.4349170923233032,
      "learning_rate": 3.097390313987053e-06,
      "loss": 0.0523,
      "step": 292300
    },
    {
      "epoch": 4.305164872387741,
      "grad_norm": 1.044699788093567,
      "learning_rate": 3.095753942020069e-06,
      "loss": 0.0517,
      "step": 292325
    },
    {
      "epoch": 4.305533055477828,
      "grad_norm": 1.0127317905426025,
      "learning_rate": 3.094117570053084e-06,
      "loss": 0.055,
      "step": 292350
    },
    {
      "epoch": 4.305901238567915,
      "grad_norm": 1.1746273040771484,
      "learning_rate": 3.0924811980861e-06,
      "loss": 0.0516,
      "step": 292375
    },
    {
      "epoch": 4.306269421658002,
      "grad_norm": 1.809509038925171,
      "learning_rate": 3.0908448261191147e-06,
      "loss": 0.0524,
      "step": 292400
    },
    {
      "epoch": 4.306637604748089,
      "grad_norm": 1.101275086402893,
      "learning_rate": 3.0892084541521305e-06,
      "loss": 0.0476,
      "step": 292425
    },
    {
      "epoch": 4.307005787838176,
      "grad_norm": 1.2939157485961914,
      "learning_rate": 3.0875720821851458e-06,
      "loss": 0.0553,
      "step": 292450
    },
    {
      "epoch": 4.307373970928263,
      "grad_norm": 0.8187004327774048,
      "learning_rate": 3.0859357102181615e-06,
      "loss": 0.0508,
      "step": 292475
    },
    {
      "epoch": 4.30774215401835,
      "grad_norm": 1.361149787902832,
      "learning_rate": 3.0842993382511764e-06,
      "loss": 0.0509,
      "step": 292500
    },
    {
      "epoch": 4.308110337108437,
      "grad_norm": 1.3855130672454834,
      "learning_rate": 3.082662966284192e-06,
      "loss": 0.0563,
      "step": 292525
    },
    {
      "epoch": 4.308478520198524,
      "grad_norm": 1.2144083976745605,
      "learning_rate": 3.0810265943172074e-06,
      "loss": 0.0488,
      "step": 292550
    },
    {
      "epoch": 4.308846703288611,
      "grad_norm": 1.3592517375946045,
      "learning_rate": 3.079390222350223e-06,
      "loss": 0.0549,
      "step": 292575
    },
    {
      "epoch": 4.309214886378698,
      "grad_norm": 1.1568927764892578,
      "learning_rate": 3.077753850383239e-06,
      "loss": 0.0546,
      "step": 292600
    },
    {
      "epoch": 4.309583069468785,
      "grad_norm": 0.8282021880149841,
      "learning_rate": 3.076117478416254e-06,
      "loss": 0.0491,
      "step": 292625
    },
    {
      "epoch": 4.309951252558872,
      "grad_norm": 1.4004698991775513,
      "learning_rate": 3.0744811064492695e-06,
      "loss": 0.0565,
      "step": 292650
    },
    {
      "epoch": 4.310319435648959,
      "grad_norm": 1.5502421855926514,
      "learning_rate": 3.0728447344822848e-06,
      "loss": 0.0481,
      "step": 292675
    },
    {
      "epoch": 4.310687618739046,
      "grad_norm": 1.4700853824615479,
      "learning_rate": 3.0712083625153005e-06,
      "loss": 0.0453,
      "step": 292700
    },
    {
      "epoch": 4.311055801829133,
      "grad_norm": 1.9841800928115845,
      "learning_rate": 3.069571990548316e-06,
      "loss": 0.0528,
      "step": 292725
    },
    {
      "epoch": 4.31142398491922,
      "grad_norm": 1.8632556200027466,
      "learning_rate": 3.0679356185813315e-06,
      "loss": 0.0571,
      "step": 292750
    },
    {
      "epoch": 4.311792168009307,
      "grad_norm": 1.39838707447052,
      "learning_rate": 3.0662992466143464e-06,
      "loss": 0.0509,
      "step": 292775
    },
    {
      "epoch": 4.312160351099394,
      "grad_norm": 0.8109171986579895,
      "learning_rate": 3.064662874647362e-06,
      "loss": 0.0543,
      "step": 292800
    },
    {
      "epoch": 4.3125285341894815,
      "grad_norm": 1.4002578258514404,
      "learning_rate": 3.0630265026803775e-06,
      "loss": 0.0584,
      "step": 292825
    },
    {
      "epoch": 4.3128967172795685,
      "grad_norm": 1.2867463827133179,
      "learning_rate": 3.061390130713393e-06,
      "loss": 0.0468,
      "step": 292850
    },
    {
      "epoch": 4.3132649003696555,
      "grad_norm": 1.6728713512420654,
      "learning_rate": 3.059753758746408e-06,
      "loss": 0.0572,
      "step": 292875
    },
    {
      "epoch": 4.3136330834597425,
      "grad_norm": 1.7604358196258545,
      "learning_rate": 3.058117386779424e-06,
      "loss": 0.0528,
      "step": 292900
    },
    {
      "epoch": 4.3140012665498295,
      "grad_norm": 1.296737790107727,
      "learning_rate": 3.056481014812439e-06,
      "loss": 0.0534,
      "step": 292925
    },
    {
      "epoch": 4.3143694496399165,
      "grad_norm": 1.0273759365081787,
      "learning_rate": 3.054844642845455e-06,
      "loss": 0.0469,
      "step": 292950
    },
    {
      "epoch": 4.314737632730004,
      "grad_norm": 1.5726728439331055,
      "learning_rate": 3.05320827087847e-06,
      "loss": 0.057,
      "step": 292975
    },
    {
      "epoch": 4.3151058158200915,
      "grad_norm": 1.0067387819290161,
      "learning_rate": 3.0515718989114854e-06,
      "loss": 0.0513,
      "step": 293000
    },
    {
      "epoch": 4.3154739989101785,
      "grad_norm": 1.455614447593689,
      "learning_rate": 3.0499355269445007e-06,
      "loss": 0.0522,
      "step": 293025
    },
    {
      "epoch": 4.3158421820002655,
      "grad_norm": 1.105013370513916,
      "learning_rate": 3.0482991549775165e-06,
      "loss": 0.0487,
      "step": 293050
    },
    {
      "epoch": 4.3162103650903525,
      "grad_norm": 1.3728548288345337,
      "learning_rate": 3.046662783010532e-06,
      "loss": 0.0498,
      "step": 293075
    },
    {
      "epoch": 4.3165785481804395,
      "grad_norm": 1.3198384046554565,
      "learning_rate": 3.0450264110435475e-06,
      "loss": 0.0517,
      "step": 293100
    },
    {
      "epoch": 4.3169467312705265,
      "grad_norm": 1.5928163528442383,
      "learning_rate": 3.0433900390765632e-06,
      "loss": 0.0512,
      "step": 293125
    },
    {
      "epoch": 4.3173149143606135,
      "grad_norm": 1.511946439743042,
      "learning_rate": 3.041753667109578e-06,
      "loss": 0.0521,
      "step": 293150
    },
    {
      "epoch": 4.317683097450701,
      "grad_norm": 1.5115729570388794,
      "learning_rate": 3.040117295142594e-06,
      "loss": 0.055,
      "step": 293175
    },
    {
      "epoch": 4.318051280540788,
      "grad_norm": 1.5589832067489624,
      "learning_rate": 3.038480923175609e-06,
      "loss": 0.0489,
      "step": 293200
    },
    {
      "epoch": 4.318419463630875,
      "grad_norm": 1.6328973770141602,
      "learning_rate": 3.036844551208625e-06,
      "loss": 0.0504,
      "step": 293225
    },
    {
      "epoch": 4.318787646720962,
      "grad_norm": 1.2181569337844849,
      "learning_rate": 3.0352081792416398e-06,
      "loss": 0.0507,
      "step": 293250
    },
    {
      "epoch": 4.319155829811049,
      "grad_norm": 1.2631444931030273,
      "learning_rate": 3.0335718072746555e-06,
      "loss": 0.0496,
      "step": 293275
    },
    {
      "epoch": 4.319524012901136,
      "grad_norm": 1.3769088983535767,
      "learning_rate": 3.0319354353076708e-06,
      "loss": 0.0548,
      "step": 293300
    },
    {
      "epoch": 4.319892195991223,
      "grad_norm": 1.1551789045333862,
      "learning_rate": 3.0302990633406865e-06,
      "loss": 0.0478,
      "step": 293325
    },
    {
      "epoch": 4.32026037908131,
      "grad_norm": 1.0351506471633911,
      "learning_rate": 3.028662691373702e-06,
      "loss": 0.0451,
      "step": 293350
    },
    {
      "epoch": 4.320628562171397,
      "grad_norm": 1.35616934299469,
      "learning_rate": 3.027026319406717e-06,
      "loss": 0.0458,
      "step": 293375
    },
    {
      "epoch": 4.320996745261484,
      "grad_norm": 1.03713858127594,
      "learning_rate": 3.0253899474397324e-06,
      "loss": 0.0511,
      "step": 293400
    },
    {
      "epoch": 4.321364928351571,
      "grad_norm": 1.6304552555084229,
      "learning_rate": 3.023753575472748e-06,
      "loss": 0.0605,
      "step": 293425
    },
    {
      "epoch": 4.321733111441658,
      "grad_norm": 0.953269362449646,
      "learning_rate": 3.0221172035057635e-06,
      "loss": 0.0505,
      "step": 293450
    },
    {
      "epoch": 4.322101294531745,
      "grad_norm": 1.4370427131652832,
      "learning_rate": 3.020480831538779e-06,
      "loss": 0.0475,
      "step": 293475
    },
    {
      "epoch": 4.322469477621832,
      "grad_norm": 1.8312782049179077,
      "learning_rate": 3.018844459571794e-06,
      "loss": 0.0509,
      "step": 293500
    },
    {
      "epoch": 4.322837660711919,
      "grad_norm": 1.1360297203063965,
      "learning_rate": 3.01720808760481e-06,
      "loss": 0.049,
      "step": 293525
    },
    {
      "epoch": 4.323205843802006,
      "grad_norm": 1.1346262693405151,
      "learning_rate": 3.0155717156378255e-06,
      "loss": 0.0483,
      "step": 293550
    },
    {
      "epoch": 4.323574026892093,
      "grad_norm": 0.7781646251678467,
      "learning_rate": 3.013935343670841e-06,
      "loss": 0.0488,
      "step": 293575
    },
    {
      "epoch": 4.32394220998218,
      "grad_norm": 1.7196122407913208,
      "learning_rate": 3.0122989717038566e-06,
      "loss": 0.0469,
      "step": 293600
    },
    {
      "epoch": 4.324310393072267,
      "grad_norm": 1.5313682556152344,
      "learning_rate": 3.0106625997368714e-06,
      "loss": 0.0511,
      "step": 293625
    },
    {
      "epoch": 4.324678576162354,
      "grad_norm": 1.2286125421524048,
      "learning_rate": 3.009026227769887e-06,
      "loss": 0.0537,
      "step": 293650
    },
    {
      "epoch": 4.325046759252441,
      "grad_norm": 1.605811357498169,
      "learning_rate": 3.0073898558029025e-06,
      "loss": 0.0443,
      "step": 293675
    },
    {
      "epoch": 4.325414942342528,
      "grad_norm": 1.1447017192840576,
      "learning_rate": 3.005753483835918e-06,
      "loss": 0.0576,
      "step": 293700
    },
    {
      "epoch": 4.325783125432615,
      "grad_norm": 1.6054688692092896,
      "learning_rate": 3.0041171118689335e-06,
      "loss": 0.0501,
      "step": 293725
    },
    {
      "epoch": 4.326151308522702,
      "grad_norm": 1.2128782272338867,
      "learning_rate": 3.002480739901949e-06,
      "loss": 0.0519,
      "step": 293750
    },
    {
      "epoch": 4.326519491612789,
      "grad_norm": 0.9511484503746033,
      "learning_rate": 3.000844367934964e-06,
      "loss": 0.0536,
      "step": 293775
    },
    {
      "epoch": 4.326887674702876,
      "grad_norm": 1.5446316003799438,
      "learning_rate": 2.99920799596798e-06,
      "loss": 0.0502,
      "step": 293800
    },
    {
      "epoch": 4.327255857792963,
      "grad_norm": 1.140790581703186,
      "learning_rate": 2.997571624000995e-06,
      "loss": 0.049,
      "step": 293825
    },
    {
      "epoch": 4.32762404088305,
      "grad_norm": 1.5022300481796265,
      "learning_rate": 2.995935252034011e-06,
      "loss": 0.0524,
      "step": 293850
    },
    {
      "epoch": 4.327992223973137,
      "grad_norm": 1.4234492778778076,
      "learning_rate": 2.9942988800670257e-06,
      "loss": 0.0564,
      "step": 293875
    },
    {
      "epoch": 4.328360407063224,
      "grad_norm": 1.4587119817733765,
      "learning_rate": 2.9926625081000415e-06,
      "loss": 0.0522,
      "step": 293900
    },
    {
      "epoch": 4.328728590153311,
      "grad_norm": 1.490440845489502,
      "learning_rate": 2.991091591011736e-06,
      "loss": 0.0499,
      "step": 293925
    },
    {
      "epoch": 4.329096773243398,
      "grad_norm": 1.5111546516418457,
      "learning_rate": 2.989455219044752e-06,
      "loss": 0.0504,
      "step": 293950
    },
    {
      "epoch": 4.329464956333485,
      "grad_norm": 1.2817327976226807,
      "learning_rate": 2.987818847077767e-06,
      "loss": 0.0525,
      "step": 293975
    },
    {
      "epoch": 4.329833139423573,
      "grad_norm": 1.8078248500823975,
      "learning_rate": 2.9861824751107825e-06,
      "loss": 0.0565,
      "step": 294000
    },
    {
      "epoch": 4.33020132251366,
      "grad_norm": 1.2433573007583618,
      "learning_rate": 2.9845461031437983e-06,
      "loss": 0.0575,
      "step": 294025
    },
    {
      "epoch": 4.330569505603747,
      "grad_norm": 1.7191497087478638,
      "learning_rate": 2.9829097311768136e-06,
      "loss": 0.057,
      "step": 294050
    },
    {
      "epoch": 4.330937688693834,
      "grad_norm": 1.2274821996688843,
      "learning_rate": 2.9812733592098293e-06,
      "loss": 0.042,
      "step": 294075
    },
    {
      "epoch": 4.331305871783921,
      "grad_norm": 1.2338813543319702,
      "learning_rate": 2.979636987242844e-06,
      "loss": 0.0453,
      "step": 294100
    },
    {
      "epoch": 4.331674054874008,
      "grad_norm": 1.0689059495925903,
      "learning_rate": 2.97800061527586e-06,
      "loss": 0.0563,
      "step": 294125
    },
    {
      "epoch": 4.332042237964095,
      "grad_norm": 1.1198010444641113,
      "learning_rate": 2.9763642433088752e-06,
      "loss": 0.0507,
      "step": 294150
    },
    {
      "epoch": 4.332410421054182,
      "grad_norm": 0.9678905010223389,
      "learning_rate": 2.974727871341891e-06,
      "loss": 0.0494,
      "step": 294175
    },
    {
      "epoch": 4.332778604144269,
      "grad_norm": 1.0130997896194458,
      "learning_rate": 2.9730914993749062e-06,
      "loss": 0.0457,
      "step": 294200
    },
    {
      "epoch": 4.333146787234356,
      "grad_norm": 1.324289083480835,
      "learning_rate": 2.9714551274079216e-06,
      "loss": 0.0534,
      "step": 294225
    },
    {
      "epoch": 4.333514970324443,
      "grad_norm": 1.6521437168121338,
      "learning_rate": 2.969818755440937e-06,
      "loss": 0.0514,
      "step": 294250
    },
    {
      "epoch": 4.33388315341453,
      "grad_norm": 1.2321314811706543,
      "learning_rate": 2.9681823834739526e-06,
      "loss": 0.0511,
      "step": 294275
    },
    {
      "epoch": 4.334251336504617,
      "grad_norm": 0.857231616973877,
      "learning_rate": 2.966546011506968e-06,
      "loss": 0.0479,
      "step": 294300
    },
    {
      "epoch": 4.334619519594704,
      "grad_norm": 1.4737480878829956,
      "learning_rate": 2.9649096395399836e-06,
      "loss": 0.0473,
      "step": 294325
    },
    {
      "epoch": 4.334987702684791,
      "grad_norm": 1.3005889654159546,
      "learning_rate": 2.9632732675729985e-06,
      "loss": 0.0481,
      "step": 294350
    },
    {
      "epoch": 4.335355885774878,
      "grad_norm": 1.1469502449035645,
      "learning_rate": 2.9616368956060142e-06,
      "loss": 0.052,
      "step": 294375
    },
    {
      "epoch": 4.335724068864965,
      "grad_norm": 1.7334963083267212,
      "learning_rate": 2.9600005236390295e-06,
      "loss": 0.0558,
      "step": 294400
    },
    {
      "epoch": 4.336092251955052,
      "grad_norm": 1.4991530179977417,
      "learning_rate": 2.9583641516720453e-06,
      "loss": 0.0485,
      "step": 294425
    },
    {
      "epoch": 4.336460435045139,
      "grad_norm": 0.9723924398422241,
      "learning_rate": 2.95672777970506e-06,
      "loss": 0.0521,
      "step": 294450
    },
    {
      "epoch": 4.336828618135226,
      "grad_norm": 1.8658486604690552,
      "learning_rate": 2.955091407738076e-06,
      "loss": 0.0581,
      "step": 294475
    },
    {
      "epoch": 4.337196801225313,
      "grad_norm": 1.2224810123443604,
      "learning_rate": 2.9534550357710916e-06,
      "loss": 0.0535,
      "step": 294500
    },
    {
      "epoch": 4.3375649843154,
      "grad_norm": 1.3490447998046875,
      "learning_rate": 2.951818663804107e-06,
      "loss": 0.056,
      "step": 294525
    },
    {
      "epoch": 4.337933167405487,
      "grad_norm": 1.211398720741272,
      "learning_rate": 2.9501822918371226e-06,
      "loss": 0.0463,
      "step": 294550
    },
    {
      "epoch": 4.338301350495574,
      "grad_norm": 1.1779062747955322,
      "learning_rate": 2.948545919870138e-06,
      "loss": 0.0514,
      "step": 294575
    },
    {
      "epoch": 4.338669533585661,
      "grad_norm": 1.1999893188476562,
      "learning_rate": 2.9469095479031532e-06,
      "loss": 0.0471,
      "step": 294600
    },
    {
      "epoch": 4.339037716675748,
      "grad_norm": 1.1773226261138916,
      "learning_rate": 2.9452731759361685e-06,
      "loss": 0.0483,
      "step": 294625
    },
    {
      "epoch": 4.339405899765835,
      "grad_norm": 1.5120632648468018,
      "learning_rate": 2.9436368039691843e-06,
      "loss": 0.0504,
      "step": 294650
    },
    {
      "epoch": 4.339774082855922,
      "grad_norm": 0.8460533618927002,
      "learning_rate": 2.9420004320021996e-06,
      "loss": 0.0518,
      "step": 294675
    },
    {
      "epoch": 4.340142265946009,
      "grad_norm": 0.9781943559646606,
      "learning_rate": 2.9403640600352153e-06,
      "loss": 0.0478,
      "step": 294700
    },
    {
      "epoch": 4.340510449036096,
      "grad_norm": 0.6635684967041016,
      "learning_rate": 2.93872768806823e-06,
      "loss": 0.0525,
      "step": 294725
    },
    {
      "epoch": 4.340878632126183,
      "grad_norm": 1.8739856481552124,
      "learning_rate": 2.937091316101246e-06,
      "loss": 0.052,
      "step": 294750
    },
    {
      "epoch": 4.34124681521627,
      "grad_norm": 1.3742173910140991,
      "learning_rate": 2.9354549441342612e-06,
      "loss": 0.0461,
      "step": 294775
    },
    {
      "epoch": 4.341614998306357,
      "grad_norm": 1.1041861772537231,
      "learning_rate": 2.933818572167277e-06,
      "loss": 0.0459,
      "step": 294800
    },
    {
      "epoch": 4.3419831813964445,
      "grad_norm": 1.3814668655395508,
      "learning_rate": 2.932182200200292e-06,
      "loss": 0.0498,
      "step": 294825
    },
    {
      "epoch": 4.3423513644865315,
      "grad_norm": 1.1698280572891235,
      "learning_rate": 2.9305458282333075e-06,
      "loss": 0.0521,
      "step": 294850
    },
    {
      "epoch": 4.3427195475766185,
      "grad_norm": 1.7867683172225952,
      "learning_rate": 2.928909456266323e-06,
      "loss": 0.0484,
      "step": 294875
    },
    {
      "epoch": 4.343087730666706,
      "grad_norm": 1.2471885681152344,
      "learning_rate": 2.9272730842993386e-06,
      "loss": 0.0532,
      "step": 294900
    },
    {
      "epoch": 4.343455913756793,
      "grad_norm": 1.2834385633468628,
      "learning_rate": 2.925636712332354e-06,
      "loss": 0.0535,
      "step": 294925
    },
    {
      "epoch": 4.34382409684688,
      "grad_norm": 1.1170151233673096,
      "learning_rate": 2.924000340365369e-06,
      "loss": 0.0503,
      "step": 294950
    },
    {
      "epoch": 4.344192279936967,
      "grad_norm": 1.467023491859436,
      "learning_rate": 2.9223639683983845e-06,
      "loss": 0.0597,
      "step": 294975
    },
    {
      "epoch": 4.3445604630270545,
      "grad_norm": 0.9308691620826721,
      "learning_rate": 2.9207275964314002e-06,
      "loss": 0.0544,
      "step": 295000
    },
    {
      "epoch": 4.3449286461171415,
      "grad_norm": 1.2788670063018799,
      "learning_rate": 2.919091224464416e-06,
      "loss": 0.0556,
      "step": 295025
    },
    {
      "epoch": 4.3452968292072285,
      "grad_norm": 1.2762491703033447,
      "learning_rate": 2.9174548524974313e-06,
      "loss": 0.051,
      "step": 295050
    },
    {
      "epoch": 4.3456650122973155,
      "grad_norm": 1.1978625059127808,
      "learning_rate": 2.915818480530447e-06,
      "loss": 0.0515,
      "step": 295075
    },
    {
      "epoch": 4.3460331953874025,
      "grad_norm": 0.8507020473480225,
      "learning_rate": 2.914182108563462e-06,
      "loss": 0.0526,
      "step": 295100
    },
    {
      "epoch": 4.3464013784774895,
      "grad_norm": 1.7318373918533325,
      "learning_rate": 2.9125457365964776e-06,
      "loss": 0.0464,
      "step": 295125
    },
    {
      "epoch": 4.3467695615675765,
      "grad_norm": 1.1228421926498413,
      "learning_rate": 2.910909364629493e-06,
      "loss": 0.0501,
      "step": 295150
    },
    {
      "epoch": 4.347137744657664,
      "grad_norm": 2.1247904300689697,
      "learning_rate": 2.9092729926625086e-06,
      "loss": 0.0619,
      "step": 295175
    },
    {
      "epoch": 4.347505927747751,
      "grad_norm": 1.4659314155578613,
      "learning_rate": 2.9076366206955235e-06,
      "loss": 0.0502,
      "step": 295200
    },
    {
      "epoch": 4.347874110837838,
      "grad_norm": 1.657680630683899,
      "learning_rate": 2.9060002487285392e-06,
      "loss": 0.0481,
      "step": 295225
    },
    {
      "epoch": 4.348242293927925,
      "grad_norm": 1.4498772621154785,
      "learning_rate": 2.9043638767615545e-06,
      "loss": 0.0475,
      "step": 295250
    },
    {
      "epoch": 4.348610477018012,
      "grad_norm": 1.4223194122314453,
      "learning_rate": 2.9027275047945703e-06,
      "loss": 0.0523,
      "step": 295275
    },
    {
      "epoch": 4.348978660108099,
      "grad_norm": 1.369748830795288,
      "learning_rate": 2.9010911328275856e-06,
      "loss": 0.0518,
      "step": 295300
    },
    {
      "epoch": 4.349346843198186,
      "grad_norm": 0.855309247970581,
      "learning_rate": 2.899454760860601e-06,
      "loss": 0.0518,
      "step": 295325
    },
    {
      "epoch": 4.349715026288273,
      "grad_norm": 1.7137930393218994,
      "learning_rate": 2.897818388893616e-06,
      "loss": 0.0505,
      "step": 295350
    },
    {
      "epoch": 4.35008320937836,
      "grad_norm": 1.4124717712402344,
      "learning_rate": 2.896182016926632e-06,
      "loss": 0.0566,
      "step": 295375
    },
    {
      "epoch": 4.350451392468447,
      "grad_norm": 1.562554121017456,
      "learning_rate": 2.894545644959647e-06,
      "loss": 0.0497,
      "step": 295400
    },
    {
      "epoch": 4.350819575558534,
      "grad_norm": 1.7596133947372437,
      "learning_rate": 2.892909272992663e-06,
      "loss": 0.0506,
      "step": 295425
    },
    {
      "epoch": 4.351187758648621,
      "grad_norm": 1.0120108127593994,
      "learning_rate": 2.891272901025678e-06,
      "loss": 0.0533,
      "step": 295450
    },
    {
      "epoch": 4.351555941738708,
      "grad_norm": 1.1392792463302612,
      "learning_rate": 2.8896365290586935e-06,
      "loss": 0.0519,
      "step": 295475
    },
    {
      "epoch": 4.351924124828795,
      "grad_norm": 1.0277249813079834,
      "learning_rate": 2.8880001570917093e-06,
      "loss": 0.0444,
      "step": 295500
    },
    {
      "epoch": 4.352292307918882,
      "grad_norm": 1.5045995712280273,
      "learning_rate": 2.8863637851247246e-06,
      "loss": 0.0473,
      "step": 295525
    },
    {
      "epoch": 4.352660491008969,
      "grad_norm": 1.440407633781433,
      "learning_rate": 2.8847274131577403e-06,
      "loss": 0.0544,
      "step": 295550
    },
    {
      "epoch": 4.353028674099056,
      "grad_norm": 1.6547504663467407,
      "learning_rate": 2.883091041190755e-06,
      "loss": 0.058,
      "step": 295575
    },
    {
      "epoch": 4.353396857189143,
      "grad_norm": 1.3485205173492432,
      "learning_rate": 2.881454669223771e-06,
      "loss": 0.0484,
      "step": 295600
    },
    {
      "epoch": 4.35376504027923,
      "grad_norm": 1.4512834548950195,
      "learning_rate": 2.8798182972567862e-06,
      "loss": 0.0565,
      "step": 295625
    },
    {
      "epoch": 4.354133223369317,
      "grad_norm": 1.4781733751296997,
      "learning_rate": 2.878181925289802e-06,
      "loss": 0.0512,
      "step": 295650
    },
    {
      "epoch": 4.354501406459404,
      "grad_norm": 0.9898490309715271,
      "learning_rate": 2.876545553322817e-06,
      "loss": 0.0546,
      "step": 295675
    },
    {
      "epoch": 4.354869589549491,
      "grad_norm": 1.4517611265182495,
      "learning_rate": 2.8749091813558326e-06,
      "loss": 0.054,
      "step": 295700
    },
    {
      "epoch": 4.355237772639578,
      "grad_norm": 1.4266328811645508,
      "learning_rate": 2.873272809388848e-06,
      "loss": 0.0442,
      "step": 295725
    },
    {
      "epoch": 4.355605955729665,
      "grad_norm": 1.6748931407928467,
      "learning_rate": 2.8716364374218636e-06,
      "loss": 0.0491,
      "step": 295750
    },
    {
      "epoch": 4.355974138819752,
      "grad_norm": 1.054490089416504,
      "learning_rate": 2.870000065454879e-06,
      "loss": 0.0582,
      "step": 295775
    },
    {
      "epoch": 4.356342321909839,
      "grad_norm": 1.383265733718872,
      "learning_rate": 2.8683636934878946e-06,
      "loss": 0.0434,
      "step": 295800
    },
    {
      "epoch": 4.356710504999926,
      "grad_norm": 1.3003169298171997,
      "learning_rate": 2.8667273215209095e-06,
      "loss": 0.0503,
      "step": 295825
    },
    {
      "epoch": 4.357078688090013,
      "grad_norm": 1.1027840375900269,
      "learning_rate": 2.8650909495539252e-06,
      "loss": 0.0531,
      "step": 295850
    },
    {
      "epoch": 4.3574468711801,
      "grad_norm": 1.10606050491333,
      "learning_rate": 2.8634545775869405e-06,
      "loss": 0.053,
      "step": 295875
    },
    {
      "epoch": 4.357815054270188,
      "grad_norm": 1.530905842781067,
      "learning_rate": 2.8618182056199563e-06,
      "loss": 0.0571,
      "step": 295900
    },
    {
      "epoch": 4.358183237360275,
      "grad_norm": 1.0879781246185303,
      "learning_rate": 2.860181833652971e-06,
      "loss": 0.0533,
      "step": 295925
    },
    {
      "epoch": 4.358551420450362,
      "grad_norm": 1.706228256225586,
      "learning_rate": 2.858545461685987e-06,
      "loss": 0.0567,
      "step": 295950
    },
    {
      "epoch": 4.358919603540449,
      "grad_norm": 1.7201368808746338,
      "learning_rate": 2.8569090897190026e-06,
      "loss": 0.0504,
      "step": 295975
    },
    {
      "epoch": 4.359287786630536,
      "grad_norm": 1.1667488813400269,
      "learning_rate": 2.855272717752018e-06,
      "loss": 0.0538,
      "step": 296000
    },
    {
      "epoch": 4.359655969720623,
      "grad_norm": 1.4120999574661255,
      "learning_rate": 2.8536363457850336e-06,
      "loss": 0.0501,
      "step": 296025
    },
    {
      "epoch": 4.36002415281071,
      "grad_norm": 1.6748450994491577,
      "learning_rate": 2.852065428696728e-06,
      "loss": 0.0581,
      "step": 296050
    },
    {
      "epoch": 4.360392335900797,
      "grad_norm": 1.5026227235794067,
      "learning_rate": 2.8504290567297437e-06,
      "loss": 0.0541,
      "step": 296075
    },
    {
      "epoch": 4.360760518990884,
      "grad_norm": 1.3919445276260376,
      "learning_rate": 2.848792684762759e-06,
      "loss": 0.0542,
      "step": 296100
    },
    {
      "epoch": 4.361128702080971,
      "grad_norm": 1.1815283298492432,
      "learning_rate": 2.8471563127957747e-06,
      "loss": 0.0596,
      "step": 296125
    },
    {
      "epoch": 4.361496885171058,
      "grad_norm": 1.4525161981582642,
      "learning_rate": 2.84551994082879e-06,
      "loss": 0.0529,
      "step": 296150
    },
    {
      "epoch": 4.361865068261145,
      "grad_norm": 1.4406558275222778,
      "learning_rate": 2.8438835688618053e-06,
      "loss": 0.0586,
      "step": 296175
    },
    {
      "epoch": 4.362233251351232,
      "grad_norm": 1.7876112461090088,
      "learning_rate": 2.8422471968948206e-06,
      "loss": 0.0545,
      "step": 296200
    },
    {
      "epoch": 4.362601434441319,
      "grad_norm": 1.2497589588165283,
      "learning_rate": 2.8406108249278363e-06,
      "loss": 0.0538,
      "step": 296225
    },
    {
      "epoch": 4.362969617531406,
      "grad_norm": 1.3769502639770508,
      "learning_rate": 2.8389744529608516e-06,
      "loss": 0.0524,
      "step": 296250
    },
    {
      "epoch": 4.363337800621493,
      "grad_norm": 1.0861910581588745,
      "learning_rate": 2.8373380809938674e-06,
      "loss": 0.0514,
      "step": 296275
    },
    {
      "epoch": 4.36370598371158,
      "grad_norm": 1.4146472215652466,
      "learning_rate": 2.8357017090268823e-06,
      "loss": 0.0558,
      "step": 296300
    },
    {
      "epoch": 4.364074166801667,
      "grad_norm": 0.7640154361724854,
      "learning_rate": 2.834065337059898e-06,
      "loss": 0.0633,
      "step": 296325
    },
    {
      "epoch": 4.364442349891754,
      "grad_norm": 1.4597136974334717,
      "learning_rate": 2.8324289650929133e-06,
      "loss": 0.0492,
      "step": 296350
    },
    {
      "epoch": 4.364810532981841,
      "grad_norm": 1.7000768184661865,
      "learning_rate": 2.830792593125929e-06,
      "loss": 0.0476,
      "step": 296375
    },
    {
      "epoch": 4.365178716071928,
      "grad_norm": 1.114599347114563,
      "learning_rate": 2.829156221158944e-06,
      "loss": 0.0589,
      "step": 296400
    },
    {
      "epoch": 4.365546899162015,
      "grad_norm": 1.2643077373504639,
      "learning_rate": 2.8275198491919596e-06,
      "loss": 0.0506,
      "step": 296425
    },
    {
      "epoch": 4.365915082252102,
      "grad_norm": 1.7858588695526123,
      "learning_rate": 2.8258834772249753e-06,
      "loss": 0.0585,
      "step": 296450
    },
    {
      "epoch": 4.366283265342189,
      "grad_norm": 1.3012501001358032,
      "learning_rate": 2.8242471052579907e-06,
      "loss": 0.0539,
      "step": 296475
    },
    {
      "epoch": 4.366651448432276,
      "grad_norm": 1.6165803670883179,
      "learning_rate": 2.8226107332910064e-06,
      "loss": 0.0523,
      "step": 296500
    },
    {
      "epoch": 4.367019631522363,
      "grad_norm": 1.2451139688491821,
      "learning_rate": 2.8209743613240213e-06,
      "loss": 0.0562,
      "step": 296525
    },
    {
      "epoch": 4.36738781461245,
      "grad_norm": 1.2229077816009521,
      "learning_rate": 2.819337989357037e-06,
      "loss": 0.0508,
      "step": 296550
    },
    {
      "epoch": 4.367755997702537,
      "grad_norm": 1.5834873914718628,
      "learning_rate": 2.8177016173900523e-06,
      "loss": 0.0521,
      "step": 296575
    },
    {
      "epoch": 4.368124180792624,
      "grad_norm": 1.4526238441467285,
      "learning_rate": 2.816065245423068e-06,
      "loss": 0.0511,
      "step": 296600
    },
    {
      "epoch": 4.368492363882711,
      "grad_norm": 0.9935227632522583,
      "learning_rate": 2.8144288734560833e-06,
      "loss": 0.052,
      "step": 296625
    },
    {
      "epoch": 4.368860546972798,
      "grad_norm": 1.1918072700500488,
      "learning_rate": 2.812792501489099e-06,
      "loss": 0.0478,
      "step": 296650
    },
    {
      "epoch": 4.369228730062885,
      "grad_norm": 1.8841264247894287,
      "learning_rate": 2.811156129522114e-06,
      "loss": 0.0499,
      "step": 296675
    },
    {
      "epoch": 4.369596913152972,
      "grad_norm": 1.350530982017517,
      "learning_rate": 2.8095197575551297e-06,
      "loss": 0.0448,
      "step": 296700
    },
    {
      "epoch": 4.369965096243059,
      "grad_norm": 1.1562445163726807,
      "learning_rate": 2.807883385588145e-06,
      "loss": 0.0511,
      "step": 296725
    },
    {
      "epoch": 4.370333279333146,
      "grad_norm": 1.1276297569274902,
      "learning_rate": 2.8062470136211607e-06,
      "loss": 0.0488,
      "step": 296750
    },
    {
      "epoch": 4.370701462423233,
      "grad_norm": 1.4236958026885986,
      "learning_rate": 2.8046106416541756e-06,
      "loss": 0.0593,
      "step": 296775
    },
    {
      "epoch": 4.371069645513321,
      "grad_norm": 1.2665432691574097,
      "learning_rate": 2.8029742696871913e-06,
      "loss": 0.0543,
      "step": 296800
    },
    {
      "epoch": 4.371437828603408,
      "grad_norm": 1.5597598552703857,
      "learning_rate": 2.8013378977202066e-06,
      "loss": 0.0512,
      "step": 296825
    },
    {
      "epoch": 4.371806011693495,
      "grad_norm": 0.6048399806022644,
      "learning_rate": 2.7997015257532223e-06,
      "loss": 0.0517,
      "step": 296850
    },
    {
      "epoch": 4.372174194783582,
      "grad_norm": 1.1398652791976929,
      "learning_rate": 2.7980651537862376e-06,
      "loss": 0.0476,
      "step": 296875
    },
    {
      "epoch": 4.372542377873669,
      "grad_norm": 1.0361242294311523,
      "learning_rate": 2.796428781819253e-06,
      "loss": 0.0544,
      "step": 296900
    },
    {
      "epoch": 4.372910560963756,
      "grad_norm": 1.6299474239349365,
      "learning_rate": 2.7947924098522683e-06,
      "loss": 0.0549,
      "step": 296925
    },
    {
      "epoch": 4.373278744053843,
      "grad_norm": 1.4158477783203125,
      "learning_rate": 2.793156037885284e-06,
      "loss": 0.0554,
      "step": 296950
    },
    {
      "epoch": 4.37364692714393,
      "grad_norm": 1.4037063121795654,
      "learning_rate": 2.7915196659182997e-06,
      "loss": 0.0497,
      "step": 296975
    },
    {
      "epoch": 4.3740151102340175,
      "grad_norm": 1.048384428024292,
      "learning_rate": 2.789883293951315e-06,
      "loss": 0.0581,
      "step": 297000
    },
    {
      "epoch": 4.3743832933241045,
      "grad_norm": 1.5785359144210815,
      "learning_rate": 2.7882469219843307e-06,
      "loss": 0.0541,
      "step": 297025
    },
    {
      "epoch": 4.3747514764141915,
      "grad_norm": 0.9885494112968445,
      "learning_rate": 2.7866105500173456e-06,
      "loss": 0.0488,
      "step": 297050
    },
    {
      "epoch": 4.3751196595042785,
      "grad_norm": 1.119652271270752,
      "learning_rate": 2.7849741780503613e-06,
      "loss": 0.054,
      "step": 297075
    },
    {
      "epoch": 4.3754878425943655,
      "grad_norm": 1.4240087270736694,
      "learning_rate": 2.7833378060833767e-06,
      "loss": 0.0523,
      "step": 297100
    },
    {
      "epoch": 4.3758560256844525,
      "grad_norm": 1.3366814851760864,
      "learning_rate": 2.7817014341163924e-06,
      "loss": 0.0538,
      "step": 297125
    },
    {
      "epoch": 4.3762242087745395,
      "grad_norm": 1.2877370119094849,
      "learning_rate": 2.7800650621494073e-06,
      "loss": 0.0583,
      "step": 297150
    },
    {
      "epoch": 4.376592391864627,
      "grad_norm": 0.7757311463356018,
      "learning_rate": 2.778428690182423e-06,
      "loss": 0.0466,
      "step": 297175
    },
    {
      "epoch": 4.376960574954714,
      "grad_norm": 1.0005812644958496,
      "learning_rate": 2.7767923182154383e-06,
      "loss": 0.0568,
      "step": 297200
    },
    {
      "epoch": 4.377328758044801,
      "grad_norm": 1.170225977897644,
      "learning_rate": 2.775155946248454e-06,
      "loss": 0.0439,
      "step": 297225
    },
    {
      "epoch": 4.377696941134888,
      "grad_norm": 1.5662139654159546,
      "learning_rate": 2.7735195742814693e-06,
      "loss": 0.0462,
      "step": 297250
    },
    {
      "epoch": 4.378065124224975,
      "grad_norm": 1.189277172088623,
      "learning_rate": 2.7718832023144846e-06,
      "loss": 0.048,
      "step": 297275
    },
    {
      "epoch": 4.378433307315062,
      "grad_norm": 1.0078911781311035,
      "learning_rate": 2.7702468303475e-06,
      "loss": 0.0522,
      "step": 297300
    },
    {
      "epoch": 4.378801490405149,
      "grad_norm": 1.310375452041626,
      "learning_rate": 2.7686104583805157e-06,
      "loss": 0.0504,
      "step": 297325
    },
    {
      "epoch": 4.379169673495236,
      "grad_norm": 1.2654945850372314,
      "learning_rate": 2.766974086413531e-06,
      "loss": 0.0478,
      "step": 297350
    },
    {
      "epoch": 4.379537856585323,
      "grad_norm": 1.029016137123108,
      "learning_rate": 2.7653377144465467e-06,
      "loss": 0.0525,
      "step": 297375
    },
    {
      "epoch": 4.37990603967541,
      "grad_norm": 1.1610445976257324,
      "learning_rate": 2.7637013424795616e-06,
      "loss": 0.0483,
      "step": 297400
    },
    {
      "epoch": 4.380274222765497,
      "grad_norm": 1.5103274583816528,
      "learning_rate": 2.7620649705125773e-06,
      "loss": 0.0506,
      "step": 297425
    },
    {
      "epoch": 4.380642405855584,
      "grad_norm": 1.6109086275100708,
      "learning_rate": 2.760428598545593e-06,
      "loss": 0.053,
      "step": 297450
    },
    {
      "epoch": 4.381010588945671,
      "grad_norm": 1.547471284866333,
      "learning_rate": 2.7587922265786083e-06,
      "loss": 0.0517,
      "step": 297475
    },
    {
      "epoch": 4.381378772035758,
      "grad_norm": 1.4506932497024536,
      "learning_rate": 2.757155854611624e-06,
      "loss": 0.0461,
      "step": 297500
    },
    {
      "epoch": 4.381746955125845,
      "grad_norm": 1.1375908851623535,
      "learning_rate": 2.755519482644639e-06,
      "loss": 0.0472,
      "step": 297525
    },
    {
      "epoch": 4.382115138215932,
      "grad_norm": 1.2325059175491333,
      "learning_rate": 2.7538831106776547e-06,
      "loss": 0.0563,
      "step": 297550
    },
    {
      "epoch": 4.382483321306019,
      "grad_norm": 1.5177874565124512,
      "learning_rate": 2.75224673871067e-06,
      "loss": 0.0511,
      "step": 297575
    },
    {
      "epoch": 4.382851504396106,
      "grad_norm": 1.8435479402542114,
      "learning_rate": 2.7506103667436857e-06,
      "loss": 0.0505,
      "step": 297600
    },
    {
      "epoch": 4.383219687486193,
      "grad_norm": 1.104292392730713,
      "learning_rate": 2.7489739947767006e-06,
      "loss": 0.0587,
      "step": 297625
    },
    {
      "epoch": 4.38358787057628,
      "grad_norm": 1.0239830017089844,
      "learning_rate": 2.7473376228097163e-06,
      "loss": 0.0481,
      "step": 297650
    },
    {
      "epoch": 4.383956053666367,
      "grad_norm": 1.1710364818572998,
      "learning_rate": 2.7457012508427316e-06,
      "loss": 0.0511,
      "step": 297675
    },
    {
      "epoch": 4.384324236756454,
      "grad_norm": 1.2445671558380127,
      "learning_rate": 2.7440648788757473e-06,
      "loss": 0.0509,
      "step": 297700
    },
    {
      "epoch": 4.384692419846541,
      "grad_norm": 0.7922903299331665,
      "learning_rate": 2.7424285069087627e-06,
      "loss": 0.0456,
      "step": 297725
    },
    {
      "epoch": 4.385060602936628,
      "grad_norm": 1.3949757814407349,
      "learning_rate": 2.7407921349417784e-06,
      "loss": 0.0484,
      "step": 297750
    },
    {
      "epoch": 4.385428786026715,
      "grad_norm": 1.1927595138549805,
      "learning_rate": 2.7391557629747933e-06,
      "loss": 0.0538,
      "step": 297775
    },
    {
      "epoch": 4.385796969116802,
      "grad_norm": 1.561627984046936,
      "learning_rate": 2.737519391007809e-06,
      "loss": 0.0484,
      "step": 297800
    },
    {
      "epoch": 4.38616515220689,
      "grad_norm": 1.2204314470291138,
      "learning_rate": 2.7358830190408243e-06,
      "loss": 0.0511,
      "step": 297825
    },
    {
      "epoch": 4.386533335296977,
      "grad_norm": 1.379150629043579,
      "learning_rate": 2.73424664707384e-06,
      "loss": 0.0572,
      "step": 297850
    },
    {
      "epoch": 4.386901518387064,
      "grad_norm": 1.0495593547821045,
      "learning_rate": 2.732610275106855e-06,
      "loss": 0.0511,
      "step": 297875
    },
    {
      "epoch": 4.387269701477151,
      "grad_norm": 1.3866804838180542,
      "learning_rate": 2.7309739031398706e-06,
      "loss": 0.0552,
      "step": 297900
    },
    {
      "epoch": 4.387637884567238,
      "grad_norm": 1.5815870761871338,
      "learning_rate": 2.7293375311728864e-06,
      "loss": 0.055,
      "step": 297925
    },
    {
      "epoch": 4.388006067657325,
      "grad_norm": 1.0184849500656128,
      "learning_rate": 2.7277011592059017e-06,
      "loss": 0.0517,
      "step": 297950
    },
    {
      "epoch": 4.388374250747412,
      "grad_norm": 1.3937221765518188,
      "learning_rate": 2.7260647872389174e-06,
      "loss": 0.048,
      "step": 297975
    },
    {
      "epoch": 4.388742433837499,
      "grad_norm": 1.2227100133895874,
      "learning_rate": 2.7244284152719323e-06,
      "loss": 0.0495,
      "step": 298000
    },
    {
      "epoch": 4.389110616927586,
      "grad_norm": 1.2483971118927002,
      "learning_rate": 2.722792043304948e-06,
      "loss": 0.0513,
      "step": 298025
    },
    {
      "epoch": 4.389478800017673,
      "grad_norm": 1.6177209615707397,
      "learning_rate": 2.7211556713379633e-06,
      "loss": 0.0486,
      "step": 298050
    },
    {
      "epoch": 4.38984698310776,
      "grad_norm": 1.45516836643219,
      "learning_rate": 2.719519299370979e-06,
      "loss": 0.0532,
      "step": 298075
    },
    {
      "epoch": 4.390215166197847,
      "grad_norm": 1.3236064910888672,
      "learning_rate": 2.7179483822826738e-06,
      "loss": 0.0606,
      "step": 298100
    },
    {
      "epoch": 4.390583349287934,
      "grad_norm": 1.5390287637710571,
      "learning_rate": 2.716312010315689e-06,
      "loss": 0.0544,
      "step": 298125
    },
    {
      "epoch": 4.390951532378021,
      "grad_norm": 1.111795425415039,
      "learning_rate": 2.7146756383487044e-06,
      "loss": 0.0546,
      "step": 298150
    },
    {
      "epoch": 4.391319715468108,
      "grad_norm": 1.445644736289978,
      "learning_rate": 2.71303926638172e-06,
      "loss": 0.0552,
      "step": 298175
    },
    {
      "epoch": 4.391687898558195,
      "grad_norm": 0.9250501394271851,
      "learning_rate": 2.7114028944147354e-06,
      "loss": 0.0521,
      "step": 298200
    },
    {
      "epoch": 4.392056081648282,
      "grad_norm": 1.3016377687454224,
      "learning_rate": 2.709766522447751e-06,
      "loss": 0.0526,
      "step": 298225
    },
    {
      "epoch": 4.392424264738369,
      "grad_norm": 1.109683632850647,
      "learning_rate": 2.708130150480766e-06,
      "loss": 0.053,
      "step": 298250
    },
    {
      "epoch": 4.392792447828456,
      "grad_norm": 1.369711995124817,
      "learning_rate": 2.7064937785137817e-06,
      "loss": 0.0501,
      "step": 298275
    },
    {
      "epoch": 4.393160630918543,
      "grad_norm": 1.0356179475784302,
      "learning_rate": 2.704857406546797e-06,
      "loss": 0.0504,
      "step": 298300
    },
    {
      "epoch": 4.39352881400863,
      "grad_norm": 1.7696492671966553,
      "learning_rate": 2.7032210345798128e-06,
      "loss": 0.0522,
      "step": 298325
    },
    {
      "epoch": 4.393896997098717,
      "grad_norm": 1.1351957321166992,
      "learning_rate": 2.7015846626128276e-06,
      "loss": 0.0515,
      "step": 298350
    },
    {
      "epoch": 4.394265180188804,
      "grad_norm": 1.5372394323349,
      "learning_rate": 2.6999482906458434e-06,
      "loss": 0.057,
      "step": 298375
    },
    {
      "epoch": 4.394633363278891,
      "grad_norm": 1.4864375591278076,
      "learning_rate": 2.698311918678859e-06,
      "loss": 0.0532,
      "step": 298400
    },
    {
      "epoch": 4.395001546368978,
      "grad_norm": 1.473694920539856,
      "learning_rate": 2.6966755467118744e-06,
      "loss": 0.0524,
      "step": 298425
    },
    {
      "epoch": 4.395369729459065,
      "grad_norm": 1.9480584859848022,
      "learning_rate": 2.69503917474489e-06,
      "loss": 0.0558,
      "step": 298450
    },
    {
      "epoch": 4.395737912549152,
      "grad_norm": 1.272112488746643,
      "learning_rate": 2.693402802777905e-06,
      "loss": 0.0504,
      "step": 298475
    },
    {
      "epoch": 4.396106095639239,
      "grad_norm": 1.0264941453933716,
      "learning_rate": 2.6917664308109207e-06,
      "loss": 0.0451,
      "step": 298500
    },
    {
      "epoch": 4.396474278729326,
      "grad_norm": 1.3120458126068115,
      "learning_rate": 2.690130058843936e-06,
      "loss": 0.0506,
      "step": 298525
    },
    {
      "epoch": 4.396842461819413,
      "grad_norm": 1.693663477897644,
      "learning_rate": 2.6884936868769518e-06,
      "loss": 0.0527,
      "step": 298550
    },
    {
      "epoch": 4.3972106449095,
      "grad_norm": 0.8481330871582031,
      "learning_rate": 2.686857314909967e-06,
      "loss": 0.0471,
      "step": 298575
    },
    {
      "epoch": 4.397578827999587,
      "grad_norm": 1.2794184684753418,
      "learning_rate": 2.685220942942983e-06,
      "loss": 0.0561,
      "step": 298600
    },
    {
      "epoch": 4.397947011089674,
      "grad_norm": 0.4349769353866577,
      "learning_rate": 2.6835845709759977e-06,
      "loss": 0.0498,
      "step": 298625
    },
    {
      "epoch": 4.398315194179761,
      "grad_norm": 1.0860872268676758,
      "learning_rate": 2.6819481990090134e-06,
      "loss": 0.0487,
      "step": 298650
    },
    {
      "epoch": 4.398683377269848,
      "grad_norm": 1.555217981338501,
      "learning_rate": 2.6803118270420287e-06,
      "loss": 0.0584,
      "step": 298675
    },
    {
      "epoch": 4.399051560359935,
      "grad_norm": 1.3808329105377197,
      "learning_rate": 2.6786754550750445e-06,
      "loss": 0.0576,
      "step": 298700
    },
    {
      "epoch": 4.399419743450023,
      "grad_norm": 1.217711329460144,
      "learning_rate": 2.6770390831080593e-06,
      "loss": 0.055,
      "step": 298725
    },
    {
      "epoch": 4.39978792654011,
      "grad_norm": 1.384169101715088,
      "learning_rate": 2.675402711141075e-06,
      "loss": 0.0506,
      "step": 298750
    },
    {
      "epoch": 4.400156109630197,
      "grad_norm": 0.9789212942123413,
      "learning_rate": 2.6737663391740904e-06,
      "loss": 0.046,
      "step": 298775
    },
    {
      "epoch": 4.400524292720284,
      "grad_norm": 2.006904125213623,
      "learning_rate": 2.672129967207106e-06,
      "loss": 0.0534,
      "step": 298800
    },
    {
      "epoch": 4.400892475810371,
      "grad_norm": 0.9747607707977295,
      "learning_rate": 2.6704935952401214e-06,
      "loss": 0.0531,
      "step": 298825
    },
    {
      "epoch": 4.401260658900458,
      "grad_norm": 0.8829006552696228,
      "learning_rate": 2.6688572232731367e-06,
      "loss": 0.0511,
      "step": 298850
    },
    {
      "epoch": 4.401628841990545,
      "grad_norm": 1.4461387395858765,
      "learning_rate": 2.667220851306152e-06,
      "loss": 0.0517,
      "step": 298875
    },
    {
      "epoch": 4.401997025080632,
      "grad_norm": 1.2935779094696045,
      "learning_rate": 2.6655844793391677e-06,
      "loss": 0.0504,
      "step": 298900
    },
    {
      "epoch": 4.402365208170719,
      "grad_norm": 1.8071701526641846,
      "learning_rate": 2.6639481073721835e-06,
      "loss": 0.0502,
      "step": 298925
    },
    {
      "epoch": 4.402733391260806,
      "grad_norm": 1.2152225971221924,
      "learning_rate": 2.6623117354051988e-06,
      "loss": 0.0516,
      "step": 298950
    },
    {
      "epoch": 4.403101574350893,
      "grad_norm": 1.7807745933532715,
      "learning_rate": 2.6606753634382145e-06,
      "loss": 0.0545,
      "step": 298975
    },
    {
      "epoch": 4.4034697574409805,
      "grad_norm": 1.3733631372451782,
      "learning_rate": 2.6590389914712294e-06,
      "loss": 0.0563,
      "step": 299000
    },
    {
      "epoch": 4.4038379405310675,
      "grad_norm": 1.5601617097854614,
      "learning_rate": 2.657402619504245e-06,
      "loss": 0.0576,
      "step": 299025
    },
    {
      "epoch": 4.4042061236211545,
      "grad_norm": 1.0742367506027222,
      "learning_rate": 2.6557662475372604e-06,
      "loss": 0.0582,
      "step": 299050
    },
    {
      "epoch": 4.4045743067112415,
      "grad_norm": 0.9975115656852722,
      "learning_rate": 2.654129875570276e-06,
      "loss": 0.0557,
      "step": 299075
    },
    {
      "epoch": 4.4049424898013285,
      "grad_norm": 1.9738733768463135,
      "learning_rate": 2.652493503603291e-06,
      "loss": 0.0523,
      "step": 299100
    },
    {
      "epoch": 4.4053106728914155,
      "grad_norm": 1.4237779378890991,
      "learning_rate": 2.6508571316363067e-06,
      "loss": 0.0516,
      "step": 299125
    },
    {
      "epoch": 4.4056788559815026,
      "grad_norm": 1.1410068273544312,
      "learning_rate": 2.649220759669322e-06,
      "loss": 0.0545,
      "step": 299150
    },
    {
      "epoch": 4.40604703907159,
      "grad_norm": 1.1840661764144897,
      "learning_rate": 2.6475843877023378e-06,
      "loss": 0.0532,
      "step": 299175
    },
    {
      "epoch": 4.406415222161677,
      "grad_norm": 1.254218339920044,
      "learning_rate": 2.645948015735353e-06,
      "loss": 0.0572,
      "step": 299200
    },
    {
      "epoch": 4.406783405251764,
      "grad_norm": 1.343315839767456,
      "learning_rate": 2.6443116437683684e-06,
      "loss": 0.0548,
      "step": 299225
    },
    {
      "epoch": 4.407151588341851,
      "grad_norm": 1.2422288656234741,
      "learning_rate": 2.6426752718013837e-06,
      "loss": 0.052,
      "step": 299250
    },
    {
      "epoch": 4.407519771431938,
      "grad_norm": 1.7081353664398193,
      "learning_rate": 2.6410388998343994e-06,
      "loss": 0.0539,
      "step": 299275
    },
    {
      "epoch": 4.407887954522025,
      "grad_norm": 1.272098183631897,
      "learning_rate": 2.6394025278674147e-06,
      "loss": 0.0461,
      "step": 299300
    },
    {
      "epoch": 4.408256137612112,
      "grad_norm": 1.390230417251587,
      "learning_rate": 2.6377661559004305e-06,
      "loss": 0.0561,
      "step": 299325
    },
    {
      "epoch": 4.408624320702199,
      "grad_norm": 1.0581073760986328,
      "learning_rate": 2.6361297839334453e-06,
      "loss": 0.0426,
      "step": 299350
    },
    {
      "epoch": 4.408992503792286,
      "grad_norm": 0.8326212167739868,
      "learning_rate": 2.634493411966461e-06,
      "loss": 0.0501,
      "step": 299375
    },
    {
      "epoch": 4.409360686882373,
      "grad_norm": 0.8531941771507263,
      "learning_rate": 2.6328570399994768e-06,
      "loss": 0.0529,
      "step": 299400
    },
    {
      "epoch": 4.40972886997246,
      "grad_norm": 2.0169854164123535,
      "learning_rate": 2.631220668032492e-06,
      "loss": 0.0551,
      "step": 299425
    },
    {
      "epoch": 4.410097053062547,
      "grad_norm": 1.4882276058197021,
      "learning_rate": 2.629584296065508e-06,
      "loss": 0.0519,
      "step": 299450
    },
    {
      "epoch": 4.410465236152634,
      "grad_norm": 1.610654592514038,
      "learning_rate": 2.6279479240985227e-06,
      "loss": 0.0532,
      "step": 299475
    },
    {
      "epoch": 4.410833419242721,
      "grad_norm": 1.629957675933838,
      "learning_rate": 2.6263115521315384e-06,
      "loss": 0.0539,
      "step": 299500
    },
    {
      "epoch": 4.411201602332808,
      "grad_norm": 1.3898857831954956,
      "learning_rate": 2.6246751801645537e-06,
      "loss": 0.0477,
      "step": 299525
    },
    {
      "epoch": 4.411569785422895,
      "grad_norm": 1.2525744438171387,
      "learning_rate": 2.6230388081975695e-06,
      "loss": 0.0543,
      "step": 299550
    },
    {
      "epoch": 4.411937968512982,
      "grad_norm": 0.7920873761177063,
      "learning_rate": 2.6214024362305843e-06,
      "loss": 0.0507,
      "step": 299575
    },
    {
      "epoch": 4.412306151603069,
      "grad_norm": 1.3093024492263794,
      "learning_rate": 2.6197660642636e-06,
      "loss": 0.0532,
      "step": 299600
    },
    {
      "epoch": 4.412674334693156,
      "grad_norm": 1.4931198358535767,
      "learning_rate": 2.6181296922966154e-06,
      "loss": 0.051,
      "step": 299625
    },
    {
      "epoch": 4.413042517783243,
      "grad_norm": 1.462104082107544,
      "learning_rate": 2.616493320329631e-06,
      "loss": 0.0631,
      "step": 299650
    },
    {
      "epoch": 4.41341070087333,
      "grad_norm": 1.5805273056030273,
      "learning_rate": 2.6148569483626464e-06,
      "loss": 0.0508,
      "step": 299675
    },
    {
      "epoch": 4.413778883963417,
      "grad_norm": 1.3844873905181885,
      "learning_rate": 2.613220576395662e-06,
      "loss": 0.0505,
      "step": 299700
    },
    {
      "epoch": 4.414147067053504,
      "grad_norm": 1.4826583862304688,
      "learning_rate": 2.611584204428677e-06,
      "loss": 0.0536,
      "step": 299725
    },
    {
      "epoch": 4.414515250143592,
      "grad_norm": 1.5089002847671509,
      "learning_rate": 2.6099478324616927e-06,
      "loss": 0.0525,
      "step": 299750
    },
    {
      "epoch": 4.414883433233679,
      "grad_norm": 1.598009467124939,
      "learning_rate": 2.608311460494708e-06,
      "loss": 0.0574,
      "step": 299775
    },
    {
      "epoch": 4.415251616323766,
      "grad_norm": 1.2918604612350464,
      "learning_rate": 2.6066750885277238e-06,
      "loss": 0.0497,
      "step": 299800
    },
    {
      "epoch": 4.415619799413853,
      "grad_norm": 1.3681979179382324,
      "learning_rate": 2.6050387165607387e-06,
      "loss": 0.0492,
      "step": 299825
    },
    {
      "epoch": 4.41598798250394,
      "grad_norm": 1.422539234161377,
      "learning_rate": 2.6034023445937544e-06,
      "loss": 0.0504,
      "step": 299850
    },
    {
      "epoch": 4.416356165594027,
      "grad_norm": 1.8677875995635986,
      "learning_rate": 2.60176597262677e-06,
      "loss": 0.0562,
      "step": 299875
    },
    {
      "epoch": 4.416724348684114,
      "grad_norm": 1.4954921007156372,
      "learning_rate": 2.6001296006597854e-06,
      "loss": 0.0495,
      "step": 299900
    },
    {
      "epoch": 4.417092531774201,
      "grad_norm": 0.998248815536499,
      "learning_rate": 2.598493228692801e-06,
      "loss": 0.0475,
      "step": 299925
    },
    {
      "epoch": 4.417460714864288,
      "grad_norm": 1.0428069829940796,
      "learning_rate": 2.596856856725816e-06,
      "loss": 0.0476,
      "step": 299950
    },
    {
      "epoch": 4.417828897954375,
      "grad_norm": 1.4951359033584595,
      "learning_rate": 2.5952204847588318e-06,
      "loss": 0.0563,
      "step": 299975
    },
    {
      "epoch": 4.418197081044462,
      "grad_norm": 1.082573413848877,
      "learning_rate": 2.593584112791847e-06,
      "loss": 0.0516,
      "step": 300000
    },
    {
      "epoch": 4.418565264134549,
      "grad_norm": 1.1197725534439087,
      "learning_rate": 2.5919477408248628e-06,
      "loss": 0.048,
      "step": 300025
    },
    {
      "epoch": 4.418933447224636,
      "grad_norm": 1.5714365243911743,
      "learning_rate": 2.590311368857878e-06,
      "loss": 0.0472,
      "step": 300050
    },
    {
      "epoch": 4.419301630314723,
      "grad_norm": 1.0920747518539429,
      "learning_rate": 2.588674996890894e-06,
      "loss": 0.0513,
      "step": 300075
    },
    {
      "epoch": 4.41966981340481,
      "grad_norm": 0.701291024684906,
      "learning_rate": 2.5870386249239087e-06,
      "loss": 0.0459,
      "step": 300100
    },
    {
      "epoch": 4.420037996494897,
      "grad_norm": 1.2260955572128296,
      "learning_rate": 2.5854022529569244e-06,
      "loss": 0.0532,
      "step": 300125
    },
    {
      "epoch": 4.420406179584984,
      "grad_norm": 1.5864808559417725,
      "learning_rate": 2.5837658809899397e-06,
      "loss": 0.0575,
      "step": 300150
    },
    {
      "epoch": 4.420774362675071,
      "grad_norm": 1.6229859590530396,
      "learning_rate": 2.5821295090229555e-06,
      "loss": 0.0479,
      "step": 300175
    },
    {
      "epoch": 4.421142545765158,
      "grad_norm": 1.4057663679122925,
      "learning_rate": 2.5804931370559703e-06,
      "loss": 0.0584,
      "step": 300200
    },
    {
      "epoch": 4.421510728855245,
      "grad_norm": 1.6148444414138794,
      "learning_rate": 2.578856765088986e-06,
      "loss": 0.0518,
      "step": 300225
    },
    {
      "epoch": 4.421878911945332,
      "grad_norm": 0.7885890603065491,
      "learning_rate": 2.5772203931220014e-06,
      "loss": 0.0474,
      "step": 300250
    },
    {
      "epoch": 4.422247095035419,
      "grad_norm": 1.4228276014328003,
      "learning_rate": 2.575584021155017e-06,
      "loss": 0.0495,
      "step": 300275
    },
    {
      "epoch": 4.422615278125506,
      "grad_norm": 1.3355226516723633,
      "learning_rate": 2.5739476491880324e-06,
      "loss": 0.0519,
      "step": 300300
    },
    {
      "epoch": 4.422983461215593,
      "grad_norm": 0.9256533980369568,
      "learning_rate": 2.5723112772210477e-06,
      "loss": 0.055,
      "step": 300325
    },
    {
      "epoch": 4.42335164430568,
      "grad_norm": 1.1552969217300415,
      "learning_rate": 2.570674905254063e-06,
      "loss": 0.0492,
      "step": 300350
    },
    {
      "epoch": 4.423719827395767,
      "grad_norm": 0.8856759071350098,
      "learning_rate": 2.5690385332870787e-06,
      "loss": 0.0494,
      "step": 300375
    },
    {
      "epoch": 4.424088010485854,
      "grad_norm": 0.7557277679443359,
      "learning_rate": 2.5674021613200945e-06,
      "loss": 0.048,
      "step": 300400
    },
    {
      "epoch": 4.424456193575941,
      "grad_norm": 1.100771188735962,
      "learning_rate": 2.5657657893531098e-06,
      "loss": 0.0521,
      "step": 300425
    },
    {
      "epoch": 4.424824376666028,
      "grad_norm": 1.4789375066757202,
      "learning_rate": 2.5641294173861255e-06,
      "loss": 0.0531,
      "step": 300450
    },
    {
      "epoch": 4.425192559756115,
      "grad_norm": 1.3899985551834106,
      "learning_rate": 2.5624930454191404e-06,
      "loss": 0.0465,
      "step": 300475
    },
    {
      "epoch": 4.425560742846202,
      "grad_norm": 1.5925854444503784,
      "learning_rate": 2.560856673452156e-06,
      "loss": 0.0512,
      "step": 300500
    },
    {
      "epoch": 4.425928925936289,
      "grad_norm": 1.3255113363265991,
      "learning_rate": 2.5592203014851714e-06,
      "loss": 0.0537,
      "step": 300525
    },
    {
      "epoch": 4.426297109026376,
      "grad_norm": 1.81946861743927,
      "learning_rate": 2.557583929518187e-06,
      "loss": 0.0564,
      "step": 300550
    },
    {
      "epoch": 4.426665292116463,
      "grad_norm": 1.4383327960968018,
      "learning_rate": 2.555947557551202e-06,
      "loss": 0.0471,
      "step": 300575
    },
    {
      "epoch": 4.42703347520655,
      "grad_norm": 1.5087847709655762,
      "learning_rate": 2.5543111855842178e-06,
      "loss": 0.0584,
      "step": 300600
    },
    {
      "epoch": 4.427401658296637,
      "grad_norm": 1.9436042308807373,
      "learning_rate": 2.552674813617233e-06,
      "loss": 0.054,
      "step": 300625
    },
    {
      "epoch": 4.427769841386725,
      "grad_norm": 0.8835351467132568,
      "learning_rate": 2.5510384416502488e-06,
      "loss": 0.048,
      "step": 300650
    },
    {
      "epoch": 4.428138024476812,
      "grad_norm": 1.235834002494812,
      "learning_rate": 2.5494020696832637e-06,
      "loss": 0.0468,
      "step": 300675
    },
    {
      "epoch": 4.428506207566899,
      "grad_norm": 1.4335917234420776,
      "learning_rate": 2.5477656977162794e-06,
      "loss": 0.0549,
      "step": 300700
    },
    {
      "epoch": 4.428874390656986,
      "grad_norm": 1.2068747282028198,
      "learning_rate": 2.5461293257492947e-06,
      "loss": 0.0568,
      "step": 300725
    },
    {
      "epoch": 4.429242573747073,
      "grad_norm": 1.0815385580062866,
      "learning_rate": 2.5444929537823104e-06,
      "loss": 0.0509,
      "step": 300750
    },
    {
      "epoch": 4.42961075683716,
      "grad_norm": 0.9886804223060608,
      "learning_rate": 2.5428565818153257e-06,
      "loss": 0.0536,
      "step": 300775
    },
    {
      "epoch": 4.429978939927247,
      "grad_norm": 1.6844570636749268,
      "learning_rate": 2.5412202098483415e-06,
      "loss": 0.0512,
      "step": 300800
    },
    {
      "epoch": 4.430347123017334,
      "grad_norm": 1.2598366737365723,
      "learning_rate": 2.5395838378813563e-06,
      "loss": 0.0509,
      "step": 300825
    },
    {
      "epoch": 4.430715306107421,
      "grad_norm": 1.3276710510253906,
      "learning_rate": 2.537947465914372e-06,
      "loss": 0.0495,
      "step": 300850
    },
    {
      "epoch": 4.431083489197508,
      "grad_norm": 0.822670578956604,
      "learning_rate": 2.536311093947388e-06,
      "loss": 0.0536,
      "step": 300875
    },
    {
      "epoch": 4.431451672287595,
      "grad_norm": 1.4422811269760132,
      "learning_rate": 2.5347401768590825e-06,
      "loss": 0.0577,
      "step": 300900
    },
    {
      "epoch": 4.431819855377682,
      "grad_norm": 1.4851795434951782,
      "learning_rate": 2.5331038048920982e-06,
      "loss": 0.057,
      "step": 300925
    },
    {
      "epoch": 4.432188038467769,
      "grad_norm": 1.595252513885498,
      "learning_rate": 2.531467432925113e-06,
      "loss": 0.047,
      "step": 300950
    },
    {
      "epoch": 4.4325562215578564,
      "grad_norm": 1.4419323205947876,
      "learning_rate": 2.529831060958129e-06,
      "loss": 0.0517,
      "step": 300975
    },
    {
      "epoch": 4.4329244046479435,
      "grad_norm": 1.0129940509796143,
      "learning_rate": 2.528194688991144e-06,
      "loss": 0.05,
      "step": 301000
    },
    {
      "epoch": 4.4332925877380305,
      "grad_norm": 1.3844878673553467,
      "learning_rate": 2.52655831702416e-06,
      "loss": 0.0524,
      "step": 301025
    },
    {
      "epoch": 4.4336607708281175,
      "grad_norm": 1.4275884628295898,
      "learning_rate": 2.5249219450571748e-06,
      "loss": 0.0441,
      "step": 301050
    },
    {
      "epoch": 4.4340289539182045,
      "grad_norm": 1.6299229860305786,
      "learning_rate": 2.5232855730901905e-06,
      "loss": 0.0533,
      "step": 301075
    },
    {
      "epoch": 4.4343971370082915,
      "grad_norm": 0.8308444023132324,
      "learning_rate": 2.521649201123206e-06,
      "loss": 0.0507,
      "step": 301100
    },
    {
      "epoch": 4.4347653200983785,
      "grad_norm": 1.664947509765625,
      "learning_rate": 2.5200128291562215e-06,
      "loss": 0.0465,
      "step": 301125
    },
    {
      "epoch": 4.4351335031884656,
      "grad_norm": 1.0470316410064697,
      "learning_rate": 2.5183764571892364e-06,
      "loss": 0.0442,
      "step": 301150
    },
    {
      "epoch": 4.435501686278553,
      "grad_norm": 0.891488790512085,
      "learning_rate": 2.516740085222252e-06,
      "loss": 0.0522,
      "step": 301175
    },
    {
      "epoch": 4.43586986936864,
      "grad_norm": 1.3064780235290527,
      "learning_rate": 2.5151037132552674e-06,
      "loss": 0.0455,
      "step": 301200
    },
    {
      "epoch": 4.436238052458727,
      "grad_norm": 1.6625359058380127,
      "learning_rate": 2.513467341288283e-06,
      "loss": 0.0641,
      "step": 301225
    },
    {
      "epoch": 4.436606235548814,
      "grad_norm": 1.353169560432434,
      "learning_rate": 2.5118309693212985e-06,
      "loss": 0.0588,
      "step": 301250
    },
    {
      "epoch": 4.436974418638901,
      "grad_norm": 1.3421375751495361,
      "learning_rate": 2.510194597354314e-06,
      "loss": 0.0512,
      "step": 301275
    },
    {
      "epoch": 4.437342601728988,
      "grad_norm": 0.7791436314582825,
      "learning_rate": 2.508558225387329e-06,
      "loss": 0.049,
      "step": 301300
    },
    {
      "epoch": 4.437710784819075,
      "grad_norm": 1.415158748626709,
      "learning_rate": 2.506921853420345e-06,
      "loss": 0.0603,
      "step": 301325
    },
    {
      "epoch": 4.438078967909162,
      "grad_norm": 1.0226030349731445,
      "learning_rate": 2.5052854814533605e-06,
      "loss": 0.048,
      "step": 301350
    },
    {
      "epoch": 4.438447150999249,
      "grad_norm": 1.190688967704773,
      "learning_rate": 2.503649109486376e-06,
      "loss": 0.0462,
      "step": 301375
    },
    {
      "epoch": 4.438815334089336,
      "grad_norm": 1.3343232870101929,
      "learning_rate": 2.5020127375193916e-06,
      "loss": 0.0551,
      "step": 301400
    },
    {
      "epoch": 4.439183517179423,
      "grad_norm": 1.429953932762146,
      "learning_rate": 2.5003763655524065e-06,
      "loss": 0.0508,
      "step": 301425
    },
    {
      "epoch": 4.43955170026951,
      "grad_norm": 1.3824851512908936,
      "learning_rate": 2.498739993585422e-06,
      "loss": 0.0502,
      "step": 301450
    },
    {
      "epoch": 4.439919883359597,
      "grad_norm": 1.5845098495483398,
      "learning_rate": 2.4971036216184375e-06,
      "loss": 0.0587,
      "step": 301475
    },
    {
      "epoch": 4.440288066449684,
      "grad_norm": 1.8588379621505737,
      "learning_rate": 2.495467249651453e-06,
      "loss": 0.0493,
      "step": 301500
    },
    {
      "epoch": 4.440656249539771,
      "grad_norm": 1.6703959703445435,
      "learning_rate": 2.493830877684468e-06,
      "loss": 0.0535,
      "step": 301525
    },
    {
      "epoch": 4.441024432629858,
      "grad_norm": 1.5778106451034546,
      "learning_rate": 2.492194505717484e-06,
      "loss": 0.0554,
      "step": 301550
    },
    {
      "epoch": 4.441392615719945,
      "grad_norm": 1.471972942352295,
      "learning_rate": 2.4905581337504996e-06,
      "loss": 0.0484,
      "step": 301575
    },
    {
      "epoch": 4.441760798810032,
      "grad_norm": 1.1855192184448242,
      "learning_rate": 2.488921761783515e-06,
      "loss": 0.0498,
      "step": 301600
    },
    {
      "epoch": 4.442128981900119,
      "grad_norm": 1.611812949180603,
      "learning_rate": 2.48728538981653e-06,
      "loss": 0.0459,
      "step": 301625
    },
    {
      "epoch": 4.442497164990206,
      "grad_norm": 1.5399442911148071,
      "learning_rate": 2.485649017849546e-06,
      "loss": 0.061,
      "step": 301650
    },
    {
      "epoch": 4.442865348080294,
      "grad_norm": 1.0918532609939575,
      "learning_rate": 2.484012645882561e-06,
      "loss": 0.0498,
      "step": 301675
    },
    {
      "epoch": 4.443233531170381,
      "grad_norm": 0.7599321007728577,
      "learning_rate": 2.4823762739155765e-06,
      "loss": 0.0529,
      "step": 301700
    },
    {
      "epoch": 4.443601714260468,
      "grad_norm": 1.6148830652236938,
      "learning_rate": 2.480739901948592e-06,
      "loss": 0.0517,
      "step": 301725
    },
    {
      "epoch": 4.443969897350555,
      "grad_norm": 1.2528841495513916,
      "learning_rate": 2.4791035299816075e-06,
      "loss": 0.0491,
      "step": 301750
    },
    {
      "epoch": 4.444338080440642,
      "grad_norm": 1.618980884552002,
      "learning_rate": 2.477467158014623e-06,
      "loss": 0.0486,
      "step": 301775
    },
    {
      "epoch": 4.444706263530729,
      "grad_norm": 0.9234166145324707,
      "learning_rate": 2.475830786047638e-06,
      "loss": 0.0524,
      "step": 301800
    },
    {
      "epoch": 4.445074446620816,
      "grad_norm": 1.7744756937026978,
      "learning_rate": 2.474194414080654e-06,
      "loss": 0.0505,
      "step": 301825
    },
    {
      "epoch": 4.445442629710903,
      "grad_norm": 1.1677004098892212,
      "learning_rate": 2.472558042113669e-06,
      "loss": 0.0591,
      "step": 301850
    },
    {
      "epoch": 4.44581081280099,
      "grad_norm": 1.5444965362548828,
      "learning_rate": 2.4709216701466845e-06,
      "loss": 0.0512,
      "step": 301875
    },
    {
      "epoch": 4.446178995891077,
      "grad_norm": 1.2752526998519897,
      "learning_rate": 2.4692852981796998e-06,
      "loss": 0.0531,
      "step": 301900
    },
    {
      "epoch": 4.446547178981164,
      "grad_norm": 1.4353508949279785,
      "learning_rate": 2.4676489262127155e-06,
      "loss": 0.0484,
      "step": 301925
    },
    {
      "epoch": 4.446915362071251,
      "grad_norm": 1.9133244752883911,
      "learning_rate": 2.466012554245731e-06,
      "loss": 0.0505,
      "step": 301950
    },
    {
      "epoch": 4.447283545161338,
      "grad_norm": 1.2566324472427368,
      "learning_rate": 2.464376182278746e-06,
      "loss": 0.0502,
      "step": 301975
    },
    {
      "epoch": 4.447651728251425,
      "grad_norm": 1.4808827638626099,
      "learning_rate": 2.462739810311762e-06,
      "loss": 0.0492,
      "step": 302000
    },
    {
      "epoch": 4.448019911341512,
      "grad_norm": 0.9286792874336243,
      "learning_rate": 2.461103438344777e-06,
      "loss": 0.0472,
      "step": 302025
    },
    {
      "epoch": 4.448388094431599,
      "grad_norm": 0.6995263695716858,
      "learning_rate": 2.4594670663777925e-06,
      "loss": 0.058,
      "step": 302050
    },
    {
      "epoch": 4.448756277521686,
      "grad_norm": 1.1706923246383667,
      "learning_rate": 2.457830694410808e-06,
      "loss": 0.0463,
      "step": 302075
    },
    {
      "epoch": 4.449124460611773,
      "grad_norm": 0.8549727201461792,
      "learning_rate": 2.4561943224438235e-06,
      "loss": 0.0556,
      "step": 302100
    },
    {
      "epoch": 4.44949264370186,
      "grad_norm": 0.8848899602890015,
      "learning_rate": 2.4545579504768392e-06,
      "loss": 0.0546,
      "step": 302125
    },
    {
      "epoch": 4.449860826791947,
      "grad_norm": 1.4121531248092651,
      "learning_rate": 2.4529215785098545e-06,
      "loss": 0.0498,
      "step": 302150
    },
    {
      "epoch": 4.450229009882034,
      "grad_norm": 1.789003610610962,
      "learning_rate": 2.45128520654287e-06,
      "loss": 0.0505,
      "step": 302175
    },
    {
      "epoch": 4.450597192972121,
      "grad_norm": 1.1204239130020142,
      "learning_rate": 2.4496488345758856e-06,
      "loss": 0.0531,
      "step": 302200
    },
    {
      "epoch": 4.450965376062208,
      "grad_norm": 1.5883513689041138,
      "learning_rate": 2.448012462608901e-06,
      "loss": 0.0558,
      "step": 302225
    },
    {
      "epoch": 4.451333559152295,
      "grad_norm": 1.1423672437667847,
      "learning_rate": 2.446376090641916e-06,
      "loss": 0.0487,
      "step": 302250
    },
    {
      "epoch": 4.451701742242382,
      "grad_norm": 1.250074028968811,
      "learning_rate": 2.4447397186749315e-06,
      "loss": 0.0483,
      "step": 302275
    },
    {
      "epoch": 4.452069925332469,
      "grad_norm": 1.2350640296936035,
      "learning_rate": 2.443103346707947e-06,
      "loss": 0.0537,
      "step": 302300
    },
    {
      "epoch": 4.452438108422556,
      "grad_norm": 1.8885859251022339,
      "learning_rate": 2.4414669747409625e-06,
      "loss": 0.0547,
      "step": 302325
    },
    {
      "epoch": 4.452806291512643,
      "grad_norm": 1.1005498170852661,
      "learning_rate": 2.439830602773978e-06,
      "loss": 0.0471,
      "step": 302350
    },
    {
      "epoch": 4.45317447460273,
      "grad_norm": 1.6805378198623657,
      "learning_rate": 2.4381942308069935e-06,
      "loss": 0.053,
      "step": 302375
    },
    {
      "epoch": 4.453542657692817,
      "grad_norm": 0.9430249929428101,
      "learning_rate": 2.436557858840009e-06,
      "loss": 0.0566,
      "step": 302400
    },
    {
      "epoch": 4.453910840782904,
      "grad_norm": 1.1696069240570068,
      "learning_rate": 2.434921486873024e-06,
      "loss": 0.0552,
      "step": 302425
    },
    {
      "epoch": 4.454279023872991,
      "grad_norm": 1.346459984779358,
      "learning_rate": 2.4332851149060394e-06,
      "loss": 0.0534,
      "step": 302450
    },
    {
      "epoch": 4.454647206963078,
      "grad_norm": 1.5569554567337036,
      "learning_rate": 2.431648742939055e-06,
      "loss": 0.052,
      "step": 302475
    },
    {
      "epoch": 4.455015390053165,
      "grad_norm": 1.1602433919906616,
      "learning_rate": 2.4300123709720705e-06,
      "loss": 0.057,
      "step": 302500
    },
    {
      "epoch": 4.455383573143252,
      "grad_norm": 1.3797434568405151,
      "learning_rate": 2.4283759990050858e-06,
      "loss": 0.0517,
      "step": 302525
    },
    {
      "epoch": 4.455751756233339,
      "grad_norm": 1.4027955532073975,
      "learning_rate": 2.4267396270381015e-06,
      "loss": 0.0515,
      "step": 302550
    },
    {
      "epoch": 4.456119939323427,
      "grad_norm": 1.0705606937408447,
      "learning_rate": 2.4251032550711172e-06,
      "loss": 0.0472,
      "step": 302575
    },
    {
      "epoch": 4.456488122413514,
      "grad_norm": 1.7451097965240479,
      "learning_rate": 2.4234668831041325e-06,
      "loss": 0.0505,
      "step": 302600
    },
    {
      "epoch": 4.456856305503601,
      "grad_norm": 1.7931379079818726,
      "learning_rate": 2.421830511137148e-06,
      "loss": 0.0539,
      "step": 302625
    },
    {
      "epoch": 4.457224488593688,
      "grad_norm": 1.5439224243164062,
      "learning_rate": 2.420194139170163e-06,
      "loss": 0.0518,
      "step": 302650
    },
    {
      "epoch": 4.457592671683775,
      "grad_norm": 1.378536581993103,
      "learning_rate": 2.418557767203179e-06,
      "loss": 0.0538,
      "step": 302675
    },
    {
      "epoch": 4.457960854773862,
      "grad_norm": 1.214421033859253,
      "learning_rate": 2.416921395236194e-06,
      "loss": 0.0491,
      "step": 302700
    },
    {
      "epoch": 4.458329037863949,
      "grad_norm": 1.3873339891433716,
      "learning_rate": 2.4152850232692095e-06,
      "loss": 0.0486,
      "step": 302725
    },
    {
      "epoch": 4.458697220954036,
      "grad_norm": 1.8396759033203125,
      "learning_rate": 2.4136486513022252e-06,
      "loss": 0.0533,
      "step": 302750
    },
    {
      "epoch": 4.459065404044123,
      "grad_norm": 0.9953267574310303,
      "learning_rate": 2.4120122793352405e-06,
      "loss": 0.0522,
      "step": 302775
    },
    {
      "epoch": 4.45943358713421,
      "grad_norm": 1.5243364572525024,
      "learning_rate": 2.410375907368256e-06,
      "loss": 0.0586,
      "step": 302800
    },
    {
      "epoch": 4.459801770224297,
      "grad_norm": 1.5203975439071655,
      "learning_rate": 2.408739535401271e-06,
      "loss": 0.0483,
      "step": 302825
    },
    {
      "epoch": 4.460169953314384,
      "grad_norm": 0.6166027784347534,
      "learning_rate": 2.407103163434287e-06,
      "loss": 0.0553,
      "step": 302850
    },
    {
      "epoch": 4.460538136404471,
      "grad_norm": 1.088309407234192,
      "learning_rate": 2.405466791467302e-06,
      "loss": 0.051,
      "step": 302875
    },
    {
      "epoch": 4.460906319494558,
      "grad_norm": 1.5143396854400635,
      "learning_rate": 2.4038304195003175e-06,
      "loss": 0.0521,
      "step": 302900
    },
    {
      "epoch": 4.461274502584645,
      "grad_norm": 1.5423846244812012,
      "learning_rate": 2.402194047533333e-06,
      "loss": 0.0538,
      "step": 302925
    },
    {
      "epoch": 4.461642685674732,
      "grad_norm": 0.7741517424583435,
      "learning_rate": 2.4005576755663485e-06,
      "loss": 0.0539,
      "step": 302950
    },
    {
      "epoch": 4.4620108687648194,
      "grad_norm": 1.6776727437973022,
      "learning_rate": 2.398921303599364e-06,
      "loss": 0.052,
      "step": 302975
    },
    {
      "epoch": 4.4623790518549065,
      "grad_norm": 2.0980570316314697,
      "learning_rate": 2.397284931632379e-06,
      "loss": 0.0594,
      "step": 303000
    },
    {
      "epoch": 4.4627472349449935,
      "grad_norm": 1.3102998733520508,
      "learning_rate": 2.395648559665395e-06,
      "loss": 0.0484,
      "step": 303025
    },
    {
      "epoch": 4.4631154180350805,
      "grad_norm": 1.2197163105010986,
      "learning_rate": 2.3940121876984106e-06,
      "loss": 0.0497,
      "step": 303050
    },
    {
      "epoch": 4.4634836011251675,
      "grad_norm": 1.6866099834442139,
      "learning_rate": 2.392375815731426e-06,
      "loss": 0.0556,
      "step": 303075
    },
    {
      "epoch": 4.4638517842152545,
      "grad_norm": 0.8207196593284607,
      "learning_rate": 2.390739443764441e-06,
      "loss": 0.0486,
      "step": 303100
    },
    {
      "epoch": 4.4642199673053415,
      "grad_norm": 1.1538947820663452,
      "learning_rate": 2.389103071797457e-06,
      "loss": 0.057,
      "step": 303125
    },
    {
      "epoch": 4.464588150395429,
      "grad_norm": 1.5906556844711304,
      "learning_rate": 2.387466699830472e-06,
      "loss": 0.0492,
      "step": 303150
    },
    {
      "epoch": 4.464956333485516,
      "grad_norm": 1.5449646711349487,
      "learning_rate": 2.3858303278634875e-06,
      "loss": 0.0587,
      "step": 303175
    },
    {
      "epoch": 4.465324516575603,
      "grad_norm": 1.6102849245071411,
      "learning_rate": 2.384193955896503e-06,
      "loss": 0.0543,
      "step": 303200
    },
    {
      "epoch": 4.46569269966569,
      "grad_norm": 0.7648277878761292,
      "learning_rate": 2.3825575839295185e-06,
      "loss": 0.0532,
      "step": 303225
    },
    {
      "epoch": 4.466060882755777,
      "grad_norm": 1.6596821546554565,
      "learning_rate": 2.380921211962534e-06,
      "loss": 0.047,
      "step": 303250
    },
    {
      "epoch": 4.466429065845864,
      "grad_norm": 1.2146966457366943,
      "learning_rate": 2.379284839995549e-06,
      "loss": 0.0461,
      "step": 303275
    },
    {
      "epoch": 4.466797248935951,
      "grad_norm": 1.5407129526138306,
      "learning_rate": 2.377648468028565e-06,
      "loss": 0.054,
      "step": 303300
    },
    {
      "epoch": 4.467165432026038,
      "grad_norm": 2.0960118770599365,
      "learning_rate": 2.37601209606158e-06,
      "loss": 0.053,
      "step": 303325
    },
    {
      "epoch": 4.467533615116125,
      "grad_norm": 1.0898778438568115,
      "learning_rate": 2.3743757240945955e-06,
      "loss": 0.0467,
      "step": 303350
    },
    {
      "epoch": 4.467901798206212,
      "grad_norm": 1.2167497873306274,
      "learning_rate": 2.372739352127611e-06,
      "loss": 0.0476,
      "step": 303375
    },
    {
      "epoch": 4.468269981296299,
      "grad_norm": 1.5868618488311768,
      "learning_rate": 2.3711029801606265e-06,
      "loss": 0.0472,
      "step": 303400
    },
    {
      "epoch": 4.468638164386386,
      "grad_norm": 1.0690587759017944,
      "learning_rate": 2.369466608193642e-06,
      "loss": 0.0556,
      "step": 303425
    },
    {
      "epoch": 4.469006347476473,
      "grad_norm": 1.8434339761734009,
      "learning_rate": 2.367830236226657e-06,
      "loss": 0.049,
      "step": 303450
    },
    {
      "epoch": 4.46937453056656,
      "grad_norm": 1.3922070264816284,
      "learning_rate": 2.366193864259673e-06,
      "loss": 0.055,
      "step": 303475
    },
    {
      "epoch": 4.469742713656647,
      "grad_norm": 1.1121543645858765,
      "learning_rate": 2.364557492292688e-06,
      "loss": 0.0454,
      "step": 303500
    },
    {
      "epoch": 4.470110896746734,
      "grad_norm": 1.2854058742523193,
      "learning_rate": 2.3629211203257035e-06,
      "loss": 0.0515,
      "step": 303525
    },
    {
      "epoch": 4.470479079836821,
      "grad_norm": 1.4660992622375488,
      "learning_rate": 2.361284748358719e-06,
      "loss": 0.0518,
      "step": 303550
    },
    {
      "epoch": 4.470847262926909,
      "grad_norm": 1.7707901000976562,
      "learning_rate": 2.3596483763917345e-06,
      "loss": 0.0505,
      "step": 303575
    },
    {
      "epoch": 4.471215446016996,
      "grad_norm": 1.2327923774719238,
      "learning_rate": 2.3580120044247502e-06,
      "loss": 0.0513,
      "step": 303600
    },
    {
      "epoch": 4.471583629107083,
      "grad_norm": 1.302258014678955,
      "learning_rate": 2.3563756324577655e-06,
      "loss": 0.0477,
      "step": 303625
    },
    {
      "epoch": 4.47195181219717,
      "grad_norm": 1.5161261558532715,
      "learning_rate": 2.354739260490781e-06,
      "loss": 0.0499,
      "step": 303650
    },
    {
      "epoch": 4.472319995287257,
      "grad_norm": 1.1093387603759766,
      "learning_rate": 2.3531028885237966e-06,
      "loss": 0.06,
      "step": 303675
    },
    {
      "epoch": 4.472688178377344,
      "grad_norm": 1.2082892656326294,
      "learning_rate": 2.351466516556812e-06,
      "loss": 0.0603,
      "step": 303700
    },
    {
      "epoch": 4.473056361467431,
      "grad_norm": 1.7611470222473145,
      "learning_rate": 2.349830144589827e-06,
      "loss": 0.0485,
      "step": 303725
    },
    {
      "epoch": 4.473424544557518,
      "grad_norm": 1.4070907831192017,
      "learning_rate": 2.3481937726228425e-06,
      "loss": 0.0555,
      "step": 303750
    },
    {
      "epoch": 4.473792727647605,
      "grad_norm": 2.345460891723633,
      "learning_rate": 2.346557400655858e-06,
      "loss": 0.049,
      "step": 303775
    },
    {
      "epoch": 4.474160910737692,
      "grad_norm": 1.7008612155914307,
      "learning_rate": 2.3449210286888735e-06,
      "loss": 0.0489,
      "step": 303800
    },
    {
      "epoch": 4.474529093827779,
      "grad_norm": 1.2351136207580566,
      "learning_rate": 2.343284656721889e-06,
      "loss": 0.0508,
      "step": 303825
    },
    {
      "epoch": 4.474897276917866,
      "grad_norm": 0.689289927482605,
      "learning_rate": 2.3416482847549045e-06,
      "loss": 0.0478,
      "step": 303850
    },
    {
      "epoch": 4.475265460007953,
      "grad_norm": 1.561794400215149,
      "learning_rate": 2.34001191278792e-06,
      "loss": 0.0507,
      "step": 303875
    },
    {
      "epoch": 4.47563364309804,
      "grad_norm": 0.8364143967628479,
      "learning_rate": 2.338375540820935e-06,
      "loss": 0.0543,
      "step": 303900
    },
    {
      "epoch": 4.476001826188127,
      "grad_norm": 0.9369179010391235,
      "learning_rate": 2.3367391688539505e-06,
      "loss": 0.0493,
      "step": 303925
    },
    {
      "epoch": 4.476370009278214,
      "grad_norm": 1.6872276067733765,
      "learning_rate": 2.335102796886966e-06,
      "loss": 0.0546,
      "step": 303950
    },
    {
      "epoch": 4.476738192368301,
      "grad_norm": 0.6905471086502075,
      "learning_rate": 2.3334664249199815e-06,
      "loss": 0.05,
      "step": 303975
    },
    {
      "epoch": 4.477106375458388,
      "grad_norm": 1.1974610090255737,
      "learning_rate": 2.331830052952997e-06,
      "loss": 0.0481,
      "step": 304000
    },
    {
      "epoch": 4.477474558548475,
      "grad_norm": 1.2794733047485352,
      "learning_rate": 2.3301936809860125e-06,
      "loss": 0.0508,
      "step": 304025
    },
    {
      "epoch": 4.477842741638562,
      "grad_norm": 1.5989645719528198,
      "learning_rate": 2.3285573090190282e-06,
      "loss": 0.0579,
      "step": 304050
    },
    {
      "epoch": 4.478210924728649,
      "grad_norm": 0.7437596321105957,
      "learning_rate": 2.3269209370520435e-06,
      "loss": 0.0554,
      "step": 304075
    },
    {
      "epoch": 4.478579107818736,
      "grad_norm": 1.1852704286575317,
      "learning_rate": 2.325284565085059e-06,
      "loss": 0.0537,
      "step": 304100
    },
    {
      "epoch": 4.478947290908823,
      "grad_norm": 1.0112836360931396,
      "learning_rate": 2.323648193118074e-06,
      "loss": 0.0504,
      "step": 304125
    },
    {
      "epoch": 4.47931547399891,
      "grad_norm": 1.5848444700241089,
      "learning_rate": 2.32201182115109e-06,
      "loss": 0.0473,
      "step": 304150
    },
    {
      "epoch": 4.479683657088997,
      "grad_norm": 1.5396461486816406,
      "learning_rate": 2.320375449184105e-06,
      "loss": 0.051,
      "step": 304175
    },
    {
      "epoch": 4.480051840179084,
      "grad_norm": 2.1263575553894043,
      "learning_rate": 2.3187390772171205e-06,
      "loss": 0.0487,
      "step": 304200
    },
    {
      "epoch": 4.480420023269171,
      "grad_norm": 1.0708191394805908,
      "learning_rate": 2.3171027052501362e-06,
      "loss": 0.0519,
      "step": 304225
    },
    {
      "epoch": 4.480788206359258,
      "grad_norm": 1.460453748703003,
      "learning_rate": 2.3154663332831515e-06,
      "loss": 0.053,
      "step": 304250
    },
    {
      "epoch": 4.481156389449345,
      "grad_norm": 1.161363124847412,
      "learning_rate": 2.313829961316167e-06,
      "loss": 0.0533,
      "step": 304275
    },
    {
      "epoch": 4.481524572539432,
      "grad_norm": 1.493945598602295,
      "learning_rate": 2.312193589349182e-06,
      "loss": 0.051,
      "step": 304300
    },
    {
      "epoch": 4.481892755629519,
      "grad_norm": 1.3790481090545654,
      "learning_rate": 2.310557217382198e-06,
      "loss": 0.048,
      "step": 304325
    },
    {
      "epoch": 4.482260938719606,
      "grad_norm": 1.0603439807891846,
      "learning_rate": 2.308920845415213e-06,
      "loss": 0.0528,
      "step": 304350
    },
    {
      "epoch": 4.482629121809693,
      "grad_norm": 1.2882375717163086,
      "learning_rate": 2.3072844734482285e-06,
      "loss": 0.0511,
      "step": 304375
    },
    {
      "epoch": 4.48299730489978,
      "grad_norm": 1.3489183187484741,
      "learning_rate": 2.305648101481244e-06,
      "loss": 0.0514,
      "step": 304400
    },
    {
      "epoch": 4.483365487989867,
      "grad_norm": 1.2654738426208496,
      "learning_rate": 2.3040117295142595e-06,
      "loss": 0.0519,
      "step": 304425
    },
    {
      "epoch": 4.483733671079954,
      "grad_norm": 1.0368967056274414,
      "learning_rate": 2.302375357547275e-06,
      "loss": 0.0489,
      "step": 304450
    },
    {
      "epoch": 4.484101854170042,
      "grad_norm": 1.6323338747024536,
      "learning_rate": 2.30073898558029e-06,
      "loss": 0.0515,
      "step": 304475
    },
    {
      "epoch": 4.484470037260129,
      "grad_norm": 1.1667799949645996,
      "learning_rate": 2.299102613613306e-06,
      "loss": 0.0566,
      "step": 304500
    },
    {
      "epoch": 4.484838220350216,
      "grad_norm": 1.8040019273757935,
      "learning_rate": 2.2974662416463216e-06,
      "loss": 0.05,
      "step": 304525
    },
    {
      "epoch": 4.485206403440303,
      "grad_norm": 1.075221300125122,
      "learning_rate": 2.295829869679337e-06,
      "loss": 0.0483,
      "step": 304550
    },
    {
      "epoch": 4.48557458653039,
      "grad_norm": 1.8594220876693726,
      "learning_rate": 2.294193497712352e-06,
      "loss": 0.0586,
      "step": 304575
    },
    {
      "epoch": 4.485942769620477,
      "grad_norm": 1.4127616882324219,
      "learning_rate": 2.292557125745368e-06,
      "loss": 0.053,
      "step": 304600
    },
    {
      "epoch": 4.486310952710564,
      "grad_norm": 2.123760938644409,
      "learning_rate": 2.290920753778383e-06,
      "loss": 0.0539,
      "step": 304625
    },
    {
      "epoch": 4.486679135800651,
      "grad_norm": 1.1200772523880005,
      "learning_rate": 2.2892843818113985e-06,
      "loss": 0.0477,
      "step": 304650
    },
    {
      "epoch": 4.487047318890738,
      "grad_norm": 1.318161964416504,
      "learning_rate": 2.287648009844414e-06,
      "loss": 0.0481,
      "step": 304675
    },
    {
      "epoch": 4.487415501980825,
      "grad_norm": 1.3724302053451538,
      "learning_rate": 2.2860116378774295e-06,
      "loss": 0.0608,
      "step": 304700
    },
    {
      "epoch": 4.487783685070912,
      "grad_norm": 1.4727050065994263,
      "learning_rate": 2.284375265910445e-06,
      "loss": 0.0519,
      "step": 304725
    },
    {
      "epoch": 4.488151868160999,
      "grad_norm": 1.287550687789917,
      "learning_rate": 2.28273889394346e-06,
      "loss": 0.0552,
      "step": 304750
    },
    {
      "epoch": 4.488520051251086,
      "grad_norm": 1.0559489727020264,
      "learning_rate": 2.281102521976476e-06,
      "loss": 0.0527,
      "step": 304775
    },
    {
      "epoch": 4.488888234341173,
      "grad_norm": 1.2488083839416504,
      "learning_rate": 2.279466150009491e-06,
      "loss": 0.0494,
      "step": 304800
    },
    {
      "epoch": 4.48925641743126,
      "grad_norm": 1.230230450630188,
      "learning_rate": 2.2778297780425065e-06,
      "loss": 0.0475,
      "step": 304825
    },
    {
      "epoch": 4.489624600521347,
      "grad_norm": 1.1732892990112305,
      "learning_rate": 2.276193406075522e-06,
      "loss": 0.0541,
      "step": 304850
    },
    {
      "epoch": 4.489992783611434,
      "grad_norm": 1.4207347631454468,
      "learning_rate": 2.2745570341085375e-06,
      "loss": 0.0487,
      "step": 304875
    },
    {
      "epoch": 4.490360966701521,
      "grad_norm": 1.2135663032531738,
      "learning_rate": 2.2729861170202323e-06,
      "loss": 0.0477,
      "step": 304900
    },
    {
      "epoch": 4.490729149791608,
      "grad_norm": 1.8241201639175415,
      "learning_rate": 2.2713497450532476e-06,
      "loss": 0.0471,
      "step": 304925
    },
    {
      "epoch": 4.491097332881695,
      "grad_norm": 1.2652854919433594,
      "learning_rate": 2.269713373086263e-06,
      "loss": 0.0578,
      "step": 304950
    },
    {
      "epoch": 4.4914655159717825,
      "grad_norm": 1.6332687139511108,
      "learning_rate": 2.2680770011192786e-06,
      "loss": 0.0456,
      "step": 304975
    },
    {
      "epoch": 4.4918336990618695,
      "grad_norm": 1.4398852586746216,
      "learning_rate": 2.2664406291522943e-06,
      "loss": 0.049,
      "step": 305000
    },
    {
      "epoch": 4.4922018821519565,
      "grad_norm": 1.6836400032043457,
      "learning_rate": 2.2648042571853096e-06,
      "loss": 0.0598,
      "step": 305025
    },
    {
      "epoch": 4.4925700652420435,
      "grad_norm": 1.1527715921401978,
      "learning_rate": 2.263167885218325e-06,
      "loss": 0.051,
      "step": 305050
    },
    {
      "epoch": 4.4929382483321305,
      "grad_norm": 1.7300193309783936,
      "learning_rate": 2.2615315132513407e-06,
      "loss": 0.0555,
      "step": 305075
    },
    {
      "epoch": 4.4933064314222175,
      "grad_norm": 1.165737271308899,
      "learning_rate": 2.259895141284356e-06,
      "loss": 0.0448,
      "step": 305100
    },
    {
      "epoch": 4.4936746145123045,
      "grad_norm": 1.4664829969406128,
      "learning_rate": 2.2582587693173713e-06,
      "loss": 0.0471,
      "step": 305125
    },
    {
      "epoch": 4.494042797602392,
      "grad_norm": 1.0854312181472778,
      "learning_rate": 2.2566223973503866e-06,
      "loss": 0.0429,
      "step": 305150
    },
    {
      "epoch": 4.494410980692479,
      "grad_norm": 1.5705701112747192,
      "learning_rate": 2.2549860253834023e-06,
      "loss": 0.0557,
      "step": 305175
    },
    {
      "epoch": 4.494779163782566,
      "grad_norm": 1.1281174421310425,
      "learning_rate": 2.2533496534164176e-06,
      "loss": 0.0488,
      "step": 305200
    },
    {
      "epoch": 4.495147346872653,
      "grad_norm": 1.310794472694397,
      "learning_rate": 2.251713281449433e-06,
      "loss": 0.0556,
      "step": 305225
    },
    {
      "epoch": 4.49551552996274,
      "grad_norm": 1.3496712446212769,
      "learning_rate": 2.2500769094824486e-06,
      "loss": 0.0484,
      "step": 305250
    },
    {
      "epoch": 4.495883713052827,
      "grad_norm": 1.17291259765625,
      "learning_rate": 2.248440537515464e-06,
      "loss": 0.0482,
      "step": 305275
    },
    {
      "epoch": 4.496251896142914,
      "grad_norm": 1.2383538484573364,
      "learning_rate": 2.2468041655484792e-06,
      "loss": 0.0461,
      "step": 305300
    },
    {
      "epoch": 4.496620079233001,
      "grad_norm": 1.3425499200820923,
      "learning_rate": 2.2451677935814945e-06,
      "loss": 0.052,
      "step": 305325
    },
    {
      "epoch": 4.496988262323088,
      "grad_norm": 1.3372963666915894,
      "learning_rate": 2.2435314216145103e-06,
      "loss": 0.0505,
      "step": 305350
    },
    {
      "epoch": 4.497356445413175,
      "grad_norm": 1.6998307704925537,
      "learning_rate": 2.2418950496475256e-06,
      "loss": 0.0523,
      "step": 305375
    },
    {
      "epoch": 4.497724628503262,
      "grad_norm": 1.1803414821624756,
      "learning_rate": 2.240258677680541e-06,
      "loss": 0.0473,
      "step": 305400
    },
    {
      "epoch": 4.498092811593349,
      "grad_norm": 1.2189404964447021,
      "learning_rate": 2.2386223057135566e-06,
      "loss": 0.0495,
      "step": 305425
    },
    {
      "epoch": 4.498460994683436,
      "grad_norm": 1.2284362316131592,
      "learning_rate": 2.236985933746572e-06,
      "loss": 0.0466,
      "step": 305450
    },
    {
      "epoch": 4.498829177773523,
      "grad_norm": 1.1205028295516968,
      "learning_rate": 2.2353495617795876e-06,
      "loss": 0.0505,
      "step": 305475
    },
    {
      "epoch": 4.499197360863611,
      "grad_norm": 1.914658784866333,
      "learning_rate": 2.233713189812603e-06,
      "loss": 0.0508,
      "step": 305500
    },
    {
      "epoch": 4.499565543953698,
      "grad_norm": 1.8325726985931396,
      "learning_rate": 2.2320768178456183e-06,
      "loss": 0.0575,
      "step": 305525
    },
    {
      "epoch": 4.499933727043785,
      "grad_norm": 1.3237731456756592,
      "learning_rate": 2.230440445878634e-06,
      "loss": 0.051,
      "step": 305550
    },
    {
      "epoch": 4.500301910133872,
      "grad_norm": 1.2312746047973633,
      "learning_rate": 2.2288040739116493e-06,
      "loss": 0.0554,
      "step": 305575
    },
    {
      "epoch": 4.500670093223959,
      "grad_norm": 1.255330204963684,
      "learning_rate": 2.2271677019446646e-06,
      "loss": 0.0512,
      "step": 305600
    },
    {
      "epoch": 4.501038276314046,
      "grad_norm": 1.2739551067352295,
      "learning_rate": 2.2255313299776803e-06,
      "loss": 0.0505,
      "step": 305625
    },
    {
      "epoch": 4.501406459404133,
      "grad_norm": 1.4140033721923828,
      "learning_rate": 2.2238949580106956e-06,
      "loss": 0.0506,
      "step": 305650
    },
    {
      "epoch": 4.50177464249422,
      "grad_norm": 1.2576498985290527,
      "learning_rate": 2.222258586043711e-06,
      "loss": 0.0467,
      "step": 305675
    },
    {
      "epoch": 4.502142825584307,
      "grad_norm": 1.3852546215057373,
      "learning_rate": 2.2206222140767262e-06,
      "loss": 0.0532,
      "step": 305700
    },
    {
      "epoch": 4.502511008674394,
      "grad_norm": 1.5580438375473022,
      "learning_rate": 2.218985842109742e-06,
      "loss": 0.0503,
      "step": 305725
    },
    {
      "epoch": 4.502879191764481,
      "grad_norm": 1.479324460029602,
      "learning_rate": 2.2173494701427573e-06,
      "loss": 0.053,
      "step": 305750
    },
    {
      "epoch": 4.503247374854568,
      "grad_norm": 1.2652708292007446,
      "learning_rate": 2.2157130981757726e-06,
      "loss": 0.0473,
      "step": 305775
    },
    {
      "epoch": 4.503615557944655,
      "grad_norm": 1.5438512563705444,
      "learning_rate": 2.2140767262087883e-06,
      "loss": 0.052,
      "step": 305800
    },
    {
      "epoch": 4.503983741034742,
      "grad_norm": 0.9680161476135254,
      "learning_rate": 2.2124403542418036e-06,
      "loss": 0.0478,
      "step": 305825
    },
    {
      "epoch": 4.504351924124829,
      "grad_norm": 1.074666976928711,
      "learning_rate": 2.210803982274819e-06,
      "loss": 0.0504,
      "step": 305850
    },
    {
      "epoch": 4.504720107214916,
      "grad_norm": 1.2749511003494263,
      "learning_rate": 2.209167610307834e-06,
      "loss": 0.0451,
      "step": 305875
    },
    {
      "epoch": 4.505088290305003,
      "grad_norm": 1.2340952157974243,
      "learning_rate": 2.20753123834085e-06,
      "loss": 0.0473,
      "step": 305900
    },
    {
      "epoch": 4.50545647339509,
      "grad_norm": 1.9128999710083008,
      "learning_rate": 2.2058948663738652e-06,
      "loss": 0.0535,
      "step": 305925
    },
    {
      "epoch": 4.505824656485177,
      "grad_norm": 1.1371002197265625,
      "learning_rate": 2.2042584944068805e-06,
      "loss": 0.0529,
      "step": 305950
    },
    {
      "epoch": 4.506192839575264,
      "grad_norm": 1.4185383319854736,
      "learning_rate": 2.2026221224398963e-06,
      "loss": 0.057,
      "step": 305975
    },
    {
      "epoch": 4.506561022665351,
      "grad_norm": 0.9934126734733582,
      "learning_rate": 2.200985750472912e-06,
      "loss": 0.0504,
      "step": 306000
    },
    {
      "epoch": 4.506929205755438,
      "grad_norm": 1.484820008277893,
      "learning_rate": 2.1993493785059273e-06,
      "loss": 0.047,
      "step": 306025
    },
    {
      "epoch": 4.507297388845525,
      "grad_norm": 0.9192208647727966,
      "learning_rate": 2.1977130065389426e-06,
      "loss": 0.048,
      "step": 306050
    },
    {
      "epoch": 4.507665571935612,
      "grad_norm": 1.314201831817627,
      "learning_rate": 2.196076634571958e-06,
      "loss": 0.0495,
      "step": 306075
    },
    {
      "epoch": 4.508033755025699,
      "grad_norm": 1.4820038080215454,
      "learning_rate": 2.1944402626049736e-06,
      "loss": 0.0489,
      "step": 306100
    },
    {
      "epoch": 4.508401938115786,
      "grad_norm": 1.5399411916732788,
      "learning_rate": 2.192803890637989e-06,
      "loss": 0.051,
      "step": 306125
    },
    {
      "epoch": 4.508770121205873,
      "grad_norm": 1.285320520401001,
      "learning_rate": 2.1911675186710042e-06,
      "loss": 0.052,
      "step": 306150
    },
    {
      "epoch": 4.50913830429596,
      "grad_norm": 1.5088833570480347,
      "learning_rate": 2.18953114670402e-06,
      "loss": 0.0555,
      "step": 306175
    },
    {
      "epoch": 4.509506487386047,
      "grad_norm": 1.1098202466964722,
      "learning_rate": 2.1878947747370353e-06,
      "loss": 0.0445,
      "step": 306200
    },
    {
      "epoch": 4.509874670476134,
      "grad_norm": 1.4007564783096313,
      "learning_rate": 2.1862584027700506e-06,
      "loss": 0.0514,
      "step": 306225
    },
    {
      "epoch": 4.510242853566221,
      "grad_norm": 1.8120425939559937,
      "learning_rate": 2.184622030803066e-06,
      "loss": 0.0543,
      "step": 306250
    },
    {
      "epoch": 4.510611036656308,
      "grad_norm": 1.083579421043396,
      "learning_rate": 2.1829856588360816e-06,
      "loss": 0.049,
      "step": 306275
    },
    {
      "epoch": 4.510979219746395,
      "grad_norm": 1.1269021034240723,
      "learning_rate": 2.181349286869097e-06,
      "loss": 0.0505,
      "step": 306300
    },
    {
      "epoch": 4.511347402836482,
      "grad_norm": 1.0835684537887573,
      "learning_rate": 2.1797129149021122e-06,
      "loss": 0.0485,
      "step": 306325
    },
    {
      "epoch": 4.511715585926569,
      "grad_norm": 1.3766403198242188,
      "learning_rate": 2.178076542935128e-06,
      "loss": 0.0505,
      "step": 306350
    },
    {
      "epoch": 4.512083769016657,
      "grad_norm": 1.0139336585998535,
      "learning_rate": 2.1764401709681433e-06,
      "loss": 0.0537,
      "step": 306375
    },
    {
      "epoch": 4.512451952106744,
      "grad_norm": 1.4512341022491455,
      "learning_rate": 2.1748037990011586e-06,
      "loss": 0.0545,
      "step": 306400
    },
    {
      "epoch": 4.512820135196831,
      "grad_norm": 1.534245491027832,
      "learning_rate": 2.173167427034174e-06,
      "loss": 0.049,
      "step": 306425
    },
    {
      "epoch": 4.513188318286918,
      "grad_norm": 1.7229063510894775,
      "learning_rate": 2.1715310550671896e-06,
      "loss": 0.0528,
      "step": 306450
    },
    {
      "epoch": 4.513556501377005,
      "grad_norm": 1.3052908182144165,
      "learning_rate": 2.1698946831002053e-06,
      "loss": 0.0502,
      "step": 306475
    },
    {
      "epoch": 4.513924684467092,
      "grad_norm": 1.6839303970336914,
      "learning_rate": 2.1682583111332206e-06,
      "loss": 0.0529,
      "step": 306500
    },
    {
      "epoch": 4.514292867557179,
      "grad_norm": 1.0376553535461426,
      "learning_rate": 2.166621939166236e-06,
      "loss": 0.0503,
      "step": 306525
    },
    {
      "epoch": 4.514661050647266,
      "grad_norm": 1.3496531248092651,
      "learning_rate": 2.1649855671992517e-06,
      "loss": 0.0544,
      "step": 306550
    },
    {
      "epoch": 4.515029233737353,
      "grad_norm": 1.4430419206619263,
      "learning_rate": 2.163349195232267e-06,
      "loss": 0.0483,
      "step": 306575
    },
    {
      "epoch": 4.51539741682744,
      "grad_norm": 1.4553323984146118,
      "learning_rate": 2.1617128232652823e-06,
      "loss": 0.0554,
      "step": 306600
    },
    {
      "epoch": 4.515765599917527,
      "grad_norm": 0.8176844716072083,
      "learning_rate": 2.1600764512982976e-06,
      "loss": 0.0571,
      "step": 306625
    },
    {
      "epoch": 4.516133783007614,
      "grad_norm": 1.2480665445327759,
      "learning_rate": 2.1584400793313133e-06,
      "loss": 0.0517,
      "step": 306650
    },
    {
      "epoch": 4.516501966097701,
      "grad_norm": 1.301805853843689,
      "learning_rate": 2.1568037073643286e-06,
      "loss": 0.0549,
      "step": 306675
    },
    {
      "epoch": 4.516870149187788,
      "grad_norm": 1.3430217504501343,
      "learning_rate": 2.155167335397344e-06,
      "loss": 0.0488,
      "step": 306700
    },
    {
      "epoch": 4.517238332277875,
      "grad_norm": 0.9315279126167297,
      "learning_rate": 2.1535309634303596e-06,
      "loss": 0.0505,
      "step": 306725
    },
    {
      "epoch": 4.517606515367962,
      "grad_norm": 1.353087306022644,
      "learning_rate": 2.1519600463420544e-06,
      "loss": 0.0482,
      "step": 306750
    },
    {
      "epoch": 4.517974698458049,
      "grad_norm": 1.4864869117736816,
      "learning_rate": 2.1503236743750697e-06,
      "loss": 0.0525,
      "step": 306775
    },
    {
      "epoch": 4.518342881548136,
      "grad_norm": 1.2863086462020874,
      "learning_rate": 2.148687302408085e-06,
      "loss": 0.0533,
      "step": 306800
    },
    {
      "epoch": 4.518711064638223,
      "grad_norm": 0.984040379524231,
      "learning_rate": 2.1470509304411007e-06,
      "loss": 0.0466,
      "step": 306825
    },
    {
      "epoch": 4.51907924772831,
      "grad_norm": 1.4597172737121582,
      "learning_rate": 2.145414558474116e-06,
      "loss": 0.0493,
      "step": 306850
    },
    {
      "epoch": 4.519447430818397,
      "grad_norm": 1.468724012374878,
      "learning_rate": 2.1437781865071313e-06,
      "loss": 0.0567,
      "step": 306875
    },
    {
      "epoch": 4.519815613908484,
      "grad_norm": 0.874403178691864,
      "learning_rate": 2.1421418145401466e-06,
      "loss": 0.0518,
      "step": 306900
    },
    {
      "epoch": 4.520183796998571,
      "grad_norm": 0.9855050444602966,
      "learning_rate": 2.1405054425731623e-06,
      "loss": 0.0489,
      "step": 306925
    },
    {
      "epoch": 4.520551980088658,
      "grad_norm": 1.2628121376037598,
      "learning_rate": 2.138869070606178e-06,
      "loss": 0.0534,
      "step": 306950
    },
    {
      "epoch": 4.5209201631787455,
      "grad_norm": 1.2815665006637573,
      "learning_rate": 2.1372326986391934e-06,
      "loss": 0.053,
      "step": 306975
    },
    {
      "epoch": 4.5212883462688325,
      "grad_norm": 1.0003927946090698,
      "learning_rate": 2.1355963266722087e-06,
      "loss": 0.0505,
      "step": 307000
    },
    {
      "epoch": 4.5216565293589195,
      "grad_norm": 1.924319863319397,
      "learning_rate": 2.1339599547052244e-06,
      "loss": 0.0541,
      "step": 307025
    },
    {
      "epoch": 4.5220247124490065,
      "grad_norm": 1.4959354400634766,
      "learning_rate": 2.1323235827382397e-06,
      "loss": 0.0501,
      "step": 307050
    },
    {
      "epoch": 4.5223928955390935,
      "grad_norm": 0.8307550549507141,
      "learning_rate": 2.130687210771255e-06,
      "loss": 0.0583,
      "step": 307075
    },
    {
      "epoch": 4.5227610786291805,
      "grad_norm": 1.1998952627182007,
      "learning_rate": 2.1290508388042703e-06,
      "loss": 0.0502,
      "step": 307100
    },
    {
      "epoch": 4.5231292617192675,
      "grad_norm": 1.3657691478729248,
      "learning_rate": 2.127414466837286e-06,
      "loss": 0.0495,
      "step": 307125
    },
    {
      "epoch": 4.523497444809355,
      "grad_norm": 0.9390334486961365,
      "learning_rate": 2.1257780948703014e-06,
      "loss": 0.0466,
      "step": 307150
    },
    {
      "epoch": 4.523865627899442,
      "grad_norm": 1.3445976972579956,
      "learning_rate": 2.1241417229033167e-06,
      "loss": 0.0578,
      "step": 307175
    },
    {
      "epoch": 4.524233810989529,
      "grad_norm": 1.182094931602478,
      "learning_rate": 2.1225053509363324e-06,
      "loss": 0.0465,
      "step": 307200
    },
    {
      "epoch": 4.524601994079616,
      "grad_norm": 0.7793055772781372,
      "learning_rate": 2.1208689789693477e-06,
      "loss": 0.0459,
      "step": 307225
    },
    {
      "epoch": 4.524970177169703,
      "grad_norm": 1.6466139554977417,
      "learning_rate": 2.119232607002363e-06,
      "loss": 0.0528,
      "step": 307250
    },
    {
      "epoch": 4.52533836025979,
      "grad_norm": 1.24486243724823,
      "learning_rate": 2.1175962350353783e-06,
      "loss": 0.0516,
      "step": 307275
    },
    {
      "epoch": 4.525706543349877,
      "grad_norm": 1.2348495721817017,
      "learning_rate": 2.115959863068394e-06,
      "loss": 0.0505,
      "step": 307300
    },
    {
      "epoch": 4.526074726439964,
      "grad_norm": 0.8380063772201538,
      "learning_rate": 2.1143234911014093e-06,
      "loss": 0.053,
      "step": 307325
    },
    {
      "epoch": 4.526442909530051,
      "grad_norm": 1.3594605922698975,
      "learning_rate": 2.1126871191344246e-06,
      "loss": 0.0558,
      "step": 307350
    },
    {
      "epoch": 4.526811092620138,
      "grad_norm": 1.736283540725708,
      "learning_rate": 2.1110507471674404e-06,
      "loss": 0.0519,
      "step": 307375
    },
    {
      "epoch": 4.527179275710225,
      "grad_norm": 1.2260088920593262,
      "learning_rate": 2.1094143752004557e-06,
      "loss": 0.0574,
      "step": 307400
    },
    {
      "epoch": 4.527547458800312,
      "grad_norm": 1.6989027261734009,
      "learning_rate": 2.1077780032334714e-06,
      "loss": 0.0501,
      "step": 307425
    },
    {
      "epoch": 4.5279156418904,
      "grad_norm": 1.137641191482544,
      "learning_rate": 2.1061416312664867e-06,
      "loss": 0.0511,
      "step": 307450
    },
    {
      "epoch": 4.528283824980487,
      "grad_norm": 1.1760953664779663,
      "learning_rate": 2.104505259299502e-06,
      "loss": 0.0488,
      "step": 307475
    },
    {
      "epoch": 4.528652008070574,
      "grad_norm": 1.7314229011535645,
      "learning_rate": 2.1028688873325177e-06,
      "loss": 0.0505,
      "step": 307500
    },
    {
      "epoch": 4.529020191160661,
      "grad_norm": 1.4109761714935303,
      "learning_rate": 2.101232515365533e-06,
      "loss": 0.0491,
      "step": 307525
    },
    {
      "epoch": 4.529388374250748,
      "grad_norm": 0.888072669506073,
      "learning_rate": 2.0995961433985483e-06,
      "loss": 0.0465,
      "step": 307550
    },
    {
      "epoch": 4.529756557340835,
      "grad_norm": 1.5533137321472168,
      "learning_rate": 2.097959771431564e-06,
      "loss": 0.0523,
      "step": 307575
    },
    {
      "epoch": 4.530124740430922,
      "grad_norm": 1.5203653573989868,
      "learning_rate": 2.0963233994645794e-06,
      "loss": 0.0531,
      "step": 307600
    },
    {
      "epoch": 4.530492923521009,
      "grad_norm": 1.265055775642395,
      "learning_rate": 2.0946870274975947e-06,
      "loss": 0.0533,
      "step": 307625
    },
    {
      "epoch": 4.530861106611096,
      "grad_norm": 1.1532315015792847,
      "learning_rate": 2.09305065553061e-06,
      "loss": 0.0437,
      "step": 307650
    },
    {
      "epoch": 4.531229289701183,
      "grad_norm": 1.476191520690918,
      "learning_rate": 2.0914142835636257e-06,
      "loss": 0.048,
      "step": 307675
    },
    {
      "epoch": 4.53159747279127,
      "grad_norm": 1.0593379735946655,
      "learning_rate": 2.089777911596641e-06,
      "loss": 0.0525,
      "step": 307700
    },
    {
      "epoch": 4.531965655881357,
      "grad_norm": 1.4708608388900757,
      "learning_rate": 2.0881415396296563e-06,
      "loss": 0.051,
      "step": 307725
    },
    {
      "epoch": 4.532333838971444,
      "grad_norm": 1.4360496997833252,
      "learning_rate": 2.086505167662672e-06,
      "loss": 0.0493,
      "step": 307750
    },
    {
      "epoch": 4.532702022061531,
      "grad_norm": 0.9608443975448608,
      "learning_rate": 2.0848687956956874e-06,
      "loss": 0.0536,
      "step": 307775
    },
    {
      "epoch": 4.533070205151618,
      "grad_norm": 1.5330548286437988,
      "learning_rate": 2.0832324237287027e-06,
      "loss": 0.0506,
      "step": 307800
    },
    {
      "epoch": 4.533438388241705,
      "grad_norm": 0.862598180770874,
      "learning_rate": 2.081596051761718e-06,
      "loss": 0.047,
      "step": 307825
    },
    {
      "epoch": 4.533806571331792,
      "grad_norm": 1.1573472023010254,
      "learning_rate": 2.0799596797947337e-06,
      "loss": 0.045,
      "step": 307850
    },
    {
      "epoch": 4.534174754421879,
      "grad_norm": 1.48971688747406,
      "learning_rate": 2.078323307827749e-06,
      "loss": 0.0482,
      "step": 307875
    },
    {
      "epoch": 4.534542937511966,
      "grad_norm": 1.1666826009750366,
      "learning_rate": 2.0766869358607643e-06,
      "loss": 0.0452,
      "step": 307900
    },
    {
      "epoch": 4.534911120602053,
      "grad_norm": 1.4414671659469604,
      "learning_rate": 2.07505056389378e-06,
      "loss": 0.0543,
      "step": 307925
    },
    {
      "epoch": 4.53527930369214,
      "grad_norm": 1.4044976234436035,
      "learning_rate": 2.0734141919267958e-06,
      "loss": 0.0586,
      "step": 307950
    },
    {
      "epoch": 4.535647486782227,
      "grad_norm": 1.4503514766693115,
      "learning_rate": 2.071777819959811e-06,
      "loss": 0.0492,
      "step": 307975
    },
    {
      "epoch": 4.536015669872314,
      "grad_norm": 1.3328460454940796,
      "learning_rate": 2.0701414479928264e-06,
      "loss": 0.0436,
      "step": 308000
    },
    {
      "epoch": 4.536383852962401,
      "grad_norm": 1.34400475025177,
      "learning_rate": 2.0685050760258417e-06,
      "loss": 0.056,
      "step": 308025
    },
    {
      "epoch": 4.536752036052488,
      "grad_norm": 1.5616716146469116,
      "learning_rate": 2.0668687040588574e-06,
      "loss": 0.0573,
      "step": 308050
    },
    {
      "epoch": 4.537120219142575,
      "grad_norm": 0.8876332640647888,
      "learning_rate": 2.0652323320918727e-06,
      "loss": 0.0512,
      "step": 308075
    },
    {
      "epoch": 4.537488402232662,
      "grad_norm": 0.9860138893127441,
      "learning_rate": 2.063595960124888e-06,
      "loss": 0.0555,
      "step": 308100
    },
    {
      "epoch": 4.537856585322749,
      "grad_norm": 1.431259036064148,
      "learning_rate": 2.0619595881579037e-06,
      "loss": 0.05,
      "step": 308125
    },
    {
      "epoch": 4.538224768412836,
      "grad_norm": 1.095428228378296,
      "learning_rate": 2.060323216190919e-06,
      "loss": 0.0485,
      "step": 308150
    },
    {
      "epoch": 4.538592951502923,
      "grad_norm": 1.5755724906921387,
      "learning_rate": 2.0586868442239343e-06,
      "loss": 0.0533,
      "step": 308175
    },
    {
      "epoch": 4.53896113459301,
      "grad_norm": 2.0550174713134766,
      "learning_rate": 2.0570504722569496e-06,
      "loss": 0.052,
      "step": 308200
    },
    {
      "epoch": 4.539329317683097,
      "grad_norm": 1.2741310596466064,
      "learning_rate": 2.0554141002899654e-06,
      "loss": 0.0529,
      "step": 308225
    },
    {
      "epoch": 4.539697500773184,
      "grad_norm": 0.894555926322937,
      "learning_rate": 2.0537777283229807e-06,
      "loss": 0.0492,
      "step": 308250
    },
    {
      "epoch": 4.540065683863271,
      "grad_norm": 1.6967883110046387,
      "learning_rate": 2.052141356355996e-06,
      "loss": 0.0483,
      "step": 308275
    },
    {
      "epoch": 4.540433866953359,
      "grad_norm": 1.0774853229522705,
      "learning_rate": 2.0505049843890117e-06,
      "loss": 0.0483,
      "step": 308300
    },
    {
      "epoch": 4.540802050043446,
      "grad_norm": 1.6280452013015747,
      "learning_rate": 2.048868612422027e-06,
      "loss": 0.056,
      "step": 308325
    },
    {
      "epoch": 4.541170233133533,
      "grad_norm": 1.453758955001831,
      "learning_rate": 2.0472322404550423e-06,
      "loss": 0.0504,
      "step": 308350
    },
    {
      "epoch": 4.54153841622362,
      "grad_norm": 1.55021333694458,
      "learning_rate": 2.0455958684880576e-06,
      "loss": 0.0469,
      "step": 308375
    },
    {
      "epoch": 4.541906599313707,
      "grad_norm": 1.4157837629318237,
      "learning_rate": 2.0439594965210734e-06,
      "loss": 0.0521,
      "step": 308400
    },
    {
      "epoch": 4.542274782403794,
      "grad_norm": 0.7594125270843506,
      "learning_rate": 2.042323124554089e-06,
      "loss": 0.0511,
      "step": 308425
    },
    {
      "epoch": 4.542642965493881,
      "grad_norm": 1.248920202255249,
      "learning_rate": 2.0406867525871044e-06,
      "loss": 0.0458,
      "step": 308450
    },
    {
      "epoch": 4.543011148583968,
      "grad_norm": 1.3545526266098022,
      "learning_rate": 2.0390503806201197e-06,
      "loss": 0.0497,
      "step": 308475
    },
    {
      "epoch": 4.543379331674055,
      "grad_norm": 0.7493758201599121,
      "learning_rate": 2.0374140086531354e-06,
      "loss": 0.0539,
      "step": 308500
    },
    {
      "epoch": 4.543747514764142,
      "grad_norm": 1.0456050634384155,
      "learning_rate": 2.0357776366861507e-06,
      "loss": 0.054,
      "step": 308525
    },
    {
      "epoch": 4.544115697854229,
      "grad_norm": 1.1004092693328857,
      "learning_rate": 2.034141264719166e-06,
      "loss": 0.0511,
      "step": 308550
    },
    {
      "epoch": 4.544483880944316,
      "grad_norm": 1.1395137310028076,
      "learning_rate": 2.0325048927521813e-06,
      "loss": 0.0467,
      "step": 308575
    },
    {
      "epoch": 4.544852064034403,
      "grad_norm": 1.5351648330688477,
      "learning_rate": 2.030868520785197e-06,
      "loss": 0.0546,
      "step": 308600
    },
    {
      "epoch": 4.54522024712449,
      "grad_norm": 1.2298682928085327,
      "learning_rate": 2.0292321488182124e-06,
      "loss": 0.0527,
      "step": 308625
    },
    {
      "epoch": 4.545588430214577,
      "grad_norm": 1.3381637334823608,
      "learning_rate": 2.0275957768512277e-06,
      "loss": 0.053,
      "step": 308650
    },
    {
      "epoch": 4.545956613304664,
      "grad_norm": 0.8341094255447388,
      "learning_rate": 2.0259594048842434e-06,
      "loss": 0.0503,
      "step": 308675
    },
    {
      "epoch": 4.546324796394751,
      "grad_norm": 1.2205511331558228,
      "learning_rate": 2.0243230329172587e-06,
      "loss": 0.0514,
      "step": 308700
    },
    {
      "epoch": 4.546692979484838,
      "grad_norm": 0.9575676918029785,
      "learning_rate": 2.022686660950274e-06,
      "loss": 0.05,
      "step": 308725
    },
    {
      "epoch": 4.547061162574925,
      "grad_norm": 1.0823307037353516,
      "learning_rate": 2.0210502889832893e-06,
      "loss": 0.0616,
      "step": 308750
    },
    {
      "epoch": 4.547429345665012,
      "grad_norm": 1.2561061382293701,
      "learning_rate": 2.019413917016305e-06,
      "loss": 0.0602,
      "step": 308775
    },
    {
      "epoch": 4.547797528755099,
      "grad_norm": 1.1245801448822021,
      "learning_rate": 2.0177775450493203e-06,
      "loss": 0.0505,
      "step": 308800
    },
    {
      "epoch": 4.548165711845186,
      "grad_norm": 1.0400805473327637,
      "learning_rate": 2.0161411730823356e-06,
      "loss": 0.0462,
      "step": 308825
    },
    {
      "epoch": 4.548533894935273,
      "grad_norm": 1.3343368768692017,
      "learning_rate": 2.0145048011153514e-06,
      "loss": 0.0594,
      "step": 308850
    },
    {
      "epoch": 4.54890207802536,
      "grad_norm": 1.1459853649139404,
      "learning_rate": 2.0128684291483667e-06,
      "loss": 0.0561,
      "step": 308875
    },
    {
      "epoch": 4.549270261115447,
      "grad_norm": 1.415501594543457,
      "learning_rate": 2.0112320571813824e-06,
      "loss": 0.052,
      "step": 308900
    },
    {
      "epoch": 4.549638444205534,
      "grad_norm": 1.3437832593917847,
      "learning_rate": 2.0095956852143977e-06,
      "loss": 0.0535,
      "step": 308925
    },
    {
      "epoch": 4.550006627295621,
      "grad_norm": 1.5965673923492432,
      "learning_rate": 2.007959313247413e-06,
      "loss": 0.0482,
      "step": 308950
    },
    {
      "epoch": 4.5503748103857085,
      "grad_norm": 1.0317606925964355,
      "learning_rate": 2.0063229412804287e-06,
      "loss": 0.043,
      "step": 308975
    },
    {
      "epoch": 4.5507429934757955,
      "grad_norm": 1.7104601860046387,
      "learning_rate": 2.004686569313444e-06,
      "loss": 0.0512,
      "step": 309000
    },
    {
      "epoch": 4.5511111765658825,
      "grad_norm": 1.38189697265625,
      "learning_rate": 2.0030501973464594e-06,
      "loss": 0.0443,
      "step": 309025
    },
    {
      "epoch": 4.5514793596559695,
      "grad_norm": 1.3474329710006714,
      "learning_rate": 2.001413825379475e-06,
      "loss": 0.0525,
      "step": 309050
    },
    {
      "epoch": 4.5518475427460565,
      "grad_norm": 1.881171703338623,
      "learning_rate": 1.9997774534124904e-06,
      "loss": 0.0546,
      "step": 309075
    },
    {
      "epoch": 4.5522157258361435,
      "grad_norm": 1.063180923461914,
      "learning_rate": 1.9981410814455057e-06,
      "loss": 0.0549,
      "step": 309100
    },
    {
      "epoch": 4.5525839089262305,
      "grad_norm": 1.3530104160308838,
      "learning_rate": 1.996504709478521e-06,
      "loss": 0.0572,
      "step": 309125
    },
    {
      "epoch": 4.552952092016318,
      "grad_norm": 1.5687769651412964,
      "learning_rate": 1.9948683375115367e-06,
      "loss": 0.05,
      "step": 309150
    },
    {
      "epoch": 4.553320275106405,
      "grad_norm": 1.1460498571395874,
      "learning_rate": 1.993231965544552e-06,
      "loss": 0.0505,
      "step": 309175
    },
    {
      "epoch": 4.553688458196492,
      "grad_norm": 1.2409263849258423,
      "learning_rate": 1.9915955935775673e-06,
      "loss": 0.0444,
      "step": 309200
    },
    {
      "epoch": 4.554056641286579,
      "grad_norm": 1.58135986328125,
      "learning_rate": 1.989959221610583e-06,
      "loss": 0.0545,
      "step": 309225
    },
    {
      "epoch": 4.554424824376666,
      "grad_norm": 0.8890162706375122,
      "learning_rate": 1.9883228496435984e-06,
      "loss": 0.0518,
      "step": 309250
    },
    {
      "epoch": 4.554793007466753,
      "grad_norm": 1.0495911836624146,
      "learning_rate": 1.9866864776766137e-06,
      "loss": 0.05,
      "step": 309275
    },
    {
      "epoch": 4.55516119055684,
      "grad_norm": 1.4100531339645386,
      "learning_rate": 1.985050105709629e-06,
      "loss": 0.0538,
      "step": 309300
    },
    {
      "epoch": 4.555529373646927,
      "grad_norm": 1.3735814094543457,
      "learning_rate": 1.9834137337426447e-06,
      "loss": 0.0507,
      "step": 309325
    },
    {
      "epoch": 4.555897556737014,
      "grad_norm": 1.9199820756912231,
      "learning_rate": 1.98177736177566e-06,
      "loss": 0.0556,
      "step": 309350
    },
    {
      "epoch": 4.556265739827102,
      "grad_norm": 1.5116608142852783,
      "learning_rate": 1.9801409898086753e-06,
      "loss": 0.0593,
      "step": 309375
    },
    {
      "epoch": 4.556633922917189,
      "grad_norm": 1.0083881616592407,
      "learning_rate": 1.978504617841691e-06,
      "loss": 0.0515,
      "step": 309400
    },
    {
      "epoch": 4.557002106007276,
      "grad_norm": 1.572373628616333,
      "learning_rate": 1.9768682458747063e-06,
      "loss": 0.0549,
      "step": 309425
    },
    {
      "epoch": 4.557370289097363,
      "grad_norm": 1.614080548286438,
      "learning_rate": 1.975231873907722e-06,
      "loss": 0.0516,
      "step": 309450
    },
    {
      "epoch": 4.55773847218745,
      "grad_norm": 1.8171714544296265,
      "learning_rate": 1.9735955019407374e-06,
      "loss": 0.0494,
      "step": 309475
    },
    {
      "epoch": 4.558106655277537,
      "grad_norm": 1.7256240844726562,
      "learning_rate": 1.9719591299737527e-06,
      "loss": 0.0553,
      "step": 309500
    },
    {
      "epoch": 4.558474838367624,
      "grad_norm": 1.19894278049469,
      "learning_rate": 1.9703227580067684e-06,
      "loss": 0.0491,
      "step": 309525
    },
    {
      "epoch": 4.558843021457711,
      "grad_norm": 1.234932541847229,
      "learning_rate": 1.9686863860397837e-06,
      "loss": 0.0449,
      "step": 309550
    },
    {
      "epoch": 4.559211204547798,
      "grad_norm": 1.6571643352508545,
      "learning_rate": 1.967050014072799e-06,
      "loss": 0.0533,
      "step": 309575
    },
    {
      "epoch": 4.559579387637885,
      "grad_norm": 1.6973224878311157,
      "learning_rate": 1.9654136421058147e-06,
      "loss": 0.0542,
      "step": 309600
    },
    {
      "epoch": 4.559947570727972,
      "grad_norm": 0.9314064383506775,
      "learning_rate": 1.96377727013883e-06,
      "loss": 0.0488,
      "step": 309625
    },
    {
      "epoch": 4.560315753818059,
      "grad_norm": 1.58635413646698,
      "learning_rate": 1.9621408981718454e-06,
      "loss": 0.047,
      "step": 309650
    },
    {
      "epoch": 4.560683936908146,
      "grad_norm": 1.5457500219345093,
      "learning_rate": 1.9605045262048607e-06,
      "loss": 0.0546,
      "step": 309675
    },
    {
      "epoch": 4.561052119998233,
      "grad_norm": 1.4691550731658936,
      "learning_rate": 1.9588681542378764e-06,
      "loss": 0.0492,
      "step": 309700
    },
    {
      "epoch": 4.56142030308832,
      "grad_norm": 1.3959052562713623,
      "learning_rate": 1.9572317822708917e-06,
      "loss": 0.054,
      "step": 309725
    },
    {
      "epoch": 4.561788486178407,
      "grad_norm": 1.1163315773010254,
      "learning_rate": 1.955595410303907e-06,
      "loss": 0.0497,
      "step": 309750
    },
    {
      "epoch": 4.562156669268494,
      "grad_norm": 1.162321925163269,
      "learning_rate": 1.9539590383369227e-06,
      "loss": 0.0563,
      "step": 309775
    },
    {
      "epoch": 4.562524852358581,
      "grad_norm": 0.9975666403770447,
      "learning_rate": 1.952322666369938e-06,
      "loss": 0.0542,
      "step": 309800
    },
    {
      "epoch": 4.562893035448668,
      "grad_norm": 1.577072024345398,
      "learning_rate": 1.9507517492816328e-06,
      "loss": 0.0551,
      "step": 309825
    },
    {
      "epoch": 4.563261218538755,
      "grad_norm": 1.5585591793060303,
      "learning_rate": 1.949115377314648e-06,
      "loss": 0.0524,
      "step": 309850
    },
    {
      "epoch": 4.563629401628842,
      "grad_norm": 1.5018388032913208,
      "learning_rate": 1.9474790053476638e-06,
      "loss": 0.0476,
      "step": 309875
    },
    {
      "epoch": 4.563997584718929,
      "grad_norm": 1.6291329860687256,
      "learning_rate": 1.9458426333806795e-06,
      "loss": 0.0526,
      "step": 309900
    },
    {
      "epoch": 4.564365767809016,
      "grad_norm": 1.3654611110687256,
      "learning_rate": 1.944206261413695e-06,
      "loss": 0.0521,
      "step": 309925
    },
    {
      "epoch": 4.564733950899103,
      "grad_norm": 1.7379506826400757,
      "learning_rate": 1.94256988944671e-06,
      "loss": 0.0492,
      "step": 309950
    },
    {
      "epoch": 4.56510213398919,
      "grad_norm": 2.03979754447937,
      "learning_rate": 1.9409335174797254e-06,
      "loss": 0.0497,
      "step": 309975
    },
    {
      "epoch": 4.565470317079277,
      "grad_norm": 1.3213950395584106,
      "learning_rate": 1.939297145512741e-06,
      "loss": 0.0525,
      "step": 310000
    },
    {
      "epoch": 4.565838500169364,
      "grad_norm": 1.017162561416626,
      "learning_rate": 1.9376607735457565e-06,
      "loss": 0.0571,
      "step": 310025
    },
    {
      "epoch": 4.566206683259451,
      "grad_norm": 1.3367708921432495,
      "learning_rate": 1.9360244015787718e-06,
      "loss": 0.0522,
      "step": 310050
    },
    {
      "epoch": 4.566574866349538,
      "grad_norm": 1.4543776512145996,
      "learning_rate": 1.9343880296117875e-06,
      "loss": 0.0469,
      "step": 310075
    },
    {
      "epoch": 4.566943049439625,
      "grad_norm": 1.4850057363510132,
      "learning_rate": 1.932751657644803e-06,
      "loss": 0.0544,
      "step": 310100
    },
    {
      "epoch": 4.567311232529712,
      "grad_norm": 1.5761196613311768,
      "learning_rate": 1.931115285677818e-06,
      "loss": 0.0495,
      "step": 310125
    },
    {
      "epoch": 4.567679415619799,
      "grad_norm": 1.6672794818878174,
      "learning_rate": 1.9294789137108334e-06,
      "loss": 0.0535,
      "step": 310150
    },
    {
      "epoch": 4.568047598709886,
      "grad_norm": 1.3041654825210571,
      "learning_rate": 1.927842541743849e-06,
      "loss": 0.051,
      "step": 310175
    },
    {
      "epoch": 4.568415781799974,
      "grad_norm": 1.3045835494995117,
      "learning_rate": 1.9262061697768644e-06,
      "loss": 0.0539,
      "step": 310200
    },
    {
      "epoch": 4.568783964890061,
      "grad_norm": 1.269605040550232,
      "learning_rate": 1.9245697978098797e-06,
      "loss": 0.0495,
      "step": 310225
    },
    {
      "epoch": 4.569152147980148,
      "grad_norm": 1.4230904579162598,
      "learning_rate": 1.9229334258428955e-06,
      "loss": 0.0528,
      "step": 310250
    },
    {
      "epoch": 4.569520331070235,
      "grad_norm": 1.7751332521438599,
      "learning_rate": 1.9212970538759108e-06,
      "loss": 0.0526,
      "step": 310275
    },
    {
      "epoch": 4.569888514160322,
      "grad_norm": 1.475242257118225,
      "learning_rate": 1.919660681908926e-06,
      "loss": 0.0535,
      "step": 310300
    },
    {
      "epoch": 4.570256697250409,
      "grad_norm": 1.2149518728256226,
      "learning_rate": 1.9180243099419414e-06,
      "loss": 0.0488,
      "step": 310325
    },
    {
      "epoch": 4.570624880340496,
      "grad_norm": 1.1104406118392944,
      "learning_rate": 1.916387937974957e-06,
      "loss": 0.0475,
      "step": 310350
    },
    {
      "epoch": 4.570993063430583,
      "grad_norm": 1.0464363098144531,
      "learning_rate": 1.914751566007973e-06,
      "loss": 0.0554,
      "step": 310375
    },
    {
      "epoch": 4.57136124652067,
      "grad_norm": 1.5187437534332275,
      "learning_rate": 1.913115194040988e-06,
      "loss": 0.0512,
      "step": 310400
    },
    {
      "epoch": 4.571729429610757,
      "grad_norm": 1.0319486856460571,
      "learning_rate": 1.9114788220740034e-06,
      "loss": 0.0488,
      "step": 310425
    },
    {
      "epoch": 4.572097612700844,
      "grad_norm": 1.4134297370910645,
      "learning_rate": 1.909842450107019e-06,
      "loss": 0.0534,
      "step": 310450
    },
    {
      "epoch": 4.572465795790931,
      "grad_norm": 1.3182697296142578,
      "learning_rate": 1.9082060781400345e-06,
      "loss": 0.0514,
      "step": 310475
    },
    {
      "epoch": 4.572833978881018,
      "grad_norm": 1.3037372827529907,
      "learning_rate": 1.9065697061730498e-06,
      "loss": 0.0478,
      "step": 310500
    },
    {
      "epoch": 4.573202161971105,
      "grad_norm": 1.3757737874984741,
      "learning_rate": 1.9049333342060653e-06,
      "loss": 0.0555,
      "step": 310525
    },
    {
      "epoch": 4.573570345061192,
      "grad_norm": 1.1345593929290771,
      "learning_rate": 1.9032969622390806e-06,
      "loss": 0.0484,
      "step": 310550
    },
    {
      "epoch": 4.573938528151279,
      "grad_norm": 1.1313515901565552,
      "learning_rate": 1.9016605902720961e-06,
      "loss": 0.0552,
      "step": 310575
    },
    {
      "epoch": 4.574306711241366,
      "grad_norm": 1.2449558973312378,
      "learning_rate": 1.9000242183051114e-06,
      "loss": 0.0585,
      "step": 310600
    },
    {
      "epoch": 4.574674894331453,
      "grad_norm": 1.5524110794067383,
      "learning_rate": 1.898387846338127e-06,
      "loss": 0.048,
      "step": 310625
    },
    {
      "epoch": 4.57504307742154,
      "grad_norm": 1.2277108430862427,
      "learning_rate": 1.8967514743711425e-06,
      "loss": 0.0526,
      "step": 310650
    },
    {
      "epoch": 4.575411260511627,
      "grad_norm": 1.7272179126739502,
      "learning_rate": 1.8951151024041578e-06,
      "loss": 0.0497,
      "step": 310675
    },
    {
      "epoch": 4.575779443601714,
      "grad_norm": 1.518348217010498,
      "learning_rate": 1.8934787304371733e-06,
      "loss": 0.0507,
      "step": 310700
    },
    {
      "epoch": 4.576147626691801,
      "grad_norm": 1.2473044395446777,
      "learning_rate": 1.8918423584701886e-06,
      "loss": 0.0571,
      "step": 310725
    },
    {
      "epoch": 4.576515809781888,
      "grad_norm": 1.1358023881912231,
      "learning_rate": 1.890205986503204e-06,
      "loss": 0.0548,
      "step": 310750
    },
    {
      "epoch": 4.576883992871975,
      "grad_norm": 1.5578036308288574,
      "learning_rate": 1.8885696145362196e-06,
      "loss": 0.058,
      "step": 310775
    },
    {
      "epoch": 4.577252175962062,
      "grad_norm": 1.411180853843689,
      "learning_rate": 1.886933242569235e-06,
      "loss": 0.0502,
      "step": 310800
    },
    {
      "epoch": 4.577620359052149,
      "grad_norm": 1.2316758632659912,
      "learning_rate": 1.8852968706022504e-06,
      "loss": 0.0485,
      "step": 310825
    },
    {
      "epoch": 4.577988542142236,
      "grad_norm": 1.708455204963684,
      "learning_rate": 1.8836604986352662e-06,
      "loss": 0.0511,
      "step": 310850
    },
    {
      "epoch": 4.578356725232323,
      "grad_norm": 1.1811246871948242,
      "learning_rate": 1.8820241266682815e-06,
      "loss": 0.0533,
      "step": 310875
    },
    {
      "epoch": 4.57872490832241,
      "grad_norm": 1.598293423652649,
      "learning_rate": 1.880387754701297e-06,
      "loss": 0.0493,
      "step": 310900
    },
    {
      "epoch": 4.579093091412497,
      "grad_norm": 1.7547820806503296,
      "learning_rate": 1.8787513827343123e-06,
      "loss": 0.0554,
      "step": 310925
    },
    {
      "epoch": 4.5794612745025844,
      "grad_norm": 1.325798749923706,
      "learning_rate": 1.8771150107673278e-06,
      "loss": 0.0487,
      "step": 310950
    },
    {
      "epoch": 4.5798294575926715,
      "grad_norm": 1.6396585702896118,
      "learning_rate": 1.8754786388003431e-06,
      "loss": 0.0553,
      "step": 310975
    },
    {
      "epoch": 4.5801976406827585,
      "grad_norm": 1.2691404819488525,
      "learning_rate": 1.8738422668333586e-06,
      "loss": 0.0538,
      "step": 311000
    },
    {
      "epoch": 4.5805658237728455,
      "grad_norm": 1.3830153942108154,
      "learning_rate": 1.8722058948663741e-06,
      "loss": 0.0483,
      "step": 311025
    },
    {
      "epoch": 4.5809340068629325,
      "grad_norm": 1.3440278768539429,
      "learning_rate": 1.8705695228993894e-06,
      "loss": 0.05,
      "step": 311050
    },
    {
      "epoch": 4.5813021899530195,
      "grad_norm": 1.1792877912521362,
      "learning_rate": 1.868933150932405e-06,
      "loss": 0.0523,
      "step": 311075
    },
    {
      "epoch": 4.5816703730431065,
      "grad_norm": 1.055132508277893,
      "learning_rate": 1.8672967789654203e-06,
      "loss": 0.0522,
      "step": 311100
    },
    {
      "epoch": 4.5820385561331936,
      "grad_norm": 1.2314695119857788,
      "learning_rate": 1.8656604069984358e-06,
      "loss": 0.0533,
      "step": 311125
    },
    {
      "epoch": 4.582406739223281,
      "grad_norm": 1.379610538482666,
      "learning_rate": 1.864024035031451e-06,
      "loss": 0.0503,
      "step": 311150
    },
    {
      "epoch": 4.582774922313368,
      "grad_norm": 1.4142709970474243,
      "learning_rate": 1.8623876630644666e-06,
      "loss": 0.0496,
      "step": 311175
    },
    {
      "epoch": 4.583143105403455,
      "grad_norm": 1.2724547386169434,
      "learning_rate": 1.8607512910974821e-06,
      "loss": 0.0533,
      "step": 311200
    },
    {
      "epoch": 4.583511288493542,
      "grad_norm": 1.617893934249878,
      "learning_rate": 1.8591149191304974e-06,
      "loss": 0.0495,
      "step": 311225
    },
    {
      "epoch": 4.583879471583629,
      "grad_norm": 0.8769577741622925,
      "learning_rate": 1.857478547163513e-06,
      "loss": 0.0472,
      "step": 311250
    },
    {
      "epoch": 4.5842476546737165,
      "grad_norm": 1.9812270402908325,
      "learning_rate": 1.8558421751965282e-06,
      "loss": 0.0519,
      "step": 311275
    },
    {
      "epoch": 4.5846158377638035,
      "grad_norm": 0.9911406636238098,
      "learning_rate": 1.8542058032295438e-06,
      "loss": 0.0534,
      "step": 311300
    },
    {
      "epoch": 4.584984020853891,
      "grad_norm": 1.3202862739562988,
      "learning_rate": 1.8525694312625593e-06,
      "loss": 0.0483,
      "step": 311325
    },
    {
      "epoch": 4.585352203943978,
      "grad_norm": 1.2891674041748047,
      "learning_rate": 1.8509330592955748e-06,
      "loss": 0.052,
      "step": 311350
    },
    {
      "epoch": 4.585720387034065,
      "grad_norm": 1.5891659259796143,
      "learning_rate": 1.8492966873285903e-06,
      "loss": 0.0552,
      "step": 311375
    },
    {
      "epoch": 4.586088570124152,
      "grad_norm": 1.1767886877059937,
      "learning_rate": 1.8476603153616058e-06,
      "loss": 0.0535,
      "step": 311400
    },
    {
      "epoch": 4.586456753214239,
      "grad_norm": 1.64137601852417,
      "learning_rate": 1.8460239433946211e-06,
      "loss": 0.0527,
      "step": 311425
    },
    {
      "epoch": 4.586824936304326,
      "grad_norm": 1.0645405054092407,
      "learning_rate": 1.8443875714276366e-06,
      "loss": 0.0512,
      "step": 311450
    },
    {
      "epoch": 4.587193119394413,
      "grad_norm": 1.5446661710739136,
      "learning_rate": 1.842751199460652e-06,
      "loss": 0.055,
      "step": 311475
    },
    {
      "epoch": 4.5875613024845,
      "grad_norm": 1.27260160446167,
      "learning_rate": 1.8411148274936675e-06,
      "loss": 0.0554,
      "step": 311500
    },
    {
      "epoch": 4.587929485574587,
      "grad_norm": 1.2854636907577515,
      "learning_rate": 1.8394784555266828e-06,
      "loss": 0.0534,
      "step": 311525
    },
    {
      "epoch": 4.588297668664674,
      "grad_norm": 1.3868088722229004,
      "learning_rate": 1.8378420835596983e-06,
      "loss": 0.0565,
      "step": 311550
    },
    {
      "epoch": 4.588665851754761,
      "grad_norm": 1.0023466348648071,
      "learning_rate": 1.8362057115927138e-06,
      "loss": 0.0583,
      "step": 311575
    },
    {
      "epoch": 4.589034034844848,
      "grad_norm": 1.0893226861953735,
      "learning_rate": 1.8345693396257291e-06,
      "loss": 0.0545,
      "step": 311600
    },
    {
      "epoch": 4.589402217934935,
      "grad_norm": 1.3062037229537964,
      "learning_rate": 1.8329329676587446e-06,
      "loss": 0.0513,
      "step": 311625
    },
    {
      "epoch": 4.589770401025022,
      "grad_norm": 1.3451515436172485,
      "learning_rate": 1.83129659569176e-06,
      "loss": 0.0526,
      "step": 311650
    },
    {
      "epoch": 4.590138584115109,
      "grad_norm": 0.977739155292511,
      "learning_rate": 1.8296602237247754e-06,
      "loss": 0.0501,
      "step": 311675
    },
    {
      "epoch": 4.590506767205196,
      "grad_norm": 1.2526726722717285,
      "learning_rate": 1.8280238517577907e-06,
      "loss": 0.0558,
      "step": 311700
    },
    {
      "epoch": 4.590874950295283,
      "grad_norm": 1.677848219871521,
      "learning_rate": 1.8263874797908063e-06,
      "loss": 0.0519,
      "step": 311725
    },
    {
      "epoch": 4.59124313338537,
      "grad_norm": 1.2457284927368164,
      "learning_rate": 1.8247511078238218e-06,
      "loss": 0.0477,
      "step": 311750
    },
    {
      "epoch": 4.591611316475457,
      "grad_norm": 0.8873688578605652,
      "learning_rate": 1.823114735856837e-06,
      "loss": 0.0543,
      "step": 311775
    },
    {
      "epoch": 4.591979499565544,
      "grad_norm": 1.4722398519515991,
      "learning_rate": 1.8214783638898526e-06,
      "loss": 0.0448,
      "step": 311800
    },
    {
      "epoch": 4.592347682655631,
      "grad_norm": 1.2841516733169556,
      "learning_rate": 1.8198419919228683e-06,
      "loss": 0.0536,
      "step": 311825
    },
    {
      "epoch": 4.592715865745718,
      "grad_norm": 1.401716709136963,
      "learning_rate": 1.8182056199558836e-06,
      "loss": 0.0564,
      "step": 311850
    },
    {
      "epoch": 4.593084048835805,
      "grad_norm": 1.0068591833114624,
      "learning_rate": 1.8165692479888991e-06,
      "loss": 0.0482,
      "step": 311875
    },
    {
      "epoch": 4.593452231925892,
      "grad_norm": 0.9521609544754028,
      "learning_rate": 1.8149328760219145e-06,
      "loss": 0.0469,
      "step": 311900
    },
    {
      "epoch": 4.593820415015979,
      "grad_norm": 0.9109975099563599,
      "learning_rate": 1.81329650405493e-06,
      "loss": 0.0498,
      "step": 311925
    },
    {
      "epoch": 4.594188598106066,
      "grad_norm": 1.2523980140686035,
      "learning_rate": 1.8116601320879455e-06,
      "loss": 0.0509,
      "step": 311950
    },
    {
      "epoch": 4.594556781196153,
      "grad_norm": 1.1682549715042114,
      "learning_rate": 1.8100237601209608e-06,
      "loss": 0.0434,
      "step": 311975
    },
    {
      "epoch": 4.59492496428624,
      "grad_norm": 1.3259446620941162,
      "learning_rate": 1.8083873881539763e-06,
      "loss": 0.0521,
      "step": 312000
    },
    {
      "epoch": 4.595293147376327,
      "grad_norm": 1.3220573663711548,
      "learning_rate": 1.8067510161869916e-06,
      "loss": 0.0517,
      "step": 312025
    },
    {
      "epoch": 4.595661330466414,
      "grad_norm": 1.246146321296692,
      "learning_rate": 1.8051146442200071e-06,
      "loss": 0.0489,
      "step": 312050
    },
    {
      "epoch": 4.596029513556501,
      "grad_norm": 1.0258642435073853,
      "learning_rate": 1.8034782722530224e-06,
      "loss": 0.0493,
      "step": 312075
    },
    {
      "epoch": 4.596397696646588,
      "grad_norm": 1.2516448497772217,
      "learning_rate": 1.801841900286038e-06,
      "loss": 0.0521,
      "step": 312100
    },
    {
      "epoch": 4.596765879736676,
      "grad_norm": 1.1295586824417114,
      "learning_rate": 1.8002055283190535e-06,
      "loss": 0.0525,
      "step": 312125
    },
    {
      "epoch": 4.597134062826763,
      "grad_norm": 1.355372667312622,
      "learning_rate": 1.7985691563520688e-06,
      "loss": 0.0479,
      "step": 312150
    },
    {
      "epoch": 4.59750224591685,
      "grad_norm": 1.616727590560913,
      "learning_rate": 1.7969327843850843e-06,
      "loss": 0.0465,
      "step": 312175
    },
    {
      "epoch": 4.597870429006937,
      "grad_norm": 1.3392568826675415,
      "learning_rate": 1.7952964124180996e-06,
      "loss": 0.0487,
      "step": 312200
    },
    {
      "epoch": 4.598238612097024,
      "grad_norm": 1.1653858423233032,
      "learning_rate": 1.793660040451115e-06,
      "loss": 0.0513,
      "step": 312225
    },
    {
      "epoch": 4.598606795187111,
      "grad_norm": 1.2011961936950684,
      "learning_rate": 1.7920236684841304e-06,
      "loss": 0.0513,
      "step": 312250
    },
    {
      "epoch": 4.598974978277198,
      "grad_norm": 1.2303119897842407,
      "learning_rate": 1.790387296517146e-06,
      "loss": 0.0435,
      "step": 312275
    },
    {
      "epoch": 4.599343161367285,
      "grad_norm": 1.6332257986068726,
      "learning_rate": 1.7887509245501614e-06,
      "loss": 0.0516,
      "step": 312300
    },
    {
      "epoch": 4.599711344457372,
      "grad_norm": 1.1437263488769531,
      "learning_rate": 1.7871145525831772e-06,
      "loss": 0.0535,
      "step": 312325
    },
    {
      "epoch": 4.600079527547459,
      "grad_norm": 0.8941243886947632,
      "learning_rate": 1.7854781806161925e-06,
      "loss": 0.0565,
      "step": 312350
    },
    {
      "epoch": 4.600447710637546,
      "grad_norm": 1.5741775035858154,
      "learning_rate": 1.783841808649208e-06,
      "loss": 0.049,
      "step": 312375
    },
    {
      "epoch": 4.600815893727633,
      "grad_norm": 1.402544617652893,
      "learning_rate": 1.7822054366822233e-06,
      "loss": 0.0568,
      "step": 312400
    },
    {
      "epoch": 4.60118407681772,
      "grad_norm": 1.5375876426696777,
      "learning_rate": 1.7805690647152388e-06,
      "loss": 0.0555,
      "step": 312425
    },
    {
      "epoch": 4.601552259907807,
      "grad_norm": 1.2158266305923462,
      "learning_rate": 1.7789326927482541e-06,
      "loss": 0.0485,
      "step": 312450
    },
    {
      "epoch": 4.601920442997894,
      "grad_norm": 0.9526785612106323,
      "learning_rate": 1.7772963207812696e-06,
      "loss": 0.0507,
      "step": 312475
    },
    {
      "epoch": 4.602288626087981,
      "grad_norm": 1.5378708839416504,
      "learning_rate": 1.7756599488142851e-06,
      "loss": 0.0482,
      "step": 312500
    },
    {
      "epoch": 4.602656809178068,
      "grad_norm": 0.9887898564338684,
      "learning_rate": 1.7740235768473005e-06,
      "loss": 0.0508,
      "step": 312525
    },
    {
      "epoch": 4.603024992268155,
      "grad_norm": 1.1639207601547241,
      "learning_rate": 1.772387204880316e-06,
      "loss": 0.0481,
      "step": 312550
    },
    {
      "epoch": 4.603393175358242,
      "grad_norm": 0.9504382014274597,
      "learning_rate": 1.7707508329133313e-06,
      "loss": 0.0481,
      "step": 312575
    },
    {
      "epoch": 4.603761358448329,
      "grad_norm": 1.2448655366897583,
      "learning_rate": 1.7691144609463468e-06,
      "loss": 0.0515,
      "step": 312600
    },
    {
      "epoch": 4.604129541538416,
      "grad_norm": 0.9778062105178833,
      "learning_rate": 1.767478088979362e-06,
      "loss": 0.0518,
      "step": 312625
    },
    {
      "epoch": 4.604497724628503,
      "grad_norm": 1.0024763345718384,
      "learning_rate": 1.7658417170123776e-06,
      "loss": 0.0449,
      "step": 312650
    },
    {
      "epoch": 4.60486590771859,
      "grad_norm": 1.8824326992034912,
      "learning_rate": 1.7642053450453931e-06,
      "loss": 0.0512,
      "step": 312675
    },
    {
      "epoch": 4.605234090808677,
      "grad_norm": 1.2657194137573242,
      "learning_rate": 1.7625689730784084e-06,
      "loss": 0.0521,
      "step": 312700
    },
    {
      "epoch": 4.605602273898764,
      "grad_norm": 1.0155348777770996,
      "learning_rate": 1.760932601111424e-06,
      "loss": 0.0544,
      "step": 312725
    },
    {
      "epoch": 4.605970456988851,
      "grad_norm": 1.4752349853515625,
      "learning_rate": 1.7592962291444393e-06,
      "loss": 0.0491,
      "step": 312750
    },
    {
      "epoch": 4.606338640078938,
      "grad_norm": 1.147755742073059,
      "learning_rate": 1.7576598571774548e-06,
      "loss": 0.0462,
      "step": 312775
    },
    {
      "epoch": 4.606706823169025,
      "grad_norm": 1.1272672414779663,
      "learning_rate": 1.75602348521047e-06,
      "loss": 0.0554,
      "step": 312800
    },
    {
      "epoch": 4.607075006259112,
      "grad_norm": 2.004840135574341,
      "learning_rate": 1.7543871132434858e-06,
      "loss": 0.0479,
      "step": 312825
    },
    {
      "epoch": 4.607443189349199,
      "grad_norm": 1.5390398502349854,
      "learning_rate": 1.7528161961551807e-06,
      "loss": 0.0546,
      "step": 312850
    },
    {
      "epoch": 4.607811372439286,
      "grad_norm": 1.48603355884552,
      "learning_rate": 1.751179824188196e-06,
      "loss": 0.049,
      "step": 312875
    },
    {
      "epoch": 4.608179555529373,
      "grad_norm": 1.4582126140594482,
      "learning_rate": 1.7495434522212116e-06,
      "loss": 0.0512,
      "step": 312900
    },
    {
      "epoch": 4.60854773861946,
      "grad_norm": 1.2588406801223755,
      "learning_rate": 1.7479070802542269e-06,
      "loss": 0.0484,
      "step": 312925
    },
    {
      "epoch": 4.6089159217095474,
      "grad_norm": 1.276369333267212,
      "learning_rate": 1.7462707082872424e-06,
      "loss": 0.0519,
      "step": 312950
    },
    {
      "epoch": 4.6092841047996345,
      "grad_norm": 1.3421517610549927,
      "learning_rate": 1.744634336320258e-06,
      "loss": 0.0573,
      "step": 312975
    },
    {
      "epoch": 4.6096522878897215,
      "grad_norm": 1.6167590618133545,
      "learning_rate": 1.7429979643532732e-06,
      "loss": 0.0493,
      "step": 313000
    },
    {
      "epoch": 4.6100204709798085,
      "grad_norm": 0.8688099384307861,
      "learning_rate": 1.7413615923862887e-06,
      "loss": 0.0486,
      "step": 313025
    },
    {
      "epoch": 4.6103886540698955,
      "grad_norm": 1.4828499555587769,
      "learning_rate": 1.739725220419304e-06,
      "loss": 0.0527,
      "step": 313050
    },
    {
      "epoch": 4.6107568371599825,
      "grad_norm": 0.9489651322364807,
      "learning_rate": 1.7380888484523195e-06,
      "loss": 0.0532,
      "step": 313075
    },
    {
      "epoch": 4.6111250202500695,
      "grad_norm": 1.2675871849060059,
      "learning_rate": 1.7364524764853348e-06,
      "loss": 0.0525,
      "step": 313100
    },
    {
      "epoch": 4.6114932033401566,
      "grad_norm": 1.261709213256836,
      "learning_rate": 1.7348161045183504e-06,
      "loss": 0.0494,
      "step": 313125
    },
    {
      "epoch": 4.611861386430244,
      "grad_norm": 1.3355411291122437,
      "learning_rate": 1.7331797325513659e-06,
      "loss": 0.0531,
      "step": 313150
    },
    {
      "epoch": 4.612229569520331,
      "grad_norm": 1.144920825958252,
      "learning_rate": 1.7315433605843812e-06,
      "loss": 0.0439,
      "step": 313175
    },
    {
      "epoch": 4.6125977526104185,
      "grad_norm": 1.3352934122085571,
      "learning_rate": 1.7299069886173967e-06,
      "loss": 0.0472,
      "step": 313200
    },
    {
      "epoch": 4.6129659357005055,
      "grad_norm": 0.980284571647644,
      "learning_rate": 1.728270616650412e-06,
      "loss": 0.0514,
      "step": 313225
    },
    {
      "epoch": 4.6133341187905925,
      "grad_norm": 1.496730089187622,
      "learning_rate": 1.7266342446834275e-06,
      "loss": 0.0524,
      "step": 313250
    },
    {
      "epoch": 4.6137023018806795,
      "grad_norm": 0.9213072657585144,
      "learning_rate": 1.7249978727164428e-06,
      "loss": 0.051,
      "step": 313275
    },
    {
      "epoch": 4.6140704849707666,
      "grad_norm": 1.0719350576400757,
      "learning_rate": 1.7233615007494585e-06,
      "loss": 0.0571,
      "step": 313300
    },
    {
      "epoch": 4.614438668060854,
      "grad_norm": 1.549264907836914,
      "learning_rate": 1.721725128782474e-06,
      "loss": 0.0554,
      "step": 313325
    },
    {
      "epoch": 4.614806851150941,
      "grad_norm": 0.8943632245063782,
      "learning_rate": 1.7200887568154896e-06,
      "loss": 0.0451,
      "step": 313350
    },
    {
      "epoch": 4.615175034241028,
      "grad_norm": 1.2267400026321411,
      "learning_rate": 1.7184523848485049e-06,
      "loss": 0.0551,
      "step": 313375
    },
    {
      "epoch": 4.615543217331115,
      "grad_norm": 1.7456096410751343,
      "learning_rate": 1.7168160128815204e-06,
      "loss": 0.0568,
      "step": 313400
    },
    {
      "epoch": 4.615911400421202,
      "grad_norm": 1.3656628131866455,
      "learning_rate": 1.7151796409145357e-06,
      "loss": 0.0497,
      "step": 313425
    },
    {
      "epoch": 4.616279583511289,
      "grad_norm": 1.5298798084259033,
      "learning_rate": 1.7135432689475512e-06,
      "loss": 0.0543,
      "step": 313450
    },
    {
      "epoch": 4.616647766601376,
      "grad_norm": 1.545066475868225,
      "learning_rate": 1.7119068969805665e-06,
      "loss": 0.0554,
      "step": 313475
    },
    {
      "epoch": 4.617015949691463,
      "grad_norm": 1.1340826749801636,
      "learning_rate": 1.710270525013582e-06,
      "loss": 0.0482,
      "step": 313500
    },
    {
      "epoch": 4.61738413278155,
      "grad_norm": 1.4046839475631714,
      "learning_rate": 1.7086341530465976e-06,
      "loss": 0.0491,
      "step": 313525
    },
    {
      "epoch": 4.617752315871637,
      "grad_norm": 1.1509557962417603,
      "learning_rate": 1.7069977810796129e-06,
      "loss": 0.048,
      "step": 313550
    },
    {
      "epoch": 4.618120498961724,
      "grad_norm": 0.8133782744407654,
      "learning_rate": 1.7053614091126284e-06,
      "loss": 0.0542,
      "step": 313575
    },
    {
      "epoch": 4.618488682051811,
      "grad_norm": 1.0698652267456055,
      "learning_rate": 1.7037250371456437e-06,
      "loss": 0.0503,
      "step": 313600
    },
    {
      "epoch": 4.618856865141898,
      "grad_norm": 1.2086094617843628,
      "learning_rate": 1.7020886651786592e-06,
      "loss": 0.0515,
      "step": 313625
    },
    {
      "epoch": 4.619225048231985,
      "grad_norm": 1.5048376321792603,
      "learning_rate": 1.7004522932116745e-06,
      "loss": 0.052,
      "step": 313650
    },
    {
      "epoch": 4.619593231322072,
      "grad_norm": 1.3776469230651855,
      "learning_rate": 1.69881592124469e-06,
      "loss": 0.0527,
      "step": 313675
    },
    {
      "epoch": 4.619961414412159,
      "grad_norm": 1.3432800769805908,
      "learning_rate": 1.6971795492777055e-06,
      "loss": 0.0582,
      "step": 313700
    },
    {
      "epoch": 4.620329597502246,
      "grad_norm": 1.5966800451278687,
      "learning_rate": 1.6955431773107208e-06,
      "loss": 0.0598,
      "step": 313725
    },
    {
      "epoch": 4.620697780592333,
      "grad_norm": 1.4164814949035645,
      "learning_rate": 1.6939068053437364e-06,
      "loss": 0.0519,
      "step": 313750
    },
    {
      "epoch": 4.62106596368242,
      "grad_norm": 1.2153713703155518,
      "learning_rate": 1.692270433376752e-06,
      "loss": 0.0453,
      "step": 313775
    },
    {
      "epoch": 4.621434146772507,
      "grad_norm": 1.4762452840805054,
      "learning_rate": 1.6906340614097674e-06,
      "loss": 0.0541,
      "step": 313800
    },
    {
      "epoch": 4.621802329862594,
      "grad_norm": 1.0811917781829834,
      "learning_rate": 1.688997689442783e-06,
      "loss": 0.0554,
      "step": 313825
    },
    {
      "epoch": 4.622170512952681,
      "grad_norm": 1.6722486019134521,
      "learning_rate": 1.6873613174757982e-06,
      "loss": 0.0451,
      "step": 313850
    },
    {
      "epoch": 4.622538696042768,
      "grad_norm": 0.5524638295173645,
      "learning_rate": 1.6857249455088137e-06,
      "loss": 0.0422,
      "step": 313875
    },
    {
      "epoch": 4.622906879132855,
      "grad_norm": 1.4932174682617188,
      "learning_rate": 1.6840885735418292e-06,
      "loss": 0.0512,
      "step": 313900
    },
    {
      "epoch": 4.623275062222942,
      "grad_norm": 1.0760420560836792,
      "learning_rate": 1.6824522015748445e-06,
      "loss": 0.0502,
      "step": 313925
    },
    {
      "epoch": 4.623643245313029,
      "grad_norm": 1.1159383058547974,
      "learning_rate": 1.68081582960786e-06,
      "loss": 0.0524,
      "step": 313950
    },
    {
      "epoch": 4.624011428403116,
      "grad_norm": 0.7388298511505127,
      "learning_rate": 1.6791794576408754e-06,
      "loss": 0.0565,
      "step": 313975
    },
    {
      "epoch": 4.624379611493203,
      "grad_norm": 1.4073553085327148,
      "learning_rate": 1.6775430856738909e-06,
      "loss": 0.0542,
      "step": 314000
    },
    {
      "epoch": 4.62474779458329,
      "grad_norm": 1.2523738145828247,
      "learning_rate": 1.6759067137069062e-06,
      "loss": 0.0544,
      "step": 314025
    },
    {
      "epoch": 4.625115977673378,
      "grad_norm": 1.7144370079040527,
      "learning_rate": 1.6742703417399217e-06,
      "loss": 0.0478,
      "step": 314050
    },
    {
      "epoch": 4.625484160763465,
      "grad_norm": 1.1651575565338135,
      "learning_rate": 1.6726339697729372e-06,
      "loss": 0.0561,
      "step": 314075
    },
    {
      "epoch": 4.625852343853552,
      "grad_norm": 1.1400476694107056,
      "learning_rate": 1.6709975978059525e-06,
      "loss": 0.0538,
      "step": 314100
    },
    {
      "epoch": 4.626220526943639,
      "grad_norm": 1.1046559810638428,
      "learning_rate": 1.669361225838968e-06,
      "loss": 0.0488,
      "step": 314125
    },
    {
      "epoch": 4.626588710033726,
      "grad_norm": 1.2454677820205688,
      "learning_rate": 1.6677248538719833e-06,
      "loss": 0.0517,
      "step": 314150
    },
    {
      "epoch": 4.626956893123813,
      "grad_norm": 1.4426357746124268,
      "learning_rate": 1.6660884819049989e-06,
      "loss": 0.0493,
      "step": 314175
    },
    {
      "epoch": 4.6273250762139,
      "grad_norm": 1.4759565591812134,
      "learning_rate": 1.6644521099380142e-06,
      "loss": 0.0509,
      "step": 314200
    },
    {
      "epoch": 4.627693259303987,
      "grad_norm": 0.9264955520629883,
      "learning_rate": 1.6628157379710297e-06,
      "loss": 0.0499,
      "step": 314225
    },
    {
      "epoch": 4.628061442394074,
      "grad_norm": 1.1288747787475586,
      "learning_rate": 1.6611793660040452e-06,
      "loss": 0.0493,
      "step": 314250
    },
    {
      "epoch": 4.628429625484161,
      "grad_norm": 1.6480368375778198,
      "learning_rate": 1.6595429940370607e-06,
      "loss": 0.0529,
      "step": 314275
    },
    {
      "epoch": 4.628797808574248,
      "grad_norm": 1.027058482170105,
      "learning_rate": 1.6579066220700762e-06,
      "loss": 0.048,
      "step": 314300
    },
    {
      "epoch": 4.629165991664335,
      "grad_norm": 1.4424378871917725,
      "learning_rate": 1.6562702501030917e-06,
      "loss": 0.0495,
      "step": 314325
    },
    {
      "epoch": 4.629534174754422,
      "grad_norm": 1.127670168876648,
      "learning_rate": 1.654633878136107e-06,
      "loss": 0.0495,
      "step": 314350
    },
    {
      "epoch": 4.629902357844509,
      "grad_norm": 0.9387210607528687,
      "learning_rate": 1.6529975061691226e-06,
      "loss": 0.0547,
      "step": 314375
    },
    {
      "epoch": 4.630270540934596,
      "grad_norm": 0.9497400522232056,
      "learning_rate": 1.6513611342021379e-06,
      "loss": 0.0506,
      "step": 314400
    },
    {
      "epoch": 4.630638724024683,
      "grad_norm": 1.5737370252609253,
      "learning_rate": 1.6497247622351534e-06,
      "loss": 0.0545,
      "step": 314425
    },
    {
      "epoch": 4.63100690711477,
      "grad_norm": 1.650532841682434,
      "learning_rate": 1.648088390268169e-06,
      "loss": 0.0512,
      "step": 314450
    },
    {
      "epoch": 4.631375090204857,
      "grad_norm": 1.9460455179214478,
      "learning_rate": 1.6464520183011842e-06,
      "loss": 0.0569,
      "step": 314475
    },
    {
      "epoch": 4.631743273294944,
      "grad_norm": 1.724442958831787,
      "learning_rate": 1.6448156463341997e-06,
      "loss": 0.0583,
      "step": 314500
    },
    {
      "epoch": 4.632111456385031,
      "grad_norm": 1.5884807109832764,
      "learning_rate": 1.643179274367215e-06,
      "loss": 0.0538,
      "step": 314525
    },
    {
      "epoch": 4.632479639475118,
      "grad_norm": 1.2943904399871826,
      "learning_rate": 1.6415429024002305e-06,
      "loss": 0.0453,
      "step": 314550
    },
    {
      "epoch": 4.632847822565205,
      "grad_norm": 0.8792909979820251,
      "learning_rate": 1.6399065304332458e-06,
      "loss": 0.0489,
      "step": 314575
    },
    {
      "epoch": 4.633216005655292,
      "grad_norm": 1.2160710096359253,
      "learning_rate": 1.6382701584662614e-06,
      "loss": 0.0534,
      "step": 314600
    },
    {
      "epoch": 4.633584188745379,
      "grad_norm": 1.3838019371032715,
      "learning_rate": 1.6366337864992769e-06,
      "loss": 0.0543,
      "step": 314625
    },
    {
      "epoch": 4.633952371835466,
      "grad_norm": 1.0022517442703247,
      "learning_rate": 1.6349974145322922e-06,
      "loss": 0.049,
      "step": 314650
    },
    {
      "epoch": 4.634320554925553,
      "grad_norm": 1.6174120903015137,
      "learning_rate": 1.6333610425653077e-06,
      "loss": 0.0526,
      "step": 314675
    },
    {
      "epoch": 4.63468873801564,
      "grad_norm": 1.340504765510559,
      "learning_rate": 1.631724670598323e-06,
      "loss": 0.0486,
      "step": 314700
    },
    {
      "epoch": 4.635056921105727,
      "grad_norm": 1.519906997680664,
      "learning_rate": 1.6300882986313385e-06,
      "loss": 0.0466,
      "step": 314725
    },
    {
      "epoch": 4.635425104195814,
      "grad_norm": 1.1377347707748413,
      "learning_rate": 1.6284519266643543e-06,
      "loss": 0.048,
      "step": 314750
    },
    {
      "epoch": 4.635793287285901,
      "grad_norm": 1.998067021369934,
      "learning_rate": 1.6268155546973696e-06,
      "loss": 0.0531,
      "step": 314775
    },
    {
      "epoch": 4.636161470375988,
      "grad_norm": 1.3252503871917725,
      "learning_rate": 1.625179182730385e-06,
      "loss": 0.0562,
      "step": 314800
    },
    {
      "epoch": 4.636529653466075,
      "grad_norm": 1.2372859716415405,
      "learning_rate": 1.6235428107634004e-06,
      "loss": 0.0478,
      "step": 314825
    },
    {
      "epoch": 4.636897836556162,
      "grad_norm": 0.8300296068191528,
      "learning_rate": 1.6219064387964159e-06,
      "loss": 0.05,
      "step": 314850
    },
    {
      "epoch": 4.637266019646249,
      "grad_norm": 1.3158036470413208,
      "learning_rate": 1.6202700668294314e-06,
      "loss": 0.046,
      "step": 314875
    },
    {
      "epoch": 4.637634202736336,
      "grad_norm": 1.2119743824005127,
      "learning_rate": 1.6186336948624467e-06,
      "loss": 0.0445,
      "step": 314900
    },
    {
      "epoch": 4.638002385826423,
      "grad_norm": 1.3168491125106812,
      "learning_rate": 1.6169973228954622e-06,
      "loss": 0.0557,
      "step": 314925
    },
    {
      "epoch": 4.6383705689165105,
      "grad_norm": 1.3508614301681519,
      "learning_rate": 1.6153609509284775e-06,
      "loss": 0.0538,
      "step": 314950
    },
    {
      "epoch": 4.6387387520065975,
      "grad_norm": 1.364932656288147,
      "learning_rate": 1.613724578961493e-06,
      "loss": 0.0449,
      "step": 314975
    },
    {
      "epoch": 4.6391069350966845,
      "grad_norm": 1.0636990070343018,
      "learning_rate": 1.6120882069945086e-06,
      "loss": 0.0598,
      "step": 315000
    },
    {
      "epoch": 4.6394751181867715,
      "grad_norm": 1.0130901336669922,
      "learning_rate": 1.6104518350275239e-06,
      "loss": 0.0501,
      "step": 315025
    },
    {
      "epoch": 4.6398433012768585,
      "grad_norm": 2.022745370864868,
      "learning_rate": 1.6088154630605394e-06,
      "loss": 0.0548,
      "step": 315050
    },
    {
      "epoch": 4.6402114843669455,
      "grad_norm": 1.4349260330200195,
      "learning_rate": 1.6071790910935547e-06,
      "loss": 0.0506,
      "step": 315075
    },
    {
      "epoch": 4.6405796674570325,
      "grad_norm": 1.2244259119033813,
      "learning_rate": 1.6055427191265702e-06,
      "loss": 0.0527,
      "step": 315100
    },
    {
      "epoch": 4.6409478505471204,
      "grad_norm": 1.2823513746261597,
      "learning_rate": 1.6039063471595855e-06,
      "loss": 0.0487,
      "step": 315125
    },
    {
      "epoch": 4.6413160336372075,
      "grad_norm": 1.5059019327163696,
      "learning_rate": 1.602269975192601e-06,
      "loss": 0.053,
      "step": 315150
    },
    {
      "epoch": 4.6416842167272945,
      "grad_norm": 1.4514081478118896,
      "learning_rate": 1.6006336032256165e-06,
      "loss": 0.0491,
      "step": 315175
    },
    {
      "epoch": 4.6420523998173815,
      "grad_norm": 1.50724458694458,
      "learning_rate": 1.5989972312586318e-06,
      "loss": 0.0556,
      "step": 315200
    },
    {
      "epoch": 4.6424205829074685,
      "grad_norm": 1.20079505443573,
      "learning_rate": 1.5973608592916474e-06,
      "loss": 0.0496,
      "step": 315225
    },
    {
      "epoch": 4.6427887659975555,
      "grad_norm": 1.6044684648513794,
      "learning_rate": 1.595724487324663e-06,
      "loss": 0.0528,
      "step": 315250
    },
    {
      "epoch": 4.6431569490876425,
      "grad_norm": 1.4655390977859497,
      "learning_rate": 1.5940881153576784e-06,
      "loss": 0.0519,
      "step": 315275
    },
    {
      "epoch": 4.6435251321777296,
      "grad_norm": 0.4617000222206116,
      "learning_rate": 1.592451743390694e-06,
      "loss": 0.0485,
      "step": 315300
    },
    {
      "epoch": 4.643893315267817,
      "grad_norm": 1.4401443004608154,
      "learning_rate": 1.5908153714237092e-06,
      "loss": 0.0491,
      "step": 315325
    },
    {
      "epoch": 4.644261498357904,
      "grad_norm": 0.8690106272697449,
      "learning_rate": 1.5891789994567247e-06,
      "loss": 0.0531,
      "step": 315350
    },
    {
      "epoch": 4.644629681447991,
      "grad_norm": 1.3379560708999634,
      "learning_rate": 1.58754262748974e-06,
      "loss": 0.0464,
      "step": 315375
    },
    {
      "epoch": 4.644997864538078,
      "grad_norm": 1.6646497249603271,
      "learning_rate": 1.5859062555227556e-06,
      "loss": 0.0479,
      "step": 315400
    },
    {
      "epoch": 4.645366047628165,
      "grad_norm": 0.9758337736129761,
      "learning_rate": 1.584269883555771e-06,
      "loss": 0.0481,
      "step": 315425
    },
    {
      "epoch": 4.645734230718252,
      "grad_norm": 1.05086350440979,
      "learning_rate": 1.5826335115887864e-06,
      "loss": 0.0477,
      "step": 315450
    },
    {
      "epoch": 4.646102413808339,
      "grad_norm": 1.4862937927246094,
      "learning_rate": 1.5809971396218019e-06,
      "loss": 0.0494,
      "step": 315475
    },
    {
      "epoch": 4.646470596898426,
      "grad_norm": 1.3824751377105713,
      "learning_rate": 1.5793607676548172e-06,
      "loss": 0.0513,
      "step": 315500
    },
    {
      "epoch": 4.646838779988513,
      "grad_norm": 1.7633616924285889,
      "learning_rate": 1.5777243956878327e-06,
      "loss": 0.0518,
      "step": 315525
    },
    {
      "epoch": 4.6472069630786,
      "grad_norm": 0.899743914604187,
      "learning_rate": 1.5760880237208482e-06,
      "loss": 0.0507,
      "step": 315550
    },
    {
      "epoch": 4.647575146168687,
      "grad_norm": 1.464272379875183,
      "learning_rate": 1.5744516517538635e-06,
      "loss": 0.0499,
      "step": 315575
    },
    {
      "epoch": 4.647943329258774,
      "grad_norm": 1.549464464187622,
      "learning_rate": 1.572815279786879e-06,
      "loss": 0.0537,
      "step": 315600
    },
    {
      "epoch": 4.648311512348861,
      "grad_norm": 1.1711963415145874,
      "learning_rate": 1.5711789078198944e-06,
      "loss": 0.0499,
      "step": 315625
    },
    {
      "epoch": 4.648679695438948,
      "grad_norm": 1.3006376028060913,
      "learning_rate": 1.5695425358529099e-06,
      "loss": 0.0516,
      "step": 315650
    },
    {
      "epoch": 4.649047878529035,
      "grad_norm": 1.8834633827209473,
      "learning_rate": 1.5679061638859252e-06,
      "loss": 0.0478,
      "step": 315675
    },
    {
      "epoch": 4.649416061619122,
      "grad_norm": 1.5423249006271362,
      "learning_rate": 1.5662697919189407e-06,
      "loss": 0.0514,
      "step": 315700
    },
    {
      "epoch": 4.649784244709209,
      "grad_norm": 0.895173966884613,
      "learning_rate": 1.5646334199519562e-06,
      "loss": 0.0579,
      "step": 315725
    },
    {
      "epoch": 4.650152427799296,
      "grad_norm": 1.0859516859054565,
      "learning_rate": 1.5629970479849717e-06,
      "loss": 0.0537,
      "step": 315750
    },
    {
      "epoch": 4.650520610889383,
      "grad_norm": 1.404670000076294,
      "learning_rate": 1.5613606760179872e-06,
      "loss": 0.0464,
      "step": 315775
    },
    {
      "epoch": 4.65088879397947,
      "grad_norm": 0.9491305351257324,
      "learning_rate": 1.5597243040510028e-06,
      "loss": 0.0511,
      "step": 315800
    },
    {
      "epoch": 4.651256977069557,
      "grad_norm": 1.3225444555282593,
      "learning_rate": 1.558087932084018e-06,
      "loss": 0.0449,
      "step": 315825
    },
    {
      "epoch": 4.651625160159644,
      "grad_norm": 1.6994314193725586,
      "learning_rate": 1.5564515601170336e-06,
      "loss": 0.0527,
      "step": 315850
    },
    {
      "epoch": 4.651993343249731,
      "grad_norm": 1.5581889152526855,
      "learning_rate": 1.5548151881500489e-06,
      "loss": 0.0568,
      "step": 315875
    },
    {
      "epoch": 4.652361526339818,
      "grad_norm": 1.135936975479126,
      "learning_rate": 1.5531788161830644e-06,
      "loss": 0.0518,
      "step": 315900
    },
    {
      "epoch": 4.652729709429905,
      "grad_norm": 1.3783987760543823,
      "learning_rate": 1.5515424442160797e-06,
      "loss": 0.0534,
      "step": 315925
    },
    {
      "epoch": 4.653097892519993,
      "grad_norm": 0.7552604079246521,
      "learning_rate": 1.5499060722490952e-06,
      "loss": 0.0506,
      "step": 315950
    },
    {
      "epoch": 4.65346607561008,
      "grad_norm": 1.5362344980239868,
      "learning_rate": 1.5482697002821107e-06,
      "loss": 0.0498,
      "step": 315975
    },
    {
      "epoch": 4.653834258700167,
      "grad_norm": 1.3423012495040894,
      "learning_rate": 1.546633328315126e-06,
      "loss": 0.0474,
      "step": 316000
    },
    {
      "epoch": 4.654202441790254,
      "grad_norm": 1.125746250152588,
      "learning_rate": 1.5449969563481416e-06,
      "loss": 0.0546,
      "step": 316025
    },
    {
      "epoch": 4.654570624880341,
      "grad_norm": 1.1130011081695557,
      "learning_rate": 1.5433605843811569e-06,
      "loss": 0.0448,
      "step": 316050
    },
    {
      "epoch": 4.654938807970428,
      "grad_norm": 0.882328987121582,
      "learning_rate": 1.5417242124141724e-06,
      "loss": 0.0573,
      "step": 316075
    },
    {
      "epoch": 4.655306991060515,
      "grad_norm": 1.3008991479873657,
      "learning_rate": 1.5400878404471879e-06,
      "loss": 0.0526,
      "step": 316100
    },
    {
      "epoch": 4.655675174150602,
      "grad_norm": 1.0555094480514526,
      "learning_rate": 1.5384514684802032e-06,
      "loss": 0.0525,
      "step": 316125
    },
    {
      "epoch": 4.656043357240689,
      "grad_norm": 1.6777547597885132,
      "learning_rate": 1.5368150965132187e-06,
      "loss": 0.0473,
      "step": 316150
    },
    {
      "epoch": 4.656411540330776,
      "grad_norm": 1.5286930799484253,
      "learning_rate": 1.535178724546234e-06,
      "loss": 0.0504,
      "step": 316175
    },
    {
      "epoch": 4.656779723420863,
      "grad_norm": 1.203751802444458,
      "learning_rate": 1.5335423525792495e-06,
      "loss": 0.0443,
      "step": 316200
    },
    {
      "epoch": 4.65714790651095,
      "grad_norm": 1.884735345840454,
      "learning_rate": 1.5319059806122653e-06,
      "loss": 0.0484,
      "step": 316225
    },
    {
      "epoch": 4.657516089601037,
      "grad_norm": 1.0852811336517334,
      "learning_rate": 1.5302696086452806e-06,
      "loss": 0.0533,
      "step": 316250
    },
    {
      "epoch": 4.657884272691124,
      "grad_norm": 0.999225914478302,
      "learning_rate": 1.528633236678296e-06,
      "loss": 0.0483,
      "step": 316275
    },
    {
      "epoch": 4.658252455781211,
      "grad_norm": 1.2926151752471924,
      "learning_rate": 1.5269968647113114e-06,
      "loss": 0.0528,
      "step": 316300
    },
    {
      "epoch": 4.658620638871298,
      "grad_norm": 1.5772346258163452,
      "learning_rate": 1.525360492744327e-06,
      "loss": 0.0542,
      "step": 316325
    },
    {
      "epoch": 4.658988821961385,
      "grad_norm": 1.1165692806243896,
      "learning_rate": 1.5237241207773424e-06,
      "loss": 0.052,
      "step": 316350
    },
    {
      "epoch": 4.659357005051472,
      "grad_norm": 1.0065970420837402,
      "learning_rate": 1.5220877488103577e-06,
      "loss": 0.0504,
      "step": 316375
    },
    {
      "epoch": 4.659725188141559,
      "grad_norm": 1.510586142539978,
      "learning_rate": 1.5204513768433732e-06,
      "loss": 0.0536,
      "step": 316400
    },
    {
      "epoch": 4.660093371231646,
      "grad_norm": 1.4851386547088623,
      "learning_rate": 1.5188150048763885e-06,
      "loss": 0.0471,
      "step": 316425
    },
    {
      "epoch": 4.660461554321733,
      "grad_norm": 1.5317485332489014,
      "learning_rate": 1.517178632909404e-06,
      "loss": 0.05,
      "step": 316450
    },
    {
      "epoch": 4.66082973741182,
      "grad_norm": 1.4782650470733643,
      "learning_rate": 1.5155422609424194e-06,
      "loss": 0.0565,
      "step": 316475
    },
    {
      "epoch": 4.661197920501907,
      "grad_norm": 1.536999225616455,
      "learning_rate": 1.5139058889754349e-06,
      "loss": 0.0485,
      "step": 316500
    },
    {
      "epoch": 4.661566103591994,
      "grad_norm": 1.1295926570892334,
      "learning_rate": 1.5122695170084504e-06,
      "loss": 0.0534,
      "step": 316525
    },
    {
      "epoch": 4.661934286682081,
      "grad_norm": 1.2453328371047974,
      "learning_rate": 1.5106331450414657e-06,
      "loss": 0.0514,
      "step": 316550
    },
    {
      "epoch": 4.662302469772168,
      "grad_norm": 1.397569179534912,
      "learning_rate": 1.5089967730744812e-06,
      "loss": 0.0512,
      "step": 316575
    },
    {
      "epoch": 4.662670652862255,
      "grad_norm": 1.7768785953521729,
      "learning_rate": 1.5073604011074965e-06,
      "loss": 0.0559,
      "step": 316600
    },
    {
      "epoch": 4.663038835952342,
      "grad_norm": 1.3191170692443848,
      "learning_rate": 1.505724029140512e-06,
      "loss": 0.0566,
      "step": 316625
    },
    {
      "epoch": 4.663407019042429,
      "grad_norm": 1.1507437229156494,
      "learning_rate": 1.5040876571735276e-06,
      "loss": 0.0528,
      "step": 316650
    },
    {
      "epoch": 4.663775202132516,
      "grad_norm": 1.3382512331008911,
      "learning_rate": 1.5024512852065429e-06,
      "loss": 0.0525,
      "step": 316675
    },
    {
      "epoch": 4.664143385222603,
      "grad_norm": 1.2620623111724854,
      "learning_rate": 1.5008149132395584e-06,
      "loss": 0.0503,
      "step": 316700
    },
    {
      "epoch": 4.66451156831269,
      "grad_norm": 1.5619930028915405,
      "learning_rate": 1.499178541272574e-06,
      "loss": 0.0506,
      "step": 316725
    },
    {
      "epoch": 4.664879751402777,
      "grad_norm": 1.284537672996521,
      "learning_rate": 1.4975421693055894e-06,
      "loss": 0.0528,
      "step": 316750
    },
    {
      "epoch": 4.665247934492864,
      "grad_norm": 1.6497796773910522,
      "learning_rate": 1.495905797338605e-06,
      "loss": 0.0492,
      "step": 316775
    },
    {
      "epoch": 4.665616117582951,
      "grad_norm": 1.3899180889129639,
      "learning_rate": 1.4942694253716202e-06,
      "loss": 0.05,
      "step": 316800
    },
    {
      "epoch": 4.665984300673038,
      "grad_norm": 1.125265121459961,
      "learning_rate": 1.4926330534046357e-06,
      "loss": 0.0456,
      "step": 316825
    },
    {
      "epoch": 4.666352483763125,
      "grad_norm": 1.2792105674743652,
      "learning_rate": 1.4910621363163305e-06,
      "loss": 0.0563,
      "step": 316850
    },
    {
      "epoch": 4.666720666853212,
      "grad_norm": 1.3837476968765259,
      "learning_rate": 1.489425764349346e-06,
      "loss": 0.051,
      "step": 316875
    },
    {
      "epoch": 4.667088849943299,
      "grad_norm": 1.2632555961608887,
      "learning_rate": 1.4877893923823613e-06,
      "loss": 0.0503,
      "step": 316900
    },
    {
      "epoch": 4.667457033033386,
      "grad_norm": 1.1787450313568115,
      "learning_rate": 1.4861530204153768e-06,
      "loss": 0.0572,
      "step": 316925
    },
    {
      "epoch": 4.6678252161234735,
      "grad_norm": 0.8252682089805603,
      "learning_rate": 1.4845166484483923e-06,
      "loss": 0.0487,
      "step": 316950
    },
    {
      "epoch": 4.6681933992135605,
      "grad_norm": 1.583823561668396,
      "learning_rate": 1.4828802764814076e-06,
      "loss": 0.0521,
      "step": 316975
    },
    {
      "epoch": 4.6685615823036475,
      "grad_norm": 0.885945200920105,
      "learning_rate": 1.4812439045144231e-06,
      "loss": 0.0522,
      "step": 317000
    },
    {
      "epoch": 4.6689297653937345,
      "grad_norm": 1.4736963510513306,
      "learning_rate": 1.4796075325474384e-06,
      "loss": 0.0496,
      "step": 317025
    },
    {
      "epoch": 4.669297948483822,
      "grad_norm": 1.4008076190948486,
      "learning_rate": 1.477971160580454e-06,
      "loss": 0.054,
      "step": 317050
    },
    {
      "epoch": 4.669666131573909,
      "grad_norm": 1.1805485486984253,
      "learning_rate": 1.4763347886134693e-06,
      "loss": 0.0479,
      "step": 317075
    },
    {
      "epoch": 4.670034314663996,
      "grad_norm": 1.7553777694702148,
      "learning_rate": 1.4746984166464848e-06,
      "loss": 0.0585,
      "step": 317100
    },
    {
      "epoch": 4.6704024977540834,
      "grad_norm": 1.246410608291626,
      "learning_rate": 1.4730620446795003e-06,
      "loss": 0.0483,
      "step": 317125
    },
    {
      "epoch": 4.6707706808441705,
      "grad_norm": 1.301881194114685,
      "learning_rate": 1.4714256727125156e-06,
      "loss": 0.0514,
      "step": 317150
    },
    {
      "epoch": 4.6711388639342575,
      "grad_norm": 1.1803393363952637,
      "learning_rate": 1.4697893007455311e-06,
      "loss": 0.0517,
      "step": 317175
    },
    {
      "epoch": 4.6715070470243445,
      "grad_norm": 1.532606601715088,
      "learning_rate": 1.4681529287785468e-06,
      "loss": 0.055,
      "step": 317200
    },
    {
      "epoch": 4.6718752301144315,
      "grad_norm": 1.0925289392471313,
      "learning_rate": 1.4665165568115622e-06,
      "loss": 0.0449,
      "step": 317225
    },
    {
      "epoch": 4.6722434132045185,
      "grad_norm": 1.2983659505844116,
      "learning_rate": 1.464945639723257e-06,
      "loss": 0.0494,
      "step": 317250
    },
    {
      "epoch": 4.6726115962946055,
      "grad_norm": 1.8341485261917114,
      "learning_rate": 1.4633092677562724e-06,
      "loss": 0.052,
      "step": 317275
    },
    {
      "epoch": 4.6729797793846926,
      "grad_norm": 1.4778763055801392,
      "learning_rate": 1.461672895789288e-06,
      "loss": 0.0533,
      "step": 317300
    },
    {
      "epoch": 4.67334796247478,
      "grad_norm": 1.5188133716583252,
      "learning_rate": 1.4600365238223032e-06,
      "loss": 0.0488,
      "step": 317325
    },
    {
      "epoch": 4.673716145564867,
      "grad_norm": 1.1263964176177979,
      "learning_rate": 1.4584001518553187e-06,
      "loss": 0.0454,
      "step": 317350
    },
    {
      "epoch": 4.674084328654954,
      "grad_norm": 1.165518879890442,
      "learning_rate": 1.456763779888334e-06,
      "loss": 0.0471,
      "step": 317375
    },
    {
      "epoch": 4.674452511745041,
      "grad_norm": 1.2447923421859741,
      "learning_rate": 1.4551274079213496e-06,
      "loss": 0.0537,
      "step": 317400
    },
    {
      "epoch": 4.674820694835128,
      "grad_norm": 1.7116609811782837,
      "learning_rate": 1.453491035954365e-06,
      "loss": 0.0519,
      "step": 317425
    },
    {
      "epoch": 4.675188877925215,
      "grad_norm": 1.2317157983779907,
      "learning_rate": 1.4518546639873804e-06,
      "loss": 0.0485,
      "step": 317450
    },
    {
      "epoch": 4.675557061015302,
      "grad_norm": 1.0174763202667236,
      "learning_rate": 1.4502182920203959e-06,
      "loss": 0.0489,
      "step": 317475
    },
    {
      "epoch": 4.675925244105389,
      "grad_norm": 1.6003159284591675,
      "learning_rate": 1.4485819200534112e-06,
      "loss": 0.05,
      "step": 317500
    },
    {
      "epoch": 4.676293427195476,
      "grad_norm": 1.542261004447937,
      "learning_rate": 1.4469455480864267e-06,
      "loss": 0.0538,
      "step": 317525
    },
    {
      "epoch": 4.676661610285563,
      "grad_norm": 1.2530255317687988,
      "learning_rate": 1.445309176119442e-06,
      "loss": 0.0564,
      "step": 317550
    },
    {
      "epoch": 4.67702979337565,
      "grad_norm": 1.2042635679244995,
      "learning_rate": 1.4436728041524575e-06,
      "loss": 0.0502,
      "step": 317575
    },
    {
      "epoch": 4.677397976465737,
      "grad_norm": 1.1547775268554688,
      "learning_rate": 1.442036432185473e-06,
      "loss": 0.0545,
      "step": 317600
    },
    {
      "epoch": 4.677766159555824,
      "grad_norm": 1.435349941253662,
      "learning_rate": 1.4404000602184884e-06,
      "loss": 0.0541,
      "step": 317625
    },
    {
      "epoch": 4.678134342645911,
      "grad_norm": 1.601083517074585,
      "learning_rate": 1.4387636882515039e-06,
      "loss": 0.0539,
      "step": 317650
    },
    {
      "epoch": 4.678502525735998,
      "grad_norm": 1.3502408266067505,
      "learning_rate": 1.4371273162845196e-06,
      "loss": 0.0572,
      "step": 317675
    },
    {
      "epoch": 4.678870708826085,
      "grad_norm": 1.0823687314987183,
      "learning_rate": 1.435490944317535e-06,
      "loss": 0.0443,
      "step": 317700
    },
    {
      "epoch": 4.679238891916172,
      "grad_norm": 1.1347333192825317,
      "learning_rate": 1.4338545723505504e-06,
      "loss": 0.0489,
      "step": 317725
    },
    {
      "epoch": 4.679607075006259,
      "grad_norm": 1.3563753366470337,
      "learning_rate": 1.4322182003835657e-06,
      "loss": 0.0511,
      "step": 317750
    },
    {
      "epoch": 4.679975258096346,
      "grad_norm": 1.3053771257400513,
      "learning_rate": 1.4305818284165812e-06,
      "loss": 0.0493,
      "step": 317775
    },
    {
      "epoch": 4.680343441186433,
      "grad_norm": 1.4208966493606567,
      "learning_rate": 1.4289454564495968e-06,
      "loss": 0.0562,
      "step": 317800
    },
    {
      "epoch": 4.68071162427652,
      "grad_norm": 1.2151519060134888,
      "learning_rate": 1.427309084482612e-06,
      "loss": 0.05,
      "step": 317825
    },
    {
      "epoch": 4.681079807366607,
      "grad_norm": 1.4731098413467407,
      "learning_rate": 1.4256727125156276e-06,
      "loss": 0.0488,
      "step": 317850
    },
    {
      "epoch": 4.681447990456695,
      "grad_norm": 1.3843494653701782,
      "learning_rate": 1.4240363405486429e-06,
      "loss": 0.0521,
      "step": 317875
    },
    {
      "epoch": 4.681816173546782,
      "grad_norm": 1.7827988862991333,
      "learning_rate": 1.4223999685816584e-06,
      "loss": 0.0505,
      "step": 317900
    },
    {
      "epoch": 4.682184356636869,
      "grad_norm": 1.2990777492523193,
      "learning_rate": 1.4207635966146737e-06,
      "loss": 0.0536,
      "step": 317925
    },
    {
      "epoch": 4.682552539726956,
      "grad_norm": 1.0244513750076294,
      "learning_rate": 1.4191272246476892e-06,
      "loss": 0.0533,
      "step": 317950
    },
    {
      "epoch": 4.682920722817043,
      "grad_norm": 1.6147840023040771,
      "learning_rate": 1.4174908526807047e-06,
      "loss": 0.0499,
      "step": 317975
    },
    {
      "epoch": 4.68328890590713,
      "grad_norm": 1.1264728307724,
      "learning_rate": 1.41585448071372e-06,
      "loss": 0.055,
      "step": 318000
    },
    {
      "epoch": 4.683657088997217,
      "grad_norm": 1.2167012691497803,
      "learning_rate": 1.4142181087467356e-06,
      "loss": 0.0556,
      "step": 318025
    },
    {
      "epoch": 4.684025272087304,
      "grad_norm": 1.3691784143447876,
      "learning_rate": 1.4125817367797509e-06,
      "loss": 0.0452,
      "step": 318050
    },
    {
      "epoch": 4.684393455177391,
      "grad_norm": 1.1392791271209717,
      "learning_rate": 1.4109453648127664e-06,
      "loss": 0.0532,
      "step": 318075
    },
    {
      "epoch": 4.684761638267478,
      "grad_norm": 1.8969905376434326,
      "learning_rate": 1.4093089928457817e-06,
      "loss": 0.0508,
      "step": 318100
    },
    {
      "epoch": 4.685129821357565,
      "grad_norm": 1.0668405294418335,
      "learning_rate": 1.4076726208787972e-06,
      "loss": 0.052,
      "step": 318125
    },
    {
      "epoch": 4.685498004447652,
      "grad_norm": 1.4562944173812866,
      "learning_rate": 1.406036248911813e-06,
      "loss": 0.0512,
      "step": 318150
    },
    {
      "epoch": 4.685866187537739,
      "grad_norm": 1.69785737991333,
      "learning_rate": 1.4043998769448282e-06,
      "loss": 0.0484,
      "step": 318175
    },
    {
      "epoch": 4.686234370627826,
      "grad_norm": 0.809941828250885,
      "learning_rate": 1.4027635049778437e-06,
      "loss": 0.0472,
      "step": 318200
    },
    {
      "epoch": 4.686602553717913,
      "grad_norm": 1.2320263385772705,
      "learning_rate": 1.4011271330108593e-06,
      "loss": 0.0493,
      "step": 318225
    },
    {
      "epoch": 4.686970736808,
      "grad_norm": 1.3246911764144897,
      "learning_rate": 1.3994907610438746e-06,
      "loss": 0.0551,
      "step": 318250
    },
    {
      "epoch": 4.687338919898087,
      "grad_norm": 1.4349480867385864,
      "learning_rate": 1.39785438907689e-06,
      "loss": 0.0475,
      "step": 318275
    },
    {
      "epoch": 4.687707102988174,
      "grad_norm": 0.7472424507141113,
      "learning_rate": 1.3962180171099054e-06,
      "loss": 0.0539,
      "step": 318300
    },
    {
      "epoch": 4.688075286078261,
      "grad_norm": 1.3324025869369507,
      "learning_rate": 1.394581645142921e-06,
      "loss": 0.049,
      "step": 318325
    },
    {
      "epoch": 4.688443469168348,
      "grad_norm": 1.3278992176055908,
      "learning_rate": 1.3929452731759362e-06,
      "loss": 0.0494,
      "step": 318350
    },
    {
      "epoch": 4.688811652258435,
      "grad_norm": 1.0949945449829102,
      "learning_rate": 1.3913089012089517e-06,
      "loss": 0.0513,
      "step": 318375
    },
    {
      "epoch": 4.689179835348522,
      "grad_norm": 0.8925934433937073,
      "learning_rate": 1.3896725292419672e-06,
      "loss": 0.0485,
      "step": 318400
    },
    {
      "epoch": 4.689548018438609,
      "grad_norm": 1.1080822944641113,
      "learning_rate": 1.3880361572749825e-06,
      "loss": 0.0493,
      "step": 318425
    },
    {
      "epoch": 4.689916201528696,
      "grad_norm": 1.1725772619247437,
      "learning_rate": 1.386399785307998e-06,
      "loss": 0.0478,
      "step": 318450
    },
    {
      "epoch": 4.690284384618783,
      "grad_norm": 1.3580622673034668,
      "learning_rate": 1.3847634133410134e-06,
      "loss": 0.0561,
      "step": 318475
    },
    {
      "epoch": 4.69065256770887,
      "grad_norm": 1.0833568572998047,
      "learning_rate": 1.3831270413740289e-06,
      "loss": 0.0476,
      "step": 318500
    },
    {
      "epoch": 4.691020750798957,
      "grad_norm": 1.0186476707458496,
      "learning_rate": 1.3814906694070444e-06,
      "loss": 0.0576,
      "step": 318525
    },
    {
      "epoch": 4.691388933889044,
      "grad_norm": 1.232149600982666,
      "learning_rate": 1.3798542974400597e-06,
      "loss": 0.0512,
      "step": 318550
    },
    {
      "epoch": 4.691757116979131,
      "grad_norm": 1.1767240762710571,
      "learning_rate": 1.3782179254730752e-06,
      "loss": 0.0464,
      "step": 318575
    },
    {
      "epoch": 4.692125300069218,
      "grad_norm": 1.4793529510498047,
      "learning_rate": 1.3765815535060905e-06,
      "loss": 0.0555,
      "step": 318600
    },
    {
      "epoch": 4.692493483159305,
      "grad_norm": 1.2782994508743286,
      "learning_rate": 1.374945181539106e-06,
      "loss": 0.0506,
      "step": 318625
    },
    {
      "epoch": 4.692861666249392,
      "grad_norm": 1.3629090785980225,
      "learning_rate": 1.3733088095721218e-06,
      "loss": 0.0581,
      "step": 318650
    },
    {
      "epoch": 4.693229849339479,
      "grad_norm": 0.9741042256355286,
      "learning_rate": 1.371672437605137e-06,
      "loss": 0.0594,
      "step": 318675
    },
    {
      "epoch": 4.693598032429566,
      "grad_norm": 1.7201316356658936,
      "learning_rate": 1.3700360656381526e-06,
      "loss": 0.0553,
      "step": 318700
    },
    {
      "epoch": 4.693966215519653,
      "grad_norm": 1.2854005098342896,
      "learning_rate": 1.3683996936711679e-06,
      "loss": 0.048,
      "step": 318725
    },
    {
      "epoch": 4.69433439860974,
      "grad_norm": 1.5507843494415283,
      "learning_rate": 1.3667633217041834e-06,
      "loss": 0.053,
      "step": 318750
    },
    {
      "epoch": 4.694702581699827,
      "grad_norm": 1.2158867120742798,
      "learning_rate": 1.365126949737199e-06,
      "loss": 0.0502,
      "step": 318775
    },
    {
      "epoch": 4.695070764789914,
      "grad_norm": 1.0017958879470825,
      "learning_rate": 1.3634905777702142e-06,
      "loss": 0.0484,
      "step": 318800
    },
    {
      "epoch": 4.695438947880001,
      "grad_norm": 1.4825832843780518,
      "learning_rate": 1.3618542058032297e-06,
      "loss": 0.0525,
      "step": 318825
    },
    {
      "epoch": 4.695807130970088,
      "grad_norm": 1.5330793857574463,
      "learning_rate": 1.360217833836245e-06,
      "loss": 0.0543,
      "step": 318850
    },
    {
      "epoch": 4.696175314060175,
      "grad_norm": 1.3737893104553223,
      "learning_rate": 1.3585814618692606e-06,
      "loss": 0.057,
      "step": 318875
    },
    {
      "epoch": 4.696543497150262,
      "grad_norm": 1.6639982461929321,
      "learning_rate": 1.3569450899022759e-06,
      "loss": 0.0535,
      "step": 318900
    },
    {
      "epoch": 4.696911680240349,
      "grad_norm": 1.3726798295974731,
      "learning_rate": 1.3553087179352914e-06,
      "loss": 0.047,
      "step": 318925
    },
    {
      "epoch": 4.697279863330437,
      "grad_norm": 1.4801117181777954,
      "learning_rate": 1.353672345968307e-06,
      "loss": 0.0511,
      "step": 318950
    },
    {
      "epoch": 4.697648046420524,
      "grad_norm": 1.135445475578308,
      "learning_rate": 1.3520359740013222e-06,
      "loss": 0.0519,
      "step": 318975
    },
    {
      "epoch": 4.698016229510611,
      "grad_norm": 1.2365450859069824,
      "learning_rate": 1.3503996020343377e-06,
      "loss": 0.0494,
      "step": 319000
    },
    {
      "epoch": 4.698384412600698,
      "grad_norm": 1.4642046689987183,
      "learning_rate": 1.348763230067353e-06,
      "loss": 0.0475,
      "step": 319025
    },
    {
      "epoch": 4.698752595690785,
      "grad_norm": 1.3627007007598877,
      "learning_rate": 1.3471268581003685e-06,
      "loss": 0.0505,
      "step": 319050
    },
    {
      "epoch": 4.699120778780872,
      "grad_norm": 1.0912110805511475,
      "learning_rate": 1.345490486133384e-06,
      "loss": 0.0557,
      "step": 319075
    },
    {
      "epoch": 4.699488961870959,
      "grad_norm": 1.7698466777801514,
      "learning_rate": 1.3438541141663994e-06,
      "loss": 0.0489,
      "step": 319100
    },
    {
      "epoch": 4.6998571449610465,
      "grad_norm": 0.9112261533737183,
      "learning_rate": 1.3422177421994149e-06,
      "loss": 0.0552,
      "step": 319125
    },
    {
      "epoch": 4.7002253280511335,
      "grad_norm": 1.3635870218276978,
      "learning_rate": 1.3405813702324306e-06,
      "loss": 0.0512,
      "step": 319150
    },
    {
      "epoch": 4.7005935111412205,
      "grad_norm": 0.8168050050735474,
      "learning_rate": 1.338944998265446e-06,
      "loss": 0.0521,
      "step": 319175
    },
    {
      "epoch": 4.7009616942313075,
      "grad_norm": 1.0441139936447144,
      "learning_rate": 1.3373086262984614e-06,
      "loss": 0.0506,
      "step": 319200
    },
    {
      "epoch": 4.7013298773213945,
      "grad_norm": 1.6629910469055176,
      "learning_rate": 1.3356722543314767e-06,
      "loss": 0.0503,
      "step": 319225
    },
    {
      "epoch": 4.7016980604114815,
      "grad_norm": 1.2574551105499268,
      "learning_rate": 1.3340358823644922e-06,
      "loss": 0.0515,
      "step": 319250
    },
    {
      "epoch": 4.7020662435015685,
      "grad_norm": 1.3636797666549683,
      "learning_rate": 1.3323995103975075e-06,
      "loss": 0.0486,
      "step": 319275
    },
    {
      "epoch": 4.702434426591656,
      "grad_norm": 0.7224562764167786,
      "learning_rate": 1.330763138430523e-06,
      "loss": 0.0509,
      "step": 319300
    },
    {
      "epoch": 4.702802609681743,
      "grad_norm": 1.9357801675796509,
      "learning_rate": 1.3291267664635386e-06,
      "loss": 0.057,
      "step": 319325
    },
    {
      "epoch": 4.70317079277183,
      "grad_norm": 0.9637460112571716,
      "learning_rate": 1.3274903944965539e-06,
      "loss": 0.0489,
      "step": 319350
    },
    {
      "epoch": 4.703538975861917,
      "grad_norm": 1.1193232536315918,
      "learning_rate": 1.3258540225295694e-06,
      "loss": 0.0565,
      "step": 319375
    },
    {
      "epoch": 4.703907158952004,
      "grad_norm": 1.5464826822280884,
      "learning_rate": 1.3242831054412641e-06,
      "loss": 0.0537,
      "step": 319400
    },
    {
      "epoch": 4.704275342042091,
      "grad_norm": 1.3487155437469482,
      "learning_rate": 1.3226467334742796e-06,
      "loss": 0.0501,
      "step": 319425
    },
    {
      "epoch": 4.704643525132178,
      "grad_norm": 1.546373724937439,
      "learning_rate": 1.321010361507295e-06,
      "loss": 0.0525,
      "step": 319450
    },
    {
      "epoch": 4.705011708222265,
      "grad_norm": 1.136923909187317,
      "learning_rate": 1.3193739895403105e-06,
      "loss": 0.0534,
      "step": 319475
    },
    {
      "epoch": 4.705379891312352,
      "grad_norm": 1.3697749376296997,
      "learning_rate": 1.3177376175733258e-06,
      "loss": 0.0518,
      "step": 319500
    },
    {
      "epoch": 4.705748074402439,
      "grad_norm": 1.372707724571228,
      "learning_rate": 1.3161012456063413e-06,
      "loss": 0.0472,
      "step": 319525
    },
    {
      "epoch": 4.706116257492526,
      "grad_norm": 1.7428264617919922,
      "learning_rate": 1.3144648736393568e-06,
      "loss": 0.0524,
      "step": 319550
    },
    {
      "epoch": 4.706484440582613,
      "grad_norm": 1.4106968641281128,
      "learning_rate": 1.312828501672372e-06,
      "loss": 0.0515,
      "step": 319575
    },
    {
      "epoch": 4.7068526236727,
      "grad_norm": 1.0190820693969727,
      "learning_rate": 1.3111921297053876e-06,
      "loss": 0.054,
      "step": 319600
    },
    {
      "epoch": 4.707220806762787,
      "grad_norm": 1.2995719909667969,
      "learning_rate": 1.3095557577384033e-06,
      "loss": 0.046,
      "step": 319625
    },
    {
      "epoch": 4.707588989852874,
      "grad_norm": 1.1809227466583252,
      "learning_rate": 1.3079193857714187e-06,
      "loss": 0.0492,
      "step": 319650
    },
    {
      "epoch": 4.707957172942961,
      "grad_norm": 1.3847019672393799,
      "learning_rate": 1.3062830138044342e-06,
      "loss": 0.048,
      "step": 319675
    },
    {
      "epoch": 4.708325356033048,
      "grad_norm": 1.6168261766433716,
      "learning_rate": 1.3046466418374495e-06,
      "loss": 0.0538,
      "step": 319700
    },
    {
      "epoch": 4.708693539123135,
      "grad_norm": 1.3727226257324219,
      "learning_rate": 1.303010269870465e-06,
      "loss": 0.0529,
      "step": 319725
    },
    {
      "epoch": 4.709061722213222,
      "grad_norm": 1.4479005336761475,
      "learning_rate": 1.3013738979034803e-06,
      "loss": 0.0504,
      "step": 319750
    },
    {
      "epoch": 4.709429905303309,
      "grad_norm": 1.310695767402649,
      "learning_rate": 1.2997375259364958e-06,
      "loss": 0.0475,
      "step": 319775
    },
    {
      "epoch": 4.709798088393397,
      "grad_norm": 0.8655557632446289,
      "learning_rate": 1.2981011539695113e-06,
      "loss": 0.0493,
      "step": 319800
    },
    {
      "epoch": 4.710166271483484,
      "grad_norm": 1.3890773057937622,
      "learning_rate": 1.2964647820025266e-06,
      "loss": 0.0424,
      "step": 319825
    },
    {
      "epoch": 4.710534454573571,
      "grad_norm": 1.1416980028152466,
      "learning_rate": 1.2948284100355421e-06,
      "loss": 0.0532,
      "step": 319850
    },
    {
      "epoch": 4.710902637663658,
      "grad_norm": 1.0013505220413208,
      "learning_rate": 1.2931920380685575e-06,
      "loss": 0.0466,
      "step": 319875
    },
    {
      "epoch": 4.711270820753745,
      "grad_norm": 1.165756106376648,
      "learning_rate": 1.291555666101573e-06,
      "loss": 0.0556,
      "step": 319900
    },
    {
      "epoch": 4.711639003843832,
      "grad_norm": 1.3401901721954346,
      "learning_rate": 1.2899192941345885e-06,
      "loss": 0.0453,
      "step": 319925
    },
    {
      "epoch": 4.712007186933919,
      "grad_norm": 1.4125206470489502,
      "learning_rate": 1.2882829221676038e-06,
      "loss": 0.0433,
      "step": 319950
    },
    {
      "epoch": 4.712375370024006,
      "grad_norm": 1.076216459274292,
      "learning_rate": 1.2866465502006193e-06,
      "loss": 0.0511,
      "step": 319975
    },
    {
      "epoch": 4.712743553114093,
      "grad_norm": 1.5663036108016968,
      "learning_rate": 1.2850101782336346e-06,
      "loss": 0.0518,
      "step": 320000
    },
    {
      "epoch": 4.71311173620418,
      "grad_norm": 1.4842246770858765,
      "learning_rate": 1.2833738062666501e-06,
      "loss": 0.0592,
      "step": 320025
    },
    {
      "epoch": 4.713479919294267,
      "grad_norm": 1.2105618715286255,
      "learning_rate": 1.2817374342996654e-06,
      "loss": 0.05,
      "step": 320050
    },
    {
      "epoch": 4.713848102384354,
      "grad_norm": 1.1992719173431396,
      "learning_rate": 1.280101062332681e-06,
      "loss": 0.0516,
      "step": 320075
    },
    {
      "epoch": 4.714216285474441,
      "grad_norm": 1.4981083869934082,
      "learning_rate": 1.2784646903656967e-06,
      "loss": 0.0522,
      "step": 320100
    },
    {
      "epoch": 4.714584468564528,
      "grad_norm": 1.3090482950210571,
      "learning_rate": 1.276828318398712e-06,
      "loss": 0.0539,
      "step": 320125
    },
    {
      "epoch": 4.714952651654615,
      "grad_norm": 1.3454090356826782,
      "learning_rate": 1.2751919464317275e-06,
      "loss": 0.0549,
      "step": 320150
    },
    {
      "epoch": 4.715320834744702,
      "grad_norm": 1.3377399444580078,
      "learning_rate": 1.273555574464743e-06,
      "loss": 0.0504,
      "step": 320175
    },
    {
      "epoch": 4.715689017834789,
      "grad_norm": 1.4805724620819092,
      "learning_rate": 1.2719192024977583e-06,
      "loss": 0.0515,
      "step": 320200
    },
    {
      "epoch": 4.716057200924876,
      "grad_norm": 1.2056387662887573,
      "learning_rate": 1.2702828305307738e-06,
      "loss": 0.0479,
      "step": 320225
    },
    {
      "epoch": 4.716425384014963,
      "grad_norm": 1.34861159324646,
      "learning_rate": 1.2686464585637891e-06,
      "loss": 0.0482,
      "step": 320250
    },
    {
      "epoch": 4.71679356710505,
      "grad_norm": 0.8846838474273682,
      "learning_rate": 1.2670100865968047e-06,
      "loss": 0.0492,
      "step": 320275
    },
    {
      "epoch": 4.717161750195137,
      "grad_norm": 1.5410267114639282,
      "learning_rate": 1.26537371462982e-06,
      "loss": 0.0478,
      "step": 320300
    },
    {
      "epoch": 4.717529933285224,
      "grad_norm": 1.4455264806747437,
      "learning_rate": 1.2637373426628355e-06,
      "loss": 0.0547,
      "step": 320325
    },
    {
      "epoch": 4.717898116375311,
      "grad_norm": 1.443855881690979,
      "learning_rate": 1.262100970695851e-06,
      "loss": 0.0474,
      "step": 320350
    },
    {
      "epoch": 4.718266299465398,
      "grad_norm": 1.4226480722427368,
      "learning_rate": 1.2604645987288663e-06,
      "loss": 0.0525,
      "step": 320375
    },
    {
      "epoch": 4.718634482555485,
      "grad_norm": 1.2856069803237915,
      "learning_rate": 1.2588282267618818e-06,
      "loss": 0.0495,
      "step": 320400
    },
    {
      "epoch": 4.719002665645572,
      "grad_norm": 1.7483409643173218,
      "learning_rate": 1.2571918547948971e-06,
      "loss": 0.0494,
      "step": 320425
    },
    {
      "epoch": 4.719370848735659,
      "grad_norm": 1.3221842050552368,
      "learning_rate": 1.2555554828279126e-06,
      "loss": 0.0481,
      "step": 320450
    },
    {
      "epoch": 4.719739031825746,
      "grad_norm": 1.5567339658737183,
      "learning_rate": 1.2539191108609281e-06,
      "loss": 0.0499,
      "step": 320475
    },
    {
      "epoch": 4.720107214915833,
      "grad_norm": 1.3565034866333008,
      "learning_rate": 1.2522827388939435e-06,
      "loss": 0.0521,
      "step": 320500
    },
    {
      "epoch": 4.72047539800592,
      "grad_norm": 0.942406415939331,
      "learning_rate": 1.250646366926959e-06,
      "loss": 0.0531,
      "step": 320525
    },
    {
      "epoch": 4.720843581096007,
      "grad_norm": 1.432540774345398,
      "learning_rate": 1.2490099949599745e-06,
      "loss": 0.0516,
      "step": 320550
    },
    {
      "epoch": 4.721211764186094,
      "grad_norm": 0.9517788887023926,
      "learning_rate": 1.2473736229929898e-06,
      "loss": 0.0521,
      "step": 320575
    },
    {
      "epoch": 4.721579947276181,
      "grad_norm": 1.0094650983810425,
      "learning_rate": 1.2457372510260053e-06,
      "loss": 0.0555,
      "step": 320600
    },
    {
      "epoch": 4.721948130366268,
      "grad_norm": 1.7948014736175537,
      "learning_rate": 1.2441008790590208e-06,
      "loss": 0.0538,
      "step": 320625
    },
    {
      "epoch": 4.722316313456355,
      "grad_norm": 1.6364896297454834,
      "learning_rate": 1.2424645070920361e-06,
      "loss": 0.0466,
      "step": 320650
    },
    {
      "epoch": 4.722684496546442,
      "grad_norm": 1.1485779285430908,
      "learning_rate": 1.2408281351250516e-06,
      "loss": 0.0502,
      "step": 320675
    },
    {
      "epoch": 4.723052679636529,
      "grad_norm": 1.5522489547729492,
      "learning_rate": 1.239191763158067e-06,
      "loss": 0.0534,
      "step": 320700
    },
    {
      "epoch": 4.723420862726616,
      "grad_norm": 1.1088215112686157,
      "learning_rate": 1.2375553911910827e-06,
      "loss": 0.049,
      "step": 320725
    },
    {
      "epoch": 4.723789045816703,
      "grad_norm": 1.452671766281128,
      "learning_rate": 1.235919019224098e-06,
      "loss": 0.0522,
      "step": 320750
    },
    {
      "epoch": 4.72415722890679,
      "grad_norm": 1.465901494026184,
      "learning_rate": 1.2342826472571135e-06,
      "loss": 0.0476,
      "step": 320775
    },
    {
      "epoch": 4.724525411996877,
      "grad_norm": 0.9885482788085938,
      "learning_rate": 1.2326462752901288e-06,
      "loss": 0.0492,
      "step": 320800
    },
    {
      "epoch": 4.724893595086964,
      "grad_norm": 1.0932338237762451,
      "learning_rate": 1.2310099033231443e-06,
      "loss": 0.0538,
      "step": 320825
    },
    {
      "epoch": 4.725261778177051,
      "grad_norm": 1.2269691228866577,
      "learning_rate": 1.2293735313561596e-06,
      "loss": 0.0466,
      "step": 320850
    },
    {
      "epoch": 4.725629961267139,
      "grad_norm": 1.564968228340149,
      "learning_rate": 1.2277371593891751e-06,
      "loss": 0.0529,
      "step": 320875
    },
    {
      "epoch": 4.725998144357226,
      "grad_norm": 1.5487475395202637,
      "learning_rate": 1.2261007874221907e-06,
      "loss": 0.0554,
      "step": 320900
    },
    {
      "epoch": 4.726366327447313,
      "grad_norm": 1.544245958328247,
      "learning_rate": 1.224464415455206e-06,
      "loss": 0.0594,
      "step": 320925
    },
    {
      "epoch": 4.7267345105374,
      "grad_norm": 1.680324673652649,
      "learning_rate": 1.2228280434882215e-06,
      "loss": 0.0536,
      "step": 320950
    },
    {
      "epoch": 4.727102693627487,
      "grad_norm": 1.075310468673706,
      "learning_rate": 1.221191671521237e-06,
      "loss": 0.0481,
      "step": 320975
    },
    {
      "epoch": 4.727470876717574,
      "grad_norm": 0.7573911547660828,
      "learning_rate": 1.2195552995542525e-06,
      "loss": 0.056,
      "step": 321000
    },
    {
      "epoch": 4.727839059807661,
      "grad_norm": 1.0710138082504272,
      "learning_rate": 1.2179189275872678e-06,
      "loss": 0.0482,
      "step": 321025
    },
    {
      "epoch": 4.728207242897748,
      "grad_norm": 0.8604738116264343,
      "learning_rate": 1.2162825556202833e-06,
      "loss": 0.0522,
      "step": 321050
    },
    {
      "epoch": 4.728575425987835,
      "grad_norm": 1.9007833003997803,
      "learning_rate": 1.2146461836532986e-06,
      "loss": 0.0536,
      "step": 321075
    },
    {
      "epoch": 4.728943609077922,
      "grad_norm": 1.7964595556259155,
      "learning_rate": 1.2130098116863141e-06,
      "loss": 0.0481,
      "step": 321100
    },
    {
      "epoch": 4.7293117921680095,
      "grad_norm": 0.9984670877456665,
      "learning_rate": 1.2113734397193295e-06,
      "loss": 0.0434,
      "step": 321125
    },
    {
      "epoch": 4.7296799752580965,
      "grad_norm": 0.8985347747802734,
      "learning_rate": 1.209737067752345e-06,
      "loss": 0.0581,
      "step": 321150
    },
    {
      "epoch": 4.7300481583481835,
      "grad_norm": 1.394180178642273,
      "learning_rate": 1.2081006957853605e-06,
      "loss": 0.0499,
      "step": 321175
    },
    {
      "epoch": 4.7304163414382705,
      "grad_norm": 0.9982278347015381,
      "learning_rate": 1.2064643238183758e-06,
      "loss": 0.0503,
      "step": 321200
    },
    {
      "epoch": 4.7307845245283575,
      "grad_norm": 1.2349838018417358,
      "learning_rate": 1.2048279518513913e-06,
      "loss": 0.0513,
      "step": 321225
    },
    {
      "epoch": 4.7311527076184445,
      "grad_norm": 1.577845573425293,
      "learning_rate": 1.2031915798844068e-06,
      "loss": 0.056,
      "step": 321250
    },
    {
      "epoch": 4.7315208907085315,
      "grad_norm": 0.9992119073867798,
      "learning_rate": 1.2015552079174223e-06,
      "loss": 0.0532,
      "step": 321275
    },
    {
      "epoch": 4.731889073798619,
      "grad_norm": 1.7987030744552612,
      "learning_rate": 1.1999188359504376e-06,
      "loss": 0.0503,
      "step": 321300
    },
    {
      "epoch": 4.732257256888706,
      "grad_norm": 1.6102665662765503,
      "learning_rate": 1.1982824639834532e-06,
      "loss": 0.0535,
      "step": 321325
    },
    {
      "epoch": 4.732625439978793,
      "grad_norm": 1.149827480316162,
      "learning_rate": 1.1966460920164685e-06,
      "loss": 0.0433,
      "step": 321350
    },
    {
      "epoch": 4.73299362306888,
      "grad_norm": 1.6134099960327148,
      "learning_rate": 1.195009720049484e-06,
      "loss": 0.0543,
      "step": 321375
    },
    {
      "epoch": 4.733361806158967,
      "grad_norm": 1.3289073705673218,
      "learning_rate": 1.1933733480824993e-06,
      "loss": 0.0524,
      "step": 321400
    },
    {
      "epoch": 4.733729989249054,
      "grad_norm": 1.0444259643554688,
      "learning_rate": 1.1917369761155148e-06,
      "loss": 0.0496,
      "step": 321425
    },
    {
      "epoch": 4.734098172339141,
      "grad_norm": 1.2919409275054932,
      "learning_rate": 1.1901006041485303e-06,
      "loss": 0.0489,
      "step": 321450
    },
    {
      "epoch": 4.734466355429228,
      "grad_norm": 1.063719391822815,
      "learning_rate": 1.1884642321815458e-06,
      "loss": 0.0524,
      "step": 321475
    },
    {
      "epoch": 4.734834538519315,
      "grad_norm": 1.7920410633087158,
      "learning_rate": 1.1868278602145611e-06,
      "loss": 0.048,
      "step": 321500
    },
    {
      "epoch": 4.735202721609402,
      "grad_norm": 1.9228638410568237,
      "learning_rate": 1.1851914882475767e-06,
      "loss": 0.0467,
      "step": 321525
    },
    {
      "epoch": 4.735570904699489,
      "grad_norm": 1.3136377334594727,
      "learning_rate": 1.1835551162805922e-06,
      "loss": 0.0473,
      "step": 321550
    },
    {
      "epoch": 4.735939087789576,
      "grad_norm": 1.263770341873169,
      "learning_rate": 1.1819187443136075e-06,
      "loss": 0.0514,
      "step": 321575
    },
    {
      "epoch": 4.736307270879663,
      "grad_norm": 1.6247344017028809,
      "learning_rate": 1.180282372346623e-06,
      "loss": 0.0545,
      "step": 321600
    },
    {
      "epoch": 4.73667545396975,
      "grad_norm": 1.4460808038711548,
      "learning_rate": 1.1786460003796383e-06,
      "loss": 0.0504,
      "step": 321625
    },
    {
      "epoch": 4.737043637059837,
      "grad_norm": 0.8617902398109436,
      "learning_rate": 1.1770096284126538e-06,
      "loss": 0.0551,
      "step": 321650
    },
    {
      "epoch": 4.737411820149924,
      "grad_norm": 1.2764588594436646,
      "learning_rate": 1.1753732564456691e-06,
      "loss": 0.0548,
      "step": 321675
    },
    {
      "epoch": 4.737780003240011,
      "grad_norm": 1.4454516172409058,
      "learning_rate": 1.1737368844786848e-06,
      "loss": 0.0488,
      "step": 321700
    },
    {
      "epoch": 4.738148186330099,
      "grad_norm": 1.0546947717666626,
      "learning_rate": 1.1721005125117001e-06,
      "loss": 0.0508,
      "step": 321725
    },
    {
      "epoch": 4.738516369420186,
      "grad_norm": 1.3404879570007324,
      "learning_rate": 1.1704641405447157e-06,
      "loss": 0.0571,
      "step": 321750
    },
    {
      "epoch": 4.738884552510273,
      "grad_norm": 1.3486171960830688,
      "learning_rate": 1.168827768577731e-06,
      "loss": 0.0529,
      "step": 321775
    },
    {
      "epoch": 4.73925273560036,
      "grad_norm": 1.0811330080032349,
      "learning_rate": 1.1671913966107465e-06,
      "loss": 0.0574,
      "step": 321800
    },
    {
      "epoch": 4.739620918690447,
      "grad_norm": 1.2224199771881104,
      "learning_rate": 1.165555024643762e-06,
      "loss": 0.0469,
      "step": 321825
    },
    {
      "epoch": 4.739989101780534,
      "grad_norm": 1.5925763845443726,
      "learning_rate": 1.1639186526767773e-06,
      "loss": 0.0521,
      "step": 321850
    },
    {
      "epoch": 4.740357284870621,
      "grad_norm": 1.9667174816131592,
      "learning_rate": 1.1622822807097928e-06,
      "loss": 0.0629,
      "step": 321875
    },
    {
      "epoch": 4.740725467960708,
      "grad_norm": 1.2191787958145142,
      "learning_rate": 1.1606459087428081e-06,
      "loss": 0.0547,
      "step": 321900
    },
    {
      "epoch": 4.741093651050795,
      "grad_norm": 1.41875422000885,
      "learning_rate": 1.1590095367758236e-06,
      "loss": 0.0511,
      "step": 321925
    },
    {
      "epoch": 4.741461834140882,
      "grad_norm": 1.2733855247497559,
      "learning_rate": 1.1573731648088392e-06,
      "loss": 0.0516,
      "step": 321950
    },
    {
      "epoch": 4.741830017230969,
      "grad_norm": 1.1774779558181763,
      "learning_rate": 1.1557367928418547e-06,
      "loss": 0.0479,
      "step": 321975
    },
    {
      "epoch": 4.742198200321056,
      "grad_norm": 1.20047128200531,
      "learning_rate": 1.15410042087487e-06,
      "loss": 0.0528,
      "step": 322000
    },
    {
      "epoch": 4.742566383411143,
      "grad_norm": 1.8065284490585327,
      "learning_rate": 1.1524640489078855e-06,
      "loss": 0.0515,
      "step": 322025
    },
    {
      "epoch": 4.74293456650123,
      "grad_norm": 1.3913235664367676,
      "learning_rate": 1.1508276769409008e-06,
      "loss": 0.0512,
      "step": 322050
    },
    {
      "epoch": 4.743302749591317,
      "grad_norm": 1.7780894041061401,
      "learning_rate": 1.1491913049739163e-06,
      "loss": 0.049,
      "step": 322075
    },
    {
      "epoch": 4.743670932681404,
      "grad_norm": 0.8062098026275635,
      "learning_rate": 1.1475549330069318e-06,
      "loss": 0.0557,
      "step": 322100
    },
    {
      "epoch": 4.744039115771491,
      "grad_norm": 1.3578803539276123,
      "learning_rate": 1.1459185610399471e-06,
      "loss": 0.0547,
      "step": 322125
    },
    {
      "epoch": 4.744407298861578,
      "grad_norm": 1.3337706327438354,
      "learning_rate": 1.1442821890729627e-06,
      "loss": 0.0597,
      "step": 322150
    },
    {
      "epoch": 4.744775481951665,
      "grad_norm": 1.9644591808319092,
      "learning_rate": 1.142645817105978e-06,
      "loss": 0.0532,
      "step": 322175
    },
    {
      "epoch": 4.745143665041752,
      "grad_norm": 1.4632443189620972,
      "learning_rate": 1.1410094451389937e-06,
      "loss": 0.0543,
      "step": 322200
    },
    {
      "epoch": 4.745511848131839,
      "grad_norm": 1.484179139137268,
      "learning_rate": 1.139373073172009e-06,
      "loss": 0.0524,
      "step": 322225
    },
    {
      "epoch": 4.745880031221926,
      "grad_norm": 1.3495906591415405,
      "learning_rate": 1.1377367012050245e-06,
      "loss": 0.054,
      "step": 322250
    },
    {
      "epoch": 4.746248214312013,
      "grad_norm": 0.9905472993850708,
      "learning_rate": 1.1361003292380398e-06,
      "loss": 0.0486,
      "step": 322275
    },
    {
      "epoch": 4.7466163974021,
      "grad_norm": 1.2342287302017212,
      "learning_rate": 1.1344639572710553e-06,
      "loss": 0.0504,
      "step": 322300
    },
    {
      "epoch": 4.746984580492187,
      "grad_norm": 0.6113099455833435,
      "learning_rate": 1.1328275853040706e-06,
      "loss": 0.0415,
      "step": 322325
    },
    {
      "epoch": 4.747352763582274,
      "grad_norm": 1.0404375791549683,
      "learning_rate": 1.1311912133370861e-06,
      "loss": 0.0504,
      "step": 322350
    },
    {
      "epoch": 4.747720946672361,
      "grad_norm": 1.4372038841247559,
      "learning_rate": 1.1295548413701017e-06,
      "loss": 0.0535,
      "step": 322375
    },
    {
      "epoch": 4.748089129762448,
      "grad_norm": 1.7727999687194824,
      "learning_rate": 1.127918469403117e-06,
      "loss": 0.0537,
      "step": 322400
    },
    {
      "epoch": 4.748457312852535,
      "grad_norm": 1.2488081455230713,
      "learning_rate": 1.1262820974361325e-06,
      "loss": 0.0501,
      "step": 322425
    },
    {
      "epoch": 4.748825495942622,
      "grad_norm": 0.9674785137176514,
      "learning_rate": 1.124645725469148e-06,
      "loss": 0.0542,
      "step": 322450
    },
    {
      "epoch": 4.749193679032709,
      "grad_norm": 1.5368746519088745,
      "learning_rate": 1.1230093535021635e-06,
      "loss": 0.053,
      "step": 322475
    },
    {
      "epoch": 4.749561862122796,
      "grad_norm": 1.7688888311386108,
      "learning_rate": 1.1213729815351788e-06,
      "loss": 0.0516,
      "step": 322500
    },
    {
      "epoch": 4.749930045212883,
      "grad_norm": 1.4937430620193481,
      "learning_rate": 1.1197366095681943e-06,
      "loss": 0.0507,
      "step": 322525
    },
    {
      "epoch": 4.75029822830297,
      "grad_norm": 1.249537706375122,
      "learning_rate": 1.1181002376012096e-06,
      "loss": 0.0505,
      "step": 322550
    },
    {
      "epoch": 4.750666411393057,
      "grad_norm": 1.3840093612670898,
      "learning_rate": 1.1164638656342252e-06,
      "loss": 0.0462,
      "step": 322575
    },
    {
      "epoch": 4.751034594483144,
      "grad_norm": 1.5492513179779053,
      "learning_rate": 1.1148274936672405e-06,
      "loss": 0.058,
      "step": 322600
    },
    {
      "epoch": 4.751402777573231,
      "grad_norm": 1.401488184928894,
      "learning_rate": 1.113191121700256e-06,
      "loss": 0.0512,
      "step": 322625
    },
    {
      "epoch": 4.751770960663318,
      "grad_norm": 1.3748246431350708,
      "learning_rate": 1.1115547497332715e-06,
      "loss": 0.061,
      "step": 322650
    },
    {
      "epoch": 4.752139143753405,
      "grad_norm": 1.383940577507019,
      "learning_rate": 1.109918377766287e-06,
      "loss": 0.0501,
      "step": 322675
    },
    {
      "epoch": 4.752507326843492,
      "grad_norm": 0.9253655672073364,
      "learning_rate": 1.1082820057993023e-06,
      "loss": 0.0528,
      "step": 322700
    },
    {
      "epoch": 4.752875509933579,
      "grad_norm": 0.9685051441192627,
      "learning_rate": 1.1066456338323178e-06,
      "loss": 0.0503,
      "step": 322725
    },
    {
      "epoch": 4.753243693023666,
      "grad_norm": 0.9926207661628723,
      "learning_rate": 1.1050092618653333e-06,
      "loss": 0.0534,
      "step": 322750
    },
    {
      "epoch": 4.753611876113753,
      "grad_norm": 1.3221259117126465,
      "learning_rate": 1.1033728898983486e-06,
      "loss": 0.0512,
      "step": 322775
    },
    {
      "epoch": 4.753980059203841,
      "grad_norm": 0.907096803188324,
      "learning_rate": 1.1017365179313642e-06,
      "loss": 0.0488,
      "step": 322800
    },
    {
      "epoch": 4.754348242293928,
      "grad_norm": 1.1337298154830933,
      "learning_rate": 1.1001001459643795e-06,
      "loss": 0.0536,
      "step": 322825
    },
    {
      "epoch": 4.754716425384015,
      "grad_norm": 1.404736042022705,
      "learning_rate": 1.098463773997395e-06,
      "loss": 0.0527,
      "step": 322850
    },
    {
      "epoch": 4.755084608474102,
      "grad_norm": 1.2021695375442505,
      "learning_rate": 1.0968274020304103e-06,
      "loss": 0.05,
      "step": 322875
    },
    {
      "epoch": 4.755452791564189,
      "grad_norm": 0.8419737815856934,
      "learning_rate": 1.0951910300634258e-06,
      "loss": 0.0549,
      "step": 322900
    },
    {
      "epoch": 4.755820974654276,
      "grad_norm": 1.5595836639404297,
      "learning_rate": 1.0935546580964413e-06,
      "loss": 0.05,
      "step": 322925
    },
    {
      "epoch": 4.756189157744363,
      "grad_norm": 1.1508982181549072,
      "learning_rate": 1.0919182861294568e-06,
      "loss": 0.0494,
      "step": 322950
    },
    {
      "epoch": 4.75655734083445,
      "grad_norm": 1.384660005569458,
      "learning_rate": 1.0902819141624721e-06,
      "loss": 0.0548,
      "step": 322975
    },
    {
      "epoch": 4.756925523924537,
      "grad_norm": 1.290677785873413,
      "learning_rate": 1.0886455421954877e-06,
      "loss": 0.0418,
      "step": 323000
    },
    {
      "epoch": 4.757293707014624,
      "grad_norm": 1.8103145360946655,
      "learning_rate": 1.0870091702285032e-06,
      "loss": 0.0547,
      "step": 323025
    },
    {
      "epoch": 4.757661890104711,
      "grad_norm": 1.6526089906692505,
      "learning_rate": 1.0853727982615185e-06,
      "loss": 0.0546,
      "step": 323050
    },
    {
      "epoch": 4.758030073194798,
      "grad_norm": 0.9802822470664978,
      "learning_rate": 1.083736426294534e-06,
      "loss": 0.0565,
      "step": 323075
    },
    {
      "epoch": 4.758398256284885,
      "grad_norm": 1.385122299194336,
      "learning_rate": 1.0821000543275493e-06,
      "loss": 0.0484,
      "step": 323100
    },
    {
      "epoch": 4.7587664393749725,
      "grad_norm": 0.9054996371269226,
      "learning_rate": 1.0804636823605648e-06,
      "loss": 0.0463,
      "step": 323125
    },
    {
      "epoch": 4.7591346224650595,
      "grad_norm": 1.4178301095962524,
      "learning_rate": 1.0788273103935801e-06,
      "loss": 0.0507,
      "step": 323150
    },
    {
      "epoch": 4.7595028055551465,
      "grad_norm": 0.997786819934845,
      "learning_rate": 1.0771909384265958e-06,
      "loss": 0.0566,
      "step": 323175
    },
    {
      "epoch": 4.7598709886452335,
      "grad_norm": 1.315610408782959,
      "learning_rate": 1.0755545664596112e-06,
      "loss": 0.0538,
      "step": 323200
    },
    {
      "epoch": 4.7602391717353205,
      "grad_norm": 1.4628230333328247,
      "learning_rate": 1.0739181944926267e-06,
      "loss": 0.0502,
      "step": 323225
    },
    {
      "epoch": 4.7606073548254075,
      "grad_norm": 1.667850136756897,
      "learning_rate": 1.072281822525642e-06,
      "loss": 0.0556,
      "step": 323250
    },
    {
      "epoch": 4.7609755379154945,
      "grad_norm": 0.65080726146698,
      "learning_rate": 1.0706454505586575e-06,
      "loss": 0.0532,
      "step": 323275
    },
    {
      "epoch": 4.761343721005582,
      "grad_norm": 0.9832432866096497,
      "learning_rate": 1.069009078591673e-06,
      "loss": 0.0473,
      "step": 323300
    },
    {
      "epoch": 4.761711904095669,
      "grad_norm": 1.011878252029419,
      "learning_rate": 1.0673727066246883e-06,
      "loss": 0.0556,
      "step": 323325
    },
    {
      "epoch": 4.762080087185756,
      "grad_norm": 0.9791728258132935,
      "learning_rate": 1.0657363346577038e-06,
      "loss": 0.0468,
      "step": 323350
    },
    {
      "epoch": 4.762448270275843,
      "grad_norm": 1.6162090301513672,
      "learning_rate": 1.0640999626907191e-06,
      "loss": 0.0514,
      "step": 323375
    },
    {
      "epoch": 4.76281645336593,
      "grad_norm": 1.2402819395065308,
      "learning_rate": 1.062529045602414e-06,
      "loss": 0.0546,
      "step": 323400
    },
    {
      "epoch": 4.763184636456017,
      "grad_norm": 1.3816293478012085,
      "learning_rate": 1.0608926736354296e-06,
      "loss": 0.0492,
      "step": 323425
    },
    {
      "epoch": 4.763552819546104,
      "grad_norm": 1.1788517236709595,
      "learning_rate": 1.0592563016684449e-06,
      "loss": 0.051,
      "step": 323450
    },
    {
      "epoch": 4.763921002636191,
      "grad_norm": 1.0061683654785156,
      "learning_rate": 1.0576199297014604e-06,
      "loss": 0.0455,
      "step": 323475
    },
    {
      "epoch": 4.764289185726278,
      "grad_norm": 1.5762227773666382,
      "learning_rate": 1.055983557734476e-06,
      "loss": 0.0527,
      "step": 323500
    },
    {
      "epoch": 4.764657368816365,
      "grad_norm": 1.380527377128601,
      "learning_rate": 1.0543471857674912e-06,
      "loss": 0.0486,
      "step": 323525
    },
    {
      "epoch": 4.765025551906452,
      "grad_norm": 1.0709291696548462,
      "learning_rate": 1.0527108138005067e-06,
      "loss": 0.054,
      "step": 323550
    },
    {
      "epoch": 4.765393734996539,
      "grad_norm": 1.018418312072754,
      "learning_rate": 1.051074441833522e-06,
      "loss": 0.0484,
      "step": 323575
    },
    {
      "epoch": 4.765761918086626,
      "grad_norm": 1.36138916015625,
      "learning_rate": 1.0494380698665376e-06,
      "loss": 0.0524,
      "step": 323600
    },
    {
      "epoch": 4.766130101176714,
      "grad_norm": 0.963590681552887,
      "learning_rate": 1.0478016978995529e-06,
      "loss": 0.0479,
      "step": 323625
    },
    {
      "epoch": 4.766498284266801,
      "grad_norm": 1.878432035446167,
      "learning_rate": 1.0461653259325686e-06,
      "loss": 0.0559,
      "step": 323650
    },
    {
      "epoch": 4.766866467356888,
      "grad_norm": 0.9522743225097656,
      "learning_rate": 1.044528953965584e-06,
      "loss": 0.0472,
      "step": 323675
    },
    {
      "epoch": 4.767234650446975,
      "grad_norm": 1.4975364208221436,
      "learning_rate": 1.0428925819985994e-06,
      "loss": 0.055,
      "step": 323700
    },
    {
      "epoch": 4.767602833537062,
      "grad_norm": 1.1804375648498535,
      "learning_rate": 1.0412562100316147e-06,
      "loss": 0.0511,
      "step": 323725
    },
    {
      "epoch": 4.767971016627149,
      "grad_norm": 1.4956169128417969,
      "learning_rate": 1.0396198380646302e-06,
      "loss": 0.0506,
      "step": 323750
    },
    {
      "epoch": 4.768339199717236,
      "grad_norm": 1.0784454345703125,
      "learning_rate": 1.0379834660976458e-06,
      "loss": 0.0491,
      "step": 323775
    },
    {
      "epoch": 4.768707382807323,
      "grad_norm": 1.3169667720794678,
      "learning_rate": 1.036347094130661e-06,
      "loss": 0.0509,
      "step": 323800
    },
    {
      "epoch": 4.76907556589741,
      "grad_norm": 1.6832619905471802,
      "learning_rate": 1.0347107221636766e-06,
      "loss": 0.0487,
      "step": 323825
    },
    {
      "epoch": 4.769443748987497,
      "grad_norm": 1.136932134628296,
      "learning_rate": 1.0330743501966919e-06,
      "loss": 0.0505,
      "step": 323850
    },
    {
      "epoch": 4.769811932077584,
      "grad_norm": 1.147109031677246,
      "learning_rate": 1.0314379782297074e-06,
      "loss": 0.0543,
      "step": 323875
    },
    {
      "epoch": 4.770180115167671,
      "grad_norm": 1.5984269380569458,
      "learning_rate": 1.029801606262723e-06,
      "loss": 0.0486,
      "step": 323900
    },
    {
      "epoch": 4.770548298257758,
      "grad_norm": 1.9397368431091309,
      "learning_rate": 1.0281652342957384e-06,
      "loss": 0.0519,
      "step": 323925
    },
    {
      "epoch": 4.770916481347845,
      "grad_norm": 1.4014109373092651,
      "learning_rate": 1.0265288623287537e-06,
      "loss": 0.0516,
      "step": 323950
    },
    {
      "epoch": 4.771284664437932,
      "grad_norm": 0.9060550928115845,
      "learning_rate": 1.0248924903617692e-06,
      "loss": 0.0489,
      "step": 323975
    },
    {
      "epoch": 4.771652847528019,
      "grad_norm": 0.9443454146385193,
      "learning_rate": 1.0232561183947846e-06,
      "loss": 0.0494,
      "step": 324000
    },
    {
      "epoch": 4.772021030618106,
      "grad_norm": 1.1277412176132202,
      "learning_rate": 1.0216197464278e-06,
      "loss": 0.0497,
      "step": 324025
    },
    {
      "epoch": 4.772389213708193,
      "grad_norm": 1.4183262586593628,
      "learning_rate": 1.0199833744608156e-06,
      "loss": 0.0546,
      "step": 324050
    },
    {
      "epoch": 4.77275739679828,
      "grad_norm": 1.7397887706756592,
      "learning_rate": 1.0183470024938309e-06,
      "loss": 0.0525,
      "step": 324075
    },
    {
      "epoch": 4.773125579888367,
      "grad_norm": 1.469543695449829,
      "learning_rate": 1.0167106305268464e-06,
      "loss": 0.0575,
      "step": 324100
    },
    {
      "epoch": 4.773493762978454,
      "grad_norm": 1.4543373584747314,
      "learning_rate": 1.0150742585598617e-06,
      "loss": 0.0514,
      "step": 324125
    },
    {
      "epoch": 4.773861946068541,
      "grad_norm": 1.2241088151931763,
      "learning_rate": 1.0134378865928774e-06,
      "loss": 0.0488,
      "step": 324150
    },
    {
      "epoch": 4.774230129158628,
      "grad_norm": 0.8647933006286621,
      "learning_rate": 1.0118015146258927e-06,
      "loss": 0.0518,
      "step": 324175
    },
    {
      "epoch": 4.774598312248715,
      "grad_norm": 0.7101517915725708,
      "learning_rate": 1.0101651426589083e-06,
      "loss": 0.0564,
      "step": 324200
    },
    {
      "epoch": 4.774966495338802,
      "grad_norm": 1.52452552318573,
      "learning_rate": 1.0085287706919236e-06,
      "loss": 0.0585,
      "step": 324225
    },
    {
      "epoch": 4.775334678428889,
      "grad_norm": 1.1051076650619507,
      "learning_rate": 1.006892398724939e-06,
      "loss": 0.0503,
      "step": 324250
    },
    {
      "epoch": 4.775702861518976,
      "grad_norm": 1.1222940683364868,
      "learning_rate": 1.0052560267579544e-06,
      "loss": 0.0514,
      "step": 324275
    },
    {
      "epoch": 4.776071044609063,
      "grad_norm": 1.3647911548614502,
      "learning_rate": 1.00361965479097e-06,
      "loss": 0.0459,
      "step": 324300
    },
    {
      "epoch": 4.77643922769915,
      "grad_norm": 0.4823213815689087,
      "learning_rate": 1.0019832828239854e-06,
      "loss": 0.0477,
      "step": 324325
    },
    {
      "epoch": 4.776807410789237,
      "grad_norm": 1.3930208683013916,
      "learning_rate": 1.0003469108570007e-06,
      "loss": 0.0523,
      "step": 324350
    },
    {
      "epoch": 4.777175593879324,
      "grad_norm": 1.2660657167434692,
      "learning_rate": 9.987759937686957e-07,
      "loss": 0.061,
      "step": 324375
    },
    {
      "epoch": 4.777543776969411,
      "grad_norm": 1.159363865852356,
      "learning_rate": 9.971396218017112e-07,
      "loss": 0.0516,
      "step": 324400
    },
    {
      "epoch": 4.777911960059498,
      "grad_norm": 1.0487005710601807,
      "learning_rate": 9.955032498347265e-07,
      "loss": 0.0479,
      "step": 324425
    },
    {
      "epoch": 4.778280143149585,
      "grad_norm": 1.370375394821167,
      "learning_rate": 9.93866877867742e-07,
      "loss": 0.0504,
      "step": 324450
    },
    {
      "epoch": 4.778648326239672,
      "grad_norm": 1.228419303894043,
      "learning_rate": 9.922305059007573e-07,
      "loss": 0.0498,
      "step": 324475
    },
    {
      "epoch": 4.779016509329759,
      "grad_norm": 0.745658814907074,
      "learning_rate": 9.905941339337728e-07,
      "loss": 0.0433,
      "step": 324500
    },
    {
      "epoch": 4.779384692419846,
      "grad_norm": 1.2757568359375,
      "learning_rate": 9.889577619667883e-07,
      "loss": 0.0562,
      "step": 324525
    },
    {
      "epoch": 4.779752875509933,
      "grad_norm": 1.2557812929153442,
      "learning_rate": 9.873213899998036e-07,
      "loss": 0.0588,
      "step": 324550
    },
    {
      "epoch": 4.78012105860002,
      "grad_norm": 1.9373170137405396,
      "learning_rate": 9.856850180328192e-07,
      "loss": 0.0595,
      "step": 324575
    },
    {
      "epoch": 4.780489241690107,
      "grad_norm": 1.338927984237671,
      "learning_rate": 9.840486460658345e-07,
      "loss": 0.0549,
      "step": 324600
    },
    {
      "epoch": 4.780857424780194,
      "grad_norm": 1.3716917037963867,
      "learning_rate": 9.824122740988502e-07,
      "loss": 0.0458,
      "step": 324625
    },
    {
      "epoch": 4.781225607870281,
      "grad_norm": 0.8940625190734863,
      "learning_rate": 9.807759021318655e-07,
      "loss": 0.0495,
      "step": 324650
    },
    {
      "epoch": 4.781593790960368,
      "grad_norm": 1.1801742315292358,
      "learning_rate": 9.79139530164881e-07,
      "loss": 0.0495,
      "step": 324675
    },
    {
      "epoch": 4.781961974050456,
      "grad_norm": 1.4311069250106812,
      "learning_rate": 9.775031581978963e-07,
      "loss": 0.0513,
      "step": 324700
    },
    {
      "epoch": 4.782330157140543,
      "grad_norm": 1.0667465925216675,
      "learning_rate": 9.758667862309118e-07,
      "loss": 0.0416,
      "step": 324725
    },
    {
      "epoch": 4.78269834023063,
      "grad_norm": 0.8358407616615295,
      "learning_rate": 9.742304142639271e-07,
      "loss": 0.0477,
      "step": 324750
    },
    {
      "epoch": 4.783066523320717,
      "grad_norm": 1.2532055377960205,
      "learning_rate": 9.725940422969426e-07,
      "loss": 0.0568,
      "step": 324775
    },
    {
      "epoch": 4.783434706410804,
      "grad_norm": 0.9507465958595276,
      "learning_rate": 9.709576703299582e-07,
      "loss": 0.0485,
      "step": 324800
    },
    {
      "epoch": 4.783802889500891,
      "grad_norm": 1.4061230421066284,
      "learning_rate": 9.693212983629735e-07,
      "loss": 0.0483,
      "step": 324825
    },
    {
      "epoch": 4.784171072590978,
      "grad_norm": 1.0594359636306763,
      "learning_rate": 9.67684926395989e-07,
      "loss": 0.0517,
      "step": 324850
    },
    {
      "epoch": 4.784539255681065,
      "grad_norm": 1.1768832206726074,
      "learning_rate": 9.660485544290045e-07,
      "loss": 0.0472,
      "step": 324875
    },
    {
      "epoch": 4.784907438771152,
      "grad_norm": 1.840198040008545,
      "learning_rate": 9.6441218246202e-07,
      "loss": 0.0556,
      "step": 324900
    },
    {
      "epoch": 4.785275621861239,
      "grad_norm": 1.4787946939468384,
      "learning_rate": 9.627758104950353e-07,
      "loss": 0.0484,
      "step": 324925
    },
    {
      "epoch": 4.785643804951326,
      "grad_norm": 1.386216402053833,
      "learning_rate": 9.611394385280508e-07,
      "loss": 0.0448,
      "step": 324950
    },
    {
      "epoch": 4.786011988041413,
      "grad_norm": 1.3710432052612305,
      "learning_rate": 9.595030665610661e-07,
      "loss": 0.0531,
      "step": 324975
    },
    {
      "epoch": 4.7863801711315,
      "grad_norm": 1.7313579320907593,
      "learning_rate": 9.578666945940817e-07,
      "loss": 0.0501,
      "step": 325000
    },
    {
      "epoch": 4.786748354221587,
      "grad_norm": 1.5621999502182007,
      "learning_rate": 9.56230322627097e-07,
      "loss": 0.0455,
      "step": 325025
    },
    {
      "epoch": 4.787116537311674,
      "grad_norm": 0.8633673191070557,
      "learning_rate": 9.545939506601125e-07,
      "loss": 0.0492,
      "step": 325050
    },
    {
      "epoch": 4.787484720401761,
      "grad_norm": 1.5263844728469849,
      "learning_rate": 9.529575786931279e-07,
      "loss": 0.0504,
      "step": 325075
    },
    {
      "epoch": 4.7878529034918484,
      "grad_norm": 1.505133867263794,
      "learning_rate": 9.513212067261435e-07,
      "loss": 0.0513,
      "step": 325100
    },
    {
      "epoch": 4.7882210865819355,
      "grad_norm": 1.5316458940505981,
      "learning_rate": 9.496848347591589e-07,
      "loss": 0.0526,
      "step": 325125
    },
    {
      "epoch": 4.7885892696720225,
      "grad_norm": 0.865847647190094,
      "learning_rate": 9.480484627921743e-07,
      "loss": 0.0479,
      "step": 325150
    },
    {
      "epoch": 4.7889574527621095,
      "grad_norm": 1.2504726648330688,
      "learning_rate": 9.464120908251897e-07,
      "loss": 0.0482,
      "step": 325175
    },
    {
      "epoch": 4.7893256358521965,
      "grad_norm": 1.563200831413269,
      "learning_rate": 9.447757188582052e-07,
      "loss": 0.0521,
      "step": 325200
    },
    {
      "epoch": 4.7896938189422835,
      "grad_norm": 1.4421212673187256,
      "learning_rate": 9.431393468912206e-07,
      "loss": 0.056,
      "step": 325225
    },
    {
      "epoch": 4.7900620020323705,
      "grad_norm": 1.4201912879943848,
      "learning_rate": 9.415029749242361e-07,
      "loss": 0.0527,
      "step": 325250
    },
    {
      "epoch": 4.7904301851224576,
      "grad_norm": 1.103562831878662,
      "learning_rate": 9.398666029572515e-07,
      "loss": 0.051,
      "step": 325275
    },
    {
      "epoch": 4.790798368212545,
      "grad_norm": 0.8093121647834778,
      "learning_rate": 9.382302309902669e-07,
      "loss": 0.046,
      "step": 325300
    },
    {
      "epoch": 4.791166551302632,
      "grad_norm": 1.7056058645248413,
      "learning_rate": 9.365938590232823e-07,
      "loss": 0.0486,
      "step": 325325
    },
    {
      "epoch": 4.791534734392719,
      "grad_norm": 1.1183526515960693,
      "learning_rate": 9.349574870562979e-07,
      "loss": 0.0478,
      "step": 325350
    },
    {
      "epoch": 4.791902917482806,
      "grad_norm": 1.2861144542694092,
      "learning_rate": 9.333211150893133e-07,
      "loss": 0.0482,
      "step": 325375
    },
    {
      "epoch": 4.792271100572893,
      "grad_norm": 1.2314645051956177,
      "learning_rate": 9.316847431223288e-07,
      "loss": 0.0504,
      "step": 325400
    },
    {
      "epoch": 4.79263928366298,
      "grad_norm": 1.6391178369522095,
      "learning_rate": 9.300483711553442e-07,
      "loss": 0.0521,
      "step": 325425
    },
    {
      "epoch": 4.793007466753067,
      "grad_norm": 1.5589710474014282,
      "learning_rate": 9.284119991883596e-07,
      "loss": 0.0503,
      "step": 325450
    },
    {
      "epoch": 4.793375649843154,
      "grad_norm": 1.2693525552749634,
      "learning_rate": 9.26775627221375e-07,
      "loss": 0.0533,
      "step": 325475
    },
    {
      "epoch": 4.793743832933241,
      "grad_norm": 0.7547500729560852,
      "learning_rate": 9.251392552543904e-07,
      "loss": 0.0519,
      "step": 325500
    },
    {
      "epoch": 4.794112016023328,
      "grad_norm": 1.2113326787948608,
      "learning_rate": 9.235028832874059e-07,
      "loss": 0.0463,
      "step": 325525
    },
    {
      "epoch": 4.794480199113416,
      "grad_norm": 1.2797787189483643,
      "learning_rate": 9.218665113204213e-07,
      "loss": 0.0561,
      "step": 325550
    },
    {
      "epoch": 4.794848382203503,
      "grad_norm": 1.2187013626098633,
      "learning_rate": 9.202301393534367e-07,
      "loss": 0.049,
      "step": 325575
    },
    {
      "epoch": 4.79521656529359,
      "grad_norm": 1.6214213371276855,
      "learning_rate": 9.185937673864522e-07,
      "loss": 0.0517,
      "step": 325600
    },
    {
      "epoch": 4.795584748383677,
      "grad_norm": 1.0034376382827759,
      "learning_rate": 9.169573954194678e-07,
      "loss": 0.0518,
      "step": 325625
    },
    {
      "epoch": 4.795952931473764,
      "grad_norm": 1.2485432624816895,
      "learning_rate": 9.153210234524832e-07,
      "loss": 0.0531,
      "step": 325650
    },
    {
      "epoch": 4.796321114563851,
      "grad_norm": 1.3036500215530396,
      "learning_rate": 9.136846514854986e-07,
      "loss": 0.0526,
      "step": 325675
    },
    {
      "epoch": 4.796689297653938,
      "grad_norm": 1.1776338815689087,
      "learning_rate": 9.12048279518514e-07,
      "loss": 0.0612,
      "step": 325700
    },
    {
      "epoch": 4.797057480744025,
      "grad_norm": 1.1754858493804932,
      "learning_rate": 9.104119075515294e-07,
      "loss": 0.0493,
      "step": 325725
    },
    {
      "epoch": 4.797425663834112,
      "grad_norm": 0.8314536809921265,
      "learning_rate": 9.087755355845448e-07,
      "loss": 0.0503,
      "step": 325750
    },
    {
      "epoch": 4.797793846924199,
      "grad_norm": 1.3675329685211182,
      "learning_rate": 9.071391636175602e-07,
      "loss": 0.0468,
      "step": 325775
    },
    {
      "epoch": 4.798162030014286,
      "grad_norm": 1.031081199645996,
      "learning_rate": 9.055027916505757e-07,
      "loss": 0.0599,
      "step": 325800
    },
    {
      "epoch": 4.798530213104373,
      "grad_norm": 0.9318610429763794,
      "learning_rate": 9.038664196835912e-07,
      "loss": 0.0526,
      "step": 325825
    },
    {
      "epoch": 4.79889839619446,
      "grad_norm": 1.7119038105010986,
      "learning_rate": 9.022300477166067e-07,
      "loss": 0.0524,
      "step": 325850
    },
    {
      "epoch": 4.799266579284547,
      "grad_norm": 1.085234522819519,
      "learning_rate": 9.005936757496221e-07,
      "loss": 0.0512,
      "step": 325875
    },
    {
      "epoch": 4.799634762374634,
      "grad_norm": 1.5839431285858154,
      "learning_rate": 8.989573037826376e-07,
      "loss": 0.0502,
      "step": 325900
    },
    {
      "epoch": 4.800002945464721,
      "grad_norm": 0.8413432836532593,
      "learning_rate": 8.97320931815653e-07,
      "loss": 0.0509,
      "step": 325925
    },
    {
      "epoch": 4.800371128554808,
      "grad_norm": 1.1245535612106323,
      "learning_rate": 8.956845598486684e-07,
      "loss": 0.0579,
      "step": 325950
    },
    {
      "epoch": 4.800739311644895,
      "grad_norm": 1.8375391960144043,
      "learning_rate": 8.940481878816838e-07,
      "loss": 0.0581,
      "step": 325975
    },
    {
      "epoch": 4.801107494734982,
      "grad_norm": 1.3693115711212158,
      "learning_rate": 8.924118159146992e-07,
      "loss": 0.0499,
      "step": 326000
    },
    {
      "epoch": 4.801475677825069,
      "grad_norm": 1.5521268844604492,
      "learning_rate": 8.907754439477146e-07,
      "loss": 0.0571,
      "step": 326025
    },
    {
      "epoch": 4.801843860915156,
      "grad_norm": 1.5986840724945068,
      "learning_rate": 8.891390719807301e-07,
      "loss": 0.0602,
      "step": 326050
    },
    {
      "epoch": 4.802212044005243,
      "grad_norm": 0.8774245381355286,
      "learning_rate": 8.875027000137457e-07,
      "loss": 0.0482,
      "step": 326075
    },
    {
      "epoch": 4.80258022709533,
      "grad_norm": 1.327412724494934,
      "learning_rate": 8.858663280467611e-07,
      "loss": 0.0568,
      "step": 326100
    },
    {
      "epoch": 4.802948410185417,
      "grad_norm": 1.7069092988967896,
      "learning_rate": 8.842299560797765e-07,
      "loss": 0.0502,
      "step": 326125
    },
    {
      "epoch": 4.803316593275504,
      "grad_norm": 0.9696958661079407,
      "learning_rate": 8.825935841127919e-07,
      "loss": 0.0564,
      "step": 326150
    },
    {
      "epoch": 4.803684776365591,
      "grad_norm": 0.9806936979293823,
      "learning_rate": 8.809572121458074e-07,
      "loss": 0.0529,
      "step": 326175
    },
    {
      "epoch": 4.804052959455678,
      "grad_norm": 1.2925121784210205,
      "learning_rate": 8.793208401788228e-07,
      "loss": 0.0476,
      "step": 326200
    },
    {
      "epoch": 4.804421142545765,
      "grad_norm": 1.307326078414917,
      "learning_rate": 8.776844682118382e-07,
      "loss": 0.0436,
      "step": 326225
    },
    {
      "epoch": 4.804789325635852,
      "grad_norm": 1.6416270732879639,
      "learning_rate": 8.760480962448537e-07,
      "loss": 0.048,
      "step": 326250
    },
    {
      "epoch": 4.805157508725939,
      "grad_norm": 2.0258803367614746,
      "learning_rate": 8.744117242778691e-07,
      "loss": 0.0577,
      "step": 326275
    },
    {
      "epoch": 4.805525691816026,
      "grad_norm": 1.1298747062683105,
      "learning_rate": 8.727753523108845e-07,
      "loss": 0.0524,
      "step": 326300
    },
    {
      "epoch": 4.805893874906113,
      "grad_norm": 1.1773145198822021,
      "learning_rate": 8.711389803439001e-07,
      "loss": 0.0468,
      "step": 326325
    },
    {
      "epoch": 4.8062620579962,
      "grad_norm": 0.7392027378082275,
      "learning_rate": 8.695026083769155e-07,
      "loss": 0.0534,
      "step": 326350
    },
    {
      "epoch": 4.806630241086287,
      "grad_norm": 1.398547649383545,
      "learning_rate": 8.678662364099309e-07,
      "loss": 0.0499,
      "step": 326375
    },
    {
      "epoch": 4.806998424176374,
      "grad_norm": 1.8869794607162476,
      "learning_rate": 8.662298644429463e-07,
      "loss": 0.0477,
      "step": 326400
    },
    {
      "epoch": 4.807366607266461,
      "grad_norm": 1.2947700023651123,
      "learning_rate": 8.645934924759617e-07,
      "loss": 0.0534,
      "step": 326425
    },
    {
      "epoch": 4.807734790356548,
      "grad_norm": 1.1962100267410278,
      "learning_rate": 8.629571205089773e-07,
      "loss": 0.0543,
      "step": 326450
    },
    {
      "epoch": 4.808102973446635,
      "grad_norm": 1.1489663124084473,
      "learning_rate": 8.613207485419927e-07,
      "loss": 0.0518,
      "step": 326475
    },
    {
      "epoch": 4.808471156536722,
      "grad_norm": 1.3486053943634033,
      "learning_rate": 8.596843765750081e-07,
      "loss": 0.0504,
      "step": 326500
    },
    {
      "epoch": 4.808839339626809,
      "grad_norm": 1.1370948553085327,
      "learning_rate": 8.580480046080235e-07,
      "loss": 0.0493,
      "step": 326525
    },
    {
      "epoch": 4.809207522716896,
      "grad_norm": 1.1146185398101807,
      "learning_rate": 8.564116326410389e-07,
      "loss": 0.0541,
      "step": 326550
    },
    {
      "epoch": 4.809575705806983,
      "grad_norm": 1.0838857889175415,
      "learning_rate": 8.547752606740545e-07,
      "loss": 0.0535,
      "step": 326575
    },
    {
      "epoch": 4.80994388889707,
      "grad_norm": 1.1172491312026978,
      "learning_rate": 8.531388887070699e-07,
      "loss": 0.0506,
      "step": 326600
    },
    {
      "epoch": 4.810312071987158,
      "grad_norm": 1.3487467765808105,
      "learning_rate": 8.515025167400853e-07,
      "loss": 0.054,
      "step": 326625
    },
    {
      "epoch": 4.810680255077245,
      "grad_norm": 1.6122474670410156,
      "learning_rate": 8.498661447731007e-07,
      "loss": 0.0452,
      "step": 326650
    },
    {
      "epoch": 4.811048438167332,
      "grad_norm": 1.456758975982666,
      "learning_rate": 8.482297728061162e-07,
      "loss": 0.0491,
      "step": 326675
    },
    {
      "epoch": 4.811416621257419,
      "grad_norm": 1.0345221757888794,
      "learning_rate": 8.465934008391316e-07,
      "loss": 0.051,
      "step": 326700
    },
    {
      "epoch": 4.811784804347506,
      "grad_norm": 1.674077033996582,
      "learning_rate": 8.449570288721471e-07,
      "loss": 0.0493,
      "step": 326725
    },
    {
      "epoch": 4.812152987437593,
      "grad_norm": 1.4658232927322388,
      "learning_rate": 8.433206569051625e-07,
      "loss": 0.0517,
      "step": 326750
    },
    {
      "epoch": 4.81252117052768,
      "grad_norm": 1.1406266689300537,
      "learning_rate": 8.416842849381779e-07,
      "loss": 0.0439,
      "step": 326775
    },
    {
      "epoch": 4.812889353617767,
      "grad_norm": 1.344596266746521,
      "learning_rate": 8.400479129711933e-07,
      "loss": 0.0466,
      "step": 326800
    },
    {
      "epoch": 4.813257536707854,
      "grad_norm": 0.862582802772522,
      "learning_rate": 8.384115410042088e-07,
      "loss": 0.0495,
      "step": 326825
    },
    {
      "epoch": 4.813625719797941,
      "grad_norm": 1.445375680923462,
      "learning_rate": 8.367751690372243e-07,
      "loss": 0.047,
      "step": 326850
    },
    {
      "epoch": 4.813993902888028,
      "grad_norm": 1.269646406173706,
      "learning_rate": 8.351387970702398e-07,
      "loss": 0.0537,
      "step": 326875
    },
    {
      "epoch": 4.814362085978115,
      "grad_norm": 1.3017737865447998,
      "learning_rate": 8.335024251032552e-07,
      "loss": 0.0527,
      "step": 326900
    },
    {
      "epoch": 4.814730269068202,
      "grad_norm": 1.1769801378250122,
      "learning_rate": 8.318660531362706e-07,
      "loss": 0.0553,
      "step": 326925
    },
    {
      "epoch": 4.815098452158289,
      "grad_norm": 0.9682812690734863,
      "learning_rate": 8.30229681169286e-07,
      "loss": 0.0523,
      "step": 326950
    },
    {
      "epoch": 4.815466635248376,
      "grad_norm": 1.2495931386947632,
      "learning_rate": 8.285933092023014e-07,
      "loss": 0.0435,
      "step": 326975
    },
    {
      "epoch": 4.815834818338463,
      "grad_norm": 1.8156076669692993,
      "learning_rate": 8.269569372353169e-07,
      "loss": 0.0533,
      "step": 327000
    },
    {
      "epoch": 4.81620300142855,
      "grad_norm": 1.5517669916152954,
      "learning_rate": 8.253205652683323e-07,
      "loss": 0.0488,
      "step": 327025
    },
    {
      "epoch": 4.816571184518637,
      "grad_norm": 1.2019058465957642,
      "learning_rate": 8.236841933013477e-07,
      "loss": 0.0573,
      "step": 327050
    },
    {
      "epoch": 4.816939367608724,
      "grad_norm": 1.6249291896820068,
      "learning_rate": 8.220478213343633e-07,
      "loss": 0.0574,
      "step": 327075
    },
    {
      "epoch": 4.8173075506988114,
      "grad_norm": 1.2429358959197998,
      "learning_rate": 8.204114493673787e-07,
      "loss": 0.0438,
      "step": 327100
    },
    {
      "epoch": 4.8176757337888985,
      "grad_norm": 0.8272768259048462,
      "learning_rate": 8.187750774003942e-07,
      "loss": 0.05,
      "step": 327125
    },
    {
      "epoch": 4.8180439168789855,
      "grad_norm": 1.6176915168762207,
      "learning_rate": 8.171387054334096e-07,
      "loss": 0.0573,
      "step": 327150
    },
    {
      "epoch": 4.8184120999690725,
      "grad_norm": 1.1579937934875488,
      "learning_rate": 8.15502333466425e-07,
      "loss": 0.0533,
      "step": 327175
    },
    {
      "epoch": 4.8187802830591595,
      "grad_norm": 1.4144437313079834,
      "learning_rate": 8.138659614994404e-07,
      "loss": 0.0474,
      "step": 327200
    },
    {
      "epoch": 4.8191484661492465,
      "grad_norm": 1.2462953329086304,
      "learning_rate": 8.122950444111352e-07,
      "loss": 0.0532,
      "step": 327225
    },
    {
      "epoch": 4.8195166492393335,
      "grad_norm": 1.2043429613113403,
      "learning_rate": 8.106586724441507e-07,
      "loss": 0.0439,
      "step": 327250
    },
    {
      "epoch": 4.8198848323294206,
      "grad_norm": 1.1414886713027954,
      "learning_rate": 8.090223004771661e-07,
      "loss": 0.0516,
      "step": 327275
    },
    {
      "epoch": 4.820253015419508,
      "grad_norm": 1.1177997589111328,
      "learning_rate": 8.073859285101816e-07,
      "loss": 0.0515,
      "step": 327300
    },
    {
      "epoch": 4.820621198509595,
      "grad_norm": 1.1892731189727783,
      "learning_rate": 8.057495565431971e-07,
      "loss": 0.0459,
      "step": 327325
    },
    {
      "epoch": 4.820989381599682,
      "grad_norm": 1.4891889095306396,
      "learning_rate": 8.041131845762125e-07,
      "loss": 0.0532,
      "step": 327350
    },
    {
      "epoch": 4.821357564689769,
      "grad_norm": 1.2619518041610718,
      "learning_rate": 8.024768126092279e-07,
      "loss": 0.0503,
      "step": 327375
    },
    {
      "epoch": 4.821725747779856,
      "grad_norm": 0.9511722326278687,
      "learning_rate": 8.008404406422433e-07,
      "loss": 0.0661,
      "step": 327400
    },
    {
      "epoch": 4.822093930869943,
      "grad_norm": 1.9508581161499023,
      "learning_rate": 7.992040686752587e-07,
      "loss": 0.0427,
      "step": 327425
    },
    {
      "epoch": 4.82246211396003,
      "grad_norm": 1.086293339729309,
      "learning_rate": 7.975676967082741e-07,
      "loss": 0.0541,
      "step": 327450
    },
    {
      "epoch": 4.822830297050118,
      "grad_norm": 0.9589499831199646,
      "learning_rate": 7.959313247412897e-07,
      "loss": 0.0482,
      "step": 327475
    },
    {
      "epoch": 4.823198480140205,
      "grad_norm": 1.375903844833374,
      "learning_rate": 7.942949527743051e-07,
      "loss": 0.0583,
      "step": 327500
    },
    {
      "epoch": 4.823566663230292,
      "grad_norm": 2.153398036956787,
      "learning_rate": 7.926585808073205e-07,
      "loss": 0.0548,
      "step": 327525
    },
    {
      "epoch": 4.823934846320379,
      "grad_norm": 1.0173044204711914,
      "learning_rate": 7.91022208840336e-07,
      "loss": 0.0463,
      "step": 327550
    },
    {
      "epoch": 4.824303029410466,
      "grad_norm": 1.4276267290115356,
      "learning_rate": 7.893858368733514e-07,
      "loss": 0.0548,
      "step": 327575
    },
    {
      "epoch": 4.824671212500553,
      "grad_norm": 0.8254579901695251,
      "learning_rate": 7.877494649063669e-07,
      "loss": 0.0555,
      "step": 327600
    },
    {
      "epoch": 4.82503939559064,
      "grad_norm": 1.2390910387039185,
      "learning_rate": 7.861130929393823e-07,
      "loss": 0.0464,
      "step": 327625
    },
    {
      "epoch": 4.825407578680727,
      "grad_norm": 1.0116550922393799,
      "learning_rate": 7.844767209723977e-07,
      "loss": 0.0515,
      "step": 327650
    },
    {
      "epoch": 4.825775761770814,
      "grad_norm": 1.6932095289230347,
      "learning_rate": 7.828403490054132e-07,
      "loss": 0.0552,
      "step": 327675
    },
    {
      "epoch": 4.826143944860901,
      "grad_norm": 1.4505504369735718,
      "learning_rate": 7.812039770384286e-07,
      "loss": 0.0512,
      "step": 327700
    },
    {
      "epoch": 4.826512127950988,
      "grad_norm": 0.9411727786064148,
      "learning_rate": 7.79567605071444e-07,
      "loss": 0.0448,
      "step": 327725
    },
    {
      "epoch": 4.826880311041075,
      "grad_norm": 1.2102724313735962,
      "learning_rate": 7.779312331044595e-07,
      "loss": 0.0542,
      "step": 327750
    },
    {
      "epoch": 4.827248494131162,
      "grad_norm": 1.5939611196517944,
      "learning_rate": 7.76294861137475e-07,
      "loss": 0.0508,
      "step": 327775
    },
    {
      "epoch": 4.827616677221249,
      "grad_norm": 1.3736553192138672,
      "learning_rate": 7.746584891704904e-07,
      "loss": 0.0487,
      "step": 327800
    },
    {
      "epoch": 4.827984860311336,
      "grad_norm": 1.20279860496521,
      "learning_rate": 7.730221172035058e-07,
      "loss": 0.0488,
      "step": 327825
    },
    {
      "epoch": 4.828353043401423,
      "grad_norm": 1.5555202960968018,
      "learning_rate": 7.713857452365212e-07,
      "loss": 0.0557,
      "step": 327850
    },
    {
      "epoch": 4.82872122649151,
      "grad_norm": 0.9337530136108398,
      "learning_rate": 7.697493732695368e-07,
      "loss": 0.0516,
      "step": 327875
    },
    {
      "epoch": 4.829089409581597,
      "grad_norm": 1.2430012226104736,
      "learning_rate": 7.681130013025522e-07,
      "loss": 0.0527,
      "step": 327900
    },
    {
      "epoch": 4.829457592671684,
      "grad_norm": 1.108012318611145,
      "learning_rate": 7.664766293355676e-07,
      "loss": 0.0474,
      "step": 327925
    },
    {
      "epoch": 4.829825775761771,
      "grad_norm": 1.0347472429275513,
      "learning_rate": 7.64840257368583e-07,
      "loss": 0.0428,
      "step": 327950
    },
    {
      "epoch": 4.830193958851858,
      "grad_norm": 1.0532573461532593,
      "learning_rate": 7.632038854015984e-07,
      "loss": 0.0494,
      "step": 327975
    },
    {
      "epoch": 4.830562141941945,
      "grad_norm": 1.7293601036071777,
      "learning_rate": 7.615675134346138e-07,
      "loss": 0.0533,
      "step": 328000
    },
    {
      "epoch": 4.830930325032032,
      "grad_norm": 0.7408385872840881,
      "learning_rate": 7.599311414676294e-07,
      "loss": 0.0456,
      "step": 328025
    },
    {
      "epoch": 4.831298508122119,
      "grad_norm": 1.4440840482711792,
      "learning_rate": 7.582947695006448e-07,
      "loss": 0.0587,
      "step": 328050
    },
    {
      "epoch": 4.831666691212206,
      "grad_norm": 1.2703137397766113,
      "learning_rate": 7.566583975336603e-07,
      "loss": 0.0549,
      "step": 328075
    },
    {
      "epoch": 4.832034874302293,
      "grad_norm": 2.1623666286468506,
      "learning_rate": 7.550220255666757e-07,
      "loss": 0.0516,
      "step": 328100
    },
    {
      "epoch": 4.83240305739238,
      "grad_norm": 0.751977801322937,
      "learning_rate": 7.533856535996911e-07,
      "loss": 0.0508,
      "step": 328125
    },
    {
      "epoch": 4.832771240482467,
      "grad_norm": 1.1010229587554932,
      "learning_rate": 7.517492816327066e-07,
      "loss": 0.0498,
      "step": 328150
    },
    {
      "epoch": 4.833139423572554,
      "grad_norm": 1.6952306032180786,
      "learning_rate": 7.50112909665722e-07,
      "loss": 0.0524,
      "step": 328175
    },
    {
      "epoch": 4.833507606662641,
      "grad_norm": 1.6126854419708252,
      "learning_rate": 7.484765376987374e-07,
      "loss": 0.0502,
      "step": 328200
    },
    {
      "epoch": 4.833875789752728,
      "grad_norm": 0.7832873463630676,
      "learning_rate": 7.468401657317528e-07,
      "loss": 0.0499,
      "step": 328225
    },
    {
      "epoch": 4.834243972842815,
      "grad_norm": 1.3976212739944458,
      "learning_rate": 7.452037937647682e-07,
      "loss": 0.0504,
      "step": 328250
    },
    {
      "epoch": 4.834612155932902,
      "grad_norm": 1.4674153327941895,
      "learning_rate": 7.435674217977839e-07,
      "loss": 0.0489,
      "step": 328275
    },
    {
      "epoch": 4.834980339022989,
      "grad_norm": 1.2507990598678589,
      "learning_rate": 7.419310498307993e-07,
      "loss": 0.0483,
      "step": 328300
    },
    {
      "epoch": 4.835348522113076,
      "grad_norm": 1.1491302251815796,
      "learning_rate": 7.402946778638147e-07,
      "loss": 0.0513,
      "step": 328325
    },
    {
      "epoch": 4.835716705203163,
      "grad_norm": 1.1399215459823608,
      "learning_rate": 7.386583058968301e-07,
      "loss": 0.046,
      "step": 328350
    },
    {
      "epoch": 4.83608488829325,
      "grad_norm": 1.4150493144989014,
      "learning_rate": 7.370219339298455e-07,
      "loss": 0.0528,
      "step": 328375
    },
    {
      "epoch": 4.836453071383337,
      "grad_norm": 1.4345759153366089,
      "learning_rate": 7.353855619628609e-07,
      "loss": 0.0508,
      "step": 328400
    },
    {
      "epoch": 4.836821254473424,
      "grad_norm": 1.3032423257827759,
      "learning_rate": 7.337491899958764e-07,
      "loss": 0.0504,
      "step": 328425
    },
    {
      "epoch": 4.837189437563511,
      "grad_norm": 1.1716992855072021,
      "learning_rate": 7.321128180288918e-07,
      "loss": 0.0504,
      "step": 328450
    },
    {
      "epoch": 4.837557620653598,
      "grad_norm": 1.040269136428833,
      "learning_rate": 7.304764460619072e-07,
      "loss": 0.0507,
      "step": 328475
    },
    {
      "epoch": 4.837925803743685,
      "grad_norm": 0.811455488204956,
      "learning_rate": 7.288400740949227e-07,
      "loss": 0.0559,
      "step": 328500
    },
    {
      "epoch": 4.838293986833772,
      "grad_norm": 1.0569169521331787,
      "learning_rate": 7.272037021279383e-07,
      "loss": 0.0473,
      "step": 328525
    },
    {
      "epoch": 4.83866216992386,
      "grad_norm": 1.2798147201538086,
      "learning_rate": 7.255673301609537e-07,
      "loss": 0.0507,
      "step": 328550
    },
    {
      "epoch": 4.839030353013947,
      "grad_norm": 1.2332454919815063,
      "learning_rate": 7.239309581939691e-07,
      "loss": 0.0455,
      "step": 328575
    },
    {
      "epoch": 4.839398536104034,
      "grad_norm": 1.310945987701416,
      "learning_rate": 7.222945862269845e-07,
      "loss": 0.0508,
      "step": 328600
    },
    {
      "epoch": 4.839766719194121,
      "grad_norm": 1.7524762153625488,
      "learning_rate": 7.206582142599999e-07,
      "loss": 0.0473,
      "step": 328625
    },
    {
      "epoch": 4.840134902284208,
      "grad_norm": 1.3432730436325073,
      "learning_rate": 7.190218422930153e-07,
      "loss": 0.0432,
      "step": 328650
    },
    {
      "epoch": 4.840503085374295,
      "grad_norm": 0.9963130354881287,
      "learning_rate": 7.173854703260307e-07,
      "loss": 0.054,
      "step": 328675
    },
    {
      "epoch": 4.840871268464382,
      "grad_norm": 1.740356206893921,
      "learning_rate": 7.157490983590463e-07,
      "loss": 0.0572,
      "step": 328700
    },
    {
      "epoch": 4.841239451554469,
      "grad_norm": 1.0979925394058228,
      "learning_rate": 7.141127263920617e-07,
      "loss": 0.0468,
      "step": 328725
    },
    {
      "epoch": 4.841607634644556,
      "grad_norm": 1.3253620862960815,
      "learning_rate": 7.124763544250771e-07,
      "loss": 0.0452,
      "step": 328750
    },
    {
      "epoch": 4.841975817734643,
      "grad_norm": 1.3796427249908447,
      "learning_rate": 7.108399824580926e-07,
      "loss": 0.0496,
      "step": 328775
    },
    {
      "epoch": 4.84234400082473,
      "grad_norm": 1.4862746000289917,
      "learning_rate": 7.092036104911081e-07,
      "loss": 0.0518,
      "step": 328800
    },
    {
      "epoch": 4.842712183914817,
      "grad_norm": 1.012390375137329,
      "learning_rate": 7.075672385241235e-07,
      "loss": 0.0492,
      "step": 328825
    },
    {
      "epoch": 4.843080367004904,
      "grad_norm": 1.2318142652511597,
      "learning_rate": 7.059308665571389e-07,
      "loss": 0.0548,
      "step": 328850
    },
    {
      "epoch": 4.843448550094991,
      "grad_norm": 1.677403211593628,
      "learning_rate": 7.042944945901543e-07,
      "loss": 0.0526,
      "step": 328875
    },
    {
      "epoch": 4.843816733185078,
      "grad_norm": 1.266858696937561,
      "learning_rate": 7.026581226231697e-07,
      "loss": 0.0523,
      "step": 328900
    },
    {
      "epoch": 4.844184916275165,
      "grad_norm": 1.00752592086792,
      "learning_rate": 7.010217506561852e-07,
      "loss": 0.0519,
      "step": 328925
    },
    {
      "epoch": 4.844553099365252,
      "grad_norm": 1.9096211194992065,
      "learning_rate": 6.993853786892006e-07,
      "loss": 0.0502,
      "step": 328950
    },
    {
      "epoch": 4.844921282455339,
      "grad_norm": 1.1299902200698853,
      "learning_rate": 6.977490067222161e-07,
      "loss": 0.0474,
      "step": 328975
    },
    {
      "epoch": 4.845289465545426,
      "grad_norm": 1.2764970064163208,
      "learning_rate": 6.961126347552315e-07,
      "loss": 0.052,
      "step": 329000
    },
    {
      "epoch": 4.845657648635513,
      "grad_norm": 1.3504260778427124,
      "learning_rate": 6.94476262788247e-07,
      "loss": 0.0594,
      "step": 329025
    },
    {
      "epoch": 4.8460258317256,
      "grad_norm": 1.3586283922195435,
      "learning_rate": 6.928398908212624e-07,
      "loss": 0.0571,
      "step": 329050
    },
    {
      "epoch": 4.846394014815687,
      "grad_norm": 1.3230373859405518,
      "learning_rate": 6.912035188542779e-07,
      "loss": 0.0519,
      "step": 329075
    },
    {
      "epoch": 4.8467621979057744,
      "grad_norm": 1.492194414138794,
      "learning_rate": 6.895671468872933e-07,
      "loss": 0.054,
      "step": 329100
    },
    {
      "epoch": 4.8471303809958615,
      "grad_norm": 1.6579687595367432,
      "learning_rate": 6.879307749203088e-07,
      "loss": 0.0491,
      "step": 329125
    },
    {
      "epoch": 4.8474985640859485,
      "grad_norm": 1.6297032833099365,
      "learning_rate": 6.862944029533242e-07,
      "loss": 0.0437,
      "step": 329150
    },
    {
      "epoch": 4.8478667471760355,
      "grad_norm": 1.1896623373031616,
      "learning_rate": 6.846580309863396e-07,
      "loss": 0.0514,
      "step": 329175
    },
    {
      "epoch": 4.8482349302661225,
      "grad_norm": 1.277224063873291,
      "learning_rate": 6.83021659019355e-07,
      "loss": 0.0488,
      "step": 329200
    },
    {
      "epoch": 4.8486031133562095,
      "grad_norm": 1.1937624216079712,
      "learning_rate": 6.813852870523704e-07,
      "loss": 0.0491,
      "step": 329225
    },
    {
      "epoch": 4.8489712964462965,
      "grad_norm": 1.638359785079956,
      "learning_rate": 6.79748915085386e-07,
      "loss": 0.0532,
      "step": 329250
    },
    {
      "epoch": 4.849339479536384,
      "grad_norm": 1.6064482927322388,
      "learning_rate": 6.781125431184014e-07,
      "loss": 0.0498,
      "step": 329275
    },
    {
      "epoch": 4.849707662626471,
      "grad_norm": 1.092834711074829,
      "learning_rate": 6.764761711514168e-07,
      "loss": 0.0491,
      "step": 329300
    },
    {
      "epoch": 4.850075845716558,
      "grad_norm": 1.2567262649536133,
      "learning_rate": 6.748397991844323e-07,
      "loss": 0.0559,
      "step": 329325
    },
    {
      "epoch": 4.850444028806645,
      "grad_norm": 1.1250033378601074,
      "learning_rate": 6.732034272174478e-07,
      "loss": 0.0533,
      "step": 329350
    },
    {
      "epoch": 4.8508122118967325,
      "grad_norm": 0.8184024691581726,
      "learning_rate": 6.715670552504632e-07,
      "loss": 0.0582,
      "step": 329375
    },
    {
      "epoch": 4.8511803949868195,
      "grad_norm": 1.5957434177398682,
      "learning_rate": 6.699306832834786e-07,
      "loss": 0.0499,
      "step": 329400
    },
    {
      "epoch": 4.8515485780769065,
      "grad_norm": 1.2781341075897217,
      "learning_rate": 6.68294311316494e-07,
      "loss": 0.0526,
      "step": 329425
    },
    {
      "epoch": 4.8519167611669936,
      "grad_norm": 1.4263216257095337,
      "learning_rate": 6.666579393495094e-07,
      "loss": 0.0479,
      "step": 329450
    },
    {
      "epoch": 4.852284944257081,
      "grad_norm": 1.428635835647583,
      "learning_rate": 6.650215673825248e-07,
      "loss": 0.0503,
      "step": 329475
    },
    {
      "epoch": 4.852653127347168,
      "grad_norm": 1.3879873752593994,
      "learning_rate": 6.633851954155404e-07,
      "loss": 0.0544,
      "step": 329500
    },
    {
      "epoch": 4.853021310437255,
      "grad_norm": 1.4613816738128662,
      "learning_rate": 6.617488234485559e-07,
      "loss": 0.0458,
      "step": 329525
    },
    {
      "epoch": 4.853389493527342,
      "grad_norm": 1.7965375185012817,
      "learning_rate": 6.601124514815713e-07,
      "loss": 0.046,
      "step": 329550
    },
    {
      "epoch": 4.853757676617429,
      "grad_norm": 1.6311373710632324,
      "learning_rate": 6.584760795145867e-07,
      "loss": 0.0519,
      "step": 329575
    },
    {
      "epoch": 4.854125859707516,
      "grad_norm": 1.1862528324127197,
      "learning_rate": 6.568397075476021e-07,
      "loss": 0.0555,
      "step": 329600
    },
    {
      "epoch": 4.854494042797603,
      "grad_norm": 1.306323528289795,
      "learning_rate": 6.552033355806176e-07,
      "loss": 0.0465,
      "step": 329625
    },
    {
      "epoch": 4.85486222588769,
      "grad_norm": 0.8707461357116699,
      "learning_rate": 6.53566963613633e-07,
      "loss": 0.0559,
      "step": 329650
    },
    {
      "epoch": 4.855230408977777,
      "grad_norm": 1.1268742084503174,
      "learning_rate": 6.519305916466484e-07,
      "loss": 0.0529,
      "step": 329675
    },
    {
      "epoch": 4.855598592067864,
      "grad_norm": 1.3436145782470703,
      "learning_rate": 6.502942196796638e-07,
      "loss": 0.0469,
      "step": 329700
    },
    {
      "epoch": 4.855966775157951,
      "grad_norm": 1.2283753156661987,
      "learning_rate": 6.486578477126792e-07,
      "loss": 0.0508,
      "step": 329725
    },
    {
      "epoch": 4.856334958248038,
      "grad_norm": 1.127272367477417,
      "learning_rate": 6.470214757456949e-07,
      "loss": 0.0514,
      "step": 329750
    },
    {
      "epoch": 4.856703141338125,
      "grad_norm": 1.8110359907150269,
      "learning_rate": 6.453851037787103e-07,
      "loss": 0.0562,
      "step": 329775
    },
    {
      "epoch": 4.857071324428212,
      "grad_norm": 0.9426853656768799,
      "learning_rate": 6.437487318117257e-07,
      "loss": 0.0506,
      "step": 329800
    },
    {
      "epoch": 4.857439507518299,
      "grad_norm": 1.3399136066436768,
      "learning_rate": 6.421123598447411e-07,
      "loss": 0.0503,
      "step": 329825
    },
    {
      "epoch": 4.857807690608386,
      "grad_norm": 1.0861376523971558,
      "learning_rate": 6.404759878777565e-07,
      "loss": 0.0502,
      "step": 329850
    },
    {
      "epoch": 4.858175873698473,
      "grad_norm": 1.4880216121673584,
      "learning_rate": 6.388396159107719e-07,
      "loss": 0.0504,
      "step": 329875
    },
    {
      "epoch": 4.85854405678856,
      "grad_norm": 1.7946224212646484,
      "learning_rate": 6.372032439437874e-07,
      "loss": 0.0488,
      "step": 329900
    },
    {
      "epoch": 4.858912239878647,
      "grad_norm": 1.44832444190979,
      "learning_rate": 6.355668719768028e-07,
      "loss": 0.0479,
      "step": 329925
    },
    {
      "epoch": 4.859280422968734,
      "grad_norm": 1.042392611503601,
      "learning_rate": 6.339305000098183e-07,
      "loss": 0.0517,
      "step": 329950
    },
    {
      "epoch": 4.859648606058821,
      "grad_norm": 1.482990026473999,
      "learning_rate": 6.322941280428337e-07,
      "loss": 0.0559,
      "step": 329975
    },
    {
      "epoch": 4.860016789148908,
      "grad_norm": 0.9654814004898071,
      "learning_rate": 6.306577560758492e-07,
      "loss": 0.0494,
      "step": 330000
    },
    {
      "epoch": 4.860384972238995,
      "grad_norm": 1.0492042303085327,
      "learning_rate": 6.290213841088647e-07,
      "loss": 0.0503,
      "step": 330025
    },
    {
      "epoch": 4.860753155329082,
      "grad_norm": 1.1498184204101562,
      "learning_rate": 6.273850121418801e-07,
      "loss": 0.0482,
      "step": 330050
    },
    {
      "epoch": 4.861121338419169,
      "grad_norm": 1.389785885810852,
      "learning_rate": 6.257486401748955e-07,
      "loss": 0.0514,
      "step": 330075
    },
    {
      "epoch": 4.861489521509256,
      "grad_norm": 1.2617695331573486,
      "learning_rate": 6.241122682079109e-07,
      "loss": 0.0476,
      "step": 330100
    },
    {
      "epoch": 4.861857704599343,
      "grad_norm": 1.6691839694976807,
      "learning_rate": 6.224758962409263e-07,
      "loss": 0.0544,
      "step": 330125
    },
    {
      "epoch": 4.86222588768943,
      "grad_norm": 1.0076056718826294,
      "learning_rate": 6.208395242739417e-07,
      "loss": 0.0482,
      "step": 330150
    },
    {
      "epoch": 4.862594070779517,
      "grad_norm": 1.3942625522613525,
      "learning_rate": 6.192031523069573e-07,
      "loss": 0.0521,
      "step": 330175
    },
    {
      "epoch": 4.862962253869604,
      "grad_norm": 1.074027419090271,
      "learning_rate": 6.175667803399727e-07,
      "loss": 0.051,
      "step": 330200
    },
    {
      "epoch": 4.863330436959691,
      "grad_norm": 0.8190245628356934,
      "learning_rate": 6.159304083729881e-07,
      "loss": 0.0555,
      "step": 330225
    },
    {
      "epoch": 4.863698620049778,
      "grad_norm": 1.3431546688079834,
      "learning_rate": 6.142940364060036e-07,
      "loss": 0.0477,
      "step": 330250
    },
    {
      "epoch": 4.864066803139865,
      "grad_norm": 1.045159935951233,
      "learning_rate": 6.12657664439019e-07,
      "loss": 0.0425,
      "step": 330275
    },
    {
      "epoch": 4.864434986229952,
      "grad_norm": 1.422806978225708,
      "learning_rate": 6.110212924720345e-07,
      "loss": 0.0493,
      "step": 330300
    },
    {
      "epoch": 4.864803169320039,
      "grad_norm": 1.7697246074676514,
      "learning_rate": 6.093849205050499e-07,
      "loss": 0.0448,
      "step": 330325
    },
    {
      "epoch": 4.865171352410126,
      "grad_norm": 1.3944896459579468,
      "learning_rate": 6.077485485380653e-07,
      "loss": 0.0596,
      "step": 330350
    },
    {
      "epoch": 4.865539535500213,
      "grad_norm": 1.082607626914978,
      "learning_rate": 6.061121765710808e-07,
      "loss": 0.0443,
      "step": 330375
    },
    {
      "epoch": 4.8659077185903,
      "grad_norm": 0.8295331597328186,
      "learning_rate": 6.044758046040962e-07,
      "loss": 0.0456,
      "step": 330400
    },
    {
      "epoch": 4.866275901680387,
      "grad_norm": 1.444100022315979,
      "learning_rate": 6.028394326371117e-07,
      "loss": 0.049,
      "step": 330425
    },
    {
      "epoch": 4.866644084770474,
      "grad_norm": 1.6951584815979004,
      "learning_rate": 6.012030606701271e-07,
      "loss": 0.0499,
      "step": 330450
    },
    {
      "epoch": 4.867012267860562,
      "grad_norm": 1.1521512269973755,
      "learning_rate": 5.995666887031425e-07,
      "loss": 0.0469,
      "step": 330475
    },
    {
      "epoch": 4.867380450950649,
      "grad_norm": 1.6782010793685913,
      "learning_rate": 5.979303167361579e-07,
      "loss": 0.0582,
      "step": 330500
    },
    {
      "epoch": 4.867748634040736,
      "grad_norm": 0.9512406587600708,
      "learning_rate": 5.962939447691734e-07,
      "loss": 0.0516,
      "step": 330525
    },
    {
      "epoch": 4.868116817130823,
      "grad_norm": 1.4517745971679688,
      "learning_rate": 5.946575728021888e-07,
      "loss": 0.0555,
      "step": 330550
    },
    {
      "epoch": 4.86848500022091,
      "grad_norm": 1.5887190103530884,
      "learning_rate": 5.930212008352044e-07,
      "loss": 0.0475,
      "step": 330575
    },
    {
      "epoch": 4.868853183310997,
      "grad_norm": 1.53322172164917,
      "learning_rate": 5.913848288682198e-07,
      "loss": 0.0469,
      "step": 330600
    },
    {
      "epoch": 4.869221366401084,
      "grad_norm": 0.5552016496658325,
      "learning_rate": 5.897484569012352e-07,
      "loss": 0.0497,
      "step": 330625
    },
    {
      "epoch": 4.869589549491171,
      "grad_norm": 1.5306885242462158,
      "learning_rate": 5.881120849342506e-07,
      "loss": 0.0468,
      "step": 330650
    },
    {
      "epoch": 4.869957732581258,
      "grad_norm": 1.3030372858047485,
      "learning_rate": 5.864757129672661e-07,
      "loss": 0.0509,
      "step": 330675
    },
    {
      "epoch": 4.870325915671345,
      "grad_norm": 1.384799838066101,
      "learning_rate": 5.848393410002815e-07,
      "loss": 0.0539,
      "step": 330700
    },
    {
      "epoch": 4.870694098761432,
      "grad_norm": 1.330909013748169,
      "learning_rate": 5.832029690332969e-07,
      "loss": 0.0527,
      "step": 330725
    },
    {
      "epoch": 4.871062281851519,
      "grad_norm": 1.3578200340270996,
      "learning_rate": 5.815665970663123e-07,
      "loss": 0.0563,
      "step": 330750
    },
    {
      "epoch": 4.871430464941606,
      "grad_norm": 1.4362881183624268,
      "learning_rate": 5.799956799780073e-07,
      "loss": 0.0525,
      "step": 330775
    },
    {
      "epoch": 4.871798648031693,
      "grad_norm": 1.0092297792434692,
      "learning_rate": 5.783593080110227e-07,
      "loss": 0.0527,
      "step": 330800
    },
    {
      "epoch": 4.87216683112178,
      "grad_norm": 1.3779137134552002,
      "learning_rate": 5.767229360440381e-07,
      "loss": 0.0487,
      "step": 330825
    },
    {
      "epoch": 4.872535014211867,
      "grad_norm": 0.7738792896270752,
      "learning_rate": 5.750865640770535e-07,
      "loss": 0.0567,
      "step": 330850
    },
    {
      "epoch": 4.872903197301954,
      "grad_norm": 0.8938313722610474,
      "learning_rate": 5.734501921100689e-07,
      "loss": 0.052,
      "step": 330875
    },
    {
      "epoch": 4.873271380392041,
      "grad_norm": 1.2754948139190674,
      "learning_rate": 5.718138201430844e-07,
      "loss": 0.0531,
      "step": 330900
    },
    {
      "epoch": 4.873639563482128,
      "grad_norm": 1.1972107887268066,
      "learning_rate": 5.701774481760998e-07,
      "loss": 0.0492,
      "step": 330925
    },
    {
      "epoch": 4.874007746572215,
      "grad_norm": 1.327114462852478,
      "learning_rate": 5.685410762091152e-07,
      "loss": 0.0523,
      "step": 330950
    },
    {
      "epoch": 4.874375929662302,
      "grad_norm": 1.462823510169983,
      "learning_rate": 5.669047042421308e-07,
      "loss": 0.0476,
      "step": 330975
    },
    {
      "epoch": 4.874744112752389,
      "grad_norm": 0.8686959743499756,
      "learning_rate": 5.652683322751462e-07,
      "loss": 0.0515,
      "step": 331000
    },
    {
      "epoch": 4.875112295842476,
      "grad_norm": 1.6701581478118896,
      "learning_rate": 5.636319603081617e-07,
      "loss": 0.0469,
      "step": 331025
    },
    {
      "epoch": 4.875480478932563,
      "grad_norm": 1.3872337341308594,
      "learning_rate": 5.619955883411771e-07,
      "loss": 0.0519,
      "step": 331050
    },
    {
      "epoch": 4.87584866202265,
      "grad_norm": 1.8140034675598145,
      "learning_rate": 5.603592163741925e-07,
      "loss": 0.0559,
      "step": 331075
    },
    {
      "epoch": 4.8762168451127375,
      "grad_norm": 1.8342561721801758,
      "learning_rate": 5.587228444072079e-07,
      "loss": 0.0514,
      "step": 331100
    },
    {
      "epoch": 4.8765850282028245,
      "grad_norm": 0.9031476378440857,
      "learning_rate": 5.570864724402233e-07,
      "loss": 0.0516,
      "step": 331125
    },
    {
      "epoch": 4.8769532112929115,
      "grad_norm": 1.5302590131759644,
      "learning_rate": 5.554501004732388e-07,
      "loss": 0.054,
      "step": 331150
    },
    {
      "epoch": 4.8773213943829985,
      "grad_norm": 1.2084879875183105,
      "learning_rate": 5.538137285062543e-07,
      "loss": 0.0526,
      "step": 331175
    },
    {
      "epoch": 4.8776895774730855,
      "grad_norm": 1.3794918060302734,
      "learning_rate": 5.521773565392697e-07,
      "loss": 0.0485,
      "step": 331200
    },
    {
      "epoch": 4.8780577605631725,
      "grad_norm": 0.9214732050895691,
      "learning_rate": 5.505409845722851e-07,
      "loss": 0.0562,
      "step": 331225
    },
    {
      "epoch": 4.8784259436532595,
      "grad_norm": 1.7100882530212402,
      "learning_rate": 5.489046126053006e-07,
      "loss": 0.0549,
      "step": 331250
    },
    {
      "epoch": 4.878794126743347,
      "grad_norm": 1.8604553937911987,
      "learning_rate": 5.47268240638316e-07,
      "loss": 0.0506,
      "step": 331275
    },
    {
      "epoch": 4.8791623098334345,
      "grad_norm": 1.4793431758880615,
      "learning_rate": 5.456318686713315e-07,
      "loss": 0.048,
      "step": 331300
    },
    {
      "epoch": 4.8795304929235215,
      "grad_norm": 0.7446379661560059,
      "learning_rate": 5.439954967043469e-07,
      "loss": 0.0492,
      "step": 331325
    },
    {
      "epoch": 4.8798986760136085,
      "grad_norm": 1.1350880861282349,
      "learning_rate": 5.423591247373623e-07,
      "loss": 0.0476,
      "step": 331350
    },
    {
      "epoch": 4.8802668591036955,
      "grad_norm": 1.6795074939727783,
      "learning_rate": 5.407227527703778e-07,
      "loss": 0.0544,
      "step": 331375
    },
    {
      "epoch": 4.8806350421937825,
      "grad_norm": 1.2859065532684326,
      "learning_rate": 5.390863808033933e-07,
      "loss": 0.0479,
      "step": 331400
    },
    {
      "epoch": 4.8810032252838695,
      "grad_norm": 0.7876410484313965,
      "learning_rate": 5.374500088364087e-07,
      "loss": 0.0475,
      "step": 331425
    },
    {
      "epoch": 4.8813714083739566,
      "grad_norm": 0.9864166975021362,
      "learning_rate": 5.358136368694241e-07,
      "loss": 0.0469,
      "step": 331450
    },
    {
      "epoch": 4.881739591464044,
      "grad_norm": 1.576211929321289,
      "learning_rate": 5.341772649024395e-07,
      "loss": 0.0533,
      "step": 331475
    },
    {
      "epoch": 4.882107774554131,
      "grad_norm": 1.6332675218582153,
      "learning_rate": 5.325408929354549e-07,
      "loss": 0.0537,
      "step": 331500
    },
    {
      "epoch": 4.882475957644218,
      "grad_norm": 1.6791424751281738,
      "learning_rate": 5.309045209684704e-07,
      "loss": 0.0492,
      "step": 331525
    },
    {
      "epoch": 4.882844140734305,
      "grad_norm": 1.1094706058502197,
      "learning_rate": 5.292681490014858e-07,
      "loss": 0.0513,
      "step": 331550
    },
    {
      "epoch": 4.883212323824392,
      "grad_norm": 1.319616675376892,
      "learning_rate": 5.276317770345014e-07,
      "loss": 0.0511,
      "step": 331575
    },
    {
      "epoch": 4.883580506914479,
      "grad_norm": 1.0090830326080322,
      "learning_rate": 5.259954050675168e-07,
      "loss": 0.0434,
      "step": 331600
    },
    {
      "epoch": 4.883948690004566,
      "grad_norm": 1.749432921409607,
      "learning_rate": 5.243590331005322e-07,
      "loss": 0.0526,
      "step": 331625
    },
    {
      "epoch": 4.884316873094653,
      "grad_norm": 1.5274707078933716,
      "learning_rate": 5.227226611335477e-07,
      "loss": 0.0473,
      "step": 331650
    },
    {
      "epoch": 4.88468505618474,
      "grad_norm": 1.4532458782196045,
      "learning_rate": 5.210862891665631e-07,
      "loss": 0.0444,
      "step": 331675
    },
    {
      "epoch": 4.885053239274827,
      "grad_norm": 1.5814168453216553,
      "learning_rate": 5.194499171995785e-07,
      "loss": 0.0503,
      "step": 331700
    },
    {
      "epoch": 4.885421422364914,
      "grad_norm": 1.417539358139038,
      "learning_rate": 5.178135452325939e-07,
      "loss": 0.0499,
      "step": 331725
    },
    {
      "epoch": 4.885789605455001,
      "grad_norm": 1.2722365856170654,
      "learning_rate": 5.161771732656094e-07,
      "loss": 0.0484,
      "step": 331750
    },
    {
      "epoch": 4.886157788545088,
      "grad_norm": 1.0175721645355225,
      "learning_rate": 5.145408012986248e-07,
      "loss": 0.0517,
      "step": 331775
    },
    {
      "epoch": 4.886525971635175,
      "grad_norm": 1.4009501934051514,
      "learning_rate": 5.129044293316403e-07,
      "loss": 0.0453,
      "step": 331800
    },
    {
      "epoch": 4.886894154725262,
      "grad_norm": 1.321657657623291,
      "learning_rate": 5.112680573646557e-07,
      "loss": 0.0486,
      "step": 331825
    },
    {
      "epoch": 4.887262337815349,
      "grad_norm": 1.2296380996704102,
      "learning_rate": 5.096316853976712e-07,
      "loss": 0.0605,
      "step": 331850
    },
    {
      "epoch": 4.887630520905436,
      "grad_norm": 1.1335623264312744,
      "learning_rate": 5.079953134306866e-07,
      "loss": 0.0493,
      "step": 331875
    },
    {
      "epoch": 4.887998703995523,
      "grad_norm": 1.815750241279602,
      "learning_rate": 5.06358941463702e-07,
      "loss": 0.0556,
      "step": 331900
    },
    {
      "epoch": 4.88836688708561,
      "grad_norm": 1.2230286598205566,
      "learning_rate": 5.047225694967175e-07,
      "loss": 0.0538,
      "step": 331925
    },
    {
      "epoch": 4.888735070175697,
      "grad_norm": 1.824168086051941,
      "learning_rate": 5.030861975297329e-07,
      "loss": 0.052,
      "step": 331950
    },
    {
      "epoch": 4.889103253265784,
      "grad_norm": 1.2683234214782715,
      "learning_rate": 5.014498255627483e-07,
      "loss": 0.0516,
      "step": 331975
    },
    {
      "epoch": 4.889471436355871,
      "grad_norm": 1.3644198179244995,
      "learning_rate": 4.998134535957639e-07,
      "loss": 0.0548,
      "step": 332000
    },
    {
      "epoch": 4.889839619445958,
      "grad_norm": 1.120449423789978,
      "learning_rate": 4.981770816287793e-07,
      "loss": 0.0507,
      "step": 332025
    },
    {
      "epoch": 4.890207802536045,
      "grad_norm": 0.8146657347679138,
      "learning_rate": 4.965407096617947e-07,
      "loss": 0.0475,
      "step": 332050
    },
    {
      "epoch": 4.890575985626132,
      "grad_norm": 1.4584808349609375,
      "learning_rate": 4.949043376948101e-07,
      "loss": 0.0547,
      "step": 332075
    },
    {
      "epoch": 4.890944168716219,
      "grad_norm": 1.1213934421539307,
      "learning_rate": 4.932679657278255e-07,
      "loss": 0.0563,
      "step": 332100
    },
    {
      "epoch": 4.891312351806306,
      "grad_norm": 1.1377229690551758,
      "learning_rate": 4.91631593760841e-07,
      "loss": 0.0496,
      "step": 332125
    },
    {
      "epoch": 4.891680534896393,
      "grad_norm": 1.2989132404327393,
      "learning_rate": 4.899952217938564e-07,
      "loss": 0.0504,
      "step": 332150
    },
    {
      "epoch": 4.89204871798648,
      "grad_norm": 1.4607120752334595,
      "learning_rate": 4.883588498268718e-07,
      "loss": 0.0501,
      "step": 332175
    },
    {
      "epoch": 4.892416901076567,
      "grad_norm": 1.5263921022415161,
      "learning_rate": 4.867224778598874e-07,
      "loss": 0.0556,
      "step": 332200
    },
    {
      "epoch": 4.892785084166654,
      "grad_norm": 0.8324810266494751,
      "learning_rate": 4.850861058929028e-07,
      "loss": 0.0534,
      "step": 332225
    },
    {
      "epoch": 4.893153267256741,
      "grad_norm": 1.6191877126693726,
      "learning_rate": 4.834497339259183e-07,
      "loss": 0.0511,
      "step": 332250
    },
    {
      "epoch": 4.893521450346828,
      "grad_norm": 1.1996718645095825,
      "learning_rate": 4.818133619589337e-07,
      "loss": 0.0494,
      "step": 332275
    },
    {
      "epoch": 4.893889633436915,
      "grad_norm": 1.200501799583435,
      "learning_rate": 4.801769899919491e-07,
      "loss": 0.0503,
      "step": 332300
    },
    {
      "epoch": 4.894257816527002,
      "grad_norm": 1.8005608320236206,
      "learning_rate": 4.785406180249645e-07,
      "loss": 0.0499,
      "step": 332325
    },
    {
      "epoch": 4.894625999617089,
      "grad_norm": 1.4183095693588257,
      "learning_rate": 4.769042460579799e-07,
      "loss": 0.0535,
      "step": 332350
    },
    {
      "epoch": 4.894994182707177,
      "grad_norm": 1.3116949796676636,
      "learning_rate": 4.7526787409099544e-07,
      "loss": 0.0541,
      "step": 332375
    },
    {
      "epoch": 4.895362365797264,
      "grad_norm": 1.670743465423584,
      "learning_rate": 4.7363150212401085e-07,
      "loss": 0.0503,
      "step": 332400
    },
    {
      "epoch": 4.895730548887351,
      "grad_norm": 1.104217290878296,
      "learning_rate": 4.719951301570263e-07,
      "loss": 0.0507,
      "step": 332425
    },
    {
      "epoch": 4.896098731977438,
      "grad_norm": 1.4548280239105225,
      "learning_rate": 4.703587581900417e-07,
      "loss": 0.0482,
      "step": 332450
    },
    {
      "epoch": 4.896466915067525,
      "grad_norm": 1.2266905307769775,
      "learning_rate": 4.6872238622305713e-07,
      "loss": 0.0492,
      "step": 332475
    },
    {
      "epoch": 4.896835098157612,
      "grad_norm": 1.3408427238464355,
      "learning_rate": 4.6708601425607265e-07,
      "loss": 0.0487,
      "step": 332500
    },
    {
      "epoch": 4.897203281247699,
      "grad_norm": 1.434448003768921,
      "learning_rate": 4.6544964228908806e-07,
      "loss": 0.0499,
      "step": 332525
    },
    {
      "epoch": 4.897571464337786,
      "grad_norm": 1.2371189594268799,
      "learning_rate": 4.6381327032210347e-07,
      "loss": 0.0558,
      "step": 332550
    },
    {
      "epoch": 4.897939647427873,
      "grad_norm": 1.33144211769104,
      "learning_rate": 4.6217689835511893e-07,
      "loss": 0.0532,
      "step": 332575
    },
    {
      "epoch": 4.89830783051796,
      "grad_norm": 1.6626392602920532,
      "learning_rate": 4.6054052638813434e-07,
      "loss": 0.0503,
      "step": 332600
    },
    {
      "epoch": 4.898676013608047,
      "grad_norm": 1.1345593929290771,
      "learning_rate": 4.5890415442114986e-07,
      "loss": 0.0535,
      "step": 332625
    },
    {
      "epoch": 4.899044196698134,
      "grad_norm": 1.310836911201477,
      "learning_rate": 4.5726778245416527e-07,
      "loss": 0.0484,
      "step": 332650
    },
    {
      "epoch": 4.899412379788221,
      "grad_norm": 1.0395610332489014,
      "learning_rate": 4.556314104871807e-07,
      "loss": 0.0558,
      "step": 332675
    },
    {
      "epoch": 4.899780562878308,
      "grad_norm": 0.6484836935997009,
      "learning_rate": 4.5399503852019614e-07,
      "loss": 0.0498,
      "step": 332700
    },
    {
      "epoch": 4.900148745968395,
      "grad_norm": 0.7665823101997375,
      "learning_rate": 4.5235866655321155e-07,
      "loss": 0.0463,
      "step": 332725
    },
    {
      "epoch": 4.900516929058482,
      "grad_norm": 1.2902981042861938,
      "learning_rate": 4.5072229458622707e-07,
      "loss": 0.0453,
      "step": 332750
    },
    {
      "epoch": 4.900885112148569,
      "grad_norm": 1.5675803422927856,
      "learning_rate": 4.490859226192425e-07,
      "loss": 0.0534,
      "step": 332775
    },
    {
      "epoch": 4.901253295238656,
      "grad_norm": 0.6525160670280457,
      "learning_rate": 4.474495506522579e-07,
      "loss": 0.0526,
      "step": 332800
    },
    {
      "epoch": 4.901621478328743,
      "grad_norm": 1.0245834589004517,
      "learning_rate": 4.458131786852733e-07,
      "loss": 0.0459,
      "step": 332825
    },
    {
      "epoch": 4.90198966141883,
      "grad_norm": 1.2377426624298096,
      "learning_rate": 4.4417680671828876e-07,
      "loss": 0.0497,
      "step": 332850
    },
    {
      "epoch": 4.902357844508917,
      "grad_norm": 1.6618950366973877,
      "learning_rate": 4.425404347513042e-07,
      "loss": 0.0512,
      "step": 332875
    },
    {
      "epoch": 4.902726027599004,
      "grad_norm": 1.6873887777328491,
      "learning_rate": 4.409040627843197e-07,
      "loss": 0.0563,
      "step": 332900
    },
    {
      "epoch": 4.903094210689091,
      "grad_norm": 0.8748207092285156,
      "learning_rate": 4.392676908173351e-07,
      "loss": 0.0445,
      "step": 332925
    },
    {
      "epoch": 4.903462393779178,
      "grad_norm": 1.0240164995193481,
      "learning_rate": 4.376313188503505e-07,
      "loss": 0.0552,
      "step": 332950
    },
    {
      "epoch": 4.903830576869265,
      "grad_norm": 1.0954469442367554,
      "learning_rate": 4.35994946883366e-07,
      "loss": 0.0533,
      "step": 332975
    },
    {
      "epoch": 4.904198759959352,
      "grad_norm": 1.4549872875213623,
      "learning_rate": 4.3435857491638144e-07,
      "loss": 0.0496,
      "step": 333000
    },
    {
      "epoch": 4.904566943049439,
      "grad_norm": 1.754491925239563,
      "learning_rate": 4.327222029493969e-07,
      "loss": 0.0557,
      "step": 333025
    },
    {
      "epoch": 4.904935126139526,
      "grad_norm": 0.7370721697807312,
      "learning_rate": 4.310858309824123e-07,
      "loss": 0.0524,
      "step": 333050
    },
    {
      "epoch": 4.905303309229613,
      "grad_norm": 1.241559386253357,
      "learning_rate": 4.294494590154277e-07,
      "loss": 0.0485,
      "step": 333075
    },
    {
      "epoch": 4.9056714923197005,
      "grad_norm": 1.6938914060592651,
      "learning_rate": 4.2781308704844324e-07,
      "loss": 0.0483,
      "step": 333100
    },
    {
      "epoch": 4.9060396754097875,
      "grad_norm": 1.006069540977478,
      "learning_rate": 4.2617671508145865e-07,
      "loss": 0.0529,
      "step": 333125
    },
    {
      "epoch": 4.9064078584998745,
      "grad_norm": 1.5768673419952393,
      "learning_rate": 4.2454034311447406e-07,
      "loss": 0.0511,
      "step": 333150
    },
    {
      "epoch": 4.9067760415899615,
      "grad_norm": 2.3626139163970947,
      "learning_rate": 4.229039711474895e-07,
      "loss": 0.0552,
      "step": 333175
    },
    {
      "epoch": 4.9071442246800485,
      "grad_norm": 0.9993570446968079,
      "learning_rate": 4.2126759918050493e-07,
      "loss": 0.0481,
      "step": 333200
    },
    {
      "epoch": 4.907512407770136,
      "grad_norm": 1.7570234537124634,
      "learning_rate": 4.196966820921998e-07,
      "loss": 0.0485,
      "step": 333225
    },
    {
      "epoch": 4.907880590860223,
      "grad_norm": 1.5358786582946777,
      "learning_rate": 4.180603101252152e-07,
      "loss": 0.0515,
      "step": 333250
    },
    {
      "epoch": 4.9082487739503105,
      "grad_norm": 1.0907758474349976,
      "learning_rate": 4.1642393815823064e-07,
      "loss": 0.0482,
      "step": 333275
    },
    {
      "epoch": 4.9086169570403975,
      "grad_norm": 1.3479907512664795,
      "learning_rate": 4.147875661912461e-07,
      "loss": 0.0566,
      "step": 333300
    },
    {
      "epoch": 4.9089851401304845,
      "grad_norm": 1.0612907409667969,
      "learning_rate": 4.1315119422426156e-07,
      "loss": 0.0474,
      "step": 333325
    },
    {
      "epoch": 4.9093533232205715,
      "grad_norm": 1.3949934244155884,
      "learning_rate": 4.11514822257277e-07,
      "loss": 0.0487,
      "step": 333350
    },
    {
      "epoch": 4.9097215063106585,
      "grad_norm": 1.223125696182251,
      "learning_rate": 4.0987845029029244e-07,
      "loss": 0.0491,
      "step": 333375
    },
    {
      "epoch": 4.9100896894007455,
      "grad_norm": 0.973293662071228,
      "learning_rate": 4.0824207832330785e-07,
      "loss": 0.0447,
      "step": 333400
    },
    {
      "epoch": 4.9104578724908325,
      "grad_norm": 1.2000644207000732,
      "learning_rate": 4.0660570635632326e-07,
      "loss": 0.0482,
      "step": 333425
    },
    {
      "epoch": 4.91082605558092,
      "grad_norm": 1.2607672214508057,
      "learning_rate": 4.0496933438933877e-07,
      "loss": 0.0527,
      "step": 333450
    },
    {
      "epoch": 4.911194238671007,
      "grad_norm": 1.2454999685287476,
      "learning_rate": 4.033329624223542e-07,
      "loss": 0.0591,
      "step": 333475
    },
    {
      "epoch": 4.911562421761094,
      "grad_norm": 1.1193015575408936,
      "learning_rate": 4.0169659045536965e-07,
      "loss": 0.0518,
      "step": 333500
    },
    {
      "epoch": 4.911930604851181,
      "grad_norm": 1.552511215209961,
      "learning_rate": 4.0006021848838506e-07,
      "loss": 0.0553,
      "step": 333525
    },
    {
      "epoch": 4.912298787941268,
      "grad_norm": 1.1871947050094604,
      "learning_rate": 3.9842384652140047e-07,
      "loss": 0.049,
      "step": 333550
    },
    {
      "epoch": 4.912666971031355,
      "grad_norm": 1.3959815502166748,
      "learning_rate": 3.96787474554416e-07,
      "loss": 0.0481,
      "step": 333575
    },
    {
      "epoch": 4.913035154121442,
      "grad_norm": 1.3606975078582764,
      "learning_rate": 3.951511025874314e-07,
      "loss": 0.0433,
      "step": 333600
    },
    {
      "epoch": 4.913403337211529,
      "grad_norm": 1.348492980003357,
      "learning_rate": 3.9351473062044686e-07,
      "loss": 0.0496,
      "step": 333625
    },
    {
      "epoch": 4.913771520301616,
      "grad_norm": 1.370913028717041,
      "learning_rate": 3.9187835865346227e-07,
      "loss": 0.0487,
      "step": 333650
    },
    {
      "epoch": 4.914139703391703,
      "grad_norm": 1.3554139137268066,
      "learning_rate": 3.902419866864777e-07,
      "loss": 0.0498,
      "step": 333675
    },
    {
      "epoch": 4.91450788648179,
      "grad_norm": 1.1865737438201904,
      "learning_rate": 3.886056147194932e-07,
      "loss": 0.0476,
      "step": 333700
    },
    {
      "epoch": 4.914876069571877,
      "grad_norm": 1.9947452545166016,
      "learning_rate": 3.869692427525086e-07,
      "loss": 0.0525,
      "step": 333725
    },
    {
      "epoch": 4.915244252661964,
      "grad_norm": 1.2996271848678589,
      "learning_rate": 3.85332870785524e-07,
      "loss": 0.0515,
      "step": 333750
    },
    {
      "epoch": 4.915612435752051,
      "grad_norm": 1.139541506767273,
      "learning_rate": 3.836964988185395e-07,
      "loss": 0.0537,
      "step": 333775
    },
    {
      "epoch": 4.915980618842138,
      "grad_norm": 1.5225929021835327,
      "learning_rate": 3.820601268515549e-07,
      "loss": 0.0495,
      "step": 333800
    },
    {
      "epoch": 4.916348801932225,
      "grad_norm": 1.0650289058685303,
      "learning_rate": 3.804237548845704e-07,
      "loss": 0.0459,
      "step": 333825
    },
    {
      "epoch": 4.916716985022312,
      "grad_norm": 1.4298961162567139,
      "learning_rate": 3.787873829175858e-07,
      "loss": 0.0582,
      "step": 333850
    },
    {
      "epoch": 4.917085168112399,
      "grad_norm": 1.447590947151184,
      "learning_rate": 3.771510109506012e-07,
      "loss": 0.0456,
      "step": 333875
    },
    {
      "epoch": 4.917453351202486,
      "grad_norm": 1.102942705154419,
      "learning_rate": 3.755146389836167e-07,
      "loss": 0.0547,
      "step": 333900
    },
    {
      "epoch": 4.917821534292573,
      "grad_norm": 0.8419262170791626,
      "learning_rate": 3.738782670166321e-07,
      "loss": 0.047,
      "step": 333925
    },
    {
      "epoch": 4.91818971738266,
      "grad_norm": 1.1380261182785034,
      "learning_rate": 3.7224189504964756e-07,
      "loss": 0.0491,
      "step": 333950
    },
    {
      "epoch": 4.918557900472747,
      "grad_norm": 1.1962227821350098,
      "learning_rate": 3.70605523082663e-07,
      "loss": 0.0506,
      "step": 333975
    },
    {
      "epoch": 4.918926083562834,
      "grad_norm": 1.627918004989624,
      "learning_rate": 3.6896915111567843e-07,
      "loss": 0.0496,
      "step": 334000
    },
    {
      "epoch": 4.919294266652921,
      "grad_norm": 1.403448462486267,
      "learning_rate": 3.6733277914869384e-07,
      "loss": 0.0497,
      "step": 334025
    },
    {
      "epoch": 4.919662449743008,
      "grad_norm": 1.3129823207855225,
      "learning_rate": 3.656964071817093e-07,
      "loss": 0.0548,
      "step": 334050
    },
    {
      "epoch": 4.920030632833095,
      "grad_norm": 0.92596435546875,
      "learning_rate": 3.6406003521472477e-07,
      "loss": 0.0456,
      "step": 334075
    },
    {
      "epoch": 4.920398815923182,
      "grad_norm": 1.1982165575027466,
      "learning_rate": 3.6242366324774023e-07,
      "loss": 0.052,
      "step": 334100
    },
    {
      "epoch": 4.920766999013269,
      "grad_norm": 0.7939584255218506,
      "learning_rate": 3.6078729128075564e-07,
      "loss": 0.0536,
      "step": 334125
    },
    {
      "epoch": 4.921135182103356,
      "grad_norm": 0.966560959815979,
      "learning_rate": 3.5915091931377106e-07,
      "loss": 0.0459,
      "step": 334150
    },
    {
      "epoch": 4.921503365193443,
      "grad_norm": 0.7333667278289795,
      "learning_rate": 3.5751454734678647e-07,
      "loss": 0.0515,
      "step": 334175
    },
    {
      "epoch": 4.92187154828353,
      "grad_norm": 1.469001054763794,
      "learning_rate": 3.55878175379802e-07,
      "loss": 0.0534,
      "step": 334200
    },
    {
      "epoch": 4.922239731373617,
      "grad_norm": 1.4328724145889282,
      "learning_rate": 3.542418034128174e-07,
      "loss": 0.0509,
      "step": 334225
    },
    {
      "epoch": 4.922607914463704,
      "grad_norm": 1.0952802896499634,
      "learning_rate": 3.5260543144583286e-07,
      "loss": 0.0514,
      "step": 334250
    },
    {
      "epoch": 4.922976097553791,
      "grad_norm": 1.4380772113800049,
      "learning_rate": 3.5096905947884827e-07,
      "loss": 0.0558,
      "step": 334275
    },
    {
      "epoch": 4.923344280643879,
      "grad_norm": 1.3975204229354858,
      "learning_rate": 3.493326875118637e-07,
      "loss": 0.0456,
      "step": 334300
    },
    {
      "epoch": 4.923712463733966,
      "grad_norm": 1.3751635551452637,
      "learning_rate": 3.476963155448792e-07,
      "loss": 0.0524,
      "step": 334325
    },
    {
      "epoch": 4.924080646824053,
      "grad_norm": 1.231284737586975,
      "learning_rate": 3.460599435778946e-07,
      "loss": 0.0475,
      "step": 334350
    },
    {
      "epoch": 4.92444882991414,
      "grad_norm": 1.2894213199615479,
      "learning_rate": 3.4442357161091007e-07,
      "loss": 0.0491,
      "step": 334375
    },
    {
      "epoch": 4.924817013004227,
      "grad_norm": 1.3960341215133667,
      "learning_rate": 3.427871996439255e-07,
      "loss": 0.0499,
      "step": 334400
    },
    {
      "epoch": 4.925185196094314,
      "grad_norm": 1.9660850763320923,
      "learning_rate": 3.411508276769409e-07,
      "loss": 0.0493,
      "step": 334425
    },
    {
      "epoch": 4.925553379184401,
      "grad_norm": 1.4844788312911987,
      "learning_rate": 3.395144557099564e-07,
      "loss": 0.056,
      "step": 334450
    },
    {
      "epoch": 4.925921562274488,
      "grad_norm": 1.1590521335601807,
      "learning_rate": 3.378780837429718e-07,
      "loss": 0.0529,
      "step": 334475
    },
    {
      "epoch": 4.926289745364575,
      "grad_norm": 1.0546517372131348,
      "learning_rate": 3.362417117759872e-07,
      "loss": 0.049,
      "step": 334500
    },
    {
      "epoch": 4.926657928454662,
      "grad_norm": 1.026505470275879,
      "learning_rate": 3.346053398090027e-07,
      "loss": 0.0501,
      "step": 334525
    },
    {
      "epoch": 4.927026111544749,
      "grad_norm": 1.2053793668746948,
      "learning_rate": 3.329689678420181e-07,
      "loss": 0.0517,
      "step": 334550
    },
    {
      "epoch": 4.927394294634836,
      "grad_norm": 0.9803434014320374,
      "learning_rate": 3.313325958750336e-07,
      "loss": 0.0474,
      "step": 334575
    },
    {
      "epoch": 4.927762477724923,
      "grad_norm": 1.159724235534668,
      "learning_rate": 3.29696223908049e-07,
      "loss": 0.0519,
      "step": 334600
    },
    {
      "epoch": 4.92813066081501,
      "grad_norm": 0.7486581802368164,
      "learning_rate": 3.2805985194106443e-07,
      "loss": 0.0479,
      "step": 334625
    },
    {
      "epoch": 4.928498843905097,
      "grad_norm": 1.1323797702789307,
      "learning_rate": 3.264234799740799e-07,
      "loss": 0.0457,
      "step": 334650
    },
    {
      "epoch": 4.928867026995184,
      "grad_norm": 1.5626558065414429,
      "learning_rate": 3.2478710800709536e-07,
      "loss": 0.0516,
      "step": 334675
    },
    {
      "epoch": 4.929235210085271,
      "grad_norm": 1.535515308380127,
      "learning_rate": 3.2315073604011077e-07,
      "loss": 0.0549,
      "step": 334700
    },
    {
      "epoch": 4.929603393175358,
      "grad_norm": 0.9101071357727051,
      "learning_rate": 3.2151436407312623e-07,
      "loss": 0.0466,
      "step": 334725
    },
    {
      "epoch": 4.929971576265445,
      "grad_norm": 0.5328569412231445,
      "learning_rate": 3.1987799210614164e-07,
      "loss": 0.0464,
      "step": 334750
    },
    {
      "epoch": 4.930339759355532,
      "grad_norm": 1.1395955085754395,
      "learning_rate": 3.1824162013915705e-07,
      "loss": 0.0555,
      "step": 334775
    },
    {
      "epoch": 4.930707942445619,
      "grad_norm": 1.7324939966201782,
      "learning_rate": 3.1660524817217257e-07,
      "loss": 0.0507,
      "step": 334800
    },
    {
      "epoch": 4.931076125535706,
      "grad_norm": 1.212140679359436,
      "learning_rate": 3.14968876205188e-07,
      "loss": 0.0514,
      "step": 334825
    },
    {
      "epoch": 4.931444308625793,
      "grad_norm": 0.6742513179779053,
      "learning_rate": 3.1333250423820344e-07,
      "loss": 0.0499,
      "step": 334850
    },
    {
      "epoch": 4.93181249171588,
      "grad_norm": 1.4140636920928955,
      "learning_rate": 3.1169613227121885e-07,
      "loss": 0.0497,
      "step": 334875
    },
    {
      "epoch": 4.932180674805967,
      "grad_norm": 1.450247883796692,
      "learning_rate": 3.100597603042343e-07,
      "loss": 0.0578,
      "step": 334900
    },
    {
      "epoch": 4.932548857896054,
      "grad_norm": 1.3651790618896484,
      "learning_rate": 3.0842338833724973e-07,
      "loss": 0.0544,
      "step": 334925
    },
    {
      "epoch": 4.932917040986141,
      "grad_norm": 1.8230345249176025,
      "learning_rate": 3.067870163702652e-07,
      "loss": 0.0494,
      "step": 334950
    },
    {
      "epoch": 4.933285224076228,
      "grad_norm": 1.5822384357452393,
      "learning_rate": 3.051506444032806e-07,
      "loss": 0.0465,
      "step": 334975
    },
    {
      "epoch": 4.933653407166315,
      "grad_norm": 1.2938799858093262,
      "learning_rate": 3.0351427243629606e-07,
      "loss": 0.0515,
      "step": 335000
    },
    {
      "epoch": 4.934021590256402,
      "grad_norm": 1.7103651762008667,
      "learning_rate": 3.0187790046931153e-07,
      "loss": 0.0571,
      "step": 335025
    },
    {
      "epoch": 4.934389773346489,
      "grad_norm": 1.286973237991333,
      "learning_rate": 3.0024152850232694e-07,
      "loss": 0.0499,
      "step": 335050
    },
    {
      "epoch": 4.934757956436576,
      "grad_norm": 1.3599498271942139,
      "learning_rate": 2.986051565353424e-07,
      "loss": 0.0511,
      "step": 335075
    },
    {
      "epoch": 4.9351261395266635,
      "grad_norm": 1.8223161697387695,
      "learning_rate": 2.969687845683578e-07,
      "loss": 0.0534,
      "step": 335100
    },
    {
      "epoch": 4.9354943226167505,
      "grad_norm": 1.3782145977020264,
      "learning_rate": 2.953324126013733e-07,
      "loss": 0.0501,
      "step": 335125
    },
    {
      "epoch": 4.935862505706838,
      "grad_norm": 1.3739384412765503,
      "learning_rate": 2.9369604063438874e-07,
      "loss": 0.0565,
      "step": 335150
    },
    {
      "epoch": 4.936230688796925,
      "grad_norm": 1.2734593152999878,
      "learning_rate": 2.9205966866740415e-07,
      "loss": 0.0468,
      "step": 335175
    },
    {
      "epoch": 4.936598871887012,
      "grad_norm": 1.7262781858444214,
      "learning_rate": 2.904232967004196e-07,
      "loss": 0.053,
      "step": 335200
    },
    {
      "epoch": 4.936967054977099,
      "grad_norm": 1.7785955667495728,
      "learning_rate": 2.88786924733435e-07,
      "loss": 0.0555,
      "step": 335225
    },
    {
      "epoch": 4.937335238067186,
      "grad_norm": 1.5030748844146729,
      "learning_rate": 2.8715055276645043e-07,
      "loss": 0.0553,
      "step": 335250
    },
    {
      "epoch": 4.9377034211572735,
      "grad_norm": 1.154745101928711,
      "learning_rate": 2.855141807994659e-07,
      "loss": 0.0511,
      "step": 335275
    },
    {
      "epoch": 4.9380716042473605,
      "grad_norm": 1.1805689334869385,
      "learning_rate": 2.8387780883248136e-07,
      "loss": 0.0536,
      "step": 335300
    },
    {
      "epoch": 4.9384397873374475,
      "grad_norm": 1.0875493288040161,
      "learning_rate": 2.822414368654968e-07,
      "loss": 0.0552,
      "step": 335325
    },
    {
      "epoch": 4.9388079704275345,
      "grad_norm": 1.270925521850586,
      "learning_rate": 2.8060506489851223e-07,
      "loss": 0.0519,
      "step": 335350
    },
    {
      "epoch": 4.9391761535176215,
      "grad_norm": 1.1785509586334229,
      "learning_rate": 2.7896869293152764e-07,
      "loss": 0.0466,
      "step": 335375
    },
    {
      "epoch": 4.9395443366077085,
      "grad_norm": 1.1776304244995117,
      "learning_rate": 2.773323209645431e-07,
      "loss": 0.0521,
      "step": 335400
    },
    {
      "epoch": 4.9399125196977955,
      "grad_norm": 0.881881833076477,
      "learning_rate": 2.7569594899755857e-07,
      "loss": 0.0508,
      "step": 335425
    },
    {
      "epoch": 4.940280702787883,
      "grad_norm": 1.2219123840332031,
      "learning_rate": 2.7405957703057403e-07,
      "loss": 0.0542,
      "step": 335450
    },
    {
      "epoch": 4.94064888587797,
      "grad_norm": 1.6511127948760986,
      "learning_rate": 2.7242320506358944e-07,
      "loss": 0.0519,
      "step": 335475
    },
    {
      "epoch": 4.941017068968057,
      "grad_norm": 1.2264586687088013,
      "learning_rate": 2.7078683309660485e-07,
      "loss": 0.0474,
      "step": 335500
    },
    {
      "epoch": 4.941385252058144,
      "grad_norm": 0.7492833137512207,
      "learning_rate": 2.691504611296203e-07,
      "loss": 0.0479,
      "step": 335525
    },
    {
      "epoch": 4.941753435148231,
      "grad_norm": 1.6098127365112305,
      "learning_rate": 2.6751408916263573e-07,
      "loss": 0.0487,
      "step": 335550
    },
    {
      "epoch": 4.942121618238318,
      "grad_norm": 1.2260299921035767,
      "learning_rate": 2.658777171956512e-07,
      "loss": 0.0469,
      "step": 335575
    },
    {
      "epoch": 4.942489801328405,
      "grad_norm": 1.1029223203659058,
      "learning_rate": 2.6424134522866665e-07,
      "loss": 0.0503,
      "step": 335600
    },
    {
      "epoch": 4.942857984418492,
      "grad_norm": 2.017568349838257,
      "learning_rate": 2.6260497326168206e-07,
      "loss": 0.0488,
      "step": 335625
    },
    {
      "epoch": 4.943226167508579,
      "grad_norm": 1.5061942338943481,
      "learning_rate": 2.6096860129469753e-07,
      "loss": 0.0533,
      "step": 335650
    },
    {
      "epoch": 4.943594350598666,
      "grad_norm": 1.5082191228866577,
      "learning_rate": 2.5933222932771294e-07,
      "loss": 0.0533,
      "step": 335675
    },
    {
      "epoch": 4.943962533688753,
      "grad_norm": 0.6955373883247375,
      "learning_rate": 2.576958573607284e-07,
      "loss": 0.0553,
      "step": 335700
    },
    {
      "epoch": 4.94433071677884,
      "grad_norm": 1.2744516134262085,
      "learning_rate": 2.5605948539374386e-07,
      "loss": 0.0514,
      "step": 335725
    },
    {
      "epoch": 4.944698899868927,
      "grad_norm": 1.1599795818328857,
      "learning_rate": 2.5442311342675933e-07,
      "loss": 0.0557,
      "step": 335750
    },
    {
      "epoch": 4.945067082959014,
      "grad_norm": 0.999093770980835,
      "learning_rate": 2.5278674145977474e-07,
      "loss": 0.0526,
      "step": 335775
    },
    {
      "epoch": 4.945435266049101,
      "grad_norm": 1.3702365159988403,
      "learning_rate": 2.5115036949279015e-07,
      "loss": 0.0489,
      "step": 335800
    },
    {
      "epoch": 4.945803449139188,
      "grad_norm": 1.4919281005859375,
      "learning_rate": 2.495139975258056e-07,
      "loss": 0.0502,
      "step": 335825
    },
    {
      "epoch": 4.946171632229275,
      "grad_norm": 1.319697380065918,
      "learning_rate": 2.47877625558821e-07,
      "loss": 0.0543,
      "step": 335850
    },
    {
      "epoch": 4.946539815319362,
      "grad_norm": 1.5539913177490234,
      "learning_rate": 2.462412535918365e-07,
      "loss": 0.0498,
      "step": 335875
    },
    {
      "epoch": 4.946907998409449,
      "grad_norm": 1.0406968593597412,
      "learning_rate": 2.4460488162485195e-07,
      "loss": 0.0492,
      "step": 335900
    },
    {
      "epoch": 4.947276181499536,
      "grad_norm": 1.4123789072036743,
      "learning_rate": 2.4296850965786736e-07,
      "loss": 0.0534,
      "step": 335925
    },
    {
      "epoch": 4.947644364589623,
      "grad_norm": 0.5169786810874939,
      "learning_rate": 2.413321376908828e-07,
      "loss": 0.0473,
      "step": 335950
    },
    {
      "epoch": 4.94801254767971,
      "grad_norm": 1.24554443359375,
      "learning_rate": 2.3969576572389823e-07,
      "loss": 0.0457,
      "step": 335975
    },
    {
      "epoch": 4.948380730769797,
      "grad_norm": 1.6909687519073486,
      "learning_rate": 2.380593937569137e-07,
      "loss": 0.0562,
      "step": 336000
    },
    {
      "epoch": 4.948748913859884,
      "grad_norm": 1.3616148233413696,
      "learning_rate": 2.3642302178992913e-07,
      "loss": 0.0527,
      "step": 336025
    },
    {
      "epoch": 4.949117096949971,
      "grad_norm": 1.5458955764770508,
      "learning_rate": 2.3478664982294457e-07,
      "loss": 0.0539,
      "step": 336050
    },
    {
      "epoch": 4.949485280040058,
      "grad_norm": 1.1470768451690674,
      "learning_rate": 2.3315027785596003e-07,
      "loss": 0.0577,
      "step": 336075
    },
    {
      "epoch": 4.949853463130145,
      "grad_norm": 1.7016922235488892,
      "learning_rate": 2.3151390588897544e-07,
      "loss": 0.0524,
      "step": 336100
    },
    {
      "epoch": 4.950221646220232,
      "grad_norm": 1.0179957151412964,
      "learning_rate": 2.298775339219909e-07,
      "loss": 0.0509,
      "step": 336125
    },
    {
      "epoch": 4.950589829310319,
      "grad_norm": 1.4813505411148071,
      "learning_rate": 2.2824116195500634e-07,
      "loss": 0.0515,
      "step": 336150
    },
    {
      "epoch": 4.950958012400406,
      "grad_norm": 1.5040459632873535,
      "learning_rate": 2.2660478998802175e-07,
      "loss": 0.0554,
      "step": 336175
    },
    {
      "epoch": 4.951326195490493,
      "grad_norm": 1.3197249174118042,
      "learning_rate": 2.2496841802103722e-07,
      "loss": 0.049,
      "step": 336200
    },
    {
      "epoch": 4.951694378580581,
      "grad_norm": 1.9273678064346313,
      "learning_rate": 2.2333204605405265e-07,
      "loss": 0.0547,
      "step": 336225
    },
    {
      "epoch": 4.952062561670668,
      "grad_norm": 1.6828017234802246,
      "learning_rate": 2.2169567408706812e-07,
      "loss": 0.0513,
      "step": 336250
    },
    {
      "epoch": 4.952430744760755,
      "grad_norm": 1.3567432165145874,
      "learning_rate": 2.2005930212008353e-07,
      "loss": 0.0522,
      "step": 336275
    },
    {
      "epoch": 4.952798927850842,
      "grad_norm": 0.9874514937400818,
      "learning_rate": 2.1842293015309896e-07,
      "loss": 0.0468,
      "step": 336300
    },
    {
      "epoch": 4.953167110940929,
      "grad_norm": 0.7054497599601746,
      "learning_rate": 2.1678655818611443e-07,
      "loss": 0.046,
      "step": 336325
    },
    {
      "epoch": 4.953535294031016,
      "grad_norm": 1.7746872901916504,
      "learning_rate": 2.1515018621912986e-07,
      "loss": 0.0511,
      "step": 336350
    },
    {
      "epoch": 4.953903477121103,
      "grad_norm": 1.3093620538711548,
      "learning_rate": 2.135138142521453e-07,
      "loss": 0.0502,
      "step": 336375
    },
    {
      "epoch": 4.95427166021119,
      "grad_norm": 1.322843074798584,
      "learning_rate": 2.1187744228516074e-07,
      "loss": 0.0518,
      "step": 336400
    },
    {
      "epoch": 4.954639843301277,
      "grad_norm": 1.187206506729126,
      "learning_rate": 2.102410703181762e-07,
      "loss": 0.051,
      "step": 336425
    },
    {
      "epoch": 4.955008026391364,
      "grad_norm": 1.5089824199676514,
      "learning_rate": 2.0860469835119164e-07,
      "loss": 0.0508,
      "step": 336450
    },
    {
      "epoch": 4.955376209481451,
      "grad_norm": 1.4597768783569336,
      "learning_rate": 2.0696832638420705e-07,
      "loss": 0.0468,
      "step": 336475
    },
    {
      "epoch": 4.955744392571538,
      "grad_norm": 1.157888650894165,
      "learning_rate": 2.053319544172225e-07,
      "loss": 0.0472,
      "step": 336500
    },
    {
      "epoch": 4.956112575661625,
      "grad_norm": 2.018395185470581,
      "learning_rate": 2.0369558245023795e-07,
      "loss": 0.0479,
      "step": 336525
    },
    {
      "epoch": 4.956480758751712,
      "grad_norm": 1.4621374607086182,
      "learning_rate": 2.020592104832534e-07,
      "loss": 0.0497,
      "step": 336550
    },
    {
      "epoch": 4.956848941841799,
      "grad_norm": 1.0740878582000732,
      "learning_rate": 2.0042283851626882e-07,
      "loss": 0.0539,
      "step": 336575
    },
    {
      "epoch": 4.957217124931886,
      "grad_norm": 1.709808111190796,
      "learning_rate": 1.9878646654928426e-07,
      "loss": 0.0572,
      "step": 336600
    },
    {
      "epoch": 4.957585308021973,
      "grad_norm": 0.843580424785614,
      "learning_rate": 1.9715009458229972e-07,
      "loss": 0.0492,
      "step": 336625
    },
    {
      "epoch": 4.95795349111206,
      "grad_norm": 1.3082584142684937,
      "learning_rate": 1.9551372261531513e-07,
      "loss": 0.0446,
      "step": 336650
    },
    {
      "epoch": 4.958321674202147,
      "grad_norm": 1.191066861152649,
      "learning_rate": 1.938773506483306e-07,
      "loss": 0.0487,
      "step": 336675
    },
    {
      "epoch": 4.958689857292234,
      "grad_norm": 1.2488881349563599,
      "learning_rate": 1.9224097868134603e-07,
      "loss": 0.0504,
      "step": 336700
    },
    {
      "epoch": 4.959058040382321,
      "grad_norm": 1.416822075843811,
      "learning_rate": 1.9060460671436147e-07,
      "loss": 0.051,
      "step": 336725
    },
    {
      "epoch": 4.959426223472408,
      "grad_norm": 1.1460826396942139,
      "learning_rate": 1.8896823474737693e-07,
      "loss": 0.0532,
      "step": 336750
    },
    {
      "epoch": 4.959794406562495,
      "grad_norm": 1.1475660800933838,
      "learning_rate": 1.8733186278039234e-07,
      "loss": 0.0505,
      "step": 336775
    },
    {
      "epoch": 4.960162589652582,
      "grad_norm": 2.0693392753601074,
      "learning_rate": 1.856954908134078e-07,
      "loss": 0.0543,
      "step": 336800
    },
    {
      "epoch": 4.960530772742669,
      "grad_norm": 1.0268577337265015,
      "learning_rate": 1.8405911884642324e-07,
      "loss": 0.0512,
      "step": 336825
    },
    {
      "epoch": 4.960898955832756,
      "grad_norm": 0.741677463054657,
      "learning_rate": 1.8242274687943865e-07,
      "loss": 0.0458,
      "step": 336850
    },
    {
      "epoch": 4.961267138922843,
      "grad_norm": 1.309356689453125,
      "learning_rate": 1.8078637491245411e-07,
      "loss": 0.0572,
      "step": 336875
    },
    {
      "epoch": 4.96163532201293,
      "grad_norm": 1.411283254623413,
      "learning_rate": 1.7915000294546955e-07,
      "loss": 0.0515,
      "step": 336900
    },
    {
      "epoch": 4.962003505103017,
      "grad_norm": 1.1230144500732422,
      "learning_rate": 1.7751363097848501e-07,
      "loss": 0.0546,
      "step": 336925
    },
    {
      "epoch": 4.962371688193104,
      "grad_norm": 1.6702622175216675,
      "learning_rate": 1.7587725901150042e-07,
      "loss": 0.049,
      "step": 336950
    },
    {
      "epoch": 4.962739871283191,
      "grad_norm": 1.3394089937210083,
      "learning_rate": 1.7424088704451586e-07,
      "loss": 0.0526,
      "step": 336975
    },
    {
      "epoch": 4.963108054373278,
      "grad_norm": 2.077453136444092,
      "learning_rate": 1.7260451507753132e-07,
      "loss": 0.0489,
      "step": 337000
    },
    {
      "epoch": 4.963476237463365,
      "grad_norm": 1.6623622179031372,
      "learning_rate": 1.7096814311054676e-07,
      "loss": 0.0518,
      "step": 337025
    },
    {
      "epoch": 4.963844420553453,
      "grad_norm": 1.215965986251831,
      "learning_rate": 1.693317711435622e-07,
      "loss": 0.0551,
      "step": 337050
    },
    {
      "epoch": 4.96421260364354,
      "grad_norm": 1.5748342275619507,
      "learning_rate": 1.6769539917657764e-07,
      "loss": 0.0516,
      "step": 337075
    },
    {
      "epoch": 4.964580786733627,
      "grad_norm": 0.9643785953521729,
      "learning_rate": 1.660590272095931e-07,
      "loss": 0.0438,
      "step": 337100
    },
    {
      "epoch": 4.964948969823714,
      "grad_norm": 1.6236388683319092,
      "learning_rate": 1.6442265524260854e-07,
      "loss": 0.0505,
      "step": 337125
    },
    {
      "epoch": 4.965317152913801,
      "grad_norm": 1.2222963571548462,
      "learning_rate": 1.6278628327562395e-07,
      "loss": 0.0492,
      "step": 337150
    },
    {
      "epoch": 4.965685336003888,
      "grad_norm": 1.48945152759552,
      "learning_rate": 1.611499113086394e-07,
      "loss": 0.0478,
      "step": 337175
    },
    {
      "epoch": 4.966053519093975,
      "grad_norm": 1.3975740671157837,
      "learning_rate": 1.5951353934165485e-07,
      "loss": 0.0528,
      "step": 337200
    },
    {
      "epoch": 4.966421702184062,
      "grad_norm": 1.1448314189910889,
      "learning_rate": 1.5794262225334968e-07,
      "loss": 0.0586,
      "step": 337225
    },
    {
      "epoch": 4.966789885274149,
      "grad_norm": 1.710077166557312,
      "learning_rate": 1.563062502863651e-07,
      "loss": 0.055,
      "step": 337250
    },
    {
      "epoch": 4.9671580683642365,
      "grad_norm": 1.4766947031021118,
      "learning_rate": 1.5466987831938055e-07,
      "loss": 0.056,
      "step": 337275
    },
    {
      "epoch": 4.9675262514543235,
      "grad_norm": 1.4366049766540527,
      "learning_rate": 1.53033506352396e-07,
      "loss": 0.0531,
      "step": 337300
    },
    {
      "epoch": 4.9678944345444105,
      "grad_norm": 1.582160234451294,
      "learning_rate": 1.5139713438541142e-07,
      "loss": 0.0491,
      "step": 337325
    },
    {
      "epoch": 4.9682626176344975,
      "grad_norm": 1.4440860748291016,
      "learning_rate": 1.4976076241842686e-07,
      "loss": 0.0535,
      "step": 337350
    },
    {
      "epoch": 4.9686308007245845,
      "grad_norm": 1.2116984128952026,
      "learning_rate": 1.4812439045144232e-07,
      "loss": 0.0497,
      "step": 337375
    },
    {
      "epoch": 4.9689989838146715,
      "grad_norm": 1.1933037042617798,
      "learning_rate": 1.4655347336313713e-07,
      "loss": 0.0461,
      "step": 337400
    },
    {
      "epoch": 4.9693671669047585,
      "grad_norm": 1.3626468181610107,
      "learning_rate": 1.4491710139615257e-07,
      "loss": 0.0552,
      "step": 337425
    },
    {
      "epoch": 4.969735349994846,
      "grad_norm": 1.5436662435531616,
      "learning_rate": 1.4328072942916803e-07,
      "loss": 0.0505,
      "step": 337450
    },
    {
      "epoch": 4.970103533084933,
      "grad_norm": 0.9573881030082703,
      "learning_rate": 1.4164435746218344e-07,
      "loss": 0.0558,
      "step": 337475
    },
    {
      "epoch": 4.97047171617502,
      "grad_norm": 1.3884872198104858,
      "learning_rate": 1.400079854951989e-07,
      "loss": 0.0497,
      "step": 337500
    },
    {
      "epoch": 4.970839899265107,
      "grad_norm": 1.5933350324630737,
      "learning_rate": 1.3837161352821434e-07,
      "loss": 0.0534,
      "step": 337525
    },
    {
      "epoch": 4.971208082355194,
      "grad_norm": 1.4374070167541504,
      "learning_rate": 1.3673524156122978e-07,
      "loss": 0.0473,
      "step": 337550
    },
    {
      "epoch": 4.971576265445281,
      "grad_norm": 1.8629683256149292,
      "learning_rate": 1.3509886959424521e-07,
      "loss": 0.0521,
      "step": 337575
    },
    {
      "epoch": 4.971944448535368,
      "grad_norm": 1.319122552871704,
      "learning_rate": 1.3346249762726068e-07,
      "loss": 0.0593,
      "step": 337600
    },
    {
      "epoch": 4.972312631625455,
      "grad_norm": 1.3031715154647827,
      "learning_rate": 1.318261256602761e-07,
      "loss": 0.0471,
      "step": 337625
    },
    {
      "epoch": 4.972680814715542,
      "grad_norm": 0.5853040814399719,
      "learning_rate": 1.3018975369329152e-07,
      "loss": 0.0565,
      "step": 337650
    },
    {
      "epoch": 4.973048997805629,
      "grad_norm": 1.227638602256775,
      "learning_rate": 1.28553381726307e-07,
      "loss": 0.0497,
      "step": 337675
    },
    {
      "epoch": 4.973417180895716,
      "grad_norm": 1.5295565128326416,
      "learning_rate": 1.2691700975932242e-07,
      "loss": 0.0521,
      "step": 337700
    },
    {
      "epoch": 4.973785363985803,
      "grad_norm": 1.1376793384552002,
      "learning_rate": 1.2528063779233786e-07,
      "loss": 0.0529,
      "step": 337725
    },
    {
      "epoch": 4.97415354707589,
      "grad_norm": 1.0767399072647095,
      "learning_rate": 1.236442658253533e-07,
      "loss": 0.0563,
      "step": 337750
    },
    {
      "epoch": 4.974521730165977,
      "grad_norm": 1.2164992094039917,
      "learning_rate": 1.2200789385836873e-07,
      "loss": 0.0512,
      "step": 337775
    },
    {
      "epoch": 4.974889913256064,
      "grad_norm": 0.9008989334106445,
      "learning_rate": 1.2037152189138417e-07,
      "loss": 0.0584,
      "step": 337800
    },
    {
      "epoch": 4.975258096346151,
      "grad_norm": 1.5048881769180298,
      "learning_rate": 1.1873514992439962e-07,
      "loss": 0.0567,
      "step": 337825
    },
    {
      "epoch": 4.975626279436238,
      "grad_norm": 1.5705159902572632,
      "learning_rate": 1.1709877795741507e-07,
      "loss": 0.056,
      "step": 337850
    },
    {
      "epoch": 4.975994462526325,
      "grad_norm": 1.4366154670715332,
      "learning_rate": 1.154624059904305e-07,
      "loss": 0.0507,
      "step": 337875
    },
    {
      "epoch": 4.976362645616412,
      "grad_norm": 1.8623098134994507,
      "learning_rate": 1.1382603402344595e-07,
      "loss": 0.0496,
      "step": 337900
    },
    {
      "epoch": 4.976730828706499,
      "grad_norm": 0.9317939877510071,
      "learning_rate": 1.1218966205646138e-07,
      "loss": 0.0531,
      "step": 337925
    },
    {
      "epoch": 4.977099011796586,
      "grad_norm": 1.711613416671753,
      "learning_rate": 1.1055329008947683e-07,
      "loss": 0.0497,
      "step": 337950
    },
    {
      "epoch": 4.977467194886673,
      "grad_norm": 1.1050227880477905,
      "learning_rate": 1.0891691812249227e-07,
      "loss": 0.0539,
      "step": 337975
    },
    {
      "epoch": 4.97783537797676,
      "grad_norm": 1.5085375308990479,
      "learning_rate": 1.0728054615550772e-07,
      "loss": 0.0529,
      "step": 338000
    },
    {
      "epoch": 4.978203561066847,
      "grad_norm": 1.560699701309204,
      "learning_rate": 1.0564417418852314e-07,
      "loss": 0.0516,
      "step": 338025
    },
    {
      "epoch": 4.978571744156934,
      "grad_norm": 1.5578539371490479,
      "learning_rate": 1.0400780222153859e-07,
      "loss": 0.0525,
      "step": 338050
    },
    {
      "epoch": 4.978939927247021,
      "grad_norm": 1.4515794515609741,
      "learning_rate": 1.0237143025455403e-07,
      "loss": 0.0551,
      "step": 338075
    },
    {
      "epoch": 4.979308110337108,
      "grad_norm": 1.3008462190628052,
      "learning_rate": 1.0073505828756948e-07,
      "loss": 0.0555,
      "step": 338100
    },
    {
      "epoch": 4.979676293427195,
      "grad_norm": 1.4472436904907227,
      "learning_rate": 9.909868632058492e-08,
      "loss": 0.0421,
      "step": 338125
    },
    {
      "epoch": 4.980044476517283,
      "grad_norm": 1.3187246322631836,
      "learning_rate": 9.746231435360034e-08,
      "loss": 0.0555,
      "step": 338150
    },
    {
      "epoch": 4.98041265960737,
      "grad_norm": 1.2117855548858643,
      "learning_rate": 9.582594238661579e-08,
      "loss": 0.0512,
      "step": 338175
    },
    {
      "epoch": 4.980780842697457,
      "grad_norm": 1.4057400226593018,
      "learning_rate": 9.418957041963124e-08,
      "loss": 0.0467,
      "step": 338200
    },
    {
      "epoch": 4.981149025787544,
      "grad_norm": 1.093895435333252,
      "learning_rate": 9.255319845264668e-08,
      "loss": 0.0543,
      "step": 338225
    },
    {
      "epoch": 4.981517208877631,
      "grad_norm": 1.2608652114868164,
      "learning_rate": 9.091682648566213e-08,
      "loss": 0.0444,
      "step": 338250
    },
    {
      "epoch": 4.981885391967718,
      "grad_norm": 1.4286466836929321,
      "learning_rate": 8.928045451867756e-08,
      "loss": 0.0521,
      "step": 338275
    },
    {
      "epoch": 4.982253575057805,
      "grad_norm": 1.9670741558074951,
      "learning_rate": 8.764408255169299e-08,
      "loss": 0.0495,
      "step": 338300
    },
    {
      "epoch": 4.982621758147892,
      "grad_norm": 1.6306285858154297,
      "learning_rate": 8.600771058470844e-08,
      "loss": 0.0485,
      "step": 338325
    },
    {
      "epoch": 4.982989941237979,
      "grad_norm": 1.510868787765503,
      "learning_rate": 8.437133861772387e-08,
      "loss": 0.0544,
      "step": 338350
    },
    {
      "epoch": 4.983358124328066,
      "grad_norm": 1.2294979095458984,
      "learning_rate": 8.273496665073932e-08,
      "loss": 0.0528,
      "step": 338375
    },
    {
      "epoch": 4.983726307418153,
      "grad_norm": 1.2693171501159668,
      "learning_rate": 8.109859468375477e-08,
      "loss": 0.0453,
      "step": 338400
    },
    {
      "epoch": 4.98409449050824,
      "grad_norm": 1.668380618095398,
      "learning_rate": 7.94622227167702e-08,
      "loss": 0.0487,
      "step": 338425
    },
    {
      "epoch": 4.984462673598327,
      "grad_norm": 1.1777913570404053,
      "learning_rate": 7.782585074978563e-08,
      "loss": 0.0512,
      "step": 338450
    },
    {
      "epoch": 4.984830856688414,
      "grad_norm": 1.0354845523834229,
      "learning_rate": 7.618947878280108e-08,
      "loss": 0.0495,
      "step": 338475
    },
    {
      "epoch": 4.985199039778501,
      "grad_norm": 1.717870831489563,
      "learning_rate": 7.455310681581652e-08,
      "loss": 0.0563,
      "step": 338500
    },
    {
      "epoch": 4.985567222868588,
      "grad_norm": 1.194688320159912,
      "learning_rate": 7.291673484883196e-08,
      "loss": 0.0522,
      "step": 338525
    },
    {
      "epoch": 4.985935405958675,
      "grad_norm": 1.4643720388412476,
      "learning_rate": 7.128036288184741e-08,
      "loss": 0.0562,
      "step": 338550
    },
    {
      "epoch": 4.986303589048762,
      "grad_norm": 1.3760372400283813,
      "learning_rate": 6.964399091486284e-08,
      "loss": 0.0546,
      "step": 338575
    },
    {
      "epoch": 4.986671772138849,
      "grad_norm": 1.822245478630066,
      "learning_rate": 6.800761894787828e-08,
      "loss": 0.0555,
      "step": 338600
    },
    {
      "epoch": 4.987039955228936,
      "grad_norm": 1.3164429664611816,
      "learning_rate": 6.637124698089373e-08,
      "loss": 0.0549,
      "step": 338625
    },
    {
      "epoch": 4.987408138319023,
      "grad_norm": 1.501807689666748,
      "learning_rate": 6.473487501390917e-08,
      "loss": 0.0497,
      "step": 338650
    },
    {
      "epoch": 4.98777632140911,
      "grad_norm": 1.3571958541870117,
      "learning_rate": 6.30985030469246e-08,
      "loss": 0.0608,
      "step": 338675
    },
    {
      "epoch": 4.988144504499197,
      "grad_norm": 0.8018175363540649,
      "learning_rate": 6.146213107994005e-08,
      "loss": 0.054,
      "step": 338700
    },
    {
      "epoch": 4.988512687589284,
      "grad_norm": 1.8292900323867798,
      "learning_rate": 5.982575911295549e-08,
      "loss": 0.0517,
      "step": 338725
    },
    {
      "epoch": 4.988880870679371,
      "grad_norm": 0.9227701425552368,
      "learning_rate": 5.818938714597093e-08,
      "loss": 0.0533,
      "step": 338750
    },
    {
      "epoch": 4.989249053769458,
      "grad_norm": 1.1791300773620605,
      "learning_rate": 5.655301517898637e-08,
      "loss": 0.0547,
      "step": 338775
    },
    {
      "epoch": 4.989617236859545,
      "grad_norm": 1.451053500175476,
      "learning_rate": 5.491664321200181e-08,
      "loss": 0.0537,
      "step": 338800
    },
    {
      "epoch": 4.989985419949632,
      "grad_norm": 0.9784480929374695,
      "learning_rate": 5.328027124501725e-08,
      "loss": 0.0505,
      "step": 338825
    },
    {
      "epoch": 4.990353603039719,
      "grad_norm": 1.47343111038208,
      "learning_rate": 5.1643899278032695e-08,
      "loss": 0.0504,
      "step": 338850
    },
    {
      "epoch": 4.990721786129806,
      "grad_norm": 1.6505411863327026,
      "learning_rate": 5.000752731104813e-08,
      "loss": 0.0546,
      "step": 338875
    },
    {
      "epoch": 4.991089969219893,
      "grad_norm": 1.5374494791030884,
      "learning_rate": 4.8371155344063575e-08,
      "loss": 0.0503,
      "step": 338900
    },
    {
      "epoch": 4.99145815230998,
      "grad_norm": 1.2402594089508057,
      "learning_rate": 4.673478337707901e-08,
      "loss": 0.0517,
      "step": 338925
    },
    {
      "epoch": 4.991826335400067,
      "grad_norm": 1.457255482673645,
      "learning_rate": 4.5098411410094456e-08,
      "loss": 0.0515,
      "step": 338950
    },
    {
      "epoch": 4.992194518490155,
      "grad_norm": 0.909200131893158,
      "learning_rate": 4.34620394431099e-08,
      "loss": 0.0453,
      "step": 338975
    },
    {
      "epoch": 4.992562701580242,
      "grad_norm": 1.0346252918243408,
      "learning_rate": 4.1825667476125336e-08,
      "loss": 0.0477,
      "step": 339000
    },
    {
      "epoch": 4.992930884670329,
      "grad_norm": 1.4849971532821655,
      "learning_rate": 4.018929550914078e-08,
      "loss": 0.0562,
      "step": 339025
    },
    {
      "epoch": 4.993299067760416,
      "grad_norm": 1.4646254777908325,
      "learning_rate": 3.8552923542156216e-08,
      "loss": 0.0545,
      "step": 339050
    },
    {
      "epoch": 4.993667250850503,
      "grad_norm": 1.4139177799224854,
      "learning_rate": 3.691655157517166e-08,
      "loss": 0.0464,
      "step": 339075
    },
    {
      "epoch": 4.99403543394059,
      "grad_norm": 1.2818245887756348,
      "learning_rate": 3.52801796081871e-08,
      "loss": 0.0506,
      "step": 339100
    },
    {
      "epoch": 4.994403617030677,
      "grad_norm": 1.8238946199417114,
      "learning_rate": 3.364380764120254e-08,
      "loss": 0.0491,
      "step": 339125
    },
    {
      "epoch": 4.994771800120764,
      "grad_norm": 1.6911935806274414,
      "learning_rate": 3.200743567421798e-08,
      "loss": 0.0551,
      "step": 339150
    },
    {
      "epoch": 4.995139983210851,
      "grad_norm": 0.8746576905250549,
      "learning_rate": 3.037106370723342e-08,
      "loss": 0.0541,
      "step": 339175
    },
    {
      "epoch": 4.995508166300938,
      "grad_norm": 1.5822333097457886,
      "learning_rate": 2.8734691740248863e-08,
      "loss": 0.0532,
      "step": 339200
    },
    {
      "epoch": 4.995876349391025,
      "grad_norm": 1.1354644298553467,
      "learning_rate": 2.7098319773264303e-08,
      "loss": 0.0522,
      "step": 339225
    },
    {
      "epoch": 4.996244532481112,
      "grad_norm": 1.8758031129837036,
      "learning_rate": 2.5461947806279743e-08,
      "loss": 0.0489,
      "step": 339250
    },
    {
      "epoch": 4.9966127155711995,
      "grad_norm": 1.827846884727478,
      "learning_rate": 2.3825575839295183e-08,
      "loss": 0.0448,
      "step": 339275
    },
    {
      "epoch": 4.9969808986612865,
      "grad_norm": 1.3098126649856567,
      "learning_rate": 2.2189203872310627e-08,
      "loss": 0.0488,
      "step": 339300
    },
    {
      "epoch": 4.9973490817513735,
      "grad_norm": 1.4571337699890137,
      "learning_rate": 2.0552831905326067e-08,
      "loss": 0.0518,
      "step": 339325
    },
    {
      "epoch": 4.9977172648414605,
      "grad_norm": 0.8155379891395569,
      "learning_rate": 1.8916459938341504e-08,
      "loss": 0.0532,
      "step": 339350
    },
    {
      "epoch": 4.9980854479315475,
      "grad_norm": 1.5733963251113892,
      "learning_rate": 1.7280087971356947e-08,
      "loss": 0.055,
      "step": 339375
    },
    {
      "epoch": 4.9984536310216345,
      "grad_norm": 1.219093918800354,
      "learning_rate": 1.5643716004372387e-08,
      "loss": 0.0562,
      "step": 339400
    },
    {
      "epoch": 4.9988218141117216,
      "grad_norm": 1.0231002569198608,
      "learning_rate": 1.4007344037387827e-08,
      "loss": 0.059,
      "step": 339425
    },
    {
      "epoch": 4.999189997201809,
      "grad_norm": 1.2934322357177734,
      "learning_rate": 1.2370972070403267e-08,
      "loss": 0.0521,
      "step": 339450
    },
    {
      "epoch": 4.999558180291896,
      "grad_norm": 0.9300215840339661,
      "learning_rate": 1.0734600103418709e-08,
      "loss": 0.0537,
      "step": 339475
    },
    {
      "epoch": 4.999926363381983,
      "grad_norm": 1.3510725498199463,
      "learning_rate": 9.09822813643415e-09,
      "loss": 0.0472,
      "step": 339500
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.056979671120643616,
      "eval_runtime": 116.4543,
      "eval_samples_per_second": 3046.242,
      "eval_steps_per_second": 5.951,
      "step": 339505
    }
  ],
  "logging_steps": 25,
  "max_steps": 339505,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}