{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 721,
  "global_step": 3608,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002771618625277162,
      "grad_norm": 68.79409790039062,
      "learning_rate": 1e-05,
      "loss": 1.9656,
      "step": 1
    },
    {
      "epoch": 0.0005543237250554324,
      "grad_norm": 8.276924133300781,
      "learning_rate": 9.99999810457472e-06,
      "loss": 1.3782,
      "step": 2
    },
    {
      "epoch": 0.0008314855875831486,
      "grad_norm": 7.675302028656006,
      "learning_rate": 9.999992418300315e-06,
      "loss": 1.1511,
      "step": 3
    },
    {
      "epoch": 0.0011086474501108647,
      "grad_norm": 3.182561159133911,
      "learning_rate": 9.999982941181095e-06,
      "loss": 1.0919,
      "step": 4
    },
    {
      "epoch": 0.001385809312638581,
      "grad_norm": 3.8787267208099365,
      "learning_rate": 9.999969673224247e-06,
      "loss": 1.1424,
      "step": 5
    },
    {
      "epoch": 0.0016629711751662971,
      "grad_norm": 2.3091750144958496,
      "learning_rate": 9.999952614439832e-06,
      "loss": 1.0477,
      "step": 6
    },
    {
      "epoch": 0.0019401330376940134,
      "grad_norm": 2.278836965560913,
      "learning_rate": 9.999931764840779e-06,
      "loss": 1.0752,
      "step": 7
    },
    {
      "epoch": 0.0022172949002217295,
      "grad_norm": 2.0411248207092285,
      "learning_rate": 9.999907124442899e-06,
      "loss": 1.0309,
      "step": 8
    },
    {
      "epoch": 0.0024944567627494456,
      "grad_norm": 1.8744409084320068,
      "learning_rate": 9.999878693264873e-06,
      "loss": 0.9693,
      "step": 9
    },
    {
      "epoch": 0.002771618625277162,
      "grad_norm": 1.8021060228347778,
      "learning_rate": 9.999846471328256e-06,
      "loss": 0.9634,
      "step": 10
    },
    {
      "epoch": 0.003048780487804878,
      "grad_norm": 2.168149471282959,
      "learning_rate": 9.999810458657478e-06,
      "loss": 1.0303,
      "step": 11
    },
    {
      "epoch": 0.0033259423503325942,
      "grad_norm": 5.173647403717041,
      "learning_rate": 9.999770655279843e-06,
      "loss": 1.0446,
      "step": 12
    },
    {
      "epoch": 0.0036031042128603103,
      "grad_norm": 1.7857084274291992,
      "learning_rate": 9.99972706122553e-06,
      "loss": 0.9503,
      "step": 13
    },
    {
      "epoch": 0.003880266075388027,
      "grad_norm": 1.6743955612182617,
      "learning_rate": 9.999679676527588e-06,
      "loss": 0.9922,
      "step": 14
    },
    {
      "epoch": 0.0041574279379157425,
      "grad_norm": 1.6995370388031006,
      "learning_rate": 9.999628501221943e-06,
      "loss": 0.9625,
      "step": 15
    },
    {
      "epoch": 0.004434589800443459,
      "grad_norm": 1.7427685260772705,
      "learning_rate": 9.999573535347396e-06,
      "loss": 0.9252,
      "step": 16
    },
    {
      "epoch": 0.0047117516629711755,
      "grad_norm": 1.665605902671814,
      "learning_rate": 9.99951477894562e-06,
      "loss": 0.949,
      "step": 17
    },
    {
      "epoch": 0.004988913525498891,
      "grad_norm": 1.681048035621643,
      "learning_rate": 9.999452232061163e-06,
      "loss": 0.9218,
      "step": 18
    },
    {
      "epoch": 0.005266075388026608,
      "grad_norm": 1.5869801044464111,
      "learning_rate": 9.999385894741445e-06,
      "loss": 0.8499,
      "step": 19
    },
    {
      "epoch": 0.005543237250554324,
      "grad_norm": 1.671985387802124,
      "learning_rate": 9.99931576703676e-06,
      "loss": 0.935,
      "step": 20
    },
    {
      "epoch": 0.00582039911308204,
      "grad_norm": 1.618288516998291,
      "learning_rate": 9.99924184900028e-06,
      "loss": 0.8348,
      "step": 21
    },
    {
      "epoch": 0.006097560975609756,
      "grad_norm": 1.5699511766433716,
      "learning_rate": 9.999164140688046e-06,
      "loss": 0.928,
      "step": 22
    },
    {
      "epoch": 0.006374722838137472,
      "grad_norm": 1.590915322303772,
      "learning_rate": 9.999082642158972e-06,
      "loss": 0.9364,
      "step": 23
    },
    {
      "epoch": 0.0066518847006651885,
      "grad_norm": 1.4195369482040405,
      "learning_rate": 9.99899735347485e-06,
      "loss": 0.7412,
      "step": 24
    },
    {
      "epoch": 0.006929046563192905,
      "grad_norm": 1.5920929908752441,
      "learning_rate": 9.998908274700344e-06,
      "loss": 0.8546,
      "step": 25
    },
    {
      "epoch": 0.007206208425720621,
      "grad_norm": 1.7091312408447266,
      "learning_rate": 9.998815405902989e-06,
      "loss": 0.9089,
      "step": 26
    },
    {
      "epoch": 0.007483370288248337,
      "grad_norm": 1.6607009172439575,
      "learning_rate": 9.998718747153196e-06,
      "loss": 0.9343,
      "step": 27
    },
    {
      "epoch": 0.007760532150776054,
      "grad_norm": 1.5094019174575806,
      "learning_rate": 9.99861829852425e-06,
      "loss": 0.8615,
      "step": 28
    },
    {
      "epoch": 0.00803769401330377,
      "grad_norm": 1.4243597984313965,
      "learning_rate": 9.998514060092306e-06,
      "loss": 0.8245,
      "step": 29
    },
    {
      "epoch": 0.008314855875831485,
      "grad_norm": 1.5735164880752563,
      "learning_rate": 9.998406031936393e-06,
      "loss": 0.9165,
      "step": 30
    },
    {
      "epoch": 0.008592017738359202,
      "grad_norm": 1.5836375951766968,
      "learning_rate": 9.99829421413842e-06,
      "loss": 0.9331,
      "step": 31
    },
    {
      "epoch": 0.008869179600886918,
      "grad_norm": 1.5047401189804077,
      "learning_rate": 9.998178606783162e-06,
      "loss": 0.896,
      "step": 32
    },
    {
      "epoch": 0.009146341463414634,
      "grad_norm": 1.5114184617996216,
      "learning_rate": 9.998059209958266e-06,
      "loss": 0.9483,
      "step": 33
    },
    {
      "epoch": 0.009423503325942351,
      "grad_norm": 1.4533095359802246,
      "learning_rate": 9.997936023754258e-06,
      "loss": 0.8846,
      "step": 34
    },
    {
      "epoch": 0.009700665188470067,
      "grad_norm": 1.4530928134918213,
      "learning_rate": 9.997809048264533e-06,
      "loss": 0.9009,
      "step": 35
    },
    {
      "epoch": 0.009977827050997782,
      "grad_norm": 1.3865830898284912,
      "learning_rate": 9.997678283585361e-06,
      "loss": 0.7918,
      "step": 36
    },
    {
      "epoch": 0.0102549889135255,
      "grad_norm": 1.5267198085784912,
      "learning_rate": 9.997543729815881e-06,
      "loss": 0.8747,
      "step": 37
    },
    {
      "epoch": 0.010532150776053215,
      "grad_norm": 1.4894969463348389,
      "learning_rate": 9.997405387058113e-06,
      "loss": 0.8212,
      "step": 38
    },
    {
      "epoch": 0.010809312638580931,
      "grad_norm": 1.5563372373580933,
      "learning_rate": 9.99726325541694e-06,
      "loss": 0.8637,
      "step": 39
    },
    {
      "epoch": 0.011086474501108648,
      "grad_norm": 1.4693702459335327,
      "learning_rate": 9.997117335000122e-06,
      "loss": 0.7932,
      "step": 40
    },
    {
      "epoch": 0.011363636363636364,
      "grad_norm": 1.428948163986206,
      "learning_rate": 9.996967625918295e-06,
      "loss": 0.87,
      "step": 41
    },
    {
      "epoch": 0.01164079822616408,
      "grad_norm": 1.564921498298645,
      "learning_rate": 9.99681412828496e-06,
      "loss": 0.9204,
      "step": 42
    },
    {
      "epoch": 0.011917960088691795,
      "grad_norm": 1.4181956052780151,
      "learning_rate": 9.996656842216496e-06,
      "loss": 0.7806,
      "step": 43
    },
    {
      "epoch": 0.012195121951219513,
      "grad_norm": 1.4746185541152954,
      "learning_rate": 9.996495767832153e-06,
      "loss": 0.8631,
      "step": 44
    },
    {
      "epoch": 0.012472283813747228,
      "grad_norm": 1.6308395862579346,
      "learning_rate": 9.99633090525405e-06,
      "loss": 0.8783,
      "step": 45
    },
    {
      "epoch": 0.012749445676274944,
      "grad_norm": 1.6226578950881958,
      "learning_rate": 9.996162254607186e-06,
      "loss": 0.8477,
      "step": 46
    },
    {
      "epoch": 0.013026607538802661,
      "grad_norm": 1.9179185628890991,
      "learning_rate": 9.995989816019422e-06,
      "loss": 0.8889,
      "step": 47
    },
    {
      "epoch": 0.013303769401330377,
      "grad_norm": 1.640040397644043,
      "learning_rate": 9.9958135896215e-06,
      "loss": 0.8544,
      "step": 48
    },
    {
      "epoch": 0.013580931263858093,
      "grad_norm": 1.4565930366516113,
      "learning_rate": 9.995633575547027e-06,
      "loss": 0.791,
      "step": 49
    },
    {
      "epoch": 0.01385809312638581,
      "grad_norm": 1.7545428276062012,
      "learning_rate": 9.995449773932484e-06,
      "loss": 0.8125,
      "step": 50
    },
    {
      "epoch": 0.014135254988913526,
      "grad_norm": 1.7211869955062866,
      "learning_rate": 9.995262184917225e-06,
      "loss": 1.006,
      "step": 51
    },
    {
      "epoch": 0.014412416851441241,
      "grad_norm": 1.4764785766601562,
      "learning_rate": 9.995070808643473e-06,
      "loss": 0.8741,
      "step": 52
    },
    {
      "epoch": 0.014689578713968959,
      "grad_norm": 2.432535171508789,
      "learning_rate": 9.994875645256326e-06,
      "loss": 0.8201,
      "step": 53
    },
    {
      "epoch": 0.014966740576496674,
      "grad_norm": 1.717855453491211,
      "learning_rate": 9.994676694903748e-06,
      "loss": 0.8288,
      "step": 54
    },
    {
      "epoch": 0.01524390243902439,
      "grad_norm": 1.4470189809799194,
      "learning_rate": 9.99447395773658e-06,
      "loss": 0.8166,
      "step": 55
    },
    {
      "epoch": 0.015521064301552107,
      "grad_norm": 1.3151928186416626,
      "learning_rate": 9.994267433908533e-06,
      "loss": 0.7535,
      "step": 56
    },
    {
      "epoch": 0.01579822616407982,
      "grad_norm": 1.5647271871566772,
      "learning_rate": 9.994057123576182e-06,
      "loss": 0.8409,
      "step": 57
    },
    {
      "epoch": 0.01607538802660754,
      "grad_norm": 1.7927865982055664,
      "learning_rate": 9.99384302689898e-06,
      "loss": 0.8922,
      "step": 58
    },
    {
      "epoch": 0.016352549889135256,
      "grad_norm": 1.4496896266937256,
      "learning_rate": 9.99362514403925e-06,
      "loss": 0.8341,
      "step": 59
    },
    {
      "epoch": 0.01662971175166297,
      "grad_norm": 1.6189762353897095,
      "learning_rate": 9.993403475162185e-06,
      "loss": 0.8493,
      "step": 60
    },
    {
      "epoch": 0.016906873614190687,
      "grad_norm": 1.6460496187210083,
      "learning_rate": 9.993178020435843e-06,
      "loss": 0.9279,
      "step": 61
    },
    {
      "epoch": 0.017184035476718405,
      "grad_norm": 1.3378429412841797,
      "learning_rate": 9.992948780031162e-06,
      "loss": 0.8327,
      "step": 62
    },
    {
      "epoch": 0.01746119733924612,
      "grad_norm": 1.5132380723953247,
      "learning_rate": 9.992715754121943e-06,
      "loss": 0.7858,
      "step": 63
    },
    {
      "epoch": 0.017738359201773836,
      "grad_norm": 1.3593744039535522,
      "learning_rate": 9.99247894288486e-06,
      "loss": 0.8544,
      "step": 64
    },
    {
      "epoch": 0.018015521064301553,
      "grad_norm": 1.43010675907135,
      "learning_rate": 9.992238346499457e-06,
      "loss": 0.8334,
      "step": 65
    },
    {
      "epoch": 0.018292682926829267,
      "grad_norm": 1.487077236175537,
      "learning_rate": 9.991993965148144e-06,
      "loss": 0.7764,
      "step": 66
    },
    {
      "epoch": 0.018569844789356985,
      "grad_norm": 1.4395357370376587,
      "learning_rate": 9.991745799016206e-06,
      "loss": 0.8821,
      "step": 67
    },
    {
      "epoch": 0.018847006651884702,
      "grad_norm": 1.3526511192321777,
      "learning_rate": 9.991493848291796e-06,
      "loss": 0.8222,
      "step": 68
    },
    {
      "epoch": 0.019124168514412416,
      "grad_norm": 1.4752482175827026,
      "learning_rate": 9.991238113165934e-06,
      "loss": 0.8166,
      "step": 69
    },
    {
      "epoch": 0.019401330376940133,
      "grad_norm": 1.5390830039978027,
      "learning_rate": 9.99097859383251e-06,
      "loss": 0.826,
      "step": 70
    },
    {
      "epoch": 0.01967849223946785,
      "grad_norm": 1.4127715826034546,
      "learning_rate": 9.990715290488284e-06,
      "loss": 0.8636,
      "step": 71
    },
    {
      "epoch": 0.019955654101995565,
      "grad_norm": 1.3301950693130493,
      "learning_rate": 9.990448203332886e-06,
      "loss": 0.7789,
      "step": 72
    },
    {
      "epoch": 0.020232815964523282,
      "grad_norm": 1.4248167276382446,
      "learning_rate": 9.990177332568813e-06,
      "loss": 0.7926,
      "step": 73
    },
    {
      "epoch": 0.020509977827051,
      "grad_norm": 1.818259358406067,
      "learning_rate": 9.989902678401433e-06,
      "loss": 0.7629,
      "step": 74
    },
    {
      "epoch": 0.020787139689578713,
      "grad_norm": 1.562551736831665,
      "learning_rate": 9.989624241038977e-06,
      "loss": 0.7959,
      "step": 75
    },
    {
      "epoch": 0.02106430155210643,
      "grad_norm": 10.578536033630371,
      "learning_rate": 9.98934202069255e-06,
      "loss": 0.7801,
      "step": 76
    },
    {
      "epoch": 0.021341463414634148,
      "grad_norm": 1.819461703300476,
      "learning_rate": 9.989056017576122e-06,
      "loss": 0.7774,
      "step": 77
    },
    {
      "epoch": 0.021618625277161862,
      "grad_norm": 1.454900860786438,
      "learning_rate": 9.988766231906532e-06,
      "loss": 0.7819,
      "step": 78
    },
    {
      "epoch": 0.02189578713968958,
      "grad_norm": 1.324053406715393,
      "learning_rate": 9.988472663903488e-06,
      "loss": 0.7801,
      "step": 79
    },
    {
      "epoch": 0.022172949002217297,
      "grad_norm": 1.5011035203933716,
      "learning_rate": 9.988175313789564e-06,
      "loss": 0.8145,
      "step": 80
    },
    {
      "epoch": 0.02245011086474501,
      "grad_norm": 1.323172926902771,
      "learning_rate": 9.987874181790203e-06,
      "loss": 0.7682,
      "step": 81
    },
    {
      "epoch": 0.022727272727272728,
      "grad_norm": 1.3663185834884644,
      "learning_rate": 9.987569268133712e-06,
      "loss": 0.7656,
      "step": 82
    },
    {
      "epoch": 0.023004434589800442,
      "grad_norm": 1.351488471031189,
      "learning_rate": 9.987260573051268e-06,
      "loss": 0.8462,
      "step": 83
    },
    {
      "epoch": 0.02328159645232816,
      "grad_norm": 1.5657989978790283,
      "learning_rate": 9.986948096776916e-06,
      "loss": 0.8156,
      "step": 84
    },
    {
      "epoch": 0.023558758314855877,
      "grad_norm": 4.385202884674072,
      "learning_rate": 9.986631839547565e-06,
      "loss": 0.7888,
      "step": 85
    },
    {
      "epoch": 0.02383592017738359,
      "grad_norm": 1.5465915203094482,
      "learning_rate": 9.986311801602992e-06,
      "loss": 0.7855,
      "step": 86
    },
    {
      "epoch": 0.024113082039911308,
      "grad_norm": 1.2775983810424805,
      "learning_rate": 9.98598798318584e-06,
      "loss": 0.7827,
      "step": 87
    },
    {
      "epoch": 0.024390243902439025,
      "grad_norm": 1.3533475399017334,
      "learning_rate": 9.985660384541619e-06,
      "loss": 0.7681,
      "step": 88
    },
    {
      "epoch": 0.02466740576496674,
      "grad_norm": 2.272099733352661,
      "learning_rate": 9.985329005918702e-06,
      "loss": 0.7885,
      "step": 89
    },
    {
      "epoch": 0.024944567627494457,
      "grad_norm": 1.50556218624115,
      "learning_rate": 9.984993847568336e-06,
      "loss": 0.8524,
      "step": 90
    },
    {
      "epoch": 0.025221729490022174,
      "grad_norm": 1.3322641849517822,
      "learning_rate": 9.984654909744623e-06,
      "loss": 0.8236,
      "step": 91
    },
    {
      "epoch": 0.025498891352549888,
      "grad_norm": 1.5123928785324097,
      "learning_rate": 9.984312192704535e-06,
      "loss": 0.8197,
      "step": 92
    },
    {
      "epoch": 0.025776053215077605,
      "grad_norm": 1.6728800535202026,
      "learning_rate": 9.983965696707913e-06,
      "loss": 0.7816,
      "step": 93
    },
    {
      "epoch": 0.026053215077605323,
      "grad_norm": 1.3960485458374023,
      "learning_rate": 9.983615422017459e-06,
      "loss": 0.7974,
      "step": 94
    },
    {
      "epoch": 0.026330376940133036,
      "grad_norm": 1.3659459352493286,
      "learning_rate": 9.983261368898742e-06,
      "loss": 0.7997,
      "step": 95
    },
    {
      "epoch": 0.026607538802660754,
      "grad_norm": 1.9001489877700806,
      "learning_rate": 9.982903537620191e-06,
      "loss": 0.7755,
      "step": 96
    },
    {
      "epoch": 0.02688470066518847,
      "grad_norm": 1.509147047996521,
      "learning_rate": 9.982541928453105e-06,
      "loss": 0.7893,
      "step": 97
    },
    {
      "epoch": 0.027161862527716185,
      "grad_norm": 1.3140685558319092,
      "learning_rate": 9.982176541671646e-06,
      "loss": 0.7995,
      "step": 98
    },
    {
      "epoch": 0.027439024390243903,
      "grad_norm": 1.4455986022949219,
      "learning_rate": 9.98180737755284e-06,
      "loss": 0.8061,
      "step": 99
    },
    {
      "epoch": 0.02771618625277162,
      "grad_norm": 1.30985426902771,
      "learning_rate": 9.981434436376572e-06,
      "loss": 0.7864,
      "step": 100
    },
    {
      "epoch": 0.027993348115299334,
      "grad_norm": 1.482763648033142,
      "learning_rate": 9.9810577184256e-06,
      "loss": 0.748,
      "step": 101
    },
    {
      "epoch": 0.02827050997782705,
      "grad_norm": 1.3551603555679321,
      "learning_rate": 9.980677223985538e-06,
      "loss": 0.824,
      "step": 102
    },
    {
      "epoch": 0.02854767184035477,
      "grad_norm": 1.378212332725525,
      "learning_rate": 9.980292953344864e-06,
      "loss": 0.8259,
      "step": 103
    },
    {
      "epoch": 0.028824833702882482,
      "grad_norm": 1.3249517679214478,
      "learning_rate": 9.979904906794922e-06,
      "loss": 0.7289,
      "step": 104
    },
    {
      "epoch": 0.0291019955654102,
      "grad_norm": 1.4336422681808472,
      "learning_rate": 9.979513084629917e-06,
      "loss": 0.8458,
      "step": 105
    },
    {
      "epoch": 0.029379157427937917,
      "grad_norm": 1.5225223302841187,
      "learning_rate": 9.979117487146919e-06,
      "loss": 0.824,
      "step": 106
    },
    {
      "epoch": 0.02965631929046563,
      "grad_norm": 1.3841367959976196,
      "learning_rate": 9.978718114645854e-06,
      "loss": 0.8725,
      "step": 107
    },
    {
      "epoch": 0.02993348115299335,
      "grad_norm": 1.3778966665267944,
      "learning_rate": 9.978314967429518e-06,
      "loss": 0.7865,
      "step": 108
    },
    {
      "epoch": 0.030210643015521066,
      "grad_norm": 1.3829749822616577,
      "learning_rate": 9.977908045803564e-06,
      "loss": 0.7929,
      "step": 109
    },
    {
      "epoch": 0.03048780487804878,
      "grad_norm": 1.24405837059021,
      "learning_rate": 9.977497350076509e-06,
      "loss": 0.7679,
      "step": 110
    },
    {
      "epoch": 0.030764966740576497,
      "grad_norm": 1.313050389289856,
      "learning_rate": 9.977082880559725e-06,
      "loss": 0.7712,
      "step": 111
    },
    {
      "epoch": 0.031042128603104215,
      "grad_norm": 1.3578225374221802,
      "learning_rate": 9.976664637567456e-06,
      "loss": 0.8019,
      "step": 112
    },
    {
      "epoch": 0.03131929046563193,
      "grad_norm": 1.2553681135177612,
      "learning_rate": 9.9762426214168e-06,
      "loss": 0.7375,
      "step": 113
    },
    {
      "epoch": 0.03159645232815964,
      "grad_norm": 1.5658224821090698,
      "learning_rate": 9.975816832427716e-06,
      "loss": 0.8494,
      "step": 114
    },
    {
      "epoch": 0.03187361419068736,
      "grad_norm": 1.3735398054122925,
      "learning_rate": 9.975387270923026e-06,
      "loss": 0.7552,
      "step": 115
    },
    {
      "epoch": 0.03215077605321508,
      "grad_norm": 1.3076518774032593,
      "learning_rate": 9.97495393722841e-06,
      "loss": 0.8107,
      "step": 116
    },
    {
      "epoch": 0.03242793791574279,
      "grad_norm": 1.353601098060608,
      "learning_rate": 9.974516831672407e-06,
      "loss": 0.7563,
      "step": 117
    },
    {
      "epoch": 0.03270509977827051,
      "grad_norm": 1.345301866531372,
      "learning_rate": 9.97407595458642e-06,
      "loss": 0.7805,
      "step": 118
    },
    {
      "epoch": 0.032982261640798226,
      "grad_norm": 1.4127788543701172,
      "learning_rate": 9.973631306304707e-06,
      "loss": 0.7957,
      "step": 119
    },
    {
      "epoch": 0.03325942350332594,
      "grad_norm": 1.2626283168792725,
      "learning_rate": 9.97318288716439e-06,
      "loss": 0.7666,
      "step": 120
    },
    {
      "epoch": 0.03353658536585366,
      "grad_norm": 1.5355186462402344,
      "learning_rate": 9.972730697505442e-06,
      "loss": 0.8153,
      "step": 121
    },
    {
      "epoch": 0.033813747228381374,
      "grad_norm": 1.284247875213623,
      "learning_rate": 9.972274737670702e-06,
      "loss": 0.7191,
      "step": 122
    },
    {
      "epoch": 0.03409090909090909,
      "grad_norm": 1.3518129587173462,
      "learning_rate": 9.971815008005867e-06,
      "loss": 0.7715,
      "step": 123
    },
    {
      "epoch": 0.03436807095343681,
      "grad_norm": 1.6222867965698242,
      "learning_rate": 9.971351508859488e-06,
      "loss": 0.8374,
      "step": 124
    },
    {
      "epoch": 0.03464523281596452,
      "grad_norm": 1.3930152654647827,
      "learning_rate": 9.970884240582976e-06,
      "loss": 0.8335,
      "step": 125
    },
    {
      "epoch": 0.03492239467849224,
      "grad_norm": 1.3259778022766113,
      "learning_rate": 9.970413203530602e-06,
      "loss": 0.8402,
      "step": 126
    },
    {
      "epoch": 0.03519955654101996,
      "grad_norm": 1.3834583759307861,
      "learning_rate": 9.969938398059491e-06,
      "loss": 0.793,
      "step": 127
    },
    {
      "epoch": 0.03547671840354767,
      "grad_norm": 1.4983466863632202,
      "learning_rate": 9.969459824529626e-06,
      "loss": 0.7192,
      "step": 128
    },
    {
      "epoch": 0.035753880266075386,
      "grad_norm": 1.2399680614471436,
      "learning_rate": 9.968977483303848e-06,
      "loss": 0.7745,
      "step": 129
    },
    {
      "epoch": 0.03603104212860311,
      "grad_norm": 1.3195708990097046,
      "learning_rate": 9.968491374747854e-06,
      "loss": 0.735,
      "step": 130
    },
    {
      "epoch": 0.03630820399113082,
      "grad_norm": 1.9255726337432861,
      "learning_rate": 9.968001499230194e-06,
      "loss": 0.8113,
      "step": 131
    },
    {
      "epoch": 0.036585365853658534,
      "grad_norm": 1.3445053100585938,
      "learning_rate": 9.96750785712228e-06,
      "loss": 0.7587,
      "step": 132
    },
    {
      "epoch": 0.036862527716186255,
      "grad_norm": 1.4876223802566528,
      "learning_rate": 9.967010448798376e-06,
      "loss": 0.7765,
      "step": 133
    },
    {
      "epoch": 0.03713968957871397,
      "grad_norm": 1.4350440502166748,
      "learning_rate": 9.966509274635603e-06,
      "loss": 0.7486,
      "step": 134
    },
    {
      "epoch": 0.03741685144124168,
      "grad_norm": 1.2283313274383545,
      "learning_rate": 9.966004335013933e-06,
      "loss": 0.7417,
      "step": 135
    },
    {
      "epoch": 0.037694013303769404,
      "grad_norm": 1.4492980241775513,
      "learning_rate": 9.9654956303162e-06,
      "loss": 0.7726,
      "step": 136
    },
    {
      "epoch": 0.03797117516629712,
      "grad_norm": 1.5209165811538696,
      "learning_rate": 9.964983160928085e-06,
      "loss": 0.781,
      "step": 137
    },
    {
      "epoch": 0.03824833702882483,
      "grad_norm": 1.41078519821167,
      "learning_rate": 9.964466927238132e-06,
      "loss": 0.799,
      "step": 138
    },
    {
      "epoch": 0.03852549889135255,
      "grad_norm": 1.6199933290481567,
      "learning_rate": 9.963946929637728e-06,
      "loss": 0.7742,
      "step": 139
    },
    {
      "epoch": 0.038802660753880266,
      "grad_norm": 1.4153028726577759,
      "learning_rate": 9.963423168521126e-06,
      "loss": 0.8217,
      "step": 140
    },
    {
      "epoch": 0.03907982261640798,
      "grad_norm": 1.2682759761810303,
      "learning_rate": 9.96289564428542e-06,
      "loss": 0.7448,
      "step": 141
    },
    {
      "epoch": 0.0393569844789357,
      "grad_norm": 1.422242522239685,
      "learning_rate": 9.962364357330567e-06,
      "loss": 0.8022,
      "step": 142
    },
    {
      "epoch": 0.039634146341463415,
      "grad_norm": 1.3762917518615723,
      "learning_rate": 9.961829308059372e-06,
      "loss": 0.8376,
      "step": 143
    },
    {
      "epoch": 0.03991130820399113,
      "grad_norm": 1.2771761417388916,
      "learning_rate": 9.961290496877492e-06,
      "loss": 0.7942,
      "step": 144
    },
    {
      "epoch": 0.04018847006651885,
      "grad_norm": 1.2935817241668701,
      "learning_rate": 9.960747924193439e-06,
      "loss": 0.778,
      "step": 145
    },
    {
      "epoch": 0.040465631929046564,
      "grad_norm": 1.1905834674835205,
      "learning_rate": 9.960201590418574e-06,
      "loss": 0.7225,
      "step": 146
    },
    {
      "epoch": 0.04074279379157428,
      "grad_norm": 1.2269259691238403,
      "learning_rate": 9.959651495967113e-06,
      "loss": 0.7384,
      "step": 147
    },
    {
      "epoch": 0.041019955654102,
      "grad_norm": 1.2341827154159546,
      "learning_rate": 9.95909764125612e-06,
      "loss": 0.7403,
      "step": 148
    },
    {
      "epoch": 0.04129711751662971,
      "grad_norm": 1.2603411674499512,
      "learning_rate": 9.95854002670551e-06,
      "loss": 0.7546,
      "step": 149
    },
    {
      "epoch": 0.041574279379157426,
      "grad_norm": 1.2231045961380005,
      "learning_rate": 9.957978652738051e-06,
      "loss": 0.7806,
      "step": 150
    },
    {
      "epoch": 0.04185144124168515,
      "grad_norm": 1.5159447193145752,
      "learning_rate": 9.95741351977936e-06,
      "loss": 0.7737,
      "step": 151
    },
    {
      "epoch": 0.04212860310421286,
      "grad_norm": 1.4394538402557373,
      "learning_rate": 9.956844628257904e-06,
      "loss": 0.7522,
      "step": 152
    },
    {
      "epoch": 0.042405764966740575,
      "grad_norm": 1.2448875904083252,
      "learning_rate": 9.956271978605e-06,
      "loss": 0.6686,
      "step": 153
    },
    {
      "epoch": 0.042682926829268296,
      "grad_norm": 1.3425490856170654,
      "learning_rate": 9.955695571254813e-06,
      "loss": 0.744,
      "step": 154
    },
    {
      "epoch": 0.04296008869179601,
      "grad_norm": 1.4192100763320923,
      "learning_rate": 9.955115406644357e-06,
      "loss": 0.7612,
      "step": 155
    },
    {
      "epoch": 0.043237250554323724,
      "grad_norm": 1.2422420978546143,
      "learning_rate": 9.954531485213497e-06,
      "loss": 0.7275,
      "step": 156
    },
    {
      "epoch": 0.043514412416851445,
      "grad_norm": 1.3473109006881714,
      "learning_rate": 9.953943807404945e-06,
      "loss": 0.7639,
      "step": 157
    },
    {
      "epoch": 0.04379157427937916,
      "grad_norm": 1.4643961191177368,
      "learning_rate": 9.953352373664259e-06,
      "loss": 0.7327,
      "step": 158
    },
    {
      "epoch": 0.04406873614190687,
      "grad_norm": 1.269757628440857,
      "learning_rate": 9.952757184439845e-06,
      "loss": 0.8316,
      "step": 159
    },
    {
      "epoch": 0.04434589800443459,
      "grad_norm": 1.2889595031738281,
      "learning_rate": 9.952158240182963e-06,
      "loss": 0.7149,
      "step": 160
    },
    {
      "epoch": 0.04462305986696231,
      "grad_norm": 1.3815184831619263,
      "learning_rate": 9.95155554134771e-06,
      "loss": 0.8287,
      "step": 161
    },
    {
      "epoch": 0.04490022172949002,
      "grad_norm": 1.4527182579040527,
      "learning_rate": 9.950949088391038e-06,
      "loss": 0.8382,
      "step": 162
    },
    {
      "epoch": 0.045177383592017735,
      "grad_norm": 1.3789409399032593,
      "learning_rate": 9.950338881772736e-06,
      "loss": 0.7423,
      "step": 163
    },
    {
      "epoch": 0.045454545454545456,
      "grad_norm": 1.2654200792312622,
      "learning_rate": 9.94972492195545e-06,
      "loss": 0.741,
      "step": 164
    },
    {
      "epoch": 0.04573170731707317,
      "grad_norm": 1.6224985122680664,
      "learning_rate": 9.949107209404664e-06,
      "loss": 0.6958,
      "step": 165
    },
    {
      "epoch": 0.046008869179600884,
      "grad_norm": 1.2471752166748047,
      "learning_rate": 9.948485744588709e-06,
      "loss": 0.7324,
      "step": 166
    },
    {
      "epoch": 0.046286031042128604,
      "grad_norm": 1.4502007961273193,
      "learning_rate": 9.94786052797876e-06,
      "loss": 0.8086,
      "step": 167
    },
    {
      "epoch": 0.04656319290465632,
      "grad_norm": 1.3897017240524292,
      "learning_rate": 9.94723156004884e-06,
      "loss": 0.7758,
      "step": 168
    },
    {
      "epoch": 0.04684035476718403,
      "grad_norm": 1.2692089080810547,
      "learning_rate": 9.946598841275812e-06,
      "loss": 0.7872,
      "step": 169
    },
    {
      "epoch": 0.04711751662971175,
      "grad_norm": 1.404775857925415,
      "learning_rate": 9.945962372139384e-06,
      "loss": 0.8186,
      "step": 170
    },
    {
      "epoch": 0.04739467849223947,
      "grad_norm": 1.517909288406372,
      "learning_rate": 9.945322153122108e-06,
      "loss": 0.8321,
      "step": 171
    },
    {
      "epoch": 0.04767184035476718,
      "grad_norm": 1.3688514232635498,
      "learning_rate": 9.94467818470938e-06,
      "loss": 0.7258,
      "step": 172
    },
    {
      "epoch": 0.0479490022172949,
      "grad_norm": 1.4006280899047852,
      "learning_rate": 9.944030467389438e-06,
      "loss": 0.8184,
      "step": 173
    },
    {
      "epoch": 0.048226164079822616,
      "grad_norm": 1.340347170829773,
      "learning_rate": 9.943379001653359e-06,
      "loss": 0.804,
      "step": 174
    },
    {
      "epoch": 0.04850332594235033,
      "grad_norm": 1.3811019659042358,
      "learning_rate": 9.94272378799507e-06,
      "loss": 0.8011,
      "step": 175
    },
    {
      "epoch": 0.04878048780487805,
      "grad_norm": 1.362228512763977,
      "learning_rate": 9.942064826911328e-06,
      "loss": 0.7009,
      "step": 176
    },
    {
      "epoch": 0.049057649667405764,
      "grad_norm": 1.2894893884658813,
      "learning_rate": 9.941402118901743e-06,
      "loss": 0.7501,
      "step": 177
    },
    {
      "epoch": 0.04933481152993348,
      "grad_norm": 1.288261890411377,
      "learning_rate": 9.940735664468758e-06,
      "loss": 0.761,
      "step": 178
    },
    {
      "epoch": 0.0496119733924612,
      "grad_norm": 1.4531985521316528,
      "learning_rate": 9.940065464117659e-06,
      "loss": 0.7772,
      "step": 179
    },
    {
      "epoch": 0.04988913525498891,
      "grad_norm": 1.3255290985107422,
      "learning_rate": 9.939391518356571e-06,
      "loss": 0.8083,
      "step": 180
    },
    {
      "epoch": 0.05016629711751663,
      "grad_norm": 1.3454551696777344,
      "learning_rate": 9.938713827696462e-06,
      "loss": 0.7886,
      "step": 181
    },
    {
      "epoch": 0.05044345898004435,
      "grad_norm": 1.4141685962677002,
      "learning_rate": 9.938032392651136e-06,
      "loss": 0.7087,
      "step": 182
    },
    {
      "epoch": 0.05072062084257206,
      "grad_norm": 1.2578986883163452,
      "learning_rate": 9.937347213737234e-06,
      "loss": 0.7627,
      "step": 183
    },
    {
      "epoch": 0.050997782705099776,
      "grad_norm": 1.4219942092895508,
      "learning_rate": 9.936658291474242e-06,
      "loss": 0.7254,
      "step": 184
    },
    {
      "epoch": 0.051274944567627496,
      "grad_norm": 1.2239954471588135,
      "learning_rate": 9.935965626384477e-06,
      "loss": 0.8263,
      "step": 185
    },
    {
      "epoch": 0.05155210643015521,
      "grad_norm": 1.3954108953475952,
      "learning_rate": 9.935269218993099e-06,
      "loss": 0.7397,
      "step": 186
    },
    {
      "epoch": 0.051829268292682924,
      "grad_norm": 1.7105181217193604,
      "learning_rate": 9.934569069828104e-06,
      "loss": 0.7931,
      "step": 187
    },
    {
      "epoch": 0.052106430155210645,
      "grad_norm": 1.2178845405578613,
      "learning_rate": 9.93386517942032e-06,
      "loss": 0.799,
      "step": 188
    },
    {
      "epoch": 0.05238359201773836,
      "grad_norm": 1.3326935768127441,
      "learning_rate": 9.93315754830342e-06,
      "loss": 0.7712,
      "step": 189
    },
    {
      "epoch": 0.05266075388026607,
      "grad_norm": 1.2806878089904785,
      "learning_rate": 9.932446177013908e-06,
      "loss": 0.7704,
      "step": 190
    },
    {
      "epoch": 0.052937915742793794,
      "grad_norm": 1.5231550931930542,
      "learning_rate": 9.931731066091123e-06,
      "loss": 0.7633,
      "step": 191
    },
    {
      "epoch": 0.05321507760532151,
      "grad_norm": 1.4274626970291138,
      "learning_rate": 9.931012216077242e-06,
      "loss": 0.7317,
      "step": 192
    },
    {
      "epoch": 0.05349223946784922,
      "grad_norm": 1.9555613994598389,
      "learning_rate": 9.930289627517275e-06,
      "loss": 0.8012,
      "step": 193
    },
    {
      "epoch": 0.05376940133037694,
      "grad_norm": 1.4476178884506226,
      "learning_rate": 9.929563300959066e-06,
      "loss": 0.7751,
      "step": 194
    },
    {
      "epoch": 0.054046563192904656,
      "grad_norm": 1.2330094575881958,
      "learning_rate": 9.928833236953297e-06,
      "loss": 0.7827,
      "step": 195
    },
    {
      "epoch": 0.05432372505543237,
      "grad_norm": 1.233737826347351,
      "learning_rate": 9.928099436053479e-06,
      "loss": 0.7567,
      "step": 196
    },
    {
      "epoch": 0.05460088691796009,
      "grad_norm": 1.6478792428970337,
      "learning_rate": 9.927361898815957e-06,
      "loss": 0.7509,
      "step": 197
    },
    {
      "epoch": 0.054878048780487805,
      "grad_norm": 1.3135851621627808,
      "learning_rate": 9.92662062579991e-06,
      "loss": 0.7153,
      "step": 198
    },
    {
      "epoch": 0.05515521064301552,
      "grad_norm": 1.738797664642334,
      "learning_rate": 9.92587561756735e-06,
      "loss": 0.8361,
      "step": 199
    },
    {
      "epoch": 0.05543237250554324,
      "grad_norm": 1.5317659378051758,
      "learning_rate": 9.92512687468312e-06,
      "loss": 0.7479,
      "step": 200
    },
    {
      "epoch": 0.055709534368070954,
      "grad_norm": 1.4397844076156616,
      "learning_rate": 9.924374397714895e-06,
      "loss": 0.8427,
      "step": 201
    },
    {
      "epoch": 0.05598669623059867,
      "grad_norm": 1.4012843370437622,
      "learning_rate": 9.923618187233177e-06,
      "loss": 0.7262,
      "step": 202
    },
    {
      "epoch": 0.05626385809312639,
      "grad_norm": 1.424390196800232,
      "learning_rate": 9.922858243811307e-06,
      "loss": 0.7184,
      "step": 203
    },
    {
      "epoch": 0.0565410199556541,
      "grad_norm": 1.4714741706848145,
      "learning_rate": 9.922094568025449e-06,
      "loss": 0.7937,
      "step": 204
    },
    {
      "epoch": 0.056818181818181816,
      "grad_norm": 1.2887104749679565,
      "learning_rate": 9.921327160454598e-06,
      "loss": 0.7678,
      "step": 205
    },
    {
      "epoch": 0.05709534368070954,
      "grad_norm": 1.342674970626831,
      "learning_rate": 9.92055602168058e-06,
      "loss": 0.7078,
      "step": 206
    },
    {
      "epoch": 0.05737250554323725,
      "grad_norm": 1.300392508506775,
      "learning_rate": 9.919781152288053e-06,
      "loss": 0.7295,
      "step": 207
    },
    {
      "epoch": 0.057649667405764965,
      "grad_norm": 1.2475287914276123,
      "learning_rate": 9.919002552864496e-06,
      "loss": 0.7958,
      "step": 208
    },
    {
      "epoch": 0.057926829268292686,
      "grad_norm": 1.523695945739746,
      "learning_rate": 9.91822022400022e-06,
      "loss": 0.7338,
      "step": 209
    },
    {
      "epoch": 0.0582039911308204,
      "grad_norm": 1.3299717903137207,
      "learning_rate": 9.917434166288364e-06,
      "loss": 0.7658,
      "step": 210
    },
    {
      "epoch": 0.058481152993348114,
      "grad_norm": 1.5577712059020996,
      "learning_rate": 9.916644380324896e-06,
      "loss": 0.767,
      "step": 211
    },
    {
      "epoch": 0.058758314855875834,
      "grad_norm": 1.234549641609192,
      "learning_rate": 9.915850866708604e-06,
      "loss": 0.7675,
      "step": 212
    },
    {
      "epoch": 0.05903547671840355,
      "grad_norm": 1.1839942932128906,
      "learning_rate": 9.915053626041107e-06,
      "loss": 0.7461,
      "step": 213
    },
    {
      "epoch": 0.05931263858093126,
      "grad_norm": 1.2250906229019165,
      "learning_rate": 9.91425265892685e-06,
      "loss": 0.789,
      "step": 214
    },
    {
      "epoch": 0.05958980044345898,
      "grad_norm": 1.413956642150879,
      "learning_rate": 9.913447965973105e-06,
      "loss": 0.7346,
      "step": 215
    },
    {
      "epoch": 0.0598669623059867,
      "grad_norm": 1.2320823669433594,
      "learning_rate": 9.912639547789962e-06,
      "loss": 0.7848,
      "step": 216
    },
    {
      "epoch": 0.06014412416851441,
      "grad_norm": 1.2523480653762817,
      "learning_rate": 9.911827404990341e-06,
      "loss": 0.78,
      "step": 217
    },
    {
      "epoch": 0.06042128603104213,
      "grad_norm": 1.2115002870559692,
      "learning_rate": 9.911011538189985e-06,
      "loss": 0.7404,
      "step": 218
    },
    {
      "epoch": 0.060698447893569846,
      "grad_norm": 1.2285375595092773,
      "learning_rate": 9.91019194800746e-06,
      "loss": 0.7051,
      "step": 219
    },
    {
      "epoch": 0.06097560975609756,
      "grad_norm": 1.2412936687469482,
      "learning_rate": 9.909368635064151e-06,
      "loss": 0.7163,
      "step": 220
    },
    {
      "epoch": 0.06125277161862528,
      "grad_norm": 1.2311959266662598,
      "learning_rate": 9.908541599984276e-06,
      "loss": 0.8172,
      "step": 221
    },
    {
      "epoch": 0.061529933481152994,
      "grad_norm": 1.535592794418335,
      "learning_rate": 9.907710843394862e-06,
      "loss": 0.765,
      "step": 222
    },
    {
      "epoch": 0.06180709534368071,
      "grad_norm": 1.3960353136062622,
      "learning_rate": 9.906876365925768e-06,
      "loss": 0.7293,
      "step": 223
    },
    {
      "epoch": 0.06208425720620843,
      "grad_norm": 1.2382304668426514,
      "learning_rate": 9.906038168209668e-06,
      "loss": 0.7264,
      "step": 224
    },
    {
      "epoch": 0.06236141906873614,
      "grad_norm": 1.6238666772842407,
      "learning_rate": 9.905196250882059e-06,
      "loss": 0.7773,
      "step": 225
    },
    {
      "epoch": 0.06263858093126386,
      "grad_norm": 2.825510263442993,
      "learning_rate": 9.904350614581256e-06,
      "loss": 0.7594,
      "step": 226
    },
    {
      "epoch": 0.06291574279379157,
      "grad_norm": 1.3431154489517212,
      "learning_rate": 9.903501259948398e-06,
      "loss": 0.7746,
      "step": 227
    },
    {
      "epoch": 0.06319290465631928,
      "grad_norm": 1.3419393301010132,
      "learning_rate": 9.902648187627439e-06,
      "loss": 0.6972,
      "step": 228
    },
    {
      "epoch": 0.06347006651884701,
      "grad_norm": 1.265593409538269,
      "learning_rate": 9.90179139826515e-06,
      "loss": 0.7664,
      "step": 229
    },
    {
      "epoch": 0.06374722838137473,
      "grad_norm": 1.2219282388687134,
      "learning_rate": 9.90093089251113e-06,
      "loss": 0.7687,
      "step": 230
    },
    {
      "epoch": 0.06402439024390244,
      "grad_norm": 1.345005989074707,
      "learning_rate": 9.90006667101778e-06,
      "loss": 0.766,
      "step": 231
    },
    {
      "epoch": 0.06430155210643015,
      "grad_norm": 1.165500283241272,
      "learning_rate": 9.899198734440335e-06,
      "loss": 0.7159,
      "step": 232
    },
    {
      "epoch": 0.06457871396895787,
      "grad_norm": 1.2363556623458862,
      "learning_rate": 9.898327083436833e-06,
      "loss": 0.7712,
      "step": 233
    },
    {
      "epoch": 0.06485587583148558,
      "grad_norm": 1.2054091691970825,
      "learning_rate": 9.897451718668137e-06,
      "loss": 0.7473,
      "step": 234
    },
    {
      "epoch": 0.06513303769401331,
      "grad_norm": 1.2169853448867798,
      "learning_rate": 9.896572640797922e-06,
      "loss": 0.7333,
      "step": 235
    },
    {
      "epoch": 0.06541019955654102,
      "grad_norm": 1.2373777627944946,
      "learning_rate": 9.895689850492676e-06,
      "loss": 0.7058,
      "step": 236
    },
    {
      "epoch": 0.06568736141906874,
      "grad_norm": 1.4917948246002197,
      "learning_rate": 9.894803348421706e-06,
      "loss": 0.7295,
      "step": 237
    },
    {
      "epoch": 0.06596452328159645,
      "grad_norm": 1.707734227180481,
      "learning_rate": 9.893913135257132e-06,
      "loss": 0.7776,
      "step": 238
    },
    {
      "epoch": 0.06624168514412417,
      "grad_norm": 1.2687124013900757,
      "learning_rate": 9.893019211673886e-06,
      "loss": 0.7463,
      "step": 239
    },
    {
      "epoch": 0.06651884700665188,
      "grad_norm": 1.2293355464935303,
      "learning_rate": 9.892121578349716e-06,
      "loss": 0.7557,
      "step": 240
    },
    {
      "epoch": 0.06679600886917961,
      "grad_norm": 1.2101538181304932,
      "learning_rate": 9.891220235965177e-06,
      "loss": 0.724,
      "step": 241
    },
    {
      "epoch": 0.06707317073170732,
      "grad_norm": 1.158309817314148,
      "learning_rate": 9.890315185203644e-06,
      "loss": 0.7259,
      "step": 242
    },
    {
      "epoch": 0.06735033259423504,
      "grad_norm": 1.2866536378860474,
      "learning_rate": 9.889406426751296e-06,
      "loss": 0.7757,
      "step": 243
    },
    {
      "epoch": 0.06762749445676275,
      "grad_norm": 1.2190892696380615,
      "learning_rate": 9.88849396129713e-06,
      "loss": 0.7604,
      "step": 244
    },
    {
      "epoch": 0.06790465631929046,
      "grad_norm": 1.3062098026275635,
      "learning_rate": 9.887577789532948e-06,
      "loss": 0.8002,
      "step": 245
    },
    {
      "epoch": 0.06818181818181818,
      "grad_norm": 1.2259243726730347,
      "learning_rate": 9.88665791215336e-06,
      "loss": 0.752,
      "step": 246
    },
    {
      "epoch": 0.0684589800443459,
      "grad_norm": 1.237217903137207,
      "learning_rate": 9.885734329855798e-06,
      "loss": 0.7743,
      "step": 247
    },
    {
      "epoch": 0.06873614190687362,
      "grad_norm": 1.2918739318847656,
      "learning_rate": 9.88480704334049e-06,
      "loss": 0.7341,
      "step": 248
    },
    {
      "epoch": 0.06901330376940133,
      "grad_norm": 1.4236030578613281,
      "learning_rate": 9.883876053310475e-06,
      "loss": 0.7386,
      "step": 249
    },
    {
      "epoch": 0.06929046563192905,
      "grad_norm": 1.368618130683899,
      "learning_rate": 9.882941360471605e-06,
      "loss": 0.7669,
      "step": 250
    },
    {
      "epoch": 0.06956762749445676,
      "grad_norm": 1.3190573453903198,
      "learning_rate": 9.882002965532535e-06,
      "loss": 0.7159,
      "step": 251
    },
    {
      "epoch": 0.06984478935698447,
      "grad_norm": 1.1929199695587158,
      "learning_rate": 9.881060869204729e-06,
      "loss": 0.7507,
      "step": 252
    },
    {
      "epoch": 0.0701219512195122,
      "grad_norm": 1.2934563159942627,
      "learning_rate": 9.880115072202454e-06,
      "loss": 0.7345,
      "step": 253
    },
    {
      "epoch": 0.07039911308203992,
      "grad_norm": 3.7547526359558105,
      "learning_rate": 9.879165575242788e-06,
      "loss": 0.7446,
      "step": 254
    },
    {
      "epoch": 0.07067627494456763,
      "grad_norm": 1.263873815536499,
      "learning_rate": 9.878212379045608e-06,
      "loss": 0.7419,
      "step": 255
    },
    {
      "epoch": 0.07095343680709534,
      "grad_norm": 1.195499300956726,
      "learning_rate": 9.877255484333602e-06,
      "loss": 0.769,
      "step": 256
    },
    {
      "epoch": 0.07123059866962306,
      "grad_norm": 1.2025052309036255,
      "learning_rate": 9.876294891832255e-06,
      "loss": 0.7087,
      "step": 257
    },
    {
      "epoch": 0.07150776053215077,
      "grad_norm": 1.2034642696380615,
      "learning_rate": 9.875330602269864e-06,
      "loss": 0.7145,
      "step": 258
    },
    {
      "epoch": 0.07178492239467849,
      "grad_norm": 1.2883434295654297,
      "learning_rate": 9.874362616377522e-06,
      "loss": 0.7147,
      "step": 259
    },
    {
      "epoch": 0.07206208425720621,
      "grad_norm": 1.1575876474380493,
      "learning_rate": 9.873390934889126e-06,
      "loss": 0.728,
      "step": 260
    },
    {
      "epoch": 0.07233924611973393,
      "grad_norm": 1.265578031539917,
      "learning_rate": 9.872415558541379e-06,
      "loss": 0.793,
      "step": 261
    },
    {
      "epoch": 0.07261640798226164,
      "grad_norm": 1.234798550605774,
      "learning_rate": 9.871436488073781e-06,
      "loss": 0.729,
      "step": 262
    },
    {
      "epoch": 0.07289356984478935,
      "grad_norm": 1.176753044128418,
      "learning_rate": 9.870453724228633e-06,
      "loss": 0.7707,
      "step": 263
    },
    {
      "epoch": 0.07317073170731707,
      "grad_norm": 1.302773118019104,
      "learning_rate": 9.869467267751037e-06,
      "loss": 0.7716,
      "step": 264
    },
    {
      "epoch": 0.07344789356984478,
      "grad_norm": 1.252744197845459,
      "learning_rate": 9.868477119388897e-06,
      "loss": 0.7762,
      "step": 265
    },
    {
      "epoch": 0.07372505543237251,
      "grad_norm": 1.1192384958267212,
      "learning_rate": 9.867483279892912e-06,
      "loss": 0.7262,
      "step": 266
    },
    {
      "epoch": 0.07400221729490022,
      "grad_norm": 1.2904949188232422,
      "learning_rate": 9.866485750016579e-06,
      "loss": 0.812,
      "step": 267
    },
    {
      "epoch": 0.07427937915742794,
      "grad_norm": 1.1470303535461426,
      "learning_rate": 9.8654845305162e-06,
      "loss": 0.7192,
      "step": 268
    },
    {
      "epoch": 0.07455654101995565,
      "grad_norm": 1.1994022130966187,
      "learning_rate": 9.86447962215087e-06,
      "loss": 0.7627,
      "step": 269
    },
    {
      "epoch": 0.07483370288248337,
      "grad_norm": 1.1994693279266357,
      "learning_rate": 9.863471025682475e-06,
      "loss": 0.7109,
      "step": 270
    },
    {
      "epoch": 0.07511086474501108,
      "grad_norm": 1.6612850427627563,
      "learning_rate": 9.862458741875707e-06,
      "loss": 0.7281,
      "step": 271
    },
    {
      "epoch": 0.07538802660753881,
      "grad_norm": 1.3092461824417114,
      "learning_rate": 9.861442771498049e-06,
      "loss": 0.7491,
      "step": 272
    },
    {
      "epoch": 0.07566518847006652,
      "grad_norm": 1.1035274267196655,
      "learning_rate": 9.860423115319778e-06,
      "loss": 0.6975,
      "step": 273
    },
    {
      "epoch": 0.07594235033259424,
      "grad_norm": 1.359500765800476,
      "learning_rate": 9.859399774113968e-06,
      "loss": 0.7433,
      "step": 274
    },
    {
      "epoch": 0.07621951219512195,
      "grad_norm": 1.2822551727294922,
      "learning_rate": 9.858372748656485e-06,
      "loss": 0.7632,
      "step": 275
    },
    {
      "epoch": 0.07649667405764966,
      "grad_norm": 1.3419774770736694,
      "learning_rate": 9.85734203972599e-06,
      "loss": 0.7475,
      "step": 276
    },
    {
      "epoch": 0.07677383592017738,
      "grad_norm": 1.1211000680923462,
      "learning_rate": 9.856307648103934e-06,
      "loss": 0.7121,
      "step": 277
    },
    {
      "epoch": 0.0770509977827051,
      "grad_norm": 1.253115177154541,
      "learning_rate": 9.855269574574565e-06,
      "loss": 0.7886,
      "step": 278
    },
    {
      "epoch": 0.07732815964523282,
      "grad_norm": 1.3522508144378662,
      "learning_rate": 9.854227819924913e-06,
      "loss": 0.6935,
      "step": 279
    },
    {
      "epoch": 0.07760532150776053,
      "grad_norm": 1.5343108177185059,
      "learning_rate": 9.853182384944812e-06,
      "loss": 0.7561,
      "step": 280
    },
    {
      "epoch": 0.07788248337028825,
      "grad_norm": 1.2900199890136719,
      "learning_rate": 9.852133270426877e-06,
      "loss": 0.7564,
      "step": 281
    },
    {
      "epoch": 0.07815964523281596,
      "grad_norm": 1.2070568799972534,
      "learning_rate": 9.851080477166515e-06,
      "loss": 0.7619,
      "step": 282
    },
    {
      "epoch": 0.07843680709534367,
      "grad_norm": 1.164781928062439,
      "learning_rate": 9.850024005961922e-06,
      "loss": 0.7731,
      "step": 283
    },
    {
      "epoch": 0.0787139689578714,
      "grad_norm": 1.5192410945892334,
      "learning_rate": 9.848963857614085e-06,
      "loss": 0.77,
      "step": 284
    },
    {
      "epoch": 0.07899113082039912,
      "grad_norm": 1.1506012678146362,
      "learning_rate": 9.847900032926774e-06,
      "loss": 0.7137,
      "step": 285
    },
    {
      "epoch": 0.07926829268292683,
      "grad_norm": 1.2045660018920898,
      "learning_rate": 9.84683253270655e-06,
      "loss": 0.7306,
      "step": 286
    },
    {
      "epoch": 0.07954545454545454,
      "grad_norm": 1.3198407888412476,
      "learning_rate": 9.84576135776276e-06,
      "loss": 0.7575,
      "step": 287
    },
    {
      "epoch": 0.07982261640798226,
      "grad_norm": 1.4521311521530151,
      "learning_rate": 9.844686508907538e-06,
      "loss": 0.8131,
      "step": 288
    },
    {
      "epoch": 0.08009977827050997,
      "grad_norm": 1.2787872552871704,
      "learning_rate": 9.8436079869558e-06,
      "loss": 0.7337,
      "step": 289
    },
    {
      "epoch": 0.0803769401330377,
      "grad_norm": 1.1939911842346191,
      "learning_rate": 9.84252579272525e-06,
      "loss": 0.7729,
      "step": 290
    },
    {
      "epoch": 0.08065410199556541,
      "grad_norm": 1.1444569826126099,
      "learning_rate": 9.841439927036378e-06,
      "loss": 0.7717,
      "step": 291
    },
    {
      "epoch": 0.08093126385809313,
      "grad_norm": 1.1840887069702148,
      "learning_rate": 9.840350390712451e-06,
      "loss": 0.7351,
      "step": 292
    },
    {
      "epoch": 0.08120842572062084,
      "grad_norm": 7.0830912590026855,
      "learning_rate": 9.839257184579524e-06,
      "loss": 0.6865,
      "step": 293
    },
    {
      "epoch": 0.08148558758314856,
      "grad_norm": 3.1618802547454834,
      "learning_rate": 9.838160309466434e-06,
      "loss": 0.7723,
      "step": 294
    },
    {
      "epoch": 0.08176274944567627,
      "grad_norm": 1.2273366451263428,
      "learning_rate": 9.8370597662048e-06,
      "loss": 0.7052,
      "step": 295
    },
    {
      "epoch": 0.082039911308204,
      "grad_norm": 1.191846251487732,
      "learning_rate": 9.835955555629018e-06,
      "loss": 0.7632,
      "step": 296
    },
    {
      "epoch": 0.08231707317073171,
      "grad_norm": 1.1624358892440796,
      "learning_rate": 9.83484767857627e-06,
      "loss": 0.7559,
      "step": 297
    },
    {
      "epoch": 0.08259423503325942,
      "grad_norm": 1.152174711227417,
      "learning_rate": 9.833736135886513e-06,
      "loss": 0.7209,
      "step": 298
    },
    {
      "epoch": 0.08287139689578714,
      "grad_norm": 1.235674262046814,
      "learning_rate": 9.832620928402487e-06,
      "loss": 0.7877,
      "step": 299
    },
    {
      "epoch": 0.08314855875831485,
      "grad_norm": 1.1706044673919678,
      "learning_rate": 9.83150205696971e-06,
      "loss": 0.7103,
      "step": 300
    },
    {
      "epoch": 0.08342572062084257,
      "grad_norm": 2.1092822551727295,
      "learning_rate": 9.830379522436474e-06,
      "loss": 0.7414,
      "step": 301
    },
    {
      "epoch": 0.0837028824833703,
      "grad_norm": 1.2987644672393799,
      "learning_rate": 9.829253325653854e-06,
      "loss": 0.7692,
      "step": 302
    },
    {
      "epoch": 0.08398004434589801,
      "grad_norm": 1.161926507949829,
      "learning_rate": 9.828123467475696e-06,
      "loss": 0.7274,
      "step": 303
    },
    {
      "epoch": 0.08425720620842572,
      "grad_norm": 1.2371591329574585,
      "learning_rate": 9.826989948758627e-06,
      "loss": 0.6701,
      "step": 304
    },
    {
      "epoch": 0.08453436807095344,
      "grad_norm": 1.197625994682312,
      "learning_rate": 9.825852770362046e-06,
      "loss": 0.8141,
      "step": 305
    },
    {
      "epoch": 0.08481152993348115,
      "grad_norm": 1.2106765508651733,
      "learning_rate": 9.824711933148128e-06,
      "loss": 0.7598,
      "step": 306
    },
    {
      "epoch": 0.08508869179600886,
      "grad_norm": 1.1379930973052979,
      "learning_rate": 9.823567437981823e-06,
      "loss": 0.6914,
      "step": 307
    },
    {
      "epoch": 0.08536585365853659,
      "grad_norm": 1.3295775651931763,
      "learning_rate": 9.82241928573085e-06,
      "loss": 0.7449,
      "step": 308
    },
    {
      "epoch": 0.0856430155210643,
      "grad_norm": 1.2388590574264526,
      "learning_rate": 9.821267477265705e-06,
      "loss": 0.7248,
      "step": 309
    },
    {
      "epoch": 0.08592017738359202,
      "grad_norm": 1.2240455150604248,
      "learning_rate": 9.820112013459656e-06,
      "loss": 0.7425,
      "step": 310
    },
    {
      "epoch": 0.08619733924611973,
      "grad_norm": 1.396546721458435,
      "learning_rate": 9.81895289518874e-06,
      "loss": 0.7368,
      "step": 311
    },
    {
      "epoch": 0.08647450110864745,
      "grad_norm": 1.3109747171401978,
      "learning_rate": 9.817790123331767e-06,
      "loss": 0.7823,
      "step": 312
    },
    {
      "epoch": 0.08675166297117516,
      "grad_norm": 1.218022108078003,
      "learning_rate": 9.816623698770314e-06,
      "loss": 0.8283,
      "step": 313
    },
    {
      "epoch": 0.08702882483370289,
      "grad_norm": 1.2392939329147339,
      "learning_rate": 9.815453622388728e-06,
      "loss": 0.699,
      "step": 314
    },
    {
      "epoch": 0.0873059866962306,
      "grad_norm": 1.2485021352767944,
      "learning_rate": 9.81427989507413e-06,
      "loss": 0.7008,
      "step": 315
    },
    {
      "epoch": 0.08758314855875832,
      "grad_norm": 1.2169044017791748,
      "learning_rate": 9.813102517716403e-06,
      "loss": 0.8102,
      "step": 316
    },
    {
      "epoch": 0.08786031042128603,
      "grad_norm": 1.4195579290390015,
      "learning_rate": 9.8119214912082e-06,
      "loss": 0.791,
      "step": 317
    },
    {
      "epoch": 0.08813747228381374,
      "grad_norm": 1.236379861831665,
      "learning_rate": 9.810736816444938e-06,
      "loss": 0.7414,
      "step": 318
    },
    {
      "epoch": 0.08841463414634146,
      "grad_norm": 1.2336901426315308,
      "learning_rate": 9.809548494324806e-06,
      "loss": 0.696,
      "step": 319
    },
    {
      "epoch": 0.08869179600886919,
      "grad_norm": 1.827335238456726,
      "learning_rate": 9.808356525748748e-06,
      "loss": 0.7106,
      "step": 320
    },
    {
      "epoch": 0.0889689578713969,
      "grad_norm": 1.2580500841140747,
      "learning_rate": 9.807160911620484e-06,
      "loss": 0.7379,
      "step": 321
    },
    {
      "epoch": 0.08924611973392461,
      "grad_norm": 1.2233524322509766,
      "learning_rate": 9.805961652846493e-06,
      "loss": 0.707,
      "step": 322
    },
    {
      "epoch": 0.08952328159645233,
      "grad_norm": 1.3095909357070923,
      "learning_rate": 9.804758750336015e-06,
      "loss": 0.7745,
      "step": 323
    },
    {
      "epoch": 0.08980044345898004,
      "grad_norm": 1.2433184385299683,
      "learning_rate": 9.803552205001054e-06,
      "loss": 0.7107,
      "step": 324
    },
    {
      "epoch": 0.09007760532150776,
      "grad_norm": 1.164352536201477,
      "learning_rate": 9.80234201775638e-06,
      "loss": 0.7869,
      "step": 325
    },
    {
      "epoch": 0.09035476718403547,
      "grad_norm": 1.1965830326080322,
      "learning_rate": 9.801128189519517e-06,
      "loss": 0.723,
      "step": 326
    },
    {
      "epoch": 0.0906319290465632,
      "grad_norm": 1.2782299518585205,
      "learning_rate": 9.799910721210755e-06,
      "loss": 0.7556,
      "step": 327
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 1.132859706878662,
      "learning_rate": 9.798689613753143e-06,
      "loss": 0.71,
      "step": 328
    },
    {
      "epoch": 0.09118625277161863,
      "grad_norm": 1.2156035900115967,
      "learning_rate": 9.797464868072489e-06,
      "loss": 0.7612,
      "step": 329
    },
    {
      "epoch": 0.09146341463414634,
      "grad_norm": 1.2431600093841553,
      "learning_rate": 9.796236485097354e-06,
      "loss": 0.7057,
      "step": 330
    },
    {
      "epoch": 0.09174057649667405,
      "grad_norm": 1.1446610689163208,
      "learning_rate": 9.795004465759067e-06,
      "loss": 0.7327,
      "step": 331
    },
    {
      "epoch": 0.09201773835920177,
      "grad_norm": 1.2144614458084106,
      "learning_rate": 9.793768810991702e-06,
      "loss": 0.7349,
      "step": 332
    },
    {
      "epoch": 0.0922949002217295,
      "grad_norm": 1.3191543817520142,
      "learning_rate": 9.792529521732101e-06,
      "loss": 0.7242,
      "step": 333
    },
    {
      "epoch": 0.09257206208425721,
      "grad_norm": 1.2725412845611572,
      "learning_rate": 9.791286598919855e-06,
      "loss": 0.7163,
      "step": 334
    },
    {
      "epoch": 0.09284922394678492,
      "grad_norm": 1.208930492401123,
      "learning_rate": 9.790040043497309e-06,
      "loss": 0.7272,
      "step": 335
    },
    {
      "epoch": 0.09312638580931264,
      "grad_norm": 1.2537230253219604,
      "learning_rate": 9.788789856409565e-06,
      "loss": 0.7645,
      "step": 336
    },
    {
      "epoch": 0.09340354767184035,
      "grad_norm": 1.2034469842910767,
      "learning_rate": 9.787536038604478e-06,
      "loss": 0.7399,
      "step": 337
    },
    {
      "epoch": 0.09368070953436806,
      "grad_norm": 1.240942358970642,
      "learning_rate": 9.786278591032653e-06,
      "loss": 0.7709,
      "step": 338
    },
    {
      "epoch": 0.09395787139689579,
      "grad_norm": 1.3362568616867065,
      "learning_rate": 9.785017514647453e-06,
      "loss": 0.7528,
      "step": 339
    },
    {
      "epoch": 0.0942350332594235,
      "grad_norm": 1.2375695705413818,
      "learning_rate": 9.783752810404984e-06,
      "loss": 0.8134,
      "step": 340
    },
    {
      "epoch": 0.09451219512195122,
      "grad_norm": 1.2192111015319824,
      "learning_rate": 9.782484479264111e-06,
      "loss": 0.8022,
      "step": 341
    },
    {
      "epoch": 0.09478935698447893,
      "grad_norm": 1.1349451541900635,
      "learning_rate": 9.781212522186442e-06,
      "loss": 0.7585,
      "step": 342
    },
    {
      "epoch": 0.09506651884700665,
      "grad_norm": 1.1353130340576172,
      "learning_rate": 9.779936940136338e-06,
      "loss": 0.6721,
      "step": 343
    },
    {
      "epoch": 0.09534368070953436,
      "grad_norm": 1.3071366548538208,
      "learning_rate": 9.778657734080908e-06,
      "loss": 0.7855,
      "step": 344
    },
    {
      "epoch": 0.09562084257206209,
      "grad_norm": 1.1780011653900146,
      "learning_rate": 9.777374904990007e-06,
      "loss": 0.7376,
      "step": 345
    },
    {
      "epoch": 0.0958980044345898,
      "grad_norm": 1.1800563335418701,
      "learning_rate": 9.776088453836236e-06,
      "loss": 0.7238,
      "step": 346
    },
    {
      "epoch": 0.09617516629711752,
      "grad_norm": 1.0810941457748413,
      "learning_rate": 9.774798381594946e-06,
      "loss": 0.7009,
      "step": 347
    },
    {
      "epoch": 0.09645232815964523,
      "grad_norm": 1.3248358964920044,
      "learning_rate": 9.773504689244231e-06,
      "loss": 0.7865,
      "step": 348
    },
    {
      "epoch": 0.09672949002217295,
      "grad_norm": 1.153160810470581,
      "learning_rate": 9.772207377764928e-06,
      "loss": 0.7981,
      "step": 349
    },
    {
      "epoch": 0.09700665188470066,
      "grad_norm": 1.2143999338150024,
      "learning_rate": 9.770906448140622e-06,
      "loss": 0.7096,
      "step": 350
    },
    {
      "epoch": 0.09728381374722839,
      "grad_norm": 1.1204257011413574,
      "learning_rate": 9.769601901357639e-06,
      "loss": 0.7973,
      "step": 351
    },
    {
      "epoch": 0.0975609756097561,
      "grad_norm": 1.1486655473709106,
      "learning_rate": 9.768293738405046e-06,
      "loss": 0.6948,
      "step": 352
    },
    {
      "epoch": 0.09783813747228381,
      "grad_norm": 1.1340482234954834,
      "learning_rate": 9.766981960274653e-06,
      "loss": 0.708,
      "step": 353
    },
    {
      "epoch": 0.09811529933481153,
      "grad_norm": 1.053807020187378,
      "learning_rate": 9.765666567961011e-06,
      "loss": 0.6948,
      "step": 354
    },
    {
      "epoch": 0.09839246119733924,
      "grad_norm": 1.3398793935775757,
      "learning_rate": 9.764347562461413e-06,
      "loss": 0.7251,
      "step": 355
    },
    {
      "epoch": 0.09866962305986696,
      "grad_norm": 1.2038933038711548,
      "learning_rate": 9.76302494477589e-06,
      "loss": 0.6702,
      "step": 356
    },
    {
      "epoch": 0.09894678492239468,
      "grad_norm": 1.1128658056259155,
      "learning_rate": 9.761698715907207e-06,
      "loss": 0.6957,
      "step": 357
    },
    {
      "epoch": 0.0992239467849224,
      "grad_norm": 1.0897446870803833,
      "learning_rate": 9.760368876860873e-06,
      "loss": 0.7213,
      "step": 358
    },
    {
      "epoch": 0.09950110864745011,
      "grad_norm": 1.3016173839569092,
      "learning_rate": 9.759035428645134e-06,
      "loss": 0.7986,
      "step": 359
    },
    {
      "epoch": 0.09977827050997783,
      "grad_norm": 1.1744967699050903,
      "learning_rate": 9.75769837227097e-06,
      "loss": 0.78,
      "step": 360
    },
    {
      "epoch": 0.10005543237250554,
      "grad_norm": 1.2591989040374756,
      "learning_rate": 9.756357708752096e-06,
      "loss": 0.7298,
      "step": 361
    },
    {
      "epoch": 0.10033259423503325,
      "grad_norm": 3.5684590339660645,
      "learning_rate": 9.755013439104963e-06,
      "loss": 0.7822,
      "step": 362
    },
    {
      "epoch": 0.10060975609756098,
      "grad_norm": 1.234763741493225,
      "learning_rate": 9.753665564348758e-06,
      "loss": 0.7252,
      "step": 363
    },
    {
      "epoch": 0.1008869179600887,
      "grad_norm": 1.1623377799987793,
      "learning_rate": 9.752314085505396e-06,
      "loss": 0.7005,
      "step": 364
    },
    {
      "epoch": 0.10116407982261641,
      "grad_norm": 1.2115342617034912,
      "learning_rate": 9.75095900359953e-06,
      "loss": 0.7076,
      "step": 365
    },
    {
      "epoch": 0.10144124168514412,
      "grad_norm": 1.1660770177841187,
      "learning_rate": 9.749600319658542e-06,
      "loss": 0.697,
      "step": 366
    },
    {
      "epoch": 0.10171840354767184,
      "grad_norm": 1.1513139009475708,
      "learning_rate": 9.748238034712548e-06,
      "loss": 0.6666,
      "step": 367
    },
    {
      "epoch": 0.10199556541019955,
      "grad_norm": 12.983518600463867,
      "learning_rate": 9.746872149794388e-06,
      "loss": 0.7807,
      "step": 368
    },
    {
      "epoch": 0.10227272727272728,
      "grad_norm": 1.3617948293685913,
      "learning_rate": 9.74550266593964e-06,
      "loss": 0.7483,
      "step": 369
    },
    {
      "epoch": 0.10254988913525499,
      "grad_norm": 1.164811372756958,
      "learning_rate": 9.744129584186599e-06,
      "loss": 0.6963,
      "step": 370
    },
    {
      "epoch": 0.1028270509977827,
      "grad_norm": 1.237431526184082,
      "learning_rate": 9.7427529055763e-06,
      "loss": 0.7501,
      "step": 371
    },
    {
      "epoch": 0.10310421286031042,
      "grad_norm": 1.3964523077011108,
      "learning_rate": 9.741372631152497e-06,
      "loss": 0.6944,
      "step": 372
    },
    {
      "epoch": 0.10338137472283813,
      "grad_norm": 1.1237542629241943,
      "learning_rate": 9.739988761961673e-06,
      "loss": 0.7605,
      "step": 373
    },
    {
      "epoch": 0.10365853658536585,
      "grad_norm": 1.1669179201126099,
      "learning_rate": 9.738601299053037e-06,
      "loss": 0.7273,
      "step": 374
    },
    {
      "epoch": 0.10393569844789358,
      "grad_norm": 1.188980221748352,
      "learning_rate": 9.737210243478522e-06,
      "loss": 0.6782,
      "step": 375
    },
    {
      "epoch": 0.10421286031042129,
      "grad_norm": 1.2021273374557495,
      "learning_rate": 9.735815596292784e-06,
      "loss": 0.7128,
      "step": 376
    },
    {
      "epoch": 0.104490022172949,
      "grad_norm": 1.4315763711929321,
      "learning_rate": 9.734417358553205e-06,
      "loss": 0.7284,
      "step": 377
    },
    {
      "epoch": 0.10476718403547672,
      "grad_norm": 1.1424318552017212,
      "learning_rate": 9.733015531319882e-06,
      "loss": 0.7533,
      "step": 378
    },
    {
      "epoch": 0.10504434589800443,
      "grad_norm": 1.087956428527832,
      "learning_rate": 9.731610115655645e-06,
      "loss": 0.645,
      "step": 379
    },
    {
      "epoch": 0.10532150776053215,
      "grad_norm": 1.249829888343811,
      "learning_rate": 9.730201112626034e-06,
      "loss": 0.6935,
      "step": 380
    },
    {
      "epoch": 0.10559866962305987,
      "grad_norm": 1.2351734638214111,
      "learning_rate": 9.728788523299313e-06,
      "loss": 0.7192,
      "step": 381
    },
    {
      "epoch": 0.10587583148558759,
      "grad_norm": 1.158217191696167,
      "learning_rate": 9.727372348746466e-06,
      "loss": 0.706,
      "step": 382
    },
    {
      "epoch": 0.1061529933481153,
      "grad_norm": 1.254313349723816,
      "learning_rate": 9.725952590041196e-06,
      "loss": 0.7521,
      "step": 383
    },
    {
      "epoch": 0.10643015521064302,
      "grad_norm": 1.1844993829727173,
      "learning_rate": 9.724529248259918e-06,
      "loss": 0.7395,
      "step": 384
    },
    {
      "epoch": 0.10670731707317073,
      "grad_norm": 1.0911539793014526,
      "learning_rate": 9.72310232448177e-06,
      "loss": 0.6577,
      "step": 385
    },
    {
      "epoch": 0.10698447893569844,
      "grad_norm": 1.32505202293396,
      "learning_rate": 9.721671819788603e-06,
      "loss": 0.7036,
      "step": 386
    },
    {
      "epoch": 0.10726164079822616,
      "grad_norm": 1.0777511596679688,
      "learning_rate": 9.72023773526498e-06,
      "loss": 0.7571,
      "step": 387
    },
    {
      "epoch": 0.10753880266075388,
      "grad_norm": 1.2890657186508179,
      "learning_rate": 9.718800071998183e-06,
      "loss": 0.7183,
      "step": 388
    },
    {
      "epoch": 0.1078159645232816,
      "grad_norm": 1.269568920135498,
      "learning_rate": 9.717358831078205e-06,
      "loss": 0.6614,
      "step": 389
    },
    {
      "epoch": 0.10809312638580931,
      "grad_norm": 1.154547095298767,
      "learning_rate": 9.715914013597753e-06,
      "loss": 0.7073,
      "step": 390
    },
    {
      "epoch": 0.10837028824833703,
      "grad_norm": 1.335741400718689,
      "learning_rate": 9.714465620652244e-06,
      "loss": 0.7995,
      "step": 391
    },
    {
      "epoch": 0.10864745011086474,
      "grad_norm": 1.6507827043533325,
      "learning_rate": 9.713013653339805e-06,
      "loss": 0.7166,
      "step": 392
    },
    {
      "epoch": 0.10892461197339245,
      "grad_norm": 1.1748288869857788,
      "learning_rate": 9.711558112761275e-06,
      "loss": 0.7514,
      "step": 393
    },
    {
      "epoch": 0.10920177383592018,
      "grad_norm": 1.1637909412384033,
      "learning_rate": 9.7100990000202e-06,
      "loss": 0.7125,
      "step": 394
    },
    {
      "epoch": 0.1094789356984479,
      "grad_norm": 1.2766393423080444,
      "learning_rate": 9.708636316222838e-06,
      "loss": 0.6723,
      "step": 395
    },
    {
      "epoch": 0.10975609756097561,
      "grad_norm": 1.126134991645813,
      "learning_rate": 9.70717006247815e-06,
      "loss": 0.72,
      "step": 396
    },
    {
      "epoch": 0.11003325942350332,
      "grad_norm": 1.191482424736023,
      "learning_rate": 9.705700239897809e-06,
      "loss": 0.6915,
      "step": 397
    },
    {
      "epoch": 0.11031042128603104,
      "grad_norm": 1.2467244863510132,
      "learning_rate": 9.704226849596186e-06,
      "loss": 0.6197,
      "step": 398
    },
    {
      "epoch": 0.11058758314855875,
      "grad_norm": 1.1040388345718384,
      "learning_rate": 9.702749892690366e-06,
      "loss": 0.7198,
      "step": 399
    },
    {
      "epoch": 0.11086474501108648,
      "grad_norm": 1.3388793468475342,
      "learning_rate": 9.701269370300131e-06,
      "loss": 0.6769,
      "step": 400
    },
    {
      "epoch": 0.1111419068736142,
      "grad_norm": 1.243288278579712,
      "learning_rate": 9.69978528354797e-06,
      "loss": 0.7332,
      "step": 401
    },
    {
      "epoch": 0.11141906873614191,
      "grad_norm": 1.1603100299835205,
      "learning_rate": 9.698297633559069e-06,
      "loss": 0.7237,
      "step": 402
    },
    {
      "epoch": 0.11169623059866962,
      "grad_norm": 1.2112748622894287,
      "learning_rate": 9.696806421461326e-06,
      "loss": 0.787,
      "step": 403
    },
    {
      "epoch": 0.11197339246119734,
      "grad_norm": 1.213304042816162,
      "learning_rate": 9.69531164838533e-06,
      "loss": 0.6571,
      "step": 404
    },
    {
      "epoch": 0.11225055432372505,
      "grad_norm": 1.1718356609344482,
      "learning_rate": 9.693813315464376e-06,
      "loss": 0.7083,
      "step": 405
    },
    {
      "epoch": 0.11252771618625278,
      "grad_norm": 1.1287505626678467,
      "learning_rate": 9.69231142383445e-06,
      "loss": 0.727,
      "step": 406
    },
    {
      "epoch": 0.11280487804878049,
      "grad_norm": 1.1594974994659424,
      "learning_rate": 9.690805974634247e-06,
      "loss": 0.7417,
      "step": 407
    },
    {
      "epoch": 0.1130820399113082,
      "grad_norm": 1.306227445602417,
      "learning_rate": 9.689296969005151e-06,
      "loss": 0.7338,
      "step": 408
    },
    {
      "epoch": 0.11335920177383592,
      "grad_norm": 1.1863619089126587,
      "learning_rate": 9.687784408091243e-06,
      "loss": 0.7127,
      "step": 409
    },
    {
      "epoch": 0.11363636363636363,
      "grad_norm": 1.0896873474121094,
      "learning_rate": 9.686268293039307e-06,
      "loss": 0.7633,
      "step": 410
    },
    {
      "epoch": 0.11391352549889135,
      "grad_norm": 3.0236032009124756,
      "learning_rate": 9.68474862499881e-06,
      "loss": 0.7303,
      "step": 411
    },
    {
      "epoch": 0.11419068736141907,
      "grad_norm": 1.412406086921692,
      "learning_rate": 9.683225405121922e-06,
      "loss": 0.8132,
      "step": 412
    },
    {
      "epoch": 0.11446784922394679,
      "grad_norm": 1.470916748046875,
      "learning_rate": 9.681698634563503e-06,
      "loss": 0.706,
      "step": 413
    },
    {
      "epoch": 0.1147450110864745,
      "grad_norm": 1.1504226922988892,
      "learning_rate": 9.680168314481104e-06,
      "loss": 0.6326,
      "step": 414
    },
    {
      "epoch": 0.11502217294900222,
      "grad_norm": 1.176234483718872,
      "learning_rate": 9.678634446034969e-06,
      "loss": 0.6707,
      "step": 415
    },
    {
      "epoch": 0.11529933481152993,
      "grad_norm": 1.292615532875061,
      "learning_rate": 9.677097030388028e-06,
      "loss": 0.7339,
      "step": 416
    },
    {
      "epoch": 0.11557649667405764,
      "grad_norm": 1.2269580364227295,
      "learning_rate": 9.675556068705908e-06,
      "loss": 0.7643,
      "step": 417
    },
    {
      "epoch": 0.11585365853658537,
      "grad_norm": 1.327617883682251,
      "learning_rate": 9.674011562156918e-06,
      "loss": 0.7274,
      "step": 418
    },
    {
      "epoch": 0.11613082039911309,
      "grad_norm": 1.2873212099075317,
      "learning_rate": 9.672463511912056e-06,
      "loss": 0.702,
      "step": 419
    },
    {
      "epoch": 0.1164079822616408,
      "grad_norm": 1.1502008438110352,
      "learning_rate": 9.670911919145007e-06,
      "loss": 0.7004,
      "step": 420
    },
    {
      "epoch": 0.11668514412416851,
      "grad_norm": 1.1371418237686157,
      "learning_rate": 9.669356785032147e-06,
      "loss": 0.6961,
      "step": 421
    },
    {
      "epoch": 0.11696230598669623,
      "grad_norm": 1.2206065654754639,
      "learning_rate": 9.667798110752526e-06,
      "loss": 0.7493,
      "step": 422
    },
    {
      "epoch": 0.11723946784922394,
      "grad_norm": 1.353977918624878,
      "learning_rate": 9.666235897487888e-06,
      "loss": 0.6722,
      "step": 423
    },
    {
      "epoch": 0.11751662971175167,
      "grad_norm": 1.2804518938064575,
      "learning_rate": 9.664670146422654e-06,
      "loss": 0.7103,
      "step": 424
    },
    {
      "epoch": 0.11779379157427938,
      "grad_norm": 1.1250946521759033,
      "learning_rate": 9.663100858743932e-06,
      "loss": 0.6533,
      "step": 425
    },
    {
      "epoch": 0.1180709534368071,
      "grad_norm": 1.2152454853057861,
      "learning_rate": 9.661528035641507e-06,
      "loss": 0.6786,
      "step": 426
    },
    {
      "epoch": 0.11834811529933481,
      "grad_norm": 1.1647899150848389,
      "learning_rate": 9.659951678307848e-06,
      "loss": 0.7406,
      "step": 427
    },
    {
      "epoch": 0.11862527716186252,
      "grad_norm": 1.2126303911209106,
      "learning_rate": 9.658371787938103e-06,
      "loss": 0.7241,
      "step": 428
    },
    {
      "epoch": 0.11890243902439024,
      "grad_norm": 1.1051273345947266,
      "learning_rate": 9.656788365730095e-06,
      "loss": 0.6904,
      "step": 429
    },
    {
      "epoch": 0.11917960088691797,
      "grad_norm": 1.0918411016464233,
      "learning_rate": 9.655201412884328e-06,
      "loss": 0.7614,
      "step": 430
    },
    {
      "epoch": 0.11945676274944568,
      "grad_norm": 1.1484405994415283,
      "learning_rate": 9.653610930603983e-06,
      "loss": 0.699,
      "step": 431
    },
    {
      "epoch": 0.1197339246119734,
      "grad_norm": 1.233140230178833,
      "learning_rate": 9.652016920094915e-06,
      "loss": 0.73,
      "step": 432
    },
    {
      "epoch": 0.12001108647450111,
      "grad_norm": 1.0900344848632812,
      "learning_rate": 9.650419382565658e-06,
      "loss": 0.6499,
      "step": 433
    },
    {
      "epoch": 0.12028824833702882,
      "grad_norm": 2.5382466316223145,
      "learning_rate": 9.648818319227413e-06,
      "loss": 0.7346,
      "step": 434
    },
    {
      "epoch": 0.12056541019955654,
      "grad_norm": 1.1844030618667603,
      "learning_rate": 9.647213731294062e-06,
      "loss": 0.7167,
      "step": 435
    },
    {
      "epoch": 0.12084257206208426,
      "grad_norm": 1.0956852436065674,
      "learning_rate": 9.645605619982154e-06,
      "loss": 0.6819,
      "step": 436
    },
    {
      "epoch": 0.12111973392461198,
      "grad_norm": 1.1735581159591675,
      "learning_rate": 9.643993986510912e-06,
      "loss": 0.7625,
      "step": 437
    },
    {
      "epoch": 0.12139689578713969,
      "grad_norm": 1.2941609621047974,
      "learning_rate": 9.642378832102226e-06,
      "loss": 0.6953,
      "step": 438
    },
    {
      "epoch": 0.1216740576496674,
      "grad_norm": 1.1904845237731934,
      "learning_rate": 9.64076015798066e-06,
      "loss": 0.6778,
      "step": 439
    },
    {
      "epoch": 0.12195121951219512,
      "grad_norm": 1.121621012687683,
      "learning_rate": 9.639137965373443e-06,
      "loss": 0.7328,
      "step": 440
    },
    {
      "epoch": 0.12222838137472283,
      "grad_norm": 1.1233983039855957,
      "learning_rate": 9.637512255510475e-06,
      "loss": 0.6914,
      "step": 441
    },
    {
      "epoch": 0.12250554323725056,
      "grad_norm": 1.1504545211791992,
      "learning_rate": 9.635883029624317e-06,
      "loss": 0.7982,
      "step": 442
    },
    {
      "epoch": 0.12278270509977827,
      "grad_norm": 1.1533653736114502,
      "learning_rate": 9.634250288950203e-06,
      "loss": 0.6754,
      "step": 443
    },
    {
      "epoch": 0.12305986696230599,
      "grad_norm": 1.0936402082443237,
      "learning_rate": 9.632614034726026e-06,
      "loss": 0.6604,
      "step": 444
    },
    {
      "epoch": 0.1233370288248337,
      "grad_norm": 1.3587909936904907,
      "learning_rate": 9.630974268192346e-06,
      "loss": 0.7246,
      "step": 445
    },
    {
      "epoch": 0.12361419068736142,
      "grad_norm": 1.188857913017273,
      "learning_rate": 9.629330990592385e-06,
      "loss": 0.7334,
      "step": 446
    },
    {
      "epoch": 0.12389135254988913,
      "grad_norm": 1.1382105350494385,
      "learning_rate": 9.627684203172027e-06,
      "loss": 0.762,
      "step": 447
    },
    {
      "epoch": 0.12416851441241686,
      "grad_norm": 1.165677547454834,
      "learning_rate": 9.626033907179818e-06,
      "loss": 0.7207,
      "step": 448
    },
    {
      "epoch": 0.12444567627494457,
      "grad_norm": 1.042955756187439,
      "learning_rate": 9.624380103866959e-06,
      "loss": 0.6692,
      "step": 449
    },
    {
      "epoch": 0.12472283813747229,
      "grad_norm": 1.2074801921844482,
      "learning_rate": 9.622722794487318e-06,
      "loss": 0.769,
      "step": 450
    },
    {
      "epoch": 0.125,
      "grad_norm": 1.1248900890350342,
      "learning_rate": 9.621061980297417e-06,
      "loss": 0.7098,
      "step": 451
    },
    {
      "epoch": 0.12527716186252771,
      "grad_norm": 1.1106007099151611,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.7364,
      "step": 452
    },
    {
      "epoch": 0.12555432372505543,
      "grad_norm": 1.0914770364761353,
      "learning_rate": 9.617729842526207e-06,
      "loss": 0.7012,
      "step": 453
    },
    {
      "epoch": 0.12583148558758314,
      "grad_norm": 1.102309226989746,
      "learning_rate": 9.616058521471227e-06,
      "loss": 0.6778,
      "step": 454
    },
    {
      "epoch": 0.12610864745011086,
      "grad_norm": 1.760117530822754,
      "learning_rate": 9.614383700658638e-06,
      "loss": 0.6929,
      "step": 455
    },
    {
      "epoch": 0.12638580931263857,
      "grad_norm": 1.5223033428192139,
      "learning_rate": 9.612705381358244e-06,
      "loss": 0.7112,
      "step": 456
    },
    {
      "epoch": 0.12666297117516628,
      "grad_norm": 1.1166366338729858,
      "learning_rate": 9.611023564842487e-06,
      "loss": 0.7332,
      "step": 457
    },
    {
      "epoch": 0.12694013303769403,
      "grad_norm": 1.088150978088379,
      "learning_rate": 9.609338252386479e-06,
      "loss": 0.7002,
      "step": 458
    },
    {
      "epoch": 0.12721729490022174,
      "grad_norm": 1.2250138521194458,
      "learning_rate": 9.60764944526797e-06,
      "loss": 0.7563,
      "step": 459
    },
    {
      "epoch": 0.12749445676274945,
      "grad_norm": 1.1011302471160889,
      "learning_rate": 9.605957144767363e-06,
      "loss": 0.6837,
      "step": 460
    },
    {
      "epoch": 0.12777161862527717,
      "grad_norm": 1.4382882118225098,
      "learning_rate": 9.60426135216771e-06,
      "loss": 0.6844,
      "step": 461
    },
    {
      "epoch": 0.12804878048780488,
      "grad_norm": 1.1931488513946533,
      "learning_rate": 9.60256206875471e-06,
      "loss": 0.7292,
      "step": 462
    },
    {
      "epoch": 0.1283259423503326,
      "grad_norm": 1.0783765316009521,
      "learning_rate": 9.600859295816708e-06,
      "loss": 0.7313,
      "step": 463
    },
    {
      "epoch": 0.1286031042128603,
      "grad_norm": 1.3118054866790771,
      "learning_rate": 9.599153034644699e-06,
      "loss": 0.725,
      "step": 464
    },
    {
      "epoch": 0.12888026607538802,
      "grad_norm": 1.1242176294326782,
      "learning_rate": 9.597443286532318e-06,
      "loss": 0.7457,
      "step": 465
    },
    {
      "epoch": 0.12915742793791574,
      "grad_norm": 1.178567886352539,
      "learning_rate": 9.595730052775842e-06,
      "loss": 0.6913,
      "step": 466
    },
    {
      "epoch": 0.12943458980044345,
      "grad_norm": 1.1237245798110962,
      "learning_rate": 9.594013334674195e-06,
      "loss": 0.6469,
      "step": 467
    },
    {
      "epoch": 0.12971175166297116,
      "grad_norm": 1.2146196365356445,
      "learning_rate": 9.592293133528942e-06,
      "loss": 0.7388,
      "step": 468
    },
    {
      "epoch": 0.12998891352549888,
      "grad_norm": 1.2083908319473267,
      "learning_rate": 9.590569450644289e-06,
      "loss": 0.7169,
      "step": 469
    },
    {
      "epoch": 0.13026607538802662,
      "grad_norm": 1.068419337272644,
      "learning_rate": 9.588842287327081e-06,
      "loss": 0.723,
      "step": 470
    },
    {
      "epoch": 0.13054323725055433,
      "grad_norm": 1.145043969154358,
      "learning_rate": 9.587111644886798e-06,
      "loss": 0.7134,
      "step": 471
    },
    {
      "epoch": 0.13082039911308205,
      "grad_norm": 1.1432920694351196,
      "learning_rate": 9.585377524635563e-06,
      "loss": 0.7131,
      "step": 472
    },
    {
      "epoch": 0.13109756097560976,
      "grad_norm": 1.1221081018447876,
      "learning_rate": 9.583639927888138e-06,
      "loss": 0.685,
      "step": 473
    },
    {
      "epoch": 0.13137472283813748,
      "grad_norm": 1.3001903295516968,
      "learning_rate": 9.581898855961911e-06,
      "loss": 0.6823,
      "step": 474
    },
    {
      "epoch": 0.1316518847006652,
      "grad_norm": 1.4029662609100342,
      "learning_rate": 9.580154310176917e-06,
      "loss": 0.7338,
      "step": 475
    },
    {
      "epoch": 0.1319290465631929,
      "grad_norm": 1.1707890033721924,
      "learning_rate": 9.578406291855811e-06,
      "loss": 0.6858,
      "step": 476
    },
    {
      "epoch": 0.13220620842572062,
      "grad_norm": 1.1698969602584839,
      "learning_rate": 9.576654802323896e-06,
      "loss": 0.7021,
      "step": 477
    },
    {
      "epoch": 0.13248337028824833,
      "grad_norm": 1.128864049911499,
      "learning_rate": 9.574899842909094e-06,
      "loss": 0.7426,
      "step": 478
    },
    {
      "epoch": 0.13276053215077604,
      "grad_norm": 1.5543546676635742,
      "learning_rate": 9.573141414941963e-06,
      "loss": 0.7066,
      "step": 479
    },
    {
      "epoch": 0.13303769401330376,
      "grad_norm": 1.2824289798736572,
      "learning_rate": 9.571379519755692e-06,
      "loss": 0.6735,
      "step": 480
    },
    {
      "epoch": 0.13331485587583147,
      "grad_norm": 1.1312131881713867,
      "learning_rate": 9.569614158686097e-06,
      "loss": 0.6654,
      "step": 481
    },
    {
      "epoch": 0.13359201773835921,
      "grad_norm": 1.2285507917404175,
      "learning_rate": 9.567845333071622e-06,
      "loss": 0.7183,
      "step": 482
    },
    {
      "epoch": 0.13386917960088693,
      "grad_norm": 1.3707958459854126,
      "learning_rate": 9.566073044253337e-06,
      "loss": 0.7114,
      "step": 483
    },
    {
      "epoch": 0.13414634146341464,
      "grad_norm": 1.1595308780670166,
      "learning_rate": 9.564297293574941e-06,
      "loss": 0.7013,
      "step": 484
    },
    {
      "epoch": 0.13442350332594236,
      "grad_norm": 1.1979457139968872,
      "learning_rate": 9.562518082382751e-06,
      "loss": 0.7201,
      "step": 485
    },
    {
      "epoch": 0.13470066518847007,
      "grad_norm": 1.2188159227371216,
      "learning_rate": 9.560735412025715e-06,
      "loss": 0.7087,
      "step": 486
    },
    {
      "epoch": 0.13497782705099778,
      "grad_norm": 1.1252472400665283,
      "learning_rate": 9.5589492838554e-06,
      "loss": 0.6717,
      "step": 487
    },
    {
      "epoch": 0.1352549889135255,
      "grad_norm": 1.151863932609558,
      "learning_rate": 9.557159699225991e-06,
      "loss": 0.6442,
      "step": 488
    },
    {
      "epoch": 0.1355321507760532,
      "grad_norm": 1.2114776372909546,
      "learning_rate": 9.555366659494303e-06,
      "loss": 0.6611,
      "step": 489
    },
    {
      "epoch": 0.13580931263858093,
      "grad_norm": 1.130232810974121,
      "learning_rate": 9.553570166019763e-06,
      "loss": 0.72,
      "step": 490
    },
    {
      "epoch": 0.13608647450110864,
      "grad_norm": 1.4085006713867188,
      "learning_rate": 9.551770220164417e-06,
      "loss": 0.7176,
      "step": 491
    },
    {
      "epoch": 0.13636363636363635,
      "grad_norm": 1.1805462837219238,
      "learning_rate": 9.549966823292935e-06,
      "loss": 0.6726,
      "step": 492
    },
    {
      "epoch": 0.13664079822616407,
      "grad_norm": 1.147813320159912,
      "learning_rate": 9.548159976772593e-06,
      "loss": 0.7625,
      "step": 493
    },
    {
      "epoch": 0.1369179600886918,
      "grad_norm": 1.2201555967330933,
      "learning_rate": 9.54634968197329e-06,
      "loss": 0.6989,
      "step": 494
    },
    {
      "epoch": 0.13719512195121952,
      "grad_norm": 1.1137393712997437,
      "learning_rate": 9.544535940267538e-06,
      "loss": 0.6884,
      "step": 495
    },
    {
      "epoch": 0.13747228381374724,
      "grad_norm": 1.0494108200073242,
      "learning_rate": 9.542718753030463e-06,
      "loss": 0.7249,
      "step": 496
    },
    {
      "epoch": 0.13774944567627495,
      "grad_norm": 1.1924444437026978,
      "learning_rate": 9.5408981216398e-06,
      "loss": 0.7259,
      "step": 497
    },
    {
      "epoch": 0.13802660753880266,
      "grad_norm": 1.076897144317627,
      "learning_rate": 9.539074047475897e-06,
      "loss": 0.7124,
      "step": 498
    },
    {
      "epoch": 0.13830376940133038,
      "grad_norm": 1.1206849813461304,
      "learning_rate": 9.537246531921715e-06,
      "loss": 0.7524,
      "step": 499
    },
    {
      "epoch": 0.1385809312638581,
      "grad_norm": 1.2192662954330444,
      "learning_rate": 9.535415576362818e-06,
      "loss": 0.7296,
      "step": 500
    },
    {
      "epoch": 0.1388580931263858,
      "grad_norm": 1.821325421333313,
      "learning_rate": 9.533581182187386e-06,
      "loss": 0.7287,
      "step": 501
    },
    {
      "epoch": 0.13913525498891352,
      "grad_norm": 1.3631850481033325,
      "learning_rate": 9.5317433507862e-06,
      "loss": 0.7493,
      "step": 502
    },
    {
      "epoch": 0.13941241685144123,
      "grad_norm": 1.1213668584823608,
      "learning_rate": 9.529902083552647e-06,
      "loss": 0.7172,
      "step": 503
    },
    {
      "epoch": 0.13968957871396895,
      "grad_norm": 1.1022928953170776,
      "learning_rate": 9.528057381882722e-06,
      "loss": 0.6509,
      "step": 504
    },
    {
      "epoch": 0.13996674057649666,
      "grad_norm": 1.132429838180542,
      "learning_rate": 9.526209247175024e-06,
      "loss": 0.7651,
      "step": 505
    },
    {
      "epoch": 0.1402439024390244,
      "grad_norm": 1.1668297052383423,
      "learning_rate": 9.524357680830752e-06,
      "loss": 0.7432,
      "step": 506
    },
    {
      "epoch": 0.14052106430155212,
      "grad_norm": 1.1074306964874268,
      "learning_rate": 9.522502684253709e-06,
      "loss": 0.6828,
      "step": 507
    },
    {
      "epoch": 0.14079822616407983,
      "grad_norm": 1.088744878768921,
      "learning_rate": 9.520644258850298e-06,
      "loss": 0.6796,
      "step": 508
    },
    {
      "epoch": 0.14107538802660755,
      "grad_norm": 1.100386619567871,
      "learning_rate": 9.518782406029521e-06,
      "loss": 0.7223,
      "step": 509
    },
    {
      "epoch": 0.14135254988913526,
      "grad_norm": 1.4519058465957642,
      "learning_rate": 9.51691712720298e-06,
      "loss": 0.6675,
      "step": 510
    },
    {
      "epoch": 0.14162971175166297,
      "grad_norm": 1.1271750926971436,
      "learning_rate": 9.515048423784872e-06,
      "loss": 0.7048,
      "step": 511
    },
    {
      "epoch": 0.1419068736141907,
      "grad_norm": 2.329416513442993,
      "learning_rate": 9.513176297191996e-06,
      "loss": 0.7316,
      "step": 512
    },
    {
      "epoch": 0.1421840354767184,
      "grad_norm": 1.1434932947158813,
      "learning_rate": 9.511300748843736e-06,
      "loss": 0.7138,
      "step": 513
    },
    {
      "epoch": 0.14246119733924612,
      "grad_norm": 2.5311226844787598,
      "learning_rate": 9.509421780162083e-06,
      "loss": 0.6423,
      "step": 514
    },
    {
      "epoch": 0.14273835920177383,
      "grad_norm": 1.1380990743637085,
      "learning_rate": 9.507539392571609e-06,
      "loss": 0.6987,
      "step": 515
    },
    {
      "epoch": 0.14301552106430154,
      "grad_norm": 1.0649292469024658,
      "learning_rate": 9.50565358749949e-06,
      "loss": 0.6741,
      "step": 516
    },
    {
      "epoch": 0.14329268292682926,
      "grad_norm": 1.399359107017517,
      "learning_rate": 9.503764366375486e-06,
      "loss": 0.7057,
      "step": 517
    },
    {
      "epoch": 0.14356984478935697,
      "grad_norm": 1.0662505626678467,
      "learning_rate": 9.501871730631944e-06,
      "loss": 0.7562,
      "step": 518
    },
    {
      "epoch": 0.1438470066518847,
      "grad_norm": 1.15830659866333,
      "learning_rate": 9.499975681703805e-06,
      "loss": 0.6761,
      "step": 519
    },
    {
      "epoch": 0.14412416851441243,
      "grad_norm": 1.1665935516357422,
      "learning_rate": 9.4980762210286e-06,
      "loss": 0.6958,
      "step": 520
    },
    {
      "epoch": 0.14440133037694014,
      "grad_norm": 1.4200971126556396,
      "learning_rate": 9.49617335004644e-06,
      "loss": 0.7352,
      "step": 521
    },
    {
      "epoch": 0.14467849223946785,
      "grad_norm": 1.1404743194580078,
      "learning_rate": 9.494267070200027e-06,
      "loss": 0.7391,
      "step": 522
    },
    {
      "epoch": 0.14495565410199557,
      "grad_norm": 1.0870524644851685,
      "learning_rate": 9.492357382934642e-06,
      "loss": 0.7512,
      "step": 523
    },
    {
      "epoch": 0.14523281596452328,
      "grad_norm": 1.0196195840835571,
      "learning_rate": 9.490444289698158e-06,
      "loss": 0.7016,
      "step": 524
    },
    {
      "epoch": 0.145509977827051,
      "grad_norm": 1.1591192483901978,
      "learning_rate": 9.488527791941021e-06,
      "loss": 0.7706,
      "step": 525
    },
    {
      "epoch": 0.1457871396895787,
      "grad_norm": 1.1802294254302979,
      "learning_rate": 9.486607891116264e-06,
      "loss": 0.672,
      "step": 526
    },
    {
      "epoch": 0.14606430155210642,
      "grad_norm": 1.0852816104888916,
      "learning_rate": 9.484684588679498e-06,
      "loss": 0.7388,
      "step": 527
    },
    {
      "epoch": 0.14634146341463414,
      "grad_norm": 1.1203935146331787,
      "learning_rate": 9.482757886088914e-06,
      "loss": 0.6977,
      "step": 528
    },
    {
      "epoch": 0.14661862527716185,
      "grad_norm": 1.128830909729004,
      "learning_rate": 9.480827784805278e-06,
      "loss": 0.6623,
      "step": 529
    },
    {
      "epoch": 0.14689578713968957,
      "grad_norm": 1.4379198551177979,
      "learning_rate": 9.478894286291939e-06,
      "loss": 0.7314,
      "step": 530
    },
    {
      "epoch": 0.1471729490022173,
      "grad_norm": 1.2827485799789429,
      "learning_rate": 9.476957392014815e-06,
      "loss": 0.6927,
      "step": 531
    },
    {
      "epoch": 0.14745011086474502,
      "grad_norm": 1.1027277708053589,
      "learning_rate": 9.475017103442403e-06,
      "loss": 0.6813,
      "step": 532
    },
    {
      "epoch": 0.14772727272727273,
      "grad_norm": 1.1824883222579956,
      "learning_rate": 9.473073422045769e-06,
      "loss": 0.6826,
      "step": 533
    },
    {
      "epoch": 0.14800443458980045,
      "grad_norm": 1.1087284088134766,
      "learning_rate": 9.471126349298557e-06,
      "loss": 0.7023,
      "step": 534
    },
    {
      "epoch": 0.14828159645232816,
      "grad_norm": 1.1807860136032104,
      "learning_rate": 9.469175886676978e-06,
      "loss": 0.7118,
      "step": 535
    },
    {
      "epoch": 0.14855875831485588,
      "grad_norm": 1.479722499847412,
      "learning_rate": 9.467222035659816e-06,
      "loss": 0.7111,
      "step": 536
    },
    {
      "epoch": 0.1488359201773836,
      "grad_norm": 1.0895475149154663,
      "learning_rate": 9.46526479772842e-06,
      "loss": 0.6732,
      "step": 537
    },
    {
      "epoch": 0.1491130820399113,
      "grad_norm": 1.2127909660339355,
      "learning_rate": 9.463304174366712e-06,
      "loss": 0.6826,
      "step": 538
    },
    {
      "epoch": 0.14939024390243902,
      "grad_norm": 1.144930362701416,
      "learning_rate": 9.461340167061175e-06,
      "loss": 0.6787,
      "step": 539
    },
    {
      "epoch": 0.14966740576496673,
      "grad_norm": 1.112392783164978,
      "learning_rate": 9.459372777300863e-06,
      "loss": 0.719,
      "step": 540
    },
    {
      "epoch": 0.14994456762749445,
      "grad_norm": 1.1307202577590942,
      "learning_rate": 9.457402006577391e-06,
      "loss": 0.6645,
      "step": 541
    },
    {
      "epoch": 0.15022172949002216,
      "grad_norm": 1.1112520694732666,
      "learning_rate": 9.45542785638494e-06,
      "loss": 0.7122,
      "step": 542
    },
    {
      "epoch": 0.1504988913525499,
      "grad_norm": 1.0987744331359863,
      "learning_rate": 9.453450328220251e-06,
      "loss": 0.7053,
      "step": 543
    },
    {
      "epoch": 0.15077605321507762,
      "grad_norm": 1.1085962057113647,
      "learning_rate": 9.451469423582625e-06,
      "loss": 0.6715,
      "step": 544
    },
    {
      "epoch": 0.15105321507760533,
      "grad_norm": 1.1974430084228516,
      "learning_rate": 9.449485143973924e-06,
      "loss": 0.7542,
      "step": 545
    },
    {
      "epoch": 0.15133037694013304,
      "grad_norm": 1.1093443632125854,
      "learning_rate": 9.447497490898574e-06,
      "loss": 0.6969,
      "step": 546
    },
    {
      "epoch": 0.15160753880266076,
      "grad_norm": 1.3354485034942627,
      "learning_rate": 9.445506465863551e-06,
      "loss": 0.7317,
      "step": 547
    },
    {
      "epoch": 0.15188470066518847,
      "grad_norm": 1.1983423233032227,
      "learning_rate": 9.443512070378392e-06,
      "loss": 0.7132,
      "step": 548
    },
    {
      "epoch": 0.15216186252771619,
      "grad_norm": 1.0577049255371094,
      "learning_rate": 9.441514305955186e-06,
      "loss": 0.6602,
      "step": 549
    },
    {
      "epoch": 0.1524390243902439,
      "grad_norm": 1.2756695747375488,
      "learning_rate": 9.439513174108582e-06,
      "loss": 0.6847,
      "step": 550
    },
    {
      "epoch": 0.1527161862527716,
      "grad_norm": 1.1317598819732666,
      "learning_rate": 9.437508676355774e-06,
      "loss": 0.7083,
      "step": 551
    },
    {
      "epoch": 0.15299334811529933,
      "grad_norm": 1.2591124773025513,
      "learning_rate": 9.435500814216514e-06,
      "loss": 0.6567,
      "step": 552
    },
    {
      "epoch": 0.15327050997782704,
      "grad_norm": 1.3038520812988281,
      "learning_rate": 9.433489589213103e-06,
      "loss": 0.734,
      "step": 553
    },
    {
      "epoch": 0.15354767184035475,
      "grad_norm": 1.0968515872955322,
      "learning_rate": 9.431475002870394e-06,
      "loss": 0.7151,
      "step": 554
    },
    {
      "epoch": 0.1538248337028825,
      "grad_norm": 1.2311168909072876,
      "learning_rate": 9.429457056715782e-06,
      "loss": 0.7018,
      "step": 555
    },
    {
      "epoch": 0.1541019955654102,
      "grad_norm": 1.128651738166809,
      "learning_rate": 9.427435752279217e-06,
      "loss": 0.6993,
      "step": 556
    },
    {
      "epoch": 0.15437915742793792,
      "grad_norm": 1.0508489608764648,
      "learning_rate": 9.42541109109319e-06,
      "loss": 0.69,
      "step": 557
    },
    {
      "epoch": 0.15465631929046564,
      "grad_norm": 1.0961440801620483,
      "learning_rate": 9.423383074692738e-06,
      "loss": 0.7235,
      "step": 558
    },
    {
      "epoch": 0.15493348115299335,
      "grad_norm": 1.0844651460647583,
      "learning_rate": 9.421351704615445e-06,
      "loss": 0.6803,
      "step": 559
    },
    {
      "epoch": 0.15521064301552107,
      "grad_norm": 1.0876169204711914,
      "learning_rate": 9.419316982401432e-06,
      "loss": 0.671,
      "step": 560
    },
    {
      "epoch": 0.15548780487804878,
      "grad_norm": 1.0772528648376465,
      "learning_rate": 9.417278909593366e-06,
      "loss": 0.6564,
      "step": 561
    },
    {
      "epoch": 0.1557649667405765,
      "grad_norm": 0.9972807765007019,
      "learning_rate": 9.415237487736452e-06,
      "loss": 0.6661,
      "step": 562
    },
    {
      "epoch": 0.1560421286031042,
      "grad_norm": 1.099785327911377,
      "learning_rate": 9.413192718378439e-06,
      "loss": 0.6702,
      "step": 563
    },
    {
      "epoch": 0.15631929046563192,
      "grad_norm": 1.0642567873001099,
      "learning_rate": 9.411144603069604e-06,
      "loss": 0.7177,
      "step": 564
    },
    {
      "epoch": 0.15659645232815964,
      "grad_norm": 1.0651732683181763,
      "learning_rate": 9.409093143362771e-06,
      "loss": 0.6874,
      "step": 565
    },
    {
      "epoch": 0.15687361419068735,
      "grad_norm": 1.1245360374450684,
      "learning_rate": 9.407038340813292e-06,
      "loss": 0.6732,
      "step": 566
    },
    {
      "epoch": 0.1571507760532151,
      "grad_norm": 1.1072710752487183,
      "learning_rate": 9.404980196979061e-06,
      "loss": 0.6835,
      "step": 567
    },
    {
      "epoch": 0.1574279379157428,
      "grad_norm": 1.1299422979354858,
      "learning_rate": 9.402918713420498e-06,
      "loss": 0.6627,
      "step": 568
    },
    {
      "epoch": 0.15770509977827052,
      "grad_norm": 1.1340341567993164,
      "learning_rate": 9.40085389170056e-06,
      "loss": 0.7145,
      "step": 569
    },
    {
      "epoch": 0.15798226164079823,
      "grad_norm": 1.084333896636963,
      "learning_rate": 9.398785733384734e-06,
      "loss": 0.7312,
      "step": 570
    },
    {
      "epoch": 0.15825942350332595,
      "grad_norm": 1.7750368118286133,
      "learning_rate": 9.396714240041032e-06,
      "loss": 0.6754,
      "step": 571
    },
    {
      "epoch": 0.15853658536585366,
      "grad_norm": 1.0820484161376953,
      "learning_rate": 9.39463941324e-06,
      "loss": 0.7096,
      "step": 572
    },
    {
      "epoch": 0.15881374722838137,
      "grad_norm": 1.0386626720428467,
      "learning_rate": 9.392561254554712e-06,
      "loss": 0.6944,
      "step": 573
    },
    {
      "epoch": 0.1590909090909091,
      "grad_norm": 1.033211350440979,
      "learning_rate": 9.390479765560765e-06,
      "loss": 0.6834,
      "step": 574
    },
    {
      "epoch": 0.1593680709534368,
      "grad_norm": 1.1400901079177856,
      "learning_rate": 9.388394947836278e-06,
      "loss": 0.7416,
      "step": 575
    },
    {
      "epoch": 0.15964523281596452,
      "grad_norm": 1.1463879346847534,
      "learning_rate": 9.386306802961903e-06,
      "loss": 0.6949,
      "step": 576
    },
    {
      "epoch": 0.15992239467849223,
      "grad_norm": 1.1249183416366577,
      "learning_rate": 9.384215332520805e-06,
      "loss": 0.6947,
      "step": 577
    },
    {
      "epoch": 0.16019955654101994,
      "grad_norm": 1.1162731647491455,
      "learning_rate": 9.382120538098675e-06,
      "loss": 0.7121,
      "step": 578
    },
    {
      "epoch": 0.16047671840354769,
      "grad_norm": 1.0953441858291626,
      "learning_rate": 9.380022421283725e-06,
      "loss": 0.7068,
      "step": 579
    },
    {
      "epoch": 0.1607538802660754,
      "grad_norm": 1.092992901802063,
      "learning_rate": 9.377920983666684e-06,
      "loss": 0.6282,
      "step": 580
    },
    {
      "epoch": 0.1610310421286031,
      "grad_norm": 1.1506720781326294,
      "learning_rate": 9.3758162268408e-06,
      "loss": 0.8145,
      "step": 581
    },
    {
      "epoch": 0.16130820399113083,
      "grad_norm": 1.251467227935791,
      "learning_rate": 9.373708152401835e-06,
      "loss": 0.688,
      "step": 582
    },
    {
      "epoch": 0.16158536585365854,
      "grad_norm": 2.624659538269043,
      "learning_rate": 9.371596761948067e-06,
      "loss": 0.6704,
      "step": 583
    },
    {
      "epoch": 0.16186252771618626,
      "grad_norm": 1.1827201843261719,
      "learning_rate": 9.369482057080293e-06,
      "loss": 0.6526,
      "step": 584
    },
    {
      "epoch": 0.16213968957871397,
      "grad_norm": 1.068268895149231,
      "learning_rate": 9.367364039401815e-06,
      "loss": 0.6666,
      "step": 585
    },
    {
      "epoch": 0.16241685144124168,
      "grad_norm": 1.0521800518035889,
      "learning_rate": 9.365242710518454e-06,
      "loss": 0.6846,
      "step": 586
    },
    {
      "epoch": 0.1626940133037694,
      "grad_norm": 1.1052156686782837,
      "learning_rate": 9.363118072038535e-06,
      "loss": 0.6605,
      "step": 587
    },
    {
      "epoch": 0.1629711751662971,
      "grad_norm": 1.1407642364501953,
      "learning_rate": 9.3609901255729e-06,
      "loss": 0.7467,
      "step": 588
    },
    {
      "epoch": 0.16324833702882482,
      "grad_norm": 1.0845444202423096,
      "learning_rate": 9.358858872734888e-06,
      "loss": 0.7022,
      "step": 589
    },
    {
      "epoch": 0.16352549889135254,
      "grad_norm": 1.105584979057312,
      "learning_rate": 9.356724315140357e-06,
      "loss": 0.7322,
      "step": 590
    },
    {
      "epoch": 0.16380266075388025,
      "grad_norm": 1.0324032306671143,
      "learning_rate": 9.354586454407662e-06,
      "loss": 0.6433,
      "step": 591
    },
    {
      "epoch": 0.164079822616408,
      "grad_norm": 1.0412020683288574,
      "learning_rate": 9.352445292157666e-06,
      "loss": 0.653,
      "step": 592
    },
    {
      "epoch": 0.1643569844789357,
      "grad_norm": 1.1130517721176147,
      "learning_rate": 9.350300830013732e-06,
      "loss": 0.712,
      "step": 593
    },
    {
      "epoch": 0.16463414634146342,
      "grad_norm": 1.1627273559570312,
      "learning_rate": 9.34815306960173e-06,
      "loss": 0.6575,
      "step": 594
    },
    {
      "epoch": 0.16491130820399114,
      "grad_norm": 1.066411018371582,
      "learning_rate": 9.346002012550027e-06,
      "loss": 0.708,
      "step": 595
    },
    {
      "epoch": 0.16518847006651885,
      "grad_norm": 1.1324329376220703,
      "learning_rate": 9.34384766048949e-06,
      "loss": 0.7031,
      "step": 596
    },
    {
      "epoch": 0.16546563192904656,
      "grad_norm": 1.0397595167160034,
      "learning_rate": 9.341690015053485e-06,
      "loss": 0.7606,
      "step": 597
    },
    {
      "epoch": 0.16574279379157428,
      "grad_norm": 1.086559534072876,
      "learning_rate": 9.339529077877871e-06,
      "loss": 0.7689,
      "step": 598
    },
    {
      "epoch": 0.166019955654102,
      "grad_norm": 1.2256852388381958,
      "learning_rate": 9.337364850601009e-06,
      "loss": 0.6801,
      "step": 599
    },
    {
      "epoch": 0.1662971175166297,
      "grad_norm": 1.1164605617523193,
      "learning_rate": 9.335197334863752e-06,
      "loss": 0.7303,
      "step": 600
    },
    {
      "epoch": 0.16657427937915742,
      "grad_norm": 1.1182153224945068,
      "learning_rate": 9.333026532309444e-06,
      "loss": 0.7108,
      "step": 601
    },
    {
      "epoch": 0.16685144124168513,
      "grad_norm": 1.1427406072616577,
      "learning_rate": 9.330852444583924e-06,
      "loss": 0.7254,
      "step": 602
    },
    {
      "epoch": 0.16712860310421285,
      "grad_norm": 1.0919369459152222,
      "learning_rate": 9.328675073335517e-06,
      "loss": 0.6997,
      "step": 603
    },
    {
      "epoch": 0.1674057649667406,
      "grad_norm": 1.2490993738174438,
      "learning_rate": 9.326494420215046e-06,
      "loss": 0.7008,
      "step": 604
    },
    {
      "epoch": 0.1676829268292683,
      "grad_norm": 1.1109461784362793,
      "learning_rate": 9.324310486875814e-06,
      "loss": 0.7488,
      "step": 605
    },
    {
      "epoch": 0.16796008869179602,
      "grad_norm": 1.0762676000595093,
      "learning_rate": 9.322123274973613e-06,
      "loss": 0.6976,
      "step": 606
    },
    {
      "epoch": 0.16823725055432373,
      "grad_norm": 1.1399937868118286,
      "learning_rate": 9.319932786166725e-06,
      "loss": 0.7013,
      "step": 607
    },
    {
      "epoch": 0.16851441241685144,
      "grad_norm": 1.1086857318878174,
      "learning_rate": 9.317739022115909e-06,
      "loss": 0.7119,
      "step": 608
    },
    {
      "epoch": 0.16879157427937916,
      "grad_norm": 1.1931015253067017,
      "learning_rate": 9.315541984484414e-06,
      "loss": 0.7274,
      "step": 609
    },
    {
      "epoch": 0.16906873614190687,
      "grad_norm": 1.175046443939209,
      "learning_rate": 9.313341674937968e-06,
      "loss": 0.6881,
      "step": 610
    },
    {
      "epoch": 0.1693458980044346,
      "grad_norm": 1.108963966369629,
      "learning_rate": 9.31113809514478e-06,
      "loss": 0.7338,
      "step": 611
    },
    {
      "epoch": 0.1696230598669623,
      "grad_norm": 1.0643583536148071,
      "learning_rate": 9.308931246775537e-06,
      "loss": 0.6823,
      "step": 612
    },
    {
      "epoch": 0.16990022172949001,
      "grad_norm": 1.1571307182312012,
      "learning_rate": 9.306721131503408e-06,
      "loss": 0.7972,
      "step": 613
    },
    {
      "epoch": 0.17017738359201773,
      "grad_norm": 1.2920901775360107,
      "learning_rate": 9.304507751004034e-06,
      "loss": 0.6912,
      "step": 614
    },
    {
      "epoch": 0.17045454545454544,
      "grad_norm": 1.0574997663497925,
      "learning_rate": 9.302291106955534e-06,
      "loss": 0.6569,
      "step": 615
    },
    {
      "epoch": 0.17073170731707318,
      "grad_norm": 1.1633069515228271,
      "learning_rate": 9.300071201038503e-06,
      "loss": 0.6852,
      "step": 616
    },
    {
      "epoch": 0.1710088691796009,
      "grad_norm": 1.0523173809051514,
      "learning_rate": 9.297848034936007e-06,
      "loss": 0.7094,
      "step": 617
    },
    {
      "epoch": 0.1712860310421286,
      "grad_norm": 1.2257211208343506,
      "learning_rate": 9.295621610333583e-06,
      "loss": 0.7159,
      "step": 618
    },
    {
      "epoch": 0.17156319290465633,
      "grad_norm": 1.1697139739990234,
      "learning_rate": 9.29339192891924e-06,
      "loss": 0.7081,
      "step": 619
    },
    {
      "epoch": 0.17184035476718404,
      "grad_norm": 1.088154911994934,
      "learning_rate": 9.291158992383454e-06,
      "loss": 0.7153,
      "step": 620
    },
    {
      "epoch": 0.17211751662971175,
      "grad_norm": 2.6553843021392822,
      "learning_rate": 9.288922802419175e-06,
      "loss": 0.6764,
      "step": 621
    },
    {
      "epoch": 0.17239467849223947,
      "grad_norm": 1.4576897621154785,
      "learning_rate": 9.28668336072181e-06,
      "loss": 0.6988,
      "step": 622
    },
    {
      "epoch": 0.17267184035476718,
      "grad_norm": 1.068582534790039,
      "learning_rate": 9.284440668989242e-06,
      "loss": 0.6646,
      "step": 623
    },
    {
      "epoch": 0.1729490022172949,
      "grad_norm": 1.084531545639038,
      "learning_rate": 9.282194728921812e-06,
      "loss": 0.6447,
      "step": 624
    },
    {
      "epoch": 0.1732261640798226,
      "grad_norm": 1.0536274909973145,
      "learning_rate": 9.279945542222321e-06,
      "loss": 0.6669,
      "step": 625
    },
    {
      "epoch": 0.17350332594235032,
      "grad_norm": 1.0535060167312622,
      "learning_rate": 9.277693110596038e-06,
      "loss": 0.6616,
      "step": 626
    },
    {
      "epoch": 0.17378048780487804,
      "grad_norm": 2.441319704055786,
      "learning_rate": 9.27543743575069e-06,
      "loss": 0.6422,
      "step": 627
    },
    {
      "epoch": 0.17405764966740578,
      "grad_norm": 1.0846524238586426,
      "learning_rate": 9.273178519396459e-06,
      "loss": 0.6594,
      "step": 628
    },
    {
      "epoch": 0.1743348115299335,
      "grad_norm": 1.096823811531067,
      "learning_rate": 9.270916363245992e-06,
      "loss": 0.7139,
      "step": 629
    },
    {
      "epoch": 0.1746119733924612,
      "grad_norm": 1.064342737197876,
      "learning_rate": 9.268650969014384e-06,
      "loss": 0.733,
      "step": 630
    },
    {
      "epoch": 0.17488913525498892,
      "grad_norm": 1.0647087097167969,
      "learning_rate": 9.26638233841919e-06,
      "loss": 0.688,
      "step": 631
    },
    {
      "epoch": 0.17516629711751663,
      "grad_norm": 1.1161932945251465,
      "learning_rate": 9.264110473180423e-06,
      "loss": 0.7077,
      "step": 632
    },
    {
      "epoch": 0.17544345898004435,
      "grad_norm": 1.0978693962097168,
      "learning_rate": 9.261835375020536e-06,
      "loss": 0.6855,
      "step": 633
    },
    {
      "epoch": 0.17572062084257206,
      "grad_norm": 1.1772342920303345,
      "learning_rate": 9.259557045664445e-06,
      "loss": 0.6504,
      "step": 634
    },
    {
      "epoch": 0.17599778270509978,
      "grad_norm": 1.0728942155838013,
      "learning_rate": 9.257275486839512e-06,
      "loss": 0.7024,
      "step": 635
    },
    {
      "epoch": 0.1762749445676275,
      "grad_norm": 1.0770776271820068,
      "learning_rate": 9.254990700275542e-06,
      "loss": 0.669,
      "step": 636
    },
    {
      "epoch": 0.1765521064301552,
      "grad_norm": 1.0961601734161377,
      "learning_rate": 9.252702687704795e-06,
      "loss": 0.6773,
      "step": 637
    },
    {
      "epoch": 0.17682926829268292,
      "grad_norm": 1.1559607982635498,
      "learning_rate": 9.250411450861976e-06,
      "loss": 0.6913,
      "step": 638
    },
    {
      "epoch": 0.17710643015521063,
      "grad_norm": 1.1064178943634033,
      "learning_rate": 9.24811699148423e-06,
      "loss": 0.6822,
      "step": 639
    },
    {
      "epoch": 0.17738359201773837,
      "grad_norm": 1.2792575359344482,
      "learning_rate": 9.245819311311143e-06,
      "loss": 0.6437,
      "step": 640
    },
    {
      "epoch": 0.1776607538802661,
      "grad_norm": 1.1433138847351074,
      "learning_rate": 9.243518412084752e-06,
      "loss": 0.6651,
      "step": 641
    },
    {
      "epoch": 0.1779379157427938,
      "grad_norm": 1.0967350006103516,
      "learning_rate": 9.241214295549533e-06,
      "loss": 0.6796,
      "step": 642
    },
    {
      "epoch": 0.17821507760532151,
      "grad_norm": 1.0110682249069214,
      "learning_rate": 9.238906963452395e-06,
      "loss": 0.6664,
      "step": 643
    },
    {
      "epoch": 0.17849223946784923,
      "grad_norm": 1.0949651002883911,
      "learning_rate": 9.236596417542685e-06,
      "loss": 0.733,
      "step": 644
    },
    {
      "epoch": 0.17876940133037694,
      "grad_norm": 1.0891575813293457,
      "learning_rate": 9.234282659572196e-06,
      "loss": 0.7108,
      "step": 645
    },
    {
      "epoch": 0.17904656319290466,
      "grad_norm": 1.0992698669433594,
      "learning_rate": 9.231965691295145e-06,
      "loss": 0.689,
      "step": 646
    },
    {
      "epoch": 0.17932372505543237,
      "grad_norm": 1.1483192443847656,
      "learning_rate": 9.229645514468193e-06,
      "loss": 0.7089,
      "step": 647
    },
    {
      "epoch": 0.17960088691796008,
      "grad_norm": 1.0684666633605957,
      "learning_rate": 9.227322130850425e-06,
      "loss": 0.7061,
      "step": 648
    },
    {
      "epoch": 0.1798780487804878,
      "grad_norm": 1.092572808265686,
      "learning_rate": 9.22499554220336e-06,
      "loss": 0.7187,
      "step": 649
    },
    {
      "epoch": 0.1801552106430155,
      "grad_norm": 1.0879347324371338,
      "learning_rate": 9.222665750290953e-06,
      "loss": 0.7037,
      "step": 650
    },
    {
      "epoch": 0.18043237250554323,
      "grad_norm": 1.0556684732437134,
      "learning_rate": 9.220332756879578e-06,
      "loss": 0.7048,
      "step": 651
    },
    {
      "epoch": 0.18070953436807094,
      "grad_norm": 1.0972034931182861,
      "learning_rate": 9.217996563738043e-06,
      "loss": 0.689,
      "step": 652
    },
    {
      "epoch": 0.18098669623059868,
      "grad_norm": 1.0601528882980347,
      "learning_rate": 9.21565717263758e-06,
      "loss": 0.6999,
      "step": 653
    },
    {
      "epoch": 0.1812638580931264,
      "grad_norm": 1.26553475856781,
      "learning_rate": 9.213314585351842e-06,
      "loss": 0.6749,
      "step": 654
    },
    {
      "epoch": 0.1815410199556541,
      "grad_norm": 1.1288368701934814,
      "learning_rate": 9.210968803656915e-06,
      "loss": 0.6934,
      "step": 655
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 1.0554332733154297,
      "learning_rate": 9.208619829331296e-06,
      "loss": 0.6877,
      "step": 656
    },
    {
      "epoch": 0.18209534368070954,
      "grad_norm": 1.0673589706420898,
      "learning_rate": 9.206267664155906e-06,
      "loss": 0.6742,
      "step": 657
    },
    {
      "epoch": 0.18237250554323725,
      "grad_norm": 1.0937588214874268,
      "learning_rate": 9.203912309914091e-06,
      "loss": 0.7097,
      "step": 658
    },
    {
      "epoch": 0.18264966740576496,
      "grad_norm": 1.1480052471160889,
      "learning_rate": 9.201553768391606e-06,
      "loss": 0.6564,
      "step": 659
    },
    {
      "epoch": 0.18292682926829268,
      "grad_norm": 1.0542210340499878,
      "learning_rate": 9.199192041376631e-06,
      "loss": 0.718,
      "step": 660
    },
    {
      "epoch": 0.1832039911308204,
      "grad_norm": 1.0505344867706299,
      "learning_rate": 9.196827130659752e-06,
      "loss": 0.6976,
      "step": 661
    },
    {
      "epoch": 0.1834811529933481,
      "grad_norm": 1.0495635271072388,
      "learning_rate": 9.194459038033974e-06,
      "loss": 0.6717,
      "step": 662
    },
    {
      "epoch": 0.18375831485587582,
      "grad_norm": 1.1337180137634277,
      "learning_rate": 9.192087765294718e-06,
      "loss": 0.763,
      "step": 663
    },
    {
      "epoch": 0.18403547671840353,
      "grad_norm": 0.9883743524551392,
      "learning_rate": 9.18971331423981e-06,
      "loss": 0.6953,
      "step": 664
    },
    {
      "epoch": 0.18431263858093128,
      "grad_norm": 1.201306939125061,
      "learning_rate": 9.187335686669487e-06,
      "loss": 0.7011,
      "step": 665
    },
    {
      "epoch": 0.184589800443459,
      "grad_norm": 1.310215711593628,
      "learning_rate": 9.184954884386397e-06,
      "loss": 0.764,
      "step": 666
    },
    {
      "epoch": 0.1848669623059867,
      "grad_norm": 1.2735015153884888,
      "learning_rate": 9.182570909195591e-06,
      "loss": 0.6819,
      "step": 667
    },
    {
      "epoch": 0.18514412416851442,
      "grad_norm": 1.138912558555603,
      "learning_rate": 9.180183762904531e-06,
      "loss": 0.7279,
      "step": 668
    },
    {
      "epoch": 0.18542128603104213,
      "grad_norm": 1.013511061668396,
      "learning_rate": 9.177793447323076e-06,
      "loss": 0.6847,
      "step": 669
    },
    {
      "epoch": 0.18569844789356985,
      "grad_norm": 1.1299934387207031,
      "learning_rate": 9.175399964263494e-06,
      "loss": 0.7849,
      "step": 670
    },
    {
      "epoch": 0.18597560975609756,
      "grad_norm": 1.104788899421692,
      "learning_rate": 9.173003315540452e-06,
      "loss": 0.673,
      "step": 671
    },
    {
      "epoch": 0.18625277161862527,
      "grad_norm": 1.1223130226135254,
      "learning_rate": 9.170603502971017e-06,
      "loss": 0.6856,
      "step": 672
    },
    {
      "epoch": 0.186529933481153,
      "grad_norm": 1.1408203840255737,
      "learning_rate": 9.168200528374656e-06,
      "loss": 0.6971,
      "step": 673
    },
    {
      "epoch": 0.1868070953436807,
      "grad_norm": 1.082505702972412,
      "learning_rate": 9.165794393573232e-06,
      "loss": 0.6917,
      "step": 674
    },
    {
      "epoch": 0.18708425720620842,
      "grad_norm": 1.1268630027770996,
      "learning_rate": 9.163385100391004e-06,
      "loss": 0.6902,
      "step": 675
    },
    {
      "epoch": 0.18736141906873613,
      "grad_norm": 1.0520730018615723,
      "learning_rate": 9.160972650654627e-06,
      "loss": 0.7013,
      "step": 676
    },
    {
      "epoch": 0.18763858093126387,
      "grad_norm": 2.059978485107422,
      "learning_rate": 9.158557046193148e-06,
      "loss": 0.6857,
      "step": 677
    },
    {
      "epoch": 0.18791574279379158,
      "grad_norm": 1.165637493133545,
      "learning_rate": 9.156138288838006e-06,
      "loss": 0.7184,
      "step": 678
    },
    {
      "epoch": 0.1881929046563193,
      "grad_norm": 1.1204204559326172,
      "learning_rate": 9.153716380423031e-06,
      "loss": 0.71,
      "step": 679
    },
    {
      "epoch": 0.188470066518847,
      "grad_norm": 3.7244086265563965,
      "learning_rate": 9.151291322784441e-06,
      "loss": 0.6766,
      "step": 680
    },
    {
      "epoch": 0.18874722838137473,
      "grad_norm": 1.306227445602417,
      "learning_rate": 9.148863117760843e-06,
      "loss": 0.6473,
      "step": 681
    },
    {
      "epoch": 0.18902439024390244,
      "grad_norm": 1.2176991701126099,
      "learning_rate": 9.146431767193227e-06,
      "loss": 0.7263,
      "step": 682
    },
    {
      "epoch": 0.18930155210643015,
      "grad_norm": 1.0712350606918335,
      "learning_rate": 9.143997272924974e-06,
      "loss": 0.7065,
      "step": 683
    },
    {
      "epoch": 0.18957871396895787,
      "grad_norm": 1.0973856449127197,
      "learning_rate": 9.141559636801842e-06,
      "loss": 0.6837,
      "step": 684
    },
    {
      "epoch": 0.18985587583148558,
      "grad_norm": 1.1923117637634277,
      "learning_rate": 9.139118860671975e-06,
      "loss": 0.6992,
      "step": 685
    },
    {
      "epoch": 0.1901330376940133,
      "grad_norm": 1.2174626588821411,
      "learning_rate": 9.136674946385896e-06,
      "loss": 0.7221,
      "step": 686
    },
    {
      "epoch": 0.190410199556541,
      "grad_norm": 1.3265727758407593,
      "learning_rate": 9.134227895796507e-06,
      "loss": 0.7248,
      "step": 687
    },
    {
      "epoch": 0.19068736141906872,
      "grad_norm": 1.2449836730957031,
      "learning_rate": 9.131777710759093e-06,
      "loss": 0.7096,
      "step": 688
    },
    {
      "epoch": 0.19096452328159647,
      "grad_norm": 1.0649595260620117,
      "learning_rate": 9.129324393131306e-06,
      "loss": 0.6414,
      "step": 689
    },
    {
      "epoch": 0.19124168514412418,
      "grad_norm": 1.1550146341323853,
      "learning_rate": 9.126867944773179e-06,
      "loss": 0.7136,
      "step": 690
    },
    {
      "epoch": 0.1915188470066519,
      "grad_norm": 1.7589013576507568,
      "learning_rate": 9.12440836754712e-06,
      "loss": 0.6788,
      "step": 691
    },
    {
      "epoch": 0.1917960088691796,
      "grad_norm": 1.0719186067581177,
      "learning_rate": 9.121945663317906e-06,
      "loss": 0.6418,
      "step": 692
    },
    {
      "epoch": 0.19207317073170732,
      "grad_norm": 1.0116682052612305,
      "learning_rate": 9.119479833952684e-06,
      "loss": 0.6469,
      "step": 693
    },
    {
      "epoch": 0.19235033259423504,
      "grad_norm": 1.072725534439087,
      "learning_rate": 9.117010881320973e-06,
      "loss": 0.6322,
      "step": 694
    },
    {
      "epoch": 0.19262749445676275,
      "grad_norm": 1.0865559577941895,
      "learning_rate": 9.11453880729466e-06,
      "loss": 0.6847,
      "step": 695
    },
    {
      "epoch": 0.19290465631929046,
      "grad_norm": 1.2960596084594727,
      "learning_rate": 9.112063613747999e-06,
      "loss": 0.6486,
      "step": 696
    },
    {
      "epoch": 0.19318181818181818,
      "grad_norm": 1.0968952178955078,
      "learning_rate": 9.109585302557605e-06,
      "loss": 0.6283,
      "step": 697
    },
    {
      "epoch": 0.1934589800443459,
      "grad_norm": 1.1286451816558838,
      "learning_rate": 9.107103875602458e-06,
      "loss": 0.7372,
      "step": 698
    },
    {
      "epoch": 0.1937361419068736,
      "grad_norm": 1.122975468635559,
      "learning_rate": 9.104619334763907e-06,
      "loss": 0.7521,
      "step": 699
    },
    {
      "epoch": 0.19401330376940132,
      "grad_norm": 1.1205012798309326,
      "learning_rate": 9.102131681925652e-06,
      "loss": 0.6881,
      "step": 700
    },
    {
      "epoch": 0.19429046563192906,
      "grad_norm": 1.1621146202087402,
      "learning_rate": 9.09964091897376e-06,
      "loss": 0.7194,
      "step": 701
    },
    {
      "epoch": 0.19456762749445677,
      "grad_norm": 1.0543018579483032,
      "learning_rate": 9.097147047796651e-06,
      "loss": 0.6675,
      "step": 702
    },
    {
      "epoch": 0.1948447893569845,
      "grad_norm": 1.145240068435669,
      "learning_rate": 9.094650070285104e-06,
      "loss": 0.7079,
      "step": 703
    },
    {
      "epoch": 0.1951219512195122,
      "grad_norm": 1.0833097696304321,
      "learning_rate": 9.092149988332256e-06,
      "loss": 0.675,
      "step": 704
    },
    {
      "epoch": 0.19539911308203992,
      "grad_norm": 1.035591721534729,
      "learning_rate": 9.089646803833589e-06,
      "loss": 0.7019,
      "step": 705
    },
    {
      "epoch": 0.19567627494456763,
      "grad_norm": 1.2432726621627808,
      "learning_rate": 9.087140518686945e-06,
      "loss": 0.6693,
      "step": 706
    },
    {
      "epoch": 0.19595343680709534,
      "grad_norm": 1.085686445236206,
      "learning_rate": 9.084631134792516e-06,
      "loss": 0.7247,
      "step": 707
    },
    {
      "epoch": 0.19623059866962306,
      "grad_norm": 1.103668212890625,
      "learning_rate": 9.082118654052841e-06,
      "loss": 0.7553,
      "step": 708
    },
    {
      "epoch": 0.19650776053215077,
      "grad_norm": 0.9990668892860413,
      "learning_rate": 9.079603078372806e-06,
      "loss": 0.6974,
      "step": 709
    },
    {
      "epoch": 0.19678492239467849,
      "grad_norm": 1.0454579591751099,
      "learning_rate": 9.077084409659648e-06,
      "loss": 0.6408,
      "step": 710
    },
    {
      "epoch": 0.1970620842572062,
      "grad_norm": 1.4945971965789795,
      "learning_rate": 9.074562649822944e-06,
      "loss": 0.5953,
      "step": 711
    },
    {
      "epoch": 0.1973392461197339,
      "grad_norm": 1.0621200799942017,
      "learning_rate": 9.072037800774618e-06,
      "loss": 0.6677,
      "step": 712
    },
    {
      "epoch": 0.19761640798226163,
      "grad_norm": 1.0579849481582642,
      "learning_rate": 9.069509864428935e-06,
      "loss": 0.6662,
      "step": 713
    },
    {
      "epoch": 0.19789356984478937,
      "grad_norm": 1.1701908111572266,
      "learning_rate": 9.066978842702501e-06,
      "loss": 0.6682,
      "step": 714
    },
    {
      "epoch": 0.19817073170731708,
      "grad_norm": 1.0179616212844849,
      "learning_rate": 9.064444737514262e-06,
      "loss": 0.705,
      "step": 715
    },
    {
      "epoch": 0.1984478935698448,
      "grad_norm": 1.0693926811218262,
      "learning_rate": 9.061907550785498e-06,
      "loss": 0.7633,
      "step": 716
    },
    {
      "epoch": 0.1987250554323725,
      "grad_norm": 1.028255581855774,
      "learning_rate": 9.05936728443983e-06,
      "loss": 0.737,
      "step": 717
    },
    {
      "epoch": 0.19900221729490022,
      "grad_norm": 1.059967279434204,
      "learning_rate": 9.056823940403213e-06,
      "loss": 0.7202,
      "step": 718
    },
    {
      "epoch": 0.19927937915742794,
      "grad_norm": 1.055762529373169,
      "learning_rate": 9.054277520603933e-06,
      "loss": 0.6945,
      "step": 719
    },
    {
      "epoch": 0.19955654101995565,
      "grad_norm": 1.007707953453064,
      "learning_rate": 9.051728026972609e-06,
      "loss": 0.6166,
      "step": 720
    },
    {
      "epoch": 0.19983370288248337,
      "grad_norm": 1.040519118309021,
      "learning_rate": 9.049175461442192e-06,
      "loss": 0.6714,
      "step": 721
    },
    {
      "epoch": 0.20011086474501108,
      "grad_norm": 1.0507467985153198,
      "learning_rate": 9.04661982594796e-06,
      "loss": 0.6736,
      "step": 722
    },
    {
      "epoch": 0.2003880266075388,
      "grad_norm": 1.0965094566345215,
      "learning_rate": 9.044061122427522e-06,
      "loss": 0.6861,
      "step": 723
    },
    {
      "epoch": 0.2006651884700665,
      "grad_norm": 1.1330515146255493,
      "learning_rate": 9.041499352820805e-06,
      "loss": 0.7634,
      "step": 724
    },
    {
      "epoch": 0.2006651884700665,
      "eval_loss": 0.6909334063529968,
      "eval_runtime": 386.6283,
      "eval_samples_per_second": 8.354,
      "eval_steps_per_second": 1.045,
      "step": 724
    },
    {
      "epoch": 0.20094235033259422,
      "grad_norm": 1.4479457139968872,
      "learning_rate": 9.038934519070071e-06,
      "loss": 0.698,
      "step": 725
    },
    {
      "epoch": 0.20121951219512196,
      "grad_norm": 1.1164449453353882,
      "learning_rate": 9.0363666231199e-06,
      "loss": 0.7053,
      "step": 726
    },
    {
      "epoch": 0.20149667405764968,
      "grad_norm": 1.1700286865234375,
      "learning_rate": 9.033795666917191e-06,
      "loss": 0.7314,
      "step": 727
    },
    {
      "epoch": 0.2017738359201774,
      "grad_norm": 1.0873788595199585,
      "learning_rate": 9.03122165241117e-06,
      "loss": 0.6987,
      "step": 728
    },
    {
      "epoch": 0.2020509977827051,
      "grad_norm": 0.9982103705406189,
      "learning_rate": 9.028644581553374e-06,
      "loss": 0.6865,
      "step": 729
    },
    {
      "epoch": 0.20232815964523282,
      "grad_norm": 1.0216984748840332,
      "learning_rate": 9.026064456297665e-06,
      "loss": 0.6616,
      "step": 730
    },
    {
      "epoch": 0.20260532150776053,
      "grad_norm": 1.349269151687622,
      "learning_rate": 9.023481278600213e-06,
      "loss": 0.7125,
      "step": 731
    },
    {
      "epoch": 0.20288248337028825,
      "grad_norm": 1.0253278017044067,
      "learning_rate": 9.020895050419507e-06,
      "loss": 0.6902,
      "step": 732
    },
    {
      "epoch": 0.20315964523281596,
      "grad_norm": 1.0118982791900635,
      "learning_rate": 9.018305773716352e-06,
      "loss": 0.6401,
      "step": 733
    },
    {
      "epoch": 0.20343680709534367,
      "grad_norm": 0.9961504936218262,
      "learning_rate": 9.015713450453854e-06,
      "loss": 0.6628,
      "step": 734
    },
    {
      "epoch": 0.2037139689578714,
      "grad_norm": 1.0494288206100464,
      "learning_rate": 9.013118082597441e-06,
      "loss": 0.6629,
      "step": 735
    },
    {
      "epoch": 0.2039911308203991,
      "grad_norm": 1.095228672027588,
      "learning_rate": 9.010519672114838e-06,
      "loss": 0.6812,
      "step": 736
    },
    {
      "epoch": 0.20426829268292682,
      "grad_norm": 1.5415910482406616,
      "learning_rate": 9.007918220976086e-06,
      "loss": 0.6606,
      "step": 737
    },
    {
      "epoch": 0.20454545454545456,
      "grad_norm": 1.0454844236373901,
      "learning_rate": 9.005313731153525e-06,
      "loss": 0.7328,
      "step": 738
    },
    {
      "epoch": 0.20482261640798227,
      "grad_norm": 1.5914939641952515,
      "learning_rate": 9.002706204621802e-06,
      "loss": 0.6987,
      "step": 739
    },
    {
      "epoch": 0.20509977827050999,
      "grad_norm": 1.0318976640701294,
      "learning_rate": 9.000095643357868e-06,
      "loss": 0.6779,
      "step": 740
    },
    {
      "epoch": 0.2053769401330377,
      "grad_norm": 1.0594877004623413,
      "learning_rate": 8.997482049340969e-06,
      "loss": 0.6706,
      "step": 741
    },
    {
      "epoch": 0.2056541019955654,
      "grad_norm": 1.0505141019821167,
      "learning_rate": 8.994865424552656e-06,
      "loss": 0.6883,
      "step": 742
    },
    {
      "epoch": 0.20593126385809313,
      "grad_norm": 1.2077686786651611,
      "learning_rate": 8.992245770976775e-06,
      "loss": 0.6871,
      "step": 743
    },
    {
      "epoch": 0.20620842572062084,
      "grad_norm": 1.0780158042907715,
      "learning_rate": 8.989623090599471e-06,
      "loss": 0.7118,
      "step": 744
    },
    {
      "epoch": 0.20648558758314856,
      "grad_norm": 1.0667814016342163,
      "learning_rate": 8.986997385409179e-06,
      "loss": 0.6733,
      "step": 745
    },
    {
      "epoch": 0.20676274944567627,
      "grad_norm": 1.0276763439178467,
      "learning_rate": 8.98436865739663e-06,
      "loss": 0.7023,
      "step": 746
    },
    {
      "epoch": 0.20703991130820398,
      "grad_norm": 1.0980539321899414,
      "learning_rate": 8.98173690855485e-06,
      "loss": 0.6566,
      "step": 747
    },
    {
      "epoch": 0.2073170731707317,
      "grad_norm": 1.198668122291565,
      "learning_rate": 8.97910214087915e-06,
      "loss": 0.6886,
      "step": 748
    },
    {
      "epoch": 0.2075942350332594,
      "grad_norm": 1.0833740234375,
      "learning_rate": 8.976464356367133e-06,
      "loss": 0.6708,
      "step": 749
    },
    {
      "epoch": 0.20787139689578715,
      "grad_norm": 1.0562046766281128,
      "learning_rate": 8.973823557018688e-06,
      "loss": 0.7297,
      "step": 750
    },
    {
      "epoch": 0.20814855875831487,
      "grad_norm": 1.1016583442687988,
      "learning_rate": 8.97117974483599e-06,
      "loss": 0.6855,
      "step": 751
    },
    {
      "epoch": 0.20842572062084258,
      "grad_norm": 1.0025273561477661,
      "learning_rate": 8.968532921823498e-06,
      "loss": 0.7333,
      "step": 752
    },
    {
      "epoch": 0.2087028824833703,
      "grad_norm": 1.1347954273223877,
      "learning_rate": 8.965883089987957e-06,
      "loss": 0.6788,
      "step": 753
    },
    {
      "epoch": 0.208980044345898,
      "grad_norm": 1.0131365060806274,
      "learning_rate": 8.963230251338387e-06,
      "loss": 0.7022,
      "step": 754
    },
    {
      "epoch": 0.20925720620842572,
      "grad_norm": 1.179019808769226,
      "learning_rate": 8.96057440788609e-06,
      "loss": 0.6682,
      "step": 755
    },
    {
      "epoch": 0.20953436807095344,
      "grad_norm": 1.5707875490188599,
      "learning_rate": 8.957915561644652e-06,
      "loss": 0.6747,
      "step": 756
    },
    {
      "epoch": 0.20981152993348115,
      "grad_norm": 1.0582066774368286,
      "learning_rate": 8.955253714629927e-06,
      "loss": 0.6257,
      "step": 757
    },
    {
      "epoch": 0.21008869179600886,
      "grad_norm": 1.028433918952942,
      "learning_rate": 8.95258886886005e-06,
      "loss": 0.6276,
      "step": 758
    },
    {
      "epoch": 0.21036585365853658,
      "grad_norm": 1.2005999088287354,
      "learning_rate": 8.949921026355423e-06,
      "loss": 0.6775,
      "step": 759
    },
    {
      "epoch": 0.2106430155210643,
      "grad_norm": 4.517018795013428,
      "learning_rate": 8.947250189138732e-06,
      "loss": 0.6541,
      "step": 760
    },
    {
      "epoch": 0.210920177383592,
      "grad_norm": 1.1633397340774536,
      "learning_rate": 8.944576359234918e-06,
      "loss": 0.7492,
      "step": 761
    },
    {
      "epoch": 0.21119733924611975,
      "grad_norm": 1.0938668251037598,
      "learning_rate": 8.941899538671203e-06,
      "loss": 0.6522,
      "step": 762
    },
    {
      "epoch": 0.21147450110864746,
      "grad_norm": 1.0546579360961914,
      "learning_rate": 8.939219729477072e-06,
      "loss": 0.6746,
      "step": 763
    },
    {
      "epoch": 0.21175166297117518,
      "grad_norm": 1.1979657411575317,
      "learning_rate": 8.936536933684277e-06,
      "loss": 0.654,
      "step": 764
    },
    {
      "epoch": 0.2120288248337029,
      "grad_norm": 0.966216504573822,
      "learning_rate": 8.933851153326833e-06,
      "loss": 0.675,
      "step": 765
    },
    {
      "epoch": 0.2123059866962306,
      "grad_norm": 1.118032455444336,
      "learning_rate": 8.931162390441017e-06,
      "loss": 0.6925,
      "step": 766
    },
    {
      "epoch": 0.21258314855875832,
      "grad_norm": 1.2527402639389038,
      "learning_rate": 8.92847064706537e-06,
      "loss": 0.6771,
      "step": 767
    },
    {
      "epoch": 0.21286031042128603,
      "grad_norm": 1.039366364479065,
      "learning_rate": 8.925775925240692e-06,
      "loss": 0.7273,
      "step": 768
    },
    {
      "epoch": 0.21313747228381374,
      "grad_norm": 1.2166990041732788,
      "learning_rate": 8.923078227010038e-06,
      "loss": 0.7074,
      "step": 769
    },
    {
      "epoch": 0.21341463414634146,
      "grad_norm": 1.0891538858413696,
      "learning_rate": 8.920377554418725e-06,
      "loss": 0.6782,
      "step": 770
    },
    {
      "epoch": 0.21369179600886917,
      "grad_norm": 1.4806143045425415,
      "learning_rate": 8.917673909514321e-06,
      "loss": 0.6714,
      "step": 771
    },
    {
      "epoch": 0.2139689578713969,
      "grad_norm": 1.1526752710342407,
      "learning_rate": 8.91496729434665e-06,
      "loss": 0.7868,
      "step": 772
    },
    {
      "epoch": 0.2142461197339246,
      "grad_norm": 1.1266826391220093,
      "learning_rate": 8.912257710967784e-06,
      "loss": 0.6724,
      "step": 773
    },
    {
      "epoch": 0.21452328159645231,
      "grad_norm": 1.0209293365478516,
      "learning_rate": 8.90954516143205e-06,
      "loss": 0.6537,
      "step": 774
    },
    {
      "epoch": 0.21480044345898006,
      "grad_norm": 1.0760411024093628,
      "learning_rate": 8.906829647796021e-06,
      "loss": 0.644,
      "step": 775
    },
    {
      "epoch": 0.21507760532150777,
      "grad_norm": 1.1462444067001343,
      "learning_rate": 8.90411117211852e-06,
      "loss": 0.6661,
      "step": 776
    },
    {
      "epoch": 0.21535476718403548,
      "grad_norm": 1.0488094091415405,
      "learning_rate": 8.901389736460611e-06,
      "loss": 0.7078,
      "step": 777
    },
    {
      "epoch": 0.2156319290465632,
      "grad_norm": 1.128150224685669,
      "learning_rate": 8.89866534288561e-06,
      "loss": 0.6941,
      "step": 778
    },
    {
      "epoch": 0.2159090909090909,
      "grad_norm": 1.0867666006088257,
      "learning_rate": 8.895937993459065e-06,
      "loss": 0.6495,
      "step": 779
    },
    {
      "epoch": 0.21618625277161863,
      "grad_norm": 1.072265625,
      "learning_rate": 8.893207690248776e-06,
      "loss": 0.7274,
      "step": 780
    },
    {
      "epoch": 0.21646341463414634,
      "grad_norm": 1.148205280303955,
      "learning_rate": 8.890474435324775e-06,
      "loss": 0.7846,
      "step": 781
    },
    {
      "epoch": 0.21674057649667405,
      "grad_norm": 1.0795115232467651,
      "learning_rate": 8.887738230759334e-06,
      "loss": 0.6838,
      "step": 782
    },
    {
      "epoch": 0.21701773835920177,
      "grad_norm": 1.135696530342102,
      "learning_rate": 8.884999078626961e-06,
      "loss": 0.7626,
      "step": 783
    },
    {
      "epoch": 0.21729490022172948,
      "grad_norm": 1.0517326593399048,
      "learning_rate": 8.8822569810044e-06,
      "loss": 0.6694,
      "step": 784
    },
    {
      "epoch": 0.2175720620842572,
      "grad_norm": 5.737823963165283,
      "learning_rate": 8.879511939970629e-06,
      "loss": 0.7176,
      "step": 785
    },
    {
      "epoch": 0.2178492239467849,
      "grad_norm": 1.1152360439300537,
      "learning_rate": 8.876763957606852e-06,
      "loss": 0.6521,
      "step": 786
    },
    {
      "epoch": 0.21812638580931265,
      "grad_norm": 3.068476915359497,
      "learning_rate": 8.874013035996514e-06,
      "loss": 0.7039,
      "step": 787
    },
    {
      "epoch": 0.21840354767184036,
      "grad_norm": 1.8737348318099976,
      "learning_rate": 8.871259177225273e-06,
      "loss": 0.7702,
      "step": 788
    },
    {
      "epoch": 0.21868070953436808,
      "grad_norm": 1.0583746433258057,
      "learning_rate": 8.868502383381027e-06,
      "loss": 0.6575,
      "step": 789
    },
    {
      "epoch": 0.2189578713968958,
      "grad_norm": 1.0765517950057983,
      "learning_rate": 8.865742656553898e-06,
      "loss": 0.7249,
      "step": 790
    },
    {
      "epoch": 0.2192350332594235,
      "grad_norm": 1.1150541305541992,
      "learning_rate": 8.86297999883622e-06,
      "loss": 0.6972,
      "step": 791
    },
    {
      "epoch": 0.21951219512195122,
      "grad_norm": 1.0635972023010254,
      "learning_rate": 8.860214412322566e-06,
      "loss": 0.6732,
      "step": 792
    },
    {
      "epoch": 0.21978935698447893,
      "grad_norm": 1.057027816772461,
      "learning_rate": 8.857445899109716e-06,
      "loss": 0.6524,
      "step": 793
    },
    {
      "epoch": 0.22006651884700665,
      "grad_norm": 2.260875940322876,
      "learning_rate": 8.854674461296675e-06,
      "loss": 0.6362,
      "step": 794
    },
    {
      "epoch": 0.22034368070953436,
      "grad_norm": 1.227053165435791,
      "learning_rate": 8.851900100984665e-06,
      "loss": 0.7461,
      "step": 795
    },
    {
      "epoch": 0.22062084257206208,
      "grad_norm": 1.0996044874191284,
      "learning_rate": 8.849122820277121e-06,
      "loss": 0.7272,
      "step": 796
    },
    {
      "epoch": 0.2208980044345898,
      "grad_norm": 1.0419882535934448,
      "learning_rate": 8.846342621279697e-06,
      "loss": 0.6575,
      "step": 797
    },
    {
      "epoch": 0.2211751662971175,
      "grad_norm": 1.188440203666687,
      "learning_rate": 8.843559506100257e-06,
      "loss": 0.7308,
      "step": 798
    },
    {
      "epoch": 0.22145232815964525,
      "grad_norm": 1.1257516145706177,
      "learning_rate": 8.840773476848871e-06,
      "loss": 0.6944,
      "step": 799
    },
    {
      "epoch": 0.22172949002217296,
      "grad_norm": 0.9660893678665161,
      "learning_rate": 8.837984535637829e-06,
      "loss": 0.6477,
      "step": 800
    },
    {
      "epoch": 0.22200665188470067,
      "grad_norm": 1.1348531246185303,
      "learning_rate": 8.835192684581621e-06,
      "loss": 0.6544,
      "step": 801
    },
    {
      "epoch": 0.2222838137472284,
      "grad_norm": 1.0433878898620605,
      "learning_rate": 8.832397925796942e-06,
      "loss": 0.6538,
      "step": 802
    },
    {
      "epoch": 0.2225609756097561,
      "grad_norm": 1.0156769752502441,
      "learning_rate": 8.8296002614027e-06,
      "loss": 0.6712,
      "step": 803
    },
    {
      "epoch": 0.22283813747228381,
      "grad_norm": 1.4438713788986206,
      "learning_rate": 8.826799693519996e-06,
      "loss": 0.7622,
      "step": 804
    },
    {
      "epoch": 0.22311529933481153,
      "grad_norm": 1.2079060077667236,
      "learning_rate": 8.823996224272139e-06,
      "loss": 0.6741,
      "step": 805
    },
    {
      "epoch": 0.22339246119733924,
      "grad_norm": 1.073943853378296,
      "learning_rate": 8.821189855784633e-06,
      "loss": 0.6476,
      "step": 806
    },
    {
      "epoch": 0.22366962305986696,
      "grad_norm": 1.1338884830474854,
      "learning_rate": 8.818380590185187e-06,
      "loss": 0.7496,
      "step": 807
    },
    {
      "epoch": 0.22394678492239467,
      "grad_norm": 1.0286699533462524,
      "learning_rate": 8.815568429603698e-06,
      "loss": 0.6506,
      "step": 808
    },
    {
      "epoch": 0.22422394678492238,
      "grad_norm": 1.0358154773712158,
      "learning_rate": 8.812753376172266e-06,
      "loss": 0.6587,
      "step": 809
    },
    {
      "epoch": 0.2245011086474501,
      "grad_norm": 1.0752719640731812,
      "learning_rate": 8.809935432025178e-06,
      "loss": 0.6781,
      "step": 810
    },
    {
      "epoch": 0.22477827050997784,
      "grad_norm": 1.0550380945205688,
      "learning_rate": 8.807114599298915e-06,
      "loss": 0.6428,
      "step": 811
    },
    {
      "epoch": 0.22505543237250555,
      "grad_norm": 1.1432865858078003,
      "learning_rate": 8.804290880132146e-06,
      "loss": 0.6854,
      "step": 812
    },
    {
      "epoch": 0.22533259423503327,
      "grad_norm": 1.0045182704925537,
      "learning_rate": 8.801464276665736e-06,
      "loss": 0.6704,
      "step": 813
    },
    {
      "epoch": 0.22560975609756098,
      "grad_norm": 1.8004993200302124,
      "learning_rate": 8.798634791042728e-06,
      "loss": 0.7352,
      "step": 814
    },
    {
      "epoch": 0.2258869179600887,
      "grad_norm": 1.125670075416565,
      "learning_rate": 8.795802425408352e-06,
      "loss": 0.654,
      "step": 815
    },
    {
      "epoch": 0.2261640798226164,
      "grad_norm": 1.0534700155258179,
      "learning_rate": 8.792967181910027e-06,
      "loss": 0.7243,
      "step": 816
    },
    {
      "epoch": 0.22644124168514412,
      "grad_norm": 0.9902735352516174,
      "learning_rate": 8.790129062697347e-06,
      "loss": 0.6927,
      "step": 817
    },
    {
      "epoch": 0.22671840354767184,
      "grad_norm": 1.1545597314834595,
      "learning_rate": 8.787288069922089e-06,
      "loss": 0.7229,
      "step": 818
    },
    {
      "epoch": 0.22699556541019955,
      "grad_norm": 1.2079600095748901,
      "learning_rate": 8.78444420573821e-06,
      "loss": 0.7018,
      "step": 819
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 1.0179967880249023,
      "learning_rate": 8.781597472301842e-06,
      "loss": 0.7191,
      "step": 820
    },
    {
      "epoch": 0.22754988913525498,
      "grad_norm": 2.1364400386810303,
      "learning_rate": 8.778747871771293e-06,
      "loss": 0.634,
      "step": 821
    },
    {
      "epoch": 0.2278270509977827,
      "grad_norm": 1.1180236339569092,
      "learning_rate": 8.775895406307044e-06,
      "loss": 0.6542,
      "step": 822
    },
    {
      "epoch": 0.22810421286031043,
      "grad_norm": 1.0954948663711548,
      "learning_rate": 8.773040078071753e-06,
      "loss": 0.6516,
      "step": 823
    },
    {
      "epoch": 0.22838137472283815,
      "grad_norm": 1.09214186668396,
      "learning_rate": 8.770181889230242e-06,
      "loss": 0.6895,
      "step": 824
    },
    {
      "epoch": 0.22865853658536586,
      "grad_norm": 1.0559109449386597,
      "learning_rate": 8.767320841949504e-06,
      "loss": 0.6948,
      "step": 825
    },
    {
      "epoch": 0.22893569844789358,
      "grad_norm": 1.0168449878692627,
      "learning_rate": 8.7644569383987e-06,
      "loss": 0.637,
      "step": 826
    },
    {
      "epoch": 0.2292128603104213,
      "grad_norm": 1.0238269567489624,
      "learning_rate": 8.761590180749155e-06,
      "loss": 0.6429,
      "step": 827
    },
    {
      "epoch": 0.229490022172949,
      "grad_norm": 1.114939570426941,
      "learning_rate": 8.758720571174362e-06,
      "loss": 0.6289,
      "step": 828
    },
    {
      "epoch": 0.22976718403547672,
      "grad_norm": 0.9702379107475281,
      "learning_rate": 8.755848111849969e-06,
      "loss": 0.6776,
      "step": 829
    },
    {
      "epoch": 0.23004434589800443,
      "grad_norm": 1.1277539730072021,
      "learning_rate": 8.752972804953793e-06,
      "loss": 0.7177,
      "step": 830
    },
    {
      "epoch": 0.23032150776053215,
      "grad_norm": 1.2088180780410767,
      "learning_rate": 8.750094652665804e-06,
      "loss": 0.726,
      "step": 831
    },
    {
      "epoch": 0.23059866962305986,
      "grad_norm": 1.1021149158477783,
      "learning_rate": 8.74721365716813e-06,
      "loss": 0.6529,
      "step": 832
    },
    {
      "epoch": 0.23087583148558757,
      "grad_norm": 1.154092788696289,
      "learning_rate": 8.744329820645055e-06,
      "loss": 0.7141,
      "step": 833
    },
    {
      "epoch": 0.2311529933481153,
      "grad_norm": 1.4882187843322754,
      "learning_rate": 8.741443145283022e-06,
      "loss": 0.7003,
      "step": 834
    },
    {
      "epoch": 0.23143015521064303,
      "grad_norm": 1.0510627031326294,
      "learning_rate": 8.738553633270616e-06,
      "loss": 0.6876,
      "step": 835
    },
    {
      "epoch": 0.23170731707317074,
      "grad_norm": 0.9811626672744751,
      "learning_rate": 8.735661286798585e-06,
      "loss": 0.6479,
      "step": 836
    },
    {
      "epoch": 0.23198447893569846,
      "grad_norm": 1.0098036527633667,
      "learning_rate": 8.732766108059814e-06,
      "loss": 0.6671,
      "step": 837
    },
    {
      "epoch": 0.23226164079822617,
      "grad_norm": 1.0715770721435547,
      "learning_rate": 8.729868099249344e-06,
      "loss": 0.6951,
      "step": 838
    },
    {
      "epoch": 0.23253880266075388,
      "grad_norm": 1.6819283962249756,
      "learning_rate": 8.726967262564358e-06,
      "loss": 0.7088,
      "step": 839
    },
    {
      "epoch": 0.2328159645232816,
      "grad_norm": 1.5650465488433838,
      "learning_rate": 8.724063600204186e-06,
      "loss": 0.6665,
      "step": 840
    },
    {
      "epoch": 0.2330931263858093,
      "grad_norm": 1.0549918413162231,
      "learning_rate": 8.721157114370293e-06,
      "loss": 0.6281,
      "step": 841
    },
    {
      "epoch": 0.23337028824833703,
      "grad_norm": 0.9835588932037354,
      "learning_rate": 8.718247807266293e-06,
      "loss": 0.6332,
      "step": 842
    },
    {
      "epoch": 0.23364745011086474,
      "grad_norm": 1.0332732200622559,
      "learning_rate": 8.715335681097936e-06,
      "loss": 0.7438,
      "step": 843
    },
    {
      "epoch": 0.23392461197339245,
      "grad_norm": 0.977677583694458,
      "learning_rate": 8.712420738073109e-06,
      "loss": 0.6592,
      "step": 844
    },
    {
      "epoch": 0.23420177383592017,
      "grad_norm": 1.0052309036254883,
      "learning_rate": 8.709502980401833e-06,
      "loss": 0.6774,
      "step": 845
    },
    {
      "epoch": 0.23447893569844788,
      "grad_norm": 0.9832924008369446,
      "learning_rate": 8.706582410296264e-06,
      "loss": 0.6904,
      "step": 846
    },
    {
      "epoch": 0.2347560975609756,
      "grad_norm": 1.057180643081665,
      "learning_rate": 8.703659029970694e-06,
      "loss": 0.7024,
      "step": 847
    },
    {
      "epoch": 0.23503325942350334,
      "grad_norm": 1.0528740882873535,
      "learning_rate": 8.700732841641542e-06,
      "loss": 0.6452,
      "step": 848
    },
    {
      "epoch": 0.23531042128603105,
      "grad_norm": 1.034043550491333,
      "learning_rate": 8.697803847527355e-06,
      "loss": 0.7101,
      "step": 849
    },
    {
      "epoch": 0.23558758314855877,
      "grad_norm": 0.9909468293190002,
      "learning_rate": 8.69487204984881e-06,
      "loss": 0.6142,
      "step": 850
    },
    {
      "epoch": 0.23586474501108648,
      "grad_norm": 1.0700955390930176,
      "learning_rate": 8.691937450828707e-06,
      "loss": 0.679,
      "step": 851
    },
    {
      "epoch": 0.2361419068736142,
      "grad_norm": 1.0137542486190796,
      "learning_rate": 8.68900005269197e-06,
      "loss": 0.6776,
      "step": 852
    },
    {
      "epoch": 0.2364190687361419,
      "grad_norm": 1.0986920595169067,
      "learning_rate": 8.686059857665651e-06,
      "loss": 0.7312,
      "step": 853
    },
    {
      "epoch": 0.23669623059866962,
      "grad_norm": 1.5366584062576294,
      "learning_rate": 8.683116867978915e-06,
      "loss": 0.7147,
      "step": 854
    },
    {
      "epoch": 0.23697339246119734,
      "grad_norm": 1.6449958086013794,
      "learning_rate": 8.680171085863052e-06,
      "loss": 0.6958,
      "step": 855
    },
    {
      "epoch": 0.23725055432372505,
      "grad_norm": 1.0987673997879028,
      "learning_rate": 8.67722251355146e-06,
      "loss": 0.6628,
      "step": 856
    },
    {
      "epoch": 0.23752771618625276,
      "grad_norm": 1.0848712921142578,
      "learning_rate": 8.674271153279663e-06,
      "loss": 0.6579,
      "step": 857
    },
    {
      "epoch": 0.23780487804878048,
      "grad_norm": 1.4163048267364502,
      "learning_rate": 8.67131700728529e-06,
      "loss": 0.6844,
      "step": 858
    },
    {
      "epoch": 0.2380820399113082,
      "grad_norm": 1.1419154405593872,
      "learning_rate": 8.668360077808093e-06,
      "loss": 0.6831,
      "step": 859
    },
    {
      "epoch": 0.23835920177383593,
      "grad_norm": 1.069930911064148,
      "learning_rate": 8.665400367089922e-06,
      "loss": 0.7183,
      "step": 860
    },
    {
      "epoch": 0.23863636363636365,
      "grad_norm": 0.9768102169036865,
      "learning_rate": 8.662437877374743e-06,
      "loss": 0.6633,
      "step": 861
    },
    {
      "epoch": 0.23891352549889136,
      "grad_norm": 1.07041597366333,
      "learning_rate": 8.659472610908628e-06,
      "loss": 0.6267,
      "step": 862
    },
    {
      "epoch": 0.23919068736141907,
      "grad_norm": 1.0292925834655762,
      "learning_rate": 8.656504569939753e-06,
      "loss": 0.665,
      "step": 863
    },
    {
      "epoch": 0.2394678492239468,
      "grad_norm": 1.0920473337173462,
      "learning_rate": 8.653533756718394e-06,
      "loss": 0.6979,
      "step": 864
    },
    {
      "epoch": 0.2397450110864745,
      "grad_norm": 1.03950834274292,
      "learning_rate": 8.650560173496937e-06,
      "loss": 0.7188,
      "step": 865
    },
    {
      "epoch": 0.24002217294900222,
      "grad_norm": 0.9998500943183899,
      "learning_rate": 8.647583822529867e-06,
      "loss": 0.6559,
      "step": 866
    },
    {
      "epoch": 0.24029933481152993,
      "grad_norm": 1.0412664413452148,
      "learning_rate": 8.644604706073756e-06,
      "loss": 0.664,
      "step": 867
    },
    {
      "epoch": 0.24057649667405764,
      "grad_norm": 0.9959930777549744,
      "learning_rate": 8.641622826387287e-06,
      "loss": 0.6876,
      "step": 868
    },
    {
      "epoch": 0.24085365853658536,
      "grad_norm": 1.0017403364181519,
      "learning_rate": 8.638638185731231e-06,
      "loss": 0.6804,
      "step": 869
    },
    {
      "epoch": 0.24113082039911307,
      "grad_norm": 1.1178396940231323,
      "learning_rate": 8.635650786368452e-06,
      "loss": 0.7413,
      "step": 870
    },
    {
      "epoch": 0.24140798226164079,
      "grad_norm": 1.0203651189804077,
      "learning_rate": 8.63266063056391e-06,
      "loss": 0.7135,
      "step": 871
    },
    {
      "epoch": 0.24168514412416853,
      "grad_norm": 1.0406657457351685,
      "learning_rate": 8.629667720584646e-06,
      "loss": 0.7001,
      "step": 872
    },
    {
      "epoch": 0.24196230598669624,
      "grad_norm": 1.023748755455017,
      "learning_rate": 8.626672058699802e-06,
      "loss": 0.6521,
      "step": 873
    },
    {
      "epoch": 0.24223946784922396,
      "grad_norm": 1.0040292739868164,
      "learning_rate": 8.623673647180593e-06,
      "loss": 0.7086,
      "step": 874
    },
    {
      "epoch": 0.24251662971175167,
      "grad_norm": 1.0326099395751953,
      "learning_rate": 8.620672488300329e-06,
      "loss": 0.7166,
      "step": 875
    },
    {
      "epoch": 0.24279379157427938,
      "grad_norm": 1.0433515310287476,
      "learning_rate": 8.617668584334398e-06,
      "loss": 0.6797,
      "step": 876
    },
    {
      "epoch": 0.2430709534368071,
      "grad_norm": 1.0207209587097168,
      "learning_rate": 8.614661937560268e-06,
      "loss": 0.7399,
      "step": 877
    },
    {
      "epoch": 0.2433481152993348,
      "grad_norm": 1.115918755531311,
      "learning_rate": 8.611652550257495e-06,
      "loss": 0.6952,
      "step": 878
    },
    {
      "epoch": 0.24362527716186252,
      "grad_norm": 1.0671745538711548,
      "learning_rate": 8.608640424707697e-06,
      "loss": 0.7001,
      "step": 879
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 0.9944931268692017,
      "learning_rate": 8.605625563194586e-06,
      "loss": 0.6677,
      "step": 880
    },
    {
      "epoch": 0.24417960088691795,
      "grad_norm": 1.0397342443466187,
      "learning_rate": 8.602607968003935e-06,
      "loss": 0.7561,
      "step": 881
    },
    {
      "epoch": 0.24445676274944567,
      "grad_norm": 1.0978014469146729,
      "learning_rate": 8.599587641423598e-06,
      "loss": 0.7106,
      "step": 882
    },
    {
      "epoch": 0.24473392461197338,
      "grad_norm": 1.285919427871704,
      "learning_rate": 8.596564585743493e-06,
      "loss": 0.6782,
      "step": 883
    },
    {
      "epoch": 0.24501108647450112,
      "grad_norm": 1.2276241779327393,
      "learning_rate": 8.593538803255612e-06,
      "loss": 0.7005,
      "step": 884
    },
    {
      "epoch": 0.24528824833702884,
      "grad_norm": 1.0876127481460571,
      "learning_rate": 8.590510296254013e-06,
      "loss": 0.6709,
      "step": 885
    },
    {
      "epoch": 0.24556541019955655,
      "grad_norm": 1.041988730430603,
      "learning_rate": 8.587479067034819e-06,
      "loss": 0.692,
      "step": 886
    },
    {
      "epoch": 0.24584257206208426,
      "grad_norm": 1.090190052986145,
      "learning_rate": 8.584445117896219e-06,
      "loss": 0.6962,
      "step": 887
    },
    {
      "epoch": 0.24611973392461198,
      "grad_norm": 1.0158236026763916,
      "learning_rate": 8.58140845113846e-06,
      "loss": 0.6668,
      "step": 888
    },
    {
      "epoch": 0.2463968957871397,
      "grad_norm": 1.1282451152801514,
      "learning_rate": 8.578369069063854e-06,
      "loss": 0.6943,
      "step": 889
    },
    {
      "epoch": 0.2466740576496674,
      "grad_norm": 1.0262227058410645,
      "learning_rate": 8.575326973976767e-06,
      "loss": 0.7026,
      "step": 890
    },
    {
      "epoch": 0.24695121951219512,
      "grad_norm": 0.9519471526145935,
      "learning_rate": 8.572282168183628e-06,
      "loss": 0.699,
      "step": 891
    },
    {
      "epoch": 0.24722838137472283,
      "grad_norm": 1.2660669088363647,
      "learning_rate": 8.569234653992916e-06,
      "loss": 0.6945,
      "step": 892
    },
    {
      "epoch": 0.24750554323725055,
      "grad_norm": 1.0465381145477295,
      "learning_rate": 8.566184433715166e-06,
      "loss": 0.6759,
      "step": 893
    },
    {
      "epoch": 0.24778270509977826,
      "grad_norm": 1.0093185901641846,
      "learning_rate": 8.563131509662963e-06,
      "loss": 0.7387,
      "step": 894
    },
    {
      "epoch": 0.24805986696230597,
      "grad_norm": 0.9591051936149597,
      "learning_rate": 8.560075884150942e-06,
      "loss": 0.65,
      "step": 895
    },
    {
      "epoch": 0.24833702882483372,
      "grad_norm": 1.0582847595214844,
      "learning_rate": 8.557017559495788e-06,
      "loss": 0.6969,
      "step": 896
    },
    {
      "epoch": 0.24861419068736143,
      "grad_norm": 0.9744759202003479,
      "learning_rate": 8.55395653801623e-06,
      "loss": 0.6403,
      "step": 897
    },
    {
      "epoch": 0.24889135254988914,
      "grad_norm": 0.9649972319602966,
      "learning_rate": 8.550892822033047e-06,
      "loss": 0.6864,
      "step": 898
    },
    {
      "epoch": 0.24916851441241686,
      "grad_norm": 0.9675922989845276,
      "learning_rate": 8.547826413869052e-06,
      "loss": 0.6548,
      "step": 899
    },
    {
      "epoch": 0.24944567627494457,
      "grad_norm": 1.0141688585281372,
      "learning_rate": 8.544757315849108e-06,
      "loss": 0.7162,
      "step": 900
    },
    {
      "epoch": 0.24972283813747229,
      "grad_norm": 1.0500422716140747,
      "learning_rate": 8.54168553030011e-06,
      "loss": 0.6722,
      "step": 901
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0784482955932617,
      "learning_rate": 8.538611059550997e-06,
      "loss": 0.6832,
      "step": 902
    },
    {
      "epoch": 0.25027716186252774,
      "grad_norm": 1.0078564882278442,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.6848,
      "step": 903
    },
    {
      "epoch": 0.25055432372505543,
      "grad_norm": 1.1089128255844116,
      "learning_rate": 8.53245407177834e-06,
      "loss": 0.7254,
      "step": 904
    },
    {
      "epoch": 0.25083148558758317,
      "grad_norm": 1.0367542505264282,
      "learning_rate": 8.529371559422843e-06,
      "loss": 0.7209,
      "step": 905
    },
    {
      "epoch": 0.25110864745011086,
      "grad_norm": 1.0461618900299072,
      "learning_rate": 8.526286371203314e-06,
      "loss": 0.7016,
      "step": 906
    },
    {
      "epoch": 0.2513858093126386,
      "grad_norm": 1.071169376373291,
      "learning_rate": 8.523198509458852e-06,
      "loss": 0.6742,
      "step": 907
    },
    {
      "epoch": 0.2516629711751663,
      "grad_norm": 0.9628375172615051,
      "learning_rate": 8.520107976530577e-06,
      "loss": 0.6915,
      "step": 908
    },
    {
      "epoch": 0.251940133037694,
      "grad_norm": 0.9911226630210876,
      "learning_rate": 8.517014774761643e-06,
      "loss": 0.7062,
      "step": 909
    },
    {
      "epoch": 0.2522172949002217,
      "grad_norm": 1.0530232191085815,
      "learning_rate": 8.513918906497222e-06,
      "loss": 0.6592,
      "step": 910
    },
    {
      "epoch": 0.25249445676274945,
      "grad_norm": 1.0125035047531128,
      "learning_rate": 8.510820374084511e-06,
      "loss": 0.6127,
      "step": 911
    },
    {
      "epoch": 0.25277161862527714,
      "grad_norm": 1.0277364253997803,
      "learning_rate": 8.50771917987272e-06,
      "loss": 0.6959,
      "step": 912
    },
    {
      "epoch": 0.2530487804878049,
      "grad_norm": 0.9462876319885254,
      "learning_rate": 8.504615326213085e-06,
      "loss": 0.636,
      "step": 913
    },
    {
      "epoch": 0.25332594235033257,
      "grad_norm": 1.0683320760726929,
      "learning_rate": 8.501508815458856e-06,
      "loss": 0.6692,
      "step": 914
    },
    {
      "epoch": 0.2536031042128603,
      "grad_norm": 1.491721510887146,
      "learning_rate": 8.498399649965295e-06,
      "loss": 0.7101,
      "step": 915
    },
    {
      "epoch": 0.25388026607538805,
      "grad_norm": 1.0101505517959595,
      "learning_rate": 8.495287832089676e-06,
      "loss": 0.6559,
      "step": 916
    },
    {
      "epoch": 0.25415742793791574,
      "grad_norm": 0.9924319386482239,
      "learning_rate": 8.492173364191288e-06,
      "loss": 0.6499,
      "step": 917
    },
    {
      "epoch": 0.2544345898004435,
      "grad_norm": 1.0329563617706299,
      "learning_rate": 8.48905624863143e-06,
      "loss": 0.7047,
      "step": 918
    },
    {
      "epoch": 0.25471175166297116,
      "grad_norm": 0.9805352687835693,
      "learning_rate": 8.485936487773402e-06,
      "loss": 0.6659,
      "step": 919
    },
    {
      "epoch": 0.2549889135254989,
      "grad_norm": 1.0327603816986084,
      "learning_rate": 8.482814083982518e-06,
      "loss": 0.6357,
      "step": 920
    },
    {
      "epoch": 0.2552660753880266,
      "grad_norm": 1.0860047340393066,
      "learning_rate": 8.479689039626088e-06,
      "loss": 0.6963,
      "step": 921
    },
    {
      "epoch": 0.25554323725055433,
      "grad_norm": 1.0468031167984009,
      "learning_rate": 8.476561357073427e-06,
      "loss": 0.6793,
      "step": 922
    },
    {
      "epoch": 0.255820399113082,
      "grad_norm": 1.0575693845748901,
      "learning_rate": 8.473431038695852e-06,
      "loss": 0.7356,
      "step": 923
    },
    {
      "epoch": 0.25609756097560976,
      "grad_norm": 0.9729621410369873,
      "learning_rate": 8.470298086866677e-06,
      "loss": 0.7075,
      "step": 924
    },
    {
      "epoch": 0.25637472283813745,
      "grad_norm": 4.950893402099609,
      "learning_rate": 8.467162503961209e-06,
      "loss": 0.6857,
      "step": 925
    },
    {
      "epoch": 0.2566518847006652,
      "grad_norm": 1.1679433584213257,
      "learning_rate": 8.464024292356759e-06,
      "loss": 0.6728,
      "step": 926
    },
    {
      "epoch": 0.25692904656319293,
      "grad_norm": 1.0587812662124634,
      "learning_rate": 8.46088345443262e-06,
      "loss": 0.6704,
      "step": 927
    },
    {
      "epoch": 0.2572062084257206,
      "grad_norm": 0.9744266867637634,
      "learning_rate": 8.457739992570083e-06,
      "loss": 0.6503,
      "step": 928
    },
    {
      "epoch": 0.25748337028824836,
      "grad_norm": 1.014914870262146,
      "learning_rate": 8.454593909152427e-06,
      "loss": 0.6698,
      "step": 929
    },
    {
      "epoch": 0.25776053215077604,
      "grad_norm": 0.9977729320526123,
      "learning_rate": 8.45144520656492e-06,
      "loss": 0.6711,
      "step": 930
    },
    {
      "epoch": 0.2580376940133038,
      "grad_norm": 0.9963119626045227,
      "learning_rate": 8.448293887194814e-06,
      "loss": 0.6857,
      "step": 931
    },
    {
      "epoch": 0.2583148558758315,
      "grad_norm": 0.9977074861526489,
      "learning_rate": 8.445139953431342e-06,
      "loss": 0.6941,
      "step": 932
    },
    {
      "epoch": 0.2585920177383592,
      "grad_norm": 1.0489494800567627,
      "learning_rate": 8.441983407665726e-06,
      "loss": 0.6864,
      "step": 933
    },
    {
      "epoch": 0.2588691796008869,
      "grad_norm": 1.2952728271484375,
      "learning_rate": 8.438824252291162e-06,
      "loss": 0.6809,
      "step": 934
    },
    {
      "epoch": 0.25914634146341464,
      "grad_norm": 0.9615367650985718,
      "learning_rate": 8.435662489702829e-06,
      "loss": 0.6489,
      "step": 935
    },
    {
      "epoch": 0.25942350332594233,
      "grad_norm": 1.0222314596176147,
      "learning_rate": 8.432498122297879e-06,
      "loss": 0.6225,
      "step": 936
    },
    {
      "epoch": 0.25970066518847007,
      "grad_norm": 1.091224193572998,
      "learning_rate": 8.429331152475443e-06,
      "loss": 0.6981,
      "step": 937
    },
    {
      "epoch": 0.25997782705099776,
      "grad_norm": 1.0282418727874756,
      "learning_rate": 8.426161582636621e-06,
      "loss": 0.6484,
      "step": 938
    },
    {
      "epoch": 0.2602549889135255,
      "grad_norm": 1.0255424976348877,
      "learning_rate": 8.422989415184489e-06,
      "loss": 0.6259,
      "step": 939
    },
    {
      "epoch": 0.26053215077605324,
      "grad_norm": 1.0709596872329712,
      "learning_rate": 8.419814652524084e-06,
      "loss": 0.6486,
      "step": 940
    },
    {
      "epoch": 0.2608093126385809,
      "grad_norm": 2.215697765350342,
      "learning_rate": 8.416637297062421e-06,
      "loss": 0.6605,
      "step": 941
    },
    {
      "epoch": 0.26108647450110867,
      "grad_norm": 1.0059655904769897,
      "learning_rate": 8.413457351208477e-06,
      "loss": 0.6992,
      "step": 942
    },
    {
      "epoch": 0.26136363636363635,
      "grad_norm": 1.0461018085479736,
      "learning_rate": 8.410274817373188e-06,
      "loss": 0.6982,
      "step": 943
    },
    {
      "epoch": 0.2616407982261641,
      "grad_norm": 1.0889177322387695,
      "learning_rate": 8.407089697969458e-06,
      "loss": 0.6854,
      "step": 944
    },
    {
      "epoch": 0.2619179600886918,
      "grad_norm": 2.238266944885254,
      "learning_rate": 8.403901995412147e-06,
      "loss": 0.6763,
      "step": 945
    },
    {
      "epoch": 0.2621951219512195,
      "grad_norm": 1.2084206342697144,
      "learning_rate": 8.400711712118079e-06,
      "loss": 0.7366,
      "step": 946
    },
    {
      "epoch": 0.2624722838137472,
      "grad_norm": 1.2617522478103638,
      "learning_rate": 8.39751885050603e-06,
      "loss": 0.6505,
      "step": 947
    },
    {
      "epoch": 0.26274944567627495,
      "grad_norm": 1.098517656326294,
      "learning_rate": 8.394323412996733e-06,
      "loss": 0.6463,
      "step": 948
    },
    {
      "epoch": 0.26302660753880264,
      "grad_norm": 1.0320173501968384,
      "learning_rate": 8.39112540201287e-06,
      "loss": 0.6743,
      "step": 949
    },
    {
      "epoch": 0.2633037694013304,
      "grad_norm": 1.0113227367401123,
      "learning_rate": 8.387924819979083e-06,
      "loss": 0.6379,
      "step": 950
    },
    {
      "epoch": 0.2635809312638581,
      "grad_norm": 1.007591962814331,
      "learning_rate": 8.384721669321952e-06,
      "loss": 0.7072,
      "step": 951
    },
    {
      "epoch": 0.2638580931263858,
      "grad_norm": 1.088312029838562,
      "learning_rate": 8.381515952470015e-06,
      "loss": 0.6634,
      "step": 952
    },
    {
      "epoch": 0.26413525498891355,
      "grad_norm": 1.0449656248092651,
      "learning_rate": 8.378307671853747e-06,
      "loss": 0.663,
      "step": 953
    },
    {
      "epoch": 0.26441241685144123,
      "grad_norm": 0.9798909425735474,
      "learning_rate": 8.375096829905573e-06,
      "loss": 0.6566,
      "step": 954
    },
    {
      "epoch": 0.264689578713969,
      "grad_norm": 1.077069878578186,
      "learning_rate": 8.371883429059856e-06,
      "loss": 0.666,
      "step": 955
    },
    {
      "epoch": 0.26496674057649666,
      "grad_norm": 1.0567377805709839,
      "learning_rate": 8.368667471752897e-06,
      "loss": 0.7205,
      "step": 956
    },
    {
      "epoch": 0.2652439024390244,
      "grad_norm": 0.9581822752952576,
      "learning_rate": 8.365448960422948e-06,
      "loss": 0.6339,
      "step": 957
    },
    {
      "epoch": 0.2655210643015521,
      "grad_norm": 1.0013914108276367,
      "learning_rate": 8.36222789751018e-06,
      "loss": 0.701,
      "step": 958
    },
    {
      "epoch": 0.26579822616407983,
      "grad_norm": 0.9505181312561035,
      "learning_rate": 8.35900428545671e-06,
      "loss": 0.6145,
      "step": 959
    },
    {
      "epoch": 0.2660753880266075,
      "grad_norm": 0.9950973391532898,
      "learning_rate": 8.355778126706582e-06,
      "loss": 0.6692,
      "step": 960
    },
    {
      "epoch": 0.26635254988913526,
      "grad_norm": 1.4571685791015625,
      "learning_rate": 8.352549423705776e-06,
      "loss": 0.6851,
      "step": 961
    },
    {
      "epoch": 0.26662971175166295,
      "grad_norm": 0.9905958771705627,
      "learning_rate": 8.349318178902195e-06,
      "loss": 0.6859,
      "step": 962
    },
    {
      "epoch": 0.2669068736141907,
      "grad_norm": 1.017879843711853,
      "learning_rate": 8.346084394745677e-06,
      "loss": 0.6595,
      "step": 963
    },
    {
      "epoch": 0.26718403547671843,
      "grad_norm": 1.0100001096725464,
      "learning_rate": 8.342848073687974e-06,
      "loss": 0.6324,
      "step": 964
    },
    {
      "epoch": 0.2674611973392461,
      "grad_norm": 1.1317713260650635,
      "learning_rate": 8.339609218182774e-06,
      "loss": 0.6757,
      "step": 965
    },
    {
      "epoch": 0.26773835920177386,
      "grad_norm": 0.9610005617141724,
      "learning_rate": 8.336367830685677e-06,
      "loss": 0.6496,
      "step": 966
    },
    {
      "epoch": 0.26801552106430154,
      "grad_norm": 1.0578080415725708,
      "learning_rate": 8.333123913654206e-06,
      "loss": 0.6839,
      "step": 967
    },
    {
      "epoch": 0.2682926829268293,
      "grad_norm": 1.066480040550232,
      "learning_rate": 8.329877469547805e-06,
      "loss": 0.7039,
      "step": 968
    },
    {
      "epoch": 0.26856984478935697,
      "grad_norm": 1.011892318725586,
      "learning_rate": 8.326628500827826e-06,
      "loss": 0.6794,
      "step": 969
    },
    {
      "epoch": 0.2688470066518847,
      "grad_norm": 1.0349020957946777,
      "learning_rate": 8.323377009957546e-06,
      "loss": 0.6869,
      "step": 970
    },
    {
      "epoch": 0.2691241685144124,
      "grad_norm": 0.9678639769554138,
      "learning_rate": 8.320122999402142e-06,
      "loss": 0.6442,
      "step": 971
    },
    {
      "epoch": 0.26940133037694014,
      "grad_norm": 1.030608892440796,
      "learning_rate": 8.316866471628713e-06,
      "loss": 0.7018,
      "step": 972
    },
    {
      "epoch": 0.2696784922394678,
      "grad_norm": 1.5167262554168701,
      "learning_rate": 8.313607429106258e-06,
      "loss": 0.7427,
      "step": 973
    },
    {
      "epoch": 0.26995565410199557,
      "grad_norm": 1.0126299858093262,
      "learning_rate": 8.310345874305686e-06,
      "loss": 0.6394,
      "step": 974
    },
    {
      "epoch": 0.27023281596452325,
      "grad_norm": 1.0805809497833252,
      "learning_rate": 8.30708180969981e-06,
      "loss": 0.6639,
      "step": 975
    },
    {
      "epoch": 0.270509977827051,
      "grad_norm": 1.0736037492752075,
      "learning_rate": 8.303815237763347e-06,
      "loss": 0.6884,
      "step": 976
    },
    {
      "epoch": 0.27078713968957874,
      "grad_norm": 0.9442486763000488,
      "learning_rate": 8.300546160972911e-06,
      "loss": 0.6516,
      "step": 977
    },
    {
      "epoch": 0.2710643015521064,
      "grad_norm": 1.0039750337600708,
      "learning_rate": 8.297274581807024e-06,
      "loss": 0.6362,
      "step": 978
    },
    {
      "epoch": 0.27134146341463417,
      "grad_norm": 1.0862104892730713,
      "learning_rate": 8.294000502746098e-06,
      "loss": 0.6773,
      "step": 979
    },
    {
      "epoch": 0.27161862527716185,
      "grad_norm": 0.9906332492828369,
      "learning_rate": 8.290723926272439e-06,
      "loss": 0.6398,
      "step": 980
    },
    {
      "epoch": 0.2718957871396896,
      "grad_norm": 1.2317073345184326,
      "learning_rate": 8.287444854870252e-06,
      "loss": 0.6717,
      "step": 981
    },
    {
      "epoch": 0.2721729490022173,
      "grad_norm": 1.0142642259597778,
      "learning_rate": 8.284163291025628e-06,
      "loss": 0.717,
      "step": 982
    },
    {
      "epoch": 0.272450110864745,
      "grad_norm": 1.0692951679229736,
      "learning_rate": 8.280879237226556e-06,
      "loss": 0.6869,
      "step": 983
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 1.0355192422866821,
      "learning_rate": 8.277592695962902e-06,
      "loss": 0.659,
      "step": 984
    },
    {
      "epoch": 0.27300443458980045,
      "grad_norm": 1.0701180696487427,
      "learning_rate": 8.274303669726427e-06,
      "loss": 0.6837,
      "step": 985
    },
    {
      "epoch": 0.27328159645232813,
      "grad_norm": 0.9856352806091309,
      "learning_rate": 8.271012161010769e-06,
      "loss": 0.641,
      "step": 986
    },
    {
      "epoch": 0.2735587583148559,
      "grad_norm": 0.9781636595726013,
      "learning_rate": 8.267718172311455e-06,
      "loss": 0.6767,
      "step": 987
    },
    {
      "epoch": 0.2738359201773836,
      "grad_norm": 1.0709030628204346,
      "learning_rate": 8.264421706125885e-06,
      "loss": 0.661,
      "step": 988
    },
    {
      "epoch": 0.2741130820399113,
      "grad_norm": 1.0208141803741455,
      "learning_rate": 8.261122764953344e-06,
      "loss": 0.6777,
      "step": 989
    },
    {
      "epoch": 0.27439024390243905,
      "grad_norm": 1.1091591119766235,
      "learning_rate": 8.25782135129499e-06,
      "loss": 0.6507,
      "step": 990
    },
    {
      "epoch": 0.27466740576496673,
      "grad_norm": 0.9993110299110413,
      "learning_rate": 8.254517467653858e-06,
      "loss": 0.6987,
      "step": 991
    },
    {
      "epoch": 0.2749445676274945,
      "grad_norm": 1.0656479597091675,
      "learning_rate": 8.25121111653485e-06,
      "loss": 0.6856,
      "step": 992
    },
    {
      "epoch": 0.27522172949002216,
      "grad_norm": 1.0330309867858887,
      "learning_rate": 8.247902300444743e-06,
      "loss": 0.7046,
      "step": 993
    },
    {
      "epoch": 0.2754988913525499,
      "grad_norm": 0.981895923614502,
      "learning_rate": 8.244591021892186e-06,
      "loss": 0.6375,
      "step": 994
    },
    {
      "epoch": 0.2757760532150776,
      "grad_norm": 1.0508733987808228,
      "learning_rate": 8.24127728338769e-06,
      "loss": 0.6109,
      "step": 995
    },
    {
      "epoch": 0.27605321507760533,
      "grad_norm": 0.970263659954071,
      "learning_rate": 8.23796108744363e-06,
      "loss": 0.6576,
      "step": 996
    },
    {
      "epoch": 0.276330376940133,
      "grad_norm": 1.025241732597351,
      "learning_rate": 8.23464243657425e-06,
      "loss": 0.6263,
      "step": 997
    },
    {
      "epoch": 0.27660753880266076,
      "grad_norm": 0.9892611503601074,
      "learning_rate": 8.23132133329565e-06,
      "loss": 0.625,
      "step": 998
    },
    {
      "epoch": 0.27688470066518844,
      "grad_norm": 1.1280057430267334,
      "learning_rate": 8.227997780125791e-06,
      "loss": 0.6533,
      "step": 999
    },
    {
      "epoch": 0.2771618625277162,
      "grad_norm": 1.0535888671875,
      "learning_rate": 8.224671779584493e-06,
      "loss": 0.6868,
      "step": 1000
    },
    {
      "epoch": 0.2774390243902439,
      "grad_norm": 1.0197157859802246,
      "learning_rate": 8.22134333419343e-06,
      "loss": 0.6998,
      "step": 1001
    },
    {
      "epoch": 0.2777161862527716,
      "grad_norm": 0.9827303886413574,
      "learning_rate": 8.218012446476128e-06,
      "loss": 0.6481,
      "step": 1002
    },
    {
      "epoch": 0.27799334811529935,
      "grad_norm": 1.130528450012207,
      "learning_rate": 8.214679118957968e-06,
      "loss": 0.6718,
      "step": 1003
    },
    {
      "epoch": 0.27827050997782704,
      "grad_norm": 0.9759430885314941,
      "learning_rate": 8.211343354166181e-06,
      "loss": 0.6675,
      "step": 1004
    },
    {
      "epoch": 0.2785476718403548,
      "grad_norm": 0.967077910900116,
      "learning_rate": 8.208005154629841e-06,
      "loss": 0.7141,
      "step": 1005
    },
    {
      "epoch": 0.27882483370288247,
      "grad_norm": 1.0265557765960693,
      "learning_rate": 8.204664522879874e-06,
      "loss": 0.6041,
      "step": 1006
    },
    {
      "epoch": 0.2791019955654102,
      "grad_norm": 1.0266109704971313,
      "learning_rate": 8.201321461449043e-06,
      "loss": 0.6587,
      "step": 1007
    },
    {
      "epoch": 0.2793791574279379,
      "grad_norm": 1.0265191793441772,
      "learning_rate": 8.197975972871962e-06,
      "loss": 0.6752,
      "step": 1008
    },
    {
      "epoch": 0.27965631929046564,
      "grad_norm": 1.0341322422027588,
      "learning_rate": 8.194628059685077e-06,
      "loss": 0.6695,
      "step": 1009
    },
    {
      "epoch": 0.2799334811529933,
      "grad_norm": 1.175136685371399,
      "learning_rate": 8.191277724426678e-06,
      "loss": 0.711,
      "step": 1010
    },
    {
      "epoch": 0.28021064301552107,
      "grad_norm": 0.9784871339797974,
      "learning_rate": 8.187924969636886e-06,
      "loss": 0.6976,
      "step": 1011
    },
    {
      "epoch": 0.2804878048780488,
      "grad_norm": 1.0755832195281982,
      "learning_rate": 8.184569797857665e-06,
      "loss": 0.6588,
      "step": 1012
    },
    {
      "epoch": 0.2807649667405765,
      "grad_norm": 1.1047717332839966,
      "learning_rate": 8.1812122116328e-06,
      "loss": 0.6122,
      "step": 1013
    },
    {
      "epoch": 0.28104212860310424,
      "grad_norm": 1.052466869354248,
      "learning_rate": 8.177852213507916e-06,
      "loss": 0.6859,
      "step": 1014
    },
    {
      "epoch": 0.2813192904656319,
      "grad_norm": 1.1258013248443604,
      "learning_rate": 8.174489806030462e-06,
      "loss": 0.7188,
      "step": 1015
    },
    {
      "epoch": 0.28159645232815966,
      "grad_norm": 1.1118065118789673,
      "learning_rate": 8.171124991749715e-06,
      "loss": 0.6198,
      "step": 1016
    },
    {
      "epoch": 0.28187361419068735,
      "grad_norm": 0.9888275861740112,
      "learning_rate": 8.167757773216776e-06,
      "loss": 0.6719,
      "step": 1017
    },
    {
      "epoch": 0.2821507760532151,
      "grad_norm": 1.090035080909729,
      "learning_rate": 8.164388152984572e-06,
      "loss": 0.703,
      "step": 1018
    },
    {
      "epoch": 0.2824279379157428,
      "grad_norm": 1.113784670829773,
      "learning_rate": 8.161016133607846e-06,
      "loss": 0.6476,
      "step": 1019
    },
    {
      "epoch": 0.2827050997782705,
      "grad_norm": 1.0172189474105835,
      "learning_rate": 8.157641717643162e-06,
      "loss": 0.6572,
      "step": 1020
    },
    {
      "epoch": 0.2829822616407982,
      "grad_norm": 1.024641513824463,
      "learning_rate": 8.154264907648902e-06,
      "loss": 0.6798,
      "step": 1021
    },
    {
      "epoch": 0.28325942350332595,
      "grad_norm": 1.0841732025146484,
      "learning_rate": 8.150885706185264e-06,
      "loss": 0.6994,
      "step": 1022
    },
    {
      "epoch": 0.28353658536585363,
      "grad_norm": 0.9887521266937256,
      "learning_rate": 8.147504115814255e-06,
      "loss": 0.6631,
      "step": 1023
    },
    {
      "epoch": 0.2838137472283814,
      "grad_norm": 0.9686548113822937,
      "learning_rate": 8.144120139099697e-06,
      "loss": 0.6383,
      "step": 1024
    },
    {
      "epoch": 0.2840909090909091,
      "grad_norm": 1.0649688243865967,
      "learning_rate": 8.140733778607219e-06,
      "loss": 0.6861,
      "step": 1025
    },
    {
      "epoch": 0.2843680709534368,
      "grad_norm": 0.9883394837379456,
      "learning_rate": 8.13734503690426e-06,
      "loss": 0.6491,
      "step": 1026
    },
    {
      "epoch": 0.28464523281596454,
      "grad_norm": 0.9586151838302612,
      "learning_rate": 8.133953916560063e-06,
      "loss": 0.6149,
      "step": 1027
    },
    {
      "epoch": 0.28492239467849223,
      "grad_norm": 1.0658714771270752,
      "learning_rate": 8.13056042014567e-06,
      "loss": 0.651,
      "step": 1028
    },
    {
      "epoch": 0.28519955654101997,
      "grad_norm": 1.0532435178756714,
      "learning_rate": 8.127164550233935e-06,
      "loss": 0.6948,
      "step": 1029
    },
    {
      "epoch": 0.28547671840354766,
      "grad_norm": 1.0832480192184448,
      "learning_rate": 8.1237663093995e-06,
      "loss": 0.7548,
      "step": 1030
    },
    {
      "epoch": 0.2857538802660754,
      "grad_norm": 1.0805360078811646,
      "learning_rate": 8.12036570021881e-06,
      "loss": 0.6843,
      "step": 1031
    },
    {
      "epoch": 0.2860310421286031,
      "grad_norm": 1.0097062587738037,
      "learning_rate": 8.11696272527011e-06,
      "loss": 0.7274,
      "step": 1032
    },
    {
      "epoch": 0.2863082039911308,
      "grad_norm": 1.0317516326904297,
      "learning_rate": 8.113557387133427e-06,
      "loss": 0.6494,
      "step": 1033
    },
    {
      "epoch": 0.2865853658536585,
      "grad_norm": 1.0504487752914429,
      "learning_rate": 8.110149688390591e-06,
      "loss": 0.6136,
      "step": 1034
    },
    {
      "epoch": 0.28686252771618626,
      "grad_norm": 0.9815801382064819,
      "learning_rate": 8.106739631625216e-06,
      "loss": 0.6715,
      "step": 1035
    },
    {
      "epoch": 0.28713968957871394,
      "grad_norm": 1.0195338726043701,
      "learning_rate": 8.103327219422706e-06,
      "loss": 0.6489,
      "step": 1036
    },
    {
      "epoch": 0.2874168514412417,
      "grad_norm": 1.1127164363861084,
      "learning_rate": 8.09991245437025e-06,
      "loss": 0.7299,
      "step": 1037
    },
    {
      "epoch": 0.2876940133037694,
      "grad_norm": 0.9896835684776306,
      "learning_rate": 8.096495339056819e-06,
      "loss": 0.6636,
      "step": 1038
    },
    {
      "epoch": 0.2879711751662971,
      "grad_norm": 0.9863495230674744,
      "learning_rate": 8.093075876073169e-06,
      "loss": 0.6167,
      "step": 1039
    },
    {
      "epoch": 0.28824833702882485,
      "grad_norm": 0.9785789847373962,
      "learning_rate": 8.089654068011834e-06,
      "loss": 0.6276,
      "step": 1040
    },
    {
      "epoch": 0.28852549889135254,
      "grad_norm": 0.9777904748916626,
      "learning_rate": 8.086229917467128e-06,
      "loss": 0.6889,
      "step": 1041
    },
    {
      "epoch": 0.2888026607538803,
      "grad_norm": 1.0210143327713013,
      "learning_rate": 8.082803427035138e-06,
      "loss": 0.6833,
      "step": 1042
    },
    {
      "epoch": 0.28907982261640797,
      "grad_norm": 0.9986875057220459,
      "learning_rate": 8.079374599313728e-06,
      "loss": 0.6772,
      "step": 1043
    },
    {
      "epoch": 0.2893569844789357,
      "grad_norm": 1.0390561819076538,
      "learning_rate": 8.075943436902534e-06,
      "loss": 0.6501,
      "step": 1044
    },
    {
      "epoch": 0.2896341463414634,
      "grad_norm": 0.9936463236808777,
      "learning_rate": 8.072509942402956e-06,
      "loss": 0.7069,
      "step": 1045
    },
    {
      "epoch": 0.28991130820399114,
      "grad_norm": 0.9883036613464355,
      "learning_rate": 8.06907411841817e-06,
      "loss": 0.6479,
      "step": 1046
    },
    {
      "epoch": 0.2901884700665188,
      "grad_norm": 1.5985538959503174,
      "learning_rate": 8.065635967553115e-06,
      "loss": 0.64,
      "step": 1047
    },
    {
      "epoch": 0.29046563192904656,
      "grad_norm": 1.0107734203338623,
      "learning_rate": 8.062195492414495e-06,
      "loss": 0.687,
      "step": 1048
    },
    {
      "epoch": 0.2907427937915743,
      "grad_norm": 1.093672752380371,
      "learning_rate": 8.058752695610772e-06,
      "loss": 0.696,
      "step": 1049
    },
    {
      "epoch": 0.291019955654102,
      "grad_norm": 0.9810365438461304,
      "learning_rate": 8.055307579752177e-06,
      "loss": 0.6329,
      "step": 1050
    },
    {
      "epoch": 0.29129711751662973,
      "grad_norm": 1.0164715051651,
      "learning_rate": 8.05186014745069e-06,
      "loss": 0.6775,
      "step": 1051
    },
    {
      "epoch": 0.2915742793791574,
      "grad_norm": 1.708271861076355,
      "learning_rate": 8.048410401320051e-06,
      "loss": 0.6978,
      "step": 1052
    },
    {
      "epoch": 0.29185144124168516,
      "grad_norm": 1.2873444557189941,
      "learning_rate": 8.044958343975757e-06,
      "loss": 0.6081,
      "step": 1053
    },
    {
      "epoch": 0.29212860310421285,
      "grad_norm": 1.0233101844787598,
      "learning_rate": 8.041503978035052e-06,
      "loss": 0.6379,
      "step": 1054
    },
    {
      "epoch": 0.2924057649667406,
      "grad_norm": 1.328210473060608,
      "learning_rate": 8.038047306116937e-06,
      "loss": 0.6766,
      "step": 1055
    },
    {
      "epoch": 0.2926829268292683,
      "grad_norm": 1.1154030561447144,
      "learning_rate": 8.034588330842151e-06,
      "loss": 0.6922,
      "step": 1056
    },
    {
      "epoch": 0.292960088691796,
      "grad_norm": 0.9995690584182739,
      "learning_rate": 8.031127054833192e-06,
      "loss": 0.6555,
      "step": 1057
    },
    {
      "epoch": 0.2932372505543237,
      "grad_norm": 1.0511906147003174,
      "learning_rate": 8.027663480714291e-06,
      "loss": 0.6675,
      "step": 1058
    },
    {
      "epoch": 0.29351441241685144,
      "grad_norm": 0.9422762989997864,
      "learning_rate": 8.02419761111143e-06,
      "loss": 0.62,
      "step": 1059
    },
    {
      "epoch": 0.29379157427937913,
      "grad_norm": 1.0947083234786987,
      "learning_rate": 8.020729448652326e-06,
      "loss": 0.6567,
      "step": 1060
    },
    {
      "epoch": 0.2940687361419069,
      "grad_norm": 0.9754561185836792,
      "learning_rate": 8.017258995966437e-06,
      "loss": 0.6858,
      "step": 1061
    },
    {
      "epoch": 0.2943458980044346,
      "grad_norm": 0.9609292149543762,
      "learning_rate": 8.013786255684955e-06,
      "loss": 0.6285,
      "step": 1062
    },
    {
      "epoch": 0.2946230598669623,
      "grad_norm": 0.9961211085319519,
      "learning_rate": 8.010311230440807e-06,
      "loss": 0.6722,
      "step": 1063
    },
    {
      "epoch": 0.29490022172949004,
      "grad_norm": 0.9477207064628601,
      "learning_rate": 8.00683392286866e-06,
      "loss": 0.6073,
      "step": 1064
    },
    {
      "epoch": 0.29517738359201773,
      "grad_norm": 1.0454493761062622,
      "learning_rate": 8.003354335604897e-06,
      "loss": 0.7454,
      "step": 1065
    },
    {
      "epoch": 0.29545454545454547,
      "grad_norm": 0.9574244022369385,
      "learning_rate": 7.99987247128764e-06,
      "loss": 0.7112,
      "step": 1066
    },
    {
      "epoch": 0.29573170731707316,
      "grad_norm": 1.1436874866485596,
      "learning_rate": 7.996388332556735e-06,
      "loss": 0.7381,
      "step": 1067
    },
    {
      "epoch": 0.2960088691796009,
      "grad_norm": 1.0004335641860962,
      "learning_rate": 7.992901922053751e-06,
      "loss": 0.6224,
      "step": 1068
    },
    {
      "epoch": 0.2962860310421286,
      "grad_norm": 0.98311448097229,
      "learning_rate": 7.98941324242198e-06,
      "loss": 0.6735,
      "step": 1069
    },
    {
      "epoch": 0.2965631929046563,
      "grad_norm": 0.9864624738693237,
      "learning_rate": 7.985922296306436e-06,
      "loss": 0.6872,
      "step": 1070
    },
    {
      "epoch": 0.296840354767184,
      "grad_norm": 0.9482383131980896,
      "learning_rate": 7.98242908635385e-06,
      "loss": 0.6645,
      "step": 1071
    },
    {
      "epoch": 0.29711751662971175,
      "grad_norm": 1.0025297403335571,
      "learning_rate": 7.978933615212667e-06,
      "loss": 0.6208,
      "step": 1072
    },
    {
      "epoch": 0.2973946784922395,
      "grad_norm": 1.036503791809082,
      "learning_rate": 7.97543588553305e-06,
      "loss": 0.6171,
      "step": 1073
    },
    {
      "epoch": 0.2976718403547672,
      "grad_norm": 0.9877637028694153,
      "learning_rate": 7.971935899966873e-06,
      "loss": 0.6509,
      "step": 1074
    },
    {
      "epoch": 0.2979490022172949,
      "grad_norm": 1.0485590696334839,
      "learning_rate": 7.968433661167723e-06,
      "loss": 0.667,
      "step": 1075
    },
    {
      "epoch": 0.2982261640798226,
      "grad_norm": 1.040682315826416,
      "learning_rate": 7.964929171790889e-06,
      "loss": 0.6705,
      "step": 1076
    },
    {
      "epoch": 0.29850332594235035,
      "grad_norm": 1.0462253093719482,
      "learning_rate": 7.961422434493372e-06,
      "loss": 0.6613,
      "step": 1077
    },
    {
      "epoch": 0.29878048780487804,
      "grad_norm": 1.0131391286849976,
      "learning_rate": 7.957913451933874e-06,
      "loss": 0.6245,
      "step": 1078
    },
    {
      "epoch": 0.2990576496674058,
      "grad_norm": 0.9953332543373108,
      "learning_rate": 7.954402226772804e-06,
      "loss": 0.7033,
      "step": 1079
    },
    {
      "epoch": 0.29933481152993346,
      "grad_norm": 0.9879525899887085,
      "learning_rate": 7.950888761672264e-06,
      "loss": 0.6336,
      "step": 1080
    },
    {
      "epoch": 0.2996119733924612,
      "grad_norm": 0.9953214526176453,
      "learning_rate": 7.947373059296061e-06,
      "loss": 0.5964,
      "step": 1081
    },
    {
      "epoch": 0.2998891352549889,
      "grad_norm": 1.1380144357681274,
      "learning_rate": 7.943855122309695e-06,
      "loss": 0.6979,
      "step": 1082
    },
    {
      "epoch": 0.30016629711751663,
      "grad_norm": 2.2891016006469727,
      "learning_rate": 7.940334953380357e-06,
      "loss": 0.6967,
      "step": 1083
    },
    {
      "epoch": 0.3004434589800443,
      "grad_norm": 1.0065667629241943,
      "learning_rate": 7.93681255517694e-06,
      "loss": 0.606,
      "step": 1084
    },
    {
      "epoch": 0.30072062084257206,
      "grad_norm": 1.0351427793502808,
      "learning_rate": 7.933287930370019e-06,
      "loss": 0.6861,
      "step": 1085
    },
    {
      "epoch": 0.3009977827050998,
      "grad_norm": 1.0012801885604858,
      "learning_rate": 7.929761081631855e-06,
      "loss": 0.6612,
      "step": 1086
    },
    {
      "epoch": 0.3012749445676275,
      "grad_norm": 0.9524402618408203,
      "learning_rate": 7.926232011636404e-06,
      "loss": 0.6338,
      "step": 1087
    },
    {
      "epoch": 0.30155210643015523,
      "grad_norm": 1.4257475137710571,
      "learning_rate": 7.9227007230593e-06,
      "loss": 0.6676,
      "step": 1088
    },
    {
      "epoch": 0.3018292682926829,
      "grad_norm": 1.126801609992981,
      "learning_rate": 7.919167218577862e-06,
      "loss": 0.6802,
      "step": 1089
    },
    {
      "epoch": 0.30210643015521066,
      "grad_norm": 0.9848793745040894,
      "learning_rate": 7.915631500871084e-06,
      "loss": 0.7022,
      "step": 1090
    },
    {
      "epoch": 0.30238359201773835,
      "grad_norm": 1.0821044445037842,
      "learning_rate": 7.912093572619644e-06,
      "loss": 0.6382,
      "step": 1091
    },
    {
      "epoch": 0.3026607538802661,
      "grad_norm": 1.01325523853302,
      "learning_rate": 7.908553436505892e-06,
      "loss": 0.6558,
      "step": 1092
    },
    {
      "epoch": 0.3029379157427938,
      "grad_norm": 1.0849889516830444,
      "learning_rate": 7.905011095213855e-06,
      "loss": 0.7149,
      "step": 1093
    },
    {
      "epoch": 0.3032150776053215,
      "grad_norm": 1.0153639316558838,
      "learning_rate": 7.901466551429229e-06,
      "loss": 0.6419,
      "step": 1094
    },
    {
      "epoch": 0.3034922394678492,
      "grad_norm": 0.9735358357429504,
      "learning_rate": 7.897919807839382e-06,
      "loss": 0.6643,
      "step": 1095
    },
    {
      "epoch": 0.30376940133037694,
      "grad_norm": 0.9817280769348145,
      "learning_rate": 7.894370867133349e-06,
      "loss": 0.6177,
      "step": 1096
    },
    {
      "epoch": 0.30404656319290463,
      "grad_norm": 0.9782675504684448,
      "learning_rate": 7.89081973200183e-06,
      "loss": 0.6446,
      "step": 1097
    },
    {
      "epoch": 0.30432372505543237,
      "grad_norm": 0.9636240005493164,
      "learning_rate": 7.88726640513719e-06,
      "loss": 0.6579,
      "step": 1098
    },
    {
      "epoch": 0.3046008869179601,
      "grad_norm": 1.085845947265625,
      "learning_rate": 7.883710889233456e-06,
      "loss": 0.6982,
      "step": 1099
    },
    {
      "epoch": 0.3048780487804878,
      "grad_norm": 1.0915851593017578,
      "learning_rate": 7.880153186986312e-06,
      "loss": 0.668,
      "step": 1100
    },
    {
      "epoch": 0.30515521064301554,
      "grad_norm": 1.0071191787719727,
      "learning_rate": 7.876593301093104e-06,
      "loss": 0.694,
      "step": 1101
    },
    {
      "epoch": 0.3054323725055432,
      "grad_norm": 1.0277234315872192,
      "learning_rate": 7.873031234252829e-06,
      "loss": 0.6655,
      "step": 1102
    },
    {
      "epoch": 0.30570953436807097,
      "grad_norm": 1.019014835357666,
      "learning_rate": 7.86946698916614e-06,
      "loss": 0.7282,
      "step": 1103
    },
    {
      "epoch": 0.30598669623059865,
      "grad_norm": 0.9821386337280273,
      "learning_rate": 7.86590056853534e-06,
      "loss": 0.6803,
      "step": 1104
    },
    {
      "epoch": 0.3062638580931264,
      "grad_norm": 1.0124874114990234,
      "learning_rate": 7.862331975064389e-06,
      "loss": 0.6701,
      "step": 1105
    },
    {
      "epoch": 0.3065410199556541,
      "grad_norm": 1.125983476638794,
      "learning_rate": 7.858761211458877e-06,
      "loss": 0.698,
      "step": 1106
    },
    {
      "epoch": 0.3068181818181818,
      "grad_norm": 1.00869619846344,
      "learning_rate": 7.855188280426059e-06,
      "loss": 0.6186,
      "step": 1107
    },
    {
      "epoch": 0.3070953436807095,
      "grad_norm": 1.0136815309524536,
      "learning_rate": 7.851613184674821e-06,
      "loss": 0.6441,
      "step": 1108
    },
    {
      "epoch": 0.30737250554323725,
      "grad_norm": 1.0768005847930908,
      "learning_rate": 7.848035926915695e-06,
      "loss": 0.6743,
      "step": 1109
    },
    {
      "epoch": 0.307649667405765,
      "grad_norm": 1.2657219171524048,
      "learning_rate": 7.84445650986085e-06,
      "loss": 0.6779,
      "step": 1110
    },
    {
      "epoch": 0.3079268292682927,
      "grad_norm": 1.0795034170150757,
      "learning_rate": 7.840874936224092e-06,
      "loss": 0.6784,
      "step": 1111
    },
    {
      "epoch": 0.3082039911308204,
      "grad_norm": 0.9717222452163696,
      "learning_rate": 7.837291208720867e-06,
      "loss": 0.6498,
      "step": 1112
    },
    {
      "epoch": 0.3084811529933481,
      "grad_norm": 0.9881088137626648,
      "learning_rate": 7.833705330068244e-06,
      "loss": 0.6655,
      "step": 1113
    },
    {
      "epoch": 0.30875831485587585,
      "grad_norm": 0.9614074230194092,
      "learning_rate": 7.830117302984937e-06,
      "loss": 0.6692,
      "step": 1114
    },
    {
      "epoch": 0.30903547671840353,
      "grad_norm": 1.0697457790374756,
      "learning_rate": 7.826527130191272e-06,
      "loss": 0.7044,
      "step": 1115
    },
    {
      "epoch": 0.3093126385809313,
      "grad_norm": 0.9808500409126282,
      "learning_rate": 7.822934814409218e-06,
      "loss": 0.6835,
      "step": 1116
    },
    {
      "epoch": 0.30958980044345896,
      "grad_norm": 1.0104738473892212,
      "learning_rate": 7.819340358362358e-06,
      "loss": 0.6581,
      "step": 1117
    },
    {
      "epoch": 0.3098669623059867,
      "grad_norm": 1.0733543634414673,
      "learning_rate": 7.8157437647759e-06,
      "loss": 0.6283,
      "step": 1118
    },
    {
      "epoch": 0.3101441241685144,
      "grad_norm": 0.9918462634086609,
      "learning_rate": 7.812145036376678e-06,
      "loss": 0.6567,
      "step": 1119
    },
    {
      "epoch": 0.31042128603104213,
      "grad_norm": 0.9426614046096802,
      "learning_rate": 7.808544175893137e-06,
      "loss": 0.6439,
      "step": 1120
    },
    {
      "epoch": 0.3106984478935698,
      "grad_norm": 1.0060877799987793,
      "learning_rate": 7.804941186055341e-06,
      "loss": 0.6223,
      "step": 1121
    },
    {
      "epoch": 0.31097560975609756,
      "grad_norm": 1.0332056283950806,
      "learning_rate": 7.801336069594974e-06,
      "loss": 0.653,
      "step": 1122
    },
    {
      "epoch": 0.3112527716186253,
      "grad_norm": 0.9826747179031372,
      "learning_rate": 7.797728829245321e-06,
      "loss": 0.6675,
      "step": 1123
    },
    {
      "epoch": 0.311529933481153,
      "grad_norm": 1.0419799089431763,
      "learning_rate": 7.794119467741288e-06,
      "loss": 0.6988,
      "step": 1124
    },
    {
      "epoch": 0.31180709534368073,
      "grad_norm": 1.1100513935089111,
      "learning_rate": 7.790507987819385e-06,
      "loss": 0.6802,
      "step": 1125
    },
    {
      "epoch": 0.3120842572062084,
      "grad_norm": 1.029111623764038,
      "learning_rate": 7.786894392217727e-06,
      "loss": 0.6718,
      "step": 1126
    },
    {
      "epoch": 0.31236141906873616,
      "grad_norm": 0.9914080500602722,
      "learning_rate": 7.783278683676035e-06,
      "loss": 0.625,
      "step": 1127
    },
    {
      "epoch": 0.31263858093126384,
      "grad_norm": 0.9214078187942505,
      "learning_rate": 7.779660864935629e-06,
      "loss": 0.6165,
      "step": 1128
    },
    {
      "epoch": 0.3129157427937916,
      "grad_norm": 1.0645792484283447,
      "learning_rate": 7.776040938739435e-06,
      "loss": 0.6275,
      "step": 1129
    },
    {
      "epoch": 0.31319290465631927,
      "grad_norm": 1.0532695055007935,
      "learning_rate": 7.77241890783197e-06,
      "loss": 0.5952,
      "step": 1130
    },
    {
      "epoch": 0.313470066518847,
      "grad_norm": 1.003064513206482,
      "learning_rate": 7.76879477495935e-06,
      "loss": 0.705,
      "step": 1131
    },
    {
      "epoch": 0.3137472283813747,
      "grad_norm": 0.9500225782394409,
      "learning_rate": 7.765168542869285e-06,
      "loss": 0.6314,
      "step": 1132
    },
    {
      "epoch": 0.31402439024390244,
      "grad_norm": 0.9641050696372986,
      "learning_rate": 7.761540214311073e-06,
      "loss": 0.6763,
      "step": 1133
    },
    {
      "epoch": 0.3143015521064302,
      "grad_norm": 1.0112738609313965,
      "learning_rate": 7.757909792035608e-06,
      "loss": 0.6791,
      "step": 1134
    },
    {
      "epoch": 0.31457871396895787,
      "grad_norm": 0.9570349454879761,
      "learning_rate": 7.754277278795365e-06,
      "loss": 0.6244,
      "step": 1135
    },
    {
      "epoch": 0.3148558758314856,
      "grad_norm": 0.9350433349609375,
      "learning_rate": 7.75064267734441e-06,
      "loss": 0.6789,
      "step": 1136
    },
    {
      "epoch": 0.3151330376940133,
      "grad_norm": 0.9699059128761292,
      "learning_rate": 7.747005990438386e-06,
      "loss": 0.6461,
      "step": 1137
    },
    {
      "epoch": 0.31541019955654104,
      "grad_norm": 0.9459288716316223,
      "learning_rate": 7.743367220834523e-06,
      "loss": 0.6148,
      "step": 1138
    },
    {
      "epoch": 0.3156873614190687,
      "grad_norm": 2.317232608795166,
      "learning_rate": 7.739726371291623e-06,
      "loss": 0.7074,
      "step": 1139
    },
    {
      "epoch": 0.31596452328159647,
      "grad_norm": 1.0124671459197998,
      "learning_rate": 7.736083444570074e-06,
      "loss": 0.6857,
      "step": 1140
    },
    {
      "epoch": 0.31624168514412415,
      "grad_norm": 1.0443675518035889,
      "learning_rate": 7.732438443431834e-06,
      "loss": 0.643,
      "step": 1141
    },
    {
      "epoch": 0.3165188470066519,
      "grad_norm": 1.003424882888794,
      "learning_rate": 7.72879137064043e-06,
      "loss": 0.6654,
      "step": 1142
    },
    {
      "epoch": 0.3167960088691796,
      "grad_norm": 0.9867714643478394,
      "learning_rate": 7.725142228960968e-06,
      "loss": 0.6516,
      "step": 1143
    },
    {
      "epoch": 0.3170731707317073,
      "grad_norm": 0.9995523691177368,
      "learning_rate": 7.721491021160114e-06,
      "loss": 0.6504,
      "step": 1144
    },
    {
      "epoch": 0.317350332594235,
      "grad_norm": 1.0890066623687744,
      "learning_rate": 7.717837750006106e-06,
      "loss": 0.7125,
      "step": 1145
    },
    {
      "epoch": 0.31762749445676275,
      "grad_norm": 1.220260500907898,
      "learning_rate": 7.714182418268748e-06,
      "loss": 0.6372,
      "step": 1146
    },
    {
      "epoch": 0.3179046563192905,
      "grad_norm": 1.0132172107696533,
      "learning_rate": 7.7105250287194e-06,
      "loss": 0.6343,
      "step": 1147
    },
    {
      "epoch": 0.3181818181818182,
      "grad_norm": 1.0059329271316528,
      "learning_rate": 7.706865584130986e-06,
      "loss": 0.6405,
      "step": 1148
    },
    {
      "epoch": 0.3184589800443459,
      "grad_norm": 1.1500821113586426,
      "learning_rate": 7.703204087277989e-06,
      "loss": 0.6845,
      "step": 1149
    },
    {
      "epoch": 0.3187361419068736,
      "grad_norm": 1.0286020040512085,
      "learning_rate": 7.699540540936444e-06,
      "loss": 0.721,
      "step": 1150
    },
    {
      "epoch": 0.31901330376940135,
      "grad_norm": 1.0051214694976807,
      "learning_rate": 7.695874947883945e-06,
      "loss": 0.6355,
      "step": 1151
    },
    {
      "epoch": 0.31929046563192903,
      "grad_norm": 0.9497034549713135,
      "learning_rate": 7.692207310899633e-06,
      "loss": 0.6131,
      "step": 1152
    },
    {
      "epoch": 0.3195676274944568,
      "grad_norm": 1.0475882291793823,
      "learning_rate": 7.688537632764204e-06,
      "loss": 0.6799,
      "step": 1153
    },
    {
      "epoch": 0.31984478935698446,
      "grad_norm": 1.063934087753296,
      "learning_rate": 7.684865916259892e-06,
      "loss": 0.635,
      "step": 1154
    },
    {
      "epoch": 0.3201219512195122,
      "grad_norm": 1.0559049844741821,
      "learning_rate": 7.68119216417049e-06,
      "loss": 0.6494,
      "step": 1155
    },
    {
      "epoch": 0.3203991130820399,
      "grad_norm": 0.9714808464050293,
      "learning_rate": 7.67751637928132e-06,
      "loss": 0.6304,
      "step": 1156
    },
    {
      "epoch": 0.32067627494456763,
      "grad_norm": 1.0343856811523438,
      "learning_rate": 7.673838564379259e-06,
      "loss": 0.6913,
      "step": 1157
    },
    {
      "epoch": 0.32095343680709537,
      "grad_norm": 0.9655337929725647,
      "learning_rate": 7.670158722252714e-06,
      "loss": 0.6465,
      "step": 1158
    },
    {
      "epoch": 0.32123059866962306,
      "grad_norm": 1.0004667043685913,
      "learning_rate": 7.666476855691628e-06,
      "loss": 0.6767,
      "step": 1159
    },
    {
      "epoch": 0.3215077605321508,
      "grad_norm": 1.0500253438949585,
      "learning_rate": 7.662792967487485e-06,
      "loss": 0.67,
      "step": 1160
    },
    {
      "epoch": 0.3217849223946785,
      "grad_norm": 1.6229087114334106,
      "learning_rate": 7.659107060433299e-06,
      "loss": 0.7315,
      "step": 1161
    },
    {
      "epoch": 0.3220620842572062,
      "grad_norm": 1.030405879020691,
      "learning_rate": 7.655419137323614e-06,
      "loss": 0.6391,
      "step": 1162
    },
    {
      "epoch": 0.3223392461197339,
      "grad_norm": 0.9996685981750488,
      "learning_rate": 7.651729200954505e-06,
      "loss": 0.6811,
      "step": 1163
    },
    {
      "epoch": 0.32261640798226165,
      "grad_norm": 3.42040753364563,
      "learning_rate": 7.648037254123569e-06,
      "loss": 0.6482,
      "step": 1164
    },
    {
      "epoch": 0.32289356984478934,
      "grad_norm": 1.1811870336532593,
      "learning_rate": 7.644343299629928e-06,
      "loss": 0.6624,
      "step": 1165
    },
    {
      "epoch": 0.3231707317073171,
      "grad_norm": 0.9915343523025513,
      "learning_rate": 7.640647340274234e-06,
      "loss": 0.6627,
      "step": 1166
    },
    {
      "epoch": 0.32344789356984477,
      "grad_norm": 0.9969204664230347,
      "learning_rate": 7.636949378858647e-06,
      "loss": 0.686,
      "step": 1167
    },
    {
      "epoch": 0.3237250554323725,
      "grad_norm": 1.0520285367965698,
      "learning_rate": 7.633249418186855e-06,
      "loss": 0.6619,
      "step": 1168
    },
    {
      "epoch": 0.3240022172949002,
      "grad_norm": 1.006005883216858,
      "learning_rate": 7.629547461064054e-06,
      "loss": 0.6842,
      "step": 1169
    },
    {
      "epoch": 0.32427937915742794,
      "grad_norm": 0.9916930794715881,
      "learning_rate": 7.62584351029696e-06,
      "loss": 0.6837,
      "step": 1170
    },
    {
      "epoch": 0.3245565410199557,
      "grad_norm": 0.9915592670440674,
      "learning_rate": 7.622137568693797e-06,
      "loss": 0.6857,
      "step": 1171
    },
    {
      "epoch": 0.32483370288248337,
      "grad_norm": 1.0919700860977173,
      "learning_rate": 7.618429639064298e-06,
      "loss": 0.6781,
      "step": 1172
    },
    {
      "epoch": 0.3251108647450111,
      "grad_norm": 0.9511833190917969,
      "learning_rate": 7.614719724219706e-06,
      "loss": 0.6401,
      "step": 1173
    },
    {
      "epoch": 0.3253880266075388,
      "grad_norm": 0.9887614846229553,
      "learning_rate": 7.611007826972766e-06,
      "loss": 0.6163,
      "step": 1174
    },
    {
      "epoch": 0.32566518847006654,
      "grad_norm": 0.9499550461769104,
      "learning_rate": 7.607293950137729e-06,
      "loss": 0.6813,
      "step": 1175
    },
    {
      "epoch": 0.3259423503325942,
      "grad_norm": 1.0149205923080444,
      "learning_rate": 7.603578096530345e-06,
      "loss": 0.6812,
      "step": 1176
    },
    {
      "epoch": 0.32621951219512196,
      "grad_norm": 1.0792814493179321,
      "learning_rate": 7.599860268967863e-06,
      "loss": 0.6689,
      "step": 1177
    },
    {
      "epoch": 0.32649667405764965,
      "grad_norm": 1.014920711517334,
      "learning_rate": 7.596140470269029e-06,
      "loss": 0.6138,
      "step": 1178
    },
    {
      "epoch": 0.3267738359201774,
      "grad_norm": 1.3245211839675903,
      "learning_rate": 7.592418703254083e-06,
      "loss": 0.6283,
      "step": 1179
    },
    {
      "epoch": 0.3270509977827051,
      "grad_norm": 0.9935864806175232,
      "learning_rate": 7.588694970744756e-06,
      "loss": 0.6531,
      "step": 1180
    },
    {
      "epoch": 0.3273281596452328,
      "grad_norm": 0.9791322350502014,
      "learning_rate": 7.584969275564274e-06,
      "loss": 0.6946,
      "step": 1181
    },
    {
      "epoch": 0.3276053215077605,
      "grad_norm": 0.9806080460548401,
      "learning_rate": 7.581241620537345e-06,
      "loss": 0.6376,
      "step": 1182
    },
    {
      "epoch": 0.32788248337028825,
      "grad_norm": 1.1917588710784912,
      "learning_rate": 7.577512008490166e-06,
      "loss": 0.6308,
      "step": 1183
    },
    {
      "epoch": 0.328159645232816,
      "grad_norm": 1.2878074645996094,
      "learning_rate": 7.573780442250419e-06,
      "loss": 0.6337,
      "step": 1184
    },
    {
      "epoch": 0.3284368070953437,
      "grad_norm": 0.9972389936447144,
      "learning_rate": 7.570046924647265e-06,
      "loss": 0.6608,
      "step": 1185
    },
    {
      "epoch": 0.3287139689578714,
      "grad_norm": 1.31184720993042,
      "learning_rate": 7.566311458511345e-06,
      "loss": 0.6838,
      "step": 1186
    },
    {
      "epoch": 0.3289911308203991,
      "grad_norm": 1.0220319032669067,
      "learning_rate": 7.562574046674777e-06,
      "loss": 0.6594,
      "step": 1187
    },
    {
      "epoch": 0.32926829268292684,
      "grad_norm": 1.020687460899353,
      "learning_rate": 7.558834691971159e-06,
      "loss": 0.6597,
      "step": 1188
    },
    {
      "epoch": 0.32954545454545453,
      "grad_norm": 1.2046128511428833,
      "learning_rate": 7.555093397235553e-06,
      "loss": 0.6462,
      "step": 1189
    },
    {
      "epoch": 0.32982261640798227,
      "grad_norm": 0.9723968505859375,
      "learning_rate": 7.5513501653045e-06,
      "loss": 0.6438,
      "step": 1190
    },
    {
      "epoch": 0.33009977827050996,
      "grad_norm": 0.9655552506446838,
      "learning_rate": 7.5476049990160026e-06,
      "loss": 0.6392,
      "step": 1191
    },
    {
      "epoch": 0.3303769401330377,
      "grad_norm": 0.9897040724754333,
      "learning_rate": 7.543857901209539e-06,
      "loss": 0.7016,
      "step": 1192
    },
    {
      "epoch": 0.3306541019955654,
      "grad_norm": 0.9563917517662048,
      "learning_rate": 7.540108874726047e-06,
      "loss": 0.6772,
      "step": 1193
    },
    {
      "epoch": 0.3309312638580931,
      "grad_norm": 1.180972933769226,
      "learning_rate": 7.53635792240792e-06,
      "loss": 0.6194,
      "step": 1194
    },
    {
      "epoch": 0.33120842572062087,
      "grad_norm": 2.2736005783081055,
      "learning_rate": 7.532605047099024e-06,
      "loss": 0.6833,
      "step": 1195
    },
    {
      "epoch": 0.33148558758314856,
      "grad_norm": 0.9830758571624756,
      "learning_rate": 7.528850251644675e-06,
      "loss": 0.6195,
      "step": 1196
    },
    {
      "epoch": 0.3317627494456763,
      "grad_norm": 1.0017986297607422,
      "learning_rate": 7.5250935388916456e-06,
      "loss": 0.6336,
      "step": 1197
    },
    {
      "epoch": 0.332039911308204,
      "grad_norm": 1.0333436727523804,
      "learning_rate": 7.521334911688165e-06,
      "loss": 0.6698,
      "step": 1198
    },
    {
      "epoch": 0.3323170731707317,
      "grad_norm": 1.0713564157485962,
      "learning_rate": 7.5175743728839115e-06,
      "loss": 0.7106,
      "step": 1199
    },
    {
      "epoch": 0.3325942350332594,
      "grad_norm": 1.0985337495803833,
      "learning_rate": 7.51381192533001e-06,
      "loss": 0.6414,
      "step": 1200
    },
    {
      "epoch": 0.33287139689578715,
      "grad_norm": 1.0069489479064941,
      "learning_rate": 7.510047571879042e-06,
      "loss": 0.5922,
      "step": 1201
    },
    {
      "epoch": 0.33314855875831484,
      "grad_norm": 1.0290392637252808,
      "learning_rate": 7.506281315385022e-06,
      "loss": 0.6358,
      "step": 1202
    },
    {
      "epoch": 0.3334257206208426,
      "grad_norm": 1.1097171306610107,
      "learning_rate": 7.502513158703417e-06,
      "loss": 0.717,
      "step": 1203
    },
    {
      "epoch": 0.33370288248337027,
      "grad_norm": 1.0760340690612793,
      "learning_rate": 7.4987431046911284e-06,
      "loss": 0.6745,
      "step": 1204
    },
    {
      "epoch": 0.333980044345898,
      "grad_norm": 1.0111151933670044,
      "learning_rate": 7.494971156206501e-06,
      "loss": 0.6835,
      "step": 1205
    },
    {
      "epoch": 0.3342572062084257,
      "grad_norm": 1.1360201835632324,
      "learning_rate": 7.491197316109311e-06,
      "loss": 0.6587,
      "step": 1206
    },
    {
      "epoch": 0.33453436807095344,
      "grad_norm": 1.0373414754867554,
      "learning_rate": 7.487421587260771e-06,
      "loss": 0.7052,
      "step": 1207
    },
    {
      "epoch": 0.3348115299334812,
      "grad_norm": 0.9397007822990417,
      "learning_rate": 7.483643972523527e-06,
      "loss": 0.6296,
      "step": 1208
    },
    {
      "epoch": 0.33508869179600886,
      "grad_norm": 0.9365803599357605,
      "learning_rate": 7.4798644747616535e-06,
      "loss": 0.646,
      "step": 1209
    },
    {
      "epoch": 0.3353658536585366,
      "grad_norm": 1.1441316604614258,
      "learning_rate": 7.476083096840653e-06,
      "loss": 0.5895,
      "step": 1210
    },
    {
      "epoch": 0.3356430155210643,
      "grad_norm": 1.0175974369049072,
      "learning_rate": 7.472299841627452e-06,
      "loss": 0.6786,
      "step": 1211
    },
    {
      "epoch": 0.33592017738359203,
      "grad_norm": 1.072439432144165,
      "learning_rate": 7.468514711990404e-06,
      "loss": 0.6811,
      "step": 1212
    },
    {
      "epoch": 0.3361973392461197,
      "grad_norm": 0.9615662097930908,
      "learning_rate": 7.464727710799276e-06,
      "loss": 0.6624,
      "step": 1213
    },
    {
      "epoch": 0.33647450110864746,
      "grad_norm": 0.9670886397361755,
      "learning_rate": 7.460938840925264e-06,
      "loss": 0.6293,
      "step": 1214
    },
    {
      "epoch": 0.33675166297117515,
      "grad_norm": 1.241652011871338,
      "learning_rate": 7.457148105240976e-06,
      "loss": 0.6638,
      "step": 1215
    },
    {
      "epoch": 0.3370288248337029,
      "grad_norm": 1.017176866531372,
      "learning_rate": 7.453355506620432e-06,
      "loss": 0.7401,
      "step": 1216
    },
    {
      "epoch": 0.3373059866962306,
      "grad_norm": 0.9744071960449219,
      "learning_rate": 7.449561047939069e-06,
      "loss": 0.6895,
      "step": 1217
    },
    {
      "epoch": 0.3375831485587583,
      "grad_norm": 1.1128416061401367,
      "learning_rate": 7.44576473207373e-06,
      "loss": 0.7139,
      "step": 1218
    },
    {
      "epoch": 0.33786031042128606,
      "grad_norm": 0.9965283274650574,
      "learning_rate": 7.44196656190267e-06,
      "loss": 0.726,
      "step": 1219
    },
    {
      "epoch": 0.33813747228381374,
      "grad_norm": 0.9873096942901611,
      "learning_rate": 7.4381665403055455e-06,
      "loss": 0.7003,
      "step": 1220
    },
    {
      "epoch": 0.3384146341463415,
      "grad_norm": 1.055173635482788,
      "learning_rate": 7.4343646701634234e-06,
      "loss": 0.667,
      "step": 1221
    },
    {
      "epoch": 0.3386917960088692,
      "grad_norm": 0.9397551417350769,
      "learning_rate": 7.430560954358764e-06,
      "loss": 0.626,
      "step": 1222
    },
    {
      "epoch": 0.3389689578713969,
      "grad_norm": 0.9773762226104736,
      "learning_rate": 7.426755395775433e-06,
      "loss": 0.6959,
      "step": 1223
    },
    {
      "epoch": 0.3392461197339246,
      "grad_norm": 0.9731864929199219,
      "learning_rate": 7.422947997298692e-06,
      "loss": 0.6632,
      "step": 1224
    },
    {
      "epoch": 0.33952328159645234,
      "grad_norm": 1.0325366258621216,
      "learning_rate": 7.419138761815195e-06,
      "loss": 0.6925,
      "step": 1225
    },
    {
      "epoch": 0.33980044345898003,
      "grad_norm": 0.999126672744751,
      "learning_rate": 7.4153276922129916e-06,
      "loss": 0.7398,
      "step": 1226
    },
    {
      "epoch": 0.34007760532150777,
      "grad_norm": 1.1250699758529663,
      "learning_rate": 7.41151479138152e-06,
      "loss": 0.6323,
      "step": 1227
    },
    {
      "epoch": 0.34035476718403546,
      "grad_norm": 0.9539690613746643,
      "learning_rate": 7.407700062211609e-06,
      "loss": 0.6353,
      "step": 1228
    },
    {
      "epoch": 0.3406319290465632,
      "grad_norm": 0.9885956645011902,
      "learning_rate": 7.403883507595471e-06,
      "loss": 0.6755,
      "step": 1229
    },
    {
      "epoch": 0.3409090909090909,
      "grad_norm": 0.9885444045066833,
      "learning_rate": 7.400065130426701e-06,
      "loss": 0.6501,
      "step": 1230
    },
    {
      "epoch": 0.3411862527716186,
      "grad_norm": 21.068147659301758,
      "learning_rate": 7.396244933600285e-06,
      "loss": 0.6662,
      "step": 1231
    },
    {
      "epoch": 0.34146341463414637,
      "grad_norm": 1.0699386596679688,
      "learning_rate": 7.3924229200125775e-06,
      "loss": 0.6532,
      "step": 1232
    },
    {
      "epoch": 0.34174057649667405,
      "grad_norm": 1.1311858892440796,
      "learning_rate": 7.388599092561315e-06,
      "loss": 0.7221,
      "step": 1233
    },
    {
      "epoch": 0.3420177383592018,
      "grad_norm": 1.0103546380996704,
      "learning_rate": 7.384773454145611e-06,
      "loss": 0.6701,
      "step": 1234
    },
    {
      "epoch": 0.3422949002217295,
      "grad_norm": 0.9485781192779541,
      "learning_rate": 7.380946007665949e-06,
      "loss": 0.6217,
      "step": 1235
    },
    {
      "epoch": 0.3425720620842572,
      "grad_norm": 0.9890924096107483,
      "learning_rate": 7.377116756024186e-06,
      "loss": 0.6203,
      "step": 1236
    },
    {
      "epoch": 0.3428492239467849,
      "grad_norm": 0.9551647305488586,
      "learning_rate": 7.3732857021235425e-06,
      "loss": 0.6513,
      "step": 1237
    },
    {
      "epoch": 0.34312638580931265,
      "grad_norm": 0.9472162127494812,
      "learning_rate": 7.369452848868613e-06,
      "loss": 0.6758,
      "step": 1238
    },
    {
      "epoch": 0.34340354767184034,
      "grad_norm": 1.880536675453186,
      "learning_rate": 7.365618199165352e-06,
      "loss": 0.6156,
      "step": 1239
    },
    {
      "epoch": 0.3436807095343681,
      "grad_norm": 0.959347128868103,
      "learning_rate": 7.361781755921073e-06,
      "loss": 0.6882,
      "step": 1240
    },
    {
      "epoch": 0.34395787139689576,
      "grad_norm": 0.954792320728302,
      "learning_rate": 7.357943522044456e-06,
      "loss": 0.6163,
      "step": 1241
    },
    {
      "epoch": 0.3442350332594235,
      "grad_norm": 0.9621318578720093,
      "learning_rate": 7.354103500445535e-06,
      "loss": 0.667,
      "step": 1242
    },
    {
      "epoch": 0.3445121951219512,
      "grad_norm": 1.0437779426574707,
      "learning_rate": 7.350261694035696e-06,
      "loss": 0.6745,
      "step": 1243
    },
    {
      "epoch": 0.34478935698447893,
      "grad_norm": 0.9759037494659424,
      "learning_rate": 7.3464181057276864e-06,
      "loss": 0.6452,
      "step": 1244
    },
    {
      "epoch": 0.3450665188470067,
      "grad_norm": 1.0703085660934448,
      "learning_rate": 7.3425727384355975e-06,
      "loss": 0.6277,
      "step": 1245
    },
    {
      "epoch": 0.34534368070953436,
      "grad_norm": 0.9777181148529053,
      "learning_rate": 7.338725595074871e-06,
      "loss": 0.6853,
      "step": 1246
    },
    {
      "epoch": 0.3456208425720621,
      "grad_norm": 1.0352790355682373,
      "learning_rate": 7.3348766785623e-06,
      "loss": 0.7168,
      "step": 1247
    },
    {
      "epoch": 0.3458980044345898,
      "grad_norm": 1.1051297187805176,
      "learning_rate": 7.331025991816012e-06,
      "loss": 0.6263,
      "step": 1248
    },
    {
      "epoch": 0.34617516629711753,
      "grad_norm": 1.0543749332427979,
      "learning_rate": 7.327173537755487e-06,
      "loss": 0.7005,
      "step": 1249
    },
    {
      "epoch": 0.3464523281596452,
      "grad_norm": 1.1085089445114136,
      "learning_rate": 7.3233193193015405e-06,
      "loss": 0.6737,
      "step": 1250
    },
    {
      "epoch": 0.34672949002217296,
      "grad_norm": 1.0004216432571411,
      "learning_rate": 7.319463339376325e-06,
      "loss": 0.6304,
      "step": 1251
    },
    {
      "epoch": 0.34700665188470065,
      "grad_norm": 0.9487253427505493,
      "learning_rate": 7.315605600903326e-06,
      "loss": 0.6192,
      "step": 1252
    },
    {
      "epoch": 0.3472838137472284,
      "grad_norm": 0.9456372857093811,
      "learning_rate": 7.3117461068073715e-06,
      "loss": 0.6958,
      "step": 1253
    },
    {
      "epoch": 0.3475609756097561,
      "grad_norm": 1.0266773700714111,
      "learning_rate": 7.307884860014609e-06,
      "loss": 0.7171,
      "step": 1254
    },
    {
      "epoch": 0.3478381374722838,
      "grad_norm": 0.9623868465423584,
      "learning_rate": 7.304021863452525e-06,
      "loss": 0.6545,
      "step": 1255
    },
    {
      "epoch": 0.34811529933481156,
      "grad_norm": 1.0270359516143799,
      "learning_rate": 7.300157120049925e-06,
      "loss": 0.6822,
      "step": 1256
    },
    {
      "epoch": 0.34839246119733924,
      "grad_norm": 1.1198663711547852,
      "learning_rate": 7.296290632736942e-06,
      "loss": 0.6929,
      "step": 1257
    },
    {
      "epoch": 0.348669623059867,
      "grad_norm": 0.9808056950569153,
      "learning_rate": 7.292422404445034e-06,
      "loss": 0.6591,
      "step": 1258
    },
    {
      "epoch": 0.34894678492239467,
      "grad_norm": 1.2076290845870972,
      "learning_rate": 7.288552438106973e-06,
      "loss": 0.6303,
      "step": 1259
    },
    {
      "epoch": 0.3492239467849224,
      "grad_norm": 0.9628764986991882,
      "learning_rate": 7.2846807366568515e-06,
      "loss": 0.6348,
      "step": 1260
    },
    {
      "epoch": 0.3495011086474501,
      "grad_norm": 1.111904263496399,
      "learning_rate": 7.28080730303008e-06,
      "loss": 0.6858,
      "step": 1261
    },
    {
      "epoch": 0.34977827050997784,
      "grad_norm": 1.0118154287338257,
      "learning_rate": 7.276932140163381e-06,
      "loss": 0.6368,
      "step": 1262
    },
    {
      "epoch": 0.3500554323725055,
      "grad_norm": 0.9061203598976135,
      "learning_rate": 7.273055250994785e-06,
      "loss": 0.5872,
      "step": 1263
    },
    {
      "epoch": 0.35033259423503327,
      "grad_norm": 1.0164272785186768,
      "learning_rate": 7.269176638463633e-06,
      "loss": 0.694,
      "step": 1264
    },
    {
      "epoch": 0.35060975609756095,
      "grad_norm": 0.940666675567627,
      "learning_rate": 7.265296305510573e-06,
      "loss": 0.6586,
      "step": 1265
    },
    {
      "epoch": 0.3508869179600887,
      "grad_norm": 0.9285731315612793,
      "learning_rate": 7.261414255077561e-06,
      "loss": 0.6424,
      "step": 1266
    },
    {
      "epoch": 0.3511640798226164,
      "grad_norm": 0.9869420528411865,
      "learning_rate": 7.257530490107847e-06,
      "loss": 0.6411,
      "step": 1267
    },
    {
      "epoch": 0.3514412416851441,
      "grad_norm": 0.9687947034835815,
      "learning_rate": 7.253645013545989e-06,
      "loss": 0.605,
      "step": 1268
    },
    {
      "epoch": 0.35171840354767187,
      "grad_norm": 0.972424328327179,
      "learning_rate": 7.249757828337838e-06,
      "loss": 0.6535,
      "step": 1269
    },
    {
      "epoch": 0.35199556541019955,
      "grad_norm": 1.2197883129119873,
      "learning_rate": 7.245868937430539e-06,
      "loss": 0.6955,
      "step": 1270
    },
    {
      "epoch": 0.3522727272727273,
      "grad_norm": 1.0794578790664673,
      "learning_rate": 7.2419783437725374e-06,
      "loss": 0.6796,
      "step": 1271
    },
    {
      "epoch": 0.352549889135255,
      "grad_norm": 0.9858092069625854,
      "learning_rate": 7.238086050313563e-06,
      "loss": 0.6441,
      "step": 1272
    },
    {
      "epoch": 0.3528270509977827,
      "grad_norm": 1.2168737649917603,
      "learning_rate": 7.234192060004636e-06,
      "loss": 0.7468,
      "step": 1273
    },
    {
      "epoch": 0.3531042128603104,
      "grad_norm": 1.0040653944015503,
      "learning_rate": 7.230296375798065e-06,
      "loss": 0.6658,
      "step": 1274
    },
    {
      "epoch": 0.35338137472283815,
      "grad_norm": 1.0651823282241821,
      "learning_rate": 7.2263990006474395e-06,
      "loss": 0.6788,
      "step": 1275
    },
    {
      "epoch": 0.35365853658536583,
      "grad_norm": 0.9986994862556458,
      "learning_rate": 7.222499937507632e-06,
      "loss": 0.6629,
      "step": 1276
    },
    {
      "epoch": 0.3539356984478936,
      "grad_norm": 0.9787682294845581,
      "learning_rate": 7.218599189334799e-06,
      "loss": 0.6422,
      "step": 1277
    },
    {
      "epoch": 0.35421286031042126,
      "grad_norm": 0.9902765154838562,
      "learning_rate": 7.214696759086369e-06,
      "loss": 0.6328,
      "step": 1278
    },
    {
      "epoch": 0.354490022172949,
      "grad_norm": 0.9646845459938049,
      "learning_rate": 7.210792649721048e-06,
      "loss": 0.6467,
      "step": 1279
    },
    {
      "epoch": 0.35476718403547675,
      "grad_norm": 0.9937499165534973,
      "learning_rate": 7.206886864198816e-06,
      "loss": 0.676,
      "step": 1280
    },
    {
      "epoch": 0.35504434589800443,
      "grad_norm": 0.9916523098945618,
      "learning_rate": 7.202979405480921e-06,
      "loss": 0.6579,
      "step": 1281
    },
    {
      "epoch": 0.3553215077605322,
      "grad_norm": 0.9660215973854065,
      "learning_rate": 7.199070276529882e-06,
      "loss": 0.7116,
      "step": 1282
    },
    {
      "epoch": 0.35559866962305986,
      "grad_norm": 1.2620660066604614,
      "learning_rate": 7.195159480309487e-06,
      "loss": 0.6115,
      "step": 1283
    },
    {
      "epoch": 0.3558758314855876,
      "grad_norm": 1.0173921585083008,
      "learning_rate": 7.1912470197847805e-06,
      "loss": 0.6799,
      "step": 1284
    },
    {
      "epoch": 0.3561529933481153,
      "grad_norm": 1.0322010517120361,
      "learning_rate": 7.187332897922076e-06,
      "loss": 0.6314,
      "step": 1285
    },
    {
      "epoch": 0.35643015521064303,
      "grad_norm": 0.9895126819610596,
      "learning_rate": 7.183417117688941e-06,
      "loss": 0.7076,
      "step": 1286
    },
    {
      "epoch": 0.3567073170731707,
      "grad_norm": 0.966982901096344,
      "learning_rate": 7.179499682054205e-06,
      "loss": 0.6906,
      "step": 1287
    },
    {
      "epoch": 0.35698447893569846,
      "grad_norm": 1.0499436855316162,
      "learning_rate": 7.175580593987952e-06,
      "loss": 0.6986,
      "step": 1288
    },
    {
      "epoch": 0.35726164079822614,
      "grad_norm": 0.9585891962051392,
      "learning_rate": 7.171659856461512e-06,
      "loss": 0.6725,
      "step": 1289
    },
    {
      "epoch": 0.3575388026607539,
      "grad_norm": 1.9402228593826294,
      "learning_rate": 7.167737472447477e-06,
      "loss": 0.6904,
      "step": 1290
    },
    {
      "epoch": 0.35781596452328157,
      "grad_norm": 1.023492693901062,
      "learning_rate": 7.163813444919679e-06,
      "loss": 0.6614,
      "step": 1291
    },
    {
      "epoch": 0.3580931263858093,
      "grad_norm": 0.9573469758033752,
      "learning_rate": 7.159887776853198e-06,
      "loss": 0.6229,
      "step": 1292
    },
    {
      "epoch": 0.35837028824833705,
      "grad_norm": 0.9239112734794617,
      "learning_rate": 7.155960471224359e-06,
      "loss": 0.6065,
      "step": 1293
    },
    {
      "epoch": 0.35864745011086474,
      "grad_norm": 0.97780442237854,
      "learning_rate": 7.1520315310107256e-06,
      "loss": 0.6766,
      "step": 1294
    },
    {
      "epoch": 0.3589246119733925,
      "grad_norm": 2.2231037616729736,
      "learning_rate": 7.148100959191106e-06,
      "loss": 0.7252,
      "step": 1295
    },
    {
      "epoch": 0.35920177383592017,
      "grad_norm": 1.176999807357788,
      "learning_rate": 7.14416875874554e-06,
      "loss": 0.6609,
      "step": 1296
    },
    {
      "epoch": 0.3594789356984479,
      "grad_norm": 1.039251685142517,
      "learning_rate": 7.140234932655307e-06,
      "loss": 0.6366,
      "step": 1297
    },
    {
      "epoch": 0.3597560975609756,
      "grad_norm": 1.0029577016830444,
      "learning_rate": 7.136299483902912e-06,
      "loss": 0.6358,
      "step": 1298
    },
    {
      "epoch": 0.36003325942350334,
      "grad_norm": 1.2515664100646973,
      "learning_rate": 7.132362415472099e-06,
      "loss": 0.6909,
      "step": 1299
    },
    {
      "epoch": 0.360310421286031,
      "grad_norm": 0.9256406426429749,
      "learning_rate": 7.128423730347833e-06,
      "loss": 0.6469,
      "step": 1300
    },
    {
      "epoch": 0.36058758314855877,
      "grad_norm": 0.9679304957389832,
      "learning_rate": 7.124483431516309e-06,
      "loss": 0.6717,
      "step": 1301
    },
    {
      "epoch": 0.36086474501108645,
      "grad_norm": 0.955081045627594,
      "learning_rate": 7.120541521964943e-06,
      "loss": 0.7175,
      "step": 1302
    },
    {
      "epoch": 0.3611419068736142,
      "grad_norm": 0.955545961856842,
      "learning_rate": 7.116598004682374e-06,
      "loss": 0.6184,
      "step": 1303
    },
    {
      "epoch": 0.3614190687361419,
      "grad_norm": 1.0974160432815552,
      "learning_rate": 7.112652882658456e-06,
      "loss": 0.7073,
      "step": 1304
    },
    {
      "epoch": 0.3616962305986696,
      "grad_norm": 1.266672134399414,
      "learning_rate": 7.108706158884265e-06,
      "loss": 0.6156,
      "step": 1305
    },
    {
      "epoch": 0.36197339246119736,
      "grad_norm": 1.3576197624206543,
      "learning_rate": 7.104757836352089e-06,
      "loss": 0.625,
      "step": 1306
    },
    {
      "epoch": 0.36225055432372505,
      "grad_norm": 0.9495280981063843,
      "learning_rate": 7.100807918055428e-06,
      "loss": 0.6953,
      "step": 1307
    },
    {
      "epoch": 0.3625277161862528,
      "grad_norm": 1.0138309001922607,
      "learning_rate": 7.096856406988991e-06,
      "loss": 0.7051,
      "step": 1308
    },
    {
      "epoch": 0.3628048780487805,
      "grad_norm": 0.9690846800804138,
      "learning_rate": 7.092903306148698e-06,
      "loss": 0.6701,
      "step": 1309
    },
    {
      "epoch": 0.3630820399113082,
      "grad_norm": 0.9900096654891968,
      "learning_rate": 7.088948618531668e-06,
      "loss": 0.6795,
      "step": 1310
    },
    {
      "epoch": 0.3633592017738359,
      "grad_norm": 1.0737160444259644,
      "learning_rate": 7.084992347136229e-06,
      "loss": 0.637,
      "step": 1311
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 1.6111465692520142,
      "learning_rate": 7.081034494961909e-06,
      "loss": 0.6283,
      "step": 1312
    },
    {
      "epoch": 0.36391352549889133,
      "grad_norm": 0.9675025939941406,
      "learning_rate": 7.0770750650094335e-06,
      "loss": 0.6823,
      "step": 1313
    },
    {
      "epoch": 0.3641906873614191,
      "grad_norm": 0.955152153968811,
      "learning_rate": 7.073114060280721e-06,
      "loss": 0.6059,
      "step": 1314
    },
    {
      "epoch": 0.36446784922394676,
      "grad_norm": 0.9653815031051636,
      "learning_rate": 7.069151483778889e-06,
      "loss": 0.634,
      "step": 1315
    },
    {
      "epoch": 0.3647450110864745,
      "grad_norm": 1.0369712114334106,
      "learning_rate": 7.065187338508244e-06,
      "loss": 0.6555,
      "step": 1316
    },
    {
      "epoch": 0.36502217294900224,
      "grad_norm": 1.1086362600326538,
      "learning_rate": 7.061221627474281e-06,
      "loss": 0.7068,
      "step": 1317
    },
    {
      "epoch": 0.36529933481152993,
      "grad_norm": 1.6846973896026611,
      "learning_rate": 7.057254353683687e-06,
      "loss": 0.6466,
      "step": 1318
    },
    {
      "epoch": 0.36557649667405767,
      "grad_norm": 1.1918679475784302,
      "learning_rate": 7.0532855201443275e-06,
      "loss": 0.6067,
      "step": 1319
    },
    {
      "epoch": 0.36585365853658536,
      "grad_norm": 0.9908503890037537,
      "learning_rate": 7.049315129865255e-06,
      "loss": 0.6611,
      "step": 1320
    },
    {
      "epoch": 0.3661308203991131,
      "grad_norm": 1.3801754713058472,
      "learning_rate": 7.045343185856701e-06,
      "loss": 0.6272,
      "step": 1321
    },
    {
      "epoch": 0.3664079822616408,
      "grad_norm": 0.9984681010246277,
      "learning_rate": 7.041369691130072e-06,
      "loss": 0.6167,
      "step": 1322
    },
    {
      "epoch": 0.3666851441241685,
      "grad_norm": 0.9567697644233704,
      "learning_rate": 7.0373946486979575e-06,
      "loss": 0.6876,
      "step": 1323
    },
    {
      "epoch": 0.3669623059866962,
      "grad_norm": 0.9922916293144226,
      "learning_rate": 7.033418061574111e-06,
      "loss": 0.6635,
      "step": 1324
    },
    {
      "epoch": 0.36723946784922396,
      "grad_norm": 3.8554611206054688,
      "learning_rate": 7.029439932773465e-06,
      "loss": 0.6211,
      "step": 1325
    },
    {
      "epoch": 0.36751662971175164,
      "grad_norm": 0.9919523000717163,
      "learning_rate": 7.025460265312117e-06,
      "loss": 0.6733,
      "step": 1326
    },
    {
      "epoch": 0.3677937915742794,
      "grad_norm": 0.9952264428138733,
      "learning_rate": 7.021479062207333e-06,
      "loss": 0.6553,
      "step": 1327
    },
    {
      "epoch": 0.36807095343680707,
      "grad_norm": 0.9605492353439331,
      "learning_rate": 7.01749632647754e-06,
      "loss": 0.7417,
      "step": 1328
    },
    {
      "epoch": 0.3683481152993348,
      "grad_norm": 1.0074713230133057,
      "learning_rate": 7.0135120611423315e-06,
      "loss": 0.7204,
      "step": 1329
    },
    {
      "epoch": 0.36862527716186255,
      "grad_norm": 1.0424383878707886,
      "learning_rate": 7.009526269222456e-06,
      "loss": 0.6866,
      "step": 1330
    },
    {
      "epoch": 0.36890243902439024,
      "grad_norm": 0.9386099576950073,
      "learning_rate": 7.005538953739824e-06,
      "loss": 0.6221,
      "step": 1331
    },
    {
      "epoch": 0.369179600886918,
      "grad_norm": 0.960460901260376,
      "learning_rate": 7.001550117717499e-06,
      "loss": 0.6919,
      "step": 1332
    },
    {
      "epoch": 0.36945676274944567,
      "grad_norm": 0.9206731915473938,
      "learning_rate": 6.9975597641796945e-06,
      "loss": 0.6555,
      "step": 1333
    },
    {
      "epoch": 0.3697339246119734,
      "grad_norm": 1.1137611865997314,
      "learning_rate": 6.993567896151781e-06,
      "loss": 0.5812,
      "step": 1334
    },
    {
      "epoch": 0.3700110864745011,
      "grad_norm": 0.9640247225761414,
      "learning_rate": 6.98957451666027e-06,
      "loss": 0.6724,
      "step": 1335
    },
    {
      "epoch": 0.37028824833702884,
      "grad_norm": 0.9465628862380981,
      "learning_rate": 6.985579628732826e-06,
      "loss": 0.6137,
      "step": 1336
    },
    {
      "epoch": 0.3705654101995565,
      "grad_norm": 0.9890480041503906,
      "learning_rate": 6.981583235398251e-06,
      "loss": 0.7293,
      "step": 1337
    },
    {
      "epoch": 0.37084257206208426,
      "grad_norm": 1.0150998830795288,
      "learning_rate": 6.977585339686492e-06,
      "loss": 0.7199,
      "step": 1338
    },
    {
      "epoch": 0.37111973392461195,
      "grad_norm": 1.0189439058303833,
      "learning_rate": 6.973585944628633e-06,
      "loss": 0.6728,
      "step": 1339
    },
    {
      "epoch": 0.3713968957871397,
      "grad_norm": 0.9699463248252869,
      "learning_rate": 6.9695850532568975e-06,
      "loss": 0.6818,
      "step": 1340
    },
    {
      "epoch": 0.37167405764966743,
      "grad_norm": 1.0160675048828125,
      "learning_rate": 6.965582668604639e-06,
      "loss": 0.6982,
      "step": 1341
    },
    {
      "epoch": 0.3719512195121951,
      "grad_norm": 1.054276466369629,
      "learning_rate": 6.961578793706349e-06,
      "loss": 0.6596,
      "step": 1342
    },
    {
      "epoch": 0.37222838137472286,
      "grad_norm": 1.0114878416061401,
      "learning_rate": 6.957573431597646e-06,
      "loss": 0.6593,
      "step": 1343
    },
    {
      "epoch": 0.37250554323725055,
      "grad_norm": 0.9151564240455627,
      "learning_rate": 6.9535665853152725e-06,
      "loss": 0.7022,
      "step": 1344
    },
    {
      "epoch": 0.3727827050997783,
      "grad_norm": 0.970537543296814,
      "learning_rate": 6.949558257897102e-06,
      "loss": 0.6398,
      "step": 1345
    },
    {
      "epoch": 0.373059866962306,
      "grad_norm": 0.9379668235778809,
      "learning_rate": 6.9455484523821286e-06,
      "loss": 0.6435,
      "step": 1346
    },
    {
      "epoch": 0.3733370288248337,
      "grad_norm": 0.9471784830093384,
      "learning_rate": 6.941537171810466e-06,
      "loss": 0.6795,
      "step": 1347
    },
    {
      "epoch": 0.3736141906873614,
      "grad_norm": 0.9729798436164856,
      "learning_rate": 6.937524419223348e-06,
      "loss": 0.6341,
      "step": 1348
    },
    {
      "epoch": 0.37389135254988914,
      "grad_norm": 1.0149915218353271,
      "learning_rate": 6.933510197663124e-06,
      "loss": 0.6431,
      "step": 1349
    },
    {
      "epoch": 0.37416851441241683,
      "grad_norm": 0.9590326547622681,
      "learning_rate": 6.929494510173255e-06,
      "loss": 0.7318,
      "step": 1350
    },
    {
      "epoch": 0.37444567627494457,
      "grad_norm": 0.9475587606430054,
      "learning_rate": 6.925477359798317e-06,
      "loss": 0.6093,
      "step": 1351
    },
    {
      "epoch": 0.37472283813747226,
      "grad_norm": 1.253885269165039,
      "learning_rate": 6.921458749583991e-06,
      "loss": 0.6961,
      "step": 1352
    },
    {
      "epoch": 0.375,
      "grad_norm": 1.0577948093414307,
      "learning_rate": 6.917438682577072e-06,
      "loss": 0.66,
      "step": 1353
    },
    {
      "epoch": 0.37527716186252774,
      "grad_norm": 0.9437140226364136,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.6573,
      "step": 1354
    },
    {
      "epoch": 0.37555432372505543,
      "grad_norm": 0.9115737676620483,
      "learning_rate": 6.909394190378123e-06,
      "loss": 0.677,
      "step": 1355
    },
    {
      "epoch": 0.37583148558758317,
      "grad_norm": 1.0059001445770264,
      "learning_rate": 6.905369771285189e-06,
      "loss": 0.6809,
      "step": 1356
    },
    {
      "epoch": 0.37610864745011086,
      "grad_norm": 1.0167832374572754,
      "learning_rate": 6.9013439075978405e-06,
      "loss": 0.6678,
      "step": 1357
    },
    {
      "epoch": 0.3763858093126386,
      "grad_norm": 1.052416443824768,
      "learning_rate": 6.8973166023683694e-06,
      "loss": 0.6974,
      "step": 1358
    },
    {
      "epoch": 0.3766629711751663,
      "grad_norm": 1.0128875970840454,
      "learning_rate": 6.893287858650155e-06,
      "loss": 0.666,
      "step": 1359
    },
    {
      "epoch": 0.376940133037694,
      "grad_norm": 1.0004562139511108,
      "learning_rate": 6.889257679497674e-06,
      "loss": 0.6722,
      "step": 1360
    },
    {
      "epoch": 0.3772172949002217,
      "grad_norm": 0.971312940120697,
      "learning_rate": 6.885226067966484e-06,
      "loss": 0.6528,
      "step": 1361
    },
    {
      "epoch": 0.37749445676274945,
      "grad_norm": 1.014846920967102,
      "learning_rate": 6.881193027113235e-06,
      "loss": 0.6675,
      "step": 1362
    },
    {
      "epoch": 0.37777161862527714,
      "grad_norm": 0.9828606843948364,
      "learning_rate": 6.877158559995657e-06,
      "loss": 0.6674,
      "step": 1363
    },
    {
      "epoch": 0.3780487804878049,
      "grad_norm": 0.9550121426582336,
      "learning_rate": 6.873122669672565e-06,
      "loss": 0.6712,
      "step": 1364
    },
    {
      "epoch": 0.37832594235033257,
      "grad_norm": 2.1919875144958496,
      "learning_rate": 6.869085359203844e-06,
      "loss": 0.6518,
      "step": 1365
    },
    {
      "epoch": 0.3786031042128603,
      "grad_norm": 0.9933090806007385,
      "learning_rate": 6.865046631650469e-06,
      "loss": 0.64,
      "step": 1366
    },
    {
      "epoch": 0.37888026607538805,
      "grad_norm": 1.0647354125976562,
      "learning_rate": 6.861006490074478e-06,
      "loss": 0.6249,
      "step": 1367
    },
    {
      "epoch": 0.37915742793791574,
      "grad_norm": 1.0500155687332153,
      "learning_rate": 6.856964937538989e-06,
      "loss": 0.6267,
      "step": 1368
    },
    {
      "epoch": 0.3794345898004435,
      "grad_norm": 1.0033810138702393,
      "learning_rate": 6.8529219771081835e-06,
      "loss": 0.6332,
      "step": 1369
    },
    {
      "epoch": 0.37971175166297116,
      "grad_norm": 1.0702896118164062,
      "learning_rate": 6.848877611847316e-06,
      "loss": 0.6781,
      "step": 1370
    },
    {
      "epoch": 0.3799889135254989,
      "grad_norm": 0.9883443713188171,
      "learning_rate": 6.8448318448227e-06,
      "loss": 0.6268,
      "step": 1371
    },
    {
      "epoch": 0.3802660753880266,
      "grad_norm": 1.0685425996780396,
      "learning_rate": 6.8407846791017174e-06,
      "loss": 0.6685,
      "step": 1372
    },
    {
      "epoch": 0.38054323725055433,
      "grad_norm": 1.0228108167648315,
      "learning_rate": 6.836736117752809e-06,
      "loss": 0.6475,
      "step": 1373
    },
    {
      "epoch": 0.380820399113082,
      "grad_norm": 1.0116190910339355,
      "learning_rate": 6.83268616384547e-06,
      "loss": 0.6673,
      "step": 1374
    },
    {
      "epoch": 0.38109756097560976,
      "grad_norm": 1.0222080945968628,
      "learning_rate": 6.828634820450258e-06,
      "loss": 0.6587,
      "step": 1375
    },
    {
      "epoch": 0.38137472283813745,
      "grad_norm": 1.0798224210739136,
      "learning_rate": 6.824582090638777e-06,
      "loss": 0.6689,
      "step": 1376
    },
    {
      "epoch": 0.3816518847006652,
      "grad_norm": 0.8958001136779785,
      "learning_rate": 6.820527977483688e-06,
      "loss": 0.6427,
      "step": 1377
    },
    {
      "epoch": 0.38192904656319293,
      "grad_norm": 1.0003738403320312,
      "learning_rate": 6.816472484058698e-06,
      "loss": 0.6537,
      "step": 1378
    },
    {
      "epoch": 0.3822062084257206,
      "grad_norm": 1.0307390689849854,
      "learning_rate": 6.81241561343856e-06,
      "loss": 0.6557,
      "step": 1379
    },
    {
      "epoch": 0.38248337028824836,
      "grad_norm": 0.93963223695755,
      "learning_rate": 6.808357368699074e-06,
      "loss": 0.605,
      "step": 1380
    },
    {
      "epoch": 0.38276053215077604,
      "grad_norm": 1.0678964853286743,
      "learning_rate": 6.804297752917077e-06,
      "loss": 0.6266,
      "step": 1381
    },
    {
      "epoch": 0.3830376940133038,
      "grad_norm": 1.0365877151489258,
      "learning_rate": 6.800236769170452e-06,
      "loss": 0.592,
      "step": 1382
    },
    {
      "epoch": 0.3833148558758315,
      "grad_norm": 1.0459866523742676,
      "learning_rate": 6.796174420538112e-06,
      "loss": 0.6987,
      "step": 1383
    },
    {
      "epoch": 0.3835920177383592,
      "grad_norm": 0.9718211889266968,
      "learning_rate": 6.792110710100011e-06,
      "loss": 0.6688,
      "step": 1384
    },
    {
      "epoch": 0.3838691796008869,
      "grad_norm": 0.9408607482910156,
      "learning_rate": 6.788045640937129e-06,
      "loss": 0.6168,
      "step": 1385
    },
    {
      "epoch": 0.38414634146341464,
      "grad_norm": 1.0024921894073486,
      "learning_rate": 6.7839792161314855e-06,
      "loss": 0.6145,
      "step": 1386
    },
    {
      "epoch": 0.38442350332594233,
      "grad_norm": 0.9389835000038147,
      "learning_rate": 6.779911438766117e-06,
      "loss": 0.6746,
      "step": 1387
    },
    {
      "epoch": 0.38470066518847007,
      "grad_norm": 1.1758266687393188,
      "learning_rate": 6.775842311925094e-06,
      "loss": 0.6432,
      "step": 1388
    },
    {
      "epoch": 0.38497782705099776,
      "grad_norm": 0.9761664271354675,
      "learning_rate": 6.771771838693506e-06,
      "loss": 0.6794,
      "step": 1389
    },
    {
      "epoch": 0.3852549889135255,
      "grad_norm": 0.9226049780845642,
      "learning_rate": 6.767700022157464e-06,
      "loss": 0.6211,
      "step": 1390
    },
    {
      "epoch": 0.38553215077605324,
      "grad_norm": 0.9875936508178711,
      "learning_rate": 6.763626865404097e-06,
      "loss": 0.6942,
      "step": 1391
    },
    {
      "epoch": 0.3858093126385809,
      "grad_norm": 1.0599123239517212,
      "learning_rate": 6.75955237152155e-06,
      "loss": 0.6909,
      "step": 1392
    },
    {
      "epoch": 0.38608647450110867,
      "grad_norm": 0.9981021881103516,
      "learning_rate": 6.755476543598986e-06,
      "loss": 0.68,
      "step": 1393
    },
    {
      "epoch": 0.38636363636363635,
      "grad_norm": 1.0852402448654175,
      "learning_rate": 6.751399384726574e-06,
      "loss": 0.6739,
      "step": 1394
    },
    {
      "epoch": 0.3866407982261641,
      "grad_norm": 0.9767455458641052,
      "learning_rate": 6.747320897995493e-06,
      "loss": 0.6966,
      "step": 1395
    },
    {
      "epoch": 0.3869179600886918,
      "grad_norm": 0.9727725386619568,
      "learning_rate": 6.74324108649793e-06,
      "loss": 0.6133,
      "step": 1396
    },
    {
      "epoch": 0.3871951219512195,
      "grad_norm": 0.9729546308517456,
      "learning_rate": 6.739159953327077e-06,
      "loss": 0.6442,
      "step": 1397
    },
    {
      "epoch": 0.3874722838137472,
      "grad_norm": 1.004780888557434,
      "learning_rate": 6.735077501577126e-06,
      "loss": 0.6877,
      "step": 1398
    },
    {
      "epoch": 0.38774944567627495,
      "grad_norm": 1.020516037940979,
      "learning_rate": 6.730993734343271e-06,
      "loss": 0.6627,
      "step": 1399
    },
    {
      "epoch": 0.38802660753880264,
      "grad_norm": 0.9457736015319824,
      "learning_rate": 6.726908654721702e-06,
      "loss": 0.6638,
      "step": 1400
    },
    {
      "epoch": 0.3883037694013304,
      "grad_norm": 0.9722868800163269,
      "learning_rate": 6.722822265809605e-06,
      "loss": 0.6269,
      "step": 1401
    },
    {
      "epoch": 0.3885809312638581,
      "grad_norm": 0.9892644286155701,
      "learning_rate": 6.718734570705156e-06,
      "loss": 0.6731,
      "step": 1402
    },
    {
      "epoch": 0.3888580931263858,
      "grad_norm": 0.9584894776344299,
      "learning_rate": 6.714645572507524e-06,
      "loss": 0.6481,
      "step": 1403
    },
    {
      "epoch": 0.38913525498891355,
      "grad_norm": 0.9501873850822449,
      "learning_rate": 6.710555274316868e-06,
      "loss": 0.6112,
      "step": 1404
    },
    {
      "epoch": 0.38941241685144123,
      "grad_norm": 1.0048390626907349,
      "learning_rate": 6.706463679234325e-06,
      "loss": 0.6694,
      "step": 1405
    },
    {
      "epoch": 0.389689578713969,
      "grad_norm": 1.0848133563995361,
      "learning_rate": 6.702370790362025e-06,
      "loss": 0.6586,
      "step": 1406
    },
    {
      "epoch": 0.38996674057649666,
      "grad_norm": 0.9528453946113586,
      "learning_rate": 6.69827661080307e-06,
      "loss": 0.6713,
      "step": 1407
    },
    {
      "epoch": 0.3902439024390244,
      "grad_norm": 0.9351996779441833,
      "learning_rate": 6.694181143661546e-06,
      "loss": 0.6291,
      "step": 1408
    },
    {
      "epoch": 0.3905210643015521,
      "grad_norm": 0.9548179507255554,
      "learning_rate": 6.690084392042514e-06,
      "loss": 0.6197,
      "step": 1409
    },
    {
      "epoch": 0.39079822616407983,
      "grad_norm": 1.0153478384017944,
      "learning_rate": 6.6859863590520106e-06,
      "loss": 0.6219,
      "step": 1410
    },
    {
      "epoch": 0.3910753880266075,
      "grad_norm": 0.9848963618278503,
      "learning_rate": 6.681887047797037e-06,
      "loss": 0.6103,
      "step": 1411
    },
    {
      "epoch": 0.39135254988913526,
      "grad_norm": 0.9871175289154053,
      "learning_rate": 6.6777864613855725e-06,
      "loss": 0.6107,
      "step": 1412
    },
    {
      "epoch": 0.39162971175166295,
      "grad_norm": 1.0368480682373047,
      "learning_rate": 6.673684602926559e-06,
      "loss": 0.6951,
      "step": 1413
    },
    {
      "epoch": 0.3919068736141907,
      "grad_norm": 0.9579537510871887,
      "learning_rate": 6.6695814755299016e-06,
      "loss": 0.6161,
      "step": 1414
    },
    {
      "epoch": 0.39218403547671843,
      "grad_norm": 0.95173180103302,
      "learning_rate": 6.665477082306469e-06,
      "loss": 0.6212,
      "step": 1415
    },
    {
      "epoch": 0.3924611973392461,
      "grad_norm": 0.9919817447662354,
      "learning_rate": 6.661371426368088e-06,
      "loss": 0.6111,
      "step": 1416
    },
    {
      "epoch": 0.39273835920177386,
      "grad_norm": 1.0151320695877075,
      "learning_rate": 6.6572645108275495e-06,
      "loss": 0.68,
      "step": 1417
    },
    {
      "epoch": 0.39301552106430154,
      "grad_norm": 1.0485632419586182,
      "learning_rate": 6.6531563387985885e-06,
      "loss": 0.6835,
      "step": 1418
    },
    {
      "epoch": 0.3932926829268293,
      "grad_norm": 1.034400463104248,
      "learning_rate": 6.6490469133959e-06,
      "loss": 0.6275,
      "step": 1419
    },
    {
      "epoch": 0.39356984478935697,
      "grad_norm": 1.4372198581695557,
      "learning_rate": 6.644936237735129e-06,
      "loss": 0.7555,
      "step": 1420
    },
    {
      "epoch": 0.3938470066518847,
      "grad_norm": 0.97006756067276,
      "learning_rate": 6.640824314932867e-06,
      "loss": 0.612,
      "step": 1421
    },
    {
      "epoch": 0.3941241685144124,
      "grad_norm": 0.9723441004753113,
      "learning_rate": 6.636711148106647e-06,
      "loss": 0.6583,
      "step": 1422
    },
    {
      "epoch": 0.39440133037694014,
      "grad_norm": 0.9524688720703125,
      "learning_rate": 6.632596740374952e-06,
      "loss": 0.6265,
      "step": 1423
    },
    {
      "epoch": 0.3946784922394678,
      "grad_norm": 1.02532958984375,
      "learning_rate": 6.628481094857205e-06,
      "loss": 0.691,
      "step": 1424
    },
    {
      "epoch": 0.39495565410199557,
      "grad_norm": 1.0523744821548462,
      "learning_rate": 6.624364214673763e-06,
      "loss": 0.6646,
      "step": 1425
    },
    {
      "epoch": 0.39523281596452325,
      "grad_norm": 1.042662262916565,
      "learning_rate": 6.620246102945922e-06,
      "loss": 0.675,
      "step": 1426
    },
    {
      "epoch": 0.395509977827051,
      "grad_norm": 1.0651161670684814,
      "learning_rate": 6.616126762795911e-06,
      "loss": 0.6218,
      "step": 1427
    },
    {
      "epoch": 0.39578713968957874,
      "grad_norm": 0.9835768342018127,
      "learning_rate": 6.612006197346889e-06,
      "loss": 0.6239,
      "step": 1428
    },
    {
      "epoch": 0.3960643015521064,
      "grad_norm": 0.9560927748680115,
      "learning_rate": 6.607884409722948e-06,
      "loss": 0.7234,
      "step": 1429
    },
    {
      "epoch": 0.39634146341463417,
      "grad_norm": 0.9194403886795044,
      "learning_rate": 6.603761403049103e-06,
      "loss": 0.6355,
      "step": 1430
    },
    {
      "epoch": 0.39661862527716185,
      "grad_norm": 1.139859914779663,
      "learning_rate": 6.599637180451295e-06,
      "loss": 0.65,
      "step": 1431
    },
    {
      "epoch": 0.3968957871396896,
      "grad_norm": 1.1439051628112793,
      "learning_rate": 6.595511745056389e-06,
      "loss": 0.6068,
      "step": 1432
    },
    {
      "epoch": 0.3971729490022173,
      "grad_norm": 0.9855517148971558,
      "learning_rate": 6.5913850999921606e-06,
      "loss": 0.6331,
      "step": 1433
    },
    {
      "epoch": 0.397450110864745,
      "grad_norm": 1.3665685653686523,
      "learning_rate": 6.587257248387313e-06,
      "loss": 0.6593,
      "step": 1434
    },
    {
      "epoch": 0.3977272727272727,
      "grad_norm": 0.9791069626808167,
      "learning_rate": 6.583128193371459e-06,
      "loss": 0.6732,
      "step": 1435
    },
    {
      "epoch": 0.39800443458980045,
      "grad_norm": 0.9057217240333557,
      "learning_rate": 6.578997938075126e-06,
      "loss": 0.6787,
      "step": 1436
    },
    {
      "epoch": 0.39828159645232813,
      "grad_norm": 1.0490965843200684,
      "learning_rate": 6.574866485629748e-06,
      "loss": 0.6673,
      "step": 1437
    },
    {
      "epoch": 0.3985587583148559,
      "grad_norm": 1.0069544315338135,
      "learning_rate": 6.5707338391676686e-06,
      "loss": 0.5965,
      "step": 1438
    },
    {
      "epoch": 0.3988359201773836,
      "grad_norm": 0.9066039323806763,
      "learning_rate": 6.566600001822139e-06,
      "loss": 0.6165,
      "step": 1439
    },
    {
      "epoch": 0.3991130820399113,
      "grad_norm": 0.9101569652557373,
      "learning_rate": 6.562464976727309e-06,
      "loss": 0.6085,
      "step": 1440
    },
    {
      "epoch": 0.39939024390243905,
      "grad_norm": 1.0806589126586914,
      "learning_rate": 6.5583287670182335e-06,
      "loss": 0.6991,
      "step": 1441
    },
    {
      "epoch": 0.39966740576496673,
      "grad_norm": 1.0659431219100952,
      "learning_rate": 6.554191375830861e-06,
      "loss": 0.6427,
      "step": 1442
    },
    {
      "epoch": 0.3999445676274945,
      "grad_norm": 1.3432010412216187,
      "learning_rate": 6.550052806302039e-06,
      "loss": 0.6912,
      "step": 1443
    },
    {
      "epoch": 0.40022172949002216,
      "grad_norm": 0.9609830975532532,
      "learning_rate": 6.545913061569505e-06,
      "loss": 0.6267,
      "step": 1444
    },
    {
      "epoch": 0.4004988913525499,
      "grad_norm": 0.9976644515991211,
      "learning_rate": 6.541772144771893e-06,
      "loss": 0.6268,
      "step": 1445
    },
    {
      "epoch": 0.4004988913525499,
      "eval_loss": 0.6487058401107788,
      "eval_runtime": 386.6827,
      "eval_samples_per_second": 8.353,
      "eval_steps_per_second": 1.045,
      "step": 1445
    },
    {
      "epoch": 0.4007760532150776,
      "grad_norm": 1.1064192056655884,
      "learning_rate": 6.537630059048721e-06,
      "loss": 0.6645,
      "step": 1446
    },
    {
      "epoch": 0.40105321507760533,
      "grad_norm": 1.0119853019714355,
      "learning_rate": 6.533486807540394e-06,
      "loss": 0.6707,
      "step": 1447
    },
    {
      "epoch": 0.401330376940133,
      "grad_norm": 0.9314111471176147,
      "learning_rate": 6.5293423933882005e-06,
      "loss": 0.633,
      "step": 1448
    },
    {
      "epoch": 0.40160753880266076,
      "grad_norm": 1.0213507413864136,
      "learning_rate": 6.525196819734314e-06,
      "loss": 0.7161,
      "step": 1449
    },
    {
      "epoch": 0.40188470066518844,
      "grad_norm": 1.0169448852539062,
      "learning_rate": 6.521050089721783e-06,
      "loss": 0.6969,
      "step": 1450
    },
    {
      "epoch": 0.4021618625277162,
      "grad_norm": 0.9786871671676636,
      "learning_rate": 6.516902206494535e-06,
      "loss": 0.6619,
      "step": 1451
    },
    {
      "epoch": 0.4024390243902439,
      "grad_norm": 0.9995800256729126,
      "learning_rate": 6.5127531731973705e-06,
      "loss": 0.6729,
      "step": 1452
    },
    {
      "epoch": 0.4027161862527716,
      "grad_norm": 1.0010552406311035,
      "learning_rate": 6.508602992975963e-06,
      "loss": 0.7314,
      "step": 1453
    },
    {
      "epoch": 0.40299334811529935,
      "grad_norm": 0.9234841465950012,
      "learning_rate": 6.504451668976854e-06,
      "loss": 0.6541,
      "step": 1454
    },
    {
      "epoch": 0.40327050997782704,
      "grad_norm": 0.9077450037002563,
      "learning_rate": 6.500299204347453e-06,
      "loss": 0.6095,
      "step": 1455
    },
    {
      "epoch": 0.4035476718403548,
      "grad_norm": 0.9172946214675903,
      "learning_rate": 6.496145602236039e-06,
      "loss": 0.5968,
      "step": 1456
    },
    {
      "epoch": 0.40382483370288247,
      "grad_norm": 0.9514081478118896,
      "learning_rate": 6.491990865791743e-06,
      "loss": 0.6019,
      "step": 1457
    },
    {
      "epoch": 0.4041019955654102,
      "grad_norm": 1.0396736860275269,
      "learning_rate": 6.487834998164566e-06,
      "loss": 0.6507,
      "step": 1458
    },
    {
      "epoch": 0.4043791574279379,
      "grad_norm": 0.9865654110908508,
      "learning_rate": 6.48367800250536e-06,
      "loss": 0.6186,
      "step": 1459
    },
    {
      "epoch": 0.40465631929046564,
      "grad_norm": 0.9157853722572327,
      "learning_rate": 6.479519881965837e-06,
      "loss": 0.6504,
      "step": 1460
    },
    {
      "epoch": 0.4049334811529933,
      "grad_norm": 0.9600792527198792,
      "learning_rate": 6.475360639698558e-06,
      "loss": 0.6992,
      "step": 1461
    },
    {
      "epoch": 0.40521064301552107,
      "grad_norm": 0.9788382649421692,
      "learning_rate": 6.471200278856939e-06,
      "loss": 0.6534,
      "step": 1462
    },
    {
      "epoch": 0.4054878048780488,
      "grad_norm": 0.974075436592102,
      "learning_rate": 6.467038802595237e-06,
      "loss": 0.6463,
      "step": 1463
    },
    {
      "epoch": 0.4057649667405765,
      "grad_norm": 1.5758569240570068,
      "learning_rate": 6.462876214068563e-06,
      "loss": 0.6058,
      "step": 1464
    },
    {
      "epoch": 0.40604212860310424,
      "grad_norm": 0.9710908532142639,
      "learning_rate": 6.458712516432865e-06,
      "loss": 0.6304,
      "step": 1465
    },
    {
      "epoch": 0.4063192904656319,
      "grad_norm": 1.0405300855636597,
      "learning_rate": 6.4545477128449315e-06,
      "loss": 0.6589,
      "step": 1466
    },
    {
      "epoch": 0.40659645232815966,
      "grad_norm": 1.0399127006530762,
      "learning_rate": 6.4503818064624e-06,
      "loss": 0.6707,
      "step": 1467
    },
    {
      "epoch": 0.40687361419068735,
      "grad_norm": 0.929868221282959,
      "learning_rate": 6.446214800443729e-06,
      "loss": 0.6191,
      "step": 1468
    },
    {
      "epoch": 0.4071507760532151,
      "grad_norm": 0.9916759729385376,
      "learning_rate": 6.44204669794822e-06,
      "loss": 0.6508,
      "step": 1469
    },
    {
      "epoch": 0.4074279379157428,
      "grad_norm": 1.1239665746688843,
      "learning_rate": 6.437877502136004e-06,
      "loss": 0.6684,
      "step": 1470
    },
    {
      "epoch": 0.4077050997782705,
      "grad_norm": 0.9552251100540161,
      "learning_rate": 6.433707216168041e-06,
      "loss": 0.6514,
      "step": 1471
    },
    {
      "epoch": 0.4079822616407982,
      "grad_norm": 0.9652640223503113,
      "learning_rate": 6.429535843206116e-06,
      "loss": 0.6155,
      "step": 1472
    },
    {
      "epoch": 0.40825942350332595,
      "grad_norm": 0.9693674445152283,
      "learning_rate": 6.4253633864128425e-06,
      "loss": 0.6326,
      "step": 1473
    },
    {
      "epoch": 0.40853658536585363,
      "grad_norm": 0.9547419548034668,
      "learning_rate": 6.4211898489516475e-06,
      "loss": 0.6422,
      "step": 1474
    },
    {
      "epoch": 0.4088137472283814,
      "grad_norm": 0.9614960551261902,
      "learning_rate": 6.417015233986786e-06,
      "loss": 0.6834,
      "step": 1475
    },
    {
      "epoch": 0.4090909090909091,
      "grad_norm": 0.9349309802055359,
      "learning_rate": 6.412839544683326e-06,
      "loss": 0.6528,
      "step": 1476
    },
    {
      "epoch": 0.4093680709534368,
      "grad_norm": 2.3739306926727295,
      "learning_rate": 6.408662784207149e-06,
      "loss": 0.666,
      "step": 1477
    },
    {
      "epoch": 0.40964523281596454,
      "grad_norm": 0.9911890625953674,
      "learning_rate": 6.404484955724951e-06,
      "loss": 0.6403,
      "step": 1478
    },
    {
      "epoch": 0.40992239467849223,
      "grad_norm": 0.9537221193313599,
      "learning_rate": 6.400306062404236e-06,
      "loss": 0.5855,
      "step": 1479
    },
    {
      "epoch": 0.41019955654101997,
      "grad_norm": 0.9836280345916748,
      "learning_rate": 6.396126107413316e-06,
      "loss": 0.6317,
      "step": 1480
    },
    {
      "epoch": 0.41047671840354766,
      "grad_norm": 1.080386757850647,
      "learning_rate": 6.391945093921309e-06,
      "loss": 0.6703,
      "step": 1481
    },
    {
      "epoch": 0.4107538802660754,
      "grad_norm": 1.222082257270813,
      "learning_rate": 6.387763025098134e-06,
      "loss": 0.5907,
      "step": 1482
    },
    {
      "epoch": 0.4110310421286031,
      "grad_norm": 0.9887967109680176,
      "learning_rate": 6.38357990411451e-06,
      "loss": 0.6223,
      "step": 1483
    },
    {
      "epoch": 0.4113082039911308,
      "grad_norm": 0.9401822090148926,
      "learning_rate": 6.3793957341419545e-06,
      "loss": 0.63,
      "step": 1484
    },
    {
      "epoch": 0.4115853658536585,
      "grad_norm": 1.0052038431167603,
      "learning_rate": 6.375210518352779e-06,
      "loss": 0.6474,
      "step": 1485
    },
    {
      "epoch": 0.41186252771618626,
      "grad_norm": 0.9514338374137878,
      "learning_rate": 6.371024259920091e-06,
      "loss": 0.6515,
      "step": 1486
    },
    {
      "epoch": 0.41213968957871394,
      "grad_norm": 1.0660734176635742,
      "learning_rate": 6.3668369620177875e-06,
      "loss": 0.6818,
      "step": 1487
    },
    {
      "epoch": 0.4124168514412417,
      "grad_norm": 1.1472899913787842,
      "learning_rate": 6.362648627820549e-06,
      "loss": 0.6967,
      "step": 1488
    },
    {
      "epoch": 0.4126940133037694,
      "grad_norm": 1.0314851999282837,
      "learning_rate": 6.3584592605038484e-06,
      "loss": 0.6278,
      "step": 1489
    },
    {
      "epoch": 0.4129711751662971,
      "grad_norm": 0.9308122396469116,
      "learning_rate": 6.354268863243937e-06,
      "loss": 0.6591,
      "step": 1490
    },
    {
      "epoch": 0.41324833702882485,
      "grad_norm": 0.9771365523338318,
      "learning_rate": 6.35007743921785e-06,
      "loss": 0.6401,
      "step": 1491
    },
    {
      "epoch": 0.41352549889135254,
      "grad_norm": 0.9872000217437744,
      "learning_rate": 6.345884991603398e-06,
      "loss": 0.6458,
      "step": 1492
    },
    {
      "epoch": 0.4138026607538803,
      "grad_norm": 1.1995009183883667,
      "learning_rate": 6.341691523579172e-06,
      "loss": 0.6623,
      "step": 1493
    },
    {
      "epoch": 0.41407982261640797,
      "grad_norm": 1.062201976776123,
      "learning_rate": 6.3374970383245326e-06,
      "loss": 0.6816,
      "step": 1494
    },
    {
      "epoch": 0.4143569844789357,
      "grad_norm": 0.9729212522506714,
      "learning_rate": 6.333301539019613e-06,
      "loss": 0.6654,
      "step": 1495
    },
    {
      "epoch": 0.4146341463414634,
      "grad_norm": 0.9871843457221985,
      "learning_rate": 6.3291050288453144e-06,
      "loss": 0.6401,
      "step": 1496
    },
    {
      "epoch": 0.41491130820399114,
      "grad_norm": 1.031113862991333,
      "learning_rate": 6.32490751098331e-06,
      "loss": 0.6749,
      "step": 1497
    },
    {
      "epoch": 0.4151884700665188,
      "grad_norm": 1.0082614421844482,
      "learning_rate": 6.320708988616027e-06,
      "loss": 0.6522,
      "step": 1498
    },
    {
      "epoch": 0.41546563192904656,
      "grad_norm": 1.014375925064087,
      "learning_rate": 6.316509464926661e-06,
      "loss": 0.6384,
      "step": 1499
    },
    {
      "epoch": 0.4157427937915743,
      "grad_norm": 0.9375274181365967,
      "learning_rate": 6.312308943099168e-06,
      "loss": 0.6479,
      "step": 1500
    },
    {
      "epoch": 0.416019955654102,
      "grad_norm": 0.984343945980072,
      "learning_rate": 6.308107426318254e-06,
      "loss": 0.6413,
      "step": 1501
    },
    {
      "epoch": 0.41629711751662973,
      "grad_norm": 0.9858793020248413,
      "learning_rate": 6.303904917769387e-06,
      "loss": 0.6388,
      "step": 1502
    },
    {
      "epoch": 0.4165742793791574,
      "grad_norm": 0.9806444644927979,
      "learning_rate": 6.299701420638782e-06,
      "loss": 0.6603,
      "step": 1503
    },
    {
      "epoch": 0.41685144124168516,
      "grad_norm": 1.219649076461792,
      "learning_rate": 6.295496938113406e-06,
      "loss": 0.6886,
      "step": 1504
    },
    {
      "epoch": 0.41712860310421285,
      "grad_norm": 0.9614108800888062,
      "learning_rate": 6.291291473380969e-06,
      "loss": 0.6253,
      "step": 1505
    },
    {
      "epoch": 0.4174057649667406,
      "grad_norm": 0.9865434765815735,
      "learning_rate": 6.287085029629931e-06,
      "loss": 0.5943,
      "step": 1506
    },
    {
      "epoch": 0.4176829268292683,
      "grad_norm": 1.1778502464294434,
      "learning_rate": 6.282877610049492e-06,
      "loss": 0.6572,
      "step": 1507
    },
    {
      "epoch": 0.417960088691796,
      "grad_norm": 0.9641556143760681,
      "learning_rate": 6.27866921782959e-06,
      "loss": 0.6082,
      "step": 1508
    },
    {
      "epoch": 0.4182372505543237,
      "grad_norm": 1.045828104019165,
      "learning_rate": 6.274459856160904e-06,
      "loss": 0.6801,
      "step": 1509
    },
    {
      "epoch": 0.41851441241685144,
      "grad_norm": 1.367441177368164,
      "learning_rate": 6.2702495282348465e-06,
      "loss": 0.6269,
      "step": 1510
    },
    {
      "epoch": 0.41879157427937913,
      "grad_norm": 1.0308990478515625,
      "learning_rate": 6.266038237243561e-06,
      "loss": 0.6281,
      "step": 1511
    },
    {
      "epoch": 0.4190687361419069,
      "grad_norm": 0.9780556559562683,
      "learning_rate": 6.261825986379924e-06,
      "loss": 0.6358,
      "step": 1512
    },
    {
      "epoch": 0.4193458980044346,
      "grad_norm": 0.985342264175415,
      "learning_rate": 6.257612778837535e-06,
      "loss": 0.6768,
      "step": 1513
    },
    {
      "epoch": 0.4196230598669623,
      "grad_norm": 0.9815990328788757,
      "learning_rate": 6.2533986178107245e-06,
      "loss": 0.6281,
      "step": 1514
    },
    {
      "epoch": 0.41990022172949004,
      "grad_norm": 0.969113290309906,
      "learning_rate": 6.249183506494544e-06,
      "loss": 0.6302,
      "step": 1515
    },
    {
      "epoch": 0.42017738359201773,
      "grad_norm": 0.9980261325836182,
      "learning_rate": 6.2449674480847634e-06,
      "loss": 0.7093,
      "step": 1516
    },
    {
      "epoch": 0.42045454545454547,
      "grad_norm": 1.010101079940796,
      "learning_rate": 6.240750445777874e-06,
      "loss": 0.608,
      "step": 1517
    },
    {
      "epoch": 0.42073170731707316,
      "grad_norm": 2.62573504447937,
      "learning_rate": 6.236532502771078e-06,
      "loss": 0.6102,
      "step": 1518
    },
    {
      "epoch": 0.4210088691796009,
      "grad_norm": 0.9677010178565979,
      "learning_rate": 6.232313622262297e-06,
      "loss": 0.6002,
      "step": 1519
    },
    {
      "epoch": 0.4212860310421286,
      "grad_norm": 1.0187071561813354,
      "learning_rate": 6.228093807450157e-06,
      "loss": 0.6905,
      "step": 1520
    },
    {
      "epoch": 0.4215631929046563,
      "grad_norm": 0.9771445393562317,
      "learning_rate": 6.223873061533998e-06,
      "loss": 0.6283,
      "step": 1521
    },
    {
      "epoch": 0.421840354767184,
      "grad_norm": 0.8986212015151978,
      "learning_rate": 6.219651387713863e-06,
      "loss": 0.6,
      "step": 1522
    },
    {
      "epoch": 0.42211751662971175,
      "grad_norm": 0.9228836894035339,
      "learning_rate": 6.215428789190498e-06,
      "loss": 0.5724,
      "step": 1523
    },
    {
      "epoch": 0.4223946784922395,
      "grad_norm": 1.0414414405822754,
      "learning_rate": 6.2112052691653504e-06,
      "loss": 0.6345,
      "step": 1524
    },
    {
      "epoch": 0.4226718403547672,
      "grad_norm": 0.9479822516441345,
      "learning_rate": 6.206980830840568e-06,
      "loss": 0.6173,
      "step": 1525
    },
    {
      "epoch": 0.4229490022172949,
      "grad_norm": 0.9595789313316345,
      "learning_rate": 6.202755477418994e-06,
      "loss": 0.675,
      "step": 1526
    },
    {
      "epoch": 0.4232261640798226,
      "grad_norm": 0.9980440735816956,
      "learning_rate": 6.198529212104164e-06,
      "loss": 0.6508,
      "step": 1527
    },
    {
      "epoch": 0.42350332594235035,
      "grad_norm": 0.9666949510574341,
      "learning_rate": 6.1943020381003075e-06,
      "loss": 0.63,
      "step": 1528
    },
    {
      "epoch": 0.42378048780487804,
      "grad_norm": 0.9362856149673462,
      "learning_rate": 6.19007395861234e-06,
      "loss": 0.6576,
      "step": 1529
    },
    {
      "epoch": 0.4240576496674058,
      "grad_norm": 0.9785576462745667,
      "learning_rate": 6.185844976845866e-06,
      "loss": 0.6798,
      "step": 1530
    },
    {
      "epoch": 0.42433481152993346,
      "grad_norm": 1.0019989013671875,
      "learning_rate": 6.181615096007172e-06,
      "loss": 0.6556,
      "step": 1531
    },
    {
      "epoch": 0.4246119733924612,
      "grad_norm": 0.9336652159690857,
      "learning_rate": 6.1773843193032275e-06,
      "loss": 0.6241,
      "step": 1532
    },
    {
      "epoch": 0.4248891352549889,
      "grad_norm": 1.0282363891601562,
      "learning_rate": 6.173152649941683e-06,
      "loss": 0.6724,
      "step": 1533
    },
    {
      "epoch": 0.42516629711751663,
      "grad_norm": 0.9802669286727905,
      "learning_rate": 6.168920091130861e-06,
      "loss": 0.7078,
      "step": 1534
    },
    {
      "epoch": 0.4254434589800443,
      "grad_norm": 0.9307838082313538,
      "learning_rate": 6.164686646079763e-06,
      "loss": 0.5936,
      "step": 1535
    },
    {
      "epoch": 0.42572062084257206,
      "grad_norm": 0.9869518876075745,
      "learning_rate": 6.160452317998058e-06,
      "loss": 0.6364,
      "step": 1536
    },
    {
      "epoch": 0.4259977827050998,
      "grad_norm": 0.9854388236999512,
      "learning_rate": 6.156217110096089e-06,
      "loss": 0.6522,
      "step": 1537
    },
    {
      "epoch": 0.4262749445676275,
      "grad_norm": 1.0097907781600952,
      "learning_rate": 6.151981025584864e-06,
      "loss": 0.616,
      "step": 1538
    },
    {
      "epoch": 0.42655210643015523,
      "grad_norm": 1.1431684494018555,
      "learning_rate": 6.147744067676057e-06,
      "loss": 0.6118,
      "step": 1539
    },
    {
      "epoch": 0.4268292682926829,
      "grad_norm": 0.965347409248352,
      "learning_rate": 6.143506239582e-06,
      "loss": 0.655,
      "step": 1540
    },
    {
      "epoch": 0.42710643015521066,
      "grad_norm": 0.9409809708595276,
      "learning_rate": 6.139267544515689e-06,
      "loss": 0.5828,
      "step": 1541
    },
    {
      "epoch": 0.42738359201773835,
      "grad_norm": 0.9613562226295471,
      "learning_rate": 6.135027985690775e-06,
      "loss": 0.6964,
      "step": 1542
    },
    {
      "epoch": 0.4276607538802661,
      "grad_norm": 0.9736526608467102,
      "learning_rate": 6.130787566321567e-06,
      "loss": 0.6852,
      "step": 1543
    },
    {
      "epoch": 0.4279379157427938,
      "grad_norm": 0.9558885097503662,
      "learning_rate": 6.126546289623021e-06,
      "loss": 0.638,
      "step": 1544
    },
    {
      "epoch": 0.4282150776053215,
      "grad_norm": 0.9968909025192261,
      "learning_rate": 6.12230415881075e-06,
      "loss": 0.6274,
      "step": 1545
    },
    {
      "epoch": 0.4284922394678492,
      "grad_norm": 0.9609183669090271,
      "learning_rate": 6.118061177101008e-06,
      "loss": 0.651,
      "step": 1546
    },
    {
      "epoch": 0.42876940133037694,
      "grad_norm": 0.937053918838501,
      "learning_rate": 6.113817347710698e-06,
      "loss": 0.6328,
      "step": 1547
    },
    {
      "epoch": 0.42904656319290463,
      "grad_norm": 0.9860870838165283,
      "learning_rate": 6.109572673857363e-06,
      "loss": 0.6594,
      "step": 1548
    },
    {
      "epoch": 0.42932372505543237,
      "grad_norm": 0.9281561970710754,
      "learning_rate": 6.105327158759191e-06,
      "loss": 0.6232,
      "step": 1549
    },
    {
      "epoch": 0.4296008869179601,
      "grad_norm": 3.1716020107269287,
      "learning_rate": 6.101080805635e-06,
      "loss": 0.6948,
      "step": 1550
    },
    {
      "epoch": 0.4298780487804878,
      "grad_norm": 0.9358204007148743,
      "learning_rate": 6.096833617704254e-06,
      "loss": 0.6406,
      "step": 1551
    },
    {
      "epoch": 0.43015521064301554,
      "grad_norm": 1.0352239608764648,
      "learning_rate": 6.09258559818704e-06,
      "loss": 0.6679,
      "step": 1552
    },
    {
      "epoch": 0.4304323725055432,
      "grad_norm": 0.9457054734230042,
      "learning_rate": 6.088336750304078e-06,
      "loss": 0.6633,
      "step": 1553
    },
    {
      "epoch": 0.43070953436807097,
      "grad_norm": 0.9639077186584473,
      "learning_rate": 6.0840870772767235e-06,
      "loss": 0.6489,
      "step": 1554
    },
    {
      "epoch": 0.43098669623059865,
      "grad_norm": 0.9737128019332886,
      "learning_rate": 6.079836582326946e-06,
      "loss": 0.6804,
      "step": 1555
    },
    {
      "epoch": 0.4312638580931264,
      "grad_norm": 0.9784910678863525,
      "learning_rate": 6.0755852686773445e-06,
      "loss": 0.6247,
      "step": 1556
    },
    {
      "epoch": 0.4315410199556541,
      "grad_norm": 6.636213779449463,
      "learning_rate": 6.071333139551141e-06,
      "loss": 0.6511,
      "step": 1557
    },
    {
      "epoch": 0.4318181818181818,
      "grad_norm": 1.287584662437439,
      "learning_rate": 6.067080198172169e-06,
      "loss": 0.6547,
      "step": 1558
    },
    {
      "epoch": 0.4320953436807095,
      "grad_norm": 0.9574753046035767,
      "learning_rate": 6.062826447764883e-06,
      "loss": 0.6874,
      "step": 1559
    },
    {
      "epoch": 0.43237250554323725,
      "grad_norm": 1.0471415519714355,
      "learning_rate": 6.058571891554352e-06,
      "loss": 0.6497,
      "step": 1560
    },
    {
      "epoch": 0.432649667405765,
      "grad_norm": 0.9488391876220703,
      "learning_rate": 6.054316532766249e-06,
      "loss": 0.6023,
      "step": 1561
    },
    {
      "epoch": 0.4329268292682927,
      "grad_norm": 0.9737764000892639,
      "learning_rate": 6.050060374626862e-06,
      "loss": 0.6526,
      "step": 1562
    },
    {
      "epoch": 0.4332039911308204,
      "grad_norm": 1.17532217502594,
      "learning_rate": 6.045803420363085e-06,
      "loss": 0.6321,
      "step": 1563
    },
    {
      "epoch": 0.4334811529933481,
      "grad_norm": 1.8308554887771606,
      "learning_rate": 6.041545673202408e-06,
      "loss": 0.5916,
      "step": 1564
    },
    {
      "epoch": 0.43375831485587585,
      "grad_norm": 1.0375925302505493,
      "learning_rate": 6.037287136372935e-06,
      "loss": 0.6188,
      "step": 1565
    },
    {
      "epoch": 0.43403547671840353,
      "grad_norm": 1.0579415559768677,
      "learning_rate": 6.033027813103354e-06,
      "loss": 0.6846,
      "step": 1566
    },
    {
      "epoch": 0.4343126385809313,
      "grad_norm": 0.9687324166297913,
      "learning_rate": 6.02876770662296e-06,
      "loss": 0.6492,
      "step": 1567
    },
    {
      "epoch": 0.43458980044345896,
      "grad_norm": 2.991565227508545,
      "learning_rate": 6.0245068201616396e-06,
      "loss": 0.6298,
      "step": 1568
    },
    {
      "epoch": 0.4348669623059867,
      "grad_norm": 1.3269773721694946,
      "learning_rate": 6.020245156949867e-06,
      "loss": 0.6821,
      "step": 1569
    },
    {
      "epoch": 0.4351441241685144,
      "grad_norm": 1.0441876649856567,
      "learning_rate": 6.01598272021871e-06,
      "loss": 0.643,
      "step": 1570
    },
    {
      "epoch": 0.43542128603104213,
      "grad_norm": 1.013378381729126,
      "learning_rate": 6.0117195131998205e-06,
      "loss": 0.6542,
      "step": 1571
    },
    {
      "epoch": 0.4356984478935698,
      "grad_norm": 1.1963896751403809,
      "learning_rate": 6.007455539125432e-06,
      "loss": 0.6716,
      "step": 1572
    },
    {
      "epoch": 0.43597560975609756,
      "grad_norm": 1.0005234479904175,
      "learning_rate": 6.003190801228367e-06,
      "loss": 0.6357,
      "step": 1573
    },
    {
      "epoch": 0.4362527716186253,
      "grad_norm": 0.9952285289764404,
      "learning_rate": 5.998925302742017e-06,
      "loss": 0.6658,
      "step": 1574
    },
    {
      "epoch": 0.436529933481153,
      "grad_norm": 0.8943706154823303,
      "learning_rate": 5.994659046900357e-06,
      "loss": 0.5941,
      "step": 1575
    },
    {
      "epoch": 0.43680709534368073,
      "grad_norm": 0.9476959109306335,
      "learning_rate": 5.99039203693794e-06,
      "loss": 0.5939,
      "step": 1576
    },
    {
      "epoch": 0.4370842572062084,
      "grad_norm": 0.9122312068939209,
      "learning_rate": 5.9861242760898775e-06,
      "loss": 0.6935,
      "step": 1577
    },
    {
      "epoch": 0.43736141906873616,
      "grad_norm": 0.9845150709152222,
      "learning_rate": 5.981855767591862e-06,
      "loss": 0.6755,
      "step": 1578
    },
    {
      "epoch": 0.43763858093126384,
      "grad_norm": 0.9237679243087769,
      "learning_rate": 5.977586514680151e-06,
      "loss": 0.6188,
      "step": 1579
    },
    {
      "epoch": 0.4379157427937916,
      "grad_norm": 0.9258509874343872,
      "learning_rate": 5.973316520591561e-06,
      "loss": 0.5885,
      "step": 1580
    },
    {
      "epoch": 0.43819290465631927,
      "grad_norm": 1.7372828722000122,
      "learning_rate": 5.9690457885634756e-06,
      "loss": 0.6308,
      "step": 1581
    },
    {
      "epoch": 0.438470066518847,
      "grad_norm": 1.028012990951538,
      "learning_rate": 5.964774321833835e-06,
      "loss": 0.6396,
      "step": 1582
    },
    {
      "epoch": 0.4387472283813747,
      "grad_norm": 1.0539612770080566,
      "learning_rate": 5.960502123641139e-06,
      "loss": 0.61,
      "step": 1583
    },
    {
      "epoch": 0.43902439024390244,
      "grad_norm": 3.2485971450805664,
      "learning_rate": 5.95622919722444e-06,
      "loss": 0.6327,
      "step": 1584
    },
    {
      "epoch": 0.4393015521064302,
      "grad_norm": 0.9735034704208374,
      "learning_rate": 5.951955545823342e-06,
      "loss": 0.6121,
      "step": 1585
    },
    {
      "epoch": 0.43957871396895787,
      "grad_norm": 1.0430376529693604,
      "learning_rate": 5.947681172678001e-06,
      "loss": 0.6263,
      "step": 1586
    },
    {
      "epoch": 0.4398558758314856,
      "grad_norm": 0.9806467294692993,
      "learning_rate": 5.943406081029122e-06,
      "loss": 0.6522,
      "step": 1587
    },
    {
      "epoch": 0.4401330376940133,
      "grad_norm": 0.9362474083900452,
      "learning_rate": 5.939130274117945e-06,
      "loss": 0.5944,
      "step": 1588
    },
    {
      "epoch": 0.44041019955654104,
      "grad_norm": 0.9484013915061951,
      "learning_rate": 5.934853755186264e-06,
      "loss": 0.6266,
      "step": 1589
    },
    {
      "epoch": 0.4406873614190687,
      "grad_norm": 1.6190098524093628,
      "learning_rate": 5.930576527476406e-06,
      "loss": 0.6253,
      "step": 1590
    },
    {
      "epoch": 0.44096452328159647,
      "grad_norm": 0.9738261699676514,
      "learning_rate": 5.9262985942312376e-06,
      "loss": 0.6323,
      "step": 1591
    },
    {
      "epoch": 0.44124168514412415,
      "grad_norm": 0.9400619268417358,
      "learning_rate": 5.9220199586941604e-06,
      "loss": 0.6064,
      "step": 1592
    },
    {
      "epoch": 0.4415188470066519,
      "grad_norm": 0.9870661497116089,
      "learning_rate": 5.917740624109107e-06,
      "loss": 0.5871,
      "step": 1593
    },
    {
      "epoch": 0.4417960088691796,
      "grad_norm": 0.918742835521698,
      "learning_rate": 5.913460593720541e-06,
      "loss": 0.6248,
      "step": 1594
    },
    {
      "epoch": 0.4420731707317073,
      "grad_norm": 0.9589046835899353,
      "learning_rate": 5.909179870773457e-06,
      "loss": 0.6016,
      "step": 1595
    },
    {
      "epoch": 0.442350332594235,
      "grad_norm": 1.0047661066055298,
      "learning_rate": 5.904898458513365e-06,
      "loss": 0.6615,
      "step": 1596
    },
    {
      "epoch": 0.44262749445676275,
      "grad_norm": 1.6050983667373657,
      "learning_rate": 5.900616360186309e-06,
      "loss": 0.6434,
      "step": 1597
    },
    {
      "epoch": 0.4429046563192905,
      "grad_norm": 0.9696361422538757,
      "learning_rate": 5.896333579038845e-06,
      "loss": 0.6393,
      "step": 1598
    },
    {
      "epoch": 0.4431818181818182,
      "grad_norm": 1.5872431993484497,
      "learning_rate": 5.892050118318049e-06,
      "loss": 0.5921,
      "step": 1599
    },
    {
      "epoch": 0.4434589800443459,
      "grad_norm": 0.9772668480873108,
      "learning_rate": 5.887765981271518e-06,
      "loss": 0.6874,
      "step": 1600
    },
    {
      "epoch": 0.4437361419068736,
      "grad_norm": 0.9553358554840088,
      "learning_rate": 5.883481171147351e-06,
      "loss": 0.5941,
      "step": 1601
    },
    {
      "epoch": 0.44401330376940135,
      "grad_norm": 0.93100506067276,
      "learning_rate": 5.879195691194165e-06,
      "loss": 0.6054,
      "step": 1602
    },
    {
      "epoch": 0.44429046563192903,
      "grad_norm": 6.235424995422363,
      "learning_rate": 5.874909544661084e-06,
      "loss": 0.7142,
      "step": 1603
    },
    {
      "epoch": 0.4445676274944568,
      "grad_norm": 1.044837236404419,
      "learning_rate": 5.8706227347977355e-06,
      "loss": 0.6338,
      "step": 1604
    },
    {
      "epoch": 0.44484478935698446,
      "grad_norm": 1.260089635848999,
      "learning_rate": 5.866335264854248e-06,
      "loss": 0.5923,
      "step": 1605
    },
    {
      "epoch": 0.4451219512195122,
      "grad_norm": 1.025605320930481,
      "learning_rate": 5.86204713808126e-06,
      "loss": 0.6677,
      "step": 1606
    },
    {
      "epoch": 0.4453991130820399,
      "grad_norm": 0.9737890362739563,
      "learning_rate": 5.857758357729892e-06,
      "loss": 0.7022,
      "step": 1607
    },
    {
      "epoch": 0.44567627494456763,
      "grad_norm": 1.294573426246643,
      "learning_rate": 5.853468927051775e-06,
      "loss": 0.6867,
      "step": 1608
    },
    {
      "epoch": 0.44595343680709537,
      "grad_norm": 1.6500320434570312,
      "learning_rate": 5.849178849299027e-06,
      "loss": 0.6711,
      "step": 1609
    },
    {
      "epoch": 0.44623059866962306,
      "grad_norm": 0.9938747882843018,
      "learning_rate": 5.844888127724253e-06,
      "loss": 0.64,
      "step": 1610
    },
    {
      "epoch": 0.4465077605321508,
      "grad_norm": 0.9788873791694641,
      "learning_rate": 5.840596765580555e-06,
      "loss": 0.6513,
      "step": 1611
    },
    {
      "epoch": 0.4467849223946785,
      "grad_norm": 1.059719204902649,
      "learning_rate": 5.83630476612151e-06,
      "loss": 0.677,
      "step": 1612
    },
    {
      "epoch": 0.4470620842572062,
      "grad_norm": 0.9614583253860474,
      "learning_rate": 5.832012132601188e-06,
      "loss": 0.6168,
      "step": 1613
    },
    {
      "epoch": 0.4473392461197339,
      "grad_norm": 1.3544195890426636,
      "learning_rate": 5.827718868274133e-06,
      "loss": 0.6884,
      "step": 1614
    },
    {
      "epoch": 0.44761640798226165,
      "grad_norm": 0.9775740504264832,
      "learning_rate": 5.823424976395373e-06,
      "loss": 0.6425,
      "step": 1615
    },
    {
      "epoch": 0.44789356984478934,
      "grad_norm": 0.9319234490394592,
      "learning_rate": 5.819130460220404e-06,
      "loss": 0.6307,
      "step": 1616
    },
    {
      "epoch": 0.4481707317073171,
      "grad_norm": 1.309570550918579,
      "learning_rate": 5.8148353230052035e-06,
      "loss": 0.6478,
      "step": 1617
    },
    {
      "epoch": 0.44844789356984477,
      "grad_norm": 1.2258167266845703,
      "learning_rate": 5.810539568006213e-06,
      "loss": 0.6765,
      "step": 1618
    },
    {
      "epoch": 0.4487250554323725,
      "grad_norm": 1.0901715755462646,
      "learning_rate": 5.806243198480348e-06,
      "loss": 0.6453,
      "step": 1619
    },
    {
      "epoch": 0.4490022172949002,
      "grad_norm": 0.9257847666740417,
      "learning_rate": 5.801946217684986e-06,
      "loss": 0.6112,
      "step": 1620
    },
    {
      "epoch": 0.44927937915742794,
      "grad_norm": 1.0322872400283813,
      "learning_rate": 5.797648628877971e-06,
      "loss": 0.6509,
      "step": 1621
    },
    {
      "epoch": 0.4495565410199557,
      "grad_norm": 1.7655582427978516,
      "learning_rate": 5.793350435317605e-06,
      "loss": 0.655,
      "step": 1622
    },
    {
      "epoch": 0.44983370288248337,
      "grad_norm": 0.9697898626327515,
      "learning_rate": 5.789051640262648e-06,
      "loss": 0.6051,
      "step": 1623
    },
    {
      "epoch": 0.4501108647450111,
      "grad_norm": 0.9892861247062683,
      "learning_rate": 5.784752246972324e-06,
      "loss": 0.619,
      "step": 1624
    },
    {
      "epoch": 0.4503880266075388,
      "grad_norm": 0.8978703618049622,
      "learning_rate": 5.7804522587062995e-06,
      "loss": 0.6534,
      "step": 1625
    },
    {
      "epoch": 0.45066518847006654,
      "grad_norm": 0.957634449005127,
      "learning_rate": 5.776151678724698e-06,
      "loss": 0.635,
      "step": 1626
    },
    {
      "epoch": 0.4509423503325942,
      "grad_norm": 0.9474959373474121,
      "learning_rate": 5.77185051028809e-06,
      "loss": 0.6144,
      "step": 1627
    },
    {
      "epoch": 0.45121951219512196,
      "grad_norm": 0.9703852534294128,
      "learning_rate": 5.767548756657496e-06,
      "loss": 0.6553,
      "step": 1628
    },
    {
      "epoch": 0.45149667405764965,
      "grad_norm": 0.9233821630477905,
      "learning_rate": 5.763246421094373e-06,
      "loss": 0.6143,
      "step": 1629
    },
    {
      "epoch": 0.4517738359201774,
      "grad_norm": 1.0111219882965088,
      "learning_rate": 5.758943506860627e-06,
      "loss": 0.688,
      "step": 1630
    },
    {
      "epoch": 0.4520509977827051,
      "grad_norm": 1.053755521774292,
      "learning_rate": 5.754640017218596e-06,
      "loss": 0.6567,
      "step": 1631
    },
    {
      "epoch": 0.4523281596452328,
      "grad_norm": 0.9531455039978027,
      "learning_rate": 5.7503359554310575e-06,
      "loss": 0.6166,
      "step": 1632
    },
    {
      "epoch": 0.4526053215077605,
      "grad_norm": 1.0708383321762085,
      "learning_rate": 5.746031324761225e-06,
      "loss": 0.6657,
      "step": 1633
    },
    {
      "epoch": 0.45288248337028825,
      "grad_norm": 1.0728577375411987,
      "learning_rate": 5.741726128472739e-06,
      "loss": 0.6575,
      "step": 1634
    },
    {
      "epoch": 0.453159645232816,
      "grad_norm": 1.1792805194854736,
      "learning_rate": 5.73742036982967e-06,
      "loss": 0.5913,
      "step": 1635
    },
    {
      "epoch": 0.4534368070953437,
      "grad_norm": 0.8835238218307495,
      "learning_rate": 5.733114052096516e-06,
      "loss": 0.5837,
      "step": 1636
    },
    {
      "epoch": 0.4537139689578714,
      "grad_norm": 0.9243692755699158,
      "learning_rate": 5.728807178538199e-06,
      "loss": 0.6018,
      "step": 1637
    },
    {
      "epoch": 0.4539911308203991,
      "grad_norm": 0.999752938747406,
      "learning_rate": 5.724499752420063e-06,
      "loss": 0.6654,
      "step": 1638
    },
    {
      "epoch": 0.45426829268292684,
      "grad_norm": 1.289741039276123,
      "learning_rate": 5.720191777007867e-06,
      "loss": 0.6317,
      "step": 1639
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 0.9556340575218201,
      "learning_rate": 5.7158832555677904e-06,
      "loss": 0.6254,
      "step": 1640
    },
    {
      "epoch": 0.45482261640798227,
      "grad_norm": 1.0347785949707031,
      "learning_rate": 5.711574191366427e-06,
      "loss": 0.696,
      "step": 1641
    },
    {
      "epoch": 0.45509977827050996,
      "grad_norm": 0.9838829636573792,
      "learning_rate": 5.707264587670778e-06,
      "loss": 0.6229,
      "step": 1642
    },
    {
      "epoch": 0.4553769401330377,
      "grad_norm": 0.9321885108947754,
      "learning_rate": 5.702954447748257e-06,
      "loss": 0.6197,
      "step": 1643
    },
    {
      "epoch": 0.4556541019955654,
      "grad_norm": 0.9708073139190674,
      "learning_rate": 5.698643774866684e-06,
      "loss": 0.617,
      "step": 1644
    },
    {
      "epoch": 0.4559312638580931,
      "grad_norm": 0.9317402839660645,
      "learning_rate": 5.69433257229428e-06,
      "loss": 0.6714,
      "step": 1645
    },
    {
      "epoch": 0.45620842572062087,
      "grad_norm": 0.9135407209396362,
      "learning_rate": 5.690020843299672e-06,
      "loss": 0.6044,
      "step": 1646
    },
    {
      "epoch": 0.45648558758314856,
      "grad_norm": 0.9951721429824829,
      "learning_rate": 5.685708591151885e-06,
      "loss": 0.6844,
      "step": 1647
    },
    {
      "epoch": 0.4567627494456763,
      "grad_norm": 0.9301499128341675,
      "learning_rate": 5.681395819120338e-06,
      "loss": 0.6208,
      "step": 1648
    },
    {
      "epoch": 0.457039911308204,
      "grad_norm": 2.524885892868042,
      "learning_rate": 5.677082530474845e-06,
      "loss": 0.6071,
      "step": 1649
    },
    {
      "epoch": 0.4573170731707317,
      "grad_norm": 1.1049906015396118,
      "learning_rate": 5.672768728485616e-06,
      "loss": 0.6829,
      "step": 1650
    },
    {
      "epoch": 0.4575942350332594,
      "grad_norm": 1.5846030712127686,
      "learning_rate": 5.668454416423243e-06,
      "loss": 0.6399,
      "step": 1651
    },
    {
      "epoch": 0.45787139689578715,
      "grad_norm": 0.997001051902771,
      "learning_rate": 5.664139597558711e-06,
      "loss": 0.6329,
      "step": 1652
    },
    {
      "epoch": 0.45814855875831484,
      "grad_norm": 0.9608299732208252,
      "learning_rate": 5.659824275163385e-06,
      "loss": 0.6301,
      "step": 1653
    },
    {
      "epoch": 0.4584257206208426,
      "grad_norm": 0.9963470697402954,
      "learning_rate": 5.655508452509014e-06,
      "loss": 0.6124,
      "step": 1654
    },
    {
      "epoch": 0.45870288248337027,
      "grad_norm": 0.9673656225204468,
      "learning_rate": 5.651192132867727e-06,
      "loss": 0.621,
      "step": 1655
    },
    {
      "epoch": 0.458980044345898,
      "grad_norm": 0.9238991737365723,
      "learning_rate": 5.646875319512026e-06,
      "loss": 0.7125,
      "step": 1656
    },
    {
      "epoch": 0.4592572062084257,
      "grad_norm": 1.0230560302734375,
      "learning_rate": 5.642558015714792e-06,
      "loss": 0.6759,
      "step": 1657
    },
    {
      "epoch": 0.45953436807095344,
      "grad_norm": 1.1747996807098389,
      "learning_rate": 5.638240224749275e-06,
      "loss": 0.6721,
      "step": 1658
    },
    {
      "epoch": 0.4598115299334812,
      "grad_norm": 0.9682770371437073,
      "learning_rate": 5.633921949889095e-06,
      "loss": 0.6692,
      "step": 1659
    },
    {
      "epoch": 0.46008869179600886,
      "grad_norm": 1.0820960998535156,
      "learning_rate": 5.629603194408238e-06,
      "loss": 0.6935,
      "step": 1660
    },
    {
      "epoch": 0.4603658536585366,
      "grad_norm": 0.9443132281303406,
      "learning_rate": 5.6252839615810586e-06,
      "loss": 0.5629,
      "step": 1661
    },
    {
      "epoch": 0.4606430155210643,
      "grad_norm": 0.9565850496292114,
      "learning_rate": 5.620964254682267e-06,
      "loss": 0.6648,
      "step": 1662
    },
    {
      "epoch": 0.46092017738359203,
      "grad_norm": 0.8753352165222168,
      "learning_rate": 5.6166440769869365e-06,
      "loss": 0.6043,
      "step": 1663
    },
    {
      "epoch": 0.4611973392461197,
      "grad_norm": 1.3888369798660278,
      "learning_rate": 5.6123234317704955e-06,
      "loss": 0.6235,
      "step": 1664
    },
    {
      "epoch": 0.46147450110864746,
      "grad_norm": 0.9881001710891724,
      "learning_rate": 5.608002322308731e-06,
      "loss": 0.6677,
      "step": 1665
    },
    {
      "epoch": 0.46175166297117515,
      "grad_norm": 1.038641333580017,
      "learning_rate": 5.603680751877777e-06,
      "loss": 0.6338,
      "step": 1666
    },
    {
      "epoch": 0.4620288248337029,
      "grad_norm": 1.0780448913574219,
      "learning_rate": 5.599358723754121e-06,
      "loss": 0.6732,
      "step": 1667
    },
    {
      "epoch": 0.4623059866962306,
      "grad_norm": 1.0981347560882568,
      "learning_rate": 5.595036241214592e-06,
      "loss": 0.6805,
      "step": 1668
    },
    {
      "epoch": 0.4625831485587583,
      "grad_norm": 0.9794098138809204,
      "learning_rate": 5.59071330753637e-06,
      "loss": 0.7075,
      "step": 1669
    },
    {
      "epoch": 0.46286031042128606,
      "grad_norm": 1.2102551460266113,
      "learning_rate": 5.586389925996972e-06,
      "loss": 0.6415,
      "step": 1670
    },
    {
      "epoch": 0.46313747228381374,
      "grad_norm": 1.0050914287567139,
      "learning_rate": 5.582066099874259e-06,
      "loss": 0.6614,
      "step": 1671
    },
    {
      "epoch": 0.4634146341463415,
      "grad_norm": 1.0354278087615967,
      "learning_rate": 5.577741832446426e-06,
      "loss": 0.6534,
      "step": 1672
    },
    {
      "epoch": 0.4636917960088692,
      "grad_norm": 0.9451271891593933,
      "learning_rate": 5.573417126992004e-06,
      "loss": 0.576,
      "step": 1673
    },
    {
      "epoch": 0.4639689578713969,
      "grad_norm": 0.990343451499939,
      "learning_rate": 5.5690919867898516e-06,
      "loss": 0.5939,
      "step": 1674
    },
    {
      "epoch": 0.4642461197339246,
      "grad_norm": 0.92983478307724,
      "learning_rate": 5.564766415119165e-06,
      "loss": 0.6721,
      "step": 1675
    },
    {
      "epoch": 0.46452328159645234,
      "grad_norm": 0.9351134896278381,
      "learning_rate": 5.560440415259463e-06,
      "loss": 0.6431,
      "step": 1676
    },
    {
      "epoch": 0.46480044345898003,
      "grad_norm": 1.2689239978790283,
      "learning_rate": 5.556113990490587e-06,
      "loss": 0.6647,
      "step": 1677
    },
    {
      "epoch": 0.46507760532150777,
      "grad_norm": 0.934180736541748,
      "learning_rate": 5.551787144092705e-06,
      "loss": 0.6266,
      "step": 1678
    },
    {
      "epoch": 0.46535476718403546,
      "grad_norm": 1.310180425643921,
      "learning_rate": 5.547459879346302e-06,
      "loss": 0.6752,
      "step": 1679
    },
    {
      "epoch": 0.4656319290465632,
      "grad_norm": 0.9672146439552307,
      "learning_rate": 5.5431321995321806e-06,
      "loss": 0.5923,
      "step": 1680
    },
    {
      "epoch": 0.4659090909090909,
      "grad_norm": 0.9535188674926758,
      "learning_rate": 5.538804107931457e-06,
      "loss": 0.6875,
      "step": 1681
    },
    {
      "epoch": 0.4661862527716186,
      "grad_norm": 0.9432002902030945,
      "learning_rate": 5.534475607825566e-06,
      "loss": 0.5858,
      "step": 1682
    },
    {
      "epoch": 0.46646341463414637,
      "grad_norm": 0.915444552898407,
      "learning_rate": 5.530146702496239e-06,
      "loss": 0.6105,
      "step": 1683
    },
    {
      "epoch": 0.46674057649667405,
      "grad_norm": 0.8882535696029663,
      "learning_rate": 5.52581739522553e-06,
      "loss": 0.5913,
      "step": 1684
    },
    {
      "epoch": 0.4670177383592018,
      "grad_norm": 0.8791733384132385,
      "learning_rate": 5.5214876892957866e-06,
      "loss": 0.6623,
      "step": 1685
    },
    {
      "epoch": 0.4672949002217295,
      "grad_norm": 1.1980395317077637,
      "learning_rate": 5.517157587989662e-06,
      "loss": 0.6182,
      "step": 1686
    },
    {
      "epoch": 0.4675720620842572,
      "grad_norm": 1.0265716314315796,
      "learning_rate": 5.512827094590113e-06,
      "loss": 0.5833,
      "step": 1687
    },
    {
      "epoch": 0.4678492239467849,
      "grad_norm": 1.0098729133605957,
      "learning_rate": 5.5084962123803865e-06,
      "loss": 0.6638,
      "step": 1688
    },
    {
      "epoch": 0.46812638580931265,
      "grad_norm": 0.9334395527839661,
      "learning_rate": 5.50416494464403e-06,
      "loss": 0.5985,
      "step": 1689
    },
    {
      "epoch": 0.46840354767184034,
      "grad_norm": 0.9940615892410278,
      "learning_rate": 5.499833294664881e-06,
      "loss": 0.6604,
      "step": 1690
    },
    {
      "epoch": 0.4686807095343681,
      "grad_norm": 0.947023332118988,
      "learning_rate": 5.495501265727067e-06,
      "loss": 0.6305,
      "step": 1691
    },
    {
      "epoch": 0.46895787139689576,
      "grad_norm": 0.9094706177711487,
      "learning_rate": 5.491168861115001e-06,
      "loss": 0.7015,
      "step": 1692
    },
    {
      "epoch": 0.4692350332594235,
      "grad_norm": 0.9953396916389465,
      "learning_rate": 5.486836084113387e-06,
      "loss": 0.6731,
      "step": 1693
    },
    {
      "epoch": 0.4695121951219512,
      "grad_norm": 1.0848197937011719,
      "learning_rate": 5.482502938007202e-06,
      "loss": 0.6537,
      "step": 1694
    },
    {
      "epoch": 0.46978935698447893,
      "grad_norm": 0.9725229740142822,
      "learning_rate": 5.478169426081712e-06,
      "loss": 0.5879,
      "step": 1695
    },
    {
      "epoch": 0.4700665188470067,
      "grad_norm": 0.95648592710495,
      "learning_rate": 5.473835551622454e-06,
      "loss": 0.6417,
      "step": 1696
    },
    {
      "epoch": 0.47034368070953436,
      "grad_norm": 0.9639899134635925,
      "learning_rate": 5.469501317915242e-06,
      "loss": 0.6486,
      "step": 1697
    },
    {
      "epoch": 0.4706208425720621,
      "grad_norm": 0.9836540222167969,
      "learning_rate": 5.465166728246164e-06,
      "loss": 0.5827,
      "step": 1698
    },
    {
      "epoch": 0.4708980044345898,
      "grad_norm": 0.932554304599762,
      "learning_rate": 5.460831785901575e-06,
      "loss": 0.6164,
      "step": 1699
    },
    {
      "epoch": 0.47117516629711753,
      "grad_norm": 0.8863709568977356,
      "learning_rate": 5.456496494168098e-06,
      "loss": 0.6192,
      "step": 1700
    },
    {
      "epoch": 0.4714523281596452,
      "grad_norm": 0.9699694514274597,
      "learning_rate": 5.452160856332624e-06,
      "loss": 0.6486,
      "step": 1701
    },
    {
      "epoch": 0.47172949002217296,
      "grad_norm": 0.9378741383552551,
      "learning_rate": 5.447824875682304e-06,
      "loss": 0.5979,
      "step": 1702
    },
    {
      "epoch": 0.47200665188470065,
      "grad_norm": 1.0518074035644531,
      "learning_rate": 5.443488555504545e-06,
      "loss": 0.6435,
      "step": 1703
    },
    {
      "epoch": 0.4722838137472284,
      "grad_norm": 0.8923550844192505,
      "learning_rate": 5.4391518990870196e-06,
      "loss": 0.6048,
      "step": 1704
    },
    {
      "epoch": 0.4725609756097561,
      "grad_norm": 1.8436543941497803,
      "learning_rate": 5.4348149097176485e-06,
      "loss": 0.6627,
      "step": 1705
    },
    {
      "epoch": 0.4728381374722838,
      "grad_norm": 0.9894528388977051,
      "learning_rate": 5.4304775906846095e-06,
      "loss": 0.6388,
      "step": 1706
    },
    {
      "epoch": 0.47311529933481156,
      "grad_norm": 0.9412456750869751,
      "learning_rate": 5.426139945276327e-06,
      "loss": 0.6059,
      "step": 1707
    },
    {
      "epoch": 0.47339246119733924,
      "grad_norm": 1.0622469186782837,
      "learning_rate": 5.421801976781473e-06,
      "loss": 0.646,
      "step": 1708
    },
    {
      "epoch": 0.473669623059867,
      "grad_norm": 0.968080461025238,
      "learning_rate": 5.417463688488969e-06,
      "loss": 0.5985,
      "step": 1709
    },
    {
      "epoch": 0.47394678492239467,
      "grad_norm": 0.906406581401825,
      "learning_rate": 5.4131250836879715e-06,
      "loss": 0.6136,
      "step": 1710
    },
    {
      "epoch": 0.4742239467849224,
      "grad_norm": 0.925296425819397,
      "learning_rate": 5.408786165667884e-06,
      "loss": 0.6537,
      "step": 1711
    },
    {
      "epoch": 0.4745011086474501,
      "grad_norm": 0.9841566681861877,
      "learning_rate": 5.404446937718343e-06,
      "loss": 0.6735,
      "step": 1712
    },
    {
      "epoch": 0.47477827050997784,
      "grad_norm": 0.9771679639816284,
      "learning_rate": 5.4001074031292225e-06,
      "loss": 0.6382,
      "step": 1713
    },
    {
      "epoch": 0.4750554323725055,
      "grad_norm": 1.2212685346603394,
      "learning_rate": 5.395767565190626e-06,
      "loss": 0.6279,
      "step": 1714
    },
    {
      "epoch": 0.47533259423503327,
      "grad_norm": 0.9823108911514282,
      "learning_rate": 5.391427427192892e-06,
      "loss": 0.6064,
      "step": 1715
    },
    {
      "epoch": 0.47560975609756095,
      "grad_norm": 0.9849187731742859,
      "learning_rate": 5.387086992426579e-06,
      "loss": 0.6365,
      "step": 1716
    },
    {
      "epoch": 0.4758869179600887,
      "grad_norm": 0.9364112019538879,
      "learning_rate": 5.38274626418248e-06,
      "loss": 0.5761,
      "step": 1717
    },
    {
      "epoch": 0.4761640798226164,
      "grad_norm": 0.9436275959014893,
      "learning_rate": 5.3784052457516e-06,
      "loss": 0.6284,
      "step": 1718
    },
    {
      "epoch": 0.4764412416851441,
      "grad_norm": 0.947232186794281,
      "learning_rate": 5.374063940425175e-06,
      "loss": 0.5896,
      "step": 1719
    },
    {
      "epoch": 0.47671840354767187,
      "grad_norm": 0.973252534866333,
      "learning_rate": 5.3697223514946485e-06,
      "loss": 0.5787,
      "step": 1720
    },
    {
      "epoch": 0.47699556541019955,
      "grad_norm": 0.9909771680831909,
      "learning_rate": 5.365380482251685e-06,
      "loss": 0.669,
      "step": 1721
    },
    {
      "epoch": 0.4772727272727273,
      "grad_norm": 1.0703457593917847,
      "learning_rate": 5.361038335988161e-06,
      "loss": 0.6683,
      "step": 1722
    },
    {
      "epoch": 0.477549889135255,
      "grad_norm": 0.9353010058403015,
      "learning_rate": 5.356695915996162e-06,
      "loss": 0.6947,
      "step": 1723
    },
    {
      "epoch": 0.4778270509977827,
      "grad_norm": 0.9475141763687134,
      "learning_rate": 5.35235322556798e-06,
      "loss": 0.6317,
      "step": 1724
    },
    {
      "epoch": 0.4781042128603104,
      "grad_norm": 1.0570001602172852,
      "learning_rate": 5.348010267996113e-06,
      "loss": 0.5947,
      "step": 1725
    },
    {
      "epoch": 0.47838137472283815,
      "grad_norm": 0.935027003288269,
      "learning_rate": 5.343667046573262e-06,
      "loss": 0.6115,
      "step": 1726
    },
    {
      "epoch": 0.47865853658536583,
      "grad_norm": 1.060848355293274,
      "learning_rate": 5.339323564592327e-06,
      "loss": 0.6314,
      "step": 1727
    },
    {
      "epoch": 0.4789356984478936,
      "grad_norm": 0.9244704842567444,
      "learning_rate": 5.334979825346409e-06,
      "loss": 0.6142,
      "step": 1728
    },
    {
      "epoch": 0.47921286031042126,
      "grad_norm": 0.9224184155464172,
      "learning_rate": 5.330635832128796e-06,
      "loss": 0.6174,
      "step": 1729
    },
    {
      "epoch": 0.479490022172949,
      "grad_norm": 0.921455442905426,
      "learning_rate": 5.32629158823298e-06,
      "loss": 0.6121,
      "step": 1730
    },
    {
      "epoch": 0.47976718403547675,
      "grad_norm": 0.9538763761520386,
      "learning_rate": 5.321947096952633e-06,
      "loss": 0.6394,
      "step": 1731
    },
    {
      "epoch": 0.48004434589800443,
      "grad_norm": 1.0409653186798096,
      "learning_rate": 5.317602361581618e-06,
      "loss": 0.669,
      "step": 1732
    },
    {
      "epoch": 0.4803215077605322,
      "grad_norm": 2.6697299480438232,
      "learning_rate": 5.313257385413986e-06,
      "loss": 0.727,
      "step": 1733
    },
    {
      "epoch": 0.48059866962305986,
      "grad_norm": 0.9733155369758606,
      "learning_rate": 5.308912171743967e-06,
      "loss": 0.6127,
      "step": 1734
    },
    {
      "epoch": 0.4808758314855876,
      "grad_norm": 1.069728136062622,
      "learning_rate": 5.3045667238659714e-06,
      "loss": 0.6252,
      "step": 1735
    },
    {
      "epoch": 0.4811529933481153,
      "grad_norm": 0.9509124159812927,
      "learning_rate": 5.300221045074589e-06,
      "loss": 0.6593,
      "step": 1736
    },
    {
      "epoch": 0.48143015521064303,
      "grad_norm": 0.9342741966247559,
      "learning_rate": 5.2958751386645835e-06,
      "loss": 0.644,
      "step": 1737
    },
    {
      "epoch": 0.4817073170731707,
      "grad_norm": 0.9112820625305176,
      "learning_rate": 5.29152900793089e-06,
      "loss": 0.6555,
      "step": 1738
    },
    {
      "epoch": 0.48198447893569846,
      "grad_norm": 1.0964763164520264,
      "learning_rate": 5.287182656168618e-06,
      "loss": 0.6208,
      "step": 1739
    },
    {
      "epoch": 0.48226164079822614,
      "grad_norm": 1.0069471597671509,
      "learning_rate": 5.282836086673037e-06,
      "loss": 0.6492,
      "step": 1740
    },
    {
      "epoch": 0.4825388026607539,
      "grad_norm": 0.9737354516983032,
      "learning_rate": 5.2784893027395876e-06,
      "loss": 0.5934,
      "step": 1741
    },
    {
      "epoch": 0.48281596452328157,
      "grad_norm": 1.1793164014816284,
      "learning_rate": 5.274142307663874e-06,
      "loss": 0.6489,
      "step": 1742
    },
    {
      "epoch": 0.4830931263858093,
      "grad_norm": 1.5903456211090088,
      "learning_rate": 5.269795104741656e-06,
      "loss": 0.643,
      "step": 1743
    },
    {
      "epoch": 0.48337028824833705,
      "grad_norm": 0.9718561768531799,
      "learning_rate": 5.265447697268854e-06,
      "loss": 0.6331,
      "step": 1744
    },
    {
      "epoch": 0.48364745011086474,
      "grad_norm": 1.1706026792526245,
      "learning_rate": 5.2611000885415385e-06,
      "loss": 0.6412,
      "step": 1745
    },
    {
      "epoch": 0.4839246119733925,
      "grad_norm": 0.9517596364021301,
      "learning_rate": 5.256752281855942e-06,
      "loss": 0.5972,
      "step": 1746
    },
    {
      "epoch": 0.48420177383592017,
      "grad_norm": 1.0255533456802368,
      "learning_rate": 5.2524042805084385e-06,
      "loss": 0.619,
      "step": 1747
    },
    {
      "epoch": 0.4844789356984479,
      "grad_norm": 0.9293138384819031,
      "learning_rate": 5.248056087795555e-06,
      "loss": 0.6188,
      "step": 1748
    },
    {
      "epoch": 0.4847560975609756,
      "grad_norm": 0.9461867809295654,
      "learning_rate": 5.243707707013956e-06,
      "loss": 0.6619,
      "step": 1749
    },
    {
      "epoch": 0.48503325942350334,
      "grad_norm": 0.9311886429786682,
      "learning_rate": 5.239359141460461e-06,
      "loss": 0.6307,
      "step": 1750
    },
    {
      "epoch": 0.485310421286031,
      "grad_norm": 0.943748950958252,
      "learning_rate": 5.2350103944320165e-06,
      "loss": 0.5702,
      "step": 1751
    },
    {
      "epoch": 0.48558758314855877,
      "grad_norm": 1.1350374221801758,
      "learning_rate": 5.230661469225715e-06,
      "loss": 0.6971,
      "step": 1752
    },
    {
      "epoch": 0.48586474501108645,
      "grad_norm": 0.8815587162971497,
      "learning_rate": 5.226312369138782e-06,
      "loss": 0.6215,
      "step": 1753
    },
    {
      "epoch": 0.4861419068736142,
      "grad_norm": 0.9106056094169617,
      "learning_rate": 5.221963097468576e-06,
      "loss": 0.6446,
      "step": 1754
    },
    {
      "epoch": 0.4864190687361419,
      "grad_norm": 0.9667460322380066,
      "learning_rate": 5.217613657512582e-06,
      "loss": 0.6351,
      "step": 1755
    },
    {
      "epoch": 0.4866962305986696,
      "grad_norm": 1.0147180557250977,
      "learning_rate": 5.213264052568416e-06,
      "loss": 0.631,
      "step": 1756
    },
    {
      "epoch": 0.48697339246119736,
      "grad_norm": 0.9304533004760742,
      "learning_rate": 5.208914285933819e-06,
      "loss": 0.628,
      "step": 1757
    },
    {
      "epoch": 0.48725055432372505,
      "grad_norm": 0.9463015794754028,
      "learning_rate": 5.204564360906657e-06,
      "loss": 0.587,
      "step": 1758
    },
    {
      "epoch": 0.4875277161862528,
      "grad_norm": 0.9149193167686462,
      "learning_rate": 5.200214280784909e-06,
      "loss": 0.6011,
      "step": 1759
    },
    {
      "epoch": 0.4878048780487805,
      "grad_norm": 0.9273519515991211,
      "learning_rate": 5.195864048866678e-06,
      "loss": 0.6281,
      "step": 1760
    },
    {
      "epoch": 0.4880820399113082,
      "grad_norm": 0.9949411153793335,
      "learning_rate": 5.191513668450178e-06,
      "loss": 0.6418,
      "step": 1761
    },
    {
      "epoch": 0.4883592017738359,
      "grad_norm": 1.160217046737671,
      "learning_rate": 5.1871631428337375e-06,
      "loss": 0.651,
      "step": 1762
    },
    {
      "epoch": 0.48863636363636365,
      "grad_norm": 0.9488372206687927,
      "learning_rate": 5.182812475315798e-06,
      "loss": 0.6322,
      "step": 1763
    },
    {
      "epoch": 0.48891352549889133,
      "grad_norm": 0.905265212059021,
      "learning_rate": 5.178461669194903e-06,
      "loss": 0.6563,
      "step": 1764
    },
    {
      "epoch": 0.4891906873614191,
      "grad_norm": 0.9566940665245056,
      "learning_rate": 5.1741107277697045e-06,
      "loss": 0.6581,
      "step": 1765
    },
    {
      "epoch": 0.48946784922394676,
      "grad_norm": 1.3952374458312988,
      "learning_rate": 5.169759654338955e-06,
      "loss": 0.5988,
      "step": 1766
    },
    {
      "epoch": 0.4897450110864745,
      "grad_norm": 0.9551557302474976,
      "learning_rate": 5.16540845220151e-06,
      "loss": 0.6574,
      "step": 1767
    },
    {
      "epoch": 0.49002217294900224,
      "grad_norm": 0.8798801302909851,
      "learning_rate": 5.161057124656319e-06,
      "loss": 0.597,
      "step": 1768
    },
    {
      "epoch": 0.49029933481152993,
      "grad_norm": 0.9633632898330688,
      "learning_rate": 5.1567056750024315e-06,
      "loss": 0.627,
      "step": 1769
    },
    {
      "epoch": 0.49057649667405767,
      "grad_norm": 1.3392114639282227,
      "learning_rate": 5.152354106538983e-06,
      "loss": 0.6207,
      "step": 1770
    },
    {
      "epoch": 0.49085365853658536,
      "grad_norm": 1.0457055568695068,
      "learning_rate": 5.148002422565207e-06,
      "loss": 0.6335,
      "step": 1771
    },
    {
      "epoch": 0.4911308203991131,
      "grad_norm": 1.0747556686401367,
      "learning_rate": 5.143650626380417e-06,
      "loss": 0.62,
      "step": 1772
    },
    {
      "epoch": 0.4914079822616408,
      "grad_norm": 0.9243521690368652,
      "learning_rate": 5.139298721284013e-06,
      "loss": 0.64,
      "step": 1773
    },
    {
      "epoch": 0.4916851441241685,
      "grad_norm": 0.9947118163108826,
      "learning_rate": 5.134946710575484e-06,
      "loss": 0.6371,
      "step": 1774
    },
    {
      "epoch": 0.4919623059866962,
      "grad_norm": 2.1055049896240234,
      "learning_rate": 5.130594597554392e-06,
      "loss": 0.6101,
      "step": 1775
    },
    {
      "epoch": 0.49223946784922396,
      "grad_norm": 0.9581360816955566,
      "learning_rate": 5.126242385520378e-06,
      "loss": 0.6092,
      "step": 1776
    },
    {
      "epoch": 0.49251662971175164,
      "grad_norm": 1.5534403324127197,
      "learning_rate": 5.121890077773162e-06,
      "loss": 0.6757,
      "step": 1777
    },
    {
      "epoch": 0.4927937915742794,
      "grad_norm": 0.9902575016021729,
      "learning_rate": 5.117537677612532e-06,
      "loss": 0.6035,
      "step": 1778
    },
    {
      "epoch": 0.49307095343680707,
      "grad_norm": 0.9931328892707825,
      "learning_rate": 5.113185188338345e-06,
      "loss": 0.6634,
      "step": 1779
    },
    {
      "epoch": 0.4933481152993348,
      "grad_norm": 0.9070310592651367,
      "learning_rate": 5.108832613250535e-06,
      "loss": 0.5938,
      "step": 1780
    },
    {
      "epoch": 0.49362527716186255,
      "grad_norm": 0.9533089399337769,
      "learning_rate": 5.1044799556490875e-06,
      "loss": 0.6277,
      "step": 1781
    },
    {
      "epoch": 0.49390243902439024,
      "grad_norm": 1.218001365661621,
      "learning_rate": 5.100127218834062e-06,
      "loss": 0.5871,
      "step": 1782
    },
    {
      "epoch": 0.494179600886918,
      "grad_norm": 1.0498141050338745,
      "learning_rate": 5.095774406105572e-06,
      "loss": 0.6842,
      "step": 1783
    },
    {
      "epoch": 0.49445676274944567,
      "grad_norm": 0.8833009600639343,
      "learning_rate": 5.091421520763788e-06,
      "loss": 0.6138,
      "step": 1784
    },
    {
      "epoch": 0.4947339246119734,
      "grad_norm": 0.9720257520675659,
      "learning_rate": 5.087068566108942e-06,
      "loss": 0.6075,
      "step": 1785
    },
    {
      "epoch": 0.4950110864745011,
      "grad_norm": 1.3731825351715088,
      "learning_rate": 5.082715545441308e-06,
      "loss": 0.6023,
      "step": 1786
    },
    {
      "epoch": 0.49528824833702884,
      "grad_norm": 0.9550261497497559,
      "learning_rate": 5.078362462061222e-06,
      "loss": 0.6547,
      "step": 1787
    },
    {
      "epoch": 0.4955654101995565,
      "grad_norm": 0.880942165851593,
      "learning_rate": 5.074009319269058e-06,
      "loss": 0.5811,
      "step": 1788
    },
    {
      "epoch": 0.49584257206208426,
      "grad_norm": 0.9849389791488647,
      "learning_rate": 5.069656120365241e-06,
      "loss": 0.5907,
      "step": 1789
    },
    {
      "epoch": 0.49611973392461195,
      "grad_norm": 0.9270283579826355,
      "learning_rate": 5.065302868650234e-06,
      "loss": 0.6098,
      "step": 1790
    },
    {
      "epoch": 0.4963968957871397,
      "grad_norm": 0.971530020236969,
      "learning_rate": 5.060949567424546e-06,
      "loss": 0.6292,
      "step": 1791
    },
    {
      "epoch": 0.49667405764966743,
      "grad_norm": 1.0050526857376099,
      "learning_rate": 5.056596219988715e-06,
      "loss": 0.6186,
      "step": 1792
    },
    {
      "epoch": 0.4969512195121951,
      "grad_norm": 0.9732151627540588,
      "learning_rate": 5.052242829643323e-06,
      "loss": 0.6798,
      "step": 1793
    },
    {
      "epoch": 0.49722838137472286,
      "grad_norm": 0.992128312587738,
      "learning_rate": 5.04788939968898e-06,
      "loss": 0.6662,
      "step": 1794
    },
    {
      "epoch": 0.49750554323725055,
      "grad_norm": 0.9050638675689697,
      "learning_rate": 5.043535933426323e-06,
      "loss": 0.6563,
      "step": 1795
    },
    {
      "epoch": 0.4977827050997783,
      "grad_norm": 0.930923342704773,
      "learning_rate": 5.0391824341560256e-06,
      "loss": 0.6476,
      "step": 1796
    },
    {
      "epoch": 0.498059866962306,
      "grad_norm": 1.0295449495315552,
      "learning_rate": 5.034828905178775e-06,
      "loss": 0.6491,
      "step": 1797
    },
    {
      "epoch": 0.4983370288248337,
      "grad_norm": 0.9762796759605408,
      "learning_rate": 5.030475349795291e-06,
      "loss": 0.6567,
      "step": 1798
    },
    {
      "epoch": 0.4986141906873614,
      "grad_norm": 0.9946879148483276,
      "learning_rate": 5.026121771306307e-06,
      "loss": 0.656,
      "step": 1799
    },
    {
      "epoch": 0.49889135254988914,
      "grad_norm": 0.9751898646354675,
      "learning_rate": 5.021768173012576e-06,
      "loss": 0.6171,
      "step": 1800
    },
    {
      "epoch": 0.49916851441241683,
      "grad_norm": 0.9071579575538635,
      "learning_rate": 5.0174145582148695e-06,
      "loss": 0.6217,
      "step": 1801
    },
    {
      "epoch": 0.49944567627494457,
      "grad_norm": 0.9689897894859314,
      "learning_rate": 5.013060930213963e-06,
      "loss": 0.6412,
      "step": 1802
    },
    {
      "epoch": 0.49972283813747226,
      "grad_norm": 1.010853886604309,
      "learning_rate": 5.008707292310649e-06,
      "loss": 0.6321,
      "step": 1803
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2724926471710205,
      "learning_rate": 5.004353647805728e-06,
      "loss": 0.6007,
      "step": 1804
    },
    {
      "epoch": 0.5002771618625277,
      "grad_norm": 1.036178469657898,
      "learning_rate": 5e-06,
      "loss": 0.6259,
      "step": 1805
    },
    {
      "epoch": 0.5005543237250555,
      "grad_norm": 0.921195924282074,
      "learning_rate": 4.9956463521942734e-06,
      "loss": 0.6567,
      "step": 1806
    },
    {
      "epoch": 0.5008314855875832,
      "grad_norm": 1.0009310245513916,
      "learning_rate": 4.991292707689351e-06,
      "loss": 0.6563,
      "step": 1807
    },
    {
      "epoch": 0.5011086474501109,
      "grad_norm": 0.9469943642616272,
      "learning_rate": 4.986939069786038e-06,
      "loss": 0.6389,
      "step": 1808
    },
    {
      "epoch": 0.5013858093126385,
      "grad_norm": 1.3667163848876953,
      "learning_rate": 4.982585441785133e-06,
      "loss": 0.6312,
      "step": 1809
    },
    {
      "epoch": 0.5016629711751663,
      "grad_norm": 0.9387463331222534,
      "learning_rate": 4.978231826987425e-06,
      "loss": 0.6277,
      "step": 1810
    },
    {
      "epoch": 0.501940133037694,
      "grad_norm": 0.9371724128723145,
      "learning_rate": 4.973878228693693e-06,
      "loss": 0.6305,
      "step": 1811
    },
    {
      "epoch": 0.5022172949002217,
      "grad_norm": 0.930037796497345,
      "learning_rate": 4.969524650204711e-06,
      "loss": 0.6096,
      "step": 1812
    },
    {
      "epoch": 0.5024944567627494,
      "grad_norm": 1.1126900911331177,
      "learning_rate": 4.965171094821226e-06,
      "loss": 0.6895,
      "step": 1813
    },
    {
      "epoch": 0.5027716186252772,
      "grad_norm": 0.9861113429069519,
      "learning_rate": 4.960817565843977e-06,
      "loss": 0.672,
      "step": 1814
    },
    {
      "epoch": 0.5030487804878049,
      "grad_norm": 0.9184641242027283,
      "learning_rate": 4.9564640665736766e-06,
      "loss": 0.6043,
      "step": 1815
    },
    {
      "epoch": 0.5033259423503326,
      "grad_norm": 0.9156967401504517,
      "learning_rate": 4.952110600311021e-06,
      "loss": 0.6391,
      "step": 1816
    },
    {
      "epoch": 0.5036031042128604,
      "grad_norm": 1.0486929416656494,
      "learning_rate": 4.947757170356678e-06,
      "loss": 0.5758,
      "step": 1817
    },
    {
      "epoch": 0.503880266075388,
      "grad_norm": 1.190490484237671,
      "learning_rate": 4.943403780011285e-06,
      "loss": 0.6569,
      "step": 1818
    },
    {
      "epoch": 0.5041574279379157,
      "grad_norm": 0.9535707235336304,
      "learning_rate": 4.939050432575455e-06,
      "loss": 0.6304,
      "step": 1819
    },
    {
      "epoch": 0.5044345898004434,
      "grad_norm": 0.9367846250534058,
      "learning_rate": 4.934697131349766e-06,
      "loss": 0.6049,
      "step": 1820
    },
    {
      "epoch": 0.5047117516629712,
      "grad_norm": 1.1276642084121704,
      "learning_rate": 4.930343879634762e-06,
      "loss": 0.6322,
      "step": 1821
    },
    {
      "epoch": 0.5049889135254989,
      "grad_norm": 0.9851393699645996,
      "learning_rate": 4.925990680730942e-06,
      "loss": 0.6568,
      "step": 1822
    },
    {
      "epoch": 0.5052660753880266,
      "grad_norm": 0.9442154169082642,
      "learning_rate": 4.921637537938781e-06,
      "loss": 0.6183,
      "step": 1823
    },
    {
      "epoch": 0.5055432372505543,
      "grad_norm": 1.128682017326355,
      "learning_rate": 4.917284454558694e-06,
      "loss": 0.6212,
      "step": 1824
    },
    {
      "epoch": 0.5058203991130821,
      "grad_norm": 1.0211834907531738,
      "learning_rate": 4.912931433891062e-06,
      "loss": 0.6833,
      "step": 1825
    },
    {
      "epoch": 0.5060975609756098,
      "grad_norm": 1.0013262033462524,
      "learning_rate": 4.908578479236213e-06,
      "loss": 0.6503,
      "step": 1826
    },
    {
      "epoch": 0.5063747228381374,
      "grad_norm": 0.9239797592163086,
      "learning_rate": 4.90422559389443e-06,
      "loss": 0.6272,
      "step": 1827
    },
    {
      "epoch": 0.5066518847006651,
      "grad_norm": 1.1298401355743408,
      "learning_rate": 4.899872781165939e-06,
      "loss": 0.6568,
      "step": 1828
    },
    {
      "epoch": 0.5069290465631929,
      "grad_norm": 0.9411318898200989,
      "learning_rate": 4.895520044350914e-06,
      "loss": 0.6083,
      "step": 1829
    },
    {
      "epoch": 0.5072062084257206,
      "grad_norm": 0.9553874731063843,
      "learning_rate": 4.891167386749466e-06,
      "loss": 0.6213,
      "step": 1830
    },
    {
      "epoch": 0.5074833702882483,
      "grad_norm": 0.990439772605896,
      "learning_rate": 4.8868148116616555e-06,
      "loss": 0.6606,
      "step": 1831
    },
    {
      "epoch": 0.5077605321507761,
      "grad_norm": 0.9374085664749146,
      "learning_rate": 4.882462322387471e-06,
      "loss": 0.6989,
      "step": 1832
    },
    {
      "epoch": 0.5080376940133038,
      "grad_norm": 0.9373449683189392,
      "learning_rate": 4.878109922226838e-06,
      "loss": 0.6003,
      "step": 1833
    },
    {
      "epoch": 0.5083148558758315,
      "grad_norm": 1.0831090211868286,
      "learning_rate": 4.873757614479623e-06,
      "loss": 0.6379,
      "step": 1834
    },
    {
      "epoch": 0.5085920177383592,
      "grad_norm": 0.9927808046340942,
      "learning_rate": 4.86940540244561e-06,
      "loss": 0.6821,
      "step": 1835
    },
    {
      "epoch": 0.508869179600887,
      "grad_norm": 0.9646443724632263,
      "learning_rate": 4.865053289424518e-06,
      "loss": 0.6284,
      "step": 1836
    },
    {
      "epoch": 0.5091463414634146,
      "grad_norm": 0.9921881556510925,
      "learning_rate": 4.860701278715987e-06,
      "loss": 0.6728,
      "step": 1837
    },
    {
      "epoch": 0.5094235033259423,
      "grad_norm": 0.9545189738273621,
      "learning_rate": 4.856349373619586e-06,
      "loss": 0.6328,
      "step": 1838
    },
    {
      "epoch": 0.50970066518847,
      "grad_norm": 0.9823859930038452,
      "learning_rate": 4.851997577434794e-06,
      "loss": 0.6768,
      "step": 1839
    },
    {
      "epoch": 0.5099778270509978,
      "grad_norm": 0.9806753993034363,
      "learning_rate": 4.847645893461018e-06,
      "loss": 0.6257,
      "step": 1840
    },
    {
      "epoch": 0.5102549889135255,
      "grad_norm": 0.9305614829063416,
      "learning_rate": 4.8432943249975685e-06,
      "loss": 0.5997,
      "step": 1841
    },
    {
      "epoch": 0.5105321507760532,
      "grad_norm": 1.3063558340072632,
      "learning_rate": 4.838942875343682e-06,
      "loss": 0.5874,
      "step": 1842
    },
    {
      "epoch": 0.510809312638581,
      "grad_norm": 0.978273332118988,
      "learning_rate": 4.834591547798492e-06,
      "loss": 0.6523,
      "step": 1843
    },
    {
      "epoch": 0.5110864745011087,
      "grad_norm": 0.9391226768493652,
      "learning_rate": 4.830240345661046e-06,
      "loss": 0.5787,
      "step": 1844
    },
    {
      "epoch": 0.5113636363636364,
      "grad_norm": 0.9401317238807678,
      "learning_rate": 4.825889272230297e-06,
      "loss": 0.6636,
      "step": 1845
    },
    {
      "epoch": 0.511640798226164,
      "grad_norm": 0.9199084043502808,
      "learning_rate": 4.821538330805098e-06,
      "loss": 0.6038,
      "step": 1846
    },
    {
      "epoch": 0.5119179600886918,
      "grad_norm": 1.0469775199890137,
      "learning_rate": 4.817187524684204e-06,
      "loss": 0.6587,
      "step": 1847
    },
    {
      "epoch": 0.5121951219512195,
      "grad_norm": 0.9811663627624512,
      "learning_rate": 4.8128368571662625e-06,
      "loss": 0.6158,
      "step": 1848
    },
    {
      "epoch": 0.5124722838137472,
      "grad_norm": 1.0672260522842407,
      "learning_rate": 4.808486331549824e-06,
      "loss": 0.6089,
      "step": 1849
    },
    {
      "epoch": 0.5127494456762749,
      "grad_norm": 0.9439910054206848,
      "learning_rate": 4.804135951133324e-06,
      "loss": 0.6399,
      "step": 1850
    },
    {
      "epoch": 0.5130266075388027,
      "grad_norm": 0.9401859045028687,
      "learning_rate": 4.799785719215094e-06,
      "loss": 0.6369,
      "step": 1851
    },
    {
      "epoch": 0.5133037694013304,
      "grad_norm": 0.9953728318214417,
      "learning_rate": 4.795435639093344e-06,
      "loss": 0.6439,
      "step": 1852
    },
    {
      "epoch": 0.5135809312638581,
      "grad_norm": 1.0834261178970337,
      "learning_rate": 4.7910857140661815e-06,
      "loss": 0.6291,
      "step": 1853
    },
    {
      "epoch": 0.5138580931263859,
      "grad_norm": 0.909688413143158,
      "learning_rate": 4.7867359474315865e-06,
      "loss": 0.6393,
      "step": 1854
    },
    {
      "epoch": 0.5141352549889135,
      "grad_norm": 0.967305600643158,
      "learning_rate": 4.78238634248742e-06,
      "loss": 0.6167,
      "step": 1855
    },
    {
      "epoch": 0.5144124168514412,
      "grad_norm": 0.9870229363441467,
      "learning_rate": 4.778036902531426e-06,
      "loss": 0.6272,
      "step": 1856
    },
    {
      "epoch": 0.5146895787139689,
      "grad_norm": 1.1316946744918823,
      "learning_rate": 4.773687630861219e-06,
      "loss": 0.6284,
      "step": 1857
    },
    {
      "epoch": 0.5149667405764967,
      "grad_norm": 0.92823326587677,
      "learning_rate": 4.769338530774286e-06,
      "loss": 0.61,
      "step": 1858
    },
    {
      "epoch": 0.5152439024390244,
      "grad_norm": 0.956476092338562,
      "learning_rate": 4.764989605567984e-06,
      "loss": 0.6056,
      "step": 1859
    },
    {
      "epoch": 0.5155210643015521,
      "grad_norm": 0.9007798433303833,
      "learning_rate": 4.760640858539541e-06,
      "loss": 0.5916,
      "step": 1860
    },
    {
      "epoch": 0.5157982261640798,
      "grad_norm": 1.0851870775222778,
      "learning_rate": 4.756292292986045e-06,
      "loss": 0.6159,
      "step": 1861
    },
    {
      "epoch": 0.5160753880266076,
      "grad_norm": 1.0149645805358887,
      "learning_rate": 4.751943912204449e-06,
      "loss": 0.6424,
      "step": 1862
    },
    {
      "epoch": 0.5163525498891353,
      "grad_norm": 0.9196805953979492,
      "learning_rate": 4.7475957194915615e-06,
      "loss": 0.6084,
      "step": 1863
    },
    {
      "epoch": 0.516629711751663,
      "grad_norm": 1.0127639770507812,
      "learning_rate": 4.7432477181440585e-06,
      "loss": 0.6218,
      "step": 1864
    },
    {
      "epoch": 0.5169068736141907,
      "grad_norm": 0.9323347806930542,
      "learning_rate": 4.738899911458462e-06,
      "loss": 0.6054,
      "step": 1865
    },
    {
      "epoch": 0.5171840354767184,
      "grad_norm": 0.9245114326477051,
      "learning_rate": 4.734552302731148e-06,
      "loss": 0.6144,
      "step": 1866
    },
    {
      "epoch": 0.5174611973392461,
      "grad_norm": 0.9755973219871521,
      "learning_rate": 4.730204895258345e-06,
      "loss": 0.6023,
      "step": 1867
    },
    {
      "epoch": 0.5177383592017738,
      "grad_norm": 1.0480046272277832,
      "learning_rate": 4.7258576923361274e-06,
      "loss": 0.6015,
      "step": 1868
    },
    {
      "epoch": 0.5180155210643016,
      "grad_norm": 0.970574140548706,
      "learning_rate": 4.721510697260413e-06,
      "loss": 0.7309,
      "step": 1869
    },
    {
      "epoch": 0.5182926829268293,
      "grad_norm": 1.0803941488265991,
      "learning_rate": 4.717163913326964e-06,
      "loss": 0.6543,
      "step": 1870
    },
    {
      "epoch": 0.518569844789357,
      "grad_norm": 0.9905227422714233,
      "learning_rate": 4.712817343831384e-06,
      "loss": 0.6258,
      "step": 1871
    },
    {
      "epoch": 0.5188470066518847,
      "grad_norm": 0.9443010091781616,
      "learning_rate": 4.708470992069112e-06,
      "loss": 0.6195,
      "step": 1872
    },
    {
      "epoch": 0.5191241685144125,
      "grad_norm": 0.9772205352783203,
      "learning_rate": 4.704124861335418e-06,
      "loss": 0.6156,
      "step": 1873
    },
    {
      "epoch": 0.5194013303769401,
      "grad_norm": 0.9519892334938049,
      "learning_rate": 4.699778954925412e-06,
      "loss": 0.6327,
      "step": 1874
    },
    {
      "epoch": 0.5196784922394678,
      "grad_norm": 1.1737631559371948,
      "learning_rate": 4.695433276134029e-06,
      "loss": 0.6296,
      "step": 1875
    },
    {
      "epoch": 0.5199556541019955,
      "grad_norm": 0.8893111348152161,
      "learning_rate": 4.6910878282560345e-06,
      "loss": 0.5684,
      "step": 1876
    },
    {
      "epoch": 0.5202328159645233,
      "grad_norm": 0.9515904784202576,
      "learning_rate": 4.686742614586015e-06,
      "loss": 0.6434,
      "step": 1877
    },
    {
      "epoch": 0.520509977827051,
      "grad_norm": 0.8826665282249451,
      "learning_rate": 4.682397638418382e-06,
      "loss": 0.6622,
      "step": 1878
    },
    {
      "epoch": 0.5207871396895787,
      "grad_norm": 1.0165308713912964,
      "learning_rate": 4.678052903047369e-06,
      "loss": 0.6979,
      "step": 1879
    },
    {
      "epoch": 0.5210643015521065,
      "grad_norm": 0.9487347602844238,
      "learning_rate": 4.673708411767021e-06,
      "loss": 0.6231,
      "step": 1880
    },
    {
      "epoch": 0.5213414634146342,
      "grad_norm": 0.9874770641326904,
      "learning_rate": 4.669364167871203e-06,
      "loss": 0.692,
      "step": 1881
    },
    {
      "epoch": 0.5216186252771619,
      "grad_norm": 1.001013159751892,
      "learning_rate": 4.6650201746535926e-06,
      "loss": 0.6462,
      "step": 1882
    },
    {
      "epoch": 0.5218957871396895,
      "grad_norm": 1.0059013366699219,
      "learning_rate": 4.660676435407674e-06,
      "loss": 0.6737,
      "step": 1883
    },
    {
      "epoch": 0.5221729490022173,
      "grad_norm": 0.9549412727355957,
      "learning_rate": 4.65633295342674e-06,
      "loss": 0.6238,
      "step": 1884
    },
    {
      "epoch": 0.522450110864745,
      "grad_norm": 0.9940391778945923,
      "learning_rate": 4.651989732003888e-06,
      "loss": 0.6872,
      "step": 1885
    },
    {
      "epoch": 0.5227272727272727,
      "grad_norm": 0.9536202549934387,
      "learning_rate": 4.647646774432021e-06,
      "loss": 0.6166,
      "step": 1886
    },
    {
      "epoch": 0.5230044345898004,
      "grad_norm": 0.9292188286781311,
      "learning_rate": 4.643304084003839e-06,
      "loss": 0.6864,
      "step": 1887
    },
    {
      "epoch": 0.5232815964523282,
      "grad_norm": 0.8910459280014038,
      "learning_rate": 4.63896166401184e-06,
      "loss": 0.6647,
      "step": 1888
    },
    {
      "epoch": 0.5235587583148559,
      "grad_norm": 0.949334979057312,
      "learning_rate": 4.634619517748315e-06,
      "loss": 0.6986,
      "step": 1889
    },
    {
      "epoch": 0.5238359201773836,
      "grad_norm": 1.2461720705032349,
      "learning_rate": 4.630277648505352e-06,
      "loss": 0.6615,
      "step": 1890
    },
    {
      "epoch": 0.5241130820399114,
      "grad_norm": 0.9033803343772888,
      "learning_rate": 4.625936059574828e-06,
      "loss": 0.658,
      "step": 1891
    },
    {
      "epoch": 0.524390243902439,
      "grad_norm": 0.9548044204711914,
      "learning_rate": 4.621594754248399e-06,
      "loss": 0.5964,
      "step": 1892
    },
    {
      "epoch": 0.5246674057649667,
      "grad_norm": 0.9296775460243225,
      "learning_rate": 4.617253735817522e-06,
      "loss": 0.6014,
      "step": 1893
    },
    {
      "epoch": 0.5249445676274944,
      "grad_norm": 0.897864580154419,
      "learning_rate": 4.6129130075734224e-06,
      "loss": 0.5876,
      "step": 1894
    },
    {
      "epoch": 0.5252217294900222,
      "grad_norm": 0.9740297198295593,
      "learning_rate": 4.608572572807111e-06,
      "loss": 0.6132,
      "step": 1895
    },
    {
      "epoch": 0.5254988913525499,
      "grad_norm": 0.9701299071311951,
      "learning_rate": 4.604232434809375e-06,
      "loss": 0.6248,
      "step": 1896
    },
    {
      "epoch": 0.5257760532150776,
      "grad_norm": 0.921424150466919,
      "learning_rate": 4.599892596870779e-06,
      "loss": 0.6212,
      "step": 1897
    },
    {
      "epoch": 0.5260532150776053,
      "grad_norm": 0.9575247168540955,
      "learning_rate": 4.595553062281658e-06,
      "loss": 0.6427,
      "step": 1898
    },
    {
      "epoch": 0.5263303769401331,
      "grad_norm": 0.9710375666618347,
      "learning_rate": 4.5912138343321176e-06,
      "loss": 0.626,
      "step": 1899
    },
    {
      "epoch": 0.5266075388026608,
      "grad_norm": 1.1599607467651367,
      "learning_rate": 4.5868749163120284e-06,
      "loss": 0.6092,
      "step": 1900
    },
    {
      "epoch": 0.5268847006651884,
      "grad_norm": 0.9210077524185181,
      "learning_rate": 4.582536311511032e-06,
      "loss": 0.6864,
      "step": 1901
    },
    {
      "epoch": 0.5271618625277162,
      "grad_norm": 0.9271044135093689,
      "learning_rate": 4.578198023218528e-06,
      "loss": 0.6346,
      "step": 1902
    },
    {
      "epoch": 0.5274390243902439,
      "grad_norm": 0.9247244000434875,
      "learning_rate": 4.573860054723673e-06,
      "loss": 0.5968,
      "step": 1903
    },
    {
      "epoch": 0.5277161862527716,
      "grad_norm": 0.9233640432357788,
      "learning_rate": 4.569522409315392e-06,
      "loss": 0.6597,
      "step": 1904
    },
    {
      "epoch": 0.5279933481152993,
      "grad_norm": 1.1118342876434326,
      "learning_rate": 4.565185090282353e-06,
      "loss": 0.6419,
      "step": 1905
    },
    {
      "epoch": 0.5282705099778271,
      "grad_norm": 1.0024704933166504,
      "learning_rate": 4.560848100912982e-06,
      "loss": 0.5834,
      "step": 1906
    },
    {
      "epoch": 0.5285476718403548,
      "grad_norm": 0.9597956538200378,
      "learning_rate": 4.556511444495456e-06,
      "loss": 0.6374,
      "step": 1907
    },
    {
      "epoch": 0.5288248337028825,
      "grad_norm": 0.9927630424499512,
      "learning_rate": 4.5521751243176985e-06,
      "loss": 0.6369,
      "step": 1908
    },
    {
      "epoch": 0.5291019955654102,
      "grad_norm": 0.9749294519424438,
      "learning_rate": 4.5478391436673765e-06,
      "loss": 0.5972,
      "step": 1909
    },
    {
      "epoch": 0.529379157427938,
      "grad_norm": 0.8978511691093445,
      "learning_rate": 4.5435035058319025e-06,
      "loss": 0.6098,
      "step": 1910
    },
    {
      "epoch": 0.5296563192904656,
      "grad_norm": 1.004485011100769,
      "learning_rate": 4.539168214098426e-06,
      "loss": 0.7026,
      "step": 1911
    },
    {
      "epoch": 0.5299334811529933,
      "grad_norm": 0.9585144519805908,
      "learning_rate": 4.534833271753837e-06,
      "loss": 0.625,
      "step": 1912
    },
    {
      "epoch": 0.530210643015521,
      "grad_norm": 0.9488794207572937,
      "learning_rate": 4.530498682084759e-06,
      "loss": 0.6481,
      "step": 1913
    },
    {
      "epoch": 0.5304878048780488,
      "grad_norm": 0.9584585428237915,
      "learning_rate": 4.526164448377549e-06,
      "loss": 0.6285,
      "step": 1914
    },
    {
      "epoch": 0.5307649667405765,
      "grad_norm": 0.9224879741668701,
      "learning_rate": 4.521830573918289e-06,
      "loss": 0.6837,
      "step": 1915
    },
    {
      "epoch": 0.5310421286031042,
      "grad_norm": 0.9391283392906189,
      "learning_rate": 4.517497061992799e-06,
      "loss": 0.6083,
      "step": 1916
    },
    {
      "epoch": 0.531319290465632,
      "grad_norm": 0.9703270196914673,
      "learning_rate": 4.5131639158866156e-06,
      "loss": 0.6272,
      "step": 1917
    },
    {
      "epoch": 0.5315964523281597,
      "grad_norm": 0.94402676820755,
      "learning_rate": 4.508831138884999e-06,
      "loss": 0.6215,
      "step": 1918
    },
    {
      "epoch": 0.5318736141906873,
      "grad_norm": 0.9292389750480652,
      "learning_rate": 4.504498734272935e-06,
      "loss": 0.6721,
      "step": 1919
    },
    {
      "epoch": 0.532150776053215,
      "grad_norm": 4.003532409667969,
      "learning_rate": 4.50016670533512e-06,
      "loss": 0.597,
      "step": 1920
    },
    {
      "epoch": 0.5324279379157428,
      "grad_norm": 0.9677640199661255,
      "learning_rate": 4.495835055355972e-06,
      "loss": 0.6365,
      "step": 1921
    },
    {
      "epoch": 0.5327050997782705,
      "grad_norm": 1.646508812904358,
      "learning_rate": 4.4915037876196135e-06,
      "loss": 0.6461,
      "step": 1922
    },
    {
      "epoch": 0.5329822616407982,
      "grad_norm": 1.2125016450881958,
      "learning_rate": 4.487172905409888e-06,
      "loss": 0.5841,
      "step": 1923
    },
    {
      "epoch": 0.5332594235033259,
      "grad_norm": 1.0309375524520874,
      "learning_rate": 4.4828424120103395e-06,
      "loss": 0.6255,
      "step": 1924
    },
    {
      "epoch": 0.5335365853658537,
      "grad_norm": 0.8604519963264465,
      "learning_rate": 4.478512310704216e-06,
      "loss": 0.6091,
      "step": 1925
    },
    {
      "epoch": 0.5338137472283814,
      "grad_norm": 0.9382449984550476,
      "learning_rate": 4.474182604774471e-06,
      "loss": 0.675,
      "step": 1926
    },
    {
      "epoch": 0.5340909090909091,
      "grad_norm": 1.0042564868927002,
      "learning_rate": 4.4698532975037615e-06,
      "loss": 0.6382,
      "step": 1927
    },
    {
      "epoch": 0.5343680709534369,
      "grad_norm": 0.9131385087966919,
      "learning_rate": 4.465524392174437e-06,
      "loss": 0.5516,
      "step": 1928
    },
    {
      "epoch": 0.5346452328159645,
      "grad_norm": 1.0916540622711182,
      "learning_rate": 4.461195892068542e-06,
      "loss": 0.6104,
      "step": 1929
    },
    {
      "epoch": 0.5349223946784922,
      "grad_norm": 0.9794466495513916,
      "learning_rate": 4.456867800467821e-06,
      "loss": 0.619,
      "step": 1930
    },
    {
      "epoch": 0.5351995565410199,
      "grad_norm": 0.9203370809555054,
      "learning_rate": 4.452540120653699e-06,
      "loss": 0.5962,
      "step": 1931
    },
    {
      "epoch": 0.5354767184035477,
      "grad_norm": 0.9847226142883301,
      "learning_rate": 4.448212855907298e-06,
      "loss": 0.5935,
      "step": 1932
    },
    {
      "epoch": 0.5357538802660754,
      "grad_norm": 0.9747816920280457,
      "learning_rate": 4.4438860095094135e-06,
      "loss": 0.6304,
      "step": 1933
    },
    {
      "epoch": 0.5360310421286031,
      "grad_norm": 0.9696584939956665,
      "learning_rate": 4.4395595847405395e-06,
      "loss": 0.6157,
      "step": 1934
    },
    {
      "epoch": 0.5363082039911308,
      "grad_norm": 0.9932990670204163,
      "learning_rate": 4.4352335848808365e-06,
      "loss": 0.6004,
      "step": 1935
    },
    {
      "epoch": 0.5365853658536586,
      "grad_norm": 0.9492164254188538,
      "learning_rate": 4.43090801321015e-06,
      "loss": 0.6508,
      "step": 1936
    },
    {
      "epoch": 0.5368625277161863,
      "grad_norm": 0.8831507563591003,
      "learning_rate": 4.426582873007999e-06,
      "loss": 0.6006,
      "step": 1937
    },
    {
      "epoch": 0.5371396895787139,
      "grad_norm": 0.9564241170883179,
      "learning_rate": 4.422258167553575e-06,
      "loss": 0.6374,
      "step": 1938
    },
    {
      "epoch": 0.5374168514412417,
      "grad_norm": 0.979810893535614,
      "learning_rate": 4.417933900125742e-06,
      "loss": 0.647,
      "step": 1939
    },
    {
      "epoch": 0.5376940133037694,
      "grad_norm": 0.9264168739318848,
      "learning_rate": 4.413610074003028e-06,
      "loss": 0.6685,
      "step": 1940
    },
    {
      "epoch": 0.5379711751662971,
      "grad_norm": 1.0026520490646362,
      "learning_rate": 4.409286692463631e-06,
      "loss": 0.6417,
      "step": 1941
    },
    {
      "epoch": 0.5382483370288248,
      "grad_norm": 0.9220601916313171,
      "learning_rate": 4.404963758785409e-06,
      "loss": 0.6201,
      "step": 1942
    },
    {
      "epoch": 0.5385254988913526,
      "grad_norm": 1.0650179386138916,
      "learning_rate": 4.400641276245882e-06,
      "loss": 0.6176,
      "step": 1943
    },
    {
      "epoch": 0.5388026607538803,
      "grad_norm": 1.8167989253997803,
      "learning_rate": 4.396319248122223e-06,
      "loss": 0.6026,
      "step": 1944
    },
    {
      "epoch": 0.539079822616408,
      "grad_norm": 0.8850213289260864,
      "learning_rate": 4.391997677691271e-06,
      "loss": 0.6101,
      "step": 1945
    },
    {
      "epoch": 0.5393569844789357,
      "grad_norm": 0.9556333422660828,
      "learning_rate": 4.387676568229506e-06,
      "loss": 0.64,
      "step": 1946
    },
    {
      "epoch": 0.5396341463414634,
      "grad_norm": 0.9353305697441101,
      "learning_rate": 4.383355923013066e-06,
      "loss": 0.7036,
      "step": 1947
    },
    {
      "epoch": 0.5399113082039911,
      "grad_norm": 0.9178584814071655,
      "learning_rate": 4.379035745317734e-06,
      "loss": 0.6492,
      "step": 1948
    },
    {
      "epoch": 0.5401884700665188,
      "grad_norm": 0.9144257307052612,
      "learning_rate": 4.374716038418942e-06,
      "loss": 0.6215,
      "step": 1949
    },
    {
      "epoch": 0.5404656319290465,
      "grad_norm": 12.930267333984375,
      "learning_rate": 4.3703968055917625e-06,
      "loss": 0.6683,
      "step": 1950
    },
    {
      "epoch": 0.5407427937915743,
      "grad_norm": 1.0284466743469238,
      "learning_rate": 4.366078050110908e-06,
      "loss": 0.5961,
      "step": 1951
    },
    {
      "epoch": 0.541019955654102,
      "grad_norm": 1.0989900827407837,
      "learning_rate": 4.361759775250726e-06,
      "loss": 0.5965,
      "step": 1952
    },
    {
      "epoch": 0.5412971175166297,
      "grad_norm": 2.1038870811462402,
      "learning_rate": 4.3574419842852085e-06,
      "loss": 0.5838,
      "step": 1953
    },
    {
      "epoch": 0.5415742793791575,
      "grad_norm": 1.0307985544204712,
      "learning_rate": 4.353124680487976e-06,
      "loss": 0.5815,
      "step": 1954
    },
    {
      "epoch": 0.5418514412416852,
      "grad_norm": 0.9098923802375793,
      "learning_rate": 4.348807867132274e-06,
      "loss": 0.5816,
      "step": 1955
    },
    {
      "epoch": 0.5421286031042128,
      "grad_norm": 1.0747097730636597,
      "learning_rate": 4.344491547490987e-06,
      "loss": 0.6318,
      "step": 1956
    },
    {
      "epoch": 0.5424057649667405,
      "grad_norm": 0.9083472490310669,
      "learning_rate": 4.340175724836617e-06,
      "loss": 0.5963,
      "step": 1957
    },
    {
      "epoch": 0.5426829268292683,
      "grad_norm": 0.9875832200050354,
      "learning_rate": 4.335860402441292e-06,
      "loss": 0.6379,
      "step": 1958
    },
    {
      "epoch": 0.542960088691796,
      "grad_norm": 0.9157746434211731,
      "learning_rate": 4.331545583576758e-06,
      "loss": 0.6402,
      "step": 1959
    },
    {
      "epoch": 0.5432372505543237,
      "grad_norm": 1.03079354763031,
      "learning_rate": 4.327231271514386e-06,
      "loss": 0.665,
      "step": 1960
    },
    {
      "epoch": 0.5435144124168514,
      "grad_norm": 0.9118499755859375,
      "learning_rate": 4.322917469525156e-06,
      "loss": 0.6285,
      "step": 1961
    },
    {
      "epoch": 0.5437915742793792,
      "grad_norm": 0.9343001842498779,
      "learning_rate": 4.318604180879665e-06,
      "loss": 0.6117,
      "step": 1962
    },
    {
      "epoch": 0.5440687361419069,
      "grad_norm": 0.9382731318473816,
      "learning_rate": 4.314291408848116e-06,
      "loss": 0.671,
      "step": 1963
    },
    {
      "epoch": 0.5443458980044346,
      "grad_norm": 1.071782112121582,
      "learning_rate": 4.30997915670033e-06,
      "loss": 0.6197,
      "step": 1964
    },
    {
      "epoch": 0.5446230598669624,
      "grad_norm": 0.9022917747497559,
      "learning_rate": 4.305667427705722e-06,
      "loss": 0.613,
      "step": 1965
    },
    {
      "epoch": 0.54490022172949,
      "grad_norm": 0.9304943084716797,
      "learning_rate": 4.301356225133317e-06,
      "loss": 0.6295,
      "step": 1966
    },
    {
      "epoch": 0.5451773835920177,
      "grad_norm": 0.8972543478012085,
      "learning_rate": 4.297045552251745e-06,
      "loss": 0.6152,
      "step": 1967
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.9338002800941467,
      "learning_rate": 4.2927354123292245e-06,
      "loss": 0.6134,
      "step": 1968
    },
    {
      "epoch": 0.5457317073170732,
      "grad_norm": 1.1097288131713867,
      "learning_rate": 4.2884258086335755e-06,
      "loss": 0.6161,
      "step": 1969
    },
    {
      "epoch": 0.5460088691796009,
      "grad_norm": 0.9571998119354248,
      "learning_rate": 4.28411674443221e-06,
      "loss": 0.6005,
      "step": 1970
    },
    {
      "epoch": 0.5462860310421286,
      "grad_norm": 0.9178850054740906,
      "learning_rate": 4.2798082229921336e-06,
      "loss": 0.608,
      "step": 1971
    },
    {
      "epoch": 0.5465631929046563,
      "grad_norm": 0.9593766927719116,
      "learning_rate": 4.275500247579939e-06,
      "loss": 0.6102,
      "step": 1972
    },
    {
      "epoch": 0.5468403547671841,
      "grad_norm": 0.9173176288604736,
      "learning_rate": 4.271192821461802e-06,
      "loss": 0.6101,
      "step": 1973
    },
    {
      "epoch": 0.5471175166297118,
      "grad_norm": 0.9552692770957947,
      "learning_rate": 4.266885947903484e-06,
      "loss": 0.6795,
      "step": 1974
    },
    {
      "epoch": 0.5473946784922394,
      "grad_norm": 0.956268846988678,
      "learning_rate": 4.262579630170333e-06,
      "loss": 0.6358,
      "step": 1975
    },
    {
      "epoch": 0.5476718403547672,
      "grad_norm": 0.9220622181892395,
      "learning_rate": 4.258273871527264e-06,
      "loss": 0.6202,
      "step": 1976
    },
    {
      "epoch": 0.5479490022172949,
      "grad_norm": 0.906059980392456,
      "learning_rate": 4.253968675238776e-06,
      "loss": 0.6613,
      "step": 1977
    },
    {
      "epoch": 0.5482261640798226,
      "grad_norm": 0.9330417513847351,
      "learning_rate": 4.249664044568943e-06,
      "loss": 0.5743,
      "step": 1978
    },
    {
      "epoch": 0.5485033259423503,
      "grad_norm": 0.9449990391731262,
      "learning_rate": 4.2453599827814065e-06,
      "loss": 0.6256,
      "step": 1979
    },
    {
      "epoch": 0.5487804878048781,
      "grad_norm": 1.0983213186264038,
      "learning_rate": 4.2410564931393754e-06,
      "loss": 0.5978,
      "step": 1980
    },
    {
      "epoch": 0.5490576496674058,
      "grad_norm": 1.1821951866149902,
      "learning_rate": 4.236753578905627e-06,
      "loss": 0.6127,
      "step": 1981
    },
    {
      "epoch": 0.5493348115299335,
      "grad_norm": 1.0114191770553589,
      "learning_rate": 4.232451243342505e-06,
      "loss": 0.6491,
      "step": 1982
    },
    {
      "epoch": 0.5496119733924612,
      "grad_norm": 0.8807015419006348,
      "learning_rate": 4.2281494897119105e-06,
      "loss": 0.6386,
      "step": 1983
    },
    {
      "epoch": 0.549889135254989,
      "grad_norm": 1.4195075035095215,
      "learning_rate": 4.223848321275305e-06,
      "loss": 0.5573,
      "step": 1984
    },
    {
      "epoch": 0.5501662971175166,
      "grad_norm": 1.0813356637954712,
      "learning_rate": 4.219547741293701e-06,
      "loss": 0.6905,
      "step": 1985
    },
    {
      "epoch": 0.5504434589800443,
      "grad_norm": 0.9149232506752014,
      "learning_rate": 4.215247753027678e-06,
      "loss": 0.6072,
      "step": 1986
    },
    {
      "epoch": 0.5507206208425721,
      "grad_norm": 0.9031868577003479,
      "learning_rate": 4.210948359737353e-06,
      "loss": 0.624,
      "step": 1987
    },
    {
      "epoch": 0.5509977827050998,
      "grad_norm": 0.9128208160400391,
      "learning_rate": 4.206649564682397e-06,
      "loss": 0.576,
      "step": 1988
    },
    {
      "epoch": 0.5512749445676275,
      "grad_norm": 0.994259774684906,
      "learning_rate": 4.202351371122031e-06,
      "loss": 0.5928,
      "step": 1989
    },
    {
      "epoch": 0.5515521064301552,
      "grad_norm": 1.0523613691329956,
      "learning_rate": 4.198053782315015e-06,
      "loss": 0.6312,
      "step": 1990
    },
    {
      "epoch": 0.551829268292683,
      "grad_norm": 0.8839083313941956,
      "learning_rate": 4.193756801519654e-06,
      "loss": 0.5778,
      "step": 1991
    },
    {
      "epoch": 0.5521064301552107,
      "grad_norm": 0.9347401261329651,
      "learning_rate": 4.189460431993788e-06,
      "loss": 0.5849,
      "step": 1992
    },
    {
      "epoch": 0.5523835920177383,
      "grad_norm": 0.9494583010673523,
      "learning_rate": 4.185164676994798e-06,
      "loss": 0.6397,
      "step": 1993
    },
    {
      "epoch": 0.552660753880266,
      "grad_norm": 1.0372991561889648,
      "learning_rate": 4.180869539779598e-06,
      "loss": 0.5909,
      "step": 1994
    },
    {
      "epoch": 0.5529379157427938,
      "grad_norm": 0.964820146560669,
      "learning_rate": 4.1765750236046296e-06,
      "loss": 0.6452,
      "step": 1995
    },
    {
      "epoch": 0.5532150776053215,
      "grad_norm": 0.9372079968452454,
      "learning_rate": 4.172281131725866e-06,
      "loss": 0.6358,
      "step": 1996
    },
    {
      "epoch": 0.5534922394678492,
      "grad_norm": 0.926743745803833,
      "learning_rate": 4.1679878673988135e-06,
      "loss": 0.6829,
      "step": 1997
    },
    {
      "epoch": 0.5537694013303769,
      "grad_norm": 1.162719488143921,
      "learning_rate": 4.163695233878491e-06,
      "loss": 0.6105,
      "step": 1998
    },
    {
      "epoch": 0.5540465631929047,
      "grad_norm": 25.200119018554688,
      "learning_rate": 4.159403234419448e-06,
      "loss": 0.6652,
      "step": 1999
    },
    {
      "epoch": 0.5543237250554324,
      "grad_norm": 0.9593657851219177,
      "learning_rate": 4.1551118722757475e-06,
      "loss": 0.5766,
      "step": 2000
    },
    {
      "epoch": 0.5546008869179601,
      "grad_norm": 0.9257370829582214,
      "learning_rate": 4.150821150700975e-06,
      "loss": 0.6325,
      "step": 2001
    },
    {
      "epoch": 0.5548780487804879,
      "grad_norm": 0.9575889706611633,
      "learning_rate": 4.146531072948226e-06,
      "loss": 0.5536,
      "step": 2002
    },
    {
      "epoch": 0.5551552106430155,
      "grad_norm": 0.9062398672103882,
      "learning_rate": 4.142241642270109e-06,
      "loss": 0.6353,
      "step": 2003
    },
    {
      "epoch": 0.5554323725055432,
      "grad_norm": 0.9405509829521179,
      "learning_rate": 4.137952861918743e-06,
      "loss": 0.5946,
      "step": 2004
    },
    {
      "epoch": 0.5557095343680709,
      "grad_norm": 0.9740002155303955,
      "learning_rate": 4.133664735145753e-06,
      "loss": 0.5658,
      "step": 2005
    },
    {
      "epoch": 0.5559866962305987,
      "grad_norm": 0.9601983428001404,
      "learning_rate": 4.129377265202267e-06,
      "loss": 0.6493,
      "step": 2006
    },
    {
      "epoch": 0.5562638580931264,
      "grad_norm": 0.943874180316925,
      "learning_rate": 4.125090455338917e-06,
      "loss": 0.6488,
      "step": 2007
    },
    {
      "epoch": 0.5565410199556541,
      "grad_norm": 0.9879426956176758,
      "learning_rate": 4.120804308805836e-06,
      "loss": 0.6056,
      "step": 2008
    },
    {
      "epoch": 0.5568181818181818,
      "grad_norm": 1.0218994617462158,
      "learning_rate": 4.116518828852652e-06,
      "loss": 0.5927,
      "step": 2009
    },
    {
      "epoch": 0.5570953436807096,
      "grad_norm": 0.9353734254837036,
      "learning_rate": 4.1122340187284845e-06,
      "loss": 0.6656,
      "step": 2010
    },
    {
      "epoch": 0.5573725055432373,
      "grad_norm": 0.934982180595398,
      "learning_rate": 4.107949881681951e-06,
      "loss": 0.6184,
      "step": 2011
    },
    {
      "epoch": 0.5576496674057649,
      "grad_norm": 0.9512789249420166,
      "learning_rate": 4.103666420961157e-06,
      "loss": 0.6191,
      "step": 2012
    },
    {
      "epoch": 0.5579268292682927,
      "grad_norm": 0.9364797472953796,
      "learning_rate": 4.0993836398136934e-06,
      "loss": 0.6641,
      "step": 2013
    },
    {
      "epoch": 0.5582039911308204,
      "grad_norm": 0.9196001887321472,
      "learning_rate": 4.095101541486636e-06,
      "loss": 0.6121,
      "step": 2014
    },
    {
      "epoch": 0.5584811529933481,
      "grad_norm": 0.9241234660148621,
      "learning_rate": 4.090820129226544e-06,
      "loss": 0.5882,
      "step": 2015
    },
    {
      "epoch": 0.5587583148558758,
      "grad_norm": 0.9520816206932068,
      "learning_rate": 4.08653940627946e-06,
      "loss": 0.6291,
      "step": 2016
    },
    {
      "epoch": 0.5590354767184036,
      "grad_norm": 0.8972570300102234,
      "learning_rate": 4.082259375890894e-06,
      "loss": 0.6127,
      "step": 2017
    },
    {
      "epoch": 0.5593126385809313,
      "grad_norm": 0.9368783831596375,
      "learning_rate": 4.07798004130584e-06,
      "loss": 0.6175,
      "step": 2018
    },
    {
      "epoch": 0.559589800443459,
      "grad_norm": 0.8947859406471252,
      "learning_rate": 4.073701405768763e-06,
      "loss": 0.6062,
      "step": 2019
    },
    {
      "epoch": 0.5598669623059866,
      "grad_norm": 0.9883090257644653,
      "learning_rate": 4.069423472523595e-06,
      "loss": 0.6228,
      "step": 2020
    },
    {
      "epoch": 0.5601441241685144,
      "grad_norm": 0.915278434753418,
      "learning_rate": 4.065146244813738e-06,
      "loss": 0.5951,
      "step": 2021
    },
    {
      "epoch": 0.5604212860310421,
      "grad_norm": 0.9109216332435608,
      "learning_rate": 4.060869725882056e-06,
      "loss": 0.5953,
      "step": 2022
    },
    {
      "epoch": 0.5606984478935698,
      "grad_norm": 1.0380358695983887,
      "learning_rate": 4.056593918970881e-06,
      "loss": 0.6054,
      "step": 2023
    },
    {
      "epoch": 0.5609756097560976,
      "grad_norm": 1.0287468433380127,
      "learning_rate": 4.0523188273219995e-06,
      "loss": 0.6685,
      "step": 2024
    },
    {
      "epoch": 0.5612527716186253,
      "grad_norm": 0.9987754225730896,
      "learning_rate": 4.048044454176658e-06,
      "loss": 0.6327,
      "step": 2025
    },
    {
      "epoch": 0.561529933481153,
      "grad_norm": 0.9561455845832825,
      "learning_rate": 4.043770802775561e-06,
      "loss": 0.6546,
      "step": 2026
    },
    {
      "epoch": 0.5618070953436807,
      "grad_norm": 0.9268828630447388,
      "learning_rate": 4.039497876358864e-06,
      "loss": 0.6485,
      "step": 2027
    },
    {
      "epoch": 0.5620842572062085,
      "grad_norm": 0.9680755138397217,
      "learning_rate": 4.035225678166167e-06,
      "loss": 0.6376,
      "step": 2028
    },
    {
      "epoch": 0.5623614190687362,
      "grad_norm": 0.92108154296875,
      "learning_rate": 4.030954211436526e-06,
      "loss": 0.6401,
      "step": 2029
    },
    {
      "epoch": 0.5626385809312638,
      "grad_norm": 0.9219558238983154,
      "learning_rate": 4.026683479408441e-06,
      "loss": 0.6103,
      "step": 2030
    },
    {
      "epoch": 0.5629157427937915,
      "grad_norm": 0.8897237777709961,
      "learning_rate": 4.022413485319851e-06,
      "loss": 0.5955,
      "step": 2031
    },
    {
      "epoch": 0.5631929046563193,
      "grad_norm": 0.9743934273719788,
      "learning_rate": 4.018144232408139e-06,
      "loss": 0.6781,
      "step": 2032
    },
    {
      "epoch": 0.563470066518847,
      "grad_norm": 0.8995400071144104,
      "learning_rate": 4.013875723910123e-06,
      "loss": 0.6061,
      "step": 2033
    },
    {
      "epoch": 0.5637472283813747,
      "grad_norm": 0.8838605284690857,
      "learning_rate": 4.009607963062062e-06,
      "loss": 0.5366,
      "step": 2034
    },
    {
      "epoch": 0.5640243902439024,
      "grad_norm": 0.9501442909240723,
      "learning_rate": 4.005340953099644e-06,
      "loss": 0.6024,
      "step": 2035
    },
    {
      "epoch": 0.5643015521064302,
      "grad_norm": 0.9234347939491272,
      "learning_rate": 4.0010746972579865e-06,
      "loss": 0.6565,
      "step": 2036
    },
    {
      "epoch": 0.5645787139689579,
      "grad_norm": 0.9640131592750549,
      "learning_rate": 3.9968091987716344e-06,
      "loss": 0.6401,
      "step": 2037
    },
    {
      "epoch": 0.5648558758314856,
      "grad_norm": 0.9318749904632568,
      "learning_rate": 3.992544460874569e-06,
      "loss": 0.6095,
      "step": 2038
    },
    {
      "epoch": 0.5651330376940134,
      "grad_norm": 0.8990483283996582,
      "learning_rate": 3.988280486800182e-06,
      "loss": 0.5508,
      "step": 2039
    },
    {
      "epoch": 0.565410199556541,
      "grad_norm": 0.9361546635627747,
      "learning_rate": 3.984017279781291e-06,
      "loss": 0.661,
      "step": 2040
    },
    {
      "epoch": 0.5656873614190687,
      "grad_norm": 1.0334283113479614,
      "learning_rate": 3.9797548430501335e-06,
      "loss": 0.5844,
      "step": 2041
    },
    {
      "epoch": 0.5659645232815964,
      "grad_norm": 1.0047529935836792,
      "learning_rate": 3.975493179838362e-06,
      "loss": 0.6572,
      "step": 2042
    },
    {
      "epoch": 0.5662416851441242,
      "grad_norm": 0.9088764190673828,
      "learning_rate": 3.971232293377042e-06,
      "loss": 0.6088,
      "step": 2043
    },
    {
      "epoch": 0.5665188470066519,
      "grad_norm": 0.9695283770561218,
      "learning_rate": 3.966972186896647e-06,
      "loss": 0.6065,
      "step": 2044
    },
    {
      "epoch": 0.5667960088691796,
      "grad_norm": 1.05171537399292,
      "learning_rate": 3.962712863627067e-06,
      "loss": 0.6451,
      "step": 2045
    },
    {
      "epoch": 0.5670731707317073,
      "grad_norm": 1.035414218902588,
      "learning_rate": 3.9584543267975925e-06,
      "loss": 0.661,
      "step": 2046
    },
    {
      "epoch": 0.5673503325942351,
      "grad_norm": 2.366011619567871,
      "learning_rate": 3.954196579636918e-06,
      "loss": 0.6978,
      "step": 2047
    },
    {
      "epoch": 0.5676274944567627,
      "grad_norm": 0.9780608415603638,
      "learning_rate": 3.9499396253731385e-06,
      "loss": 0.6166,
      "step": 2048
    },
    {
      "epoch": 0.5679046563192904,
      "grad_norm": 0.9367539882659912,
      "learning_rate": 3.945683467233752e-06,
      "loss": 0.6571,
      "step": 2049
    },
    {
      "epoch": 0.5681818181818182,
      "grad_norm": 0.9372556209564209,
      "learning_rate": 3.94142810844565e-06,
      "loss": 0.6266,
      "step": 2050
    },
    {
      "epoch": 0.5684589800443459,
      "grad_norm": 0.9685991406440735,
      "learning_rate": 3.937173552235117e-06,
      "loss": 0.6285,
      "step": 2051
    },
    {
      "epoch": 0.5687361419068736,
      "grad_norm": 0.9765574932098389,
      "learning_rate": 3.932919801827832e-06,
      "loss": 0.6178,
      "step": 2052
    },
    {
      "epoch": 0.5690133037694013,
      "grad_norm": 0.9493586421012878,
      "learning_rate": 3.928666860448861e-06,
      "loss": 0.61,
      "step": 2053
    },
    {
      "epoch": 0.5692904656319291,
      "grad_norm": 1.2602441310882568,
      "learning_rate": 3.924414731322656e-06,
      "loss": 0.5989,
      "step": 2054
    },
    {
      "epoch": 0.5695676274944568,
      "grad_norm": 1.028871774673462,
      "learning_rate": 3.920163417673055e-06,
      "loss": 0.6557,
      "step": 2055
    },
    {
      "epoch": 0.5698447893569845,
      "grad_norm": 0.9398410320281982,
      "learning_rate": 3.915912922723278e-06,
      "loss": 0.6542,
      "step": 2056
    },
    {
      "epoch": 0.5701219512195121,
      "grad_norm": 0.9321939945220947,
      "learning_rate": 3.9116632496959224e-06,
      "loss": 0.6364,
      "step": 2057
    },
    {
      "epoch": 0.5703991130820399,
      "grad_norm": 0.9160268902778625,
      "learning_rate": 3.907414401812963e-06,
      "loss": 0.6173,
      "step": 2058
    },
    {
      "epoch": 0.5706762749445676,
      "grad_norm": 3.194167137145996,
      "learning_rate": 3.903166382295748e-06,
      "loss": 0.6408,
      "step": 2059
    },
    {
      "epoch": 0.5709534368070953,
      "grad_norm": 0.9748038649559021,
      "learning_rate": 3.898919194365001e-06,
      "loss": 0.6311,
      "step": 2060
    },
    {
      "epoch": 0.5712305986696231,
      "grad_norm": 0.9689584970474243,
      "learning_rate": 3.894672841240812e-06,
      "loss": 0.6276,
      "step": 2061
    },
    {
      "epoch": 0.5715077605321508,
      "grad_norm": 0.8838222026824951,
      "learning_rate": 3.890427326142638e-06,
      "loss": 0.5892,
      "step": 2062
    },
    {
      "epoch": 0.5717849223946785,
      "grad_norm": 0.9741097688674927,
      "learning_rate": 3.8861826522893044e-06,
      "loss": 0.7018,
      "step": 2063
    },
    {
      "epoch": 0.5720620842572062,
      "grad_norm": 0.9057713747024536,
      "learning_rate": 3.881938822898994e-06,
      "loss": 0.643,
      "step": 2064
    },
    {
      "epoch": 0.572339246119734,
      "grad_norm": 0.9110400080680847,
      "learning_rate": 3.877695841189253e-06,
      "loss": 0.6559,
      "step": 2065
    },
    {
      "epoch": 0.5726164079822617,
      "grad_norm": 0.9857572317123413,
      "learning_rate": 3.873453710376979e-06,
      "loss": 0.6323,
      "step": 2066
    },
    {
      "epoch": 0.5728935698447893,
      "grad_norm": 1.254204511642456,
      "learning_rate": 3.869212433678434e-06,
      "loss": 0.6012,
      "step": 2067
    },
    {
      "epoch": 0.573170731707317,
      "grad_norm": 0.9410615563392639,
      "learning_rate": 3.864972014309227e-06,
      "loss": 0.6438,
      "step": 2068
    },
    {
      "epoch": 0.5734478935698448,
      "grad_norm": 1.0086312294006348,
      "learning_rate": 3.860732455484314e-06,
      "loss": 0.626,
      "step": 2069
    },
    {
      "epoch": 0.5737250554323725,
      "grad_norm": 0.9180527329444885,
      "learning_rate": 3.856493760418001e-06,
      "loss": 0.5847,
      "step": 2070
    },
    {
      "epoch": 0.5740022172949002,
      "grad_norm": 0.8774058818817139,
      "learning_rate": 3.852255932323945e-06,
      "loss": 0.5913,
      "step": 2071
    },
    {
      "epoch": 0.5742793791574279,
      "grad_norm": 0.8988126516342163,
      "learning_rate": 3.848018974415137e-06,
      "loss": 0.5618,
      "step": 2072
    },
    {
      "epoch": 0.5745565410199557,
      "grad_norm": 0.971570611000061,
      "learning_rate": 3.843782889903912e-06,
      "loss": 0.6235,
      "step": 2073
    },
    {
      "epoch": 0.5748337028824834,
      "grad_norm": 0.9541396498680115,
      "learning_rate": 3.8395476820019425e-06,
      "loss": 0.6422,
      "step": 2074
    },
    {
      "epoch": 0.575110864745011,
      "grad_norm": 0.9387429356575012,
      "learning_rate": 3.8353133539202395e-06,
      "loss": 0.6047,
      "step": 2075
    },
    {
      "epoch": 0.5753880266075388,
      "grad_norm": 1.0678415298461914,
      "learning_rate": 3.831079908869142e-06,
      "loss": 0.5878,
      "step": 2076
    },
    {
      "epoch": 0.5756651884700665,
      "grad_norm": 0.9405881762504578,
      "learning_rate": 3.826847350058318e-06,
      "loss": 0.653,
      "step": 2077
    },
    {
      "epoch": 0.5759423503325942,
      "grad_norm": 1.1655982732772827,
      "learning_rate": 3.822615680696773e-06,
      "loss": 0.6265,
      "step": 2078
    },
    {
      "epoch": 0.5762195121951219,
      "grad_norm": 0.9040281772613525,
      "learning_rate": 3.81838490399283e-06,
      "loss": 0.5957,
      "step": 2079
    },
    {
      "epoch": 0.5764966740576497,
      "grad_norm": 0.8968104720115662,
      "learning_rate": 3.814155023154136e-06,
      "loss": 0.6059,
      "step": 2080
    },
    {
      "epoch": 0.5767738359201774,
      "grad_norm": 0.922947347164154,
      "learning_rate": 3.8099260413876616e-06,
      "loss": 0.6406,
      "step": 2081
    },
    {
      "epoch": 0.5770509977827051,
      "grad_norm": 0.9375759363174438,
      "learning_rate": 3.805697961899694e-06,
      "loss": 0.6346,
      "step": 2082
    },
    {
      "epoch": 0.5773281596452328,
      "grad_norm": 0.9134944677352905,
      "learning_rate": 3.8014707878958372e-06,
      "loss": 0.6135,
      "step": 2083
    },
    {
      "epoch": 0.5776053215077606,
      "grad_norm": 0.9046237468719482,
      "learning_rate": 3.797244522581008e-06,
      "loss": 0.6226,
      "step": 2084
    },
    {
      "epoch": 0.5778824833702882,
      "grad_norm": 0.9711705446243286,
      "learning_rate": 3.7930191691594327e-06,
      "loss": 0.69,
      "step": 2085
    },
    {
      "epoch": 0.5781596452328159,
      "grad_norm": 3.3938992023468018,
      "learning_rate": 3.788794730834651e-06,
      "loss": 0.6031,
      "step": 2086
    },
    {
      "epoch": 0.5784368070953437,
      "grad_norm": 0.940595269203186,
      "learning_rate": 3.7845712108095053e-06,
      "loss": 0.6222,
      "step": 2087
    },
    {
      "epoch": 0.5787139689578714,
      "grad_norm": 1.0249360799789429,
      "learning_rate": 3.7803486122861377e-06,
      "loss": 0.6803,
      "step": 2088
    },
    {
      "epoch": 0.5789911308203991,
      "grad_norm": 1.3544152975082397,
      "learning_rate": 3.776126938466003e-06,
      "loss": 0.6418,
      "step": 2089
    },
    {
      "epoch": 0.5792682926829268,
      "grad_norm": 0.9240651726722717,
      "learning_rate": 3.7719061925498445e-06,
      "loss": 0.5801,
      "step": 2090
    },
    {
      "epoch": 0.5795454545454546,
      "grad_norm": 0.9353915452957153,
      "learning_rate": 3.7676863777377055e-06,
      "loss": 0.6966,
      "step": 2091
    },
    {
      "epoch": 0.5798226164079823,
      "grad_norm": 0.9819472432136536,
      "learning_rate": 3.7634674972289227e-06,
      "loss": 0.5562,
      "step": 2092
    },
    {
      "epoch": 0.58009977827051,
      "grad_norm": 1.5322479009628296,
      "learning_rate": 3.759249554222128e-06,
      "loss": 0.731,
      "step": 2093
    },
    {
      "epoch": 0.5803769401330376,
      "grad_norm": 0.9671079516410828,
      "learning_rate": 3.7550325519152374e-06,
      "loss": 0.6664,
      "step": 2094
    },
    {
      "epoch": 0.5806541019955654,
      "grad_norm": 0.9898436665534973,
      "learning_rate": 3.7508164935054586e-06,
      "loss": 0.6038,
      "step": 2095
    },
    {
      "epoch": 0.5809312638580931,
      "grad_norm": 0.940621554851532,
      "learning_rate": 3.746601382189276e-06,
      "loss": 0.5802,
      "step": 2096
    },
    {
      "epoch": 0.5812084257206208,
      "grad_norm": 0.9115685820579529,
      "learning_rate": 3.742387221162466e-06,
      "loss": 0.6113,
      "step": 2097
    },
    {
      "epoch": 0.5814855875831486,
      "grad_norm": 0.9867151379585266,
      "learning_rate": 3.7381740136200795e-06,
      "loss": 0.6107,
      "step": 2098
    },
    {
      "epoch": 0.5817627494456763,
      "grad_norm": 1.3976001739501953,
      "learning_rate": 3.7339617627564395e-06,
      "loss": 0.65,
      "step": 2099
    },
    {
      "epoch": 0.582039911308204,
      "grad_norm": 1.0150848627090454,
      "learning_rate": 3.7297504717651552e-06,
      "loss": 0.6424,
      "step": 2100
    },
    {
      "epoch": 0.5823170731707317,
      "grad_norm": 0.9045628309249878,
      "learning_rate": 3.725540143839097e-06,
      "loss": 0.6094,
      "step": 2101
    },
    {
      "epoch": 0.5825942350332595,
      "grad_norm": 0.9201278686523438,
      "learning_rate": 3.7213307821704115e-06,
      "loss": 0.6534,
      "step": 2102
    },
    {
      "epoch": 0.5828713968957872,
      "grad_norm": 0.9796575903892517,
      "learning_rate": 3.717122389950509e-06,
      "loss": 0.6652,
      "step": 2103
    },
    {
      "epoch": 0.5831485587583148,
      "grad_norm": 1.0790433883666992,
      "learning_rate": 3.7129149703700708e-06,
      "loss": 0.6235,
      "step": 2104
    },
    {
      "epoch": 0.5834257206208425,
      "grad_norm": 0.9277780652046204,
      "learning_rate": 3.7087085266190324e-06,
      "loss": 0.6506,
      "step": 2105
    },
    {
      "epoch": 0.5837028824833703,
      "grad_norm": 0.8882150053977966,
      "learning_rate": 3.7045030618865973e-06,
      "loss": 0.6171,
      "step": 2106
    },
    {
      "epoch": 0.583980044345898,
      "grad_norm": 0.9362030625343323,
      "learning_rate": 3.700298579361218e-06,
      "loss": 0.5882,
      "step": 2107
    },
    {
      "epoch": 0.5842572062084257,
      "grad_norm": 1.0750738382339478,
      "learning_rate": 3.6960950822306143e-06,
      "loss": 0.5964,
      "step": 2108
    },
    {
      "epoch": 0.5845343680709535,
      "grad_norm": 0.91434645652771,
      "learning_rate": 3.6918925736817474e-06,
      "loss": 0.6227,
      "step": 2109
    },
    {
      "epoch": 0.5848115299334812,
      "grad_norm": 0.88181072473526,
      "learning_rate": 3.6876910569008327e-06,
      "loss": 0.6145,
      "step": 2110
    },
    {
      "epoch": 0.5850886917960089,
      "grad_norm": 0.9677065014839172,
      "learning_rate": 3.6834905350733395e-06,
      "loss": 0.5764,
      "step": 2111
    },
    {
      "epoch": 0.5853658536585366,
      "grad_norm": 0.9593001008033752,
      "learning_rate": 3.6792910113839755e-06,
      "loss": 0.634,
      "step": 2112
    },
    {
      "epoch": 0.5856430155210643,
      "grad_norm": 0.8889330625534058,
      "learning_rate": 3.675092489016693e-06,
      "loss": 0.6039,
      "step": 2113
    },
    {
      "epoch": 0.585920177383592,
      "grad_norm": 0.9357283711433411,
      "learning_rate": 3.6708949711546855e-06,
      "loss": 0.6395,
      "step": 2114
    },
    {
      "epoch": 0.5861973392461197,
      "grad_norm": 0.9623759984970093,
      "learning_rate": 3.6666984609803884e-06,
      "loss": 0.613,
      "step": 2115
    },
    {
      "epoch": 0.5864745011086474,
      "grad_norm": 0.882182240486145,
      "learning_rate": 3.662502961675469e-06,
      "loss": 0.5691,
      "step": 2116
    },
    {
      "epoch": 0.5867516629711752,
      "grad_norm": 0.8748295903205872,
      "learning_rate": 3.658308476420831e-06,
      "loss": 0.6343,
      "step": 2117
    },
    {
      "epoch": 0.5870288248337029,
      "grad_norm": 0.8942072987556458,
      "learning_rate": 3.654115008396602e-06,
      "loss": 0.6209,
      "step": 2118
    },
    {
      "epoch": 0.5873059866962306,
      "grad_norm": 1.8384547233581543,
      "learning_rate": 3.649922560782152e-06,
      "loss": 0.62,
      "step": 2119
    },
    {
      "epoch": 0.5875831485587583,
      "grad_norm": 0.9456958770751953,
      "learning_rate": 3.645731136756065e-06,
      "loss": 0.6372,
      "step": 2120
    },
    {
      "epoch": 0.5878603104212861,
      "grad_norm": 1.021805763244629,
      "learning_rate": 3.6415407394961536e-06,
      "loss": 0.5992,
      "step": 2121
    },
    {
      "epoch": 0.5881374722838137,
      "grad_norm": 0.9746824502944946,
      "learning_rate": 3.637351372179452e-06,
      "loss": 0.5927,
      "step": 2122
    },
    {
      "epoch": 0.5884146341463414,
      "grad_norm": 0.9158193469047546,
      "learning_rate": 3.633163037982214e-06,
      "loss": 0.6161,
      "step": 2123
    },
    {
      "epoch": 0.5886917960088692,
      "grad_norm": 0.9691551923751831,
      "learning_rate": 3.62897574007991e-06,
      "loss": 0.5857,
      "step": 2124
    },
    {
      "epoch": 0.5889689578713969,
      "grad_norm": 1.6825636625289917,
      "learning_rate": 3.624789481647221e-06,
      "loss": 0.6417,
      "step": 2125
    },
    {
      "epoch": 0.5892461197339246,
      "grad_norm": 1.1903324127197266,
      "learning_rate": 3.620604265858047e-06,
      "loss": 0.64,
      "step": 2126
    },
    {
      "epoch": 0.5895232815964523,
      "grad_norm": 0.9687173962593079,
      "learning_rate": 3.6164200958854916e-06,
      "loss": 0.6295,
      "step": 2127
    },
    {
      "epoch": 0.5898004434589801,
      "grad_norm": 0.9228530526161194,
      "learning_rate": 3.6122369749018686e-06,
      "loss": 0.5821,
      "step": 2128
    },
    {
      "epoch": 0.5900776053215078,
      "grad_norm": 3.92156720161438,
      "learning_rate": 3.6080549060786914e-06,
      "loss": 0.6334,
      "step": 2129
    },
    {
      "epoch": 0.5903547671840355,
      "grad_norm": 1.1126471757888794,
      "learning_rate": 3.603873892586685e-06,
      "loss": 0.5728,
      "step": 2130
    },
    {
      "epoch": 0.5906319290465631,
      "grad_norm": 1.0013856887817383,
      "learning_rate": 3.5996939375957657e-06,
      "loss": 0.6406,
      "step": 2131
    },
    {
      "epoch": 0.5909090909090909,
      "grad_norm": 0.904280424118042,
      "learning_rate": 3.5955150442750507e-06,
      "loss": 0.5624,
      "step": 2132
    },
    {
      "epoch": 0.5911862527716186,
      "grad_norm": 0.873333752155304,
      "learning_rate": 3.5913372157928515e-06,
      "loss": 0.6134,
      "step": 2133
    },
    {
      "epoch": 0.5914634146341463,
      "grad_norm": 1.030680537223816,
      "learning_rate": 3.587160455316675e-06,
      "loss": 0.6894,
      "step": 2134
    },
    {
      "epoch": 0.5917405764966741,
      "grad_norm": 0.9405762553215027,
      "learning_rate": 3.582984766013215e-06,
      "loss": 0.6569,
      "step": 2135
    },
    {
      "epoch": 0.5920177383592018,
      "grad_norm": 0.9706565141677856,
      "learning_rate": 3.5788101510483525e-06,
      "loss": 0.5676,
      "step": 2136
    },
    {
      "epoch": 0.5922949002217295,
      "grad_norm": 0.9130007028579712,
      "learning_rate": 3.574636613587159e-06,
      "loss": 0.6104,
      "step": 2137
    },
    {
      "epoch": 0.5925720620842572,
      "grad_norm": 0.9499661922454834,
      "learning_rate": 3.5704641567938845e-06,
      "loss": 0.5767,
      "step": 2138
    },
    {
      "epoch": 0.592849223946785,
      "grad_norm": 0.933824896812439,
      "learning_rate": 3.566292783831961e-06,
      "loss": 0.6669,
      "step": 2139
    },
    {
      "epoch": 0.5931263858093127,
      "grad_norm": 0.9359897375106812,
      "learning_rate": 3.5621224978639964e-06,
      "loss": 0.5833,
      "step": 2140
    },
    {
      "epoch": 0.5934035476718403,
      "grad_norm": 0.9414730072021484,
      "learning_rate": 3.557953302051782e-06,
      "loss": 0.5889,
      "step": 2141
    },
    {
      "epoch": 0.593680709534368,
      "grad_norm": 0.9129725694656372,
      "learning_rate": 3.553785199556273e-06,
      "loss": 0.6325,
      "step": 2142
    },
    {
      "epoch": 0.5939578713968958,
      "grad_norm": 0.8998770117759705,
      "learning_rate": 3.549618193537602e-06,
      "loss": 0.5699,
      "step": 2143
    },
    {
      "epoch": 0.5942350332594235,
      "grad_norm": 0.9255713224411011,
      "learning_rate": 3.545452287155068e-06,
      "loss": 0.5728,
      "step": 2144
    },
    {
      "epoch": 0.5945121951219512,
      "grad_norm": 0.9057931303977966,
      "learning_rate": 3.541287483567137e-06,
      "loss": 0.5973,
      "step": 2145
    },
    {
      "epoch": 0.594789356984479,
      "grad_norm": 0.9988406300544739,
      "learning_rate": 3.5371237859314387e-06,
      "loss": 0.6079,
      "step": 2146
    },
    {
      "epoch": 0.5950665188470067,
      "grad_norm": 0.8964307904243469,
      "learning_rate": 3.532961197404763e-06,
      "loss": 0.57,
      "step": 2147
    },
    {
      "epoch": 0.5953436807095344,
      "grad_norm": 0.8950427174568176,
      "learning_rate": 3.5287997211430623e-06,
      "loss": 0.5617,
      "step": 2148
    },
    {
      "epoch": 0.595620842572062,
      "grad_norm": 0.9923955202102661,
      "learning_rate": 3.5246393603014435e-06,
      "loss": 0.6769,
      "step": 2149
    },
    {
      "epoch": 0.5958980044345898,
      "grad_norm": 0.9337422847747803,
      "learning_rate": 3.520480118034165e-06,
      "loss": 0.6502,
      "step": 2150
    },
    {
      "epoch": 0.5961751662971175,
      "grad_norm": 0.9283485412597656,
      "learning_rate": 3.516321997494641e-06,
      "loss": 0.5652,
      "step": 2151
    },
    {
      "epoch": 0.5964523281596452,
      "grad_norm": 0.9011205434799194,
      "learning_rate": 3.512165001835436e-06,
      "loss": 0.6072,
      "step": 2152
    },
    {
      "epoch": 0.5967294900221729,
      "grad_norm": 0.9607095122337341,
      "learning_rate": 3.508009134208259e-06,
      "loss": 0.5936,
      "step": 2153
    },
    {
      "epoch": 0.5970066518847007,
      "grad_norm": 0.9380379915237427,
      "learning_rate": 3.5038543977639637e-06,
      "loss": 0.6554,
      "step": 2154
    },
    {
      "epoch": 0.5972838137472284,
      "grad_norm": 1.0177109241485596,
      "learning_rate": 3.4997007956525463e-06,
      "loss": 0.6169,
      "step": 2155
    },
    {
      "epoch": 0.5975609756097561,
      "grad_norm": 1.0223538875579834,
      "learning_rate": 3.4955483310231476e-06,
      "loss": 0.6535,
      "step": 2156
    },
    {
      "epoch": 0.5978381374722838,
      "grad_norm": 1.0145108699798584,
      "learning_rate": 3.4913970070240388e-06,
      "loss": 0.634,
      "step": 2157
    },
    {
      "epoch": 0.5981152993348116,
      "grad_norm": 0.905261754989624,
      "learning_rate": 3.487246826802632e-06,
      "loss": 0.595,
      "step": 2158
    },
    {
      "epoch": 0.5983924611973392,
      "grad_norm": 0.9475619196891785,
      "learning_rate": 3.4830977935054653e-06,
      "loss": 0.6234,
      "step": 2159
    },
    {
      "epoch": 0.5986696230598669,
      "grad_norm": 1.0885118246078491,
      "learning_rate": 3.478949910278219e-06,
      "loss": 0.6302,
      "step": 2160
    },
    {
      "epoch": 0.5989467849223947,
      "grad_norm": 0.897792398929596,
      "learning_rate": 3.474803180265688e-06,
      "loss": 0.6412,
      "step": 2161
    },
    {
      "epoch": 0.5992239467849224,
      "grad_norm": 0.9830977320671082,
      "learning_rate": 3.4706576066118003e-06,
      "loss": 0.6216,
      "step": 2162
    },
    {
      "epoch": 0.5995011086474501,
      "grad_norm": 0.9700034260749817,
      "learning_rate": 3.466513192459608e-06,
      "loss": 0.6392,
      "step": 2163
    },
    {
      "epoch": 0.5997782705099778,
      "grad_norm": 0.9777757525444031,
      "learning_rate": 3.4623699409512806e-06,
      "loss": 0.5908,
      "step": 2164
    },
    {
      "epoch": 0.6000554323725056,
      "grad_norm": 0.9921084642410278,
      "learning_rate": 3.458227855228108e-06,
      "loss": 0.6506,
      "step": 2165
    },
    {
      "epoch": 0.6003325942350333,
      "grad_norm": 0.924079954624176,
      "learning_rate": 3.4540869384304955e-06,
      "loss": 0.6166,
      "step": 2166
    },
    {
      "epoch": 0.6003325942350333,
      "eval_loss": 0.6206091046333313,
      "eval_runtime": 387.6032,
      "eval_samples_per_second": 8.333,
      "eval_steps_per_second": 1.042,
      "step": 2166
    },
    {
      "epoch": 0.600609756097561,
      "grad_norm": 0.9212763905525208,
      "learning_rate": 3.449947193697962e-06,
      "loss": 0.6212,
      "step": 2167
    },
    {
      "epoch": 0.6008869179600886,
      "grad_norm": 1.1291515827178955,
      "learning_rate": 3.445808624169141e-06,
      "loss": 0.6387,
      "step": 2168
    },
    {
      "epoch": 0.6011640798226164,
      "grad_norm": 0.908785879611969,
      "learning_rate": 3.4416712329817686e-06,
      "loss": 0.6638,
      "step": 2169
    },
    {
      "epoch": 0.6014412416851441,
      "grad_norm": 1.0202382802963257,
      "learning_rate": 3.4375350232726906e-06,
      "loss": 0.6497,
      "step": 2170
    },
    {
      "epoch": 0.6017184035476718,
      "grad_norm": 0.9061534404754639,
      "learning_rate": 3.4333999981778627e-06,
      "loss": 0.5628,
      "step": 2171
    },
    {
      "epoch": 0.6019955654101996,
      "grad_norm": 0.9259747862815857,
      "learning_rate": 3.429266160832333e-06,
      "loss": 0.6216,
      "step": 2172
    },
    {
      "epoch": 0.6022727272727273,
      "grad_norm": 0.9232959747314453,
      "learning_rate": 3.4251335143702536e-06,
      "loss": 0.6245,
      "step": 2173
    },
    {
      "epoch": 0.602549889135255,
      "grad_norm": 0.975280225276947,
      "learning_rate": 3.4210020619248762e-06,
      "loss": 0.6507,
      "step": 2174
    },
    {
      "epoch": 0.6028270509977827,
      "grad_norm": 0.9640636444091797,
      "learning_rate": 3.4168718066285423e-06,
      "loss": 0.6707,
      "step": 2175
    },
    {
      "epoch": 0.6031042128603105,
      "grad_norm": 0.9069342613220215,
      "learning_rate": 3.4127427516126887e-06,
      "loss": 0.6345,
      "step": 2176
    },
    {
      "epoch": 0.6033813747228381,
      "grad_norm": 0.9064071178436279,
      "learning_rate": 3.4086149000078407e-06,
      "loss": 0.6109,
      "step": 2177
    },
    {
      "epoch": 0.6036585365853658,
      "grad_norm": 0.9506678581237793,
      "learning_rate": 3.404488254943613e-06,
      "loss": 0.621,
      "step": 2178
    },
    {
      "epoch": 0.6039356984478935,
      "grad_norm": 0.9344003796577454,
      "learning_rate": 3.400362819548706e-06,
      "loss": 0.5815,
      "step": 2179
    },
    {
      "epoch": 0.6042128603104213,
      "grad_norm": 0.9407650828361511,
      "learning_rate": 3.396238596950899e-06,
      "loss": 0.6311,
      "step": 2180
    },
    {
      "epoch": 0.604490022172949,
      "grad_norm": 1.0108115673065186,
      "learning_rate": 3.3921155902770536e-06,
      "loss": 0.5766,
      "step": 2181
    },
    {
      "epoch": 0.6047671840354767,
      "grad_norm": 0.9091290831565857,
      "learning_rate": 3.3879938026531134e-06,
      "loss": 0.6222,
      "step": 2182
    },
    {
      "epoch": 0.6050443458980045,
      "grad_norm": 1.0003244876861572,
      "learning_rate": 3.3838732372040927e-06,
      "loss": 0.5903,
      "step": 2183
    },
    {
      "epoch": 0.6053215077605322,
      "grad_norm": 0.9043534398078918,
      "learning_rate": 3.37975389705408e-06,
      "loss": 0.6317,
      "step": 2184
    },
    {
      "epoch": 0.6055986696230599,
      "grad_norm": 0.917094349861145,
      "learning_rate": 3.3756357853262386e-06,
      "loss": 0.5978,
      "step": 2185
    },
    {
      "epoch": 0.6058758314855875,
      "grad_norm": 0.911262035369873,
      "learning_rate": 3.371518905142796e-06,
      "loss": 0.6138,
      "step": 2186
    },
    {
      "epoch": 0.6061529933481153,
      "grad_norm": 1.1653263568878174,
      "learning_rate": 3.3674032596250483e-06,
      "loss": 0.6599,
      "step": 2187
    },
    {
      "epoch": 0.606430155210643,
      "grad_norm": 1.071895718574524,
      "learning_rate": 3.3632888518933537e-06,
      "loss": 0.6293,
      "step": 2188
    },
    {
      "epoch": 0.6067073170731707,
      "grad_norm": 0.993876039981842,
      "learning_rate": 3.359175685067135e-06,
      "loss": 0.6791,
      "step": 2189
    },
    {
      "epoch": 0.6069844789356984,
      "grad_norm": 0.942128598690033,
      "learning_rate": 3.355063762264873e-06,
      "loss": 0.5931,
      "step": 2190
    },
    {
      "epoch": 0.6072616407982262,
      "grad_norm": 0.9455112814903259,
      "learning_rate": 3.350953086604102e-06,
      "loss": 0.6676,
      "step": 2191
    },
    {
      "epoch": 0.6075388026607539,
      "grad_norm": 0.911321759223938,
      "learning_rate": 3.346843661201413e-06,
      "loss": 0.6048,
      "step": 2192
    },
    {
      "epoch": 0.6078159645232816,
      "grad_norm": 1.333901047706604,
      "learning_rate": 3.342735489172453e-06,
      "loss": 0.597,
      "step": 2193
    },
    {
      "epoch": 0.6080931263858093,
      "grad_norm": 0.9772914052009583,
      "learning_rate": 3.3386285736319123e-06,
      "loss": 0.5795,
      "step": 2194
    },
    {
      "epoch": 0.608370288248337,
      "grad_norm": 0.9488179683685303,
      "learning_rate": 3.3345229176935332e-06,
      "loss": 0.6567,
      "step": 2195
    },
    {
      "epoch": 0.6086474501108647,
      "grad_norm": 1.0120118856430054,
      "learning_rate": 3.3304185244701005e-06,
      "loss": 0.5635,
      "step": 2196
    },
    {
      "epoch": 0.6089246119733924,
      "grad_norm": 0.8673779368400574,
      "learning_rate": 3.3263153970734425e-06,
      "loss": 0.585,
      "step": 2197
    },
    {
      "epoch": 0.6092017738359202,
      "grad_norm": 0.9002432823181152,
      "learning_rate": 3.3222135386144296e-06,
      "loss": 0.6155,
      "step": 2198
    },
    {
      "epoch": 0.6094789356984479,
      "grad_norm": 0.9242809414863586,
      "learning_rate": 3.318112952202963e-06,
      "loss": 0.5919,
      "step": 2199
    },
    {
      "epoch": 0.6097560975609756,
      "grad_norm": 1.0841050148010254,
      "learning_rate": 3.314013640947991e-06,
      "loss": 0.5728,
      "step": 2200
    },
    {
      "epoch": 0.6100332594235033,
      "grad_norm": 0.925150454044342,
      "learning_rate": 3.309915607957487e-06,
      "loss": 0.6179,
      "step": 2201
    },
    {
      "epoch": 0.6103104212860311,
      "grad_norm": 0.9567393660545349,
      "learning_rate": 3.3058188563384557e-06,
      "loss": 0.5782,
      "step": 2202
    },
    {
      "epoch": 0.6105875831485588,
      "grad_norm": 0.9342221021652222,
      "learning_rate": 3.3017233891969315e-06,
      "loss": 0.5937,
      "step": 2203
    },
    {
      "epoch": 0.6108647450110865,
      "grad_norm": 0.9257034063339233,
      "learning_rate": 3.297629209637977e-06,
      "loss": 0.6016,
      "step": 2204
    },
    {
      "epoch": 0.6111419068736141,
      "grad_norm": 0.9247617721557617,
      "learning_rate": 3.2935363207656758e-06,
      "loss": 0.6173,
      "step": 2205
    },
    {
      "epoch": 0.6114190687361419,
      "grad_norm": 0.981719970703125,
      "learning_rate": 3.289444725683134e-06,
      "loss": 0.6604,
      "step": 2206
    },
    {
      "epoch": 0.6116962305986696,
      "grad_norm": 1.4826394319534302,
      "learning_rate": 3.2853544274924754e-06,
      "loss": 0.5549,
      "step": 2207
    },
    {
      "epoch": 0.6119733924611973,
      "grad_norm": 0.919156551361084,
      "learning_rate": 3.2812654292948446e-06,
      "loss": 0.6113,
      "step": 2208
    },
    {
      "epoch": 0.6122505543237251,
      "grad_norm": 1.0025298595428467,
      "learning_rate": 3.277177734190398e-06,
      "loss": 0.7166,
      "step": 2209
    },
    {
      "epoch": 0.6125277161862528,
      "grad_norm": 0.9497832655906677,
      "learning_rate": 3.273091345278298e-06,
      "loss": 0.5784,
      "step": 2210
    },
    {
      "epoch": 0.6128048780487805,
      "grad_norm": 0.9400694966316223,
      "learning_rate": 3.26900626565673e-06,
      "loss": 0.6503,
      "step": 2211
    },
    {
      "epoch": 0.6130820399113082,
      "grad_norm": 1.0779929161071777,
      "learning_rate": 3.2649224984228756e-06,
      "loss": 0.6648,
      "step": 2212
    },
    {
      "epoch": 0.613359201773836,
      "grad_norm": 1.1921908855438232,
      "learning_rate": 3.2608400466729257e-06,
      "loss": 0.5857,
      "step": 2213
    },
    {
      "epoch": 0.6136363636363636,
      "grad_norm": 0.9252128005027771,
      "learning_rate": 3.256758913502072e-06,
      "loss": 0.6274,
      "step": 2214
    },
    {
      "epoch": 0.6139135254988913,
      "grad_norm": 1.0689862966537476,
      "learning_rate": 3.252679102004509e-06,
      "loss": 0.63,
      "step": 2215
    },
    {
      "epoch": 0.614190687361419,
      "grad_norm": 1.3509533405303955,
      "learning_rate": 3.2486006152734277e-06,
      "loss": 0.6054,
      "step": 2216
    },
    {
      "epoch": 0.6144678492239468,
      "grad_norm": 0.8919309973716736,
      "learning_rate": 3.2445234564010154e-06,
      "loss": 0.6158,
      "step": 2217
    },
    {
      "epoch": 0.6147450110864745,
      "grad_norm": 1.00626540184021,
      "learning_rate": 3.240447628478449e-06,
      "loss": 0.6107,
      "step": 2218
    },
    {
      "epoch": 0.6150221729490022,
      "grad_norm": 0.9172050952911377,
      "learning_rate": 3.2363731345959044e-06,
      "loss": 0.6139,
      "step": 2219
    },
    {
      "epoch": 0.61529933481153,
      "grad_norm": 1.0072675943374634,
      "learning_rate": 3.232299977842539e-06,
      "loss": 0.6721,
      "step": 2220
    },
    {
      "epoch": 0.6155764966740577,
      "grad_norm": 0.9451637864112854,
      "learning_rate": 3.228228161306495e-06,
      "loss": 0.6093,
      "step": 2221
    },
    {
      "epoch": 0.6158536585365854,
      "grad_norm": 0.9737775921821594,
      "learning_rate": 3.224157688074907e-06,
      "loss": 0.6371,
      "step": 2222
    },
    {
      "epoch": 0.616130820399113,
      "grad_norm": 1.0541902780532837,
      "learning_rate": 3.2200885612338846e-06,
      "loss": 0.6506,
      "step": 2223
    },
    {
      "epoch": 0.6164079822616408,
      "grad_norm": 1.0016262531280518,
      "learning_rate": 3.216020783868517e-06,
      "loss": 0.6423,
      "step": 2224
    },
    {
      "epoch": 0.6166851441241685,
      "grad_norm": 0.9521600008010864,
      "learning_rate": 3.211954359062871e-06,
      "loss": 0.5951,
      "step": 2225
    },
    {
      "epoch": 0.6169623059866962,
      "grad_norm": 0.9885860085487366,
      "learning_rate": 3.2078892898999914e-06,
      "loss": 0.5916,
      "step": 2226
    },
    {
      "epoch": 0.6172394678492239,
      "grad_norm": 1.526144027709961,
      "learning_rate": 3.203825579461889e-06,
      "loss": 0.6612,
      "step": 2227
    },
    {
      "epoch": 0.6175166297117517,
      "grad_norm": 1.012352466583252,
      "learning_rate": 3.1997632308295507e-06,
      "loss": 0.6819,
      "step": 2228
    },
    {
      "epoch": 0.6177937915742794,
      "grad_norm": 1.0651805400848389,
      "learning_rate": 3.1957022470829224e-06,
      "loss": 0.6281,
      "step": 2229
    },
    {
      "epoch": 0.6180709534368071,
      "grad_norm": 0.933239221572876,
      "learning_rate": 3.1916426313009265e-06,
      "loss": 0.6212,
      "step": 2230
    },
    {
      "epoch": 0.6183481152993349,
      "grad_norm": 1.041180968284607,
      "learning_rate": 3.187584386561441e-06,
      "loss": 0.612,
      "step": 2231
    },
    {
      "epoch": 0.6186252771618626,
      "grad_norm": 0.94051194190979,
      "learning_rate": 3.1835275159413014e-06,
      "loss": 0.567,
      "step": 2232
    },
    {
      "epoch": 0.6189024390243902,
      "grad_norm": 0.9452318549156189,
      "learning_rate": 3.1794720225163133e-06,
      "loss": 0.5351,
      "step": 2233
    },
    {
      "epoch": 0.6191796008869179,
      "grad_norm": 0.9381417632102966,
      "learning_rate": 3.1754179093612246e-06,
      "loss": 0.6176,
      "step": 2234
    },
    {
      "epoch": 0.6194567627494457,
      "grad_norm": 1.8218015432357788,
      "learning_rate": 3.1713651795497445e-06,
      "loss": 0.625,
      "step": 2235
    },
    {
      "epoch": 0.6197339246119734,
      "grad_norm": 0.974257230758667,
      "learning_rate": 3.167313836154531e-06,
      "loss": 0.5989,
      "step": 2236
    },
    {
      "epoch": 0.6200110864745011,
      "grad_norm": 0.9133803248405457,
      "learning_rate": 3.163263882247193e-06,
      "loss": 0.5962,
      "step": 2237
    },
    {
      "epoch": 0.6202882483370288,
      "grad_norm": 0.9205320477485657,
      "learning_rate": 3.1592153208982834e-06,
      "loss": 0.659,
      "step": 2238
    },
    {
      "epoch": 0.6205654101995566,
      "grad_norm": 0.8690792918205261,
      "learning_rate": 3.1551681551773023e-06,
      "loss": 0.5983,
      "step": 2239
    },
    {
      "epoch": 0.6208425720620843,
      "grad_norm": 0.9533085227012634,
      "learning_rate": 3.1511223881526854e-06,
      "loss": 0.6531,
      "step": 2240
    },
    {
      "epoch": 0.621119733924612,
      "grad_norm": 0.9318209290504456,
      "learning_rate": 3.1470780228918173e-06,
      "loss": 0.5945,
      "step": 2241
    },
    {
      "epoch": 0.6213968957871396,
      "grad_norm": 0.9481802582740784,
      "learning_rate": 3.1430350624610127e-06,
      "loss": 0.6228,
      "step": 2242
    },
    {
      "epoch": 0.6216740576496674,
      "grad_norm": 0.8784264326095581,
      "learning_rate": 3.1389935099255233e-06,
      "loss": 0.5843,
      "step": 2243
    },
    {
      "epoch": 0.6219512195121951,
      "grad_norm": 1.0005888938903809,
      "learning_rate": 3.1349533683495327e-06,
      "loss": 0.5983,
      "step": 2244
    },
    {
      "epoch": 0.6222283813747228,
      "grad_norm": 0.923927366733551,
      "learning_rate": 3.1309146407961565e-06,
      "loss": 0.6175,
      "step": 2245
    },
    {
      "epoch": 0.6225055432372506,
      "grad_norm": 0.8980122208595276,
      "learning_rate": 3.126877330327438e-06,
      "loss": 0.5938,
      "step": 2246
    },
    {
      "epoch": 0.6227827050997783,
      "grad_norm": 0.8575908541679382,
      "learning_rate": 3.1228414400043427e-06,
      "loss": 0.6405,
      "step": 2247
    },
    {
      "epoch": 0.623059866962306,
      "grad_norm": 0.8704333901405334,
      "learning_rate": 3.1188069728867656e-06,
      "loss": 0.5626,
      "step": 2248
    },
    {
      "epoch": 0.6233370288248337,
      "grad_norm": 0.9436385035514832,
      "learning_rate": 3.1147739320335168e-06,
      "loss": 0.6236,
      "step": 2249
    },
    {
      "epoch": 0.6236141906873615,
      "grad_norm": 0.9219684600830078,
      "learning_rate": 3.110742320502329e-06,
      "loss": 0.6065,
      "step": 2250
    },
    {
      "epoch": 0.6238913525498891,
      "grad_norm": 0.9098445177078247,
      "learning_rate": 3.1067121413498447e-06,
      "loss": 0.6346,
      "step": 2251
    },
    {
      "epoch": 0.6241685144124168,
      "grad_norm": 0.9612181186676025,
      "learning_rate": 3.1026833976316327e-06,
      "loss": 0.6427,
      "step": 2252
    },
    {
      "epoch": 0.6244456762749445,
      "grad_norm": 0.9452755451202393,
      "learning_rate": 3.0986560924021607e-06,
      "loss": 0.5732,
      "step": 2253
    },
    {
      "epoch": 0.6247228381374723,
      "grad_norm": 0.976834774017334,
      "learning_rate": 3.094630228714813e-06,
      "loss": 0.6973,
      "step": 2254
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.973034143447876,
      "learning_rate": 3.0906058096218784e-06,
      "loss": 0.5599,
      "step": 2255
    },
    {
      "epoch": 0.6252771618625277,
      "grad_norm": 0.9148198962211609,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.6064,
      "step": 2256
    },
    {
      "epoch": 0.6255543237250555,
      "grad_norm": 0.9199146628379822,
      "learning_rate": 3.08256131742293e-06,
      "loss": 0.5747,
      "step": 2257
    },
    {
      "epoch": 0.6258314855875832,
      "grad_norm": 0.975209653377533,
      "learning_rate": 3.0785412504160086e-06,
      "loss": 0.7056,
      "step": 2258
    },
    {
      "epoch": 0.6261086474501109,
      "grad_norm": 1.0050715208053589,
      "learning_rate": 3.074522640201685e-06,
      "loss": 0.5994,
      "step": 2259
    },
    {
      "epoch": 0.6263858093126385,
      "grad_norm": 0.9690459966659546,
      "learning_rate": 3.070505489826746e-06,
      "loss": 0.6065,
      "step": 2260
    },
    {
      "epoch": 0.6266629711751663,
      "grad_norm": 0.9836888313293457,
      "learning_rate": 3.0664898023368792e-06,
      "loss": 0.5947,
      "step": 2261
    },
    {
      "epoch": 0.626940133037694,
      "grad_norm": 0.923062264919281,
      "learning_rate": 3.062475580776653e-06,
      "loss": 0.6296,
      "step": 2262
    },
    {
      "epoch": 0.6272172949002217,
      "grad_norm": 1.0103152990341187,
      "learning_rate": 3.0584628281895358e-06,
      "loss": 0.6267,
      "step": 2263
    },
    {
      "epoch": 0.6274944567627494,
      "grad_norm": 0.9340530633926392,
      "learning_rate": 3.054451547617873e-06,
      "loss": 0.6649,
      "step": 2264
    },
    {
      "epoch": 0.6277716186252772,
      "grad_norm": 1.9745525121688843,
      "learning_rate": 3.0504417421028997e-06,
      "loss": 0.6128,
      "step": 2265
    },
    {
      "epoch": 0.6280487804878049,
      "grad_norm": 0.8800843358039856,
      "learning_rate": 3.0464334146847284e-06,
      "loss": 0.6102,
      "step": 2266
    },
    {
      "epoch": 0.6283259423503326,
      "grad_norm": 2.4883599281311035,
      "learning_rate": 3.0424265684023556e-06,
      "loss": 0.6293,
      "step": 2267
    },
    {
      "epoch": 0.6286031042128604,
      "grad_norm": 1.0130128860473633,
      "learning_rate": 3.038421206293652e-06,
      "loss": 0.6103,
      "step": 2268
    },
    {
      "epoch": 0.628880266075388,
      "grad_norm": 0.9179598689079285,
      "learning_rate": 3.034417331395361e-06,
      "loss": 0.6112,
      "step": 2269
    },
    {
      "epoch": 0.6291574279379157,
      "grad_norm": 0.9985660314559937,
      "learning_rate": 3.030414946743104e-06,
      "loss": 0.6444,
      "step": 2270
    },
    {
      "epoch": 0.6294345898004434,
      "grad_norm": 0.9793429970741272,
      "learning_rate": 3.0264140553713683e-06,
      "loss": 0.6379,
      "step": 2271
    },
    {
      "epoch": 0.6297117516629712,
      "grad_norm": 0.9218000769615173,
      "learning_rate": 3.022414660313511e-06,
      "loss": 0.6127,
      "step": 2272
    },
    {
      "epoch": 0.6299889135254989,
      "grad_norm": 0.9548899531364441,
      "learning_rate": 3.01841676460175e-06,
      "loss": 0.5859,
      "step": 2273
    },
    {
      "epoch": 0.6302660753880266,
      "grad_norm": 0.9474799036979675,
      "learning_rate": 3.014420371267176e-06,
      "loss": 0.6072,
      "step": 2274
    },
    {
      "epoch": 0.6305432372505543,
      "grad_norm": 0.9072204232215881,
      "learning_rate": 3.0104254833397307e-06,
      "loss": 0.659,
      "step": 2275
    },
    {
      "epoch": 0.6308203991130821,
      "grad_norm": 0.9421378374099731,
      "learning_rate": 3.0064321038482216e-06,
      "loss": 0.6306,
      "step": 2276
    },
    {
      "epoch": 0.6310975609756098,
      "grad_norm": 0.9201748967170715,
      "learning_rate": 3.002440235820306e-06,
      "loss": 0.5532,
      "step": 2277
    },
    {
      "epoch": 0.6313747228381374,
      "grad_norm": 0.9818258881568909,
      "learning_rate": 2.9984498822825027e-06,
      "loss": 0.6231,
      "step": 2278
    },
    {
      "epoch": 0.6316518847006651,
      "grad_norm": 0.958289623260498,
      "learning_rate": 2.994461046260177e-06,
      "loss": 0.5978,
      "step": 2279
    },
    {
      "epoch": 0.6319290465631929,
      "grad_norm": 0.9228447675704956,
      "learning_rate": 2.990473730777547e-06,
      "loss": 0.6282,
      "step": 2280
    },
    {
      "epoch": 0.6322062084257206,
      "grad_norm": 0.9315418004989624,
      "learning_rate": 2.9864879388576693e-06,
      "loss": 0.5969,
      "step": 2281
    },
    {
      "epoch": 0.6324833702882483,
      "grad_norm": 0.9290876388549805,
      "learning_rate": 2.982503673522462e-06,
      "loss": 0.5809,
      "step": 2282
    },
    {
      "epoch": 0.6327605321507761,
      "grad_norm": 1.0890594720840454,
      "learning_rate": 2.97852093779267e-06,
      "loss": 0.5751,
      "step": 2283
    },
    {
      "epoch": 0.6330376940133038,
      "grad_norm": 0.9416822195053101,
      "learning_rate": 2.9745397346878835e-06,
      "loss": 0.65,
      "step": 2284
    },
    {
      "epoch": 0.6333148558758315,
      "grad_norm": 0.9181316494941711,
      "learning_rate": 2.9705600672265365e-06,
      "loss": 0.6072,
      "step": 2285
    },
    {
      "epoch": 0.6335920177383592,
      "grad_norm": 0.885188102722168,
      "learning_rate": 2.9665819384258907e-06,
      "loss": 0.6126,
      "step": 2286
    },
    {
      "epoch": 0.633869179600887,
      "grad_norm": 0.9093576073646545,
      "learning_rate": 2.9626053513020454e-06,
      "loss": 0.5767,
      "step": 2287
    },
    {
      "epoch": 0.6341463414634146,
      "grad_norm": 0.9119716882705688,
      "learning_rate": 2.9586303088699284e-06,
      "loss": 0.6439,
      "step": 2288
    },
    {
      "epoch": 0.6344235033259423,
      "grad_norm": 0.9192542433738708,
      "learning_rate": 2.9546568141433007e-06,
      "loss": 0.679,
      "step": 2289
    },
    {
      "epoch": 0.63470066518847,
      "grad_norm": 0.9141254425048828,
      "learning_rate": 2.9506848701347464e-06,
      "loss": 0.5827,
      "step": 2290
    },
    {
      "epoch": 0.6349778270509978,
      "grad_norm": 0.9209197163581848,
      "learning_rate": 2.9467144798556746e-06,
      "loss": 0.5856,
      "step": 2291
    },
    {
      "epoch": 0.6352549889135255,
      "grad_norm": 0.910414457321167,
      "learning_rate": 2.942745646316314e-06,
      "loss": 0.5945,
      "step": 2292
    },
    {
      "epoch": 0.6355321507760532,
      "grad_norm": 1.3773835897445679,
      "learning_rate": 2.9387783725257203e-06,
      "loss": 0.6245,
      "step": 2293
    },
    {
      "epoch": 0.635809312638581,
      "grad_norm": 0.955291211605072,
      "learning_rate": 2.934812661491759e-06,
      "loss": 0.63,
      "step": 2294
    },
    {
      "epoch": 0.6360864745011087,
      "grad_norm": 0.9982396960258484,
      "learning_rate": 2.930848516221113e-06,
      "loss": 0.6389,
      "step": 2295
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 0.9088571071624756,
      "learning_rate": 2.9268859397192807e-06,
      "loss": 0.6368,
      "step": 2296
    },
    {
      "epoch": 0.636640798226164,
      "grad_norm": 25.85850715637207,
      "learning_rate": 2.9229249349905686e-06,
      "loss": 0.6434,
      "step": 2297
    },
    {
      "epoch": 0.6369179600886918,
      "grad_norm": 0.9728751182556152,
      "learning_rate": 2.918965505038093e-06,
      "loss": 0.6416,
      "step": 2298
    },
    {
      "epoch": 0.6371951219512195,
      "grad_norm": 1.1737561225891113,
      "learning_rate": 2.915007652863771e-06,
      "loss": 0.652,
      "step": 2299
    },
    {
      "epoch": 0.6374722838137472,
      "grad_norm": 0.9412201046943665,
      "learning_rate": 2.9110513814683335e-06,
      "loss": 0.5809,
      "step": 2300
    },
    {
      "epoch": 0.6377494456762749,
      "grad_norm": 0.9374039173126221,
      "learning_rate": 2.9070966938513044e-06,
      "loss": 0.594,
      "step": 2301
    },
    {
      "epoch": 0.6380266075388027,
      "grad_norm": 1.0018386840820312,
      "learning_rate": 2.9031435930110104e-06,
      "loss": 0.6323,
      "step": 2302
    },
    {
      "epoch": 0.6383037694013304,
      "grad_norm": 0.9752061367034912,
      "learning_rate": 2.899192081944574e-06,
      "loss": 0.6242,
      "step": 2303
    },
    {
      "epoch": 0.6385809312638581,
      "grad_norm": 1.098096489906311,
      "learning_rate": 2.895242163647913e-06,
      "loss": 0.5984,
      "step": 2304
    },
    {
      "epoch": 0.6388580931263859,
      "grad_norm": 0.8896121382713318,
      "learning_rate": 2.8912938411157366e-06,
      "loss": 0.6412,
      "step": 2305
    },
    {
      "epoch": 0.6391352549889135,
      "grad_norm": 1.1872313022613525,
      "learning_rate": 2.8873471173415444e-06,
      "loss": 0.6154,
      "step": 2306
    },
    {
      "epoch": 0.6394124168514412,
      "grad_norm": 0.9439737796783447,
      "learning_rate": 2.883401995317627e-06,
      "loss": 0.6105,
      "step": 2307
    },
    {
      "epoch": 0.6396895787139689,
      "grad_norm": 0.902180552482605,
      "learning_rate": 2.879458478035059e-06,
      "loss": 0.5939,
      "step": 2308
    },
    {
      "epoch": 0.6399667405764967,
      "grad_norm": 0.9350372552871704,
      "learning_rate": 2.875516568483693e-06,
      "loss": 0.6515,
      "step": 2309
    },
    {
      "epoch": 0.6402439024390244,
      "grad_norm": 0.9625951051712036,
      "learning_rate": 2.871576269652167e-06,
      "loss": 0.6602,
      "step": 2310
    },
    {
      "epoch": 0.6405210643015521,
      "grad_norm": 0.90188068151474,
      "learning_rate": 2.8676375845279013e-06,
      "loss": 0.6139,
      "step": 2311
    },
    {
      "epoch": 0.6407982261640798,
      "grad_norm": 0.871262788772583,
      "learning_rate": 2.8637005160970886e-06,
      "loss": 0.5644,
      "step": 2312
    },
    {
      "epoch": 0.6410753880266076,
      "grad_norm": 0.9545223116874695,
      "learning_rate": 2.859765067344695e-06,
      "loss": 0.6255,
      "step": 2313
    },
    {
      "epoch": 0.6413525498891353,
      "grad_norm": 0.9503719210624695,
      "learning_rate": 2.855831241254461e-06,
      "loss": 0.6265,
      "step": 2314
    },
    {
      "epoch": 0.641629711751663,
      "grad_norm": 0.8915459513664246,
      "learning_rate": 2.8518990408088954e-06,
      "loss": 0.5989,
      "step": 2315
    },
    {
      "epoch": 0.6419068736141907,
      "grad_norm": 1.0214285850524902,
      "learning_rate": 2.847968468989276e-06,
      "loss": 0.6146,
      "step": 2316
    },
    {
      "epoch": 0.6421840354767184,
      "grad_norm": 0.9842682480812073,
      "learning_rate": 2.8440395287756417e-06,
      "loss": 0.6433,
      "step": 2317
    },
    {
      "epoch": 0.6424611973392461,
      "grad_norm": 0.8752260208129883,
      "learning_rate": 2.8401122231468028e-06,
      "loss": 0.6407,
      "step": 2318
    },
    {
      "epoch": 0.6427383592017738,
      "grad_norm": 0.909376323223114,
      "learning_rate": 2.8361865550803234e-06,
      "loss": 0.6263,
      "step": 2319
    },
    {
      "epoch": 0.6430155210643016,
      "grad_norm": 0.9622334241867065,
      "learning_rate": 2.832262527552525e-06,
      "loss": 0.6335,
      "step": 2320
    },
    {
      "epoch": 0.6432926829268293,
      "grad_norm": 0.8845048546791077,
      "learning_rate": 2.828340143538488e-06,
      "loss": 0.6048,
      "step": 2321
    },
    {
      "epoch": 0.643569844789357,
      "grad_norm": 0.9259751439094543,
      "learning_rate": 2.82441940601205e-06,
      "loss": 0.6287,
      "step": 2322
    },
    {
      "epoch": 0.6438470066518847,
      "grad_norm": 1.089139699935913,
      "learning_rate": 2.820500317945796e-06,
      "loss": 0.5795,
      "step": 2323
    },
    {
      "epoch": 0.6441241685144125,
      "grad_norm": 1.1091558933258057,
      "learning_rate": 2.8165828823110595e-06,
      "loss": 0.6235,
      "step": 2324
    },
    {
      "epoch": 0.6444013303769401,
      "grad_norm": 1.1082777976989746,
      "learning_rate": 2.8126671020779263e-06,
      "loss": 0.588,
      "step": 2325
    },
    {
      "epoch": 0.6446784922394678,
      "grad_norm": 0.9081136584281921,
      "learning_rate": 2.8087529802152203e-06,
      "loss": 0.6451,
      "step": 2326
    },
    {
      "epoch": 0.6449556541019955,
      "grad_norm": 0.9630493521690369,
      "learning_rate": 2.8048405196905147e-06,
      "loss": 0.6365,
      "step": 2327
    },
    {
      "epoch": 0.6452328159645233,
      "grad_norm": 0.8991391062736511,
      "learning_rate": 2.800929723470118e-06,
      "loss": 0.5832,
      "step": 2328
    },
    {
      "epoch": 0.645509977827051,
      "grad_norm": 0.9502070546150208,
      "learning_rate": 2.7970205945190788e-06,
      "loss": 0.6047,
      "step": 2329
    },
    {
      "epoch": 0.6457871396895787,
      "grad_norm": 0.8755717873573303,
      "learning_rate": 2.7931131358011863e-06,
      "loss": 0.6098,
      "step": 2330
    },
    {
      "epoch": 0.6460643015521065,
      "grad_norm": 0.9199130535125732,
      "learning_rate": 2.789207350278954e-06,
      "loss": 0.5356,
      "step": 2331
    },
    {
      "epoch": 0.6463414634146342,
      "grad_norm": 0.9074609279632568,
      "learning_rate": 2.7853032409136315e-06,
      "loss": 0.6342,
      "step": 2332
    },
    {
      "epoch": 0.6466186252771619,
      "grad_norm": 0.9696486592292786,
      "learning_rate": 2.781400810665201e-06,
      "loss": 0.596,
      "step": 2333
    },
    {
      "epoch": 0.6468957871396895,
      "grad_norm": 1.3411980867385864,
      "learning_rate": 2.777500062492368e-06,
      "loss": 0.632,
      "step": 2334
    },
    {
      "epoch": 0.6471729490022173,
      "grad_norm": 0.8985353112220764,
      "learning_rate": 2.773600999352562e-06,
      "loss": 0.6676,
      "step": 2335
    },
    {
      "epoch": 0.647450110864745,
      "grad_norm": 0.8677486181259155,
      "learning_rate": 2.7697036242019364e-06,
      "loss": 0.611,
      "step": 2336
    },
    {
      "epoch": 0.6477272727272727,
      "grad_norm": 0.9613577127456665,
      "learning_rate": 2.765807939995365e-06,
      "loss": 0.6371,
      "step": 2337
    },
    {
      "epoch": 0.6480044345898004,
      "grad_norm": 0.9791407585144043,
      "learning_rate": 2.761913949686438e-06,
      "loss": 0.6175,
      "step": 2338
    },
    {
      "epoch": 0.6482815964523282,
      "grad_norm": 0.9672890901565552,
      "learning_rate": 2.758021656227464e-06,
      "loss": 0.5785,
      "step": 2339
    },
    {
      "epoch": 0.6485587583148559,
      "grad_norm": 0.9837719202041626,
      "learning_rate": 2.7541310625694618e-06,
      "loss": 0.554,
      "step": 2340
    },
    {
      "epoch": 0.6488359201773836,
      "grad_norm": 0.9042434096336365,
      "learning_rate": 2.750242171662165e-06,
      "loss": 0.6344,
      "step": 2341
    },
    {
      "epoch": 0.6491130820399114,
      "grad_norm": 0.9383901953697205,
      "learning_rate": 2.746354986454014e-06,
      "loss": 0.6517,
      "step": 2342
    },
    {
      "epoch": 0.649390243902439,
      "grad_norm": 1.137465476989746,
      "learning_rate": 2.7424695098921534e-06,
      "loss": 0.5784,
      "step": 2343
    },
    {
      "epoch": 0.6496674057649667,
      "grad_norm": 1.726316213607788,
      "learning_rate": 2.7385857449224408e-06,
      "loss": 0.6095,
      "step": 2344
    },
    {
      "epoch": 0.6499445676274944,
      "grad_norm": 1.0230807065963745,
      "learning_rate": 2.7347036944894274e-06,
      "loss": 0.5605,
      "step": 2345
    },
    {
      "epoch": 0.6502217294900222,
      "grad_norm": 0.9231581091880798,
      "learning_rate": 2.730823361536369e-06,
      "loss": 0.5529,
      "step": 2346
    },
    {
      "epoch": 0.6504988913525499,
      "grad_norm": 1.283384919166565,
      "learning_rate": 2.726944749005217e-06,
      "loss": 0.6443,
      "step": 2347
    },
    {
      "epoch": 0.6507760532150776,
      "grad_norm": 0.9603099226951599,
      "learning_rate": 2.72306785983662e-06,
      "loss": 0.6122,
      "step": 2348
    },
    {
      "epoch": 0.6510532150776053,
      "grad_norm": 0.9532773494720459,
      "learning_rate": 2.71919269696992e-06,
      "loss": 0.6034,
      "step": 2349
    },
    {
      "epoch": 0.6513303769401331,
      "grad_norm": 0.949319064617157,
      "learning_rate": 2.715319263343149e-06,
      "loss": 0.6632,
      "step": 2350
    },
    {
      "epoch": 0.6516075388026608,
      "grad_norm": 0.9295929670333862,
      "learning_rate": 2.711447561893029e-06,
      "loss": 0.6011,
      "step": 2351
    },
    {
      "epoch": 0.6518847006651884,
      "grad_norm": 0.9418200254440308,
      "learning_rate": 2.7075775955549682e-06,
      "loss": 0.6027,
      "step": 2352
    },
    {
      "epoch": 0.6521618625277162,
      "grad_norm": 0.9280131459236145,
      "learning_rate": 2.7037093672630595e-06,
      "loss": 0.5859,
      "step": 2353
    },
    {
      "epoch": 0.6524390243902439,
      "grad_norm": 0.8904016613960266,
      "learning_rate": 2.699842879950075e-06,
      "loss": 0.6148,
      "step": 2354
    },
    {
      "epoch": 0.6527161862527716,
      "grad_norm": 1.3310836553573608,
      "learning_rate": 2.695978136547476e-06,
      "loss": 0.6437,
      "step": 2355
    },
    {
      "epoch": 0.6529933481152993,
      "grad_norm": 0.9601128697395325,
      "learning_rate": 2.6921151399853907e-06,
      "loss": 0.5844,
      "step": 2356
    },
    {
      "epoch": 0.6532705099778271,
      "grad_norm": 1.0560967922210693,
      "learning_rate": 2.6882538931926315e-06,
      "loss": 0.601,
      "step": 2357
    },
    {
      "epoch": 0.6535476718403548,
      "grad_norm": 0.9117201566696167,
      "learning_rate": 2.6843943990966738e-06,
      "loss": 0.6166,
      "step": 2358
    },
    {
      "epoch": 0.6538248337028825,
      "grad_norm": 0.9628698229789734,
      "learning_rate": 2.680536660623677e-06,
      "loss": 0.5823,
      "step": 2359
    },
    {
      "epoch": 0.6541019955654102,
      "grad_norm": 4.3640031814575195,
      "learning_rate": 2.6766806806984608e-06,
      "loss": 0.5915,
      "step": 2360
    },
    {
      "epoch": 0.654379157427938,
      "grad_norm": 0.9255430102348328,
      "learning_rate": 2.672826462244514e-06,
      "loss": 0.6392,
      "step": 2361
    },
    {
      "epoch": 0.6546563192904656,
      "grad_norm": 0.890258252620697,
      "learning_rate": 2.6689740081839895e-06,
      "loss": 0.5741,
      "step": 2362
    },
    {
      "epoch": 0.6549334811529933,
      "grad_norm": 0.994884729385376,
      "learning_rate": 2.6651233214377033e-06,
      "loss": 0.6246,
      "step": 2363
    },
    {
      "epoch": 0.655210643015521,
      "grad_norm": 0.9001596570014954,
      "learning_rate": 2.6612744049251304e-06,
      "loss": 0.5916,
      "step": 2364
    },
    {
      "epoch": 0.6554878048780488,
      "grad_norm": 0.8928593993186951,
      "learning_rate": 2.6574272615644054e-06,
      "loss": 0.6446,
      "step": 2365
    },
    {
      "epoch": 0.6557649667405765,
      "grad_norm": 0.961046040058136,
      "learning_rate": 2.653581894272315e-06,
      "loss": 0.6346,
      "step": 2366
    },
    {
      "epoch": 0.6560421286031042,
      "grad_norm": 0.998016893863678,
      "learning_rate": 2.649738305964304e-06,
      "loss": 0.5889,
      "step": 2367
    },
    {
      "epoch": 0.656319290465632,
      "grad_norm": 1.014095664024353,
      "learning_rate": 2.6458964995544682e-06,
      "loss": 0.6802,
      "step": 2368
    },
    {
      "epoch": 0.6565964523281597,
      "grad_norm": 0.9567813277244568,
      "learning_rate": 2.6420564779555447e-06,
      "loss": 0.668,
      "step": 2369
    },
    {
      "epoch": 0.6568736141906873,
      "grad_norm": 0.8805619478225708,
      "learning_rate": 2.638218244078927e-06,
      "loss": 0.6155,
      "step": 2370
    },
    {
      "epoch": 0.657150776053215,
      "grad_norm": 1.043816089630127,
      "learning_rate": 2.6343818008346495e-06,
      "loss": 0.6122,
      "step": 2371
    },
    {
      "epoch": 0.6574279379157428,
      "grad_norm": 1.0211701393127441,
      "learning_rate": 2.630547151131387e-06,
      "loss": 0.6239,
      "step": 2372
    },
    {
      "epoch": 0.6577050997782705,
      "grad_norm": 0.9593768119812012,
      "learning_rate": 2.6267142978764583e-06,
      "loss": 0.5928,
      "step": 2373
    },
    {
      "epoch": 0.6579822616407982,
      "grad_norm": 0.9443894624710083,
      "learning_rate": 2.622883243975817e-06,
      "loss": 0.6627,
      "step": 2374
    },
    {
      "epoch": 0.6582594235033259,
      "grad_norm": 0.9818089604377747,
      "learning_rate": 2.619053992334053e-06,
      "loss": 0.6538,
      "step": 2375
    },
    {
      "epoch": 0.6585365853658537,
      "grad_norm": 0.9003406167030334,
      "learning_rate": 2.615226545854391e-06,
      "loss": 0.632,
      "step": 2376
    },
    {
      "epoch": 0.6588137472283814,
      "grad_norm": 0.8801334500312805,
      "learning_rate": 2.611400907438685e-06,
      "loss": 0.5484,
      "step": 2377
    },
    {
      "epoch": 0.6590909090909091,
      "grad_norm": 1.007947325706482,
      "learning_rate": 2.6075770799874234e-06,
      "loss": 0.6636,
      "step": 2378
    },
    {
      "epoch": 0.6593680709534369,
      "grad_norm": 1.0059397220611572,
      "learning_rate": 2.603755066399718e-06,
      "loss": 0.5873,
      "step": 2379
    },
    {
      "epoch": 0.6596452328159645,
      "grad_norm": 0.9073856472969055,
      "learning_rate": 2.5999348695732986e-06,
      "loss": 0.5893,
      "step": 2380
    },
    {
      "epoch": 0.6599223946784922,
      "grad_norm": 0.9740723371505737,
      "learning_rate": 2.596116492404531e-06,
      "loss": 0.6045,
      "step": 2381
    },
    {
      "epoch": 0.6601995565410199,
      "grad_norm": 0.8926793932914734,
      "learning_rate": 2.5922999377883927e-06,
      "loss": 0.6002,
      "step": 2382
    },
    {
      "epoch": 0.6604767184035477,
      "grad_norm": 1.1488016843795776,
      "learning_rate": 2.588485208618481e-06,
      "loss": 0.6235,
      "step": 2383
    },
    {
      "epoch": 0.6607538802660754,
      "grad_norm": 0.9780311584472656,
      "learning_rate": 2.5846723077870097e-06,
      "loss": 0.6211,
      "step": 2384
    },
    {
      "epoch": 0.6610310421286031,
      "grad_norm": 0.9628798365592957,
      "learning_rate": 2.5808612381848064e-06,
      "loss": 0.6397,
      "step": 2385
    },
    {
      "epoch": 0.6613082039911308,
      "grad_norm": 0.9107243418693542,
      "learning_rate": 2.5770520027013094e-06,
      "loss": 0.6014,
      "step": 2386
    },
    {
      "epoch": 0.6615853658536586,
      "grad_norm": 0.9829089641571045,
      "learning_rate": 2.5732446042245686e-06,
      "loss": 0.625,
      "step": 2387
    },
    {
      "epoch": 0.6618625277161863,
      "grad_norm": 1.1884398460388184,
      "learning_rate": 2.569439045641236e-06,
      "loss": 0.5657,
      "step": 2388
    },
    {
      "epoch": 0.6621396895787139,
      "grad_norm": 0.9791149497032166,
      "learning_rate": 2.565635329836578e-06,
      "loss": 0.731,
      "step": 2389
    },
    {
      "epoch": 0.6624168514412417,
      "grad_norm": 0.9547834992408752,
      "learning_rate": 2.5618334596944566e-06,
      "loss": 0.6197,
      "step": 2390
    },
    {
      "epoch": 0.6626940133037694,
      "grad_norm": 1.9295960664749146,
      "learning_rate": 2.5580334380973316e-06,
      "loss": 0.5959,
      "step": 2391
    },
    {
      "epoch": 0.6629711751662971,
      "grad_norm": 0.9225706458091736,
      "learning_rate": 2.554235267926271e-06,
      "loss": 0.6011,
      "step": 2392
    },
    {
      "epoch": 0.6632483370288248,
      "grad_norm": 0.8912270069122314,
      "learning_rate": 2.550438952060932e-06,
      "loss": 0.5961,
      "step": 2393
    },
    {
      "epoch": 0.6635254988913526,
      "grad_norm": 1.2017149925231934,
      "learning_rate": 2.546644493379569e-06,
      "loss": 0.6579,
      "step": 2394
    },
    {
      "epoch": 0.6638026607538803,
      "grad_norm": 0.8656323552131653,
      "learning_rate": 2.5428518947590246e-06,
      "loss": 0.5354,
      "step": 2395
    },
    {
      "epoch": 0.664079822616408,
      "grad_norm": 0.9028607606887817,
      "learning_rate": 2.539061159074736e-06,
      "loss": 0.6263,
      "step": 2396
    },
    {
      "epoch": 0.6643569844789357,
      "grad_norm": 0.9497581720352173,
      "learning_rate": 2.5352722892007255e-06,
      "loss": 0.5784,
      "step": 2397
    },
    {
      "epoch": 0.6646341463414634,
      "grad_norm": 0.8713258504867554,
      "learning_rate": 2.5314852880095992e-06,
      "loss": 0.5659,
      "step": 2398
    },
    {
      "epoch": 0.6649113082039911,
      "grad_norm": 0.9382534623146057,
      "learning_rate": 2.527700158372548e-06,
      "loss": 0.6653,
      "step": 2399
    },
    {
      "epoch": 0.6651884700665188,
      "grad_norm": 0.9334493279457092,
      "learning_rate": 2.52391690315935e-06,
      "loss": 0.6554,
      "step": 2400
    },
    {
      "epoch": 0.6654656319290465,
      "grad_norm": 0.9509286880493164,
      "learning_rate": 2.5201355252383485e-06,
      "loss": 0.5807,
      "step": 2401
    },
    {
      "epoch": 0.6657427937915743,
      "grad_norm": 0.9798305034637451,
      "learning_rate": 2.5163560274764755e-06,
      "loss": 0.5919,
      "step": 2402
    },
    {
      "epoch": 0.666019955654102,
      "grad_norm": 0.8690276741981506,
      "learning_rate": 2.51257841273923e-06,
      "loss": 0.5608,
      "step": 2403
    },
    {
      "epoch": 0.6662971175166297,
      "grad_norm": 0.8986737728118896,
      "learning_rate": 2.508802683890691e-06,
      "loss": 0.6507,
      "step": 2404
    },
    {
      "epoch": 0.6665742793791575,
      "grad_norm": 0.8858300447463989,
      "learning_rate": 2.5050288437935003e-06,
      "loss": 0.5472,
      "step": 2405
    },
    {
      "epoch": 0.6668514412416852,
      "grad_norm": 1.0452501773834229,
      "learning_rate": 2.5012568953088724e-06,
      "loss": 0.6493,
      "step": 2406
    },
    {
      "epoch": 0.6671286031042128,
      "grad_norm": 1.2422797679901123,
      "learning_rate": 2.497486841296584e-06,
      "loss": 0.5793,
      "step": 2407
    },
    {
      "epoch": 0.6674057649667405,
      "grad_norm": 0.9606922268867493,
      "learning_rate": 2.4937186846149793e-06,
      "loss": 0.6207,
      "step": 2408
    },
    {
      "epoch": 0.6676829268292683,
      "grad_norm": 0.95497727394104,
      "learning_rate": 2.4899524281209602e-06,
      "loss": 0.6511,
      "step": 2409
    },
    {
      "epoch": 0.667960088691796,
      "grad_norm": 0.9187065958976746,
      "learning_rate": 2.486188074669989e-06,
      "loss": 0.5977,
      "step": 2410
    },
    {
      "epoch": 0.6682372505543237,
      "grad_norm": 0.9201993942260742,
      "learning_rate": 2.482425627116092e-06,
      "loss": 0.6263,
      "step": 2411
    },
    {
      "epoch": 0.6685144124168514,
      "grad_norm": 1.333795189857483,
      "learning_rate": 2.478665088311838e-06,
      "loss": 0.6306,
      "step": 2412
    },
    {
      "epoch": 0.6687915742793792,
      "grad_norm": 1.1742695569992065,
      "learning_rate": 2.474906461108357e-06,
      "loss": 0.62,
      "step": 2413
    },
    {
      "epoch": 0.6690687361419069,
      "grad_norm": 0.988561749458313,
      "learning_rate": 2.4711497483553265e-06,
      "loss": 0.6915,
      "step": 2414
    },
    {
      "epoch": 0.6693458980044346,
      "grad_norm": 1.1129121780395508,
      "learning_rate": 2.4673949529009773e-06,
      "loss": 0.5479,
      "step": 2415
    },
    {
      "epoch": 0.6696230598669624,
      "grad_norm": 0.8774926066398621,
      "learning_rate": 2.4636420775920803e-06,
      "loss": 0.6185,
      "step": 2416
    },
    {
      "epoch": 0.66990022172949,
      "grad_norm": 0.9125105738639832,
      "learning_rate": 2.4598911252739553e-06,
      "loss": 0.6204,
      "step": 2417
    },
    {
      "epoch": 0.6701773835920177,
      "grad_norm": 0.8906736969947815,
      "learning_rate": 2.4561420987904614e-06,
      "loss": 0.6647,
      "step": 2418
    },
    {
      "epoch": 0.6704545454545454,
      "grad_norm": 0.958116888999939,
      "learning_rate": 2.4523950009839983e-06,
      "loss": 0.6095,
      "step": 2419
    },
    {
      "epoch": 0.6707317073170732,
      "grad_norm": 0.9968855381011963,
      "learning_rate": 2.448649834695503e-06,
      "loss": 0.6183,
      "step": 2420
    },
    {
      "epoch": 0.6710088691796009,
      "grad_norm": 0.8793277144432068,
      "learning_rate": 2.4449066027644473e-06,
      "loss": 0.5541,
      "step": 2421
    },
    {
      "epoch": 0.6712860310421286,
      "grad_norm": 0.9026990532875061,
      "learning_rate": 2.441165308028844e-06,
      "loss": 0.597,
      "step": 2422
    },
    {
      "epoch": 0.6715631929046563,
      "grad_norm": 0.9662331938743591,
      "learning_rate": 2.4374259533252244e-06,
      "loss": 0.6046,
      "step": 2423
    },
    {
      "epoch": 0.6718403547671841,
      "grad_norm": 0.9669737815856934,
      "learning_rate": 2.4336885414886575e-06,
      "loss": 0.6737,
      "step": 2424
    },
    {
      "epoch": 0.6721175166297118,
      "grad_norm": 1.2011749744415283,
      "learning_rate": 2.4299530753527356e-06,
      "loss": 0.6447,
      "step": 2425
    },
    {
      "epoch": 0.6723946784922394,
      "grad_norm": 0.8892765641212463,
      "learning_rate": 2.4262195577495812e-06,
      "loss": 0.5655,
      "step": 2426
    },
    {
      "epoch": 0.6726718403547672,
      "grad_norm": 2.6837399005889893,
      "learning_rate": 2.4224879915098345e-06,
      "loss": 0.6462,
      "step": 2427
    },
    {
      "epoch": 0.6729490022172949,
      "grad_norm": 0.9050636291503906,
      "learning_rate": 2.418758379462656e-06,
      "loss": 0.5748,
      "step": 2428
    },
    {
      "epoch": 0.6732261640798226,
      "grad_norm": 0.9088660478591919,
      "learning_rate": 2.4150307244357275e-06,
      "loss": 0.6102,
      "step": 2429
    },
    {
      "epoch": 0.6735033259423503,
      "grad_norm": 0.8926207423210144,
      "learning_rate": 2.411305029255245e-06,
      "loss": 0.6001,
      "step": 2430
    },
    {
      "epoch": 0.6737804878048781,
      "grad_norm": 2.767646312713623,
      "learning_rate": 2.4075812967459196e-06,
      "loss": 0.6428,
      "step": 2431
    },
    {
      "epoch": 0.6740576496674058,
      "grad_norm": 2.0327162742614746,
      "learning_rate": 2.4038595297309712e-06,
      "loss": 0.6898,
      "step": 2432
    },
    {
      "epoch": 0.6743348115299335,
      "grad_norm": 1.0321258306503296,
      "learning_rate": 2.400139731032139e-06,
      "loss": 0.5819,
      "step": 2433
    },
    {
      "epoch": 0.6746119733924612,
      "grad_norm": 0.9737773537635803,
      "learning_rate": 2.3964219034696573e-06,
      "loss": 0.6507,
      "step": 2434
    },
    {
      "epoch": 0.674889135254989,
      "grad_norm": 2.3519890308380127,
      "learning_rate": 2.3927060498622734e-06,
      "loss": 0.5898,
      "step": 2435
    },
    {
      "epoch": 0.6751662971175166,
      "grad_norm": 0.8883225917816162,
      "learning_rate": 2.388992173027235e-06,
      "loss": 0.5632,
      "step": 2436
    },
    {
      "epoch": 0.6754434589800443,
      "grad_norm": 0.8912991881370544,
      "learning_rate": 2.385280275780295e-06,
      "loss": 0.6157,
      "step": 2437
    },
    {
      "epoch": 0.6757206208425721,
      "grad_norm": 1.181156873703003,
      "learning_rate": 2.381570360935703e-06,
      "loss": 0.6117,
      "step": 2438
    },
    {
      "epoch": 0.6759977827050998,
      "grad_norm": 0.9026086330413818,
      "learning_rate": 2.377862431306205e-06,
      "loss": 0.5456,
      "step": 2439
    },
    {
      "epoch": 0.6762749445676275,
      "grad_norm": 0.9072038531303406,
      "learning_rate": 2.3741564897030413e-06,
      "loss": 0.5422,
      "step": 2440
    },
    {
      "epoch": 0.6765521064301552,
      "grad_norm": 0.9225761890411377,
      "learning_rate": 2.3704525389359473e-06,
      "loss": 0.622,
      "step": 2441
    },
    {
      "epoch": 0.676829268292683,
      "grad_norm": 0.9022672176361084,
      "learning_rate": 2.366750581813147e-06,
      "loss": 0.6382,
      "step": 2442
    },
    {
      "epoch": 0.6771064301552107,
      "grad_norm": 0.9899957776069641,
      "learning_rate": 2.363050621141354e-06,
      "loss": 0.5857,
      "step": 2443
    },
    {
      "epoch": 0.6773835920177383,
      "grad_norm": 0.8846986293792725,
      "learning_rate": 2.3593526597257676e-06,
      "loss": 0.6178,
      "step": 2444
    },
    {
      "epoch": 0.677660753880266,
      "grad_norm": 0.9573742747306824,
      "learning_rate": 2.3556567003700727e-06,
      "loss": 0.7066,
      "step": 2445
    },
    {
      "epoch": 0.6779379157427938,
      "grad_norm": 0.9746490716934204,
      "learning_rate": 2.351962745876434e-06,
      "loss": 0.6857,
      "step": 2446
    },
    {
      "epoch": 0.6782150776053215,
      "grad_norm": 0.9136871695518494,
      "learning_rate": 2.3482707990454955e-06,
      "loss": 0.6236,
      "step": 2447
    },
    {
      "epoch": 0.6784922394678492,
      "grad_norm": 0.8982982635498047,
      "learning_rate": 2.3445808626763857e-06,
      "loss": 0.6224,
      "step": 2448
    },
    {
      "epoch": 0.6787694013303769,
      "grad_norm": 0.9058036804199219,
      "learning_rate": 2.3408929395667013e-06,
      "loss": 0.6003,
      "step": 2449
    },
    {
      "epoch": 0.6790465631929047,
      "grad_norm": 1.2122752666473389,
      "learning_rate": 2.3372070325125175e-06,
      "loss": 0.5914,
      "step": 2450
    },
    {
      "epoch": 0.6793237250554324,
      "grad_norm": 0.8823323845863342,
      "learning_rate": 2.3335231443083732e-06,
      "loss": 0.6112,
      "step": 2451
    },
    {
      "epoch": 0.6796008869179601,
      "grad_norm": 0.9507444500923157,
      "learning_rate": 2.3298412777472883e-06,
      "loss": 0.6025,
      "step": 2452
    },
    {
      "epoch": 0.6798780487804879,
      "grad_norm": 0.8672029376029968,
      "learning_rate": 2.326161435620742e-06,
      "loss": 0.5818,
      "step": 2453
    },
    {
      "epoch": 0.6801552106430155,
      "grad_norm": 0.9387220740318298,
      "learning_rate": 2.32248362071868e-06,
      "loss": 0.5924,
      "step": 2454
    },
    {
      "epoch": 0.6804323725055432,
      "grad_norm": 0.9134634733200073,
      "learning_rate": 2.3188078358295124e-06,
      "loss": 0.5919,
      "step": 2455
    },
    {
      "epoch": 0.6807095343680709,
      "grad_norm": 0.8871769905090332,
      "learning_rate": 2.315134083740109e-06,
      "loss": 0.5676,
      "step": 2456
    },
    {
      "epoch": 0.6809866962305987,
      "grad_norm": 0.8845345377922058,
      "learning_rate": 2.3114623672357993e-06,
      "loss": 0.6239,
      "step": 2457
    },
    {
      "epoch": 0.6812638580931264,
      "grad_norm": 0.9062137007713318,
      "learning_rate": 2.3077926891003664e-06,
      "loss": 0.6018,
      "step": 2458
    },
    {
      "epoch": 0.6815410199556541,
      "grad_norm": 0.9140833616256714,
      "learning_rate": 2.3041250521160553e-06,
      "loss": 0.6547,
      "step": 2459
    },
    {
      "epoch": 0.6818181818181818,
      "grad_norm": 0.9058482646942139,
      "learning_rate": 2.3004594590635582e-06,
      "loss": 0.6165,
      "step": 2460
    },
    {
      "epoch": 0.6820953436807096,
      "grad_norm": 0.878648579120636,
      "learning_rate": 2.296795912722014e-06,
      "loss": 0.5889,
      "step": 2461
    },
    {
      "epoch": 0.6823725055432373,
      "grad_norm": 0.891660213470459,
      "learning_rate": 2.293134415869015e-06,
      "loss": 0.6354,
      "step": 2462
    },
    {
      "epoch": 0.6826496674057649,
      "grad_norm": 0.940909743309021,
      "learning_rate": 2.2894749712806015e-06,
      "loss": 0.6099,
      "step": 2463
    },
    {
      "epoch": 0.6829268292682927,
      "grad_norm": 1.4032608270645142,
      "learning_rate": 2.285817581731254e-06,
      "loss": 0.6689,
      "step": 2464
    },
    {
      "epoch": 0.6832039911308204,
      "grad_norm": 0.9373987913131714,
      "learning_rate": 2.282162249993895e-06,
      "loss": 0.6072,
      "step": 2465
    },
    {
      "epoch": 0.6834811529933481,
      "grad_norm": 0.9362706542015076,
      "learning_rate": 2.2785089788398883e-06,
      "loss": 0.6026,
      "step": 2466
    },
    {
      "epoch": 0.6837583148558758,
      "grad_norm": 0.9612498879432678,
      "learning_rate": 2.2748577710390353e-06,
      "loss": 0.6521,
      "step": 2467
    },
    {
      "epoch": 0.6840354767184036,
      "grad_norm": 1.0066217184066772,
      "learning_rate": 2.271208629359572e-06,
      "loss": 0.6088,
      "step": 2468
    },
    {
      "epoch": 0.6843126385809313,
      "grad_norm": 0.9177899956703186,
      "learning_rate": 2.267561556568167e-06,
      "loss": 0.5815,
      "step": 2469
    },
    {
      "epoch": 0.684589800443459,
      "grad_norm": 0.9818257093429565,
      "learning_rate": 2.263916555429926e-06,
      "loss": 0.6329,
      "step": 2470
    },
    {
      "epoch": 0.6848669623059866,
      "grad_norm": 1.0053770542144775,
      "learning_rate": 2.2602736287083786e-06,
      "loss": 0.5946,
      "step": 2471
    },
    {
      "epoch": 0.6851441241685144,
      "grad_norm": 0.9074699282646179,
      "learning_rate": 2.2566327791654807e-06,
      "loss": 0.5989,
      "step": 2472
    },
    {
      "epoch": 0.6854212860310421,
      "grad_norm": 0.9504579305648804,
      "learning_rate": 2.2529940095616142e-06,
      "loss": 0.604,
      "step": 2473
    },
    {
      "epoch": 0.6856984478935698,
      "grad_norm": 0.9115659594535828,
      "learning_rate": 2.249357322655591e-06,
      "loss": 0.5436,
      "step": 2474
    },
    {
      "epoch": 0.6859756097560976,
      "grad_norm": 1.0428122282028198,
      "learning_rate": 2.245722721204635e-06,
      "loss": 0.617,
      "step": 2475
    },
    {
      "epoch": 0.6862527716186253,
      "grad_norm": 0.915018618106842,
      "learning_rate": 2.242090207964393e-06,
      "loss": 0.6364,
      "step": 2476
    },
    {
      "epoch": 0.686529933481153,
      "grad_norm": 2.2257471084594727,
      "learning_rate": 2.238459785688928e-06,
      "loss": 0.5671,
      "step": 2477
    },
    {
      "epoch": 0.6868070953436807,
      "grad_norm": 1.1521894931793213,
      "learning_rate": 2.2348314571307174e-06,
      "loss": 0.5633,
      "step": 2478
    },
    {
      "epoch": 0.6870842572062085,
      "grad_norm": 1.3343416452407837,
      "learning_rate": 2.2312052250406518e-06,
      "loss": 0.6711,
      "step": 2479
    },
    {
      "epoch": 0.6873614190687362,
      "grad_norm": 0.9535503387451172,
      "learning_rate": 2.22758109216803e-06,
      "loss": 0.6124,
      "step": 2480
    },
    {
      "epoch": 0.6876385809312638,
      "grad_norm": 0.9176021218299866,
      "learning_rate": 2.2239590612605645e-06,
      "loss": 0.5674,
      "step": 2481
    },
    {
      "epoch": 0.6879157427937915,
      "grad_norm": 0.9608571529388428,
      "learning_rate": 2.2203391350643715e-06,
      "loss": 0.6011,
      "step": 2482
    },
    {
      "epoch": 0.6881929046563193,
      "grad_norm": 0.9583219289779663,
      "learning_rate": 2.216721316323967e-06,
      "loss": 0.6161,
      "step": 2483
    },
    {
      "epoch": 0.688470066518847,
      "grad_norm": 1.1677453517913818,
      "learning_rate": 2.213105607782273e-06,
      "loss": 0.5723,
      "step": 2484
    },
    {
      "epoch": 0.6887472283813747,
      "grad_norm": 0.9204335808753967,
      "learning_rate": 2.209492012180616e-06,
      "loss": 0.6119,
      "step": 2485
    },
    {
      "epoch": 0.6890243902439024,
      "grad_norm": 1.1052603721618652,
      "learning_rate": 2.2058805322587123e-06,
      "loss": 0.5937,
      "step": 2486
    },
    {
      "epoch": 0.6893015521064302,
      "grad_norm": 0.9117688536643982,
      "learning_rate": 2.20227117075468e-06,
      "loss": 0.574,
      "step": 2487
    },
    {
      "epoch": 0.6895787139689579,
      "grad_norm": 1.0832945108413696,
      "learning_rate": 2.1986639304050287e-06,
      "loss": 0.5895,
      "step": 2488
    },
    {
      "epoch": 0.6898558758314856,
      "grad_norm": 0.9870914816856384,
      "learning_rate": 2.19505881394466e-06,
      "loss": 0.6238,
      "step": 2489
    },
    {
      "epoch": 0.6901330376940134,
      "grad_norm": 0.8943236470222473,
      "learning_rate": 2.1914558241068654e-06,
      "loss": 0.6268,
      "step": 2490
    },
    {
      "epoch": 0.690410199556541,
      "grad_norm": 0.9509558081626892,
      "learning_rate": 2.1878549636233227e-06,
      "loss": 0.6458,
      "step": 2491
    },
    {
      "epoch": 0.6906873614190687,
      "grad_norm": 1.8221944570541382,
      "learning_rate": 2.1842562352240997e-06,
      "loss": 0.5651,
      "step": 2492
    },
    {
      "epoch": 0.6909645232815964,
      "grad_norm": 0.9451060891151428,
      "learning_rate": 2.1806596416376445e-06,
      "loss": 0.5847,
      "step": 2493
    },
    {
      "epoch": 0.6912416851441242,
      "grad_norm": 0.8858688473701477,
      "learning_rate": 2.177065185590785e-06,
      "loss": 0.5996,
      "step": 2494
    },
    {
      "epoch": 0.6915188470066519,
      "grad_norm": 0.9197235703468323,
      "learning_rate": 2.173472869808728e-06,
      "loss": 0.5907,
      "step": 2495
    },
    {
      "epoch": 0.6917960088691796,
      "grad_norm": 0.944354772567749,
      "learning_rate": 2.1698826970150653e-06,
      "loss": 0.5722,
      "step": 2496
    },
    {
      "epoch": 0.6920731707317073,
      "grad_norm": 0.9132809042930603,
      "learning_rate": 2.166294669931756e-06,
      "loss": 0.5688,
      "step": 2497
    },
    {
      "epoch": 0.6923503325942351,
      "grad_norm": 0.9641174077987671,
      "learning_rate": 2.162708791279135e-06,
      "loss": 0.6712,
      "step": 2498
    },
    {
      "epoch": 0.6926274944567627,
      "grad_norm": 0.9505220651626587,
      "learning_rate": 2.1591250637759086e-06,
      "loss": 0.6004,
      "step": 2499
    },
    {
      "epoch": 0.6929046563192904,
      "grad_norm": 0.8659893274307251,
      "learning_rate": 2.155543490139152e-06,
      "loss": 0.5812,
      "step": 2500
    },
    {
      "epoch": 0.6931818181818182,
      "grad_norm": 0.920311689376831,
      "learning_rate": 2.1519640730843072e-06,
      "loss": 0.6852,
      "step": 2501
    },
    {
      "epoch": 0.6934589800443459,
      "grad_norm": 0.8610700368881226,
      "learning_rate": 2.148386815325179e-06,
      "loss": 0.5841,
      "step": 2502
    },
    {
      "epoch": 0.6937361419068736,
      "grad_norm": 0.8715665340423584,
      "learning_rate": 2.144811719573941e-06,
      "loss": 0.5858,
      "step": 2503
    },
    {
      "epoch": 0.6940133037694013,
      "grad_norm": 0.9390292763710022,
      "learning_rate": 2.1412387885411245e-06,
      "loss": 0.5937,
      "step": 2504
    },
    {
      "epoch": 0.6942904656319291,
      "grad_norm": 0.97919762134552,
      "learning_rate": 2.137668024935615e-06,
      "loss": 0.6694,
      "step": 2505
    },
    {
      "epoch": 0.6945676274944568,
      "grad_norm": 0.92188960313797,
      "learning_rate": 2.134099431464659e-06,
      "loss": 0.6094,
      "step": 2506
    },
    {
      "epoch": 0.6948447893569845,
      "grad_norm": 0.9230710864067078,
      "learning_rate": 2.1305330108338606e-06,
      "loss": 0.6251,
      "step": 2507
    },
    {
      "epoch": 0.6951219512195121,
      "grad_norm": 0.9229340553283691,
      "learning_rate": 2.1269687657471722e-06,
      "loss": 0.5857,
      "step": 2508
    },
    {
      "epoch": 0.6953991130820399,
      "grad_norm": 0.938339352607727,
      "learning_rate": 2.1234066989068972e-06,
      "loss": 0.6042,
      "step": 2509
    },
    {
      "epoch": 0.6956762749445676,
      "grad_norm": 1.0110208988189697,
      "learning_rate": 2.1198468130136884e-06,
      "loss": 0.5914,
      "step": 2510
    },
    {
      "epoch": 0.6959534368070953,
      "grad_norm": 0.924217939376831,
      "learning_rate": 2.116289110766545e-06,
      "loss": 0.6178,
      "step": 2511
    },
    {
      "epoch": 0.6962305986696231,
      "grad_norm": 0.9366999864578247,
      "learning_rate": 2.112733594862811e-06,
      "loss": 0.5854,
      "step": 2512
    },
    {
      "epoch": 0.6965077605321508,
      "grad_norm": 0.9499722719192505,
      "learning_rate": 2.109180267998169e-06,
      "loss": 0.5679,
      "step": 2513
    },
    {
      "epoch": 0.6967849223946785,
      "grad_norm": 0.9244142770767212,
      "learning_rate": 2.105629132866653e-06,
      "loss": 0.6161,
      "step": 2514
    },
    {
      "epoch": 0.6970620842572062,
      "grad_norm": 0.9088588356971741,
      "learning_rate": 2.1020801921606198e-06,
      "loss": 0.5978,
      "step": 2515
    },
    {
      "epoch": 0.697339246119734,
      "grad_norm": 0.9687389135360718,
      "learning_rate": 2.0985334485707733e-06,
      "loss": 0.5499,
      "step": 2516
    },
    {
      "epoch": 0.6976164079822617,
      "grad_norm": 0.9124114513397217,
      "learning_rate": 2.094988904786146e-06,
      "loss": 0.5873,
      "step": 2517
    },
    {
      "epoch": 0.6978935698447893,
      "grad_norm": 0.9514687657356262,
      "learning_rate": 2.091446563494109e-06,
      "loss": 0.6779,
      "step": 2518
    },
    {
      "epoch": 0.698170731707317,
      "grad_norm": 0.914942741394043,
      "learning_rate": 2.0879064273803577e-06,
      "loss": 0.6085,
      "step": 2519
    },
    {
      "epoch": 0.6984478935698448,
      "grad_norm": 0.9589675068855286,
      "learning_rate": 2.0843684991289177e-06,
      "loss": 0.6131,
      "step": 2520
    },
    {
      "epoch": 0.6987250554323725,
      "grad_norm": 0.9328190088272095,
      "learning_rate": 2.0808327814221403e-06,
      "loss": 0.6083,
      "step": 2521
    },
    {
      "epoch": 0.6990022172949002,
      "grad_norm": 1.2690070867538452,
      "learning_rate": 2.0772992769407007e-06,
      "loss": 0.5711,
      "step": 2522
    },
    {
      "epoch": 0.6992793791574279,
      "grad_norm": 0.8913567662239075,
      "learning_rate": 2.073767988363597e-06,
      "loss": 0.636,
      "step": 2523
    },
    {
      "epoch": 0.6995565410199557,
      "grad_norm": 0.909529983997345,
      "learning_rate": 2.0702389183681464e-06,
      "loss": 0.5886,
      "step": 2524
    },
    {
      "epoch": 0.6998337028824834,
      "grad_norm": 0.9342029690742493,
      "learning_rate": 2.0667120696299842e-06,
      "loss": 0.6575,
      "step": 2525
    },
    {
      "epoch": 0.700110864745011,
      "grad_norm": 0.9127923250198364,
      "learning_rate": 2.0631874448230617e-06,
      "loss": 0.6191,
      "step": 2526
    },
    {
      "epoch": 0.7003880266075388,
      "grad_norm": 0.9679656028747559,
      "learning_rate": 2.059665046619644e-06,
      "loss": 0.6282,
      "step": 2527
    },
    {
      "epoch": 0.7006651884700665,
      "grad_norm": 1.0941179990768433,
      "learning_rate": 2.0561448776903066e-06,
      "loss": 0.6008,
      "step": 2528
    },
    {
      "epoch": 0.7009423503325942,
      "grad_norm": 0.9392668604850769,
      "learning_rate": 2.0526269407039394e-06,
      "loss": 0.6318,
      "step": 2529
    },
    {
      "epoch": 0.7012195121951219,
      "grad_norm": 0.9506534934043884,
      "learning_rate": 2.049111238327737e-06,
      "loss": 0.6153,
      "step": 2530
    },
    {
      "epoch": 0.7014966740576497,
      "grad_norm": 0.982219398021698,
      "learning_rate": 2.045597773227199e-06,
      "loss": 0.5902,
      "step": 2531
    },
    {
      "epoch": 0.7017738359201774,
      "grad_norm": 0.988243043422699,
      "learning_rate": 2.042086548066126e-06,
      "loss": 0.6982,
      "step": 2532
    },
    {
      "epoch": 0.7020509977827051,
      "grad_norm": 0.9205905199050903,
      "learning_rate": 2.038577565506629e-06,
      "loss": 0.6696,
      "step": 2533
    },
    {
      "epoch": 0.7023281596452328,
      "grad_norm": 1.060985803604126,
      "learning_rate": 2.0350708282091124e-06,
      "loss": 0.6958,
      "step": 2534
    },
    {
      "epoch": 0.7026053215077606,
      "grad_norm": 1.1833977699279785,
      "learning_rate": 2.0315663388322787e-06,
      "loss": 0.5993,
      "step": 2535
    },
    {
      "epoch": 0.7028824833702882,
      "grad_norm": 0.9333755970001221,
      "learning_rate": 2.0280641000331274e-06,
      "loss": 0.6273,
      "step": 2536
    },
    {
      "epoch": 0.7031596452328159,
      "grad_norm": 1.022659420967102,
      "learning_rate": 2.0245641144669516e-06,
      "loss": 0.6258,
      "step": 2537
    },
    {
      "epoch": 0.7034368070953437,
      "grad_norm": 1.191434383392334,
      "learning_rate": 2.0210663847873354e-06,
      "loss": 0.525,
      "step": 2538
    },
    {
      "epoch": 0.7037139689578714,
      "grad_norm": 0.9467064142227173,
      "learning_rate": 2.017570913646151e-06,
      "loss": 0.5868,
      "step": 2539
    },
    {
      "epoch": 0.7039911308203991,
      "grad_norm": 0.952989399433136,
      "learning_rate": 2.0140777036935637e-06,
      "loss": 0.5692,
      "step": 2540
    },
    {
      "epoch": 0.7042682926829268,
      "grad_norm": 1.1088733673095703,
      "learning_rate": 2.0105867575780195e-06,
      "loss": 0.6018,
      "step": 2541
    },
    {
      "epoch": 0.7045454545454546,
      "grad_norm": 1.0089728832244873,
      "learning_rate": 2.0070980779462513e-06,
      "loss": 0.6693,
      "step": 2542
    },
    {
      "epoch": 0.7048226164079823,
      "grad_norm": 1.1321909427642822,
      "learning_rate": 2.0036116674432653e-06,
      "loss": 0.599,
      "step": 2543
    },
    {
      "epoch": 0.70509977827051,
      "grad_norm": 0.9014486074447632,
      "learning_rate": 2.0001275287123607e-06,
      "loss": 0.6115,
      "step": 2544
    },
    {
      "epoch": 0.7053769401330376,
      "grad_norm": 0.9105014801025391,
      "learning_rate": 1.996645664395104e-06,
      "loss": 0.6305,
      "step": 2545
    },
    {
      "epoch": 0.7056541019955654,
      "grad_norm": 0.8885296583175659,
      "learning_rate": 1.9931660771313417e-06,
      "loss": 0.6413,
      "step": 2546
    },
    {
      "epoch": 0.7059312638580931,
      "grad_norm": 0.9203810095787048,
      "learning_rate": 1.9896887695591925e-06,
      "loss": 0.6064,
      "step": 2547
    },
    {
      "epoch": 0.7062084257206208,
      "grad_norm": 0.8912733793258667,
      "learning_rate": 1.986213744315047e-06,
      "loss": 0.5586,
      "step": 2548
    },
    {
      "epoch": 0.7064855875831486,
      "grad_norm": 0.939007580280304,
      "learning_rate": 1.982741004033566e-06,
      "loss": 0.6081,
      "step": 2549
    },
    {
      "epoch": 0.7067627494456763,
      "grad_norm": 0.8949007987976074,
      "learning_rate": 1.9792705513476747e-06,
      "loss": 0.6123,
      "step": 2550
    },
    {
      "epoch": 0.707039911308204,
      "grad_norm": 1.9172075986862183,
      "learning_rate": 1.975802388888571e-06,
      "loss": 0.5941,
      "step": 2551
    },
    {
      "epoch": 0.7073170731707317,
      "grad_norm": 1.3964992761611938,
      "learning_rate": 1.97233651928571e-06,
      "loss": 0.6568,
      "step": 2552
    },
    {
      "epoch": 0.7075942350332595,
      "grad_norm": 0.9398753643035889,
      "learning_rate": 1.9688729451668116e-06,
      "loss": 0.6532,
      "step": 2553
    },
    {
      "epoch": 0.7078713968957872,
      "grad_norm": 0.9429638981819153,
      "learning_rate": 1.9654116691578494e-06,
      "loss": 0.5843,
      "step": 2554
    },
    {
      "epoch": 0.7081485587583148,
      "grad_norm": 1.3068598508834839,
      "learning_rate": 1.961952693883065e-06,
      "loss": 0.6086,
      "step": 2555
    },
    {
      "epoch": 0.7084257206208425,
      "grad_norm": 0.925593912601471,
      "learning_rate": 1.9584960219649487e-06,
      "loss": 0.574,
      "step": 2556
    },
    {
      "epoch": 0.7087028824833703,
      "grad_norm": 0.968411922454834,
      "learning_rate": 1.955041656024244e-06,
      "loss": 0.5765,
      "step": 2557
    },
    {
      "epoch": 0.708980044345898,
      "grad_norm": 0.8858674168586731,
      "learning_rate": 1.95158959867995e-06,
      "loss": 0.6079,
      "step": 2558
    },
    {
      "epoch": 0.7092572062084257,
      "grad_norm": 0.985417366027832,
      "learning_rate": 1.9481398525493116e-06,
      "loss": 0.6182,
      "step": 2559
    },
    {
      "epoch": 0.7095343680709535,
      "grad_norm": 1.5624934434890747,
      "learning_rate": 1.944692420247825e-06,
      "loss": 0.639,
      "step": 2560
    },
    {
      "epoch": 0.7098115299334812,
      "grad_norm": 0.972358763217926,
      "learning_rate": 1.941247304389227e-06,
      "loss": 0.5898,
      "step": 2561
    },
    {
      "epoch": 0.7100886917960089,
      "grad_norm": 0.9473497867584229,
      "learning_rate": 1.9378045075855063e-06,
      "loss": 0.6021,
      "step": 2562
    },
    {
      "epoch": 0.7103658536585366,
      "grad_norm": 0.977231502532959,
      "learning_rate": 1.9343640324468853e-06,
      "loss": 0.5936,
      "step": 2563
    },
    {
      "epoch": 0.7106430155210643,
      "grad_norm": 0.9414117932319641,
      "learning_rate": 1.9309258815818326e-06,
      "loss": 0.6623,
      "step": 2564
    },
    {
      "epoch": 0.710920177383592,
      "grad_norm": 1.168657660484314,
      "learning_rate": 1.9274900575970456e-06,
      "loss": 0.6058,
      "step": 2565
    },
    {
      "epoch": 0.7111973392461197,
      "grad_norm": 0.9354054927825928,
      "learning_rate": 1.9240565630974685e-06,
      "loss": 0.5886,
      "step": 2566
    },
    {
      "epoch": 0.7114745011086474,
      "grad_norm": 1.1029391288757324,
      "learning_rate": 1.9206254006862726e-06,
      "loss": 0.5884,
      "step": 2567
    },
    {
      "epoch": 0.7117516629711752,
      "grad_norm": 0.906823992729187,
      "learning_rate": 1.9171965729648623e-06,
      "loss": 0.5973,
      "step": 2568
    },
    {
      "epoch": 0.7120288248337029,
      "grad_norm": 0.9374582767486572,
      "learning_rate": 1.913770082532873e-06,
      "loss": 0.6166,
      "step": 2569
    },
    {
      "epoch": 0.7123059866962306,
      "grad_norm": 0.957998514175415,
      "learning_rate": 1.9103459319881673e-06,
      "loss": 0.605,
      "step": 2570
    },
    {
      "epoch": 0.7125831485587583,
      "grad_norm": 0.914169430732727,
      "learning_rate": 1.9069241239268326e-06,
      "loss": 0.5812,
      "step": 2571
    },
    {
      "epoch": 0.7128603104212861,
      "grad_norm": 1.2697951793670654,
      "learning_rate": 1.9035046609431835e-06,
      "loss": 0.6114,
      "step": 2572
    },
    {
      "epoch": 0.7131374722838137,
      "grad_norm": 1.062964916229248,
      "learning_rate": 1.9000875456297506e-06,
      "loss": 0.647,
      "step": 2573
    },
    {
      "epoch": 0.7134146341463414,
      "grad_norm": 0.9897979497909546,
      "learning_rate": 1.896672780577296e-06,
      "loss": 0.6193,
      "step": 2574
    },
    {
      "epoch": 0.7136917960088692,
      "grad_norm": 0.8968859910964966,
      "learning_rate": 1.8932603683747858e-06,
      "loss": 0.5787,
      "step": 2575
    },
    {
      "epoch": 0.7139689578713969,
      "grad_norm": 0.9092812538146973,
      "learning_rate": 1.8898503116094097e-06,
      "loss": 0.6414,
      "step": 2576
    },
    {
      "epoch": 0.7142461197339246,
      "grad_norm": 1.1782081127166748,
      "learning_rate": 1.8864426128665736e-06,
      "loss": 0.5863,
      "step": 2577
    },
    {
      "epoch": 0.7145232815964523,
      "grad_norm": 0.949711263179779,
      "learning_rate": 1.883037274729892e-06,
      "loss": 0.6301,
      "step": 2578
    },
    {
      "epoch": 0.7148004434589801,
      "grad_norm": 1.1857211589813232,
      "learning_rate": 1.87963429978119e-06,
      "loss": 0.5906,
      "step": 2579
    },
    {
      "epoch": 0.7150776053215078,
      "grad_norm": 0.9244367480278015,
      "learning_rate": 1.8762336906005012e-06,
      "loss": 0.558,
      "step": 2580
    },
    {
      "epoch": 0.7153547671840355,
      "grad_norm": 1.0326879024505615,
      "learning_rate": 1.8728354497660666e-06,
      "loss": 0.6718,
      "step": 2581
    },
    {
      "epoch": 0.7156319290465631,
      "grad_norm": 0.9409862160682678,
      "learning_rate": 1.8694395798543303e-06,
      "loss": 0.581,
      "step": 2582
    },
    {
      "epoch": 0.7159090909090909,
      "grad_norm": 0.917054295539856,
      "learning_rate": 1.8660460834399392e-06,
      "loss": 0.6481,
      "step": 2583
    },
    {
      "epoch": 0.7161862527716186,
      "grad_norm": 0.9486770629882812,
      "learning_rate": 1.8626549630957397e-06,
      "loss": 0.6145,
      "step": 2584
    },
    {
      "epoch": 0.7164634146341463,
      "grad_norm": 0.9608169198036194,
      "learning_rate": 1.8592662213927826e-06,
      "loss": 0.6044,
      "step": 2585
    },
    {
      "epoch": 0.7167405764966741,
      "grad_norm": 0.8771957755088806,
      "learning_rate": 1.8558798609003053e-06,
      "loss": 0.5559,
      "step": 2586
    },
    {
      "epoch": 0.7170177383592018,
      "grad_norm": 0.919573962688446,
      "learning_rate": 1.852495884185746e-06,
      "loss": 0.5941,
      "step": 2587
    },
    {
      "epoch": 0.7172949002217295,
      "grad_norm": 1.0577229261398315,
      "learning_rate": 1.8491142938147371e-06,
      "loss": 0.6105,
      "step": 2588
    },
    {
      "epoch": 0.7175720620842572,
      "grad_norm": 0.9977592825889587,
      "learning_rate": 1.8457350923510985e-06,
      "loss": 0.6524,
      "step": 2589
    },
    {
      "epoch": 0.717849223946785,
      "grad_norm": 0.9757591485977173,
      "learning_rate": 1.8423582823568393e-06,
      "loss": 0.6005,
      "step": 2590
    },
    {
      "epoch": 0.7181263858093127,
      "grad_norm": 0.9630820155143738,
      "learning_rate": 1.8389838663921555e-06,
      "loss": 0.605,
      "step": 2591
    },
    {
      "epoch": 0.7184035476718403,
      "grad_norm": 0.9591258764266968,
      "learning_rate": 1.8356118470154295e-06,
      "loss": 0.6562,
      "step": 2592
    },
    {
      "epoch": 0.718680709534368,
      "grad_norm": 0.9422085881233215,
      "learning_rate": 1.8322422267832246e-06,
      "loss": 0.637,
      "step": 2593
    },
    {
      "epoch": 0.7189578713968958,
      "grad_norm": 0.8897464275360107,
      "learning_rate": 1.8288750082502865e-06,
      "loss": 0.6381,
      "step": 2594
    },
    {
      "epoch": 0.7192350332594235,
      "grad_norm": 0.9025383591651917,
      "learning_rate": 1.825510193969538e-06,
      "loss": 0.5835,
      "step": 2595
    },
    {
      "epoch": 0.7195121951219512,
      "grad_norm": 0.9557996392250061,
      "learning_rate": 1.8221477864920861e-06,
      "loss": 0.5855,
      "step": 2596
    },
    {
      "epoch": 0.719789356984479,
      "grad_norm": 0.9291855096817017,
      "learning_rate": 1.8187877883672024e-06,
      "loss": 0.6127,
      "step": 2597
    },
    {
      "epoch": 0.7200665188470067,
      "grad_norm": 0.8927591443061829,
      "learning_rate": 1.8154302021423364e-06,
      "loss": 0.5775,
      "step": 2598
    },
    {
      "epoch": 0.7203436807095344,
      "grad_norm": 0.9637985229492188,
      "learning_rate": 1.8120750303631135e-06,
      "loss": 0.6301,
      "step": 2599
    },
    {
      "epoch": 0.720620842572062,
      "grad_norm": 1.0334081649780273,
      "learning_rate": 1.8087222755733235e-06,
      "loss": 0.6016,
      "step": 2600
    },
    {
      "epoch": 0.7208980044345898,
      "grad_norm": 0.9419227242469788,
      "learning_rate": 1.8053719403149245e-06,
      "loss": 0.6229,
      "step": 2601
    },
    {
      "epoch": 0.7211751662971175,
      "grad_norm": 0.8835015892982483,
      "learning_rate": 1.8020240271280397e-06,
      "loss": 0.6482,
      "step": 2602
    },
    {
      "epoch": 0.7214523281596452,
      "grad_norm": 0.8873974680900574,
      "learning_rate": 1.798678538550958e-06,
      "loss": 0.5704,
      "step": 2603
    },
    {
      "epoch": 0.7217294900221729,
      "grad_norm": 0.9216687083244324,
      "learning_rate": 1.7953354771201286e-06,
      "loss": 0.6179,
      "step": 2604
    },
    {
      "epoch": 0.7220066518847007,
      "grad_norm": 0.9324672818183899,
      "learning_rate": 1.7919948453701608e-06,
      "loss": 0.5921,
      "step": 2605
    },
    {
      "epoch": 0.7222838137472284,
      "grad_norm": 0.9387713074684143,
      "learning_rate": 1.788656645833819e-06,
      "loss": 0.6502,
      "step": 2606
    },
    {
      "epoch": 0.7225609756097561,
      "grad_norm": 0.8915704488754272,
      "learning_rate": 1.785320881042033e-06,
      "loss": 0.6036,
      "step": 2607
    },
    {
      "epoch": 0.7228381374722838,
      "grad_norm": 0.9070380330085754,
      "learning_rate": 1.7819875535238739e-06,
      "loss": 0.575,
      "step": 2608
    },
    {
      "epoch": 0.7231152993348116,
      "grad_norm": 0.8776170015335083,
      "learning_rate": 1.7786566658065723e-06,
      "loss": 0.5819,
      "step": 2609
    },
    {
      "epoch": 0.7233924611973392,
      "grad_norm": 1.1294704675674438,
      "learning_rate": 1.7753282204155076e-06,
      "loss": 0.5936,
      "step": 2610
    },
    {
      "epoch": 0.7236696230598669,
      "grad_norm": 1.3992184400558472,
      "learning_rate": 1.7720022198742092e-06,
      "loss": 0.6028,
      "step": 2611
    },
    {
      "epoch": 0.7239467849223947,
      "grad_norm": 0.9199864268302917,
      "learning_rate": 1.7686786667043509e-06,
      "loss": 0.5992,
      "step": 2612
    },
    {
      "epoch": 0.7242239467849224,
      "grad_norm": 0.9233824610710144,
      "learning_rate": 1.7653575634257513e-06,
      "loss": 0.6274,
      "step": 2613
    },
    {
      "epoch": 0.7245011086474501,
      "grad_norm": 0.8464730381965637,
      "learning_rate": 1.7620389125563708e-06,
      "loss": 0.5472,
      "step": 2614
    },
    {
      "epoch": 0.7247782705099778,
      "grad_norm": 0.8823688626289368,
      "learning_rate": 1.7587227166123117e-06,
      "loss": 0.6052,
      "step": 2615
    },
    {
      "epoch": 0.7250554323725056,
      "grad_norm": 0.910723865032196,
      "learning_rate": 1.755408978107815e-06,
      "loss": 0.6424,
      "step": 2616
    },
    {
      "epoch": 0.7253325942350333,
      "grad_norm": 0.9244400262832642,
      "learning_rate": 1.752097699555258e-06,
      "loss": 0.6195,
      "step": 2617
    },
    {
      "epoch": 0.725609756097561,
      "grad_norm": 0.9056515693664551,
      "learning_rate": 1.748788883465153e-06,
      "loss": 0.5672,
      "step": 2618
    },
    {
      "epoch": 0.7258869179600886,
      "grad_norm": 0.9662718176841736,
      "learning_rate": 1.745482532346145e-06,
      "loss": 0.6405,
      "step": 2619
    },
    {
      "epoch": 0.7261640798226164,
      "grad_norm": 0.909351646900177,
      "learning_rate": 1.7421786487050112e-06,
      "loss": 0.6501,
      "step": 2620
    },
    {
      "epoch": 0.7264412416851441,
      "grad_norm": 0.8854562640190125,
      "learning_rate": 1.738877235046656e-06,
      "loss": 0.5802,
      "step": 2621
    },
    {
      "epoch": 0.7267184035476718,
      "grad_norm": 1.0422521829605103,
      "learning_rate": 1.7355782938741156e-06,
      "loss": 0.6206,
      "step": 2622
    },
    {
      "epoch": 0.7269955654101996,
      "grad_norm": 1.1050196886062622,
      "learning_rate": 1.7322818276885473e-06,
      "loss": 0.5936,
      "step": 2623
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.9484232068061829,
      "learning_rate": 1.7289878389892322e-06,
      "loss": 0.6059,
      "step": 2624
    },
    {
      "epoch": 0.727549889135255,
      "grad_norm": 0.879736065864563,
      "learning_rate": 1.7256963302735752e-06,
      "loss": 0.6634,
      "step": 2625
    },
    {
      "epoch": 0.7278270509977827,
      "grad_norm": 0.9157165288925171,
      "learning_rate": 1.7224073040370992e-06,
      "loss": 0.5811,
      "step": 2626
    },
    {
      "epoch": 0.7281042128603105,
      "grad_norm": 0.9170553088188171,
      "learning_rate": 1.7191207627734452e-06,
      "loss": 0.5869,
      "step": 2627
    },
    {
      "epoch": 0.7283813747228381,
      "grad_norm": 0.9620257616043091,
      "learning_rate": 1.7158367089743722e-06,
      "loss": 0.5954,
      "step": 2628
    },
    {
      "epoch": 0.7286585365853658,
      "grad_norm": 0.902377724647522,
      "learning_rate": 1.7125551451297495e-06,
      "loss": 0.6173,
      "step": 2629
    },
    {
      "epoch": 0.7289356984478935,
      "grad_norm": 0.8985111713409424,
      "learning_rate": 1.7092760737275626e-06,
      "loss": 0.6017,
      "step": 2630
    },
    {
      "epoch": 0.7292128603104213,
      "grad_norm": 0.9750831723213196,
      "learning_rate": 1.705999497253904e-06,
      "loss": 0.6373,
      "step": 2631
    },
    {
      "epoch": 0.729490022172949,
      "grad_norm": 0.9468337297439575,
      "learning_rate": 1.7027254181929754e-06,
      "loss": 0.5883,
      "step": 2632
    },
    {
      "epoch": 0.7297671840354767,
      "grad_norm": 0.8804526329040527,
      "learning_rate": 1.6994538390270887e-06,
      "loss": 0.6079,
      "step": 2633
    },
    {
      "epoch": 0.7300443458980045,
      "grad_norm": 1.053310513496399,
      "learning_rate": 1.6961847622366574e-06,
      "loss": 0.6164,
      "step": 2634
    },
    {
      "epoch": 0.7303215077605322,
      "grad_norm": 0.9017496109008789,
      "learning_rate": 1.6929181903001923e-06,
      "loss": 0.5737,
      "step": 2635
    },
    {
      "epoch": 0.7305986696230599,
      "grad_norm": 0.904101550579071,
      "learning_rate": 1.6896541256943165e-06,
      "loss": 0.6133,
      "step": 2636
    },
    {
      "epoch": 0.7308758314855875,
      "grad_norm": 0.9971810579299927,
      "learning_rate": 1.686392570893744e-06,
      "loss": 0.6381,
      "step": 2637
    },
    {
      "epoch": 0.7311529933481153,
      "grad_norm": 1.0173929929733276,
      "learning_rate": 1.6831335283712885e-06,
      "loss": 0.5744,
      "step": 2638
    },
    {
      "epoch": 0.731430155210643,
      "grad_norm": 1.0616050958633423,
      "learning_rate": 1.679877000597858e-06,
      "loss": 0.6452,
      "step": 2639
    },
    {
      "epoch": 0.7317073170731707,
      "grad_norm": 0.9606106877326965,
      "learning_rate": 1.6766229900424558e-06,
      "loss": 0.5761,
      "step": 2640
    },
    {
      "epoch": 0.7319844789356984,
      "grad_norm": 0.9274082779884338,
      "learning_rate": 1.6733714991721738e-06,
      "loss": 0.6229,
      "step": 2641
    },
    {
      "epoch": 0.7322616407982262,
      "grad_norm": 0.929862380027771,
      "learning_rate": 1.6701225304521967e-06,
      "loss": 0.6195,
      "step": 2642
    },
    {
      "epoch": 0.7325388026607539,
      "grad_norm": 0.9097288846969604,
      "learning_rate": 1.6668760863457929e-06,
      "loss": 0.641,
      "step": 2643
    },
    {
      "epoch": 0.7328159645232816,
      "grad_norm": 0.9352868795394897,
      "learning_rate": 1.663632169314323e-06,
      "loss": 0.6285,
      "step": 2644
    },
    {
      "epoch": 0.7330931263858093,
      "grad_norm": 0.933367908000946,
      "learning_rate": 1.6603907818172277e-06,
      "loss": 0.6294,
      "step": 2645
    },
    {
      "epoch": 0.733370288248337,
      "grad_norm": 1.0231188535690308,
      "learning_rate": 1.6571519263120273e-06,
      "loss": 0.5705,
      "step": 2646
    },
    {
      "epoch": 0.7336474501108647,
      "grad_norm": 0.9520381689071655,
      "learning_rate": 1.6539156052543243e-06,
      "loss": 0.6154,
      "step": 2647
    },
    {
      "epoch": 0.7339246119733924,
      "grad_norm": 0.915677547454834,
      "learning_rate": 1.650681821097805e-06,
      "loss": 0.6711,
      "step": 2648
    },
    {
      "epoch": 0.7342017738359202,
      "grad_norm": 0.9358387589454651,
      "learning_rate": 1.647450576294225e-06,
      "loss": 0.6173,
      "step": 2649
    },
    {
      "epoch": 0.7344789356984479,
      "grad_norm": 1.0654709339141846,
      "learning_rate": 1.6442218732934189e-06,
      "loss": 0.5591,
      "step": 2650
    },
    {
      "epoch": 0.7347560975609756,
      "grad_norm": 0.9490320086479187,
      "learning_rate": 1.640995714543292e-06,
      "loss": 0.5769,
      "step": 2651
    },
    {
      "epoch": 0.7350332594235033,
      "grad_norm": 0.9451633095741272,
      "learning_rate": 1.6377721024898214e-06,
      "loss": 0.6207,
      "step": 2652
    },
    {
      "epoch": 0.7353104212860311,
      "grad_norm": 0.8734908103942871,
      "learning_rate": 1.6345510395770536e-06,
      "loss": 0.5708,
      "step": 2653
    },
    {
      "epoch": 0.7355875831485588,
      "grad_norm": 0.9219774603843689,
      "learning_rate": 1.6313325282471009e-06,
      "loss": 0.6221,
      "step": 2654
    },
    {
      "epoch": 0.7358647450110865,
      "grad_norm": 0.9642323851585388,
      "learning_rate": 1.6281165709401453e-06,
      "loss": 0.6291,
      "step": 2655
    },
    {
      "epoch": 0.7361419068736141,
      "grad_norm": 0.9722159504890442,
      "learning_rate": 1.6249031700944302e-06,
      "loss": 0.5997,
      "step": 2656
    },
    {
      "epoch": 0.7364190687361419,
      "grad_norm": 0.9837990403175354,
      "learning_rate": 1.6216923281462555e-06,
      "loss": 0.6311,
      "step": 2657
    },
    {
      "epoch": 0.7366962305986696,
      "grad_norm": 0.8908426761627197,
      "learning_rate": 1.6184840475299863e-06,
      "loss": 0.6231,
      "step": 2658
    },
    {
      "epoch": 0.7369733924611973,
      "grad_norm": 0.9759544730186462,
      "learning_rate": 1.6152783306780484e-06,
      "loss": 0.639,
      "step": 2659
    },
    {
      "epoch": 0.7372505543237251,
      "grad_norm": 0.8646494746208191,
      "learning_rate": 1.612075180020919e-06,
      "loss": 0.6139,
      "step": 2660
    },
    {
      "epoch": 0.7375277161862528,
      "grad_norm": 0.9379385113716125,
      "learning_rate": 1.6088745979871312e-06,
      "loss": 0.6395,
      "step": 2661
    },
    {
      "epoch": 0.7378048780487805,
      "grad_norm": 1.032354474067688,
      "learning_rate": 1.6056765870032698e-06,
      "loss": 0.5884,
      "step": 2662
    },
    {
      "epoch": 0.7380820399113082,
      "grad_norm": 0.9390593767166138,
      "learning_rate": 1.6024811494939723e-06,
      "loss": 0.5817,
      "step": 2663
    },
    {
      "epoch": 0.738359201773836,
      "grad_norm": 0.8791359066963196,
      "learning_rate": 1.5992882878819233e-06,
      "loss": 0.5878,
      "step": 2664
    },
    {
      "epoch": 0.7386363636363636,
      "grad_norm": 1.079493522644043,
      "learning_rate": 1.5960980045878538e-06,
      "loss": 0.5622,
      "step": 2665
    },
    {
      "epoch": 0.7389135254988913,
      "grad_norm": 0.8706197738647461,
      "learning_rate": 1.5929103020305441e-06,
      "loss": 0.6095,
      "step": 2666
    },
    {
      "epoch": 0.739190687361419,
      "grad_norm": 0.8931334614753723,
      "learning_rate": 1.589725182626815e-06,
      "loss": 0.5886,
      "step": 2667
    },
    {
      "epoch": 0.7394678492239468,
      "grad_norm": 0.9231024980545044,
      "learning_rate": 1.586542648791526e-06,
      "loss": 0.6425,
      "step": 2668
    },
    {
      "epoch": 0.7397450110864745,
      "grad_norm": 0.9225329160690308,
      "learning_rate": 1.5833627029375786e-06,
      "loss": 0.6399,
      "step": 2669
    },
    {
      "epoch": 0.7400221729490022,
      "grad_norm": 0.9641125202178955,
      "learning_rate": 1.5801853474759165e-06,
      "loss": 0.6575,
      "step": 2670
    },
    {
      "epoch": 0.74029933481153,
      "grad_norm": 1.0995537042617798,
      "learning_rate": 1.5770105848155131e-06,
      "loss": 0.6313,
      "step": 2671
    },
    {
      "epoch": 0.7405764966740577,
      "grad_norm": 1.1289243698120117,
      "learning_rate": 1.5738384173633803e-06,
      "loss": 0.6693,
      "step": 2672
    },
    {
      "epoch": 0.7408536585365854,
      "grad_norm": 0.9171227812767029,
      "learning_rate": 1.5706688475245584e-06,
      "loss": 0.6189,
      "step": 2673
    },
    {
      "epoch": 0.741130820399113,
      "grad_norm": 0.9017335176467896,
      "learning_rate": 1.5675018777021223e-06,
      "loss": 0.6275,
      "step": 2674
    },
    {
      "epoch": 0.7414079822616408,
      "grad_norm": 1.199042558670044,
      "learning_rate": 1.5643375102971736e-06,
      "loss": 0.5946,
      "step": 2675
    },
    {
      "epoch": 0.7416851441241685,
      "grad_norm": 0.9708437323570251,
      "learning_rate": 1.5611757477088384e-06,
      "loss": 0.646,
      "step": 2676
    },
    {
      "epoch": 0.7419623059866962,
      "grad_norm": 0.8897635340690613,
      "learning_rate": 1.5580165923342767e-06,
      "loss": 0.6124,
      "step": 2677
    },
    {
      "epoch": 0.7422394678492239,
      "grad_norm": 9.644840240478516,
      "learning_rate": 1.5548600465686603e-06,
      "loss": 0.6057,
      "step": 2678
    },
    {
      "epoch": 0.7425166297117517,
      "grad_norm": 1.1001945734024048,
      "learning_rate": 1.5517061128051891e-06,
      "loss": 0.5899,
      "step": 2679
    },
    {
      "epoch": 0.7427937915742794,
      "grad_norm": 0.9036157727241516,
      "learning_rate": 1.5485547934350803e-06,
      "loss": 0.5939,
      "step": 2680
    },
    {
      "epoch": 0.7430709534368071,
      "grad_norm": 0.9791507124900818,
      "learning_rate": 1.5454060908475733e-06,
      "loss": 0.6197,
      "step": 2681
    },
    {
      "epoch": 0.7433481152993349,
      "grad_norm": 0.9112191796302795,
      "learning_rate": 1.5422600074299183e-06,
      "loss": 0.6445,
      "step": 2682
    },
    {
      "epoch": 0.7436252771618626,
      "grad_norm": 0.8660767674446106,
      "learning_rate": 1.5391165455673823e-06,
      "loss": 0.5687,
      "step": 2683
    },
    {
      "epoch": 0.7439024390243902,
      "grad_norm": 0.8678001165390015,
      "learning_rate": 1.5359757076432434e-06,
      "loss": 0.5719,
      "step": 2684
    },
    {
      "epoch": 0.7441796008869179,
      "grad_norm": 0.9091851711273193,
      "learning_rate": 1.532837496038792e-06,
      "loss": 0.5793,
      "step": 2685
    },
    {
      "epoch": 0.7444567627494457,
      "grad_norm": 1.1939051151275635,
      "learning_rate": 1.5297019131333257e-06,
      "loss": 0.6175,
      "step": 2686
    },
    {
      "epoch": 0.7447339246119734,
      "grad_norm": 0.9126074314117432,
      "learning_rate": 1.5265689613041478e-06,
      "loss": 0.6229,
      "step": 2687
    },
    {
      "epoch": 0.7450110864745011,
      "grad_norm": 0.911407470703125,
      "learning_rate": 1.523438642926574e-06,
      "loss": 0.5743,
      "step": 2688
    },
    {
      "epoch": 0.7452882483370288,
      "grad_norm": 0.8981878757476807,
      "learning_rate": 1.5203109603739136e-06,
      "loss": 0.5471,
      "step": 2689
    },
    {
      "epoch": 0.7455654101995566,
      "grad_norm": 0.8748145699501038,
      "learning_rate": 1.5171859160174834e-06,
      "loss": 0.6081,
      "step": 2690
    },
    {
      "epoch": 0.7458425720620843,
      "grad_norm": 0.8360415101051331,
      "learning_rate": 1.5140635122265968e-06,
      "loss": 0.5517,
      "step": 2691
    },
    {
      "epoch": 0.746119733924612,
      "grad_norm": 1.0377824306488037,
      "learning_rate": 1.5109437513685705e-06,
      "loss": 0.6183,
      "step": 2692
    },
    {
      "epoch": 0.7463968957871396,
      "grad_norm": 0.9267407655715942,
      "learning_rate": 1.5078266358087123e-06,
      "loss": 0.62,
      "step": 2693
    },
    {
      "epoch": 0.7466740576496674,
      "grad_norm": 1.2559822797775269,
      "learning_rate": 1.5047121679103276e-06,
      "loss": 0.567,
      "step": 2694
    },
    {
      "epoch": 0.7469512195121951,
      "grad_norm": 0.85604327917099,
      "learning_rate": 1.5016003500347075e-06,
      "loss": 0.5941,
      "step": 2695
    },
    {
      "epoch": 0.7472283813747228,
      "grad_norm": 0.9598928689956665,
      "learning_rate": 1.4984911845411453e-06,
      "loss": 0.6254,
      "step": 2696
    },
    {
      "epoch": 0.7475055432372506,
      "grad_norm": 1.2117916345596313,
      "learning_rate": 1.4953846737869155e-06,
      "loss": 0.5769,
      "step": 2697
    },
    {
      "epoch": 0.7477827050997783,
      "grad_norm": 0.9220390915870667,
      "learning_rate": 1.4922808201272792e-06,
      "loss": 0.6081,
      "step": 2698
    },
    {
      "epoch": 0.748059866962306,
      "grad_norm": 1.0042617321014404,
      "learning_rate": 1.4891796259154905e-06,
      "loss": 0.6396,
      "step": 2699
    },
    {
      "epoch": 0.7483370288248337,
      "grad_norm": 0.9186859130859375,
      "learning_rate": 1.486081093502778e-06,
      "loss": 0.5787,
      "step": 2700
    },
    {
      "epoch": 0.7486141906873615,
      "grad_norm": 0.9615282416343689,
      "learning_rate": 1.4829852252383582e-06,
      "loss": 0.6398,
      "step": 2701
    },
    {
      "epoch": 0.7488913525498891,
      "grad_norm": 0.8974078893661499,
      "learning_rate": 1.479892023469423e-06,
      "loss": 0.5924,
      "step": 2702
    },
    {
      "epoch": 0.7491685144124168,
      "grad_norm": 0.9370240569114685,
      "learning_rate": 1.4768014905411498e-06,
      "loss": 0.6584,
      "step": 2703
    },
    {
      "epoch": 0.7494456762749445,
      "grad_norm": 1.317137598991394,
      "learning_rate": 1.4737136287966864e-06,
      "loss": 0.5887,
      "step": 2704
    },
    {
      "epoch": 0.7497228381374723,
      "grad_norm": 0.9515214562416077,
      "learning_rate": 1.4706284405771587e-06,
      "loss": 0.6117,
      "step": 2705
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8844241499900818,
      "learning_rate": 1.4675459282216597e-06,
      "loss": 0.6171,
      "step": 2706
    },
    {
      "epoch": 0.7502771618625277,
      "grad_norm": 0.8989660143852234,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.602,
      "step": 2707
    },
    {
      "epoch": 0.7505543237250555,
      "grad_norm": 0.9225838780403137,
      "learning_rate": 1.4613889404490045e-06,
      "loss": 0.582,
      "step": 2708
    },
    {
      "epoch": 0.7508314855875832,
      "grad_norm": 0.9069251418113708,
      "learning_rate": 1.4583144696998897e-06,
      "loss": 0.5848,
      "step": 2709
    },
    {
      "epoch": 0.7511086474501109,
      "grad_norm": 0.8755104541778564,
      "learning_rate": 1.4552426841508937e-06,
      "loss": 0.5681,
      "step": 2710
    },
    {
      "epoch": 0.7513858093126385,
      "grad_norm": 1.0188369750976562,
      "learning_rate": 1.452173586130949e-06,
      "loss": 0.5946,
      "step": 2711
    },
    {
      "epoch": 0.7516629711751663,
      "grad_norm": 1.0545951128005981,
      "learning_rate": 1.4491071779669551e-06,
      "loss": 0.6042,
      "step": 2712
    },
    {
      "epoch": 0.751940133037694,
      "grad_norm": 0.9170023202896118,
      "learning_rate": 1.4460434619837693e-06,
      "loss": 0.6435,
      "step": 2713
    },
    {
      "epoch": 0.7522172949002217,
      "grad_norm": 0.9123980402946472,
      "learning_rate": 1.4429824405042131e-06,
      "loss": 0.5737,
      "step": 2714
    },
    {
      "epoch": 0.7524944567627494,
      "grad_norm": 0.9425738453865051,
      "learning_rate": 1.4399241158490596e-06,
      "loss": 0.6268,
      "step": 2715
    },
    {
      "epoch": 0.7527716186252772,
      "grad_norm": 0.8926444053649902,
      "learning_rate": 1.4368684903370406e-06,
      "loss": 0.6416,
      "step": 2716
    },
    {
      "epoch": 0.7530487804878049,
      "grad_norm": 0.8768932819366455,
      "learning_rate": 1.4338155662848357e-06,
      "loss": 0.5978,
      "step": 2717
    },
    {
      "epoch": 0.7533259423503326,
      "grad_norm": 0.935009777545929,
      "learning_rate": 1.4307653460070848e-06,
      "loss": 0.5529,
      "step": 2718
    },
    {
      "epoch": 0.7536031042128604,
      "grad_norm": 0.9306527376174927,
      "learning_rate": 1.4277178318163732e-06,
      "loss": 0.5703,
      "step": 2719
    },
    {
      "epoch": 0.753880266075388,
      "grad_norm": 0.9080113768577576,
      "learning_rate": 1.4246730260232323e-06,
      "loss": 0.6311,
      "step": 2720
    },
    {
      "epoch": 0.7541574279379157,
      "grad_norm": 0.8746107220649719,
      "learning_rate": 1.421630930936148e-06,
      "loss": 0.6555,
      "step": 2721
    },
    {
      "epoch": 0.7544345898004434,
      "grad_norm": 0.9089754223823547,
      "learning_rate": 1.4185915488615415e-06,
      "loss": 0.5573,
      "step": 2722
    },
    {
      "epoch": 0.7547117516629712,
      "grad_norm": 0.9230754971504211,
      "learning_rate": 1.4155548821037828e-06,
      "loss": 0.6122,
      "step": 2723
    },
    {
      "epoch": 0.7549889135254989,
      "grad_norm": 1.0566086769104004,
      "learning_rate": 1.4125209329651806e-06,
      "loss": 0.6029,
      "step": 2724
    },
    {
      "epoch": 0.7552660753880266,
      "grad_norm": 0.9110480546951294,
      "learning_rate": 1.409489703745987e-06,
      "loss": 0.635,
      "step": 2725
    },
    {
      "epoch": 0.7555432372505543,
      "grad_norm": 0.9055542945861816,
      "learning_rate": 1.406461196744388e-06,
      "loss": 0.592,
      "step": 2726
    },
    {
      "epoch": 0.7558203991130821,
      "grad_norm": 0.9680756330490112,
      "learning_rate": 1.4034354142565092e-06,
      "loss": 0.6264,
      "step": 2727
    },
    {
      "epoch": 0.7560975609756098,
      "grad_norm": 1.0295230150222778,
      "learning_rate": 1.4004123585764035e-06,
      "loss": 0.6037,
      "step": 2728
    },
    {
      "epoch": 0.7563747228381374,
      "grad_norm": 0.9069114327430725,
      "learning_rate": 1.3973920319960654e-06,
      "loss": 0.5994,
      "step": 2729
    },
    {
      "epoch": 0.7566518847006651,
      "grad_norm": 0.881405770778656,
      "learning_rate": 1.3943744368054152e-06,
      "loss": 0.6241,
      "step": 2730
    },
    {
      "epoch": 0.7569290465631929,
      "grad_norm": 0.8689480423927307,
      "learning_rate": 1.3913595752923038e-06,
      "loss": 0.5842,
      "step": 2731
    },
    {
      "epoch": 0.7572062084257206,
      "grad_norm": 0.8979754447937012,
      "learning_rate": 1.3883474497425082e-06,
      "loss": 0.586,
      "step": 2732
    },
    {
      "epoch": 0.7574833702882483,
      "grad_norm": 0.8968691229820251,
      "learning_rate": 1.3853380624397322e-06,
      "loss": 0.5754,
      "step": 2733
    },
    {
      "epoch": 0.7577605321507761,
      "grad_norm": 0.899472713470459,
      "learning_rate": 1.3823314156656042e-06,
      "loss": 0.6499,
      "step": 2734
    },
    {
      "epoch": 0.7580376940133038,
      "grad_norm": 0.9182565212249756,
      "learning_rate": 1.3793275116996713e-06,
      "loss": 0.6406,
      "step": 2735
    },
    {
      "epoch": 0.7583148558758315,
      "grad_norm": 0.9605724215507507,
      "learning_rate": 1.3763263528194071e-06,
      "loss": 0.6145,
      "step": 2736
    },
    {
      "epoch": 0.7585920177383592,
      "grad_norm": 1.0388187170028687,
      "learning_rate": 1.3733279413001998e-06,
      "loss": 0.6205,
      "step": 2737
    },
    {
      "epoch": 0.758869179600887,
      "grad_norm": 0.9141704440116882,
      "learning_rate": 1.3703322794153557e-06,
      "loss": 0.5912,
      "step": 2738
    },
    {
      "epoch": 0.7591463414634146,
      "grad_norm": 0.9080764055252075,
      "learning_rate": 1.367339369436092e-06,
      "loss": 0.6085,
      "step": 2739
    },
    {
      "epoch": 0.7594235033259423,
      "grad_norm": 0.8896963000297546,
      "learning_rate": 1.3643492136315483e-06,
      "loss": 0.609,
      "step": 2740
    },
    {
      "epoch": 0.75970066518847,
      "grad_norm": 0.9679001569747925,
      "learning_rate": 1.36136181426877e-06,
      "loss": 0.5837,
      "step": 2741
    },
    {
      "epoch": 0.7599778270509978,
      "grad_norm": 2.3503236770629883,
      "learning_rate": 1.3583771736127139e-06,
      "loss": 0.6107,
      "step": 2742
    },
    {
      "epoch": 0.7602549889135255,
      "grad_norm": 0.9228477478027344,
      "learning_rate": 1.3553952939262455e-06,
      "loss": 0.6426,
      "step": 2743
    },
    {
      "epoch": 0.7605321507760532,
      "grad_norm": 0.9899008870124817,
      "learning_rate": 1.352416177470136e-06,
      "loss": 0.647,
      "step": 2744
    },
    {
      "epoch": 0.760809312638581,
      "grad_norm": 0.9604774713516235,
      "learning_rate": 1.3494398265030635e-06,
      "loss": 0.6333,
      "step": 2745
    },
    {
      "epoch": 0.7610864745011087,
      "grad_norm": 1.1392303705215454,
      "learning_rate": 1.346466243281606e-06,
      "loss": 0.5765,
      "step": 2746
    },
    {
      "epoch": 0.7613636363636364,
      "grad_norm": 1.122330665588379,
      "learning_rate": 1.3434954300602487e-06,
      "loss": 0.6348,
      "step": 2747
    },
    {
      "epoch": 0.761640798226164,
      "grad_norm": 0.9127699732780457,
      "learning_rate": 1.340527389091374e-06,
      "loss": 0.6507,
      "step": 2748
    },
    {
      "epoch": 0.7619179600886918,
      "grad_norm": 0.8936976194381714,
      "learning_rate": 1.3375621226252583e-06,
      "loss": 0.5802,
      "step": 2749
    },
    {
      "epoch": 0.7621951219512195,
      "grad_norm": 0.903390109539032,
      "learning_rate": 1.334599632910078e-06,
      "loss": 0.5789,
      "step": 2750
    },
    {
      "epoch": 0.7624722838137472,
      "grad_norm": 0.9042217135429382,
      "learning_rate": 1.3316399221919075e-06,
      "loss": 0.6102,
      "step": 2751
    },
    {
      "epoch": 0.7627494456762749,
      "grad_norm": 0.8959875702857971,
      "learning_rate": 1.3286829927147093e-06,
      "loss": 0.5911,
      "step": 2752
    },
    {
      "epoch": 0.7630266075388027,
      "grad_norm": 0.8885048031806946,
      "learning_rate": 1.325728846720339e-06,
      "loss": 0.6088,
      "step": 2753
    },
    {
      "epoch": 0.7633037694013304,
      "grad_norm": 0.907653272151947,
      "learning_rate": 1.3227774864485416e-06,
      "loss": 0.6289,
      "step": 2754
    },
    {
      "epoch": 0.7635809312638581,
      "grad_norm": 0.9569871425628662,
      "learning_rate": 1.3198289141369507e-06,
      "loss": 0.616,
      "step": 2755
    },
    {
      "epoch": 0.7638580931263859,
      "grad_norm": 0.936529278755188,
      "learning_rate": 1.3168831320210857e-06,
      "loss": 0.6385,
      "step": 2756
    },
    {
      "epoch": 0.7641352549889135,
      "grad_norm": 0.8818163871765137,
      "learning_rate": 1.3139401423343484e-06,
      "loss": 0.6275,
      "step": 2757
    },
    {
      "epoch": 0.7644124168514412,
      "grad_norm": 0.954918384552002,
      "learning_rate": 1.3109999473080292e-06,
      "loss": 0.6781,
      "step": 2758
    },
    {
      "epoch": 0.7646895787139689,
      "grad_norm": 1.012184977531433,
      "learning_rate": 1.3080625491712957e-06,
      "loss": 0.5962,
      "step": 2759
    },
    {
      "epoch": 0.7649667405764967,
      "grad_norm": 0.9162760376930237,
      "learning_rate": 1.3051279501511926e-06,
      "loss": 0.591,
      "step": 2760
    },
    {
      "epoch": 0.7652439024390244,
      "grad_norm": 1.0084331035614014,
      "learning_rate": 1.3021961524726457e-06,
      "loss": 0.6276,
      "step": 2761
    },
    {
      "epoch": 0.7655210643015521,
      "grad_norm": 2.0890440940856934,
      "learning_rate": 1.2992671583584587e-06,
      "loss": 0.6092,
      "step": 2762
    },
    {
      "epoch": 0.7657982261640798,
      "grad_norm": 1.0363022089004517,
      "learning_rate": 1.2963409700293056e-06,
      "loss": 0.6156,
      "step": 2763
    },
    {
      "epoch": 0.7660753880266076,
      "grad_norm": 0.927970290184021,
      "learning_rate": 1.293417589703736e-06,
      "loss": 0.5699,
      "step": 2764
    },
    {
      "epoch": 0.7663525498891353,
      "grad_norm": 0.9648489952087402,
      "learning_rate": 1.2904970195981686e-06,
      "loss": 0.5751,
      "step": 2765
    },
    {
      "epoch": 0.766629711751663,
      "grad_norm": 0.8916625380516052,
      "learning_rate": 1.2875792619268923e-06,
      "loss": 0.5934,
      "step": 2766
    },
    {
      "epoch": 0.7669068736141907,
      "grad_norm": 1.0316390991210938,
      "learning_rate": 1.2846643189020646e-06,
      "loss": 0.6165,
      "step": 2767
    },
    {
      "epoch": 0.7671840354767184,
      "grad_norm": 1.2923777103424072,
      "learning_rate": 1.2817521927337062e-06,
      "loss": 0.653,
      "step": 2768
    },
    {
      "epoch": 0.7674611973392461,
      "grad_norm": 0.9731424450874329,
      "learning_rate": 1.278842885629707e-06,
      "loss": 0.5688,
      "step": 2769
    },
    {
      "epoch": 0.7677383592017738,
      "grad_norm": 0.9841533899307251,
      "learning_rate": 1.2759363997958168e-06,
      "loss": 0.6008,
      "step": 2770
    },
    {
      "epoch": 0.7680155210643016,
      "grad_norm": 19.405967712402344,
      "learning_rate": 1.2730327374356432e-06,
      "loss": 0.6182,
      "step": 2771
    },
    {
      "epoch": 0.7682926829268293,
      "grad_norm": 0.9332964420318604,
      "learning_rate": 1.2701319007506563e-06,
      "loss": 0.5514,
      "step": 2772
    },
    {
      "epoch": 0.768569844789357,
      "grad_norm": 0.8506276607513428,
      "learning_rate": 1.2672338919401866e-06,
      "loss": 0.5768,
      "step": 2773
    },
    {
      "epoch": 0.7688470066518847,
      "grad_norm": 0.9194921255111694,
      "learning_rate": 1.2643387132014169e-06,
      "loss": 0.6215,
      "step": 2774
    },
    {
      "epoch": 0.7691241685144125,
      "grad_norm": 0.897859513759613,
      "learning_rate": 1.2614463667293847e-06,
      "loss": 0.6168,
      "step": 2775
    },
    {
      "epoch": 0.7694013303769401,
      "grad_norm": 0.9653081297874451,
      "learning_rate": 1.2585568547169802e-06,
      "loss": 0.6139,
      "step": 2776
    },
    {
      "epoch": 0.7696784922394678,
      "grad_norm": 2.7514216899871826,
      "learning_rate": 1.2556701793549458e-06,
      "loss": 0.5999,
      "step": 2777
    },
    {
      "epoch": 0.7699556541019955,
      "grad_norm": 0.9534131288528442,
      "learning_rate": 1.2527863428318726e-06,
      "loss": 0.6153,
      "step": 2778
    },
    {
      "epoch": 0.7702328159645233,
      "grad_norm": 1.0352519750595093,
      "learning_rate": 1.2499053473341982e-06,
      "loss": 0.6252,
      "step": 2779
    },
    {
      "epoch": 0.770509977827051,
      "grad_norm": 0.8873860836029053,
      "learning_rate": 1.247027195046207e-06,
      "loss": 0.6409,
      "step": 2780
    },
    {
      "epoch": 0.7707871396895787,
      "grad_norm": 0.8832753300666809,
      "learning_rate": 1.244151888150032e-06,
      "loss": 0.5752,
      "step": 2781
    },
    {
      "epoch": 0.7710643015521065,
      "grad_norm": 0.8507212996482849,
      "learning_rate": 1.2412794288256403e-06,
      "loss": 0.5867,
      "step": 2782
    },
    {
      "epoch": 0.7713414634146342,
      "grad_norm": 1.0074859857559204,
      "learning_rate": 1.2384098192508448e-06,
      "loss": 0.6194,
      "step": 2783
    },
    {
      "epoch": 0.7716186252771619,
      "grad_norm": 0.9031928777694702,
      "learning_rate": 1.235543061601301e-06,
      "loss": 0.6507,
      "step": 2784
    },
    {
      "epoch": 0.7718957871396895,
      "grad_norm": 0.8621088862419128,
      "learning_rate": 1.232679158050497e-06,
      "loss": 0.5982,
      "step": 2785
    },
    {
      "epoch": 0.7721729490022173,
      "grad_norm": 0.8918038010597229,
      "learning_rate": 1.2298181107697587e-06,
      "loss": 0.6197,
      "step": 2786
    },
    {
      "epoch": 0.772450110864745,
      "grad_norm": 0.904039740562439,
      "learning_rate": 1.2269599219282474e-06,
      "loss": 0.5792,
      "step": 2787
    },
    {
      "epoch": 0.7727272727272727,
      "grad_norm": 0.8730078339576721,
      "learning_rate": 1.2241045936929563e-06,
      "loss": 0.5868,
      "step": 2788
    },
    {
      "epoch": 0.7730044345898004,
      "grad_norm": 0.8887600302696228,
      "learning_rate": 1.2212521282287093e-06,
      "loss": 0.6152,
      "step": 2789
    },
    {
      "epoch": 0.7732815964523282,
      "grad_norm": 0.9191362857818604,
      "learning_rate": 1.2184025276981604e-06,
      "loss": 0.5911,
      "step": 2790
    },
    {
      "epoch": 0.7735587583148559,
      "grad_norm": 0.9001409411430359,
      "learning_rate": 1.2155557942617925e-06,
      "loss": 0.5485,
      "step": 2791
    },
    {
      "epoch": 0.7738359201773836,
      "grad_norm": 0.9301763772964478,
      "learning_rate": 1.2127119300779134e-06,
      "loss": 0.6378,
      "step": 2792
    },
    {
      "epoch": 0.7741130820399114,
      "grad_norm": 1.1996046304702759,
      "learning_rate": 1.2098709373026552e-06,
      "loss": 0.5607,
      "step": 2793
    },
    {
      "epoch": 0.774390243902439,
      "grad_norm": 0.9698590040206909,
      "learning_rate": 1.2070328180899732e-06,
      "loss": 0.638,
      "step": 2794
    },
    {
      "epoch": 0.7746674057649667,
      "grad_norm": 0.940263569355011,
      "learning_rate": 1.2041975745916474e-06,
      "loss": 0.6306,
      "step": 2795
    },
    {
      "epoch": 0.7749445676274944,
      "grad_norm": 0.9340899586677551,
      "learning_rate": 1.2013652089572724e-06,
      "loss": 0.581,
      "step": 2796
    },
    {
      "epoch": 0.7752217294900222,
      "grad_norm": 0.8662264347076416,
      "learning_rate": 1.1985357233342643e-06,
      "loss": 0.5647,
      "step": 2797
    },
    {
      "epoch": 0.7754988913525499,
      "grad_norm": 0.903910756111145,
      "learning_rate": 1.1957091198678538e-06,
      "loss": 0.6341,
      "step": 2798
    },
    {
      "epoch": 0.7757760532150776,
      "grad_norm": 1.0029243230819702,
      "learning_rate": 1.192885400701087e-06,
      "loss": 0.6547,
      "step": 2799
    },
    {
      "epoch": 0.7760532150776053,
      "grad_norm": 1.0451689958572388,
      "learning_rate": 1.1900645679748241e-06,
      "loss": 0.6242,
      "step": 2800
    },
    {
      "epoch": 0.7763303769401331,
      "grad_norm": 0.8692581057548523,
      "learning_rate": 1.1872466238277357e-06,
      "loss": 0.5432,
      "step": 2801
    },
    {
      "epoch": 0.7766075388026608,
      "grad_norm": 0.9086769223213196,
      "learning_rate": 1.1844315703963028e-06,
      "loss": 0.6104,
      "step": 2802
    },
    {
      "epoch": 0.7768847006651884,
      "grad_norm": 1.4469305276870728,
      "learning_rate": 1.1816194098148148e-06,
      "loss": 0.6463,
      "step": 2803
    },
    {
      "epoch": 0.7771618625277162,
      "grad_norm": 0.8784159421920776,
      "learning_rate": 1.1788101442153683e-06,
      "loss": 0.5629,
      "step": 2804
    },
    {
      "epoch": 0.7774390243902439,
      "grad_norm": 0.9766161441802979,
      "learning_rate": 1.1760037757278618e-06,
      "loss": 0.6012,
      "step": 2805
    },
    {
      "epoch": 0.7777161862527716,
      "grad_norm": 0.901168704032898,
      "learning_rate": 1.1732003064800046e-06,
      "loss": 0.5986,
      "step": 2806
    },
    {
      "epoch": 0.7779933481152993,
      "grad_norm": 0.9314857721328735,
      "learning_rate": 1.170399738597301e-06,
      "loss": 0.5666,
      "step": 2807
    },
    {
      "epoch": 0.7782705099778271,
      "grad_norm": 0.9054350256919861,
      "learning_rate": 1.1676020742030585e-06,
      "loss": 0.5876,
      "step": 2808
    },
    {
      "epoch": 0.7785476718403548,
      "grad_norm": 2.843311071395874,
      "learning_rate": 1.1648073154183799e-06,
      "loss": 0.6079,
      "step": 2809
    },
    {
      "epoch": 0.7788248337028825,
      "grad_norm": 1.161486268043518,
      "learning_rate": 1.1620154643621712e-06,
      "loss": 0.5721,
      "step": 2810
    },
    {
      "epoch": 0.7791019955654102,
      "grad_norm": 0.9468228816986084,
      "learning_rate": 1.1592265231511291e-06,
      "loss": 0.6036,
      "step": 2811
    },
    {
      "epoch": 0.779379157427938,
      "grad_norm": 0.956716775894165,
      "learning_rate": 1.1564404938997454e-06,
      "loss": 0.6172,
      "step": 2812
    },
    {
      "epoch": 0.7796563192904656,
      "grad_norm": 0.8925110101699829,
      "learning_rate": 1.1536573787203042e-06,
      "loss": 0.6009,
      "step": 2813
    },
    {
      "epoch": 0.7799334811529933,
      "grad_norm": 0.8762784600257874,
      "learning_rate": 1.1508771797228801e-06,
      "loss": 0.6119,
      "step": 2814
    },
    {
      "epoch": 0.780210643015521,
      "grad_norm": 0.8965146541595459,
      "learning_rate": 1.1480998990153375e-06,
      "loss": 0.5651,
      "step": 2815
    },
    {
      "epoch": 0.7804878048780488,
      "grad_norm": 0.9722217321395874,
      "learning_rate": 1.1453255387033274e-06,
      "loss": 0.6424,
      "step": 2816
    },
    {
      "epoch": 0.7807649667405765,
      "grad_norm": 1.0520968437194824,
      "learning_rate": 1.1425541008902852e-06,
      "loss": 0.5967,
      "step": 2817
    },
    {
      "epoch": 0.7810421286031042,
      "grad_norm": 0.9263484477996826,
      "learning_rate": 1.139785587677435e-06,
      "loss": 0.6038,
      "step": 2818
    },
    {
      "epoch": 0.781319290465632,
      "grad_norm": 0.949636697769165,
      "learning_rate": 1.1370200011637811e-06,
      "loss": 0.6218,
      "step": 2819
    },
    {
      "epoch": 0.7815964523281597,
      "grad_norm": 1.1339818239212036,
      "learning_rate": 1.1342573434461041e-06,
      "loss": 0.5644,
      "step": 2820
    },
    {
      "epoch": 0.7818736141906873,
      "grad_norm": 0.8858981132507324,
      "learning_rate": 1.1314976166189722e-06,
      "loss": 0.5778,
      "step": 2821
    },
    {
      "epoch": 0.782150776053215,
      "grad_norm": 0.872136652469635,
      "learning_rate": 1.1287408227747276e-06,
      "loss": 0.5597,
      "step": 2822
    },
    {
      "epoch": 0.7824279379157428,
      "grad_norm": 1.0090842247009277,
      "learning_rate": 1.1259869640034881e-06,
      "loss": 0.6352,
      "step": 2823
    },
    {
      "epoch": 0.7827050997782705,
      "grad_norm": 1.175087571144104,
      "learning_rate": 1.1232360423931476e-06,
      "loss": 0.6215,
      "step": 2824
    },
    {
      "epoch": 0.7829822616407982,
      "grad_norm": 0.9332869052886963,
      "learning_rate": 1.1204880600293728e-06,
      "loss": 0.5815,
      "step": 2825
    },
    {
      "epoch": 0.7832594235033259,
      "grad_norm": 0.8935688138008118,
      "learning_rate": 1.1177430189956007e-06,
      "loss": 0.5904,
      "step": 2826
    },
    {
      "epoch": 0.7835365853658537,
      "grad_norm": 0.9881109595298767,
      "learning_rate": 1.1150009213730407e-06,
      "loss": 0.5981,
      "step": 2827
    },
    {
      "epoch": 0.7838137472283814,
      "grad_norm": 7.601603984832764,
      "learning_rate": 1.1122617692406667e-06,
      "loss": 0.5685,
      "step": 2828
    },
    {
      "epoch": 0.7840909090909091,
      "grad_norm": 1.0429744720458984,
      "learning_rate": 1.1095255646752256e-06,
      "loss": 0.6192,
      "step": 2829
    },
    {
      "epoch": 0.7843680709534369,
      "grad_norm": 0.8771736025810242,
      "learning_rate": 1.1067923097512256e-06,
      "loss": 0.5799,
      "step": 2830
    },
    {
      "epoch": 0.7846452328159645,
      "grad_norm": 0.9138473272323608,
      "learning_rate": 1.1040620065409347e-06,
      "loss": 0.6195,
      "step": 2831
    },
    {
      "epoch": 0.7849223946784922,
      "grad_norm": 0.9057692289352417,
      "learning_rate": 1.1013346571143911e-06,
      "loss": 0.5967,
      "step": 2832
    },
    {
      "epoch": 0.7851995565410199,
      "grad_norm": 3.365168809890747,
      "learning_rate": 1.0986102635393891e-06,
      "loss": 0.6102,
      "step": 2833
    },
    {
      "epoch": 0.7854767184035477,
      "grad_norm": 1.0161689519882202,
      "learning_rate": 1.0958888278814818e-06,
      "loss": 0.5844,
      "step": 2834
    },
    {
      "epoch": 0.7857538802660754,
      "grad_norm": 0.9107481241226196,
      "learning_rate": 1.0931703522039806e-06,
      "loss": 0.5588,
      "step": 2835
    },
    {
      "epoch": 0.7860310421286031,
      "grad_norm": 0.9658748507499695,
      "learning_rate": 1.0904548385679526e-06,
      "loss": 0.6454,
      "step": 2836
    },
    {
      "epoch": 0.7863082039911308,
      "grad_norm": 0.9087231755256653,
      "learning_rate": 1.0877422890322181e-06,
      "loss": 0.6048,
      "step": 2837
    },
    {
      "epoch": 0.7865853658536586,
      "grad_norm": 0.8574754595756531,
      "learning_rate": 1.0850327056533527e-06,
      "loss": 0.5965,
      "step": 2838
    },
    {
      "epoch": 0.7868625277161863,
      "grad_norm": 0.9230338335037231,
      "learning_rate": 1.0823260904856791e-06,
      "loss": 0.5904,
      "step": 2839
    },
    {
      "epoch": 0.7871396895787139,
      "grad_norm": 0.9008994698524475,
      "learning_rate": 1.0796224455812753e-06,
      "loss": 0.6204,
      "step": 2840
    },
    {
      "epoch": 0.7874168514412417,
      "grad_norm": 0.9164884090423584,
      "learning_rate": 1.0769217729899634e-06,
      "loss": 0.6223,
      "step": 2841
    },
    {
      "epoch": 0.7876940133037694,
      "grad_norm": 0.9149385690689087,
      "learning_rate": 1.074224074759309e-06,
      "loss": 0.6049,
      "step": 2842
    },
    {
      "epoch": 0.7879711751662971,
      "grad_norm": 0.9520339369773865,
      "learning_rate": 1.0715293529346305e-06,
      "loss": 0.6159,
      "step": 2843
    },
    {
      "epoch": 0.7882483370288248,
      "grad_norm": 0.930320143699646,
      "learning_rate": 1.0688376095589837e-06,
      "loss": 0.5576,
      "step": 2844
    },
    {
      "epoch": 0.7885254988913526,
      "grad_norm": 0.9795324802398682,
      "learning_rate": 1.0661488466731678e-06,
      "loss": 0.6092,
      "step": 2845
    },
    {
      "epoch": 0.7888026607538803,
      "grad_norm": 0.9603287577629089,
      "learning_rate": 1.0634630663157236e-06,
      "loss": 0.654,
      "step": 2846
    },
    {
      "epoch": 0.789079822616408,
      "grad_norm": 0.885473906993866,
      "learning_rate": 1.0607802705229282e-06,
      "loss": 0.6357,
      "step": 2847
    },
    {
      "epoch": 0.7893569844789357,
      "grad_norm": 0.9706182479858398,
      "learning_rate": 1.0581004613287976e-06,
      "loss": 0.5778,
      "step": 2848
    },
    {
      "epoch": 0.7896341463414634,
      "grad_norm": 0.9267870783805847,
      "learning_rate": 1.0554236407650837e-06,
      "loss": 0.569,
      "step": 2849
    },
    {
      "epoch": 0.7899113082039911,
      "grad_norm": 0.8881688117980957,
      "learning_rate": 1.0527498108612694e-06,
      "loss": 0.5821,
      "step": 2850
    },
    {
      "epoch": 0.7901884700665188,
      "grad_norm": 1.1143068075180054,
      "learning_rate": 1.0500789736445776e-06,
      "loss": 0.5678,
      "step": 2851
    },
    {
      "epoch": 0.7904656319290465,
      "grad_norm": 0.9259443283081055,
      "learning_rate": 1.0474111311399527e-06,
      "loss": 0.5736,
      "step": 2852
    },
    {
      "epoch": 0.7907427937915743,
      "grad_norm": 0.9418421387672424,
      "learning_rate": 1.0447462853700746e-06,
      "loss": 0.6181,
      "step": 2853
    },
    {
      "epoch": 0.791019955654102,
      "grad_norm": 1.3612916469573975,
      "learning_rate": 1.042084438355348e-06,
      "loss": 0.63,
      "step": 2854
    },
    {
      "epoch": 0.7912971175166297,
      "grad_norm": 0.9086347818374634,
      "learning_rate": 1.039425592113909e-06,
      "loss": 0.5763,
      "step": 2855
    },
    {
      "epoch": 0.7915742793791575,
      "grad_norm": 0.9217668771743774,
      "learning_rate": 1.036769748661614e-06,
      "loss": 0.5765,
      "step": 2856
    },
    {
      "epoch": 0.7918514412416852,
      "grad_norm": 0.8801589012145996,
      "learning_rate": 1.034116910012044e-06,
      "loss": 0.5826,
      "step": 2857
    },
    {
      "epoch": 0.7921286031042128,
      "grad_norm": 1.0377134084701538,
      "learning_rate": 1.0314670781765018e-06,
      "loss": 0.5879,
      "step": 2858
    },
    {
      "epoch": 0.7924057649667405,
      "grad_norm": 0.9336162209510803,
      "learning_rate": 1.0288202551640109e-06,
      "loss": 0.6015,
      "step": 2859
    },
    {
      "epoch": 0.7926829268292683,
      "grad_norm": 1.0927352905273438,
      "learning_rate": 1.0261764429813136e-06,
      "loss": 0.5343,
      "step": 2860
    },
    {
      "epoch": 0.792960088691796,
      "grad_norm": 0.9454530477523804,
      "learning_rate": 1.0235356436328675e-06,
      "loss": 0.6088,
      "step": 2861
    },
    {
      "epoch": 0.7932372505543237,
      "grad_norm": 0.9535458087921143,
      "learning_rate": 1.020897859120852e-06,
      "loss": 0.6074,
      "step": 2862
    },
    {
      "epoch": 0.7935144124168514,
      "grad_norm": 0.9658083915710449,
      "learning_rate": 1.018263091445152e-06,
      "loss": 0.5716,
      "step": 2863
    },
    {
      "epoch": 0.7937915742793792,
      "grad_norm": 1.1349619626998901,
      "learning_rate": 1.0156313426033715e-06,
      "loss": 0.6285,
      "step": 2864
    },
    {
      "epoch": 0.7940687361419069,
      "grad_norm": 0.954574704170227,
      "learning_rate": 1.0130026145908222e-06,
      "loss": 0.6264,
      "step": 2865
    },
    {
      "epoch": 0.7943458980044346,
      "grad_norm": 0.9433424472808838,
      "learning_rate": 1.01037690940053e-06,
      "loss": 0.5813,
      "step": 2866
    },
    {
      "epoch": 0.7946230598669624,
      "grad_norm": 0.8976333737373352,
      "learning_rate": 1.007754229023225e-06,
      "loss": 0.5947,
      "step": 2867
    },
    {
      "epoch": 0.79490022172949,
      "grad_norm": 0.8800441026687622,
      "learning_rate": 1.0051345754473446e-06,
      "loss": 0.5597,
      "step": 2868
    },
    {
      "epoch": 0.7951773835920177,
      "grad_norm": 0.8776037693023682,
      "learning_rate": 1.0025179506590315e-06,
      "loss": 0.5728,
      "step": 2869
    },
    {
      "epoch": 0.7954545454545454,
      "grad_norm": 0.8595742583274841,
      "learning_rate": 9.999043566421334e-07,
      "loss": 0.5584,
      "step": 2870
    },
    {
      "epoch": 0.7957317073170732,
      "grad_norm": 0.975944459438324,
      "learning_rate": 9.972937953781985e-07,
      "loss": 0.5713,
      "step": 2871
    },
    {
      "epoch": 0.7960088691796009,
      "grad_norm": 0.9173409938812256,
      "learning_rate": 9.946862688464753e-07,
      "loss": 0.5668,
      "step": 2872
    },
    {
      "epoch": 0.7962860310421286,
      "grad_norm": 0.9533830285072327,
      "learning_rate": 9.920817790239163e-07,
      "loss": 0.6036,
      "step": 2873
    },
    {
      "epoch": 0.7965631929046563,
      "grad_norm": 0.971548318862915,
      "learning_rate": 9.89480327885164e-07,
      "loss": 0.6746,
      "step": 2874
    },
    {
      "epoch": 0.7968403547671841,
      "grad_norm": 1.123424768447876,
      "learning_rate": 9.868819174025617e-07,
      "loss": 0.6175,
      "step": 2875
    },
    {
      "epoch": 0.7971175166297118,
      "grad_norm": 0.9111065864562988,
      "learning_rate": 9.842865495461462e-07,
      "loss": 0.5985,
      "step": 2876
    },
    {
      "epoch": 0.7973946784922394,
      "grad_norm": 0.8669419288635254,
      "learning_rate": 9.816942262836494e-07,
      "loss": 0.577,
      "step": 2877
    },
    {
      "epoch": 0.7976718403547672,
      "grad_norm": 0.8944295048713684,
      "learning_rate": 9.791049495804928e-07,
      "loss": 0.6118,
      "step": 2878
    },
    {
      "epoch": 0.7979490022172949,
      "grad_norm": 0.8854717016220093,
      "learning_rate": 9.765187213997885e-07,
      "loss": 0.5779,
      "step": 2879
    },
    {
      "epoch": 0.7982261640798226,
      "grad_norm": 0.8605630993843079,
      "learning_rate": 9.739355437023373e-07,
      "loss": 0.5962,
      "step": 2880
    },
    {
      "epoch": 0.7985033259423503,
      "grad_norm": 0.9200057983398438,
      "learning_rate": 9.713554184466267e-07,
      "loss": 0.5828,
      "step": 2881
    },
    {
      "epoch": 0.7987804878048781,
      "grad_norm": 0.8976324200630188,
      "learning_rate": 9.687783475888317e-07,
      "loss": 0.5596,
      "step": 2882
    },
    {
      "epoch": 0.7990576496674058,
      "grad_norm": 0.9166635274887085,
      "learning_rate": 9.662043330828086e-07,
      "loss": 0.5637,
      "step": 2883
    },
    {
      "epoch": 0.7993348115299335,
      "grad_norm": 1.117083191871643,
      "learning_rate": 9.63633376880102e-07,
      "loss": 0.629,
      "step": 2884
    },
    {
      "epoch": 0.7996119733924612,
      "grad_norm": 0.9207936525344849,
      "learning_rate": 9.610654809299303e-07,
      "loss": 0.5664,
      "step": 2885
    },
    {
      "epoch": 0.799889135254989,
      "grad_norm": 0.8483346104621887,
      "learning_rate": 9.585006471791963e-07,
      "loss": 0.5793,
      "step": 2886
    },
    {
      "epoch": 0.8001662971175166,
      "grad_norm": 0.9296838641166687,
      "learning_rate": 9.559388775724798e-07,
      "loss": 0.6213,
      "step": 2887
    },
    {
      "epoch": 0.8001662971175166,
      "eval_loss": 0.6023418307304382,
      "eval_runtime": 387.3245,
      "eval_samples_per_second": 8.339,
      "eval_steps_per_second": 1.043,
      "step": 2887
    },
    {
      "epoch": 0.8004434589800443,
      "grad_norm": 0.9403005242347717,
      "learning_rate": 9.533801740520405e-07,
      "loss": 0.6139,
      "step": 2888
    },
    {
      "epoch": 0.8007206208425721,
      "grad_norm": 2.8383069038391113,
      "learning_rate": 9.508245385578085e-07,
      "loss": 0.5442,
      "step": 2889
    },
    {
      "epoch": 0.8009977827050998,
      "grad_norm": 0.9428724646568298,
      "learning_rate": 9.482719730273921e-07,
      "loss": 0.6295,
      "step": 2890
    },
    {
      "epoch": 0.8012749445676275,
      "grad_norm": 0.904663622379303,
      "learning_rate": 9.457224793960684e-07,
      "loss": 0.6145,
      "step": 2891
    },
    {
      "epoch": 0.8015521064301552,
      "grad_norm": 0.9360557198524475,
      "learning_rate": 9.431760595967882e-07,
      "loss": 0.5769,
      "step": 2892
    },
    {
      "epoch": 0.801829268292683,
      "grad_norm": 0.9010534882545471,
      "learning_rate": 9.406327155601708e-07,
      "loss": 0.6488,
      "step": 2893
    },
    {
      "epoch": 0.8021064301552107,
      "grad_norm": 0.8938716650009155,
      "learning_rate": 9.380924492145033e-07,
      "loss": 0.6279,
      "step": 2894
    },
    {
      "epoch": 0.8023835920177383,
      "grad_norm": 0.9596985578536987,
      "learning_rate": 9.355552624857401e-07,
      "loss": 0.6327,
      "step": 2895
    },
    {
      "epoch": 0.802660753880266,
      "grad_norm": 0.909292995929718,
      "learning_rate": 9.330211572975001e-07,
      "loss": 0.6021,
      "step": 2896
    },
    {
      "epoch": 0.8029379157427938,
      "grad_norm": 0.8601246476173401,
      "learning_rate": 9.304901355710666e-07,
      "loss": 0.5678,
      "step": 2897
    },
    {
      "epoch": 0.8032150776053215,
      "grad_norm": 0.9699132442474365,
      "learning_rate": 9.279621992253824e-07,
      "loss": 0.6235,
      "step": 2898
    },
    {
      "epoch": 0.8034922394678492,
      "grad_norm": 1.073152780532837,
      "learning_rate": 9.254373501770569e-07,
      "loss": 0.6229,
      "step": 2899
    },
    {
      "epoch": 0.8037694013303769,
      "grad_norm": 0.9197423458099365,
      "learning_rate": 9.229155903403531e-07,
      "loss": 0.6219,
      "step": 2900
    },
    {
      "epoch": 0.8040465631929047,
      "grad_norm": 0.8869557976722717,
      "learning_rate": 9.203969216271958e-07,
      "loss": 0.6306,
      "step": 2901
    },
    {
      "epoch": 0.8043237250554324,
      "grad_norm": 0.9480774998664856,
      "learning_rate": 9.178813459471603e-07,
      "loss": 0.6477,
      "step": 2902
    },
    {
      "epoch": 0.8046008869179601,
      "grad_norm": 0.8836572766304016,
      "learning_rate": 9.153688652074844e-07,
      "loss": 0.5494,
      "step": 2903
    },
    {
      "epoch": 0.8048780487804879,
      "grad_norm": 0.860690712928772,
      "learning_rate": 9.128594813130554e-07,
      "loss": 0.5983,
      "step": 2904
    },
    {
      "epoch": 0.8051552106430155,
      "grad_norm": 0.8752050399780273,
      "learning_rate": 9.10353196166412e-07,
      "loss": 0.601,
      "step": 2905
    },
    {
      "epoch": 0.8054323725055432,
      "grad_norm": 1.285021185874939,
      "learning_rate": 9.078500116677458e-07,
      "loss": 0.6095,
      "step": 2906
    },
    {
      "epoch": 0.8057095343680709,
      "grad_norm": 0.9997894167900085,
      "learning_rate": 9.05349929714896e-07,
      "loss": 0.5788,
      "step": 2907
    },
    {
      "epoch": 0.8059866962305987,
      "grad_norm": 1.1161974668502808,
      "learning_rate": 9.028529522033502e-07,
      "loss": 0.5532,
      "step": 2908
    },
    {
      "epoch": 0.8062638580931264,
      "grad_norm": 0.8917732238769531,
      "learning_rate": 9.003590810262402e-07,
      "loss": 0.5619,
      "step": 2909
    },
    {
      "epoch": 0.8065410199556541,
      "grad_norm": 1.0807230472564697,
      "learning_rate": 8.978683180743481e-07,
      "loss": 0.6109,
      "step": 2910
    },
    {
      "epoch": 0.8068181818181818,
      "grad_norm": 0.975057065486908,
      "learning_rate": 8.953806652360953e-07,
      "loss": 0.675,
      "step": 2911
    },
    {
      "epoch": 0.8070953436807096,
      "grad_norm": 0.9096999168395996,
      "learning_rate": 8.928961243975437e-07,
      "loss": 0.5961,
      "step": 2912
    },
    {
      "epoch": 0.8073725055432373,
      "grad_norm": 0.8946964740753174,
      "learning_rate": 8.904146974423972e-07,
      "loss": 0.6067,
      "step": 2913
    },
    {
      "epoch": 0.8076496674057649,
      "grad_norm": 0.9662375450134277,
      "learning_rate": 8.87936386252003e-07,
      "loss": 0.5766,
      "step": 2914
    },
    {
      "epoch": 0.8079268292682927,
      "grad_norm": 0.8835971355438232,
      "learning_rate": 8.854611927053403e-07,
      "loss": 0.6563,
      "step": 2915
    },
    {
      "epoch": 0.8082039911308204,
      "grad_norm": 2.0184857845306396,
      "learning_rate": 8.82989118679028e-07,
      "loss": 0.6119,
      "step": 2916
    },
    {
      "epoch": 0.8084811529933481,
      "grad_norm": 0.9034411311149597,
      "learning_rate": 8.805201660473179e-07,
      "loss": 0.5834,
      "step": 2917
    },
    {
      "epoch": 0.8087583148558758,
      "grad_norm": 0.8922361135482788,
      "learning_rate": 8.780543366820965e-07,
      "loss": 0.6124,
      "step": 2918
    },
    {
      "epoch": 0.8090354767184036,
      "grad_norm": 0.9450308680534363,
      "learning_rate": 8.755916324528813e-07,
      "loss": 0.6371,
      "step": 2919
    },
    {
      "epoch": 0.8093126385809313,
      "grad_norm": 1.5803946256637573,
      "learning_rate": 8.731320552268208e-07,
      "loss": 0.6263,
      "step": 2920
    },
    {
      "epoch": 0.809589800443459,
      "grad_norm": 0.8892036080360413,
      "learning_rate": 8.70675606868695e-07,
      "loss": 0.5571,
      "step": 2921
    },
    {
      "epoch": 0.8098669623059866,
      "grad_norm": 0.9580626487731934,
      "learning_rate": 8.682222892409093e-07,
      "loss": 0.6823,
      "step": 2922
    },
    {
      "epoch": 0.8101441241685144,
      "grad_norm": 0.9158246517181396,
      "learning_rate": 8.657721042034934e-07,
      "loss": 0.5912,
      "step": 2923
    },
    {
      "epoch": 0.8104212860310421,
      "grad_norm": 1.082554578781128,
      "learning_rate": 8.633250536141047e-07,
      "loss": 0.5909,
      "step": 2924
    },
    {
      "epoch": 0.8106984478935698,
      "grad_norm": 0.972333550453186,
      "learning_rate": 8.608811393280264e-07,
      "loss": 0.6256,
      "step": 2925
    },
    {
      "epoch": 0.8109756097560976,
      "grad_norm": 0.9838047027587891,
      "learning_rate": 8.584403631981598e-07,
      "loss": 0.6342,
      "step": 2926
    },
    {
      "epoch": 0.8112527716186253,
      "grad_norm": 0.9203789830207825,
      "learning_rate": 8.560027270750276e-07,
      "loss": 0.5745,
      "step": 2927
    },
    {
      "epoch": 0.811529933481153,
      "grad_norm": 1.0449795722961426,
      "learning_rate": 8.53568232806774e-07,
      "loss": 0.6075,
      "step": 2928
    },
    {
      "epoch": 0.8118070953436807,
      "grad_norm": 0.8675444722175598,
      "learning_rate": 8.511368822391592e-07,
      "loss": 0.5969,
      "step": 2929
    },
    {
      "epoch": 0.8120842572062085,
      "grad_norm": 0.8838660717010498,
      "learning_rate": 8.487086772155606e-07,
      "loss": 0.5976,
      "step": 2930
    },
    {
      "epoch": 0.8123614190687362,
      "grad_norm": 0.8774744272232056,
      "learning_rate": 8.462836195769691e-07,
      "loss": 0.5727,
      "step": 2931
    },
    {
      "epoch": 0.8126385809312638,
      "grad_norm": 0.9328903555870056,
      "learning_rate": 8.43861711161994e-07,
      "loss": 0.6064,
      "step": 2932
    },
    {
      "epoch": 0.8129157427937915,
      "grad_norm": 0.8853145837783813,
      "learning_rate": 8.414429538068536e-07,
      "loss": 0.5976,
      "step": 2933
    },
    {
      "epoch": 0.8131929046563193,
      "grad_norm": 1.01247239112854,
      "learning_rate": 8.390273493453743e-07,
      "loss": 0.6057,
      "step": 2934
    },
    {
      "epoch": 0.813470066518847,
      "grad_norm": 0.9435156583786011,
      "learning_rate": 8.366148996089968e-07,
      "loss": 0.6025,
      "step": 2935
    },
    {
      "epoch": 0.8137472283813747,
      "grad_norm": 0.8783007264137268,
      "learning_rate": 8.34205606426769e-07,
      "loss": 0.5481,
      "step": 2936
    },
    {
      "epoch": 0.8140243902439024,
      "grad_norm": 0.9048199653625488,
      "learning_rate": 8.317994716253453e-07,
      "loss": 0.5873,
      "step": 2937
    },
    {
      "epoch": 0.8143015521064302,
      "grad_norm": 0.877376139163971,
      "learning_rate": 8.293964970289841e-07,
      "loss": 0.5736,
      "step": 2938
    },
    {
      "epoch": 0.8145787139689579,
      "grad_norm": 0.8848210573196411,
      "learning_rate": 8.269966844595495e-07,
      "loss": 0.623,
      "step": 2939
    },
    {
      "epoch": 0.8148558758314856,
      "grad_norm": 1.7820807695388794,
      "learning_rate": 8.246000357365075e-07,
      "loss": 0.5926,
      "step": 2940
    },
    {
      "epoch": 0.8151330376940134,
      "grad_norm": 0.939264178276062,
      "learning_rate": 8.222065526769257e-07,
      "loss": 0.6127,
      "step": 2941
    },
    {
      "epoch": 0.815410199556541,
      "grad_norm": 0.9397186040878296,
      "learning_rate": 8.198162370954698e-07,
      "loss": 0.6772,
      "step": 2942
    },
    {
      "epoch": 0.8156873614190687,
      "grad_norm": 0.9674282670021057,
      "learning_rate": 8.17429090804408e-07,
      "loss": 0.6018,
      "step": 2943
    },
    {
      "epoch": 0.8159645232815964,
      "grad_norm": 1.0511608123779297,
      "learning_rate": 8.150451156136041e-07,
      "loss": 0.6395,
      "step": 2944
    },
    {
      "epoch": 0.8162416851441242,
      "grad_norm": 0.9168019890785217,
      "learning_rate": 8.12664313330514e-07,
      "loss": 0.5656,
      "step": 2945
    },
    {
      "epoch": 0.8165188470066519,
      "grad_norm": 0.904909074306488,
      "learning_rate": 8.102866857601904e-07,
      "loss": 0.641,
      "step": 2946
    },
    {
      "epoch": 0.8167960088691796,
      "grad_norm": 0.929472029209137,
      "learning_rate": 8.079122347052826e-07,
      "loss": 0.5651,
      "step": 2947
    },
    {
      "epoch": 0.8170731707317073,
      "grad_norm": 0.908782422542572,
      "learning_rate": 8.055409619660265e-07,
      "loss": 0.6143,
      "step": 2948
    },
    {
      "epoch": 0.8173503325942351,
      "grad_norm": 0.9369438886642456,
      "learning_rate": 8.031728693402502e-07,
      "loss": 0.5829,
      "step": 2949
    },
    {
      "epoch": 0.8176274944567627,
      "grad_norm": 0.8760191798210144,
      "learning_rate": 8.008079586233719e-07,
      "loss": 0.5984,
      "step": 2950
    },
    {
      "epoch": 0.8179046563192904,
      "grad_norm": 0.8470495939254761,
      "learning_rate": 7.984462316083946e-07,
      "loss": 0.5566,
      "step": 2951
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 0.8832671046257019,
      "learning_rate": 7.960876900859104e-07,
      "loss": 0.5975,
      "step": 2952
    },
    {
      "epoch": 0.8184589800443459,
      "grad_norm": 0.8597487807273865,
      "learning_rate": 7.937323358440935e-07,
      "loss": 0.6262,
      "step": 2953
    },
    {
      "epoch": 0.8187361419068736,
      "grad_norm": 0.9326977729797363,
      "learning_rate": 7.913801706687052e-07,
      "loss": 0.5905,
      "step": 2954
    },
    {
      "epoch": 0.8190133037694013,
      "grad_norm": 0.9568464159965515,
      "learning_rate": 7.890311963430868e-07,
      "loss": 0.6271,
      "step": 2955
    },
    {
      "epoch": 0.8192904656319291,
      "grad_norm": 0.9429141879081726,
      "learning_rate": 7.866854146481583e-07,
      "loss": 0.6015,
      "step": 2956
    },
    {
      "epoch": 0.8195676274944568,
      "grad_norm": 0.8989124894142151,
      "learning_rate": 7.843428273624216e-07,
      "loss": 0.6404,
      "step": 2957
    },
    {
      "epoch": 0.8198447893569845,
      "grad_norm": 1.0577901601791382,
      "learning_rate": 7.820034362619577e-07,
      "loss": 0.5889,
      "step": 2958
    },
    {
      "epoch": 0.8201219512195121,
      "grad_norm": 1.1544501781463623,
      "learning_rate": 7.796672431204228e-07,
      "loss": 0.5745,
      "step": 2959
    },
    {
      "epoch": 0.8203991130820399,
      "grad_norm": 0.8843512535095215,
      "learning_rate": 7.773342497090486e-07,
      "loss": 0.5818,
      "step": 2960
    },
    {
      "epoch": 0.8206762749445676,
      "grad_norm": 0.889343798160553,
      "learning_rate": 7.750044577966404e-07,
      "loss": 0.5922,
      "step": 2961
    },
    {
      "epoch": 0.8209534368070953,
      "grad_norm": 1.032057285308838,
      "learning_rate": 7.726778691495773e-07,
      "loss": 0.612,
      "step": 2962
    },
    {
      "epoch": 0.8212305986696231,
      "grad_norm": 0.9290952682495117,
      "learning_rate": 7.70354485531809e-07,
      "loss": 0.6363,
      "step": 2963
    },
    {
      "epoch": 0.8215077605321508,
      "grad_norm": 0.8757408261299133,
      "learning_rate": 7.680343087048542e-07,
      "loss": 0.5911,
      "step": 2964
    },
    {
      "epoch": 0.8217849223946785,
      "grad_norm": 1.097544550895691,
      "learning_rate": 7.657173404278057e-07,
      "loss": 0.6069,
      "step": 2965
    },
    {
      "epoch": 0.8220620842572062,
      "grad_norm": 0.8809067010879517,
      "learning_rate": 7.634035824573161e-07,
      "loss": 0.5659,
      "step": 2966
    },
    {
      "epoch": 0.822339246119734,
      "grad_norm": 0.9556673169136047,
      "learning_rate": 7.610930365476077e-07,
      "loss": 0.6171,
      "step": 2967
    },
    {
      "epoch": 0.8226164079822617,
      "grad_norm": 1.2251789569854736,
      "learning_rate": 7.587857044504676e-07,
      "loss": 0.5548,
      "step": 2968
    },
    {
      "epoch": 0.8228935698447893,
      "grad_norm": 0.8849217295646667,
      "learning_rate": 7.56481587915247e-07,
      "loss": 0.5683,
      "step": 2969
    },
    {
      "epoch": 0.823170731707317,
      "grad_norm": 1.3557699918746948,
      "learning_rate": 7.541806886888581e-07,
      "loss": 0.5709,
      "step": 2970
    },
    {
      "epoch": 0.8234478935698448,
      "grad_norm": 0.973290741443634,
      "learning_rate": 7.518830085157735e-07,
      "loss": 0.6626,
      "step": 2971
    },
    {
      "epoch": 0.8237250554323725,
      "grad_norm": 0.934116005897522,
      "learning_rate": 7.495885491380256e-07,
      "loss": 0.6021,
      "step": 2972
    },
    {
      "epoch": 0.8240022172949002,
      "grad_norm": 0.9660149216651917,
      "learning_rate": 7.472973122952049e-07,
      "loss": 0.5896,
      "step": 2973
    },
    {
      "epoch": 0.8242793791574279,
      "grad_norm": 1.1267499923706055,
      "learning_rate": 7.45009299724459e-07,
      "loss": 0.5836,
      "step": 2974
    },
    {
      "epoch": 0.8245565410199557,
      "grad_norm": 0.9896127581596375,
      "learning_rate": 7.427245131604899e-07,
      "loss": 0.6175,
      "step": 2975
    },
    {
      "epoch": 0.8248337028824834,
      "grad_norm": 0.9330118298530579,
      "learning_rate": 7.404429543355557e-07,
      "loss": 0.6574,
      "step": 2976
    },
    {
      "epoch": 0.825110864745011,
      "grad_norm": 0.9217411279678345,
      "learning_rate": 7.381646249794649e-07,
      "loss": 0.6043,
      "step": 2977
    },
    {
      "epoch": 0.8253880266075388,
      "grad_norm": 0.9388583898544312,
      "learning_rate": 7.358895268195793e-07,
      "loss": 0.6219,
      "step": 2978
    },
    {
      "epoch": 0.8256651884700665,
      "grad_norm": 1.1909815073013306,
      "learning_rate": 7.336176615808094e-07,
      "loss": 0.6107,
      "step": 2979
    },
    {
      "epoch": 0.8259423503325942,
      "grad_norm": 0.9236672520637512,
      "learning_rate": 7.313490309856169e-07,
      "loss": 0.6074,
      "step": 2980
    },
    {
      "epoch": 0.8262195121951219,
      "grad_norm": 1.0435806512832642,
      "learning_rate": 7.2908363675401e-07,
      "loss": 0.593,
      "step": 2981
    },
    {
      "epoch": 0.8264966740576497,
      "grad_norm": 0.9081978797912598,
      "learning_rate": 7.268214806035423e-07,
      "loss": 0.5809,
      "step": 2982
    },
    {
      "epoch": 0.8267738359201774,
      "grad_norm": 0.8854150176048279,
      "learning_rate": 7.245625642493115e-07,
      "loss": 0.5561,
      "step": 2983
    },
    {
      "epoch": 0.8270509977827051,
      "grad_norm": 0.9030740857124329,
      "learning_rate": 7.22306889403962e-07,
      "loss": 0.573,
      "step": 2984
    },
    {
      "epoch": 0.8273281596452328,
      "grad_norm": 0.8904342651367188,
      "learning_rate": 7.200544577776791e-07,
      "loss": 0.6064,
      "step": 2985
    },
    {
      "epoch": 0.8276053215077606,
      "grad_norm": 1.0130550861358643,
      "learning_rate": 7.178052710781891e-07,
      "loss": 0.6289,
      "step": 2986
    },
    {
      "epoch": 0.8278824833702882,
      "grad_norm": 1.1549328565597534,
      "learning_rate": 7.155593310107578e-07,
      "loss": 0.6388,
      "step": 2987
    },
    {
      "epoch": 0.8281596452328159,
      "grad_norm": 0.900002121925354,
      "learning_rate": 7.133166392781899e-07,
      "loss": 0.5525,
      "step": 2988
    },
    {
      "epoch": 0.8284368070953437,
      "grad_norm": 0.8821041584014893,
      "learning_rate": 7.110771975808267e-07,
      "loss": 0.5657,
      "step": 2989
    },
    {
      "epoch": 0.8287139689578714,
      "grad_norm": 1.0247434377670288,
      "learning_rate": 7.088410076165458e-07,
      "loss": 0.647,
      "step": 2990
    },
    {
      "epoch": 0.8289911308203991,
      "grad_norm": 0.8705650568008423,
      "learning_rate": 7.066080710807616e-07,
      "loss": 0.5814,
      "step": 2991
    },
    {
      "epoch": 0.8292682926829268,
      "grad_norm": 1.0124642848968506,
      "learning_rate": 7.043783896664186e-07,
      "loss": 0.6603,
      "step": 2992
    },
    {
      "epoch": 0.8295454545454546,
      "grad_norm": 0.8919931650161743,
      "learning_rate": 7.021519650639952e-07,
      "loss": 0.6011,
      "step": 2993
    },
    {
      "epoch": 0.8298226164079823,
      "grad_norm": 0.9636069536209106,
      "learning_rate": 6.999287989614972e-07,
      "loss": 0.594,
      "step": 2994
    },
    {
      "epoch": 0.83009977827051,
      "grad_norm": 0.8765805959701538,
      "learning_rate": 6.977088930444665e-07,
      "loss": 0.5589,
      "step": 2995
    },
    {
      "epoch": 0.8303769401330376,
      "grad_norm": 0.9367631077766418,
      "learning_rate": 6.954922489959671e-07,
      "loss": 0.6149,
      "step": 2996
    },
    {
      "epoch": 0.8306541019955654,
      "grad_norm": 2.0685477256774902,
      "learning_rate": 6.932788684965935e-07,
      "loss": 0.6082,
      "step": 2997
    },
    {
      "epoch": 0.8309312638580931,
      "grad_norm": 0.9323012828826904,
      "learning_rate": 6.910687532244636e-07,
      "loss": 0.6326,
      "step": 2998
    },
    {
      "epoch": 0.8312084257206208,
      "grad_norm": 0.9778831005096436,
      "learning_rate": 6.888619048552214e-07,
      "loss": 0.6173,
      "step": 2999
    },
    {
      "epoch": 0.8314855875831486,
      "grad_norm": 0.9090682864189148,
      "learning_rate": 6.866583250620335e-07,
      "loss": 0.6057,
      "step": 3000
    },
    {
      "epoch": 0.8317627494456763,
      "grad_norm": 1.6415265798568726,
      "learning_rate": 6.844580155155861e-07,
      "loss": 0.5706,
      "step": 3001
    },
    {
      "epoch": 0.832039911308204,
      "grad_norm": 1.3576527833938599,
      "learning_rate": 6.822609778840916e-07,
      "loss": 0.5993,
      "step": 3002
    },
    {
      "epoch": 0.8323170731707317,
      "grad_norm": 0.9539833068847656,
      "learning_rate": 6.800672138332764e-07,
      "loss": 0.5913,
      "step": 3003
    },
    {
      "epoch": 0.8325942350332595,
      "grad_norm": 0.8659795522689819,
      "learning_rate": 6.778767250263884e-07,
      "loss": 0.5834,
      "step": 3004
    },
    {
      "epoch": 0.8328713968957872,
      "grad_norm": 0.8646303415298462,
      "learning_rate": 6.756895131241869e-07,
      "loss": 0.5618,
      "step": 3005
    },
    {
      "epoch": 0.8331485587583148,
      "grad_norm": 0.8914692401885986,
      "learning_rate": 6.735055797849543e-07,
      "loss": 0.5902,
      "step": 3006
    },
    {
      "epoch": 0.8334257206208425,
      "grad_norm": 0.8860941529273987,
      "learning_rate": 6.713249266644828e-07,
      "loss": 0.6138,
      "step": 3007
    },
    {
      "epoch": 0.8337028824833703,
      "grad_norm": 0.8851166367530823,
      "learning_rate": 6.691475554160775e-07,
      "loss": 0.6034,
      "step": 3008
    },
    {
      "epoch": 0.833980044345898,
      "grad_norm": 0.9062486886978149,
      "learning_rate": 6.669734676905571e-07,
      "loss": 0.5986,
      "step": 3009
    },
    {
      "epoch": 0.8342572062084257,
      "grad_norm": 0.9255950450897217,
      "learning_rate": 6.648026651362489e-07,
      "loss": 0.6135,
      "step": 3010
    },
    {
      "epoch": 0.8345343680709535,
      "grad_norm": 0.9378454089164734,
      "learning_rate": 6.626351493989918e-07,
      "loss": 0.5947,
      "step": 3011
    },
    {
      "epoch": 0.8348115299334812,
      "grad_norm": 0.8815849423408508,
      "learning_rate": 6.604709221221295e-07,
      "loss": 0.5886,
      "step": 3012
    },
    {
      "epoch": 0.8350886917960089,
      "grad_norm": 1.0878963470458984,
      "learning_rate": 6.583099849465169e-07,
      "loss": 0.6189,
      "step": 3013
    },
    {
      "epoch": 0.8353658536585366,
      "grad_norm": 0.940568745136261,
      "learning_rate": 6.561523395105108e-07,
      "loss": 0.6452,
      "step": 3014
    },
    {
      "epoch": 0.8356430155210643,
      "grad_norm": 0.9190086722373962,
      "learning_rate": 6.539979874499747e-07,
      "loss": 0.625,
      "step": 3015
    },
    {
      "epoch": 0.835920177383592,
      "grad_norm": 0.8718287944793701,
      "learning_rate": 6.518469303982705e-07,
      "loss": 0.6068,
      "step": 3016
    },
    {
      "epoch": 0.8361973392461197,
      "grad_norm": 0.9296078085899353,
      "learning_rate": 6.496991699862687e-07,
      "loss": 0.5864,
      "step": 3017
    },
    {
      "epoch": 0.8364745011086474,
      "grad_norm": 0.8663907051086426,
      "learning_rate": 6.475547078423361e-07,
      "loss": 0.5731,
      "step": 3018
    },
    {
      "epoch": 0.8367516629711752,
      "grad_norm": 0.9033164978027344,
      "learning_rate": 6.454135455923394e-07,
      "loss": 0.599,
      "step": 3019
    },
    {
      "epoch": 0.8370288248337029,
      "grad_norm": 0.9046875834465027,
      "learning_rate": 6.432756848596445e-07,
      "loss": 0.5759,
      "step": 3020
    },
    {
      "epoch": 0.8373059866962306,
      "grad_norm": 0.9035504460334778,
      "learning_rate": 6.41141127265113e-07,
      "loss": 0.5878,
      "step": 3021
    },
    {
      "epoch": 0.8375831485587583,
      "grad_norm": 0.9053865671157837,
      "learning_rate": 6.390098744271028e-07,
      "loss": 0.5758,
      "step": 3022
    },
    {
      "epoch": 0.8378603104212861,
      "grad_norm": 1.3063606023788452,
      "learning_rate": 6.368819279614658e-07,
      "loss": 0.6401,
      "step": 3023
    },
    {
      "epoch": 0.8381374722838137,
      "grad_norm": 0.9161602258682251,
      "learning_rate": 6.347572894815468e-07,
      "loss": 0.5974,
      "step": 3024
    },
    {
      "epoch": 0.8384146341463414,
      "grad_norm": 0.9932770729064941,
      "learning_rate": 6.326359605981863e-07,
      "loss": 0.6094,
      "step": 3025
    },
    {
      "epoch": 0.8386917960088692,
      "grad_norm": 0.8887829184532166,
      "learning_rate": 6.305179429197094e-07,
      "loss": 0.5798,
      "step": 3026
    },
    {
      "epoch": 0.8389689578713969,
      "grad_norm": 1.0603339672088623,
      "learning_rate": 6.284032380519334e-07,
      "loss": 0.5429,
      "step": 3027
    },
    {
      "epoch": 0.8392461197339246,
      "grad_norm": 0.9347849488258362,
      "learning_rate": 6.262918475981666e-07,
      "loss": 0.5887,
      "step": 3028
    },
    {
      "epoch": 0.8395232815964523,
      "grad_norm": 0.9598890542984009,
      "learning_rate": 6.24183773159201e-07,
      "loss": 0.5817,
      "step": 3029
    },
    {
      "epoch": 0.8398004434589801,
      "grad_norm": 0.9092518091201782,
      "learning_rate": 6.220790163333162e-07,
      "loss": 0.5702,
      "step": 3030
    },
    {
      "epoch": 0.8400776053215078,
      "grad_norm": 0.9842642545700073,
      "learning_rate": 6.199775787162749e-07,
      "loss": 0.6257,
      "step": 3031
    },
    {
      "epoch": 0.8403547671840355,
      "grad_norm": 0.9325064420700073,
      "learning_rate": 6.178794619013251e-07,
      "loss": 0.5906,
      "step": 3032
    },
    {
      "epoch": 0.8406319290465631,
      "grad_norm": 0.8709660172462463,
      "learning_rate": 6.157846674791961e-07,
      "loss": 0.588,
      "step": 3033
    },
    {
      "epoch": 0.8409090909090909,
      "grad_norm": 0.8986772894859314,
      "learning_rate": 6.136931970380988e-07,
      "loss": 0.5855,
      "step": 3034
    },
    {
      "epoch": 0.8411862527716186,
      "grad_norm": 0.9421331286430359,
      "learning_rate": 6.116050521637218e-07,
      "loss": 0.6143,
      "step": 3035
    },
    {
      "epoch": 0.8414634146341463,
      "grad_norm": 1.028441071510315,
      "learning_rate": 6.095202344392371e-07,
      "loss": 0.5379,
      "step": 3036
    },
    {
      "epoch": 0.8417405764966741,
      "grad_norm": 0.8975226283073425,
      "learning_rate": 6.074387454452891e-07,
      "loss": 0.676,
      "step": 3037
    },
    {
      "epoch": 0.8420177383592018,
      "grad_norm": 1.015478253364563,
      "learning_rate": 6.053605867599999e-07,
      "loss": 0.6072,
      "step": 3038
    },
    {
      "epoch": 0.8422949002217295,
      "grad_norm": 0.9013005495071411,
      "learning_rate": 6.032857599589692e-07,
      "loss": 0.6049,
      "step": 3039
    },
    {
      "epoch": 0.8425720620842572,
      "grad_norm": 0.888767659664154,
      "learning_rate": 6.012142666152676e-07,
      "loss": 0.6129,
      "step": 3040
    },
    {
      "epoch": 0.842849223946785,
      "grad_norm": 0.9180863499641418,
      "learning_rate": 5.991461082994404e-07,
      "loss": 0.5872,
      "step": 3041
    },
    {
      "epoch": 0.8431263858093127,
      "grad_norm": 0.915776252746582,
      "learning_rate": 5.970812865795023e-07,
      "loss": 0.5962,
      "step": 3042
    },
    {
      "epoch": 0.8434035476718403,
      "grad_norm": 0.8733999729156494,
      "learning_rate": 5.950198030209398e-07,
      "loss": 0.5536,
      "step": 3043
    },
    {
      "epoch": 0.843680709534368,
      "grad_norm": 1.6952321529388428,
      "learning_rate": 5.929616591867082e-07,
      "loss": 0.5875,
      "step": 3044
    },
    {
      "epoch": 0.8439578713968958,
      "grad_norm": 0.9307530522346497,
      "learning_rate": 5.909068566372312e-07,
      "loss": 0.6574,
      "step": 3045
    },
    {
      "epoch": 0.8442350332594235,
      "grad_norm": 2.1602160930633545,
      "learning_rate": 5.888553969303967e-07,
      "loss": 0.6252,
      "step": 3046
    },
    {
      "epoch": 0.8445121951219512,
      "grad_norm": 0.9739928245544434,
      "learning_rate": 5.868072816215631e-07,
      "loss": 0.5886,
      "step": 3047
    },
    {
      "epoch": 0.844789356984479,
      "grad_norm": 0.9440780282020569,
      "learning_rate": 5.847625122635481e-07,
      "loss": 0.6097,
      "step": 3048
    },
    {
      "epoch": 0.8450665188470067,
      "grad_norm": 0.9449934959411621,
      "learning_rate": 5.827210904066344e-07,
      "loss": 0.5739,
      "step": 3049
    },
    {
      "epoch": 0.8453436807095344,
      "grad_norm": 0.8726100325584412,
      "learning_rate": 5.80683017598569e-07,
      "loss": 0.6268,
      "step": 3050
    },
    {
      "epoch": 0.845620842572062,
      "grad_norm": 0.8874340653419495,
      "learning_rate": 5.786482953845563e-07,
      "loss": 0.5955,
      "step": 3051
    },
    {
      "epoch": 0.8458980044345898,
      "grad_norm": 0.8904621601104736,
      "learning_rate": 5.766169253072624e-07,
      "loss": 0.5641,
      "step": 3052
    },
    {
      "epoch": 0.8461751662971175,
      "grad_norm": 0.8842699527740479,
      "learning_rate": 5.745889089068113e-07,
      "loss": 0.6108,
      "step": 3053
    },
    {
      "epoch": 0.8464523281596452,
      "grad_norm": 0.8713987469673157,
      "learning_rate": 5.725642477207838e-07,
      "loss": 0.5946,
      "step": 3054
    },
    {
      "epoch": 0.8467294900221729,
      "grad_norm": 0.9701282382011414,
      "learning_rate": 5.705429432842185e-07,
      "loss": 0.6338,
      "step": 3055
    },
    {
      "epoch": 0.8470066518847007,
      "grad_norm": 0.9430311322212219,
      "learning_rate": 5.685249971296081e-07,
      "loss": 0.6143,
      "step": 3056
    },
    {
      "epoch": 0.8472838137472284,
      "grad_norm": 0.8861961364746094,
      "learning_rate": 5.665104107868969e-07,
      "loss": 0.5657,
      "step": 3057
    },
    {
      "epoch": 0.8475609756097561,
      "grad_norm": 0.9051007628440857,
      "learning_rate": 5.644991857834881e-07,
      "loss": 0.5907,
      "step": 3058
    },
    {
      "epoch": 0.8478381374722838,
      "grad_norm": 1.055182695388794,
      "learning_rate": 5.624913236442287e-07,
      "loss": 0.5934,
      "step": 3059
    },
    {
      "epoch": 0.8481152993348116,
      "grad_norm": 0.9025108218193054,
      "learning_rate": 5.604868258914209e-07,
      "loss": 0.6264,
      "step": 3060
    },
    {
      "epoch": 0.8483924611973392,
      "grad_norm": 0.9061220288276672,
      "learning_rate": 5.584856940448141e-07,
      "loss": 0.5809,
      "step": 3061
    },
    {
      "epoch": 0.8486696230598669,
      "grad_norm": 0.8891677856445312,
      "learning_rate": 5.564879296216091e-07,
      "loss": 0.5819,
      "step": 3062
    },
    {
      "epoch": 0.8489467849223947,
      "grad_norm": 0.9663417935371399,
      "learning_rate": 5.544935341364494e-07,
      "loss": 0.6109,
      "step": 3063
    },
    {
      "epoch": 0.8492239467849224,
      "grad_norm": 2.387605667114258,
      "learning_rate": 5.525025091014269e-07,
      "loss": 0.5998,
      "step": 3064
    },
    {
      "epoch": 0.8495011086474501,
      "grad_norm": 2.560267686843872,
      "learning_rate": 5.505148560260765e-07,
      "loss": 0.5505,
      "step": 3065
    },
    {
      "epoch": 0.8497782705099778,
      "grad_norm": 0.9106470346450806,
      "learning_rate": 5.485305764173776e-07,
      "loss": 0.6154,
      "step": 3066
    },
    {
      "epoch": 0.8500554323725056,
      "grad_norm": 0.8921628594398499,
      "learning_rate": 5.465496717797514e-07,
      "loss": 0.5434,
      "step": 3067
    },
    {
      "epoch": 0.8503325942350333,
      "grad_norm": 0.9205003380775452,
      "learning_rate": 5.445721436150614e-07,
      "loss": 0.5929,
      "step": 3068
    },
    {
      "epoch": 0.850609756097561,
      "grad_norm": 0.9085372686386108,
      "learning_rate": 5.4259799342261e-07,
      "loss": 0.6052,
      "step": 3069
    },
    {
      "epoch": 0.8508869179600886,
      "grad_norm": 0.9360840320587158,
      "learning_rate": 5.406272226991383e-07,
      "loss": 0.6177,
      "step": 3070
    },
    {
      "epoch": 0.8511640798226164,
      "grad_norm": 0.8828532099723816,
      "learning_rate": 5.386598329388265e-07,
      "loss": 0.6009,
      "step": 3071
    },
    {
      "epoch": 0.8514412416851441,
      "grad_norm": 0.8840983510017395,
      "learning_rate": 5.366958256332893e-07,
      "loss": 0.5965,
      "step": 3072
    },
    {
      "epoch": 0.8517184035476718,
      "grad_norm": 0.8991380333900452,
      "learning_rate": 5.347352022715802e-07,
      "loss": 0.5719,
      "step": 3073
    },
    {
      "epoch": 0.8519955654101996,
      "grad_norm": 0.8952696919441223,
      "learning_rate": 5.327779643401849e-07,
      "loss": 0.6889,
      "step": 3074
    },
    {
      "epoch": 0.8522727272727273,
      "grad_norm": 0.8828235268592834,
      "learning_rate": 5.308241133230224e-07,
      "loss": 0.6188,
      "step": 3075
    },
    {
      "epoch": 0.852549889135255,
      "grad_norm": 0.8854100704193115,
      "learning_rate": 5.288736507014436e-07,
      "loss": 0.5669,
      "step": 3076
    },
    {
      "epoch": 0.8528270509977827,
      "grad_norm": 0.9610351324081421,
      "learning_rate": 5.269265779542321e-07,
      "loss": 0.606,
      "step": 3077
    },
    {
      "epoch": 0.8531042128603105,
      "grad_norm": 0.9233687520027161,
      "learning_rate": 5.249828965575987e-07,
      "loss": 0.5807,
      "step": 3078
    },
    {
      "epoch": 0.8533813747228381,
      "grad_norm": 1.0030553340911865,
      "learning_rate": 5.230426079851858e-07,
      "loss": 0.577,
      "step": 3079
    },
    {
      "epoch": 0.8536585365853658,
      "grad_norm": 0.8985577821731567,
      "learning_rate": 5.211057137080622e-07,
      "loss": 0.6396,
      "step": 3080
    },
    {
      "epoch": 0.8539356984478935,
      "grad_norm": 0.8572052717208862,
      "learning_rate": 5.191722151947227e-07,
      "loss": 0.5491,
      "step": 3081
    },
    {
      "epoch": 0.8542128603104213,
      "grad_norm": 0.9241860508918762,
      "learning_rate": 5.172421139110878e-07,
      "loss": 0.6265,
      "step": 3082
    },
    {
      "epoch": 0.854490022172949,
      "grad_norm": 0.8954199552536011,
      "learning_rate": 5.153154113205028e-07,
      "loss": 0.6329,
      "step": 3083
    },
    {
      "epoch": 0.8547671840354767,
      "grad_norm": 0.8850727081298828,
      "learning_rate": 5.133921088837368e-07,
      "loss": 0.5841,
      "step": 3084
    },
    {
      "epoch": 0.8550443458980045,
      "grad_norm": 0.8948554396629333,
      "learning_rate": 5.114722080589807e-07,
      "loss": 0.6464,
      "step": 3085
    },
    {
      "epoch": 0.8553215077605322,
      "grad_norm": 0.8707226514816284,
      "learning_rate": 5.095557103018434e-07,
      "loss": 0.5835,
      "step": 3086
    },
    {
      "epoch": 0.8555986696230599,
      "grad_norm": 0.9451014995574951,
      "learning_rate": 5.076426170653581e-07,
      "loss": 0.6275,
      "step": 3087
    },
    {
      "epoch": 0.8558758314855875,
      "grad_norm": 0.8599836826324463,
      "learning_rate": 5.057329297999747e-07,
      "loss": 0.5926,
      "step": 3088
    },
    {
      "epoch": 0.8561529933481153,
      "grad_norm": 0.8965653777122498,
      "learning_rate": 5.03826649953561e-07,
      "loss": 0.5558,
      "step": 3089
    },
    {
      "epoch": 0.856430155210643,
      "grad_norm": 0.8715825080871582,
      "learning_rate": 5.019237789714015e-07,
      "loss": 0.6164,
      "step": 3090
    },
    {
      "epoch": 0.8567073170731707,
      "grad_norm": 0.9669959545135498,
      "learning_rate": 5.000243182961956e-07,
      "loss": 0.6142,
      "step": 3091
    },
    {
      "epoch": 0.8569844789356984,
      "grad_norm": 0.9049342274665833,
      "learning_rate": 4.981282693680584e-07,
      "loss": 0.5826,
      "step": 3092
    },
    {
      "epoch": 0.8572616407982262,
      "grad_norm": 0.998245120048523,
      "learning_rate": 4.962356336245166e-07,
      "loss": 0.6241,
      "step": 3093
    },
    {
      "epoch": 0.8575388026607539,
      "grad_norm": 1.2676723003387451,
      "learning_rate": 4.943464125005093e-07,
      "loss": 0.5901,
      "step": 3094
    },
    {
      "epoch": 0.8578159645232816,
      "grad_norm": 0.9595057368278503,
      "learning_rate": 4.9246060742839e-07,
      "loss": 0.623,
      "step": 3095
    },
    {
      "epoch": 0.8580931263858093,
      "grad_norm": 0.8981519937515259,
      "learning_rate": 4.905782198379194e-07,
      "loss": 0.5777,
      "step": 3096
    },
    {
      "epoch": 0.858370288248337,
      "grad_norm": 0.8928995132446289,
      "learning_rate": 4.886992511562655e-07,
      "loss": 0.5795,
      "step": 3097
    },
    {
      "epoch": 0.8586474501108647,
      "grad_norm": 0.9994025826454163,
      "learning_rate": 4.868237028080064e-07,
      "loss": 0.6444,
      "step": 3098
    },
    {
      "epoch": 0.8589246119733924,
      "grad_norm": 0.932578444480896,
      "learning_rate": 4.849515762151286e-07,
      "loss": 0.635,
      "step": 3099
    },
    {
      "epoch": 0.8592017738359202,
      "grad_norm": 0.8810226321220398,
      "learning_rate": 4.830828727970216e-07,
      "loss": 0.5907,
      "step": 3100
    },
    {
      "epoch": 0.8594789356984479,
      "grad_norm": 0.8843953609466553,
      "learning_rate": 4.8121759397048e-07,
      "loss": 0.5594,
      "step": 3101
    },
    {
      "epoch": 0.8597560975609756,
      "grad_norm": 0.9085744619369507,
      "learning_rate": 4.79355741149703e-07,
      "loss": 0.576,
      "step": 3102
    },
    {
      "epoch": 0.8600332594235033,
      "grad_norm": 0.8972387313842773,
      "learning_rate": 4.77497315746292e-07,
      "loss": 0.5575,
      "step": 3103
    },
    {
      "epoch": 0.8603104212860311,
      "grad_norm": 0.9024810791015625,
      "learning_rate": 4.756423191692494e-07,
      "loss": 0.5728,
      "step": 3104
    },
    {
      "epoch": 0.8605875831485588,
      "grad_norm": 0.8609679341316223,
      "learning_rate": 4.7379075282497635e-07,
      "loss": 0.536,
      "step": 3105
    },
    {
      "epoch": 0.8608647450110865,
      "grad_norm": 0.882659912109375,
      "learning_rate": 4.7194261811727817e-07,
      "loss": 0.6002,
      "step": 3106
    },
    {
      "epoch": 0.8611419068736141,
      "grad_norm": 0.9215378761291504,
      "learning_rate": 4.700979164473551e-07,
      "loss": 0.5848,
      "step": 3107
    },
    {
      "epoch": 0.8614190687361419,
      "grad_norm": 0.902506411075592,
      "learning_rate": 4.6825664921380245e-07,
      "loss": 0.6615,
      "step": 3108
    },
    {
      "epoch": 0.8616962305986696,
      "grad_norm": 0.8506453633308411,
      "learning_rate": 4.6641881781261447e-07,
      "loss": 0.5592,
      "step": 3109
    },
    {
      "epoch": 0.8619733924611973,
      "grad_norm": 0.9339303374290466,
      "learning_rate": 4.645844236371816e-07,
      "loss": 0.6067,
      "step": 3110
    },
    {
      "epoch": 0.8622505543237251,
      "grad_norm": 0.9149566292762756,
      "learning_rate": 4.627534680782858e-07,
      "loss": 0.5688,
      "step": 3111
    },
    {
      "epoch": 0.8625277161862528,
      "grad_norm": 0.8643877506256104,
      "learning_rate": 4.6092595252410335e-07,
      "loss": 0.6516,
      "step": 3112
    },
    {
      "epoch": 0.8628048780487805,
      "grad_norm": 0.9572612643241882,
      "learning_rate": 4.591018783602008e-07,
      "loss": 0.602,
      "step": 3113
    },
    {
      "epoch": 0.8630820399113082,
      "grad_norm": 0.9675074219703674,
      "learning_rate": 4.5728124696953815e-07,
      "loss": 0.6354,
      "step": 3114
    },
    {
      "epoch": 0.863359201773836,
      "grad_norm": 0.882343053817749,
      "learning_rate": 4.554640597324622e-07,
      "loss": 0.6413,
      "step": 3115
    },
    {
      "epoch": 0.8636363636363636,
      "grad_norm": 1.1098073720932007,
      "learning_rate": 4.536503180267099e-07,
      "loss": 0.6423,
      "step": 3116
    },
    {
      "epoch": 0.8639135254988913,
      "grad_norm": 2.582703113555908,
      "learning_rate": 4.5184002322740784e-07,
      "loss": 0.5962,
      "step": 3117
    },
    {
      "epoch": 0.864190687361419,
      "grad_norm": 1.0651459693908691,
      "learning_rate": 4.500331767070676e-07,
      "loss": 0.5469,
      "step": 3118
    },
    {
      "epoch": 0.8644678492239468,
      "grad_norm": 0.9216436147689819,
      "learning_rate": 4.4822977983558337e-07,
      "loss": 0.5461,
      "step": 3119
    },
    {
      "epoch": 0.8647450110864745,
      "grad_norm": 0.8736562132835388,
      "learning_rate": 4.464298339802381e-07,
      "loss": 0.5895,
      "step": 3120
    },
    {
      "epoch": 0.8650221729490022,
      "grad_norm": 0.8831925988197327,
      "learning_rate": 4.4463334050569796e-07,
      "loss": 0.5927,
      "step": 3121
    },
    {
      "epoch": 0.86529933481153,
      "grad_norm": 0.943334698677063,
      "learning_rate": 4.4284030077400954e-07,
      "loss": 0.616,
      "step": 3122
    },
    {
      "epoch": 0.8655764966740577,
      "grad_norm": 0.924526572227478,
      "learning_rate": 4.410507161446026e-07,
      "loss": 0.6075,
      "step": 3123
    },
    {
      "epoch": 0.8658536585365854,
      "grad_norm": 1.094819188117981,
      "learning_rate": 4.3926458797428683e-07,
      "loss": 0.6336,
      "step": 3124
    },
    {
      "epoch": 0.866130820399113,
      "grad_norm": 0.9040202498435974,
      "learning_rate": 4.374819176172501e-07,
      "loss": 0.5837,
      "step": 3125
    },
    {
      "epoch": 0.8664079822616408,
      "grad_norm": 0.9692425727844238,
      "learning_rate": 4.357027064250613e-07,
      "loss": 0.5885,
      "step": 3126
    },
    {
      "epoch": 0.8666851441241685,
      "grad_norm": 1.0382074117660522,
      "learning_rate": 4.3392695574666297e-07,
      "loss": 0.6153,
      "step": 3127
    },
    {
      "epoch": 0.8669623059866962,
      "grad_norm": 0.9221178889274597,
      "learning_rate": 4.321546669283794e-07,
      "loss": 0.5957,
      "step": 3128
    },
    {
      "epoch": 0.8672394678492239,
      "grad_norm": 0.9339891076087952,
      "learning_rate": 4.3038584131390446e-07,
      "loss": 0.5875,
      "step": 3129
    },
    {
      "epoch": 0.8675166297117517,
      "grad_norm": 0.9576802253723145,
      "learning_rate": 4.2862048024431e-07,
      "loss": 0.6536,
      "step": 3130
    },
    {
      "epoch": 0.8677937915742794,
      "grad_norm": 0.8980457186698914,
      "learning_rate": 4.268585850580381e-07,
      "loss": 0.6217,
      "step": 3131
    },
    {
      "epoch": 0.8680709534368071,
      "grad_norm": 0.9049385786056519,
      "learning_rate": 4.25100157090908e-07,
      "loss": 0.5502,
      "step": 3132
    },
    {
      "epoch": 0.8683481152993349,
      "grad_norm": 0.8616086840629578,
      "learning_rate": 4.233451976761055e-07,
      "loss": 0.5409,
      "step": 3133
    },
    {
      "epoch": 0.8686252771618626,
      "grad_norm": 0.920098602771759,
      "learning_rate": 4.21593708144189e-07,
      "loss": 0.5916,
      "step": 3134
    },
    {
      "epoch": 0.8689024390243902,
      "grad_norm": 0.9143387079238892,
      "learning_rate": 4.198456898230852e-07,
      "loss": 0.5956,
      "step": 3135
    },
    {
      "epoch": 0.8691796008869179,
      "grad_norm": 0.9339837431907654,
      "learning_rate": 4.181011440380889e-07,
      "loss": 0.5895,
      "step": 3136
    },
    {
      "epoch": 0.8694567627494457,
      "grad_norm": 0.8980045318603516,
      "learning_rate": 4.163600721118638e-07,
      "loss": 0.6599,
      "step": 3137
    },
    {
      "epoch": 0.8697339246119734,
      "grad_norm": 0.8912875056266785,
      "learning_rate": 4.146224753644362e-07,
      "loss": 0.5665,
      "step": 3138
    },
    {
      "epoch": 0.8700110864745011,
      "grad_norm": 0.9057996273040771,
      "learning_rate": 4.12888355113204e-07,
      "loss": 0.6271,
      "step": 3139
    },
    {
      "epoch": 0.8702882483370288,
      "grad_norm": 0.9385129809379578,
      "learning_rate": 4.111577126729216e-07,
      "loss": 0.6747,
      "step": 3140
    },
    {
      "epoch": 0.8705654101995566,
      "grad_norm": 0.9020197987556458,
      "learning_rate": 4.0943054935571214e-07,
      "loss": 0.6201,
      "step": 3141
    },
    {
      "epoch": 0.8708425720620843,
      "grad_norm": 0.89072185754776,
      "learning_rate": 4.0770686647105816e-07,
      "loss": 0.5751,
      "step": 3142
    },
    {
      "epoch": 0.871119733924612,
      "grad_norm": 0.9446536302566528,
      "learning_rate": 4.059866653258054e-07,
      "loss": 0.6306,
      "step": 3143
    },
    {
      "epoch": 0.8713968957871396,
      "grad_norm": 0.9637663960456848,
      "learning_rate": 4.0426994722415957e-07,
      "loss": 0.6249,
      "step": 3144
    },
    {
      "epoch": 0.8716740576496674,
      "grad_norm": 0.9755100607872009,
      "learning_rate": 4.0255671346768444e-07,
      "loss": 0.6187,
      "step": 3145
    },
    {
      "epoch": 0.8719512195121951,
      "grad_norm": 0.948525607585907,
      "learning_rate": 4.00846965355301e-07,
      "loss": 0.6638,
      "step": 3146
    },
    {
      "epoch": 0.8722283813747228,
      "grad_norm": 0.864585280418396,
      "learning_rate": 3.9914070418329123e-07,
      "loss": 0.6041,
      "step": 3147
    },
    {
      "epoch": 0.8725055432372506,
      "grad_norm": 1.1280852556228638,
      "learning_rate": 3.974379312452914e-07,
      "loss": 0.5611,
      "step": 3148
    },
    {
      "epoch": 0.8727827050997783,
      "grad_norm": 0.8907719254493713,
      "learning_rate": 3.9573864783229053e-07,
      "loss": 0.608,
      "step": 3149
    },
    {
      "epoch": 0.873059866962306,
      "grad_norm": 0.9615777134895325,
      "learning_rate": 3.9404285523263864e-07,
      "loss": 0.6049,
      "step": 3150
    },
    {
      "epoch": 0.8733370288248337,
      "grad_norm": 2.1850686073303223,
      "learning_rate": 3.9235055473203164e-07,
      "loss": 0.5435,
      "step": 3151
    },
    {
      "epoch": 0.8736141906873615,
      "grad_norm": 0.9337279796600342,
      "learning_rate": 3.906617476135227e-07,
      "loss": 0.6067,
      "step": 3152
    },
    {
      "epoch": 0.8738913525498891,
      "grad_norm": 1.0221915245056152,
      "learning_rate": 3.8897643515751315e-07,
      "loss": 0.6375,
      "step": 3153
    },
    {
      "epoch": 0.8741685144124168,
      "grad_norm": 1.4337308406829834,
      "learning_rate": 3.872946186417592e-07,
      "loss": 0.5624,
      "step": 3154
    },
    {
      "epoch": 0.8744456762749445,
      "grad_norm": 0.9208903312683105,
      "learning_rate": 3.85616299341362e-07,
      "loss": 0.5475,
      "step": 3155
    },
    {
      "epoch": 0.8747228381374723,
      "grad_norm": 0.8909422159194946,
      "learning_rate": 3.839414785287743e-07,
      "loss": 0.6059,
      "step": 3156
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.920901358127594,
      "learning_rate": 3.8227015747379305e-07,
      "loss": 0.5653,
      "step": 3157
    },
    {
      "epoch": 0.8752771618625277,
      "grad_norm": 0.865265965461731,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.5494,
      "step": 3158
    },
    {
      "epoch": 0.8755543237250555,
      "grad_norm": 0.8890305757522583,
      "learning_rate": 3.789380197025844e-07,
      "loss": 0.5616,
      "step": 3159
    },
    {
      "epoch": 0.8758314855875832,
      "grad_norm": 0.8838022351264954,
      "learning_rate": 3.7727720551268223e-07,
      "loss": 0.5523,
      "step": 3160
    },
    {
      "epoch": 0.8761086474501109,
      "grad_norm": 0.9336024522781372,
      "learning_rate": 3.7561989613304264e-07,
      "loss": 0.5862,
      "step": 3161
    },
    {
      "epoch": 0.8763858093126385,
      "grad_norm": 0.8348010182380676,
      "learning_rate": 3.7396609282018493e-07,
      "loss": 0.5649,
      "step": 3162
    },
    {
      "epoch": 0.8766629711751663,
      "grad_norm": 0.945239782333374,
      "learning_rate": 3.723157968279745e-07,
      "loss": 0.6238,
      "step": 3163
    },
    {
      "epoch": 0.876940133037694,
      "grad_norm": 0.9231823682785034,
      "learning_rate": 3.706690094076154e-07,
      "loss": 0.6191,
      "step": 3164
    },
    {
      "epoch": 0.8772172949002217,
      "grad_norm": 0.8679295778274536,
      "learning_rate": 3.69025731807654e-07,
      "loss": 0.5565,
      "step": 3165
    },
    {
      "epoch": 0.8774944567627494,
      "grad_norm": 0.8968644738197327,
      "learning_rate": 3.6738596527397475e-07,
      "loss": 0.63,
      "step": 3166
    },
    {
      "epoch": 0.8777716186252772,
      "grad_norm": 0.8925013542175293,
      "learning_rate": 3.657497110497987e-07,
      "loss": 0.5756,
      "step": 3167
    },
    {
      "epoch": 0.8780487804878049,
      "grad_norm": 1.0145095586776733,
      "learning_rate": 3.641169703756836e-07,
      "loss": 0.5992,
      "step": 3168
    },
    {
      "epoch": 0.8783259423503326,
      "grad_norm": 0.9143484234809875,
      "learning_rate": 3.6248774448952695e-07,
      "loss": 0.6169,
      "step": 3169
    },
    {
      "epoch": 0.8786031042128604,
      "grad_norm": 1.9589005708694458,
      "learning_rate": 3.608620346265579e-07,
      "loss": 0.5591,
      "step": 3170
    },
    {
      "epoch": 0.878880266075388,
      "grad_norm": 0.9512463212013245,
      "learning_rate": 3.592398420193405e-07,
      "loss": 0.6244,
      "step": 3171
    },
    {
      "epoch": 0.8791574279379157,
      "grad_norm": 0.897696852684021,
      "learning_rate": 3.5762116789777535e-07,
      "loss": 0.6194,
      "step": 3172
    },
    {
      "epoch": 0.8794345898004434,
      "grad_norm": 0.9666314721107483,
      "learning_rate": 3.560060134890897e-07,
      "loss": 0.6404,
      "step": 3173
    },
    {
      "epoch": 0.8797117516629712,
      "grad_norm": 0.8712226748466492,
      "learning_rate": 3.5439438001784676e-07,
      "loss": 0.5551,
      "step": 3174
    },
    {
      "epoch": 0.8799889135254989,
      "grad_norm": 0.9037732481956482,
      "learning_rate": 3.52786268705938e-07,
      "loss": 0.5953,
      "step": 3175
    },
    {
      "epoch": 0.8802660753880266,
      "grad_norm": 0.9434072375297546,
      "learning_rate": 3.5118168077258707e-07,
      "loss": 0.6104,
      "step": 3176
    },
    {
      "epoch": 0.8805432372505543,
      "grad_norm": 0.964867115020752,
      "learning_rate": 3.4958061743434304e-07,
      "loss": 0.6039,
      "step": 3177
    },
    {
      "epoch": 0.8808203991130821,
      "grad_norm": 0.886447012424469,
      "learning_rate": 3.4798307990508596e-07,
      "loss": 0.5991,
      "step": 3178
    },
    {
      "epoch": 0.8810975609756098,
      "grad_norm": 1.186274528503418,
      "learning_rate": 3.463890693960181e-07,
      "loss": 0.6521,
      "step": 3179
    },
    {
      "epoch": 0.8813747228381374,
      "grad_norm": 1.1206927299499512,
      "learning_rate": 3.447985871156734e-07,
      "loss": 0.5738,
      "step": 3180
    },
    {
      "epoch": 0.8816518847006651,
      "grad_norm": 0.8783342242240906,
      "learning_rate": 3.432116342699071e-07,
      "loss": 0.5725,
      "step": 3181
    },
    {
      "epoch": 0.8819290465631929,
      "grad_norm": 0.9446502327919006,
      "learning_rate": 3.4162821206189857e-07,
      "loss": 0.6395,
      "step": 3182
    },
    {
      "epoch": 0.8822062084257206,
      "grad_norm": 0.9326784610748291,
      "learning_rate": 3.400483216921524e-07,
      "loss": 0.5922,
      "step": 3183
    },
    {
      "epoch": 0.8824833702882483,
      "grad_norm": 1.0650842189788818,
      "learning_rate": 3.3847196435849385e-07,
      "loss": 0.6295,
      "step": 3184
    },
    {
      "epoch": 0.8827605321507761,
      "grad_norm": 0.8696801662445068,
      "learning_rate": 3.368991412560696e-07,
      "loss": 0.5981,
      "step": 3185
    },
    {
      "epoch": 0.8830376940133038,
      "grad_norm": 0.9241025447845459,
      "learning_rate": 3.3532985357734694e-07,
      "loss": 0.6291,
      "step": 3186
    },
    {
      "epoch": 0.8833148558758315,
      "grad_norm": 0.8954719305038452,
      "learning_rate": 3.337641025121136e-07,
      "loss": 0.6162,
      "step": 3187
    },
    {
      "epoch": 0.8835920177383592,
      "grad_norm": 0.8846049308776855,
      "learning_rate": 3.3220188924747486e-07,
      "loss": 0.6018,
      "step": 3188
    },
    {
      "epoch": 0.883869179600887,
      "grad_norm": 0.889862060546875,
      "learning_rate": 3.306432149678557e-07,
      "loss": 0.5705,
      "step": 3189
    },
    {
      "epoch": 0.8841463414634146,
      "grad_norm": 1.077681541442871,
      "learning_rate": 3.290880808549929e-07,
      "loss": 0.5886,
      "step": 3190
    },
    {
      "epoch": 0.8844235033259423,
      "grad_norm": 0.9393237233161926,
      "learning_rate": 3.2753648808794505e-07,
      "loss": 0.6367,
      "step": 3191
    },
    {
      "epoch": 0.88470066518847,
      "grad_norm": 1.1079896688461304,
      "learning_rate": 3.259884378430839e-07,
      "loss": 0.5954,
      "step": 3192
    },
    {
      "epoch": 0.8849778270509978,
      "grad_norm": 0.9378155469894409,
      "learning_rate": 3.2444393129409336e-07,
      "loss": 0.6016,
      "step": 3193
    },
    {
      "epoch": 0.8852549889135255,
      "grad_norm": 0.9072238206863403,
      "learning_rate": 3.2290296961197287e-07,
      "loss": 0.6158,
      "step": 3194
    },
    {
      "epoch": 0.8855321507760532,
      "grad_norm": 0.876338005065918,
      "learning_rate": 3.2136555396503346e-07,
      "loss": 0.6337,
      "step": 3195
    },
    {
      "epoch": 0.885809312638581,
      "grad_norm": 0.9981226325035095,
      "learning_rate": 3.198316855188977e-07,
      "loss": 0.6234,
      "step": 3196
    },
    {
      "epoch": 0.8860864745011087,
      "grad_norm": 0.8923183679580688,
      "learning_rate": 3.18301365436498e-07,
      "loss": 0.5887,
      "step": 3197
    },
    {
      "epoch": 0.8863636363636364,
      "grad_norm": 0.9313464164733887,
      "learning_rate": 3.167745948780787e-07,
      "loss": 0.571,
      "step": 3198
    },
    {
      "epoch": 0.886640798226164,
      "grad_norm": 0.962217390537262,
      "learning_rate": 3.1525137500119207e-07,
      "loss": 0.5661,
      "step": 3199
    },
    {
      "epoch": 0.8869179600886918,
      "grad_norm": 0.8958284854888916,
      "learning_rate": 3.1373170696069597e-07,
      "loss": 0.5606,
      "step": 3200
    },
    {
      "epoch": 0.8871951219512195,
      "grad_norm": 1.5315319299697876,
      "learning_rate": 3.1221559190875716e-07,
      "loss": 0.6366,
      "step": 3201
    },
    {
      "epoch": 0.8874722838137472,
      "grad_norm": 0.8770660758018494,
      "learning_rate": 3.1070303099485055e-07,
      "loss": 0.6058,
      "step": 3202
    },
    {
      "epoch": 0.8877494456762749,
      "grad_norm": 0.8936622142791748,
      "learning_rate": 3.091940253657538e-07,
      "loss": 0.6107,
      "step": 3203
    },
    {
      "epoch": 0.8880266075388027,
      "grad_norm": 1.000054955482483,
      "learning_rate": 3.0768857616555015e-07,
      "loss": 0.5893,
      "step": 3204
    },
    {
      "epoch": 0.8883037694013304,
      "grad_norm": 0.871757447719574,
      "learning_rate": 3.0618668453562595e-07,
      "loss": 0.5508,
      "step": 3205
    },
    {
      "epoch": 0.8885809312638581,
      "grad_norm": 1.0230610370635986,
      "learning_rate": 3.046883516146704e-07,
      "loss": 0.585,
      "step": 3206
    },
    {
      "epoch": 0.8888580931263859,
      "grad_norm": 0.9172118902206421,
      "learning_rate": 3.031935785386747e-07,
      "loss": 0.6155,
      "step": 3207
    },
    {
      "epoch": 0.8891352549889135,
      "grad_norm": 0.8957566022872925,
      "learning_rate": 3.0170236644093074e-07,
      "loss": 0.6148,
      "step": 3208
    },
    {
      "epoch": 0.8894124168514412,
      "grad_norm": 4.542577266693115,
      "learning_rate": 3.002147164520319e-07,
      "loss": 0.5961,
      "step": 3209
    },
    {
      "epoch": 0.8896895787139689,
      "grad_norm": 0.8909934163093567,
      "learning_rate": 2.98730629699871e-07,
      "loss": 0.5766,
      "step": 3210
    },
    {
      "epoch": 0.8899667405764967,
      "grad_norm": 1.0035943984985352,
      "learning_rate": 2.9725010730963566e-07,
      "loss": 0.6149,
      "step": 3211
    },
    {
      "epoch": 0.8902439024390244,
      "grad_norm": 0.8604962825775146,
      "learning_rate": 2.957731504038136e-07,
      "loss": 0.5307,
      "step": 3212
    },
    {
      "epoch": 0.8905210643015521,
      "grad_norm": 0.9013574719429016,
      "learning_rate": 2.942997601021924e-07,
      "loss": 0.6046,
      "step": 3213
    },
    {
      "epoch": 0.8907982261640798,
      "grad_norm": 1.7110899686813354,
      "learning_rate": 2.9282993752185016e-07,
      "loss": 0.6174,
      "step": 3214
    },
    {
      "epoch": 0.8910753880266076,
      "grad_norm": 0.9859708547592163,
      "learning_rate": 2.9136368377716327e-07,
      "loss": 0.6201,
      "step": 3215
    },
    {
      "epoch": 0.8913525498891353,
      "grad_norm": 0.9034924507141113,
      "learning_rate": 2.899009999798008e-07,
      "loss": 0.5808,
      "step": 3216
    },
    {
      "epoch": 0.891629711751663,
      "grad_norm": 0.8715956807136536,
      "learning_rate": 2.8844188723872737e-07,
      "loss": 0.5397,
      "step": 3217
    },
    {
      "epoch": 0.8919068736141907,
      "grad_norm": 1.541416883468628,
      "learning_rate": 2.869863466601969e-07,
      "loss": 0.565,
      "step": 3218
    },
    {
      "epoch": 0.8921840354767184,
      "grad_norm": 0.8920409083366394,
      "learning_rate": 2.855343793477572e-07,
      "loss": 0.5535,
      "step": 3219
    },
    {
      "epoch": 0.8924611973392461,
      "grad_norm": 0.9265097975730896,
      "learning_rate": 2.8408598640224715e-07,
      "loss": 0.6088,
      "step": 3220
    },
    {
      "epoch": 0.8927383592017738,
      "grad_norm": 0.9068689942359924,
      "learning_rate": 2.826411689217956e-07,
      "loss": 0.5499,
      "step": 3221
    },
    {
      "epoch": 0.8930155210643016,
      "grad_norm": 0.9220547080039978,
      "learning_rate": 2.811999280018185e-07,
      "loss": 0.6118,
      "step": 3222
    },
    {
      "epoch": 0.8932926829268293,
      "grad_norm": 0.8870977759361267,
      "learning_rate": 2.797622647350212e-07,
      "loss": 0.5659,
      "step": 3223
    },
    {
      "epoch": 0.893569844789357,
      "grad_norm": 0.9130967855453491,
      "learning_rate": 2.783281802113985e-07,
      "loss": 0.6011,
      "step": 3224
    },
    {
      "epoch": 0.8938470066518847,
      "grad_norm": 0.8833882212638855,
      "learning_rate": 2.7689767551823067e-07,
      "loss": 0.6456,
      "step": 3225
    },
    {
      "epoch": 0.8941241685144125,
      "grad_norm": 0.8868809938430786,
      "learning_rate": 2.754707517400823e-07,
      "loss": 0.6137,
      "step": 3226
    },
    {
      "epoch": 0.8944013303769401,
      "grad_norm": 1.0524964332580566,
      "learning_rate": 2.740474099588053e-07,
      "loss": 0.6718,
      "step": 3227
    },
    {
      "epoch": 0.8946784922394678,
      "grad_norm": 0.9088641405105591,
      "learning_rate": 2.726276512535342e-07,
      "loss": 0.5822,
      "step": 3228
    },
    {
      "epoch": 0.8949556541019955,
      "grad_norm": 0.8602390885353088,
      "learning_rate": 2.712114767006879e-07,
      "loss": 0.5697,
      "step": 3229
    },
    {
      "epoch": 0.8952328159645233,
      "grad_norm": 0.862032413482666,
      "learning_rate": 2.697988873739676e-07,
      "loss": 0.6317,
      "step": 3230
    },
    {
      "epoch": 0.895509977827051,
      "grad_norm": 0.8910408616065979,
      "learning_rate": 2.683898843443555e-07,
      "loss": 0.561,
      "step": 3231
    },
    {
      "epoch": 0.8957871396895787,
      "grad_norm": 0.8637416958808899,
      "learning_rate": 2.669844686801182e-07,
      "loss": 0.5574,
      "step": 3232
    },
    {
      "epoch": 0.8960643015521065,
      "grad_norm": 0.986998438835144,
      "learning_rate": 2.6558264144679715e-07,
      "loss": 0.5684,
      "step": 3233
    },
    {
      "epoch": 0.8963414634146342,
      "grad_norm": 0.9156070947647095,
      "learning_rate": 2.641844037072161e-07,
      "loss": 0.5663,
      "step": 3234
    },
    {
      "epoch": 0.8966186252771619,
      "grad_norm": 71.44258880615234,
      "learning_rate": 2.6278975652147875e-07,
      "loss": 0.5939,
      "step": 3235
    },
    {
      "epoch": 0.8968957871396895,
      "grad_norm": 0.9284784197807312,
      "learning_rate": 2.613987009469637e-07,
      "loss": 0.6114,
      "step": 3236
    },
    {
      "epoch": 0.8971729490022173,
      "grad_norm": 0.9667542576789856,
      "learning_rate": 2.60011238038328e-07,
      "loss": 0.6237,
      "step": 3237
    },
    {
      "epoch": 0.897450110864745,
      "grad_norm": 0.9795854687690735,
      "learning_rate": 2.586273688475044e-07,
      "loss": 0.6418,
      "step": 3238
    },
    {
      "epoch": 0.8977272727272727,
      "grad_norm": 2.3076579570770264,
      "learning_rate": 2.5724709442370155e-07,
      "loss": 0.6288,
      "step": 3239
    },
    {
      "epoch": 0.8980044345898004,
      "grad_norm": 0.98255455493927,
      "learning_rate": 2.5587041581340235e-07,
      "loss": 0.5939,
      "step": 3240
    },
    {
      "epoch": 0.8982815964523282,
      "grad_norm": 0.885590136051178,
      "learning_rate": 2.5449733406036235e-07,
      "loss": 0.5995,
      "step": 3241
    },
    {
      "epoch": 0.8985587583148559,
      "grad_norm": 0.915796160697937,
      "learning_rate": 2.5312785020561225e-07,
      "loss": 0.6526,
      "step": 3242
    },
    {
      "epoch": 0.8988359201773836,
      "grad_norm": 1.097190022468567,
      "learning_rate": 2.5176196528745313e-07,
      "loss": 0.5627,
      "step": 3243
    },
    {
      "epoch": 0.8991130820399114,
      "grad_norm": 1.059183955192566,
      "learning_rate": 2.503996803414582e-07,
      "loss": 0.5641,
      "step": 3244
    },
    {
      "epoch": 0.899390243902439,
      "grad_norm": 0.9504148960113525,
      "learning_rate": 2.490409964004703e-07,
      "loss": 0.6272,
      "step": 3245
    },
    {
      "epoch": 0.8996674057649667,
      "grad_norm": 0.9209854006767273,
      "learning_rate": 2.476859144946053e-07,
      "loss": 0.5601,
      "step": 3246
    },
    {
      "epoch": 0.8999445676274944,
      "grad_norm": 0.9400312304496765,
      "learning_rate": 2.463344356512437e-07,
      "loss": 0.5709,
      "step": 3247
    },
    {
      "epoch": 0.9002217294900222,
      "grad_norm": 0.8447540402412415,
      "learning_rate": 2.4498656089503805e-07,
      "loss": 0.5966,
      "step": 3248
    },
    {
      "epoch": 0.9004988913525499,
      "grad_norm": 0.8677281141281128,
      "learning_rate": 2.436422912479053e-07,
      "loss": 0.553,
      "step": 3249
    },
    {
      "epoch": 0.9007760532150776,
      "grad_norm": 1.5341401100158691,
      "learning_rate": 2.4230162772903074e-07,
      "loss": 0.5872,
      "step": 3250
    },
    {
      "epoch": 0.9010532150776053,
      "grad_norm": 0.895592212677002,
      "learning_rate": 2.409645713548664e-07,
      "loss": 0.6143,
      "step": 3251
    },
    {
      "epoch": 0.9013303769401331,
      "grad_norm": 0.9074665904045105,
      "learning_rate": 2.3963112313912684e-07,
      "loss": 0.5943,
      "step": 3252
    },
    {
      "epoch": 0.9016075388026608,
      "grad_norm": 0.9578946232795715,
      "learning_rate": 2.3830128409279395e-07,
      "loss": 0.5942,
      "step": 3253
    },
    {
      "epoch": 0.9018847006651884,
      "grad_norm": 0.9234129190444946,
      "learning_rate": 2.369750552241118e-07,
      "loss": 0.5811,
      "step": 3254
    },
    {
      "epoch": 0.9021618625277162,
      "grad_norm": 0.8978248834609985,
      "learning_rate": 2.3565243753858658e-07,
      "loss": 0.6114,
      "step": 3255
    },
    {
      "epoch": 0.9024390243902439,
      "grad_norm": 1.0330461263656616,
      "learning_rate": 2.3433343203898783e-07,
      "loss": 0.6462,
      "step": 3256
    },
    {
      "epoch": 0.9027161862527716,
      "grad_norm": 0.8822248578071594,
      "learning_rate": 2.330180397253473e-07,
      "loss": 0.6205,
      "step": 3257
    },
    {
      "epoch": 0.9029933481152993,
      "grad_norm": 0.9551835656166077,
      "learning_rate": 2.31706261594955e-07,
      "loss": 0.5893,
      "step": 3258
    },
    {
      "epoch": 0.9032705099778271,
      "grad_norm": 1.0977249145507812,
      "learning_rate": 2.303980986423626e-07,
      "loss": 0.6534,
      "step": 3259
    },
    {
      "epoch": 0.9035476718403548,
      "grad_norm": 0.9334918260574341,
      "learning_rate": 2.2909355185937844e-07,
      "loss": 0.6142,
      "step": 3260
    },
    {
      "epoch": 0.9038248337028825,
      "grad_norm": 0.8930160999298096,
      "learning_rate": 2.2779262223507238e-07,
      "loss": 0.5832,
      "step": 3261
    },
    {
      "epoch": 0.9041019955654102,
      "grad_norm": 0.8896879553794861,
      "learning_rate": 2.2649531075577047e-07,
      "loss": 0.5675,
      "step": 3262
    },
    {
      "epoch": 0.904379157427938,
      "grad_norm": 0.8879204392433167,
      "learning_rate": 2.252016184050554e-07,
      "loss": 0.5858,
      "step": 3263
    },
    {
      "epoch": 0.9046563192904656,
      "grad_norm": 0.8994612693786621,
      "learning_rate": 2.2391154616376532e-07,
      "loss": 0.5967,
      "step": 3264
    },
    {
      "epoch": 0.9049334811529933,
      "grad_norm": 0.9033707976341248,
      "learning_rate": 2.2262509500999507e-07,
      "loss": 0.6729,
      "step": 3265
    },
    {
      "epoch": 0.905210643015521,
      "grad_norm": 0.9183560013771057,
      "learning_rate": 2.2134226591909335e-07,
      "loss": 0.6081,
      "step": 3266
    },
    {
      "epoch": 0.9054878048780488,
      "grad_norm": 2.5113847255706787,
      "learning_rate": 2.2006305986366272e-07,
      "loss": 0.5962,
      "step": 3267
    },
    {
      "epoch": 0.9057649667405765,
      "grad_norm": 0.9004520177841187,
      "learning_rate": 2.1878747781355847e-07,
      "loss": 0.6453,
      "step": 3268
    },
    {
      "epoch": 0.9060421286031042,
      "grad_norm": 0.91668301820755,
      "learning_rate": 2.1751552073589032e-07,
      "loss": 0.5594,
      "step": 3269
    },
    {
      "epoch": 0.906319290465632,
      "grad_norm": 0.8704707026481628,
      "learning_rate": 2.1624718959501745e-07,
      "loss": 0.5801,
      "step": 3270
    },
    {
      "epoch": 0.9065964523281597,
      "grad_norm": 2.0776472091674805,
      "learning_rate": 2.1498248535254895e-07,
      "loss": 0.5643,
      "step": 3271
    },
    {
      "epoch": 0.9068736141906873,
      "grad_norm": 0.9032646417617798,
      "learning_rate": 2.1372140896734782e-07,
      "loss": 0.5803,
      "step": 3272
    },
    {
      "epoch": 0.907150776053215,
      "grad_norm": 1.0452221632003784,
      "learning_rate": 2.1246396139552426e-07,
      "loss": 0.5968,
      "step": 3273
    },
    {
      "epoch": 0.9074279379157428,
      "grad_norm": 0.9451413154602051,
      "learning_rate": 2.112101435904368e-07,
      "loss": 0.5949,
      "step": 3274
    },
    {
      "epoch": 0.9077050997782705,
      "grad_norm": 0.9008277058601379,
      "learning_rate": 2.0995995650269275e-07,
      "loss": 0.5882,
      "step": 3275
    },
    {
      "epoch": 0.9079822616407982,
      "grad_norm": 0.8833110332489014,
      "learning_rate": 2.087134010801467e-07,
      "loss": 0.6499,
      "step": 3276
    },
    {
      "epoch": 0.9082594235033259,
      "grad_norm": 3.0891177654266357,
      "learning_rate": 2.074704782678999e-07,
      "loss": 0.5884,
      "step": 3277
    },
    {
      "epoch": 0.9085365853658537,
      "grad_norm": 0.8932285904884338,
      "learning_rate": 2.0623118900829853e-07,
      "loss": 0.6041,
      "step": 3278
    },
    {
      "epoch": 0.9088137472283814,
      "grad_norm": 0.8929831385612488,
      "learning_rate": 2.049955342409349e-07,
      "loss": 0.5661,
      "step": 3279
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.8975732326507568,
      "learning_rate": 2.0376351490264678e-07,
      "loss": 0.6115,
      "step": 3280
    },
    {
      "epoch": 0.9093680709534369,
      "grad_norm": 0.8626673221588135,
      "learning_rate": 2.0253513192751374e-07,
      "loss": 0.5717,
      "step": 3281
    },
    {
      "epoch": 0.9096452328159645,
      "grad_norm": 1.3280938863754272,
      "learning_rate": 2.013103862468574e-07,
      "loss": 0.6102,
      "step": 3282
    },
    {
      "epoch": 0.9099223946784922,
      "grad_norm": 0.9030762910842896,
      "learning_rate": 2.0008927878924557e-07,
      "loss": 0.6006,
      "step": 3283
    },
    {
      "epoch": 0.9101995565410199,
      "grad_norm": 0.8815461993217468,
      "learning_rate": 1.9887181048048432e-07,
      "loss": 0.6288,
      "step": 3284
    },
    {
      "epoch": 0.9104767184035477,
      "grad_norm": 1.3317636251449585,
      "learning_rate": 1.9765798224362197e-07,
      "loss": 0.5949,
      "step": 3285
    },
    {
      "epoch": 0.9107538802660754,
      "grad_norm": 0.979107677936554,
      "learning_rate": 1.964477949989474e-07,
      "loss": 0.5908,
      "step": 3286
    },
    {
      "epoch": 0.9110310421286031,
      "grad_norm": 0.8687864542007446,
      "learning_rate": 1.9524124966398717e-07,
      "loss": 0.5614,
      "step": 3287
    },
    {
      "epoch": 0.9113082039911308,
      "grad_norm": 0.9358117580413818,
      "learning_rate": 1.9403834715350844e-07,
      "loss": 0.616,
      "step": 3288
    },
    {
      "epoch": 0.9115853658536586,
      "grad_norm": 1.5283809900283813,
      "learning_rate": 1.928390883795167e-07,
      "loss": 0.5917,
      "step": 3289
    },
    {
      "epoch": 0.9118625277161863,
      "grad_norm": 0.8779503703117371,
      "learning_rate": 1.916434742512524e-07,
      "loss": 0.5875,
      "step": 3290
    },
    {
      "epoch": 0.9121396895787139,
      "grad_norm": 0.9456507563591003,
      "learning_rate": 1.90451505675196e-07,
      "loss": 0.6097,
      "step": 3291
    },
    {
      "epoch": 0.9124168514412417,
      "grad_norm": 0.8666355609893799,
      "learning_rate": 1.892631835550629e-07,
      "loss": 0.6112,
      "step": 3292
    },
    {
      "epoch": 0.9126940133037694,
      "grad_norm": 0.911436915397644,
      "learning_rate": 1.8807850879180078e-07,
      "loss": 0.5682,
      "step": 3293
    },
    {
      "epoch": 0.9129711751662971,
      "grad_norm": 0.9080280065536499,
      "learning_rate": 1.8689748228359728e-07,
      "loss": 0.5768,
      "step": 3294
    },
    {
      "epoch": 0.9132483370288248,
      "grad_norm": 0.9375961422920227,
      "learning_rate": 1.8572010492587001e-07,
      "loss": 0.6425,
      "step": 3295
    },
    {
      "epoch": 0.9135254988913526,
      "grad_norm": 1.2250362634658813,
      "learning_rate": 1.8454637761127213e-07,
      "loss": 0.6394,
      "step": 3296
    },
    {
      "epoch": 0.9138026607538803,
      "grad_norm": 0.947627604007721,
      "learning_rate": 1.8337630122968796e-07,
      "loss": 0.6038,
      "step": 3297
    },
    {
      "epoch": 0.914079822616408,
      "grad_norm": 0.8502238988876343,
      "learning_rate": 1.8220987666823508e-07,
      "loss": 0.5723,
      "step": 3298
    },
    {
      "epoch": 0.9143569844789357,
      "grad_norm": 0.9765928387641907,
      "learning_rate": 1.8104710481126052e-07,
      "loss": 0.651,
      "step": 3299
    },
    {
      "epoch": 0.9146341463414634,
      "grad_norm": 0.9094823002815247,
      "learning_rate": 1.7988798654034466e-07,
      "loss": 0.6374,
      "step": 3300
    },
    {
      "epoch": 0.9149113082039911,
      "grad_norm": 1.5822863578796387,
      "learning_rate": 1.787325227342951e-07,
      "loss": 0.5446,
      "step": 3301
    },
    {
      "epoch": 0.9151884700665188,
      "grad_norm": 0.875126838684082,
      "learning_rate": 1.7758071426915102e-07,
      "loss": 0.6043,
      "step": 3302
    },
    {
      "epoch": 0.9154656319290465,
      "grad_norm": 0.9265002012252808,
      "learning_rate": 1.764325620181784e-07,
      "loss": 0.6264,
      "step": 3303
    },
    {
      "epoch": 0.9157427937915743,
      "grad_norm": 0.8867161273956299,
      "learning_rate": 1.752880668518725e-07,
      "loss": 0.6001,
      "step": 3304
    },
    {
      "epoch": 0.916019955654102,
      "grad_norm": 0.9153332114219666,
      "learning_rate": 1.7414722963795426e-07,
      "loss": 0.5833,
      "step": 3305
    },
    {
      "epoch": 0.9162971175166297,
      "grad_norm": 0.8688998222351074,
      "learning_rate": 1.7301005124137348e-07,
      "loss": 0.5946,
      "step": 3306
    },
    {
      "epoch": 0.9165742793791575,
      "grad_norm": 0.8976378440856934,
      "learning_rate": 1.7187653252430492e-07,
      "loss": 0.5856,
      "step": 3307
    },
    {
      "epoch": 0.9168514412416852,
      "grad_norm": 0.9155281186103821,
      "learning_rate": 1.7074667434614777e-07,
      "loss": 0.5963,
      "step": 3308
    },
    {
      "epoch": 0.9171286031042128,
      "grad_norm": 0.9036497473716736,
      "learning_rate": 1.696204775635274e-07,
      "loss": 0.6073,
      "step": 3309
    },
    {
      "epoch": 0.9174057649667405,
      "grad_norm": 0.9359136819839478,
      "learning_rate": 1.6849794303029187e-07,
      "loss": 0.5948,
      "step": 3310
    },
    {
      "epoch": 0.9176829268292683,
      "grad_norm": 0.9224042892456055,
      "learning_rate": 1.6737907159751432e-07,
      "loss": 0.6218,
      "step": 3311
    },
    {
      "epoch": 0.917960088691796,
      "grad_norm": 1.1248661279678345,
      "learning_rate": 1.6626386411348783e-07,
      "loss": 0.6376,
      "step": 3312
    },
    {
      "epoch": 0.9182372505543237,
      "grad_norm": 0.8539760708808899,
      "learning_rate": 1.6515232142373217e-07,
      "loss": 0.5662,
      "step": 3313
    },
    {
      "epoch": 0.9185144124168514,
      "grad_norm": 1.0661745071411133,
      "learning_rate": 1.6404444437098376e-07,
      "loss": 0.568,
      "step": 3314
    },
    {
      "epoch": 0.9187915742793792,
      "grad_norm": 1.019711971282959,
      "learning_rate": 1.6294023379520186e-07,
      "loss": 0.5693,
      "step": 3315
    },
    {
      "epoch": 0.9190687361419069,
      "grad_norm": 0.8921720385551453,
      "learning_rate": 1.6183969053356618e-07,
      "loss": 0.5533,
      "step": 3316
    },
    {
      "epoch": 0.9193458980044346,
      "grad_norm": 1.401977777481079,
      "learning_rate": 1.6074281542047654e-07,
      "loss": 0.5803,
      "step": 3317
    },
    {
      "epoch": 0.9196230598669624,
      "grad_norm": 0.8679239749908447,
      "learning_rate": 1.596496092875499e-07,
      "loss": 0.5766,
      "step": 3318
    },
    {
      "epoch": 0.91990022172949,
      "grad_norm": 0.868754506111145,
      "learning_rate": 1.585600729636233e-07,
      "loss": 0.5901,
      "step": 3319
    },
    {
      "epoch": 0.9201773835920177,
      "grad_norm": 0.8776298761367798,
      "learning_rate": 1.5747420727474983e-07,
      "loss": 0.5854,
      "step": 3320
    },
    {
      "epoch": 0.9204545454545454,
      "grad_norm": 0.9194466471672058,
      "learning_rate": 1.5639201304420094e-07,
      "loss": 0.5929,
      "step": 3321
    },
    {
      "epoch": 0.9207317073170732,
      "grad_norm": 3.052128791809082,
      "learning_rate": 1.5531349109246364e-07,
      "loss": 0.6085,
      "step": 3322
    },
    {
      "epoch": 0.9210088691796009,
      "grad_norm": 0.9028221964836121,
      "learning_rate": 1.542386422372405e-07,
      "loss": 0.5918,
      "step": 3323
    },
    {
      "epoch": 0.9212860310421286,
      "grad_norm": 0.9225941300392151,
      "learning_rate": 1.5316746729345188e-07,
      "loss": 0.6131,
      "step": 3324
    },
    {
      "epoch": 0.9215631929046563,
      "grad_norm": 1.0857294797897339,
      "learning_rate": 1.5209996707322817e-07,
      "loss": 0.5975,
      "step": 3325
    },
    {
      "epoch": 0.9218403547671841,
      "grad_norm": 0.8855028748512268,
      "learning_rate": 1.5103614238591746e-07,
      "loss": 0.6068,
      "step": 3326
    },
    {
      "epoch": 0.9221175166297118,
      "grad_norm": 0.904399573802948,
      "learning_rate": 1.499759940380785e-07,
      "loss": 0.6286,
      "step": 3327
    },
    {
      "epoch": 0.9223946784922394,
      "grad_norm": 0.9075015783309937,
      "learning_rate": 1.4891952283348608e-07,
      "loss": 0.6559,
      "step": 3328
    },
    {
      "epoch": 0.9226718403547672,
      "grad_norm": 1.155478596687317,
      "learning_rate": 1.47866729573124e-07,
      "loss": 0.5828,
      "step": 3329
    },
    {
      "epoch": 0.9229490022172949,
      "grad_norm": 1.087980031967163,
      "learning_rate": 1.4681761505518866e-07,
      "loss": 0.624,
      "step": 3330
    },
    {
      "epoch": 0.9232261640798226,
      "grad_norm": 0.9388636350631714,
      "learning_rate": 1.457721800750872e-07,
      "loss": 0.5459,
      "step": 3331
    },
    {
      "epoch": 0.9235033259423503,
      "grad_norm": 0.9056477546691895,
      "learning_rate": 1.4473042542543782e-07,
      "loss": 0.6535,
      "step": 3332
    },
    {
      "epoch": 0.9237804878048781,
      "grad_norm": 2.370955228805542,
      "learning_rate": 1.43692351896067e-07,
      "loss": 0.6059,
      "step": 3333
    },
    {
      "epoch": 0.9240576496674058,
      "grad_norm": 0.8677951097488403,
      "learning_rate": 1.426579602740108e-07,
      "loss": 0.5409,
      "step": 3334
    },
    {
      "epoch": 0.9243348115299335,
      "grad_norm": 0.9404576420783997,
      "learning_rate": 1.4162725134351573e-07,
      "loss": 0.6332,
      "step": 3335
    },
    {
      "epoch": 0.9246119733924612,
      "grad_norm": 0.8663669228553772,
      "learning_rate": 1.4060022588603283e-07,
      "loss": 0.5927,
      "step": 3336
    },
    {
      "epoch": 0.924889135254989,
      "grad_norm": 0.9205547571182251,
      "learning_rate": 1.3957688468022313e-07,
      "loss": 0.6423,
      "step": 3337
    },
    {
      "epoch": 0.9251662971175166,
      "grad_norm": 0.8978030681610107,
      "learning_rate": 1.385572285019521e-07,
      "loss": 0.6088,
      "step": 3338
    },
    {
      "epoch": 0.9254434589800443,
      "grad_norm": 1.1959782838821411,
      "learning_rate": 1.3754125812429352e-07,
      "loss": 0.6776,
      "step": 3339
    },
    {
      "epoch": 0.9257206208425721,
      "grad_norm": 0.9296877384185791,
      "learning_rate": 1.3652897431752576e-07,
      "loss": 0.6308,
      "step": 3340
    },
    {
      "epoch": 0.9259977827050998,
      "grad_norm": 0.8530126214027405,
      "learning_rate": 1.3552037784913208e-07,
      "loss": 0.5231,
      "step": 3341
    },
    {
      "epoch": 0.9262749445676275,
      "grad_norm": 0.9269120097160339,
      "learning_rate": 1.3451546948380022e-07,
      "loss": 0.6704,
      "step": 3342
    },
    {
      "epoch": 0.9265521064301552,
      "grad_norm": 0.9044782519340515,
      "learning_rate": 1.3351424998342134e-07,
      "loss": 0.5879,
      "step": 3343
    },
    {
      "epoch": 0.926829268292683,
      "grad_norm": 0.89072585105896,
      "learning_rate": 1.3251672010709094e-07,
      "loss": 0.5815,
      "step": 3344
    },
    {
      "epoch": 0.9271064301552107,
      "grad_norm": 0.8945850729942322,
      "learning_rate": 1.3152288061110518e-07,
      "loss": 0.5852,
      "step": 3345
    },
    {
      "epoch": 0.9273835920177383,
      "grad_norm": 0.9248208403587341,
      "learning_rate": 1.3053273224896413e-07,
      "loss": 0.6167,
      "step": 3346
    },
    {
      "epoch": 0.927660753880266,
      "grad_norm": 0.9458051919937134,
      "learning_rate": 1.2954627577136835e-07,
      "loss": 0.5518,
      "step": 3347
    },
    {
      "epoch": 0.9279379157427938,
      "grad_norm": 0.8808291554450989,
      "learning_rate": 1.285635119262202e-07,
      "loss": 0.5854,
      "step": 3348
    },
    {
      "epoch": 0.9282150776053215,
      "grad_norm": 0.9023213386535645,
      "learning_rate": 1.275844414586208e-07,
      "loss": 0.6127,
      "step": 3349
    },
    {
      "epoch": 0.9284922394678492,
      "grad_norm": 0.9000441431999207,
      "learning_rate": 1.2660906511087366e-07,
      "loss": 0.5821,
      "step": 3350
    },
    {
      "epoch": 0.9287694013303769,
      "grad_norm": 0.91656494140625,
      "learning_rate": 1.2563738362247892e-07,
      "loss": 0.5956,
      "step": 3351
    },
    {
      "epoch": 0.9290465631929047,
      "grad_norm": 1.247296929359436,
      "learning_rate": 1.2466939773013774e-07,
      "loss": 0.5797,
      "step": 3352
    },
    {
      "epoch": 0.9293237250554324,
      "grad_norm": 0.9189592003822327,
      "learning_rate": 1.237051081677454e-07,
      "loss": 0.5634,
      "step": 3353
    },
    {
      "epoch": 0.9296008869179601,
      "grad_norm": 0.907017707824707,
      "learning_rate": 1.2274451566640033e-07,
      "loss": 0.6286,
      "step": 3354
    },
    {
      "epoch": 0.9298780487804879,
      "grad_norm": 0.9345569610595703,
      "learning_rate": 1.2178762095439334e-07,
      "loss": 0.564,
      "step": 3355
    },
    {
      "epoch": 0.9301552106430155,
      "grad_norm": 2.047731637954712,
      "learning_rate": 1.2083442475721353e-07,
      "loss": 0.5947,
      "step": 3356
    },
    {
      "epoch": 0.9304323725055432,
      "grad_norm": 1.080769658088684,
      "learning_rate": 1.1988492779754668e-07,
      "loss": 0.5796,
      "step": 3357
    },
    {
      "epoch": 0.9307095343680709,
      "grad_norm": 0.8791870474815369,
      "learning_rate": 1.1893913079527252e-07,
      "loss": 0.5857,
      "step": 3358
    },
    {
      "epoch": 0.9309866962305987,
      "grad_norm": 0.8703077435493469,
      "learning_rate": 1.179970344674658e-07,
      "loss": 0.5806,
      "step": 3359
    },
    {
      "epoch": 0.9312638580931264,
      "grad_norm": 0.9094696044921875,
      "learning_rate": 1.1705863952839513e-07,
      "loss": 0.5699,
      "step": 3360
    },
    {
      "epoch": 0.9315410199556541,
      "grad_norm": 1.3129992485046387,
      "learning_rate": 1.1612394668952531e-07,
      "loss": 0.6346,
      "step": 3361
    },
    {
      "epoch": 0.9318181818181818,
      "grad_norm": 0.8581680059432983,
      "learning_rate": 1.1519295665951169e-07,
      "loss": 0.5644,
      "step": 3362
    },
    {
      "epoch": 0.9320953436807096,
      "grad_norm": 0.8981721997261047,
      "learning_rate": 1.1426567014420297e-07,
      "loss": 0.6017,
      "step": 3363
    },
    {
      "epoch": 0.9323725055432373,
      "grad_norm": 0.8988456130027771,
      "learning_rate": 1.1334208784663903e-07,
      "loss": 0.5731,
      "step": 3364
    },
    {
      "epoch": 0.9326496674057649,
      "grad_norm": 0.8890488743782043,
      "learning_rate": 1.1242221046705415e-07,
      "loss": 0.5694,
      "step": 3365
    },
    {
      "epoch": 0.9329268292682927,
      "grad_norm": 0.9228889346122742,
      "learning_rate": 1.1150603870287158e-07,
      "loss": 0.5767,
      "step": 3366
    },
    {
      "epoch": 0.9332039911308204,
      "grad_norm": 0.8663297295570374,
      "learning_rate": 1.1059357324870456e-07,
      "loss": 0.54,
      "step": 3367
    },
    {
      "epoch": 0.9334811529933481,
      "grad_norm": 0.9155005812644958,
      "learning_rate": 1.0968481479635696e-07,
      "loss": 0.5791,
      "step": 3368
    },
    {
      "epoch": 0.9337583148558758,
      "grad_norm": 0.8805655241012573,
      "learning_rate": 1.0877976403482316e-07,
      "loss": 0.5979,
      "step": 3369
    },
    {
      "epoch": 0.9340354767184036,
      "grad_norm": 0.887981653213501,
      "learning_rate": 1.0787842165028539e-07,
      "loss": 0.543,
      "step": 3370
    },
    {
      "epoch": 0.9343126385809313,
      "grad_norm": 0.891690731048584,
      "learning_rate": 1.0698078832611369e-07,
      "loss": 0.5609,
      "step": 3371
    },
    {
      "epoch": 0.934589800443459,
      "grad_norm": 0.9002558588981628,
      "learning_rate": 1.0608686474286812e-07,
      "loss": 0.6617,
      "step": 3372
    },
    {
      "epoch": 0.9348669623059866,
      "grad_norm": 15.70753288269043,
      "learning_rate": 1.0519665157829483e-07,
      "loss": 0.559,
      "step": 3373
    },
    {
      "epoch": 0.9351441241685144,
      "grad_norm": 0.949597954750061,
      "learning_rate": 1.043101495073251e-07,
      "loss": 0.6567,
      "step": 3374
    },
    {
      "epoch": 0.9354212860310421,
      "grad_norm": 0.9315176606178284,
      "learning_rate": 1.0342735920207958e-07,
      "loss": 0.5845,
      "step": 3375
    },
    {
      "epoch": 0.9356984478935698,
      "grad_norm": 0.955216109752655,
      "learning_rate": 1.0254828133186346e-07,
      "loss": 0.6058,
      "step": 3376
    },
    {
      "epoch": 0.9359756097560976,
      "grad_norm": 0.9852487444877625,
      "learning_rate": 1.0167291656316691e-07,
      "loss": 0.594,
      "step": 3377
    },
    {
      "epoch": 0.9362527716186253,
      "grad_norm": 0.8792576193809509,
      "learning_rate": 1.0080126555966574e-07,
      "loss": 0.5802,
      "step": 3378
    },
    {
      "epoch": 0.936529933481153,
      "grad_norm": 0.9451266527175903,
      "learning_rate": 9.993332898221964e-08,
      "loss": 0.6355,
      "step": 3379
    },
    {
      "epoch": 0.9368070953436807,
      "grad_norm": 0.911716878414154,
      "learning_rate": 9.906910748887222e-08,
      "loss": 0.6101,
      "step": 3380
    },
    {
      "epoch": 0.9370842572062085,
      "grad_norm": 0.9273044466972351,
      "learning_rate": 9.820860173484992e-08,
      "loss": 0.6073,
      "step": 3381
    },
    {
      "epoch": 0.9373614190687362,
      "grad_norm": 0.9155735373497009,
      "learning_rate": 9.735181237256252e-08,
      "loss": 0.6335,
      "step": 3382
    },
    {
      "epoch": 0.9376385809312638,
      "grad_norm": 0.9472571015357971,
      "learning_rate": 9.649874005160264e-08,
      "loss": 0.5595,
      "step": 3383
    },
    {
      "epoch": 0.9379157427937915,
      "grad_norm": 0.9151954650878906,
      "learning_rate": 9.564938541874457e-08,
      "loss": 0.5903,
      "step": 3384
    },
    {
      "epoch": 0.9381929046563193,
      "grad_norm": 0.8622840046882629,
      "learning_rate": 9.480374911794265e-08,
      "loss": 0.6163,
      "step": 3385
    },
    {
      "epoch": 0.938470066518847,
      "grad_norm": 0.9745652079582214,
      "learning_rate": 9.396183179033292e-08,
      "loss": 0.6045,
      "step": 3386
    },
    {
      "epoch": 0.9387472283813747,
      "grad_norm": 0.8505103588104248,
      "learning_rate": 9.312363407423253e-08,
      "loss": 0.5403,
      "step": 3387
    },
    {
      "epoch": 0.9390243902439024,
      "grad_norm": 0.9271870255470276,
      "learning_rate": 9.228915660513815e-08,
      "loss": 0.6057,
      "step": 3388
    },
    {
      "epoch": 0.9393015521064302,
      "grad_norm": 0.9313729405403137,
      "learning_rate": 9.145840001572537e-08,
      "loss": 0.5956,
      "step": 3389
    },
    {
      "epoch": 0.9395787139689579,
      "grad_norm": 1.0042086839675903,
      "learning_rate": 9.063136493584868e-08,
      "loss": 0.5936,
      "step": 3390
    },
    {
      "epoch": 0.9398558758314856,
      "grad_norm": 0.8627570271492004,
      "learning_rate": 8.980805199254206e-08,
      "loss": 0.5659,
      "step": 3391
    },
    {
      "epoch": 0.9401330376940134,
      "grad_norm": 1.1236099004745483,
      "learning_rate": 8.898846181001619e-08,
      "loss": 0.5667,
      "step": 3392
    },
    {
      "epoch": 0.940410199556541,
      "grad_norm": 0.9118255376815796,
      "learning_rate": 8.817259500965958e-08,
      "loss": 0.6271,
      "step": 3393
    },
    {
      "epoch": 0.9406873614190687,
      "grad_norm": 0.8905192017555237,
      "learning_rate": 8.736045221003853e-08,
      "loss": 0.5894,
      "step": 3394
    },
    {
      "epoch": 0.9409645232815964,
      "grad_norm": 0.891627311706543,
      "learning_rate": 8.655203402689605e-08,
      "loss": 0.5978,
      "step": 3395
    },
    {
      "epoch": 0.9412416851441242,
      "grad_norm": 0.8852179050445557,
      "learning_rate": 8.574734107314963e-08,
      "loss": 0.6133,
      "step": 3396
    },
    {
      "epoch": 0.9415188470066519,
      "grad_norm": 0.9906052350997925,
      "learning_rate": 8.49463739588935e-08,
      "loss": 0.5974,
      "step": 3397
    },
    {
      "epoch": 0.9417960088691796,
      "grad_norm": 0.890288233757019,
      "learning_rate": 8.414913329139796e-08,
      "loss": 0.6484,
      "step": 3398
    },
    {
      "epoch": 0.9420731707317073,
      "grad_norm": 1.7841448783874512,
      "learning_rate": 8.335561967510563e-08,
      "loss": 0.5843,
      "step": 3399
    },
    {
      "epoch": 0.9423503325942351,
      "grad_norm": 0.9049604535102844,
      "learning_rate": 8.256583371163585e-08,
      "loss": 0.6002,
      "step": 3400
    },
    {
      "epoch": 0.9426274944567627,
      "grad_norm": 0.9577404856681824,
      "learning_rate": 8.177977599978071e-08,
      "loss": 0.6505,
      "step": 3401
    },
    {
      "epoch": 0.9429046563192904,
      "grad_norm": 0.8825478553771973,
      "learning_rate": 8.099744713550517e-08,
      "loss": 0.5554,
      "step": 3402
    },
    {
      "epoch": 0.9431818181818182,
      "grad_norm": 0.8711130023002625,
      "learning_rate": 8.02188477119481e-08,
      "loss": 0.5612,
      "step": 3403
    },
    {
      "epoch": 0.9434589800443459,
      "grad_norm": 1.2089091539382935,
      "learning_rate": 7.944397831941952e-08,
      "loss": 0.633,
      "step": 3404
    },
    {
      "epoch": 0.9437361419068736,
      "grad_norm": 0.8960710167884827,
      "learning_rate": 7.867283954540284e-08,
      "loss": 0.5413,
      "step": 3405
    },
    {
      "epoch": 0.9440133037694013,
      "grad_norm": 0.9669003486633301,
      "learning_rate": 7.790543197455313e-08,
      "loss": 0.635,
      "step": 3406
    },
    {
      "epoch": 0.9442904656319291,
      "grad_norm": 0.913794755935669,
      "learning_rate": 7.714175618869446e-08,
      "loss": 0.6196,
      "step": 3407
    },
    {
      "epoch": 0.9445676274944568,
      "grad_norm": 0.8649624586105347,
      "learning_rate": 7.63818127668231e-08,
      "loss": 0.5733,
      "step": 3408
    },
    {
      "epoch": 0.9448447893569845,
      "grad_norm": 0.8951343894004822,
      "learning_rate": 7.562560228510652e-08,
      "loss": 0.5837,
      "step": 3409
    },
    {
      "epoch": 0.9451219512195121,
      "grad_norm": 0.9343759417533875,
      "learning_rate": 7.48731253168805e-08,
      "loss": 0.6516,
      "step": 3410
    },
    {
      "epoch": 0.9453991130820399,
      "grad_norm": 0.8966418504714966,
      "learning_rate": 7.41243824326504e-08,
      "loss": 0.6207,
      "step": 3411
    },
    {
      "epoch": 0.9456762749445676,
      "grad_norm": 1.1440954208374023,
      "learning_rate": 7.337937420009044e-08,
      "loss": 0.6094,
      "step": 3412
    },
    {
      "epoch": 0.9459534368070953,
      "grad_norm": 1.0098376274108887,
      "learning_rate": 7.263810118404435e-08,
      "loss": 0.6246,
      "step": 3413
    },
    {
      "epoch": 0.9462305986696231,
      "grad_norm": 1.3510669469833374,
      "learning_rate": 7.1900563946522e-08,
      "loss": 0.6335,
      "step": 3414
    },
    {
      "epoch": 0.9465077605321508,
      "grad_norm": 1.03458833694458,
      "learning_rate": 7.116676304670334e-08,
      "loss": 0.6316,
      "step": 3415
    },
    {
      "epoch": 0.9467849223946785,
      "grad_norm": 0.8990644216537476,
      "learning_rate": 7.043669904093386e-08,
      "loss": 0.5875,
      "step": 3416
    },
    {
      "epoch": 0.9470620842572062,
      "grad_norm": 0.8902076482772827,
      "learning_rate": 6.971037248272583e-08,
      "loss": 0.5811,
      "step": 3417
    },
    {
      "epoch": 0.947339246119734,
      "grad_norm": 0.8797398805618286,
      "learning_rate": 6.898778392275873e-08,
      "loss": 0.6079,
      "step": 3418
    },
    {
      "epoch": 0.9476164079822617,
      "grad_norm": 0.9325666427612305,
      "learning_rate": 6.826893390887712e-08,
      "loss": 0.6057,
      "step": 3419
    },
    {
      "epoch": 0.9478935698447893,
      "grad_norm": 0.9186514019966125,
      "learning_rate": 6.755382298609281e-08,
      "loss": 0.6165,
      "step": 3420
    },
    {
      "epoch": 0.948170731707317,
      "grad_norm": 0.9292814135551453,
      "learning_rate": 6.684245169657988e-08,
      "loss": 0.6012,
      "step": 3421
    },
    {
      "epoch": 0.9484478935698448,
      "grad_norm": 2.4947423934936523,
      "learning_rate": 6.613482057968023e-08,
      "loss": 0.5954,
      "step": 3422
    },
    {
      "epoch": 0.9487250554323725,
      "grad_norm": 0.8948094248771667,
      "learning_rate": 6.543093017189805e-08,
      "loss": 0.577,
      "step": 3423
    },
    {
      "epoch": 0.9490022172949002,
      "grad_norm": 0.9272820353507996,
      "learning_rate": 6.4730781006902e-08,
      "loss": 0.6198,
      "step": 3424
    },
    {
      "epoch": 0.9492793791574279,
      "grad_norm": 0.9738540053367615,
      "learning_rate": 6.403437361552412e-08,
      "loss": 0.5773,
      "step": 3425
    },
    {
      "epoch": 0.9495565410199557,
      "grad_norm": 0.9050450921058655,
      "learning_rate": 6.334170852575926e-08,
      "loss": 0.6166,
      "step": 3426
    },
    {
      "epoch": 0.9498337028824834,
      "grad_norm": 0.9083531498908997,
      "learning_rate": 6.265278626276683e-08,
      "loss": 0.6087,
      "step": 3427
    },
    {
      "epoch": 0.950110864745011,
      "grad_norm": 4.646412372589111,
      "learning_rate": 6.196760734886564e-08,
      "loss": 0.5743,
      "step": 3428
    },
    {
      "epoch": 0.9503880266075388,
      "grad_norm": 0.9124096035957336,
      "learning_rate": 6.12861723035385e-08,
      "loss": 0.5844,
      "step": 3429
    },
    {
      "epoch": 0.9506651884700665,
      "grad_norm": 0.9227046966552734,
      "learning_rate": 6.060848164342881e-08,
      "loss": 0.5669,
      "step": 3430
    },
    {
      "epoch": 0.9509423503325942,
      "grad_norm": 0.998755693435669,
      "learning_rate": 5.993453588234166e-08,
      "loss": 0.6087,
      "step": 3431
    },
    {
      "epoch": 0.9512195121951219,
      "grad_norm": 0.8861397504806519,
      "learning_rate": 5.926433553124278e-08,
      "loss": 0.5942,
      "step": 3432
    },
    {
      "epoch": 0.9514966740576497,
      "grad_norm": 0.9331203699111938,
      "learning_rate": 5.8597881098257924e-08,
      "loss": 0.5913,
      "step": 3433
    },
    {
      "epoch": 0.9517738359201774,
      "grad_norm": 0.8650485873222351,
      "learning_rate": 5.7935173088671804e-08,
      "loss": 0.5736,
      "step": 3434
    },
    {
      "epoch": 0.9520509977827051,
      "grad_norm": 1.031334400177002,
      "learning_rate": 5.7276212004931384e-08,
      "loss": 0.683,
      "step": 3435
    },
    {
      "epoch": 0.9523281596452328,
      "grad_norm": 1.1049033403396606,
      "learning_rate": 5.662099834664092e-08,
      "loss": 0.5853,
      "step": 3436
    },
    {
      "epoch": 0.9526053215077606,
      "grad_norm": 0.9087440371513367,
      "learning_rate": 5.5969532610563034e-08,
      "loss": 0.6085,
      "step": 3437
    },
    {
      "epoch": 0.9528824833702882,
      "grad_norm": 1.0402244329452515,
      "learning_rate": 5.5321815290620976e-08,
      "loss": 0.5807,
      "step": 3438
    },
    {
      "epoch": 0.9531596452328159,
      "grad_norm": 2.851515531539917,
      "learning_rate": 5.4677846877893017e-08,
      "loss": 0.6104,
      "step": 3439
    },
    {
      "epoch": 0.9534368070953437,
      "grad_norm": 0.9116148352622986,
      "learning_rate": 5.4037627860617504e-08,
      "loss": 0.6262,
      "step": 3440
    },
    {
      "epoch": 0.9537139689578714,
      "grad_norm": 0.8848214149475098,
      "learning_rate": 5.340115872418949e-08,
      "loss": 0.601,
      "step": 3441
    },
    {
      "epoch": 0.9539911308203991,
      "grad_norm": 1.620965838432312,
      "learning_rate": 5.276843995116076e-08,
      "loss": 0.5903,
      "step": 3442
    },
    {
      "epoch": 0.9542682926829268,
      "grad_norm": 0.9350171685218811,
      "learning_rate": 5.213947202124037e-08,
      "loss": 0.6273,
      "step": 3443
    },
    {
      "epoch": 0.9545454545454546,
      "grad_norm": 0.9030200242996216,
      "learning_rate": 5.151425541129185e-08,
      "loss": 0.6245,
      "step": 3444
    },
    {
      "epoch": 0.9548226164079823,
      "grad_norm": 0.9086594581604004,
      "learning_rate": 5.089279059533658e-08,
      "loss": 0.5825,
      "step": 3445
    },
    {
      "epoch": 0.95509977827051,
      "grad_norm": 1.034669280052185,
      "learning_rate": 5.0275078044549893e-08,
      "loss": 0.6004,
      "step": 3446
    },
    {
      "epoch": 0.9553769401330376,
      "grad_norm": 0.876911461353302,
      "learning_rate": 4.9661118227263825e-08,
      "loss": 0.592,
      "step": 3447
    },
    {
      "epoch": 0.9556541019955654,
      "grad_norm": 0.8943122029304504,
      "learning_rate": 4.9050911608963804e-08,
      "loss": 0.6481,
      "step": 3448
    },
    {
      "epoch": 0.9559312638580931,
      "grad_norm": 0.9308010339736938,
      "learning_rate": 4.8444458652290306e-08,
      "loss": 0.6272,
      "step": 3449
    },
    {
      "epoch": 0.9562084257206208,
      "grad_norm": 0.9264365434646606,
      "learning_rate": 4.784175981703831e-08,
      "loss": 0.6265,
      "step": 3450
    },
    {
      "epoch": 0.9564855875831486,
      "grad_norm": 0.8741782307624817,
      "learning_rate": 4.724281556015564e-08,
      "loss": 0.5969,
      "step": 3451
    },
    {
      "epoch": 0.9567627494456763,
      "grad_norm": 0.9619364142417908,
      "learning_rate": 4.664762633574349e-08,
      "loss": 0.5851,
      "step": 3452
    },
    {
      "epoch": 0.957039911308204,
      "grad_norm": 0.9681580066680908,
      "learning_rate": 4.6056192595057004e-08,
      "loss": 0.6009,
      "step": 3453
    },
    {
      "epoch": 0.9573170731707317,
      "grad_norm": 0.9815784096717834,
      "learning_rate": 4.546851478650416e-08,
      "loss": 0.6128,
      "step": 3454
    },
    {
      "epoch": 0.9575942350332595,
      "grad_norm": 1.1929432153701782,
      "learning_rate": 4.48845933556441e-08,
      "loss": 0.593,
      "step": 3455
    },
    {
      "epoch": 0.9578713968957872,
      "grad_norm": 1.0790307521820068,
      "learning_rate": 4.4304428745188255e-08,
      "loss": 0.5898,
      "step": 3456
    },
    {
      "epoch": 0.9581485587583148,
      "grad_norm": 0.8950027823448181,
      "learning_rate": 4.3728021395000874e-08,
      "loss": 0.5683,
      "step": 3457
    },
    {
      "epoch": 0.9584257206208425,
      "grad_norm": 0.8418469429016113,
      "learning_rate": 4.315537174209627e-08,
      "loss": 0.5807,
      "step": 3458
    },
    {
      "epoch": 0.9587028824833703,
      "grad_norm": 0.9923114776611328,
      "learning_rate": 4.2586480220639936e-08,
      "loss": 0.616,
      "step": 3459
    },
    {
      "epoch": 0.958980044345898,
      "grad_norm": 1.1454704999923706,
      "learning_rate": 4.202134726194962e-08,
      "loss": 0.5758,
      "step": 3460
    },
    {
      "epoch": 0.9592572062084257,
      "grad_norm": 0.9801614880561829,
      "learning_rate": 4.145997329449092e-08,
      "loss": 0.5548,
      "step": 3461
    },
    {
      "epoch": 0.9595343680709535,
      "grad_norm": 0.8844665288925171,
      "learning_rate": 4.090235874388171e-08,
      "loss": 0.6046,
      "step": 3462
    },
    {
      "epoch": 0.9598115299334812,
      "grad_norm": 0.8907404541969299,
      "learning_rate": 4.034850403288826e-08,
      "loss": 0.5655,
      "step": 3463
    },
    {
      "epoch": 0.9600886917960089,
      "grad_norm": 0.958256721496582,
      "learning_rate": 3.979840958142689e-08,
      "loss": 0.5685,
      "step": 3464
    },
    {
      "epoch": 0.9603658536585366,
      "grad_norm": 0.993432879447937,
      "learning_rate": 3.9252075806562316e-08,
      "loss": 0.6319,
      "step": 3465
    },
    {
      "epoch": 0.9606430155210643,
      "grad_norm": 0.9005811214447021,
      "learning_rate": 3.8709503122509873e-08,
      "loss": 0.6667,
      "step": 3466
    },
    {
      "epoch": 0.960920177383592,
      "grad_norm": 0.9536755681037903,
      "learning_rate": 3.817069194062939e-08,
      "loss": 0.6019,
      "step": 3467
    },
    {
      "epoch": 0.9611973392461197,
      "grad_norm": 0.8933367133140564,
      "learning_rate": 3.7635642669433536e-08,
      "loss": 0.5884,
      "step": 3468
    },
    {
      "epoch": 0.9614745011086474,
      "grad_norm": 0.9920310974121094,
      "learning_rate": 3.7104355714580595e-08,
      "loss": 0.5942,
      "step": 3469
    },
    {
      "epoch": 0.9617516629711752,
      "grad_norm": 0.9644724130630493,
      "learning_rate": 3.657683147887503e-08,
      "loss": 0.6128,
      "step": 3470
    },
    {
      "epoch": 0.9620288248337029,
      "grad_norm": 1.1009955406188965,
      "learning_rate": 3.605307036227135e-08,
      "loss": 0.5485,
      "step": 3471
    },
    {
      "epoch": 0.9623059866962306,
      "grad_norm": 0.9142611026763916,
      "learning_rate": 3.5533072761869126e-08,
      "loss": 0.5898,
      "step": 3472
    },
    {
      "epoch": 0.9625831485587583,
      "grad_norm": 0.9762865304946899,
      "learning_rate": 3.5016839071914666e-08,
      "loss": 0.599,
      "step": 3473
    },
    {
      "epoch": 0.9628603104212861,
      "grad_norm": 1.0416306257247925,
      "learning_rate": 3.4504369683801e-08,
      "loss": 0.5672,
      "step": 3474
    },
    {
      "epoch": 0.9631374722838137,
      "grad_norm": 1.8042806386947632,
      "learning_rate": 3.3995664986067325e-08,
      "loss": 0.6224,
      "step": 3475
    },
    {
      "epoch": 0.9634146341463414,
      "grad_norm": 0.8975894451141357,
      "learning_rate": 3.3490725364399014e-08,
      "loss": 0.6328,
      "step": 3476
    },
    {
      "epoch": 0.9636917960088692,
      "grad_norm": 0.890964686870575,
      "learning_rate": 3.2989551201624836e-08,
      "loss": 0.5952,
      "step": 3477
    },
    {
      "epoch": 0.9639689578713969,
      "grad_norm": 0.9686786532402039,
      "learning_rate": 3.249214287772085e-08,
      "loss": 0.6064,
      "step": 3478
    },
    {
      "epoch": 0.9642461197339246,
      "grad_norm": 0.9028735756874084,
      "learning_rate": 3.199850076980704e-08,
      "loss": 0.5872,
      "step": 3479
    },
    {
      "epoch": 0.9645232815964523,
      "grad_norm": 0.9032389521598816,
      "learning_rate": 3.150862525214793e-08,
      "loss": 0.623,
      "step": 3480
    },
    {
      "epoch": 0.9648004434589801,
      "grad_norm": 1.3688398599624634,
      "learning_rate": 3.102251669615253e-08,
      "loss": 0.5351,
      "step": 3481
    },
    {
      "epoch": 0.9650776053215078,
      "grad_norm": 0.8587451577186584,
      "learning_rate": 3.0540175470374356e-08,
      "loss": 0.596,
      "step": 3482
    },
    {
      "epoch": 0.9653547671840355,
      "grad_norm": 0.8981419205665588,
      "learning_rate": 3.006160194050978e-08,
      "loss": 0.5958,
      "step": 3483
    },
    {
      "epoch": 0.9656319290465631,
      "grad_norm": 0.8762111663818359,
      "learning_rate": 2.9586796469398548e-08,
      "loss": 0.5742,
      "step": 3484
    },
    {
      "epoch": 0.9659090909090909,
      "grad_norm": 1.1123936176300049,
      "learning_rate": 2.911575941702438e-08,
      "loss": 0.5581,
      "step": 3485
    },
    {
      "epoch": 0.9661862527716186,
      "grad_norm": 1.7221845388412476,
      "learning_rate": 2.8648491140513267e-08,
      "loss": 0.6571,
      "step": 3486
    },
    {
      "epoch": 0.9664634146341463,
      "grad_norm": 0.8611127734184265,
      "learning_rate": 2.8184991994134603e-08,
      "loss": 0.5669,
      "step": 3487
    },
    {
      "epoch": 0.9667405764966741,
      "grad_norm": 0.846963107585907,
      "learning_rate": 2.7725262329298953e-08,
      "loss": 0.565,
      "step": 3488
    },
    {
      "epoch": 0.9670177383592018,
      "grad_norm": 0.9561575055122375,
      "learning_rate": 2.7269302494559725e-08,
      "loss": 0.5839,
      "step": 3489
    },
    {
      "epoch": 0.9672949002217295,
      "grad_norm": 2.289795160293579,
      "learning_rate": 2.681711283561206e-08,
      "loss": 0.6333,
      "step": 3490
    },
    {
      "epoch": 0.9675720620842572,
      "grad_norm": 0.9678146839141846,
      "learning_rate": 2.6368693695293378e-08,
      "loss": 0.623,
      "step": 3491
    },
    {
      "epoch": 0.967849223946785,
      "grad_norm": 1.0525860786437988,
      "learning_rate": 2.5924045413580624e-08,
      "loss": 0.6763,
      "step": 3492
    },
    {
      "epoch": 0.9681263858093127,
      "grad_norm": 0.887487530708313,
      "learning_rate": 2.548316832759301e-08,
      "loss": 0.5804,
      "step": 3493
    },
    {
      "epoch": 0.9684035476718403,
      "grad_norm": 0.8936427235603333,
      "learning_rate": 2.504606277159094e-08,
      "loss": 0.6112,
      "step": 3494
    },
    {
      "epoch": 0.968680709534368,
      "grad_norm": 0.8722772002220154,
      "learning_rate": 2.4612729076974317e-08,
      "loss": 0.5557,
      "step": 3495
    },
    {
      "epoch": 0.9689578713968958,
      "grad_norm": 0.8424680829048157,
      "learning_rate": 2.4183167572283672e-08,
      "loss": 0.5861,
      "step": 3496
    },
    {
      "epoch": 0.9692350332594235,
      "grad_norm": 1.2040367126464844,
      "learning_rate": 2.375737858320015e-08,
      "loss": 0.5955,
      "step": 3497
    },
    {
      "epoch": 0.9695121951219512,
      "grad_norm": 0.9035731554031372,
      "learning_rate": 2.333536243254442e-08,
      "loss": 0.5707,
      "step": 3498
    },
    {
      "epoch": 0.969789356984479,
      "grad_norm": 0.8816320896148682,
      "learning_rate": 2.2917119440275524e-08,
      "loss": 0.5958,
      "step": 3499
    },
    {
      "epoch": 0.9700665188470067,
      "grad_norm": 0.8713551759719849,
      "learning_rate": 2.250264992349316e-08,
      "loss": 0.5804,
      "step": 3500
    },
    {
      "epoch": 0.9703436807095344,
      "grad_norm": 0.8806124925613403,
      "learning_rate": 2.2091954196436505e-08,
      "loss": 0.5746,
      "step": 3501
    },
    {
      "epoch": 0.970620842572062,
      "grad_norm": 0.9343396425247192,
      "learning_rate": 2.1685032570482046e-08,
      "loss": 0.5996,
      "step": 3502
    },
    {
      "epoch": 0.9708980044345898,
      "grad_norm": 0.8904646039009094,
      "learning_rate": 2.1281885354146326e-08,
      "loss": 0.6244,
      "step": 3503
    },
    {
      "epoch": 0.9711751662971175,
      "grad_norm": 0.9168252944946289,
      "learning_rate": 2.0882512853082624e-08,
      "loss": 0.5636,
      "step": 3504
    },
    {
      "epoch": 0.9714523281596452,
      "grad_norm": 1.5927091836929321,
      "learning_rate": 2.0486915370083183e-08,
      "loss": 0.6142,
      "step": 3505
    },
    {
      "epoch": 0.9717294900221729,
      "grad_norm": 0.964887797832489,
      "learning_rate": 2.0095093205079186e-08,
      "loss": 0.6077,
      "step": 3506
    },
    {
      "epoch": 0.9720066518847007,
      "grad_norm": 1.6768697500228882,
      "learning_rate": 1.970704665513745e-08,
      "loss": 0.612,
      "step": 3507
    },
    {
      "epoch": 0.9722838137472284,
      "grad_norm": 1.155440330505371,
      "learning_rate": 1.932277601446375e-08,
      "loss": 0.6281,
      "step": 3508
    },
    {
      "epoch": 0.9725609756097561,
      "grad_norm": 0.8378085494041443,
      "learning_rate": 1.8942281574401145e-08,
      "loss": 0.5591,
      "step": 3509
    },
    {
      "epoch": 0.9728381374722838,
      "grad_norm": 0.9334925413131714,
      "learning_rate": 1.8565563623428318e-08,
      "loss": 0.5635,
      "step": 3510
    },
    {
      "epoch": 0.9731152993348116,
      "grad_norm": 0.9063448905944824,
      "learning_rate": 1.81926224471618e-08,
      "loss": 0.575,
      "step": 3511
    },
    {
      "epoch": 0.9733924611973392,
      "grad_norm": 0.9131854772567749,
      "learning_rate": 1.7823458328354305e-08,
      "loss": 0.6015,
      "step": 3512
    },
    {
      "epoch": 0.9736696230598669,
      "grad_norm": 0.9752041697502136,
      "learning_rate": 1.7458071546895272e-08,
      "loss": 0.5823,
      "step": 3513
    },
    {
      "epoch": 0.9739467849223947,
      "grad_norm": 0.9722270369529724,
      "learning_rate": 1.7096462379809773e-08,
      "loss": 0.641,
      "step": 3514
    },
    {
      "epoch": 0.9742239467849224,
      "grad_norm": 0.8641842007637024,
      "learning_rate": 1.6738631101259616e-08,
      "loss": 0.5567,
      "step": 3515
    },
    {
      "epoch": 0.9745011086474501,
      "grad_norm": 0.9012433290481567,
      "learning_rate": 1.6384577982541118e-08,
      "loss": 0.6583,
      "step": 3516
    },
    {
      "epoch": 0.9747782705099778,
      "grad_norm": 0.8677822351455688,
      "learning_rate": 1.6034303292087328e-08,
      "loss": 0.5521,
      "step": 3517
    },
    {
      "epoch": 0.9750554323725056,
      "grad_norm": 0.9607548713684082,
      "learning_rate": 1.5687807295465817e-08,
      "loss": 0.5919,
      "step": 3518
    },
    {
      "epoch": 0.9753325942350333,
      "grad_norm": 0.9430091381072998,
      "learning_rate": 1.5345090255379224e-08,
      "loss": 0.5977,
      "step": 3519
    },
    {
      "epoch": 0.975609756097561,
      "grad_norm": 0.9004371762275696,
      "learning_rate": 1.5006152431665812e-08,
      "loss": 0.5363,
      "step": 3520
    },
    {
      "epoch": 0.9758869179600886,
      "grad_norm": 1.0419788360595703,
      "learning_rate": 1.4670994081297796e-08,
      "loss": 0.6664,
      "step": 3521
    },
    {
      "epoch": 0.9761640798226164,
      "grad_norm": 0.8517856597900391,
      "learning_rate": 1.4339615458382472e-08,
      "loss": 0.6094,
      "step": 3522
    },
    {
      "epoch": 0.9764412416851441,
      "grad_norm": 0.934958815574646,
      "learning_rate": 1.4012016814160533e-08,
      "loss": 0.5811,
      "step": 3523
    },
    {
      "epoch": 0.9767184035476718,
      "grad_norm": 0.8463792204856873,
      "learning_rate": 1.3688198397008856e-08,
      "loss": 0.5809,
      "step": 3524
    },
    {
      "epoch": 0.9769955654101996,
      "grad_norm": 0.8876455426216125,
      "learning_rate": 1.3368160452435498e-08,
      "loss": 0.5875,
      "step": 3525
    },
    {
      "epoch": 0.9772727272727273,
      "grad_norm": 1.6891173124313354,
      "learning_rate": 1.3051903223084694e-08,
      "loss": 0.6221,
      "step": 3526
    },
    {
      "epoch": 0.977549889135255,
      "grad_norm": 0.8838277459144592,
      "learning_rate": 1.2739426948732426e-08,
      "loss": 0.5799,
      "step": 3527
    },
    {
      "epoch": 0.9778270509977827,
      "grad_norm": 0.9114553928375244,
      "learning_rate": 1.243073186628918e-08,
      "loss": 0.6191,
      "step": 3528
    },
    {
      "epoch": 0.9781042128603105,
      "grad_norm": 0.9148756265640259,
      "learning_rate": 1.21258182097983e-08,
      "loss": 0.5662,
      "step": 3529
    },
    {
      "epoch": 0.9783813747228381,
      "grad_norm": 0.9138510823249817,
      "learning_rate": 1.182468621043653e-08,
      "loss": 0.6247,
      "step": 3530
    },
    {
      "epoch": 0.9786585365853658,
      "grad_norm": 0.9224982261657715,
      "learning_rate": 1.1527336096512354e-08,
      "loss": 0.6312,
      "step": 3531
    },
    {
      "epoch": 0.9789356984478935,
      "grad_norm": 0.9430356025695801,
      "learning_rate": 1.1233768093468766e-08,
      "loss": 0.5634,
      "step": 3532
    },
    {
      "epoch": 0.9792128603104213,
      "grad_norm": 0.9138108491897583,
      "learning_rate": 1.0943982423879951e-08,
      "loss": 0.5574,
      "step": 3533
    },
    {
      "epoch": 0.979490022172949,
      "grad_norm": 0.9977015852928162,
      "learning_rate": 1.0657979307451827e-08,
      "loss": 0.5855,
      "step": 3534
    },
    {
      "epoch": 0.9797671840354767,
      "grad_norm": 0.9731780886650085,
      "learning_rate": 1.0375758961024274e-08,
      "loss": 0.5948,
      "step": 3535
    },
    {
      "epoch": 0.9800443458980045,
      "grad_norm": 0.8592071533203125,
      "learning_rate": 1.009732159856891e-08,
      "loss": 0.5878,
      "step": 3536
    },
    {
      "epoch": 0.9803215077605322,
      "grad_norm": 0.9075479507446289,
      "learning_rate": 9.822667431186873e-09,
      "loss": 0.5794,
      "step": 3537
    },
    {
      "epoch": 0.9805986696230599,
      "grad_norm": 0.9515004754066467,
      "learning_rate": 9.551796667114366e-09,
      "loss": 0.5777,
      "step": 3538
    },
    {
      "epoch": 0.9808758314855875,
      "grad_norm": 0.9165856838226318,
      "learning_rate": 9.28470951171656e-09,
      "loss": 0.5386,
      "step": 3539
    },
    {
      "epoch": 0.9811529933481153,
      "grad_norm": 1.0869779586791992,
      "learning_rate": 9.021406167492031e-09,
      "loss": 0.587,
      "step": 3540
    },
    {
      "epoch": 0.981430155210643,
      "grad_norm": 1.093483567237854,
      "learning_rate": 8.761886834067756e-09,
      "loss": 0.6827,
      "step": 3541
    },
    {
      "epoch": 0.9817073170731707,
      "grad_norm": 0.952051043510437,
      "learning_rate": 8.506151708205235e-09,
      "loss": 0.6173,
      "step": 3542
    },
    {
      "epoch": 0.9819844789356984,
      "grad_norm": 0.88568514585495,
      "learning_rate": 8.254200983794369e-09,
      "loss": 0.5771,
      "step": 3543
    },
    {
      "epoch": 0.9822616407982262,
      "grad_norm": 0.9030840992927551,
      "learning_rate": 8.006034851856803e-09,
      "loss": 0.5674,
      "step": 3544
    },
    {
      "epoch": 0.9825388026607539,
      "grad_norm": 0.8676204085350037,
      "learning_rate": 7.761653500544252e-09,
      "loss": 0.6068,
      "step": 3545
    },
    {
      "epoch": 0.9828159645232816,
      "grad_norm": 0.9229354858398438,
      "learning_rate": 7.521057115140173e-09,
      "loss": 0.6368,
      "step": 3546
    },
    {
      "epoch": 0.9830931263858093,
      "grad_norm": 0.923760712146759,
      "learning_rate": 7.284245878057539e-09,
      "loss": 0.5673,
      "step": 3547
    },
    {
      "epoch": 0.983370288248337,
      "grad_norm": 0.9230203032493591,
      "learning_rate": 7.05121996883884e-09,
      "loss": 0.6056,
      "step": 3548
    },
    {
      "epoch": 0.9836474501108647,
      "grad_norm": 1.07217538356781,
      "learning_rate": 6.821979564157199e-09,
      "loss": 0.5809,
      "step": 3549
    },
    {
      "epoch": 0.9839246119733924,
      "grad_norm": 0.8872886896133423,
      "learning_rate": 6.59652483781692e-09,
      "loss": 0.55,
      "step": 3550
    },
    {
      "epoch": 0.9842017738359202,
      "grad_norm": 0.9426319599151611,
      "learning_rate": 6.37485596075016e-09,
      "loss": 0.609,
      "step": 3551
    },
    {
      "epoch": 0.9844789356984479,
      "grad_norm": 0.8695278763771057,
      "learning_rate": 6.1569731010202585e-09,
      "loss": 0.56,
      "step": 3552
    },
    {
      "epoch": 0.9847560975609756,
      "grad_norm": 1.0143616199493408,
      "learning_rate": 5.942876423818966e-09,
      "loss": 0.6297,
      "step": 3553
    },
    {
      "epoch": 0.9850332594235033,
      "grad_norm": 0.860474705696106,
      "learning_rate": 5.732566091468106e-09,
      "loss": 0.5944,
      "step": 3554
    },
    {
      "epoch": 0.9853104212860311,
      "grad_norm": 0.8931911587715149,
      "learning_rate": 5.526042263419018e-09,
      "loss": 0.5925,
      "step": 3555
    },
    {
      "epoch": 0.9855875831485588,
      "grad_norm": 0.9927111268043518,
      "learning_rate": 5.323305096251452e-09,
      "loss": 0.5881,
      "step": 3556
    },
    {
      "epoch": 0.9858647450110865,
      "grad_norm": 0.9674209356307983,
      "learning_rate": 5.1243547436746754e-09,
      "loss": 0.6359,
      "step": 3557
    },
    {
      "epoch": 0.9861419068736141,
      "grad_norm": 1.0103023052215576,
      "learning_rate": 4.929191356527474e-09,
      "loss": 0.5925,
      "step": 3558
    },
    {
      "epoch": 0.9864190687361419,
      "grad_norm": 0.8641563057899475,
      "learning_rate": 4.737815082776487e-09,
      "loss": 0.5894,
      "step": 3559
    },
    {
      "epoch": 0.9866962305986696,
      "grad_norm": 0.9298277497291565,
      "learning_rate": 4.550226067516761e-09,
      "loss": 0.5675,
      "step": 3560
    },
    {
      "epoch": 0.9869733924611973,
      "grad_norm": 0.9430758357048035,
      "learning_rate": 4.366424452973972e-09,
      "loss": 0.5798,
      "step": 3561
    },
    {
      "epoch": 0.9872505543237251,
      "grad_norm": 0.9556924700737,
      "learning_rate": 4.186410378500538e-09,
      "loss": 0.6777,
      "step": 3562
    },
    {
      "epoch": 0.9875277161862528,
      "grad_norm": 0.8679664731025696,
      "learning_rate": 4.010183980577842e-09,
      "loss": 0.586,
      "step": 3563
    },
    {
      "epoch": 0.9878048780487805,
      "grad_norm": 1.6462360620498657,
      "learning_rate": 3.8377453928145625e-09,
      "loss": 0.601,
      "step": 3564
    },
    {
      "epoch": 0.9880820399113082,
      "grad_norm": 0.9156757593154907,
      "learning_rate": 3.669094745950008e-09,
      "loss": 0.5735,
      "step": 3565
    },
    {
      "epoch": 0.988359201773836,
      "grad_norm": 0.8749805092811584,
      "learning_rate": 3.50423216784912e-09,
      "loss": 0.5954,
      "step": 3566
    },
    {
      "epoch": 0.9886363636363636,
      "grad_norm": 1.1100130081176758,
      "learning_rate": 3.3431577835052464e-09,
      "loss": 0.6281,
      "step": 3567
    },
    {
      "epoch": 0.9889135254988913,
      "grad_norm": 0.9379473328590393,
      "learning_rate": 3.1858717150412554e-09,
      "loss": 0.6306,
      "step": 3568
    },
    {
      "epoch": 0.989190687361419,
      "grad_norm": 0.8439226150512695,
      "learning_rate": 3.0323740817067572e-09,
      "loss": 0.5724,
      "step": 3569
    },
    {
      "epoch": 0.9894678492239468,
      "grad_norm": 0.916608989238739,
      "learning_rate": 2.882664999878104e-09,
      "loss": 0.6127,
      "step": 3570
    },
    {
      "epoch": 0.9897450110864745,
      "grad_norm": 0.9239919781684875,
      "learning_rate": 2.7367445830611683e-09,
      "loss": 0.5782,
      "step": 3571
    },
    {
      "epoch": 0.9900221729490022,
      "grad_norm": 0.9859599471092224,
      "learning_rate": 2.5946129418880086e-09,
      "loss": 0.5734,
      "step": 3572
    },
    {
      "epoch": 0.99029933481153,
      "grad_norm": 0.9288823008537292,
      "learning_rate": 2.4562701841185366e-09,
      "loss": 0.557,
      "step": 3573
    },
    {
      "epoch": 0.9905764966740577,
      "grad_norm": 0.8267753720283508,
      "learning_rate": 2.321716414639963e-09,
      "loss": 0.5512,
      "step": 3574
    },
    {
      "epoch": 0.9908536585365854,
      "grad_norm": 0.9709101319313049,
      "learning_rate": 2.1909517354679056e-09,
      "loss": 0.6631,
      "step": 3575
    },
    {
      "epoch": 0.991130820399113,
      "grad_norm": 4.0659356117248535,
      "learning_rate": 2.06397624574306e-09,
      "loss": 0.5926,
      "step": 3576
    },
    {
      "epoch": 0.9914079822616408,
      "grad_norm": 0.9452905654907227,
      "learning_rate": 1.9407900417345304e-09,
      "loss": 0.6351,
      "step": 3577
    },
    {
      "epoch": 0.9916851441241685,
      "grad_norm": 0.9612874388694763,
      "learning_rate": 1.8213932168392733e-09,
      "loss": 0.5873,
      "step": 3578
    },
    {
      "epoch": 0.9919623059866962,
      "grad_norm": 1.1576651334762573,
      "learning_rate": 1.7057858615798783e-09,
      "loss": 0.5859,
      "step": 3579
    },
    {
      "epoch": 0.9922394678492239,
      "grad_norm": 0.9337642192840576,
      "learning_rate": 1.5939680636062327e-09,
      "loss": 0.608,
      "step": 3580
    },
    {
      "epoch": 0.9925166297117517,
      "grad_norm": 0.928565263748169,
      "learning_rate": 1.4859399076955217e-09,
      "loss": 0.5916,
      "step": 3581
    },
    {
      "epoch": 0.9927937915742794,
      "grad_norm": 0.9376745223999023,
      "learning_rate": 1.3817014757516734e-09,
      "loss": 0.6121,
      "step": 3582
    },
    {
      "epoch": 0.9930709534368071,
      "grad_norm": 0.8959884643554688,
      "learning_rate": 1.2812528468042485e-09,
      "loss": 0.6067,
      "step": 3583
    },
    {
      "epoch": 0.9933481152993349,
      "grad_norm": 0.9543491005897522,
      "learning_rate": 1.184594097011771e-09,
      "loss": 0.5866,
      "step": 3584
    },
    {
      "epoch": 0.9936252771618626,
      "grad_norm": 1.0075677633285522,
      "learning_rate": 1.091725299656732e-09,
      "loss": 0.6203,
      "step": 3585
    },
    {
      "epoch": 0.9939024390243902,
      "grad_norm": 0.8757171630859375,
      "learning_rate": 1.0026465251500305e-09,
      "loss": 0.5935,
      "step": 3586
    },
    {
      "epoch": 0.9941796008869179,
      "grad_norm": 0.8487471342086792,
      "learning_rate": 9.173578410281992e-10,
      "loss": 0.5362,
      "step": 3587
    },
    {
      "epoch": 0.9944567627494457,
      "grad_norm": 0.9618258476257324,
      "learning_rate": 8.358593119550673e-10,
      "loss": 0.6207,
      "step": 3588
    },
    {
      "epoch": 0.9947339246119734,
      "grad_norm": 0.8749884366989136,
      "learning_rate": 7.581509997206527e-10,
      "loss": 0.5586,
      "step": 3589
    },
    {
      "epoch": 0.9950110864745011,
      "grad_norm": 1.136684775352478,
      "learning_rate": 6.842329632400502e-10,
      "loss": 0.5555,
      "step": 3590
    },
    {
      "epoch": 0.9952882483370288,
      "grad_norm": 1.9581853151321411,
      "learning_rate": 6.14105258556763e-10,
      "loss": 0.596,
      "step": 3591
    },
    {
      "epoch": 0.9955654101995566,
      "grad_norm": 0.857920229434967,
      "learning_rate": 5.477679388382617e-10,
      "loss": 0.6195,
      "step": 3592
    },
    {
      "epoch": 0.9958425720620843,
      "grad_norm": 1.9047707319259644,
      "learning_rate": 4.852210543809799e-10,
      "loss": 0.6216,
      "step": 3593
    },
    {
      "epoch": 0.996119733924612,
      "grad_norm": 1.2083275318145752,
      "learning_rate": 4.2646465260476334e-10,
      "loss": 0.5819,
      "step": 3594
    },
    {
      "epoch": 0.9963968957871396,
      "grad_norm": 0.8850510716438293,
      "learning_rate": 3.7149877805786606e-10,
      "loss": 0.6185,
      "step": 3595
    },
    {
      "epoch": 0.9966740576496674,
      "grad_norm": 0.92436283826828,
      "learning_rate": 3.203234724136195e-10,
      "loss": 0.6091,
      "step": 3596
    },
    {
      "epoch": 0.9969512195121951,
      "grad_norm": 0.973864734172821,
      "learning_rate": 2.729387744715428e-10,
      "loss": 0.626,
      "step": 3597
    },
    {
      "epoch": 0.9972283813747228,
      "grad_norm": 0.9695550799369812,
      "learning_rate": 2.2934472015734287e-10,
      "loss": 0.6164,
      "step": 3598
    },
    {
      "epoch": 0.9975055432372506,
      "grad_norm": 0.9131198525428772,
      "learning_rate": 1.8954134252235911e-10,
      "loss": 0.6078,
      "step": 3599
    },
    {
      "epoch": 0.9977827050997783,
      "grad_norm": 0.842063844203949,
      "learning_rate": 1.535286717446738e-10,
      "loss": 0.6249,
      "step": 3600
    },
    {
      "epoch": 0.998059866962306,
      "grad_norm": 0.890779972076416,
      "learning_rate": 1.2130673512744661e-10,
      "loss": 0.564,
      "step": 3601
    },
    {
      "epoch": 0.9983370288248337,
      "grad_norm": 0.8939942121505737,
      "learning_rate": 9.287555710169039e-11,
      "loss": 0.5432,
      "step": 3602
    },
    {
      "epoch": 0.9986141906873615,
      "grad_norm": 0.8454970121383667,
      "learning_rate": 6.823515922182999e-11,
      "loss": 0.577,
      "step": 3603
    },
    {
      "epoch": 0.9988913525498891,
      "grad_norm": 0.8610948920249939,
      "learning_rate": 4.7385560169588285e-11,
      "loss": 0.5879,
      "step": 3604
    },
    {
      "epoch": 0.9991685144124168,
      "grad_norm": 0.9244651794433594,
      "learning_rate": 3.03267757534309e-11,
      "loss": 0.6021,
      "step": 3605
    },
    {
      "epoch": 0.9994456762749445,
      "grad_norm": 0.8801454305648804,
      "learning_rate": 1.7058818905790753e-11,
      "loss": 0.6066,
      "step": 3606
    },
    {
      "epoch": 0.9997228381374723,
      "grad_norm": 0.9328412413597107,
      "learning_rate": 7.581699686398658e-12,
      "loss": 0.6019,
      "step": 3607
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8727502822875977,
      "learning_rate": 1.8954252806180353e-12,
      "loss": 0.5698,
      "step": 3608
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.5976117849349976,
      "eval_runtime": 386.5673,
      "eval_samples_per_second": 8.356,
      "eval_steps_per_second": 1.045,
      "step": 3608
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 3608,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.812089585840685e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}