{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 1334,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014992503748125937,
      "grad_norm": 0.027869803162182624,
      "learning_rate": 1.4925373134328358e-06,
      "loss": 0.0328,
      "step": 1
    },
    {
      "epoch": 0.0029985007496251873,
      "grad_norm": 0.03034633120050999,
      "learning_rate": 2.9850746268656716e-06,
      "loss": 0.0521,
      "step": 2
    },
    {
      "epoch": 0.004497751124437781,
      "grad_norm": 0.033013192126779396,
      "learning_rate": 4.477611940298508e-06,
      "loss": 0.0481,
      "step": 3
    },
    {
      "epoch": 0.005997001499250375,
      "grad_norm": 0.02964454530876554,
      "learning_rate": 5.970149253731343e-06,
      "loss": 0.0348,
      "step": 4
    },
    {
      "epoch": 0.0074962518740629685,
      "grad_norm": 0.04545013411384094,
      "learning_rate": 7.4626865671641785e-06,
      "loss": 0.0634,
      "step": 5
    },
    {
      "epoch": 0.008995502248875561,
      "grad_norm": 0.037827090237667615,
      "learning_rate": 8.955223880597016e-06,
      "loss": 0.0511,
      "step": 6
    },
    {
      "epoch": 0.010494752623688156,
      "grad_norm": 0.04024496430992649,
      "learning_rate": 1.0447761194029851e-05,
      "loss": 0.0902,
      "step": 7
    },
    {
      "epoch": 0.01199400299850075,
      "grad_norm": 0.02784723970702267,
      "learning_rate": 1.1940298507462686e-05,
      "loss": 0.0346,
      "step": 8
    },
    {
      "epoch": 0.013493253373313344,
      "grad_norm": 0.04073024185229626,
      "learning_rate": 1.3432835820895523e-05,
      "loss": 0.0551,
      "step": 9
    },
    {
      "epoch": 0.014992503748125937,
      "grad_norm": 0.03872731587077291,
      "learning_rate": 1.4925373134328357e-05,
      "loss": 0.0457,
      "step": 10
    },
    {
      "epoch": 0.01649175412293853,
      "grad_norm": 0.036983272179927376,
      "learning_rate": 1.6417910447761194e-05,
      "loss": 0.0466,
      "step": 11
    },
    {
      "epoch": 0.017991004497751123,
      "grad_norm": 0.03450782898126308,
      "learning_rate": 1.791044776119403e-05,
      "loss": 0.0402,
      "step": 12
    },
    {
      "epoch": 0.019490254872563718,
      "grad_norm": 0.0392817087888048,
      "learning_rate": 1.9402985074626868e-05,
      "loss": 0.0679,
      "step": 13
    },
    {
      "epoch": 0.020989505247376312,
      "grad_norm": 0.044590345998053095,
      "learning_rate": 2.0895522388059702e-05,
      "loss": 0.052,
      "step": 14
    },
    {
      "epoch": 0.022488755622188907,
      "grad_norm": 0.0356527349265605,
      "learning_rate": 2.238805970149254e-05,
      "loss": 0.0468,
      "step": 15
    },
    {
      "epoch": 0.0239880059970015,
      "grad_norm": 0.03207250371592397,
      "learning_rate": 2.3880597014925373e-05,
      "loss": 0.037,
      "step": 16
    },
    {
      "epoch": 0.025487256371814093,
      "grad_norm": 0.03631797856767525,
      "learning_rate": 2.537313432835821e-05,
      "loss": 0.0606,
      "step": 17
    },
    {
      "epoch": 0.026986506746626688,
      "grad_norm": 0.042515444732217036,
      "learning_rate": 2.6865671641791047e-05,
      "loss": 0.0527,
      "step": 18
    },
    {
      "epoch": 0.02848575712143928,
      "grad_norm": 0.03859829754974758,
      "learning_rate": 2.835820895522388e-05,
      "loss": 0.0577,
      "step": 19
    },
    {
      "epoch": 0.029985007496251874,
      "grad_norm": 0.04677384374732077,
      "learning_rate": 2.9850746268656714e-05,
      "loss": 0.056,
      "step": 20
    },
    {
      "epoch": 0.031484257871064465,
      "grad_norm": 0.04751009104400587,
      "learning_rate": 3.1343283582089554e-05,
      "loss": 0.0623,
      "step": 21
    },
    {
      "epoch": 0.03298350824587706,
      "grad_norm": 0.029581915526506828,
      "learning_rate": 3.283582089552239e-05,
      "loss": 0.0468,
      "step": 22
    },
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 0.03579505353381419,
      "learning_rate": 3.432835820895522e-05,
      "loss": 0.0587,
      "step": 23
    },
    {
      "epoch": 0.035982008995502246,
      "grad_norm": 0.04751415291633006,
      "learning_rate": 3.582089552238806e-05,
      "loss": 0.1013,
      "step": 24
    },
    {
      "epoch": 0.037481259370314844,
      "grad_norm": 0.04449100113636761,
      "learning_rate": 3.73134328358209e-05,
      "loss": 0.0661,
      "step": 25
    },
    {
      "epoch": 0.038980509745127435,
      "grad_norm": 0.042733677260132046,
      "learning_rate": 3.8805970149253736e-05,
      "loss": 0.0885,
      "step": 26
    },
    {
      "epoch": 0.04047976011994003,
      "grad_norm": 0.02626605876758245,
      "learning_rate": 4.029850746268657e-05,
      "loss": 0.0423,
      "step": 27
    },
    {
      "epoch": 0.041979010494752625,
      "grad_norm": 0.03414998677776291,
      "learning_rate": 4.1791044776119404e-05,
      "loss": 0.0484,
      "step": 28
    },
    {
      "epoch": 0.043478260869565216,
      "grad_norm": 0.023465307923170507,
      "learning_rate": 4.328358208955224e-05,
      "loss": 0.0249,
      "step": 29
    },
    {
      "epoch": 0.044977511244377814,
      "grad_norm": 0.03238010472154882,
      "learning_rate": 4.477611940298508e-05,
      "loss": 0.0544,
      "step": 30
    },
    {
      "epoch": 0.046476761619190406,
      "grad_norm": 0.031312824540825665,
      "learning_rate": 4.626865671641791e-05,
      "loss": 0.046,
      "step": 31
    },
    {
      "epoch": 0.047976011994003,
      "grad_norm": 0.03219959967609331,
      "learning_rate": 4.7761194029850745e-05,
      "loss": 0.058,
      "step": 32
    },
    {
      "epoch": 0.049475262368815595,
      "grad_norm": 0.05253387545719208,
      "learning_rate": 4.9253731343283586e-05,
      "loss": 0.0935,
      "step": 33
    },
    {
      "epoch": 0.050974512743628186,
      "grad_norm": 0.04906400811846751,
      "learning_rate": 5.074626865671642e-05,
      "loss": 0.0592,
      "step": 34
    },
    {
      "epoch": 0.05247376311844078,
      "grad_norm": 0.028913111785186865,
      "learning_rate": 5.223880597014925e-05,
      "loss": 0.0422,
      "step": 35
    },
    {
      "epoch": 0.053973013493253376,
      "grad_norm": 0.036628261788675495,
      "learning_rate": 5.373134328358209e-05,
      "loss": 0.056,
      "step": 36
    },
    {
      "epoch": 0.05547226386806597,
      "grad_norm": 0.03723553635398014,
      "learning_rate": 5.5223880597014934e-05,
      "loss": 0.0514,
      "step": 37
    },
    {
      "epoch": 0.05697151424287856,
      "grad_norm": 0.030481090328732696,
      "learning_rate": 5.671641791044776e-05,
      "loss": 0.0505,
      "step": 38
    },
    {
      "epoch": 0.05847076461769116,
      "grad_norm": 0.06319461018053213,
      "learning_rate": 5.82089552238806e-05,
      "loss": 0.0932,
      "step": 39
    },
    {
      "epoch": 0.05997001499250375,
      "grad_norm": 0.04375744045965157,
      "learning_rate": 5.970149253731343e-05,
      "loss": 0.052,
      "step": 40
    },
    {
      "epoch": 0.06146926536731634,
      "grad_norm": 0.02716485170444979,
      "learning_rate": 6.119402985074628e-05,
      "loss": 0.0394,
      "step": 41
    },
    {
      "epoch": 0.06296851574212893,
      "grad_norm": 0.03380373227357743,
      "learning_rate": 6.268656716417911e-05,
      "loss": 0.0444,
      "step": 42
    },
    {
      "epoch": 0.06446776611694154,
      "grad_norm": 0.023682114706538047,
      "learning_rate": 6.417910447761194e-05,
      "loss": 0.0318,
      "step": 43
    },
    {
      "epoch": 0.06596701649175413,
      "grad_norm": 0.03717832489837232,
      "learning_rate": 6.567164179104478e-05,
      "loss": 0.0542,
      "step": 44
    },
    {
      "epoch": 0.06746626686656672,
      "grad_norm": 0.03849400793350328,
      "learning_rate": 6.716417910447762e-05,
      "loss": 0.066,
      "step": 45
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 0.039557900524499395,
      "learning_rate": 6.865671641791044e-05,
      "loss": 0.0599,
      "step": 46
    },
    {
      "epoch": 0.0704647676161919,
      "grad_norm": 0.04183512559026415,
      "learning_rate": 7.014925373134329e-05,
      "loss": 0.0559,
      "step": 47
    },
    {
      "epoch": 0.07196401799100449,
      "grad_norm": 0.023819063243281605,
      "learning_rate": 7.164179104477612e-05,
      "loss": 0.0334,
      "step": 48
    },
    {
      "epoch": 0.0734632683658171,
      "grad_norm": 0.036181513868464375,
      "learning_rate": 7.313432835820896e-05,
      "loss": 0.053,
      "step": 49
    },
    {
      "epoch": 0.07496251874062969,
      "grad_norm": 0.038961606945728214,
      "learning_rate": 7.46268656716418e-05,
      "loss": 0.0469,
      "step": 50
    },
    {
      "epoch": 0.07646176911544228,
      "grad_norm": 0.047892810949924564,
      "learning_rate": 7.611940298507463e-05,
      "loss": 0.0585,
      "step": 51
    },
    {
      "epoch": 0.07796101949025487,
      "grad_norm": 0.06267828987629875,
      "learning_rate": 7.761194029850747e-05,
      "loss": 0.0671,
      "step": 52
    },
    {
      "epoch": 0.07946026986506746,
      "grad_norm": 0.027197819167560006,
      "learning_rate": 7.910447761194029e-05,
      "loss": 0.0384,
      "step": 53
    },
    {
      "epoch": 0.08095952023988005,
      "grad_norm": 0.062135219556912424,
      "learning_rate": 8.059701492537314e-05,
      "loss": 0.0724,
      "step": 54
    },
    {
      "epoch": 0.08245877061469266,
      "grad_norm": 0.04159649809292002,
      "learning_rate": 8.208955223880597e-05,
      "loss": 0.0528,
      "step": 55
    },
    {
      "epoch": 0.08395802098950525,
      "grad_norm": 0.04806451958909231,
      "learning_rate": 8.358208955223881e-05,
      "loss": 0.0522,
      "step": 56
    },
    {
      "epoch": 0.08545727136431784,
      "grad_norm": 0.0328648844097681,
      "learning_rate": 8.507462686567164e-05,
      "loss": 0.0529,
      "step": 57
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 0.03499304511129161,
      "learning_rate": 8.656716417910447e-05,
      "loss": 0.0457,
      "step": 58
    },
    {
      "epoch": 0.08845577211394302,
      "grad_norm": 0.03632609832686457,
      "learning_rate": 8.805970149253732e-05,
      "loss": 0.0511,
      "step": 59
    },
    {
      "epoch": 0.08995502248875563,
      "grad_norm": 0.03286113779513323,
      "learning_rate": 8.955223880597016e-05,
      "loss": 0.0605,
      "step": 60
    },
    {
      "epoch": 0.09145427286356822,
      "grad_norm": 0.05626334745781654,
      "learning_rate": 9.104477611940299e-05,
      "loss": 0.0591,
      "step": 61
    },
    {
      "epoch": 0.09295352323838081,
      "grad_norm": 0.04452407950397946,
      "learning_rate": 9.253731343283582e-05,
      "loss": 0.0769,
      "step": 62
    },
    {
      "epoch": 0.0944527736131934,
      "grad_norm": 0.03625308210216529,
      "learning_rate": 9.402985074626867e-05,
      "loss": 0.0468,
      "step": 63
    },
    {
      "epoch": 0.095952023988006,
      "grad_norm": 0.029791030013327606,
      "learning_rate": 9.552238805970149e-05,
      "loss": 0.0584,
      "step": 64
    },
    {
      "epoch": 0.09745127436281859,
      "grad_norm": 0.03780831906836486,
      "learning_rate": 9.701492537313434e-05,
      "loss": 0.0333,
      "step": 65
    },
    {
      "epoch": 0.09895052473763119,
      "grad_norm": 0.032397150776764055,
      "learning_rate": 9.850746268656717e-05,
      "loss": 0.0447,
      "step": 66
    },
    {
      "epoch": 0.10044977511244378,
      "grad_norm": 0.03634540437911119,
      "learning_rate": 0.0001,
      "loss": 0.0602,
      "step": 67
    },
    {
      "epoch": 0.10194902548725637,
      "grad_norm": 0.025903306639537493,
      "learning_rate": 0.00010149253731343284,
      "loss": 0.0334,
      "step": 68
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 0.05692249372761759,
      "learning_rate": 0.00010298507462686569,
      "loss": 0.0897,
      "step": 69
    },
    {
      "epoch": 0.10494752623688156,
      "grad_norm": 0.04193568812333293,
      "learning_rate": 0.0001044776119402985,
      "loss": 0.035,
      "step": 70
    },
    {
      "epoch": 0.10644677661169415,
      "grad_norm": 0.03978190062644848,
      "learning_rate": 0.00010597014925373134,
      "loss": 0.0442,
      "step": 71
    },
    {
      "epoch": 0.10794602698650675,
      "grad_norm": 0.041177480827902055,
      "learning_rate": 0.00010746268656716419,
      "loss": 0.0695,
      "step": 72
    },
    {
      "epoch": 0.10944527736131934,
      "grad_norm": 0.044042393308047885,
      "learning_rate": 0.00010895522388059702,
      "loss": 0.061,
      "step": 73
    },
    {
      "epoch": 0.11094452773613193,
      "grad_norm": 0.027180434415964242,
      "learning_rate": 0.00011044776119402987,
      "loss": 0.0316,
      "step": 74
    },
    {
      "epoch": 0.11244377811094453,
      "grad_norm": 0.03300393578125543,
      "learning_rate": 0.00011194029850746269,
      "loss": 0.0587,
      "step": 75
    },
    {
      "epoch": 0.11394302848575712,
      "grad_norm": 0.028528364848481542,
      "learning_rate": 0.00011343283582089552,
      "loss": 0.0352,
      "step": 76
    },
    {
      "epoch": 0.11544227886056972,
      "grad_norm": 0.030828012521923,
      "learning_rate": 0.00011492537313432837,
      "loss": 0.0481,
      "step": 77
    },
    {
      "epoch": 0.11694152923538231,
      "grad_norm": 0.04083313566006794,
      "learning_rate": 0.0001164179104477612,
      "loss": 0.0604,
      "step": 78
    },
    {
      "epoch": 0.1184407796101949,
      "grad_norm": 0.025320856260068247,
      "learning_rate": 0.00011791044776119405,
      "loss": 0.0425,
      "step": 79
    },
    {
      "epoch": 0.1199400299850075,
      "grad_norm": 0.025013025974113073,
      "learning_rate": 0.00011940298507462686,
      "loss": 0.0485,
      "step": 80
    },
    {
      "epoch": 0.12143928035982009,
      "grad_norm": 0.02829433473565749,
      "learning_rate": 0.0001208955223880597,
      "loss": 0.0475,
      "step": 81
    },
    {
      "epoch": 0.12293853073463268,
      "grad_norm": 0.04936619613683576,
      "learning_rate": 0.00012238805970149255,
      "loss": 0.0766,
      "step": 82
    },
    {
      "epoch": 0.12443778110944528,
      "grad_norm": 0.06517476388380389,
      "learning_rate": 0.00012388059701492538,
      "loss": 0.0801,
      "step": 83
    },
    {
      "epoch": 0.12593703148425786,
      "grad_norm": 0.045016042881905784,
      "learning_rate": 0.00012537313432835822,
      "loss": 0.051,
      "step": 84
    },
    {
      "epoch": 0.12743628185907047,
      "grad_norm": 0.03884803395791614,
      "learning_rate": 0.00012686567164179105,
      "loss": 0.0594,
      "step": 85
    },
    {
      "epoch": 0.12893553223388307,
      "grad_norm": 0.028365426279383703,
      "learning_rate": 0.00012835820895522389,
      "loss": 0.0436,
      "step": 86
    },
    {
      "epoch": 0.13043478260869565,
      "grad_norm": 0.030909522037005376,
      "learning_rate": 0.00012985074626865672,
      "loss": 0.0348,
      "step": 87
    },
    {
      "epoch": 0.13193403298350825,
      "grad_norm": 0.0651832129849852,
      "learning_rate": 0.00013134328358208955,
      "loss": 0.0687,
      "step": 88
    },
    {
      "epoch": 0.13343328335832083,
      "grad_norm": 0.043557085072315455,
      "learning_rate": 0.0001328358208955224,
      "loss": 0.0473,
      "step": 89
    },
    {
      "epoch": 0.13493253373313344,
      "grad_norm": 0.028845028778833763,
      "learning_rate": 0.00013432835820895525,
      "loss": 0.0387,
      "step": 90
    },
    {
      "epoch": 0.136431784107946,
      "grad_norm": 0.04331404459497082,
      "learning_rate": 0.00013582089552238805,
      "loss": 0.0619,
      "step": 91
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 0.03133711164003795,
      "learning_rate": 0.0001373134328358209,
      "loss": 0.0315,
      "step": 92
    },
    {
      "epoch": 0.13943028485757122,
      "grad_norm": 0.04502206273984752,
      "learning_rate": 0.00013880597014925375,
      "loss": 0.0575,
      "step": 93
    },
    {
      "epoch": 0.1409295352323838,
      "grad_norm": 0.032917699035736186,
      "learning_rate": 0.00014029850746268658,
      "loss": 0.0398,
      "step": 94
    },
    {
      "epoch": 0.1424287856071964,
      "grad_norm": 0.04144739884037838,
      "learning_rate": 0.00014179104477611942,
      "loss": 0.0569,
      "step": 95
    },
    {
      "epoch": 0.14392803598200898,
      "grad_norm": 0.03128670188478762,
      "learning_rate": 0.00014328358208955225,
      "loss": 0.039,
      "step": 96
    },
    {
      "epoch": 0.1454272863568216,
      "grad_norm": 0.044978141143991145,
      "learning_rate": 0.00014477611940298508,
      "loss": 0.0607,
      "step": 97
    },
    {
      "epoch": 0.1469265367316342,
      "grad_norm": 0.049596838088832966,
      "learning_rate": 0.00014626865671641792,
      "loss": 0.053,
      "step": 98
    },
    {
      "epoch": 0.14842578710644677,
      "grad_norm": 0.03497503808856175,
      "learning_rate": 0.00014776119402985075,
      "loss": 0.0513,
      "step": 99
    },
    {
      "epoch": 0.14992503748125938,
      "grad_norm": 0.03494679451699245,
      "learning_rate": 0.0001492537313432836,
      "loss": 0.032,
      "step": 100
    },
    {
      "epoch": 0.15142428785607195,
      "grad_norm": 0.0572890539993266,
      "learning_rate": 0.00015074626865671642,
      "loss": 0.0665,
      "step": 101
    },
    {
      "epoch": 0.15292353823088456,
      "grad_norm": 0.03898694652987806,
      "learning_rate": 0.00015223880597014925,
      "loss": 0.0374,
      "step": 102
    },
    {
      "epoch": 0.15442278860569716,
      "grad_norm": 0.04199773920037622,
      "learning_rate": 0.00015373134328358208,
      "loss": 0.0461,
      "step": 103
    },
    {
      "epoch": 0.15592203898050974,
      "grad_norm": 0.057829415794355014,
      "learning_rate": 0.00015522388059701495,
      "loss": 0.0755,
      "step": 104
    },
    {
      "epoch": 0.15742128935532235,
      "grad_norm": 0.050148540491503274,
      "learning_rate": 0.00015671641791044778,
      "loss": 0.0507,
      "step": 105
    },
    {
      "epoch": 0.15892053973013492,
      "grad_norm": 0.04386621541760065,
      "learning_rate": 0.00015820895522388059,
      "loss": 0.0452,
      "step": 106
    },
    {
      "epoch": 0.16041979010494753,
      "grad_norm": 0.03554877367553605,
      "learning_rate": 0.00015970149253731345,
      "loss": 0.0428,
      "step": 107
    },
    {
      "epoch": 0.1619190404797601,
      "grad_norm": 0.11161835133361775,
      "learning_rate": 0.00016119402985074628,
      "loss": 0.0681,
      "step": 108
    },
    {
      "epoch": 0.1634182908545727,
      "grad_norm": 0.04645296691253517,
      "learning_rate": 0.00016268656716417911,
      "loss": 0.0538,
      "step": 109
    },
    {
      "epoch": 0.16491754122938532,
      "grad_norm": 0.02803028019747501,
      "learning_rate": 0.00016417910447761195,
      "loss": 0.0272,
      "step": 110
    },
    {
      "epoch": 0.1664167916041979,
      "grad_norm": 0.03415893742686059,
      "learning_rate": 0.00016567164179104478,
      "loss": 0.0254,
      "step": 111
    },
    {
      "epoch": 0.1679160419790105,
      "grad_norm": 0.05809631567240609,
      "learning_rate": 0.00016716417910447761,
      "loss": 0.0611,
      "step": 112
    },
    {
      "epoch": 0.16941529235382308,
      "grad_norm": 0.05603865215472538,
      "learning_rate": 0.00016865671641791045,
      "loss": 0.0394,
      "step": 113
    },
    {
      "epoch": 0.17091454272863568,
      "grad_norm": 0.045375774062388276,
      "learning_rate": 0.00017014925373134328,
      "loss": 0.0538,
      "step": 114
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 0.03960126146668028,
      "learning_rate": 0.00017164179104477614,
      "loss": 0.0423,
      "step": 115
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 0.037163479499863115,
      "learning_rate": 0.00017313432835820895,
      "loss": 0.0481,
      "step": 116
    },
    {
      "epoch": 0.17541229385307347,
      "grad_norm": 0.039738692107982816,
      "learning_rate": 0.00017462686567164178,
      "loss": 0.0345,
      "step": 117
    },
    {
      "epoch": 0.17691154422788605,
      "grad_norm": 0.03833182770101284,
      "learning_rate": 0.00017611940298507464,
      "loss": 0.0402,
      "step": 118
    },
    {
      "epoch": 0.17841079460269865,
      "grad_norm": 0.047548298417749796,
      "learning_rate": 0.00017761194029850748,
      "loss": 0.0542,
      "step": 119
    },
    {
      "epoch": 0.17991004497751126,
      "grad_norm": 0.033462493688485306,
      "learning_rate": 0.0001791044776119403,
      "loss": 0.0303,
      "step": 120
    },
    {
      "epoch": 0.18140929535232383,
      "grad_norm": 0.08098170529150021,
      "learning_rate": 0.00018059701492537314,
      "loss": 0.0802,
      "step": 121
    },
    {
      "epoch": 0.18290854572713644,
      "grad_norm": 0.031967576513773445,
      "learning_rate": 0.00018208955223880598,
      "loss": 0.0291,
      "step": 122
    },
    {
      "epoch": 0.18440779610194902,
      "grad_norm": 0.07620934156104307,
      "learning_rate": 0.0001835820895522388,
      "loss": 0.0737,
      "step": 123
    },
    {
      "epoch": 0.18590704647676162,
      "grad_norm": 0.05731449355361744,
      "learning_rate": 0.00018507462686567165,
      "loss": 0.0504,
      "step": 124
    },
    {
      "epoch": 0.1874062968515742,
      "grad_norm": 0.024809079949469532,
      "learning_rate": 0.00018656716417910448,
      "loss": 0.0302,
      "step": 125
    },
    {
      "epoch": 0.1889055472263868,
      "grad_norm": 0.03621287076524117,
      "learning_rate": 0.00018805970149253734,
      "loss": 0.0557,
      "step": 126
    },
    {
      "epoch": 0.1904047976011994,
      "grad_norm": 0.0495647456653575,
      "learning_rate": 0.00018955223880597015,
      "loss": 0.0547,
      "step": 127
    },
    {
      "epoch": 0.191904047976012,
      "grad_norm": 0.03485762608266449,
      "learning_rate": 0.00019104477611940298,
      "loss": 0.041,
      "step": 128
    },
    {
      "epoch": 0.1934032983508246,
      "grad_norm": 0.059266654938214335,
      "learning_rate": 0.00019253731343283584,
      "loss": 0.0909,
      "step": 129
    },
    {
      "epoch": 0.19490254872563717,
      "grad_norm": 0.08172903239953559,
      "learning_rate": 0.00019402985074626867,
      "loss": 0.1242,
      "step": 130
    },
    {
      "epoch": 0.19640179910044978,
      "grad_norm": 0.05505103331536185,
      "learning_rate": 0.0001955223880597015,
      "loss": 0.08,
      "step": 131
    },
    {
      "epoch": 0.19790104947526238,
      "grad_norm": 0.03964613621186028,
      "learning_rate": 0.00019701492537313434,
      "loss": 0.0465,
      "step": 132
    },
    {
      "epoch": 0.19940029985007496,
      "grad_norm": 0.04992604106896031,
      "learning_rate": 0.00019850746268656718,
      "loss": 0.0503,
      "step": 133
    },
    {
      "epoch": 0.20089955022488756,
      "grad_norm": 0.049289471075707976,
      "learning_rate": 0.0002,
      "loss": 0.0636,
      "step": 134
    },
    {
      "epoch": 0.20239880059970014,
      "grad_norm": 0.07369356962323476,
      "learning_rate": 0.00019999965730559848,
      "loss": 0.0732,
      "step": 135
    },
    {
      "epoch": 0.20389805097451275,
      "grad_norm": 0.047181541580997785,
      "learning_rate": 0.00019999862922474268,
      "loss": 0.0606,
      "step": 136
    },
    {
      "epoch": 0.20539730134932535,
      "grad_norm": 0.03887614440789255,
      "learning_rate": 0.00019999691576447898,
      "loss": 0.036,
      "step": 137
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 0.05259010448351047,
      "learning_rate": 0.00019999451693655123,
      "loss": 0.0631,
      "step": 138
    },
    {
      "epoch": 0.20839580209895053,
      "grad_norm": 0.04113206912303282,
      "learning_rate": 0.00019999143275740072,
      "loss": 0.052,
      "step": 139
    },
    {
      "epoch": 0.2098950524737631,
      "grad_norm": 0.04177105607090762,
      "learning_rate": 0.00019998766324816607,
      "loss": 0.0404,
      "step": 140
    },
    {
      "epoch": 0.21139430284857572,
      "grad_norm": 0.08334534578136313,
      "learning_rate": 0.0001999832084346831,
      "loss": 0.0897,
      "step": 141
    },
    {
      "epoch": 0.2128935532233883,
      "grad_norm": 0.04268046411364645,
      "learning_rate": 0.00019997806834748456,
      "loss": 0.0427,
      "step": 142
    },
    {
      "epoch": 0.2143928035982009,
      "grad_norm": 0.03598851664774015,
      "learning_rate": 0.00019997224302180006,
      "loss": 0.0302,
      "step": 143
    },
    {
      "epoch": 0.2158920539730135,
      "grad_norm": 0.05555632220145266,
      "learning_rate": 0.00019996573249755572,
      "loss": 0.0564,
      "step": 144
    },
    {
      "epoch": 0.21739130434782608,
      "grad_norm": 0.06965699310085238,
      "learning_rate": 0.00019995853681937397,
      "loss": 0.071,
      "step": 145
    },
    {
      "epoch": 0.21889055472263869,
      "grad_norm": 0.07939241103941817,
      "learning_rate": 0.00019995065603657316,
      "loss": 0.0907,
      "step": 146
    },
    {
      "epoch": 0.22038980509745126,
      "grad_norm": 0.050804495965929154,
      "learning_rate": 0.0001999420902031673,
      "loss": 0.048,
      "step": 147
    },
    {
      "epoch": 0.22188905547226387,
      "grad_norm": 0.0590827091873444,
      "learning_rate": 0.00019993283937786563,
      "loss": 0.0705,
      "step": 148
    },
    {
      "epoch": 0.22338830584707647,
      "grad_norm": 0.06023624941065103,
      "learning_rate": 0.0001999229036240723,
      "loss": 0.0527,
      "step": 149
    },
    {
      "epoch": 0.22488755622188905,
      "grad_norm": 0.03486164684165847,
      "learning_rate": 0.00019991228300988585,
      "loss": 0.0442,
      "step": 150
    },
    {
      "epoch": 0.22638680659670166,
      "grad_norm": 0.051399897395934625,
      "learning_rate": 0.00019990097760809876,
      "loss": 0.0975,
      "step": 151
    },
    {
      "epoch": 0.22788605697151423,
      "grad_norm": 0.03483187240018451,
      "learning_rate": 0.00019988898749619702,
      "loss": 0.0366,
      "step": 152
    },
    {
      "epoch": 0.22938530734632684,
      "grad_norm": 0.10145745147250426,
      "learning_rate": 0.00019987631275635948,
      "loss": 0.0709,
      "step": 153
    },
    {
      "epoch": 0.23088455772113944,
      "grad_norm": 0.0738726285389296,
      "learning_rate": 0.0001998629534754574,
      "loss": 0.1201,
      "step": 154
    },
    {
      "epoch": 0.23238380809595202,
      "grad_norm": 0.04853210541031046,
      "learning_rate": 0.00019984890974505381,
      "loss": 0.0628,
      "step": 155
    },
    {
      "epoch": 0.23388305847076463,
      "grad_norm": 0.05279238629760989,
      "learning_rate": 0.00019983418166140285,
      "loss": 0.0395,
      "step": 156
    },
    {
      "epoch": 0.2353823088455772,
      "grad_norm": 0.02918089043238743,
      "learning_rate": 0.00019981876932544917,
      "loss": 0.0295,
      "step": 157
    },
    {
      "epoch": 0.2368815592203898,
      "grad_norm": 0.06770346724424872,
      "learning_rate": 0.00019980267284282717,
      "loss": 0.0638,
      "step": 158
    },
    {
      "epoch": 0.2383808095952024,
      "grad_norm": 0.04833446518136561,
      "learning_rate": 0.00019978589232386035,
      "loss": 0.0519,
      "step": 159
    },
    {
      "epoch": 0.239880059970015,
      "grad_norm": 0.05186898701512186,
      "learning_rate": 0.00019976842788356055,
      "loss": 0.0478,
      "step": 160
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 0.03886911667428048,
      "learning_rate": 0.00019975027964162702,
      "loss": 0.0371,
      "step": 161
    },
    {
      "epoch": 0.24287856071964017,
      "grad_norm": 0.05937400672895507,
      "learning_rate": 0.00019973144772244582,
      "loss": 0.0645,
      "step": 162
    },
    {
      "epoch": 0.24437781109445278,
      "grad_norm": 0.046719542118188725,
      "learning_rate": 0.00019971193225508879,
      "loss": 0.0428,
      "step": 163
    },
    {
      "epoch": 0.24587706146926536,
      "grad_norm": 0.06610712938750751,
      "learning_rate": 0.0001996917333733128,
      "loss": 0.0482,
      "step": 164
    },
    {
      "epoch": 0.24737631184407796,
      "grad_norm": 0.050519212755809846,
      "learning_rate": 0.0001996708512155587,
      "loss": 0.0402,
      "step": 165
    },
    {
      "epoch": 0.24887556221889057,
      "grad_norm": 0.047755461648101266,
      "learning_rate": 0.00019964928592495045,
      "loss": 0.048,
      "step": 166
    },
    {
      "epoch": 0.25037481259370314,
      "grad_norm": 0.04539061348669229,
      "learning_rate": 0.00019962703764929413,
      "loss": 0.0539,
      "step": 167
    },
    {
      "epoch": 0.2518740629685157,
      "grad_norm": 0.044079218147863274,
      "learning_rate": 0.00019960410654107697,
      "loss": 0.0338,
      "step": 168
    },
    {
      "epoch": 0.25337331334332835,
      "grad_norm": 0.15940447460493573,
      "learning_rate": 0.0001995804927574662,
      "loss": 0.1907,
      "step": 169
    },
    {
      "epoch": 0.25487256371814093,
      "grad_norm": 0.14142827538567665,
      "learning_rate": 0.00019955619646030802,
      "loss": 0.1205,
      "step": 170
    },
    {
      "epoch": 0.2563718140929535,
      "grad_norm": 0.044572048202989434,
      "learning_rate": 0.00019953121781612657,
      "loss": 0.0398,
      "step": 171
    },
    {
      "epoch": 0.25787106446776614,
      "grad_norm": 0.06581882047357028,
      "learning_rate": 0.00019950555699612267,
      "loss": 0.0598,
      "step": 172
    },
    {
      "epoch": 0.2593703148425787,
      "grad_norm": 0.058819185905422434,
      "learning_rate": 0.00019947921417617267,
      "loss": 0.0356,
      "step": 173
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 0.07245731651154715,
      "learning_rate": 0.00019945218953682734,
      "loss": 0.0554,
      "step": 174
    },
    {
      "epoch": 0.2623688155922039,
      "grad_norm": 0.0365300697607886,
      "learning_rate": 0.00019942448326331055,
      "loss": 0.0274,
      "step": 175
    },
    {
      "epoch": 0.2638680659670165,
      "grad_norm": 0.06298974261981975,
      "learning_rate": 0.000199396095545518,
      "loss": 0.0607,
      "step": 176
    },
    {
      "epoch": 0.2653673163418291,
      "grad_norm": 0.11713127968683351,
      "learning_rate": 0.00019936702657801587,
      "loss": 0.1058,
      "step": 177
    },
    {
      "epoch": 0.26686656671664166,
      "grad_norm": 0.06199101074736703,
      "learning_rate": 0.00019933727656003963,
      "loss": 0.0491,
      "step": 178
    },
    {
      "epoch": 0.2683658170914543,
      "grad_norm": 0.06866049851783687,
      "learning_rate": 0.00019930684569549264,
      "loss": 0.0536,
      "step": 179
    },
    {
      "epoch": 0.2698650674662669,
      "grad_norm": 0.09605253534909008,
      "learning_rate": 0.00019927573419294456,
      "loss": 0.07,
      "step": 180
    },
    {
      "epoch": 0.27136431784107945,
      "grad_norm": 0.08723233109826037,
      "learning_rate": 0.00019924394226563018,
      "loss": 0.0728,
      "step": 181
    },
    {
      "epoch": 0.272863568215892,
      "grad_norm": 0.08669179137921693,
      "learning_rate": 0.0001992114701314478,
      "loss": 0.0691,
      "step": 182
    },
    {
      "epoch": 0.27436281859070466,
      "grad_norm": 0.05913118898122669,
      "learning_rate": 0.0001991783180129578,
      "loss": 0.066,
      "step": 183
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 0.051263995779607295,
      "learning_rate": 0.00019914448613738106,
      "loss": 0.0404,
      "step": 184
    },
    {
      "epoch": 0.2773613193403298,
      "grad_norm": 0.05477400514379348,
      "learning_rate": 0.0001991099747365975,
      "loss": 0.0427,
      "step": 185
    },
    {
      "epoch": 0.27886056971514245,
      "grad_norm": 0.06920109883069289,
      "learning_rate": 0.00019907478404714436,
      "loss": 0.0474,
      "step": 186
    },
    {
      "epoch": 0.280359820089955,
      "grad_norm": 0.03826038209148185,
      "learning_rate": 0.00019903891431021477,
      "loss": 0.0419,
      "step": 187
    },
    {
      "epoch": 0.2818590704647676,
      "grad_norm": 0.10991221849444072,
      "learning_rate": 0.00019900236577165576,
      "loss": 0.0858,
      "step": 188
    },
    {
      "epoch": 0.28335832083958024,
      "grad_norm": 0.031681176003979745,
      "learning_rate": 0.00019896513868196704,
      "loss": 0.035,
      "step": 189
    },
    {
      "epoch": 0.2848575712143928,
      "grad_norm": 0.0519531999433836,
      "learning_rate": 0.00019892723329629887,
      "loss": 0.0464,
      "step": 190
    },
    {
      "epoch": 0.2863568215892054,
      "grad_norm": 0.05571628339788405,
      "learning_rate": 0.0001988886498744505,
      "loss": 0.0518,
      "step": 191
    },
    {
      "epoch": 0.28785607196401797,
      "grad_norm": 0.06914430596733225,
      "learning_rate": 0.00019884938868086835,
      "loss": 0.0865,
      "step": 192
    },
    {
      "epoch": 0.2893553223388306,
      "grad_norm": 0.07723650037620701,
      "learning_rate": 0.00019880944998464434,
      "loss": 0.07,
      "step": 193
    },
    {
      "epoch": 0.2908545727136432,
      "grad_norm": 0.03868436237329038,
      "learning_rate": 0.00019876883405951377,
      "loss": 0.0405,
      "step": 194
    },
    {
      "epoch": 0.29235382308845576,
      "grad_norm": 0.03616727905905788,
      "learning_rate": 0.00019872754118385367,
      "loss": 0.0353,
      "step": 195
    },
    {
      "epoch": 0.2938530734632684,
      "grad_norm": 0.06407436416080277,
      "learning_rate": 0.00019868557164068074,
      "loss": 0.0518,
      "step": 196
    },
    {
      "epoch": 0.29535232383808097,
      "grad_norm": 0.0885276429432359,
      "learning_rate": 0.00019864292571764955,
      "loss": 0.071,
      "step": 197
    },
    {
      "epoch": 0.29685157421289354,
      "grad_norm": 0.06517887639083755,
      "learning_rate": 0.0001985996037070505,
      "loss": 0.0472,
      "step": 198
    },
    {
      "epoch": 0.2983508245877061,
      "grad_norm": 0.03262350195450329,
      "learning_rate": 0.00019855560590580778,
      "loss": 0.0357,
      "step": 199
    },
    {
      "epoch": 0.29985007496251875,
      "grad_norm": 0.06352918641533918,
      "learning_rate": 0.0001985109326154774,
      "loss": 0.0636,
      "step": 200
    },
    {
      "epoch": 0.30134932533733133,
      "grad_norm": 0.06275349247389327,
      "learning_rate": 0.0001984655841422451,
      "loss": 0.0746,
      "step": 201
    },
    {
      "epoch": 0.3028485757121439,
      "grad_norm": 0.04177370401732468,
      "learning_rate": 0.0001984195607969242,
      "loss": 0.0483,
      "step": 202
    },
    {
      "epoch": 0.30434782608695654,
      "grad_norm": 0.09421608937730566,
      "learning_rate": 0.00019837286289495361,
      "loss": 0.0731,
      "step": 203
    },
    {
      "epoch": 0.3058470764617691,
      "grad_norm": 0.06309582839396509,
      "learning_rate": 0.0001983254907563955,
      "loss": 0.0806,
      "step": 204
    },
    {
      "epoch": 0.3073463268365817,
      "grad_norm": 0.043040241993361705,
      "learning_rate": 0.00019827744470593314,
      "loss": 0.042,
      "step": 205
    },
    {
      "epoch": 0.30884557721139433,
      "grad_norm": 0.07592171510511146,
      "learning_rate": 0.0001982287250728689,
      "loss": 0.0672,
      "step": 206
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 0.04998022421825792,
      "learning_rate": 0.00019817933219112158,
      "loss": 0.0598,
      "step": 207
    },
    {
      "epoch": 0.3118440779610195,
      "grad_norm": 0.06445241515366612,
      "learning_rate": 0.0001981292663992245,
      "loss": 0.042,
      "step": 208
    },
    {
      "epoch": 0.31334332833583206,
      "grad_norm": 0.05511385292159757,
      "learning_rate": 0.00019807852804032305,
      "loss": 0.0679,
      "step": 209
    },
    {
      "epoch": 0.3148425787106447,
      "grad_norm": 0.10911657904245481,
      "learning_rate": 0.00019802711746217218,
      "loss": 0.0992,
      "step": 210
    },
    {
      "epoch": 0.31634182908545727,
      "grad_norm": 0.04607736230949591,
      "learning_rate": 0.00019797503501713427,
      "loss": 0.0386,
      "step": 211
    },
    {
      "epoch": 0.31784107946026985,
      "grad_norm": 0.055336925627864554,
      "learning_rate": 0.00019792228106217658,
      "loss": 0.0482,
      "step": 212
    },
    {
      "epoch": 0.3193403298350825,
      "grad_norm": 0.0367000018321132,
      "learning_rate": 0.0001978688559588688,
      "loss": 0.0468,
      "step": 213
    },
    {
      "epoch": 0.32083958020989506,
      "grad_norm": 0.03475056457429854,
      "learning_rate": 0.00019781476007338058,
      "loss": 0.0303,
      "step": 214
    },
    {
      "epoch": 0.32233883058470764,
      "grad_norm": 0.06222566098960902,
      "learning_rate": 0.0001977599937764791,
      "loss": 0.0643,
      "step": 215
    },
    {
      "epoch": 0.3238380809595202,
      "grad_norm": 0.08945096189461864,
      "learning_rate": 0.0001977045574435264,
      "loss": 0.1043,
      "step": 216
    },
    {
      "epoch": 0.32533733133433285,
      "grad_norm": 0.056441998030072674,
      "learning_rate": 0.00019764845145447689,
      "loss": 0.0486,
      "step": 217
    },
    {
      "epoch": 0.3268365817091454,
      "grad_norm": 0.062483693636514896,
      "learning_rate": 0.00019759167619387476,
      "loss": 0.0606,
      "step": 218
    },
    {
      "epoch": 0.328335832083958,
      "grad_norm": 0.11698538531516464,
      "learning_rate": 0.00019753423205085127,
      "loss": 0.0507,
      "step": 219
    },
    {
      "epoch": 0.32983508245877063,
      "grad_norm": 0.07112373998312807,
      "learning_rate": 0.0001974761194191222,
      "loss": 0.0504,
      "step": 220
    },
    {
      "epoch": 0.3313343328335832,
      "grad_norm": 0.04204879182316876,
      "learning_rate": 0.00019741733869698495,
      "loss": 0.0457,
      "step": 221
    },
    {
      "epoch": 0.3328335832083958,
      "grad_norm": 0.047081193320726994,
      "learning_rate": 0.00019735789028731604,
      "loss": 0.0562,
      "step": 222
    },
    {
      "epoch": 0.3343328335832084,
      "grad_norm": 0.046700620652771065,
      "learning_rate": 0.0001972977745975682,
      "loss": 0.0387,
      "step": 223
    },
    {
      "epoch": 0.335832083958021,
      "grad_norm": 0.05658161180730498,
      "learning_rate": 0.00019723699203976766,
      "loss": 0.075,
      "step": 224
    },
    {
      "epoch": 0.3373313343328336,
      "grad_norm": 0.04886519491828949,
      "learning_rate": 0.00019717554303051127,
      "loss": 0.0438,
      "step": 225
    },
    {
      "epoch": 0.33883058470764615,
      "grad_norm": 0.061882324104049265,
      "learning_rate": 0.00019711342799096361,
      "loss": 0.107,
      "step": 226
    },
    {
      "epoch": 0.3403298350824588,
      "grad_norm": 0.05128064620446334,
      "learning_rate": 0.00019705064734685425,
      "loss": 0.0499,
      "step": 227
    },
    {
      "epoch": 0.34182908545727136,
      "grad_norm": 0.037511885925608925,
      "learning_rate": 0.00019698720152847468,
      "loss": 0.0354,
      "step": 228
    },
    {
      "epoch": 0.34332833583208394,
      "grad_norm": 0.054518869510984926,
      "learning_rate": 0.00019692309097067546,
      "loss": 0.0623,
      "step": 229
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 0.04111726041766873,
      "learning_rate": 0.0001968583161128631,
      "loss": 0.0399,
      "step": 230
    },
    {
      "epoch": 0.34632683658170915,
      "grad_norm": 0.05016557394284265,
      "learning_rate": 0.0001967928773989973,
      "loss": 0.0399,
      "step": 231
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 0.054953298674171765,
      "learning_rate": 0.0001967267752775877,
      "loss": 0.0453,
      "step": 232
    },
    {
      "epoch": 0.3493253373313343,
      "grad_norm": 0.04128614741818805,
      "learning_rate": 0.00019666001020169073,
      "loss": 0.0437,
      "step": 233
    },
    {
      "epoch": 0.35082458770614694,
      "grad_norm": 0.0711824564667931,
      "learning_rate": 0.00019659258262890683,
      "loss": 0.0849,
      "step": 234
    },
    {
      "epoch": 0.3523238380809595,
      "grad_norm": 0.042020248359914844,
      "learning_rate": 0.00019652449302137704,
      "loss": 0.0386,
      "step": 235
    },
    {
      "epoch": 0.3538230884557721,
      "grad_norm": 0.046481177796938444,
      "learning_rate": 0.00019645574184577982,
      "loss": 0.0383,
      "step": 236
    },
    {
      "epoch": 0.3553223388305847,
      "grad_norm": 0.08165767843883412,
      "learning_rate": 0.0001963863295733281,
      "loss": 0.042,
      "step": 237
    },
    {
      "epoch": 0.3568215892053973,
      "grad_norm": 0.038903250983456907,
      "learning_rate": 0.00019631625667976583,
      "loss": 0.0341,
      "step": 238
    },
    {
      "epoch": 0.3583208395802099,
      "grad_norm": 0.04053657985889596,
      "learning_rate": 0.00019624552364536473,
      "loss": 0.0348,
      "step": 239
    },
    {
      "epoch": 0.3598200899550225,
      "grad_norm": 0.04790719363132042,
      "learning_rate": 0.00019617413095492114,
      "loss": 0.0386,
      "step": 240
    },
    {
      "epoch": 0.3613193403298351,
      "grad_norm": 0.053683248937234807,
      "learning_rate": 0.00019610207909775255,
      "loss": 0.0714,
      "step": 241
    },
    {
      "epoch": 0.36281859070464767,
      "grad_norm": 0.038744106761683335,
      "learning_rate": 0.0001960293685676943,
      "loss": 0.0456,
      "step": 242
    },
    {
      "epoch": 0.36431784107946025,
      "grad_norm": 0.05154181261803494,
      "learning_rate": 0.00019595599986309628,
      "loss": 0.0581,
      "step": 243
    },
    {
      "epoch": 0.3658170914542729,
      "grad_norm": 0.04824861446452559,
      "learning_rate": 0.0001958819734868193,
      "loss": 0.0706,
      "step": 244
    },
    {
      "epoch": 0.36731634182908546,
      "grad_norm": 0.03758555173738341,
      "learning_rate": 0.00019580728994623195,
      "loss": 0.0305,
      "step": 245
    },
    {
      "epoch": 0.36881559220389803,
      "grad_norm": 0.03835047499522843,
      "learning_rate": 0.00019573194975320673,
      "loss": 0.032,
      "step": 246
    },
    {
      "epoch": 0.37031484257871067,
      "grad_norm": 0.03475293565575896,
      "learning_rate": 0.000195655953424117,
      "loss": 0.0376,
      "step": 247
    },
    {
      "epoch": 0.37181409295352325,
      "grad_norm": 0.04896052744776511,
      "learning_rate": 0.00019557930147983302,
      "loss": 0.0398,
      "step": 248
    },
    {
      "epoch": 0.3733133433283358,
      "grad_norm": 0.04797369299914517,
      "learning_rate": 0.0001955019944457187,
      "loss": 0.0454,
      "step": 249
    },
    {
      "epoch": 0.3748125937031484,
      "grad_norm": 0.07781910571114095,
      "learning_rate": 0.0001954240328516277,
      "loss": 0.0698,
      "step": 250
    },
    {
      "epoch": 0.37631184407796103,
      "grad_norm": 0.04338750887929876,
      "learning_rate": 0.0001953454172319001,
      "loss": 0.0456,
      "step": 251
    },
    {
      "epoch": 0.3778110944527736,
      "grad_norm": 0.04454900811666872,
      "learning_rate": 0.00019526614812535864,
      "loss": 0.0282,
      "step": 252
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 0.05266395324382892,
      "learning_rate": 0.0001951862260753048,
      "loss": 0.0672,
      "step": 253
    },
    {
      "epoch": 0.3808095952023988,
      "grad_norm": 0.057494318285817976,
      "learning_rate": 0.00019510565162951537,
      "loss": 0.0731,
      "step": 254
    },
    {
      "epoch": 0.3823088455772114,
      "grad_norm": 0.04735808897837354,
      "learning_rate": 0.00019502442534023858,
      "loss": 0.0456,
      "step": 255
    },
    {
      "epoch": 0.383808095952024,
      "grad_norm": 0.1006133724153584,
      "learning_rate": 0.0001949425477641904,
      "loss": 0.0872,
      "step": 256
    },
    {
      "epoch": 0.3853073463268366,
      "grad_norm": 0.02920756237692682,
      "learning_rate": 0.00019486001946255046,
      "loss": 0.0309,
      "step": 257
    },
    {
      "epoch": 0.3868065967016492,
      "grad_norm": 0.04990145786446464,
      "learning_rate": 0.0001947768410009586,
      "loss": 0.033,
      "step": 258
    },
    {
      "epoch": 0.38830584707646176,
      "grad_norm": 0.050701365717084494,
      "learning_rate": 0.0001946930129495106,
      "loss": 0.0367,
      "step": 259
    },
    {
      "epoch": 0.38980509745127434,
      "grad_norm": 0.03698562204942488,
      "learning_rate": 0.00019460853588275454,
      "loss": 0.034,
      "step": 260
    },
    {
      "epoch": 0.391304347826087,
      "grad_norm": 0.07391856974626904,
      "learning_rate": 0.00019452341037968682,
      "loss": 0.0489,
      "step": 261
    },
    {
      "epoch": 0.39280359820089955,
      "grad_norm": 0.05717733621263349,
      "learning_rate": 0.00019443763702374812,
      "loss": 0.0448,
      "step": 262
    },
    {
      "epoch": 0.39430284857571213,
      "grad_norm": 0.044077421896152096,
      "learning_rate": 0.00019435121640281938,
      "loss": 0.0289,
      "step": 263
    },
    {
      "epoch": 0.39580209895052476,
      "grad_norm": 0.07710766944383475,
      "learning_rate": 0.00019426414910921787,
      "loss": 0.062,
      "step": 264
    },
    {
      "epoch": 0.39730134932533734,
      "grad_norm": 0.06386202461201915,
      "learning_rate": 0.00019417643573969303,
      "loss": 0.0545,
      "step": 265
    },
    {
      "epoch": 0.3988005997001499,
      "grad_norm": 0.035388492021471624,
      "learning_rate": 0.00019408807689542257,
      "loss": 0.0338,
      "step": 266
    },
    {
      "epoch": 0.4002998500749625,
      "grad_norm": 0.10927492127823649,
      "learning_rate": 0.00019399907318200802,
      "loss": 0.0893,
      "step": 267
    },
    {
      "epoch": 0.4017991004497751,
      "grad_norm": 0.06012238034598342,
      "learning_rate": 0.0001939094252094709,
      "loss": 0.0527,
      "step": 268
    },
    {
      "epoch": 0.4032983508245877,
      "grad_norm": 0.03391978832858692,
      "learning_rate": 0.00019381913359224842,
      "loss": 0.0374,
      "step": 269
    },
    {
      "epoch": 0.4047976011994003,
      "grad_norm": 0.059440210121053584,
      "learning_rate": 0.00019372819894918915,
      "loss": 0.0709,
      "step": 270
    },
    {
      "epoch": 0.4062968515742129,
      "grad_norm": 0.05188231618768116,
      "learning_rate": 0.000193636621903549,
      "loss": 0.0683,
      "step": 271
    },
    {
      "epoch": 0.4077961019490255,
      "grad_norm": 0.04719133608642286,
      "learning_rate": 0.00019354440308298675,
      "loss": 0.0477,
      "step": 272
    },
    {
      "epoch": 0.40929535232383807,
      "grad_norm": 0.07053359056135224,
      "learning_rate": 0.00019345154311955985,
      "loss": 0.0917,
      "step": 273
    },
    {
      "epoch": 0.4107946026986507,
      "grad_norm": 0.03312060738445713,
      "learning_rate": 0.00019335804264972018,
      "loss": 0.0342,
      "step": 274
    },
    {
      "epoch": 0.4122938530734633,
      "grad_norm": 0.05192332591417718,
      "learning_rate": 0.00019326390231430942,
      "loss": 0.0621,
      "step": 275
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 0.06103626606110473,
      "learning_rate": 0.0001931691227585549,
      "loss": 0.0544,
      "step": 276
    },
    {
      "epoch": 0.41529235382308843,
      "grad_norm": 0.035426733024569466,
      "learning_rate": 0.0001930737046320651,
      "loss": 0.0386,
      "step": 277
    },
    {
      "epoch": 0.41679160419790107,
      "grad_norm": 0.03649997902640354,
      "learning_rate": 0.00019297764858882514,
      "loss": 0.0299,
      "step": 278
    },
    {
      "epoch": 0.41829085457271364,
      "grad_norm": 0.044934938886158564,
      "learning_rate": 0.00019288095528719243,
      "loss": 0.043,
      "step": 279
    },
    {
      "epoch": 0.4197901049475262,
      "grad_norm": 0.056516397554641565,
      "learning_rate": 0.000192783625389892,
      "loss": 0.0491,
      "step": 280
    },
    {
      "epoch": 0.42128935532233885,
      "grad_norm": 0.03896533163368287,
      "learning_rate": 0.00019268565956401208,
      "loss": 0.0358,
      "step": 281
    },
    {
      "epoch": 0.42278860569715143,
      "grad_norm": 0.04117282780665747,
      "learning_rate": 0.0001925870584809995,
      "loss": 0.046,
      "step": 282
    },
    {
      "epoch": 0.424287856071964,
      "grad_norm": 0.3604260889224257,
      "learning_rate": 0.00019248782281665498,
      "loss": 0.1119,
      "step": 283
    },
    {
      "epoch": 0.4257871064467766,
      "grad_norm": 0.04720518482542838,
      "learning_rate": 0.0001923879532511287,
      "loss": 0.0427,
      "step": 284
    },
    {
      "epoch": 0.4272863568215892,
      "grad_norm": 0.04236046102866595,
      "learning_rate": 0.00019228745046891542,
      "loss": 0.026,
      "step": 285
    },
    {
      "epoch": 0.4287856071964018,
      "grad_norm": 0.06797148449245759,
      "learning_rate": 0.00019218631515885006,
      "loss": 0.0449,
      "step": 286
    },
    {
      "epoch": 0.4302848575712144,
      "grad_norm": 0.07816741282791993,
      "learning_rate": 0.00019208454801410266,
      "loss": 0.0696,
      "step": 287
    },
    {
      "epoch": 0.431784107946027,
      "grad_norm": 0.05675040921166316,
      "learning_rate": 0.00019198214973217378,
      "loss": 0.0434,
      "step": 288
    },
    {
      "epoch": 0.4332833583208396,
      "grad_norm": 0.06237548709245271,
      "learning_rate": 0.00019187912101488984,
      "loss": 0.0696,
      "step": 289
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 0.07002008527028293,
      "learning_rate": 0.00019177546256839812,
      "loss": 0.1012,
      "step": 290
    },
    {
      "epoch": 0.4362818590704648,
      "grad_norm": 0.04893068129720385,
      "learning_rate": 0.000191671175103162,
      "loss": 0.0409,
      "step": 291
    },
    {
      "epoch": 0.43778110944527737,
      "grad_norm": 0.04601357359317086,
      "learning_rate": 0.00019156625933395614,
      "loss": 0.0519,
      "step": 292
    },
    {
      "epoch": 0.43928035982008995,
      "grad_norm": 0.05337895694784327,
      "learning_rate": 0.00019146071597986138,
      "loss": 0.0372,
      "step": 293
    },
    {
      "epoch": 0.4407796101949025,
      "grad_norm": 0.03995143875131533,
      "learning_rate": 0.0001913545457642601,
      "loss": 0.0368,
      "step": 294
    },
    {
      "epoch": 0.44227886056971516,
      "grad_norm": 0.09850621381875037,
      "learning_rate": 0.00019124774941483107,
      "loss": 0.0761,
      "step": 295
    },
    {
      "epoch": 0.44377811094452774,
      "grad_norm": 0.038785537167516496,
      "learning_rate": 0.00019114032766354453,
      "loss": 0.041,
      "step": 296
    },
    {
      "epoch": 0.4452773613193403,
      "grad_norm": 0.066120691647694,
      "learning_rate": 0.00019103228124665712,
      "loss": 0.0747,
      "step": 297
    },
    {
      "epoch": 0.44677661169415295,
      "grad_norm": 0.11147859567607986,
      "learning_rate": 0.00019092361090470688,
      "loss": 0.0508,
      "step": 298
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 0.09831159493729676,
      "learning_rate": 0.00019081431738250814,
      "loss": 0.0562,
      "step": 299
    },
    {
      "epoch": 0.4497751124437781,
      "grad_norm": 0.07264006526009004,
      "learning_rate": 0.0001907044014291465,
      "loss": 0.0557,
      "step": 300
    },
    {
      "epoch": 0.4512743628185907,
      "grad_norm": 0.07996560398829375,
      "learning_rate": 0.0001905938637979736,
      "loss": 0.0482,
      "step": 301
    },
    {
      "epoch": 0.4527736131934033,
      "grad_norm": 0.0686136548292036,
      "learning_rate": 0.00019048270524660196,
      "loss": 0.0503,
      "step": 302
    },
    {
      "epoch": 0.4542728635682159,
      "grad_norm": 0.045762027803397255,
      "learning_rate": 0.00019037092653689987,
      "loss": 0.0385,
      "step": 303
    },
    {
      "epoch": 0.45577211394302847,
      "grad_norm": 0.06709124946128789,
      "learning_rate": 0.00019025852843498607,
      "loss": 0.0678,
      "step": 304
    },
    {
      "epoch": 0.4572713643178411,
      "grad_norm": 0.05610882393237923,
      "learning_rate": 0.00019014551171122457,
      "loss": 0.0482,
      "step": 305
    },
    {
      "epoch": 0.4587706146926537,
      "grad_norm": 0.06482964052411654,
      "learning_rate": 0.00019003187714021938,
      "loss": 0.0531,
      "step": 306
    },
    {
      "epoch": 0.46026986506746626,
      "grad_norm": 0.07856511913650002,
      "learning_rate": 0.00018991762550080906,
      "loss": 0.0531,
      "step": 307
    },
    {
      "epoch": 0.4617691154422789,
      "grad_norm": 0.08727827018786473,
      "learning_rate": 0.00018980275757606157,
      "loss": 0.0696,
      "step": 308
    },
    {
      "epoch": 0.46326836581709147,
      "grad_norm": 0.054829790468602986,
      "learning_rate": 0.00018968727415326884,
      "loss": 0.0522,
      "step": 309
    },
    {
      "epoch": 0.46476761619190404,
      "grad_norm": 0.0533722415191346,
      "learning_rate": 0.0001895711760239413,
      "loss": 0.0455,
      "step": 310
    },
    {
      "epoch": 0.4662668665667166,
      "grad_norm": 0.0404960196283668,
      "learning_rate": 0.0001894544639838025,
      "loss": 0.0432,
      "step": 311
    },
    {
      "epoch": 0.46776611694152925,
      "grad_norm": 0.044028946635248145,
      "learning_rate": 0.00018933713883278376,
      "loss": 0.0425,
      "step": 312
    },
    {
      "epoch": 0.46926536731634183,
      "grad_norm": 0.06293150170038057,
      "learning_rate": 0.00018921920137501849,
      "loss": 0.0508,
      "step": 313
    },
    {
      "epoch": 0.4707646176911544,
      "grad_norm": 0.06810220996600896,
      "learning_rate": 0.0001891006524188368,
      "loss": 0.0581,
      "step": 314
    },
    {
      "epoch": 0.47226386806596704,
      "grad_norm": 0.06959342447747006,
      "learning_rate": 0.00018898149277675997,
      "loss": 0.0725,
      "step": 315
    },
    {
      "epoch": 0.4737631184407796,
      "grad_norm": 0.061826777912191176,
      "learning_rate": 0.0001888617232654949,
      "loss": 0.0598,
      "step": 316
    },
    {
      "epoch": 0.4752623688155922,
      "grad_norm": 0.066670766072721,
      "learning_rate": 0.00018874134470592835,
      "loss": 0.0387,
      "step": 317
    },
    {
      "epoch": 0.4767616191904048,
      "grad_norm": 0.05284481473477404,
      "learning_rate": 0.00018862035792312147,
      "loss": 0.0353,
      "step": 318
    },
    {
      "epoch": 0.4782608695652174,
      "grad_norm": 0.03546158471045893,
      "learning_rate": 0.0001884987637463042,
      "loss": 0.0353,
      "step": 319
    },
    {
      "epoch": 0.47976011994003,
      "grad_norm": 0.03695047012775239,
      "learning_rate": 0.00018837656300886937,
      "loss": 0.033,
      "step": 320
    },
    {
      "epoch": 0.48125937031484256,
      "grad_norm": 0.09946531845424002,
      "learning_rate": 0.00018825375654836713,
      "loss": 0.068,
      "step": 321
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 0.0525789693495926,
      "learning_rate": 0.0001881303452064992,
      "loss": 0.0603,
      "step": 322
    },
    {
      "epoch": 0.48425787106446777,
      "grad_norm": 0.061737118720852,
      "learning_rate": 0.00018800632982911322,
      "loss": 0.0606,
      "step": 323
    },
    {
      "epoch": 0.48575712143928035,
      "grad_norm": 0.036553688772480226,
      "learning_rate": 0.00018788171126619653,
      "loss": 0.0299,
      "step": 324
    },
    {
      "epoch": 0.487256371814093,
      "grad_norm": 0.04923145685241048,
      "learning_rate": 0.00018775649037187093,
      "loss": 0.0475,
      "step": 325
    },
    {
      "epoch": 0.48875562218890556,
      "grad_norm": 0.06336340360281395,
      "learning_rate": 0.00018763066800438636,
      "loss": 0.0689,
      "step": 326
    },
    {
      "epoch": 0.49025487256371814,
      "grad_norm": 0.044393763008170536,
      "learning_rate": 0.00018750424502611527,
      "loss": 0.0413,
      "step": 327
    },
    {
      "epoch": 0.4917541229385307,
      "grad_norm": 0.04138603160844389,
      "learning_rate": 0.00018737722230354655,
      "loss": 0.0445,
      "step": 328
    },
    {
      "epoch": 0.49325337331334335,
      "grad_norm": 0.08068404550662704,
      "learning_rate": 0.00018724960070727972,
      "loss": 0.0507,
      "step": 329
    },
    {
      "epoch": 0.4947526236881559,
      "grad_norm": 0.05448885894958413,
      "learning_rate": 0.00018712138111201895,
      "loss": 0.0581,
      "step": 330
    },
    {
      "epoch": 0.4962518740629685,
      "grad_norm": 0.0541786649288355,
      "learning_rate": 0.00018699256439656697,
      "loss": 0.05,
      "step": 331
    },
    {
      "epoch": 0.49775112443778113,
      "grad_norm": 0.09333363889478805,
      "learning_rate": 0.00018686315144381913,
      "loss": 0.0634,
      "step": 332
    },
    {
      "epoch": 0.4992503748125937,
      "grad_norm": 0.06046721622165493,
      "learning_rate": 0.00018673314314075732,
      "loss": 0.0393,
      "step": 333
    },
    {
      "epoch": 0.5007496251874063,
      "grad_norm": 0.22984713729553005,
      "learning_rate": 0.00018660254037844388,
      "loss": 0.1195,
      "step": 334
    },
    {
      "epoch": 0.5022488755622189,
      "grad_norm": 0.03840858370215027,
      "learning_rate": 0.0001864713440520155,
      "loss": 0.038,
      "step": 335
    },
    {
      "epoch": 0.5037481259370314,
      "grad_norm": 0.0460739408926456,
      "learning_rate": 0.00018633955506067718,
      "loss": 0.0277,
      "step": 336
    },
    {
      "epoch": 0.5052473763118441,
      "grad_norm": 0.058586326621381794,
      "learning_rate": 0.00018620717430769586,
      "loss": 0.0418,
      "step": 337
    },
    {
      "epoch": 0.5067466266866567,
      "grad_norm": 0.0691170892823961,
      "learning_rate": 0.0001860742027003944,
      "loss": 0.0611,
      "step": 338
    },
    {
      "epoch": 0.5082458770614693,
      "grad_norm": 0.054450391696896736,
      "learning_rate": 0.0001859406411501453,
      "loss": 0.0389,
      "step": 339
    },
    {
      "epoch": 0.5097451274362819,
      "grad_norm": 0.05989403502334338,
      "learning_rate": 0.00018580649057236447,
      "loss": 0.0656,
      "step": 340
    },
    {
      "epoch": 0.5112443778110944,
      "grad_norm": 0.05812418432414411,
      "learning_rate": 0.00018567175188650498,
      "loss": 0.0516,
      "step": 341
    },
    {
      "epoch": 0.512743628185907,
      "grad_norm": 0.1024472803798587,
      "learning_rate": 0.00018553642601605068,
      "loss": 0.0489,
      "step": 342
    },
    {
      "epoch": 0.5142428785607196,
      "grad_norm": 0.06418199886991582,
      "learning_rate": 0.00018540051388850993,
      "loss": 0.0641,
      "step": 343
    },
    {
      "epoch": 0.5157421289355323,
      "grad_norm": 0.05354265643676874,
      "learning_rate": 0.00018526401643540922,
      "loss": 0.0342,
      "step": 344
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 0.060234343510686926,
      "learning_rate": 0.00018512693459228684,
      "loss": 0.0441,
      "step": 345
    },
    {
      "epoch": 0.5187406296851574,
      "grad_norm": 0.05502477922945836,
      "learning_rate": 0.00018498926929868642,
      "loss": 0.044,
      "step": 346
    },
    {
      "epoch": 0.52023988005997,
      "grad_norm": 0.05316566519953199,
      "learning_rate": 0.00018485102149815038,
      "loss": 0.0407,
      "step": 347
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 0.08251700451700679,
      "learning_rate": 0.00018471219213821375,
      "loss": 0.0475,
      "step": 348
    },
    {
      "epoch": 0.5232383808095952,
      "grad_norm": 0.044332263163009075,
      "learning_rate": 0.00018457278217039736,
      "loss": 0.0339,
      "step": 349
    },
    {
      "epoch": 0.5247376311844077,
      "grad_norm": 0.06598873060964301,
      "learning_rate": 0.00018443279255020152,
      "loss": 0.0623,
      "step": 350
    },
    {
      "epoch": 0.5262368815592204,
      "grad_norm": 0.09432658184269146,
      "learning_rate": 0.00018429222423709947,
      "loss": 0.0681,
      "step": 351
    },
    {
      "epoch": 0.527736131934033,
      "grad_norm": 0.048137123345461545,
      "learning_rate": 0.00018415107819453062,
      "loss": 0.0379,
      "step": 352
    },
    {
      "epoch": 0.5292353823088456,
      "grad_norm": 0.048726928699549,
      "learning_rate": 0.0001840093553898942,
      "loss": 0.0517,
      "step": 353
    },
    {
      "epoch": 0.5307346326836582,
      "grad_norm": 0.03886288744382217,
      "learning_rate": 0.00018386705679454242,
      "loss": 0.0309,
      "step": 354
    },
    {
      "epoch": 0.5322338830584707,
      "grad_norm": 0.04487311032788653,
      "learning_rate": 0.0001837241833837739,
      "loss": 0.0294,
      "step": 355
    },
    {
      "epoch": 0.5337331334332833,
      "grad_norm": 0.03798327068043393,
      "learning_rate": 0.00018358073613682706,
      "loss": 0.031,
      "step": 356
    },
    {
      "epoch": 0.5352323838080959,
      "grad_norm": 0.050882637106700504,
      "learning_rate": 0.00018343671603687317,
      "loss": 0.039,
      "step": 357
    },
    {
      "epoch": 0.5367316341829086,
      "grad_norm": 0.044214998902001985,
      "learning_rate": 0.00018329212407100994,
      "loss": 0.0333,
      "step": 358
    },
    {
      "epoch": 0.5382308845577212,
      "grad_norm": 0.06411346885005922,
      "learning_rate": 0.00018314696123025454,
      "loss": 0.066,
      "step": 359
    },
    {
      "epoch": 0.5397301349325337,
      "grad_norm": 0.042800963513781314,
      "learning_rate": 0.00018300122850953675,
      "loss": 0.0303,
      "step": 360
    },
    {
      "epoch": 0.5412293853073463,
      "grad_norm": 0.051572469675010446,
      "learning_rate": 0.00018285492690769237,
      "loss": 0.0459,
      "step": 361
    },
    {
      "epoch": 0.5427286356821589,
      "grad_norm": 0.06594725487157567,
      "learning_rate": 0.00018270805742745617,
      "loss": 0.0554,
      "step": 362
    },
    {
      "epoch": 0.5442278860569715,
      "grad_norm": 0.08966668405728594,
      "learning_rate": 0.00018256062107545518,
      "loss": 0.0504,
      "step": 363
    },
    {
      "epoch": 0.545727136431784,
      "grad_norm": 0.05705712395021745,
      "learning_rate": 0.00018241261886220154,
      "loss": 0.0542,
      "step": 364
    },
    {
      "epoch": 0.5472263868065967,
      "grad_norm": 0.046841858897893975,
      "learning_rate": 0.000182264051802086,
      "loss": 0.046,
      "step": 365
    },
    {
      "epoch": 0.5487256371814093,
      "grad_norm": 0.05043321212035937,
      "learning_rate": 0.00018211492091337042,
      "loss": 0.0515,
      "step": 366
    },
    {
      "epoch": 0.5502248875562219,
      "grad_norm": 0.05643206918749337,
      "learning_rate": 0.00018196522721818128,
      "loss": 0.0398,
      "step": 367
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.10718050662244094,
      "learning_rate": 0.00018181497174250236,
      "loss": 0.0234,
      "step": 368
    },
    {
      "epoch": 0.553223388305847,
      "grad_norm": 0.07740102244845123,
      "learning_rate": 0.00018166415551616792,
      "loss": 0.0523,
      "step": 369
    },
    {
      "epoch": 0.5547226386806596,
      "grad_norm": 0.06095682806386288,
      "learning_rate": 0.00018151277957285543,
      "loss": 0.0434,
      "step": 370
    },
    {
      "epoch": 0.5562218890554723,
      "grad_norm": 0.06932915829661977,
      "learning_rate": 0.00018136084495007872,
      "loss": 0.0531,
      "step": 371
    },
    {
      "epoch": 0.5577211394302849,
      "grad_norm": 0.04400477443557516,
      "learning_rate": 0.00018120835268918063,
      "loss": 0.0387,
      "step": 372
    },
    {
      "epoch": 0.5592203898050975,
      "grad_norm": 0.05568457669143035,
      "learning_rate": 0.00018105530383532607,
      "loss": 0.0359,
      "step": 373
    },
    {
      "epoch": 0.56071964017991,
      "grad_norm": 0.09375047939362961,
      "learning_rate": 0.00018090169943749476,
      "loss": 0.0613,
      "step": 374
    },
    {
      "epoch": 0.5622188905547226,
      "grad_norm": 0.09342104698058959,
      "learning_rate": 0.000180747540548474,
      "loss": 0.0508,
      "step": 375
    },
    {
      "epoch": 0.5637181409295352,
      "grad_norm": 0.06980360470280776,
      "learning_rate": 0.00018059282822485158,
      "loss": 0.0588,
      "step": 376
    },
    {
      "epoch": 0.5652173913043478,
      "grad_norm": 0.04834116988503355,
      "learning_rate": 0.00018043756352700846,
      "loss": 0.0407,
      "step": 377
    },
    {
      "epoch": 0.5667166416791605,
      "grad_norm": 0.0559433504674298,
      "learning_rate": 0.00018028174751911146,
      "loss": 0.0487,
      "step": 378
    },
    {
      "epoch": 0.568215892053973,
      "grad_norm": 0.04299769695781603,
      "learning_rate": 0.00018012538126910608,
      "loss": 0.0289,
      "step": 379
    },
    {
      "epoch": 0.5697151424287856,
      "grad_norm": 0.061484741377732285,
      "learning_rate": 0.00017996846584870908,
      "loss": 0.0545,
      "step": 380
    },
    {
      "epoch": 0.5712143928035982,
      "grad_norm": 0.04596560988589707,
      "learning_rate": 0.00017981100233340115,
      "loss": 0.0381,
      "step": 381
    },
    {
      "epoch": 0.5727136431784108,
      "grad_norm": 0.03818801182308544,
      "learning_rate": 0.00017965299180241963,
      "loss": 0.0348,
      "step": 382
    },
    {
      "epoch": 0.5742128935532234,
      "grad_norm": 0.050232900629482914,
      "learning_rate": 0.000179494435338751,
      "loss": 0.0265,
      "step": 383
    },
    {
      "epoch": 0.5757121439280359,
      "grad_norm": 0.0740772370322547,
      "learning_rate": 0.00017933533402912354,
      "loss": 0.0629,
      "step": 384
    },
    {
      "epoch": 0.5772113943028486,
      "grad_norm": 0.03602072499212665,
      "learning_rate": 0.00017917568896399973,
      "loss": 0.0303,
      "step": 385
    },
    {
      "epoch": 0.5787106446776612,
      "grad_norm": 0.04732712114588865,
      "learning_rate": 0.00017901550123756906,
      "loss": 0.0533,
      "step": 386
    },
    {
      "epoch": 0.5802098950524738,
      "grad_norm": 0.0438295195621145,
      "learning_rate": 0.0001788547719477402,
      "loss": 0.033,
      "step": 387
    },
    {
      "epoch": 0.5817091454272864,
      "grad_norm": 0.045622862481622464,
      "learning_rate": 0.00017869350219613375,
      "loss": 0.0436,
      "step": 388
    },
    {
      "epoch": 0.5832083958020989,
      "grad_norm": 0.04964994949451908,
      "learning_rate": 0.00017853169308807448,
      "loss": 0.0375,
      "step": 389
    },
    {
      "epoch": 0.5847076461769115,
      "grad_norm": 0.05078556363206241,
      "learning_rate": 0.000178369345732584,
      "loss": 0.0345,
      "step": 390
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 0.05074730082524994,
      "learning_rate": 0.00017820646124237278,
      "loss": 0.0364,
      "step": 391
    },
    {
      "epoch": 0.5877061469265368,
      "grad_norm": 0.04603113674846868,
      "learning_rate": 0.000178043040733833,
      "loss": 0.0302,
      "step": 392
    },
    {
      "epoch": 0.5892053973013494,
      "grad_norm": 0.058291821623076764,
      "learning_rate": 0.00017787908532703042,
      "loss": 0.0538,
      "step": 393
    },
    {
      "epoch": 0.5907046476761619,
      "grad_norm": 0.06209286221938147,
      "learning_rate": 0.0001777145961456971,
      "loss": 0.0437,
      "step": 394
    },
    {
      "epoch": 0.5922038980509745,
      "grad_norm": 0.035508544978000475,
      "learning_rate": 0.00017754957431722346,
      "loss": 0.0361,
      "step": 395
    },
    {
      "epoch": 0.5937031484257871,
      "grad_norm": 0.045021307147078936,
      "learning_rate": 0.00017738402097265064,
      "loss": 0.0314,
      "step": 396
    },
    {
      "epoch": 0.5952023988005997,
      "grad_norm": 0.045117529238231965,
      "learning_rate": 0.00017721793724666268,
      "loss": 0.0286,
      "step": 397
    },
    {
      "epoch": 0.5967016491754122,
      "grad_norm": 0.04284437809907477,
      "learning_rate": 0.00017705132427757895,
      "loss": 0.0316,
      "step": 398
    },
    {
      "epoch": 0.5982008995502249,
      "grad_norm": 0.05363788525676639,
      "learning_rate": 0.00017688418320734598,
      "loss": 0.0575,
      "step": 399
    },
    {
      "epoch": 0.5997001499250375,
      "grad_norm": 0.039147054068539446,
      "learning_rate": 0.00017671651518153,
      "loss": 0.0298,
      "step": 400
    },
    {
      "epoch": 0.6011994002998501,
      "grad_norm": 0.047103082273761164,
      "learning_rate": 0.00017654832134930882,
      "loss": 0.0332,
      "step": 401
    },
    {
      "epoch": 0.6026986506746627,
      "grad_norm": 0.03434268067696955,
      "learning_rate": 0.00017637960286346425,
      "loss": 0.0224,
      "step": 402
    },
    {
      "epoch": 0.6041979010494752,
      "grad_norm": 0.05227048698626022,
      "learning_rate": 0.00017621036088037378,
      "loss": 0.0518,
      "step": 403
    },
    {
      "epoch": 0.6056971514242878,
      "grad_norm": 0.18818259332861206,
      "learning_rate": 0.0001760405965600031,
      "loss": 0.0758,
      "step": 404
    },
    {
      "epoch": 0.6071964017991005,
      "grad_norm": 0.05885735282398511,
      "learning_rate": 0.00017587031106589782,
      "loss": 0.0361,
      "step": 405
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 0.08962878091622765,
      "learning_rate": 0.00017569950556517566,
      "loss": 0.0523,
      "step": 406
    },
    {
      "epoch": 0.6101949025487257,
      "grad_norm": 0.07406938968015182,
      "learning_rate": 0.00017552818122851838,
      "loss": 0.0372,
      "step": 407
    },
    {
      "epoch": 0.6116941529235382,
      "grad_norm": 0.0589541107505507,
      "learning_rate": 0.0001753563392301638,
      "loss": 0.0518,
      "step": 408
    },
    {
      "epoch": 0.6131934032983508,
      "grad_norm": 0.0493031824176551,
      "learning_rate": 0.00017518398074789775,
      "loss": 0.0404,
      "step": 409
    },
    {
      "epoch": 0.6146926536731634,
      "grad_norm": 0.060948463938890016,
      "learning_rate": 0.00017501110696304596,
      "loss": 0.0434,
      "step": 410
    },
    {
      "epoch": 0.616191904047976,
      "grad_norm": 0.05214572496276531,
      "learning_rate": 0.00017483771906046602,
      "loss": 0.0384,
      "step": 411
    },
    {
      "epoch": 0.6176911544227887,
      "grad_norm": 0.05632303506730485,
      "learning_rate": 0.00017466381822853915,
      "loss": 0.032,
      "step": 412
    },
    {
      "epoch": 0.6191904047976012,
      "grad_norm": 0.06363985932901699,
      "learning_rate": 0.00017448940565916222,
      "loss": 0.0532,
      "step": 413
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 0.045655772233298454,
      "learning_rate": 0.00017431448254773944,
      "loss": 0.0335,
      "step": 414
    },
    {
      "epoch": 0.6221889055472264,
      "grad_norm": 0.05496397917447876,
      "learning_rate": 0.00017413905009317423,
      "loss": 0.0432,
      "step": 415
    },
    {
      "epoch": 0.623688155922039,
      "grad_norm": 0.0542650773216564,
      "learning_rate": 0.000173963109497861,
      "loss": 0.0299,
      "step": 416
    },
    {
      "epoch": 0.6251874062968515,
      "grad_norm": 0.053416335915512446,
      "learning_rate": 0.00017378666196767685,
      "loss": 0.0384,
      "step": 417
    },
    {
      "epoch": 0.6266866566716641,
      "grad_norm": 0.053317686220209964,
      "learning_rate": 0.00017360970871197346,
      "loss": 0.0355,
      "step": 418
    },
    {
      "epoch": 0.6281859070464768,
      "grad_norm": 0.058288847449129474,
      "learning_rate": 0.00017343225094356855,
      "loss": 0.052,
      "step": 419
    },
    {
      "epoch": 0.6296851574212894,
      "grad_norm": 0.042009952684143474,
      "learning_rate": 0.0001732542898787379,
      "loss": 0.0452,
      "step": 420
    },
    {
      "epoch": 0.631184407796102,
      "grad_norm": 0.04265378963968585,
      "learning_rate": 0.00017307582673720663,
      "loss": 0.0314,
      "step": 421
    },
    {
      "epoch": 0.6326836581709145,
      "grad_norm": 0.0473560166231087,
      "learning_rate": 0.00017289686274214118,
      "loss": 0.0481,
      "step": 422
    },
    {
      "epoch": 0.6341829085457271,
      "grad_norm": 0.04316822944816478,
      "learning_rate": 0.00017271739912014068,
      "loss": 0.035,
      "step": 423
    },
    {
      "epoch": 0.6356821589205397,
      "grad_norm": 0.04821805663840275,
      "learning_rate": 0.00017253743710122875,
      "loss": 0.0353,
      "step": 424
    },
    {
      "epoch": 0.6371814092953523,
      "grad_norm": 0.05766150770706173,
      "learning_rate": 0.00017235697791884494,
      "loss": 0.0518,
      "step": 425
    },
    {
      "epoch": 0.638680659670165,
      "grad_norm": 0.09048982025586255,
      "learning_rate": 0.00017217602280983623,
      "loss": 0.0477,
      "step": 426
    },
    {
      "epoch": 0.6401799100449775,
      "grad_norm": 0.04016296087640653,
      "learning_rate": 0.00017199457301444868,
      "loss": 0.0284,
      "step": 427
    },
    {
      "epoch": 0.6416791604197901,
      "grad_norm": 0.07151231389242098,
      "learning_rate": 0.00017181262977631888,
      "loss": 0.0534,
      "step": 428
    },
    {
      "epoch": 0.6431784107946027,
      "grad_norm": 0.0521164376523254,
      "learning_rate": 0.00017163019434246547,
      "loss": 0.0413,
      "step": 429
    },
    {
      "epoch": 0.6446776611694153,
      "grad_norm": 0.06650764446753712,
      "learning_rate": 0.00017144726796328034,
      "loss": 0.0532,
      "step": 430
    },
    {
      "epoch": 0.6461769115442278,
      "grad_norm": 0.05631153402061295,
      "learning_rate": 0.00017126385189252053,
      "loss": 0.04,
      "step": 431
    },
    {
      "epoch": 0.6476761619190404,
      "grad_norm": 0.07908703283560595,
      "learning_rate": 0.00017107994738729926,
      "loss": 0.0546,
      "step": 432
    },
    {
      "epoch": 0.6491754122938531,
      "grad_norm": 0.053983476730184235,
      "learning_rate": 0.00017089555570807736,
      "loss": 0.0428,
      "step": 433
    },
    {
      "epoch": 0.6506746626686657,
      "grad_norm": 0.04223485749838613,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.039,
      "step": 434
    },
    {
      "epoch": 0.6521739130434783,
      "grad_norm": 0.05289448745150468,
      "learning_rate": 0.00017052531588616178,
      "loss": 0.0369,
      "step": 435
    },
    {
      "epoch": 0.6536731634182908,
      "grad_norm": 0.04254188142407335,
      "learning_rate": 0.00017033947028105039,
      "loss": 0.0482,
      "step": 436
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 0.03642799958156187,
      "learning_rate": 0.0001701531425770856,
      "loss": 0.0227,
      "step": 437
    },
    {
      "epoch": 0.656671664167916,
      "grad_norm": 0.044290202162954866,
      "learning_rate": 0.00016996633405133655,
      "loss": 0.0317,
      "step": 438
    },
    {
      "epoch": 0.6581709145427287,
      "grad_norm": 0.03328713625145707,
      "learning_rate": 0.00016977904598416803,
      "loss": 0.0193,
      "step": 439
    },
    {
      "epoch": 0.6596701649175413,
      "grad_norm": 0.06825872867733464,
      "learning_rate": 0.00016959127965923142,
      "loss": 0.0524,
      "step": 440
    },
    {
      "epoch": 0.6611694152923538,
      "grad_norm": 0.05737656538911755,
      "learning_rate": 0.00016940303636345618,
      "loss": 0.0384,
      "step": 441
    },
    {
      "epoch": 0.6626686656671664,
      "grad_norm": 0.05217169033909698,
      "learning_rate": 0.0001692143173870407,
      "loss": 0.0398,
      "step": 442
    },
    {
      "epoch": 0.664167916041979,
      "grad_norm": 0.08029746938521902,
      "learning_rate": 0.00016902512402344373,
      "loss": 0.0996,
      "step": 443
    },
    {
      "epoch": 0.6656671664167916,
      "grad_norm": 0.05023109850591579,
      "learning_rate": 0.0001688354575693754,
      "loss": 0.0444,
      "step": 444
    },
    {
      "epoch": 0.6671664167916042,
      "grad_norm": 0.05007534396139674,
      "learning_rate": 0.00016864531932478838,
      "loss": 0.0495,
      "step": 445
    },
    {
      "epoch": 0.6686656671664168,
      "grad_norm": 0.06850886410730117,
      "learning_rate": 0.00016845471059286887,
      "loss": 0.0491,
      "step": 446
    },
    {
      "epoch": 0.6701649175412294,
      "grad_norm": 0.0709049222949615,
      "learning_rate": 0.00016826363268002782,
      "loss": 0.086,
      "step": 447
    },
    {
      "epoch": 0.671664167916042,
      "grad_norm": 0.07625185163942867,
      "learning_rate": 0.0001680720868958918,
      "loss": 0.0612,
      "step": 448
    },
    {
      "epoch": 0.6731634182908546,
      "grad_norm": 0.03903647995204454,
      "learning_rate": 0.0001678800745532942,
      "loss": 0.0286,
      "step": 449
    },
    {
      "epoch": 0.6746626686656672,
      "grad_norm": 0.04875227954727952,
      "learning_rate": 0.00016768759696826608,
      "loss": 0.0452,
      "step": 450
    },
    {
      "epoch": 0.6761619190404797,
      "grad_norm": 0.045509076234092845,
      "learning_rate": 0.0001674946554600273,
      "loss": 0.0401,
      "step": 451
    },
    {
      "epoch": 0.6776611694152923,
      "grad_norm": 0.038954973560474074,
      "learning_rate": 0.00016730125135097735,
      "loss": 0.0428,
      "step": 452
    },
    {
      "epoch": 0.679160419790105,
      "grad_norm": 0.05064178738289288,
      "learning_rate": 0.00016710738596668632,
      "loss": 0.0481,
      "step": 453
    },
    {
      "epoch": 0.6806596701649176,
      "grad_norm": 0.03652262635297044,
      "learning_rate": 0.00016691306063588583,
      "loss": 0.0341,
      "step": 454
    },
    {
      "epoch": 0.6821589205397302,
      "grad_norm": 0.0469218301221786,
      "learning_rate": 0.00016671827669045998,
      "loss": 0.0388,
      "step": 455
    },
    {
      "epoch": 0.6836581709145427,
      "grad_norm": 0.05748273212579017,
      "learning_rate": 0.00016652303546543608,
      "loss": 0.0424,
      "step": 456
    },
    {
      "epoch": 0.6851574212893553,
      "grad_norm": 0.029734098462886188,
      "learning_rate": 0.00016632733829897566,
      "loss": 0.0221,
      "step": 457
    },
    {
      "epoch": 0.6866566716641679,
      "grad_norm": 0.03162041588945904,
      "learning_rate": 0.00016613118653236518,
      "loss": 0.0284,
      "step": 458
    },
    {
      "epoch": 0.6881559220389805,
      "grad_norm": 0.0770559271143547,
      "learning_rate": 0.00016593458151000688,
      "loss": 0.0647,
      "step": 459
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 0.0806809324128278,
      "learning_rate": 0.0001657375245794096,
      "loss": 0.0507,
      "step": 460
    },
    {
      "epoch": 0.6911544227886057,
      "grad_norm": 0.0623039707244334,
      "learning_rate": 0.0001655400170911794,
      "loss": 0.0443,
      "step": 461
    },
    {
      "epoch": 0.6926536731634183,
      "grad_norm": 0.04046812960090323,
      "learning_rate": 0.00016534206039901057,
      "loss": 0.047,
      "step": 462
    },
    {
      "epoch": 0.6941529235382309,
      "grad_norm": 0.09437551047944975,
      "learning_rate": 0.00016514365585967603,
      "loss": 0.0626,
      "step": 463
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.07162777105996744,
      "learning_rate": 0.00016494480483301836,
      "loss": 0.0503,
      "step": 464
    },
    {
      "epoch": 0.697151424287856,
      "grad_norm": 0.04109223749911729,
      "learning_rate": 0.00016474550868194023,
      "loss": 0.0315,
      "step": 465
    },
    {
      "epoch": 0.6986506746626686,
      "grad_norm": 0.06077734184621873,
      "learning_rate": 0.00016454576877239507,
      "loss": 0.0422,
      "step": 466
    },
    {
      "epoch": 0.7001499250374813,
      "grad_norm": 0.19884111549839148,
      "learning_rate": 0.0001643455864733779,
      "loss": 0.0926,
      "step": 467
    },
    {
      "epoch": 0.7016491754122939,
      "grad_norm": 0.04642784378882251,
      "learning_rate": 0.00016414496315691581,
      "loss": 0.0385,
      "step": 468
    },
    {
      "epoch": 0.7031484257871065,
      "grad_norm": 0.4393223814451174,
      "learning_rate": 0.00016394390019805848,
      "loss": 0.0993,
      "step": 469
    },
    {
      "epoch": 0.704647676161919,
      "grad_norm": 0.07482643775181062,
      "learning_rate": 0.000163742398974869,
      "loss": 0.0489,
      "step": 470
    },
    {
      "epoch": 0.7061469265367316,
      "grad_norm": 0.04885944436540397,
      "learning_rate": 0.0001635404608684141,
      "loss": 0.0348,
      "step": 471
    },
    {
      "epoch": 0.7076461769115442,
      "grad_norm": 0.049664390973816655,
      "learning_rate": 0.000163338087262755,
      "loss": 0.0391,
      "step": 472
    },
    {
      "epoch": 0.7091454272863568,
      "grad_norm": 0.09885583055337595,
      "learning_rate": 0.00016313527954493778,
      "loss": 0.0642,
      "step": 473
    },
    {
      "epoch": 0.7106446776611695,
      "grad_norm": 0.06584763072102895,
      "learning_rate": 0.00016293203910498376,
      "loss": 0.0446,
      "step": 474
    },
    {
      "epoch": 0.712143928035982,
      "grad_norm": 0.052090571949195315,
      "learning_rate": 0.00016272836733588017,
      "loss": 0.0449,
      "step": 475
    },
    {
      "epoch": 0.7136431784107946,
      "grad_norm": 0.060369511284079136,
      "learning_rate": 0.00016252426563357055,
      "loss": 0.0498,
      "step": 476
    },
    {
      "epoch": 0.7151424287856072,
      "grad_norm": 0.07350090323203447,
      "learning_rate": 0.00016231973539694504,
      "loss": 0.0594,
      "step": 477
    },
    {
      "epoch": 0.7166416791604198,
      "grad_norm": 0.07477778304256645,
      "learning_rate": 0.00016211477802783103,
      "loss": 0.0339,
      "step": 478
    },
    {
      "epoch": 0.7181409295352323,
      "grad_norm": 0.044909007459069075,
      "learning_rate": 0.00016190939493098344,
      "loss": 0.0379,
      "step": 479
    },
    {
      "epoch": 0.719640179910045,
      "grad_norm": 0.062190968309072295,
      "learning_rate": 0.00016170358751407487,
      "loss": 0.0758,
      "step": 480
    },
    {
      "epoch": 0.7211394302848576,
      "grad_norm": 0.05377348813315943,
      "learning_rate": 0.00016149735718768644,
      "loss": 0.0329,
      "step": 481
    },
    {
      "epoch": 0.7226386806596702,
      "grad_norm": 0.053690857438836906,
      "learning_rate": 0.00016129070536529766,
      "loss": 0.037,
      "step": 482
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 0.053806227909165436,
      "learning_rate": 0.000161083633463277,
      "loss": 0.0429,
      "step": 483
    },
    {
      "epoch": 0.7256371814092953,
      "grad_norm": 0.058550946973171134,
      "learning_rate": 0.00016087614290087208,
      "loss": 0.0573,
      "step": 484
    },
    {
      "epoch": 0.7271364317841079,
      "grad_norm": 0.05208602315147625,
      "learning_rate": 0.00016066823510019998,
      "loss": 0.0322,
      "step": 485
    },
    {
      "epoch": 0.7286356821589205,
      "grad_norm": 0.048780008222607175,
      "learning_rate": 0.0001604599114862375,
      "loss": 0.0391,
      "step": 486
    },
    {
      "epoch": 0.7301349325337332,
      "grad_norm": 0.04257645362681531,
      "learning_rate": 0.00016025117348681132,
      "loss": 0.0326,
      "step": 487
    },
    {
      "epoch": 0.7316341829085458,
      "grad_norm": 0.049273729257806875,
      "learning_rate": 0.00016004202253258842,
      "loss": 0.0446,
      "step": 488
    },
    {
      "epoch": 0.7331334332833583,
      "grad_norm": 0.05564516630711474,
      "learning_rate": 0.00015983246005706593,
      "loss": 0.0567,
      "step": 489
    },
    {
      "epoch": 0.7346326836581709,
      "grad_norm": 0.05015891963593426,
      "learning_rate": 0.0001596224874965616,
      "loss": 0.0365,
      "step": 490
    },
    {
      "epoch": 0.7361319340329835,
      "grad_norm": 0.04227849702908071,
      "learning_rate": 0.00015941210629020388,
      "loss": 0.0278,
      "step": 491
    },
    {
      "epoch": 0.7376311844077961,
      "grad_norm": 0.09190608644561755,
      "learning_rate": 0.00015920131787992197,
      "loss": 0.1024,
      "step": 492
    },
    {
      "epoch": 0.7391304347826086,
      "grad_norm": 0.11304924052426712,
      "learning_rate": 0.00015899012371043604,
      "loss": 0.0675,
      "step": 493
    },
    {
      "epoch": 0.7406296851574213,
      "grad_norm": 0.05613518703388969,
      "learning_rate": 0.00015877852522924732,
      "loss": 0.0407,
      "step": 494
    },
    {
      "epoch": 0.7421289355322339,
      "grad_norm": 0.04754947301267445,
      "learning_rate": 0.00015856652388662808,
      "loss": 0.0305,
      "step": 495
    },
    {
      "epoch": 0.7436281859070465,
      "grad_norm": 0.0487917467215864,
      "learning_rate": 0.00015835412113561175,
      "loss": 0.0372,
      "step": 496
    },
    {
      "epoch": 0.7451274362818591,
      "grad_norm": 0.056051719216029486,
      "learning_rate": 0.00015814131843198308,
      "loss": 0.033,
      "step": 497
    },
    {
      "epoch": 0.7466266866566716,
      "grad_norm": 0.08355105349400162,
      "learning_rate": 0.0001579281172342679,
      "loss": 0.0618,
      "step": 498
    },
    {
      "epoch": 0.7481259370314842,
      "grad_norm": 0.09016322211305738,
      "learning_rate": 0.0001577145190037234,
      "loss": 0.047,
      "step": 499
    },
    {
      "epoch": 0.7496251874062968,
      "grad_norm": 0.04982867112956479,
      "learning_rate": 0.00015750052520432787,
      "loss": 0.0313,
      "step": 500
    },
    {
      "epoch": 0.7511244377811095,
      "grad_norm": 0.04381518020575622,
      "learning_rate": 0.0001572861373027709,
      "loss": 0.0348,
      "step": 501
    },
    {
      "epoch": 0.7526236881559221,
      "grad_norm": 0.05899789350925751,
      "learning_rate": 0.0001570713567684432,
      "loss": 0.0503,
      "step": 502
    },
    {
      "epoch": 0.7541229385307346,
      "grad_norm": 0.06316265954005644,
      "learning_rate": 0.0001568561850734264,
      "loss": 0.0291,
      "step": 503
    },
    {
      "epoch": 0.7556221889055472,
      "grad_norm": 0.08632078705291048,
      "learning_rate": 0.00015664062369248328,
      "loss": 0.0639,
      "step": 504
    },
    {
      "epoch": 0.7571214392803598,
      "grad_norm": 0.050182212371613015,
      "learning_rate": 0.00015642467410304742,
      "loss": 0.0305,
      "step": 505
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 0.07158241120198346,
      "learning_rate": 0.00015620833778521307,
      "loss": 0.0573,
      "step": 506
    },
    {
      "epoch": 0.760119940029985,
      "grad_norm": 0.08175078694638557,
      "learning_rate": 0.00015599161622172517,
      "loss": 0.0462,
      "step": 507
    },
    {
      "epoch": 0.7616191904047976,
      "grad_norm": 0.0475536897913194,
      "learning_rate": 0.00015577451089796905,
      "loss": 0.0501,
      "step": 508
    },
    {
      "epoch": 0.7631184407796102,
      "grad_norm": 0.05682015512029848,
      "learning_rate": 0.00015555702330196023,
      "loss": 0.0313,
      "step": 509
    },
    {
      "epoch": 0.7646176911544228,
      "grad_norm": 0.06137429357639015,
      "learning_rate": 0.00015533915492433443,
      "loss": 0.0519,
      "step": 510
    },
    {
      "epoch": 0.7661169415292354,
      "grad_norm": 0.038591640884982854,
      "learning_rate": 0.00015512090725833705,
      "loss": 0.0414,
      "step": 511
    },
    {
      "epoch": 0.767616191904048,
      "grad_norm": 0.055149664940849136,
      "learning_rate": 0.0001549022817998132,
      "loss": 0.0318,
      "step": 512
    },
    {
      "epoch": 0.7691154422788605,
      "grad_norm": 0.04445225405210936,
      "learning_rate": 0.00015468328004719726,
      "loss": 0.041,
      "step": 513
    },
    {
      "epoch": 0.7706146926536732,
      "grad_norm": 0.05691241429851254,
      "learning_rate": 0.00015446390350150273,
      "loss": 0.0379,
      "step": 514
    },
    {
      "epoch": 0.7721139430284858,
      "grad_norm": 0.05761804723554256,
      "learning_rate": 0.00015424415366631188,
      "loss": 0.0363,
      "step": 515
    },
    {
      "epoch": 0.7736131934032984,
      "grad_norm": 0.04929119952169534,
      "learning_rate": 0.0001540240320477655,
      "loss": 0.0426,
      "step": 516
    },
    {
      "epoch": 0.775112443778111,
      "grad_norm": 0.044305931962162155,
      "learning_rate": 0.0001538035401545525,
      "loss": 0.0422,
      "step": 517
    },
    {
      "epoch": 0.7766116941529235,
      "grad_norm": 0.06205510180982029,
      "learning_rate": 0.00015358267949789966,
      "loss": 0.0603,
      "step": 518
    },
    {
      "epoch": 0.7781109445277361,
      "grad_norm": 0.07453649068372155,
      "learning_rate": 0.00015336145159156115,
      "loss": 0.0543,
      "step": 519
    },
    {
      "epoch": 0.7796101949025487,
      "grad_norm": 0.0378467187255617,
      "learning_rate": 0.00015313985795180828,
      "loss": 0.04,
      "step": 520
    },
    {
      "epoch": 0.7811094452773614,
      "grad_norm": 0.057596547352352555,
      "learning_rate": 0.00015291790009741907,
      "loss": 0.0451,
      "step": 521
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 0.04081926428494982,
      "learning_rate": 0.00015269557954966778,
      "loss": 0.0288,
      "step": 522
    },
    {
      "epoch": 0.7841079460269865,
      "grad_norm": 0.041887534030808915,
      "learning_rate": 0.00015247289783231453,
      "loss": 0.035,
      "step": 523
    },
    {
      "epoch": 0.7856071964017991,
      "grad_norm": 0.06860023926519257,
      "learning_rate": 0.0001522498564715949,
      "loss": 0.049,
      "step": 524
    },
    {
      "epoch": 0.7871064467766117,
      "grad_norm": 0.04260551619067463,
      "learning_rate": 0.0001520264569962094,
      "loss": 0.0427,
      "step": 525
    },
    {
      "epoch": 0.7886056971514243,
      "grad_norm": 0.03773306965490992,
      "learning_rate": 0.00015180270093731303,
      "loss": 0.0245,
      "step": 526
    },
    {
      "epoch": 0.7901049475262368,
      "grad_norm": 0.03431124298471051,
      "learning_rate": 0.00015157858982850475,
      "loss": 0.0289,
      "step": 527
    },
    {
      "epoch": 0.7916041979010495,
      "grad_norm": 0.07692552283154377,
      "learning_rate": 0.00015135412520581702,
      "loss": 0.0661,
      "step": 528
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 0.0347019843571733,
      "learning_rate": 0.0001511293086077052,
      "loss": 0.0267,
      "step": 529
    },
    {
      "epoch": 0.7946026986506747,
      "grad_norm": 0.05889548183130433,
      "learning_rate": 0.00015090414157503714,
      "loss": 0.0321,
      "step": 530
    },
    {
      "epoch": 0.7961019490254873,
      "grad_norm": 0.07066481807133856,
      "learning_rate": 0.0001506786256510824,
      "loss": 0.052,
      "step": 531
    },
    {
      "epoch": 0.7976011994002998,
      "grad_norm": 0.06186361286136524,
      "learning_rate": 0.00015045276238150192,
      "loss": 0.0479,
      "step": 532
    },
    {
      "epoch": 0.7991004497751124,
      "grad_norm": 0.061605402385295015,
      "learning_rate": 0.00015022655331433727,
      "loss": 0.0523,
      "step": 533
    },
    {
      "epoch": 0.800599700149925,
      "grad_norm": 0.04168389823340015,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.0293,
      "step": 534
    },
    {
      "epoch": 0.8020989505247377,
      "grad_norm": 0.04409075971831587,
      "learning_rate": 0.00014977310399126123,
      "loss": 0.0294,
      "step": 535
    },
    {
      "epoch": 0.8035982008995503,
      "grad_norm": 0.03449538576446893,
      "learning_rate": 0.00014954586684324078,
      "loss": 0.0273,
      "step": 536
    },
    {
      "epoch": 0.8050974512743628,
      "grad_norm": 0.06487922260997829,
      "learning_rate": 0.00014931829011339659,
      "loss": 0.1013,
      "step": 537
    },
    {
      "epoch": 0.8065967016491754,
      "grad_norm": 0.04080309597694205,
      "learning_rate": 0.00014909037536151409,
      "loss": 0.0271,
      "step": 538
    },
    {
      "epoch": 0.808095952023988,
      "grad_norm": 0.039495017673297265,
      "learning_rate": 0.00014886212414969553,
      "loss": 0.0274,
      "step": 539
    },
    {
      "epoch": 0.8095952023988006,
      "grad_norm": 0.08062160395086124,
      "learning_rate": 0.00014863353804234905,
      "loss": 0.0417,
      "step": 540
    },
    {
      "epoch": 0.8110944527736131,
      "grad_norm": 0.05152809022416075,
      "learning_rate": 0.00014840461860617834,
      "loss": 0.0359,
      "step": 541
    },
    {
      "epoch": 0.8125937031484258,
      "grad_norm": 0.04922307205252768,
      "learning_rate": 0.00014817536741017152,
      "loss": 0.0308,
      "step": 542
    },
    {
      "epoch": 0.8140929535232384,
      "grad_norm": 0.07044648065491671,
      "learning_rate": 0.00014794578602559067,
      "loss": 0.0566,
      "step": 543
    },
    {
      "epoch": 0.815592203898051,
      "grad_norm": 0.0465206984979631,
      "learning_rate": 0.00014771587602596084,
      "loss": 0.0513,
      "step": 544
    },
    {
      "epoch": 0.8170914542728636,
      "grad_norm": 0.09746747148479479,
      "learning_rate": 0.00014748563898705946,
      "loss": 0.0377,
      "step": 545
    },
    {
      "epoch": 0.8185907046476761,
      "grad_norm": 0.050647534375834816,
      "learning_rate": 0.00014725507648690543,
      "loss": 0.048,
      "step": 546
    },
    {
      "epoch": 0.8200899550224887,
      "grad_norm": 0.05429153224198516,
      "learning_rate": 0.00014702419010574825,
      "loss": 0.0523,
      "step": 547
    },
    {
      "epoch": 0.8215892053973014,
      "grad_norm": 0.036339453217671985,
      "learning_rate": 0.00014679298142605734,
      "loss": 0.0233,
      "step": 548
    },
    {
      "epoch": 0.823088455772114,
      "grad_norm": 0.08317522768191385,
      "learning_rate": 0.00014656145203251114,
      "loss": 0.0619,
      "step": 549
    },
    {
      "epoch": 0.8245877061469266,
      "grad_norm": 0.04231199661380915,
      "learning_rate": 0.00014632960351198618,
      "loss": 0.0477,
      "step": 550
    },
    {
      "epoch": 0.8260869565217391,
      "grad_norm": 0.1779970750798073,
      "learning_rate": 0.00014609743745354624,
      "loss": 0.0978,
      "step": 551
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 0.06735162872665944,
      "learning_rate": 0.00014586495544843152,
      "loss": 0.0435,
      "step": 552
    },
    {
      "epoch": 0.8290854572713643,
      "grad_norm": 0.06794928340973848,
      "learning_rate": 0.0001456321590900476,
      "loss": 0.0548,
      "step": 553
    },
    {
      "epoch": 0.8305847076461769,
      "grad_norm": 0.040691650099011545,
      "learning_rate": 0.00014539904997395468,
      "loss": 0.0307,
      "step": 554
    },
    {
      "epoch": 0.8320839580209896,
      "grad_norm": 0.06310347101508698,
      "learning_rate": 0.00014516562969785657,
      "loss": 0.0518,
      "step": 555
    },
    {
      "epoch": 0.8335832083958021,
      "grad_norm": 0.06600280603320972,
      "learning_rate": 0.00014493189986158965,
      "loss": 0.0283,
      "step": 556
    },
    {
      "epoch": 0.8350824587706147,
      "grad_norm": 0.048741740097207054,
      "learning_rate": 0.00014469786206711214,
      "loss": 0.0369,
      "step": 557
    },
    {
      "epoch": 0.8365817091454273,
      "grad_norm": 0.0541340846516407,
      "learning_rate": 0.00014446351791849276,
      "loss": 0.0408,
      "step": 558
    },
    {
      "epoch": 0.8380809595202399,
      "grad_norm": 0.06569812050813671,
      "learning_rate": 0.00014422886902190014,
      "loss": 0.0559,
      "step": 559
    },
    {
      "epoch": 0.8395802098950524,
      "grad_norm": 0.07697796259986985,
      "learning_rate": 0.00014399391698559152,
      "loss": 0.0359,
      "step": 560
    },
    {
      "epoch": 0.841079460269865,
      "grad_norm": 0.043380007173197774,
      "learning_rate": 0.00014375866341990187,
      "loss": 0.0307,
      "step": 561
    },
    {
      "epoch": 0.8425787106446777,
      "grad_norm": 0.06012928930562709,
      "learning_rate": 0.00014352310993723277,
      "loss": 0.0359,
      "step": 562
    },
    {
      "epoch": 0.8440779610194903,
      "grad_norm": 0.05916041881920503,
      "learning_rate": 0.00014328725815204144,
      "loss": 0.0451,
      "step": 563
    },
    {
      "epoch": 0.8455772113943029,
      "grad_norm": 0.04584559728661184,
      "learning_rate": 0.00014305110968082952,
      "loss": 0.0295,
      "step": 564
    },
    {
      "epoch": 0.8470764617691154,
      "grad_norm": 0.048484148060844064,
      "learning_rate": 0.0001428146661421323,
      "loss": 0.0413,
      "step": 565
    },
    {
      "epoch": 0.848575712143928,
      "grad_norm": 0.051087086585204396,
      "learning_rate": 0.00014257792915650728,
      "loss": 0.0377,
      "step": 566
    },
    {
      "epoch": 0.8500749625187406,
      "grad_norm": 0.04873404112533848,
      "learning_rate": 0.00014234090034652324,
      "loss": 0.0426,
      "step": 567
    },
    {
      "epoch": 0.8515742128935532,
      "grad_norm": 0.05766048192964452,
      "learning_rate": 0.00014210358133674912,
      "loss": 0.033,
      "step": 568
    },
    {
      "epoch": 0.8530734632683659,
      "grad_norm": 0.054308917320710616,
      "learning_rate": 0.0001418659737537428,
      "loss": 0.0379,
      "step": 569
    },
    {
      "epoch": 0.8545727136431784,
      "grad_norm": 0.0460844103779593,
      "learning_rate": 0.00014162807922604012,
      "loss": 0.0289,
      "step": 570
    },
    {
      "epoch": 0.856071964017991,
      "grad_norm": 0.04438781665413021,
      "learning_rate": 0.00014138989938414348,
      "loss": 0.0265,
      "step": 571
    },
    {
      "epoch": 0.8575712143928036,
      "grad_norm": 0.1327712159880055,
      "learning_rate": 0.00014115143586051088,
      "loss": 0.0676,
      "step": 572
    },
    {
      "epoch": 0.8590704647676162,
      "grad_norm": 0.050894530499255325,
      "learning_rate": 0.00014091269028954458,
      "loss": 0.0438,
      "step": 573
    },
    {
      "epoch": 0.8605697151424287,
      "grad_norm": 0.05468020442615518,
      "learning_rate": 0.00014067366430758004,
      "loss": 0.042,
      "step": 574
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 0.09435379755279395,
      "learning_rate": 0.00014043435955287452,
      "loss": 0.0599,
      "step": 575
    },
    {
      "epoch": 0.863568215892054,
      "grad_norm": 0.046063192994359826,
      "learning_rate": 0.00014019477766559604,
      "loss": 0.0348,
      "step": 576
    },
    {
      "epoch": 0.8650674662668666,
      "grad_norm": 0.052806710216832004,
      "learning_rate": 0.00013995492028781202,
      "loss": 0.0432,
      "step": 577
    },
    {
      "epoch": 0.8665667166416792,
      "grad_norm": 0.05319984444801645,
      "learning_rate": 0.00013971478906347806,
      "loss": 0.0362,
      "step": 578
    },
    {
      "epoch": 0.8680659670164917,
      "grad_norm": 0.04452467143786472,
      "learning_rate": 0.0001394743856384267,
      "loss": 0.0314,
      "step": 579
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 0.05329898478614524,
      "learning_rate": 0.00013923371166035616,
      "loss": 0.0464,
      "step": 580
    },
    {
      "epoch": 0.8710644677661169,
      "grad_norm": 0.060886833574884516,
      "learning_rate": 0.00013899276877881884,
      "loss": 0.059,
      "step": 581
    },
    {
      "epoch": 0.8725637181409296,
      "grad_norm": 0.05294256278470512,
      "learning_rate": 0.0001387515586452103,
      "loss": 0.046,
      "step": 582
    },
    {
      "epoch": 0.8740629685157422,
      "grad_norm": 0.07449191398493356,
      "learning_rate": 0.00013851008291275782,
      "loss": 0.0343,
      "step": 583
    },
    {
      "epoch": 0.8755622188905547,
      "grad_norm": 0.06943351016979123,
      "learning_rate": 0.000138268343236509,
      "loss": 0.0558,
      "step": 584
    },
    {
      "epoch": 0.8770614692653673,
      "grad_norm": 0.06652499644935615,
      "learning_rate": 0.0001380263412733205,
      "loss": 0.0451,
      "step": 585
    },
    {
      "epoch": 0.8785607196401799,
      "grad_norm": 0.05833246316842718,
      "learning_rate": 0.00013778407868184672,
      "loss": 0.0457,
      "step": 586
    },
    {
      "epoch": 0.8800599700149925,
      "grad_norm": 0.04003719271959368,
      "learning_rate": 0.00013754155712252832,
      "loss": 0.0283,
      "step": 587
    },
    {
      "epoch": 0.881559220389805,
      "grad_norm": 0.05607671087323687,
      "learning_rate": 0.0001372987782575809,
      "loss": 0.04,
      "step": 588
    },
    {
      "epoch": 0.8830584707646177,
      "grad_norm": 0.040134500394464416,
      "learning_rate": 0.00013705574375098365,
      "loss": 0.0385,
      "step": 589
    },
    {
      "epoch": 0.8845577211394303,
      "grad_norm": 0.06555335932088245,
      "learning_rate": 0.00013681245526846783,
      "loss": 0.072,
      "step": 590
    },
    {
      "epoch": 0.8860569715142429,
      "grad_norm": 0.05066618534250665,
      "learning_rate": 0.00013656891447750544,
      "loss": 0.0428,
      "step": 591
    },
    {
      "epoch": 0.8875562218890555,
      "grad_norm": 0.04391780563799797,
      "learning_rate": 0.00013632512304729785,
      "loss": 0.0367,
      "step": 592
    },
    {
      "epoch": 0.889055472263868,
      "grad_norm": 0.06116296036386768,
      "learning_rate": 0.0001360810826487642,
      "loss": 0.0462,
      "step": 593
    },
    {
      "epoch": 0.8905547226386806,
      "grad_norm": 0.053934832910471985,
      "learning_rate": 0.00013583679495453,
      "loss": 0.046,
      "step": 594
    },
    {
      "epoch": 0.8920539730134932,
      "grad_norm": 0.045163829899199405,
      "learning_rate": 0.0001355922616389159,
      "loss": 0.0362,
      "step": 595
    },
    {
      "epoch": 0.8935532233883059,
      "grad_norm": 0.08655848208121547,
      "learning_rate": 0.00013534748437792573,
      "loss": 0.0406,
      "step": 596
    },
    {
      "epoch": 0.8950524737631185,
      "grad_norm": 0.05391530518298683,
      "learning_rate": 0.00013510246484923547,
      "loss": 0.0293,
      "step": 597
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 0.04845707242796894,
      "learning_rate": 0.00013485720473218154,
      "loss": 0.0299,
      "step": 598
    },
    {
      "epoch": 0.8980509745127436,
      "grad_norm": 0.04916815685149742,
      "learning_rate": 0.0001346117057077493,
      "loss": 0.0386,
      "step": 599
    },
    {
      "epoch": 0.8995502248875562,
      "grad_norm": 0.06617662307537282,
      "learning_rate": 0.00013436596945856164,
      "loss": 0.0501,
      "step": 600
    },
    {
      "epoch": 0.9010494752623688,
      "grad_norm": 0.03991924636314799,
      "learning_rate": 0.0001341199976688672,
      "loss": 0.0282,
      "step": 601
    },
    {
      "epoch": 0.9025487256371814,
      "grad_norm": 0.050388728813716206,
      "learning_rate": 0.00013387379202452917,
      "loss": 0.0352,
      "step": 602
    },
    {
      "epoch": 0.904047976011994,
      "grad_norm": 0.03893791860248331,
      "learning_rate": 0.0001336273542130134,
      "loss": 0.021,
      "step": 603
    },
    {
      "epoch": 0.9055472263868066,
      "grad_norm": 0.059548986139220876,
      "learning_rate": 0.0001333806859233771,
      "loss": 0.0798,
      "step": 604
    },
    {
      "epoch": 0.9070464767616192,
      "grad_norm": 0.04599439337421781,
      "learning_rate": 0.0001331337888462571,
      "loss": 0.0351,
      "step": 605
    },
    {
      "epoch": 0.9085457271364318,
      "grad_norm": 0.05203471422944932,
      "learning_rate": 0.00013288666467385833,
      "loss": 0.0295,
      "step": 606
    },
    {
      "epoch": 0.9100449775112444,
      "grad_norm": 0.055165249926359204,
      "learning_rate": 0.0001326393150999422,
      "loss": 0.0345,
      "step": 607
    },
    {
      "epoch": 0.9115442278860569,
      "grad_norm": 0.13139481496590286,
      "learning_rate": 0.00013239174181981495,
      "loss": 0.0534,
      "step": 608
    },
    {
      "epoch": 0.9130434782608695,
      "grad_norm": 0.04500674501515753,
      "learning_rate": 0.00013214394653031616,
      "loss": 0.0203,
      "step": 609
    },
    {
      "epoch": 0.9145427286356822,
      "grad_norm": 0.05584447753073881,
      "learning_rate": 0.00013189593092980702,
      "loss": 0.0344,
      "step": 610
    },
    {
      "epoch": 0.9160419790104948,
      "grad_norm": 0.041591690138825445,
      "learning_rate": 0.00013164769671815862,
      "loss": 0.0201,
      "step": 611
    },
    {
      "epoch": 0.9175412293853074,
      "grad_norm": 0.05297004561152026,
      "learning_rate": 0.00013139924559674052,
      "loss": 0.0381,
      "step": 612
    },
    {
      "epoch": 0.9190404797601199,
      "grad_norm": 0.04848305915244054,
      "learning_rate": 0.00013115057926840884,
      "loss": 0.0349,
      "step": 613
    },
    {
      "epoch": 0.9205397301349325,
      "grad_norm": 0.03324583015768157,
      "learning_rate": 0.00013090169943749476,
      "loss": 0.028,
      "step": 614
    },
    {
      "epoch": 0.9220389805097451,
      "grad_norm": 0.06071140630728308,
      "learning_rate": 0.0001306526078097928,
      "loss": 0.0407,
      "step": 615
    },
    {
      "epoch": 0.9235382308845578,
      "grad_norm": 0.055677239200326835,
      "learning_rate": 0.00013040330609254903,
      "loss": 0.0362,
      "step": 616
    },
    {
      "epoch": 0.9250374812593704,
      "grad_norm": 0.05138799744779293,
      "learning_rate": 0.00013015379599444957,
      "loss": 0.0549,
      "step": 617
    },
    {
      "epoch": 0.9265367316341829,
      "grad_norm": 0.053161721439074654,
      "learning_rate": 0.00012990407922560868,
      "loss": 0.0468,
      "step": 618
    },
    {
      "epoch": 0.9280359820089955,
      "grad_norm": 0.0638192639644002,
      "learning_rate": 0.00012965415749755709,
      "loss": 0.0518,
      "step": 619
    },
    {
      "epoch": 0.9295352323838081,
      "grad_norm": 0.047780862073805754,
      "learning_rate": 0.0001294040325232304,
      "loss": 0.0462,
      "step": 620
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 0.04020955585417902,
      "learning_rate": 0.00012915370601695715,
      "loss": 0.0331,
      "step": 621
    },
    {
      "epoch": 0.9325337331334332,
      "grad_norm": 0.037639677657712055,
      "learning_rate": 0.00012890317969444716,
      "loss": 0.0248,
      "step": 622
    },
    {
      "epoch": 0.9340329835082459,
      "grad_norm": 0.05345077130247162,
      "learning_rate": 0.00012865245527277986,
      "loss": 0.0478,
      "step": 623
    },
    {
      "epoch": 0.9355322338830585,
      "grad_norm": 0.03711053033129508,
      "learning_rate": 0.00012840153447039228,
      "loss": 0.0199,
      "step": 624
    },
    {
      "epoch": 0.9370314842578711,
      "grad_norm": 0.04911139204212449,
      "learning_rate": 0.00012815041900706754,
      "loss": 0.0301,
      "step": 625
    },
    {
      "epoch": 0.9385307346326837,
      "grad_norm": 0.07752327109727443,
      "learning_rate": 0.00012789911060392294,
      "loss": 0.0692,
      "step": 626
    },
    {
      "epoch": 0.9400299850074962,
      "grad_norm": 0.13232474472513436,
      "learning_rate": 0.0001276476109833981,
      "loss": 0.0634,
      "step": 627
    },
    {
      "epoch": 0.9415292353823088,
      "grad_norm": 0.061926243742957965,
      "learning_rate": 0.00012739592186924328,
      "loss": 0.0397,
      "step": 628
    },
    {
      "epoch": 0.9430284857571214,
      "grad_norm": 0.04356242072040339,
      "learning_rate": 0.00012714404498650743,
      "loss": 0.0297,
      "step": 629
    },
    {
      "epoch": 0.9445277361319341,
      "grad_norm": 0.0376190098528388,
      "learning_rate": 0.00012689198206152657,
      "loss": 0.0271,
      "step": 630
    },
    {
      "epoch": 0.9460269865067467,
      "grad_norm": 0.05455511757321388,
      "learning_rate": 0.00012663973482191177,
      "loss": 0.0482,
      "step": 631
    },
    {
      "epoch": 0.9475262368815592,
      "grad_norm": 0.04442045716631641,
      "learning_rate": 0.0001263873049965373,
      "loss": 0.0291,
      "step": 632
    },
    {
      "epoch": 0.9490254872563718,
      "grad_norm": 0.054450506231905636,
      "learning_rate": 0.00012613469431552895,
      "loss": 0.0428,
      "step": 633
    },
    {
      "epoch": 0.9505247376311844,
      "grad_norm": 0.044327380794719165,
      "learning_rate": 0.00012588190451025207,
      "loss": 0.0286,
      "step": 634
    },
    {
      "epoch": 0.952023988005997,
      "grad_norm": 0.047204720536083225,
      "learning_rate": 0.00012562893731329967,
      "loss": 0.0406,
      "step": 635
    },
    {
      "epoch": 0.9535232383808095,
      "grad_norm": 0.19462933782113725,
      "learning_rate": 0.00012537579445848058,
      "loss": 0.1233,
      "step": 636
    },
    {
      "epoch": 0.9550224887556222,
      "grad_norm": 0.06994302712351169,
      "learning_rate": 0.00012512247768080756,
      "loss": 0.0403,
      "step": 637
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 0.06646995144905621,
      "learning_rate": 0.0001248689887164855,
      "loss": 0.0394,
      "step": 638
    },
    {
      "epoch": 0.9580209895052474,
      "grad_norm": 0.06551860726050548,
      "learning_rate": 0.00012461532930289933,
      "loss": 0.0613,
      "step": 639
    },
    {
      "epoch": 0.95952023988006,
      "grad_norm": 0.04577051878676127,
      "learning_rate": 0.00012436150117860225,
      "loss": 0.0267,
      "step": 640
    },
    {
      "epoch": 0.9610194902548725,
      "grad_norm": 0.0501593890528279,
      "learning_rate": 0.00012410750608330388,
      "loss": 0.0415,
      "step": 641
    },
    {
      "epoch": 0.9625187406296851,
      "grad_norm": 0.04213838081722136,
      "learning_rate": 0.0001238533457578581,
      "loss": 0.0363,
      "step": 642
    },
    {
      "epoch": 0.9640179910044977,
      "grad_norm": 0.04618274978280509,
      "learning_rate": 0.00012359902194425133,
      "loss": 0.0268,
      "step": 643
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 0.054945133222623746,
      "learning_rate": 0.00012334453638559057,
      "loss": 0.0411,
      "step": 644
    },
    {
      "epoch": 0.967016491754123,
      "grad_norm": 0.07629757994067972,
      "learning_rate": 0.00012308989082609122,
      "loss": 0.0363,
      "step": 645
    },
    {
      "epoch": 0.9685157421289355,
      "grad_norm": 0.040863705327469,
      "learning_rate": 0.00012283508701106557,
      "loss": 0.0302,
      "step": 646
    },
    {
      "epoch": 0.9700149925037481,
      "grad_norm": 0.07100180343789458,
      "learning_rate": 0.0001225801266869104,
      "loss": 0.044,
      "step": 647
    },
    {
      "epoch": 0.9715142428785607,
      "grad_norm": 0.06298768167998063,
      "learning_rate": 0.00012232501160109514,
      "loss": 0.0396,
      "step": 648
    },
    {
      "epoch": 0.9730134932533733,
      "grad_norm": 0.05510372575164882,
      "learning_rate": 0.00012206974350215015,
      "loss": 0.044,
      "step": 649
    },
    {
      "epoch": 0.974512743628186,
      "grad_norm": 0.0635198930062088,
      "learning_rate": 0.00012181432413965428,
      "loss": 0.0552,
      "step": 650
    },
    {
      "epoch": 0.9760119940029985,
      "grad_norm": 0.06043612042070538,
      "learning_rate": 0.00012155875526422332,
      "loss": 0.0617,
      "step": 651
    },
    {
      "epoch": 0.9775112443778111,
      "grad_norm": 0.040855526708167285,
      "learning_rate": 0.00012130303862749767,
      "loss": 0.0268,
      "step": 652
    },
    {
      "epoch": 0.9790104947526237,
      "grad_norm": 0.0496157050205137,
      "learning_rate": 0.00012104717598213056,
      "loss": 0.0225,
      "step": 653
    },
    {
      "epoch": 0.9805097451274363,
      "grad_norm": 0.04029704523356512,
      "learning_rate": 0.00012079116908177593,
      "loss": 0.0339,
      "step": 654
    },
    {
      "epoch": 0.9820089955022488,
      "grad_norm": 0.04963288277577189,
      "learning_rate": 0.00012053501968107641,
      "loss": 0.0346,
      "step": 655
    },
    {
      "epoch": 0.9835082458770614,
      "grad_norm": 0.06455026512135428,
      "learning_rate": 0.00012027872953565125,
      "loss": 0.0545,
      "step": 656
    },
    {
      "epoch": 0.9850074962518741,
      "grad_norm": 0.044681736834349894,
      "learning_rate": 0.00012002230040208447,
      "loss": 0.0377,
      "step": 657
    },
    {
      "epoch": 0.9865067466266867,
      "grad_norm": 0.06016026349629435,
      "learning_rate": 0.00011976573403791262,
      "loss": 0.0439,
      "step": 658
    },
    {
      "epoch": 0.9880059970014993,
      "grad_norm": 0.03916055819189204,
      "learning_rate": 0.00011950903220161285,
      "loss": 0.0319,
      "step": 659
    },
    {
      "epoch": 0.9895052473763118,
      "grad_norm": 0.0695326756638391,
      "learning_rate": 0.00011925219665259075,
      "loss": 0.0406,
      "step": 660
    },
    {
      "epoch": 0.9910044977511244,
      "grad_norm": 0.05484556659479051,
      "learning_rate": 0.00011899522915116848,
      "loss": 0.0413,
      "step": 661
    },
    {
      "epoch": 0.992503748125937,
      "grad_norm": 0.06761504502922643,
      "learning_rate": 0.00011873813145857249,
      "loss": 0.0502,
      "step": 662
    },
    {
      "epoch": 0.9940029985007496,
      "grad_norm": 0.03769561443583786,
      "learning_rate": 0.00011848090533692155,
      "loss": 0.0351,
      "step": 663
    },
    {
      "epoch": 0.9955022488755623,
      "grad_norm": 0.05884790570998015,
      "learning_rate": 0.00011822355254921478,
      "loss": 0.0499,
      "step": 664
    },
    {
      "epoch": 0.9970014992503748,
      "grad_norm": 0.05692462358820348,
      "learning_rate": 0.00011796607485931928,
      "loss": 0.0367,
      "step": 665
    },
    {
      "epoch": 0.9985007496251874,
      "grad_norm": 0.04183931910506595,
      "learning_rate": 0.00011770847403195834,
      "loss": 0.0362,
      "step": 666
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.05719242886094664,
      "learning_rate": 0.0001174507518326992,
      "loss": 0.0373,
      "step": 667
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.0589199960231781,
      "eval_runtime": 760.1358,
      "eval_samples_per_second": 13.65,
      "eval_steps_per_second": 0.854,
      "step": 667
    },
    {
      "epoch": 1.0014992503748126,
      "grad_norm": 0.03762475183995106,
      "learning_rate": 0.00011719291002794096,
      "loss": 0.0313,
      "step": 668
    },
    {
      "epoch": 1.0029985007496252,
      "grad_norm": 0.027248795473336453,
      "learning_rate": 0.00011693495038490245,
      "loss": 0.0202,
      "step": 669
    },
    {
      "epoch": 1.0044977511244377,
      "grad_norm": 0.041298897947596655,
      "learning_rate": 0.00011667687467161024,
      "loss": 0.0318,
      "step": 670
    },
    {
      "epoch": 1.0059970014992503,
      "grad_norm": 0.041340873666624624,
      "learning_rate": 0.0001164186846568863,
      "loss": 0.0192,
      "step": 671
    },
    {
      "epoch": 1.0074962518740629,
      "grad_norm": 0.03416279632561669,
      "learning_rate": 0.00011616038211033613,
      "loss": 0.0194,
      "step": 672
    },
    {
      "epoch": 1.0089955022488755,
      "grad_norm": 0.04269382987865334,
      "learning_rate": 0.00011590196880233644,
      "loss": 0.0306,
      "step": 673
    },
    {
      "epoch": 1.0104947526236883,
      "grad_norm": 0.04456460780740358,
      "learning_rate": 0.0001156434465040231,
      "loss": 0.0358,
      "step": 674
    },
    {
      "epoch": 1.0119940029985008,
      "grad_norm": 0.037486633011275576,
      "learning_rate": 0.00011538481698727904,
      "loss": 0.03,
      "step": 675
    },
    {
      "epoch": 1.0134932533733134,
      "grad_norm": 0.07944497636169431,
      "learning_rate": 0.00011512608202472194,
      "loss": 0.047,
      "step": 676
    },
    {
      "epoch": 1.014992503748126,
      "grad_norm": 0.07167365891576202,
      "learning_rate": 0.00011486724338969232,
      "loss": 0.0281,
      "step": 677
    },
    {
      "epoch": 1.0164917541229386,
      "grad_norm": 0.030600334090124463,
      "learning_rate": 0.00011460830285624118,
      "loss": 0.0185,
      "step": 678
    },
    {
      "epoch": 1.0179910044977512,
      "grad_norm": 0.0334860041071292,
      "learning_rate": 0.00011434926219911793,
      "loss": 0.0205,
      "step": 679
    },
    {
      "epoch": 1.0194902548725637,
      "grad_norm": 0.07452867174589614,
      "learning_rate": 0.00011409012319375827,
      "loss": 0.024,
      "step": 680
    },
    {
      "epoch": 1.0209895052473763,
      "grad_norm": 0.05352241311204091,
      "learning_rate": 0.00011383088761627192,
      "loss": 0.0153,
      "step": 681
    },
    {
      "epoch": 1.0224887556221889,
      "grad_norm": 0.05264974398072599,
      "learning_rate": 0.00011357155724343045,
      "loss": 0.0297,
      "step": 682
    },
    {
      "epoch": 1.0239880059970015,
      "grad_norm": 0.043453829569375796,
      "learning_rate": 0.00011331213385265524,
      "loss": 0.0285,
      "step": 683
    },
    {
      "epoch": 1.025487256371814,
      "grad_norm": 0.04694358278440352,
      "learning_rate": 0.00011305261922200519,
      "loss": 0.0308,
      "step": 684
    },
    {
      "epoch": 1.0269865067466266,
      "grad_norm": 0.052332835831569786,
      "learning_rate": 0.00011279301513016444,
      "loss": 0.0276,
      "step": 685
    },
    {
      "epoch": 1.0284857571214392,
      "grad_norm": 0.05731903299379123,
      "learning_rate": 0.00011253332335643043,
      "loss": 0.0368,
      "step": 686
    },
    {
      "epoch": 1.0299850074962518,
      "grad_norm": 0.03577714273632661,
      "learning_rate": 0.0001122735456807015,
      "loss": 0.0159,
      "step": 687
    },
    {
      "epoch": 1.0314842578710646,
      "grad_norm": 0.030987774585815354,
      "learning_rate": 0.00011201368388346471,
      "loss": 0.014,
      "step": 688
    },
    {
      "epoch": 1.0329835082458771,
      "grad_norm": 0.04126960363600113,
      "learning_rate": 0.00011175373974578378,
      "loss": 0.0224,
      "step": 689
    },
    {
      "epoch": 1.0344827586206897,
      "grad_norm": 0.03027314685021926,
      "learning_rate": 0.00011149371504928668,
      "loss": 0.0187,
      "step": 690
    },
    {
      "epoch": 1.0359820089955023,
      "grad_norm": 0.059016723534456986,
      "learning_rate": 0.00011123361157615355,
      "loss": 0.0336,
      "step": 691
    },
    {
      "epoch": 1.0374812593703149,
      "grad_norm": 0.03460241895483622,
      "learning_rate": 0.00011097343110910452,
      "loss": 0.0172,
      "step": 692
    },
    {
      "epoch": 1.0389805097451275,
      "grad_norm": 0.04905984564105954,
      "learning_rate": 0.00011071317543138738,
      "loss": 0.0202,
      "step": 693
    },
    {
      "epoch": 1.04047976011994,
      "grad_norm": 0.0477670312888383,
      "learning_rate": 0.00011045284632676536,
      "loss": 0.0257,
      "step": 694
    },
    {
      "epoch": 1.0419790104947526,
      "grad_norm": 0.0649026742045147,
      "learning_rate": 0.000110192445579505,
      "loss": 0.0528,
      "step": 695
    },
    {
      "epoch": 1.0434782608695652,
      "grad_norm": 0.03838342057763881,
      "learning_rate": 0.00010993197497436391,
      "loss": 0.0176,
      "step": 696
    },
    {
      "epoch": 1.0449775112443778,
      "grad_norm": 0.05081060598139383,
      "learning_rate": 0.00010967143629657842,
      "loss": 0.024,
      "step": 697
    },
    {
      "epoch": 1.0464767616191903,
      "grad_norm": 0.07793509916227749,
      "learning_rate": 0.00010941083133185146,
      "loss": 0.0475,
      "step": 698
    },
    {
      "epoch": 1.047976011994003,
      "grad_norm": 0.045160852102844276,
      "learning_rate": 0.00010915016186634026,
      "loss": 0.0169,
      "step": 699
    },
    {
      "epoch": 1.0494752623688155,
      "grad_norm": 0.08881360543272801,
      "learning_rate": 0.00010888942968664417,
      "loss": 0.023,
      "step": 700
    },
    {
      "epoch": 1.050974512743628,
      "grad_norm": 0.038412054850676614,
      "learning_rate": 0.00010862863657979237,
      "loss": 0.018,
      "step": 701
    },
    {
      "epoch": 1.0524737631184409,
      "grad_norm": 0.051809463503864034,
      "learning_rate": 0.00010836778433323158,
      "loss": 0.0266,
      "step": 702
    },
    {
      "epoch": 1.0539730134932535,
      "grad_norm": 0.059635918308077476,
      "learning_rate": 0.00010810687473481389,
      "loss": 0.0293,
      "step": 703
    },
    {
      "epoch": 1.055472263868066,
      "grad_norm": 0.05223829787695676,
      "learning_rate": 0.0001078459095727845,
      "loss": 0.0333,
      "step": 704
    },
    {
      "epoch": 1.0569715142428786,
      "grad_norm": 0.04788415350708223,
      "learning_rate": 0.00010758489063576939,
      "loss": 0.0225,
      "step": 705
    },
    {
      "epoch": 1.0584707646176912,
      "grad_norm": 0.07606032889772184,
      "learning_rate": 0.00010732381971276318,
      "loss": 0.0504,
      "step": 706
    },
    {
      "epoch": 1.0599700149925038,
      "grad_norm": 0.041830044788424676,
      "learning_rate": 0.00010706269859311669,
      "loss": 0.0339,
      "step": 707
    },
    {
      "epoch": 1.0614692653673163,
      "grad_norm": 0.056499034345126806,
      "learning_rate": 0.00010680152906652483,
      "loss": 0.0797,
      "step": 708
    },
    {
      "epoch": 1.062968515742129,
      "grad_norm": 0.050667538240860205,
      "learning_rate": 0.00010654031292301432,
      "loss": 0.0385,
      "step": 709
    },
    {
      "epoch": 1.0644677661169415,
      "grad_norm": 0.03606037951763869,
      "learning_rate": 0.00010627905195293135,
      "loss": 0.0249,
      "step": 710
    },
    {
      "epoch": 1.065967016491754,
      "grad_norm": 0.06482813154069911,
      "learning_rate": 0.00010601774794692935,
      "loss": 0.0322,
      "step": 711
    },
    {
      "epoch": 1.0674662668665666,
      "grad_norm": 0.031472797559762344,
      "learning_rate": 0.00010575640269595675,
      "loss": 0.0163,
      "step": 712
    },
    {
      "epoch": 1.0689655172413792,
      "grad_norm": 0.04865531174873738,
      "learning_rate": 0.0001054950179912446,
      "loss": 0.0265,
      "step": 713
    },
    {
      "epoch": 1.0704647676161918,
      "grad_norm": 0.035242178613106155,
      "learning_rate": 0.0001052335956242944,
      "loss": 0.017,
      "step": 714
    },
    {
      "epoch": 1.0719640179910046,
      "grad_norm": 0.10969495785314283,
      "learning_rate": 0.0001049721373868658,
      "loss": 0.051,
      "step": 715
    },
    {
      "epoch": 1.0734632683658172,
      "grad_norm": 0.055243920108870075,
      "learning_rate": 0.00010471064507096426,
      "loss": 0.0299,
      "step": 716
    },
    {
      "epoch": 1.0749625187406298,
      "grad_norm": 0.027766012133857647,
      "learning_rate": 0.00010444912046882888,
      "loss": 0.0121,
      "step": 717
    },
    {
      "epoch": 1.0764617691154423,
      "grad_norm": 0.05149363616690094,
      "learning_rate": 0.00010418756537291996,
      "loss": 0.0237,
      "step": 718
    },
    {
      "epoch": 1.077961019490255,
      "grad_norm": 0.0435854981748732,
      "learning_rate": 0.00010392598157590688,
      "loss": 0.0216,
      "step": 719
    },
    {
      "epoch": 1.0794602698650675,
      "grad_norm": 0.048516448094552064,
      "learning_rate": 0.00010366437087065564,
      "loss": 0.03,
      "step": 720
    },
    {
      "epoch": 1.08095952023988,
      "grad_norm": 0.05499780130973038,
      "learning_rate": 0.00010340273505021674,
      "loss": 0.0263,
      "step": 721
    },
    {
      "epoch": 1.0824587706146926,
      "grad_norm": 0.06511989459756257,
      "learning_rate": 0.00010314107590781284,
      "loss": 0.0374,
      "step": 722
    },
    {
      "epoch": 1.0839580209895052,
      "grad_norm": 0.05279506351690483,
      "learning_rate": 0.00010287939523682634,
      "loss": 0.026,
      "step": 723
    },
    {
      "epoch": 1.0854572713643178,
      "grad_norm": 0.049766335190166044,
      "learning_rate": 0.00010261769483078733,
      "loss": 0.0362,
      "step": 724
    },
    {
      "epoch": 1.0869565217391304,
      "grad_norm": 0.0616925740357659,
      "learning_rate": 0.00010235597648336104,
      "loss": 0.0272,
      "step": 725
    },
    {
      "epoch": 1.088455772113943,
      "grad_norm": 0.035580584914971236,
      "learning_rate": 0.0001020942419883357,
      "loss": 0.0246,
      "step": 726
    },
    {
      "epoch": 1.0899550224887555,
      "grad_norm": 0.05651314515303107,
      "learning_rate": 0.0001018324931396103,
      "loss": 0.0389,
      "step": 727
    },
    {
      "epoch": 1.0914542728635683,
      "grad_norm": 0.03403275870606777,
      "learning_rate": 0.00010157073173118208,
      "loss": 0.017,
      "step": 728
    },
    {
      "epoch": 1.092953523238381,
      "grad_norm": 0.04001508080874846,
      "learning_rate": 0.00010130895955713445,
      "loss": 0.0344,
      "step": 729
    },
    {
      "epoch": 1.0944527736131935,
      "grad_norm": 0.038898635048905054,
      "learning_rate": 0.00010104717841162458,
      "loss": 0.0139,
      "step": 730
    },
    {
      "epoch": 1.095952023988006,
      "grad_norm": 0.04852969796297334,
      "learning_rate": 0.00010078539008887114,
      "loss": 0.0303,
      "step": 731
    },
    {
      "epoch": 1.0974512743628186,
      "grad_norm": 0.030415262260471615,
      "learning_rate": 0.00010052359638314195,
      "loss": 0.0209,
      "step": 732
    },
    {
      "epoch": 1.0989505247376312,
      "grad_norm": 0.03296035338712638,
      "learning_rate": 0.00010026179908874183,
      "loss": 0.0176,
      "step": 733
    },
    {
      "epoch": 1.1004497751124438,
      "grad_norm": 0.06912112785698081,
      "learning_rate": 0.0001,
      "loss": 0.0911,
      "step": 734
    },
    {
      "epoch": 1.1019490254872564,
      "grad_norm": 0.04126417647100554,
      "learning_rate": 9.973820091125821e-05,
      "loss": 0.0278,
      "step": 735
    },
    {
      "epoch": 1.103448275862069,
      "grad_norm": 0.06337110286285316,
      "learning_rate": 9.947640361685804e-05,
      "loss": 0.0423,
      "step": 736
    },
    {
      "epoch": 1.1049475262368815,
      "grad_norm": 0.04785214604294027,
      "learning_rate": 9.921460991112891e-05,
      "loss": 0.019,
      "step": 737
    },
    {
      "epoch": 1.106446776611694,
      "grad_norm": 0.03813515686538096,
      "learning_rate": 9.895282158837545e-05,
      "loss": 0.0147,
      "step": 738
    },
    {
      "epoch": 1.1079460269865067,
      "grad_norm": 0.04010157831438066,
      "learning_rate": 9.869104044286558e-05,
      "loss": 0.0276,
      "step": 739
    },
    {
      "epoch": 1.1094452773613193,
      "grad_norm": 0.03406965983248011,
      "learning_rate": 9.842926826881796e-05,
      "loss": 0.0149,
      "step": 740
    },
    {
      "epoch": 1.1109445277361318,
      "grad_norm": 0.05122540759813054,
      "learning_rate": 9.816750686038973e-05,
      "loss": 0.0218,
      "step": 741
    },
    {
      "epoch": 1.1124437781109444,
      "grad_norm": 0.046947765546721,
      "learning_rate": 9.790575801166432e-05,
      "loss": 0.025,
      "step": 742
    },
    {
      "epoch": 1.1139430284857572,
      "grad_norm": 0.07498886528603434,
      "learning_rate": 9.764402351663901e-05,
      "loss": 0.0434,
      "step": 743
    },
    {
      "epoch": 1.1154422788605698,
      "grad_norm": 0.03187418225048272,
      "learning_rate": 9.73823051692127e-05,
      "loss": 0.0164,
      "step": 744
    },
    {
      "epoch": 1.1169415292353824,
      "grad_norm": 0.05124246170818971,
      "learning_rate": 9.712060476317367e-05,
      "loss": 0.027,
      "step": 745
    },
    {
      "epoch": 1.118440779610195,
      "grad_norm": 0.04064336939991715,
      "learning_rate": 9.685892409218717e-05,
      "loss": 0.0312,
      "step": 746
    },
    {
      "epoch": 1.1199400299850075,
      "grad_norm": 0.02940931109265308,
      "learning_rate": 9.659726494978325e-05,
      "loss": 0.0138,
      "step": 747
    },
    {
      "epoch": 1.12143928035982,
      "grad_norm": 0.07160694268820639,
      "learning_rate": 9.633562912934436e-05,
      "loss": 0.0638,
      "step": 748
    },
    {
      "epoch": 1.1229385307346327,
      "grad_norm": 0.03268947669426935,
      "learning_rate": 9.607401842409317e-05,
      "loss": 0.0152,
      "step": 749
    },
    {
      "epoch": 1.1244377811094453,
      "grad_norm": 0.04469947279909564,
      "learning_rate": 9.581243462708006e-05,
      "loss": 0.0519,
      "step": 750
    },
    {
      "epoch": 1.1259370314842578,
      "grad_norm": 0.03733568199384118,
      "learning_rate": 9.555087953117114e-05,
      "loss": 0.0191,
      "step": 751
    },
    {
      "epoch": 1.1274362818590704,
      "grad_norm": 0.04957459904827682,
      "learning_rate": 9.528935492903575e-05,
      "loss": 0.0258,
      "step": 752
    },
    {
      "epoch": 1.128935532233883,
      "grad_norm": 0.04291232620005789,
      "learning_rate": 9.502786261313423e-05,
      "loss": 0.0186,
      "step": 753
    },
    {
      "epoch": 1.1304347826086956,
      "grad_norm": 0.03645205093124093,
      "learning_rate": 9.476640437570562e-05,
      "loss": 0.0179,
      "step": 754
    },
    {
      "epoch": 1.1319340329835081,
      "grad_norm": 0.036883531854657765,
      "learning_rate": 9.450498200875546e-05,
      "loss": 0.0181,
      "step": 755
    },
    {
      "epoch": 1.133433283358321,
      "grad_norm": 0.03611209493776261,
      "learning_rate": 9.424359730404329e-05,
      "loss": 0.0149,
      "step": 756
    },
    {
      "epoch": 1.1349325337331335,
      "grad_norm": 0.03543885857842899,
      "learning_rate": 9.398225205307066e-05,
      "loss": 0.0179,
      "step": 757
    },
    {
      "epoch": 1.136431784107946,
      "grad_norm": 0.04885756178865602,
      "learning_rate": 9.372094804706867e-05,
      "loss": 0.0363,
      "step": 758
    },
    {
      "epoch": 1.1379310344827587,
      "grad_norm": 0.045201667917713226,
      "learning_rate": 9.345968707698569e-05,
      "loss": 0.015,
      "step": 759
    },
    {
      "epoch": 1.1394302848575713,
      "grad_norm": 0.045725248124270736,
      "learning_rate": 9.319847093347522e-05,
      "loss": 0.0242,
      "step": 760
    },
    {
      "epoch": 1.1409295352323838,
      "grad_norm": 0.03246760399537495,
      "learning_rate": 9.293730140688336e-05,
      "loss": 0.0136,
      "step": 761
    },
    {
      "epoch": 1.1424287856071964,
      "grad_norm": 0.04842555683642214,
      "learning_rate": 9.267618028723686e-05,
      "loss": 0.0178,
      "step": 762
    },
    {
      "epoch": 1.143928035982009,
      "grad_norm": 0.09178292202784064,
      "learning_rate": 9.241510936423062e-05,
      "loss": 0.0293,
      "step": 763
    },
    {
      "epoch": 1.1454272863568216,
      "grad_norm": 0.03732797810025069,
      "learning_rate": 9.215409042721552e-05,
      "loss": 0.032,
      "step": 764
    },
    {
      "epoch": 1.1469265367316341,
      "grad_norm": 0.06084284216603883,
      "learning_rate": 9.189312526518613e-05,
      "loss": 0.0298,
      "step": 765
    },
    {
      "epoch": 1.1484257871064467,
      "grad_norm": 0.05070843744675052,
      "learning_rate": 9.163221566676847e-05,
      "loss": 0.0115,
      "step": 766
    },
    {
      "epoch": 1.1499250374812593,
      "grad_norm": 0.040941353399746576,
      "learning_rate": 9.137136342020768e-05,
      "loss": 0.0219,
      "step": 767
    },
    {
      "epoch": 1.1514242878560719,
      "grad_norm": 0.056138839618842994,
      "learning_rate": 9.111057031335585e-05,
      "loss": 0.0209,
      "step": 768
    },
    {
      "epoch": 1.1529235382308847,
      "grad_norm": 0.053017337582656095,
      "learning_rate": 9.084983813365978e-05,
      "loss": 0.0139,
      "step": 769
    },
    {
      "epoch": 1.1544227886056972,
      "grad_norm": 0.0437160406734324,
      "learning_rate": 9.058916866814858e-05,
      "loss": 0.0292,
      "step": 770
    },
    {
      "epoch": 1.1559220389805098,
      "grad_norm": 0.06517357369678477,
      "learning_rate": 9.032856370342158e-05,
      "loss": 0.0314,
      "step": 771
    },
    {
      "epoch": 1.1574212893553224,
      "grad_norm": 0.05616399006235695,
      "learning_rate": 9.006802502563612e-05,
      "loss": 0.0251,
      "step": 772
    },
    {
      "epoch": 1.158920539730135,
      "grad_norm": 0.046115228142336546,
      "learning_rate": 8.980755442049502e-05,
      "loss": 0.0198,
      "step": 773
    },
    {
      "epoch": 1.1604197901049476,
      "grad_norm": 0.03835516395560399,
      "learning_rate": 8.954715367323468e-05,
      "loss": 0.0264,
      "step": 774
    },
    {
      "epoch": 1.1619190404797601,
      "grad_norm": 0.05643079022619911,
      "learning_rate": 8.928682456861265e-05,
      "loss": 0.0255,
      "step": 775
    },
    {
      "epoch": 1.1634182908545727,
      "grad_norm": 0.03736421317747863,
      "learning_rate": 8.902656889089548e-05,
      "loss": 0.0185,
      "step": 776
    },
    {
      "epoch": 1.1649175412293853,
      "grad_norm": 0.051100575541428585,
      "learning_rate": 8.876638842384644e-05,
      "loss": 0.0201,
      "step": 777
    },
    {
      "epoch": 1.1664167916041979,
      "grad_norm": 0.03430703468298361,
      "learning_rate": 8.850628495071336e-05,
      "loss": 0.0152,
      "step": 778
    },
    {
      "epoch": 1.1679160419790104,
      "grad_norm": 0.030283377580051823,
      "learning_rate": 8.824626025421626e-05,
      "loss": 0.0189,
      "step": 779
    },
    {
      "epoch": 1.169415292353823,
      "grad_norm": 0.04693707293564595,
      "learning_rate": 8.79863161165353e-05,
      "loss": 0.0265,
      "step": 780
    },
    {
      "epoch": 1.1709145427286356,
      "grad_norm": 0.0670635890351688,
      "learning_rate": 8.772645431929851e-05,
      "loss": 0.043,
      "step": 781
    },
    {
      "epoch": 1.1724137931034484,
      "grad_norm": 0.043848670372825894,
      "learning_rate": 8.746667664356956e-05,
      "loss": 0.0217,
      "step": 782
    },
    {
      "epoch": 1.1739130434782608,
      "grad_norm": 0.05412472149445996,
      "learning_rate": 8.720698486983557e-05,
      "loss": 0.0197,
      "step": 783
    },
    {
      "epoch": 1.1754122938530736,
      "grad_norm": 0.019649617759211542,
      "learning_rate": 8.694738077799488e-05,
      "loss": 0.0073,
      "step": 784
    },
    {
      "epoch": 1.1769115442278861,
      "grad_norm": 0.05225640536933059,
      "learning_rate": 8.668786614734478e-05,
      "loss": 0.0276,
      "step": 785
    },
    {
      "epoch": 1.1784107946026987,
      "grad_norm": 0.04315118247949963,
      "learning_rate": 8.642844275656957e-05,
      "loss": 0.0324,
      "step": 786
    },
    {
      "epoch": 1.1799100449775113,
      "grad_norm": 0.029412653837457406,
      "learning_rate": 8.616911238372812e-05,
      "loss": 0.014,
      "step": 787
    },
    {
      "epoch": 1.1814092953523239,
      "grad_norm": 0.04133482214444302,
      "learning_rate": 8.590987680624174e-05,
      "loss": 0.0202,
      "step": 788
    },
    {
      "epoch": 1.1829085457271364,
      "grad_norm": 0.06765748401829244,
      "learning_rate": 8.565073780088208e-05,
      "loss": 0.0441,
      "step": 789
    },
    {
      "epoch": 1.184407796101949,
      "grad_norm": 0.05271546437062786,
      "learning_rate": 8.539169714375885e-05,
      "loss": 0.0256,
      "step": 790
    },
    {
      "epoch": 1.1859070464767616,
      "grad_norm": 0.040199086711978846,
      "learning_rate": 8.51327566103077e-05,
      "loss": 0.0139,
      "step": 791
    },
    {
      "epoch": 1.1874062968515742,
      "grad_norm": 0.04400948652648687,
      "learning_rate": 8.487391797527808e-05,
      "loss": 0.0298,
      "step": 792
    },
    {
      "epoch": 1.1889055472263867,
      "grad_norm": 0.04023525299067876,
      "learning_rate": 8.461518301272099e-05,
      "loss": 0.025,
      "step": 793
    },
    {
      "epoch": 1.1904047976011993,
      "grad_norm": 0.03582585876670016,
      "learning_rate": 8.435655349597689e-05,
      "loss": 0.0206,
      "step": 794
    },
    {
      "epoch": 1.191904047976012,
      "grad_norm": 0.04826655494275668,
      "learning_rate": 8.409803119766361e-05,
      "loss": 0.0242,
      "step": 795
    },
    {
      "epoch": 1.1934032983508245,
      "grad_norm": 0.05078096150994414,
      "learning_rate": 8.383961788966391e-05,
      "loss": 0.0212,
      "step": 796
    },
    {
      "epoch": 1.1949025487256373,
      "grad_norm": 0.04974688478126328,
      "learning_rate": 8.358131534311372e-05,
      "loss": 0.0262,
      "step": 797
    },
    {
      "epoch": 1.1964017991004499,
      "grad_norm": 0.03592037995042853,
      "learning_rate": 8.332312532838978e-05,
      "loss": 0.0159,
      "step": 798
    },
    {
      "epoch": 1.1979010494752624,
      "grad_norm": 0.0491861501338674,
      "learning_rate": 8.306504961509754e-05,
      "loss": 0.0234,
      "step": 799
    },
    {
      "epoch": 1.199400299850075,
      "grad_norm": 0.027759792015070672,
      "learning_rate": 8.280708997205904e-05,
      "loss": 0.0136,
      "step": 800
    },
    {
      "epoch": 1.2008995502248876,
      "grad_norm": 0.0548378934219744,
      "learning_rate": 8.254924816730082e-05,
      "loss": 0.0251,
      "step": 801
    },
    {
      "epoch": 1.2023988005997002,
      "grad_norm": 0.031042646875865575,
      "learning_rate": 8.229152596804168e-05,
      "loss": 0.0114,
      "step": 802
    },
    {
      "epoch": 1.2038980509745127,
      "grad_norm": 0.04185558020402479,
      "learning_rate": 8.203392514068074e-05,
      "loss": 0.0196,
      "step": 803
    },
    {
      "epoch": 1.2053973013493253,
      "grad_norm": 0.03573774524393669,
      "learning_rate": 8.177644745078526e-05,
      "loss": 0.029,
      "step": 804
    },
    {
      "epoch": 1.206896551724138,
      "grad_norm": 0.04500572749598161,
      "learning_rate": 8.151909466307845e-05,
      "loss": 0.0283,
      "step": 805
    },
    {
      "epoch": 1.2083958020989505,
      "grad_norm": 0.055472853240349465,
      "learning_rate": 8.126186854142752e-05,
      "loss": 0.0278,
      "step": 806
    },
    {
      "epoch": 1.209895052473763,
      "grad_norm": 0.04737096525754299,
      "learning_rate": 8.100477084883156e-05,
      "loss": 0.0212,
      "step": 807
    },
    {
      "epoch": 1.2113943028485756,
      "grad_norm": 0.06377978614115022,
      "learning_rate": 8.074780334740928e-05,
      "loss": 0.0264,
      "step": 808
    },
    {
      "epoch": 1.2128935532233882,
      "grad_norm": 0.1037734762277636,
      "learning_rate": 8.049096779838719e-05,
      "loss": 0.0653,
      "step": 809
    },
    {
      "epoch": 1.214392803598201,
      "grad_norm": 0.04938456491428775,
      "learning_rate": 8.023426596208739e-05,
      "loss": 0.0359,
      "step": 810
    },
    {
      "epoch": 1.2158920539730136,
      "grad_norm": 0.03871245605295673,
      "learning_rate": 7.997769959791554e-05,
      "loss": 0.0189,
      "step": 811
    },
    {
      "epoch": 1.2173913043478262,
      "grad_norm": 0.04836114902803376,
      "learning_rate": 7.972127046434878e-05,
      "loss": 0.0216,
      "step": 812
    },
    {
      "epoch": 1.2188905547226387,
      "grad_norm": 0.06566867333027651,
      "learning_rate": 7.946498031892364e-05,
      "loss": 0.0289,
      "step": 813
    },
    {
      "epoch": 1.2203898050974513,
      "grad_norm": 0.07837421636117799,
      "learning_rate": 7.920883091822408e-05,
      "loss": 0.0312,
      "step": 814
    },
    {
      "epoch": 1.221889055472264,
      "grad_norm": 0.05650900430884517,
      "learning_rate": 7.895282401786945e-05,
      "loss": 0.025,
      "step": 815
    },
    {
      "epoch": 1.2233883058470765,
      "grad_norm": 0.04878788564349848,
      "learning_rate": 7.869696137250235e-05,
      "loss": 0.0268,
      "step": 816
    },
    {
      "epoch": 1.224887556221889,
      "grad_norm": 0.048966519858149045,
      "learning_rate": 7.844124473577672e-05,
      "loss": 0.0315,
      "step": 817
    },
    {
      "epoch": 1.2263868065967016,
      "grad_norm": 0.03824407231228468,
      "learning_rate": 7.818567586034577e-05,
      "loss": 0.0198,
      "step": 818
    },
    {
      "epoch": 1.2278860569715142,
      "grad_norm": 0.0529744116915096,
      "learning_rate": 7.79302564978499e-05,
      "loss": 0.0381,
      "step": 819
    },
    {
      "epoch": 1.2293853073463268,
      "grad_norm": 0.05426023606207575,
      "learning_rate": 7.767498839890488e-05,
      "loss": 0.0295,
      "step": 820
    },
    {
      "epoch": 1.2308845577211394,
      "grad_norm": 0.03534659066797283,
      "learning_rate": 7.741987331308964e-05,
      "loss": 0.0197,
      "step": 821
    },
    {
      "epoch": 1.232383808095952,
      "grad_norm": 0.15173262879838179,
      "learning_rate": 7.716491298893442e-05,
      "loss": 0.0299,
      "step": 822
    },
    {
      "epoch": 1.2338830584707647,
      "grad_norm": 0.04753989562431895,
      "learning_rate": 7.691010917390874e-05,
      "loss": 0.021,
      "step": 823
    },
    {
      "epoch": 1.235382308845577,
      "grad_norm": 0.045421468563024,
      "learning_rate": 7.66554636144095e-05,
      "loss": 0.0315,
      "step": 824
    },
    {
      "epoch": 1.23688155922039,
      "grad_norm": 0.047073846935776716,
      "learning_rate": 7.64009780557487e-05,
      "loss": 0.0214,
      "step": 825
    },
    {
      "epoch": 1.2383808095952025,
      "grad_norm": 0.05224603067353402,
      "learning_rate": 7.614665424214193e-05,
      "loss": 0.0217,
      "step": 826
    },
    {
      "epoch": 1.239880059970015,
      "grad_norm": 0.04520300461414657,
      "learning_rate": 7.589249391669616e-05,
      "loss": 0.0213,
      "step": 827
    },
    {
      "epoch": 1.2413793103448276,
      "grad_norm": 0.04736894258409455,
      "learning_rate": 7.563849882139776e-05,
      "loss": 0.0237,
      "step": 828
    },
    {
      "epoch": 1.2428785607196402,
      "grad_norm": 0.053766180923070186,
      "learning_rate": 7.53846706971007e-05,
      "loss": 0.0323,
      "step": 829
    },
    {
      "epoch": 1.2443778110944528,
      "grad_norm": 0.05528412145749755,
      "learning_rate": 7.513101128351454e-05,
      "loss": 0.0487,
      "step": 830
    },
    {
      "epoch": 1.2458770614692654,
      "grad_norm": 0.05566014947606353,
      "learning_rate": 7.487752231919245e-05,
      "loss": 0.0267,
      "step": 831
    },
    {
      "epoch": 1.247376311844078,
      "grad_norm": 0.04083106652022665,
      "learning_rate": 7.462420554151944e-05,
      "loss": 0.0278,
      "step": 832
    },
    {
      "epoch": 1.2488755622188905,
      "grad_norm": 0.06891562106271396,
      "learning_rate": 7.437106268670034e-05,
      "loss": 0.0294,
      "step": 833
    },
    {
      "epoch": 1.250374812593703,
      "grad_norm": 0.07342638900103611,
      "learning_rate": 7.411809548974792e-05,
      "loss": 0.0845,
      "step": 834
    },
    {
      "epoch": 1.2518740629685157,
      "grad_norm": 0.050466963864324595,
      "learning_rate": 7.386530568447107e-05,
      "loss": 0.0298,
      "step": 835
    },
    {
      "epoch": 1.2533733133433285,
      "grad_norm": 0.06686910207648407,
      "learning_rate": 7.361269500346274e-05,
      "loss": 0.0275,
      "step": 836
    },
    {
      "epoch": 1.2548725637181408,
      "grad_norm": 0.046075881059396175,
      "learning_rate": 7.336026517808827e-05,
      "loss": 0.0214,
      "step": 837
    },
    {
      "epoch": 1.2563718140929536,
      "grad_norm": 0.049483313231029974,
      "learning_rate": 7.310801793847344e-05,
      "loss": 0.0295,
      "step": 838
    },
    {
      "epoch": 1.2578710644677662,
      "grad_norm": 0.06374973164823877,
      "learning_rate": 7.285595501349258e-05,
      "loss": 0.0544,
      "step": 839
    },
    {
      "epoch": 1.2593703148425788,
      "grad_norm": 0.2122029055345762,
      "learning_rate": 7.260407813075676e-05,
      "loss": 0.099,
      "step": 840
    },
    {
      "epoch": 1.2608695652173914,
      "grad_norm": 0.02852293268003914,
      "learning_rate": 7.235238901660195e-05,
      "loss": 0.0166,
      "step": 841
    },
    {
      "epoch": 1.262368815592204,
      "grad_norm": 0.04115201623040188,
      "learning_rate": 7.210088939607708e-05,
      "loss": 0.0254,
      "step": 842
    },
    {
      "epoch": 1.2638680659670165,
      "grad_norm": 0.05628533766124335,
      "learning_rate": 7.184958099293246e-05,
      "loss": 0.0365,
      "step": 843
    },
    {
      "epoch": 1.265367316341829,
      "grad_norm": 0.08094057068973667,
      "learning_rate": 7.159846552960774e-05,
      "loss": 0.0222,
      "step": 844
    },
    {
      "epoch": 1.2668665667166417,
      "grad_norm": 0.03799623111533529,
      "learning_rate": 7.134754472722017e-05,
      "loss": 0.0224,
      "step": 845
    },
    {
      "epoch": 1.2683658170914542,
      "grad_norm": 0.04700845879809844,
      "learning_rate": 7.109682030555283e-05,
      "loss": 0.0224,
      "step": 846
    },
    {
      "epoch": 1.2698650674662668,
      "grad_norm": 0.048949756388667436,
      "learning_rate": 7.08462939830429e-05,
      "loss": 0.018,
      "step": 847
    },
    {
      "epoch": 1.2713643178410794,
      "grad_norm": 0.047029812926556,
      "learning_rate": 7.059596747676962e-05,
      "loss": 0.0266,
      "step": 848
    },
    {
      "epoch": 1.272863568215892,
      "grad_norm": 0.030351552812581083,
      "learning_rate": 7.034584250244291e-05,
      "loss": 0.0151,
      "step": 849
    },
    {
      "epoch": 1.2743628185907045,
      "grad_norm": 0.052275221441315155,
      "learning_rate": 7.009592077439134e-05,
      "loss": 0.0198,
      "step": 850
    },
    {
      "epoch": 1.2758620689655173,
      "grad_norm": 0.04376783013783442,
      "learning_rate": 6.984620400555044e-05,
      "loss": 0.0167,
      "step": 851
    },
    {
      "epoch": 1.2773613193403297,
      "grad_norm": 0.054785397673240276,
      "learning_rate": 6.959669390745097e-05,
      "loss": 0.0209,
      "step": 852
    },
    {
      "epoch": 1.2788605697151425,
      "grad_norm": 0.04062686089747498,
      "learning_rate": 6.934739219020726e-05,
      "loss": 0.015,
      "step": 853
    },
    {
      "epoch": 1.280359820089955,
      "grad_norm": 0.06936614542121676,
      "learning_rate": 6.909830056250527e-05,
      "loss": 0.0272,
      "step": 854
    },
    {
      "epoch": 1.2818590704647677,
      "grad_norm": 0.06026615690676764,
      "learning_rate": 6.88494207315912e-05,
      "loss": 0.0271,
      "step": 855
    },
    {
      "epoch": 1.2833583208395802,
      "grad_norm": 0.07021407853748118,
      "learning_rate": 6.860075440325951e-05,
      "loss": 0.0308,
      "step": 856
    },
    {
      "epoch": 1.2848575712143928,
      "grad_norm": 0.052574527846576825,
      "learning_rate": 6.835230328184138e-05,
      "loss": 0.0175,
      "step": 857
    },
    {
      "epoch": 1.2863568215892054,
      "grad_norm": 0.05298669412741438,
      "learning_rate": 6.8104069070193e-05,
      "loss": 0.0176,
      "step": 858
    },
    {
      "epoch": 1.287856071964018,
      "grad_norm": 0.04213736228746085,
      "learning_rate": 6.785605346968386e-05,
      "loss": 0.0175,
      "step": 859
    },
    {
      "epoch": 1.2893553223388305,
      "grad_norm": 0.03297170359235013,
      "learning_rate": 6.760825818018508e-05,
      "loss": 0.0293,
      "step": 860
    },
    {
      "epoch": 1.2908545727136431,
      "grad_norm": 0.07532993960457951,
      "learning_rate": 6.736068490005783e-05,
      "loss": 0.0361,
      "step": 861
    },
    {
      "epoch": 1.2923538230884557,
      "grad_norm": 0.06949245335822227,
      "learning_rate": 6.711333532614168e-05,
      "loss": 0.0428,
      "step": 862
    },
    {
      "epoch": 1.2938530734632683,
      "grad_norm": 0.033794502875320105,
      "learning_rate": 6.68662111537429e-05,
      "loss": 0.019,
      "step": 863
    },
    {
      "epoch": 1.295352323838081,
      "grad_norm": 0.042862124096238796,
      "learning_rate": 6.661931407662292e-05,
      "loss": 0.0165,
      "step": 864
    },
    {
      "epoch": 1.2968515742128934,
      "grad_norm": 0.03762469114322251,
      "learning_rate": 6.637264578698664e-05,
      "loss": 0.0177,
      "step": 865
    },
    {
      "epoch": 1.2983508245877062,
      "grad_norm": 0.06681907271371797,
      "learning_rate": 6.612620797547087e-05,
      "loss": 0.0425,
      "step": 866
    },
    {
      "epoch": 1.2998500749625188,
      "grad_norm": 0.06461028945930743,
      "learning_rate": 6.588000233113282e-05,
      "loss": 0.033,
      "step": 867
    },
    {
      "epoch": 1.3013493253373314,
      "grad_norm": 0.03602981021835219,
      "learning_rate": 6.56340305414384e-05,
      "loss": 0.0185,
      "step": 868
    },
    {
      "epoch": 1.302848575712144,
      "grad_norm": 0.07041331964732868,
      "learning_rate": 6.538829429225069e-05,
      "loss": 0.0375,
      "step": 869
    },
    {
      "epoch": 1.3043478260869565,
      "grad_norm": 0.053845021312289056,
      "learning_rate": 6.51427952678185e-05,
      "loss": 0.0264,
      "step": 870
    },
    {
      "epoch": 1.3058470764617691,
      "grad_norm": 0.03459818578421578,
      "learning_rate": 6.489753515076456e-05,
      "loss": 0.0145,
      "step": 871
    },
    {
      "epoch": 1.3073463268365817,
      "grad_norm": 0.04351524337603825,
      "learning_rate": 6.465251562207431e-05,
      "loss": 0.0207,
      "step": 872
    },
    {
      "epoch": 1.3088455772113943,
      "grad_norm": 0.04456815741914844,
      "learning_rate": 6.440773836108413e-05,
      "loss": 0.0278,
      "step": 873
    },
    {
      "epoch": 1.3103448275862069,
      "grad_norm": 0.061834383611745806,
      "learning_rate": 6.416320504546997e-05,
      "loss": 0.0332,
      "step": 874
    },
    {
      "epoch": 1.3118440779610194,
      "grad_norm": 0.045688059219150534,
      "learning_rate": 6.391891735123582e-05,
      "loss": 0.0435,
      "step": 875
    },
    {
      "epoch": 1.313343328335832,
      "grad_norm": 0.0499097200922593,
      "learning_rate": 6.367487695270217e-05,
      "loss": 0.0225,
      "step": 876
    },
    {
      "epoch": 1.3148425787106448,
      "grad_norm": 0.03930515307322255,
      "learning_rate": 6.343108552249457e-05,
      "loss": 0.0213,
      "step": 877
    },
    {
      "epoch": 1.3163418290854572,
      "grad_norm": 0.04750578901510908,
      "learning_rate": 6.318754473153221e-05,
      "loss": 0.0344,
      "step": 878
    },
    {
      "epoch": 1.31784107946027,
      "grad_norm": 0.05610290200169092,
      "learning_rate": 6.294425624901638e-05,
      "loss": 0.0322,
      "step": 879
    },
    {
      "epoch": 1.3193403298350825,
      "grad_norm": 0.029473494282681995,
      "learning_rate": 6.27012217424191e-05,
      "loss": 0.0134,
      "step": 880
    },
    {
      "epoch": 1.3208395802098951,
      "grad_norm": 0.05781224980544649,
      "learning_rate": 6.245844287747168e-05,
      "loss": 0.0354,
      "step": 881
    },
    {
      "epoch": 1.3223388305847077,
      "grad_norm": 0.06267338674733103,
      "learning_rate": 6.22159213181533e-05,
      "loss": 0.0361,
      "step": 882
    },
    {
      "epoch": 1.3238380809595203,
      "grad_norm": 0.038066306071301584,
      "learning_rate": 6.197365872667952e-05,
      "loss": 0.0181,
      "step": 883
    },
    {
      "epoch": 1.3253373313343328,
      "grad_norm": 0.07878387260295196,
      "learning_rate": 6.173165676349103e-05,
      "loss": 0.0682,
      "step": 884
    },
    {
      "epoch": 1.3268365817091454,
      "grad_norm": 0.04524683438782359,
      "learning_rate": 6.14899170872422e-05,
      "loss": 0.0295,
      "step": 885
    },
    {
      "epoch": 1.328335832083958,
      "grad_norm": 0.034393149243530124,
      "learning_rate": 6.12484413547897e-05,
      "loss": 0.0129,
      "step": 886
    },
    {
      "epoch": 1.3298350824587706,
      "grad_norm": 0.04589569278929691,
      "learning_rate": 6.1007231221181206e-05,
      "loss": 0.0214,
      "step": 887
    },
    {
      "epoch": 1.3313343328335832,
      "grad_norm": 0.0999532474797914,
      "learning_rate": 6.076628833964388e-05,
      "loss": 0.0241,
      "step": 888
    },
    {
      "epoch": 1.3328335832083957,
      "grad_norm": 0.06084071636014458,
      "learning_rate": 6.052561436157329e-05,
      "loss": 0.0242,
      "step": 889
    },
    {
      "epoch": 1.3343328335832085,
      "grad_norm": 0.04917934868115785,
      "learning_rate": 6.0285210936521955e-05,
      "loss": 0.0197,
      "step": 890
    },
    {
      "epoch": 1.3358320839580209,
      "grad_norm": 0.069354361338318,
      "learning_rate": 6.0045079712188014e-05,
      "loss": 0.0275,
      "step": 891
    },
    {
      "epoch": 1.3373313343328337,
      "grad_norm": 0.048448533551978,
      "learning_rate": 5.9805222334404e-05,
      "loss": 0.0167,
      "step": 892
    },
    {
      "epoch": 1.338830584707646,
      "grad_norm": 0.052410701670247924,
      "learning_rate": 5.956564044712551e-05,
      "loss": 0.0207,
      "step": 893
    },
    {
      "epoch": 1.3403298350824588,
      "grad_norm": 0.042448711771357485,
      "learning_rate": 5.9326335692419995e-05,
      "loss": 0.0192,
      "step": 894
    },
    {
      "epoch": 1.3418290854572714,
      "grad_norm": 0.05234893125921864,
      "learning_rate": 5.908730971045542e-05,
      "loss": 0.0317,
      "step": 895
    },
    {
      "epoch": 1.343328335832084,
      "grad_norm": 0.04331314810299154,
      "learning_rate": 5.884856413948913e-05,
      "loss": 0.0249,
      "step": 896
    },
    {
      "epoch": 1.3448275862068966,
      "grad_norm": 0.05916456115329105,
      "learning_rate": 5.8610100615856524e-05,
      "loss": 0.0462,
      "step": 897
    },
    {
      "epoch": 1.3463268365817092,
      "grad_norm": 0.062115100364669586,
      "learning_rate": 5.83719207739599e-05,
      "loss": 0.0256,
      "step": 898
    },
    {
      "epoch": 1.3478260869565217,
      "grad_norm": 0.035880225141386626,
      "learning_rate": 5.8134026246257225e-05,
      "loss": 0.0191,
      "step": 899
    },
    {
      "epoch": 1.3493253373313343,
      "grad_norm": 0.0573517254033885,
      "learning_rate": 5.789641866325091e-05,
      "loss": 0.0262,
      "step": 900
    },
    {
      "epoch": 1.3508245877061469,
      "grad_norm": 0.060303599881880644,
      "learning_rate": 5.76590996534768e-05,
      "loss": 0.0278,
      "step": 901
    },
    {
      "epoch": 1.3523238380809595,
      "grad_norm": 0.09628521288179971,
      "learning_rate": 5.7422070843492734e-05,
      "loss": 0.0356,
      "step": 902
    },
    {
      "epoch": 1.353823088455772,
      "grad_norm": 0.03617057375661126,
      "learning_rate": 5.718533385786772e-05,
      "loss": 0.02,
      "step": 903
    },
    {
      "epoch": 1.3553223388305846,
      "grad_norm": 0.044174128599893586,
      "learning_rate": 5.694889031917047e-05,
      "loss": 0.022,
      "step": 904
    },
    {
      "epoch": 1.3568215892053974,
      "grad_norm": 0.06221884291802812,
      "learning_rate": 5.671274184795865e-05,
      "loss": 0.0233,
      "step": 905
    },
    {
      "epoch": 1.3583208395802098,
      "grad_norm": 0.0711661250860426,
      "learning_rate": 5.647689006276726e-05,
      "loss": 0.0449,
      "step": 906
    },
    {
      "epoch": 1.3598200899550226,
      "grad_norm": 0.03490331687529149,
      "learning_rate": 5.624133658009817e-05,
      "loss": 0.0192,
      "step": 907
    },
    {
      "epoch": 1.3613193403298351,
      "grad_norm": 0.06280166629494677,
      "learning_rate": 5.6006083014408484e-05,
      "loss": 0.0257,
      "step": 908
    },
    {
      "epoch": 1.3628185907046477,
      "grad_norm": 0.049297420747513795,
      "learning_rate": 5.577113097809989e-05,
      "loss": 0.0287,
      "step": 909
    },
    {
      "epoch": 1.3643178410794603,
      "grad_norm": 0.04383360423444908,
      "learning_rate": 5.553648208150728e-05,
      "loss": 0.0159,
      "step": 910
    },
    {
      "epoch": 1.3658170914542729,
      "grad_norm": 0.06660129279871392,
      "learning_rate": 5.53021379328879e-05,
      "loss": 0.0316,
      "step": 911
    },
    {
      "epoch": 1.3673163418290855,
      "grad_norm": 0.030935221991709352,
      "learning_rate": 5.506810013841036e-05,
      "loss": 0.0141,
      "step": 912
    },
    {
      "epoch": 1.368815592203898,
      "grad_norm": 0.06598665791404248,
      "learning_rate": 5.483437030214343e-05,
      "loss": 0.0238,
      "step": 913
    },
    {
      "epoch": 1.3703148425787106,
      "grad_norm": 0.05185709390744075,
      "learning_rate": 5.4600950026045326e-05,
      "loss": 0.0268,
      "step": 914
    },
    {
      "epoch": 1.3718140929535232,
      "grad_norm": 0.06112957554123978,
      "learning_rate": 5.436784090995239e-05,
      "loss": 0.0384,
      "step": 915
    },
    {
      "epoch": 1.3733133433283358,
      "grad_norm": 0.02981664047266952,
      "learning_rate": 5.4135044551568546e-05,
      "loss": 0.0179,
      "step": 916
    },
    {
      "epoch": 1.3748125937031483,
      "grad_norm": 0.03404310873892391,
      "learning_rate": 5.390256254645378e-05,
      "loss": 0.0133,
      "step": 917
    },
    {
      "epoch": 1.3763118440779611,
      "grad_norm": 0.02748593423758078,
      "learning_rate": 5.3670396488013854e-05,
      "loss": 0.01,
      "step": 918
    },
    {
      "epoch": 1.3778110944527735,
      "grad_norm": 0.044108908050431356,
      "learning_rate": 5.343854796748886e-05,
      "loss": 0.016,
      "step": 919
    },
    {
      "epoch": 1.3793103448275863,
      "grad_norm": 0.04555240266490368,
      "learning_rate": 5.320701857394268e-05,
      "loss": 0.0242,
      "step": 920
    },
    {
      "epoch": 1.3808095952023989,
      "grad_norm": 0.025073591589012855,
      "learning_rate": 5.297580989425177e-05,
      "loss": 0.0102,
      "step": 921
    },
    {
      "epoch": 1.3823088455772115,
      "grad_norm": 0.03799373887428538,
      "learning_rate": 5.274492351309461e-05,
      "loss": 0.0199,
      "step": 922
    },
    {
      "epoch": 1.383808095952024,
      "grad_norm": 0.0634147924268712,
      "learning_rate": 5.251436101294056e-05,
      "loss": 0.0261,
      "step": 923
    },
    {
      "epoch": 1.3853073463268366,
      "grad_norm": 0.05771328219018494,
      "learning_rate": 5.2284123974039154e-05,
      "loss": 0.0283,
      "step": 924
    },
    {
      "epoch": 1.3868065967016492,
      "grad_norm": 0.04619876933398743,
      "learning_rate": 5.205421397440935e-05,
      "loss": 0.0218,
      "step": 925
    },
    {
      "epoch": 1.3883058470764618,
      "grad_norm": 0.05211266829503708,
      "learning_rate": 5.182463258982846e-05,
      "loss": 0.0273,
      "step": 926
    },
    {
      "epoch": 1.3898050974512743,
      "grad_norm": 0.0405536781310804,
      "learning_rate": 5.159538139382167e-05,
      "loss": 0.0163,
      "step": 927
    },
    {
      "epoch": 1.391304347826087,
      "grad_norm": 0.05228157708952931,
      "learning_rate": 5.1366461957650954e-05,
      "loss": 0.0148,
      "step": 928
    },
    {
      "epoch": 1.3928035982008995,
      "grad_norm": 0.051865023170566474,
      "learning_rate": 5.113787585030454e-05,
      "loss": 0.0253,
      "step": 929
    },
    {
      "epoch": 1.394302848575712,
      "grad_norm": 0.08414586524418576,
      "learning_rate": 5.090962463848592e-05,
      "loss": 0.0256,
      "step": 930
    },
    {
      "epoch": 1.3958020989505249,
      "grad_norm": 0.04333393205488659,
      "learning_rate": 5.068170988660346e-05,
      "loss": 0.0155,
      "step": 931
    },
    {
      "epoch": 1.3973013493253372,
      "grad_norm": 0.033783013394866904,
      "learning_rate": 5.045413315675924e-05,
      "loss": 0.0178,
      "step": 932
    },
    {
      "epoch": 1.39880059970015,
      "grad_norm": 0.041794554720093746,
      "learning_rate": 5.0226896008738786e-05,
      "loss": 0.0134,
      "step": 933
    },
    {
      "epoch": 1.4002998500749624,
      "grad_norm": 0.04766610890656114,
      "learning_rate": 5.000000000000002e-05,
      "loss": 0.0275,
      "step": 934
    },
    {
      "epoch": 1.4017991004497752,
      "grad_norm": 0.03981235051153443,
      "learning_rate": 4.977344668566275e-05,
      "loss": 0.0171,
      "step": 935
    },
    {
      "epoch": 1.4032983508245878,
      "grad_norm": 0.050230959070964275,
      "learning_rate": 4.9547237618498085e-05,
      "loss": 0.0247,
      "step": 936
    },
    {
      "epoch": 1.4047976011994003,
      "grad_norm": 0.04796385344740444,
      "learning_rate": 4.9321374348917585e-05,
      "loss": 0.0199,
      "step": 937
    },
    {
      "epoch": 1.406296851574213,
      "grad_norm": 0.03650112855229006,
      "learning_rate": 4.909585842496287e-05,
      "loss": 0.0197,
      "step": 938
    },
    {
      "epoch": 1.4077961019490255,
      "grad_norm": 0.058389964584798164,
      "learning_rate": 4.887069139229481e-05,
      "loss": 0.0305,
      "step": 939
    },
    {
      "epoch": 1.409295352323838,
      "grad_norm": 0.03657890885550993,
      "learning_rate": 4.864587479418302e-05,
      "loss": 0.0172,
      "step": 940
    },
    {
      "epoch": 1.4107946026986506,
      "grad_norm": 0.039034241468565725,
      "learning_rate": 4.842141017149526e-05,
      "loss": 0.0289,
      "step": 941
    },
    {
      "epoch": 1.4122938530734632,
      "grad_norm": 0.05779842475140324,
      "learning_rate": 4.8197299062686995e-05,
      "loss": 0.0194,
      "step": 942
    },
    {
      "epoch": 1.4137931034482758,
      "grad_norm": 0.04423489804284638,
      "learning_rate": 4.79735430037906e-05,
      "loss": 0.0167,
      "step": 943
    },
    {
      "epoch": 1.4152923538230884,
      "grad_norm": 0.06732550480304075,
      "learning_rate": 4.7750143528405126e-05,
      "loss": 0.0234,
      "step": 944
    },
    {
      "epoch": 1.416791604197901,
      "grad_norm": 0.048358500900887474,
      "learning_rate": 4.752710216768551e-05,
      "loss": 0.0126,
      "step": 945
    },
    {
      "epoch": 1.4182908545727138,
      "grad_norm": 0.0618555856514782,
      "learning_rate": 4.7304420450332244e-05,
      "loss": 0.0531,
      "step": 946
    },
    {
      "epoch": 1.419790104947526,
      "grad_norm": 0.021844937288201197,
      "learning_rate": 4.708209990258095e-05,
      "loss": 0.0092,
      "step": 947
    },
    {
      "epoch": 1.421289355322339,
      "grad_norm": 0.043762847247643975,
      "learning_rate": 4.68601420481917e-05,
      "loss": 0.0186,
      "step": 948
    },
    {
      "epoch": 1.4227886056971515,
      "grad_norm": 0.08941262921253182,
      "learning_rate": 4.6638548408438856e-05,
      "loss": 0.033,
      "step": 949
    },
    {
      "epoch": 1.424287856071964,
      "grad_norm": 0.07137680432326722,
      "learning_rate": 4.6417320502100316e-05,
      "loss": 0.019,
      "step": 950
    },
    {
      "epoch": 1.4257871064467766,
      "grad_norm": 0.05581448919899184,
      "learning_rate": 4.619645984544752e-05,
      "loss": 0.0175,
      "step": 951
    },
    {
      "epoch": 1.4272863568215892,
      "grad_norm": 0.029522071224618904,
      "learning_rate": 4.59759679522345e-05,
      "loss": 0.0191,
      "step": 952
    },
    {
      "epoch": 1.4287856071964018,
      "grad_norm": 0.03786913455549227,
      "learning_rate": 4.575584633368815e-05,
      "loss": 0.0197,
      "step": 953
    },
    {
      "epoch": 1.4302848575712144,
      "grad_norm": 0.059428163125673714,
      "learning_rate": 4.5536096498497295e-05,
      "loss": 0.0296,
      "step": 954
    },
    {
      "epoch": 1.431784107946027,
      "grad_norm": 0.03276712389103159,
      "learning_rate": 4.5316719952802776e-05,
      "loss": 0.0206,
      "step": 955
    },
    {
      "epoch": 1.4332833583208395,
      "grad_norm": 0.0411950835656051,
      "learning_rate": 4.5097718200186814e-05,
      "loss": 0.0147,
      "step": 956
    },
    {
      "epoch": 1.434782608695652,
      "grad_norm": 0.03726680229656808,
      "learning_rate": 4.4879092741663e-05,
      "loss": 0.0135,
      "step": 957
    },
    {
      "epoch": 1.4362818590704647,
      "grad_norm": 0.051339763257998224,
      "learning_rate": 4.46608450756656e-05,
      "loss": 0.0294,
      "step": 958
    },
    {
      "epoch": 1.4377811094452775,
      "grad_norm": 0.06516988257924636,
      "learning_rate": 4.444297669803981e-05,
      "loss": 0.0377,
      "step": 959
    },
    {
      "epoch": 1.4392803598200898,
      "grad_norm": 0.04522777428706767,
      "learning_rate": 4.4225489102030995e-05,
      "loss": 0.0164,
      "step": 960
    },
    {
      "epoch": 1.4407796101949026,
      "grad_norm": 0.057957991417301825,
      "learning_rate": 4.400838377827483e-05,
      "loss": 0.0292,
      "step": 961
    },
    {
      "epoch": 1.4422788605697152,
      "grad_norm": 0.04382325188747058,
      "learning_rate": 4.379166221478697e-05,
      "loss": 0.013,
      "step": 962
    },
    {
      "epoch": 1.4437781109445278,
      "grad_norm": 0.045850857607972945,
      "learning_rate": 4.35753258969526e-05,
      "loss": 0.0292,
      "step": 963
    },
    {
      "epoch": 1.4452773613193404,
      "grad_norm": 0.052319076851113866,
      "learning_rate": 4.335937630751674e-05,
      "loss": 0.0203,
      "step": 964
    },
    {
      "epoch": 1.446776611694153,
      "grad_norm": 0.04430763147098959,
      "learning_rate": 4.31438149265736e-05,
      "loss": 0.0174,
      "step": 965
    },
    {
      "epoch": 1.4482758620689655,
      "grad_norm": 0.037336734388071224,
      "learning_rate": 4.2928643231556844e-05,
      "loss": 0.0187,
      "step": 966
    },
    {
      "epoch": 1.449775112443778,
      "grad_norm": 0.03488717623172854,
      "learning_rate": 4.271386269722909e-05,
      "loss": 0.0215,
      "step": 967
    },
    {
      "epoch": 1.4512743628185907,
      "grad_norm": 0.11146756348958964,
      "learning_rate": 4.249947479567218e-05,
      "loss": 0.0632,
      "step": 968
    },
    {
      "epoch": 1.4527736131934033,
      "grad_norm": 0.0723346597125285,
      "learning_rate": 4.228548099627665e-05,
      "loss": 0.022,
      "step": 969
    },
    {
      "epoch": 1.4542728635682158,
      "grad_norm": 0.05077819533323603,
      "learning_rate": 4.207188276573214e-05,
      "loss": 0.0262,
      "step": 970
    },
    {
      "epoch": 1.4557721139430284,
      "grad_norm": 0.04220137883748374,
      "learning_rate": 4.185868156801694e-05,
      "loss": 0.022,
      "step": 971
    },
    {
      "epoch": 1.4572713643178412,
      "grad_norm": 0.04257073910089388,
      "learning_rate": 4.1645878864388266e-05,
      "loss": 0.0188,
      "step": 972
    },
    {
      "epoch": 1.4587706146926536,
      "grad_norm": 0.07051244032099359,
      "learning_rate": 4.143347611337194e-05,
      "loss": 0.0229,
      "step": 973
    },
    {
      "epoch": 1.4602698650674664,
      "grad_norm": 0.04924278467634398,
      "learning_rate": 4.12214747707527e-05,
      "loss": 0.02,
      "step": 974
    },
    {
      "epoch": 1.461769115442279,
      "grad_norm": 0.04465214846691316,
      "learning_rate": 4.1009876289563976e-05,
      "loss": 0.0204,
      "step": 975
    },
    {
      "epoch": 1.4632683658170915,
      "grad_norm": 0.04262641492883479,
      "learning_rate": 4.0798682120078044e-05,
      "loss": 0.0192,
      "step": 976
    },
    {
      "epoch": 1.464767616191904,
      "grad_norm": 0.03647273601340537,
      "learning_rate": 4.058789370979615e-05,
      "loss": 0.0184,
      "step": 977
    },
    {
      "epoch": 1.4662668665667167,
      "grad_norm": 0.04885135157810087,
      "learning_rate": 4.037751250343841e-05,
      "loss": 0.0303,
      "step": 978
    },
    {
      "epoch": 1.4677661169415293,
      "grad_norm": 0.08387069269883196,
      "learning_rate": 4.01675399429341e-05,
      "loss": 0.0234,
      "step": 979
    },
    {
      "epoch": 1.4692653673163418,
      "grad_norm": 0.038204055040209454,
      "learning_rate": 3.9957977467411615e-05,
      "loss": 0.0179,
      "step": 980
    },
    {
      "epoch": 1.4707646176911544,
      "grad_norm": 0.03786765301339409,
      "learning_rate": 3.9748826513188686e-05,
      "loss": 0.0165,
      "step": 981
    },
    {
      "epoch": 1.472263868065967,
      "grad_norm": 0.05231654154725003,
      "learning_rate": 3.954008851376252e-05,
      "loss": 0.0172,
      "step": 982
    },
    {
      "epoch": 1.4737631184407796,
      "grad_norm": 0.03335030350491901,
      "learning_rate": 3.933176489980005e-05,
      "loss": 0.0106,
      "step": 983
    },
    {
      "epoch": 1.4752623688155921,
      "grad_norm": 0.03979273183143895,
      "learning_rate": 3.9123857099127936e-05,
      "loss": 0.0212,
      "step": 984
    },
    {
      "epoch": 1.4767616191904047,
      "grad_norm": 0.05391309807180965,
      "learning_rate": 3.891636653672302e-05,
      "loss": 0.0157,
      "step": 985
    },
    {
      "epoch": 1.4782608695652173,
      "grad_norm": 0.03361291341690922,
      "learning_rate": 3.8709294634702376e-05,
      "loss": 0.0139,
      "step": 986
    },
    {
      "epoch": 1.47976011994003,
      "grad_norm": 0.035585947784928226,
      "learning_rate": 3.850264281231357e-05,
      "loss": 0.0219,
      "step": 987
    },
    {
      "epoch": 1.4812593703148424,
      "grad_norm": 0.03333439340678102,
      "learning_rate": 3.829641248592515e-05,
      "loss": 0.0129,
      "step": 988
    },
    {
      "epoch": 1.4827586206896552,
      "grad_norm": 0.040345561974199404,
      "learning_rate": 3.8090605069016595e-05,
      "loss": 0.0234,
      "step": 989
    },
    {
      "epoch": 1.4842578710644678,
      "grad_norm": 0.07403742521938174,
      "learning_rate": 3.788522197216897e-05,
      "loss": 0.02,
      "step": 990
    },
    {
      "epoch": 1.4857571214392804,
      "grad_norm": 0.05711176669065304,
      "learning_rate": 3.7680264603054994e-05,
      "loss": 0.0397,
      "step": 991
    },
    {
      "epoch": 1.487256371814093,
      "grad_norm": 0.06610246075931671,
      "learning_rate": 3.747573436642951e-05,
      "loss": 0.0277,
      "step": 992
    },
    {
      "epoch": 1.4887556221889056,
      "grad_norm": 0.03232609404738584,
      "learning_rate": 3.7271632664119846e-05,
      "loss": 0.0091,
      "step": 993
    },
    {
      "epoch": 1.4902548725637181,
      "grad_norm": 0.03787459790735476,
      "learning_rate": 3.7067960895016275e-05,
      "loss": 0.0209,
      "step": 994
    },
    {
      "epoch": 1.4917541229385307,
      "grad_norm": 0.04786121404829169,
      "learning_rate": 3.686472045506223e-05,
      "loss": 0.013,
      "step": 995
    },
    {
      "epoch": 1.4932533733133433,
      "grad_norm": 0.04350531988383918,
      "learning_rate": 3.6661912737245e-05,
      "loss": 0.0292,
      "step": 996
    },
    {
      "epoch": 1.4947526236881559,
      "grad_norm": 0.03931742385537951,
      "learning_rate": 3.645953913158593e-05,
      "loss": 0.027,
      "step": 997
    },
    {
      "epoch": 1.4962518740629684,
      "grad_norm": 0.041467982689201134,
      "learning_rate": 3.6257601025131026e-05,
      "loss": 0.0238,
      "step": 998
    },
    {
      "epoch": 1.497751124437781,
      "grad_norm": 0.04379142918036115,
      "learning_rate": 3.6056099801941534e-05,
      "loss": 0.0192,
      "step": 999
    },
    {
      "epoch": 1.4992503748125938,
      "grad_norm": 0.03952572570808585,
      "learning_rate": 3.585503684308421e-05,
      "loss": 0.0185,
      "step": 1000
    },
    {
      "epoch": 1.5007496251874062,
      "grad_norm": 0.04682486078750102,
      "learning_rate": 3.565441352662211e-05,
      "loss": 0.0172,
      "step": 1001
    },
    {
      "epoch": 1.502248875562219,
      "grad_norm": 0.05532999211305211,
      "learning_rate": 3.545423122760493e-05,
      "loss": 0.0277,
      "step": 1002
    },
    {
      "epoch": 1.5037481259370313,
      "grad_norm": 0.06235019222714452,
      "learning_rate": 3.525449131805982e-05,
      "loss": 0.0319,
      "step": 1003
    },
    {
      "epoch": 1.5052473763118441,
      "grad_norm": 0.06024901707215258,
      "learning_rate": 3.5055195166981645e-05,
      "loss": 0.0375,
      "step": 1004
    },
    {
      "epoch": 1.5067466266866567,
      "grad_norm": 0.0678387358265547,
      "learning_rate": 3.4856344140323985e-05,
      "loss": 0.0281,
      "step": 1005
    },
    {
      "epoch": 1.5082458770614693,
      "grad_norm": 0.054204263066234194,
      "learning_rate": 3.465793960098945e-05,
      "loss": 0.0434,
      "step": 1006
    },
    {
      "epoch": 1.5097451274362819,
      "grad_norm": 0.045371529886557356,
      "learning_rate": 3.445998290882062e-05,
      "loss": 0.0225,
      "step": 1007
    },
    {
      "epoch": 1.5112443778110944,
      "grad_norm": 0.12719949916957105,
      "learning_rate": 3.426247542059041e-05,
      "loss": 0.0482,
      "step": 1008
    },
    {
      "epoch": 1.512743628185907,
      "grad_norm": 0.039729045535986905,
      "learning_rate": 3.406541848999312e-05,
      "loss": 0.0135,
      "step": 1009
    },
    {
      "epoch": 1.5142428785607196,
      "grad_norm": 0.07776064383164068,
      "learning_rate": 3.386881346763483e-05,
      "loss": 0.0344,
      "step": 1010
    },
    {
      "epoch": 1.5157421289355324,
      "grad_norm": 0.031920259092681355,
      "learning_rate": 3.367266170102432e-05,
      "loss": 0.0123,
      "step": 1011
    },
    {
      "epoch": 1.5172413793103448,
      "grad_norm": 0.03993466785133325,
      "learning_rate": 3.347696453456393e-05,
      "loss": 0.0195,
      "step": 1012
    },
    {
      "epoch": 1.5187406296851576,
      "grad_norm": 0.03649095428535925,
      "learning_rate": 3.328172330954001e-05,
      "loss": 0.0127,
      "step": 1013
    },
    {
      "epoch": 1.52023988005997,
      "grad_norm": 0.05184896776912534,
      "learning_rate": 3.308693936411421e-05,
      "loss": 0.0389,
      "step": 1014
    },
    {
      "epoch": 1.5217391304347827,
      "grad_norm": 0.041105616161891526,
      "learning_rate": 3.289261403331372e-05,
      "loss": 0.0273,
      "step": 1015
    },
    {
      "epoch": 1.523238380809595,
      "grad_norm": 0.05332829161038912,
      "learning_rate": 3.269874864902269e-05,
      "loss": 0.0312,
      "step": 1016
    },
    {
      "epoch": 1.5247376311844079,
      "grad_norm": 0.029829522522220115,
      "learning_rate": 3.2505344539972705e-05,
      "loss": 0.0122,
      "step": 1017
    },
    {
      "epoch": 1.5262368815592204,
      "grad_norm": 0.05136530388164031,
      "learning_rate": 3.231240303173394e-05,
      "loss": 0.0128,
      "step": 1018
    },
    {
      "epoch": 1.527736131934033,
      "grad_norm": 0.026747951417845865,
      "learning_rate": 3.211992544670582e-05,
      "loss": 0.0139,
      "step": 1019
    },
    {
      "epoch": 1.5292353823088456,
      "grad_norm": 0.043634723111218424,
      "learning_rate": 3.192791310410822e-05,
      "loss": 0.0145,
      "step": 1020
    },
    {
      "epoch": 1.5307346326836582,
      "grad_norm": 0.032266616085346804,
      "learning_rate": 3.173636731997222e-05,
      "loss": 0.0136,
      "step": 1021
    },
    {
      "epoch": 1.5322338830584707,
      "grad_norm": 0.05249669591805282,
      "learning_rate": 3.154528940713113e-05,
      "loss": 0.0216,
      "step": 1022
    },
    {
      "epoch": 1.5337331334332833,
      "grad_norm": 0.04769500104005904,
      "learning_rate": 3.1354680675211635e-05,
      "loss": 0.0165,
      "step": 1023
    },
    {
      "epoch": 1.535232383808096,
      "grad_norm": 0.035229159668666236,
      "learning_rate": 3.116454243062459e-05,
      "loss": 0.0194,
      "step": 1024
    },
    {
      "epoch": 1.5367316341829085,
      "grad_norm": 0.05407112124934083,
      "learning_rate": 3.0974875976556284e-05,
      "loss": 0.0236,
      "step": 1025
    },
    {
      "epoch": 1.5382308845577213,
      "grad_norm": 0.025074397773416118,
      "learning_rate": 3.078568261295933e-05,
      "loss": 0.0141,
      "step": 1026
    },
    {
      "epoch": 1.5397301349325336,
      "grad_norm": 0.0445437291021861,
      "learning_rate": 3.059696363654386e-05,
      "loss": 0.0186,
      "step": 1027
    },
    {
      "epoch": 1.5412293853073464,
      "grad_norm": 0.0441384879409754,
      "learning_rate": 3.0408720340768572e-05,
      "loss": 0.0139,
      "step": 1028
    },
    {
      "epoch": 1.5427286356821588,
      "grad_norm": 0.0479088329800761,
      "learning_rate": 3.0220954015832003e-05,
      "loss": 0.0302,
      "step": 1029
    },
    {
      "epoch": 1.5442278860569716,
      "grad_norm": 0.034357781660932715,
      "learning_rate": 3.0033665948663448e-05,
      "loss": 0.0179,
      "step": 1030
    },
    {
      "epoch": 1.545727136431784,
      "grad_norm": 0.04351013918478569,
      "learning_rate": 2.9846857422914433e-05,
      "loss": 0.0237,
      "step": 1031
    },
    {
      "epoch": 1.5472263868065967,
      "grad_norm": 0.052463537989510735,
      "learning_rate": 2.9660529718949627e-05,
      "loss": 0.0264,
      "step": 1032
    },
    {
      "epoch": 1.5487256371814093,
      "grad_norm": 0.03561598179282845,
      "learning_rate": 2.9474684113838257e-05,
      "loss": 0.0199,
      "step": 1033
    },
    {
      "epoch": 1.550224887556222,
      "grad_norm": 0.03365987985086509,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 0.0122,
      "step": 1034
    },
    {
      "epoch": 1.5517241379310345,
      "grad_norm": 0.043573488595870076,
      "learning_rate": 2.910444429192265e-05,
      "loss": 0.0203,
      "step": 1035
    },
    {
      "epoch": 1.553223388305847,
      "grad_norm": 0.03592975923317004,
      "learning_rate": 2.8920052612700754e-05,
      "loss": 0.0148,
      "step": 1036
    },
    {
      "epoch": 1.5547226386806596,
      "grad_norm": 0.03740369119589898,
      "learning_rate": 2.8736148107479467e-05,
      "loss": 0.0122,
      "step": 1037
    },
    {
      "epoch": 1.5562218890554722,
      "grad_norm": 0.05866623225512895,
      "learning_rate": 2.8552732036719687e-05,
      "loss": 0.026,
      "step": 1038
    },
    {
      "epoch": 1.557721139430285,
      "grad_norm": 0.037514551907470446,
      "learning_rate": 2.8369805657534575e-05,
      "loss": 0.0101,
      "step": 1039
    },
    {
      "epoch": 1.5592203898050974,
      "grad_norm": 0.06919743559839049,
      "learning_rate": 2.8187370223681132e-05,
      "loss": 0.0186,
      "step": 1040
    },
    {
      "epoch": 1.5607196401799102,
      "grad_norm": 0.10860031837919593,
      "learning_rate": 2.800542698555132e-05,
      "loss": 0.0331,
      "step": 1041
    },
    {
      "epoch": 1.5622188905547225,
      "grad_norm": 0.04053934454593709,
      "learning_rate": 2.7823977190163786e-05,
      "loss": 0.0304,
      "step": 1042
    },
    {
      "epoch": 1.5637181409295353,
      "grad_norm": 0.04589812353371874,
      "learning_rate": 2.764302208115508e-05,
      "loss": 0.0288,
      "step": 1043
    },
    {
      "epoch": 1.5652173913043477,
      "grad_norm": 0.051759071563514204,
      "learning_rate": 2.746256289877126e-05,
      "loss": 0.0221,
      "step": 1044
    },
    {
      "epoch": 1.5667166416791605,
      "grad_norm": 0.04419463652932288,
      "learning_rate": 2.728260087985933e-05,
      "loss": 0.021,
      "step": 1045
    },
    {
      "epoch": 1.568215892053973,
      "grad_norm": 0.044966668706100936,
      "learning_rate": 2.7103137257858868e-05,
      "loss": 0.0337,
      "step": 1046
    },
    {
      "epoch": 1.5697151424287856,
      "grad_norm": 0.04302495760541486,
      "learning_rate": 2.6924173262793383e-05,
      "loss": 0.0137,
      "step": 1047
    },
    {
      "epoch": 1.5712143928035982,
      "grad_norm": 0.08218579196147292,
      "learning_rate": 2.6745710121262136e-05,
      "loss": 0.0365,
      "step": 1048
    },
    {
      "epoch": 1.5727136431784108,
      "grad_norm": 0.061167493252746986,
      "learning_rate": 2.6567749056431467e-05,
      "loss": 0.0258,
      "step": 1049
    },
    {
      "epoch": 1.5742128935532234,
      "grad_norm": 0.13364423709623616,
      "learning_rate": 2.639029128802657e-05,
      "loss": 0.0626,
      "step": 1050
    },
    {
      "epoch": 1.575712143928036,
      "grad_norm": 0.03214675852865434,
      "learning_rate": 2.6213338032323176e-05,
      "loss": 0.0133,
      "step": 1051
    },
    {
      "epoch": 1.5772113943028487,
      "grad_norm": 0.06328373661244126,
      "learning_rate": 2.603689050213902e-05,
      "loss": 0.0243,
      "step": 1052
    },
    {
      "epoch": 1.578710644677661,
      "grad_norm": 0.04742194356052198,
      "learning_rate": 2.5860949906825795e-05,
      "loss": 0.0285,
      "step": 1053
    },
    {
      "epoch": 1.580209895052474,
      "grad_norm": 0.048588116559285674,
      "learning_rate": 2.5685517452260567e-05,
      "loss": 0.0307,
      "step": 1054
    },
    {
      "epoch": 1.5817091454272862,
      "grad_norm": 0.04993527437674772,
      "learning_rate": 2.5510594340837824e-05,
      "loss": 0.0232,
      "step": 1055
    },
    {
      "epoch": 1.583208395802099,
      "grad_norm": 0.043690403745079286,
      "learning_rate": 2.5336181771460876e-05,
      "loss": 0.0205,
      "step": 1056
    },
    {
      "epoch": 1.5847076461769114,
      "grad_norm": 0.07438950423299034,
      "learning_rate": 2.516228093953402e-05,
      "loss": 0.0306,
      "step": 1057
    },
    {
      "epoch": 1.5862068965517242,
      "grad_norm": 0.03036817361902045,
      "learning_rate": 2.4988893036954043e-05,
      "loss": 0.0148,
      "step": 1058
    },
    {
      "epoch": 1.5877061469265368,
      "grad_norm": 0.0330965964673103,
      "learning_rate": 2.4816019252102273e-05,
      "loss": 0.0195,
      "step": 1059
    },
    {
      "epoch": 1.5892053973013494,
      "grad_norm": 0.035050676413594786,
      "learning_rate": 2.464366076983623e-05,
      "loss": 0.0136,
      "step": 1060
    },
    {
      "epoch": 1.590704647676162,
      "grad_norm": 0.04101060544019866,
      "learning_rate": 2.4471818771481648e-05,
      "loss": 0.0141,
      "step": 1061
    },
    {
      "epoch": 1.5922038980509745,
      "grad_norm": 0.06280027292019925,
      "learning_rate": 2.4300494434824373e-05,
      "loss": 0.0224,
      "step": 1062
    },
    {
      "epoch": 1.593703148425787,
      "grad_norm": 0.04131360603164168,
      "learning_rate": 2.412968893410218e-05,
      "loss": 0.017,
      "step": 1063
    },
    {
      "epoch": 1.5952023988005997,
      "grad_norm": 0.03827142257211483,
      "learning_rate": 2.3959403439996907e-05,
      "loss": 0.0221,
      "step": 1064
    },
    {
      "epoch": 1.5967016491754122,
      "grad_norm": 0.06504367581713787,
      "learning_rate": 2.3789639119626216e-05,
      "loss": 0.0267,
      "step": 1065
    },
    {
      "epoch": 1.5982008995502248,
      "grad_norm": 0.06818969255684648,
      "learning_rate": 2.362039713653581e-05,
      "loss": 0.0228,
      "step": 1066
    },
    {
      "epoch": 1.5997001499250376,
      "grad_norm": 0.038392151336867984,
      "learning_rate": 2.34516786506912e-05,
      "loss": 0.0154,
      "step": 1067
    },
    {
      "epoch": 1.60119940029985,
      "grad_norm": 0.06573629069388562,
      "learning_rate": 2.328348481847006e-05,
      "loss": 0.0251,
      "step": 1068
    },
    {
      "epoch": 1.6026986506746628,
      "grad_norm": 0.06148804102636816,
      "learning_rate": 2.3115816792654056e-05,
      "loss": 0.0311,
      "step": 1069
    },
    {
      "epoch": 1.6041979010494751,
      "grad_norm": 0.04471759363710799,
      "learning_rate": 2.2948675722421086e-05,
      "loss": 0.0191,
      "step": 1070
    },
    {
      "epoch": 1.605697151424288,
      "grad_norm": 0.0415533636509639,
      "learning_rate": 2.2782062753337306e-05,
      "loss": 0.0184,
      "step": 1071
    },
    {
      "epoch": 1.6071964017991005,
      "grad_norm": 0.05794733183745205,
      "learning_rate": 2.2615979027349387e-05,
      "loss": 0.0255,
      "step": 1072
    },
    {
      "epoch": 1.608695652173913,
      "grad_norm": 0.0397678166261267,
      "learning_rate": 2.2450425682776565e-05,
      "loss": 0.0218,
      "step": 1073
    },
    {
      "epoch": 1.6101949025487257,
      "grad_norm": 0.04761970893285829,
      "learning_rate": 2.2285403854302912e-05,
      "loss": 0.0195,
      "step": 1074
    },
    {
      "epoch": 1.6116941529235382,
      "grad_norm": 0.054708750805224075,
      "learning_rate": 2.2120914672969606e-05,
      "loss": 0.0207,
      "step": 1075
    },
    {
      "epoch": 1.6131934032983508,
      "grad_norm": 0.04536721690981569,
      "learning_rate": 2.195695926616702e-05,
      "loss": 0.0115,
      "step": 1076
    },
    {
      "epoch": 1.6146926536731634,
      "grad_norm": 0.03693822651785582,
      "learning_rate": 2.1793538757627218e-05,
      "loss": 0.0269,
      "step": 1077
    },
    {
      "epoch": 1.616191904047976,
      "grad_norm": 0.05430171107827179,
      "learning_rate": 2.163065426741603e-05,
      "loss": 0.0234,
      "step": 1078
    },
    {
      "epoch": 1.6176911544227885,
      "grad_norm": 0.03499504155734339,
      "learning_rate": 2.146830691192553e-05,
      "loss": 0.0215,
      "step": 1079
    },
    {
      "epoch": 1.6191904047976013,
      "grad_norm": 0.041779929626990946,
      "learning_rate": 2.1306497803866277e-05,
      "loss": 0.0219,
      "step": 1080
    },
    {
      "epoch": 1.6206896551724137,
      "grad_norm": 0.05662420578938302,
      "learning_rate": 2.1145228052259824e-05,
      "loss": 0.0266,
      "step": 1081
    },
    {
      "epoch": 1.6221889055472265,
      "grad_norm": 0.0632358405041289,
      "learning_rate": 2.098449876243096e-05,
      "loss": 0.0513,
      "step": 1082
    },
    {
      "epoch": 1.6236881559220389,
      "grad_norm": 0.03370533424123817,
      "learning_rate": 2.0824311036000276e-05,
      "loss": 0.0164,
      "step": 1083
    },
    {
      "epoch": 1.6251874062968517,
      "grad_norm": 0.03757323666893171,
      "learning_rate": 2.0664665970876496e-05,
      "loss": 0.0158,
      "step": 1084
    },
    {
      "epoch": 1.626686656671664,
      "grad_norm": 0.03313883661650599,
      "learning_rate": 2.0505564661249e-05,
      "loss": 0.0164,
      "step": 1085
    },
    {
      "epoch": 1.6281859070464768,
      "grad_norm": 0.04129568741796651,
      "learning_rate": 2.0347008197580374e-05,
      "loss": 0.0262,
      "step": 1086
    },
    {
      "epoch": 1.6296851574212894,
      "grad_norm": 0.050633574006442895,
      "learning_rate": 2.0188997666598854e-05,
      "loss": 0.0263,
      "step": 1087
    },
    {
      "epoch": 1.631184407796102,
      "grad_norm": 0.062148357793015474,
      "learning_rate": 2.0031534151290943e-05,
      "loss": 0.0322,
      "step": 1088
    },
    {
      "epoch": 1.6326836581709145,
      "grad_norm": 0.045969526657587925,
      "learning_rate": 1.9874618730893946e-05,
      "loss": 0.0168,
      "step": 1089
    },
    {
      "epoch": 1.6341829085457271,
      "grad_norm": 0.04612858174233097,
      "learning_rate": 1.9718252480888566e-05,
      "loss": 0.0286,
      "step": 1090
    },
    {
      "epoch": 1.6356821589205397,
      "grad_norm": 0.040575925623268116,
      "learning_rate": 1.9562436472991552e-05,
      "loss": 0.019,
      "step": 1091
    },
    {
      "epoch": 1.6371814092953523,
      "grad_norm": 0.04417146500100262,
      "learning_rate": 1.9407171775148436e-05,
      "loss": 0.0202,
      "step": 1092
    },
    {
      "epoch": 1.638680659670165,
      "grad_norm": 0.036770078126215765,
      "learning_rate": 1.9252459451526007e-05,
      "loss": 0.0129,
      "step": 1093
    },
    {
      "epoch": 1.6401799100449774,
      "grad_norm": 0.0364801904957448,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 0.015,
      "step": 1094
    },
    {
      "epoch": 1.6416791604197902,
      "grad_norm": 0.0254972319896911,
      "learning_rate": 1.8944696164673946e-05,
      "loss": 0.0139,
      "step": 1095
    },
    {
      "epoch": 1.6431784107946026,
      "grad_norm": 0.047093621027718184,
      "learning_rate": 1.879164731081937e-05,
      "loss": 0.0252,
      "step": 1096
    },
    {
      "epoch": 1.6446776611694154,
      "grad_norm": 0.04485052170740798,
      "learning_rate": 1.863915504992131e-05,
      "loss": 0.0202,
      "step": 1097
    },
    {
      "epoch": 1.6461769115442277,
      "grad_norm": 0.0387698802952675,
      "learning_rate": 1.848722042714457e-05,
      "loss": 0.0117,
      "step": 1098
    },
    {
      "epoch": 1.6476761619190405,
      "grad_norm": 0.0690313351636082,
      "learning_rate": 1.833584448383211e-05,
      "loss": 0.0225,
      "step": 1099
    },
    {
      "epoch": 1.6491754122938531,
      "grad_norm": 0.041762472042913064,
      "learning_rate": 1.818502825749764e-05,
      "loss": 0.0189,
      "step": 1100
    },
    {
      "epoch": 1.6506746626686657,
      "grad_norm": 0.043282701400431646,
      "learning_rate": 1.8034772781818775e-05,
      "loss": 0.0352,
      "step": 1101
    },
    {
      "epoch": 1.6521739130434783,
      "grad_norm": 0.05165034058427283,
      "learning_rate": 1.78850790866296e-05,
      "loss": 0.0225,
      "step": 1102
    },
    {
      "epoch": 1.6536731634182908,
      "grad_norm": 0.02717101883001689,
      "learning_rate": 1.773594819791403e-05,
      "loss": 0.0193,
      "step": 1103
    },
    {
      "epoch": 1.6551724137931034,
      "grad_norm": 0.04662587755796276,
      "learning_rate": 1.7587381137798432e-05,
      "loss": 0.0178,
      "step": 1104
    },
    {
      "epoch": 1.656671664167916,
      "grad_norm": 0.031822487513949387,
      "learning_rate": 1.7439378924544835e-05,
      "loss": 0.0189,
      "step": 1105
    },
    {
      "epoch": 1.6581709145427288,
      "grad_norm": 0.03280026362696025,
      "learning_rate": 1.7291942572543807e-05,
      "loss": 0.0123,
      "step": 1106
    },
    {
      "epoch": 1.6596701649175412,
      "grad_norm": 0.030180458183959057,
      "learning_rate": 1.714507309230766e-05,
      "loss": 0.0141,
      "step": 1107
    },
    {
      "epoch": 1.661169415292354,
      "grad_norm": 0.05658463863566536,
      "learning_rate": 1.6998771490463262e-05,
      "loss": 0.0647,
      "step": 1108
    },
    {
      "epoch": 1.6626686656671663,
      "grad_norm": 0.040781014716140526,
      "learning_rate": 1.6853038769745467e-05,
      "loss": 0.016,
      "step": 1109
    },
    {
      "epoch": 1.6641679160419791,
      "grad_norm": 0.04738429854592745,
      "learning_rate": 1.6707875928990058e-05,
      "loss": 0.03,
      "step": 1110
    },
    {
      "epoch": 1.6656671664167915,
      "grad_norm": 0.025248986437211247,
      "learning_rate": 1.656328396312682e-05,
      "loss": 0.0119,
      "step": 1111
    },
    {
      "epoch": 1.6671664167916043,
      "grad_norm": 0.08248141981476365,
      "learning_rate": 1.6419263863172997e-05,
      "loss": 0.0401,
      "step": 1112
    },
    {
      "epoch": 1.6686656671664168,
      "grad_norm": 0.040376633264695476,
      "learning_rate": 1.6275816616226113e-05,
      "loss": 0.0162,
      "step": 1113
    },
    {
      "epoch": 1.6701649175412294,
      "grad_norm": 0.04326524362791187,
      "learning_rate": 1.6132943205457606e-05,
      "loss": 0.0174,
      "step": 1114
    },
    {
      "epoch": 1.671664167916042,
      "grad_norm": 0.04807125621491688,
      "learning_rate": 1.599064461010581e-05,
      "loss": 0.0324,
      "step": 1115
    },
    {
      "epoch": 1.6731634182908546,
      "grad_norm": 0.03729442007384064,
      "learning_rate": 1.5848921805469397e-05,
      "loss": 0.0211,
      "step": 1116
    },
    {
      "epoch": 1.6746626686656672,
      "grad_norm": 0.029001614577088476,
      "learning_rate": 1.5707775762900547e-05,
      "loss": 0.0129,
      "step": 1117
    },
    {
      "epoch": 1.6761619190404797,
      "grad_norm": 0.030669347972443403,
      "learning_rate": 1.5567207449798515e-05,
      "loss": 0.01,
      "step": 1118
    },
    {
      "epoch": 1.6776611694152923,
      "grad_norm": 0.04829682712396594,
      "learning_rate": 1.542721782960268e-05,
      "loss": 0.0232,
      "step": 1119
    },
    {
      "epoch": 1.6791604197901049,
      "grad_norm": 0.04806923915457913,
      "learning_rate": 1.528780786178631e-05,
      "loss": 0.0202,
      "step": 1120
    },
    {
      "epoch": 1.6806596701649177,
      "grad_norm": 0.02440255176031142,
      "learning_rate": 1.5148978501849642e-05,
      "loss": 0.0128,
      "step": 1121
    },
    {
      "epoch": 1.68215892053973,
      "grad_norm": 0.04063406158834547,
      "learning_rate": 1.5010730701313625e-05,
      "loss": 0.0171,
      "step": 1122
    },
    {
      "epoch": 1.6836581709145428,
      "grad_norm": 0.04035147945576393,
      "learning_rate": 1.4873065407713149e-05,
      "loss": 0.031,
      "step": 1123
    },
    {
      "epoch": 1.6851574212893552,
      "grad_norm": 0.04027060084795129,
      "learning_rate": 1.4735983564590783e-05,
      "loss": 0.0182,
      "step": 1124
    },
    {
      "epoch": 1.686656671664168,
      "grad_norm": 0.045280521178547195,
      "learning_rate": 1.4599486111490102e-05,
      "loss": 0.0198,
      "step": 1125
    },
    {
      "epoch": 1.6881559220389803,
      "grad_norm": 0.04881923000165193,
      "learning_rate": 1.4463573983949341e-05,
      "loss": 0.0167,
      "step": 1126
    },
    {
      "epoch": 1.6896551724137931,
      "grad_norm": 0.03529851016922019,
      "learning_rate": 1.4328248113495047e-05,
      "loss": 0.0165,
      "step": 1127
    },
    {
      "epoch": 1.6911544227886057,
      "grad_norm": 0.04376517979604228,
      "learning_rate": 1.4193509427635543e-05,
      "loss": 0.0266,
      "step": 1128
    },
    {
      "epoch": 1.6926536731634183,
      "grad_norm": 0.03839629594972581,
      "learning_rate": 1.405935884985473e-05,
      "loss": 0.0109,
      "step": 1129
    },
    {
      "epoch": 1.6941529235382309,
      "grad_norm": 0.05394784900443594,
      "learning_rate": 1.3925797299605647e-05,
      "loss": 0.0289,
      "step": 1130
    },
    {
      "epoch": 1.6956521739130435,
      "grad_norm": 0.04473750186048263,
      "learning_rate": 1.379282569230419e-05,
      "loss": 0.0183,
      "step": 1131
    },
    {
      "epoch": 1.697151424287856,
      "grad_norm": 0.03971094977709959,
      "learning_rate": 1.3660444939322836e-05,
      "loss": 0.0131,
      "step": 1132
    },
    {
      "epoch": 1.6986506746626686,
      "grad_norm": 0.044781528569248615,
      "learning_rate": 1.3528655947984504e-05,
      "loss": 0.0606,
      "step": 1133
    },
    {
      "epoch": 1.7001499250374814,
      "grad_norm": 0.05687647115833445,
      "learning_rate": 1.339745962155613e-05,
      "loss": 0.0261,
      "step": 1134
    },
    {
      "epoch": 1.7016491754122938,
      "grad_norm": 0.051596153458091924,
      "learning_rate": 1.3266856859242705e-05,
      "loss": 0.0217,
      "step": 1135
    },
    {
      "epoch": 1.7031484257871066,
      "grad_norm": 0.05150733833063215,
      "learning_rate": 1.3136848556180892e-05,
      "loss": 0.0306,
      "step": 1136
    },
    {
      "epoch": 1.704647676161919,
      "grad_norm": 0.042545970063710185,
      "learning_rate": 1.3007435603433039e-05,
      "loss": 0.0195,
      "step": 1137
    },
    {
      "epoch": 1.7061469265367317,
      "grad_norm": 0.044603161701574794,
      "learning_rate": 1.2878618887981064e-05,
      "loss": 0.0241,
      "step": 1138
    },
    {
      "epoch": 1.707646176911544,
      "grad_norm": 0.05764350301977242,
      "learning_rate": 1.2750399292720283e-05,
      "loss": 0.0216,
      "step": 1139
    },
    {
      "epoch": 1.7091454272863569,
      "grad_norm": 0.05335928285880417,
      "learning_rate": 1.262277769645348e-05,
      "loss": 0.0375,
      "step": 1140
    },
    {
      "epoch": 1.7106446776611695,
      "grad_norm": 0.036542466613040704,
      "learning_rate": 1.2495754973884766e-05,
      "loss": 0.0217,
      "step": 1141
    },
    {
      "epoch": 1.712143928035982,
      "grad_norm": 0.0317543900475218,
      "learning_rate": 1.2369331995613665e-05,
      "loss": 0.0089,
      "step": 1142
    },
    {
      "epoch": 1.7136431784107946,
      "grad_norm": 0.043953143850443466,
      "learning_rate": 1.224350962812908e-05,
      "loss": 0.0221,
      "step": 1143
    },
    {
      "epoch": 1.7151424287856072,
      "grad_norm": 0.04475451692804134,
      "learning_rate": 1.2118288733803473e-05,
      "loss": 0.0269,
      "step": 1144
    },
    {
      "epoch": 1.7166416791604198,
      "grad_norm": 0.0468528264899152,
      "learning_rate": 1.1993670170886806e-05,
      "loss": 0.0251,
      "step": 1145
    },
    {
      "epoch": 1.7181409295352323,
      "grad_norm": 0.04567111932909807,
      "learning_rate": 1.1869654793500784e-05,
      "loss": 0.0256,
      "step": 1146
    },
    {
      "epoch": 1.7196401799100451,
      "grad_norm": 0.04665397165143054,
      "learning_rate": 1.17462434516329e-05,
      "loss": 0.0287,
      "step": 1147
    },
    {
      "epoch": 1.7211394302848575,
      "grad_norm": 0.043566933804619214,
      "learning_rate": 1.1623436991130654e-05,
      "loss": 0.0193,
      "step": 1148
    },
    {
      "epoch": 1.7226386806596703,
      "grad_norm": 0.03910118483915222,
      "learning_rate": 1.1501236253695823e-05,
      "loss": 0.019,
      "step": 1149
    },
    {
      "epoch": 1.7241379310344827,
      "grad_norm": 0.042811788583233146,
      "learning_rate": 1.1379642076878527e-05,
      "loss": 0.0192,
      "step": 1150
    },
    {
      "epoch": 1.7256371814092955,
      "grad_norm": 0.03259518496802405,
      "learning_rate": 1.1258655294071685e-05,
      "loss": 0.0114,
      "step": 1151
    },
    {
      "epoch": 1.7271364317841078,
      "grad_norm": 0.03610690907310535,
      "learning_rate": 1.1138276734505104e-05,
      "loss": 0.0088,
      "step": 1152
    },
    {
      "epoch": 1.7286356821589206,
      "grad_norm": 0.058981199182703795,
      "learning_rate": 1.1018507223240038e-05,
      "loss": 0.0303,
      "step": 1153
    },
    {
      "epoch": 1.7301349325337332,
      "grad_norm": 0.04288242904489575,
      "learning_rate": 1.0899347581163221e-05,
      "loss": 0.0316,
      "step": 1154
    },
    {
      "epoch": 1.7316341829085458,
      "grad_norm": 0.0322212652007745,
      "learning_rate": 1.0780798624981547e-05,
      "loss": 0.0127,
      "step": 1155
    },
    {
      "epoch": 1.7331334332833583,
      "grad_norm": 0.04036547160797265,
      "learning_rate": 1.0662861167216243e-05,
      "loss": 0.0268,
      "step": 1156
    },
    {
      "epoch": 1.734632683658171,
      "grad_norm": 0.027977601513989023,
      "learning_rate": 1.0545536016197499e-05,
      "loss": 0.0106,
      "step": 1157
    },
    {
      "epoch": 1.7361319340329835,
      "grad_norm": 0.04130410258519725,
      "learning_rate": 1.042882397605871e-05,
      "loss": 0.0164,
      "step": 1158
    },
    {
      "epoch": 1.737631184407796,
      "grad_norm": 0.09110404984904212,
      "learning_rate": 1.0312725846731175e-05,
      "loss": 0.032,
      "step": 1159
    },
    {
      "epoch": 1.7391304347826086,
      "grad_norm": 0.04675587064403581,
      "learning_rate": 1.0197242423938446e-05,
      "loss": 0.0112,
      "step": 1160
    },
    {
      "epoch": 1.7406296851574212,
      "grad_norm": 0.05310767217545907,
      "learning_rate": 1.008237449919096e-05,
      "loss": 0.0208,
      "step": 1161
    },
    {
      "epoch": 1.742128935532234,
      "grad_norm": 0.040653637020288395,
      "learning_rate": 9.968122859780648e-06,
      "loss": 0.0206,
      "step": 1162
    },
    {
      "epoch": 1.7436281859070464,
      "grad_norm": 0.04403126184179531,
      "learning_rate": 9.854488288775422e-06,
      "loss": 0.0151,
      "step": 1163
    },
    {
      "epoch": 1.7451274362818592,
      "grad_norm": 0.030076096749230896,
      "learning_rate": 9.74147156501396e-06,
      "loss": 0.0113,
      "step": 1164
    },
    {
      "epoch": 1.7466266866566715,
      "grad_norm": 0.029651097446938043,
      "learning_rate": 9.629073463100146e-06,
      "loss": 0.0082,
      "step": 1165
    },
    {
      "epoch": 1.7481259370314843,
      "grad_norm": 0.03931051532613931,
      "learning_rate": 9.517294753398064e-06,
      "loss": 0.0206,
      "step": 1166
    },
    {
      "epoch": 1.7496251874062967,
      "grad_norm": 0.04890773527042702,
      "learning_rate": 9.406136202026416e-06,
      "loss": 0.0186,
      "step": 1167
    },
    {
      "epoch": 1.7511244377811095,
      "grad_norm": 0.05681064457174664,
      "learning_rate": 9.295598570853514e-06,
      "loss": 0.0241,
      "step": 1168
    },
    {
      "epoch": 1.752623688155922,
      "grad_norm": 0.033925497614483074,
      "learning_rate": 9.185682617491863e-06,
      "loss": 0.0169,
      "step": 1169
    },
    {
      "epoch": 1.7541229385307346,
      "grad_norm": 0.050500607910895993,
      "learning_rate": 9.076389095293148e-06,
      "loss": 0.03,
      "step": 1170
    },
    {
      "epoch": 1.7556221889055472,
      "grad_norm": 0.03528769870947365,
      "learning_rate": 8.967718753342902e-06,
      "loss": 0.015,
      "step": 1171
    },
    {
      "epoch": 1.7571214392803598,
      "grad_norm": 0.03966138334448402,
      "learning_rate": 8.85967233645547e-06,
      "loss": 0.0151,
      "step": 1172
    },
    {
      "epoch": 1.7586206896551724,
      "grad_norm": 0.05472974585600949,
      "learning_rate": 8.752250585168931e-06,
      "loss": 0.0272,
      "step": 1173
    },
    {
      "epoch": 1.760119940029985,
      "grad_norm": 0.04658097856564392,
      "learning_rate": 8.645454235739903e-06,
      "loss": 0.0273,
      "step": 1174
    },
    {
      "epoch": 1.7616191904047978,
      "grad_norm": 0.04039569121015692,
      "learning_rate": 8.539284020138638e-06,
      "loss": 0.0137,
      "step": 1175
    },
    {
      "epoch": 1.76311844077961,
      "grad_norm": 0.04821730129601109,
      "learning_rate": 8.433740666043898e-06,
      "loss": 0.034,
      "step": 1176
    },
    {
      "epoch": 1.764617691154423,
      "grad_norm": 0.04511525491973174,
      "learning_rate": 8.328824896838006e-06,
      "loss": 0.0105,
      "step": 1177
    },
    {
      "epoch": 1.7661169415292353,
      "grad_norm": 0.04234182612968715,
      "learning_rate": 8.224537431601886e-06,
      "loss": 0.0159,
      "step": 1178
    },
    {
      "epoch": 1.767616191904048,
      "grad_norm": 0.0321447391016331,
      "learning_rate": 8.12087898511018e-06,
      "loss": 0.0133,
      "step": 1179
    },
    {
      "epoch": 1.7691154422788604,
      "grad_norm": 0.07341997220230499,
      "learning_rate": 8.017850267826232e-06,
      "loss": 0.0343,
      "step": 1180
    },
    {
      "epoch": 1.7706146926536732,
      "grad_norm": 0.05203785895813134,
      "learning_rate": 7.915451985897382e-06,
      "loss": 0.0321,
      "step": 1181
    },
    {
      "epoch": 1.7721139430284858,
      "grad_norm": 0.041084567895159414,
      "learning_rate": 7.81368484114996e-06,
      "loss": 0.0147,
      "step": 1182
    },
    {
      "epoch": 1.7736131934032984,
      "grad_norm": 0.031087393200696114,
      "learning_rate": 7.71254953108458e-06,
      "loss": 0.0095,
      "step": 1183
    },
    {
      "epoch": 1.775112443778111,
      "grad_norm": 0.058786196696954986,
      "learning_rate": 7.612046748871327e-06,
      "loss": 0.0343,
      "step": 1184
    },
    {
      "epoch": 1.7766116941529235,
      "grad_norm": 0.04113739414695195,
      "learning_rate": 7.512177183345026e-06,
      "loss": 0.0181,
      "step": 1185
    },
    {
      "epoch": 1.778110944527736,
      "grad_norm": 0.03883727700834806,
      "learning_rate": 7.412941519000527e-06,
      "loss": 0.0145,
      "step": 1186
    },
    {
      "epoch": 1.7796101949025487,
      "grad_norm": 0.041402364797904934,
      "learning_rate": 7.314340435987921e-06,
      "loss": 0.0147,
      "step": 1187
    },
    {
      "epoch": 1.7811094452773615,
      "grad_norm": 0.05299280675746249,
      "learning_rate": 7.216374610108012e-06,
      "loss": 0.0199,
      "step": 1188
    },
    {
      "epoch": 1.7826086956521738,
      "grad_norm": 0.02456671956395044,
      "learning_rate": 7.119044712807577e-06,
      "loss": 0.0107,
      "step": 1189
    },
    {
      "epoch": 1.7841079460269866,
      "grad_norm": 0.03171980923691591,
      "learning_rate": 7.022351411174866e-06,
      "loss": 0.0142,
      "step": 1190
    },
    {
      "epoch": 1.785607196401799,
      "grad_norm": 0.05325976441184379,
      "learning_rate": 6.92629536793491e-06,
      "loss": 0.0212,
      "step": 1191
    },
    {
      "epoch": 1.7871064467766118,
      "grad_norm": 0.04370623276607864,
      "learning_rate": 6.830877241445111e-06,
      "loss": 0.0246,
      "step": 1192
    },
    {
      "epoch": 1.7886056971514241,
      "grad_norm": 0.06432358936521287,
      "learning_rate": 6.736097685690601e-06,
      "loss": 0.0208,
      "step": 1193
    },
    {
      "epoch": 1.790104947526237,
      "grad_norm": 0.03404343307449647,
      "learning_rate": 6.6419573502798374e-06,
      "loss": 0.0125,
      "step": 1194
    },
    {
      "epoch": 1.7916041979010495,
      "grad_norm": 0.056376443801887534,
      "learning_rate": 6.548456880440135e-06,
      "loss": 0.0234,
      "step": 1195
    },
    {
      "epoch": 1.793103448275862,
      "grad_norm": 0.04235044723804527,
      "learning_rate": 6.455596917013273e-06,
      "loss": 0.0231,
      "step": 1196
    },
    {
      "epoch": 1.7946026986506747,
      "grad_norm": 0.03653428510070535,
      "learning_rate": 6.363378096451011e-06,
      "loss": 0.0245,
      "step": 1197
    },
    {
      "epoch": 1.7961019490254873,
      "grad_norm": 0.03721457535492326,
      "learning_rate": 6.2718010508108545e-06,
      "loss": 0.0133,
      "step": 1198
    },
    {
      "epoch": 1.7976011994002998,
      "grad_norm": 0.038261575822808695,
      "learning_rate": 6.180866407751595e-06,
      "loss": 0.0169,
      "step": 1199
    },
    {
      "epoch": 1.7991004497751124,
      "grad_norm": 0.044065953447000386,
      "learning_rate": 6.090574790529091e-06,
      "loss": 0.0167,
      "step": 1200
    },
    {
      "epoch": 1.800599700149925,
      "grad_norm": 0.04771815395917711,
      "learning_rate": 6.000926817991992e-06,
      "loss": 0.0181,
      "step": 1201
    },
    {
      "epoch": 1.8020989505247376,
      "grad_norm": 0.03612809486970171,
      "learning_rate": 5.911923104577455e-06,
      "loss": 0.0143,
      "step": 1202
    },
    {
      "epoch": 1.8035982008995504,
      "grad_norm": 0.06286516442714955,
      "learning_rate": 5.823564260306968e-06,
      "loss": 0.0287,
      "step": 1203
    },
    {
      "epoch": 1.8050974512743627,
      "grad_norm": 0.040689385196554954,
      "learning_rate": 5.735850890782157e-06,
      "loss": 0.0154,
      "step": 1204
    },
    {
      "epoch": 1.8065967016491755,
      "grad_norm": 0.05935355542909139,
      "learning_rate": 5.6487835971806515e-06,
      "loss": 0.0249,
      "step": 1205
    },
    {
      "epoch": 1.8080959520239879,
      "grad_norm": 0.026213533109652836,
      "learning_rate": 5.562362976251901e-06,
      "loss": 0.0102,
      "step": 1206
    },
    {
      "epoch": 1.8095952023988007,
      "grad_norm": 0.0235654511150688,
      "learning_rate": 5.476589620313188e-06,
      "loss": 0.0105,
      "step": 1207
    },
    {
      "epoch": 1.811094452773613,
      "grad_norm": 0.0635366912137273,
      "learning_rate": 5.39146411724547e-06,
      "loss": 0.018,
      "step": 1208
    },
    {
      "epoch": 1.8125937031484258,
      "grad_norm": 0.041487867509385566,
      "learning_rate": 5.306987050489442e-06,
      "loss": 0.0366,
      "step": 1209
    },
    {
      "epoch": 1.8140929535232384,
      "grad_norm": 0.04638926153042607,
      "learning_rate": 5.223158999041444e-06,
      "loss": 0.0252,
      "step": 1210
    },
    {
      "epoch": 1.815592203898051,
      "grad_norm": 0.060453405404883885,
      "learning_rate": 5.13998053744954e-06,
      "loss": 0.033,
      "step": 1211
    },
    {
      "epoch": 1.8170914542728636,
      "grad_norm": 0.026690496589016864,
      "learning_rate": 5.057452235809624e-06,
      "loss": 0.0136,
      "step": 1212
    },
    {
      "epoch": 1.8185907046476761,
      "grad_norm": 0.04363010283975346,
      "learning_rate": 4.975574659761406e-06,
      "loss": 0.0228,
      "step": 1213
    },
    {
      "epoch": 1.8200899550224887,
      "grad_norm": 0.046092550493469546,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 0.0261,
      "step": 1214
    },
    {
      "epoch": 1.8215892053973013,
      "grad_norm": 0.05531628192606062,
      "learning_rate": 4.8137739246952195e-06,
      "loss": 0.0145,
      "step": 1215
    },
    {
      "epoch": 1.823088455772114,
      "grad_norm": 0.08123174441052526,
      "learning_rate": 4.733851874641382e-06,
      "loss": 0.0317,
      "step": 1216
    },
    {
      "epoch": 1.8245877061469264,
      "grad_norm": 0.04288348187685808,
      "learning_rate": 4.65458276809988e-06,
      "loss": 0.0203,
      "step": 1217
    },
    {
      "epoch": 1.8260869565217392,
      "grad_norm": 0.04334355640597478,
      "learning_rate": 4.575967148372317e-06,
      "loss": 0.0158,
      "step": 1218
    },
    {
      "epoch": 1.8275862068965516,
      "grad_norm": 0.037010861171650276,
      "learning_rate": 4.498005554281337e-06,
      "loss": 0.0173,
      "step": 1219
    },
    {
      "epoch": 1.8290854572713644,
      "grad_norm": 0.0389453296408419,
      "learning_rate": 4.420698520166988e-06,
      "loss": 0.0125,
      "step": 1220
    },
    {
      "epoch": 1.8305847076461768,
      "grad_norm": 0.04714273729670721,
      "learning_rate": 4.34404657588302e-06,
      "loss": 0.0214,
      "step": 1221
    },
    {
      "epoch": 1.8320839580209896,
      "grad_norm": 0.07183803253489271,
      "learning_rate": 4.268050246793276e-06,
      "loss": 0.0633,
      "step": 1222
    },
    {
      "epoch": 1.8335832083958021,
      "grad_norm": 0.03354509726743248,
      "learning_rate": 4.19271005376809e-06,
      "loss": 0.0209,
      "step": 1223
    },
    {
      "epoch": 1.8350824587706147,
      "grad_norm": 0.09728906022869646,
      "learning_rate": 4.118026513180695e-06,
      "loss": 0.0475,
      "step": 1224
    },
    {
      "epoch": 1.8365817091454273,
      "grad_norm": 0.03431271996396712,
      "learning_rate": 4.044000136903736e-06,
      "loss": 0.013,
      "step": 1225
    },
    {
      "epoch": 1.8380809595202399,
      "grad_norm": 0.041831243188060474,
      "learning_rate": 3.970631432305694e-06,
      "loss": 0.0266,
      "step": 1226
    },
    {
      "epoch": 1.8395802098950524,
      "grad_norm": 0.057870391772892306,
      "learning_rate": 3.897920902247465e-06,
      "loss": 0.0248,
      "step": 1227
    },
    {
      "epoch": 1.841079460269865,
      "grad_norm": 0.04691162453728428,
      "learning_rate": 3.825869045078867e-06,
      "loss": 0.0132,
      "step": 1228
    },
    {
      "epoch": 1.8425787106446778,
      "grad_norm": 0.03562764414364893,
      "learning_rate": 3.7544763546352834e-06,
      "loss": 0.0235,
      "step": 1229
    },
    {
      "epoch": 1.8440779610194902,
      "grad_norm": 0.06764557500741161,
      "learning_rate": 3.68374332023419e-06,
      "loss": 0.0416,
      "step": 1230
    },
    {
      "epoch": 1.845577211394303,
      "grad_norm": 0.042910585008751875,
      "learning_rate": 3.6136704266719113e-06,
      "loss": 0.0221,
      "step": 1231
    },
    {
      "epoch": 1.8470764617691153,
      "grad_norm": 0.038280889879914855,
      "learning_rate": 3.5442581542201923e-06,
      "loss": 0.0162,
      "step": 1232
    },
    {
      "epoch": 1.8485757121439281,
      "grad_norm": 0.05556984768768352,
      "learning_rate": 3.475506978623e-06,
      "loss": 0.023,
      "step": 1233
    },
    {
      "epoch": 1.8500749625187405,
      "grad_norm": 0.0503673855715429,
      "learning_rate": 3.40741737109318e-06,
      "loss": 0.016,
      "step": 1234
    },
    {
      "epoch": 1.8515742128935533,
      "grad_norm": 0.03274390981831604,
      "learning_rate": 3.339989798309273e-06,
      "loss": 0.0115,
      "step": 1235
    },
    {
      "epoch": 1.8530734632683659,
      "grad_norm": 0.036087042920603704,
      "learning_rate": 3.273224722412327e-06,
      "loss": 0.0136,
      "step": 1236
    },
    {
      "epoch": 1.8545727136431784,
      "grad_norm": 0.06008954234162343,
      "learning_rate": 3.2071226010026765e-06,
      "loss": 0.0199,
      "step": 1237
    },
    {
      "epoch": 1.856071964017991,
      "grad_norm": 0.04454845739153237,
      "learning_rate": 3.1416838871368924e-06,
      "loss": 0.0258,
      "step": 1238
    },
    {
      "epoch": 1.8575712143928036,
      "grad_norm": 0.04443343456113566,
      "learning_rate": 3.0769090293245705e-06,
      "loss": 0.0199,
      "step": 1239
    },
    {
      "epoch": 1.8590704647676162,
      "grad_norm": 0.06673692405990893,
      "learning_rate": 3.012798471525324e-06,
      "loss": 0.0633,
      "step": 1240
    },
    {
      "epoch": 1.8605697151424287,
      "grad_norm": 0.049212673297310765,
      "learning_rate": 2.949352653145754e-06,
      "loss": 0.0262,
      "step": 1241
    },
    {
      "epoch": 1.8620689655172413,
      "grad_norm": 0.06719806910277995,
      "learning_rate": 2.8865720090364034e-06,
      "loss": 0.0232,
      "step": 1242
    },
    {
      "epoch": 1.863568215892054,
      "grad_norm": 0.05588717349611258,
      "learning_rate": 2.8244569694887426e-06,
      "loss": 0.0136,
      "step": 1243
    },
    {
      "epoch": 1.8650674662668667,
      "grad_norm": 0.0404242627373259,
      "learning_rate": 2.7630079602323442e-06,
      "loss": 0.012,
      "step": 1244
    },
    {
      "epoch": 1.866566716641679,
      "grad_norm": 0.03967545521601158,
      "learning_rate": 2.7022254024318015e-06,
      "loss": 0.0175,
      "step": 1245
    },
    {
      "epoch": 1.8680659670164919,
      "grad_norm": 0.026424239288222608,
      "learning_rate": 2.6421097126839712e-06,
      "loss": 0.0143,
      "step": 1246
    },
    {
      "epoch": 1.8695652173913042,
      "grad_norm": 0.02958120459046709,
      "learning_rate": 2.582661303015066e-06,
      "loss": 0.0089,
      "step": 1247
    },
    {
      "epoch": 1.871064467766117,
      "grad_norm": 0.029911673211796365,
      "learning_rate": 2.5238805808778242e-06,
      "loss": 0.0098,
      "step": 1248
    },
    {
      "epoch": 1.8725637181409296,
      "grad_norm": 0.056981287060817974,
      "learning_rate": 2.465767949148734e-06,
      "loss": 0.0227,
      "step": 1249
    },
    {
      "epoch": 1.8740629685157422,
      "grad_norm": 0.0376261021688721,
      "learning_rate": 2.4083238061252567e-06,
      "loss": 0.0174,
      "step": 1250
    },
    {
      "epoch": 1.8755622188905547,
      "grad_norm": 0.034644589104120896,
      "learning_rate": 2.351548545523141e-06,
      "loss": 0.0131,
      "step": 1251
    },
    {
      "epoch": 1.8770614692653673,
      "grad_norm": 0.039399746567982966,
      "learning_rate": 2.295442556473637e-06,
      "loss": 0.0162,
      "step": 1252
    },
    {
      "epoch": 1.87856071964018,
      "grad_norm": 0.03207964462958592,
      "learning_rate": 2.240006223520941e-06,
      "loss": 0.0078,
      "step": 1253
    },
    {
      "epoch": 1.8800599700149925,
      "grad_norm": 0.05260614234208255,
      "learning_rate": 2.1852399266194314e-06,
      "loss": 0.0185,
      "step": 1254
    },
    {
      "epoch": 1.881559220389805,
      "grad_norm": 0.05106484042867738,
      "learning_rate": 2.1311440411312166e-06,
      "loss": 0.0226,
      "step": 1255
    },
    {
      "epoch": 1.8830584707646176,
      "grad_norm": 0.04263872143801973,
      "learning_rate": 2.0777189378234143e-06,
      "loss": 0.0117,
      "step": 1256
    },
    {
      "epoch": 1.8845577211394304,
      "grad_norm": 0.07846972233061164,
      "learning_rate": 2.024964982865729e-06,
      "loss": 0.0339,
      "step": 1257
    },
    {
      "epoch": 1.8860569715142428,
      "grad_norm": 0.029117608897231563,
      "learning_rate": 1.9728825378278246e-06,
      "loss": 0.0071,
      "step": 1258
    },
    {
      "epoch": 1.8875562218890556,
      "grad_norm": 0.0388167533672085,
      "learning_rate": 1.921471959676957e-06,
      "loss": 0.0123,
      "step": 1259
    },
    {
      "epoch": 1.889055472263868,
      "grad_norm": 0.04539482718528663,
      "learning_rate": 1.8707336007754873e-06,
      "loss": 0.0179,
      "step": 1260
    },
    {
      "epoch": 1.8905547226386807,
      "grad_norm": 0.045445265904142454,
      "learning_rate": 1.820667808878429e-06,
      "loss": 0.0219,
      "step": 1261
    },
    {
      "epoch": 1.892053973013493,
      "grad_norm": 0.028894890651339714,
      "learning_rate": 1.771274927131139e-06,
      "loss": 0.0145,
      "step": 1262
    },
    {
      "epoch": 1.893553223388306,
      "grad_norm": 0.042388630611865694,
      "learning_rate": 1.7225552940668632e-06,
      "loss": 0.0157,
      "step": 1263
    },
    {
      "epoch": 1.8950524737631185,
      "grad_norm": 0.06876839316091303,
      "learning_rate": 1.6745092436045494e-06,
      "loss": 0.0351,
      "step": 1264
    },
    {
      "epoch": 1.896551724137931,
      "grad_norm": 0.03573171364688593,
      "learning_rate": 1.6271371050464056e-06,
      "loss": 0.0237,
      "step": 1265
    },
    {
      "epoch": 1.8980509745127436,
      "grad_norm": 0.038482317398168285,
      "learning_rate": 1.580439203075812e-06,
      "loss": 0.0124,
      "step": 1266
    },
    {
      "epoch": 1.8995502248875562,
      "grad_norm": 0.05737954330837195,
      "learning_rate": 1.5344158577549229e-06,
      "loss": 0.0364,
      "step": 1267
    },
    {
      "epoch": 1.9010494752623688,
      "grad_norm": 0.04099948506323669,
      "learning_rate": 1.4890673845226133e-06,
      "loss": 0.0197,
      "step": 1268
    },
    {
      "epoch": 1.9025487256371814,
      "grad_norm": 0.03313456374887663,
      "learning_rate": 1.444394094192225e-06,
      "loss": 0.02,
      "step": 1269
    },
    {
      "epoch": 1.9040479760119942,
      "grad_norm": 0.043427692868520065,
      "learning_rate": 1.400396292949513e-06,
      "loss": 0.0202,
      "step": 1270
    },
    {
      "epoch": 1.9055472263868065,
      "grad_norm": 0.07327623247350938,
      "learning_rate": 1.3570742823504567e-06,
      "loss": 0.0245,
      "step": 1271
    },
    {
      "epoch": 1.9070464767616193,
      "grad_norm": 0.05525179025128676,
      "learning_rate": 1.3144283593192752e-06,
      "loss": 0.0246,
      "step": 1272
    },
    {
      "epoch": 1.9085457271364317,
      "grad_norm": 0.06295368734028475,
      "learning_rate": 1.272458816146349e-06,
      "loss": 0.0246,
      "step": 1273
    },
    {
      "epoch": 1.9100449775112445,
      "grad_norm": 0.034583926367543656,
      "learning_rate": 1.231165940486234e-06,
      "loss": 0.0155,
      "step": 1274
    },
    {
      "epoch": 1.9115442278860568,
      "grad_norm": 0.0437209002683007,
      "learning_rate": 1.1905500153556625e-06,
      "loss": 0.0162,
      "step": 1275
    },
    {
      "epoch": 1.9130434782608696,
      "grad_norm": 0.03904804270714857,
      "learning_rate": 1.1506113191316447e-06,
      "loss": 0.0154,
      "step": 1276
    },
    {
      "epoch": 1.9145427286356822,
      "grad_norm": 0.05420766292701498,
      "learning_rate": 1.1113501255495485e-06,
      "loss": 0.0195,
      "step": 1277
    },
    {
      "epoch": 1.9160419790104948,
      "grad_norm": 0.04566901515231431,
      "learning_rate": 1.0727667037011668e-06,
      "loss": 0.0125,
      "step": 1278
    },
    {
      "epoch": 1.9175412293853074,
      "grad_norm": 0.07198157977225833,
      "learning_rate": 1.0348613180329757e-06,
      "loss": 0.0219,
      "step": 1279
    },
    {
      "epoch": 1.91904047976012,
      "grad_norm": 0.032622544247452506,
      "learning_rate": 9.976342283442463e-07,
      "loss": 0.0085,
      "step": 1280
    },
    {
      "epoch": 1.9205397301349325,
      "grad_norm": 0.03912032612952811,
      "learning_rate": 9.610856897852682e-07,
      "loss": 0.0097,
      "step": 1281
    },
    {
      "epoch": 1.922038980509745,
      "grad_norm": 0.05117531076928059,
      "learning_rate": 9.252159528556403e-07,
      "loss": 0.0219,
      "step": 1282
    },
    {
      "epoch": 1.9235382308845579,
      "grad_norm": 0.048504658124269,
      "learning_rate": 8.900252634025274e-07,
      "loss": 0.0203,
      "step": 1283
    },
    {
      "epoch": 1.9250374812593702,
      "grad_norm": 0.06377251552840921,
      "learning_rate": 8.555138626189618e-07,
      "loss": 0.0235,
      "step": 1284
    },
    {
      "epoch": 1.926536731634183,
      "grad_norm": 0.0353549864848296,
      "learning_rate": 8.216819870422332e-07,
      "loss": 0.0141,
      "step": 1285
    },
    {
      "epoch": 1.9280359820089954,
      "grad_norm": 0.039798781616309425,
      "learning_rate": 7.885298685522235e-07,
      "loss": 0.0147,
      "step": 1286
    },
    {
      "epoch": 1.9295352323838082,
      "grad_norm": 0.03541629405787621,
      "learning_rate": 7.560577343698305e-07,
      "loss": 0.0102,
      "step": 1287
    },
    {
      "epoch": 1.9310344827586206,
      "grad_norm": 0.05561642747147643,
      "learning_rate": 7.242658070554464e-07,
      "loss": 0.0206,
      "step": 1288
    },
    {
      "epoch": 1.9325337331334334,
      "grad_norm": 0.03534178862057278,
      "learning_rate": 6.931543045073708e-07,
      "loss": 0.0095,
      "step": 1289
    },
    {
      "epoch": 1.934032983508246,
      "grad_norm": 0.04209596571556163,
      "learning_rate": 6.627234399603555e-07,
      "loss": 0.0245,
      "step": 1290
    },
    {
      "epoch": 1.9355322338830585,
      "grad_norm": 0.06513898157808579,
      "learning_rate": 6.32973421984151e-07,
      "loss": 0.026,
      "step": 1291
    },
    {
      "epoch": 1.937031484257871,
      "grad_norm": 0.04921099872743036,
      "learning_rate": 6.039044544820404e-07,
      "loss": 0.0194,
      "step": 1292
    },
    {
      "epoch": 1.9385307346326837,
      "grad_norm": 0.031094335152381827,
      "learning_rate": 5.755167366894632e-07,
      "loss": 0.0121,
      "step": 1293
    },
    {
      "epoch": 1.9400299850074962,
      "grad_norm": 0.050223040034524,
      "learning_rate": 5.478104631726711e-07,
      "loss": 0.0173,
      "step": 1294
    },
    {
      "epoch": 1.9415292353823088,
      "grad_norm": 0.03520172367952888,
      "learning_rate": 5.207858238273522e-07,
      "loss": 0.019,
      "step": 1295
    },
    {
      "epoch": 1.9430284857571214,
      "grad_norm": 0.054375401258943634,
      "learning_rate": 4.944430038773762e-07,
      "loss": 0.0335,
      "step": 1296
    },
    {
      "epoch": 1.944527736131934,
      "grad_norm": 0.04736400384851895,
      "learning_rate": 4.6878218387346183e-07,
      "loss": 0.0245,
      "step": 1297
    },
    {
      "epoch": 1.9460269865067468,
      "grad_norm": 0.05631107325781845,
      "learning_rate": 4.438035396920004e-07,
      "loss": 0.0173,
      "step": 1298
    },
    {
      "epoch": 1.9475262368815591,
      "grad_norm": 0.02312088515804648,
      "learning_rate": 4.1950724253383423e-07,
      "loss": 0.0098,
      "step": 1299
    },
    {
      "epoch": 1.949025487256372,
      "grad_norm": 0.06415814349625676,
      "learning_rate": 3.958934589230467e-07,
      "loss": 0.0137,
      "step": 1300
    },
    {
      "epoch": 1.9505247376311843,
      "grad_norm": 0.042944785560430666,
      "learning_rate": 3.7296235070587435e-07,
      "loss": 0.0129,
      "step": 1301
    },
    {
      "epoch": 1.952023988005997,
      "grad_norm": 0.03691120345897048,
      "learning_rate": 3.50714075049563e-07,
      "loss": 0.0165,
      "step": 1302
    },
    {
      "epoch": 1.9535232383808094,
      "grad_norm": 0.047643378112865786,
      "learning_rate": 3.2914878444131326e-07,
      "loss": 0.0191,
      "step": 1303
    },
    {
      "epoch": 1.9550224887556222,
      "grad_norm": 0.04866045710270131,
      "learning_rate": 3.0826662668720364e-07,
      "loss": 0.0211,
      "step": 1304
    },
    {
      "epoch": 1.9565217391304348,
      "grad_norm": 0.035743634358242155,
      "learning_rate": 2.8806774491120234e-07,
      "loss": 0.0188,
      "step": 1305
    },
    {
      "epoch": 1.9580209895052474,
      "grad_norm": 0.0499970530068033,
      "learning_rate": 2.685522775541904e-07,
      "loss": 0.0275,
      "step": 1306
    },
    {
      "epoch": 1.95952023988006,
      "grad_norm": 0.043717836582837635,
      "learning_rate": 2.497203583729957e-07,
      "loss": 0.0223,
      "step": 1307
    },
    {
      "epoch": 1.9610194902548725,
      "grad_norm": 0.06115844054156324,
      "learning_rate": 2.315721164394713e-07,
      "loss": 0.0259,
      "step": 1308
    },
    {
      "epoch": 1.9625187406296851,
      "grad_norm": 0.04674640262799417,
      "learning_rate": 2.141076761396521e-07,
      "loss": 0.0296,
      "step": 1309
    },
    {
      "epoch": 1.9640179910044977,
      "grad_norm": 0.04102373844741974,
      "learning_rate": 1.973271571728441e-07,
      "loss": 0.0215,
      "step": 1310
    },
    {
      "epoch": 1.9655172413793105,
      "grad_norm": 0.05296227342955121,
      "learning_rate": 1.8123067455084741e-07,
      "loss": 0.0193,
      "step": 1311
    },
    {
      "epoch": 1.9670164917541229,
      "grad_norm": 0.04607714250992321,
      "learning_rate": 1.6581833859716788e-07,
      "loss": 0.0233,
      "step": 1312
    },
    {
      "epoch": 1.9685157421289357,
      "grad_norm": 0.05305402469014974,
      "learning_rate": 1.5109025494620677e-07,
      "loss": 0.0186,
      "step": 1313
    },
    {
      "epoch": 1.970014992503748,
      "grad_norm": 0.027178810058074054,
      "learning_rate": 1.3704652454261668e-07,
      "loss": 0.008,
      "step": 1314
    },
    {
      "epoch": 1.9715142428785608,
      "grad_norm": 0.037694527414950885,
      "learning_rate": 1.2368724364053563e-07,
      "loss": 0.0095,
      "step": 1315
    },
    {
      "epoch": 1.9730134932533732,
      "grad_norm": 0.05169984692035686,
      "learning_rate": 1.1101250380300965e-07,
      "loss": 0.022,
      "step": 1316
    },
    {
      "epoch": 1.974512743628186,
      "grad_norm": 0.05203649072494641,
      "learning_rate": 9.902239190124896e-08,
      "loss": 0.0205,
      "step": 1317
    },
    {
      "epoch": 1.9760119940029985,
      "grad_norm": 0.05317970284385352,
      "learning_rate": 8.771699011416168e-08,
      "loss": 0.0305,
      "step": 1318
    },
    {
      "epoch": 1.9775112443778111,
      "grad_norm": 0.046052084092380735,
      "learning_rate": 7.709637592770991e-08,
      "loss": 0.02,
      "step": 1319
    },
    {
      "epoch": 1.9790104947526237,
      "grad_norm": 0.020664658138814455,
      "learning_rate": 6.71606221343768e-08,
      "loss": 0.0056,
      "step": 1320
    },
    {
      "epoch": 1.9805097451274363,
      "grad_norm": 0.04220144244776692,
      "learning_rate": 5.790979683271136e-08,
      "loss": 0.0151,
      "step": 1321
    },
    {
      "epoch": 1.9820089955022488,
      "grad_norm": 0.04084288034944265,
      "learning_rate": 4.934396342684e-08,
      "loss": 0.0142,
      "step": 1322
    },
    {
      "epoch": 1.9835082458770614,
      "grad_norm": 0.05769798222724036,
      "learning_rate": 4.146318062603349e-08,
      "loss": 0.0226,
      "step": 1323
    },
    {
      "epoch": 1.9850074962518742,
      "grad_norm": 0.1100081838811883,
      "learning_rate": 3.4267502444274015e-08,
      "loss": 0.0365,
      "step": 1324
    },
    {
      "epoch": 1.9865067466266866,
      "grad_norm": 0.08427099553965144,
      "learning_rate": 2.7756978199944273e-08,
      "loss": 0.0217,
      "step": 1325
    },
    {
      "epoch": 1.9880059970014994,
      "grad_norm": 0.040887716983706905,
      "learning_rate": 2.193165251545004e-08,
      "loss": 0.0154,
      "step": 1326
    },
    {
      "epoch": 1.9895052473763117,
      "grad_norm": 0.057352690068278346,
      "learning_rate": 1.6791565316920388e-08,
      "loss": 0.0193,
      "step": 1327
    },
    {
      "epoch": 1.9910044977511245,
      "grad_norm": 0.03068600118328415,
      "learning_rate": 1.2336751833941229e-08,
      "loss": 0.0175,
      "step": 1328
    },
    {
      "epoch": 1.992503748125937,
      "grad_norm": 0.04984601232484485,
      "learning_rate": 8.567242599299973e-09,
      "loss": 0.0183,
      "step": 1329
    },
    {
      "epoch": 1.9940029985007497,
      "grad_norm": 0.04524382063881194,
      "learning_rate": 5.483063448785686e-09,
      "loss": 0.0199,
      "step": 1330
    },
    {
      "epoch": 1.9955022488755623,
      "grad_norm": 0.047304184345328626,
      "learning_rate": 3.0842355210336515e-09,
      "loss": 0.0228,
      "step": 1331
    },
    {
      "epoch": 1.9970014992503748,
      "grad_norm": 0.057086267630760426,
      "learning_rate": 1.3707752573255405e-09,
      "loss": 0.0232,
      "step": 1332
    },
    {
      "epoch": 1.9985007496251874,
      "grad_norm": 0.037788074538081265,
      "learning_rate": 3.426944015227918e-10,
      "loss": 0.0204,
      "step": 1333
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.060579723830781834,
      "learning_rate": 0.0,
      "loss": 0.0344,
      "step": 1334
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.05502813309431076,
      "eval_runtime": 759.9841,
      "eval_samples_per_second": 13.653,
      "eval_steps_per_second": 0.854,
      "step": 1334
    },
    {
      "epoch": 2.0,
      "step": 1334,
      "total_flos": 2659138639233024.0,
      "train_loss": 0.03610908869839646,
      "train_runtime": 12784.2292,
      "train_samples_per_second": 3.339,
      "train_steps_per_second": 0.104
    }
  ],
  "logging_steps": 1,
  "max_steps": 1334,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2659138639233024.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}