{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.3504548900682334,
  "eval_steps": 500,
  "global_step": 7125,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003790750568612585,
      "grad_norm": 3.831867218017578,
      "learning_rate": 5.050505050505052e-07,
      "loss": 0.24115557968616486,
      "step": 2
    },
    {
      "epoch": 0.000758150113722517,
      "grad_norm": 2.999687433242798,
      "learning_rate": 1.5151515151515152e-06,
      "loss": 0.2288808822631836,
      "step": 4
    },
    {
      "epoch": 0.0011372251705837756,
      "grad_norm": 2.8589773178100586,
      "learning_rate": 2.5252525252525253e-06,
      "loss": 0.21898214519023895,
      "step": 6
    },
    {
      "epoch": 0.001516300227445034,
      "grad_norm": 2.9575161933898926,
      "learning_rate": 3.5353535353535352e-06,
      "loss": 0.2008015364408493,
      "step": 8
    },
    {
      "epoch": 0.0018953752843062926,
      "grad_norm": 2.7476305961608887,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.21072110533714294,
      "step": 10
    },
    {
      "epoch": 0.002274450341167551,
      "grad_norm": 3.1676814556121826,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.22474601864814758,
      "step": 12
    },
    {
      "epoch": 0.00265352539802881,
      "grad_norm": 3.457512855529785,
      "learning_rate": 6.565656565656567e-06,
      "loss": 0.22432611882686615,
      "step": 14
    },
    {
      "epoch": 0.003032600454890068,
      "grad_norm": 3.278862714767456,
      "learning_rate": 7.5757575757575764e-06,
      "loss": 0.2039806842803955,
      "step": 16
    },
    {
      "epoch": 0.003411675511751327,
      "grad_norm": 2.994616985321045,
      "learning_rate": 8.585858585858587e-06,
      "loss": 0.22946196794509888,
      "step": 18
    },
    {
      "epoch": 0.0037907505686125853,
      "grad_norm": 3.4709763526916504,
      "learning_rate": 9.595959595959595e-06,
      "loss": 0.19840861856937408,
      "step": 20
    },
    {
      "epoch": 0.004169825625473844,
      "grad_norm": 3.2057032585144043,
      "learning_rate": 1.0606060606060607e-05,
      "loss": 0.22237884998321533,
      "step": 22
    },
    {
      "epoch": 0.004548900682335102,
      "grad_norm": 3.481503486633301,
      "learning_rate": 1.1616161616161616e-05,
      "loss": 0.2224944829940796,
      "step": 24
    },
    {
      "epoch": 0.004927975739196361,
      "grad_norm": 3.329615831375122,
      "learning_rate": 1.2626262626262628e-05,
      "loss": 0.21955694258213043,
      "step": 26
    },
    {
      "epoch": 0.00530705079605762,
      "grad_norm": 4.500723838806152,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 0.22555610537528992,
      "step": 28
    },
    {
      "epoch": 0.005686125852918878,
      "grad_norm": 2.477372884750366,
      "learning_rate": 1.4646464646464647e-05,
      "loss": 0.20175020396709442,
      "step": 30
    },
    {
      "epoch": 0.006065200909780136,
      "grad_norm": 3.3791396617889404,
      "learning_rate": 1.565656565656566e-05,
      "loss": 0.23708373308181763,
      "step": 32
    },
    {
      "epoch": 0.006444275966641395,
      "grad_norm": 2.715676784515381,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.22042585909366608,
      "step": 34
    },
    {
      "epoch": 0.006823351023502654,
      "grad_norm": 3.085118532180786,
      "learning_rate": 1.7676767676767676e-05,
      "loss": 0.20185081660747528,
      "step": 36
    },
    {
      "epoch": 0.007202426080363912,
      "grad_norm": 2.5242998600006104,
      "learning_rate": 1.8686868686868688e-05,
      "loss": 0.21545518934726715,
      "step": 38
    },
    {
      "epoch": 0.0075815011372251705,
      "grad_norm": 2.4197964668273926,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 0.2256695032119751,
      "step": 40
    },
    {
      "epoch": 0.00796057619408643,
      "grad_norm": 2.668548583984375,
      "learning_rate": 2.070707070707071e-05,
      "loss": 0.20934747159481049,
      "step": 42
    },
    {
      "epoch": 0.008339651250947688,
      "grad_norm": 2.519810199737549,
      "learning_rate": 2.171717171717172e-05,
      "loss": 0.2023160457611084,
      "step": 44
    },
    {
      "epoch": 0.008718726307808946,
      "grad_norm": 1.8102171421051025,
      "learning_rate": 2.272727272727273e-05,
      "loss": 0.1914549022912979,
      "step": 46
    },
    {
      "epoch": 0.009097801364670205,
      "grad_norm": 2.6659693717956543,
      "learning_rate": 2.3737373737373738e-05,
      "loss": 0.2156388759613037,
      "step": 48
    },
    {
      "epoch": 0.009476876421531463,
      "grad_norm": 2.638444423675537,
      "learning_rate": 2.474747474747475e-05,
      "loss": 0.2040487825870514,
      "step": 50
    },
    {
      "epoch": 0.009855951478392721,
      "grad_norm": 2.5430164337158203,
      "learning_rate": 2.575757575757576e-05,
      "loss": 0.1811923384666443,
      "step": 52
    },
    {
      "epoch": 0.01023502653525398,
      "grad_norm": 2.096179246902466,
      "learning_rate": 2.676767676767677e-05,
      "loss": 0.19041013717651367,
      "step": 54
    },
    {
      "epoch": 0.01061410159211524,
      "grad_norm": 2.39115047454834,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.2149028778076172,
      "step": 56
    },
    {
      "epoch": 0.010993176648976498,
      "grad_norm": 2.0140838623046875,
      "learning_rate": 2.878787878787879e-05,
      "loss": 0.20050320029258728,
      "step": 58
    },
    {
      "epoch": 0.011372251705837756,
      "grad_norm": 2.796888828277588,
      "learning_rate": 2.9797979797979796e-05,
      "loss": 0.19733527302742004,
      "step": 60
    },
    {
      "epoch": 0.011751326762699014,
      "grad_norm": 2.1936731338500977,
      "learning_rate": 3.080808080808081e-05,
      "loss": 0.19901502132415771,
      "step": 62
    },
    {
      "epoch": 0.012130401819560273,
      "grad_norm": 2.0025813579559326,
      "learning_rate": 3.181818181818182e-05,
      "loss": 0.19021570682525635,
      "step": 64
    },
    {
      "epoch": 0.012509476876421531,
      "grad_norm": 1.7600773572921753,
      "learning_rate": 3.282828282828283e-05,
      "loss": 0.1594085991382599,
      "step": 66
    },
    {
      "epoch": 0.01288855193328279,
      "grad_norm": 1.4796334505081177,
      "learning_rate": 3.3838383838383844e-05,
      "loss": 0.18020717799663544,
      "step": 68
    },
    {
      "epoch": 0.013267626990144048,
      "grad_norm": 1.9786773920059204,
      "learning_rate": 3.484848484848485e-05,
      "loss": 0.1799086481332779,
      "step": 70
    },
    {
      "epoch": 0.013646702047005308,
      "grad_norm": 1.7947726249694824,
      "learning_rate": 3.5858585858585855e-05,
      "loss": 0.15451493859291077,
      "step": 72
    },
    {
      "epoch": 0.014025777103866566,
      "grad_norm": 1.534415602684021,
      "learning_rate": 3.686868686868687e-05,
      "loss": 0.16322581470012665,
      "step": 74
    },
    {
      "epoch": 0.014404852160727824,
      "grad_norm": 2.0888843536376953,
      "learning_rate": 3.787878787878788e-05,
      "loss": 0.18872690200805664,
      "step": 76
    },
    {
      "epoch": 0.014783927217589083,
      "grad_norm": 1.7359966039657593,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.17405053973197937,
      "step": 78
    },
    {
      "epoch": 0.015163002274450341,
      "grad_norm": 1.621677279472351,
      "learning_rate": 3.98989898989899e-05,
      "loss": 0.17236940562725067,
      "step": 80
    },
    {
      "epoch": 0.0155420773313116,
      "grad_norm": 1.6410354375839233,
      "learning_rate": 4.0909090909090915e-05,
      "loss": 0.15769261121749878,
      "step": 82
    },
    {
      "epoch": 0.01592115238817286,
      "grad_norm": 2.0459330081939697,
      "learning_rate": 4.191919191919192e-05,
      "loss": 0.16512681543827057,
      "step": 84
    },
    {
      "epoch": 0.016300227445034118,
      "grad_norm": 1.849062442779541,
      "learning_rate": 4.292929292929293e-05,
      "loss": 0.15917335450649261,
      "step": 86
    },
    {
      "epoch": 0.016679302501895376,
      "grad_norm": 1.5138572454452515,
      "learning_rate": 4.3939393939393944e-05,
      "loss": 0.16564102470874786,
      "step": 88
    },
    {
      "epoch": 0.017058377558756634,
      "grad_norm": 1.7878366708755493,
      "learning_rate": 4.494949494949495e-05,
      "loss": 0.15585318207740784,
      "step": 90
    },
    {
      "epoch": 0.017437452615617893,
      "grad_norm": 1.5637410879135132,
      "learning_rate": 4.595959595959596e-05,
      "loss": 0.15551325678825378,
      "step": 92
    },
    {
      "epoch": 0.01781652767247915,
      "grad_norm": 1.418735146522522,
      "learning_rate": 4.696969696969697e-05,
      "loss": 0.16084995865821838,
      "step": 94
    },
    {
      "epoch": 0.01819560272934041,
      "grad_norm": 1.3928037881851196,
      "learning_rate": 4.797979797979798e-05,
      "loss": 0.13635674118995667,
      "step": 96
    },
    {
      "epoch": 0.018574677786201668,
      "grad_norm": 1.8615429401397705,
      "learning_rate": 4.898989898989899e-05,
      "loss": 0.1436282843351364,
      "step": 98
    },
    {
      "epoch": 0.018953752843062926,
      "grad_norm": 1.7022007703781128,
      "learning_rate": 5e-05,
      "loss": 0.14871130883693695,
      "step": 100
    },
    {
      "epoch": 0.019332827899924184,
      "grad_norm": 1.7174115180969238,
      "learning_rate": 5.101010101010101e-05,
      "loss": 0.14212347567081451,
      "step": 102
    },
    {
      "epoch": 0.019711902956785442,
      "grad_norm": 1.759929895401001,
      "learning_rate": 5.2020202020202026e-05,
      "loss": 0.1418125331401825,
      "step": 104
    },
    {
      "epoch": 0.0200909780136467,
      "grad_norm": 1.3723875284194946,
      "learning_rate": 5.303030303030303e-05,
      "loss": 0.12501539289951324,
      "step": 106
    },
    {
      "epoch": 0.02047005307050796,
      "grad_norm": 1.260796308517456,
      "learning_rate": 5.4040404040404044e-05,
      "loss": 0.12885957956314087,
      "step": 108
    },
    {
      "epoch": 0.02084912812736922,
      "grad_norm": 1.3781832456588745,
      "learning_rate": 5.5050505050505056e-05,
      "loss": 0.14464299380779266,
      "step": 110
    },
    {
      "epoch": 0.02122820318423048,
      "grad_norm": 1.1771231889724731,
      "learning_rate": 5.606060606060606e-05,
      "loss": 0.12694981694221497,
      "step": 112
    },
    {
      "epoch": 0.021607278241091737,
      "grad_norm": 1.1477068662643433,
      "learning_rate": 5.707070707070707e-05,
      "loss": 0.1178375855088234,
      "step": 114
    },
    {
      "epoch": 0.021986353297952996,
      "grad_norm": 1.8767154216766357,
      "learning_rate": 5.808080808080808e-05,
      "loss": 0.12230765074491501,
      "step": 116
    },
    {
      "epoch": 0.022365428354814254,
      "grad_norm": 1.4404174089431763,
      "learning_rate": 5.90909090909091e-05,
      "loss": 0.1126122921705246,
      "step": 118
    },
    {
      "epoch": 0.022744503411675512,
      "grad_norm": 1.161832571029663,
      "learning_rate": 6.01010101010101e-05,
      "loss": 0.12581869959831238,
      "step": 120
    },
    {
      "epoch": 0.02312357846853677,
      "grad_norm": 1.077945351600647,
      "learning_rate": 6.111111111111112e-05,
      "loss": 0.09817113727331161,
      "step": 122
    },
    {
      "epoch": 0.02350265352539803,
      "grad_norm": 1.2167441844940186,
      "learning_rate": 6.212121212121213e-05,
      "loss": 0.11549479514360428,
      "step": 124
    },
    {
      "epoch": 0.023881728582259287,
      "grad_norm": 1.0958034992218018,
      "learning_rate": 6.313131313131313e-05,
      "loss": 0.12976986169815063,
      "step": 126
    },
    {
      "epoch": 0.024260803639120546,
      "grad_norm": 0.9334296584129333,
      "learning_rate": 6.414141414141415e-05,
      "loss": 0.10869947075843811,
      "step": 128
    },
    {
      "epoch": 0.024639878695981804,
      "grad_norm": 1.0688090324401855,
      "learning_rate": 6.515151515151516e-05,
      "loss": 0.09889855980873108,
      "step": 130
    },
    {
      "epoch": 0.025018953752843062,
      "grad_norm": 1.3466724157333374,
      "learning_rate": 6.616161616161617e-05,
      "loss": 0.10494759678840637,
      "step": 132
    },
    {
      "epoch": 0.02539802880970432,
      "grad_norm": 1.1844414472579956,
      "learning_rate": 6.717171717171718e-05,
      "loss": 0.09326791763305664,
      "step": 134
    },
    {
      "epoch": 0.02577710386656558,
      "grad_norm": 1.1364573240280151,
      "learning_rate": 6.818181818181818e-05,
      "loss": 0.08711981028318405,
      "step": 136
    },
    {
      "epoch": 0.026156178923426837,
      "grad_norm": 1.0187207460403442,
      "learning_rate": 6.91919191919192e-05,
      "loss": 0.09687816351652145,
      "step": 138
    },
    {
      "epoch": 0.026535253980288095,
      "grad_norm": 1.1344724893569946,
      "learning_rate": 7.020202020202021e-05,
      "loss": 0.0869758278131485,
      "step": 140
    },
    {
      "epoch": 0.026914329037149357,
      "grad_norm": 1.1279665231704712,
      "learning_rate": 7.121212121212121e-05,
      "loss": 0.11362449824810028,
      "step": 142
    },
    {
      "epoch": 0.027293404094010616,
      "grad_norm": 0.9359419345855713,
      "learning_rate": 7.222222222222222e-05,
      "loss": 0.08251458406448364,
      "step": 144
    },
    {
      "epoch": 0.027672479150871874,
      "grad_norm": 0.9044331312179565,
      "learning_rate": 7.323232323232324e-05,
      "loss": 0.07683239132165909,
      "step": 146
    },
    {
      "epoch": 0.028051554207733132,
      "grad_norm": 0.8708617091178894,
      "learning_rate": 7.424242424242424e-05,
      "loss": 0.06908036023378372,
      "step": 148
    },
    {
      "epoch": 0.02843062926459439,
      "grad_norm": 1.1326483488082886,
      "learning_rate": 7.525252525252525e-05,
      "loss": 0.0962204560637474,
      "step": 150
    },
    {
      "epoch": 0.02880970432145565,
      "grad_norm": 1.1926954984664917,
      "learning_rate": 7.626262626262627e-05,
      "loss": 0.07143861800432205,
      "step": 152
    },
    {
      "epoch": 0.029188779378316907,
      "grad_norm": 1.2476686239242554,
      "learning_rate": 7.727272727272727e-05,
      "loss": 0.08484694361686707,
      "step": 154
    },
    {
      "epoch": 0.029567854435178165,
      "grad_norm": 1.1337658166885376,
      "learning_rate": 7.828282828282829e-05,
      "loss": 0.08729302883148193,
      "step": 156
    },
    {
      "epoch": 0.029946929492039424,
      "grad_norm": 1.4720388650894165,
      "learning_rate": 7.92929292929293e-05,
      "loss": 0.08332853764295578,
      "step": 158
    },
    {
      "epoch": 0.030326004548900682,
      "grad_norm": 0.9728957414627075,
      "learning_rate": 8.03030303030303e-05,
      "loss": 0.07362374663352966,
      "step": 160
    },
    {
      "epoch": 0.03070507960576194,
      "grad_norm": 1.6131110191345215,
      "learning_rate": 8.131313131313132e-05,
      "loss": 0.07593365013599396,
      "step": 162
    },
    {
      "epoch": 0.0310841546626232,
      "grad_norm": 1.1353189945220947,
      "learning_rate": 8.232323232323233e-05,
      "loss": 0.08282628655433655,
      "step": 164
    },
    {
      "epoch": 0.03146322971948446,
      "grad_norm": 1.2331316471099854,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.06936199218034744,
      "step": 166
    },
    {
      "epoch": 0.03184230477634572,
      "grad_norm": 1.0283564329147339,
      "learning_rate": 8.434343434343435e-05,
      "loss": 0.06914687156677246,
      "step": 168
    },
    {
      "epoch": 0.03222137983320698,
      "grad_norm": 1.0510566234588623,
      "learning_rate": 8.535353535353535e-05,
      "loss": 0.07401993870735168,
      "step": 170
    },
    {
      "epoch": 0.032600454890068235,
      "grad_norm": 0.8792195916175842,
      "learning_rate": 8.636363636363637e-05,
      "loss": 0.055682070553302765,
      "step": 172
    },
    {
      "epoch": 0.032979529946929494,
      "grad_norm": 1.3492839336395264,
      "learning_rate": 8.737373737373738e-05,
      "loss": 0.07071048766374588,
      "step": 174
    },
    {
      "epoch": 0.03335860500379075,
      "grad_norm": 1.0557173490524292,
      "learning_rate": 8.83838383838384e-05,
      "loss": 0.050383713096380234,
      "step": 176
    },
    {
      "epoch": 0.03373768006065201,
      "grad_norm": 1.3090928792953491,
      "learning_rate": 8.93939393939394e-05,
      "loss": 0.058576375246047974,
      "step": 178
    },
    {
      "epoch": 0.03411675511751327,
      "grad_norm": 1.1909061670303345,
      "learning_rate": 9.040404040404041e-05,
      "loss": 0.07507508993148804,
      "step": 180
    },
    {
      "epoch": 0.03449583017437453,
      "grad_norm": 0.9435303211212158,
      "learning_rate": 9.141414141414141e-05,
      "loss": 0.06046615540981293,
      "step": 182
    },
    {
      "epoch": 0.034874905231235785,
      "grad_norm": 1.0906531810760498,
      "learning_rate": 9.242424242424242e-05,
      "loss": 0.07550626993179321,
      "step": 184
    },
    {
      "epoch": 0.035253980288097043,
      "grad_norm": 1.089685082435608,
      "learning_rate": 9.343434343434344e-05,
      "loss": 0.056082721799612045,
      "step": 186
    },
    {
      "epoch": 0.0356330553449583,
      "grad_norm": 0.9265295267105103,
      "learning_rate": 9.444444444444444e-05,
      "loss": 0.05575732886791229,
      "step": 188
    },
    {
      "epoch": 0.03601213040181956,
      "grad_norm": 1.4318554401397705,
      "learning_rate": 9.545454545454546e-05,
      "loss": 0.07835233211517334,
      "step": 190
    },
    {
      "epoch": 0.03639120545868082,
      "grad_norm": 1.0656442642211914,
      "learning_rate": 9.646464646464647e-05,
      "loss": 0.04160471633076668,
      "step": 192
    },
    {
      "epoch": 0.03677028051554208,
      "grad_norm": 0.951963484287262,
      "learning_rate": 9.747474747474747e-05,
      "loss": 0.044328123331069946,
      "step": 194
    },
    {
      "epoch": 0.037149355572403335,
      "grad_norm": 0.9746479988098145,
      "learning_rate": 9.848484848484849e-05,
      "loss": 0.04572984576225281,
      "step": 196
    },
    {
      "epoch": 0.03752843062926459,
      "grad_norm": 0.9184594750404358,
      "learning_rate": 9.94949494949495e-05,
      "loss": 0.045363590121269226,
      "step": 198
    },
    {
      "epoch": 0.03790750568612585,
      "grad_norm": 1.3963510990142822,
      "learning_rate": 0.0001005050505050505,
      "loss": 0.05668392404913902,
      "step": 200
    },
    {
      "epoch": 0.03828658074298711,
      "grad_norm": 0.8344203233718872,
      "learning_rate": 0.00010151515151515152,
      "loss": 0.03731667995452881,
      "step": 202
    },
    {
      "epoch": 0.03866565579984837,
      "grad_norm": 1.4892925024032593,
      "learning_rate": 0.00010252525252525254,
      "loss": 0.050027329474687576,
      "step": 204
    },
    {
      "epoch": 0.03904473085670963,
      "grad_norm": 1.0475103855133057,
      "learning_rate": 0.00010353535353535353,
      "loss": 0.034702885895967484,
      "step": 206
    },
    {
      "epoch": 0.039423805913570885,
      "grad_norm": 1.1677299737930298,
      "learning_rate": 0.00010454545454545455,
      "loss": 0.05555498227477074,
      "step": 208
    },
    {
      "epoch": 0.03980288097043214,
      "grad_norm": 0.7712841629981995,
      "learning_rate": 0.00010555555555555557,
      "loss": 0.04197040572762489,
      "step": 210
    },
    {
      "epoch": 0.0401819560272934,
      "grad_norm": 1.1144565343856812,
      "learning_rate": 0.00010656565656565659,
      "loss": 0.039502520114183426,
      "step": 212
    },
    {
      "epoch": 0.04056103108415466,
      "grad_norm": 1.1026651859283447,
      "learning_rate": 0.00010757575757575758,
      "loss": 0.03712320700287819,
      "step": 214
    },
    {
      "epoch": 0.04094010614101592,
      "grad_norm": 0.9852290749549866,
      "learning_rate": 0.0001085858585858586,
      "loss": 0.06103947386145592,
      "step": 216
    },
    {
      "epoch": 0.041319181197877176,
      "grad_norm": 1.064296007156372,
      "learning_rate": 0.0001095959595959596,
      "loss": 0.05159662663936615,
      "step": 218
    },
    {
      "epoch": 0.04169825625473844,
      "grad_norm": 1.2568323612213135,
      "learning_rate": 0.00011060606060606061,
      "loss": 0.03378737345337868,
      "step": 220
    },
    {
      "epoch": 0.0420773313115997,
      "grad_norm": 0.9639132618904114,
      "learning_rate": 0.00011161616161616161,
      "loss": 0.0256947074085474,
      "step": 222
    },
    {
      "epoch": 0.04245640636846096,
      "grad_norm": 0.8631940484046936,
      "learning_rate": 0.00011262626262626263,
      "loss": 0.025856733322143555,
      "step": 224
    },
    {
      "epoch": 0.04283548142532222,
      "grad_norm": 0.7811000943183899,
      "learning_rate": 0.00011363636363636365,
      "loss": 0.03477013483643532,
      "step": 226
    },
    {
      "epoch": 0.043214556482183475,
      "grad_norm": 0.7270809412002563,
      "learning_rate": 0.00011464646464646464,
      "loss": 0.024124350398778915,
      "step": 228
    },
    {
      "epoch": 0.04359363153904473,
      "grad_norm": 0.892697811126709,
      "learning_rate": 0.00011565656565656566,
      "loss": 0.05322834849357605,
      "step": 230
    },
    {
      "epoch": 0.04397270659590599,
      "grad_norm": 0.7388175129890442,
      "learning_rate": 0.00011666666666666668,
      "loss": 0.02327428199350834,
      "step": 232
    },
    {
      "epoch": 0.04435178165276725,
      "grad_norm": 0.726128876209259,
      "learning_rate": 0.00011767676767676767,
      "loss": 0.023403741419315338,
      "step": 234
    },
    {
      "epoch": 0.04473085670962851,
      "grad_norm": 1.0080806016921997,
      "learning_rate": 0.00011868686868686869,
      "loss": 0.04749135300517082,
      "step": 236
    },
    {
      "epoch": 0.045109931766489766,
      "grad_norm": 0.7969107627868652,
      "learning_rate": 0.00011969696969696971,
      "loss": 0.046364620327949524,
      "step": 238
    },
    {
      "epoch": 0.045489006823351025,
      "grad_norm": 0.758381187915802,
      "learning_rate": 0.0001207070707070707,
      "loss": 0.04588283225893974,
      "step": 240
    },
    {
      "epoch": 0.04586808188021228,
      "grad_norm": 0.6660889387130737,
      "learning_rate": 0.00012171717171717172,
      "loss": 0.028649073094129562,
      "step": 242
    },
    {
      "epoch": 0.04624715693707354,
      "grad_norm": 0.6515665650367737,
      "learning_rate": 0.00012272727272727272,
      "loss": 0.020845184102654457,
      "step": 244
    },
    {
      "epoch": 0.0466262319939348,
      "grad_norm": 0.659670352935791,
      "learning_rate": 0.00012373737373737374,
      "loss": 0.023756979033350945,
      "step": 246
    },
    {
      "epoch": 0.04700530705079606,
      "grad_norm": 1.0604625940322876,
      "learning_rate": 0.00012474747474747473,
      "loss": 0.025589866563677788,
      "step": 248
    },
    {
      "epoch": 0.047384382107657316,
      "grad_norm": 0.6442307233810425,
      "learning_rate": 0.00012575757575757575,
      "loss": 0.027212204411625862,
      "step": 250
    },
    {
      "epoch": 0.047763457164518575,
      "grad_norm": 0.5536343455314636,
      "learning_rate": 0.00012676767676767677,
      "loss": 0.03487740457057953,
      "step": 252
    },
    {
      "epoch": 0.04814253222137983,
      "grad_norm": 0.7550069093704224,
      "learning_rate": 0.00012777777777777776,
      "loss": 0.029325518757104874,
      "step": 254
    },
    {
      "epoch": 0.04852160727824109,
      "grad_norm": 0.628350019454956,
      "learning_rate": 0.00012878787878787878,
      "loss": 0.045246679335832596,
      "step": 256
    },
    {
      "epoch": 0.04890068233510235,
      "grad_norm": 1.3373678922653198,
      "learning_rate": 0.0001297979797979798,
      "loss": 0.03796561807394028,
      "step": 258
    },
    {
      "epoch": 0.04927975739196361,
      "grad_norm": 0.7238209247589111,
      "learning_rate": 0.00013080808080808082,
      "loss": 0.028048960492014885,
      "step": 260
    },
    {
      "epoch": 0.049658832448824866,
      "grad_norm": 0.7324119210243225,
      "learning_rate": 0.0001318181818181818,
      "loss": 0.021303502842783928,
      "step": 262
    },
    {
      "epoch": 0.050037907505686124,
      "grad_norm": 0.6468061804771423,
      "learning_rate": 0.00013282828282828283,
      "loss": 0.03246187046170235,
      "step": 264
    },
    {
      "epoch": 0.05041698256254738,
      "grad_norm": 0.6783881187438965,
      "learning_rate": 0.00013383838383838385,
      "loss": 0.029054652899503708,
      "step": 266
    },
    {
      "epoch": 0.05079605761940864,
      "grad_norm": 0.7682046890258789,
      "learning_rate": 0.00013484848484848484,
      "loss": 0.044934988021850586,
      "step": 268
    },
    {
      "epoch": 0.0511751326762699,
      "grad_norm": 1.6431066989898682,
      "learning_rate": 0.00013585858585858586,
      "loss": 0.023320386186242104,
      "step": 270
    },
    {
      "epoch": 0.05155420773313116,
      "grad_norm": 0.6246185302734375,
      "learning_rate": 0.00013686868686868688,
      "loss": 0.027497459203004837,
      "step": 272
    },
    {
      "epoch": 0.051933282789992416,
      "grad_norm": 0.4422309696674347,
      "learning_rate": 0.0001378787878787879,
      "loss": 0.01773752085864544,
      "step": 274
    },
    {
      "epoch": 0.052312357846853674,
      "grad_norm": 0.7280117869377136,
      "learning_rate": 0.0001388888888888889,
      "loss": 0.0557699017226696,
      "step": 276
    },
    {
      "epoch": 0.05269143290371493,
      "grad_norm": 0.7154707312583923,
      "learning_rate": 0.0001398989898989899,
      "loss": 0.034791167825460434,
      "step": 278
    },
    {
      "epoch": 0.05307050796057619,
      "grad_norm": 0.5650017857551575,
      "learning_rate": 0.00014090909090909093,
      "loss": 0.03123454377055168,
      "step": 280
    },
    {
      "epoch": 0.05344958301743745,
      "grad_norm": 0.7572164535522461,
      "learning_rate": 0.00014191919191919192,
      "loss": 0.031812675297260284,
      "step": 282
    },
    {
      "epoch": 0.053828658074298714,
      "grad_norm": 0.697638988494873,
      "learning_rate": 0.00014292929292929294,
      "loss": 0.05757517367601395,
      "step": 284
    },
    {
      "epoch": 0.05420773313115997,
      "grad_norm": 0.48484864830970764,
      "learning_rate": 0.00014393939393939396,
      "loss": 0.03307062387466431,
      "step": 286
    },
    {
      "epoch": 0.05458680818802123,
      "grad_norm": 0.7399836182594299,
      "learning_rate": 0.00014494949494949495,
      "loss": 0.02571089379489422,
      "step": 288
    },
    {
      "epoch": 0.05496588324488249,
      "grad_norm": 0.5058830380439758,
      "learning_rate": 0.00014595959595959597,
      "loss": 0.04124552756547928,
      "step": 290
    },
    {
      "epoch": 0.05534495830174375,
      "grad_norm": 0.5523101091384888,
      "learning_rate": 0.00014696969696969698,
      "loss": 0.030898327007889748,
      "step": 292
    },
    {
      "epoch": 0.055724033358605006,
      "grad_norm": 0.6221122741699219,
      "learning_rate": 0.000147979797979798,
      "loss": 0.029636915773153305,
      "step": 294
    },
    {
      "epoch": 0.056103108415466264,
      "grad_norm": 0.6969261169433594,
      "learning_rate": 0.000148989898989899,
      "loss": 0.03281944617629051,
      "step": 296
    },
    {
      "epoch": 0.05648218347232752,
      "grad_norm": 0.45685869455337524,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.03504050150513649,
      "step": 298
    },
    {
      "epoch": 0.05686125852918878,
      "grad_norm": 0.4080333411693573,
      "learning_rate": 0.00015101010101010103,
      "loss": 0.026671521365642548,
      "step": 300
    },
    {
      "epoch": 0.05724033358605004,
      "grad_norm": 0.5489761829376221,
      "learning_rate": 0.00015202020202020202,
      "loss": 0.036205630749464035,
      "step": 302
    },
    {
      "epoch": 0.0576194086429113,
      "grad_norm": 0.8397519588470459,
      "learning_rate": 0.00015303030303030304,
      "loss": 0.0514635369181633,
      "step": 304
    },
    {
      "epoch": 0.057998483699772556,
      "grad_norm": 0.5239524245262146,
      "learning_rate": 0.00015404040404040406,
      "loss": 0.03308585658669472,
      "step": 306
    },
    {
      "epoch": 0.058377558756633814,
      "grad_norm": 0.3567987084388733,
      "learning_rate": 0.00015505050505050508,
      "loss": 0.017442354932427406,
      "step": 308
    },
    {
      "epoch": 0.05875663381349507,
      "grad_norm": 0.6197696328163147,
      "learning_rate": 0.00015606060606060607,
      "loss": 0.033882804214954376,
      "step": 310
    },
    {
      "epoch": 0.05913570887035633,
      "grad_norm": 0.7303937077522278,
      "learning_rate": 0.0001570707070707071,
      "loss": 0.032125506550073624,
      "step": 312
    },
    {
      "epoch": 0.05951478392721759,
      "grad_norm": 0.4686358869075775,
      "learning_rate": 0.00015808080808080808,
      "loss": 0.017301857471466064,
      "step": 314
    },
    {
      "epoch": 0.05989385898407885,
      "grad_norm": 0.6178193092346191,
      "learning_rate": 0.0001590909090909091,
      "loss": 0.03314616158604622,
      "step": 316
    },
    {
      "epoch": 0.060272934040940106,
      "grad_norm": 0.34282252192497253,
      "learning_rate": 0.00016010101010101012,
      "loss": 0.02810189500451088,
      "step": 318
    },
    {
      "epoch": 0.060652009097801364,
      "grad_norm": 0.5432535409927368,
      "learning_rate": 0.0001611111111111111,
      "loss": 0.021061308681964874,
      "step": 320
    },
    {
      "epoch": 0.06103108415466262,
      "grad_norm": 0.5055902600288391,
      "learning_rate": 0.00016212121212121213,
      "loss": 0.041348859667778015,
      "step": 322
    },
    {
      "epoch": 0.06141015921152388,
      "grad_norm": 0.5389487147331238,
      "learning_rate": 0.00016313131313131312,
      "loss": 0.02693920023739338,
      "step": 324
    },
    {
      "epoch": 0.06178923426838514,
      "grad_norm": 0.48516300320625305,
      "learning_rate": 0.00016414141414141414,
      "loss": 0.04365244880318642,
      "step": 326
    },
    {
      "epoch": 0.0621683093252464,
      "grad_norm": 0.6730494499206543,
      "learning_rate": 0.00016515151515151516,
      "loss": 0.06355004757642746,
      "step": 328
    },
    {
      "epoch": 0.06254738438210766,
      "grad_norm": 0.3694400191307068,
      "learning_rate": 0.00016616161616161615,
      "loss": 0.021104227751493454,
      "step": 330
    },
    {
      "epoch": 0.06292645943896892,
      "grad_norm": 0.45879530906677246,
      "learning_rate": 0.00016717171717171717,
      "loss": 0.02241661213338375,
      "step": 332
    },
    {
      "epoch": 0.06330553449583018,
      "grad_norm": 0.5688877701759338,
      "learning_rate": 0.0001681818181818182,
      "loss": 0.020102601498365402,
      "step": 334
    },
    {
      "epoch": 0.06368460955269144,
      "grad_norm": 1.1296149492263794,
      "learning_rate": 0.00016919191919191918,
      "loss": 0.024005958810448647,
      "step": 336
    },
    {
      "epoch": 0.0640636846095527,
      "grad_norm": 0.5033518075942993,
      "learning_rate": 0.0001702020202020202,
      "loss": 0.02501414343714714,
      "step": 338
    },
    {
      "epoch": 0.06444275966641395,
      "grad_norm": 0.2962978780269623,
      "learning_rate": 0.00017121212121212122,
      "loss": 0.029946187511086464,
      "step": 340
    },
    {
      "epoch": 0.06482183472327521,
      "grad_norm": 0.5404835939407349,
      "learning_rate": 0.00017222222222222224,
      "loss": 0.03533448651432991,
      "step": 342
    },
    {
      "epoch": 0.06520090978013647,
      "grad_norm": 0.4712458550930023,
      "learning_rate": 0.00017323232323232323,
      "loss": 0.035269349813461304,
      "step": 344
    },
    {
      "epoch": 0.06557998483699773,
      "grad_norm": 0.4501020610332489,
      "learning_rate": 0.00017424242424242425,
      "loss": 0.01828707568347454,
      "step": 346
    },
    {
      "epoch": 0.06595905989385899,
      "grad_norm": 0.6508148908615112,
      "learning_rate": 0.00017525252525252527,
      "loss": 0.03412401303648949,
      "step": 348
    },
    {
      "epoch": 0.06633813495072025,
      "grad_norm": 0.640232264995575,
      "learning_rate": 0.00017626262626262626,
      "loss": 0.04506737366318703,
      "step": 350
    },
    {
      "epoch": 0.0667172100075815,
      "grad_norm": 0.6278156042098999,
      "learning_rate": 0.00017727272727272728,
      "loss": 0.05109577253460884,
      "step": 352
    },
    {
      "epoch": 0.06709628506444276,
      "grad_norm": 0.6648157835006714,
      "learning_rate": 0.0001782828282828283,
      "loss": 0.03467921167612076,
      "step": 354
    },
    {
      "epoch": 0.06747536012130402,
      "grad_norm": 0.34390759468078613,
      "learning_rate": 0.00017929292929292931,
      "loss": 0.016725102439522743,
      "step": 356
    },
    {
      "epoch": 0.06785443517816528,
      "grad_norm": 0.6199392676353455,
      "learning_rate": 0.0001803030303030303,
      "loss": 0.03566098213195801,
      "step": 358
    },
    {
      "epoch": 0.06823351023502654,
      "grad_norm": 0.35154300928115845,
      "learning_rate": 0.00018131313131313132,
      "loss": 0.03368252143263817,
      "step": 360
    },
    {
      "epoch": 0.0686125852918878,
      "grad_norm": 0.565058171749115,
      "learning_rate": 0.00018232323232323234,
      "loss": 0.03137116879224777,
      "step": 362
    },
    {
      "epoch": 0.06899166034874905,
      "grad_norm": 0.5647825002670288,
      "learning_rate": 0.00018333333333333334,
      "loss": 0.02478904090821743,
      "step": 364
    },
    {
      "epoch": 0.06937073540561031,
      "grad_norm": 0.48646360635757446,
      "learning_rate": 0.00018434343434343435,
      "loss": 0.018994053825736046,
      "step": 366
    },
    {
      "epoch": 0.06974981046247157,
      "grad_norm": 0.6416683793067932,
      "learning_rate": 0.00018535353535353537,
      "loss": 0.0325634740293026,
      "step": 368
    },
    {
      "epoch": 0.07012888551933283,
      "grad_norm": 0.663541316986084,
      "learning_rate": 0.00018636363636363636,
      "loss": 0.037587862461805344,
      "step": 370
    },
    {
      "epoch": 0.07050796057619409,
      "grad_norm": 0.6606209874153137,
      "learning_rate": 0.00018737373737373738,
      "loss": 0.05762958526611328,
      "step": 372
    },
    {
      "epoch": 0.07088703563305535,
      "grad_norm": 0.5284402966499329,
      "learning_rate": 0.0001883838383838384,
      "loss": 0.05038515105843544,
      "step": 374
    },
    {
      "epoch": 0.0712661106899166,
      "grad_norm": 0.3701823055744171,
      "learning_rate": 0.00018939393939393942,
      "loss": 0.015717800706624985,
      "step": 376
    },
    {
      "epoch": 0.07164518574677786,
      "grad_norm": 0.5930253267288208,
      "learning_rate": 0.0001904040404040404,
      "loss": 0.03529056906700134,
      "step": 378
    },
    {
      "epoch": 0.07202426080363912,
      "grad_norm": 0.40528714656829834,
      "learning_rate": 0.00019141414141414143,
      "loss": 0.039576999843120575,
      "step": 380
    },
    {
      "epoch": 0.07240333586050038,
      "grad_norm": 0.45489421486854553,
      "learning_rate": 0.00019242424242424245,
      "loss": 0.027852796018123627,
      "step": 382
    },
    {
      "epoch": 0.07278241091736164,
      "grad_norm": 0.5172394514083862,
      "learning_rate": 0.00019343434343434344,
      "loss": 0.02752961777150631,
      "step": 384
    },
    {
      "epoch": 0.0731614859742229,
      "grad_norm": 0.3172379732131958,
      "learning_rate": 0.00019444444444444446,
      "loss": 0.030919566750526428,
      "step": 386
    },
    {
      "epoch": 0.07354056103108415,
      "grad_norm": 0.26674485206604004,
      "learning_rate": 0.00019545454545454548,
      "loss": 0.024898799136281013,
      "step": 388
    },
    {
      "epoch": 0.07391963608794541,
      "grad_norm": 0.3241264820098877,
      "learning_rate": 0.0001964646464646465,
      "loss": 0.014084752649068832,
      "step": 390
    },
    {
      "epoch": 0.07429871114480667,
      "grad_norm": 0.5632086396217346,
      "learning_rate": 0.0001974747474747475,
      "loss": 0.030649710446596146,
      "step": 392
    },
    {
      "epoch": 0.07467778620166793,
      "grad_norm": 0.40097731351852417,
      "learning_rate": 0.0001984848484848485,
      "loss": 0.021177582442760468,
      "step": 394
    },
    {
      "epoch": 0.07505686125852919,
      "grad_norm": 0.34898918867111206,
      "learning_rate": 0.0001994949494949495,
      "loss": 0.03656506538391113,
      "step": 396
    },
    {
      "epoch": 0.07543593631539044,
      "grad_norm": 0.4035665690898895,
      "learning_rate": 0.0002,
      "loss": 0.01831536926329136,
      "step": 398
    },
    {
      "epoch": 0.0758150113722517,
      "grad_norm": 0.45603471994400024,
      "learning_rate": 0.0002,
      "loss": 0.036029066890478134,
      "step": 400
    },
    {
      "epoch": 0.07619408642911296,
      "grad_norm": 0.2934604585170746,
      "learning_rate": 0.0002,
      "loss": 0.02903594821691513,
      "step": 402
    },
    {
      "epoch": 0.07657316148597422,
      "grad_norm": 0.4999607503414154,
      "learning_rate": 0.0002,
      "loss": 0.03906955197453499,
      "step": 404
    },
    {
      "epoch": 0.07695223654283548,
      "grad_norm": 0.40500256419181824,
      "learning_rate": 0.0002,
      "loss": 0.020890168845653534,
      "step": 406
    },
    {
      "epoch": 0.07733131159969674,
      "grad_norm": 0.6000659465789795,
      "learning_rate": 0.0002,
      "loss": 0.04048675298690796,
      "step": 408
    },
    {
      "epoch": 0.077710386656558,
      "grad_norm": 0.2876969277858734,
      "learning_rate": 0.0002,
      "loss": 0.018003463745117188,
      "step": 410
    },
    {
      "epoch": 0.07808946171341925,
      "grad_norm": 0.26975446939468384,
      "learning_rate": 0.0002,
      "loss": 0.01824376918375492,
      "step": 412
    },
    {
      "epoch": 0.07846853677028051,
      "grad_norm": 0.2751922309398651,
      "learning_rate": 0.0002,
      "loss": 0.020754685625433922,
      "step": 414
    },
    {
      "epoch": 0.07884761182714177,
      "grad_norm": 0.30965426564216614,
      "learning_rate": 0.0002,
      "loss": 0.02495865523815155,
      "step": 416
    },
    {
      "epoch": 0.07922668688400303,
      "grad_norm": 0.48935362696647644,
      "learning_rate": 0.0002,
      "loss": 0.025911930948495865,
      "step": 418
    },
    {
      "epoch": 0.07960576194086429,
      "grad_norm": 0.3824619948863983,
      "learning_rate": 0.0002,
      "loss": 0.02747204527258873,
      "step": 420
    },
    {
      "epoch": 0.07998483699772554,
      "grad_norm": 0.5974108576774597,
      "learning_rate": 0.0002,
      "loss": 0.033621422946453094,
      "step": 422
    },
    {
      "epoch": 0.0803639120545868,
      "grad_norm": 0.3670952022075653,
      "learning_rate": 0.0002,
      "loss": 0.03134021535515785,
      "step": 424
    },
    {
      "epoch": 0.08074298711144806,
      "grad_norm": 0.5022920370101929,
      "learning_rate": 0.0002,
      "loss": 0.04992576315999031,
      "step": 426
    },
    {
      "epoch": 0.08112206216830932,
      "grad_norm": 0.27531135082244873,
      "learning_rate": 0.0002,
      "loss": 0.03650656342506409,
      "step": 428
    },
    {
      "epoch": 0.08150113722517058,
      "grad_norm": 0.21311552822589874,
      "learning_rate": 0.0002,
      "loss": 0.029359037056565285,
      "step": 430
    },
    {
      "epoch": 0.08188021228203184,
      "grad_norm": 0.5420831441879272,
      "learning_rate": 0.0002,
      "loss": 0.03439466655254364,
      "step": 432
    },
    {
      "epoch": 0.0822592873388931,
      "grad_norm": 0.4299890100955963,
      "learning_rate": 0.0002,
      "loss": 0.023038797080516815,
      "step": 434
    },
    {
      "epoch": 0.08263836239575435,
      "grad_norm": 0.3432794213294983,
      "learning_rate": 0.0002,
      "loss": 0.018038302659988403,
      "step": 436
    },
    {
      "epoch": 0.08301743745261561,
      "grad_norm": 0.5037815570831299,
      "learning_rate": 0.0002,
      "loss": 0.05114838108420372,
      "step": 438
    },
    {
      "epoch": 0.08339651250947688,
      "grad_norm": 0.2410070151090622,
      "learning_rate": 0.0002,
      "loss": 0.019249601289629936,
      "step": 440
    },
    {
      "epoch": 0.08377558756633814,
      "grad_norm": 0.5636169910430908,
      "learning_rate": 0.0002,
      "loss": 0.029105940833687782,
      "step": 442
    },
    {
      "epoch": 0.0841546626231994,
      "grad_norm": 0.38531336188316345,
      "learning_rate": 0.0002,
      "loss": 0.023919686675071716,
      "step": 444
    },
    {
      "epoch": 0.08453373768006066,
      "grad_norm": 0.533788800239563,
      "learning_rate": 0.0002,
      "loss": 0.053428102284669876,
      "step": 446
    },
    {
      "epoch": 0.08491281273692192,
      "grad_norm": 0.35856422781944275,
      "learning_rate": 0.0002,
      "loss": 0.01831984706223011,
      "step": 448
    },
    {
      "epoch": 0.08529188779378317,
      "grad_norm": 0.3086874783039093,
      "learning_rate": 0.0002,
      "loss": 0.028258061036467552,
      "step": 450
    },
    {
      "epoch": 0.08567096285064443,
      "grad_norm": 0.5657638311386108,
      "learning_rate": 0.0002,
      "loss": 0.03994130343198776,
      "step": 452
    },
    {
      "epoch": 0.08605003790750569,
      "grad_norm": 0.21996910870075226,
      "learning_rate": 0.0002,
      "loss": 0.01609434001147747,
      "step": 454
    },
    {
      "epoch": 0.08642911296436695,
      "grad_norm": 0.5379889011383057,
      "learning_rate": 0.0002,
      "loss": 0.048554565757513046,
      "step": 456
    },
    {
      "epoch": 0.08680818802122821,
      "grad_norm": 0.5755592584609985,
      "learning_rate": 0.0002,
      "loss": 0.04151491820812225,
      "step": 458
    },
    {
      "epoch": 0.08718726307808947,
      "grad_norm": 0.688117504119873,
      "learning_rate": 0.0002,
      "loss": 0.04011768847703934,
      "step": 460
    },
    {
      "epoch": 0.08756633813495072,
      "grad_norm": 0.48710131645202637,
      "learning_rate": 0.0002,
      "loss": 0.01748761720955372,
      "step": 462
    },
    {
      "epoch": 0.08794541319181198,
      "grad_norm": 0.37196093797683716,
      "learning_rate": 0.0002,
      "loss": 0.04196912422776222,
      "step": 464
    },
    {
      "epoch": 0.08832448824867324,
      "grad_norm": 0.4389767050743103,
      "learning_rate": 0.0002,
      "loss": 0.02894565463066101,
      "step": 466
    },
    {
      "epoch": 0.0887035633055345,
      "grad_norm": 0.34431779384613037,
      "learning_rate": 0.0002,
      "loss": 0.028105024248361588,
      "step": 468
    },
    {
      "epoch": 0.08908263836239576,
      "grad_norm": 0.39655619859695435,
      "learning_rate": 0.0002,
      "loss": 0.02502216026186943,
      "step": 470
    },
    {
      "epoch": 0.08946171341925702,
      "grad_norm": 0.605216383934021,
      "learning_rate": 0.0002,
      "loss": 0.031098224222660065,
      "step": 472
    },
    {
      "epoch": 0.08984078847611827,
      "grad_norm": 0.3806992769241333,
      "learning_rate": 0.0002,
      "loss": 0.020990340039134026,
      "step": 474
    },
    {
      "epoch": 0.09021986353297953,
      "grad_norm": 0.5819154381752014,
      "learning_rate": 0.0002,
      "loss": 0.04054046794772148,
      "step": 476
    },
    {
      "epoch": 0.09059893858984079,
      "grad_norm": 0.3327048718929291,
      "learning_rate": 0.0002,
      "loss": 0.01634993776679039,
      "step": 478
    },
    {
      "epoch": 0.09097801364670205,
      "grad_norm": 0.28937605023384094,
      "learning_rate": 0.0002,
      "loss": 0.015452735126018524,
      "step": 480
    },
    {
      "epoch": 0.09135708870356331,
      "grad_norm": 0.17362546920776367,
      "learning_rate": 0.0002,
      "loss": 0.02722182124853134,
      "step": 482
    },
    {
      "epoch": 0.09173616376042457,
      "grad_norm": 0.21261173486709595,
      "learning_rate": 0.0002,
      "loss": 0.029629327356815338,
      "step": 484
    },
    {
      "epoch": 0.09211523881728582,
      "grad_norm": 0.3077642321586609,
      "learning_rate": 0.0002,
      "loss": 0.032117582857608795,
      "step": 486
    },
    {
      "epoch": 0.09249431387414708,
      "grad_norm": 0.29473599791526794,
      "learning_rate": 0.0002,
      "loss": 0.01726091466844082,
      "step": 488
    },
    {
      "epoch": 0.09287338893100834,
      "grad_norm": 0.4163473844528198,
      "learning_rate": 0.0002,
      "loss": 0.025915110483765602,
      "step": 490
    },
    {
      "epoch": 0.0932524639878696,
      "grad_norm": 0.3271524906158447,
      "learning_rate": 0.0002,
      "loss": 0.021333537995815277,
      "step": 492
    },
    {
      "epoch": 0.09363153904473086,
      "grad_norm": 0.6723935604095459,
      "learning_rate": 0.0002,
      "loss": 0.04749723896384239,
      "step": 494
    },
    {
      "epoch": 0.09401061410159212,
      "grad_norm": 0.3260122537612915,
      "learning_rate": 0.0002,
      "loss": 0.03195744752883911,
      "step": 496
    },
    {
      "epoch": 0.09438968915845337,
      "grad_norm": 0.6022078394889832,
      "learning_rate": 0.0002,
      "loss": 0.030410440638661385,
      "step": 498
    },
    {
      "epoch": 0.09476876421531463,
      "grad_norm": 0.2818964719772339,
      "learning_rate": 0.0002,
      "loss": 0.035177163779735565,
      "step": 500
    },
    {
      "epoch": 0.09514783927217589,
      "grad_norm": 0.24321120977401733,
      "learning_rate": 0.0002,
      "loss": 0.01301200408488512,
      "step": 502
    },
    {
      "epoch": 0.09552691432903715,
      "grad_norm": 0.6208547949790955,
      "learning_rate": 0.0002,
      "loss": 0.035799793899059296,
      "step": 504
    },
    {
      "epoch": 0.09590598938589841,
      "grad_norm": 0.3829166889190674,
      "learning_rate": 0.0002,
      "loss": 0.035714041441679,
      "step": 506
    },
    {
      "epoch": 0.09628506444275967,
      "grad_norm": 0.6473880410194397,
      "learning_rate": 0.0002,
      "loss": 0.04498967528343201,
      "step": 508
    },
    {
      "epoch": 0.09666413949962092,
      "grad_norm": 0.4147353768348694,
      "learning_rate": 0.0002,
      "loss": 0.03037058562040329,
      "step": 510
    },
    {
      "epoch": 0.09704321455648218,
      "grad_norm": 0.46487733721733093,
      "learning_rate": 0.0002,
      "loss": 0.04653283208608627,
      "step": 512
    },
    {
      "epoch": 0.09742228961334344,
      "grad_norm": 0.41490641236305237,
      "learning_rate": 0.0002,
      "loss": 0.032938532531261444,
      "step": 514
    },
    {
      "epoch": 0.0978013646702047,
      "grad_norm": 0.2730114758014679,
      "learning_rate": 0.0002,
      "loss": 0.021666234359145164,
      "step": 516
    },
    {
      "epoch": 0.09818043972706596,
      "grad_norm": 0.3387247323989868,
      "learning_rate": 0.0002,
      "loss": 0.022410286590456963,
      "step": 518
    },
    {
      "epoch": 0.09855951478392722,
      "grad_norm": 0.2233457863330841,
      "learning_rate": 0.0002,
      "loss": 0.01501537300646305,
      "step": 520
    },
    {
      "epoch": 0.09893858984078847,
      "grad_norm": 0.3914678394794464,
      "learning_rate": 0.0002,
      "loss": 0.03032861463725567,
      "step": 522
    },
    {
      "epoch": 0.09931766489764973,
      "grad_norm": 0.3760395646095276,
      "learning_rate": 0.0002,
      "loss": 0.01654687523841858,
      "step": 524
    },
    {
      "epoch": 0.09969673995451099,
      "grad_norm": 0.535559892654419,
      "learning_rate": 0.0002,
      "loss": 0.03173451125621796,
      "step": 526
    },
    {
      "epoch": 0.10007581501137225,
      "grad_norm": 0.6054198741912842,
      "learning_rate": 0.0002,
      "loss": 0.0524178110063076,
      "step": 528
    },
    {
      "epoch": 0.10045489006823351,
      "grad_norm": 0.24288570880889893,
      "learning_rate": 0.0002,
      "loss": 0.020605528727173805,
      "step": 530
    },
    {
      "epoch": 0.10083396512509477,
      "grad_norm": 0.7124440670013428,
      "learning_rate": 0.0002,
      "loss": 0.052000828087329865,
      "step": 532
    },
    {
      "epoch": 0.10121304018195602,
      "grad_norm": 0.2223275750875473,
      "learning_rate": 0.0002,
      "loss": 0.016339987516403198,
      "step": 534
    },
    {
      "epoch": 0.10159211523881728,
      "grad_norm": 0.4922110140323639,
      "learning_rate": 0.0002,
      "loss": 0.037921976298093796,
      "step": 536
    },
    {
      "epoch": 0.10197119029567854,
      "grad_norm": 0.4628094732761383,
      "learning_rate": 0.0002,
      "loss": 0.03449274227023125,
      "step": 538
    },
    {
      "epoch": 0.1023502653525398,
      "grad_norm": 0.23278965055942535,
      "learning_rate": 0.0002,
      "loss": 0.022770900279283524,
      "step": 540
    },
    {
      "epoch": 0.10272934040940106,
      "grad_norm": 0.3020157814025879,
      "learning_rate": 0.0002,
      "loss": 0.028837986290454865,
      "step": 542
    },
    {
      "epoch": 0.10310841546626232,
      "grad_norm": 0.2503639757633209,
      "learning_rate": 0.0002,
      "loss": 0.023405035957694054,
      "step": 544
    },
    {
      "epoch": 0.10348749052312357,
      "grad_norm": 0.232551708817482,
      "learning_rate": 0.0002,
      "loss": 0.01887509971857071,
      "step": 546
    },
    {
      "epoch": 0.10386656557998483,
      "grad_norm": 0.3253386616706848,
      "learning_rate": 0.0002,
      "loss": 0.03918432071805,
      "step": 548
    },
    {
      "epoch": 0.10424564063684609,
      "grad_norm": 0.5936732888221741,
      "learning_rate": 0.0002,
      "loss": 0.03646896779537201,
      "step": 550
    },
    {
      "epoch": 0.10462471569370735,
      "grad_norm": 0.5108752250671387,
      "learning_rate": 0.0002,
      "loss": 0.03972802311182022,
      "step": 552
    },
    {
      "epoch": 0.1050037907505686,
      "grad_norm": 0.4022607207298279,
      "learning_rate": 0.0002,
      "loss": 0.024879850447177887,
      "step": 554
    },
    {
      "epoch": 0.10538286580742987,
      "grad_norm": 0.38569962978363037,
      "learning_rate": 0.0002,
      "loss": 0.02764149196445942,
      "step": 556
    },
    {
      "epoch": 0.10576194086429112,
      "grad_norm": 0.31124696135520935,
      "learning_rate": 0.0002,
      "loss": 0.03711552917957306,
      "step": 558
    },
    {
      "epoch": 0.10614101592115238,
      "grad_norm": 0.42483556270599365,
      "learning_rate": 0.0002,
      "loss": 0.04631171002984047,
      "step": 560
    },
    {
      "epoch": 0.10652009097801364,
      "grad_norm": 0.2938518524169922,
      "learning_rate": 0.0002,
      "loss": 0.030097246170043945,
      "step": 562
    },
    {
      "epoch": 0.1068991660348749,
      "grad_norm": 0.22558753192424774,
      "learning_rate": 0.0002,
      "loss": 0.01914200745522976,
      "step": 564
    },
    {
      "epoch": 0.10727824109173617,
      "grad_norm": 0.491737425327301,
      "learning_rate": 0.0002,
      "loss": 0.02886776067316532,
      "step": 566
    },
    {
      "epoch": 0.10765731614859743,
      "grad_norm": 0.538254976272583,
      "learning_rate": 0.0002,
      "loss": 0.056141335517168045,
      "step": 568
    },
    {
      "epoch": 0.10803639120545869,
      "grad_norm": 0.4184367060661316,
      "learning_rate": 0.0002,
      "loss": 0.02515416592359543,
      "step": 570
    },
    {
      "epoch": 0.10841546626231995,
      "grad_norm": 0.2020062357187271,
      "learning_rate": 0.0002,
      "loss": 0.031089870259165764,
      "step": 572
    },
    {
      "epoch": 0.1087945413191812,
      "grad_norm": 0.4681042432785034,
      "learning_rate": 0.0002,
      "loss": 0.02873258665204048,
      "step": 574
    },
    {
      "epoch": 0.10917361637604246,
      "grad_norm": 0.21195966005325317,
      "learning_rate": 0.0002,
      "loss": 0.014048480428755283,
      "step": 576
    },
    {
      "epoch": 0.10955269143290372,
      "grad_norm": 0.25848039984703064,
      "learning_rate": 0.0002,
      "loss": 0.02520585246384144,
      "step": 578
    },
    {
      "epoch": 0.10993176648976498,
      "grad_norm": 0.20653173327445984,
      "learning_rate": 0.0002,
      "loss": 0.014574659988284111,
      "step": 580
    },
    {
      "epoch": 0.11031084154662624,
      "grad_norm": 0.2899819016456604,
      "learning_rate": 0.0002,
      "loss": 0.013969087041914463,
      "step": 582
    },
    {
      "epoch": 0.1106899166034875,
      "grad_norm": 0.4556408226490021,
      "learning_rate": 0.0002,
      "loss": 0.04095076024532318,
      "step": 584
    },
    {
      "epoch": 0.11106899166034875,
      "grad_norm": 0.4273016154766083,
      "learning_rate": 0.0002,
      "loss": 0.043477077037096024,
      "step": 586
    },
    {
      "epoch": 0.11144806671721001,
      "grad_norm": 0.4144064486026764,
      "learning_rate": 0.0002,
      "loss": 0.027273865416646004,
      "step": 588
    },
    {
      "epoch": 0.11182714177407127,
      "grad_norm": 0.5851113200187683,
      "learning_rate": 0.0002,
      "loss": 0.047203924506902695,
      "step": 590
    },
    {
      "epoch": 0.11220621683093253,
      "grad_norm": 0.22309322655200958,
      "learning_rate": 0.0002,
      "loss": 0.014738938771188259,
      "step": 592
    },
    {
      "epoch": 0.11258529188779379,
      "grad_norm": 0.23289616405963898,
      "learning_rate": 0.0002,
      "loss": 0.018414178863167763,
      "step": 594
    },
    {
      "epoch": 0.11296436694465505,
      "grad_norm": 0.2816562056541443,
      "learning_rate": 0.0002,
      "loss": 0.02737065590918064,
      "step": 596
    },
    {
      "epoch": 0.1133434420015163,
      "grad_norm": 0.21728035807609558,
      "learning_rate": 0.0002,
      "loss": 0.031949568539857864,
      "step": 598
    },
    {
      "epoch": 0.11372251705837756,
      "grad_norm": 0.5699847936630249,
      "learning_rate": 0.0002,
      "loss": 0.039268191903829575,
      "step": 600
    },
    {
      "epoch": 0.11410159211523882,
      "grad_norm": 0.6229617595672607,
      "learning_rate": 0.0002,
      "loss": 0.04195765405893326,
      "step": 602
    },
    {
      "epoch": 0.11448066717210008,
      "grad_norm": 0.48147881031036377,
      "learning_rate": 0.0002,
      "loss": 0.028711115941405296,
      "step": 604
    },
    {
      "epoch": 0.11485974222896134,
      "grad_norm": 0.2070382982492447,
      "learning_rate": 0.0002,
      "loss": 0.02777196280658245,
      "step": 606
    },
    {
      "epoch": 0.1152388172858226,
      "grad_norm": 0.5017819404602051,
      "learning_rate": 0.0002,
      "loss": 0.02908783033490181,
      "step": 608
    },
    {
      "epoch": 0.11561789234268385,
      "grad_norm": 0.24335645139217377,
      "learning_rate": 0.0002,
      "loss": 0.014941886067390442,
      "step": 610
    },
    {
      "epoch": 0.11599696739954511,
      "grad_norm": 0.27532389760017395,
      "learning_rate": 0.0002,
      "loss": 0.024764984846115112,
      "step": 612
    },
    {
      "epoch": 0.11637604245640637,
      "grad_norm": 0.6199448704719543,
      "learning_rate": 0.0002,
      "loss": 0.04351656883955002,
      "step": 614
    },
    {
      "epoch": 0.11675511751326763,
      "grad_norm": 0.23394182324409485,
      "learning_rate": 0.0002,
      "loss": 0.04405692219734192,
      "step": 616
    },
    {
      "epoch": 0.11713419257012889,
      "grad_norm": 0.6148048639297485,
      "learning_rate": 0.0002,
      "loss": 0.03493811935186386,
      "step": 618
    },
    {
      "epoch": 0.11751326762699014,
      "grad_norm": 0.20812927186489105,
      "learning_rate": 0.0002,
      "loss": 0.030433928593993187,
      "step": 620
    },
    {
      "epoch": 0.1178923426838514,
      "grad_norm": 0.2601549029350281,
      "learning_rate": 0.0002,
      "loss": 0.03058495745062828,
      "step": 622
    },
    {
      "epoch": 0.11827141774071266,
      "grad_norm": 0.46609920263290405,
      "learning_rate": 0.0002,
      "loss": 0.032530758529901505,
      "step": 624
    },
    {
      "epoch": 0.11865049279757392,
      "grad_norm": 0.19339662790298462,
      "learning_rate": 0.0002,
      "loss": 0.020470114424824715,
      "step": 626
    },
    {
      "epoch": 0.11902956785443518,
      "grad_norm": 0.3398362398147583,
      "learning_rate": 0.0002,
      "loss": 0.042829982936382294,
      "step": 628
    },
    {
      "epoch": 0.11940864291129644,
      "grad_norm": 0.3699742555618286,
      "learning_rate": 0.0002,
      "loss": 0.01966828852891922,
      "step": 630
    },
    {
      "epoch": 0.1197877179681577,
      "grad_norm": 0.18957608938217163,
      "learning_rate": 0.0002,
      "loss": 0.02468428947031498,
      "step": 632
    },
    {
      "epoch": 0.12016679302501895,
      "grad_norm": 0.31358736753463745,
      "learning_rate": 0.0002,
      "loss": 0.017731202766299248,
      "step": 634
    },
    {
      "epoch": 0.12054586808188021,
      "grad_norm": 0.3797684907913208,
      "learning_rate": 0.0002,
      "loss": 0.025686761364340782,
      "step": 636
    },
    {
      "epoch": 0.12092494313874147,
      "grad_norm": 0.45164015889167786,
      "learning_rate": 0.0002,
      "loss": 0.026916328817605972,
      "step": 638
    },
    {
      "epoch": 0.12130401819560273,
      "grad_norm": 0.1858924925327301,
      "learning_rate": 0.0002,
      "loss": 0.015608190558850765,
      "step": 640
    },
    {
      "epoch": 0.12168309325246399,
      "grad_norm": 0.2156955450773239,
      "learning_rate": 0.0002,
      "loss": 0.019905690103769302,
      "step": 642
    },
    {
      "epoch": 0.12206216830932524,
      "grad_norm": 0.5047284960746765,
      "learning_rate": 0.0002,
      "loss": 0.039454616606235504,
      "step": 644
    },
    {
      "epoch": 0.1224412433661865,
      "grad_norm": 0.44826728105545044,
      "learning_rate": 0.0002,
      "loss": 0.05717497691512108,
      "step": 646
    },
    {
      "epoch": 0.12282031842304776,
      "grad_norm": 0.1931070238351822,
      "learning_rate": 0.0002,
      "loss": 0.01800347864627838,
      "step": 648
    },
    {
      "epoch": 0.12319939347990902,
      "grad_norm": 0.18938183784484863,
      "learning_rate": 0.0002,
      "loss": 0.01548789907246828,
      "step": 650
    },
    {
      "epoch": 0.12357846853677028,
      "grad_norm": 0.24963738024234772,
      "learning_rate": 0.0002,
      "loss": 0.014380771666765213,
      "step": 652
    },
    {
      "epoch": 0.12395754359363154,
      "grad_norm": 0.5348266959190369,
      "learning_rate": 0.0002,
      "loss": 0.03668570891022682,
      "step": 654
    },
    {
      "epoch": 0.1243366186504928,
      "grad_norm": 0.5190284252166748,
      "learning_rate": 0.0002,
      "loss": 0.03683571517467499,
      "step": 656
    },
    {
      "epoch": 0.12471569370735405,
      "grad_norm": 0.48402974009513855,
      "learning_rate": 0.0002,
      "loss": 0.03364763408899307,
      "step": 658
    },
    {
      "epoch": 0.12509476876421532,
      "grad_norm": 0.5165347456932068,
      "learning_rate": 0.0002,
      "loss": 0.030482126399874687,
      "step": 660
    },
    {
      "epoch": 0.12547384382107657,
      "grad_norm": 0.22087602317333221,
      "learning_rate": 0.0002,
      "loss": 0.021282417699694633,
      "step": 662
    },
    {
      "epoch": 0.12585291887793784,
      "grad_norm": 0.2965761721134186,
      "learning_rate": 0.0002,
      "loss": 0.022634640336036682,
      "step": 664
    },
    {
      "epoch": 0.12623199393479909,
      "grad_norm": 0.355797678232193,
      "learning_rate": 0.0002,
      "loss": 0.019156092777848244,
      "step": 666
    },
    {
      "epoch": 0.12661106899166036,
      "grad_norm": 0.6843011975288391,
      "learning_rate": 0.0002,
      "loss": 0.041870761662721634,
      "step": 668
    },
    {
      "epoch": 0.1269901440485216,
      "grad_norm": 0.16296321153640747,
      "learning_rate": 0.0002,
      "loss": 0.012693938799202442,
      "step": 670
    },
    {
      "epoch": 0.12736921910538287,
      "grad_norm": 0.520448625087738,
      "learning_rate": 0.0002,
      "loss": 0.03822709992527962,
      "step": 672
    },
    {
      "epoch": 0.12774829416224412,
      "grad_norm": 0.20806805789470673,
      "learning_rate": 0.0002,
      "loss": 0.026458093896508217,
      "step": 674
    },
    {
      "epoch": 0.1281273692191054,
      "grad_norm": 0.48958539962768555,
      "learning_rate": 0.0002,
      "loss": 0.0279215257614851,
      "step": 676
    },
    {
      "epoch": 0.12850644427596664,
      "grad_norm": 0.42718052864074707,
      "learning_rate": 0.0002,
      "loss": 0.015449983067810535,
      "step": 678
    },
    {
      "epoch": 0.1288855193328279,
      "grad_norm": 0.4777415096759796,
      "learning_rate": 0.0002,
      "loss": 0.03371915966272354,
      "step": 680
    },
    {
      "epoch": 0.12926459438968915,
      "grad_norm": 0.1919013112783432,
      "learning_rate": 0.0002,
      "loss": 0.013474739156663418,
      "step": 682
    },
    {
      "epoch": 0.12964366944655042,
      "grad_norm": 0.206001877784729,
      "learning_rate": 0.0002,
      "loss": 0.016890231519937515,
      "step": 684
    },
    {
      "epoch": 0.13002274450341167,
      "grad_norm": 0.4648573100566864,
      "learning_rate": 0.0002,
      "loss": 0.036212775856256485,
      "step": 686
    },
    {
      "epoch": 0.13040181956027294,
      "grad_norm": 0.16812044382095337,
      "learning_rate": 0.0002,
      "loss": 0.019974766299128532,
      "step": 688
    },
    {
      "epoch": 0.13078089461713419,
      "grad_norm": 0.4050150513648987,
      "learning_rate": 0.0002,
      "loss": 0.02375943772494793,
      "step": 690
    },
    {
      "epoch": 0.13115996967399546,
      "grad_norm": 0.32235583662986755,
      "learning_rate": 0.0002,
      "loss": 0.018055399879813194,
      "step": 692
    },
    {
      "epoch": 0.1315390447308567,
      "grad_norm": 0.6430607438087463,
      "learning_rate": 0.0002,
      "loss": 0.050872452557086945,
      "step": 694
    },
    {
      "epoch": 0.13191811978771797,
      "grad_norm": 0.16732773184776306,
      "learning_rate": 0.0002,
      "loss": 0.03410002961754799,
      "step": 696
    },
    {
      "epoch": 0.13229719484457922,
      "grad_norm": 0.43800118565559387,
      "learning_rate": 0.0002,
      "loss": 0.03963862732052803,
      "step": 698
    },
    {
      "epoch": 0.1326762699014405,
      "grad_norm": 0.20501519739627838,
      "learning_rate": 0.0002,
      "loss": 0.015204418450593948,
      "step": 700
    },
    {
      "epoch": 0.13305534495830174,
      "grad_norm": 0.3210892975330353,
      "learning_rate": 0.0002,
      "loss": 0.020870478823781013,
      "step": 702
    },
    {
      "epoch": 0.133434420015163,
      "grad_norm": 0.20072193443775177,
      "learning_rate": 0.0002,
      "loss": 0.03594975173473358,
      "step": 704
    },
    {
      "epoch": 0.13381349507202425,
      "grad_norm": 0.18484672904014587,
      "learning_rate": 0.0002,
      "loss": 0.011448661796748638,
      "step": 706
    },
    {
      "epoch": 0.13419257012888552,
      "grad_norm": 0.3297073245048523,
      "learning_rate": 0.0002,
      "loss": 0.025027839466929436,
      "step": 708
    },
    {
      "epoch": 0.13457164518574677,
      "grad_norm": 0.1889140009880066,
      "learning_rate": 0.0002,
      "loss": 0.024971093982458115,
      "step": 710
    },
    {
      "epoch": 0.13495072024260804,
      "grad_norm": 0.428009569644928,
      "learning_rate": 0.0002,
      "loss": 0.027155818417668343,
      "step": 712
    },
    {
      "epoch": 0.13532979529946929,
      "grad_norm": 0.4199044704437256,
      "learning_rate": 0.0002,
      "loss": 0.016979189589619637,
      "step": 714
    },
    {
      "epoch": 0.13570887035633056,
      "grad_norm": 0.5097037553787231,
      "learning_rate": 0.0002,
      "loss": 0.05768119916319847,
      "step": 716
    },
    {
      "epoch": 0.1360879454131918,
      "grad_norm": 0.42648422718048096,
      "learning_rate": 0.0002,
      "loss": 0.04554327204823494,
      "step": 718
    },
    {
      "epoch": 0.13646702047005307,
      "grad_norm": 0.3927448093891144,
      "learning_rate": 0.0002,
      "loss": 0.03952817618846893,
      "step": 720
    },
    {
      "epoch": 0.13684609552691432,
      "grad_norm": 0.5094998478889465,
      "learning_rate": 0.0002,
      "loss": 0.04147524759173393,
      "step": 722
    },
    {
      "epoch": 0.1372251705837756,
      "grad_norm": 0.5257030129432678,
      "learning_rate": 0.0002,
      "loss": 0.03971239551901817,
      "step": 724
    },
    {
      "epoch": 0.13760424564063684,
      "grad_norm": 0.2604212164878845,
      "learning_rate": 0.0002,
      "loss": 0.025074565783143044,
      "step": 726
    },
    {
      "epoch": 0.1379833206974981,
      "grad_norm": 0.18971319496631622,
      "learning_rate": 0.0002,
      "loss": 0.029028845950961113,
      "step": 728
    },
    {
      "epoch": 0.13836239575435935,
      "grad_norm": 0.1613638699054718,
      "learning_rate": 0.0002,
      "loss": 0.015578022226691246,
      "step": 730
    },
    {
      "epoch": 0.13874147081122062,
      "grad_norm": 0.19823560118675232,
      "learning_rate": 0.0002,
      "loss": 0.016313405707478523,
      "step": 732
    },
    {
      "epoch": 0.13912054586808187,
      "grad_norm": 0.23491230607032776,
      "learning_rate": 0.0002,
      "loss": 0.02603190951049328,
      "step": 734
    },
    {
      "epoch": 0.13949962092494314,
      "grad_norm": 0.32774049043655396,
      "learning_rate": 0.0002,
      "loss": 0.01729796640574932,
      "step": 736
    },
    {
      "epoch": 0.13987869598180439,
      "grad_norm": 0.21101008355617523,
      "learning_rate": 0.0002,
      "loss": 0.03305757790803909,
      "step": 738
    },
    {
      "epoch": 0.14025777103866566,
      "grad_norm": 0.20502084493637085,
      "learning_rate": 0.0002,
      "loss": 0.029380058869719505,
      "step": 740
    },
    {
      "epoch": 0.1406368460955269,
      "grad_norm": 0.12938687205314636,
      "learning_rate": 0.0002,
      "loss": 0.01285756565630436,
      "step": 742
    },
    {
      "epoch": 0.14101592115238817,
      "grad_norm": 0.2085622102022171,
      "learning_rate": 0.0002,
      "loss": 0.013858580030500889,
      "step": 744
    },
    {
      "epoch": 0.14139499620924942,
      "grad_norm": 0.17494770884513855,
      "learning_rate": 0.0002,
      "loss": 0.01559480745345354,
      "step": 746
    },
    {
      "epoch": 0.1417740712661107,
      "grad_norm": 0.222487211227417,
      "learning_rate": 0.0002,
      "loss": 0.015432793647050858,
      "step": 748
    },
    {
      "epoch": 0.14215314632297193,
      "grad_norm": 0.8851020932197571,
      "learning_rate": 0.0002,
      "loss": 0.043594565242528915,
      "step": 750
    },
    {
      "epoch": 0.1425322213798332,
      "grad_norm": 0.16391736268997192,
      "learning_rate": 0.0002,
      "loss": 0.014918128959834576,
      "step": 752
    },
    {
      "epoch": 0.14291129643669448,
      "grad_norm": 0.4164302349090576,
      "learning_rate": 0.0002,
      "loss": 0.03950285166501999,
      "step": 754
    },
    {
      "epoch": 0.14329037149355572,
      "grad_norm": 0.5083655118942261,
      "learning_rate": 0.0002,
      "loss": 0.0527968592941761,
      "step": 756
    },
    {
      "epoch": 0.143669446550417,
      "grad_norm": 0.556408166885376,
      "learning_rate": 0.0002,
      "loss": 0.035845376551151276,
      "step": 758
    },
    {
      "epoch": 0.14404852160727824,
      "grad_norm": 0.4845661520957947,
      "learning_rate": 0.0002,
      "loss": 0.027131814509630203,
      "step": 760
    },
    {
      "epoch": 0.1444275966641395,
      "grad_norm": 0.27998489141464233,
      "learning_rate": 0.0002,
      "loss": 0.028548330068588257,
      "step": 762
    },
    {
      "epoch": 0.14480667172100076,
      "grad_norm": 0.5767595767974854,
      "learning_rate": 0.0002,
      "loss": 0.03950953483581543,
      "step": 764
    },
    {
      "epoch": 0.14518574677786203,
      "grad_norm": 0.16699405014514923,
      "learning_rate": 0.0002,
      "loss": 0.019353531301021576,
      "step": 766
    },
    {
      "epoch": 0.14556482183472327,
      "grad_norm": 0.5418617725372314,
      "learning_rate": 0.0002,
      "loss": 0.05147811397910118,
      "step": 768
    },
    {
      "epoch": 0.14594389689158455,
      "grad_norm": 0.27640998363494873,
      "learning_rate": 0.0002,
      "loss": 0.02035924233496189,
      "step": 770
    },
    {
      "epoch": 0.1463229719484458,
      "grad_norm": 0.1959608793258667,
      "learning_rate": 0.0002,
      "loss": 0.019617455080151558,
      "step": 772
    },
    {
      "epoch": 0.14670204700530706,
      "grad_norm": 0.40992066264152527,
      "learning_rate": 0.0002,
      "loss": 0.029625743627548218,
      "step": 774
    },
    {
      "epoch": 0.1470811220621683,
      "grad_norm": 0.2797467112541199,
      "learning_rate": 0.0002,
      "loss": 0.03774365037679672,
      "step": 776
    },
    {
      "epoch": 0.14746019711902958,
      "grad_norm": 0.44566917419433594,
      "learning_rate": 0.0002,
      "loss": 0.0342901349067688,
      "step": 778
    },
    {
      "epoch": 0.14783927217589082,
      "grad_norm": 0.3128335475921631,
      "learning_rate": 0.0002,
      "loss": 0.01762254908680916,
      "step": 780
    },
    {
      "epoch": 0.1482183472327521,
      "grad_norm": 0.21120303869247437,
      "learning_rate": 0.0002,
      "loss": 0.016332169994711876,
      "step": 782
    },
    {
      "epoch": 0.14859742228961334,
      "grad_norm": 0.2822227478027344,
      "learning_rate": 0.0002,
      "loss": 0.020924067124724388,
      "step": 784
    },
    {
      "epoch": 0.1489764973464746,
      "grad_norm": 0.2059127241373062,
      "learning_rate": 0.0002,
      "loss": 0.01805771514773369,
      "step": 786
    },
    {
      "epoch": 0.14935557240333586,
      "grad_norm": 0.1606138050556183,
      "learning_rate": 0.0002,
      "loss": 0.013953215442597866,
      "step": 788
    },
    {
      "epoch": 0.14973464746019713,
      "grad_norm": 0.3935791552066803,
      "learning_rate": 0.0002,
      "loss": 0.022712744772434235,
      "step": 790
    },
    {
      "epoch": 0.15011372251705837,
      "grad_norm": 0.2900366485118866,
      "learning_rate": 0.0002,
      "loss": 0.020139053463935852,
      "step": 792
    },
    {
      "epoch": 0.15049279757391965,
      "grad_norm": 0.28215262293815613,
      "learning_rate": 0.0002,
      "loss": 0.025736834853887558,
      "step": 794
    },
    {
      "epoch": 0.1508718726307809,
      "grad_norm": 0.2824196219444275,
      "learning_rate": 0.0002,
      "loss": 0.03822169825434685,
      "step": 796
    },
    {
      "epoch": 0.15125094768764216,
      "grad_norm": 0.3952607810497284,
      "learning_rate": 0.0002,
      "loss": 0.023479409515857697,
      "step": 798
    },
    {
      "epoch": 0.1516300227445034,
      "grad_norm": 0.22573865950107574,
      "learning_rate": 0.0002,
      "loss": 0.026915449649095535,
      "step": 800
    },
    {
      "epoch": 0.15200909780136468,
      "grad_norm": 0.42073655128479004,
      "learning_rate": 0.0002,
      "loss": 0.03607917204499245,
      "step": 802
    },
    {
      "epoch": 0.15238817285822592,
      "grad_norm": 0.39296823740005493,
      "learning_rate": 0.0002,
      "loss": 0.035393815487623215,
      "step": 804
    },
    {
      "epoch": 0.1527672479150872,
      "grad_norm": 0.15439341962337494,
      "learning_rate": 0.0002,
      "loss": 0.014964856207370758,
      "step": 806
    },
    {
      "epoch": 0.15314632297194844,
      "grad_norm": 0.5265119075775146,
      "learning_rate": 0.0002,
      "loss": 0.03295372053980827,
      "step": 808
    },
    {
      "epoch": 0.1535253980288097,
      "grad_norm": 0.3580765426158905,
      "learning_rate": 0.0002,
      "loss": 0.03868171200156212,
      "step": 810
    },
    {
      "epoch": 0.15390447308567096,
      "grad_norm": 0.46609780192375183,
      "learning_rate": 0.0002,
      "loss": 0.03573188930749893,
      "step": 812
    },
    {
      "epoch": 0.15428354814253223,
      "grad_norm": 0.23661619424819946,
      "learning_rate": 0.0002,
      "loss": 0.014220339246094227,
      "step": 814
    },
    {
      "epoch": 0.15466262319939347,
      "grad_norm": 0.17906050384044647,
      "learning_rate": 0.0002,
      "loss": 0.017339101061224937,
      "step": 816
    },
    {
      "epoch": 0.15504169825625475,
      "grad_norm": 0.14934410154819489,
      "learning_rate": 0.0002,
      "loss": 0.016141701489686966,
      "step": 818
    },
    {
      "epoch": 0.155420773313116,
      "grad_norm": 0.3761049509048462,
      "learning_rate": 0.0002,
      "loss": 0.02728787064552307,
      "step": 820
    },
    {
      "epoch": 0.15579984836997726,
      "grad_norm": 0.5973101258277893,
      "learning_rate": 0.0002,
      "loss": 0.05032341927289963,
      "step": 822
    },
    {
      "epoch": 0.1561789234268385,
      "grad_norm": 0.5155751705169678,
      "learning_rate": 0.0002,
      "loss": 0.05607166513800621,
      "step": 824
    },
    {
      "epoch": 0.15655799848369978,
      "grad_norm": 0.17700494825839996,
      "learning_rate": 0.0002,
      "loss": 0.013834198005497456,
      "step": 826
    },
    {
      "epoch": 0.15693707354056102,
      "grad_norm": 0.463019460439682,
      "learning_rate": 0.0002,
      "loss": 0.030374575406312943,
      "step": 828
    },
    {
      "epoch": 0.1573161485974223,
      "grad_norm": 0.5239229202270508,
      "learning_rate": 0.0002,
      "loss": 0.03492042422294617,
      "step": 830
    },
    {
      "epoch": 0.15769522365428354,
      "grad_norm": 0.19757749140262604,
      "learning_rate": 0.0002,
      "loss": 0.013787648640573025,
      "step": 832
    },
    {
      "epoch": 0.1580742987111448,
      "grad_norm": 0.16166947782039642,
      "learning_rate": 0.0002,
      "loss": 0.015328834764659405,
      "step": 834
    },
    {
      "epoch": 0.15845337376800606,
      "grad_norm": 0.2141238898038864,
      "learning_rate": 0.0002,
      "loss": 0.01616821624338627,
      "step": 836
    },
    {
      "epoch": 0.15883244882486733,
      "grad_norm": 0.5520454049110413,
      "learning_rate": 0.0002,
      "loss": 0.04373732954263687,
      "step": 838
    },
    {
      "epoch": 0.15921152388172857,
      "grad_norm": 0.23446916043758392,
      "learning_rate": 0.0002,
      "loss": 0.01704292558133602,
      "step": 840
    },
    {
      "epoch": 0.15959059893858984,
      "grad_norm": 0.39795413613319397,
      "learning_rate": 0.0002,
      "loss": 0.03359442204236984,
      "step": 842
    },
    {
      "epoch": 0.1599696739954511,
      "grad_norm": 0.2551341950893402,
      "learning_rate": 0.0002,
      "loss": 0.021616891026496887,
      "step": 844
    },
    {
      "epoch": 0.16034874905231236,
      "grad_norm": 0.4329982101917267,
      "learning_rate": 0.0002,
      "loss": 0.028922230005264282,
      "step": 846
    },
    {
      "epoch": 0.1607278241091736,
      "grad_norm": 0.32268935441970825,
      "learning_rate": 0.0002,
      "loss": 0.017331285402178764,
      "step": 848
    },
    {
      "epoch": 0.16110689916603488,
      "grad_norm": 0.192164808511734,
      "learning_rate": 0.0002,
      "loss": 0.014898553490638733,
      "step": 850
    },
    {
      "epoch": 0.16148597422289612,
      "grad_norm": 0.3638659119606018,
      "learning_rate": 0.0002,
      "loss": 0.0455300509929657,
      "step": 852
    },
    {
      "epoch": 0.1618650492797574,
      "grad_norm": 0.34963327646255493,
      "learning_rate": 0.0002,
      "loss": 0.022118110209703445,
      "step": 854
    },
    {
      "epoch": 0.16224412433661864,
      "grad_norm": 0.48588207364082336,
      "learning_rate": 0.0002,
      "loss": 0.03968464955687523,
      "step": 856
    },
    {
      "epoch": 0.1626231993934799,
      "grad_norm": 0.174435093998909,
      "learning_rate": 0.0002,
      "loss": 0.01789679564535618,
      "step": 858
    },
    {
      "epoch": 0.16300227445034116,
      "grad_norm": 0.3277144730091095,
      "learning_rate": 0.0002,
      "loss": 0.0192416962236166,
      "step": 860
    },
    {
      "epoch": 0.16338134950720243,
      "grad_norm": 0.14818210899829865,
      "learning_rate": 0.0002,
      "loss": 0.032830026000738144,
      "step": 862
    },
    {
      "epoch": 0.16376042456406367,
      "grad_norm": 0.473153293132782,
      "learning_rate": 0.0002,
      "loss": 0.0415210947394371,
      "step": 864
    },
    {
      "epoch": 0.16413949962092494,
      "grad_norm": 0.32705825567245483,
      "learning_rate": 0.0002,
      "loss": 0.019977573305368423,
      "step": 866
    },
    {
      "epoch": 0.1645185746777862,
      "grad_norm": 0.26092529296875,
      "learning_rate": 0.0002,
      "loss": 0.03465384244918823,
      "step": 868
    },
    {
      "epoch": 0.16489764973464746,
      "grad_norm": 0.3672652542591095,
      "learning_rate": 0.0002,
      "loss": 0.020774519070982933,
      "step": 870
    },
    {
      "epoch": 0.1652767247915087,
      "grad_norm": 0.3195513188838959,
      "learning_rate": 0.0002,
      "loss": 0.019471392035484314,
      "step": 872
    },
    {
      "epoch": 0.16565579984836998,
      "grad_norm": 0.1685757339000702,
      "learning_rate": 0.0002,
      "loss": 0.016216494143009186,
      "step": 874
    },
    {
      "epoch": 0.16603487490523122,
      "grad_norm": 0.14554302394390106,
      "learning_rate": 0.0002,
      "loss": 0.014610818587243557,
      "step": 876
    },
    {
      "epoch": 0.1664139499620925,
      "grad_norm": 0.7545990347862244,
      "learning_rate": 0.0002,
      "loss": 0.03460363298654556,
      "step": 878
    },
    {
      "epoch": 0.16679302501895377,
      "grad_norm": 0.4334106147289276,
      "learning_rate": 0.0002,
      "loss": 0.03627653419971466,
      "step": 880
    },
    {
      "epoch": 0.167172100075815,
      "grad_norm": 0.1313071995973587,
      "learning_rate": 0.0002,
      "loss": 0.03117945045232773,
      "step": 882
    },
    {
      "epoch": 0.16755117513267628,
      "grad_norm": 0.19411614537239075,
      "learning_rate": 0.0002,
      "loss": 0.025972707197070122,
      "step": 884
    },
    {
      "epoch": 0.16793025018953753,
      "grad_norm": 0.3259235620498657,
      "learning_rate": 0.0002,
      "loss": 0.028867168352007866,
      "step": 886
    },
    {
      "epoch": 0.1683093252463988,
      "grad_norm": 0.15235097706317902,
      "learning_rate": 0.0002,
      "loss": 0.014279675669968128,
      "step": 888
    },
    {
      "epoch": 0.16868840030326004,
      "grad_norm": 0.29882943630218506,
      "learning_rate": 0.0002,
      "loss": 0.028727473691105843,
      "step": 890
    },
    {
      "epoch": 0.16906747536012132,
      "grad_norm": 0.40746158361434937,
      "learning_rate": 0.0002,
      "loss": 0.0401567742228508,
      "step": 892
    },
    {
      "epoch": 0.16944655041698256,
      "grad_norm": 0.49056240916252136,
      "learning_rate": 0.0002,
      "loss": 0.03595411777496338,
      "step": 894
    },
    {
      "epoch": 0.16982562547384383,
      "grad_norm": 0.3808828294277191,
      "learning_rate": 0.0002,
      "loss": 0.025514105334877968,
      "step": 896
    },
    {
      "epoch": 0.17020470053070508,
      "grad_norm": 0.18481013178825378,
      "learning_rate": 0.0002,
      "loss": 0.025887150317430496,
      "step": 898
    },
    {
      "epoch": 0.17058377558756635,
      "grad_norm": 0.48471981287002563,
      "learning_rate": 0.0002,
      "loss": 0.036403313279151917,
      "step": 900
    },
    {
      "epoch": 0.1709628506444276,
      "grad_norm": 0.4137479066848755,
      "learning_rate": 0.0002,
      "loss": 0.022203141823410988,
      "step": 902
    },
    {
      "epoch": 0.17134192570128887,
      "grad_norm": 0.5495334267616272,
      "learning_rate": 0.0002,
      "loss": 0.03433135896921158,
      "step": 904
    },
    {
      "epoch": 0.1717210007581501,
      "grad_norm": 0.46607109904289246,
      "learning_rate": 0.0002,
      "loss": 0.020335154607892036,
      "step": 906
    },
    {
      "epoch": 0.17210007581501138,
      "grad_norm": 0.20998910069465637,
      "learning_rate": 0.0002,
      "loss": 0.019366079941391945,
      "step": 908
    },
    {
      "epoch": 0.17247915087187263,
      "grad_norm": 0.15787310898303986,
      "learning_rate": 0.0002,
      "loss": 0.027786120772361755,
      "step": 910
    },
    {
      "epoch": 0.1728582259287339,
      "grad_norm": 0.4805816411972046,
      "learning_rate": 0.0002,
      "loss": 0.03142945468425751,
      "step": 912
    },
    {
      "epoch": 0.17323730098559514,
      "grad_norm": 0.27658218145370483,
      "learning_rate": 0.0002,
      "loss": 0.025804150849580765,
      "step": 914
    },
    {
      "epoch": 0.17361637604245642,
      "grad_norm": 0.15161961317062378,
      "learning_rate": 0.0002,
      "loss": 0.028010597452521324,
      "step": 916
    },
    {
      "epoch": 0.17399545109931766,
      "grad_norm": 0.17970381677150726,
      "learning_rate": 0.0002,
      "loss": 0.03571527078747749,
      "step": 918
    },
    {
      "epoch": 0.17437452615617893,
      "grad_norm": 0.3615803122520447,
      "learning_rate": 0.0002,
      "loss": 0.02821771241724491,
      "step": 920
    },
    {
      "epoch": 0.17475360121304018,
      "grad_norm": 0.4722934663295746,
      "learning_rate": 0.0002,
      "loss": 0.027622580528259277,
      "step": 922
    },
    {
      "epoch": 0.17513267626990145,
      "grad_norm": 0.3511749804019928,
      "learning_rate": 0.0002,
      "loss": 0.01574043557047844,
      "step": 924
    },
    {
      "epoch": 0.1755117513267627,
      "grad_norm": 0.517323911190033,
      "learning_rate": 0.0002,
      "loss": 0.041282616555690765,
      "step": 926
    },
    {
      "epoch": 0.17589082638362397,
      "grad_norm": 0.31259870529174805,
      "learning_rate": 0.0002,
      "loss": 0.022952917963266373,
      "step": 928
    },
    {
      "epoch": 0.1762699014404852,
      "grad_norm": 0.3582378625869751,
      "learning_rate": 0.0002,
      "loss": 0.03770346939563751,
      "step": 930
    },
    {
      "epoch": 0.17664897649734648,
      "grad_norm": 0.168486550450325,
      "learning_rate": 0.0002,
      "loss": 0.013697076588869095,
      "step": 932
    },
    {
      "epoch": 0.17702805155420773,
      "grad_norm": 0.4889657497406006,
      "learning_rate": 0.0002,
      "loss": 0.03705749660730362,
      "step": 934
    },
    {
      "epoch": 0.177407126611069,
      "grad_norm": 0.4104210436344147,
      "learning_rate": 0.0002,
      "loss": 0.030090676620602608,
      "step": 936
    },
    {
      "epoch": 0.17778620166793024,
      "grad_norm": 0.1999150812625885,
      "learning_rate": 0.0002,
      "loss": 0.01491877157241106,
      "step": 938
    },
    {
      "epoch": 0.17816527672479152,
      "grad_norm": 0.47530725598335266,
      "learning_rate": 0.0002,
      "loss": 0.0280255526304245,
      "step": 940
    },
    {
      "epoch": 0.17854435178165276,
      "grad_norm": 0.37718576192855835,
      "learning_rate": 0.0002,
      "loss": 0.036472778767347336,
      "step": 942
    },
    {
      "epoch": 0.17892342683851403,
      "grad_norm": 0.3084065020084381,
      "learning_rate": 0.0002,
      "loss": 0.017929354682564735,
      "step": 944
    },
    {
      "epoch": 0.17930250189537528,
      "grad_norm": 0.4129011631011963,
      "learning_rate": 0.0002,
      "loss": 0.024281833320856094,
      "step": 946
    },
    {
      "epoch": 0.17968157695223655,
      "grad_norm": 0.5921041369438171,
      "learning_rate": 0.0002,
      "loss": 0.058901600539684296,
      "step": 948
    },
    {
      "epoch": 0.1800606520090978,
      "grad_norm": 0.45201990008354187,
      "learning_rate": 0.0002,
      "loss": 0.029276639223098755,
      "step": 950
    },
    {
      "epoch": 0.18043972706595907,
      "grad_norm": 0.38334590196609497,
      "learning_rate": 0.0002,
      "loss": 0.014062942005693913,
      "step": 952
    },
    {
      "epoch": 0.1808188021228203,
      "grad_norm": 0.17091278731822968,
      "learning_rate": 0.0002,
      "loss": 0.02697864919900894,
      "step": 954
    },
    {
      "epoch": 0.18119787717968158,
      "grad_norm": 0.17652741074562073,
      "learning_rate": 0.0002,
      "loss": 0.013573378324508667,
      "step": 956
    },
    {
      "epoch": 0.18157695223654283,
      "grad_norm": 0.4162633717060089,
      "learning_rate": 0.0002,
      "loss": 0.037839170545339584,
      "step": 958
    },
    {
      "epoch": 0.1819560272934041,
      "grad_norm": 0.1850610077381134,
      "learning_rate": 0.0002,
      "loss": 0.01515909843146801,
      "step": 960
    },
    {
      "epoch": 0.18233510235026534,
      "grad_norm": 0.15553788840770721,
      "learning_rate": 0.0002,
      "loss": 0.024147771298885345,
      "step": 962
    },
    {
      "epoch": 0.18271417740712662,
      "grad_norm": 0.5542011260986328,
      "learning_rate": 0.0002,
      "loss": 0.041745010763406754,
      "step": 964
    },
    {
      "epoch": 0.18309325246398786,
      "grad_norm": 0.3536843955516815,
      "learning_rate": 0.0002,
      "loss": 0.02222476899623871,
      "step": 966
    },
    {
      "epoch": 0.18347232752084913,
      "grad_norm": 0.27479830384254456,
      "learning_rate": 0.0002,
      "loss": 0.0306385587900877,
      "step": 968
    },
    {
      "epoch": 0.18385140257771038,
      "grad_norm": 0.32002177834510803,
      "learning_rate": 0.0002,
      "loss": 0.019492637366056442,
      "step": 970
    },
    {
      "epoch": 0.18423047763457165,
      "grad_norm": 0.39961451292037964,
      "learning_rate": 0.0002,
      "loss": 0.03355245292186737,
      "step": 972
    },
    {
      "epoch": 0.1846095526914329,
      "grad_norm": 0.1900455802679062,
      "learning_rate": 0.0002,
      "loss": 0.025758838281035423,
      "step": 974
    },
    {
      "epoch": 0.18498862774829417,
      "grad_norm": 0.23303648829460144,
      "learning_rate": 0.0002,
      "loss": 0.0296570286154747,
      "step": 976
    },
    {
      "epoch": 0.1853677028051554,
      "grad_norm": 0.11677033454179764,
      "learning_rate": 0.0002,
      "loss": 0.019421905279159546,
      "step": 978
    },
    {
      "epoch": 0.18574677786201668,
      "grad_norm": 0.21048873662948608,
      "learning_rate": 0.0002,
      "loss": 0.03341106325387955,
      "step": 980
    },
    {
      "epoch": 0.18612585291887793,
      "grad_norm": 0.4954831898212433,
      "learning_rate": 0.0002,
      "loss": 0.045417968183755875,
      "step": 982
    },
    {
      "epoch": 0.1865049279757392,
      "grad_norm": 0.40720614790916443,
      "learning_rate": 0.0002,
      "loss": 0.024235138669610023,
      "step": 984
    },
    {
      "epoch": 0.18688400303260044,
      "grad_norm": 0.45604026317596436,
      "learning_rate": 0.0002,
      "loss": 0.027927320450544357,
      "step": 986
    },
    {
      "epoch": 0.18726307808946172,
      "grad_norm": 0.5153806805610657,
      "learning_rate": 0.0002,
      "loss": 0.031180649995803833,
      "step": 988
    },
    {
      "epoch": 0.18764215314632296,
      "grad_norm": 0.374468594789505,
      "learning_rate": 0.0002,
      "loss": 0.02738489955663681,
      "step": 990
    },
    {
      "epoch": 0.18802122820318423,
      "grad_norm": 0.13685302436351776,
      "learning_rate": 0.0002,
      "loss": 0.028205839917063713,
      "step": 992
    },
    {
      "epoch": 0.18840030326004548,
      "grad_norm": 0.39598268270492554,
      "learning_rate": 0.0002,
      "loss": 0.03009963594377041,
      "step": 994
    },
    {
      "epoch": 0.18877937831690675,
      "grad_norm": 0.3017390966415405,
      "learning_rate": 0.0002,
      "loss": 0.018373258411884308,
      "step": 996
    },
    {
      "epoch": 0.189158453373768,
      "grad_norm": 0.47517889738082886,
      "learning_rate": 0.0002,
      "loss": 0.024969086050987244,
      "step": 998
    },
    {
      "epoch": 0.18953752843062927,
      "grad_norm": 0.18933281302452087,
      "learning_rate": 0.0002,
      "loss": 0.02427705004811287,
      "step": 1000
    },
    {
      "epoch": 0.1899166034874905,
      "grad_norm": 0.17442548274993896,
      "learning_rate": 0.0002,
      "loss": 0.03813823312520981,
      "step": 1002
    },
    {
      "epoch": 0.19029567854435178,
      "grad_norm": 0.5236710906028748,
      "learning_rate": 0.0002,
      "loss": 0.03336897864937782,
      "step": 1004
    },
    {
      "epoch": 0.19067475360121305,
      "grad_norm": 0.4650385081768036,
      "learning_rate": 0.0002,
      "loss": 0.05045532435178757,
      "step": 1006
    },
    {
      "epoch": 0.1910538286580743,
      "grad_norm": 0.4482308328151703,
      "learning_rate": 0.0002,
      "loss": 0.02459673210978508,
      "step": 1008
    },
    {
      "epoch": 0.19143290371493557,
      "grad_norm": 0.5231438279151917,
      "learning_rate": 0.0002,
      "loss": 0.0314045175909996,
      "step": 1010
    },
    {
      "epoch": 0.19181197877179681,
      "grad_norm": 0.20980830490589142,
      "learning_rate": 0.0002,
      "loss": 0.04040207341313362,
      "step": 1012
    },
    {
      "epoch": 0.1921910538286581,
      "grad_norm": 0.465640127658844,
      "learning_rate": 0.0002,
      "loss": 0.037633590400218964,
      "step": 1014
    },
    {
      "epoch": 0.19257012888551933,
      "grad_norm": 0.34506601095199585,
      "learning_rate": 0.0002,
      "loss": 0.029547519981861115,
      "step": 1016
    },
    {
      "epoch": 0.1929492039423806,
      "grad_norm": 0.3737991452217102,
      "learning_rate": 0.0002,
      "loss": 0.026226969435811043,
      "step": 1018
    },
    {
      "epoch": 0.19332827899924185,
      "grad_norm": 0.12815317511558533,
      "learning_rate": 0.0002,
      "loss": 0.012681324034929276,
      "step": 1020
    },
    {
      "epoch": 0.19370735405610312,
      "grad_norm": 0.5799553394317627,
      "learning_rate": 0.0002,
      "loss": 0.057865556329488754,
      "step": 1022
    },
    {
      "epoch": 0.19408642911296436,
      "grad_norm": 0.4656653106212616,
      "learning_rate": 0.0002,
      "loss": 0.0398462750017643,
      "step": 1024
    },
    {
      "epoch": 0.19446550416982564,
      "grad_norm": 0.15832734107971191,
      "learning_rate": 0.0002,
      "loss": 0.023395728319883347,
      "step": 1026
    },
    {
      "epoch": 0.19484457922668688,
      "grad_norm": 0.3733989894390106,
      "learning_rate": 0.0002,
      "loss": 0.02530939131975174,
      "step": 1028
    },
    {
      "epoch": 0.19522365428354815,
      "grad_norm": 0.2360105663537979,
      "learning_rate": 0.0002,
      "loss": 0.017093271017074585,
      "step": 1030
    },
    {
      "epoch": 0.1956027293404094,
      "grad_norm": 0.5019320845603943,
      "learning_rate": 0.0002,
      "loss": 0.04170316085219383,
      "step": 1032
    },
    {
      "epoch": 0.19598180439727067,
      "grad_norm": 0.418171763420105,
      "learning_rate": 0.0002,
      "loss": 0.01600993610918522,
      "step": 1034
    },
    {
      "epoch": 0.19636087945413191,
      "grad_norm": 0.24843889474868774,
      "learning_rate": 0.0002,
      "loss": 0.019397562369704247,
      "step": 1036
    },
    {
      "epoch": 0.1967399545109932,
      "grad_norm": 0.23631003499031067,
      "learning_rate": 0.0002,
      "loss": 0.01794506423175335,
      "step": 1038
    },
    {
      "epoch": 0.19711902956785443,
      "grad_norm": 0.14922305941581726,
      "learning_rate": 0.0002,
      "loss": 0.014477409422397614,
      "step": 1040
    },
    {
      "epoch": 0.1974981046247157,
      "grad_norm": 0.6032777428627014,
      "learning_rate": 0.0002,
      "loss": 0.025451600551605225,
      "step": 1042
    },
    {
      "epoch": 0.19787717968157695,
      "grad_norm": 0.2516152262687683,
      "learning_rate": 0.0002,
      "loss": 0.03702351078391075,
      "step": 1044
    },
    {
      "epoch": 0.19825625473843822,
      "grad_norm": 0.1284353882074356,
      "learning_rate": 0.0002,
      "loss": 0.02339061163365841,
      "step": 1046
    },
    {
      "epoch": 0.19863532979529946,
      "grad_norm": 0.1269228607416153,
      "learning_rate": 0.0002,
      "loss": 0.012414056807756424,
      "step": 1048
    },
    {
      "epoch": 0.19901440485216074,
      "grad_norm": 0.1801559180021286,
      "learning_rate": 0.0002,
      "loss": 0.013697893358767033,
      "step": 1050
    },
    {
      "epoch": 0.19939347990902198,
      "grad_norm": 0.14652559161186218,
      "learning_rate": 0.0002,
      "loss": 0.026144472882151604,
      "step": 1052
    },
    {
      "epoch": 0.19977255496588325,
      "grad_norm": 0.16796112060546875,
      "learning_rate": 0.0002,
      "loss": 0.022053087130188942,
      "step": 1054
    },
    {
      "epoch": 0.2001516300227445,
      "grad_norm": 0.24088618159294128,
      "learning_rate": 0.0002,
      "loss": 0.02511882223188877,
      "step": 1056
    },
    {
      "epoch": 0.20053070507960577,
      "grad_norm": 0.39973267912864685,
      "learning_rate": 0.0002,
      "loss": 0.029609696939587593,
      "step": 1058
    },
    {
      "epoch": 0.20090978013646701,
      "grad_norm": 0.4267011284828186,
      "learning_rate": 0.0002,
      "loss": 0.035159192979335785,
      "step": 1060
    },
    {
      "epoch": 0.2012888551933283,
      "grad_norm": 0.3776701092720032,
      "learning_rate": 0.0002,
      "loss": 0.03557874262332916,
      "step": 1062
    },
    {
      "epoch": 0.20166793025018953,
      "grad_norm": 0.6586447358131409,
      "learning_rate": 0.0002,
      "loss": 0.03392799198627472,
      "step": 1064
    },
    {
      "epoch": 0.2020470053070508,
      "grad_norm": 0.5256524682044983,
      "learning_rate": 0.0002,
      "loss": 0.03053244948387146,
      "step": 1066
    },
    {
      "epoch": 0.20242608036391205,
      "grad_norm": 0.39960190653800964,
      "learning_rate": 0.0002,
      "loss": 0.028431696817278862,
      "step": 1068
    },
    {
      "epoch": 0.20280515542077332,
      "grad_norm": 0.167899489402771,
      "learning_rate": 0.0002,
      "loss": 0.028788940981030464,
      "step": 1070
    },
    {
      "epoch": 0.20318423047763456,
      "grad_norm": 0.37918820977211,
      "learning_rate": 0.0002,
      "loss": 0.03250772878527641,
      "step": 1072
    },
    {
      "epoch": 0.20356330553449584,
      "grad_norm": 0.41917797923088074,
      "learning_rate": 0.0002,
      "loss": 0.026176655665040016,
      "step": 1074
    },
    {
      "epoch": 0.20394238059135708,
      "grad_norm": 0.13173426687717438,
      "learning_rate": 0.0002,
      "loss": 0.01946214959025383,
      "step": 1076
    },
    {
      "epoch": 0.20432145564821835,
      "grad_norm": 0.4758298099040985,
      "learning_rate": 0.0002,
      "loss": 0.05166337639093399,
      "step": 1078
    },
    {
      "epoch": 0.2047005307050796,
      "grad_norm": 0.14329728484153748,
      "learning_rate": 0.0002,
      "loss": 0.024442167952656746,
      "step": 1080
    },
    {
      "epoch": 0.20507960576194087,
      "grad_norm": 0.3563183844089508,
      "learning_rate": 0.0002,
      "loss": 0.02090851590037346,
      "step": 1082
    },
    {
      "epoch": 0.20545868081880211,
      "grad_norm": 0.1373371183872223,
      "learning_rate": 0.0002,
      "loss": 0.015078917145729065,
      "step": 1084
    },
    {
      "epoch": 0.2058377558756634,
      "grad_norm": 0.48479700088500977,
      "learning_rate": 0.0002,
      "loss": 0.024862686172127724,
      "step": 1086
    },
    {
      "epoch": 0.20621683093252463,
      "grad_norm": 0.5952326059341431,
      "learning_rate": 0.0002,
      "loss": 0.06471777707338333,
      "step": 1088
    },
    {
      "epoch": 0.2065959059893859,
      "grad_norm": 0.1521560102701187,
      "learning_rate": 0.0002,
      "loss": 0.0265179593116045,
      "step": 1090
    },
    {
      "epoch": 0.20697498104624715,
      "grad_norm": 0.35660138726234436,
      "learning_rate": 0.0002,
      "loss": 0.024818971753120422,
      "step": 1092
    },
    {
      "epoch": 0.20735405610310842,
      "grad_norm": 0.28938737511634827,
      "learning_rate": 0.0002,
      "loss": 0.030998462811112404,
      "step": 1094
    },
    {
      "epoch": 0.20773313115996966,
      "grad_norm": 0.5063716769218445,
      "learning_rate": 0.0002,
      "loss": 0.03379755839705467,
      "step": 1096
    },
    {
      "epoch": 0.20811220621683094,
      "grad_norm": 0.40287089347839355,
      "learning_rate": 0.0002,
      "loss": 0.03909891098737717,
      "step": 1098
    },
    {
      "epoch": 0.20849128127369218,
      "grad_norm": 0.49371182918548584,
      "learning_rate": 0.0002,
      "loss": 0.03448515012860298,
      "step": 1100
    },
    {
      "epoch": 0.20887035633055345,
      "grad_norm": 0.5586045384407043,
      "learning_rate": 0.0002,
      "loss": 0.029398340731859207,
      "step": 1102
    },
    {
      "epoch": 0.2092494313874147,
      "grad_norm": 0.3637477457523346,
      "learning_rate": 0.0002,
      "loss": 0.02711344324052334,
      "step": 1104
    },
    {
      "epoch": 0.20962850644427597,
      "grad_norm": 0.38242021203041077,
      "learning_rate": 0.0002,
      "loss": 0.03348962590098381,
      "step": 1106
    },
    {
      "epoch": 0.2100075815011372,
      "grad_norm": 0.15062281489372253,
      "learning_rate": 0.0002,
      "loss": 0.02110227383673191,
      "step": 1108
    },
    {
      "epoch": 0.21038665655799849,
      "grad_norm": 0.3985094428062439,
      "learning_rate": 0.0002,
      "loss": 0.030812544748187065,
      "step": 1110
    },
    {
      "epoch": 0.21076573161485973,
      "grad_norm": 0.5836614370346069,
      "learning_rate": 0.0002,
      "loss": 0.03328777477145195,
      "step": 1112
    },
    {
      "epoch": 0.211144806671721,
      "grad_norm": 0.484948068857193,
      "learning_rate": 0.0002,
      "loss": 0.04669022187590599,
      "step": 1114
    },
    {
      "epoch": 0.21152388172858225,
      "grad_norm": 0.5525700449943542,
      "learning_rate": 0.0002,
      "loss": 0.04808837175369263,
      "step": 1116
    },
    {
      "epoch": 0.21190295678544352,
      "grad_norm": 0.22838833928108215,
      "learning_rate": 0.0002,
      "loss": 0.028385240584611893,
      "step": 1118
    },
    {
      "epoch": 0.21228203184230476,
      "grad_norm": 0.18436360359191895,
      "learning_rate": 0.0002,
      "loss": 0.014112116768956184,
      "step": 1120
    },
    {
      "epoch": 0.21266110689916604,
      "grad_norm": 0.19809383153915405,
      "learning_rate": 0.0002,
      "loss": 0.0179629847407341,
      "step": 1122
    },
    {
      "epoch": 0.21304018195602728,
      "grad_norm": 0.15212291479110718,
      "learning_rate": 0.0002,
      "loss": 0.013926190324127674,
      "step": 1124
    },
    {
      "epoch": 0.21341925701288855,
      "grad_norm": 0.14080139994621277,
      "learning_rate": 0.0002,
      "loss": 0.03054637648165226,
      "step": 1126
    },
    {
      "epoch": 0.2137983320697498,
      "grad_norm": 0.30687448382377625,
      "learning_rate": 0.0002,
      "loss": 0.039678651839494705,
      "step": 1128
    },
    {
      "epoch": 0.21417740712661107,
      "grad_norm": 0.23746328055858612,
      "learning_rate": 0.0002,
      "loss": 0.0262290108948946,
      "step": 1130
    },
    {
      "epoch": 0.21455648218347234,
      "grad_norm": 0.18126142024993896,
      "learning_rate": 0.0002,
      "loss": 0.03609790652990341,
      "step": 1132
    },
    {
      "epoch": 0.21493555724033359,
      "grad_norm": 0.17882676422595978,
      "learning_rate": 0.0002,
      "loss": 0.011591329239308834,
      "step": 1134
    },
    {
      "epoch": 0.21531463229719486,
      "grad_norm": 0.22797441482543945,
      "learning_rate": 0.0002,
      "loss": 0.024538228288292885,
      "step": 1136
    },
    {
      "epoch": 0.2156937073540561,
      "grad_norm": 0.1787935495376587,
      "learning_rate": 0.0002,
      "loss": 0.023172354325652122,
      "step": 1138
    },
    {
      "epoch": 0.21607278241091737,
      "grad_norm": 0.26429131627082825,
      "learning_rate": 0.0002,
      "loss": 0.02260321006178856,
      "step": 1140
    },
    {
      "epoch": 0.21645185746777862,
      "grad_norm": 0.17137642204761505,
      "learning_rate": 0.0002,
      "loss": 0.01733425445854664,
      "step": 1142
    },
    {
      "epoch": 0.2168309325246399,
      "grad_norm": 0.2210787683725357,
      "learning_rate": 0.0002,
      "loss": 0.039520103484392166,
      "step": 1144
    },
    {
      "epoch": 0.21721000758150114,
      "grad_norm": 0.44062182307243347,
      "learning_rate": 0.0002,
      "loss": 0.0380689799785614,
      "step": 1146
    },
    {
      "epoch": 0.2175890826383624,
      "grad_norm": 0.16497430205345154,
      "learning_rate": 0.0002,
      "loss": 0.022658389061689377,
      "step": 1148
    },
    {
      "epoch": 0.21796815769522365,
      "grad_norm": 0.43936342000961304,
      "learning_rate": 0.0002,
      "loss": 0.030738933011889458,
      "step": 1150
    },
    {
      "epoch": 0.21834723275208492,
      "grad_norm": 0.46343061327934265,
      "learning_rate": 0.0002,
      "loss": 0.031555287539958954,
      "step": 1152
    },
    {
      "epoch": 0.21872630780894617,
      "grad_norm": 0.22043898701667786,
      "learning_rate": 0.0002,
      "loss": 0.032388836145401,
      "step": 1154
    },
    {
      "epoch": 0.21910538286580744,
      "grad_norm": 0.2596970796585083,
      "learning_rate": 0.0002,
      "loss": 0.01837373524904251,
      "step": 1156
    },
    {
      "epoch": 0.21948445792266869,
      "grad_norm": 0.5640380382537842,
      "learning_rate": 0.0002,
      "loss": 0.034656114876270294,
      "step": 1158
    },
    {
      "epoch": 0.21986353297952996,
      "grad_norm": 0.5247886776924133,
      "learning_rate": 0.0002,
      "loss": 0.04841337725520134,
      "step": 1160
    },
    {
      "epoch": 0.2202426080363912,
      "grad_norm": 0.16683104634284973,
      "learning_rate": 0.0002,
      "loss": 0.015878068283200264,
      "step": 1162
    },
    {
      "epoch": 0.22062168309325247,
      "grad_norm": 0.1393366903066635,
      "learning_rate": 0.0002,
      "loss": 0.0303562693297863,
      "step": 1164
    },
    {
      "epoch": 0.22100075815011372,
      "grad_norm": 0.1945030391216278,
      "learning_rate": 0.0002,
      "loss": 0.021281063556671143,
      "step": 1166
    },
    {
      "epoch": 0.221379833206975,
      "grad_norm": 0.4534865915775299,
      "learning_rate": 0.0002,
      "loss": 0.0359162837266922,
      "step": 1168
    },
    {
      "epoch": 0.22175890826383624,
      "grad_norm": 0.30880698561668396,
      "learning_rate": 0.0002,
      "loss": 0.01827920600771904,
      "step": 1170
    },
    {
      "epoch": 0.2221379833206975,
      "grad_norm": 0.396329790353775,
      "learning_rate": 0.0002,
      "loss": 0.033222660422325134,
      "step": 1172
    },
    {
      "epoch": 0.22251705837755875,
      "grad_norm": 0.2285938560962677,
      "learning_rate": 0.0002,
      "loss": 0.024670055136084557,
      "step": 1174
    },
    {
      "epoch": 0.22289613343442002,
      "grad_norm": 0.15449249744415283,
      "learning_rate": 0.0002,
      "loss": 0.014631115831434727,
      "step": 1176
    },
    {
      "epoch": 0.22327520849128127,
      "grad_norm": 0.18768326938152313,
      "learning_rate": 0.0002,
      "loss": 0.013712453655898571,
      "step": 1178
    },
    {
      "epoch": 0.22365428354814254,
      "grad_norm": 0.14142046868801117,
      "learning_rate": 0.0002,
      "loss": 0.025548120960593224,
      "step": 1180
    },
    {
      "epoch": 0.22403335860500379,
      "grad_norm": 0.32655003666877747,
      "learning_rate": 0.0002,
      "loss": 0.026756027713418007,
      "step": 1182
    },
    {
      "epoch": 0.22441243366186506,
      "grad_norm": 0.15273264050483704,
      "learning_rate": 0.0002,
      "loss": 0.029948454350233078,
      "step": 1184
    },
    {
      "epoch": 0.2247915087187263,
      "grad_norm": 0.39609622955322266,
      "learning_rate": 0.0002,
      "loss": 0.04466019570827484,
      "step": 1186
    },
    {
      "epoch": 0.22517058377558757,
      "grad_norm": 0.1447295993566513,
      "learning_rate": 0.0002,
      "loss": 0.011695250868797302,
      "step": 1188
    },
    {
      "epoch": 0.22554965883244882,
      "grad_norm": 0.37672528624534607,
      "learning_rate": 0.0002,
      "loss": 0.03341302275657654,
      "step": 1190
    },
    {
      "epoch": 0.2259287338893101,
      "grad_norm": 0.38075247406959534,
      "learning_rate": 0.0002,
      "loss": 0.031232422217726707,
      "step": 1192
    },
    {
      "epoch": 0.22630780894617133,
      "grad_norm": 0.24605412781238556,
      "learning_rate": 0.0002,
      "loss": 0.01984924077987671,
      "step": 1194
    },
    {
      "epoch": 0.2266868840030326,
      "grad_norm": 0.2939987778663635,
      "learning_rate": 0.0002,
      "loss": 0.014360908418893814,
      "step": 1196
    },
    {
      "epoch": 0.22706595905989385,
      "grad_norm": 0.14628028869628906,
      "learning_rate": 0.0002,
      "loss": 0.02225297875702381,
      "step": 1198
    },
    {
      "epoch": 0.22744503411675512,
      "grad_norm": 0.1320495903491974,
      "learning_rate": 0.0002,
      "loss": 0.013845865614712238,
      "step": 1200
    },
    {
      "epoch": 0.22782410917361637,
      "grad_norm": 0.409278005361557,
      "learning_rate": 0.0002,
      "loss": 0.03546469658613205,
      "step": 1202
    },
    {
      "epoch": 0.22820318423047764,
      "grad_norm": 0.17586016654968262,
      "learning_rate": 0.0002,
      "loss": 0.01668589375913143,
      "step": 1204
    },
    {
      "epoch": 0.22858225928733888,
      "grad_norm": 0.2557964026927948,
      "learning_rate": 0.0002,
      "loss": 0.021131785586476326,
      "step": 1206
    },
    {
      "epoch": 0.22896133434420016,
      "grad_norm": 0.4954834282398224,
      "learning_rate": 0.0002,
      "loss": 0.0572214238345623,
      "step": 1208
    },
    {
      "epoch": 0.2293404094010614,
      "grad_norm": 0.4361141324043274,
      "learning_rate": 0.0002,
      "loss": 0.02942194975912571,
      "step": 1210
    },
    {
      "epoch": 0.22971948445792267,
      "grad_norm": 0.2852785289287567,
      "learning_rate": 0.0002,
      "loss": 0.029871495440602303,
      "step": 1212
    },
    {
      "epoch": 0.23009855951478392,
      "grad_norm": 0.23743358254432678,
      "learning_rate": 0.0002,
      "loss": 0.0472852997481823,
      "step": 1214
    },
    {
      "epoch": 0.2304776345716452,
      "grad_norm": 0.328448086977005,
      "learning_rate": 0.0002,
      "loss": 0.02036426216363907,
      "step": 1216
    },
    {
      "epoch": 0.23085670962850643,
      "grad_norm": 0.2324218899011612,
      "learning_rate": 0.0002,
      "loss": 0.019669516012072563,
      "step": 1218
    },
    {
      "epoch": 0.2312357846853677,
      "grad_norm": 0.1499684900045395,
      "learning_rate": 0.0002,
      "loss": 0.03180664777755737,
      "step": 1220
    },
    {
      "epoch": 0.23161485974222895,
      "grad_norm": 0.41234803199768066,
      "learning_rate": 0.0002,
      "loss": 0.026189886033535004,
      "step": 1222
    },
    {
      "epoch": 0.23199393479909022,
      "grad_norm": 0.21643425524234772,
      "learning_rate": 0.0002,
      "loss": 0.023773767054080963,
      "step": 1224
    },
    {
      "epoch": 0.23237300985595147,
      "grad_norm": 0.2384602129459381,
      "learning_rate": 0.0002,
      "loss": 0.016462331637740135,
      "step": 1226
    },
    {
      "epoch": 0.23275208491281274,
      "grad_norm": 0.4421197772026062,
      "learning_rate": 0.0002,
      "loss": 0.039935361593961716,
      "step": 1228
    },
    {
      "epoch": 0.23313115996967398,
      "grad_norm": 0.24957142770290375,
      "learning_rate": 0.0002,
      "loss": 0.013218197971582413,
      "step": 1230
    },
    {
      "epoch": 0.23351023502653526,
      "grad_norm": 0.4590010344982147,
      "learning_rate": 0.0002,
      "loss": 0.03060908429324627,
      "step": 1232
    },
    {
      "epoch": 0.2338893100833965,
      "grad_norm": 0.29953858256340027,
      "learning_rate": 0.0002,
      "loss": 0.03190047666430473,
      "step": 1234
    },
    {
      "epoch": 0.23426838514025777,
      "grad_norm": 0.5027996897697449,
      "learning_rate": 0.0002,
      "loss": 0.04100735858082771,
      "step": 1236
    },
    {
      "epoch": 0.23464746019711902,
      "grad_norm": 0.20189468562602997,
      "learning_rate": 0.0002,
      "loss": 0.02487613819539547,
      "step": 1238
    },
    {
      "epoch": 0.2350265352539803,
      "grad_norm": 0.782785952091217,
      "learning_rate": 0.0002,
      "loss": 0.05953974649310112,
      "step": 1240
    },
    {
      "epoch": 0.23540561031084153,
      "grad_norm": 0.12561416625976562,
      "learning_rate": 0.0002,
      "loss": 0.01429492887109518,
      "step": 1242
    },
    {
      "epoch": 0.2357846853677028,
      "grad_norm": 0.12042157351970673,
      "learning_rate": 0.0002,
      "loss": 0.027772052213549614,
      "step": 1244
    },
    {
      "epoch": 0.23616376042456405,
      "grad_norm": 0.5351229906082153,
      "learning_rate": 0.0002,
      "loss": 0.044081274420022964,
      "step": 1246
    },
    {
      "epoch": 0.23654283548142532,
      "grad_norm": 0.24318385124206543,
      "learning_rate": 0.0002,
      "loss": 0.020654842257499695,
      "step": 1248
    },
    {
      "epoch": 0.23692191053828657,
      "grad_norm": 0.1350153684616089,
      "learning_rate": 0.0002,
      "loss": 0.014363355003297329,
      "step": 1250
    },
    {
      "epoch": 0.23730098559514784,
      "grad_norm": 0.3868752717971802,
      "learning_rate": 0.0002,
      "loss": 0.021097306162118912,
      "step": 1252
    },
    {
      "epoch": 0.23768006065200908,
      "grad_norm": 0.5173314809799194,
      "learning_rate": 0.0002,
      "loss": 0.032926421612501144,
      "step": 1254
    },
    {
      "epoch": 0.23805913570887036,
      "grad_norm": 0.21989795565605164,
      "learning_rate": 0.0002,
      "loss": 0.016411757096648216,
      "step": 1256
    },
    {
      "epoch": 0.23843821076573163,
      "grad_norm": 0.6634283661842346,
      "learning_rate": 0.0002,
      "loss": 0.02098790369927883,
      "step": 1258
    },
    {
      "epoch": 0.23881728582259287,
      "grad_norm": 0.45667916536331177,
      "learning_rate": 0.0002,
      "loss": 0.03331917151808739,
      "step": 1260
    },
    {
      "epoch": 0.23919636087945415,
      "grad_norm": 0.11330067366361618,
      "learning_rate": 0.0002,
      "loss": 0.02642682194709778,
      "step": 1262
    },
    {
      "epoch": 0.2395754359363154,
      "grad_norm": 0.13656266033649445,
      "learning_rate": 0.0002,
      "loss": 0.012369459494948387,
      "step": 1264
    },
    {
      "epoch": 0.23995451099317666,
      "grad_norm": 0.16169367730617523,
      "learning_rate": 0.0002,
      "loss": 0.020634982734918594,
      "step": 1266
    },
    {
      "epoch": 0.2403335860500379,
      "grad_norm": 0.1562567800283432,
      "learning_rate": 0.0002,
      "loss": 0.012752935290336609,
      "step": 1268
    },
    {
      "epoch": 0.24071266110689918,
      "grad_norm": 0.3609163165092468,
      "learning_rate": 0.0002,
      "loss": 0.04412787780165672,
      "step": 1270
    },
    {
      "epoch": 0.24109173616376042,
      "grad_norm": 0.16767564415931702,
      "learning_rate": 0.0002,
      "loss": 0.027312785387039185,
      "step": 1272
    },
    {
      "epoch": 0.2414708112206217,
      "grad_norm": 0.2709888219833374,
      "learning_rate": 0.0002,
      "loss": 0.017061755061149597,
      "step": 1274
    },
    {
      "epoch": 0.24184988627748294,
      "grad_norm": 0.2680244445800781,
      "learning_rate": 0.0002,
      "loss": 0.022410478442907333,
      "step": 1276
    },
    {
      "epoch": 0.2422289613343442,
      "grad_norm": 0.12890921533107758,
      "learning_rate": 0.0002,
      "loss": 0.014289403334259987,
      "step": 1278
    },
    {
      "epoch": 0.24260803639120546,
      "grad_norm": 0.10968673974275589,
      "learning_rate": 0.0002,
      "loss": 0.029877368360757828,
      "step": 1280
    },
    {
      "epoch": 0.24298711144806673,
      "grad_norm": 0.5929157137870789,
      "learning_rate": 0.0002,
      "loss": 0.03570236638188362,
      "step": 1282
    },
    {
      "epoch": 0.24336618650492797,
      "grad_norm": 0.42857807874679565,
      "learning_rate": 0.0002,
      "loss": 0.019389566034078598,
      "step": 1284
    },
    {
      "epoch": 0.24374526156178924,
      "grad_norm": 0.14969076216220856,
      "learning_rate": 0.0002,
      "loss": 0.013053362257778645,
      "step": 1286
    },
    {
      "epoch": 0.2441243366186505,
      "grad_norm": 0.5093541145324707,
      "learning_rate": 0.0002,
      "loss": 0.04423750564455986,
      "step": 1288
    },
    {
      "epoch": 0.24450341167551176,
      "grad_norm": 0.4911527931690216,
      "learning_rate": 0.0002,
      "loss": 0.040008723735809326,
      "step": 1290
    },
    {
      "epoch": 0.244882486732373,
      "grad_norm": 0.36300885677337646,
      "learning_rate": 0.0002,
      "loss": 0.026185927912592888,
      "step": 1292
    },
    {
      "epoch": 0.24526156178923428,
      "grad_norm": 0.23996871709823608,
      "learning_rate": 0.0002,
      "loss": 0.02597566321492195,
      "step": 1294
    },
    {
      "epoch": 0.24564063684609552,
      "grad_norm": 0.219227135181427,
      "learning_rate": 0.0002,
      "loss": 0.03484094887971878,
      "step": 1296
    },
    {
      "epoch": 0.2460197119029568,
      "grad_norm": 0.15526732802391052,
      "learning_rate": 0.0002,
      "loss": 0.014038034714758396,
      "step": 1298
    },
    {
      "epoch": 0.24639878695981804,
      "grad_norm": 0.17858320474624634,
      "learning_rate": 0.0002,
      "loss": 0.020416291430592537,
      "step": 1300
    },
    {
      "epoch": 0.2467778620166793,
      "grad_norm": 0.5510589480400085,
      "learning_rate": 0.0002,
      "loss": 0.05620286241173744,
      "step": 1302
    },
    {
      "epoch": 0.24715693707354056,
      "grad_norm": 0.1669646054506302,
      "learning_rate": 0.0002,
      "loss": 0.015420370735228062,
      "step": 1304
    },
    {
      "epoch": 0.24753601213040183,
      "grad_norm": 0.09812512993812561,
      "learning_rate": 0.0002,
      "loss": 0.01996256783604622,
      "step": 1306
    },
    {
      "epoch": 0.24791508718726307,
      "grad_norm": 0.34046483039855957,
      "learning_rate": 0.0002,
      "loss": 0.02475627511739731,
      "step": 1308
    },
    {
      "epoch": 0.24829416224412434,
      "grad_norm": 0.2773810029029846,
      "learning_rate": 0.0002,
      "loss": 0.01753375492990017,
      "step": 1310
    },
    {
      "epoch": 0.2486732373009856,
      "grad_norm": 0.5444158315658569,
      "learning_rate": 0.0002,
      "loss": 0.04099429026246071,
      "step": 1312
    },
    {
      "epoch": 0.24905231235784686,
      "grad_norm": 0.3318602442741394,
      "learning_rate": 0.0002,
      "loss": 0.028686461970210075,
      "step": 1314
    },
    {
      "epoch": 0.2494313874147081,
      "grad_norm": 0.3848995864391327,
      "learning_rate": 0.0002,
      "loss": 0.04933010786771774,
      "step": 1316
    },
    {
      "epoch": 0.24981046247156938,
      "grad_norm": 0.13651327788829803,
      "learning_rate": 0.0002,
      "loss": 0.028859058395028114,
      "step": 1318
    },
    {
      "epoch": 0.25018953752843065,
      "grad_norm": 0.5570463538169861,
      "learning_rate": 0.0002,
      "loss": 0.053565751761198044,
      "step": 1320
    },
    {
      "epoch": 0.2505686125852919,
      "grad_norm": 0.20876778662204742,
      "learning_rate": 0.0002,
      "loss": 0.015588083304464817,
      "step": 1322
    },
    {
      "epoch": 0.25094768764215314,
      "grad_norm": 0.5141584277153015,
      "learning_rate": 0.0002,
      "loss": 0.04471775144338608,
      "step": 1324
    },
    {
      "epoch": 0.2513267626990144,
      "grad_norm": 0.1515473872423172,
      "learning_rate": 0.0002,
      "loss": 0.028739696368575096,
      "step": 1326
    },
    {
      "epoch": 0.2517058377558757,
      "grad_norm": 0.33069488406181335,
      "learning_rate": 0.0002,
      "loss": 0.020965704694390297,
      "step": 1328
    },
    {
      "epoch": 0.2520849128127369,
      "grad_norm": 0.18122830986976624,
      "learning_rate": 0.0002,
      "loss": 0.01566014438867569,
      "step": 1330
    },
    {
      "epoch": 0.25246398786959817,
      "grad_norm": 0.4942258298397064,
      "learning_rate": 0.0002,
      "loss": 0.03379614278674126,
      "step": 1332
    },
    {
      "epoch": 0.2528430629264594,
      "grad_norm": 0.17477580904960632,
      "learning_rate": 0.0002,
      "loss": 0.012574037536978722,
      "step": 1334
    },
    {
      "epoch": 0.2532221379833207,
      "grad_norm": 0.45184680819511414,
      "learning_rate": 0.0002,
      "loss": 0.029008837416768074,
      "step": 1336
    },
    {
      "epoch": 0.25360121304018196,
      "grad_norm": 0.37187090516090393,
      "learning_rate": 0.0002,
      "loss": 0.04534810036420822,
      "step": 1338
    },
    {
      "epoch": 0.2539802880970432,
      "grad_norm": 0.1438458114862442,
      "learning_rate": 0.0002,
      "loss": 0.01356600783765316,
      "step": 1340
    },
    {
      "epoch": 0.25435936315390445,
      "grad_norm": 0.2622271180152893,
      "learning_rate": 0.0002,
      "loss": 0.033437155187129974,
      "step": 1342
    },
    {
      "epoch": 0.25473843821076575,
      "grad_norm": 0.14569099247455597,
      "learning_rate": 0.0002,
      "loss": 0.018014775589108467,
      "step": 1344
    },
    {
      "epoch": 0.255117513267627,
      "grad_norm": 0.3481868505477905,
      "learning_rate": 0.0002,
      "loss": 0.04088563844561577,
      "step": 1346
    },
    {
      "epoch": 0.25549658832448824,
      "grad_norm": 0.12131579965353012,
      "learning_rate": 0.0002,
      "loss": 0.024967307224869728,
      "step": 1348
    },
    {
      "epoch": 0.2558756633813495,
      "grad_norm": 0.12678959965705872,
      "learning_rate": 0.0002,
      "loss": 0.01493467204272747,
      "step": 1350
    },
    {
      "epoch": 0.2562547384382108,
      "grad_norm": 0.3829241096973419,
      "learning_rate": 0.0002,
      "loss": 0.03174349293112755,
      "step": 1352
    },
    {
      "epoch": 0.256633813495072,
      "grad_norm": 0.3604947328567505,
      "learning_rate": 0.0002,
      "loss": 0.025653939694166183,
      "step": 1354
    },
    {
      "epoch": 0.25701288855193327,
      "grad_norm": 0.1305903196334839,
      "learning_rate": 0.0002,
      "loss": 0.01280367374420166,
      "step": 1356
    },
    {
      "epoch": 0.2573919636087945,
      "grad_norm": 0.17293137311935425,
      "learning_rate": 0.0002,
      "loss": 0.012824755162000656,
      "step": 1358
    },
    {
      "epoch": 0.2577710386656558,
      "grad_norm": 0.12544064223766327,
      "learning_rate": 0.0002,
      "loss": 0.01408946793526411,
      "step": 1360
    },
    {
      "epoch": 0.25815011372251706,
      "grad_norm": 0.3684629499912262,
      "learning_rate": 0.0002,
      "loss": 0.019628416746854782,
      "step": 1362
    },
    {
      "epoch": 0.2585291887793783,
      "grad_norm": 0.1704927533864975,
      "learning_rate": 0.0002,
      "loss": 0.014785442501306534,
      "step": 1364
    },
    {
      "epoch": 0.25890826383623955,
      "grad_norm": 0.1462058275938034,
      "learning_rate": 0.0002,
      "loss": 0.029853198677301407,
      "step": 1366
    },
    {
      "epoch": 0.25928733889310085,
      "grad_norm": 0.41396602988243103,
      "learning_rate": 0.0002,
      "loss": 0.036266375333070755,
      "step": 1368
    },
    {
      "epoch": 0.2596664139499621,
      "grad_norm": 0.24247600138187408,
      "learning_rate": 0.0002,
      "loss": 0.01800282672047615,
      "step": 1370
    },
    {
      "epoch": 0.26004548900682334,
      "grad_norm": 0.14393870532512665,
      "learning_rate": 0.0002,
      "loss": 0.022550111636519432,
      "step": 1372
    },
    {
      "epoch": 0.2604245640636846,
      "grad_norm": 0.46248528361320496,
      "learning_rate": 0.0002,
      "loss": 0.0335487499833107,
      "step": 1374
    },
    {
      "epoch": 0.2608036391205459,
      "grad_norm": 0.3807598650455475,
      "learning_rate": 0.0002,
      "loss": 0.032364603132009506,
      "step": 1376
    },
    {
      "epoch": 0.2611827141774071,
      "grad_norm": 0.37710118293762207,
      "learning_rate": 0.0002,
      "loss": 0.02654591202735901,
      "step": 1378
    },
    {
      "epoch": 0.26156178923426837,
      "grad_norm": 0.11521491408348083,
      "learning_rate": 0.0002,
      "loss": 0.014118420891463757,
      "step": 1380
    },
    {
      "epoch": 0.26194086429112967,
      "grad_norm": 0.46899598836898804,
      "learning_rate": 0.0002,
      "loss": 0.028397759422659874,
      "step": 1382
    },
    {
      "epoch": 0.2623199393479909,
      "grad_norm": 0.13468901813030243,
      "learning_rate": 0.0002,
      "loss": 0.03262486681342125,
      "step": 1384
    },
    {
      "epoch": 0.26269901440485216,
      "grad_norm": 0.33410701155662537,
      "learning_rate": 0.0002,
      "loss": 0.019016897305846214,
      "step": 1386
    },
    {
      "epoch": 0.2630780894617134,
      "grad_norm": 0.2950461804866791,
      "learning_rate": 0.0002,
      "loss": 0.026379629969596863,
      "step": 1388
    },
    {
      "epoch": 0.2634571645185747,
      "grad_norm": 0.46277308464050293,
      "learning_rate": 0.0002,
      "loss": 0.02824045531451702,
      "step": 1390
    },
    {
      "epoch": 0.26383623957543595,
      "grad_norm": 0.41617241501808167,
      "learning_rate": 0.0002,
      "loss": 0.0297885499894619,
      "step": 1392
    },
    {
      "epoch": 0.2642153146322972,
      "grad_norm": 0.12658989429473877,
      "learning_rate": 0.0002,
      "loss": 0.032457273453474045,
      "step": 1394
    },
    {
      "epoch": 0.26459438968915844,
      "grad_norm": 0.25802353024482727,
      "learning_rate": 0.0002,
      "loss": 0.02149409055709839,
      "step": 1396
    },
    {
      "epoch": 0.26497346474601974,
      "grad_norm": 0.47693750262260437,
      "learning_rate": 0.0002,
      "loss": 0.02975631318986416,
      "step": 1398
    },
    {
      "epoch": 0.265352539802881,
      "grad_norm": 0.37225547432899475,
      "learning_rate": 0.0002,
      "loss": 0.032973483204841614,
      "step": 1400
    },
    {
      "epoch": 0.2657316148597422,
      "grad_norm": 0.36715471744537354,
      "learning_rate": 0.0002,
      "loss": 0.02206313982605934,
      "step": 1402
    },
    {
      "epoch": 0.26611068991660347,
      "grad_norm": 0.16459710896015167,
      "learning_rate": 0.0002,
      "loss": 0.023533688858151436,
      "step": 1404
    },
    {
      "epoch": 0.26648976497346477,
      "grad_norm": 0.5138043761253357,
      "learning_rate": 0.0002,
      "loss": 0.04955093562602997,
      "step": 1406
    },
    {
      "epoch": 0.266868840030326,
      "grad_norm": 0.1723688542842865,
      "learning_rate": 0.0002,
      "loss": 0.014032145962119102,
      "step": 1408
    },
    {
      "epoch": 0.26724791508718726,
      "grad_norm": 0.09199069440364838,
      "learning_rate": 0.0002,
      "loss": 0.016924314200878143,
      "step": 1410
    },
    {
      "epoch": 0.2676269901440485,
      "grad_norm": 0.4538004994392395,
      "learning_rate": 0.0002,
      "loss": 0.05237695202231407,
      "step": 1412
    },
    {
      "epoch": 0.2680060652009098,
      "grad_norm": 0.1579502671957016,
      "learning_rate": 0.0002,
      "loss": 0.014107112772762775,
      "step": 1414
    },
    {
      "epoch": 0.26838514025777105,
      "grad_norm": 0.12711958587169647,
      "learning_rate": 0.0002,
      "loss": 0.02932284027338028,
      "step": 1416
    },
    {
      "epoch": 0.2687642153146323,
      "grad_norm": 0.2655789256095886,
      "learning_rate": 0.0002,
      "loss": 0.028610428795218468,
      "step": 1418
    },
    {
      "epoch": 0.26914329037149354,
      "grad_norm": 0.43105027079582214,
      "learning_rate": 0.0002,
      "loss": 0.04115430638194084,
      "step": 1420
    },
    {
      "epoch": 0.26952236542835484,
      "grad_norm": 0.41622909903526306,
      "learning_rate": 0.0002,
      "loss": 0.040929585695266724,
      "step": 1422
    },
    {
      "epoch": 0.2699014404852161,
      "grad_norm": 0.33965978026390076,
      "learning_rate": 0.0002,
      "loss": 0.021732809022068977,
      "step": 1424
    },
    {
      "epoch": 0.2702805155420773,
      "grad_norm": 0.3470750153064728,
      "learning_rate": 0.0002,
      "loss": 0.03712059557437897,
      "step": 1426
    },
    {
      "epoch": 0.27065959059893857,
      "grad_norm": 0.33910220861434937,
      "learning_rate": 0.0002,
      "loss": 0.01736084558069706,
      "step": 1428
    },
    {
      "epoch": 0.27103866565579987,
      "grad_norm": 0.5011537075042725,
      "learning_rate": 0.0002,
      "loss": 0.03664974495768547,
      "step": 1430
    },
    {
      "epoch": 0.2714177407126611,
      "grad_norm": 0.29871681332588196,
      "learning_rate": 0.0002,
      "loss": 0.02039661817252636,
      "step": 1432
    },
    {
      "epoch": 0.27179681576952236,
      "grad_norm": 0.12158619612455368,
      "learning_rate": 0.0002,
      "loss": 0.01439282763749361,
      "step": 1434
    },
    {
      "epoch": 0.2721758908263836,
      "grad_norm": 0.2847796380519867,
      "learning_rate": 0.0002,
      "loss": 0.04608976095914841,
      "step": 1436
    },
    {
      "epoch": 0.2725549658832449,
      "grad_norm": 0.12210452556610107,
      "learning_rate": 0.0002,
      "loss": 0.019499335438013077,
      "step": 1438
    },
    {
      "epoch": 0.27293404094010615,
      "grad_norm": 0.19829772412776947,
      "learning_rate": 0.0002,
      "loss": 0.012894745916128159,
      "step": 1440
    },
    {
      "epoch": 0.2733131159969674,
      "grad_norm": 0.4982815384864807,
      "learning_rate": 0.0002,
      "loss": 0.029639270156621933,
      "step": 1442
    },
    {
      "epoch": 0.27369219105382864,
      "grad_norm": 0.31689736247062683,
      "learning_rate": 0.0002,
      "loss": 0.023977773264050484,
      "step": 1444
    },
    {
      "epoch": 0.27407126611068994,
      "grad_norm": 0.20953544974327087,
      "learning_rate": 0.0002,
      "loss": 0.030343137681484222,
      "step": 1446
    },
    {
      "epoch": 0.2744503411675512,
      "grad_norm": 0.2163022756576538,
      "learning_rate": 0.0002,
      "loss": 0.018982771784067154,
      "step": 1448
    },
    {
      "epoch": 0.2748294162244124,
      "grad_norm": 0.4207800328731537,
      "learning_rate": 0.0002,
      "loss": 0.028210869058966637,
      "step": 1450
    },
    {
      "epoch": 0.27520849128127367,
      "grad_norm": 0.5213370323181152,
      "learning_rate": 0.0002,
      "loss": 0.05761299282312393,
      "step": 1452
    },
    {
      "epoch": 0.27558756633813497,
      "grad_norm": 0.2046635001897812,
      "learning_rate": 0.0002,
      "loss": 0.012518886476755142,
      "step": 1454
    },
    {
      "epoch": 0.2759666413949962,
      "grad_norm": 0.29638004302978516,
      "learning_rate": 0.0002,
      "loss": 0.02391839027404785,
      "step": 1456
    },
    {
      "epoch": 0.27634571645185746,
      "grad_norm": 0.45606544613838196,
      "learning_rate": 0.0002,
      "loss": 0.034800246357917786,
      "step": 1458
    },
    {
      "epoch": 0.2767247915087187,
      "grad_norm": 0.12763471901416779,
      "learning_rate": 0.0002,
      "loss": 0.01923462748527527,
      "step": 1460
    },
    {
      "epoch": 0.27710386656558,
      "grad_norm": 0.15307757258415222,
      "learning_rate": 0.0002,
      "loss": 0.026512853801250458,
      "step": 1462
    },
    {
      "epoch": 0.27748294162244125,
      "grad_norm": 0.41160839796066284,
      "learning_rate": 0.0002,
      "loss": 0.028861582279205322,
      "step": 1464
    },
    {
      "epoch": 0.2778620166793025,
      "grad_norm": 0.22475704550743103,
      "learning_rate": 0.0002,
      "loss": 0.01705140620470047,
      "step": 1466
    },
    {
      "epoch": 0.27824109173616374,
      "grad_norm": 0.14492644369602203,
      "learning_rate": 0.0002,
      "loss": 0.02556610479950905,
      "step": 1468
    },
    {
      "epoch": 0.27862016679302504,
      "grad_norm": 0.5987374782562256,
      "learning_rate": 0.0002,
      "loss": 0.03936241567134857,
      "step": 1470
    },
    {
      "epoch": 0.2789992418498863,
      "grad_norm": 0.17927633225917816,
      "learning_rate": 0.0002,
      "loss": 0.014172183349728584,
      "step": 1472
    },
    {
      "epoch": 0.2793783169067475,
      "grad_norm": 0.46151667833328247,
      "learning_rate": 0.0002,
      "loss": 0.03046492673456669,
      "step": 1474
    },
    {
      "epoch": 0.27975739196360877,
      "grad_norm": 0.14469172060489655,
      "learning_rate": 0.0002,
      "loss": 0.015109979547560215,
      "step": 1476
    },
    {
      "epoch": 0.28013646702047007,
      "grad_norm": 0.37196364998817444,
      "learning_rate": 0.0002,
      "loss": 0.03070731833577156,
      "step": 1478
    },
    {
      "epoch": 0.2805155420773313,
      "grad_norm": 0.3579059839248657,
      "learning_rate": 0.0002,
      "loss": 0.03069712221622467,
      "step": 1480
    },
    {
      "epoch": 0.28089461713419256,
      "grad_norm": 0.12433185428380966,
      "learning_rate": 0.0002,
      "loss": 0.023986034095287323,
      "step": 1482
    },
    {
      "epoch": 0.2812736921910538,
      "grad_norm": 0.334940105676651,
      "learning_rate": 0.0002,
      "loss": 0.026163354516029358,
      "step": 1484
    },
    {
      "epoch": 0.2816527672479151,
      "grad_norm": 0.1390274465084076,
      "learning_rate": 0.0002,
      "loss": 0.025685563683509827,
      "step": 1486
    },
    {
      "epoch": 0.28203184230477635,
      "grad_norm": 0.4700069725513458,
      "learning_rate": 0.0002,
      "loss": 0.023738304153084755,
      "step": 1488
    },
    {
      "epoch": 0.2824109173616376,
      "grad_norm": 0.22838054597377777,
      "learning_rate": 0.0002,
      "loss": 0.030710551887750626,
      "step": 1490
    },
    {
      "epoch": 0.28278999241849884,
      "grad_norm": 0.4302392601966858,
      "learning_rate": 0.0002,
      "loss": 0.01418632734566927,
      "step": 1492
    },
    {
      "epoch": 0.28316906747536014,
      "grad_norm": 0.27317333221435547,
      "learning_rate": 0.0002,
      "loss": 0.04203774034976959,
      "step": 1494
    },
    {
      "epoch": 0.2835481425322214,
      "grad_norm": 0.23483392596244812,
      "learning_rate": 0.0002,
      "loss": 0.025907479226589203,
      "step": 1496
    },
    {
      "epoch": 0.2839272175890826,
      "grad_norm": 0.24170690774917603,
      "learning_rate": 0.0002,
      "loss": 0.03571818023920059,
      "step": 1498
    },
    {
      "epoch": 0.28430629264594387,
      "grad_norm": 0.41121146082878113,
      "learning_rate": 0.0002,
      "loss": 0.027514813467860222,
      "step": 1500
    },
    {
      "epoch": 0.28468536770280517,
      "grad_norm": 0.25932326912879944,
      "learning_rate": 0.0002,
      "loss": 0.016760284081101418,
      "step": 1502
    },
    {
      "epoch": 0.2850644427596664,
      "grad_norm": 0.5006921887397766,
      "learning_rate": 0.0002,
      "loss": 0.04705590382218361,
      "step": 1504
    },
    {
      "epoch": 0.28544351781652766,
      "grad_norm": 0.16624368727207184,
      "learning_rate": 0.0002,
      "loss": 0.03840074688196182,
      "step": 1506
    },
    {
      "epoch": 0.28582259287338896,
      "grad_norm": 0.2899216115474701,
      "learning_rate": 0.0002,
      "loss": 0.019632529467344284,
      "step": 1508
    },
    {
      "epoch": 0.2862016679302502,
      "grad_norm": 0.3590693771839142,
      "learning_rate": 0.0002,
      "loss": 0.02993917651474476,
      "step": 1510
    },
    {
      "epoch": 0.28658074298711145,
      "grad_norm": 0.1543245166540146,
      "learning_rate": 0.0002,
      "loss": 0.015201310627162457,
      "step": 1512
    },
    {
      "epoch": 0.2869598180439727,
      "grad_norm": 0.4352371394634247,
      "learning_rate": 0.0002,
      "loss": 0.026553241536021233,
      "step": 1514
    },
    {
      "epoch": 0.287338893100834,
      "grad_norm": 0.45238736271858215,
      "learning_rate": 0.0002,
      "loss": 0.03952331095933914,
      "step": 1516
    },
    {
      "epoch": 0.28771796815769524,
      "grad_norm": 0.4397946298122406,
      "learning_rate": 0.0002,
      "loss": 0.03599841147661209,
      "step": 1518
    },
    {
      "epoch": 0.2880970432145565,
      "grad_norm": 0.12472062557935715,
      "learning_rate": 0.0002,
      "loss": 0.01927502080798149,
      "step": 1520
    },
    {
      "epoch": 0.2884761182714177,
      "grad_norm": 0.3964513838291168,
      "learning_rate": 0.0002,
      "loss": 0.029821723699569702,
      "step": 1522
    },
    {
      "epoch": 0.288855193328279,
      "grad_norm": 0.4703337252140045,
      "learning_rate": 0.0002,
      "loss": 0.034038856625556946,
      "step": 1524
    },
    {
      "epoch": 0.28923426838514027,
      "grad_norm": 0.35713642835617065,
      "learning_rate": 0.0002,
      "loss": 0.034018319100141525,
      "step": 1526
    },
    {
      "epoch": 0.2896133434420015,
      "grad_norm": 0.28145450353622437,
      "learning_rate": 0.0002,
      "loss": 0.037787653505802155,
      "step": 1528
    },
    {
      "epoch": 0.28999241849886276,
      "grad_norm": 0.1263909786939621,
      "learning_rate": 0.0002,
      "loss": 0.020594941452145576,
      "step": 1530
    },
    {
      "epoch": 0.29037149355572406,
      "grad_norm": 0.41144609451293945,
      "learning_rate": 0.0002,
      "loss": 0.028030630201101303,
      "step": 1532
    },
    {
      "epoch": 0.2907505686125853,
      "grad_norm": 0.3816780149936676,
      "learning_rate": 0.0002,
      "loss": 0.05349470674991608,
      "step": 1534
    },
    {
      "epoch": 0.29112964366944655,
      "grad_norm": 0.4808156192302704,
      "learning_rate": 0.0002,
      "loss": 0.035557426512241364,
      "step": 1536
    },
    {
      "epoch": 0.2915087187263078,
      "grad_norm": 0.4582657217979431,
      "learning_rate": 0.0002,
      "loss": 0.050364576280117035,
      "step": 1538
    },
    {
      "epoch": 0.2918877937831691,
      "grad_norm": 0.5775293111801147,
      "learning_rate": 0.0002,
      "loss": 0.04030098393559456,
      "step": 1540
    },
    {
      "epoch": 0.29226686884003034,
      "grad_norm": 0.4349782466888428,
      "learning_rate": 0.0002,
      "loss": 0.03602564334869385,
      "step": 1542
    },
    {
      "epoch": 0.2926459438968916,
      "grad_norm": 0.5109832882881165,
      "learning_rate": 0.0002,
      "loss": 0.03534889593720436,
      "step": 1544
    },
    {
      "epoch": 0.2930250189537528,
      "grad_norm": 0.142647385597229,
      "learning_rate": 0.0002,
      "loss": 0.02213236689567566,
      "step": 1546
    },
    {
      "epoch": 0.2934040940106141,
      "grad_norm": 0.5220096111297607,
      "learning_rate": 0.0002,
      "loss": 0.042803917080163956,
      "step": 1548
    },
    {
      "epoch": 0.29378316906747537,
      "grad_norm": 0.23127217590808868,
      "learning_rate": 0.0002,
      "loss": 0.015523104928433895,
      "step": 1550
    },
    {
      "epoch": 0.2941622441243366,
      "grad_norm": 0.45516011118888855,
      "learning_rate": 0.0002,
      "loss": 0.029321083799004555,
      "step": 1552
    },
    {
      "epoch": 0.29454131918119786,
      "grad_norm": 0.4499432146549225,
      "learning_rate": 0.0002,
      "loss": 0.03624111786484718,
      "step": 1554
    },
    {
      "epoch": 0.29492039423805916,
      "grad_norm": 0.4849935472011566,
      "learning_rate": 0.0002,
      "loss": 0.04496414214372635,
      "step": 1556
    },
    {
      "epoch": 0.2952994692949204,
      "grad_norm": 0.4390590786933899,
      "learning_rate": 0.0002,
      "loss": 0.03553665429353714,
      "step": 1558
    },
    {
      "epoch": 0.29567854435178165,
      "grad_norm": 0.15762703120708466,
      "learning_rate": 0.0002,
      "loss": 0.026528365910053253,
      "step": 1560
    },
    {
      "epoch": 0.2960576194086429,
      "grad_norm": 0.4439156651496887,
      "learning_rate": 0.0002,
      "loss": 0.03529759868979454,
      "step": 1562
    },
    {
      "epoch": 0.2964366944655042,
      "grad_norm": 0.11462212353944778,
      "learning_rate": 0.0002,
      "loss": 0.015349839814007282,
      "step": 1564
    },
    {
      "epoch": 0.29681576952236544,
      "grad_norm": 0.4126538932323456,
      "learning_rate": 0.0002,
      "loss": 0.029474152252078056,
      "step": 1566
    },
    {
      "epoch": 0.2971948445792267,
      "grad_norm": 0.43119680881500244,
      "learning_rate": 0.0002,
      "loss": 0.029810212552547455,
      "step": 1568
    },
    {
      "epoch": 0.2975739196360879,
      "grad_norm": 0.3878445327281952,
      "learning_rate": 0.0002,
      "loss": 0.027787763625383377,
      "step": 1570
    },
    {
      "epoch": 0.2979529946929492,
      "grad_norm": 0.13126155734062195,
      "learning_rate": 0.0002,
      "loss": 0.014948749914765358,
      "step": 1572
    },
    {
      "epoch": 0.29833206974981047,
      "grad_norm": 0.1286592185497284,
      "learning_rate": 0.0002,
      "loss": 0.027014950290322304,
      "step": 1574
    },
    {
      "epoch": 0.2987111448066717,
      "grad_norm": 0.32967162132263184,
      "learning_rate": 0.0002,
      "loss": 0.02137107402086258,
      "step": 1576
    },
    {
      "epoch": 0.29909021986353296,
      "grad_norm": 0.4696953594684601,
      "learning_rate": 0.0002,
      "loss": 0.05843757092952728,
      "step": 1578
    },
    {
      "epoch": 0.29946929492039426,
      "grad_norm": 0.16578447818756104,
      "learning_rate": 0.0002,
      "loss": 0.015499980188906193,
      "step": 1580
    },
    {
      "epoch": 0.2998483699772555,
      "grad_norm": 0.3015766739845276,
      "learning_rate": 0.0002,
      "loss": 0.034838125109672546,
      "step": 1582
    },
    {
      "epoch": 0.30022744503411675,
      "grad_norm": 0.4285444915294647,
      "learning_rate": 0.0002,
      "loss": 0.028759334236383438,
      "step": 1584
    },
    {
      "epoch": 0.300606520090978,
      "grad_norm": 0.23633123934268951,
      "learning_rate": 0.0002,
      "loss": 0.015604455024003983,
      "step": 1586
    },
    {
      "epoch": 0.3009855951478393,
      "grad_norm": 0.10616131126880646,
      "learning_rate": 0.0002,
      "loss": 0.017495624721050262,
      "step": 1588
    },
    {
      "epoch": 0.30136467020470054,
      "grad_norm": 0.19752971827983856,
      "learning_rate": 0.0002,
      "loss": 0.023869015276432037,
      "step": 1590
    },
    {
      "epoch": 0.3017437452615618,
      "grad_norm": 0.14859068393707275,
      "learning_rate": 0.0002,
      "loss": 0.01664596237242222,
      "step": 1592
    },
    {
      "epoch": 0.302122820318423,
      "grad_norm": 0.1894073337316513,
      "learning_rate": 0.0002,
      "loss": 0.014259884133934975,
      "step": 1594
    },
    {
      "epoch": 0.3025018953752843,
      "grad_norm": 0.46087679266929626,
      "learning_rate": 0.0002,
      "loss": 0.06328590959310532,
      "step": 1596
    },
    {
      "epoch": 0.30288097043214557,
      "grad_norm": 0.42721787095069885,
      "learning_rate": 0.0002,
      "loss": 0.03390883281826973,
      "step": 1598
    },
    {
      "epoch": 0.3032600454890068,
      "grad_norm": 0.10636258870363235,
      "learning_rate": 0.0002,
      "loss": 0.021469537168741226,
      "step": 1600
    },
    {
      "epoch": 0.30363912054586806,
      "grad_norm": 0.44673389196395874,
      "learning_rate": 0.0002,
      "loss": 0.055185429751873016,
      "step": 1602
    },
    {
      "epoch": 0.30401819560272936,
      "grad_norm": 0.3242854177951813,
      "learning_rate": 0.0002,
      "loss": 0.038696642965078354,
      "step": 1604
    },
    {
      "epoch": 0.3043972706595906,
      "grad_norm": 0.14299720525741577,
      "learning_rate": 0.0002,
      "loss": 0.02176748216152191,
      "step": 1606
    },
    {
      "epoch": 0.30477634571645185,
      "grad_norm": 0.3609693944454193,
      "learning_rate": 0.0002,
      "loss": 0.02342384122312069,
      "step": 1608
    },
    {
      "epoch": 0.3051554207733131,
      "grad_norm": 0.3814120590686798,
      "learning_rate": 0.0002,
      "loss": 0.03696173056960106,
      "step": 1610
    },
    {
      "epoch": 0.3055344958301744,
      "grad_norm": 0.15312209725379944,
      "learning_rate": 0.0002,
      "loss": 0.021324345842003822,
      "step": 1612
    },
    {
      "epoch": 0.30591357088703564,
      "grad_norm": 0.1223847046494484,
      "learning_rate": 0.0002,
      "loss": 0.03714865446090698,
      "step": 1614
    },
    {
      "epoch": 0.3062926459438969,
      "grad_norm": 0.4449210464954376,
      "learning_rate": 0.0002,
      "loss": 0.025119107216596603,
      "step": 1616
    },
    {
      "epoch": 0.3066717210007581,
      "grad_norm": 0.11647365242242813,
      "learning_rate": 0.0002,
      "loss": 0.01677449233829975,
      "step": 1618
    },
    {
      "epoch": 0.3070507960576194,
      "grad_norm": 0.5005423426628113,
      "learning_rate": 0.0002,
      "loss": 0.030065903440117836,
      "step": 1620
    },
    {
      "epoch": 0.30742987111448067,
      "grad_norm": 0.24645188450813293,
      "learning_rate": 0.0002,
      "loss": 0.021039554849267006,
      "step": 1622
    },
    {
      "epoch": 0.3078089461713419,
      "grad_norm": 0.31103023886680603,
      "learning_rate": 0.0002,
      "loss": 0.021373309195041656,
      "step": 1624
    },
    {
      "epoch": 0.30818802122820316,
      "grad_norm": 0.3194957673549652,
      "learning_rate": 0.0002,
      "loss": 0.027270454913377762,
      "step": 1626
    },
    {
      "epoch": 0.30856709628506446,
      "grad_norm": 0.31171083450317383,
      "learning_rate": 0.0002,
      "loss": 0.03203478455543518,
      "step": 1628
    },
    {
      "epoch": 0.3089461713419257,
      "grad_norm": 0.2349575012922287,
      "learning_rate": 0.0002,
      "loss": 0.01739334873855114,
      "step": 1630
    },
    {
      "epoch": 0.30932524639878695,
      "grad_norm": 0.14006008207798004,
      "learning_rate": 0.0002,
      "loss": 0.0186357032507658,
      "step": 1632
    },
    {
      "epoch": 0.30970432145564825,
      "grad_norm": 0.4937635362148285,
      "learning_rate": 0.0002,
      "loss": 0.0345446951687336,
      "step": 1634
    },
    {
      "epoch": 0.3100833965125095,
      "grad_norm": 0.127630814909935,
      "learning_rate": 0.0002,
      "loss": 0.014230228029191494,
      "step": 1636
    },
    {
      "epoch": 0.31046247156937073,
      "grad_norm": 0.19497992098331451,
      "learning_rate": 0.0002,
      "loss": 0.02199695073068142,
      "step": 1638
    },
    {
      "epoch": 0.310841546626232,
      "grad_norm": 0.4816901981830597,
      "learning_rate": 0.0002,
      "loss": 0.0380391851067543,
      "step": 1640
    },
    {
      "epoch": 0.3112206216830933,
      "grad_norm": 0.4850144386291504,
      "learning_rate": 0.0002,
      "loss": 0.04754983261227608,
      "step": 1642
    },
    {
      "epoch": 0.3115996967399545,
      "grad_norm": 0.4656154215335846,
      "learning_rate": 0.0002,
      "loss": 0.02403002418577671,
      "step": 1644
    },
    {
      "epoch": 0.31197877179681577,
      "grad_norm": 0.4365699589252472,
      "learning_rate": 0.0002,
      "loss": 0.04247819632291794,
      "step": 1646
    },
    {
      "epoch": 0.312357846853677,
      "grad_norm": 0.2113034427165985,
      "learning_rate": 0.0002,
      "loss": 0.03211423382163048,
      "step": 1648
    },
    {
      "epoch": 0.3127369219105383,
      "grad_norm": 0.37794023752212524,
      "learning_rate": 0.0002,
      "loss": 0.043343354016542435,
      "step": 1650
    },
    {
      "epoch": 0.31311599696739956,
      "grad_norm": 0.12304718792438507,
      "learning_rate": 0.0002,
      "loss": 0.033650919795036316,
      "step": 1652
    },
    {
      "epoch": 0.3134950720242608,
      "grad_norm": 0.1743556261062622,
      "learning_rate": 0.0002,
      "loss": 0.014318058267235756,
      "step": 1654
    },
    {
      "epoch": 0.31387414708112205,
      "grad_norm": 0.1908075213432312,
      "learning_rate": 0.0002,
      "loss": 0.030543411150574684,
      "step": 1656
    },
    {
      "epoch": 0.31425322213798335,
      "grad_norm": 0.12400814890861511,
      "learning_rate": 0.0002,
      "loss": 0.020124029368162155,
      "step": 1658
    },
    {
      "epoch": 0.3146322971948446,
      "grad_norm": 0.12953273952007294,
      "learning_rate": 0.0002,
      "loss": 0.029818708077073097,
      "step": 1660
    },
    {
      "epoch": 0.31501137225170583,
      "grad_norm": 0.47322577238082886,
      "learning_rate": 0.0002,
      "loss": 0.035968195647001266,
      "step": 1662
    },
    {
      "epoch": 0.3153904473085671,
      "grad_norm": 0.2159811556339264,
      "learning_rate": 0.0002,
      "loss": 0.0177186019718647,
      "step": 1664
    },
    {
      "epoch": 0.3157695223654284,
      "grad_norm": 0.1963239461183548,
      "learning_rate": 0.0002,
      "loss": 0.016818663105368614,
      "step": 1666
    },
    {
      "epoch": 0.3161485974222896,
      "grad_norm": 0.26109495759010315,
      "learning_rate": 0.0002,
      "loss": 0.031019479036331177,
      "step": 1668
    },
    {
      "epoch": 0.31652767247915087,
      "grad_norm": 0.47929567098617554,
      "learning_rate": 0.0002,
      "loss": 0.04887539893388748,
      "step": 1670
    },
    {
      "epoch": 0.3169067475360121,
      "grad_norm": 0.3924487233161926,
      "learning_rate": 0.0002,
      "loss": 0.028053753077983856,
      "step": 1672
    },
    {
      "epoch": 0.3172858225928734,
      "grad_norm": 0.1136970967054367,
      "learning_rate": 0.0002,
      "loss": 0.03109646588563919,
      "step": 1674
    },
    {
      "epoch": 0.31766489764973466,
      "grad_norm": 0.2781054377555847,
      "learning_rate": 0.0002,
      "loss": 0.013624976389110088,
      "step": 1676
    },
    {
      "epoch": 0.3180439727065959,
      "grad_norm": 0.4607812464237213,
      "learning_rate": 0.0002,
      "loss": 0.04482942447066307,
      "step": 1678
    },
    {
      "epoch": 0.31842304776345715,
      "grad_norm": 0.3209092319011688,
      "learning_rate": 0.0002,
      "loss": 0.01909898780286312,
      "step": 1680
    },
    {
      "epoch": 0.31880212282031845,
      "grad_norm": 0.33151909708976746,
      "learning_rate": 0.0002,
      "loss": 0.02815149538218975,
      "step": 1682
    },
    {
      "epoch": 0.3191811978771797,
      "grad_norm": 0.33256322145462036,
      "learning_rate": 0.0002,
      "loss": 0.015690382570028305,
      "step": 1684
    },
    {
      "epoch": 0.31956027293404093,
      "grad_norm": 0.1559314876794815,
      "learning_rate": 0.0002,
      "loss": 0.02657286636531353,
      "step": 1686
    },
    {
      "epoch": 0.3199393479909022,
      "grad_norm": 0.12976674735546112,
      "learning_rate": 0.0002,
      "loss": 0.03133563697338104,
      "step": 1688
    },
    {
      "epoch": 0.3203184230477635,
      "grad_norm": 0.1337515115737915,
      "learning_rate": 0.0002,
      "loss": 0.013705960474908352,
      "step": 1690
    },
    {
      "epoch": 0.3206974981046247,
      "grad_norm": 0.2993910312652588,
      "learning_rate": 0.0002,
      "loss": 0.021465182304382324,
      "step": 1692
    },
    {
      "epoch": 0.32107657316148597,
      "grad_norm": 0.15996798872947693,
      "learning_rate": 0.0002,
      "loss": 0.015075287781655788,
      "step": 1694
    },
    {
      "epoch": 0.3214556482183472,
      "grad_norm": 0.1623569130897522,
      "learning_rate": 0.0002,
      "loss": 0.014453302137553692,
      "step": 1696
    },
    {
      "epoch": 0.3218347232752085,
      "grad_norm": 0.2793128788471222,
      "learning_rate": 0.0002,
      "loss": 0.017390741035342216,
      "step": 1698
    },
    {
      "epoch": 0.32221379833206976,
      "grad_norm": 0.12293332070112228,
      "learning_rate": 0.0002,
      "loss": 0.026852834969758987,
      "step": 1700
    },
    {
      "epoch": 0.322592873388931,
      "grad_norm": 0.12948665022850037,
      "learning_rate": 0.0002,
      "loss": 0.014915118925273418,
      "step": 1702
    },
    {
      "epoch": 0.32297194844579225,
      "grad_norm": 0.36127346754074097,
      "learning_rate": 0.0002,
      "loss": 0.032164011150598526,
      "step": 1704
    },
    {
      "epoch": 0.32335102350265355,
      "grad_norm": 0.2470552623271942,
      "learning_rate": 0.0002,
      "loss": 0.01428802590817213,
      "step": 1706
    },
    {
      "epoch": 0.3237300985595148,
      "grad_norm": 0.4427942633628845,
      "learning_rate": 0.0002,
      "loss": 0.03362748399376869,
      "step": 1708
    },
    {
      "epoch": 0.32410917361637603,
      "grad_norm": 0.13136041164398193,
      "learning_rate": 0.0002,
      "loss": 0.013468136079609394,
      "step": 1710
    },
    {
      "epoch": 0.3244882486732373,
      "grad_norm": 0.13474689424037933,
      "learning_rate": 0.0002,
      "loss": 0.02110585942864418,
      "step": 1712
    },
    {
      "epoch": 0.3248673237300986,
      "grad_norm": 0.4810140132904053,
      "learning_rate": 0.0002,
      "loss": 0.029578913003206253,
      "step": 1714
    },
    {
      "epoch": 0.3252463987869598,
      "grad_norm": 0.4723875820636749,
      "learning_rate": 0.0002,
      "loss": 0.029462087899446487,
      "step": 1716
    },
    {
      "epoch": 0.32562547384382107,
      "grad_norm": 0.10433496534824371,
      "learning_rate": 0.0002,
      "loss": 0.01887589693069458,
      "step": 1718
    },
    {
      "epoch": 0.3260045489006823,
      "grad_norm": 0.1080699935555458,
      "learning_rate": 0.0002,
      "loss": 0.01593579724431038,
      "step": 1720
    },
    {
      "epoch": 0.3263836239575436,
      "grad_norm": 0.12263483554124832,
      "learning_rate": 0.0002,
      "loss": 0.025032320991158485,
      "step": 1722
    },
    {
      "epoch": 0.32676269901440486,
      "grad_norm": 0.14497624337673187,
      "learning_rate": 0.0002,
      "loss": 0.03342824429273605,
      "step": 1724
    },
    {
      "epoch": 0.3271417740712661,
      "grad_norm": 0.332999050617218,
      "learning_rate": 0.0002,
      "loss": 0.039342015981674194,
      "step": 1726
    },
    {
      "epoch": 0.32752084912812734,
      "grad_norm": 0.40725457668304443,
      "learning_rate": 0.0002,
      "loss": 0.031406648457050323,
      "step": 1728
    },
    {
      "epoch": 0.32789992418498864,
      "grad_norm": 0.27598118782043457,
      "learning_rate": 0.0002,
      "loss": 0.01668478362262249,
      "step": 1730
    },
    {
      "epoch": 0.3282789992418499,
      "grad_norm": 0.15291550755500793,
      "learning_rate": 0.0002,
      "loss": 0.01373656652867794,
      "step": 1732
    },
    {
      "epoch": 0.32865807429871113,
      "grad_norm": 0.1303073614835739,
      "learning_rate": 0.0002,
      "loss": 0.02650412917137146,
      "step": 1734
    },
    {
      "epoch": 0.3290371493555724,
      "grad_norm": 0.1952270120382309,
      "learning_rate": 0.0002,
      "loss": 0.015457339584827423,
      "step": 1736
    },
    {
      "epoch": 0.3294162244124337,
      "grad_norm": 0.33008480072021484,
      "learning_rate": 0.0002,
      "loss": 0.022093115374445915,
      "step": 1738
    },
    {
      "epoch": 0.3297952994692949,
      "grad_norm": 0.3261886239051819,
      "learning_rate": 0.0002,
      "loss": 0.025691049173474312,
      "step": 1740
    },
    {
      "epoch": 0.33017437452615617,
      "grad_norm": 0.4512615501880646,
      "learning_rate": 0.0002,
      "loss": 0.03771752864122391,
      "step": 1742
    },
    {
      "epoch": 0.3305534495830174,
      "grad_norm": 0.2671006917953491,
      "learning_rate": 0.0002,
      "loss": 0.027542635798454285,
      "step": 1744
    },
    {
      "epoch": 0.3309325246398787,
      "grad_norm": 0.35332053899765015,
      "learning_rate": 0.0002,
      "loss": 0.022809166461229324,
      "step": 1746
    },
    {
      "epoch": 0.33131159969673996,
      "grad_norm": 0.12849749624729156,
      "learning_rate": 0.0002,
      "loss": 0.021876340731978416,
      "step": 1748
    },
    {
      "epoch": 0.3316906747536012,
      "grad_norm": 0.1153123676776886,
      "learning_rate": 0.0002,
      "loss": 0.016138147562742233,
      "step": 1750
    },
    {
      "epoch": 0.33206974981046244,
      "grad_norm": 0.1245860680937767,
      "learning_rate": 0.0002,
      "loss": 0.012701584957540035,
      "step": 1752
    },
    {
      "epoch": 0.33244882486732374,
      "grad_norm": 0.2997908592224121,
      "learning_rate": 0.0002,
      "loss": 0.02454487420618534,
      "step": 1754
    },
    {
      "epoch": 0.332827899924185,
      "grad_norm": 0.5024239420890808,
      "learning_rate": 0.0002,
      "loss": 0.03482493385672569,
      "step": 1756
    },
    {
      "epoch": 0.33320697498104623,
      "grad_norm": 0.3544254004955292,
      "learning_rate": 0.0002,
      "loss": 0.023102061823010445,
      "step": 1758
    },
    {
      "epoch": 0.33358605003790753,
      "grad_norm": 0.13101163506507874,
      "learning_rate": 0.0002,
      "loss": 0.017560632899403572,
      "step": 1760
    },
    {
      "epoch": 0.3339651250947688,
      "grad_norm": 0.5881673693656921,
      "learning_rate": 0.0002,
      "loss": 0.043129101395606995,
      "step": 1762
    },
    {
      "epoch": 0.33434420015163,
      "grad_norm": 0.3551969826221466,
      "learning_rate": 0.0002,
      "loss": 0.034549448639154434,
      "step": 1764
    },
    {
      "epoch": 0.33472327520849127,
      "grad_norm": 0.20401477813720703,
      "learning_rate": 0.0002,
      "loss": 0.025416167452931404,
      "step": 1766
    },
    {
      "epoch": 0.33510235026535257,
      "grad_norm": 0.46242186427116394,
      "learning_rate": 0.0002,
      "loss": 0.04505079984664917,
      "step": 1768
    },
    {
      "epoch": 0.3354814253222138,
      "grad_norm": 0.11528559029102325,
      "learning_rate": 0.0002,
      "loss": 0.025189828127622604,
      "step": 1770
    },
    {
      "epoch": 0.33586050037907506,
      "grad_norm": 0.2408856302499771,
      "learning_rate": 0.0002,
      "loss": 0.017544172704219818,
      "step": 1772
    },
    {
      "epoch": 0.3362395754359363,
      "grad_norm": 0.44442471861839294,
      "learning_rate": 0.0002,
      "loss": 0.041101258248090744,
      "step": 1774
    },
    {
      "epoch": 0.3366186504927976,
      "grad_norm": 0.1815728098154068,
      "learning_rate": 0.0002,
      "loss": 0.031430162489414215,
      "step": 1776
    },
    {
      "epoch": 0.33699772554965884,
      "grad_norm": 0.09460137039422989,
      "learning_rate": 0.0002,
      "loss": 0.022873617708683014,
      "step": 1778
    },
    {
      "epoch": 0.3373768006065201,
      "grad_norm": 0.3797334134578705,
      "learning_rate": 0.0002,
      "loss": 0.02664305828511715,
      "step": 1780
    },
    {
      "epoch": 0.33775587566338133,
      "grad_norm": 0.1240098848938942,
      "learning_rate": 0.0002,
      "loss": 0.013462704606354237,
      "step": 1782
    },
    {
      "epoch": 0.33813495072024263,
      "grad_norm": 0.12685416638851166,
      "learning_rate": 0.0002,
      "loss": 0.013538607396185398,
      "step": 1784
    },
    {
      "epoch": 0.3385140257771039,
      "grad_norm": 0.13458463549613953,
      "learning_rate": 0.0002,
      "loss": 0.019251609221100807,
      "step": 1786
    },
    {
      "epoch": 0.3388931008339651,
      "grad_norm": 0.15207310020923615,
      "learning_rate": 0.0002,
      "loss": 0.036093819886446,
      "step": 1788
    },
    {
      "epoch": 0.33927217589082637,
      "grad_norm": 0.16056448221206665,
      "learning_rate": 0.0002,
      "loss": 0.0215620044618845,
      "step": 1790
    },
    {
      "epoch": 0.33965125094768767,
      "grad_norm": 0.4048640727996826,
      "learning_rate": 0.0002,
      "loss": 0.0328492745757103,
      "step": 1792
    },
    {
      "epoch": 0.3400303260045489,
      "grad_norm": 0.11924159526824951,
      "learning_rate": 0.0002,
      "loss": 0.014519160613417625,
      "step": 1794
    },
    {
      "epoch": 0.34040940106141016,
      "grad_norm": 0.4102783203125,
      "learning_rate": 0.0002,
      "loss": 0.028805414214730263,
      "step": 1796
    },
    {
      "epoch": 0.3407884761182714,
      "grad_norm": 0.12864531576633453,
      "learning_rate": 0.0002,
      "loss": 0.013738314621150494,
      "step": 1798
    },
    {
      "epoch": 0.3411675511751327,
      "grad_norm": 0.34946244955062866,
      "learning_rate": 0.0002,
      "loss": 0.02228681556880474,
      "step": 1800
    },
    {
      "epoch": 0.34154662623199394,
      "grad_norm": 0.10797599703073502,
      "learning_rate": 0.0002,
      "loss": 0.030994905158877373,
      "step": 1802
    },
    {
      "epoch": 0.3419257012888552,
      "grad_norm": 0.12431894987821579,
      "learning_rate": 0.0002,
      "loss": 0.017048167064785957,
      "step": 1804
    },
    {
      "epoch": 0.34230477634571643,
      "grad_norm": 0.30579692125320435,
      "learning_rate": 0.0002,
      "loss": 0.028480367735028267,
      "step": 1806
    },
    {
      "epoch": 0.34268385140257773,
      "grad_norm": 0.11287450045347214,
      "learning_rate": 0.0002,
      "loss": 0.026433609426021576,
      "step": 1808
    },
    {
      "epoch": 0.343062926459439,
      "grad_norm": 0.47243309020996094,
      "learning_rate": 0.0002,
      "loss": 0.03775624930858612,
      "step": 1810
    },
    {
      "epoch": 0.3434420015163002,
      "grad_norm": 0.09306806325912476,
      "learning_rate": 0.0002,
      "loss": 0.012662287801504135,
      "step": 1812
    },
    {
      "epoch": 0.34382107657316147,
      "grad_norm": 0.17652657628059387,
      "learning_rate": 0.0002,
      "loss": 0.027501389384269714,
      "step": 1814
    },
    {
      "epoch": 0.34420015163002277,
      "grad_norm": 0.5555701851844788,
      "learning_rate": 0.0002,
      "loss": 0.04666243493556976,
      "step": 1816
    },
    {
      "epoch": 0.344579226686884,
      "grad_norm": 0.0979434922337532,
      "learning_rate": 0.0002,
      "loss": 0.028335893526673317,
      "step": 1818
    },
    {
      "epoch": 0.34495830174374525,
      "grad_norm": 0.29753708839416504,
      "learning_rate": 0.0002,
      "loss": 0.03895462304353714,
      "step": 1820
    },
    {
      "epoch": 0.3453373768006065,
      "grad_norm": 0.10650414228439331,
      "learning_rate": 0.0002,
      "loss": 0.019142314791679382,
      "step": 1822
    },
    {
      "epoch": 0.3457164518574678,
      "grad_norm": 0.1194157749414444,
      "learning_rate": 0.0002,
      "loss": 0.012043087743222713,
      "step": 1824
    },
    {
      "epoch": 0.34609552691432904,
      "grad_norm": 0.40988045930862427,
      "learning_rate": 0.0002,
      "loss": 0.03706936165690422,
      "step": 1826
    },
    {
      "epoch": 0.3464746019711903,
      "grad_norm": 0.41879385709762573,
      "learning_rate": 0.0002,
      "loss": 0.030927592888474464,
      "step": 1828
    },
    {
      "epoch": 0.34685367702805153,
      "grad_norm": 0.38007715344429016,
      "learning_rate": 0.0002,
      "loss": 0.02468494325876236,
      "step": 1830
    },
    {
      "epoch": 0.34723275208491283,
      "grad_norm": 0.42285582423210144,
      "learning_rate": 0.0002,
      "loss": 0.037105269730091095,
      "step": 1832
    },
    {
      "epoch": 0.3476118271417741,
      "grad_norm": 0.38064712285995483,
      "learning_rate": 0.0002,
      "loss": 0.03318294137716293,
      "step": 1834
    },
    {
      "epoch": 0.3479909021986353,
      "grad_norm": 0.4045671224594116,
      "learning_rate": 0.0002,
      "loss": 0.033585283905267715,
      "step": 1836
    },
    {
      "epoch": 0.34836997725549657,
      "grad_norm": 0.3657643496990204,
      "learning_rate": 0.0002,
      "loss": 0.03689344599843025,
      "step": 1838
    },
    {
      "epoch": 0.34874905231235787,
      "grad_norm": 0.2788217067718506,
      "learning_rate": 0.0002,
      "loss": 0.019537638872861862,
      "step": 1840
    },
    {
      "epoch": 0.3491281273692191,
      "grad_norm": 0.29316917061805725,
      "learning_rate": 0.0002,
      "loss": 0.03128159046173096,
      "step": 1842
    },
    {
      "epoch": 0.34950720242608035,
      "grad_norm": 0.4648869037628174,
      "learning_rate": 0.0002,
      "loss": 0.027758440002799034,
      "step": 1844
    },
    {
      "epoch": 0.3498862774829416,
      "grad_norm": 0.1455068737268448,
      "learning_rate": 0.0002,
      "loss": 0.019724393263459206,
      "step": 1846
    },
    {
      "epoch": 0.3502653525398029,
      "grad_norm": 0.4376871585845947,
      "learning_rate": 0.0002,
      "loss": 0.02932368591427803,
      "step": 1848
    },
    {
      "epoch": 0.35064442759666414,
      "grad_norm": 0.43592357635498047,
      "learning_rate": 0.0002,
      "loss": 0.02893572486937046,
      "step": 1850
    },
    {
      "epoch": 0.3510235026535254,
      "grad_norm": 0.48956096172332764,
      "learning_rate": 0.0002,
      "loss": 0.04212665930390358,
      "step": 1852
    },
    {
      "epoch": 0.35140257771038663,
      "grad_norm": 0.14418356120586395,
      "learning_rate": 0.0002,
      "loss": 0.013608591631054878,
      "step": 1854
    },
    {
      "epoch": 0.35178165276724793,
      "grad_norm": 0.5508156418800354,
      "learning_rate": 0.0002,
      "loss": 0.02972152829170227,
      "step": 1856
    },
    {
      "epoch": 0.3521607278241092,
      "grad_norm": 0.3310849964618683,
      "learning_rate": 0.0002,
      "loss": 0.02480008825659752,
      "step": 1858
    },
    {
      "epoch": 0.3525398028809704,
      "grad_norm": 0.14336304366588593,
      "learning_rate": 0.0002,
      "loss": 0.012660844251513481,
      "step": 1860
    },
    {
      "epoch": 0.35291887793783167,
      "grad_norm": 0.14318352937698364,
      "learning_rate": 0.0002,
      "loss": 0.022337911650538445,
      "step": 1862
    },
    {
      "epoch": 0.35329795299469297,
      "grad_norm": 0.23057079315185547,
      "learning_rate": 0.0002,
      "loss": 0.017445318400859833,
      "step": 1864
    },
    {
      "epoch": 0.3536770280515542,
      "grad_norm": 0.3046879470348358,
      "learning_rate": 0.0002,
      "loss": 0.017304126173257828,
      "step": 1866
    },
    {
      "epoch": 0.35405610310841545,
      "grad_norm": 0.3655117154121399,
      "learning_rate": 0.0002,
      "loss": 0.0341893695294857,
      "step": 1868
    },
    {
      "epoch": 0.3544351781652767,
      "grad_norm": 0.4636896848678589,
      "learning_rate": 0.0002,
      "loss": 0.03991377726197243,
      "step": 1870
    },
    {
      "epoch": 0.354814253222138,
      "grad_norm": 0.30543792247772217,
      "learning_rate": 0.0002,
      "loss": 0.0363551490008831,
      "step": 1872
    },
    {
      "epoch": 0.35519332827899924,
      "grad_norm": 0.09562942385673523,
      "learning_rate": 0.0002,
      "loss": 0.012620219960808754,
      "step": 1874
    },
    {
      "epoch": 0.3555724033358605,
      "grad_norm": 0.16613568365573883,
      "learning_rate": 0.0002,
      "loss": 0.01414234284311533,
      "step": 1876
    },
    {
      "epoch": 0.35595147839272173,
      "grad_norm": 0.3637671172618866,
      "learning_rate": 0.0002,
      "loss": 0.024448510259389877,
      "step": 1878
    },
    {
      "epoch": 0.35633055344958303,
      "grad_norm": 0.2921352982521057,
      "learning_rate": 0.0002,
      "loss": 0.03209369257092476,
      "step": 1880
    },
    {
      "epoch": 0.3567096285064443,
      "grad_norm": 0.26941418647766113,
      "learning_rate": 0.0002,
      "loss": 0.020260943099856377,
      "step": 1882
    },
    {
      "epoch": 0.3570887035633055,
      "grad_norm": 0.10331138968467712,
      "learning_rate": 0.0002,
      "loss": 0.029520699754357338,
      "step": 1884
    },
    {
      "epoch": 0.3574677786201668,
      "grad_norm": 0.3732324242591858,
      "learning_rate": 0.0002,
      "loss": 0.031058765947818756,
      "step": 1886
    },
    {
      "epoch": 0.35784685367702807,
      "grad_norm": 0.4463680684566498,
      "learning_rate": 0.0002,
      "loss": 0.03501344844698906,
      "step": 1888
    },
    {
      "epoch": 0.3582259287338893,
      "grad_norm": 0.128362238407135,
      "learning_rate": 0.0002,
      "loss": 0.023634864017367363,
      "step": 1890
    },
    {
      "epoch": 0.35860500379075055,
      "grad_norm": 0.10930440574884415,
      "learning_rate": 0.0002,
      "loss": 0.027152052149176598,
      "step": 1892
    },
    {
      "epoch": 0.35898407884761185,
      "grad_norm": 0.1292339265346527,
      "learning_rate": 0.0002,
      "loss": 0.01296201627701521,
      "step": 1894
    },
    {
      "epoch": 0.3593631539044731,
      "grad_norm": 0.24244926869869232,
      "learning_rate": 0.0002,
      "loss": 0.03891376405954361,
      "step": 1896
    },
    {
      "epoch": 0.35974222896133434,
      "grad_norm": 0.28948384523391724,
      "learning_rate": 0.0002,
      "loss": 0.03951626271009445,
      "step": 1898
    },
    {
      "epoch": 0.3601213040181956,
      "grad_norm": 0.46286430954933167,
      "learning_rate": 0.0002,
      "loss": 0.03130920231342316,
      "step": 1900
    },
    {
      "epoch": 0.3605003790750569,
      "grad_norm": 0.2901206612586975,
      "learning_rate": 0.0002,
      "loss": 0.03691709414124489,
      "step": 1902
    },
    {
      "epoch": 0.36087945413191813,
      "grad_norm": 0.4095093607902527,
      "learning_rate": 0.0002,
      "loss": 0.033189527690410614,
      "step": 1904
    },
    {
      "epoch": 0.3612585291887794,
      "grad_norm": 0.0933108776807785,
      "learning_rate": 0.0002,
      "loss": 0.016015058383345604,
      "step": 1906
    },
    {
      "epoch": 0.3616376042456406,
      "grad_norm": 0.3092695474624634,
      "learning_rate": 0.0002,
      "loss": 0.03115110471844673,
      "step": 1908
    },
    {
      "epoch": 0.3620166793025019,
      "grad_norm": 0.2582691013813019,
      "learning_rate": 0.0002,
      "loss": 0.02032007835805416,
      "step": 1910
    },
    {
      "epoch": 0.36239575435936316,
      "grad_norm": 0.1423620581626892,
      "learning_rate": 0.0002,
      "loss": 0.014029956422746181,
      "step": 1912
    },
    {
      "epoch": 0.3627748294162244,
      "grad_norm": 0.17330744862556458,
      "learning_rate": 0.0002,
      "loss": 0.034380678087472916,
      "step": 1914
    },
    {
      "epoch": 0.36315390447308565,
      "grad_norm": 0.27506157755851746,
      "learning_rate": 0.0002,
      "loss": 0.020600339397788048,
      "step": 1916
    },
    {
      "epoch": 0.36353297952994695,
      "grad_norm": 0.12912391126155853,
      "learning_rate": 0.0002,
      "loss": 0.01266078557819128,
      "step": 1918
    },
    {
      "epoch": 0.3639120545868082,
      "grad_norm": 0.29344063997268677,
      "learning_rate": 0.0002,
      "loss": 0.026390453800559044,
      "step": 1920
    },
    {
      "epoch": 0.36429112964366944,
      "grad_norm": 0.41078484058380127,
      "learning_rate": 0.0002,
      "loss": 0.058809950947761536,
      "step": 1922
    },
    {
      "epoch": 0.3646702047005307,
      "grad_norm": 0.16844700276851654,
      "learning_rate": 0.0002,
      "loss": 0.03320324420928955,
      "step": 1924
    },
    {
      "epoch": 0.365049279757392,
      "grad_norm": 0.2506592571735382,
      "learning_rate": 0.0002,
      "loss": 0.019306106492877007,
      "step": 1926
    },
    {
      "epoch": 0.36542835481425323,
      "grad_norm": 0.11695031821727753,
      "learning_rate": 0.0002,
      "loss": 0.012690147385001183,
      "step": 1928
    },
    {
      "epoch": 0.3658074298711145,
      "grad_norm": 0.3262481391429901,
      "learning_rate": 0.0002,
      "loss": 0.038317132741212845,
      "step": 1930
    },
    {
      "epoch": 0.3661865049279757,
      "grad_norm": 0.23561982810497284,
      "learning_rate": 0.0002,
      "loss": 0.01827545464038849,
      "step": 1932
    },
    {
      "epoch": 0.366565579984837,
      "grad_norm": 0.4862864017486572,
      "learning_rate": 0.0002,
      "loss": 0.038538858294487,
      "step": 1934
    },
    {
      "epoch": 0.36694465504169826,
      "grad_norm": 0.13508577644824982,
      "learning_rate": 0.0002,
      "loss": 0.021105043590068817,
      "step": 1936
    },
    {
      "epoch": 0.3673237300985595,
      "grad_norm": 0.438109815120697,
      "learning_rate": 0.0002,
      "loss": 0.029882531613111496,
      "step": 1938
    },
    {
      "epoch": 0.36770280515542075,
      "grad_norm": 0.4898236393928528,
      "learning_rate": 0.0002,
      "loss": 0.0351562574505806,
      "step": 1940
    },
    {
      "epoch": 0.36808188021228205,
      "grad_norm": 0.3217776417732239,
      "learning_rate": 0.0002,
      "loss": 0.022685833275318146,
      "step": 1942
    },
    {
      "epoch": 0.3684609552691433,
      "grad_norm": 0.36467981338500977,
      "learning_rate": 0.0002,
      "loss": 0.020401323214173317,
      "step": 1944
    },
    {
      "epoch": 0.36884003032600454,
      "grad_norm": 0.12152090668678284,
      "learning_rate": 0.0002,
      "loss": 0.026591187343001366,
      "step": 1946
    },
    {
      "epoch": 0.3692191053828658,
      "grad_norm": 0.12106894701719284,
      "learning_rate": 0.0002,
      "loss": 0.014585360884666443,
      "step": 1948
    },
    {
      "epoch": 0.3695981804397271,
      "grad_norm": 0.1549663245677948,
      "learning_rate": 0.0002,
      "loss": 0.015073969960212708,
      "step": 1950
    },
    {
      "epoch": 0.36997725549658833,
      "grad_norm": 0.35977232456207275,
      "learning_rate": 0.0002,
      "loss": 0.03718845173716545,
      "step": 1952
    },
    {
      "epoch": 0.3703563305534496,
      "grad_norm": 0.27561914920806885,
      "learning_rate": 0.0002,
      "loss": 0.0318228080868721,
      "step": 1954
    },
    {
      "epoch": 0.3707354056103108,
      "grad_norm": 0.3265073299407959,
      "learning_rate": 0.0002,
      "loss": 0.03985099494457245,
      "step": 1956
    },
    {
      "epoch": 0.3711144806671721,
      "grad_norm": 0.1277337223291397,
      "learning_rate": 0.0002,
      "loss": 0.02454950474202633,
      "step": 1958
    },
    {
      "epoch": 0.37149355572403336,
      "grad_norm": 0.40479832887649536,
      "learning_rate": 0.0002,
      "loss": 0.03887586295604706,
      "step": 1960
    },
    {
      "epoch": 0.3718726307808946,
      "grad_norm": 0.3196302056312561,
      "learning_rate": 0.0002,
      "loss": 0.02165071666240692,
      "step": 1962
    },
    {
      "epoch": 0.37225170583775585,
      "grad_norm": 0.10979686677455902,
      "learning_rate": 0.0002,
      "loss": 0.033365074545145035,
      "step": 1964
    },
    {
      "epoch": 0.37263078089461715,
      "grad_norm": 0.16850513219833374,
      "learning_rate": 0.0002,
      "loss": 0.020703036338090897,
      "step": 1966
    },
    {
      "epoch": 0.3730098559514784,
      "grad_norm": 0.4189241826534271,
      "learning_rate": 0.0002,
      "loss": 0.03563521057367325,
      "step": 1968
    },
    {
      "epoch": 0.37338893100833964,
      "grad_norm": 0.4477117359638214,
      "learning_rate": 0.0002,
      "loss": 0.02964857406914234,
      "step": 1970
    },
    {
      "epoch": 0.3737680060652009,
      "grad_norm": 0.43876639008522034,
      "learning_rate": 0.0002,
      "loss": 0.04768936708569527,
      "step": 1972
    },
    {
      "epoch": 0.3741470811220622,
      "grad_norm": 0.4803165793418884,
      "learning_rate": 0.0002,
      "loss": 0.03479911386966705,
      "step": 1974
    },
    {
      "epoch": 0.37452615617892343,
      "grad_norm": 0.1471567153930664,
      "learning_rate": 0.0002,
      "loss": 0.01935398019850254,
      "step": 1976
    },
    {
      "epoch": 0.3749052312357847,
      "grad_norm": 0.32167235016822815,
      "learning_rate": 0.0002,
      "loss": 0.021323002874851227,
      "step": 1978
    },
    {
      "epoch": 0.3752843062926459,
      "grad_norm": 0.31261950731277466,
      "learning_rate": 0.0002,
      "loss": 0.01899336650967598,
      "step": 1980
    },
    {
      "epoch": 0.3756633813495072,
      "grad_norm": 0.13916146755218506,
      "learning_rate": 0.0002,
      "loss": 0.01532474160194397,
      "step": 1982
    },
    {
      "epoch": 0.37604245640636846,
      "grad_norm": 0.11882118135690689,
      "learning_rate": 0.0002,
      "loss": 0.034881751984357834,
      "step": 1984
    },
    {
      "epoch": 0.3764215314632297,
      "grad_norm": 0.09415015578269958,
      "learning_rate": 0.0002,
      "loss": 0.024889327585697174,
      "step": 1986
    },
    {
      "epoch": 0.37680060652009095,
      "grad_norm": 0.3665996193885803,
      "learning_rate": 0.0002,
      "loss": 0.03834991902112961,
      "step": 1988
    },
    {
      "epoch": 0.37717968157695225,
      "grad_norm": 0.26396360993385315,
      "learning_rate": 0.0002,
      "loss": 0.017181742936372757,
      "step": 1990
    },
    {
      "epoch": 0.3775587566338135,
      "grad_norm": 0.10991029441356659,
      "learning_rate": 0.0002,
      "loss": 0.02855796366930008,
      "step": 1992
    },
    {
      "epoch": 0.37793783169067474,
      "grad_norm": 0.13103090226650238,
      "learning_rate": 0.0002,
      "loss": 0.02030666545033455,
      "step": 1994
    },
    {
      "epoch": 0.378316906747536,
      "grad_norm": 0.30240243673324585,
      "learning_rate": 0.0002,
      "loss": 0.030572878196835518,
      "step": 1996
    },
    {
      "epoch": 0.3786959818043973,
      "grad_norm": 0.11657755821943283,
      "learning_rate": 0.0002,
      "loss": 0.017095213755965233,
      "step": 1998
    },
    {
      "epoch": 0.37907505686125853,
      "grad_norm": 0.15121635794639587,
      "learning_rate": 0.0002,
      "loss": 0.02291691116988659,
      "step": 2000
    },
    {
      "epoch": 0.3794541319181198,
      "grad_norm": 0.2626281976699829,
      "learning_rate": 0.0002,
      "loss": 0.025745728984475136,
      "step": 2002
    },
    {
      "epoch": 0.379833206974981,
      "grad_norm": 0.27462896704673767,
      "learning_rate": 0.0002,
      "loss": 0.017245162278413773,
      "step": 2004
    },
    {
      "epoch": 0.3802122820318423,
      "grad_norm": 0.1085989847779274,
      "learning_rate": 0.0002,
      "loss": 0.015067651867866516,
      "step": 2006
    },
    {
      "epoch": 0.38059135708870356,
      "grad_norm": 0.42288899421691895,
      "learning_rate": 0.0002,
      "loss": 0.055154286324977875,
      "step": 2008
    },
    {
      "epoch": 0.3809704321455648,
      "grad_norm": 0.14434966444969177,
      "learning_rate": 0.0002,
      "loss": 0.02522304281592369,
      "step": 2010
    },
    {
      "epoch": 0.3813495072024261,
      "grad_norm": 0.36957916617393494,
      "learning_rate": 0.0002,
      "loss": 0.03993171080946922,
      "step": 2012
    },
    {
      "epoch": 0.38172858225928735,
      "grad_norm": 0.3523724377155304,
      "learning_rate": 0.0002,
      "loss": 0.0405685231089592,
      "step": 2014
    },
    {
      "epoch": 0.3821076573161486,
      "grad_norm": 0.26990360021591187,
      "learning_rate": 0.0002,
      "loss": 0.024248778820037842,
      "step": 2016
    },
    {
      "epoch": 0.38248673237300984,
      "grad_norm": 0.3949528634548187,
      "learning_rate": 0.0002,
      "loss": 0.03128862753510475,
      "step": 2018
    },
    {
      "epoch": 0.38286580742987114,
      "grad_norm": 0.12943001091480255,
      "learning_rate": 0.0002,
      "loss": 0.01404862105846405,
      "step": 2020
    },
    {
      "epoch": 0.3832448824867324,
      "grad_norm": 0.3834663927555084,
      "learning_rate": 0.0002,
      "loss": 0.04263802617788315,
      "step": 2022
    },
    {
      "epoch": 0.38362395754359363,
      "grad_norm": 0.16173094511032104,
      "learning_rate": 0.0002,
      "loss": 0.03936188295483589,
      "step": 2024
    },
    {
      "epoch": 0.3840030326004549,
      "grad_norm": 0.18683169782161713,
      "learning_rate": 0.0002,
      "loss": 0.013477638363838196,
      "step": 2026
    },
    {
      "epoch": 0.3843821076573162,
      "grad_norm": 0.3919936418533325,
      "learning_rate": 0.0002,
      "loss": 0.017762772738933563,
      "step": 2028
    },
    {
      "epoch": 0.3847611827141774,
      "grad_norm": 0.17391254007816315,
      "learning_rate": 0.0002,
      "loss": 0.013754913583397865,
      "step": 2030
    },
    {
      "epoch": 0.38514025777103866,
      "grad_norm": 0.14119277894496918,
      "learning_rate": 0.0002,
      "loss": 0.021773869171738625,
      "step": 2032
    },
    {
      "epoch": 0.3855193328278999,
      "grad_norm": 0.1023949459195137,
      "learning_rate": 0.0002,
      "loss": 0.01227244921028614,
      "step": 2034
    },
    {
      "epoch": 0.3858984078847612,
      "grad_norm": 0.18287064135074615,
      "learning_rate": 0.0002,
      "loss": 0.01646113581955433,
      "step": 2036
    },
    {
      "epoch": 0.38627748294162245,
      "grad_norm": 0.13911961019039154,
      "learning_rate": 0.0002,
      "loss": 0.01776352897286415,
      "step": 2038
    },
    {
      "epoch": 0.3866565579984837,
      "grad_norm": 0.39054280519485474,
      "learning_rate": 0.0002,
      "loss": 0.02789497748017311,
      "step": 2040
    },
    {
      "epoch": 0.38703563305534494,
      "grad_norm": 0.25265198945999146,
      "learning_rate": 0.0002,
      "loss": 0.02780643291771412,
      "step": 2042
    },
    {
      "epoch": 0.38741470811220624,
      "grad_norm": 0.27706241607666016,
      "learning_rate": 0.0002,
      "loss": 0.033393215388059616,
      "step": 2044
    },
    {
      "epoch": 0.3877937831690675,
      "grad_norm": 0.3776288628578186,
      "learning_rate": 0.0002,
      "loss": 0.046053797006607056,
      "step": 2046
    },
    {
      "epoch": 0.38817285822592873,
      "grad_norm": 0.45423415303230286,
      "learning_rate": 0.0002,
      "loss": 0.03768612816929817,
      "step": 2048
    },
    {
      "epoch": 0.38855193328279,
      "grad_norm": 0.320165753364563,
      "learning_rate": 0.0002,
      "loss": 0.020449314266443253,
      "step": 2050
    },
    {
      "epoch": 0.3889310083396513,
      "grad_norm": 0.3741523027420044,
      "learning_rate": 0.0002,
      "loss": 0.023256242275238037,
      "step": 2052
    },
    {
      "epoch": 0.3893100833965125,
      "grad_norm": 0.37912416458129883,
      "learning_rate": 0.0002,
      "loss": 0.04624989628791809,
      "step": 2054
    },
    {
      "epoch": 0.38968915845337376,
      "grad_norm": 0.21046513319015503,
      "learning_rate": 0.0002,
      "loss": 0.024260440841317177,
      "step": 2056
    },
    {
      "epoch": 0.390068233510235,
      "grad_norm": 0.5015721321105957,
      "learning_rate": 0.0002,
      "loss": 0.037919219583272934,
      "step": 2058
    },
    {
      "epoch": 0.3904473085670963,
      "grad_norm": 0.2805930972099304,
      "learning_rate": 0.0002,
      "loss": 0.027042517438530922,
      "step": 2060
    },
    {
      "epoch": 0.39082638362395755,
      "grad_norm": 0.4114140570163727,
      "learning_rate": 0.0002,
      "loss": 0.03425253927707672,
      "step": 2062
    },
    {
      "epoch": 0.3912054586808188,
      "grad_norm": 0.0991516262292862,
      "learning_rate": 0.0002,
      "loss": 0.013607734814286232,
      "step": 2064
    },
    {
      "epoch": 0.39158453373768004,
      "grad_norm": 0.27370280027389526,
      "learning_rate": 0.0002,
      "loss": 0.036977093666791916,
      "step": 2066
    },
    {
      "epoch": 0.39196360879454134,
      "grad_norm": 0.0846765786409378,
      "learning_rate": 0.0002,
      "loss": 0.024732746183872223,
      "step": 2068
    },
    {
      "epoch": 0.3923426838514026,
      "grad_norm": 0.0955590084195137,
      "learning_rate": 0.0002,
      "loss": 0.030948543921113014,
      "step": 2070
    },
    {
      "epoch": 0.39272175890826383,
      "grad_norm": 0.09481050074100494,
      "learning_rate": 0.0002,
      "loss": 0.01990809291601181,
      "step": 2072
    },
    {
      "epoch": 0.3931008339651251,
      "grad_norm": 0.1513253003358841,
      "learning_rate": 0.0002,
      "loss": 0.01790001057088375,
      "step": 2074
    },
    {
      "epoch": 0.3934799090219864,
      "grad_norm": 0.10297688841819763,
      "learning_rate": 0.0002,
      "loss": 0.025916097685694695,
      "step": 2076
    },
    {
      "epoch": 0.3938589840788476,
      "grad_norm": 0.1344214826822281,
      "learning_rate": 0.0002,
      "loss": 0.014968045055866241,
      "step": 2078
    },
    {
      "epoch": 0.39423805913570886,
      "grad_norm": 0.33775320649147034,
      "learning_rate": 0.0002,
      "loss": 0.03413890674710274,
      "step": 2080
    },
    {
      "epoch": 0.3946171341925701,
      "grad_norm": 0.36094173789024353,
      "learning_rate": 0.0002,
      "loss": 0.02451964095234871,
      "step": 2082
    },
    {
      "epoch": 0.3949962092494314,
      "grad_norm": 0.10452409088611603,
      "learning_rate": 0.0002,
      "loss": 0.017213761806488037,
      "step": 2084
    },
    {
      "epoch": 0.39537528430629265,
      "grad_norm": 0.25656500458717346,
      "learning_rate": 0.0002,
      "loss": 0.03587236627936363,
      "step": 2086
    },
    {
      "epoch": 0.3957543593631539,
      "grad_norm": 0.5169637799263,
      "learning_rate": 0.0002,
      "loss": 0.04861602187156677,
      "step": 2088
    },
    {
      "epoch": 0.39613343442001514,
      "grad_norm": 0.4878418445587158,
      "learning_rate": 0.0002,
      "loss": 0.03760712966322899,
      "step": 2090
    },
    {
      "epoch": 0.39651250947687644,
      "grad_norm": 0.18121254444122314,
      "learning_rate": 0.0002,
      "loss": 0.027085449546575546,
      "step": 2092
    },
    {
      "epoch": 0.3968915845337377,
      "grad_norm": 0.2966654896736145,
      "learning_rate": 0.0002,
      "loss": 0.02023075334727764,
      "step": 2094
    },
    {
      "epoch": 0.39727065959059893,
      "grad_norm": 0.1380729377269745,
      "learning_rate": 0.0002,
      "loss": 0.02784651890397072,
      "step": 2096
    },
    {
      "epoch": 0.3976497346474602,
      "grad_norm": 0.1999477595090866,
      "learning_rate": 0.0002,
      "loss": 0.02414214238524437,
      "step": 2098
    },
    {
      "epoch": 0.3980288097043215,
      "grad_norm": 0.3237474262714386,
      "learning_rate": 0.0002,
      "loss": 0.019914811477065086,
      "step": 2100
    },
    {
      "epoch": 0.3984078847611827,
      "grad_norm": 0.2173006385564804,
      "learning_rate": 0.0002,
      "loss": 0.034865546971559525,
      "step": 2102
    },
    {
      "epoch": 0.39878695981804396,
      "grad_norm": 0.31463563442230225,
      "learning_rate": 0.0002,
      "loss": 0.025972316041588783,
      "step": 2104
    },
    {
      "epoch": 0.3991660348749052,
      "grad_norm": 0.4048980176448822,
      "learning_rate": 0.0002,
      "loss": 0.0242728553712368,
      "step": 2106
    },
    {
      "epoch": 0.3995451099317665,
      "grad_norm": 0.12209809571504593,
      "learning_rate": 0.0002,
      "loss": 0.014455070719122887,
      "step": 2108
    },
    {
      "epoch": 0.39992418498862775,
      "grad_norm": 0.12938538193702698,
      "learning_rate": 0.0002,
      "loss": 0.014087260700762272,
      "step": 2110
    },
    {
      "epoch": 0.400303260045489,
      "grad_norm": 0.12431218475103378,
      "learning_rate": 0.0002,
      "loss": 0.028165653347969055,
      "step": 2112
    },
    {
      "epoch": 0.40068233510235024,
      "grad_norm": 0.30564501881599426,
      "learning_rate": 0.0002,
      "loss": 0.02145165018737316,
      "step": 2114
    },
    {
      "epoch": 0.40106141015921154,
      "grad_norm": 0.21738608181476593,
      "learning_rate": 0.0002,
      "loss": 0.015794849023222923,
      "step": 2116
    },
    {
      "epoch": 0.4014404852160728,
      "grad_norm": 0.3053167462348938,
      "learning_rate": 0.0002,
      "loss": 0.033887311816215515,
      "step": 2118
    },
    {
      "epoch": 0.40181956027293403,
      "grad_norm": 0.16374848783016205,
      "learning_rate": 0.0002,
      "loss": 0.01312731672078371,
      "step": 2120
    },
    {
      "epoch": 0.4021986353297953,
      "grad_norm": 0.10522999614477158,
      "learning_rate": 0.0002,
      "loss": 0.0188737940043211,
      "step": 2122
    },
    {
      "epoch": 0.4025777103866566,
      "grad_norm": 0.24129241704940796,
      "learning_rate": 0.0002,
      "loss": 0.020017845556139946,
      "step": 2124
    },
    {
      "epoch": 0.4029567854435178,
      "grad_norm": 0.15406428277492523,
      "learning_rate": 0.0002,
      "loss": 0.02100120298564434,
      "step": 2126
    },
    {
      "epoch": 0.40333586050037906,
      "grad_norm": 0.12972204387187958,
      "learning_rate": 0.0002,
      "loss": 0.012693260796368122,
      "step": 2128
    },
    {
      "epoch": 0.4037149355572403,
      "grad_norm": 0.5841611623764038,
      "learning_rate": 0.0002,
      "loss": 0.03728344291448593,
      "step": 2130
    },
    {
      "epoch": 0.4040940106141016,
      "grad_norm": 0.1685679703950882,
      "learning_rate": 0.0002,
      "loss": 0.025032328441739082,
      "step": 2132
    },
    {
      "epoch": 0.40447308567096285,
      "grad_norm": 0.203375905752182,
      "learning_rate": 0.0002,
      "loss": 0.016173316165804863,
      "step": 2134
    },
    {
      "epoch": 0.4048521607278241,
      "grad_norm": 0.16643205285072327,
      "learning_rate": 0.0002,
      "loss": 0.019061198458075523,
      "step": 2136
    },
    {
      "epoch": 0.4052312357846854,
      "grad_norm": 0.09982550144195557,
      "learning_rate": 0.0002,
      "loss": 0.013121056370437145,
      "step": 2138
    },
    {
      "epoch": 0.40561031084154664,
      "grad_norm": 0.3624897599220276,
      "learning_rate": 0.0002,
      "loss": 0.03745858743786812,
      "step": 2140
    },
    {
      "epoch": 0.4059893858984079,
      "grad_norm": 0.10503695905208588,
      "learning_rate": 0.0002,
      "loss": 0.0242761243134737,
      "step": 2142
    },
    {
      "epoch": 0.40636846095526913,
      "grad_norm": 0.49977946281433105,
      "learning_rate": 0.0002,
      "loss": 0.033382635563611984,
      "step": 2144
    },
    {
      "epoch": 0.40674753601213043,
      "grad_norm": 0.2771686911582947,
      "learning_rate": 0.0002,
      "loss": 0.03381459787487984,
      "step": 2146
    },
    {
      "epoch": 0.4071266110689917,
      "grad_norm": 0.12011824548244476,
      "learning_rate": 0.0002,
      "loss": 0.02958730049431324,
      "step": 2148
    },
    {
      "epoch": 0.4075056861258529,
      "grad_norm": 0.14134030044078827,
      "learning_rate": 0.0002,
      "loss": 0.03248757869005203,
      "step": 2150
    },
    {
      "epoch": 0.40788476118271416,
      "grad_norm": 0.13984698057174683,
      "learning_rate": 0.0002,
      "loss": 0.026591405272483826,
      "step": 2152
    },
    {
      "epoch": 0.40826383623957546,
      "grad_norm": 0.3800804316997528,
      "learning_rate": 0.0002,
      "loss": 0.051375824958086014,
      "step": 2154
    },
    {
      "epoch": 0.4086429112964367,
      "grad_norm": 0.16466543078422546,
      "learning_rate": 0.0002,
      "loss": 0.0202625822275877,
      "step": 2156
    },
    {
      "epoch": 0.40902198635329795,
      "grad_norm": 0.23973695933818817,
      "learning_rate": 0.0002,
      "loss": 0.01618131436407566,
      "step": 2158
    },
    {
      "epoch": 0.4094010614101592,
      "grad_norm": 0.4876498281955719,
      "learning_rate": 0.0002,
      "loss": 0.04498758167028427,
      "step": 2160
    },
    {
      "epoch": 0.4097801364670205,
      "grad_norm": 0.3359684646129608,
      "learning_rate": 0.0002,
      "loss": 0.03581078350543976,
      "step": 2162
    },
    {
      "epoch": 0.41015921152388174,
      "grad_norm": 0.17303654551506042,
      "learning_rate": 0.0002,
      "loss": 0.0251467302441597,
      "step": 2164
    },
    {
      "epoch": 0.410538286580743,
      "grad_norm": 0.2494642585515976,
      "learning_rate": 0.0002,
      "loss": 0.04230346530675888,
      "step": 2166
    },
    {
      "epoch": 0.41091736163760423,
      "grad_norm": 0.16362398862838745,
      "learning_rate": 0.0002,
      "loss": 0.014015881344676018,
      "step": 2168
    },
    {
      "epoch": 0.41129643669446553,
      "grad_norm": 0.294528990983963,
      "learning_rate": 0.0002,
      "loss": 0.021655144169926643,
      "step": 2170
    },
    {
      "epoch": 0.4116755117513268,
      "grad_norm": 0.34171566367149353,
      "learning_rate": 0.0002,
      "loss": 0.021504057571291924,
      "step": 2172
    },
    {
      "epoch": 0.412054586808188,
      "grad_norm": 0.447454035282135,
      "learning_rate": 0.0002,
      "loss": 0.05285695567727089,
      "step": 2174
    },
    {
      "epoch": 0.41243366186504926,
      "grad_norm": 0.10589074343442917,
      "learning_rate": 0.0002,
      "loss": 0.01979535073041916,
      "step": 2176
    },
    {
      "epoch": 0.41281273692191056,
      "grad_norm": 0.2854958772659302,
      "learning_rate": 0.0002,
      "loss": 0.024306582286953926,
      "step": 2178
    },
    {
      "epoch": 0.4131918119787718,
      "grad_norm": 0.19719398021697998,
      "learning_rate": 0.0002,
      "loss": 0.01923295482993126,
      "step": 2180
    },
    {
      "epoch": 0.41357088703563305,
      "grad_norm": 0.4988541305065155,
      "learning_rate": 0.0002,
      "loss": 0.04280329495668411,
      "step": 2182
    },
    {
      "epoch": 0.4139499620924943,
      "grad_norm": 0.6112233996391296,
      "learning_rate": 0.0002,
      "loss": 0.059212543070316315,
      "step": 2184
    },
    {
      "epoch": 0.4143290371493556,
      "grad_norm": 0.3590647876262665,
      "learning_rate": 0.0002,
      "loss": 0.023816654458642006,
      "step": 2186
    },
    {
      "epoch": 0.41470811220621684,
      "grad_norm": 0.4340093731880188,
      "learning_rate": 0.0002,
      "loss": 0.03184362128376961,
      "step": 2188
    },
    {
      "epoch": 0.4150871872630781,
      "grad_norm": 0.3399653732776642,
      "learning_rate": 0.0002,
      "loss": 0.03489590808749199,
      "step": 2190
    },
    {
      "epoch": 0.41546626231993933,
      "grad_norm": 0.12037285417318344,
      "learning_rate": 0.0002,
      "loss": 0.013541767373681068,
      "step": 2192
    },
    {
      "epoch": 0.41584533737680063,
      "grad_norm": 0.26347583532333374,
      "learning_rate": 0.0002,
      "loss": 0.018305379897356033,
      "step": 2194
    },
    {
      "epoch": 0.4162244124336619,
      "grad_norm": 0.17667879164218903,
      "learning_rate": 0.0002,
      "loss": 0.043291542679071426,
      "step": 2196
    },
    {
      "epoch": 0.4166034874905231,
      "grad_norm": 0.142020121216774,
      "learning_rate": 0.0002,
      "loss": 0.01717674918472767,
      "step": 2198
    },
    {
      "epoch": 0.41698256254738436,
      "grad_norm": 0.11874847114086151,
      "learning_rate": 0.0002,
      "loss": 0.018443461507558823,
      "step": 2200
    },
    {
      "epoch": 0.41736163760424566,
      "grad_norm": 0.1118677631020546,
      "learning_rate": 0.0002,
      "loss": 0.01679084077477455,
      "step": 2202
    },
    {
      "epoch": 0.4177407126611069,
      "grad_norm": 0.39147645235061646,
      "learning_rate": 0.0002,
      "loss": 0.0396742969751358,
      "step": 2204
    },
    {
      "epoch": 0.41811978771796815,
      "grad_norm": 0.3887481093406677,
      "learning_rate": 0.0002,
      "loss": 0.025089547038078308,
      "step": 2206
    },
    {
      "epoch": 0.4184988627748294,
      "grad_norm": 0.38606157898902893,
      "learning_rate": 0.0002,
      "loss": 0.03878406062722206,
      "step": 2208
    },
    {
      "epoch": 0.4188779378316907,
      "grad_norm": 0.10412921756505966,
      "learning_rate": 0.0002,
      "loss": 0.013634041883051395,
      "step": 2210
    },
    {
      "epoch": 0.41925701288855194,
      "grad_norm": 0.38007768988609314,
      "learning_rate": 0.0002,
      "loss": 0.04874414578080177,
      "step": 2212
    },
    {
      "epoch": 0.4196360879454132,
      "grad_norm": 0.09020747989416122,
      "learning_rate": 0.0002,
      "loss": 0.014755737036466599,
      "step": 2214
    },
    {
      "epoch": 0.4200151630022744,
      "grad_norm": 0.46710842847824097,
      "learning_rate": 0.0002,
      "loss": 0.0402301661670208,
      "step": 2216
    },
    {
      "epoch": 0.4203942380591357,
      "grad_norm": 0.4948008358478546,
      "learning_rate": 0.0002,
      "loss": 0.02833659015595913,
      "step": 2218
    },
    {
      "epoch": 0.42077331311599697,
      "grad_norm": 0.30455854535102844,
      "learning_rate": 0.0002,
      "loss": 0.027570180594921112,
      "step": 2220
    },
    {
      "epoch": 0.4211523881728582,
      "grad_norm": 0.11195392161607742,
      "learning_rate": 0.0002,
      "loss": 0.024108845740556717,
      "step": 2222
    },
    {
      "epoch": 0.42153146322971946,
      "grad_norm": 0.28630396723747253,
      "learning_rate": 0.0002,
      "loss": 0.04454338550567627,
      "step": 2224
    },
    {
      "epoch": 0.42191053828658076,
      "grad_norm": 0.12726990878582,
      "learning_rate": 0.0002,
      "loss": 0.016565794125199318,
      "step": 2226
    },
    {
      "epoch": 0.422289613343442,
      "grad_norm": 0.15536756813526154,
      "learning_rate": 0.0002,
      "loss": 0.024351682513952255,
      "step": 2228
    },
    {
      "epoch": 0.42266868840030325,
      "grad_norm": 0.2552931606769562,
      "learning_rate": 0.0002,
      "loss": 0.012890986166894436,
      "step": 2230
    },
    {
      "epoch": 0.4230477634571645,
      "grad_norm": 0.34179627895355225,
      "learning_rate": 0.0002,
      "loss": 0.024758636951446533,
      "step": 2232
    },
    {
      "epoch": 0.4234268385140258,
      "grad_norm": 0.10349880158901215,
      "learning_rate": 0.0002,
      "loss": 0.028499584645032883,
      "step": 2234
    },
    {
      "epoch": 0.42380591357088704,
      "grad_norm": 0.16802942752838135,
      "learning_rate": 0.0002,
      "loss": 0.020813260227441788,
      "step": 2236
    },
    {
      "epoch": 0.4241849886277483,
      "grad_norm": 0.41569310426712036,
      "learning_rate": 0.0002,
      "loss": 0.026287134736776352,
      "step": 2238
    },
    {
      "epoch": 0.4245640636846095,
      "grad_norm": 0.284469336271286,
      "learning_rate": 0.0002,
      "loss": 0.03456566855311394,
      "step": 2240
    },
    {
      "epoch": 0.4249431387414708,
      "grad_norm": 0.31550589203834534,
      "learning_rate": 0.0002,
      "loss": 0.032671522349119186,
      "step": 2242
    },
    {
      "epoch": 0.42532221379833207,
      "grad_norm": 0.31817400455474854,
      "learning_rate": 0.0002,
      "loss": 0.014538212679326534,
      "step": 2244
    },
    {
      "epoch": 0.4257012888551933,
      "grad_norm": 0.2953798770904541,
      "learning_rate": 0.0002,
      "loss": 0.01868695579469204,
      "step": 2246
    },
    {
      "epoch": 0.42608036391205456,
      "grad_norm": 0.34703800082206726,
      "learning_rate": 0.0002,
      "loss": 0.045497242361307144,
      "step": 2248
    },
    {
      "epoch": 0.42645943896891586,
      "grad_norm": 0.15105131268501282,
      "learning_rate": 0.0002,
      "loss": 0.013192162849009037,
      "step": 2250
    },
    {
      "epoch": 0.4268385140257771,
      "grad_norm": 0.486936092376709,
      "learning_rate": 0.0002,
      "loss": 0.017041634768247604,
      "step": 2252
    },
    {
      "epoch": 0.42721758908263835,
      "grad_norm": 0.4455849230289459,
      "learning_rate": 0.0002,
      "loss": 0.04588693752884865,
      "step": 2254
    },
    {
      "epoch": 0.4275966641394996,
      "grad_norm": 0.3686142563819885,
      "learning_rate": 0.0002,
      "loss": 0.026409994810819626,
      "step": 2256
    },
    {
      "epoch": 0.4279757391963609,
      "grad_norm": 0.1015091985464096,
      "learning_rate": 0.0002,
      "loss": 0.012055656872689724,
      "step": 2258
    },
    {
      "epoch": 0.42835481425322214,
      "grad_norm": 0.3865933418273926,
      "learning_rate": 0.0002,
      "loss": 0.03182142972946167,
      "step": 2260
    },
    {
      "epoch": 0.4287338893100834,
      "grad_norm": 0.25574997067451477,
      "learning_rate": 0.0002,
      "loss": 0.017053771764039993,
      "step": 2262
    },
    {
      "epoch": 0.4291129643669447,
      "grad_norm": 0.10296644270420074,
      "learning_rate": 0.0002,
      "loss": 0.026721619069576263,
      "step": 2264
    },
    {
      "epoch": 0.4294920394238059,
      "grad_norm": 0.11528647691011429,
      "learning_rate": 0.0002,
      "loss": 0.024054858833551407,
      "step": 2266
    },
    {
      "epoch": 0.42987111448066717,
      "grad_norm": 0.37490880489349365,
      "learning_rate": 0.0002,
      "loss": 0.03667489066720009,
      "step": 2268
    },
    {
      "epoch": 0.4302501895375284,
      "grad_norm": 0.5749694108963013,
      "learning_rate": 0.0002,
      "loss": 0.04543526470661163,
      "step": 2270
    },
    {
      "epoch": 0.4306292645943897,
      "grad_norm": 0.14483197033405304,
      "learning_rate": 0.0002,
      "loss": 0.010389521718025208,
      "step": 2272
    },
    {
      "epoch": 0.43100833965125096,
      "grad_norm": 0.4152519702911377,
      "learning_rate": 0.0002,
      "loss": 0.02182708866894245,
      "step": 2274
    },
    {
      "epoch": 0.4313874147081122,
      "grad_norm": 0.5196089744567871,
      "learning_rate": 0.0002,
      "loss": 0.04205603152513504,
      "step": 2276
    },
    {
      "epoch": 0.43176648976497345,
      "grad_norm": 0.32008442282676697,
      "learning_rate": 0.0002,
      "loss": 0.025270448997616768,
      "step": 2278
    },
    {
      "epoch": 0.43214556482183475,
      "grad_norm": 0.27811282873153687,
      "learning_rate": 0.0002,
      "loss": 0.013962823897600174,
      "step": 2280
    },
    {
      "epoch": 0.432524639878696,
      "grad_norm": 0.10612110048532486,
      "learning_rate": 0.0002,
      "loss": 0.023924291133880615,
      "step": 2282
    },
    {
      "epoch": 0.43290371493555724,
      "grad_norm": 0.09162341058254242,
      "learning_rate": 0.0002,
      "loss": 0.01650424860417843,
      "step": 2284
    },
    {
      "epoch": 0.4332827899924185,
      "grad_norm": 0.12283685803413391,
      "learning_rate": 0.0002,
      "loss": 0.018591823056340218,
      "step": 2286
    },
    {
      "epoch": 0.4336618650492798,
      "grad_norm": 0.13837744295597076,
      "learning_rate": 0.0002,
      "loss": 0.03668246790766716,
      "step": 2288
    },
    {
      "epoch": 0.434040940106141,
      "grad_norm": 0.17239834368228912,
      "learning_rate": 0.0002,
      "loss": 0.0357171855866909,
      "step": 2290
    },
    {
      "epoch": 0.43442001516300227,
      "grad_norm": 0.3778376579284668,
      "learning_rate": 0.0002,
      "loss": 0.03032105229794979,
      "step": 2292
    },
    {
      "epoch": 0.4347990902198635,
      "grad_norm": 0.164651021361351,
      "learning_rate": 0.0002,
      "loss": 0.014588196761906147,
      "step": 2294
    },
    {
      "epoch": 0.4351781652767248,
      "grad_norm": 0.21229249238967896,
      "learning_rate": 0.0002,
      "loss": 0.015660881996154785,
      "step": 2296
    },
    {
      "epoch": 0.43555724033358606,
      "grad_norm": 0.4840068221092224,
      "learning_rate": 0.0002,
      "loss": 0.03365841880440712,
      "step": 2298
    },
    {
      "epoch": 0.4359363153904473,
      "grad_norm": 0.2948763370513916,
      "learning_rate": 0.0002,
      "loss": 0.03675883263349533,
      "step": 2300
    },
    {
      "epoch": 0.43631539044730855,
      "grad_norm": 0.12205833196640015,
      "learning_rate": 0.0002,
      "loss": 0.02388700470328331,
      "step": 2302
    },
    {
      "epoch": 0.43669446550416985,
      "grad_norm": 0.2225075513124466,
      "learning_rate": 0.0002,
      "loss": 0.020768333226442337,
      "step": 2304
    },
    {
      "epoch": 0.4370735405610311,
      "grad_norm": 0.24555152654647827,
      "learning_rate": 0.0002,
      "loss": 0.02815476804971695,
      "step": 2306
    },
    {
      "epoch": 0.43745261561789234,
      "grad_norm": 0.1273302286863327,
      "learning_rate": 0.0002,
      "loss": 0.02735110931098461,
      "step": 2308
    },
    {
      "epoch": 0.4378316906747536,
      "grad_norm": 0.14780940115451813,
      "learning_rate": 0.0002,
      "loss": 0.021796755492687225,
      "step": 2310
    },
    {
      "epoch": 0.4382107657316149,
      "grad_norm": 0.2913970649242401,
      "learning_rate": 0.0002,
      "loss": 0.020407000556588173,
      "step": 2312
    },
    {
      "epoch": 0.4385898407884761,
      "grad_norm": 0.22214241325855255,
      "learning_rate": 0.0002,
      "loss": 0.019297264516353607,
      "step": 2314
    },
    {
      "epoch": 0.43896891584533737,
      "grad_norm": 0.12101015448570251,
      "learning_rate": 0.0002,
      "loss": 0.014043323695659637,
      "step": 2316
    },
    {
      "epoch": 0.4393479909021986,
      "grad_norm": 0.4345197379589081,
      "learning_rate": 0.0002,
      "loss": 0.02638918161392212,
      "step": 2318
    },
    {
      "epoch": 0.4397270659590599,
      "grad_norm": 0.37989699840545654,
      "learning_rate": 0.0002,
      "loss": 0.022709643468260765,
      "step": 2320
    },
    {
      "epoch": 0.44010614101592116,
      "grad_norm": 0.09894780814647675,
      "learning_rate": 0.0002,
      "loss": 0.01954069174826145,
      "step": 2322
    },
    {
      "epoch": 0.4404852160727824,
      "grad_norm": 0.13957110047340393,
      "learning_rate": 0.0002,
      "loss": 0.015580825507640839,
      "step": 2324
    },
    {
      "epoch": 0.44086429112964365,
      "grad_norm": 0.22013543546199799,
      "learning_rate": 0.0002,
      "loss": 0.017574070021510124,
      "step": 2326
    },
    {
      "epoch": 0.44124336618650495,
      "grad_norm": 0.37112197279930115,
      "learning_rate": 0.0002,
      "loss": 0.027101965621113777,
      "step": 2328
    },
    {
      "epoch": 0.4416224412433662,
      "grad_norm": 0.49284279346466064,
      "learning_rate": 0.0002,
      "loss": 0.03835270553827286,
      "step": 2330
    },
    {
      "epoch": 0.44200151630022744,
      "grad_norm": 0.12277010828256607,
      "learning_rate": 0.0002,
      "loss": 0.021444527432322502,
      "step": 2332
    },
    {
      "epoch": 0.4423805913570887,
      "grad_norm": 0.13368040323257446,
      "learning_rate": 0.0002,
      "loss": 0.01521667093038559,
      "step": 2334
    },
    {
      "epoch": 0.44275966641395,
      "grad_norm": 0.48285192251205444,
      "learning_rate": 0.0002,
      "loss": 0.0370807908475399,
      "step": 2336
    },
    {
      "epoch": 0.4431387414708112,
      "grad_norm": 0.12532734870910645,
      "learning_rate": 0.0002,
      "loss": 0.025063110515475273,
      "step": 2338
    },
    {
      "epoch": 0.44351781652767247,
      "grad_norm": 0.14227524399757385,
      "learning_rate": 0.0002,
      "loss": 0.02790374495089054,
      "step": 2340
    },
    {
      "epoch": 0.4438968915845337,
      "grad_norm": 0.251965194940567,
      "learning_rate": 0.0002,
      "loss": 0.0187508687376976,
      "step": 2342
    },
    {
      "epoch": 0.444275966641395,
      "grad_norm": 0.17517860233783722,
      "learning_rate": 0.0002,
      "loss": 0.0146974828094244,
      "step": 2344
    },
    {
      "epoch": 0.44465504169825626,
      "grad_norm": 0.25292181968688965,
      "learning_rate": 0.0002,
      "loss": 0.01858201064169407,
      "step": 2346
    },
    {
      "epoch": 0.4450341167551175,
      "grad_norm": 0.3896804749965668,
      "learning_rate": 0.0002,
      "loss": 0.041532982140779495,
      "step": 2348
    },
    {
      "epoch": 0.44541319181197875,
      "grad_norm": 0.2885786294937134,
      "learning_rate": 0.0002,
      "loss": 0.029649903997778893,
      "step": 2350
    },
    {
      "epoch": 0.44579226686884005,
      "grad_norm": 0.4140695631504059,
      "learning_rate": 0.0002,
      "loss": 0.037261225283145905,
      "step": 2352
    },
    {
      "epoch": 0.4461713419257013,
      "grad_norm": 0.10602843016386032,
      "learning_rate": 0.0002,
      "loss": 0.018051261082291603,
      "step": 2354
    },
    {
      "epoch": 0.44655041698256254,
      "grad_norm": 0.209665909409523,
      "learning_rate": 0.0002,
      "loss": 0.024220814928412437,
      "step": 2356
    },
    {
      "epoch": 0.4469294920394238,
      "grad_norm": 0.45807892084121704,
      "learning_rate": 0.0002,
      "loss": 0.03252968564629555,
      "step": 2358
    },
    {
      "epoch": 0.4473085670962851,
      "grad_norm": 0.18300233781337738,
      "learning_rate": 0.0002,
      "loss": 0.013130188919603825,
      "step": 2360
    },
    {
      "epoch": 0.4476876421531463,
      "grad_norm": 0.30264613032341003,
      "learning_rate": 0.0002,
      "loss": 0.02002056874334812,
      "step": 2362
    },
    {
      "epoch": 0.44806671721000757,
      "grad_norm": 0.10990407317876816,
      "learning_rate": 0.0002,
      "loss": 0.025429120287299156,
      "step": 2364
    },
    {
      "epoch": 0.4484457922668688,
      "grad_norm": 0.12857642769813538,
      "learning_rate": 0.0002,
      "loss": 0.014776235446333885,
      "step": 2366
    },
    {
      "epoch": 0.4488248673237301,
      "grad_norm": 0.3092147707939148,
      "learning_rate": 0.0002,
      "loss": 0.02407577820122242,
      "step": 2368
    },
    {
      "epoch": 0.44920394238059136,
      "grad_norm": 0.11340469121932983,
      "learning_rate": 0.0002,
      "loss": 0.01865079440176487,
      "step": 2370
    },
    {
      "epoch": 0.4495830174374526,
      "grad_norm": 0.26708266139030457,
      "learning_rate": 0.0002,
      "loss": 0.020554468035697937,
      "step": 2372
    },
    {
      "epoch": 0.44996209249431385,
      "grad_norm": 0.10331190377473831,
      "learning_rate": 0.0002,
      "loss": 0.0214383527636528,
      "step": 2374
    },
    {
      "epoch": 0.45034116755117515,
      "grad_norm": 0.4664556384086609,
      "learning_rate": 0.0002,
      "loss": 0.04275020211935043,
      "step": 2376
    },
    {
      "epoch": 0.4507202426080364,
      "grad_norm": 0.3783089518547058,
      "learning_rate": 0.0002,
      "loss": 0.024038225412368774,
      "step": 2378
    },
    {
      "epoch": 0.45109931766489764,
      "grad_norm": 0.1557580679655075,
      "learning_rate": 0.0002,
      "loss": 0.02062215283513069,
      "step": 2380
    },
    {
      "epoch": 0.4514783927217589,
      "grad_norm": 0.35908350348472595,
      "learning_rate": 0.0002,
      "loss": 0.02231615036725998,
      "step": 2382
    },
    {
      "epoch": 0.4518574677786202,
      "grad_norm": 0.3981534242630005,
      "learning_rate": 0.0002,
      "loss": 0.033430490642786026,
      "step": 2384
    },
    {
      "epoch": 0.4522365428354814,
      "grad_norm": 0.4027352035045624,
      "learning_rate": 0.0002,
      "loss": 0.033243417739868164,
      "step": 2386
    },
    {
      "epoch": 0.45261561789234267,
      "grad_norm": 0.13106557726860046,
      "learning_rate": 0.0002,
      "loss": 0.015053030103445053,
      "step": 2388
    },
    {
      "epoch": 0.45299469294920397,
      "grad_norm": 0.3947465717792511,
      "learning_rate": 0.0002,
      "loss": 0.02906070277094841,
      "step": 2390
    },
    {
      "epoch": 0.4533737680060652,
      "grad_norm": 0.4785026013851166,
      "learning_rate": 0.0002,
      "loss": 0.03691861033439636,
      "step": 2392
    },
    {
      "epoch": 0.45375284306292646,
      "grad_norm": 0.4016043245792389,
      "learning_rate": 0.0002,
      "loss": 0.029727725312113762,
      "step": 2394
    },
    {
      "epoch": 0.4541319181197877,
      "grad_norm": 0.10675439238548279,
      "learning_rate": 0.0002,
      "loss": 0.028287936002016068,
      "step": 2396
    },
    {
      "epoch": 0.454510993176649,
      "grad_norm": 0.20129506289958954,
      "learning_rate": 0.0002,
      "loss": 0.03435128554701805,
      "step": 2398
    },
    {
      "epoch": 0.45489006823351025,
      "grad_norm": 0.13506336510181427,
      "learning_rate": 0.0002,
      "loss": 0.023977553471922874,
      "step": 2400
    },
    {
      "epoch": 0.4552691432903715,
      "grad_norm": 0.448820561170578,
      "learning_rate": 0.0002,
      "loss": 0.042170632630586624,
      "step": 2402
    },
    {
      "epoch": 0.45564821834723274,
      "grad_norm": 0.12426264584064484,
      "learning_rate": 0.0002,
      "loss": 0.021638302132487297,
      "step": 2404
    },
    {
      "epoch": 0.45602729340409404,
      "grad_norm": 0.1555943787097931,
      "learning_rate": 0.0002,
      "loss": 0.02458009496331215,
      "step": 2406
    },
    {
      "epoch": 0.4564063684609553,
      "grad_norm": 0.32456174492836,
      "learning_rate": 0.0002,
      "loss": 0.023433122783899307,
      "step": 2408
    },
    {
      "epoch": 0.4567854435178165,
      "grad_norm": 0.3600495159626007,
      "learning_rate": 0.0002,
      "loss": 0.025607166811823845,
      "step": 2410
    },
    {
      "epoch": 0.45716451857467777,
      "grad_norm": 0.1378241777420044,
      "learning_rate": 0.0002,
      "loss": 0.024819616228342056,
      "step": 2412
    },
    {
      "epoch": 0.45754359363153907,
      "grad_norm": 0.47085538506507874,
      "learning_rate": 0.0002,
      "loss": 0.0411951057612896,
      "step": 2414
    },
    {
      "epoch": 0.4579226686884003,
      "grad_norm": 0.4002079665660858,
      "learning_rate": 0.0002,
      "loss": 0.0360286608338356,
      "step": 2416
    },
    {
      "epoch": 0.45830174374526156,
      "grad_norm": 0.3116513788700104,
      "learning_rate": 0.0002,
      "loss": 0.02177092432975769,
      "step": 2418
    },
    {
      "epoch": 0.4586808188021228,
      "grad_norm": 0.3212018311023712,
      "learning_rate": 0.0002,
      "loss": 0.01923115737736225,
      "step": 2420
    },
    {
      "epoch": 0.4590598938589841,
      "grad_norm": 0.37858259677886963,
      "learning_rate": 0.0002,
      "loss": 0.02471487782895565,
      "step": 2422
    },
    {
      "epoch": 0.45943896891584535,
      "grad_norm": 0.14719854295253754,
      "learning_rate": 0.0002,
      "loss": 0.01251075230538845,
      "step": 2424
    },
    {
      "epoch": 0.4598180439727066,
      "grad_norm": 0.5650354027748108,
      "learning_rate": 0.0002,
      "loss": 0.0402895025908947,
      "step": 2426
    },
    {
      "epoch": 0.46019711902956784,
      "grad_norm": 0.30454301834106445,
      "learning_rate": 0.0002,
      "loss": 0.014005903154611588,
      "step": 2428
    },
    {
      "epoch": 0.46057619408642914,
      "grad_norm": 0.33392730355262756,
      "learning_rate": 0.0002,
      "loss": 0.02729184739291668,
      "step": 2430
    },
    {
      "epoch": 0.4609552691432904,
      "grad_norm": 0.15049965679645538,
      "learning_rate": 0.0002,
      "loss": 0.03203646093606949,
      "step": 2432
    },
    {
      "epoch": 0.4613343442001516,
      "grad_norm": 0.14126455783843994,
      "learning_rate": 0.0002,
      "loss": 0.011853653937578201,
      "step": 2434
    },
    {
      "epoch": 0.46171341925701287,
      "grad_norm": 0.13984334468841553,
      "learning_rate": 0.0002,
      "loss": 0.016745394095778465,
      "step": 2436
    },
    {
      "epoch": 0.46209249431387417,
      "grad_norm": 0.46248307824134827,
      "learning_rate": 0.0002,
      "loss": 0.03846224397420883,
      "step": 2438
    },
    {
      "epoch": 0.4624715693707354,
      "grad_norm": 0.12422369420528412,
      "learning_rate": 0.0002,
      "loss": 0.01289450004696846,
      "step": 2440
    },
    {
      "epoch": 0.46285064442759666,
      "grad_norm": 0.09357895702123642,
      "learning_rate": 0.0002,
      "loss": 0.013986271806061268,
      "step": 2442
    },
    {
      "epoch": 0.4632297194844579,
      "grad_norm": 0.43572095036506653,
      "learning_rate": 0.0002,
      "loss": 0.03939120098948479,
      "step": 2444
    },
    {
      "epoch": 0.4636087945413192,
      "grad_norm": 0.11214498430490494,
      "learning_rate": 0.0002,
      "loss": 0.03058185987174511,
      "step": 2446
    },
    {
      "epoch": 0.46398786959818045,
      "grad_norm": 0.2558038830757141,
      "learning_rate": 0.0002,
      "loss": 0.03080921806395054,
      "step": 2448
    },
    {
      "epoch": 0.4643669446550417,
      "grad_norm": 0.17213119566440582,
      "learning_rate": 0.0002,
      "loss": 0.028543634340167046,
      "step": 2450
    },
    {
      "epoch": 0.46474601971190294,
      "grad_norm": 0.4556381404399872,
      "learning_rate": 0.0002,
      "loss": 0.03427255526185036,
      "step": 2452
    },
    {
      "epoch": 0.46512509476876424,
      "grad_norm": 0.08073530346155167,
      "learning_rate": 0.0002,
      "loss": 0.038199808448553085,
      "step": 2454
    },
    {
      "epoch": 0.4655041698256255,
      "grad_norm": 0.370016872882843,
      "learning_rate": 0.0002,
      "loss": 0.03307857736945152,
      "step": 2456
    },
    {
      "epoch": 0.4658832448824867,
      "grad_norm": 0.3882402777671814,
      "learning_rate": 0.0002,
      "loss": 0.02492375113070011,
      "step": 2458
    },
    {
      "epoch": 0.46626231993934797,
      "grad_norm": 0.432855486869812,
      "learning_rate": 0.0002,
      "loss": 0.021265940740704536,
      "step": 2460
    },
    {
      "epoch": 0.46664139499620927,
      "grad_norm": 0.43947964906692505,
      "learning_rate": 0.0002,
      "loss": 0.03196370229125023,
      "step": 2462
    },
    {
      "epoch": 0.4670204700530705,
      "grad_norm": 0.09703647345304489,
      "learning_rate": 0.0002,
      "loss": 0.015638839453458786,
      "step": 2464
    },
    {
      "epoch": 0.46739954510993176,
      "grad_norm": 0.2094346433877945,
      "learning_rate": 0.0002,
      "loss": 0.01691771112382412,
      "step": 2466
    },
    {
      "epoch": 0.467778620166793,
      "grad_norm": 0.3072778582572937,
      "learning_rate": 0.0002,
      "loss": 0.023778459057211876,
      "step": 2468
    },
    {
      "epoch": 0.4681576952236543,
      "grad_norm": 0.48535844683647156,
      "learning_rate": 0.0002,
      "loss": 0.034591350704431534,
      "step": 2470
    },
    {
      "epoch": 0.46853677028051555,
      "grad_norm": 0.4037223160266876,
      "learning_rate": 0.0002,
      "loss": 0.04422210901975632,
      "step": 2472
    },
    {
      "epoch": 0.4689158453373768,
      "grad_norm": 0.08801353722810745,
      "learning_rate": 0.0002,
      "loss": 0.02818048559129238,
      "step": 2474
    },
    {
      "epoch": 0.46929492039423804,
      "grad_norm": 0.17665427923202515,
      "learning_rate": 0.0002,
      "loss": 0.015133410692214966,
      "step": 2476
    },
    {
      "epoch": 0.46967399545109934,
      "grad_norm": 0.14134792983531952,
      "learning_rate": 0.0002,
      "loss": 0.01999078318476677,
      "step": 2478
    },
    {
      "epoch": 0.4700530705079606,
      "grad_norm": 0.11938256770372391,
      "learning_rate": 0.0002,
      "loss": 0.012712026946246624,
      "step": 2480
    },
    {
      "epoch": 0.4704321455648218,
      "grad_norm": 0.44495633244514465,
      "learning_rate": 0.0002,
      "loss": 0.04010077565908432,
      "step": 2482
    },
    {
      "epoch": 0.47081122062168307,
      "grad_norm": 0.48785775899887085,
      "learning_rate": 0.0002,
      "loss": 0.0181258637458086,
      "step": 2484
    },
    {
      "epoch": 0.47119029567854437,
      "grad_norm": 0.37636590003967285,
      "learning_rate": 0.0002,
      "loss": 0.04164917767047882,
      "step": 2486
    },
    {
      "epoch": 0.4715693707354056,
      "grad_norm": 0.45095106959342957,
      "learning_rate": 0.0002,
      "loss": 0.03497348353266716,
      "step": 2488
    },
    {
      "epoch": 0.47194844579226686,
      "grad_norm": 0.11939138174057007,
      "learning_rate": 0.0002,
      "loss": 0.015201631002128124,
      "step": 2490
    },
    {
      "epoch": 0.4723275208491281,
      "grad_norm": 0.3502998650074005,
      "learning_rate": 0.0002,
      "loss": 0.03251519799232483,
      "step": 2492
    },
    {
      "epoch": 0.4727065959059894,
      "grad_norm": 0.18448609113693237,
      "learning_rate": 0.0002,
      "loss": 0.018447518348693848,
      "step": 2494
    },
    {
      "epoch": 0.47308567096285065,
      "grad_norm": 0.11366496235132217,
      "learning_rate": 0.0002,
      "loss": 0.02976544201374054,
      "step": 2496
    },
    {
      "epoch": 0.4734647460197119,
      "grad_norm": 0.2914569675922394,
      "learning_rate": 0.0002,
      "loss": 0.023976128548383713,
      "step": 2498
    },
    {
      "epoch": 0.47384382107657314,
      "grad_norm": 0.3390690088272095,
      "learning_rate": 0.0002,
      "loss": 0.03946123644709587,
      "step": 2500
    },
    {
      "epoch": 0.47422289613343444,
      "grad_norm": 0.27203699946403503,
      "learning_rate": 0.0002,
      "loss": 0.01610974408686161,
      "step": 2502
    },
    {
      "epoch": 0.4746019711902957,
      "grad_norm": 0.31854259967803955,
      "learning_rate": 0.0002,
      "loss": 0.02296721190214157,
      "step": 2504
    },
    {
      "epoch": 0.4749810462471569,
      "grad_norm": 0.1318994015455246,
      "learning_rate": 0.0002,
      "loss": 0.01531603466719389,
      "step": 2506
    },
    {
      "epoch": 0.47536012130401817,
      "grad_norm": 0.1410343199968338,
      "learning_rate": 0.0002,
      "loss": 0.017761701717972755,
      "step": 2508
    },
    {
      "epoch": 0.47573919636087947,
      "grad_norm": 0.46691879630088806,
      "learning_rate": 0.0002,
      "loss": 0.05405057221651077,
      "step": 2510
    },
    {
      "epoch": 0.4761182714177407,
      "grad_norm": 0.3625180423259735,
      "learning_rate": 0.0002,
      "loss": 0.0297694094479084,
      "step": 2512
    },
    {
      "epoch": 0.47649734647460196,
      "grad_norm": 0.10899563878774643,
      "learning_rate": 0.0002,
      "loss": 0.013208113610744476,
      "step": 2514
    },
    {
      "epoch": 0.47687642153146326,
      "grad_norm": 0.11782027781009674,
      "learning_rate": 0.0002,
      "loss": 0.02340904250741005,
      "step": 2516
    },
    {
      "epoch": 0.4772554965883245,
      "grad_norm": 0.13397538661956787,
      "learning_rate": 0.0002,
      "loss": 0.019740834832191467,
      "step": 2518
    },
    {
      "epoch": 0.47763457164518575,
      "grad_norm": 0.5969170331954956,
      "learning_rate": 0.0002,
      "loss": 0.06077110394835472,
      "step": 2520
    },
    {
      "epoch": 0.478013646702047,
      "grad_norm": 0.12052387744188309,
      "learning_rate": 0.0002,
      "loss": 0.013634641654789448,
      "step": 2522
    },
    {
      "epoch": 0.4783927217589083,
      "grad_norm": 0.4338090121746063,
      "learning_rate": 0.0002,
      "loss": 0.02507498487830162,
      "step": 2524
    },
    {
      "epoch": 0.47877179681576953,
      "grad_norm": 0.0958455428481102,
      "learning_rate": 0.0002,
      "loss": 0.03981749713420868,
      "step": 2526
    },
    {
      "epoch": 0.4791508718726308,
      "grad_norm": 0.4840330481529236,
      "learning_rate": 0.0002,
      "loss": 0.027304627001285553,
      "step": 2528
    },
    {
      "epoch": 0.479529946929492,
      "grad_norm": 0.3676588237285614,
      "learning_rate": 0.0002,
      "loss": 0.02126859873533249,
      "step": 2530
    },
    {
      "epoch": 0.4799090219863533,
      "grad_norm": 0.379626601934433,
      "learning_rate": 0.0002,
      "loss": 0.036524634808301926,
      "step": 2532
    },
    {
      "epoch": 0.48028809704321457,
      "grad_norm": 0.2914204001426697,
      "learning_rate": 0.0002,
      "loss": 0.026256782934069633,
      "step": 2534
    },
    {
      "epoch": 0.4806671721000758,
      "grad_norm": 0.25586655735969543,
      "learning_rate": 0.0002,
      "loss": 0.0256264116615057,
      "step": 2536
    },
    {
      "epoch": 0.48104624715693706,
      "grad_norm": 0.20103764533996582,
      "learning_rate": 0.0002,
      "loss": 0.014733297750353813,
      "step": 2538
    },
    {
      "epoch": 0.48142532221379836,
      "grad_norm": 0.20276252925395966,
      "learning_rate": 0.0002,
      "loss": 0.022997181862592697,
      "step": 2540
    },
    {
      "epoch": 0.4818043972706596,
      "grad_norm": 0.0989723727107048,
      "learning_rate": 0.0002,
      "loss": 0.014114322140812874,
      "step": 2542
    },
    {
      "epoch": 0.48218347232752085,
      "grad_norm": 0.10162290185689926,
      "learning_rate": 0.0002,
      "loss": 0.014615937136113644,
      "step": 2544
    },
    {
      "epoch": 0.4825625473843821,
      "grad_norm": 0.4135981798171997,
      "learning_rate": 0.0002,
      "loss": 0.03778018802404404,
      "step": 2546
    },
    {
      "epoch": 0.4829416224412434,
      "grad_norm": 0.13332928717136383,
      "learning_rate": 0.0002,
      "loss": 0.023062707856297493,
      "step": 2548
    },
    {
      "epoch": 0.48332069749810463,
      "grad_norm": 0.4872605502605438,
      "learning_rate": 0.0002,
      "loss": 0.0355270616710186,
      "step": 2550
    },
    {
      "epoch": 0.4836997725549659,
      "grad_norm": 0.16766686737537384,
      "learning_rate": 0.0002,
      "loss": 0.03351651504635811,
      "step": 2552
    },
    {
      "epoch": 0.4840788476118271,
      "grad_norm": 0.47870638966560364,
      "learning_rate": 0.0002,
      "loss": 0.0321989431977272,
      "step": 2554
    },
    {
      "epoch": 0.4844579226686884,
      "grad_norm": 0.14759448170661926,
      "learning_rate": 0.0002,
      "loss": 0.029338672757148743,
      "step": 2556
    },
    {
      "epoch": 0.48483699772554967,
      "grad_norm": 0.5038665533065796,
      "learning_rate": 0.0002,
      "loss": 0.04380141943693161,
      "step": 2558
    },
    {
      "epoch": 0.4852160727824109,
      "grad_norm": 0.14351506531238556,
      "learning_rate": 0.0002,
      "loss": 0.013832271099090576,
      "step": 2560
    },
    {
      "epoch": 0.48559514783927216,
      "grad_norm": 0.09428493678569794,
      "learning_rate": 0.0002,
      "loss": 0.01774647831916809,
      "step": 2562
    },
    {
      "epoch": 0.48597422289613346,
      "grad_norm": 0.4084102213382721,
      "learning_rate": 0.0002,
      "loss": 0.04000246897339821,
      "step": 2564
    },
    {
      "epoch": 0.4863532979529947,
      "grad_norm": 0.49127379059791565,
      "learning_rate": 0.0002,
      "loss": 0.04239773377776146,
      "step": 2566
    },
    {
      "epoch": 0.48673237300985595,
      "grad_norm": 0.325455904006958,
      "learning_rate": 0.0002,
      "loss": 0.024101028218865395,
      "step": 2568
    },
    {
      "epoch": 0.4871114480667172,
      "grad_norm": 0.10583784431219101,
      "learning_rate": 0.0002,
      "loss": 0.03253728523850441,
      "step": 2570
    },
    {
      "epoch": 0.4874905231235785,
      "grad_norm": 0.1326970010995865,
      "learning_rate": 0.0002,
      "loss": 0.03136704862117767,
      "step": 2572
    },
    {
      "epoch": 0.48786959818043973,
      "grad_norm": 0.48683568835258484,
      "learning_rate": 0.0002,
      "loss": 0.040107160806655884,
      "step": 2574
    },
    {
      "epoch": 0.488248673237301,
      "grad_norm": 0.11136152595281601,
      "learning_rate": 0.0002,
      "loss": 0.014762411825358868,
      "step": 2576
    },
    {
      "epoch": 0.4886277482941622,
      "grad_norm": 0.4100545346736908,
      "learning_rate": 0.0002,
      "loss": 0.03800969943404198,
      "step": 2578
    },
    {
      "epoch": 0.4890068233510235,
      "grad_norm": 0.5779994130134583,
      "learning_rate": 0.0002,
      "loss": 0.038393326103687286,
      "step": 2580
    },
    {
      "epoch": 0.48938589840788477,
      "grad_norm": 0.37115898728370667,
      "learning_rate": 0.0002,
      "loss": 0.02928253635764122,
      "step": 2582
    },
    {
      "epoch": 0.489764973464746,
      "grad_norm": 0.35431724786758423,
      "learning_rate": 0.0002,
      "loss": 0.028558382764458656,
      "step": 2584
    },
    {
      "epoch": 0.49014404852160726,
      "grad_norm": 0.1925714612007141,
      "learning_rate": 0.0002,
      "loss": 0.027282359078526497,
      "step": 2586
    },
    {
      "epoch": 0.49052312357846856,
      "grad_norm": 0.11705954372882843,
      "learning_rate": 0.0002,
      "loss": 0.013633070513606071,
      "step": 2588
    },
    {
      "epoch": 0.4909021986353298,
      "grad_norm": 0.19168049097061157,
      "learning_rate": 0.0002,
      "loss": 0.027471257373690605,
      "step": 2590
    },
    {
      "epoch": 0.49128127369219105,
      "grad_norm": 0.3797657787799835,
      "learning_rate": 0.0002,
      "loss": 0.02444637380540371,
      "step": 2592
    },
    {
      "epoch": 0.4916603487490523,
      "grad_norm": 0.3310498297214508,
      "learning_rate": 0.0002,
      "loss": 0.04344906657934189,
      "step": 2594
    },
    {
      "epoch": 0.4920394238059136,
      "grad_norm": 0.3092232942581177,
      "learning_rate": 0.0002,
      "loss": 0.0196697860956192,
      "step": 2596
    },
    {
      "epoch": 0.49241849886277483,
      "grad_norm": 0.4163341820240021,
      "learning_rate": 0.0002,
      "loss": 0.03265901282429695,
      "step": 2598
    },
    {
      "epoch": 0.4927975739196361,
      "grad_norm": 0.11072506755590439,
      "learning_rate": 0.0002,
      "loss": 0.012868964113295078,
      "step": 2600
    },
    {
      "epoch": 0.4931766489764973,
      "grad_norm": 0.4834946393966675,
      "learning_rate": 0.0002,
      "loss": 0.03940189257264137,
      "step": 2602
    },
    {
      "epoch": 0.4935557240333586,
      "grad_norm": 0.11319993436336517,
      "learning_rate": 0.0002,
      "loss": 0.032637689262628555,
      "step": 2604
    },
    {
      "epoch": 0.49393479909021987,
      "grad_norm": 0.44500693678855896,
      "learning_rate": 0.0002,
      "loss": 0.029925568029284477,
      "step": 2606
    },
    {
      "epoch": 0.4943138741470811,
      "grad_norm": 0.39070838689804077,
      "learning_rate": 0.0002,
      "loss": 0.03414827585220337,
      "step": 2608
    },
    {
      "epoch": 0.49469294920394236,
      "grad_norm": 0.09182345867156982,
      "learning_rate": 0.0002,
      "loss": 0.013335403986275196,
      "step": 2610
    },
    {
      "epoch": 0.49507202426080366,
      "grad_norm": 0.1918291300535202,
      "learning_rate": 0.0002,
      "loss": 0.01758739724755287,
      "step": 2612
    },
    {
      "epoch": 0.4954510993176649,
      "grad_norm": 0.13049209117889404,
      "learning_rate": 0.0002,
      "loss": 0.024699898436665535,
      "step": 2614
    },
    {
      "epoch": 0.49583017437452614,
      "grad_norm": 0.10733942687511444,
      "learning_rate": 0.0002,
      "loss": 0.014559868723154068,
      "step": 2616
    },
    {
      "epoch": 0.4962092494313874,
      "grad_norm": 0.3473381996154785,
      "learning_rate": 0.0002,
      "loss": 0.02471317909657955,
      "step": 2618
    },
    {
      "epoch": 0.4965883244882487,
      "grad_norm": 0.400401771068573,
      "learning_rate": 0.0002,
      "loss": 0.025865066796541214,
      "step": 2620
    },
    {
      "epoch": 0.49696739954510993,
      "grad_norm": 0.29537704586982727,
      "learning_rate": 0.0002,
      "loss": 0.03356604650616646,
      "step": 2622
    },
    {
      "epoch": 0.4973464746019712,
      "grad_norm": 0.23059333860874176,
      "learning_rate": 0.0002,
      "loss": 0.017749998718500137,
      "step": 2624
    },
    {
      "epoch": 0.4977255496588324,
      "grad_norm": 0.09631752967834473,
      "learning_rate": 0.0002,
      "loss": 0.013304158113896847,
      "step": 2626
    },
    {
      "epoch": 0.4981046247156937,
      "grad_norm": 0.1731824427843094,
      "learning_rate": 0.0002,
      "loss": 0.016401968896389008,
      "step": 2628
    },
    {
      "epoch": 0.49848369977255497,
      "grad_norm": 0.25634849071502686,
      "learning_rate": 0.0002,
      "loss": 0.028656886890530586,
      "step": 2630
    },
    {
      "epoch": 0.4988627748294162,
      "grad_norm": 0.18101827800273895,
      "learning_rate": 0.0002,
      "loss": 0.015489146113395691,
      "step": 2632
    },
    {
      "epoch": 0.49924184988627746,
      "grad_norm": 0.11734180152416229,
      "learning_rate": 0.0002,
      "loss": 0.024658039212226868,
      "step": 2634
    },
    {
      "epoch": 0.49962092494313876,
      "grad_norm": 0.3797662854194641,
      "learning_rate": 0.0002,
      "loss": 0.03987612575292587,
      "step": 2636
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.18938790261745453,
      "learning_rate": 0.0002,
      "loss": 0.03628025949001312,
      "step": 2638
    },
    {
      "epoch": 0.5003790750568613,
      "grad_norm": 0.34816789627075195,
      "learning_rate": 0.0002,
      "loss": 0.03244170546531677,
      "step": 2640
    },
    {
      "epoch": 0.5007581501137225,
      "grad_norm": 0.2174636721611023,
      "learning_rate": 0.0002,
      "loss": 0.014512361958622932,
      "step": 2642
    },
    {
      "epoch": 0.5011372251705838,
      "grad_norm": 0.35368940234184265,
      "learning_rate": 0.0002,
      "loss": 0.03735716640949249,
      "step": 2644
    },
    {
      "epoch": 0.5015163002274451,
      "grad_norm": 0.4972464442253113,
      "learning_rate": 0.0002,
      "loss": 0.06218751519918442,
      "step": 2646
    },
    {
      "epoch": 0.5018953752843063,
      "grad_norm": 0.416360467672348,
      "learning_rate": 0.0002,
      "loss": 0.03834819793701172,
      "step": 2648
    },
    {
      "epoch": 0.5022744503411676,
      "grad_norm": 0.27326011657714844,
      "learning_rate": 0.0002,
      "loss": 0.03100012056529522,
      "step": 2650
    },
    {
      "epoch": 0.5026535253980288,
      "grad_norm": 0.3449169993400574,
      "learning_rate": 0.0002,
      "loss": 0.02217106893658638,
      "step": 2652
    },
    {
      "epoch": 0.5030326004548901,
      "grad_norm": 0.10954468697309494,
      "learning_rate": 0.0002,
      "loss": 0.014838792383670807,
      "step": 2654
    },
    {
      "epoch": 0.5034116755117514,
      "grad_norm": 0.31920281052589417,
      "learning_rate": 0.0002,
      "loss": 0.041914407163858414,
      "step": 2656
    },
    {
      "epoch": 0.5037907505686126,
      "grad_norm": 0.32559922337532043,
      "learning_rate": 0.0002,
      "loss": 0.02929573878645897,
      "step": 2658
    },
    {
      "epoch": 0.5041698256254739,
      "grad_norm": 0.1552201509475708,
      "learning_rate": 0.0002,
      "loss": 0.02041621506214142,
      "step": 2660
    },
    {
      "epoch": 0.5045489006823352,
      "grad_norm": 0.5211811065673828,
      "learning_rate": 0.0002,
      "loss": 0.03607896342873573,
      "step": 2662
    },
    {
      "epoch": 0.5049279757391963,
      "grad_norm": 0.2822529375553131,
      "learning_rate": 0.0002,
      "loss": 0.018534131348133087,
      "step": 2664
    },
    {
      "epoch": 0.5053070507960576,
      "grad_norm": 0.2124861478805542,
      "learning_rate": 0.0002,
      "loss": 0.019625460729002953,
      "step": 2666
    },
    {
      "epoch": 0.5056861258529188,
      "grad_norm": 0.18837082386016846,
      "learning_rate": 0.0002,
      "loss": 0.020228300243616104,
      "step": 2668
    },
    {
      "epoch": 0.5060652009097801,
      "grad_norm": 0.22595816850662231,
      "learning_rate": 0.0002,
      "loss": 0.01623946987092495,
      "step": 2670
    },
    {
      "epoch": 0.5064442759666414,
      "grad_norm": 0.18192455172538757,
      "learning_rate": 0.0002,
      "loss": 0.01403221394866705,
      "step": 2672
    },
    {
      "epoch": 0.5068233510235026,
      "grad_norm": 0.4674341678619385,
      "learning_rate": 0.0002,
      "loss": 0.026833366602659225,
      "step": 2674
    },
    {
      "epoch": 0.5072024260803639,
      "grad_norm": 0.20551268756389618,
      "learning_rate": 0.0002,
      "loss": 0.01584843173623085,
      "step": 2676
    },
    {
      "epoch": 0.5075815011372252,
      "grad_norm": 0.3460980951786041,
      "learning_rate": 0.0002,
      "loss": 0.028599051758646965,
      "step": 2678
    },
    {
      "epoch": 0.5079605761940864,
      "grad_norm": 0.13473540544509888,
      "learning_rate": 0.0002,
      "loss": 0.028226137161254883,
      "step": 2680
    },
    {
      "epoch": 0.5083396512509477,
      "grad_norm": 0.20844313502311707,
      "learning_rate": 0.0002,
      "loss": 0.028350090608000755,
      "step": 2682
    },
    {
      "epoch": 0.5087187263078089,
      "grad_norm": 0.17823286354541779,
      "learning_rate": 0.0002,
      "loss": 0.021697230637073517,
      "step": 2684
    },
    {
      "epoch": 0.5090978013646702,
      "grad_norm": 0.13022580742835999,
      "learning_rate": 0.0002,
      "loss": 0.02549426071345806,
      "step": 2686
    },
    {
      "epoch": 0.5094768764215315,
      "grad_norm": 0.1504010409116745,
      "learning_rate": 0.0002,
      "loss": 0.03557804971933365,
      "step": 2688
    },
    {
      "epoch": 0.5098559514783927,
      "grad_norm": 0.4446730315685272,
      "learning_rate": 0.0002,
      "loss": 0.04615699499845505,
      "step": 2690
    },
    {
      "epoch": 0.510235026535254,
      "grad_norm": 0.36164918541908264,
      "learning_rate": 0.0002,
      "loss": 0.018956296145915985,
      "step": 2692
    },
    {
      "epoch": 0.5106141015921153,
      "grad_norm": 0.22170980274677277,
      "learning_rate": 0.0002,
      "loss": 0.03483164682984352,
      "step": 2694
    },
    {
      "epoch": 0.5109931766489765,
      "grad_norm": 0.256124883890152,
      "learning_rate": 0.0002,
      "loss": 0.023329023271799088,
      "step": 2696
    },
    {
      "epoch": 0.5113722517058378,
      "grad_norm": 0.4422956705093384,
      "learning_rate": 0.0002,
      "loss": 0.0459827221930027,
      "step": 2698
    },
    {
      "epoch": 0.511751326762699,
      "grad_norm": 0.10442612320184708,
      "learning_rate": 0.0002,
      "loss": 0.025473615154623985,
      "step": 2700
    },
    {
      "epoch": 0.5121304018195603,
      "grad_norm": 0.09374447911977768,
      "learning_rate": 0.0002,
      "loss": 0.02684011124074459,
      "step": 2702
    },
    {
      "epoch": 0.5125094768764216,
      "grad_norm": 0.3810797333717346,
      "learning_rate": 0.0002,
      "loss": 0.025179684162139893,
      "step": 2704
    },
    {
      "epoch": 0.5128885519332828,
      "grad_norm": 0.27601832151412964,
      "learning_rate": 0.0002,
      "loss": 0.027225859463214874,
      "step": 2706
    },
    {
      "epoch": 0.513267626990144,
      "grad_norm": 0.3744622766971588,
      "learning_rate": 0.0002,
      "loss": 0.04036348685622215,
      "step": 2708
    },
    {
      "epoch": 0.5136467020470054,
      "grad_norm": 0.27423444390296936,
      "learning_rate": 0.0002,
      "loss": 0.03496896103024483,
      "step": 2710
    },
    {
      "epoch": 0.5140257771038665,
      "grad_norm": 0.37431105971336365,
      "learning_rate": 0.0002,
      "loss": 0.02917499467730522,
      "step": 2712
    },
    {
      "epoch": 0.5144048521607278,
      "grad_norm": 0.4777616262435913,
      "learning_rate": 0.0002,
      "loss": 0.03456306830048561,
      "step": 2714
    },
    {
      "epoch": 0.514783927217589,
      "grad_norm": 0.10231836140155792,
      "learning_rate": 0.0002,
      "loss": 0.019918756559491158,
      "step": 2716
    },
    {
      "epoch": 0.5151630022744503,
      "grad_norm": 0.41008543968200684,
      "learning_rate": 0.0002,
      "loss": 0.029876763001084328,
      "step": 2718
    },
    {
      "epoch": 0.5155420773313116,
      "grad_norm": 0.23067401349544525,
      "learning_rate": 0.0002,
      "loss": 0.015179025009274483,
      "step": 2720
    },
    {
      "epoch": 0.5159211523881728,
      "grad_norm": 0.48182517290115356,
      "learning_rate": 0.0002,
      "loss": 0.056966252624988556,
      "step": 2722
    },
    {
      "epoch": 0.5163002274450341,
      "grad_norm": 0.14167267084121704,
      "learning_rate": 0.0002,
      "loss": 0.013796834275126457,
      "step": 2724
    },
    {
      "epoch": 0.5166793025018954,
      "grad_norm": 0.08879217505455017,
      "learning_rate": 0.0002,
      "loss": 0.02098218724131584,
      "step": 2726
    },
    {
      "epoch": 0.5170583775587566,
      "grad_norm": 0.46563720703125,
      "learning_rate": 0.0002,
      "loss": 0.03291868045926094,
      "step": 2728
    },
    {
      "epoch": 0.5174374526156179,
      "grad_norm": 0.375046044588089,
      "learning_rate": 0.0002,
      "loss": 0.03902749344706535,
      "step": 2730
    },
    {
      "epoch": 0.5178165276724791,
      "grad_norm": 0.11073099821805954,
      "learning_rate": 0.0002,
      "loss": 0.014218519441783428,
      "step": 2732
    },
    {
      "epoch": 0.5181956027293404,
      "grad_norm": 0.10968394577503204,
      "learning_rate": 0.0002,
      "loss": 0.02464091219007969,
      "step": 2734
    },
    {
      "epoch": 0.5185746777862017,
      "grad_norm": 0.31474095582962036,
      "learning_rate": 0.0002,
      "loss": 0.022032957524061203,
      "step": 2736
    },
    {
      "epoch": 0.5189537528430629,
      "grad_norm": 0.552055299282074,
      "learning_rate": 0.0002,
      "loss": 0.02711760252714157,
      "step": 2738
    },
    {
      "epoch": 0.5193328278999242,
      "grad_norm": 0.3780840039253235,
      "learning_rate": 0.0002,
      "loss": 0.031147412955760956,
      "step": 2740
    },
    {
      "epoch": 0.5197119029567855,
      "grad_norm": 0.244361013174057,
      "learning_rate": 0.0002,
      "loss": 0.028957484290003777,
      "step": 2742
    },
    {
      "epoch": 0.5200909780136467,
      "grad_norm": 0.1257915049791336,
      "learning_rate": 0.0002,
      "loss": 0.011347204446792603,
      "step": 2744
    },
    {
      "epoch": 0.520470053070508,
      "grad_norm": 0.12425623834133148,
      "learning_rate": 0.0002,
      "loss": 0.030317025259137154,
      "step": 2746
    },
    {
      "epoch": 0.5208491281273692,
      "grad_norm": 0.11603008210659027,
      "learning_rate": 0.0002,
      "loss": 0.019390903413295746,
      "step": 2748
    },
    {
      "epoch": 0.5212282031842305,
      "grad_norm": 0.19393377006053925,
      "learning_rate": 0.0002,
      "loss": 0.013525665737688541,
      "step": 2750
    },
    {
      "epoch": 0.5216072782410918,
      "grad_norm": 0.16514454782009125,
      "learning_rate": 0.0002,
      "loss": 0.019473757594823837,
      "step": 2752
    },
    {
      "epoch": 0.521986353297953,
      "grad_norm": 0.29176485538482666,
      "learning_rate": 0.0002,
      "loss": 0.020253553986549377,
      "step": 2754
    },
    {
      "epoch": 0.5223654283548143,
      "grad_norm": 0.4606558382511139,
      "learning_rate": 0.0002,
      "loss": 0.0351075604557991,
      "step": 2756
    },
    {
      "epoch": 0.5227445034116756,
      "grad_norm": 0.3522665500640869,
      "learning_rate": 0.0002,
      "loss": 0.039258312433958054,
      "step": 2758
    },
    {
      "epoch": 0.5231235784685367,
      "grad_norm": 0.09744761884212494,
      "learning_rate": 0.0002,
      "loss": 0.039418064057826996,
      "step": 2760
    },
    {
      "epoch": 0.523502653525398,
      "grad_norm": 0.3469116687774658,
      "learning_rate": 0.0002,
      "loss": 0.02800964005291462,
      "step": 2762
    },
    {
      "epoch": 0.5238817285822593,
      "grad_norm": 0.1326630711555481,
      "learning_rate": 0.0002,
      "loss": 0.027681225910782814,
      "step": 2764
    },
    {
      "epoch": 0.5242608036391205,
      "grad_norm": 0.3843727111816406,
      "learning_rate": 0.0002,
      "loss": 0.03553593531250954,
      "step": 2766
    },
    {
      "epoch": 0.5246398786959818,
      "grad_norm": 0.31355658173561096,
      "learning_rate": 0.0002,
      "loss": 0.0297207310795784,
      "step": 2768
    },
    {
      "epoch": 0.525018953752843,
      "grad_norm": 0.4037788510322571,
      "learning_rate": 0.0002,
      "loss": 0.0373581200838089,
      "step": 2770
    },
    {
      "epoch": 0.5253980288097043,
      "grad_norm": 0.4511270523071289,
      "learning_rate": 0.0001999611197511664,
      "loss": 0.04203470051288605,
      "step": 2772
    },
    {
      "epoch": 0.5257771038665656,
      "grad_norm": 0.432315856218338,
      "learning_rate": 0.00019988335925349925,
      "loss": 0.05148214474320412,
      "step": 2774
    },
    {
      "epoch": 0.5261561789234268,
      "grad_norm": 0.17812836170196533,
      "learning_rate": 0.00019980559875583205,
      "loss": 0.03254704922437668,
      "step": 2776
    },
    {
      "epoch": 0.5265352539802881,
      "grad_norm": 0.13714241981506348,
      "learning_rate": 0.00019972783825816486,
      "loss": 0.032962631434202194,
      "step": 2778
    },
    {
      "epoch": 0.5269143290371494,
      "grad_norm": 0.12699323892593384,
      "learning_rate": 0.00019965007776049767,
      "loss": 0.027281397953629494,
      "step": 2780
    },
    {
      "epoch": 0.5272934040940106,
      "grad_norm": 0.31825634837150574,
      "learning_rate": 0.0001995723172628305,
      "loss": 0.024133969098329544,
      "step": 2782
    },
    {
      "epoch": 0.5276724791508719,
      "grad_norm": 0.10589946061372757,
      "learning_rate": 0.0001994945567651633,
      "loss": 0.033730827271938324,
      "step": 2784
    },
    {
      "epoch": 0.5280515542077331,
      "grad_norm": 0.14599478244781494,
      "learning_rate": 0.00019941679626749612,
      "loss": 0.014181027188897133,
      "step": 2786
    },
    {
      "epoch": 0.5284306292645944,
      "grad_norm": 0.5414167642593384,
      "learning_rate": 0.00019933903576982892,
      "loss": 0.04769860580563545,
      "step": 2788
    },
    {
      "epoch": 0.5288097043214557,
      "grad_norm": 0.39109140634536743,
      "learning_rate": 0.00019926127527216173,
      "loss": 0.022391051054000854,
      "step": 2790
    },
    {
      "epoch": 0.5291887793783169,
      "grad_norm": 0.3638787567615509,
      "learning_rate": 0.00019918351477449456,
      "loss": 0.025486595928668976,
      "step": 2792
    },
    {
      "epoch": 0.5295678544351782,
      "grad_norm": 0.45001232624053955,
      "learning_rate": 0.00019910575427682737,
      "loss": 0.04080301895737648,
      "step": 2794
    },
    {
      "epoch": 0.5299469294920395,
      "grad_norm": 0.33985230326652527,
      "learning_rate": 0.0001990279937791602,
      "loss": 0.0389416441321373,
      "step": 2796
    },
    {
      "epoch": 0.5303260045489007,
      "grad_norm": 0.11851966381072998,
      "learning_rate": 0.000198950233281493,
      "loss": 0.014843798242509365,
      "step": 2798
    },
    {
      "epoch": 0.530705079605762,
      "grad_norm": 0.16964101791381836,
      "learning_rate": 0.00019887247278382584,
      "loss": 0.01443694531917572,
      "step": 2800
    },
    {
      "epoch": 0.5310841546626232,
      "grad_norm": 0.11177535355091095,
      "learning_rate": 0.00019879471228615865,
      "loss": 0.023533713072538376,
      "step": 2802
    },
    {
      "epoch": 0.5314632297194845,
      "grad_norm": 0.10738697648048401,
      "learning_rate": 0.00019871695178849146,
      "loss": 0.015398849733173847,
      "step": 2804
    },
    {
      "epoch": 0.5318423047763458,
      "grad_norm": 0.36862343549728394,
      "learning_rate": 0.00019863919129082427,
      "loss": 0.024860743433237076,
      "step": 2806
    },
    {
      "epoch": 0.5322213798332069,
      "grad_norm": 0.37062329053878784,
      "learning_rate": 0.00019856143079315707,
      "loss": 0.02728521078824997,
      "step": 2808
    },
    {
      "epoch": 0.5326004548900682,
      "grad_norm": 0.0927887037396431,
      "learning_rate": 0.0001984836702954899,
      "loss": 0.026024116203188896,
      "step": 2810
    },
    {
      "epoch": 0.5329795299469295,
      "grad_norm": 0.4272201657295227,
      "learning_rate": 0.0001984059097978227,
      "loss": 0.04901313781738281,
      "step": 2812
    },
    {
      "epoch": 0.5333586050037907,
      "grad_norm": 0.12368964403867722,
      "learning_rate": 0.00019832814930015552,
      "loss": 0.024019483476877213,
      "step": 2814
    },
    {
      "epoch": 0.533737680060652,
      "grad_norm": 0.09634125232696533,
      "learning_rate": 0.00019825038880248833,
      "loss": 0.01179907750338316,
      "step": 2816
    },
    {
      "epoch": 0.5341167551175132,
      "grad_norm": 0.3115689754486084,
      "learning_rate": 0.00019817262830482116,
      "loss": 0.04008672758936882,
      "step": 2818
    },
    {
      "epoch": 0.5344958301743745,
      "grad_norm": 0.313693106174469,
      "learning_rate": 0.000198094867807154,
      "loss": 0.035981882363557816,
      "step": 2820
    },
    {
      "epoch": 0.5348749052312358,
      "grad_norm": 0.4013250172138214,
      "learning_rate": 0.0001980171073094868,
      "loss": 0.029134362936019897,
      "step": 2822
    },
    {
      "epoch": 0.535253980288097,
      "grad_norm": 1.6934521198272705,
      "learning_rate": 0.0001979393468118196,
      "loss": 0.024958157911896706,
      "step": 2824
    },
    {
      "epoch": 0.5356330553449583,
      "grad_norm": 0.5811415314674377,
      "learning_rate": 0.00019786158631415242,
      "loss": 0.03645217791199684,
      "step": 2826
    },
    {
      "epoch": 0.5360121304018196,
      "grad_norm": 0.3092136085033417,
      "learning_rate": 0.00019778382581648522,
      "loss": 0.018008269369602203,
      "step": 2828
    },
    {
      "epoch": 0.5363912054586808,
      "grad_norm": 0.429431676864624,
      "learning_rate": 0.00019770606531881806,
      "loss": 0.03400115296244621,
      "step": 2830
    },
    {
      "epoch": 0.5367702805155421,
      "grad_norm": 0.43019020557403564,
      "learning_rate": 0.00019762830482115086,
      "loss": 0.02818351425230503,
      "step": 2832
    },
    {
      "epoch": 0.5371493555724033,
      "grad_norm": 0.3260646462440491,
      "learning_rate": 0.00019755054432348367,
      "loss": 0.0475258007645607,
      "step": 2834
    },
    {
      "epoch": 0.5375284306292646,
      "grad_norm": 0.49879613518714905,
      "learning_rate": 0.00019747278382581648,
      "loss": 0.04219025745987892,
      "step": 2836
    },
    {
      "epoch": 0.5379075056861259,
      "grad_norm": 0.3876096308231354,
      "learning_rate": 0.0001973950233281493,
      "loss": 0.023995403200387955,
      "step": 2838
    },
    {
      "epoch": 0.5382865807429871,
      "grad_norm": 0.09154628217220306,
      "learning_rate": 0.00019731726283048212,
      "loss": 0.026344360783696175,
      "step": 2840
    },
    {
      "epoch": 0.5386656557998484,
      "grad_norm": 0.43084052205085754,
      "learning_rate": 0.00019723950233281495,
      "loss": 0.038162507116794586,
      "step": 2842
    },
    {
      "epoch": 0.5390447308567097,
      "grad_norm": 0.1457781046628952,
      "learning_rate": 0.00019716174183514776,
      "loss": 0.018057167530059814,
      "step": 2844
    },
    {
      "epoch": 0.5394238059135709,
      "grad_norm": 0.0826980471611023,
      "learning_rate": 0.00019708398133748057,
      "loss": 0.012461635284125805,
      "step": 2846
    },
    {
      "epoch": 0.5398028809704322,
      "grad_norm": 0.4999379813671112,
      "learning_rate": 0.0001970062208398134,
      "loss": 0.04290381819009781,
      "step": 2848
    },
    {
      "epoch": 0.5401819560272934,
      "grad_norm": 0.3568330407142639,
      "learning_rate": 0.0001969284603421462,
      "loss": 0.022499704733490944,
      "step": 2850
    },
    {
      "epoch": 0.5405610310841547,
      "grad_norm": 0.30095213651657104,
      "learning_rate": 0.00019685069984447901,
      "loss": 0.02980036847293377,
      "step": 2852
    },
    {
      "epoch": 0.540940106141016,
      "grad_norm": 0.3561791479587555,
      "learning_rate": 0.00019677293934681182,
      "loss": 0.020320072770118713,
      "step": 2854
    },
    {
      "epoch": 0.5413191811978771,
      "grad_norm": 0.1398552507162094,
      "learning_rate": 0.00019669517884914463,
      "loss": 0.029429102316498756,
      "step": 2856
    },
    {
      "epoch": 0.5416982562547384,
      "grad_norm": 0.07729952782392502,
      "learning_rate": 0.00019661741835147746,
      "loss": 0.01447925716638565,
      "step": 2858
    },
    {
      "epoch": 0.5420773313115997,
      "grad_norm": 0.1737392544746399,
      "learning_rate": 0.00019653965785381027,
      "loss": 0.015956709161400795,
      "step": 2860
    },
    {
      "epoch": 0.5424564063684609,
      "grad_norm": 0.13527318835258484,
      "learning_rate": 0.00019646189735614308,
      "loss": 0.01221142616122961,
      "step": 2862
    },
    {
      "epoch": 0.5428354814253222,
      "grad_norm": 0.13430286943912506,
      "learning_rate": 0.0001963841368584759,
      "loss": 0.016219358891248703,
      "step": 2864
    },
    {
      "epoch": 0.5432145564821834,
      "grad_norm": 0.44608739018440247,
      "learning_rate": 0.00019630637636080872,
      "loss": 0.0387016236782074,
      "step": 2866
    },
    {
      "epoch": 0.5435936315390447,
      "grad_norm": 0.29726365208625793,
      "learning_rate": 0.00019622861586314155,
      "loss": 0.03028523549437523,
      "step": 2868
    },
    {
      "epoch": 0.543972706595906,
      "grad_norm": 0.3266139328479767,
      "learning_rate": 0.00019615085536547436,
      "loss": 0.0318429134786129,
      "step": 2870
    },
    {
      "epoch": 0.5443517816527672,
      "grad_norm": 0.15744459629058838,
      "learning_rate": 0.00019607309486780716,
      "loss": 0.03084368258714676,
      "step": 2872
    },
    {
      "epoch": 0.5447308567096285,
      "grad_norm": 0.3423956632614136,
      "learning_rate": 0.00019599533437013997,
      "loss": 0.030789172276854515,
      "step": 2874
    },
    {
      "epoch": 0.5451099317664898,
      "grad_norm": 0.3364187180995941,
      "learning_rate": 0.0001959175738724728,
      "loss": 0.025088448077440262,
      "step": 2876
    },
    {
      "epoch": 0.545489006823351,
      "grad_norm": 0.1581997126340866,
      "learning_rate": 0.0001958398133748056,
      "loss": 0.01591775193810463,
      "step": 2878
    },
    {
      "epoch": 0.5458680818802123,
      "grad_norm": 0.30122074484825134,
      "learning_rate": 0.00019576205287713842,
      "loss": 0.02581249363720417,
      "step": 2880
    },
    {
      "epoch": 0.5462471569370735,
      "grad_norm": 0.13854409754276276,
      "learning_rate": 0.00019568429237947123,
      "loss": 0.012843429110944271,
      "step": 2882
    },
    {
      "epoch": 0.5466262319939348,
      "grad_norm": 0.09312371164560318,
      "learning_rate": 0.00019560653188180403,
      "loss": 0.01785885915160179,
      "step": 2884
    },
    {
      "epoch": 0.5470053070507961,
      "grad_norm": 0.07186237722635269,
      "learning_rate": 0.00019552877138413687,
      "loss": 0.012889943085610867,
      "step": 2886
    },
    {
      "epoch": 0.5473843821076573,
      "grad_norm": 0.10102088749408722,
      "learning_rate": 0.0001954510108864697,
      "loss": 0.013085881248116493,
      "step": 2888
    },
    {
      "epoch": 0.5477634571645186,
      "grad_norm": 0.10978246480226517,
      "learning_rate": 0.0001953732503888025,
      "loss": 0.020684150978922844,
      "step": 2890
    },
    {
      "epoch": 0.5481425322213799,
      "grad_norm": 0.36384889483451843,
      "learning_rate": 0.00019529548989113531,
      "loss": 0.04394169896841049,
      "step": 2892
    },
    {
      "epoch": 0.5485216072782411,
      "grad_norm": 0.41054531931877136,
      "learning_rate": 0.00019521772939346812,
      "loss": 0.034910548478364944,
      "step": 2894
    },
    {
      "epoch": 0.5489006823351024,
      "grad_norm": 0.46372461318969727,
      "learning_rate": 0.00019513996889580095,
      "loss": 0.05939662456512451,
      "step": 2896
    },
    {
      "epoch": 0.5492797573919637,
      "grad_norm": 0.10024144500494003,
      "learning_rate": 0.00019506220839813376,
      "loss": 0.01921691559255123,
      "step": 2898
    },
    {
      "epoch": 0.5496588324488249,
      "grad_norm": 0.132353276014328,
      "learning_rate": 0.00019498444790046657,
      "loss": 0.020812345668673515,
      "step": 2900
    },
    {
      "epoch": 0.5500379075056862,
      "grad_norm": 0.16173256933689117,
      "learning_rate": 0.00019490668740279938,
      "loss": 0.028947891667485237,
      "step": 2902
    },
    {
      "epoch": 0.5504169825625473,
      "grad_norm": 0.3491940200328827,
      "learning_rate": 0.00019482892690513218,
      "loss": 0.043473221361637115,
      "step": 2904
    },
    {
      "epoch": 0.5507960576194086,
      "grad_norm": 0.37259459495544434,
      "learning_rate": 0.00019475116640746502,
      "loss": 0.025933310389518738,
      "step": 2906
    },
    {
      "epoch": 0.5511751326762699,
      "grad_norm": 0.23021326959133148,
      "learning_rate": 0.00019467340590979782,
      "loss": 0.021500544622540474,
      "step": 2908
    },
    {
      "epoch": 0.5515542077331311,
      "grad_norm": 0.3934917449951172,
      "learning_rate": 0.00019459564541213066,
      "loss": 0.054400134831666946,
      "step": 2910
    },
    {
      "epoch": 0.5519332827899924,
      "grad_norm": 0.24869954586029053,
      "learning_rate": 0.00019451788491446346,
      "loss": 0.034735701978206635,
      "step": 2912
    },
    {
      "epoch": 0.5523123578468537,
      "grad_norm": 0.11804301291704178,
      "learning_rate": 0.0001944401244167963,
      "loss": 0.022755170240998268,
      "step": 2914
    },
    {
      "epoch": 0.5526914329037149,
      "grad_norm": 0.3166675865650177,
      "learning_rate": 0.0001943623639191291,
      "loss": 0.017435993999242783,
      "step": 2916
    },
    {
      "epoch": 0.5530705079605762,
      "grad_norm": 0.18581609427928925,
      "learning_rate": 0.0001942846034214619,
      "loss": 0.030139993876218796,
      "step": 2918
    },
    {
      "epoch": 0.5534495830174374,
      "grad_norm": 0.3183881938457489,
      "learning_rate": 0.00019420684292379472,
      "loss": 0.023322341963648796,
      "step": 2920
    },
    {
      "epoch": 0.5538286580742987,
      "grad_norm": 0.1775256246328354,
      "learning_rate": 0.00019412908242612753,
      "loss": 0.020853254944086075,
      "step": 2922
    },
    {
      "epoch": 0.55420773313116,
      "grad_norm": 0.21420994400978088,
      "learning_rate": 0.00019405132192846036,
      "loss": 0.017971239984035492,
      "step": 2924
    },
    {
      "epoch": 0.5545868081880212,
      "grad_norm": 0.36475980281829834,
      "learning_rate": 0.00019397356143079317,
      "loss": 0.03106292523443699,
      "step": 2926
    },
    {
      "epoch": 0.5549658832448825,
      "grad_norm": 0.3011978566646576,
      "learning_rate": 0.00019389580093312597,
      "loss": 0.04678582400083542,
      "step": 2928
    },
    {
      "epoch": 0.5553449583017438,
      "grad_norm": 0.27350324392318726,
      "learning_rate": 0.00019381804043545878,
      "loss": 0.039826542139053345,
      "step": 2930
    },
    {
      "epoch": 0.555724033358605,
      "grad_norm": 0.11436368525028229,
      "learning_rate": 0.00019374027993779161,
      "loss": 0.024605626240372658,
      "step": 2932
    },
    {
      "epoch": 0.5561031084154663,
      "grad_norm": 0.35529109835624695,
      "learning_rate": 0.00019366251944012442,
      "loss": 0.027540510520339012,
      "step": 2934
    },
    {
      "epoch": 0.5564821834723275,
      "grad_norm": 0.210056871175766,
      "learning_rate": 0.00019358475894245726,
      "loss": 0.02086304873228073,
      "step": 2936
    },
    {
      "epoch": 0.5568612585291888,
      "grad_norm": 0.40498390793800354,
      "learning_rate": 0.00019350699844479006,
      "loss": 0.055738430470228195,
      "step": 2938
    },
    {
      "epoch": 0.5572403335860501,
      "grad_norm": 0.2909494638442993,
      "learning_rate": 0.00019342923794712287,
      "loss": 0.02463047206401825,
      "step": 2940
    },
    {
      "epoch": 0.5576194086429113,
      "grad_norm": 0.0954369455575943,
      "learning_rate": 0.00019335147744945568,
      "loss": 0.026400623843073845,
      "step": 2942
    },
    {
      "epoch": 0.5579984836997726,
      "grad_norm": 0.4529297947883606,
      "learning_rate": 0.0001932737169517885,
      "loss": 0.027888603508472443,
      "step": 2944
    },
    {
      "epoch": 0.5583775587566339,
      "grad_norm": 0.18297646939754486,
      "learning_rate": 0.00019319595645412132,
      "loss": 0.020410310477018356,
      "step": 2946
    },
    {
      "epoch": 0.558756633813495,
      "grad_norm": 0.3178445100784302,
      "learning_rate": 0.00019311819595645412,
      "loss": 0.03054523654282093,
      "step": 2948
    },
    {
      "epoch": 0.5591357088703564,
      "grad_norm": 0.2524225115776062,
      "learning_rate": 0.00019304043545878693,
      "loss": 0.035650115460157394,
      "step": 2950
    },
    {
      "epoch": 0.5595147839272175,
      "grad_norm": 0.173721045255661,
      "learning_rate": 0.00019296267496111976,
      "loss": 0.01472693681716919,
      "step": 2952
    },
    {
      "epoch": 0.5598938589840788,
      "grad_norm": 0.15295498073101044,
      "learning_rate": 0.00019288491446345257,
      "loss": 0.013474557548761368,
      "step": 2954
    },
    {
      "epoch": 0.5602729340409401,
      "grad_norm": 0.3594950735569,
      "learning_rate": 0.0001928071539657854,
      "loss": 0.030594035983085632,
      "step": 2956
    },
    {
      "epoch": 0.5606520090978013,
      "grad_norm": 0.46896204352378845,
      "learning_rate": 0.0001927293934681182,
      "loss": 0.04241933301091194,
      "step": 2958
    },
    {
      "epoch": 0.5610310841546626,
      "grad_norm": 0.09555663168430328,
      "learning_rate": 0.00019265163297045102,
      "loss": 0.013709360733628273,
      "step": 2960
    },
    {
      "epoch": 0.5614101592115239,
      "grad_norm": 0.18933486938476562,
      "learning_rate": 0.00019257387247278385,
      "loss": 0.03924727439880371,
      "step": 2962
    },
    {
      "epoch": 0.5617892342683851,
      "grad_norm": 0.31719541549682617,
      "learning_rate": 0.00019249611197511666,
      "loss": 0.024767499417066574,
      "step": 2964
    },
    {
      "epoch": 0.5621683093252464,
      "grad_norm": 0.31606683135032654,
      "learning_rate": 0.00019241835147744947,
      "loss": 0.034522876143455505,
      "step": 2966
    },
    {
      "epoch": 0.5625473843821076,
      "grad_norm": 0.09722900390625,
      "learning_rate": 0.00019234059097978227,
      "loss": 0.020095737650990486,
      "step": 2968
    },
    {
      "epoch": 0.5629264594389689,
      "grad_norm": 0.2208389788866043,
      "learning_rate": 0.00019226283048211508,
      "loss": 0.016096502542495728,
      "step": 2970
    },
    {
      "epoch": 0.5633055344958302,
      "grad_norm": 0.4846627116203308,
      "learning_rate": 0.00019218506998444791,
      "loss": 0.04836711287498474,
      "step": 2972
    },
    {
      "epoch": 0.5636846095526914,
      "grad_norm": 0.12509289383888245,
      "learning_rate": 0.00019210730948678072,
      "loss": 0.0120081203058362,
      "step": 2974
    },
    {
      "epoch": 0.5640636846095527,
      "grad_norm": 0.2849941551685333,
      "learning_rate": 0.00019202954898911353,
      "loss": 0.042794886976480484,
      "step": 2976
    },
    {
      "epoch": 0.564442759666414,
      "grad_norm": 0.14325959980487823,
      "learning_rate": 0.00019195178849144636,
      "loss": 0.027663826942443848,
      "step": 2978
    },
    {
      "epoch": 0.5648218347232752,
      "grad_norm": 0.08576830476522446,
      "learning_rate": 0.00019187402799377917,
      "loss": 0.011677498929202557,
      "step": 2980
    },
    {
      "epoch": 0.5652009097801365,
      "grad_norm": 0.11389067023992538,
      "learning_rate": 0.000191796267496112,
      "loss": 0.023893656209111214,
      "step": 2982
    },
    {
      "epoch": 0.5655799848369977,
      "grad_norm": 0.08725640177726746,
      "learning_rate": 0.0001917185069984448,
      "loss": 0.01694301888346672,
      "step": 2984
    },
    {
      "epoch": 0.565959059893859,
      "grad_norm": 0.449332058429718,
      "learning_rate": 0.00019164074650077762,
      "loss": 0.05006073787808418,
      "step": 2986
    },
    {
      "epoch": 0.5663381349507203,
      "grad_norm": 0.3881208896636963,
      "learning_rate": 0.00019156298600311042,
      "loss": 0.03843165561556816,
      "step": 2988
    },
    {
      "epoch": 0.5667172100075815,
      "grad_norm": 0.10273662954568863,
      "learning_rate": 0.00019148522550544323,
      "loss": 0.01977883279323578,
      "step": 2990
    },
    {
      "epoch": 0.5670962850644428,
      "grad_norm": 0.3323620557785034,
      "learning_rate": 0.00019140746500777606,
      "loss": 0.023318080231547356,
      "step": 2992
    },
    {
      "epoch": 0.5674753601213041,
      "grad_norm": 0.08815070986747742,
      "learning_rate": 0.00019132970451010887,
      "loss": 0.013169223442673683,
      "step": 2994
    },
    {
      "epoch": 0.5678544351781653,
      "grad_norm": 0.3278101682662964,
      "learning_rate": 0.00019125194401244168,
      "loss": 0.021679667755961418,
      "step": 2996
    },
    {
      "epoch": 0.5682335102350266,
      "grad_norm": 0.3978005051612854,
      "learning_rate": 0.00019117418351477449,
      "loss": 0.03125254437327385,
      "step": 2998
    },
    {
      "epoch": 0.5686125852918877,
      "grad_norm": 0.31462362408638,
      "learning_rate": 0.00019109642301710732,
      "loss": 0.03708317130804062,
      "step": 3000
    },
    {
      "epoch": 0.568991660348749,
      "grad_norm": 0.45969051122665405,
      "learning_rate": 0.00019101866251944013,
      "loss": 0.03735610842704773,
      "step": 3002
    },
    {
      "epoch": 0.5693707354056103,
      "grad_norm": 0.10202889144420624,
      "learning_rate": 0.00019094090202177296,
      "loss": 0.013903950341045856,
      "step": 3004
    },
    {
      "epoch": 0.5697498104624715,
      "grad_norm": 0.3321998417377472,
      "learning_rate": 0.00019086314152410577,
      "loss": 0.019682157784700394,
      "step": 3006
    },
    {
      "epoch": 0.5701288855193328,
      "grad_norm": 0.07730302959680557,
      "learning_rate": 0.00019078538102643857,
      "loss": 0.012272226624190807,
      "step": 3008
    },
    {
      "epoch": 0.5705079605761941,
      "grad_norm": 0.09099100530147552,
      "learning_rate": 0.0001907076205287714,
      "loss": 0.013379717245697975,
      "step": 3010
    },
    {
      "epoch": 0.5708870356330553,
      "grad_norm": 0.2914208173751831,
      "learning_rate": 0.00019062986003110422,
      "loss": 0.019616827368736267,
      "step": 3012
    },
    {
      "epoch": 0.5712661106899166,
      "grad_norm": 0.13826191425323486,
      "learning_rate": 0.00019055209953343702,
      "loss": 0.01659713312983513,
      "step": 3014
    },
    {
      "epoch": 0.5716451857467779,
      "grad_norm": 0.24966391921043396,
      "learning_rate": 0.00019047433903576983,
      "loss": 0.01902940310537815,
      "step": 3016
    },
    {
      "epoch": 0.5720242608036391,
      "grad_norm": 0.11575381457805634,
      "learning_rate": 0.00019039657853810264,
      "loss": 0.035127367824316025,
      "step": 3018
    },
    {
      "epoch": 0.5724033358605004,
      "grad_norm": 0.25495052337646484,
      "learning_rate": 0.00019031881804043547,
      "loss": 0.0417012944817543,
      "step": 3020
    },
    {
      "epoch": 0.5727824109173616,
      "grad_norm": 0.37073713541030884,
      "learning_rate": 0.00019024105754276828,
      "loss": 0.037815019488334656,
      "step": 3022
    },
    {
      "epoch": 0.5731614859742229,
      "grad_norm": 0.14134618639945984,
      "learning_rate": 0.00019016329704510108,
      "loss": 0.025771846994757652,
      "step": 3024
    },
    {
      "epoch": 0.5735405610310842,
      "grad_norm": 0.43300384283065796,
      "learning_rate": 0.00019008553654743392,
      "loss": 0.03193839266896248,
      "step": 3026
    },
    {
      "epoch": 0.5739196360879454,
      "grad_norm": 0.10708079487085342,
      "learning_rate": 0.00019000777604976672,
      "loss": 0.035056427121162415,
      "step": 3028
    },
    {
      "epoch": 0.5742987111448067,
      "grad_norm": 0.3524445593357086,
      "learning_rate": 0.00018993001555209956,
      "loss": 0.03395860642194748,
      "step": 3030
    },
    {
      "epoch": 0.574677786201668,
      "grad_norm": 0.288965106010437,
      "learning_rate": 0.00018985225505443237,
      "loss": 0.039001915603876114,
      "step": 3032
    },
    {
      "epoch": 0.5750568612585292,
      "grad_norm": 0.25555264949798584,
      "learning_rate": 0.00018977449455676517,
      "loss": 0.03237171098589897,
      "step": 3034
    },
    {
      "epoch": 0.5754359363153905,
      "grad_norm": 0.09389345347881317,
      "learning_rate": 0.00018969673405909798,
      "loss": 0.013503765687346458,
      "step": 3036
    },
    {
      "epoch": 0.5758150113722517,
      "grad_norm": 0.13648808002471924,
      "learning_rate": 0.0001896189735614308,
      "loss": 0.03036203235387802,
      "step": 3038
    },
    {
      "epoch": 0.576194086429113,
      "grad_norm": 0.1801147311925888,
      "learning_rate": 0.00018954121306376362,
      "loss": 0.01219261810183525,
      "step": 3040
    },
    {
      "epoch": 0.5765731614859743,
      "grad_norm": 0.3579302728176117,
      "learning_rate": 0.00018946345256609643,
      "loss": 0.031760022044181824,
      "step": 3042
    },
    {
      "epoch": 0.5769522365428355,
      "grad_norm": 0.35200974345207214,
      "learning_rate": 0.00018938569206842923,
      "loss": 0.021255306899547577,
      "step": 3044
    },
    {
      "epoch": 0.5773313115996968,
      "grad_norm": 0.12586814165115356,
      "learning_rate": 0.00018930793157076204,
      "loss": 0.012994576245546341,
      "step": 3046
    },
    {
      "epoch": 0.577710386656558,
      "grad_norm": 0.23936529457569122,
      "learning_rate": 0.00018923017107309487,
      "loss": 0.031451720744371414,
      "step": 3048
    },
    {
      "epoch": 0.5780894617134192,
      "grad_norm": 0.09575216472148895,
      "learning_rate": 0.0001891524105754277,
      "loss": 0.02248726226389408,
      "step": 3050
    },
    {
      "epoch": 0.5784685367702805,
      "grad_norm": 0.27156904339790344,
      "learning_rate": 0.00018907465007776052,
      "loss": 0.02591857500374317,
      "step": 3052
    },
    {
      "epoch": 0.5788476118271417,
      "grad_norm": 0.3718796968460083,
      "learning_rate": 0.00018899688958009332,
      "loss": 0.03378124162554741,
      "step": 3054
    },
    {
      "epoch": 0.579226686884003,
      "grad_norm": 0.09923847019672394,
      "learning_rate": 0.00018891912908242613,
      "loss": 0.013234860263764858,
      "step": 3056
    },
    {
      "epoch": 0.5796057619408643,
      "grad_norm": 0.5019669532775879,
      "learning_rate": 0.00018884136858475896,
      "loss": 0.0445699580013752,
      "step": 3058
    },
    {
      "epoch": 0.5799848369977255,
      "grad_norm": 0.3549627959728241,
      "learning_rate": 0.00018876360808709177,
      "loss": 0.02657456323504448,
      "step": 3060
    },
    {
      "epoch": 0.5803639120545868,
      "grad_norm": 0.0830220878124237,
      "learning_rate": 0.00018868584758942458,
      "loss": 0.013005539774894714,
      "step": 3062
    },
    {
      "epoch": 0.5807429871114481,
      "grad_norm": 0.3376346528530121,
      "learning_rate": 0.00018860808709175738,
      "loss": 0.028359180316329002,
      "step": 3064
    },
    {
      "epoch": 0.5811220621683093,
      "grad_norm": 0.11597637832164764,
      "learning_rate": 0.0001885303265940902,
      "loss": 0.014026428572833538,
      "step": 3066
    },
    {
      "epoch": 0.5815011372251706,
      "grad_norm": 0.4103584289550781,
      "learning_rate": 0.00018845256609642302,
      "loss": 0.04285062104463577,
      "step": 3068
    },
    {
      "epoch": 0.5818802122820318,
      "grad_norm": 0.3364398181438446,
      "learning_rate": 0.00018837480559875583,
      "loss": 0.015217581763863564,
      "step": 3070
    },
    {
      "epoch": 0.5822592873388931,
      "grad_norm": 0.12742096185684204,
      "learning_rate": 0.00018829704510108867,
      "loss": 0.01577661745250225,
      "step": 3072
    },
    {
      "epoch": 0.5826383623957544,
      "grad_norm": 0.27025651931762695,
      "learning_rate": 0.00018821928460342147,
      "loss": 0.0330677255988121,
      "step": 3074
    },
    {
      "epoch": 0.5830174374526156,
      "grad_norm": 0.21803684532642365,
      "learning_rate": 0.0001881415241057543,
      "loss": 0.032739877700805664,
      "step": 3076
    },
    {
      "epoch": 0.5833965125094769,
      "grad_norm": 0.09379786252975464,
      "learning_rate": 0.0001880637636080871,
      "loss": 0.02152804099023342,
      "step": 3078
    },
    {
      "epoch": 0.5837755875663382,
      "grad_norm": 0.4009365439414978,
      "learning_rate": 0.00018798600311041992,
      "loss": 0.02588506229221821,
      "step": 3080
    },
    {
      "epoch": 0.5841546626231994,
      "grad_norm": 0.19567550718784332,
      "learning_rate": 0.00018790824261275273,
      "loss": 0.015504585579037666,
      "step": 3082
    },
    {
      "epoch": 0.5845337376800607,
      "grad_norm": 0.1627129316329956,
      "learning_rate": 0.00018783048211508553,
      "loss": 0.01207062415778637,
      "step": 3084
    },
    {
      "epoch": 0.5849128127369219,
      "grad_norm": 0.19477516412734985,
      "learning_rate": 0.00018775272161741837,
      "loss": 0.02056151255965233,
      "step": 3086
    },
    {
      "epoch": 0.5852918877937832,
      "grad_norm": 0.1147218719124794,
      "learning_rate": 0.00018767496111975117,
      "loss": 0.014521225355565548,
      "step": 3088
    },
    {
      "epoch": 0.5856709628506445,
      "grad_norm": 0.29134219884872437,
      "learning_rate": 0.00018759720062208398,
      "loss": 0.035216059535741806,
      "step": 3090
    },
    {
      "epoch": 0.5860500379075056,
      "grad_norm": 0.13189879059791565,
      "learning_rate": 0.0001875194401244168,
      "loss": 0.014396123588085175,
      "step": 3092
    },
    {
      "epoch": 0.586429112964367,
      "grad_norm": 0.45639294385910034,
      "learning_rate": 0.00018744167962674962,
      "loss": 0.04394407197833061,
      "step": 3094
    },
    {
      "epoch": 0.5868081880212282,
      "grad_norm": 0.4147135317325592,
      "learning_rate": 0.00018736391912908246,
      "loss": 0.033765144646167755,
      "step": 3096
    },
    {
      "epoch": 0.5871872630780894,
      "grad_norm": 0.2378246933221817,
      "learning_rate": 0.00018728615863141526,
      "loss": 0.020217545330524445,
      "step": 3098
    },
    {
      "epoch": 0.5875663381349507,
      "grad_norm": 0.3092693090438843,
      "learning_rate": 0.00018720839813374807,
      "loss": 0.024601690471172333,
      "step": 3100
    },
    {
      "epoch": 0.5879454131918119,
      "grad_norm": 0.21365715563297272,
      "learning_rate": 0.00018713063763608088,
      "loss": 0.030572770163416862,
      "step": 3102
    },
    {
      "epoch": 0.5883244882486732,
      "grad_norm": 0.08912160247564316,
      "learning_rate": 0.00018705287713841368,
      "loss": 0.01841745339334011,
      "step": 3104
    },
    {
      "epoch": 0.5887035633055345,
      "grad_norm": 0.27206242084503174,
      "learning_rate": 0.00018697511664074652,
      "loss": 0.02893420122563839,
      "step": 3106
    },
    {
      "epoch": 0.5890826383623957,
      "grad_norm": 0.16308802366256714,
      "learning_rate": 0.00018689735614307933,
      "loss": 0.014687544666230679,
      "step": 3108
    },
    {
      "epoch": 0.589461713419257,
      "grad_norm": 0.22963321208953857,
      "learning_rate": 0.00018681959564541213,
      "loss": 0.02201784960925579,
      "step": 3110
    },
    {
      "epoch": 0.5898407884761183,
      "grad_norm": 0.3977583944797516,
      "learning_rate": 0.00018674183514774494,
      "loss": 0.030358120799064636,
      "step": 3112
    },
    {
      "epoch": 0.5902198635329795,
      "grad_norm": 0.34960347414016724,
      "learning_rate": 0.00018666407465007777,
      "loss": 0.05083079636096954,
      "step": 3114
    },
    {
      "epoch": 0.5905989385898408,
      "grad_norm": 0.3732808232307434,
      "learning_rate": 0.00018658631415241058,
      "loss": 0.026019098237156868,
      "step": 3116
    },
    {
      "epoch": 0.590978013646702,
      "grad_norm": 0.37979891896247864,
      "learning_rate": 0.00018650855365474341,
      "loss": 0.029404407367110252,
      "step": 3118
    },
    {
      "epoch": 0.5913570887035633,
      "grad_norm": 0.3497915267944336,
      "learning_rate": 0.00018643079315707622,
      "loss": 0.024342525750398636,
      "step": 3120
    },
    {
      "epoch": 0.5917361637604246,
      "grad_norm": 0.35514673590660095,
      "learning_rate": 0.00018635303265940903,
      "loss": 0.028196992352604866,
      "step": 3122
    },
    {
      "epoch": 0.5921152388172858,
      "grad_norm": 0.09599101543426514,
      "learning_rate": 0.00018627527216174186,
      "loss": 0.012727697379887104,
      "step": 3124
    },
    {
      "epoch": 0.5924943138741471,
      "grad_norm": 0.36839383840560913,
      "learning_rate": 0.00018619751166407467,
      "loss": 0.026311542838811874,
      "step": 3126
    },
    {
      "epoch": 0.5928733889310084,
      "grad_norm": 0.37662625312805176,
      "learning_rate": 0.00018611975116640748,
      "loss": 0.03342700004577637,
      "step": 3128
    },
    {
      "epoch": 0.5932524639878696,
      "grad_norm": 0.3146347105503082,
      "learning_rate": 0.00018604199066874028,
      "loss": 0.03405747562646866,
      "step": 3130
    },
    {
      "epoch": 0.5936315390447309,
      "grad_norm": 0.3311243951320648,
      "learning_rate": 0.0001859642301710731,
      "loss": 0.052918534725904465,
      "step": 3132
    },
    {
      "epoch": 0.5940106141015921,
      "grad_norm": 0.10206177830696106,
      "learning_rate": 0.00018588646967340592,
      "loss": 0.028567396104335785,
      "step": 3134
    },
    {
      "epoch": 0.5943896891584534,
      "grad_norm": 0.07761494815349579,
      "learning_rate": 0.00018580870917573873,
      "loss": 0.01277000829577446,
      "step": 3136
    },
    {
      "epoch": 0.5947687642153147,
      "grad_norm": 0.30610546469688416,
      "learning_rate": 0.00018573094867807154,
      "loss": 0.029441725462675095,
      "step": 3138
    },
    {
      "epoch": 0.5951478392721758,
      "grad_norm": 0.11078057438135147,
      "learning_rate": 0.00018565318818040437,
      "loss": 0.031417135149240494,
      "step": 3140
    },
    {
      "epoch": 0.5955269143290371,
      "grad_norm": 0.17563241720199585,
      "learning_rate": 0.00018557542768273718,
      "loss": 0.015541979111731052,
      "step": 3142
    },
    {
      "epoch": 0.5959059893858984,
      "grad_norm": 0.4149199426174164,
      "learning_rate": 0.00018549766718507,
      "loss": 0.030194785445928574,
      "step": 3144
    },
    {
      "epoch": 0.5962850644427596,
      "grad_norm": 0.17346762120723724,
      "learning_rate": 0.00018541990668740282,
      "loss": 0.015065610408782959,
      "step": 3146
    },
    {
      "epoch": 0.5966641394996209,
      "grad_norm": 0.38052526116371155,
      "learning_rate": 0.00018534214618973563,
      "loss": 0.03680136054754257,
      "step": 3148
    },
    {
      "epoch": 0.5970432145564822,
      "grad_norm": 0.3846559226512909,
      "learning_rate": 0.00018526438569206843,
      "loss": 0.03749571740627289,
      "step": 3150
    },
    {
      "epoch": 0.5974222896133434,
      "grad_norm": 0.12652672827243805,
      "learning_rate": 0.00018518662519440124,
      "loss": 0.01754942163825035,
      "step": 3152
    },
    {
      "epoch": 0.5978013646702047,
      "grad_norm": 0.3539716601371765,
      "learning_rate": 0.00018510886469673407,
      "loss": 0.022329706698656082,
      "step": 3154
    },
    {
      "epoch": 0.5981804397270659,
      "grad_norm": 0.11853446066379547,
      "learning_rate": 0.00018503110419906688,
      "loss": 0.028700780123472214,
      "step": 3156
    },
    {
      "epoch": 0.5985595147839272,
      "grad_norm": 0.41758278012275696,
      "learning_rate": 0.0001849533437013997,
      "loss": 0.033206041902303696,
      "step": 3158
    },
    {
      "epoch": 0.5989385898407885,
      "grad_norm": 0.1051613986492157,
      "learning_rate": 0.0001848755832037325,
      "loss": 0.023325731977820396,
      "step": 3160
    },
    {
      "epoch": 0.5993176648976497,
      "grad_norm": 0.4089949131011963,
      "learning_rate": 0.00018479782270606533,
      "loss": 0.03672812134027481,
      "step": 3162
    },
    {
      "epoch": 0.599696739954511,
      "grad_norm": 0.34592390060424805,
      "learning_rate": 0.00018472006220839813,
      "loss": 0.03478826954960823,
      "step": 3164
    },
    {
      "epoch": 0.6000758150113723,
      "grad_norm": 0.19838707149028778,
      "learning_rate": 0.00018464230171073097,
      "loss": 0.017147162929177284,
      "step": 3166
    },
    {
      "epoch": 0.6004548900682335,
      "grad_norm": 0.282516211271286,
      "learning_rate": 0.00018456454121306378,
      "loss": 0.02622508630156517,
      "step": 3168
    },
    {
      "epoch": 0.6008339651250948,
      "grad_norm": 0.2580282390117645,
      "learning_rate": 0.00018448678071539658,
      "loss": 0.019449392333626747,
      "step": 3170
    },
    {
      "epoch": 0.601213040181956,
      "grad_norm": 0.19560405611991882,
      "learning_rate": 0.00018440902021772942,
      "loss": 0.03301525115966797,
      "step": 3172
    },
    {
      "epoch": 0.6015921152388173,
      "grad_norm": 0.3395759165287018,
      "learning_rate": 0.00018433125972006222,
      "loss": 0.03669044375419617,
      "step": 3174
    },
    {
      "epoch": 0.6019711902956786,
      "grad_norm": 0.1176770031452179,
      "learning_rate": 0.00018425349922239503,
      "loss": 0.015516012907028198,
      "step": 3176
    },
    {
      "epoch": 0.6023502653525398,
      "grad_norm": 0.10190972685813904,
      "learning_rate": 0.00018417573872472784,
      "loss": 0.016489677131175995,
      "step": 3178
    },
    {
      "epoch": 0.6027293404094011,
      "grad_norm": 0.4667470455169678,
      "learning_rate": 0.00018409797822706064,
      "loss": 0.04544667899608612,
      "step": 3180
    },
    {
      "epoch": 0.6031084154662624,
      "grad_norm": 0.24938105046749115,
      "learning_rate": 0.00018402021772939348,
      "loss": 0.030063917860388756,
      "step": 3182
    },
    {
      "epoch": 0.6034874905231236,
      "grad_norm": 0.2971332371234894,
      "learning_rate": 0.00018394245723172628,
      "loss": 0.020577870309352875,
      "step": 3184
    },
    {
      "epoch": 0.6038665655799849,
      "grad_norm": 0.08335688710212708,
      "learning_rate": 0.00018386469673405912,
      "loss": 0.013894853182137012,
      "step": 3186
    },
    {
      "epoch": 0.604245640636846,
      "grad_norm": 0.11553027480840683,
      "learning_rate": 0.00018378693623639193,
      "loss": 0.0182367954403162,
      "step": 3188
    },
    {
      "epoch": 0.6046247156937073,
      "grad_norm": 0.35293152928352356,
      "learning_rate": 0.00018370917573872473,
      "loss": 0.025110723450779915,
      "step": 3190
    },
    {
      "epoch": 0.6050037907505686,
      "grad_norm": 0.09876160323619843,
      "learning_rate": 0.00018363141524105757,
      "loss": 0.032887786626815796,
      "step": 3192
    },
    {
      "epoch": 0.6053828658074298,
      "grad_norm": 0.4536014199256897,
      "learning_rate": 0.00018355365474339037,
      "loss": 0.04348665848374367,
      "step": 3194
    },
    {
      "epoch": 0.6057619408642911,
      "grad_norm": 0.3623504936695099,
      "learning_rate": 0.00018347589424572318,
      "loss": 0.022850794717669487,
      "step": 3196
    },
    {
      "epoch": 0.6061410159211524,
      "grad_norm": 0.4449654221534729,
      "learning_rate": 0.000183398133748056,
      "loss": 0.04120400920510292,
      "step": 3198
    },
    {
      "epoch": 0.6065200909780136,
      "grad_norm": 0.1364438533782959,
      "learning_rate": 0.00018332037325038882,
      "loss": 0.01615896448493004,
      "step": 3200
    },
    {
      "epoch": 0.6068991660348749,
      "grad_norm": 0.339385986328125,
      "learning_rate": 0.00018324261275272163,
      "loss": 0.037376053631305695,
      "step": 3202
    },
    {
      "epoch": 0.6072782410917361,
      "grad_norm": 0.4637225270271301,
      "learning_rate": 0.00018316485225505443,
      "loss": 0.04689769074320793,
      "step": 3204
    },
    {
      "epoch": 0.6076573161485974,
      "grad_norm": 0.10357335954904556,
      "learning_rate": 0.00018308709175738724,
      "loss": 0.024310026317834854,
      "step": 3206
    },
    {
      "epoch": 0.6080363912054587,
      "grad_norm": 0.4082357585430145,
      "learning_rate": 0.00018300933125972008,
      "loss": 0.03663690388202667,
      "step": 3208
    },
    {
      "epoch": 0.6084154662623199,
      "grad_norm": 0.14762769639492035,
      "learning_rate": 0.00018293157076205288,
      "loss": 0.01808006688952446,
      "step": 3210
    },
    {
      "epoch": 0.6087945413191812,
      "grad_norm": 0.19179895520210266,
      "learning_rate": 0.00018285381026438572,
      "loss": 0.01584501378238201,
      "step": 3212
    },
    {
      "epoch": 0.6091736163760425,
      "grad_norm": 0.0827220231294632,
      "learning_rate": 0.00018277604976671852,
      "loss": 0.023803936317563057,
      "step": 3214
    },
    {
      "epoch": 0.6095526914329037,
      "grad_norm": 0.08579916507005692,
      "learning_rate": 0.00018269828926905133,
      "loss": 0.011877713724970818,
      "step": 3216
    },
    {
      "epoch": 0.609931766489765,
      "grad_norm": 0.3505338132381439,
      "learning_rate": 0.00018262052877138414,
      "loss": 0.03014727123081684,
      "step": 3218
    },
    {
      "epoch": 0.6103108415466262,
      "grad_norm": 0.5168899297714233,
      "learning_rate": 0.00018254276827371697,
      "loss": 0.04813176766037941,
      "step": 3220
    },
    {
      "epoch": 0.6106899166034875,
      "grad_norm": 0.5227036476135254,
      "learning_rate": 0.00018246500777604978,
      "loss": 0.04630754515528679,
      "step": 3222
    },
    {
      "epoch": 0.6110689916603488,
      "grad_norm": 0.11704640835523605,
      "learning_rate": 0.00018238724727838259,
      "loss": 0.012548375874757767,
      "step": 3224
    },
    {
      "epoch": 0.61144806671721,
      "grad_norm": 0.35705655813217163,
      "learning_rate": 0.0001823094867807154,
      "loss": 0.026378462091088295,
      "step": 3226
    },
    {
      "epoch": 0.6118271417740713,
      "grad_norm": 0.11677544564008713,
      "learning_rate": 0.0001822317262830482,
      "loss": 0.01682409457862377,
      "step": 3228
    },
    {
      "epoch": 0.6122062168309326,
      "grad_norm": 0.30060335993766785,
      "learning_rate": 0.00018215396578538103,
      "loss": 0.040233317762613297,
      "step": 3230
    },
    {
      "epoch": 0.6125852918877938,
      "grad_norm": 0.27935293316841125,
      "learning_rate": 0.00018207620528771384,
      "loss": 0.023907221853733063,
      "step": 3232
    },
    {
      "epoch": 0.6129643669446551,
      "grad_norm": 0.32262933254241943,
      "learning_rate": 0.00018199844479004667,
      "loss": 0.04721497371792793,
      "step": 3234
    },
    {
      "epoch": 0.6133434420015162,
      "grad_norm": 0.11894410848617554,
      "learning_rate": 0.00018192068429237948,
      "loss": 0.02418765425682068,
      "step": 3236
    },
    {
      "epoch": 0.6137225170583775,
      "grad_norm": 0.34787479043006897,
      "learning_rate": 0.00018184292379471231,
      "loss": 0.033630166202783585,
      "step": 3238
    },
    {
      "epoch": 0.6141015921152388,
      "grad_norm": 0.31326207518577576,
      "learning_rate": 0.00018176516329704512,
      "loss": 0.02226308546960354,
      "step": 3240
    },
    {
      "epoch": 0.6144806671721,
      "grad_norm": 0.10183638334274292,
      "learning_rate": 0.00018168740279937793,
      "loss": 0.013545099645853043,
      "step": 3242
    },
    {
      "epoch": 0.6148597422289613,
      "grad_norm": 0.11163836717605591,
      "learning_rate": 0.00018160964230171074,
      "loss": 0.014058856293559074,
      "step": 3244
    },
    {
      "epoch": 0.6152388172858226,
      "grad_norm": 0.10245490819215775,
      "learning_rate": 0.00018153188180404354,
      "loss": 0.013094826601445675,
      "step": 3246
    },
    {
      "epoch": 0.6156178923426838,
      "grad_norm": 0.278318852186203,
      "learning_rate": 0.00018145412130637638,
      "loss": 0.02157502807676792,
      "step": 3248
    },
    {
      "epoch": 0.6159969673995451,
      "grad_norm": 0.13279351592063904,
      "learning_rate": 0.00018137636080870918,
      "loss": 0.013720561750233173,
      "step": 3250
    },
    {
      "epoch": 0.6163760424564063,
      "grad_norm": 0.4894479215145111,
      "learning_rate": 0.000181298600311042,
      "loss": 0.0401906780898571,
      "step": 3252
    },
    {
      "epoch": 0.6167551175132676,
      "grad_norm": 0.31210002303123474,
      "learning_rate": 0.0001812208398133748,
      "loss": 0.0237442534416914,
      "step": 3254
    },
    {
      "epoch": 0.6171341925701289,
      "grad_norm": 0.21782945096492767,
      "learning_rate": 0.00018114307931570763,
      "loss": 0.03405555710196495,
      "step": 3256
    },
    {
      "epoch": 0.6175132676269901,
      "grad_norm": 0.38893750309944153,
      "learning_rate": 0.00018106531881804046,
      "loss": 0.027631498873233795,
      "step": 3258
    },
    {
      "epoch": 0.6178923426838514,
      "grad_norm": 0.5601509213447571,
      "learning_rate": 0.00018098755832037327,
      "loss": 0.037605781108140945,
      "step": 3260
    },
    {
      "epoch": 0.6182714177407127,
      "grad_norm": 0.1345251500606537,
      "learning_rate": 0.00018090979782270608,
      "loss": 0.024992814287543297,
      "step": 3262
    },
    {
      "epoch": 0.6186504927975739,
      "grad_norm": 0.1337820589542389,
      "learning_rate": 0.00018083203732503889,
      "loss": 0.02079440839588642,
      "step": 3264
    },
    {
      "epoch": 0.6190295678544352,
      "grad_norm": 0.12292345613241196,
      "learning_rate": 0.0001807542768273717,
      "loss": 0.03422113507986069,
      "step": 3266
    },
    {
      "epoch": 0.6194086429112965,
      "grad_norm": 0.16273459792137146,
      "learning_rate": 0.00018067651632970453,
      "loss": 0.011351805180311203,
      "step": 3268
    },
    {
      "epoch": 0.6197877179681577,
      "grad_norm": 0.13817422091960907,
      "learning_rate": 0.00018059875583203733,
      "loss": 0.030624890699982643,
      "step": 3270
    },
    {
      "epoch": 0.620166793025019,
      "grad_norm": 0.2407684326171875,
      "learning_rate": 0.00018052099533437014,
      "loss": 0.023189114406704903,
      "step": 3272
    },
    {
      "epoch": 0.6205458680818802,
      "grad_norm": 0.37191858887672424,
      "learning_rate": 0.00018044323483670295,
      "loss": 0.028864353895187378,
      "step": 3274
    },
    {
      "epoch": 0.6209249431387415,
      "grad_norm": 0.19669829308986664,
      "learning_rate": 0.00018036547433903578,
      "loss": 0.03190074488520622,
      "step": 3276
    },
    {
      "epoch": 0.6213040181956028,
      "grad_norm": 0.23308239877223969,
      "learning_rate": 0.0001802877138413686,
      "loss": 0.020435918122529984,
      "step": 3278
    },
    {
      "epoch": 0.621683093252464,
      "grad_norm": 0.09816472232341766,
      "learning_rate": 0.00018020995334370142,
      "loss": 0.012697177939116955,
      "step": 3280
    },
    {
      "epoch": 0.6220621683093253,
      "grad_norm": 0.19482895731925964,
      "learning_rate": 0.00018013219284603423,
      "loss": 0.013726494275033474,
      "step": 3282
    },
    {
      "epoch": 0.6224412433661866,
      "grad_norm": 0.09408392012119293,
      "learning_rate": 0.00018005443234836704,
      "loss": 0.03470344468951225,
      "step": 3284
    },
    {
      "epoch": 0.6228203184230477,
      "grad_norm": 0.21068313717842102,
      "learning_rate": 0.00017997667185069987,
      "loss": 0.030638786032795906,
      "step": 3286
    },
    {
      "epoch": 0.623199393479909,
      "grad_norm": 0.09732513874769211,
      "learning_rate": 0.00017989891135303268,
      "loss": 0.02776995860040188,
      "step": 3288
    },
    {
      "epoch": 0.6235784685367702,
      "grad_norm": 0.30780792236328125,
      "learning_rate": 0.00017982115085536548,
      "loss": 0.023205164819955826,
      "step": 3290
    },
    {
      "epoch": 0.6239575435936315,
      "grad_norm": 0.4203367829322815,
      "learning_rate": 0.0001797433903576983,
      "loss": 0.03377435728907585,
      "step": 3292
    },
    {
      "epoch": 0.6243366186504928,
      "grad_norm": 0.24539099633693695,
      "learning_rate": 0.0001796656298600311,
      "loss": 0.016636446118354797,
      "step": 3294
    },
    {
      "epoch": 0.624715693707354,
      "grad_norm": 0.3888906240463257,
      "learning_rate": 0.00017958786936236393,
      "loss": 0.026143047958612442,
      "step": 3296
    },
    {
      "epoch": 0.6250947687642153,
      "grad_norm": 0.0874098688364029,
      "learning_rate": 0.00017951010886469674,
      "loss": 0.03971945866942406,
      "step": 3298
    },
    {
      "epoch": 0.6254738438210766,
      "grad_norm": 0.5302427411079407,
      "learning_rate": 0.00017943234836702954,
      "loss": 0.04364989697933197,
      "step": 3300
    },
    {
      "epoch": 0.6258529188779378,
      "grad_norm": 0.3801578879356384,
      "learning_rate": 0.00017935458786936238,
      "loss": 0.037642061710357666,
      "step": 3302
    },
    {
      "epoch": 0.6262319939347991,
      "grad_norm": 0.17464666068553925,
      "learning_rate": 0.00017927682737169519,
      "loss": 0.01344422996044159,
      "step": 3304
    },
    {
      "epoch": 0.6266110689916603,
      "grad_norm": 0.3734140992164612,
      "learning_rate": 0.00017919906687402802,
      "loss": 0.03140664100646973,
      "step": 3306
    },
    {
      "epoch": 0.6269901440485216,
      "grad_norm": 0.3008832037448883,
      "learning_rate": 0.00017912130637636083,
      "loss": 0.030551627278327942,
      "step": 3308
    },
    {
      "epoch": 0.6273692191053829,
      "grad_norm": 0.15485943853855133,
      "learning_rate": 0.00017904354587869363,
      "loss": 0.038201138377189636,
      "step": 3310
    },
    {
      "epoch": 0.6277482941622441,
      "grad_norm": 0.3221513330936432,
      "learning_rate": 0.00017896578538102644,
      "loss": 0.034735485911369324,
      "step": 3312
    },
    {
      "epoch": 0.6281273692191054,
      "grad_norm": 0.3764536678791046,
      "learning_rate": 0.00017888802488335927,
      "loss": 0.03777703642845154,
      "step": 3314
    },
    {
      "epoch": 0.6285064442759667,
      "grad_norm": 0.18536259233951569,
      "learning_rate": 0.00017881026438569208,
      "loss": 0.029762912541627884,
      "step": 3316
    },
    {
      "epoch": 0.6288855193328279,
      "grad_norm": 0.1189325675368309,
      "learning_rate": 0.0001787325038880249,
      "loss": 0.014698104932904243,
      "step": 3318
    },
    {
      "epoch": 0.6292645943896892,
      "grad_norm": 0.14629042148590088,
      "learning_rate": 0.0001786547433903577,
      "loss": 0.01890277862548828,
      "step": 3320
    },
    {
      "epoch": 0.6296436694465504,
      "grad_norm": 0.3379471004009247,
      "learning_rate": 0.0001785769828926905,
      "loss": 0.034759897738695145,
      "step": 3322
    },
    {
      "epoch": 0.6300227445034117,
      "grad_norm": 0.11488811671733856,
      "learning_rate": 0.00017849922239502334,
      "loss": 0.03274564445018768,
      "step": 3324
    },
    {
      "epoch": 0.630401819560273,
      "grad_norm": 0.4276312291622162,
      "learning_rate": 0.00017842146189735617,
      "loss": 0.05421638488769531,
      "step": 3326
    },
    {
      "epoch": 0.6307808946171342,
      "grad_norm": 0.11401869356632233,
      "learning_rate": 0.00017834370139968898,
      "loss": 0.023169375956058502,
      "step": 3328
    },
    {
      "epoch": 0.6311599696739955,
      "grad_norm": 0.10652242600917816,
      "learning_rate": 0.00017826594090202178,
      "loss": 0.041071269661188126,
      "step": 3330
    },
    {
      "epoch": 0.6315390447308568,
      "grad_norm": 0.278795063495636,
      "learning_rate": 0.0001781881804043546,
      "loss": 0.02549360878765583,
      "step": 3332
    },
    {
      "epoch": 0.631918119787718,
      "grad_norm": 0.21662183105945587,
      "learning_rate": 0.00017811041990668742,
      "loss": 0.02440902777016163,
      "step": 3334
    },
    {
      "epoch": 0.6322971948445792,
      "grad_norm": 0.463920533657074,
      "learning_rate": 0.00017803265940902023,
      "loss": 0.04711126908659935,
      "step": 3336
    },
    {
      "epoch": 0.6326762699014404,
      "grad_norm": 0.28764528036117554,
      "learning_rate": 0.00017795489891135304,
      "loss": 0.026707220822572708,
      "step": 3338
    },
    {
      "epoch": 0.6330553449583017,
      "grad_norm": 0.517977774143219,
      "learning_rate": 0.00017787713841368585,
      "loss": 0.039234671741724014,
      "step": 3340
    },
    {
      "epoch": 0.633434420015163,
      "grad_norm": 0.30359581112861633,
      "learning_rate": 0.00017779937791601865,
      "loss": 0.028800586238503456,
      "step": 3342
    },
    {
      "epoch": 0.6338134950720242,
      "grad_norm": 0.1340775042772293,
      "learning_rate": 0.00017772161741835149,
      "loss": 0.013774420134723186,
      "step": 3344
    },
    {
      "epoch": 0.6341925701288855,
      "grad_norm": 0.11138175427913666,
      "learning_rate": 0.0001776438569206843,
      "loss": 0.015218285843729973,
      "step": 3346
    },
    {
      "epoch": 0.6345716451857468,
      "grad_norm": 0.4569263756275177,
      "learning_rate": 0.00017756609642301713,
      "loss": 0.05299424007534981,
      "step": 3348
    },
    {
      "epoch": 0.634950720242608,
      "grad_norm": 0.396645188331604,
      "learning_rate": 0.00017748833592534993,
      "loss": 0.024439195170998573,
      "step": 3350
    },
    {
      "epoch": 0.6353297952994693,
      "grad_norm": 0.1020454540848732,
      "learning_rate": 0.00017741057542768274,
      "loss": 0.018166743218898773,
      "step": 3352
    },
    {
      "epoch": 0.6357088703563305,
      "grad_norm": 0.08204886317253113,
      "learning_rate": 0.00017733281493001557,
      "loss": 0.021728072315454483,
      "step": 3354
    },
    {
      "epoch": 0.6360879454131918,
      "grad_norm": 0.32785701751708984,
      "learning_rate": 0.00017725505443234838,
      "loss": 0.02478598617017269,
      "step": 3356
    },
    {
      "epoch": 0.6364670204700531,
      "grad_norm": 0.3905608654022217,
      "learning_rate": 0.0001771772939346812,
      "loss": 0.02952464297413826,
      "step": 3358
    },
    {
      "epoch": 0.6368460955269143,
      "grad_norm": 0.4114357829093933,
      "learning_rate": 0.000177099533437014,
      "loss": 0.049961674958467484,
      "step": 3360
    },
    {
      "epoch": 0.6372251705837756,
      "grad_norm": 0.14502963423728943,
      "learning_rate": 0.00017702177293934683,
      "loss": 0.023696573451161385,
      "step": 3362
    },
    {
      "epoch": 0.6376042456406369,
      "grad_norm": 0.2875111699104309,
      "learning_rate": 0.00017694401244167964,
      "loss": 0.019258739426732063,
      "step": 3364
    },
    {
      "epoch": 0.6379833206974981,
      "grad_norm": 0.44048789143562317,
      "learning_rate": 0.00017686625194401244,
      "loss": 0.039614833891391754,
      "step": 3366
    },
    {
      "epoch": 0.6383623957543594,
      "grad_norm": 0.12283027172088623,
      "learning_rate": 0.00017678849144634525,
      "loss": 0.018071522936224937,
      "step": 3368
    },
    {
      "epoch": 0.6387414708112206,
      "grad_norm": 0.3707711100578308,
      "learning_rate": 0.00017671073094867808,
      "loss": 0.03960196301341057,
      "step": 3370
    },
    {
      "epoch": 0.6391205458680819,
      "grad_norm": 0.09816562384366989,
      "learning_rate": 0.0001766329704510109,
      "loss": 0.028645750135183334,
      "step": 3372
    },
    {
      "epoch": 0.6394996209249432,
      "grad_norm": 0.24374711513519287,
      "learning_rate": 0.00017655520995334372,
      "loss": 0.017832301557064056,
      "step": 3374
    },
    {
      "epoch": 0.6398786959818044,
      "grad_norm": 0.37717247009277344,
      "learning_rate": 0.00017647744945567653,
      "loss": 0.05300360545516014,
      "step": 3376
    },
    {
      "epoch": 0.6402577710386657,
      "grad_norm": 0.2986949384212494,
      "learning_rate": 0.00017639968895800934,
      "loss": 0.024783741682767868,
      "step": 3378
    },
    {
      "epoch": 0.640636846095527,
      "grad_norm": 0.09412368386983871,
      "learning_rate": 0.00017632192846034215,
      "loss": 0.023975394666194916,
      "step": 3380
    },
    {
      "epoch": 0.6410159211523881,
      "grad_norm": 0.22220611572265625,
      "learning_rate": 0.00017624416796267498,
      "loss": 0.012907742522656918,
      "step": 3382
    },
    {
      "epoch": 0.6413949962092494,
      "grad_norm": 0.09353835135698318,
      "learning_rate": 0.0001761664074650078,
      "loss": 0.019241249188780785,
      "step": 3384
    },
    {
      "epoch": 0.6417740712661106,
      "grad_norm": 0.3198024332523346,
      "learning_rate": 0.0001760886469673406,
      "loss": 0.023146186023950577,
      "step": 3386
    },
    {
      "epoch": 0.6421531463229719,
      "grad_norm": 0.31293871998786926,
      "learning_rate": 0.0001760108864696734,
      "loss": 0.0252363458275795,
      "step": 3388
    },
    {
      "epoch": 0.6425322213798332,
      "grad_norm": 0.23196402192115784,
      "learning_rate": 0.0001759331259720062,
      "loss": 0.02843927964568138,
      "step": 3390
    },
    {
      "epoch": 0.6429112964366944,
      "grad_norm": 0.20190340280532837,
      "learning_rate": 0.00017585536547433904,
      "loss": 0.022255972027778625,
      "step": 3392
    },
    {
      "epoch": 0.6432903714935557,
      "grad_norm": 0.10050095617771149,
      "learning_rate": 0.00017577760497667185,
      "loss": 0.026586124673485756,
      "step": 3394
    },
    {
      "epoch": 0.643669446550417,
      "grad_norm": 0.12878257036209106,
      "learning_rate": 0.00017569984447900468,
      "loss": 0.013756700791418552,
      "step": 3396
    },
    {
      "epoch": 0.6440485216072782,
      "grad_norm": 0.4154371917247772,
      "learning_rate": 0.0001756220839813375,
      "loss": 0.030919451266527176,
      "step": 3398
    },
    {
      "epoch": 0.6444275966641395,
      "grad_norm": 0.38877013325691223,
      "learning_rate": 0.00017554432348367032,
      "loss": 0.04369905963540077,
      "step": 3400
    },
    {
      "epoch": 0.6448066717210008,
      "grad_norm": 0.0748712345957756,
      "learning_rate": 0.00017546656298600313,
      "loss": 0.03188316151499748,
      "step": 3402
    },
    {
      "epoch": 0.645185746777862,
      "grad_norm": 0.23602420091629028,
      "learning_rate": 0.00017538880248833594,
      "loss": 0.016716131940484047,
      "step": 3404
    },
    {
      "epoch": 0.6455648218347233,
      "grad_norm": 0.26962774991989136,
      "learning_rate": 0.00017531104199066874,
      "loss": 0.0179863590747118,
      "step": 3406
    },
    {
      "epoch": 0.6459438968915845,
      "grad_norm": 0.1541454792022705,
      "learning_rate": 0.00017523328149300155,
      "loss": 0.01325973216444254,
      "step": 3408
    },
    {
      "epoch": 0.6463229719484458,
      "grad_norm": 0.07856134325265884,
      "learning_rate": 0.00017515552099533438,
      "loss": 0.01674441620707512,
      "step": 3410
    },
    {
      "epoch": 0.6467020470053071,
      "grad_norm": 0.13531701266765594,
      "learning_rate": 0.0001750777604976672,
      "loss": 0.021194355562329292,
      "step": 3412
    },
    {
      "epoch": 0.6470811220621683,
      "grad_norm": 0.11325180530548096,
      "learning_rate": 0.000175,
      "loss": 0.024266084656119347,
      "step": 3414
    },
    {
      "epoch": 0.6474601971190296,
      "grad_norm": 0.5441402196884155,
      "learning_rate": 0.0001749222395023328,
      "loss": 0.0468016155064106,
      "step": 3416
    },
    {
      "epoch": 0.6478392721758909,
      "grad_norm": 0.09842861443758011,
      "learning_rate": 0.00017484447900466564,
      "loss": 0.017690490931272507,
      "step": 3418
    },
    {
      "epoch": 0.6482183472327521,
      "grad_norm": 0.09100332856178284,
      "learning_rate": 0.00017476671850699847,
      "loss": 0.018599586561322212,
      "step": 3420
    },
    {
      "epoch": 0.6485974222896134,
      "grad_norm": 0.316900372505188,
      "learning_rate": 0.00017468895800933128,
      "loss": 0.02100774087011814,
      "step": 3422
    },
    {
      "epoch": 0.6489764973464746,
      "grad_norm": 0.27801504731178284,
      "learning_rate": 0.0001746111975116641,
      "loss": 0.034158188849687576,
      "step": 3424
    },
    {
      "epoch": 0.6493555724033359,
      "grad_norm": 0.241963729262352,
      "learning_rate": 0.0001745334370139969,
      "loss": 0.01585143245756626,
      "step": 3426
    },
    {
      "epoch": 0.6497346474601972,
      "grad_norm": 0.431583046913147,
      "learning_rate": 0.0001744556765163297,
      "loss": 0.042163483798503876,
      "step": 3428
    },
    {
      "epoch": 0.6501137225170583,
      "grad_norm": 0.1175161823630333,
      "learning_rate": 0.00017437791601866253,
      "loss": 0.02009773626923561,
      "step": 3430
    },
    {
      "epoch": 0.6504927975739196,
      "grad_norm": 0.06938846409320831,
      "learning_rate": 0.00017430015552099534,
      "loss": 0.019581951200962067,
      "step": 3432
    },
    {
      "epoch": 0.650871872630781,
      "grad_norm": 0.13978637754917145,
      "learning_rate": 0.00017422239502332815,
      "loss": 0.013395245186984539,
      "step": 3434
    },
    {
      "epoch": 0.6512509476876421,
      "grad_norm": 0.11233848333358765,
      "learning_rate": 0.00017414463452566096,
      "loss": 0.026737915351986885,
      "step": 3436
    },
    {
      "epoch": 0.6516300227445034,
      "grad_norm": 0.4428749084472656,
      "learning_rate": 0.0001740668740279938,
      "loss": 0.03789572790265083,
      "step": 3438
    },
    {
      "epoch": 0.6520090978013646,
      "grad_norm": 0.3380175232887268,
      "learning_rate": 0.0001739891135303266,
      "loss": 0.01963229663670063,
      "step": 3440
    },
    {
      "epoch": 0.6523881728582259,
      "grad_norm": 0.13347052037715912,
      "learning_rate": 0.00017391135303265943,
      "loss": 0.013603758998215199,
      "step": 3442
    },
    {
      "epoch": 0.6527672479150872,
      "grad_norm": 0.36488765478134155,
      "learning_rate": 0.00017383359253499224,
      "loss": 0.02233785018324852,
      "step": 3444
    },
    {
      "epoch": 0.6531463229719484,
      "grad_norm": 0.4152196943759918,
      "learning_rate": 0.00017375583203732504,
      "loss": 0.031349245458841324,
      "step": 3446
    },
    {
      "epoch": 0.6535253980288097,
      "grad_norm": 0.08245529234409332,
      "learning_rate": 0.00017367807153965788,
      "loss": 0.016957350075244904,
      "step": 3448
    },
    {
      "epoch": 0.653904473085671,
      "grad_norm": 0.1218654066324234,
      "learning_rate": 0.00017360031104199068,
      "loss": 0.01512941624969244,
      "step": 3450
    },
    {
      "epoch": 0.6542835481425322,
      "grad_norm": 0.1360698789358139,
      "learning_rate": 0.0001735225505443235,
      "loss": 0.025745587423443794,
      "step": 3452
    },
    {
      "epoch": 0.6546626231993935,
      "grad_norm": 0.2280663102865219,
      "learning_rate": 0.0001734447900466563,
      "loss": 0.01923779770731926,
      "step": 3454
    },
    {
      "epoch": 0.6550416982562547,
      "grad_norm": 0.39935532212257385,
      "learning_rate": 0.0001733670295489891,
      "loss": 0.03644709289073944,
      "step": 3456
    },
    {
      "epoch": 0.655420773313116,
      "grad_norm": 0.20266395807266235,
      "learning_rate": 0.00017328926905132194,
      "loss": 0.03484073281288147,
      "step": 3458
    },
    {
      "epoch": 0.6557998483699773,
      "grad_norm": 0.37619510293006897,
      "learning_rate": 0.00017321150855365475,
      "loss": 0.027734171599149704,
      "step": 3460
    },
    {
      "epoch": 0.6561789234268385,
      "grad_norm": 0.20406295359134674,
      "learning_rate": 0.00017313374805598755,
      "loss": 0.01723676174879074,
      "step": 3462
    },
    {
      "epoch": 0.6565579984836998,
      "grad_norm": 0.22849975526332855,
      "learning_rate": 0.0001730559875583204,
      "loss": 0.01754373125731945,
      "step": 3464
    },
    {
      "epoch": 0.6569370735405611,
      "grad_norm": 0.12225779891014099,
      "learning_rate": 0.0001729782270606532,
      "loss": 0.02431299537420273,
      "step": 3466
    },
    {
      "epoch": 0.6573161485974223,
      "grad_norm": 0.35298362374305725,
      "learning_rate": 0.00017290046656298603,
      "loss": 0.03120112419128418,
      "step": 3468
    },
    {
      "epoch": 0.6576952236542836,
      "grad_norm": 0.12309639155864716,
      "learning_rate": 0.00017282270606531883,
      "loss": 0.01199385803192854,
      "step": 3470
    },
    {
      "epoch": 0.6580742987111448,
      "grad_norm": 0.08981367945671082,
      "learning_rate": 0.00017274494556765164,
      "loss": 0.013332163915038109,
      "step": 3472
    },
    {
      "epoch": 0.6584533737680061,
      "grad_norm": 0.08065543323755264,
      "learning_rate": 0.00017266718506998445,
      "loss": 0.011098528280854225,
      "step": 3474
    },
    {
      "epoch": 0.6588324488248674,
      "grad_norm": 0.10327451676130295,
      "learning_rate": 0.00017258942457231728,
      "loss": 0.013030455447733402,
      "step": 3476
    },
    {
      "epoch": 0.6592115238817285,
      "grad_norm": 0.09056972712278366,
      "learning_rate": 0.0001725116640746501,
      "loss": 0.018093513324856758,
      "step": 3478
    },
    {
      "epoch": 0.6595905989385898,
      "grad_norm": 0.37239548563957214,
      "learning_rate": 0.0001724339035769829,
      "loss": 0.027799587696790695,
      "step": 3480
    },
    {
      "epoch": 0.6599696739954511,
      "grad_norm": 0.3230436146259308,
      "learning_rate": 0.0001723561430793157,
      "loss": 0.02649436704814434,
      "step": 3482
    },
    {
      "epoch": 0.6603487490523123,
      "grad_norm": 0.0923476442694664,
      "learning_rate": 0.0001722783825816485,
      "loss": 0.015032801777124405,
      "step": 3484
    },
    {
      "epoch": 0.6607278241091736,
      "grad_norm": 0.09116597473621368,
      "learning_rate": 0.00017220062208398134,
      "loss": 0.012479313649237156,
      "step": 3486
    },
    {
      "epoch": 0.6611068991660348,
      "grad_norm": 0.09168840944766998,
      "learning_rate": 0.00017212286158631418,
      "loss": 0.01147851999849081,
      "step": 3488
    },
    {
      "epoch": 0.6614859742228961,
      "grad_norm": 0.429232656955719,
      "learning_rate": 0.00017204510108864699,
      "loss": 0.050790488719940186,
      "step": 3490
    },
    {
      "epoch": 0.6618650492797574,
      "grad_norm": 0.1758628934621811,
      "learning_rate": 0.0001719673405909798,
      "loss": 0.013604709878563881,
      "step": 3492
    },
    {
      "epoch": 0.6622441243366186,
      "grad_norm": 0.2061578631401062,
      "learning_rate": 0.0001718895800933126,
      "loss": 0.03629439324140549,
      "step": 3494
    },
    {
      "epoch": 0.6626231993934799,
      "grad_norm": 0.207191601395607,
      "learning_rate": 0.00017181181959564543,
      "loss": 0.01857171021401882,
      "step": 3496
    },
    {
      "epoch": 0.6630022744503412,
      "grad_norm": 0.11149947345256805,
      "learning_rate": 0.00017173405909797824,
      "loss": 0.013156065717339516,
      "step": 3498
    },
    {
      "epoch": 0.6633813495072024,
      "grad_norm": 0.08852235972881317,
      "learning_rate": 0.00017165629860031105,
      "loss": 0.021524205803871155,
      "step": 3500
    },
    {
      "epoch": 0.6637604245640637,
      "grad_norm": 0.3610161542892456,
      "learning_rate": 0.00017157853810264385,
      "loss": 0.0373663604259491,
      "step": 3502
    },
    {
      "epoch": 0.6641394996209249,
      "grad_norm": 0.24614308774471283,
      "learning_rate": 0.00017150077760497666,
      "loss": 0.032031215727329254,
      "step": 3504
    },
    {
      "epoch": 0.6645185746777862,
      "grad_norm": 0.12059824168682098,
      "learning_rate": 0.0001714230171073095,
      "loss": 0.03620374947786331,
      "step": 3506
    },
    {
      "epoch": 0.6648976497346475,
      "grad_norm": 0.35474225878715515,
      "learning_rate": 0.0001713452566096423,
      "loss": 0.048281650990247726,
      "step": 3508
    },
    {
      "epoch": 0.6652767247915087,
      "grad_norm": 0.30771970748901367,
      "learning_rate": 0.00017126749611197514,
      "loss": 0.02021416649222374,
      "step": 3510
    },
    {
      "epoch": 0.66565579984837,
      "grad_norm": 0.11138124763965607,
      "learning_rate": 0.00017118973561430794,
      "loss": 0.02034640870988369,
      "step": 3512
    },
    {
      "epoch": 0.6660348749052313,
      "grad_norm": 0.11345259100198746,
      "learning_rate": 0.00017111197511664075,
      "loss": 0.015663381665945053,
      "step": 3514
    },
    {
      "epoch": 0.6664139499620925,
      "grad_norm": 0.2762690782546997,
      "learning_rate": 0.00017103421461897358,
      "loss": 0.031187361106276512,
      "step": 3516
    },
    {
      "epoch": 0.6667930250189538,
      "grad_norm": 0.31891068816185,
      "learning_rate": 0.0001709564541213064,
      "loss": 0.0321759395301342,
      "step": 3518
    },
    {
      "epoch": 0.6671721000758151,
      "grad_norm": 0.39584121108055115,
      "learning_rate": 0.0001708786936236392,
      "loss": 0.036889705806970596,
      "step": 3520
    },
    {
      "epoch": 0.6675511751326763,
      "grad_norm": 0.184176966547966,
      "learning_rate": 0.000170800933125972,
      "loss": 0.01688426546752453,
      "step": 3522
    },
    {
      "epoch": 0.6679302501895376,
      "grad_norm": 0.22528673708438873,
      "learning_rate": 0.00017072317262830484,
      "loss": 0.017172308638691902,
      "step": 3524
    },
    {
      "epoch": 0.6683093252463987,
      "grad_norm": 0.3061615824699402,
      "learning_rate": 0.00017064541213063764,
      "loss": 0.03337349370121956,
      "step": 3526
    },
    {
      "epoch": 0.66868840030326,
      "grad_norm": 0.2640726566314697,
      "learning_rate": 0.00017056765163297045,
      "loss": 0.018484819680452347,
      "step": 3528
    },
    {
      "epoch": 0.6690674753601213,
      "grad_norm": 0.08962294459342957,
      "learning_rate": 0.00017048989113530326,
      "loss": 0.011996286921203136,
      "step": 3530
    },
    {
      "epoch": 0.6694465504169825,
      "grad_norm": 0.39165830612182617,
      "learning_rate": 0.0001704121306376361,
      "loss": 0.03995180130004883,
      "step": 3532
    },
    {
      "epoch": 0.6698256254738438,
      "grad_norm": 0.13084030151367188,
      "learning_rate": 0.00017033437013996893,
      "loss": 0.022127198055386543,
      "step": 3534
    },
    {
      "epoch": 0.6702047005307051,
      "grad_norm": 0.0984172448515892,
      "learning_rate": 0.00017025660964230173,
      "loss": 0.01846393197774887,
      "step": 3536
    },
    {
      "epoch": 0.6705837755875663,
      "grad_norm": 0.23148265480995178,
      "learning_rate": 0.00017017884914463454,
      "loss": 0.028343241661787033,
      "step": 3538
    },
    {
      "epoch": 0.6709628506444276,
      "grad_norm": 0.11444826424121857,
      "learning_rate": 0.00017010108864696735,
      "loss": 0.01499917358160019,
      "step": 3540
    },
    {
      "epoch": 0.6713419257012888,
      "grad_norm": 0.28989484906196594,
      "learning_rate": 0.00017002332814930015,
      "loss": 0.02209066040813923,
      "step": 3542
    },
    {
      "epoch": 0.6717210007581501,
      "grad_norm": 0.09038116782903671,
      "learning_rate": 0.000169945567651633,
      "loss": 0.01118813268840313,
      "step": 3544
    },
    {
      "epoch": 0.6721000758150114,
      "grad_norm": 0.34947654604911804,
      "learning_rate": 0.0001698678071539658,
      "loss": 0.021143821999430656,
      "step": 3546
    },
    {
      "epoch": 0.6724791508718726,
      "grad_norm": 0.4094589352607727,
      "learning_rate": 0.0001697900466562986,
      "loss": 0.02443072572350502,
      "step": 3548
    },
    {
      "epoch": 0.6728582259287339,
      "grad_norm": 0.13793820142745972,
      "learning_rate": 0.0001697122861586314,
      "loss": 0.019618451595306396,
      "step": 3550
    },
    {
      "epoch": 0.6732373009855952,
      "grad_norm": 0.11049696803092957,
      "learning_rate": 0.00016963452566096422,
      "loss": 0.03059825673699379,
      "step": 3552
    },
    {
      "epoch": 0.6736163760424564,
      "grad_norm": 0.16386811435222626,
      "learning_rate": 0.00016955676516329705,
      "loss": 0.020082591101527214,
      "step": 3554
    },
    {
      "epoch": 0.6739954510993177,
      "grad_norm": 0.3109893798828125,
      "learning_rate": 0.00016947900466562988,
      "loss": 0.0354599803686142,
      "step": 3556
    },
    {
      "epoch": 0.6743745261561789,
      "grad_norm": 0.11823097616434097,
      "learning_rate": 0.0001694012441679627,
      "loss": 0.020967554301023483,
      "step": 3558
    },
    {
      "epoch": 0.6747536012130402,
      "grad_norm": 0.12166599184274673,
      "learning_rate": 0.0001693234836702955,
      "loss": 0.026991460472345352,
      "step": 3560
    },
    {
      "epoch": 0.6751326762699015,
      "grad_norm": 0.22176285088062286,
      "learning_rate": 0.00016924572317262833,
      "loss": 0.012458092533051968,
      "step": 3562
    },
    {
      "epoch": 0.6755117513267627,
      "grad_norm": 0.3549883961677551,
      "learning_rate": 0.00016916796267496114,
      "loss": 0.04119597747921944,
      "step": 3564
    },
    {
      "epoch": 0.675890826383624,
      "grad_norm": 0.328512966632843,
      "learning_rate": 0.00016909020217729394,
      "loss": 0.03839333355426788,
      "step": 3566
    },
    {
      "epoch": 0.6762699014404853,
      "grad_norm": 0.38386231660842896,
      "learning_rate": 0.00016901244167962675,
      "loss": 0.023743662983179092,
      "step": 3568
    },
    {
      "epoch": 0.6766489764973465,
      "grad_norm": 0.3365882933139801,
      "learning_rate": 0.00016893468118195956,
      "loss": 0.024985207244753838,
      "step": 3570
    },
    {
      "epoch": 0.6770280515542078,
      "grad_norm": 0.302367627620697,
      "learning_rate": 0.0001688569206842924,
      "loss": 0.02163739502429962,
      "step": 3572
    },
    {
      "epoch": 0.6774071266110689,
      "grad_norm": 0.24074001610279083,
      "learning_rate": 0.0001687791601866252,
      "loss": 0.01443205215036869,
      "step": 3574
    },
    {
      "epoch": 0.6777862016679302,
      "grad_norm": 0.10016684979200363,
      "learning_rate": 0.000168701399688958,
      "loss": 0.018765566870570183,
      "step": 3576
    },
    {
      "epoch": 0.6781652767247915,
      "grad_norm": 0.08951256424188614,
      "learning_rate": 0.00016862363919129084,
      "loss": 0.012963931076228619,
      "step": 3578
    },
    {
      "epoch": 0.6785443517816527,
      "grad_norm": 0.2944927513599396,
      "learning_rate": 0.00016854587869362365,
      "loss": 0.027621448040008545,
      "step": 3580
    },
    {
      "epoch": 0.678923426838514,
      "grad_norm": 0.3897378444671631,
      "learning_rate": 0.00016846811819595648,
      "loss": 0.03757451847195625,
      "step": 3582
    },
    {
      "epoch": 0.6793025018953753,
      "grad_norm": 0.10395447164773941,
      "learning_rate": 0.0001683903576982893,
      "loss": 0.03959803283214569,
      "step": 3584
    },
    {
      "epoch": 0.6796815769522365,
      "grad_norm": 0.1616179347038269,
      "learning_rate": 0.0001683125972006221,
      "loss": 0.014667574316263199,
      "step": 3586
    },
    {
      "epoch": 0.6800606520090978,
      "grad_norm": 0.38340073823928833,
      "learning_rate": 0.0001682348367029549,
      "loss": 0.025527482852339745,
      "step": 3588
    },
    {
      "epoch": 0.680439727065959,
      "grad_norm": 0.49462518095970154,
      "learning_rate": 0.0001681570762052877,
      "loss": 0.02576283924281597,
      "step": 3590
    },
    {
      "epoch": 0.6808188021228203,
      "grad_norm": 0.11535271257162094,
      "learning_rate": 0.00016807931570762054,
      "loss": 0.011362140066921711,
      "step": 3592
    },
    {
      "epoch": 0.6811978771796816,
      "grad_norm": 0.10844149440526962,
      "learning_rate": 0.00016800155520995335,
      "loss": 0.01676822453737259,
      "step": 3594
    },
    {
      "epoch": 0.6815769522365428,
      "grad_norm": 0.10799943655729294,
      "learning_rate": 0.00016792379471228616,
      "loss": 0.0249034333974123,
      "step": 3596
    },
    {
      "epoch": 0.6819560272934041,
      "grad_norm": 0.4497305452823639,
      "learning_rate": 0.00016784603421461896,
      "loss": 0.034122172743082047,
      "step": 3598
    },
    {
      "epoch": 0.6823351023502654,
      "grad_norm": 0.47513675689697266,
      "learning_rate": 0.0001677682737169518,
      "loss": 0.02865596115589142,
      "step": 3600
    },
    {
      "epoch": 0.6827141774071266,
      "grad_norm": 0.3015705645084381,
      "learning_rate": 0.0001676905132192846,
      "loss": 0.028854206204414368,
      "step": 3602
    },
    {
      "epoch": 0.6830932524639879,
      "grad_norm": 0.4533705413341522,
      "learning_rate": 0.00016761275272161744,
      "loss": 0.03788186237215996,
      "step": 3604
    },
    {
      "epoch": 0.6834723275208491,
      "grad_norm": 0.29058638215065,
      "learning_rate": 0.00016753499222395025,
      "loss": 0.031038133427500725,
      "step": 3606
    },
    {
      "epoch": 0.6838514025777104,
      "grad_norm": 1.1718566417694092,
      "learning_rate": 0.00016745723172628305,
      "loss": 0.02531834878027439,
      "step": 3608
    },
    {
      "epoch": 0.6842304776345717,
      "grad_norm": 0.12408395111560822,
      "learning_rate": 0.00016737947122861589,
      "loss": 0.012513711117208004,
      "step": 3610
    },
    {
      "epoch": 0.6846095526914329,
      "grad_norm": 0.3866102397441864,
      "learning_rate": 0.0001673017107309487,
      "loss": 0.033992234617471695,
      "step": 3612
    },
    {
      "epoch": 0.6849886277482942,
      "grad_norm": 0.08195605129003525,
      "learning_rate": 0.0001672239502332815,
      "loss": 0.011205393821001053,
      "step": 3614
    },
    {
      "epoch": 0.6853677028051555,
      "grad_norm": 0.11382758617401123,
      "learning_rate": 0.0001671461897356143,
      "loss": 0.015316440723836422,
      "step": 3616
    },
    {
      "epoch": 0.6857467778620167,
      "grad_norm": 0.08455821871757507,
      "learning_rate": 0.00016706842923794711,
      "loss": 0.01152700837701559,
      "step": 3618
    },
    {
      "epoch": 0.686125852918878,
      "grad_norm": 0.35380762815475464,
      "learning_rate": 0.00016699066874027995,
      "loss": 0.03162272647023201,
      "step": 3620
    },
    {
      "epoch": 0.6865049279757391,
      "grad_norm": 0.3822330832481384,
      "learning_rate": 0.00016691290824261275,
      "loss": 0.022625241428613663,
      "step": 3622
    },
    {
      "epoch": 0.6868840030326004,
      "grad_norm": 0.13503678143024445,
      "learning_rate": 0.00016683514774494556,
      "loss": 0.014141300693154335,
      "step": 3624
    },
    {
      "epoch": 0.6872630780894617,
      "grad_norm": 0.41439199447631836,
      "learning_rate": 0.0001667573872472784,
      "loss": 0.03090517409145832,
      "step": 3626
    },
    {
      "epoch": 0.6876421531463229,
      "grad_norm": 0.3364662528038025,
      "learning_rate": 0.0001666796267496112,
      "loss": 0.02934015542268753,
      "step": 3628
    },
    {
      "epoch": 0.6880212282031842,
      "grad_norm": 0.4365597367286682,
      "learning_rate": 0.00016660186625194404,
      "loss": 0.03437846526503563,
      "step": 3630
    },
    {
      "epoch": 0.6884003032600455,
      "grad_norm": 0.09241114556789398,
      "learning_rate": 0.00016652410575427684,
      "loss": 0.01282141637057066,
      "step": 3632
    },
    {
      "epoch": 0.6887793783169067,
      "grad_norm": 0.3592415153980255,
      "learning_rate": 0.00016644634525660965,
      "loss": 0.025976400822401047,
      "step": 3634
    },
    {
      "epoch": 0.689158453373768,
      "grad_norm": 0.13298451900482178,
      "learning_rate": 0.00016636858475894246,
      "loss": 0.020652223378419876,
      "step": 3636
    },
    {
      "epoch": 0.6895375284306292,
      "grad_norm": 0.19080232083797455,
      "learning_rate": 0.0001662908242612753,
      "loss": 0.01649324968457222,
      "step": 3638
    },
    {
      "epoch": 0.6899166034874905,
      "grad_norm": 0.33017876744270325,
      "learning_rate": 0.0001662130637636081,
      "loss": 0.021176347509026527,
      "step": 3640
    },
    {
      "epoch": 0.6902956785443518,
      "grad_norm": 0.0742780864238739,
      "learning_rate": 0.0001661353032659409,
      "loss": 0.013201558962464333,
      "step": 3642
    },
    {
      "epoch": 0.690674753601213,
      "grad_norm": 0.38392266631126404,
      "learning_rate": 0.0001660575427682737,
      "loss": 0.032338012009859085,
      "step": 3644
    },
    {
      "epoch": 0.6910538286580743,
      "grad_norm": 0.4604906439781189,
      "learning_rate": 0.00016597978227060652,
      "loss": 0.02214246243238449,
      "step": 3646
    },
    {
      "epoch": 0.6914329037149356,
      "grad_norm": 0.1163201555609703,
      "learning_rate": 0.00016590202177293935,
      "loss": 0.02169300988316536,
      "step": 3648
    },
    {
      "epoch": 0.6918119787717968,
      "grad_norm": 0.19555918872356415,
      "learning_rate": 0.0001658242612752722,
      "loss": 0.018949491903185844,
      "step": 3650
    },
    {
      "epoch": 0.6921910538286581,
      "grad_norm": 0.3452649414539337,
      "learning_rate": 0.000165746500777605,
      "loss": 0.03284279257059097,
      "step": 3652
    },
    {
      "epoch": 0.6925701288855194,
      "grad_norm": 0.07689487189054489,
      "learning_rate": 0.0001656687402799378,
      "loss": 0.012615790590643883,
      "step": 3654
    },
    {
      "epoch": 0.6929492039423806,
      "grad_norm": 0.3599770665168762,
      "learning_rate": 0.0001655909797822706,
      "loss": 0.036964766681194305,
      "step": 3656
    },
    {
      "epoch": 0.6933282789992419,
      "grad_norm": 0.11197979748249054,
      "learning_rate": 0.00016551321928460344,
      "loss": 0.01306288130581379,
      "step": 3658
    },
    {
      "epoch": 0.6937073540561031,
      "grad_norm": 0.24143452942371368,
      "learning_rate": 0.00016543545878693625,
      "loss": 0.030776962637901306,
      "step": 3660
    },
    {
      "epoch": 0.6940864291129644,
      "grad_norm": 0.19540826976299286,
      "learning_rate": 0.00016535769828926905,
      "loss": 0.04549889266490936,
      "step": 3662
    },
    {
      "epoch": 0.6944655041698257,
      "grad_norm": 0.09301254898309708,
      "learning_rate": 0.00016527993779160186,
      "loss": 0.01253389474004507,
      "step": 3664
    },
    {
      "epoch": 0.6948445792266869,
      "grad_norm": 0.3538283705711365,
      "learning_rate": 0.00016520217729393467,
      "loss": 0.02301037684082985,
      "step": 3666
    },
    {
      "epoch": 0.6952236542835482,
      "grad_norm": 0.08886724710464478,
      "learning_rate": 0.0001651244167962675,
      "loss": 0.012419159524142742,
      "step": 3668
    },
    {
      "epoch": 0.6956027293404095,
      "grad_norm": 0.10832777619361877,
      "learning_rate": 0.0001650466562986003,
      "loss": 0.02710283175110817,
      "step": 3670
    },
    {
      "epoch": 0.6959818043972706,
      "grad_norm": 0.35245493054389954,
      "learning_rate": 0.00016496889580093314,
      "loss": 0.02095375582575798,
      "step": 3672
    },
    {
      "epoch": 0.6963608794541319,
      "grad_norm": 0.3430188298225403,
      "learning_rate": 0.00016489113530326595,
      "loss": 0.024711713194847107,
      "step": 3674
    },
    {
      "epoch": 0.6967399545109931,
      "grad_norm": 0.23034483194351196,
      "learning_rate": 0.00016481337480559878,
      "loss": 0.023822126910090446,
      "step": 3676
    },
    {
      "epoch": 0.6971190295678544,
      "grad_norm": 0.11144204437732697,
      "learning_rate": 0.0001647356143079316,
      "loss": 0.018018702045083046,
      "step": 3678
    },
    {
      "epoch": 0.6974981046247157,
      "grad_norm": 0.23723697662353516,
      "learning_rate": 0.0001646578538102644,
      "loss": 0.03408048301935196,
      "step": 3680
    },
    {
      "epoch": 0.6978771796815769,
      "grad_norm": 0.4101530909538269,
      "learning_rate": 0.0001645800933125972,
      "loss": 0.02767627313733101,
      "step": 3682
    },
    {
      "epoch": 0.6982562547384382,
      "grad_norm": 0.2891739308834076,
      "learning_rate": 0.00016450233281493,
      "loss": 0.018062982708215714,
      "step": 3684
    },
    {
      "epoch": 0.6986353297952995,
      "grad_norm": 0.4803006649017334,
      "learning_rate": 0.00016442457231726285,
      "loss": 0.041973747313022614,
      "step": 3686
    },
    {
      "epoch": 0.6990144048521607,
      "grad_norm": 0.09366986900568008,
      "learning_rate": 0.00016434681181959565,
      "loss": 0.02036505565047264,
      "step": 3688
    },
    {
      "epoch": 0.699393479909022,
      "grad_norm": 0.3003244400024414,
      "learning_rate": 0.00016426905132192846,
      "loss": 0.015203995630145073,
      "step": 3690
    },
    {
      "epoch": 0.6997725549658832,
      "grad_norm": 0.12939901649951935,
      "learning_rate": 0.00016419129082426127,
      "loss": 0.012823432683944702,
      "step": 3692
    },
    {
      "epoch": 0.7001516300227445,
      "grad_norm": 0.10884949564933777,
      "learning_rate": 0.0001641135303265941,
      "loss": 0.01875373348593712,
      "step": 3694
    },
    {
      "epoch": 0.7005307050796058,
      "grad_norm": 0.1036163941025734,
      "learning_rate": 0.00016403576982892693,
      "loss": 0.018994901329278946,
      "step": 3696
    },
    {
      "epoch": 0.700909780136467,
      "grad_norm": 0.20548662543296814,
      "learning_rate": 0.00016395800933125974,
      "loss": 0.014273257926106453,
      "step": 3698
    },
    {
      "epoch": 0.7012888551933283,
      "grad_norm": 0.373201459646225,
      "learning_rate": 0.00016388024883359255,
      "loss": 0.025826815515756607,
      "step": 3700
    },
    {
      "epoch": 0.7016679302501896,
      "grad_norm": 0.19456113874912262,
      "learning_rate": 0.00016380248833592536,
      "loss": 0.03305621072649956,
      "step": 3702
    },
    {
      "epoch": 0.7020470053070508,
      "grad_norm": 0.10107021778821945,
      "learning_rate": 0.00016372472783825816,
      "loss": 0.024068117141723633,
      "step": 3704
    },
    {
      "epoch": 0.7024260803639121,
      "grad_norm": 0.08600705862045288,
      "learning_rate": 0.000163646967340591,
      "loss": 0.022169120609760284,
      "step": 3706
    },
    {
      "epoch": 0.7028051554207733,
      "grad_norm": 0.11414072662591934,
      "learning_rate": 0.0001635692068429238,
      "loss": 0.01389405969530344,
      "step": 3708
    },
    {
      "epoch": 0.7031842304776346,
      "grad_norm": 0.2980480194091797,
      "learning_rate": 0.0001634914463452566,
      "loss": 0.01971915364265442,
      "step": 3710
    },
    {
      "epoch": 0.7035633055344959,
      "grad_norm": 0.10974039882421494,
      "learning_rate": 0.00016341368584758942,
      "loss": 0.015716614201664925,
      "step": 3712
    },
    {
      "epoch": 0.703942380591357,
      "grad_norm": 0.09690206497907639,
      "learning_rate": 0.00016333592534992222,
      "loss": 0.015969397500157356,
      "step": 3714
    },
    {
      "epoch": 0.7043214556482184,
      "grad_norm": 0.24730677902698517,
      "learning_rate": 0.00016325816485225506,
      "loss": 0.018580010160803795,
      "step": 3716
    },
    {
      "epoch": 0.7047005307050797,
      "grad_norm": 0.2108290195465088,
      "learning_rate": 0.0001631804043545879,
      "loss": 0.03133012354373932,
      "step": 3718
    },
    {
      "epoch": 0.7050796057619408,
      "grad_norm": 0.3113643229007721,
      "learning_rate": 0.0001631026438569207,
      "loss": 0.028511473909020424,
      "step": 3720
    },
    {
      "epoch": 0.7054586808188021,
      "grad_norm": 0.4051872193813324,
      "learning_rate": 0.0001630248833592535,
      "loss": 0.044911209493875504,
      "step": 3722
    },
    {
      "epoch": 0.7058377558756633,
      "grad_norm": 0.45286911725997925,
      "learning_rate": 0.00016294712286158634,
      "loss": 0.026746004819869995,
      "step": 3724
    },
    {
      "epoch": 0.7062168309325246,
      "grad_norm": 0.4300079047679901,
      "learning_rate": 0.00016286936236391915,
      "loss": 0.0486912801861763,
      "step": 3726
    },
    {
      "epoch": 0.7065959059893859,
      "grad_norm": 0.08381729573011398,
      "learning_rate": 0.00016279160186625195,
      "loss": 0.012622766196727753,
      "step": 3728
    },
    {
      "epoch": 0.7069749810462471,
      "grad_norm": 0.08030020445585251,
      "learning_rate": 0.00016271384136858476,
      "loss": 0.024250347167253494,
      "step": 3730
    },
    {
      "epoch": 0.7073540561031084,
      "grad_norm": 0.14577342569828033,
      "learning_rate": 0.00016263608087091757,
      "loss": 0.011291931383311749,
      "step": 3732
    },
    {
      "epoch": 0.7077331311599697,
      "grad_norm": 0.11862042546272278,
      "learning_rate": 0.0001625583203732504,
      "loss": 0.02827451191842556,
      "step": 3734
    },
    {
      "epoch": 0.7081122062168309,
      "grad_norm": 0.1121063157916069,
      "learning_rate": 0.0001624805598755832,
      "loss": 0.01433574315160513,
      "step": 3736
    },
    {
      "epoch": 0.7084912812736922,
      "grad_norm": 0.2726222276687622,
      "learning_rate": 0.00016240279937791601,
      "loss": 0.026545867323875427,
      "step": 3738
    },
    {
      "epoch": 0.7088703563305534,
      "grad_norm": 0.12574708461761475,
      "learning_rate": 0.00016232503888024885,
      "loss": 0.01807837001979351,
      "step": 3740
    },
    {
      "epoch": 0.7092494313874147,
      "grad_norm": 0.3361816108226776,
      "learning_rate": 0.00016224727838258166,
      "loss": 0.030352279543876648,
      "step": 3742
    },
    {
      "epoch": 0.709628506444276,
      "grad_norm": 0.13197819888591766,
      "learning_rate": 0.0001621695178849145,
      "loss": 0.029450759291648865,
      "step": 3744
    },
    {
      "epoch": 0.7100075815011372,
      "grad_norm": 0.2460751235485077,
      "learning_rate": 0.0001620917573872473,
      "loss": 0.020430805161595345,
      "step": 3746
    },
    {
      "epoch": 0.7103866565579985,
      "grad_norm": 0.2651704251766205,
      "learning_rate": 0.0001620139968895801,
      "loss": 0.031504470854997635,
      "step": 3748
    },
    {
      "epoch": 0.7107657316148598,
      "grad_norm": 0.38346654176712036,
      "learning_rate": 0.0001619362363919129,
      "loss": 0.0430711954832077,
      "step": 3750
    },
    {
      "epoch": 0.711144806671721,
      "grad_norm": 0.2703673243522644,
      "learning_rate": 0.00016185847589424572,
      "loss": 0.02743263728916645,
      "step": 3752
    },
    {
      "epoch": 0.7115238817285823,
      "grad_norm": 0.2821315824985504,
      "learning_rate": 0.00016178071539657855,
      "loss": 0.02252320572733879,
      "step": 3754
    },
    {
      "epoch": 0.7119029567854435,
      "grad_norm": 0.07775415480136871,
      "learning_rate": 0.00016170295489891136,
      "loss": 0.011281284503638744,
      "step": 3756
    },
    {
      "epoch": 0.7122820318423048,
      "grad_norm": 0.31607306003570557,
      "learning_rate": 0.00016162519440124416,
      "loss": 0.030728325247764587,
      "step": 3758
    },
    {
      "epoch": 0.7126611068991661,
      "grad_norm": 0.12120571732521057,
      "learning_rate": 0.00016154743390357697,
      "loss": 0.0143818911164999,
      "step": 3760
    },
    {
      "epoch": 0.7130401819560273,
      "grad_norm": 0.10300187021493912,
      "learning_rate": 0.0001614696734059098,
      "loss": 0.012627165764570236,
      "step": 3762
    },
    {
      "epoch": 0.7134192570128886,
      "grad_norm": 0.22659681737422943,
      "learning_rate": 0.00016139191290824264,
      "loss": 0.032022733241319656,
      "step": 3764
    },
    {
      "epoch": 0.7137983320697499,
      "grad_norm": 0.343531996011734,
      "learning_rate": 0.00016131415241057545,
      "loss": 0.0321887768805027,
      "step": 3766
    },
    {
      "epoch": 0.714177407126611,
      "grad_norm": 0.12817032635211945,
      "learning_rate": 0.00016123639191290825,
      "loss": 0.018340948969125748,
      "step": 3768
    },
    {
      "epoch": 0.7145564821834723,
      "grad_norm": 0.37473294138908386,
      "learning_rate": 0.00016115863141524106,
      "loss": 0.04471946507692337,
      "step": 3770
    },
    {
      "epoch": 0.7149355572403336,
      "grad_norm": 0.126593217253685,
      "learning_rate": 0.0001610808709175739,
      "loss": 0.01660117693245411,
      "step": 3772
    },
    {
      "epoch": 0.7153146322971948,
      "grad_norm": 0.3316359221935272,
      "learning_rate": 0.0001610031104199067,
      "loss": 0.027885504066944122,
      "step": 3774
    },
    {
      "epoch": 0.7156937073540561,
      "grad_norm": 0.10666915029287338,
      "learning_rate": 0.0001609253499222395,
      "loss": 0.014085742644965649,
      "step": 3776
    },
    {
      "epoch": 0.7160727824109173,
      "grad_norm": 0.36249080300331116,
      "learning_rate": 0.00016084758942457232,
      "loss": 0.023693718016147614,
      "step": 3778
    },
    {
      "epoch": 0.7164518574677786,
      "grad_norm": 0.136233851313591,
      "learning_rate": 0.00016076982892690512,
      "loss": 0.023260997608304024,
      "step": 3780
    },
    {
      "epoch": 0.7168309325246399,
      "grad_norm": 0.2733556628227234,
      "learning_rate": 0.00016069206842923796,
      "loss": 0.021041139960289,
      "step": 3782
    },
    {
      "epoch": 0.7172100075815011,
      "grad_norm": 0.2523189187049866,
      "learning_rate": 0.00016061430793157076,
      "loss": 0.019526466727256775,
      "step": 3784
    },
    {
      "epoch": 0.7175890826383624,
      "grad_norm": 0.28742000460624695,
      "learning_rate": 0.0001605365474339036,
      "loss": 0.02106403559446335,
      "step": 3786
    },
    {
      "epoch": 0.7179681576952237,
      "grad_norm": 0.10891538113355637,
      "learning_rate": 0.0001604587869362364,
      "loss": 0.015154009684920311,
      "step": 3788
    },
    {
      "epoch": 0.7183472327520849,
      "grad_norm": 0.09880595654249191,
      "learning_rate": 0.0001603810264385692,
      "loss": 0.03316599875688553,
      "step": 3790
    },
    {
      "epoch": 0.7187263078089462,
      "grad_norm": 0.4140206575393677,
      "learning_rate": 0.00016030326594090204,
      "loss": 0.029138769954442978,
      "step": 3792
    },
    {
      "epoch": 0.7191053828658074,
      "grad_norm": 0.3687931299209595,
      "learning_rate": 0.00016022550544323485,
      "loss": 0.034241583198308945,
      "step": 3794
    },
    {
      "epoch": 0.7194844579226687,
      "grad_norm": 0.40678465366363525,
      "learning_rate": 0.00016014774494556766,
      "loss": 0.04178975149989128,
      "step": 3796
    },
    {
      "epoch": 0.71986353297953,
      "grad_norm": 0.29976075887680054,
      "learning_rate": 0.00016006998444790047,
      "loss": 0.02377939783036709,
      "step": 3798
    },
    {
      "epoch": 0.7202426080363912,
      "grad_norm": 0.5536606311798096,
      "learning_rate": 0.0001599922239502333,
      "loss": 0.05173732340335846,
      "step": 3800
    },
    {
      "epoch": 0.7206216830932525,
      "grad_norm": 0.08457151800394058,
      "learning_rate": 0.0001599144634525661,
      "loss": 0.0364716574549675,
      "step": 3802
    },
    {
      "epoch": 0.7210007581501138,
      "grad_norm": 0.10762257128953934,
      "learning_rate": 0.0001598367029548989,
      "loss": 0.021300500258803368,
      "step": 3804
    },
    {
      "epoch": 0.721379833206975,
      "grad_norm": 0.4102235734462738,
      "learning_rate": 0.00015975894245723172,
      "loss": 0.03497566282749176,
      "step": 3806
    },
    {
      "epoch": 0.7217589082638363,
      "grad_norm": 0.4111984670162201,
      "learning_rate": 0.00015968118195956455,
      "loss": 0.04120643064379692,
      "step": 3808
    },
    {
      "epoch": 0.7221379833206975,
      "grad_norm": 0.13950464129447937,
      "learning_rate": 0.00015960342146189736,
      "loss": 0.014654318802058697,
      "step": 3810
    },
    {
      "epoch": 0.7225170583775588,
      "grad_norm": 0.37208816409111023,
      "learning_rate": 0.0001595256609642302,
      "loss": 0.030188193544745445,
      "step": 3812
    },
    {
      "epoch": 0.72289613343442,
      "grad_norm": 0.15438085794448853,
      "learning_rate": 0.000159447900466563,
      "loss": 0.020579542964696884,
      "step": 3814
    },
    {
      "epoch": 0.7232752084912812,
      "grad_norm": 0.4008321762084961,
      "learning_rate": 0.0001593701399688958,
      "loss": 0.029975702986121178,
      "step": 3816
    },
    {
      "epoch": 0.7236542835481425,
      "grad_norm": 0.10657744854688644,
      "learning_rate": 0.00015929237947122862,
      "loss": 0.03508894518017769,
      "step": 3818
    },
    {
      "epoch": 0.7240333586050038,
      "grad_norm": 0.12319596856832504,
      "learning_rate": 0.00015921461897356145,
      "loss": 0.025779077783226967,
      "step": 3820
    },
    {
      "epoch": 0.724412433661865,
      "grad_norm": 0.26622989773750305,
      "learning_rate": 0.00015913685847589426,
      "loss": 0.02996867522597313,
      "step": 3822
    },
    {
      "epoch": 0.7247915087187263,
      "grad_norm": 0.43218162655830383,
      "learning_rate": 0.00015905909797822706,
      "loss": 0.029675763100385666,
      "step": 3824
    },
    {
      "epoch": 0.7251705837755875,
      "grad_norm": 0.11112133413553238,
      "learning_rate": 0.00015898133748055987,
      "loss": 0.013373862951993942,
      "step": 3826
    },
    {
      "epoch": 0.7255496588324488,
      "grad_norm": 0.39504167437553406,
      "learning_rate": 0.00015890357698289268,
      "loss": 0.03539038449525833,
      "step": 3828
    },
    {
      "epoch": 0.7259287338893101,
      "grad_norm": 0.1472347229719162,
      "learning_rate": 0.0001588258164852255,
      "loss": 0.030102940276265144,
      "step": 3830
    },
    {
      "epoch": 0.7263078089461713,
      "grad_norm": 0.3436129689216614,
      "learning_rate": 0.00015874805598755832,
      "loss": 0.040433868765830994,
      "step": 3832
    },
    {
      "epoch": 0.7266868840030326,
      "grad_norm": 0.3201497197151184,
      "learning_rate": 0.00015867029548989115,
      "loss": 0.02126275934278965,
      "step": 3834
    },
    {
      "epoch": 0.7270659590598939,
      "grad_norm": 0.12892794609069824,
      "learning_rate": 0.00015859253499222396,
      "loss": 0.014339407905936241,
      "step": 3836
    },
    {
      "epoch": 0.7274450341167551,
      "grad_norm": 0.3839438259601593,
      "learning_rate": 0.0001585147744945568,
      "loss": 0.02723880112171173,
      "step": 3838
    },
    {
      "epoch": 0.7278241091736164,
      "grad_norm": 0.2957480251789093,
      "learning_rate": 0.0001584370139968896,
      "loss": 0.02294890396296978,
      "step": 3840
    },
    {
      "epoch": 0.7282031842304776,
      "grad_norm": 0.3064419627189636,
      "learning_rate": 0.0001583592534992224,
      "loss": 0.01813615672290325,
      "step": 3842
    },
    {
      "epoch": 0.7285822592873389,
      "grad_norm": 0.38042503595352173,
      "learning_rate": 0.0001582814930015552,
      "loss": 0.027306433767080307,
      "step": 3844
    },
    {
      "epoch": 0.7289613343442002,
      "grad_norm": 0.11692070960998535,
      "learning_rate": 0.00015820373250388802,
      "loss": 0.022313255816698074,
      "step": 3846
    },
    {
      "epoch": 0.7293404094010614,
      "grad_norm": 0.10903311520814896,
      "learning_rate": 0.00015812597200622085,
      "loss": 0.014873048290610313,
      "step": 3848
    },
    {
      "epoch": 0.7297194844579227,
      "grad_norm": 0.3715738356113434,
      "learning_rate": 0.00015804821150855366,
      "loss": 0.04430793225765228,
      "step": 3850
    },
    {
      "epoch": 0.730098559514784,
      "grad_norm": 0.11875716596841812,
      "learning_rate": 0.00015797045101088647,
      "loss": 0.028359418734908104,
      "step": 3852
    },
    {
      "epoch": 0.7304776345716452,
      "grad_norm": 0.13353659212589264,
      "learning_rate": 0.00015789269051321927,
      "loss": 0.025622282177209854,
      "step": 3854
    },
    {
      "epoch": 0.7308567096285065,
      "grad_norm": 0.32123708724975586,
      "learning_rate": 0.0001578149300155521,
      "loss": 0.028729157522320747,
      "step": 3856
    },
    {
      "epoch": 0.7312357846853677,
      "grad_norm": 0.1547655165195465,
      "learning_rate": 0.00015773716951788494,
      "loss": 0.013496311381459236,
      "step": 3858
    },
    {
      "epoch": 0.731614859742229,
      "grad_norm": 0.32637232542037964,
      "learning_rate": 0.00015765940902021775,
      "loss": 0.029955901205539703,
      "step": 3860
    },
    {
      "epoch": 0.7319939347990903,
      "grad_norm": 0.09802603721618652,
      "learning_rate": 0.00015758164852255056,
      "loss": 0.02087332494556904,
      "step": 3862
    },
    {
      "epoch": 0.7323730098559514,
      "grad_norm": 0.3801679015159607,
      "learning_rate": 0.00015750388802488336,
      "loss": 0.024995850399136543,
      "step": 3864
    },
    {
      "epoch": 0.7327520849128127,
      "grad_norm": 0.127773255109787,
      "learning_rate": 0.00015742612752721617,
      "loss": 0.013783286325633526,
      "step": 3866
    },
    {
      "epoch": 0.733131159969674,
      "grad_norm": 0.31125953793525696,
      "learning_rate": 0.000157348367029549,
      "loss": 0.04006776958703995,
      "step": 3868
    },
    {
      "epoch": 0.7335102350265352,
      "grad_norm": 0.4542815089225769,
      "learning_rate": 0.0001572706065318818,
      "loss": 0.0350923128426075,
      "step": 3870
    },
    {
      "epoch": 0.7338893100833965,
      "grad_norm": 0.4068615734577179,
      "learning_rate": 0.00015719284603421462,
      "loss": 0.04839121177792549,
      "step": 3872
    },
    {
      "epoch": 0.7342683851402577,
      "grad_norm": 0.39573371410369873,
      "learning_rate": 0.00015711508553654743,
      "loss": 0.04728904739022255,
      "step": 3874
    },
    {
      "epoch": 0.734647460197119,
      "grad_norm": 0.27655676007270813,
      "learning_rate": 0.00015703732503888023,
      "loss": 0.02139093726873398,
      "step": 3876
    },
    {
      "epoch": 0.7350265352539803,
      "grad_norm": 0.3649783730506897,
      "learning_rate": 0.00015695956454121307,
      "loss": 0.027860263362526894,
      "step": 3878
    },
    {
      "epoch": 0.7354056103108415,
      "grad_norm": 0.10016981512308121,
      "learning_rate": 0.0001568818040435459,
      "loss": 0.019468888640403748,
      "step": 3880
    },
    {
      "epoch": 0.7357846853677028,
      "grad_norm": 0.4380815029144287,
      "learning_rate": 0.0001568040435458787,
      "loss": 0.040987540036439896,
      "step": 3882
    },
    {
      "epoch": 0.7361637604245641,
      "grad_norm": 0.24645602703094482,
      "learning_rate": 0.00015672628304821151,
      "loss": 0.019828470423817635,
      "step": 3884
    },
    {
      "epoch": 0.7365428354814253,
      "grad_norm": 0.09301592409610748,
      "learning_rate": 0.00015664852255054435,
      "loss": 0.013527280651032925,
      "step": 3886
    },
    {
      "epoch": 0.7369219105382866,
      "grad_norm": 0.08736240863800049,
      "learning_rate": 0.00015657076205287715,
      "loss": 0.021154891699552536,
      "step": 3888
    },
    {
      "epoch": 0.7373009855951478,
      "grad_norm": 0.32928285002708435,
      "learning_rate": 0.00015649300155520996,
      "loss": 0.025198573246598244,
      "step": 3890
    },
    {
      "epoch": 0.7376800606520091,
      "grad_norm": 0.352622389793396,
      "learning_rate": 0.00015641524105754277,
      "loss": 0.04133247211575508,
      "step": 3892
    },
    {
      "epoch": 0.7380591357088704,
      "grad_norm": 0.34188640117645264,
      "learning_rate": 0.00015633748055987558,
      "loss": 0.040303293615579605,
      "step": 3894
    },
    {
      "epoch": 0.7384382107657316,
      "grad_norm": 0.30073443055152893,
      "learning_rate": 0.0001562597200622084,
      "loss": 0.02087089791893959,
      "step": 3896
    },
    {
      "epoch": 0.7388172858225929,
      "grad_norm": 0.10108694434165955,
      "learning_rate": 0.00015618195956454122,
      "loss": 0.017457904294133186,
      "step": 3898
    },
    {
      "epoch": 0.7391963608794542,
      "grad_norm": 0.19065077602863312,
      "learning_rate": 0.00015610419906687402,
      "loss": 0.021667150780558586,
      "step": 3900
    },
    {
      "epoch": 0.7395754359363154,
      "grad_norm": 0.38414692878723145,
      "learning_rate": 0.00015602643856920686,
      "loss": 0.03917527571320534,
      "step": 3902
    },
    {
      "epoch": 0.7399545109931767,
      "grad_norm": 0.18734820187091827,
      "learning_rate": 0.00015594867807153966,
      "loss": 0.023370904847979546,
      "step": 3904
    },
    {
      "epoch": 0.740333586050038,
      "grad_norm": 0.2414512038230896,
      "learning_rate": 0.0001558709175738725,
      "loss": 0.031007058918476105,
      "step": 3906
    },
    {
      "epoch": 0.7407126611068992,
      "grad_norm": 0.28416913747787476,
      "learning_rate": 0.0001557931570762053,
      "loss": 0.022831708192825317,
      "step": 3908
    },
    {
      "epoch": 0.7410917361637605,
      "grad_norm": 0.13254405558109283,
      "learning_rate": 0.0001557153965785381,
      "loss": 0.021845471113920212,
      "step": 3910
    },
    {
      "epoch": 0.7414708112206216,
      "grad_norm": 0.1074175089597702,
      "learning_rate": 0.00015563763608087092,
      "loss": 0.018406201153993607,
      "step": 3912
    },
    {
      "epoch": 0.7418498862774829,
      "grad_norm": 0.41257408261299133,
      "learning_rate": 0.00015555987558320373,
      "loss": 0.03959622234106064,
      "step": 3914
    },
    {
      "epoch": 0.7422289613343442,
      "grad_norm": 0.07631173729896545,
      "learning_rate": 0.00015548211508553656,
      "loss": 0.017356498166918755,
      "step": 3916
    },
    {
      "epoch": 0.7426080363912054,
      "grad_norm": 0.0949835479259491,
      "learning_rate": 0.00015540435458786937,
      "loss": 0.01744546927511692,
      "step": 3918
    },
    {
      "epoch": 0.7429871114480667,
      "grad_norm": 0.33840835094451904,
      "learning_rate": 0.00015532659409020217,
      "loss": 0.02010497823357582,
      "step": 3920
    },
    {
      "epoch": 0.743366186504928,
      "grad_norm": 0.4543622136116028,
      "learning_rate": 0.00015524883359253498,
      "loss": 0.03465965762734413,
      "step": 3922
    },
    {
      "epoch": 0.7437452615617892,
      "grad_norm": 0.4029660224914551,
      "learning_rate": 0.00015517107309486781,
      "loss": 0.025007862597703934,
      "step": 3924
    },
    {
      "epoch": 0.7441243366186505,
      "grad_norm": 0.10410217195749283,
      "learning_rate": 0.00015509331259720065,
      "loss": 0.019747868180274963,
      "step": 3926
    },
    {
      "epoch": 0.7445034116755117,
      "grad_norm": 0.3705401122570038,
      "learning_rate": 0.00015501555209953345,
      "loss": 0.028779810294508934,
      "step": 3928
    },
    {
      "epoch": 0.744882486732373,
      "grad_norm": 0.17581459879875183,
      "learning_rate": 0.00015493779160186626,
      "loss": 0.021180234849452972,
      "step": 3930
    },
    {
      "epoch": 0.7452615617892343,
      "grad_norm": 0.31276312470436096,
      "learning_rate": 0.00015486003110419907,
      "loss": 0.018826093524694443,
      "step": 3932
    },
    {
      "epoch": 0.7456406368460955,
      "grad_norm": 0.09388747811317444,
      "learning_rate": 0.0001547822706065319,
      "loss": 0.013018107041716576,
      "step": 3934
    },
    {
      "epoch": 0.7460197119029568,
      "grad_norm": 0.3735581338405609,
      "learning_rate": 0.0001547045101088647,
      "loss": 0.0358201302587986,
      "step": 3936
    },
    {
      "epoch": 0.7463987869598181,
      "grad_norm": 0.515340268611908,
      "learning_rate": 0.00015462674961119752,
      "loss": 0.04731396958231926,
      "step": 3938
    },
    {
      "epoch": 0.7467778620166793,
      "grad_norm": 0.13163183629512787,
      "learning_rate": 0.00015454898911353032,
      "loss": 0.03286414593458176,
      "step": 3940
    },
    {
      "epoch": 0.7471569370735406,
      "grad_norm": 0.37440189719200134,
      "learning_rate": 0.00015447122861586313,
      "loss": 0.02211233787238598,
      "step": 3942
    },
    {
      "epoch": 0.7475360121304018,
      "grad_norm": 0.24613113701343536,
      "learning_rate": 0.00015439346811819596,
      "loss": 0.014796094037592411,
      "step": 3944
    },
    {
      "epoch": 0.7479150871872631,
      "grad_norm": 0.3150129020214081,
      "learning_rate": 0.00015431570762052877,
      "loss": 0.026067500934004784,
      "step": 3946
    },
    {
      "epoch": 0.7482941622441244,
      "grad_norm": 0.12359709292650223,
      "learning_rate": 0.0001542379471228616,
      "loss": 0.01673927530646324,
      "step": 3948
    },
    {
      "epoch": 0.7486732373009856,
      "grad_norm": 0.3149224519729614,
      "learning_rate": 0.0001541601866251944,
      "loss": 0.0294203981757164,
      "step": 3950
    },
    {
      "epoch": 0.7490523123578469,
      "grad_norm": 0.12521107494831085,
      "learning_rate": 0.00015408242612752722,
      "loss": 0.013356301002204418,
      "step": 3952
    },
    {
      "epoch": 0.7494313874147082,
      "grad_norm": 0.15365296602249146,
      "learning_rate": 0.00015400466562986005,
      "loss": 0.01536143384873867,
      "step": 3954
    },
    {
      "epoch": 0.7498104624715694,
      "grad_norm": 0.4804651141166687,
      "learning_rate": 0.00015392690513219286,
      "loss": 0.03612521290779114,
      "step": 3956
    },
    {
      "epoch": 0.7501895375284306,
      "grad_norm": 0.11883759498596191,
      "learning_rate": 0.00015384914463452567,
      "loss": 0.020559778437018394,
      "step": 3958
    },
    {
      "epoch": 0.7505686125852918,
      "grad_norm": 0.09018052369356155,
      "learning_rate": 0.00015377138413685847,
      "loss": 0.01731124147772789,
      "step": 3960
    },
    {
      "epoch": 0.7509476876421531,
      "grad_norm": 0.09127313643693924,
      "learning_rate": 0.0001536936236391913,
      "loss": 0.011628419160842896,
      "step": 3962
    },
    {
      "epoch": 0.7513267626990144,
      "grad_norm": 0.26581165194511414,
      "learning_rate": 0.00015361586314152411,
      "loss": 0.019836803898215294,
      "step": 3964
    },
    {
      "epoch": 0.7517058377558756,
      "grad_norm": 0.16984854638576508,
      "learning_rate": 0.00015353810264385692,
      "loss": 0.020157037302851677,
      "step": 3966
    },
    {
      "epoch": 0.7520849128127369,
      "grad_norm": 0.16032937169075012,
      "learning_rate": 0.00015346034214618973,
      "loss": 0.02272959053516388,
      "step": 3968
    },
    {
      "epoch": 0.7524639878695982,
      "grad_norm": 0.23603838682174683,
      "learning_rate": 0.00015338258164852256,
      "loss": 0.0195081178098917,
      "step": 3970
    },
    {
      "epoch": 0.7528430629264594,
      "grad_norm": 0.3147512972354889,
      "learning_rate": 0.00015330482115085537,
      "loss": 0.03443726897239685,
      "step": 3972
    },
    {
      "epoch": 0.7532221379833207,
      "grad_norm": 0.40106621384620667,
      "learning_rate": 0.0001532270606531882,
      "loss": 0.02362012304365635,
      "step": 3974
    },
    {
      "epoch": 0.7536012130401819,
      "grad_norm": 0.2806083858013153,
      "learning_rate": 0.000153149300155521,
      "loss": 0.025656728073954582,
      "step": 3976
    },
    {
      "epoch": 0.7539802880970432,
      "grad_norm": 0.48330816626548767,
      "learning_rate": 0.00015307153965785382,
      "loss": 0.03327082470059395,
      "step": 3978
    },
    {
      "epoch": 0.7543593631539045,
      "grad_norm": 0.09118037670850754,
      "learning_rate": 0.00015299377916018662,
      "loss": 0.018486827611923218,
      "step": 3980
    },
    {
      "epoch": 0.7547384382107657,
      "grad_norm": 0.10719195008277893,
      "learning_rate": 0.00015291601866251946,
      "loss": 0.016303857788443565,
      "step": 3982
    },
    {
      "epoch": 0.755117513267627,
      "grad_norm": 0.18872256577014923,
      "learning_rate": 0.00015283825816485226,
      "loss": 0.02621084451675415,
      "step": 3984
    },
    {
      "epoch": 0.7554965883244883,
      "grad_norm": 0.4830758571624756,
      "learning_rate": 0.00015276049766718507,
      "loss": 0.03137216344475746,
      "step": 3986
    },
    {
      "epoch": 0.7558756633813495,
      "grad_norm": 0.4428825378417969,
      "learning_rate": 0.00015268273716951788,
      "loss": 0.041971299797296524,
      "step": 3988
    },
    {
      "epoch": 0.7562547384382108,
      "grad_norm": 0.24404580891132355,
      "learning_rate": 0.00015260497667185069,
      "loss": 0.02492944523692131,
      "step": 3990
    },
    {
      "epoch": 0.756633813495072,
      "grad_norm": 0.2067522257566452,
      "learning_rate": 0.00015252721617418352,
      "loss": 0.023579254746437073,
      "step": 3992
    },
    {
      "epoch": 0.7570128885519333,
      "grad_norm": 0.20602256059646606,
      "learning_rate": 0.00015244945567651633,
      "loss": 0.017534788697957993,
      "step": 3994
    },
    {
      "epoch": 0.7573919636087946,
      "grad_norm": 0.26311540603637695,
      "learning_rate": 0.00015237169517884916,
      "loss": 0.027192916721105576,
      "step": 3996
    },
    {
      "epoch": 0.7577710386656558,
      "grad_norm": 0.08048859238624573,
      "learning_rate": 0.00015229393468118197,
      "loss": 0.03735966980457306,
      "step": 3998
    },
    {
      "epoch": 0.7581501137225171,
      "grad_norm": 0.441552996635437,
      "learning_rate": 0.0001522161741835148,
      "loss": 0.04623119905591011,
      "step": 4000
    },
    {
      "epoch": 0.7585291887793784,
      "grad_norm": 0.354604572057724,
      "learning_rate": 0.0001521384136858476,
      "loss": 0.044628482311964035,
      "step": 4002
    },
    {
      "epoch": 0.7589082638362395,
      "grad_norm": 0.09432460367679596,
      "learning_rate": 0.00015206065318818041,
      "loss": 0.028055187314748764,
      "step": 4004
    },
    {
      "epoch": 0.7592873388931008,
      "grad_norm": 0.12522973120212555,
      "learning_rate": 0.00015198289269051322,
      "loss": 0.0262571033090353,
      "step": 4006
    },
    {
      "epoch": 0.759666413949962,
      "grad_norm": 0.2729003131389618,
      "learning_rate": 0.00015190513219284603,
      "loss": 0.033609434962272644,
      "step": 4008
    },
    {
      "epoch": 0.7600454890068233,
      "grad_norm": 0.09898972511291504,
      "learning_rate": 0.00015182737169517886,
      "loss": 0.011849031783640385,
      "step": 4010
    },
    {
      "epoch": 0.7604245640636846,
      "grad_norm": 0.3998548686504364,
      "learning_rate": 0.00015174961119751167,
      "loss": 0.03685874491930008,
      "step": 4012
    },
    {
      "epoch": 0.7608036391205458,
      "grad_norm": 0.1699802428483963,
      "learning_rate": 0.00015167185069984448,
      "loss": 0.015329774469137192,
      "step": 4014
    },
    {
      "epoch": 0.7611827141774071,
      "grad_norm": 0.3757225573062897,
      "learning_rate": 0.0001515940902021773,
      "loss": 0.0249441210180521,
      "step": 4016
    },
    {
      "epoch": 0.7615617892342684,
      "grad_norm": 0.3648105561733246,
      "learning_rate": 0.00015151632970451012,
      "loss": 0.024121036753058434,
      "step": 4018
    },
    {
      "epoch": 0.7619408642911296,
      "grad_norm": 0.12334334850311279,
      "learning_rate": 0.00015143856920684295,
      "loss": 0.014487909153103828,
      "step": 4020
    },
    {
      "epoch": 0.7623199393479909,
      "grad_norm": 0.35869139432907104,
      "learning_rate": 0.00015136080870917576,
      "loss": 0.03468037024140358,
      "step": 4022
    },
    {
      "epoch": 0.7626990144048522,
      "grad_norm": 0.4787271320819855,
      "learning_rate": 0.00015128304821150856,
      "loss": 0.03561337664723396,
      "step": 4024
    },
    {
      "epoch": 0.7630780894617134,
      "grad_norm": 0.2975921630859375,
      "learning_rate": 0.00015120528771384137,
      "loss": 0.024220174178481102,
      "step": 4026
    },
    {
      "epoch": 0.7634571645185747,
      "grad_norm": 0.3175036907196045,
      "learning_rate": 0.00015112752721617418,
      "loss": 0.03781888633966446,
      "step": 4028
    },
    {
      "epoch": 0.7638362395754359,
      "grad_norm": 0.1638782024383545,
      "learning_rate": 0.000151049766718507,
      "loss": 0.02941550500690937,
      "step": 4030
    },
    {
      "epoch": 0.7642153146322972,
      "grad_norm": 0.1029331162571907,
      "learning_rate": 0.00015097200622083982,
      "loss": 0.01872831955552101,
      "step": 4032
    },
    {
      "epoch": 0.7645943896891585,
      "grad_norm": 0.24669474363327026,
      "learning_rate": 0.00015089424572317263,
      "loss": 0.017219258472323418,
      "step": 4034
    },
    {
      "epoch": 0.7649734647460197,
      "grad_norm": 0.344036728143692,
      "learning_rate": 0.00015081648522550543,
      "loss": 0.028363095596432686,
      "step": 4036
    },
    {
      "epoch": 0.765352539802881,
      "grad_norm": 0.08787023276090622,
      "learning_rate": 0.00015073872472783827,
      "loss": 0.02914881706237793,
      "step": 4038
    },
    {
      "epoch": 0.7657316148597423,
      "grad_norm": 0.1866435706615448,
      "learning_rate": 0.00015066096423017107,
      "loss": 0.015346669591963291,
      "step": 4040
    },
    {
      "epoch": 0.7661106899166035,
      "grad_norm": 0.3040603995323181,
      "learning_rate": 0.0001505832037325039,
      "loss": 0.024850239977240562,
      "step": 4042
    },
    {
      "epoch": 0.7664897649734648,
      "grad_norm": 0.3039913773536682,
      "learning_rate": 0.00015050544323483671,
      "loss": 0.04210745915770531,
      "step": 4044
    },
    {
      "epoch": 0.766868840030326,
      "grad_norm": 0.14281949400901794,
      "learning_rate": 0.00015042768273716952,
      "loss": 0.022492632269859314,
      "step": 4046
    },
    {
      "epoch": 0.7672479150871873,
      "grad_norm": 0.3758687376976013,
      "learning_rate": 0.00015034992223950236,
      "loss": 0.02488720417022705,
      "step": 4048
    },
    {
      "epoch": 0.7676269901440486,
      "grad_norm": 0.09285403043031693,
      "learning_rate": 0.00015027216174183516,
      "loss": 0.028162294998764992,
      "step": 4050
    },
    {
      "epoch": 0.7680060652009097,
      "grad_norm": 0.23906932771205902,
      "learning_rate": 0.00015019440124416797,
      "loss": 0.018858008086681366,
      "step": 4052
    },
    {
      "epoch": 0.768385140257771,
      "grad_norm": 0.5127546787261963,
      "learning_rate": 0.00015011664074650078,
      "loss": 0.03426947817206383,
      "step": 4054
    },
    {
      "epoch": 0.7687642153146323,
      "grad_norm": 0.1041359007358551,
      "learning_rate": 0.00015003888024883358,
      "loss": 0.0315568707883358,
      "step": 4056
    },
    {
      "epoch": 0.7691432903714935,
      "grad_norm": 0.13705438375473022,
      "learning_rate": 0.00014996111975116642,
      "loss": 0.025585979223251343,
      "step": 4058
    },
    {
      "epoch": 0.7695223654283548,
      "grad_norm": 0.17112568020820618,
      "learning_rate": 0.00014988335925349922,
      "loss": 0.02050529047846794,
      "step": 4060
    },
    {
      "epoch": 0.769901440485216,
      "grad_norm": 0.08994106203317642,
      "learning_rate": 0.00014980559875583203,
      "loss": 0.016972316429018974,
      "step": 4062
    },
    {
      "epoch": 0.7702805155420773,
      "grad_norm": 0.3451199531555176,
      "learning_rate": 0.00014972783825816487,
      "loss": 0.02240954153239727,
      "step": 4064
    },
    {
      "epoch": 0.7706595905989386,
      "grad_norm": 0.23435987532138824,
      "learning_rate": 0.00014965007776049767,
      "loss": 0.015702374279499054,
      "step": 4066
    },
    {
      "epoch": 0.7710386656557998,
      "grad_norm": 0.11549916118383408,
      "learning_rate": 0.0001495723172628305,
      "loss": 0.035111721605062485,
      "step": 4068
    },
    {
      "epoch": 0.7714177407126611,
      "grad_norm": 0.36172544956207275,
      "learning_rate": 0.0001494945567651633,
      "loss": 0.03800908103585243,
      "step": 4070
    },
    {
      "epoch": 0.7717968157695224,
      "grad_norm": 0.16232416033744812,
      "learning_rate": 0.00014941679626749612,
      "loss": 0.01726338639855385,
      "step": 4072
    },
    {
      "epoch": 0.7721758908263836,
      "grad_norm": 1.0883269309997559,
      "learning_rate": 0.00014933903576982893,
      "loss": 0.023416833952069283,
      "step": 4074
    },
    {
      "epoch": 0.7725549658832449,
      "grad_norm": 0.116034135222435,
      "learning_rate": 0.00014926127527216173,
      "loss": 0.027708731591701508,
      "step": 4076
    },
    {
      "epoch": 0.7729340409401061,
      "grad_norm": 0.087105393409729,
      "learning_rate": 0.00014918351477449457,
      "loss": 0.037236109375953674,
      "step": 4078
    },
    {
      "epoch": 0.7733131159969674,
      "grad_norm": 0.2927720248699188,
      "learning_rate": 0.00014910575427682737,
      "loss": 0.017965523526072502,
      "step": 4080
    },
    {
      "epoch": 0.7736921910538287,
      "grad_norm": 0.12538392841815948,
      "learning_rate": 0.00014902799377916018,
      "loss": 0.018100906163454056,
      "step": 4082
    },
    {
      "epoch": 0.7740712661106899,
      "grad_norm": 0.4409235417842865,
      "learning_rate": 0.000148950233281493,
      "loss": 0.03334690257906914,
      "step": 4084
    },
    {
      "epoch": 0.7744503411675512,
      "grad_norm": 0.33477553725242615,
      "learning_rate": 0.00014887247278382582,
      "loss": 0.025339730083942413,
      "step": 4086
    },
    {
      "epoch": 0.7748294162244125,
      "grad_norm": 0.10463062673807144,
      "learning_rate": 0.00014879471228615866,
      "loss": 0.013993374072015285,
      "step": 4088
    },
    {
      "epoch": 0.7752084912812737,
      "grad_norm": 0.09666959196329117,
      "learning_rate": 0.00014871695178849146,
      "loss": 0.022186772897839546,
      "step": 4090
    },
    {
      "epoch": 0.775587566338135,
      "grad_norm": 0.1312924027442932,
      "learning_rate": 0.00014863919129082427,
      "loss": 0.013046500273048878,
      "step": 4092
    },
    {
      "epoch": 0.7759666413949962,
      "grad_norm": 0.5004321336746216,
      "learning_rate": 0.00014856143079315708,
      "loss": 0.0409611351788044,
      "step": 4094
    },
    {
      "epoch": 0.7763457164518575,
      "grad_norm": 0.11120400577783585,
      "learning_rate": 0.0001484836702954899,
      "loss": 0.019754944369196892,
      "step": 4096
    },
    {
      "epoch": 0.7767247915087188,
      "grad_norm": 0.0908280462026596,
      "learning_rate": 0.00014840590979782272,
      "loss": 0.028069013729691505,
      "step": 4098
    },
    {
      "epoch": 0.77710386656558,
      "grad_norm": 0.11773878335952759,
      "learning_rate": 0.00014832814930015552,
      "loss": 0.022659875452518463,
      "step": 4100
    },
    {
      "epoch": 0.7774829416224412,
      "grad_norm": 0.08691461384296417,
      "learning_rate": 0.00014825038880248833,
      "loss": 0.017922570928931236,
      "step": 4102
    },
    {
      "epoch": 0.7778620166793025,
      "grad_norm": 0.10607599467039108,
      "learning_rate": 0.00014817262830482114,
      "loss": 0.03566766530275345,
      "step": 4104
    },
    {
      "epoch": 0.7782410917361637,
      "grad_norm": 0.2818552255630493,
      "learning_rate": 0.00014809486780715397,
      "loss": 0.019274527207016945,
      "step": 4106
    },
    {
      "epoch": 0.778620166793025,
      "grad_norm": 0.15609435737133026,
      "learning_rate": 0.00014801710730948678,
      "loss": 0.024681299924850464,
      "step": 4108
    },
    {
      "epoch": 0.7789992418498862,
      "grad_norm": 0.45389267802238464,
      "learning_rate": 0.0001479393468118196,
      "loss": 0.03439616411924362,
      "step": 4110
    },
    {
      "epoch": 0.7793783169067475,
      "grad_norm": 0.3728964030742645,
      "learning_rate": 0.00014786158631415242,
      "loss": 0.026823285967111588,
      "step": 4112
    },
    {
      "epoch": 0.7797573919636088,
      "grad_norm": 0.24034075438976288,
      "learning_rate": 0.00014778382581648523,
      "loss": 0.022101012989878654,
      "step": 4114
    },
    {
      "epoch": 0.78013646702047,
      "grad_norm": 0.09134727716445923,
      "learning_rate": 0.00014770606531881806,
      "loss": 0.04422621428966522,
      "step": 4116
    },
    {
      "epoch": 0.7805155420773313,
      "grad_norm": 0.10632926225662231,
      "learning_rate": 0.00014762830482115087,
      "loss": 0.012636770494282246,
      "step": 4118
    },
    {
      "epoch": 0.7808946171341926,
      "grad_norm": 0.1037563756108284,
      "learning_rate": 0.00014755054432348367,
      "loss": 0.014507236890494823,
      "step": 4120
    },
    {
      "epoch": 0.7812736921910538,
      "grad_norm": 0.12498272955417633,
      "learning_rate": 0.00014747278382581648,
      "loss": 0.028698265552520752,
      "step": 4122
    },
    {
      "epoch": 0.7816527672479151,
      "grad_norm": 0.1374005824327469,
      "learning_rate": 0.00014739502332814932,
      "loss": 0.012324733659625053,
      "step": 4124
    },
    {
      "epoch": 0.7820318423047763,
      "grad_norm": 0.23156264424324036,
      "learning_rate": 0.00014731726283048212,
      "loss": 0.0315374918282032,
      "step": 4126
    },
    {
      "epoch": 0.7824109173616376,
      "grad_norm": 0.4028889536857605,
      "learning_rate": 0.00014723950233281493,
      "loss": 0.02604810893535614,
      "step": 4128
    },
    {
      "epoch": 0.7827899924184989,
      "grad_norm": 0.3824390470981598,
      "learning_rate": 0.00014716174183514774,
      "loss": 0.027330825105309486,
      "step": 4130
    },
    {
      "epoch": 0.7831690674753601,
      "grad_norm": 0.11424584686756134,
      "learning_rate": 0.00014708398133748057,
      "loss": 0.01387232169508934,
      "step": 4132
    },
    {
      "epoch": 0.7835481425322214,
      "grad_norm": 0.2862674295902252,
      "learning_rate": 0.0001470062208398134,
      "loss": 0.021760625764727592,
      "step": 4134
    },
    {
      "epoch": 0.7839272175890827,
      "grad_norm": 0.15307672321796417,
      "learning_rate": 0.0001469284603421462,
      "loss": 0.022060051560401917,
      "step": 4136
    },
    {
      "epoch": 0.7843062926459439,
      "grad_norm": 0.3839503526687622,
      "learning_rate": 0.00014685069984447902,
      "loss": 0.027501288801431656,
      "step": 4138
    },
    {
      "epoch": 0.7846853677028052,
      "grad_norm": 0.09439737349748611,
      "learning_rate": 0.00014677293934681182,
      "loss": 0.026191677898168564,
      "step": 4140
    },
    {
      "epoch": 0.7850644427596664,
      "grad_norm": 0.3293452560901642,
      "learning_rate": 0.00014669517884914463,
      "loss": 0.023145576938986778,
      "step": 4142
    },
    {
      "epoch": 0.7854435178165277,
      "grad_norm": 0.10310211032629013,
      "learning_rate": 0.00014661741835147747,
      "loss": 0.02213893085718155,
      "step": 4144
    },
    {
      "epoch": 0.785822592873389,
      "grad_norm": 0.40012338757514954,
      "learning_rate": 0.00014653965785381027,
      "loss": 0.03726082295179367,
      "step": 4146
    },
    {
      "epoch": 0.7862016679302501,
      "grad_norm": 0.09514353424310684,
      "learning_rate": 0.00014646189735614308,
      "loss": 0.014966962859034538,
      "step": 4148
    },
    {
      "epoch": 0.7865807429871114,
      "grad_norm": 0.24874810874462128,
      "learning_rate": 0.0001463841368584759,
      "loss": 0.019291367381811142,
      "step": 4150
    },
    {
      "epoch": 0.7869598180439727,
      "grad_norm": 0.2617069184780121,
      "learning_rate": 0.0001463063763608087,
      "loss": 0.018816612660884857,
      "step": 4152
    },
    {
      "epoch": 0.7873388931008339,
      "grad_norm": 0.08434536308050156,
      "learning_rate": 0.00014622861586314153,
      "loss": 0.01286565326154232,
      "step": 4154
    },
    {
      "epoch": 0.7877179681576952,
      "grad_norm": 0.08556289225816727,
      "learning_rate": 0.00014615085536547436,
      "loss": 0.02261766791343689,
      "step": 4156
    },
    {
      "epoch": 0.7880970432145565,
      "grad_norm": 0.10751739144325256,
      "learning_rate": 0.00014607309486780717,
      "loss": 0.02484571747481823,
      "step": 4158
    },
    {
      "epoch": 0.7884761182714177,
      "grad_norm": 0.4505607783794403,
      "learning_rate": 0.00014599533437013998,
      "loss": 0.03476109355688095,
      "step": 4160
    },
    {
      "epoch": 0.788855193328279,
      "grad_norm": 0.10992085188627243,
      "learning_rate": 0.0001459175738724728,
      "loss": 0.02764667011797428,
      "step": 4162
    },
    {
      "epoch": 0.7892342683851402,
      "grad_norm": 0.10657194256782532,
      "learning_rate": 0.00014583981337480562,
      "loss": 0.018599964678287506,
      "step": 4164
    },
    {
      "epoch": 0.7896133434420015,
      "grad_norm": 0.11758055537939072,
      "learning_rate": 0.00014576205287713842,
      "loss": 0.022491328418254852,
      "step": 4166
    },
    {
      "epoch": 0.7899924184988628,
      "grad_norm": 0.3799620270729065,
      "learning_rate": 0.00014568429237947123,
      "loss": 0.02689218707382679,
      "step": 4168
    },
    {
      "epoch": 0.790371493555724,
      "grad_norm": 0.09631983190774918,
      "learning_rate": 0.00014560653188180404,
      "loss": 0.027401652187108994,
      "step": 4170
    },
    {
      "epoch": 0.7907505686125853,
      "grad_norm": 0.12333094328641891,
      "learning_rate": 0.00014552877138413687,
      "loss": 0.02578526735305786,
      "step": 4172
    },
    {
      "epoch": 0.7911296436694466,
      "grad_norm": 0.4027808606624603,
      "learning_rate": 0.00014545101088646968,
      "loss": 0.030051114037632942,
      "step": 4174
    },
    {
      "epoch": 0.7915087187263078,
      "grad_norm": 0.0913926213979721,
      "learning_rate": 0.00014537325038880248,
      "loss": 0.029553785920143127,
      "step": 4176
    },
    {
      "epoch": 0.7918877937831691,
      "grad_norm": 0.33775919675827026,
      "learning_rate": 0.00014529548989113532,
      "loss": 0.020305240526795387,
      "step": 4178
    },
    {
      "epoch": 0.7922668688400303,
      "grad_norm": 0.269371896982193,
      "learning_rate": 0.00014521772939346813,
      "loss": 0.030247190967202187,
      "step": 4180
    },
    {
      "epoch": 0.7926459438968916,
      "grad_norm": 0.1176525428891182,
      "learning_rate": 0.00014513996889580096,
      "loss": 0.02377990446984768,
      "step": 4182
    },
    {
      "epoch": 0.7930250189537529,
      "grad_norm": 0.18686066567897797,
      "learning_rate": 0.00014506220839813377,
      "loss": 0.017997074872255325,
      "step": 4184
    },
    {
      "epoch": 0.7934040940106141,
      "grad_norm": 0.3788314163684845,
      "learning_rate": 0.00014498444790046657,
      "loss": 0.02533004991710186,
      "step": 4186
    },
    {
      "epoch": 0.7937831690674754,
      "grad_norm": 0.10860533267259598,
      "learning_rate": 0.00014490668740279938,
      "loss": 0.013837635517120361,
      "step": 4188
    },
    {
      "epoch": 0.7941622441243367,
      "grad_norm": 0.11729373782873154,
      "learning_rate": 0.0001448289269051322,
      "loss": 0.013233192265033722,
      "step": 4190
    },
    {
      "epoch": 0.7945413191811979,
      "grad_norm": 0.4010216295719147,
      "learning_rate": 0.00014475116640746502,
      "loss": 0.032141704112291336,
      "step": 4192
    },
    {
      "epoch": 0.7949203942380592,
      "grad_norm": 0.14376893639564514,
      "learning_rate": 0.00014467340590979783,
      "loss": 0.020979242399334908,
      "step": 4194
    },
    {
      "epoch": 0.7952994692949203,
      "grad_norm": 0.3479453921318054,
      "learning_rate": 0.00014459564541213063,
      "loss": 0.02591726928949356,
      "step": 4196
    },
    {
      "epoch": 0.7956785443517816,
      "grad_norm": 0.3379877507686615,
      "learning_rate": 0.00014451788491446344,
      "loss": 0.037894412875175476,
      "step": 4198
    },
    {
      "epoch": 0.796057619408643,
      "grad_norm": 0.10460545867681503,
      "learning_rate": 0.00014444012441679628,
      "loss": 0.015401800163090229,
      "step": 4200
    },
    {
      "epoch": 0.7964366944655041,
      "grad_norm": 0.49596214294433594,
      "learning_rate": 0.00014436236391912908,
      "loss": 0.041820187121629715,
      "step": 4202
    },
    {
      "epoch": 0.7968157695223654,
      "grad_norm": 0.10190409421920776,
      "learning_rate": 0.00014428460342146192,
      "loss": 0.040367212146520615,
      "step": 4204
    },
    {
      "epoch": 0.7971948445792267,
      "grad_norm": 0.38980719447135925,
      "learning_rate": 0.00014420684292379472,
      "loss": 0.026935836300253868,
      "step": 4206
    },
    {
      "epoch": 0.7975739196360879,
      "grad_norm": 0.16889168322086334,
      "learning_rate": 0.00014412908242612753,
      "loss": 0.017140384763479233,
      "step": 4208
    },
    {
      "epoch": 0.7979529946929492,
      "grad_norm": 0.25761711597442627,
      "learning_rate": 0.00014405132192846036,
      "loss": 0.017803853377699852,
      "step": 4210
    },
    {
      "epoch": 0.7983320697498104,
      "grad_norm": 0.37673112750053406,
      "learning_rate": 0.00014397356143079317,
      "loss": 0.03314870223402977,
      "step": 4212
    },
    {
      "epoch": 0.7987111448066717,
      "grad_norm": 0.0929860770702362,
      "learning_rate": 0.00014389580093312598,
      "loss": 0.012837284244596958,
      "step": 4214
    },
    {
      "epoch": 0.799090219863533,
      "grad_norm": 0.40609219670295715,
      "learning_rate": 0.00014381804043545878,
      "loss": 0.03423410654067993,
      "step": 4216
    },
    {
      "epoch": 0.7994692949203942,
      "grad_norm": 0.26346302032470703,
      "learning_rate": 0.0001437402799377916,
      "loss": 0.028953490778803825,
      "step": 4218
    },
    {
      "epoch": 0.7998483699772555,
      "grad_norm": 0.3214341998100281,
      "learning_rate": 0.00014366251944012443,
      "loss": 0.026126142591238022,
      "step": 4220
    },
    {
      "epoch": 0.8002274450341168,
      "grad_norm": 0.17495547235012054,
      "learning_rate": 0.00014358475894245723,
      "loss": 0.011442892253398895,
      "step": 4222
    },
    {
      "epoch": 0.800606520090978,
      "grad_norm": 0.3762495219707489,
      "learning_rate": 0.00014350699844479004,
      "loss": 0.030169358476996422,
      "step": 4224
    },
    {
      "epoch": 0.8009855951478393,
      "grad_norm": 0.10486695915460587,
      "learning_rate": 0.00014342923794712287,
      "loss": 0.02091860957443714,
      "step": 4226
    },
    {
      "epoch": 0.8013646702047005,
      "grad_norm": 0.368920236825943,
      "learning_rate": 0.00014335147744945568,
      "loss": 0.030182458460330963,
      "step": 4228
    },
    {
      "epoch": 0.8017437452615618,
      "grad_norm": 0.09973981976509094,
      "learning_rate": 0.00014327371695178851,
      "loss": 0.014081391505897045,
      "step": 4230
    },
    {
      "epoch": 0.8021228203184231,
      "grad_norm": 0.09008440375328064,
      "learning_rate": 0.00014319595645412132,
      "loss": 0.01591040939092636,
      "step": 4232
    },
    {
      "epoch": 0.8025018953752843,
      "grad_norm": 0.12224497646093369,
      "learning_rate": 0.00014311819595645413,
      "loss": 0.022195061668753624,
      "step": 4234
    },
    {
      "epoch": 0.8028809704321456,
      "grad_norm": 0.10285373777151108,
      "learning_rate": 0.00014304043545878693,
      "loss": 0.017064020037651062,
      "step": 4236
    },
    {
      "epoch": 0.8032600454890069,
      "grad_norm": 0.10185997933149338,
      "learning_rate": 0.00014296267496111977,
      "loss": 0.011848432943224907,
      "step": 4238
    },
    {
      "epoch": 0.8036391205458681,
      "grad_norm": 0.3729724884033203,
      "learning_rate": 0.00014288491446345258,
      "loss": 0.023378368467092514,
      "step": 4240
    },
    {
      "epoch": 0.8040181956027294,
      "grad_norm": 0.30720970034599304,
      "learning_rate": 0.00014280715396578538,
      "loss": 0.044285740703344345,
      "step": 4242
    },
    {
      "epoch": 0.8043972706595905,
      "grad_norm": 0.21975113451480865,
      "learning_rate": 0.0001427293934681182,
      "loss": 0.021674688905477524,
      "step": 4244
    },
    {
      "epoch": 0.8047763457164518,
      "grad_norm": 0.227890744805336,
      "learning_rate": 0.00014265163297045102,
      "loss": 0.03196880966424942,
      "step": 4246
    },
    {
      "epoch": 0.8051554207733131,
      "grad_norm": 0.08544481545686722,
      "learning_rate": 0.00014257387247278383,
      "loss": 0.011958323419094086,
      "step": 4248
    },
    {
      "epoch": 0.8055344958301743,
      "grad_norm": 0.48977723717689514,
      "learning_rate": 0.00014249611197511666,
      "loss": 0.042461782693862915,
      "step": 4250
    },
    {
      "epoch": 0.8059135708870356,
      "grad_norm": 0.31451189517974854,
      "learning_rate": 0.00014241835147744947,
      "loss": 0.03325169160962105,
      "step": 4252
    },
    {
      "epoch": 0.8062926459438969,
      "grad_norm": 0.4889377951622009,
      "learning_rate": 0.00014234059097978228,
      "loss": 0.05005112290382385,
      "step": 4254
    },
    {
      "epoch": 0.8066717210007581,
      "grad_norm": 0.1076573058962822,
      "learning_rate": 0.00014226283048211509,
      "loss": 0.029809042811393738,
      "step": 4256
    },
    {
      "epoch": 0.8070507960576194,
      "grad_norm": 0.1062655821442604,
      "learning_rate": 0.00014218506998444792,
      "loss": 0.02284146659076214,
      "step": 4258
    },
    {
      "epoch": 0.8074298711144806,
      "grad_norm": 0.25030985474586487,
      "learning_rate": 0.00014210730948678073,
      "loss": 0.02703806199133396,
      "step": 4260
    },
    {
      "epoch": 0.8078089461713419,
      "grad_norm": 0.09990652650594711,
      "learning_rate": 0.00014202954898911353,
      "loss": 0.015173729509115219,
      "step": 4262
    },
    {
      "epoch": 0.8081880212282032,
      "grad_norm": 0.33066073060035706,
      "learning_rate": 0.00014195178849144634,
      "loss": 0.03736316040158272,
      "step": 4264
    },
    {
      "epoch": 0.8085670962850644,
      "grad_norm": 0.10677255690097809,
      "learning_rate": 0.00014187402799377915,
      "loss": 0.017434947192668915,
      "step": 4266
    },
    {
      "epoch": 0.8089461713419257,
      "grad_norm": 0.28705117106437683,
      "learning_rate": 0.00014179626749611198,
      "loss": 0.017762530595064163,
      "step": 4268
    },
    {
      "epoch": 0.809325246398787,
      "grad_norm": 0.08984316140413284,
      "learning_rate": 0.0001417185069984448,
      "loss": 0.012005153112113476,
      "step": 4270
    },
    {
      "epoch": 0.8097043214556482,
      "grad_norm": 0.4202665090560913,
      "learning_rate": 0.00014164074650077762,
      "loss": 0.02740001305937767,
      "step": 4272
    },
    {
      "epoch": 0.8100833965125095,
      "grad_norm": 0.1108364388346672,
      "learning_rate": 0.00014156298600311043,
      "loss": 0.012195546180009842,
      "step": 4274
    },
    {
      "epoch": 0.8104624715693708,
      "grad_norm": 0.09435643255710602,
      "learning_rate": 0.00014148522550544324,
      "loss": 0.012926739640533924,
      "step": 4276
    },
    {
      "epoch": 0.810841546626232,
      "grad_norm": 0.3146926760673523,
      "learning_rate": 0.00014140746500777607,
      "loss": 0.02362186275422573,
      "step": 4278
    },
    {
      "epoch": 0.8112206216830933,
      "grad_norm": 0.07858849316835403,
      "learning_rate": 0.00014132970451010888,
      "loss": 0.013824142515659332,
      "step": 4280
    },
    {
      "epoch": 0.8115996967399545,
      "grad_norm": 0.39317429065704346,
      "learning_rate": 0.00014125194401244168,
      "loss": 0.030340123921632767,
      "step": 4282
    },
    {
      "epoch": 0.8119787717968158,
      "grad_norm": 0.46109268069267273,
      "learning_rate": 0.0001411741835147745,
      "loss": 0.03250723332166672,
      "step": 4284
    },
    {
      "epoch": 0.8123578468536771,
      "grad_norm": 0.3550931513309479,
      "learning_rate": 0.00014109642301710732,
      "loss": 0.04114225506782532,
      "step": 4286
    },
    {
      "epoch": 0.8127369219105383,
      "grad_norm": 0.1155422180891037,
      "learning_rate": 0.00014101866251944013,
      "loss": 0.020653733983635902,
      "step": 4288
    },
    {
      "epoch": 0.8131159969673996,
      "grad_norm": 0.356442391872406,
      "learning_rate": 0.00014094090202177294,
      "loss": 0.040943216532468796,
      "step": 4290
    },
    {
      "epoch": 0.8134950720242609,
      "grad_norm": 0.39152830839157104,
      "learning_rate": 0.00014086314152410574,
      "loss": 0.03916826844215393,
      "step": 4292
    },
    {
      "epoch": 0.813874147081122,
      "grad_norm": 0.3244084119796753,
      "learning_rate": 0.00014078538102643858,
      "loss": 0.02477354370057583,
      "step": 4294
    },
    {
      "epoch": 0.8142532221379833,
      "grad_norm": 0.25789207220077515,
      "learning_rate": 0.0001407076205287714,
      "loss": 0.0215616337954998,
      "step": 4296
    },
    {
      "epoch": 0.8146322971948445,
      "grad_norm": 0.2470623254776001,
      "learning_rate": 0.00014062986003110422,
      "loss": 0.015563497319817543,
      "step": 4298
    },
    {
      "epoch": 0.8150113722517058,
      "grad_norm": 0.21999500691890717,
      "learning_rate": 0.00014055209953343703,
      "loss": 0.03388287127017975,
      "step": 4300
    },
    {
      "epoch": 0.8153904473085671,
      "grad_norm": 0.42384079098701477,
      "learning_rate": 0.00014047433903576983,
      "loss": 0.03708489239215851,
      "step": 4302
    },
    {
      "epoch": 0.8157695223654283,
      "grad_norm": 0.11002500355243683,
      "learning_rate": 0.00014039657853810264,
      "loss": 0.019896458834409714,
      "step": 4304
    },
    {
      "epoch": 0.8161485974222896,
      "grad_norm": 0.27982333302497864,
      "learning_rate": 0.00014031881804043547,
      "loss": 0.018242690712213516,
      "step": 4306
    },
    {
      "epoch": 0.8165276724791509,
      "grad_norm": 0.12198425084352493,
      "learning_rate": 0.00014024105754276828,
      "loss": 0.026219114661216736,
      "step": 4308
    },
    {
      "epoch": 0.8169067475360121,
      "grad_norm": 0.09812919050455093,
      "learning_rate": 0.0001401632970451011,
      "loss": 0.016406740993261337,
      "step": 4310
    },
    {
      "epoch": 0.8172858225928734,
      "grad_norm": 0.2901313304901123,
      "learning_rate": 0.0001400855365474339,
      "loss": 0.02239890769124031,
      "step": 4312
    },
    {
      "epoch": 0.8176648976497346,
      "grad_norm": 0.3967798352241516,
      "learning_rate": 0.0001400077760497667,
      "loss": 0.02776387892663479,
      "step": 4314
    },
    {
      "epoch": 0.8180439727065959,
      "grad_norm": 0.3789183497428894,
      "learning_rate": 0.00013993001555209954,
      "loss": 0.04729348421096802,
      "step": 4316
    },
    {
      "epoch": 0.8184230477634572,
      "grad_norm": 0.3675053119659424,
      "learning_rate": 0.00013985225505443237,
      "loss": 0.02471025660634041,
      "step": 4318
    },
    {
      "epoch": 0.8188021228203184,
      "grad_norm": 0.5037147998809814,
      "learning_rate": 0.00013977449455676518,
      "loss": 0.04605262354016304,
      "step": 4320
    },
    {
      "epoch": 0.8191811978771797,
      "grad_norm": 0.18142114579677582,
      "learning_rate": 0.00013969673405909798,
      "loss": 0.04002253711223602,
      "step": 4322
    },
    {
      "epoch": 0.819560272934041,
      "grad_norm": 0.5310032367706299,
      "learning_rate": 0.00013961897356143082,
      "loss": 0.043760936707258224,
      "step": 4324
    },
    {
      "epoch": 0.8199393479909022,
      "grad_norm": 0.2262151837348938,
      "learning_rate": 0.00013954121306376362,
      "loss": 0.02215321734547615,
      "step": 4326
    },
    {
      "epoch": 0.8203184230477635,
      "grad_norm": 0.1045028567314148,
      "learning_rate": 0.00013946345256609643,
      "loss": 0.021101834252476692,
      "step": 4328
    },
    {
      "epoch": 0.8206974981046247,
      "grad_norm": 0.4157829284667969,
      "learning_rate": 0.00013938569206842924,
      "loss": 0.06953919678926468,
      "step": 4330
    },
    {
      "epoch": 0.821076573161486,
      "grad_norm": 0.07444193959236145,
      "learning_rate": 0.00013930793157076204,
      "loss": 0.014405930414795876,
      "step": 4332
    },
    {
      "epoch": 0.8214556482183473,
      "grad_norm": 0.3744310736656189,
      "learning_rate": 0.00013923017107309488,
      "loss": 0.030993670225143433,
      "step": 4334
    },
    {
      "epoch": 0.8218347232752085,
      "grad_norm": 0.3952319324016571,
      "learning_rate": 0.00013915241057542769,
      "loss": 0.025573227554559708,
      "step": 4336
    },
    {
      "epoch": 0.8222137983320698,
      "grad_norm": 0.24030299484729767,
      "learning_rate": 0.0001390746500777605,
      "loss": 0.019772673025727272,
      "step": 4338
    },
    {
      "epoch": 0.8225928733889311,
      "grad_norm": 0.39885634183883667,
      "learning_rate": 0.00013899688958009333,
      "loss": 0.03792164474725723,
      "step": 4340
    },
    {
      "epoch": 0.8229719484457922,
      "grad_norm": 0.26384949684143066,
      "learning_rate": 0.00013891912908242613,
      "loss": 0.01734163984656334,
      "step": 4342
    },
    {
      "epoch": 0.8233510235026535,
      "grad_norm": 0.5239940285682678,
      "learning_rate": 0.00013884136858475897,
      "loss": 0.028188340365886688,
      "step": 4344
    },
    {
      "epoch": 0.8237300985595147,
      "grad_norm": 0.34959277510643005,
      "learning_rate": 0.00013876360808709177,
      "loss": 0.02462555095553398,
      "step": 4346
    },
    {
      "epoch": 0.824109173616376,
      "grad_norm": 0.10534098744392395,
      "learning_rate": 0.00013868584758942458,
      "loss": 0.015423848293721676,
      "step": 4348
    },
    {
      "epoch": 0.8244882486732373,
      "grad_norm": 0.09737388044595718,
      "learning_rate": 0.0001386080870917574,
      "loss": 0.011800343170762062,
      "step": 4350
    },
    {
      "epoch": 0.8248673237300985,
      "grad_norm": 0.12148918211460114,
      "learning_rate": 0.0001385303265940902,
      "loss": 0.04821183905005455,
      "step": 4352
    },
    {
      "epoch": 0.8252463987869598,
      "grad_norm": 0.10757593810558319,
      "learning_rate": 0.00013845256609642303,
      "loss": 0.012984521687030792,
      "step": 4354
    },
    {
      "epoch": 0.8256254738438211,
      "grad_norm": 0.08920573443174362,
      "learning_rate": 0.00013837480559875584,
      "loss": 0.02124905027449131,
      "step": 4356
    },
    {
      "epoch": 0.8260045489006823,
      "grad_norm": 0.3687925636768341,
      "learning_rate": 0.00013829704510108864,
      "loss": 0.0313914455473423,
      "step": 4358
    },
    {
      "epoch": 0.8263836239575436,
      "grad_norm": 0.3556499779224396,
      "learning_rate": 0.00013821928460342145,
      "loss": 0.024266783148050308,
      "step": 4360
    },
    {
      "epoch": 0.8267626990144048,
      "grad_norm": 0.20208154618740082,
      "learning_rate": 0.00013814152410575428,
      "loss": 0.01689288392663002,
      "step": 4362
    },
    {
      "epoch": 0.8271417740712661,
      "grad_norm": 0.40708041191101074,
      "learning_rate": 0.00013806376360808712,
      "loss": 0.0348040945827961,
      "step": 4364
    },
    {
      "epoch": 0.8275208491281274,
      "grad_norm": 0.4907212257385254,
      "learning_rate": 0.00013798600311041992,
      "loss": 0.038788747042417526,
      "step": 4366
    },
    {
      "epoch": 0.8278999241849886,
      "grad_norm": 0.2675367593765259,
      "learning_rate": 0.00013790824261275273,
      "loss": 0.01956852898001671,
      "step": 4368
    },
    {
      "epoch": 0.8282789992418499,
      "grad_norm": 0.32813605666160583,
      "learning_rate": 0.00013783048211508554,
      "loss": 0.02708805352449417,
      "step": 4370
    },
    {
      "epoch": 0.8286580742987112,
      "grad_norm": 0.1292354315519333,
      "learning_rate": 0.00013775272161741837,
      "loss": 0.019924677908420563,
      "step": 4372
    },
    {
      "epoch": 0.8290371493555724,
      "grad_norm": 0.21928013861179352,
      "learning_rate": 0.00013767496111975118,
      "loss": 0.013094400987029076,
      "step": 4374
    },
    {
      "epoch": 0.8294162244124337,
      "grad_norm": 0.42736995220184326,
      "learning_rate": 0.00013759720062208399,
      "loss": 0.03523483872413635,
      "step": 4376
    },
    {
      "epoch": 0.8297952994692949,
      "grad_norm": 0.09782935678958893,
      "learning_rate": 0.0001375194401244168,
      "loss": 0.012306333519518375,
      "step": 4378
    },
    {
      "epoch": 0.8301743745261562,
      "grad_norm": 0.19446714222431183,
      "learning_rate": 0.0001374416796267496,
      "loss": 0.019797228276729584,
      "step": 4380
    },
    {
      "epoch": 0.8305534495830175,
      "grad_norm": 0.2759411036968231,
      "learning_rate": 0.00013736391912908243,
      "loss": 0.02189849503338337,
      "step": 4382
    },
    {
      "epoch": 0.8309325246398787,
      "grad_norm": 0.40344491600990295,
      "learning_rate": 0.00013728615863141524,
      "loss": 0.03986942768096924,
      "step": 4384
    },
    {
      "epoch": 0.83131159969674,
      "grad_norm": 0.36305689811706543,
      "learning_rate": 0.00013720839813374807,
      "loss": 0.026030398905277252,
      "step": 4386
    },
    {
      "epoch": 0.8316906747536013,
      "grad_norm": 0.10638339072465897,
      "learning_rate": 0.00013713063763608088,
      "loss": 0.0277200099080801,
      "step": 4388
    },
    {
      "epoch": 0.8320697498104624,
      "grad_norm": 0.3224691152572632,
      "learning_rate": 0.0001370528771384137,
      "loss": 0.02416539005935192,
      "step": 4390
    },
    {
      "epoch": 0.8324488248673237,
      "grad_norm": 0.08977731317281723,
      "learning_rate": 0.00013697511664074652,
      "loss": 0.02662798762321472,
      "step": 4392
    },
    {
      "epoch": 0.8328278999241849,
      "grad_norm": 0.32930463552474976,
      "learning_rate": 0.00013689735614307933,
      "loss": 0.018684634938836098,
      "step": 4394
    },
    {
      "epoch": 0.8332069749810462,
      "grad_norm": 0.09861790388822556,
      "learning_rate": 0.00013681959564541214,
      "loss": 0.013711466453969479,
      "step": 4396
    },
    {
      "epoch": 0.8335860500379075,
      "grad_norm": 0.46695059537887573,
      "learning_rate": 0.00013674183514774494,
      "loss": 0.07011602073907852,
      "step": 4398
    },
    {
      "epoch": 0.8339651250947687,
      "grad_norm": 0.06987374275922775,
      "learning_rate": 0.00013666407465007778,
      "loss": 0.036618515849113464,
      "step": 4400
    },
    {
      "epoch": 0.83434420015163,
      "grad_norm": 0.09591788798570633,
      "learning_rate": 0.00013658631415241058,
      "loss": 0.02980014681816101,
      "step": 4402
    },
    {
      "epoch": 0.8347232752084913,
      "grad_norm": 0.2728998064994812,
      "learning_rate": 0.0001365085536547434,
      "loss": 0.04276042804121971,
      "step": 4404
    },
    {
      "epoch": 0.8351023502653525,
      "grad_norm": 0.34606635570526123,
      "learning_rate": 0.0001364307931570762,
      "loss": 0.03530118986964226,
      "step": 4406
    },
    {
      "epoch": 0.8354814253222138,
      "grad_norm": 0.4140183627605438,
      "learning_rate": 0.00013635303265940903,
      "loss": 0.04073764383792877,
      "step": 4408
    },
    {
      "epoch": 0.8358605003790751,
      "grad_norm": 0.435512512922287,
      "learning_rate": 0.00013627527216174184,
      "loss": 0.04255305975675583,
      "step": 4410
    },
    {
      "epoch": 0.8362395754359363,
      "grad_norm": 0.3449312448501587,
      "learning_rate": 0.00013619751166407467,
      "loss": 0.022109322249889374,
      "step": 4412
    },
    {
      "epoch": 0.8366186504927976,
      "grad_norm": 0.07434176653623581,
      "learning_rate": 0.00013611975116640748,
      "loss": 0.018481817096471786,
      "step": 4414
    },
    {
      "epoch": 0.8369977255496588,
      "grad_norm": 0.38459423184394836,
      "learning_rate": 0.0001360419906687403,
      "loss": 0.0372607558965683,
      "step": 4416
    },
    {
      "epoch": 0.8373768006065201,
      "grad_norm": 0.11316949874162674,
      "learning_rate": 0.0001359642301710731,
      "loss": 0.01690658926963806,
      "step": 4418
    },
    {
      "epoch": 0.8377558756633814,
      "grad_norm": 0.11802089214324951,
      "learning_rate": 0.00013588646967340593,
      "loss": 0.012698840349912643,
      "step": 4420
    },
    {
      "epoch": 0.8381349507202426,
      "grad_norm": 0.25802603363990784,
      "learning_rate": 0.00013580870917573873,
      "loss": 0.018923556432127953,
      "step": 4422
    },
    {
      "epoch": 0.8385140257771039,
      "grad_norm": 0.22581714391708374,
      "learning_rate": 0.00013573094867807154,
      "loss": 0.022825269028544426,
      "step": 4424
    },
    {
      "epoch": 0.8388931008339652,
      "grad_norm": 0.12377673387527466,
      "learning_rate": 0.00013565318818040435,
      "loss": 0.02482597902417183,
      "step": 4426
    },
    {
      "epoch": 0.8392721758908264,
      "grad_norm": 0.20969638228416443,
      "learning_rate": 0.00013557542768273715,
      "loss": 0.02066046930849552,
      "step": 4428
    },
    {
      "epoch": 0.8396512509476877,
      "grad_norm": 0.4492226839065552,
      "learning_rate": 0.00013549766718507,
      "loss": 0.03632405400276184,
      "step": 4430
    },
    {
      "epoch": 0.8400303260045489,
      "grad_norm": 0.09719350188970566,
      "learning_rate": 0.0001354199066874028,
      "loss": 0.014409128576517105,
      "step": 4432
    },
    {
      "epoch": 0.8404094010614102,
      "grad_norm": 0.09347815811634064,
      "learning_rate": 0.00013534214618973563,
      "loss": 0.013651976361870766,
      "step": 4434
    },
    {
      "epoch": 0.8407884761182715,
      "grad_norm": 0.07443592697381973,
      "learning_rate": 0.00013526438569206844,
      "loss": 0.01078952755779028,
      "step": 4436
    },
    {
      "epoch": 0.8411675511751326,
      "grad_norm": 0.1043194979429245,
      "learning_rate": 0.00013518662519440124,
      "loss": 0.013086033053696156,
      "step": 4438
    },
    {
      "epoch": 0.8415466262319939,
      "grad_norm": 0.36061301827430725,
      "learning_rate": 0.00013510886469673408,
      "loss": 0.02370685338973999,
      "step": 4440
    },
    {
      "epoch": 0.8419257012888552,
      "grad_norm": 0.3519911766052246,
      "learning_rate": 0.00013503110419906688,
      "loss": 0.030874164775013924,
      "step": 4442
    },
    {
      "epoch": 0.8423047763457164,
      "grad_norm": 0.48107102513313293,
      "learning_rate": 0.0001349533437013997,
      "loss": 0.03973960131406784,
      "step": 4444
    },
    {
      "epoch": 0.8426838514025777,
      "grad_norm": 0.3258921504020691,
      "learning_rate": 0.0001348755832037325,
      "loss": 0.022536512464284897,
      "step": 4446
    },
    {
      "epoch": 0.8430629264594389,
      "grad_norm": 0.37295472621917725,
      "learning_rate": 0.00013479782270606533,
      "loss": 0.02945304661989212,
      "step": 4448
    },
    {
      "epoch": 0.8434420015163002,
      "grad_norm": 0.11703449487686157,
      "learning_rate": 0.00013472006220839814,
      "loss": 0.02750067412853241,
      "step": 4450
    },
    {
      "epoch": 0.8438210765731615,
      "grad_norm": 0.25563663244247437,
      "learning_rate": 0.00013464230171073095,
      "loss": 0.028032898902893066,
      "step": 4452
    },
    {
      "epoch": 0.8442001516300227,
      "grad_norm": 0.16193749010562897,
      "learning_rate": 0.00013456454121306375,
      "loss": 0.016091417521238327,
      "step": 4454
    },
    {
      "epoch": 0.844579226686884,
      "grad_norm": 0.4281306266784668,
      "learning_rate": 0.0001344867807153966,
      "loss": 0.03770504146814346,
      "step": 4456
    },
    {
      "epoch": 0.8449583017437453,
      "grad_norm": 0.2304157316684723,
      "learning_rate": 0.00013440902021772942,
      "loss": 0.015486686490476131,
      "step": 4458
    },
    {
      "epoch": 0.8453373768006065,
      "grad_norm": 0.40869230031967163,
      "learning_rate": 0.00013433125972006223,
      "loss": 0.041143495589494705,
      "step": 4460
    },
    {
      "epoch": 0.8457164518574678,
      "grad_norm": 0.4185390770435333,
      "learning_rate": 0.00013425349922239503,
      "loss": 0.030955152586102486,
      "step": 4462
    },
    {
      "epoch": 0.846095526914329,
      "grad_norm": 0.14748385548591614,
      "learning_rate": 0.00013417573872472784,
      "loss": 0.029843419790267944,
      "step": 4464
    },
    {
      "epoch": 0.8464746019711903,
      "grad_norm": 0.35029736161231995,
      "learning_rate": 0.00013409797822706065,
      "loss": 0.02075488120317459,
      "step": 4466
    },
    {
      "epoch": 0.8468536770280516,
      "grad_norm": 0.22949282824993134,
      "learning_rate": 0.00013402021772939348,
      "loss": 0.019487028941512108,
      "step": 4468
    },
    {
      "epoch": 0.8472327520849128,
      "grad_norm": 0.23734430968761444,
      "learning_rate": 0.0001339424572317263,
      "loss": 0.022808019071817398,
      "step": 4470
    },
    {
      "epoch": 0.8476118271417741,
      "grad_norm": 0.09193255007266998,
      "learning_rate": 0.0001338646967340591,
      "loss": 0.03174440190196037,
      "step": 4472
    },
    {
      "epoch": 0.8479909021986354,
      "grad_norm": 0.2804880440235138,
      "learning_rate": 0.0001337869362363919,
      "loss": 0.023214269429445267,
      "step": 4474
    },
    {
      "epoch": 0.8483699772554966,
      "grad_norm": 0.08554807305335999,
      "learning_rate": 0.00013370917573872474,
      "loss": 0.02603004314005375,
      "step": 4476
    },
    {
      "epoch": 0.8487490523123579,
      "grad_norm": 0.0899711325764656,
      "learning_rate": 0.00013363141524105754,
      "loss": 0.013417608104646206,
      "step": 4478
    },
    {
      "epoch": 0.849128127369219,
      "grad_norm": 0.07570893317461014,
      "learning_rate": 0.00013355365474339038,
      "loss": 0.024523306638002396,
      "step": 4480
    },
    {
      "epoch": 0.8495072024260804,
      "grad_norm": 0.19448092579841614,
      "learning_rate": 0.00013347589424572318,
      "loss": 0.018823759630322456,
      "step": 4482
    },
    {
      "epoch": 0.8498862774829417,
      "grad_norm": 0.09303537011146545,
      "learning_rate": 0.000133398133748056,
      "loss": 0.01896170899271965,
      "step": 4484
    },
    {
      "epoch": 0.8502653525398028,
      "grad_norm": 0.10059264302253723,
      "learning_rate": 0.00013332037325038883,
      "loss": 0.011867736466228962,
      "step": 4486
    },
    {
      "epoch": 0.8506444275966641,
      "grad_norm": 0.13391229510307312,
      "learning_rate": 0.00013324261275272163,
      "loss": 0.013533761724829674,
      "step": 4488
    },
    {
      "epoch": 0.8510235026535254,
      "grad_norm": 0.07898584753274918,
      "learning_rate": 0.00013316485225505444,
      "loss": 0.012582832016050816,
      "step": 4490
    },
    {
      "epoch": 0.8514025777103866,
      "grad_norm": 0.4087202548980713,
      "learning_rate": 0.00013308709175738725,
      "loss": 0.028443947434425354,
      "step": 4492
    },
    {
      "epoch": 0.8517816527672479,
      "grad_norm": 0.18707993626594543,
      "learning_rate": 0.00013300933125972005,
      "loss": 0.016255727037787437,
      "step": 4494
    },
    {
      "epoch": 0.8521607278241091,
      "grad_norm": 0.36226099729537964,
      "learning_rate": 0.0001329315707620529,
      "loss": 0.0309246014803648,
      "step": 4496
    },
    {
      "epoch": 0.8525398028809704,
      "grad_norm": 0.09924179315567017,
      "learning_rate": 0.0001328538102643857,
      "loss": 0.011732785031199455,
      "step": 4498
    },
    {
      "epoch": 0.8529188779378317,
      "grad_norm": 0.10287216305732727,
      "learning_rate": 0.0001327760497667185,
      "loss": 0.018323851749300957,
      "step": 4500
    },
    {
      "epoch": 0.8532979529946929,
      "grad_norm": 0.2774707078933716,
      "learning_rate": 0.00013269828926905133,
      "loss": 0.02944786846637726,
      "step": 4502
    },
    {
      "epoch": 0.8536770280515542,
      "grad_norm": 0.35099586844444275,
      "learning_rate": 0.00013262052877138414,
      "loss": 0.02835758775472641,
      "step": 4504
    },
    {
      "epoch": 0.8540561031084155,
      "grad_norm": 0.21530432999134064,
      "learning_rate": 0.00013254276827371698,
      "loss": 0.03599095344543457,
      "step": 4506
    },
    {
      "epoch": 0.8544351781652767,
      "grad_norm": 0.09614920616149902,
      "learning_rate": 0.00013246500777604978,
      "loss": 0.023616863414645195,
      "step": 4508
    },
    {
      "epoch": 0.854814253222138,
      "grad_norm": 0.11624462902545929,
      "learning_rate": 0.0001323872472783826,
      "loss": 0.02274082601070404,
      "step": 4510
    },
    {
      "epoch": 0.8551933282789992,
      "grad_norm": 0.31832486391067505,
      "learning_rate": 0.0001323094867807154,
      "loss": 0.018632184714078903,
      "step": 4512
    },
    {
      "epoch": 0.8555724033358605,
      "grad_norm": 0.09525017440319061,
      "learning_rate": 0.0001322317262830482,
      "loss": 0.01997692510485649,
      "step": 4514
    },
    {
      "epoch": 0.8559514783927218,
      "grad_norm": 0.0880674421787262,
      "learning_rate": 0.00013215396578538104,
      "loss": 0.014976948499679565,
      "step": 4516
    },
    {
      "epoch": 0.856330553449583,
      "grad_norm": 0.1063113585114479,
      "learning_rate": 0.00013207620528771384,
      "loss": 0.01864754594862461,
      "step": 4518
    },
    {
      "epoch": 0.8567096285064443,
      "grad_norm": 0.214843288064003,
      "learning_rate": 0.00013199844479004665,
      "loss": 0.033147141337394714,
      "step": 4520
    },
    {
      "epoch": 0.8570887035633056,
      "grad_norm": 0.09947187453508377,
      "learning_rate": 0.00013192068429237946,
      "loss": 0.010454531759023666,
      "step": 4522
    },
    {
      "epoch": 0.8574677786201668,
      "grad_norm": 0.3747418522834778,
      "learning_rate": 0.0001318429237947123,
      "loss": 0.028446100652217865,
      "step": 4524
    },
    {
      "epoch": 0.8578468536770281,
      "grad_norm": 0.1285761296749115,
      "learning_rate": 0.00013176516329704513,
      "loss": 0.01802978664636612,
      "step": 4526
    },
    {
      "epoch": 0.8582259287338894,
      "grad_norm": 0.35383620858192444,
      "learning_rate": 0.00013168740279937793,
      "loss": 0.03674844279885292,
      "step": 4528
    },
    {
      "epoch": 0.8586050037907506,
      "grad_norm": 0.11894360929727554,
      "learning_rate": 0.00013160964230171074,
      "loss": 0.0267255250364542,
      "step": 4530
    },
    {
      "epoch": 0.8589840788476119,
      "grad_norm": 0.3282104730606079,
      "learning_rate": 0.00013153188180404355,
      "loss": 0.03110314905643463,
      "step": 4532
    },
    {
      "epoch": 0.859363153904473,
      "grad_norm": 0.3574288785457611,
      "learning_rate": 0.00013145412130637638,
      "loss": 0.040117938071489334,
      "step": 4534
    },
    {
      "epoch": 0.8597422289613343,
      "grad_norm": 0.1757321059703827,
      "learning_rate": 0.0001313763608087092,
      "loss": 0.02696852758526802,
      "step": 4536
    },
    {
      "epoch": 0.8601213040181956,
      "grad_norm": 0.37682539224624634,
      "learning_rate": 0.000131298600311042,
      "loss": 0.023671699687838554,
      "step": 4538
    },
    {
      "epoch": 0.8605003790750568,
      "grad_norm": 0.11149023473262787,
      "learning_rate": 0.0001312208398133748,
      "loss": 0.019572295248508453,
      "step": 4540
    },
    {
      "epoch": 0.8608794541319181,
      "grad_norm": 0.1063656359910965,
      "learning_rate": 0.0001311430793157076,
      "loss": 0.012821602635085583,
      "step": 4542
    },
    {
      "epoch": 0.8612585291887794,
      "grad_norm": 0.10011891275644302,
      "learning_rate": 0.00013106531881804044,
      "loss": 0.026829537004232407,
      "step": 4544
    },
    {
      "epoch": 0.8616376042456406,
      "grad_norm": 0.4006015658378601,
      "learning_rate": 0.00013098755832037325,
      "loss": 0.022934820502996445,
      "step": 4546
    },
    {
      "epoch": 0.8620166793025019,
      "grad_norm": 0.13728828728199005,
      "learning_rate": 0.00013090979782270608,
      "loss": 0.014411495998501778,
      "step": 4548
    },
    {
      "epoch": 0.8623957543593631,
      "grad_norm": 0.2338375598192215,
      "learning_rate": 0.0001308320373250389,
      "loss": 0.018246255815029144,
      "step": 4550
    },
    {
      "epoch": 0.8627748294162244,
      "grad_norm": 0.33900222182273865,
      "learning_rate": 0.0001307542768273717,
      "loss": 0.024502664804458618,
      "step": 4552
    },
    {
      "epoch": 0.8631539044730857,
      "grad_norm": 0.30338773131370544,
      "learning_rate": 0.00013067651632970453,
      "loss": 0.02253567799925804,
      "step": 4554
    },
    {
      "epoch": 0.8635329795299469,
      "grad_norm": 0.23682041466236115,
      "learning_rate": 0.00013059875583203734,
      "loss": 0.01973959617316723,
      "step": 4556
    },
    {
      "epoch": 0.8639120545868082,
      "grad_norm": 0.3236151337623596,
      "learning_rate": 0.00013052099533437014,
      "loss": 0.024152647703886032,
      "step": 4558
    },
    {
      "epoch": 0.8642911296436695,
      "grad_norm": 0.34635379910469055,
      "learning_rate": 0.00013044323483670295,
      "loss": 0.03223646432161331,
      "step": 4560
    },
    {
      "epoch": 0.8646702047005307,
      "grad_norm": 0.11227469891309738,
      "learning_rate": 0.00013036547433903579,
      "loss": 0.011880194768309593,
      "step": 4562
    },
    {
      "epoch": 0.865049279757392,
      "grad_norm": 0.09735177457332611,
      "learning_rate": 0.0001302877138413686,
      "loss": 0.01411335077136755,
      "step": 4564
    },
    {
      "epoch": 0.8654283548142532,
      "grad_norm": 0.11490258574485779,
      "learning_rate": 0.0001302099533437014,
      "loss": 0.016486335545778275,
      "step": 4566
    },
    {
      "epoch": 0.8658074298711145,
      "grad_norm": 0.1606936752796173,
      "learning_rate": 0.0001301321928460342,
      "loss": 0.024123918265104294,
      "step": 4568
    },
    {
      "epoch": 0.8661865049279758,
      "grad_norm": 0.3286927044391632,
      "learning_rate": 0.00013005443234836704,
      "loss": 0.023733846843242645,
      "step": 4570
    },
    {
      "epoch": 0.866565579984837,
      "grad_norm": 0.21634040772914886,
      "learning_rate": 0.00012997667185069985,
      "loss": 0.02350144274532795,
      "step": 4572
    },
    {
      "epoch": 0.8669446550416983,
      "grad_norm": 0.21931473910808563,
      "learning_rate": 0.00012989891135303268,
      "loss": 0.015952089801430702,
      "step": 4574
    },
    {
      "epoch": 0.8673237300985596,
      "grad_norm": 0.43583595752716064,
      "learning_rate": 0.0001298211508553655,
      "loss": 0.026675386354327202,
      "step": 4576
    },
    {
      "epoch": 0.8677028051554208,
      "grad_norm": 0.3976868987083435,
      "learning_rate": 0.0001297433903576983,
      "loss": 0.04027927294373512,
      "step": 4578
    },
    {
      "epoch": 0.868081880212282,
      "grad_norm": 0.1984059065580368,
      "learning_rate": 0.0001296656298600311,
      "loss": 0.01921430416405201,
      "step": 4580
    },
    {
      "epoch": 0.8684609552691432,
      "grad_norm": 0.36068233847618103,
      "learning_rate": 0.00012958786936236394,
      "loss": 0.03256803750991821,
      "step": 4582
    },
    {
      "epoch": 0.8688400303260045,
      "grad_norm": 0.3202504515647888,
      "learning_rate": 0.00012951010886469674,
      "loss": 0.032152142375707626,
      "step": 4584
    },
    {
      "epoch": 0.8692191053828658,
      "grad_norm": 0.2334008663892746,
      "learning_rate": 0.00012943234836702955,
      "loss": 0.02871026284992695,
      "step": 4586
    },
    {
      "epoch": 0.869598180439727,
      "grad_norm": 0.25329047441482544,
      "learning_rate": 0.00012935458786936236,
      "loss": 0.022526338696479797,
      "step": 4588
    },
    {
      "epoch": 0.8699772554965883,
      "grad_norm": 0.1259855329990387,
      "learning_rate": 0.00012927682737169516,
      "loss": 0.019344914704561234,
      "step": 4590
    },
    {
      "epoch": 0.8703563305534496,
      "grad_norm": 0.3956488370895386,
      "learning_rate": 0.000129199066874028,
      "loss": 0.0343489944934845,
      "step": 4592
    },
    {
      "epoch": 0.8707354056103108,
      "grad_norm": 0.3809451162815094,
      "learning_rate": 0.00012912130637636083,
      "loss": 0.03539884462952614,
      "step": 4594
    },
    {
      "epoch": 0.8711144806671721,
      "grad_norm": 0.44948506355285645,
      "learning_rate": 0.00012904354587869364,
      "loss": 0.03972838819026947,
      "step": 4596
    },
    {
      "epoch": 0.8714935557240333,
      "grad_norm": 0.07355493307113647,
      "learning_rate": 0.00012896578538102644,
      "loss": 0.02301087975502014,
      "step": 4598
    },
    {
      "epoch": 0.8718726307808946,
      "grad_norm": 0.3511672019958496,
      "learning_rate": 0.00012888802488335928,
      "loss": 0.029885580763220787,
      "step": 4600
    },
    {
      "epoch": 0.8722517058377559,
      "grad_norm": 0.3300544023513794,
      "learning_rate": 0.00012881026438569209,
      "loss": 0.01929057016968727,
      "step": 4602
    },
    {
      "epoch": 0.8726307808946171,
      "grad_norm": 0.09111789613962173,
      "learning_rate": 0.0001287325038880249,
      "loss": 0.01628243550658226,
      "step": 4604
    },
    {
      "epoch": 0.8730098559514784,
      "grad_norm": 0.08462627232074738,
      "learning_rate": 0.0001286547433903577,
      "loss": 0.012797451578080654,
      "step": 4606
    },
    {
      "epoch": 0.8733889310083397,
      "grad_norm": 0.29888227581977844,
      "learning_rate": 0.0001285769828926905,
      "loss": 0.026108818128705025,
      "step": 4608
    },
    {
      "epoch": 0.8737680060652009,
      "grad_norm": 0.09600190073251724,
      "learning_rate": 0.00012849922239502334,
      "loss": 0.01222743559628725,
      "step": 4610
    },
    {
      "epoch": 0.8741470811220622,
      "grad_norm": 0.11509885638952255,
      "learning_rate": 0.00012842146189735615,
      "loss": 0.037521325051784515,
      "step": 4612
    },
    {
      "epoch": 0.8745261561789234,
      "grad_norm": 0.33295461535453796,
      "learning_rate": 0.00012834370139968895,
      "loss": 0.02839794009923935,
      "step": 4614
    },
    {
      "epoch": 0.8749052312357847,
      "grad_norm": 0.22269675135612488,
      "learning_rate": 0.0001282659409020218,
      "loss": 0.016826733946800232,
      "step": 4616
    },
    {
      "epoch": 0.875284306292646,
      "grad_norm": 0.3596697449684143,
      "learning_rate": 0.0001281881804043546,
      "loss": 0.030626488849520683,
      "step": 4618
    },
    {
      "epoch": 0.8756633813495072,
      "grad_norm": 0.10139650851488113,
      "learning_rate": 0.00012811041990668743,
      "loss": 0.035207200795412064,
      "step": 4620
    },
    {
      "epoch": 0.8760424564063685,
      "grad_norm": 0.09467852115631104,
      "learning_rate": 0.00012803265940902024,
      "loss": 0.02734067663550377,
      "step": 4622
    },
    {
      "epoch": 0.8764215314632298,
      "grad_norm": 0.3609621524810791,
      "learning_rate": 0.00012795489891135304,
      "loss": 0.031579434871673584,
      "step": 4624
    },
    {
      "epoch": 0.876800606520091,
      "grad_norm": 0.09246067702770233,
      "learning_rate": 0.00012787713841368585,
      "loss": 0.01724037528038025,
      "step": 4626
    },
    {
      "epoch": 0.8771796815769523,
      "grad_norm": 0.33493754267692566,
      "learning_rate": 0.00012779937791601866,
      "loss": 0.03316587954759598,
      "step": 4628
    },
    {
      "epoch": 0.8775587566338134,
      "grad_norm": 0.0982179045677185,
      "learning_rate": 0.0001277216174183515,
      "loss": 0.014298757538199425,
      "step": 4630
    },
    {
      "epoch": 0.8779378316906747,
      "grad_norm": 0.4779360294342041,
      "learning_rate": 0.0001276438569206843,
      "loss": 0.05148039013147354,
      "step": 4632
    },
    {
      "epoch": 0.878316906747536,
      "grad_norm": 0.5973948240280151,
      "learning_rate": 0.0001275660964230171,
      "loss": 0.017740976065397263,
      "step": 4634
    },
    {
      "epoch": 0.8786959818043972,
      "grad_norm": 0.4067806601524353,
      "learning_rate": 0.0001274883359253499,
      "loss": 0.03166181966662407,
      "step": 4636
    },
    {
      "epoch": 0.8790750568612585,
      "grad_norm": 0.11494223773479462,
      "learning_rate": 0.00012741057542768275,
      "loss": 0.017717745155096054,
      "step": 4638
    },
    {
      "epoch": 0.8794541319181198,
      "grad_norm": 0.20511393249034882,
      "learning_rate": 0.00012733281493001555,
      "loss": 0.02694590575993061,
      "step": 4640
    },
    {
      "epoch": 0.879833206974981,
      "grad_norm": 0.13087792694568634,
      "learning_rate": 0.00012725505443234839,
      "loss": 0.012989732436835766,
      "step": 4642
    },
    {
      "epoch": 0.8802122820318423,
      "grad_norm": 0.3550160229206085,
      "learning_rate": 0.0001271772939346812,
      "loss": 0.025795046240091324,
      "step": 4644
    },
    {
      "epoch": 0.8805913570887035,
      "grad_norm": 0.3212029039859772,
      "learning_rate": 0.000127099533437014,
      "loss": 0.020270399749279022,
      "step": 4646
    },
    {
      "epoch": 0.8809704321455648,
      "grad_norm": 0.1365114003419876,
      "learning_rate": 0.00012702177293934683,
      "loss": 0.018074724823236465,
      "step": 4648
    },
    {
      "epoch": 0.8813495072024261,
      "grad_norm": 0.09264892339706421,
      "learning_rate": 0.00012694401244167964,
      "loss": 0.014760605990886688,
      "step": 4650
    },
    {
      "epoch": 0.8817285822592873,
      "grad_norm": 0.4335683286190033,
      "learning_rate": 0.00012686625194401245,
      "loss": 0.030234171077609062,
      "step": 4652
    },
    {
      "epoch": 0.8821076573161486,
      "grad_norm": 0.11134231835603714,
      "learning_rate": 0.00012678849144634525,
      "loss": 0.021064547821879387,
      "step": 4654
    },
    {
      "epoch": 0.8824867323730099,
      "grad_norm": 0.33842727541923523,
      "learning_rate": 0.00012671073094867806,
      "loss": 0.02291310951113701,
      "step": 4656
    },
    {
      "epoch": 0.8828658074298711,
      "grad_norm": 0.41571107506752014,
      "learning_rate": 0.0001266329704510109,
      "loss": 0.04038507863879204,
      "step": 4658
    },
    {
      "epoch": 0.8832448824867324,
      "grad_norm": 0.23663735389709473,
      "learning_rate": 0.0001265552099533437,
      "loss": 0.014847910962998867,
      "step": 4660
    },
    {
      "epoch": 0.8836239575435937,
      "grad_norm": 0.1199328824877739,
      "learning_rate": 0.0001264774494556765,
      "loss": 0.013912336900830269,
      "step": 4662
    },
    {
      "epoch": 0.8840030326004549,
      "grad_norm": 0.10813847184181213,
      "learning_rate": 0.00012639968895800934,
      "loss": 0.014927000738680363,
      "step": 4664
    },
    {
      "epoch": 0.8843821076573162,
      "grad_norm": 0.33676066994667053,
      "learning_rate": 0.00012632192846034215,
      "loss": 0.03761905059218407,
      "step": 4666
    },
    {
      "epoch": 0.8847611827141774,
      "grad_norm": 0.3795348107814789,
      "learning_rate": 0.00012624416796267498,
      "loss": 0.033696193248033524,
      "step": 4668
    },
    {
      "epoch": 0.8851402577710387,
      "grad_norm": 0.3736579418182373,
      "learning_rate": 0.0001261664074650078,
      "loss": 0.02164621092379093,
      "step": 4670
    },
    {
      "epoch": 0.8855193328279,
      "grad_norm": 0.0953088328242302,
      "learning_rate": 0.0001260886469673406,
      "loss": 0.02494518831372261,
      "step": 4672
    },
    {
      "epoch": 0.8858984078847612,
      "grad_norm": 0.28228360414505005,
      "learning_rate": 0.0001260108864696734,
      "loss": 0.018053803592920303,
      "step": 4674
    },
    {
      "epoch": 0.8862774829416225,
      "grad_norm": 0.08527150005102158,
      "learning_rate": 0.0001259331259720062,
      "loss": 0.01241916511207819,
      "step": 4676
    },
    {
      "epoch": 0.8866565579984838,
      "grad_norm": 0.11384999752044678,
      "learning_rate": 0.00012585536547433905,
      "loss": 0.035884540528059006,
      "step": 4678
    },
    {
      "epoch": 0.8870356330553449,
      "grad_norm": 0.06662053614854813,
      "learning_rate": 0.00012577760497667185,
      "loss": 0.019640527665615082,
      "step": 4680
    },
    {
      "epoch": 0.8874147081122062,
      "grad_norm": 0.4360368549823761,
      "learning_rate": 0.00012569984447900466,
      "loss": 0.028648480772972107,
      "step": 4682
    },
    {
      "epoch": 0.8877937831690674,
      "grad_norm": 0.10071860253810883,
      "learning_rate": 0.00012562208398133747,
      "loss": 0.014662239700555801,
      "step": 4684
    },
    {
      "epoch": 0.8881728582259287,
      "grad_norm": 0.09873034060001373,
      "learning_rate": 0.0001255443234836703,
      "loss": 0.021126318722963333,
      "step": 4686
    },
    {
      "epoch": 0.88855193328279,
      "grad_norm": 0.06910929828882217,
      "learning_rate": 0.00012546656298600313,
      "loss": 0.01317006815224886,
      "step": 4688
    },
    {
      "epoch": 0.8889310083396512,
      "grad_norm": 0.3262355625629425,
      "learning_rate": 0.00012538880248833594,
      "loss": 0.04290953278541565,
      "step": 4690
    },
    {
      "epoch": 0.8893100833965125,
      "grad_norm": 0.3542686998844147,
      "learning_rate": 0.00012531104199066875,
      "loss": 0.03302775323390961,
      "step": 4692
    },
    {
      "epoch": 0.8896891584533738,
      "grad_norm": 0.23003241419792175,
      "learning_rate": 0.00012523328149300155,
      "loss": 0.016399221494793892,
      "step": 4694
    },
    {
      "epoch": 0.890068233510235,
      "grad_norm": 0.3384794592857361,
      "learning_rate": 0.0001251555209953344,
      "loss": 0.029622511938214302,
      "step": 4696
    },
    {
      "epoch": 0.8904473085670963,
      "grad_norm": 0.10281483829021454,
      "learning_rate": 0.0001250777604976672,
      "loss": 0.016190946102142334,
      "step": 4698
    },
    {
      "epoch": 0.8908263836239575,
      "grad_norm": 0.347922682762146,
      "learning_rate": 0.000125,
      "loss": 0.030195459723472595,
      "step": 4700
    },
    {
      "epoch": 0.8912054586808188,
      "grad_norm": 0.34769606590270996,
      "learning_rate": 0.0001249222395023328,
      "loss": 0.02167971059679985,
      "step": 4702
    },
    {
      "epoch": 0.8915845337376801,
      "grad_norm": 0.38026779890060425,
      "learning_rate": 0.00012484447900466562,
      "loss": 0.0437047965824604,
      "step": 4704
    },
    {
      "epoch": 0.8919636087945413,
      "grad_norm": 0.08572964370250702,
      "learning_rate": 0.00012476671850699845,
      "loss": 0.017361773177981377,
      "step": 4706
    },
    {
      "epoch": 0.8923426838514026,
      "grad_norm": 0.29015177488327026,
      "learning_rate": 0.00012468895800933126,
      "loss": 0.023232465609908104,
      "step": 4708
    },
    {
      "epoch": 0.8927217589082639,
      "grad_norm": 0.26593825221061707,
      "learning_rate": 0.0001246111975116641,
      "loss": 0.01810300350189209,
      "step": 4710
    },
    {
      "epoch": 0.8931008339651251,
      "grad_norm": 0.26088377833366394,
      "learning_rate": 0.0001245334370139969,
      "loss": 0.02731812745332718,
      "step": 4712
    },
    {
      "epoch": 0.8934799090219864,
      "grad_norm": 0.25717639923095703,
      "learning_rate": 0.0001244556765163297,
      "loss": 0.03502586483955383,
      "step": 4714
    },
    {
      "epoch": 0.8938589840788476,
      "grad_norm": 0.13542860746383667,
      "learning_rate": 0.00012437791601866254,
      "loss": 0.018746720626950264,
      "step": 4716
    },
    {
      "epoch": 0.8942380591357089,
      "grad_norm": 0.20647838711738586,
      "learning_rate": 0.00012430015552099535,
      "loss": 0.014497400261461735,
      "step": 4718
    },
    {
      "epoch": 0.8946171341925702,
      "grad_norm": 0.35598528385162354,
      "learning_rate": 0.00012422239502332815,
      "loss": 0.02556982822716236,
      "step": 4720
    },
    {
      "epoch": 0.8949962092494314,
      "grad_norm": 0.12049107998609543,
      "learning_rate": 0.00012414463452566096,
      "loss": 0.018134113401174545,
      "step": 4722
    },
    {
      "epoch": 0.8953752843062927,
      "grad_norm": 0.1943730115890503,
      "learning_rate": 0.0001240668740279938,
      "loss": 0.019145874306559563,
      "step": 4724
    },
    {
      "epoch": 0.895754359363154,
      "grad_norm": 0.24105724692344666,
      "learning_rate": 0.0001239891135303266,
      "loss": 0.02222561463713646,
      "step": 4726
    },
    {
      "epoch": 0.8961334344200151,
      "grad_norm": 0.2861897051334381,
      "learning_rate": 0.0001239113530326594,
      "loss": 0.02343447133898735,
      "step": 4728
    },
    {
      "epoch": 0.8965125094768764,
      "grad_norm": 0.268832802772522,
      "learning_rate": 0.00012383359253499221,
      "loss": 0.028345877304673195,
      "step": 4730
    },
    {
      "epoch": 0.8968915845337376,
      "grad_norm": 0.09304618090391159,
      "learning_rate": 0.00012375583203732505,
      "loss": 0.02007635496556759,
      "step": 4732
    },
    {
      "epoch": 0.8972706595905989,
      "grad_norm": 0.12733669579029083,
      "learning_rate": 0.00012367807153965788,
      "loss": 0.01891431026160717,
      "step": 4734
    },
    {
      "epoch": 0.8976497346474602,
      "grad_norm": 0.07602265477180481,
      "learning_rate": 0.0001236003110419907,
      "loss": 0.024686617776751518,
      "step": 4736
    },
    {
      "epoch": 0.8980288097043214,
      "grad_norm": 0.11511634290218353,
      "learning_rate": 0.0001235225505443235,
      "loss": 0.020143665373325348,
      "step": 4738
    },
    {
      "epoch": 0.8984078847611827,
      "grad_norm": 0.30390486121177673,
      "learning_rate": 0.0001234447900466563,
      "loss": 0.03229751065373421,
      "step": 4740
    },
    {
      "epoch": 0.898786959818044,
      "grad_norm": 0.2719845771789551,
      "learning_rate": 0.0001233670295489891,
      "loss": 0.018324006348848343,
      "step": 4742
    },
    {
      "epoch": 0.8991660348749052,
      "grad_norm": 0.3010357618331909,
      "learning_rate": 0.00012328926905132194,
      "loss": 0.017082426697015762,
      "step": 4744
    },
    {
      "epoch": 0.8995451099317665,
      "grad_norm": 0.11084313690662384,
      "learning_rate": 0.00012321150855365475,
      "loss": 0.015838757157325745,
      "step": 4746
    },
    {
      "epoch": 0.8999241849886277,
      "grad_norm": 0.5447332262992859,
      "learning_rate": 0.00012313374805598756,
      "loss": 0.04568726569414139,
      "step": 4748
    },
    {
      "epoch": 0.900303260045489,
      "grad_norm": 0.18465696275234222,
      "learning_rate": 0.00012305598755832036,
      "loss": 0.025286488234996796,
      "step": 4750
    },
    {
      "epoch": 0.9006823351023503,
      "grad_norm": 0.08977997303009033,
      "learning_rate": 0.00012297822706065317,
      "loss": 0.027966773137450218,
      "step": 4752
    },
    {
      "epoch": 0.9010614101592115,
      "grad_norm": 0.08820124715566635,
      "learning_rate": 0.000122900466562986,
      "loss": 0.011947380378842354,
      "step": 4754
    },
    {
      "epoch": 0.9014404852160728,
      "grad_norm": 0.0836939662694931,
      "learning_rate": 0.00012282270606531884,
      "loss": 0.012747850269079208,
      "step": 4756
    },
    {
      "epoch": 0.9018195602729341,
      "grad_norm": 0.3172588050365448,
      "learning_rate": 0.00012274494556765165,
      "loss": 0.021435949951410294,
      "step": 4758
    },
    {
      "epoch": 0.9021986353297953,
      "grad_norm": 0.10854597389698029,
      "learning_rate": 0.00012266718506998445,
      "loss": 0.018285391852259636,
      "step": 4760
    },
    {
      "epoch": 0.9025777103866566,
      "grad_norm": 0.11760808527469635,
      "learning_rate": 0.0001225894245723173,
      "loss": 0.01651492901146412,
      "step": 4762
    },
    {
      "epoch": 0.9029567854435178,
      "grad_norm": 0.09066929668188095,
      "learning_rate": 0.0001225116640746501,
      "loss": 0.014670286327600479,
      "step": 4764
    },
    {
      "epoch": 0.9033358605003791,
      "grad_norm": 0.36775606870651245,
      "learning_rate": 0.0001224339035769829,
      "loss": 0.02351599559187889,
      "step": 4766
    },
    {
      "epoch": 0.9037149355572404,
      "grad_norm": 0.1868634968996048,
      "learning_rate": 0.0001223561430793157,
      "loss": 0.017168838530778885,
      "step": 4768
    },
    {
      "epoch": 0.9040940106141016,
      "grad_norm": 0.08891164511442184,
      "learning_rate": 0.00012227838258164851,
      "loss": 0.01923687383532524,
      "step": 4770
    },
    {
      "epoch": 0.9044730856709629,
      "grad_norm": 0.25330910086631775,
      "learning_rate": 0.00012220062208398135,
      "loss": 0.017186865210533142,
      "step": 4772
    },
    {
      "epoch": 0.9048521607278242,
      "grad_norm": 0.10209764540195465,
      "learning_rate": 0.00012212286158631416,
      "loss": 0.022540083155035973,
      "step": 4774
    },
    {
      "epoch": 0.9052312357846853,
      "grad_norm": 0.09469939023256302,
      "learning_rate": 0.00012204510108864698,
      "loss": 0.02042205259203911,
      "step": 4776
    },
    {
      "epoch": 0.9056103108415466,
      "grad_norm": 0.11047276109457016,
      "learning_rate": 0.00012196734059097978,
      "loss": 0.015126630663871765,
      "step": 4778
    },
    {
      "epoch": 0.9059893858984079,
      "grad_norm": 0.3405984342098236,
      "learning_rate": 0.00012188958009331259,
      "loss": 0.043596744537353516,
      "step": 4780
    },
    {
      "epoch": 0.9063684609552691,
      "grad_norm": 0.1094244122505188,
      "learning_rate": 0.00012181181959564542,
      "loss": 0.015975045040249825,
      "step": 4782
    },
    {
      "epoch": 0.9067475360121304,
      "grad_norm": 0.11506441980600357,
      "learning_rate": 0.00012173405909797823,
      "loss": 0.017167771235108376,
      "step": 4784
    },
    {
      "epoch": 0.9071266110689916,
      "grad_norm": 0.3079836964607239,
      "learning_rate": 0.00012165629860031105,
      "loss": 0.020778505131602287,
      "step": 4786
    },
    {
      "epoch": 0.9075056861258529,
      "grad_norm": 0.43154582381248474,
      "learning_rate": 0.00012157853810264386,
      "loss": 0.03496735543012619,
      "step": 4788
    },
    {
      "epoch": 0.9078847611827142,
      "grad_norm": 0.19403250515460968,
      "learning_rate": 0.00012150077760497666,
      "loss": 0.01556428987532854,
      "step": 4790
    },
    {
      "epoch": 0.9082638362395754,
      "grad_norm": 0.3485799729824066,
      "learning_rate": 0.0001214230171073095,
      "loss": 0.018308809027075768,
      "step": 4792
    },
    {
      "epoch": 0.9086429112964367,
      "grad_norm": 0.28087955713272095,
      "learning_rate": 0.0001213452566096423,
      "loss": 0.02375912293791771,
      "step": 4794
    },
    {
      "epoch": 0.909021986353298,
      "grad_norm": 0.29939934611320496,
      "learning_rate": 0.00012126749611197513,
      "loss": 0.025254981592297554,
      "step": 4796
    },
    {
      "epoch": 0.9094010614101592,
      "grad_norm": 0.09879150241613388,
      "learning_rate": 0.00012118973561430793,
      "loss": 0.012360273860394955,
      "step": 4798
    },
    {
      "epoch": 0.9097801364670205,
      "grad_norm": 0.11423557996749878,
      "learning_rate": 0.00012111197511664074,
      "loss": 0.030264750123023987,
      "step": 4800
    },
    {
      "epoch": 0.9101592115238817,
      "grad_norm": 0.30111032724380493,
      "learning_rate": 0.00012103421461897357,
      "loss": 0.02435140497982502,
      "step": 4802
    },
    {
      "epoch": 0.910538286580743,
      "grad_norm": 0.08741040527820587,
      "learning_rate": 0.00012095645412130638,
      "loss": 0.013593969866633415,
      "step": 4804
    },
    {
      "epoch": 0.9109173616376043,
      "grad_norm": 0.4134092628955841,
      "learning_rate": 0.00012087869362363919,
      "loss": 0.04107526317238808,
      "step": 4806
    },
    {
      "epoch": 0.9112964366944655,
      "grad_norm": 0.37939226627349854,
      "learning_rate": 0.00012080093312597201,
      "loss": 0.04229620844125748,
      "step": 4808
    },
    {
      "epoch": 0.9116755117513268,
      "grad_norm": 0.305838942527771,
      "learning_rate": 0.00012072317262830484,
      "loss": 0.028919409960508347,
      "step": 4810
    },
    {
      "epoch": 0.9120545868081881,
      "grad_norm": 0.28467026352882385,
      "learning_rate": 0.00012064541213063765,
      "loss": 0.019650034606456757,
      "step": 4812
    },
    {
      "epoch": 0.9124336618650493,
      "grad_norm": 0.26889175176620483,
      "learning_rate": 0.00012056765163297046,
      "loss": 0.019862528890371323,
      "step": 4814
    },
    {
      "epoch": 0.9128127369219106,
      "grad_norm": 0.18718042969703674,
      "learning_rate": 0.00012048989113530326,
      "loss": 0.016704890877008438,
      "step": 4816
    },
    {
      "epoch": 0.9131918119787718,
      "grad_norm": 0.3286634385585785,
      "learning_rate": 0.00012041213063763608,
      "loss": 0.03511321172118187,
      "step": 4818
    },
    {
      "epoch": 0.913570887035633,
      "grad_norm": 0.1390841156244278,
      "learning_rate": 0.00012033437013996892,
      "loss": 0.017997337505221367,
      "step": 4820
    },
    {
      "epoch": 0.9139499620924944,
      "grad_norm": 0.20266053080558777,
      "learning_rate": 0.00012025660964230172,
      "loss": 0.02322385087609291,
      "step": 4822
    },
    {
      "epoch": 0.9143290371493555,
      "grad_norm": 0.2996003329753876,
      "learning_rate": 0.00012017884914463453,
      "loss": 0.02220533788204193,
      "step": 4824
    },
    {
      "epoch": 0.9147081122062168,
      "grad_norm": 0.16885150969028473,
      "learning_rate": 0.00012010108864696734,
      "loss": 0.014305281452834606,
      "step": 4826
    },
    {
      "epoch": 0.9150871872630781,
      "grad_norm": 0.18584294617176056,
      "learning_rate": 0.00012002332814930016,
      "loss": 0.01324379537254572,
      "step": 4828
    },
    {
      "epoch": 0.9154662623199393,
      "grad_norm": 0.11375819146633148,
      "learning_rate": 0.00011994556765163298,
      "loss": 0.014249395579099655,
      "step": 4830
    },
    {
      "epoch": 0.9158453373768006,
      "grad_norm": 0.43544653058052063,
      "learning_rate": 0.0001198678071539658,
      "loss": 0.03134123608469963,
      "step": 4832
    },
    {
      "epoch": 0.9162244124336618,
      "grad_norm": 0.11616012454032898,
      "learning_rate": 0.0001197900466562986,
      "loss": 0.020563887432217598,
      "step": 4834
    },
    {
      "epoch": 0.9166034874905231,
      "grad_norm": 0.09021292626857758,
      "learning_rate": 0.00011971228615863141,
      "loss": 0.013597190380096436,
      "step": 4836
    },
    {
      "epoch": 0.9169825625473844,
      "grad_norm": 0.37724170088768005,
      "learning_rate": 0.00011963452566096422,
      "loss": 0.022771576419472694,
      "step": 4838
    },
    {
      "epoch": 0.9173616376042456,
      "grad_norm": 0.34938350319862366,
      "learning_rate": 0.00011955676516329705,
      "loss": 0.04184842109680176,
      "step": 4840
    },
    {
      "epoch": 0.9177407126611069,
      "grad_norm": 0.08478841930627823,
      "learning_rate": 0.00011947900466562987,
      "loss": 0.01571592688560486,
      "step": 4842
    },
    {
      "epoch": 0.9181197877179682,
      "grad_norm": 0.08701690286397934,
      "learning_rate": 0.00011940124416796268,
      "loss": 0.011752170510590076,
      "step": 4844
    },
    {
      "epoch": 0.9184988627748294,
      "grad_norm": 0.07776515185832977,
      "learning_rate": 0.00011932348367029549,
      "loss": 0.023012813180685043,
      "step": 4846
    },
    {
      "epoch": 0.9188779378316907,
      "grad_norm": 0.3229367136955261,
      "learning_rate": 0.00011924572317262832,
      "loss": 0.014834106899797916,
      "step": 4848
    },
    {
      "epoch": 0.9192570128885519,
      "grad_norm": 0.08595360070466995,
      "learning_rate": 0.00011916796267496113,
      "loss": 0.012587101198732853,
      "step": 4850
    },
    {
      "epoch": 0.9196360879454132,
      "grad_norm": 0.23178155720233917,
      "learning_rate": 0.00011909020217729394,
      "loss": 0.012188398279249668,
      "step": 4852
    },
    {
      "epoch": 0.9200151630022745,
      "grad_norm": 0.29562345147132874,
      "learning_rate": 0.00011901244167962676,
      "loss": 0.03794550895690918,
      "step": 4854
    },
    {
      "epoch": 0.9203942380591357,
      "grad_norm": 0.10319071263074875,
      "learning_rate": 0.00011893468118195956,
      "loss": 0.011929251253604889,
      "step": 4856
    },
    {
      "epoch": 0.920773313115997,
      "grad_norm": 0.11863153427839279,
      "learning_rate": 0.0001188569206842924,
      "loss": 0.021210823208093643,
      "step": 4858
    },
    {
      "epoch": 0.9211523881728583,
      "grad_norm": 0.09216062724590302,
      "learning_rate": 0.0001187791601866252,
      "loss": 0.01195535808801651,
      "step": 4860
    },
    {
      "epoch": 0.9215314632297195,
      "grad_norm": 0.4216978847980499,
      "learning_rate": 0.00011870139968895801,
      "loss": 0.0414099246263504,
      "step": 4862
    },
    {
      "epoch": 0.9219105382865808,
      "grad_norm": 0.27651602029800415,
      "learning_rate": 0.00011862363919129083,
      "loss": 0.03660212457180023,
      "step": 4864
    },
    {
      "epoch": 0.922289613343442,
      "grad_norm": 0.20093125104904175,
      "learning_rate": 0.00011854587869362364,
      "loss": 0.01743401400744915,
      "step": 4866
    },
    {
      "epoch": 0.9226686884003032,
      "grad_norm": 0.34571871161460876,
      "learning_rate": 0.00011846811819595647,
      "loss": 0.02855294570326805,
      "step": 4868
    },
    {
      "epoch": 0.9230477634571645,
      "grad_norm": 0.08522697538137436,
      "learning_rate": 0.00011839035769828928,
      "loss": 0.01885554939508438,
      "step": 4870
    },
    {
      "epoch": 0.9234268385140257,
      "grad_norm": 0.0780453085899353,
      "learning_rate": 0.00011831259720062209,
      "loss": 0.016927143558859825,
      "step": 4872
    },
    {
      "epoch": 0.923805913570887,
      "grad_norm": 0.14450372755527496,
      "learning_rate": 0.00011823483670295489,
      "loss": 0.013237196952104568,
      "step": 4874
    },
    {
      "epoch": 0.9241849886277483,
      "grad_norm": 0.09669701009988785,
      "learning_rate": 0.00011815707620528771,
      "loss": 0.016254419460892677,
      "step": 4876
    },
    {
      "epoch": 0.9245640636846095,
      "grad_norm": 0.19501034915447235,
      "learning_rate": 0.00011807931570762055,
      "loss": 0.014222484081983566,
      "step": 4878
    },
    {
      "epoch": 0.9249431387414708,
      "grad_norm": 0.25318852066993713,
      "learning_rate": 0.00011800155520995335,
      "loss": 0.023323725908994675,
      "step": 4880
    },
    {
      "epoch": 0.925322213798332,
      "grad_norm": 0.27721381187438965,
      "learning_rate": 0.00011792379471228616,
      "loss": 0.027046924456954002,
      "step": 4882
    },
    {
      "epoch": 0.9257012888551933,
      "grad_norm": 0.06618180125951767,
      "learning_rate": 0.00011784603421461897,
      "loss": 0.03125859797000885,
      "step": 4884
    },
    {
      "epoch": 0.9260803639120546,
      "grad_norm": 0.2952975630760193,
      "learning_rate": 0.0001177682737169518,
      "loss": 0.027029238641262054,
      "step": 4886
    },
    {
      "epoch": 0.9264594389689158,
      "grad_norm": 0.40849998593330383,
      "learning_rate": 0.00011769051321928461,
      "loss": 0.03215637430548668,
      "step": 4888
    },
    {
      "epoch": 0.9268385140257771,
      "grad_norm": 0.3802756369113922,
      "learning_rate": 0.00011761275272161743,
      "loss": 0.026449324563145638,
      "step": 4890
    },
    {
      "epoch": 0.9272175890826384,
      "grad_norm": 0.11265728622674942,
      "learning_rate": 0.00011753499222395024,
      "loss": 0.03121352568268776,
      "step": 4892
    },
    {
      "epoch": 0.9275966641394996,
      "grad_norm": 0.21977321803569794,
      "learning_rate": 0.00011745723172628304,
      "loss": 0.02111108787357807,
      "step": 4894
    },
    {
      "epoch": 0.9279757391963609,
      "grad_norm": 0.10784134268760681,
      "learning_rate": 0.00011737947122861588,
      "loss": 0.012711403891444206,
      "step": 4896
    },
    {
      "epoch": 0.9283548142532221,
      "grad_norm": 0.3026321828365326,
      "learning_rate": 0.00011730171073094868,
      "loss": 0.03130694106221199,
      "step": 4898
    },
    {
      "epoch": 0.9287338893100834,
      "grad_norm": 0.11288869380950928,
      "learning_rate": 0.0001172239502332815,
      "loss": 0.025479581207036972,
      "step": 4900
    },
    {
      "epoch": 0.9291129643669447,
      "grad_norm": 0.0802822932600975,
      "learning_rate": 0.00011714618973561431,
      "loss": 0.020771805197000504,
      "step": 4902
    },
    {
      "epoch": 0.9294920394238059,
      "grad_norm": 0.4269200563430786,
      "learning_rate": 0.00011706842923794712,
      "loss": 0.03204229474067688,
      "step": 4904
    },
    {
      "epoch": 0.9298711144806672,
      "grad_norm": 0.07792746275663376,
      "learning_rate": 0.00011699066874027995,
      "loss": 0.015683870762586594,
      "step": 4906
    },
    {
      "epoch": 0.9302501895375285,
      "grad_norm": 0.33430904150009155,
      "learning_rate": 0.00011691290824261276,
      "loss": 0.022286659106612206,
      "step": 4908
    },
    {
      "epoch": 0.9306292645943897,
      "grad_norm": 0.08044560998678207,
      "learning_rate": 0.00011683514774494557,
      "loss": 0.013169285841286182,
      "step": 4910
    },
    {
      "epoch": 0.931008339651251,
      "grad_norm": 0.11767961084842682,
      "learning_rate": 0.00011675738724727839,
      "loss": 0.0176282599568367,
      "step": 4912
    },
    {
      "epoch": 0.9313874147081123,
      "grad_norm": 0.2808260917663574,
      "learning_rate": 0.00011667962674961119,
      "loss": 0.03198181092739105,
      "step": 4914
    },
    {
      "epoch": 0.9317664897649734,
      "grad_norm": 0.0927247405052185,
      "learning_rate": 0.00011660186625194403,
      "loss": 0.01780868135392666,
      "step": 4916
    },
    {
      "epoch": 0.9321455648218347,
      "grad_norm": 0.11085844784975052,
      "learning_rate": 0.00011652410575427683,
      "loss": 0.01625320315361023,
      "step": 4918
    },
    {
      "epoch": 0.9325246398786959,
      "grad_norm": 0.15586325526237488,
      "learning_rate": 0.00011644634525660964,
      "loss": 0.022082053124904633,
      "step": 4920
    },
    {
      "epoch": 0.9329037149355572,
      "grad_norm": 0.3556426763534546,
      "learning_rate": 0.00011636858475894246,
      "loss": 0.02617257833480835,
      "step": 4922
    },
    {
      "epoch": 0.9332827899924185,
      "grad_norm": 0.2605738639831543,
      "learning_rate": 0.00011629082426127528,
      "loss": 0.019233493134379387,
      "step": 4924
    },
    {
      "epoch": 0.9336618650492797,
      "grad_norm": 0.09175051003694534,
      "learning_rate": 0.0001162130637636081,
      "loss": 0.0223687831312418,
      "step": 4926
    },
    {
      "epoch": 0.934040940106141,
      "grad_norm": 0.3446679711341858,
      "learning_rate": 0.00011613530326594091,
      "loss": 0.029870562255382538,
      "step": 4928
    },
    {
      "epoch": 0.9344200151630023,
      "grad_norm": 0.4167662262916565,
      "learning_rate": 0.00011605754276827372,
      "loss": 0.028264150023460388,
      "step": 4930
    },
    {
      "epoch": 0.9347990902198635,
      "grad_norm": 0.1155393198132515,
      "learning_rate": 0.00011597978227060652,
      "loss": 0.012684740126132965,
      "step": 4932
    },
    {
      "epoch": 0.9351781652767248,
      "grad_norm": 0.0734013095498085,
      "learning_rate": 0.00011590202177293936,
      "loss": 0.014267846941947937,
      "step": 4934
    },
    {
      "epoch": 0.935557240333586,
      "grad_norm": 0.33633556962013245,
      "learning_rate": 0.00011582426127527218,
      "loss": 0.01716134324669838,
      "step": 4936
    },
    {
      "epoch": 0.9359363153904473,
      "grad_norm": 0.10030391067266464,
      "learning_rate": 0.00011574650077760498,
      "loss": 0.023834167048335075,
      "step": 4938
    },
    {
      "epoch": 0.9363153904473086,
      "grad_norm": 0.38975581526756287,
      "learning_rate": 0.00011566874027993779,
      "loss": 0.025169335305690765,
      "step": 4940
    },
    {
      "epoch": 0.9366944655041698,
      "grad_norm": 0.4263734817504883,
      "learning_rate": 0.0001155909797822706,
      "loss": 0.04931062459945679,
      "step": 4942
    },
    {
      "epoch": 0.9370735405610311,
      "grad_norm": 0.40675443410873413,
      "learning_rate": 0.00011551321928460343,
      "loss": 0.036400727927684784,
      "step": 4944
    },
    {
      "epoch": 0.9374526156178924,
      "grad_norm": 0.08500052243471146,
      "learning_rate": 0.00011543545878693625,
      "loss": 0.02551363967359066,
      "step": 4946
    },
    {
      "epoch": 0.9378316906747536,
      "grad_norm": 0.35708948969841003,
      "learning_rate": 0.00011535769828926906,
      "loss": 0.03289925679564476,
      "step": 4948
    },
    {
      "epoch": 0.9382107657316149,
      "grad_norm": 0.12973976135253906,
      "learning_rate": 0.00011527993779160187,
      "loss": 0.020301323384046555,
      "step": 4950
    },
    {
      "epoch": 0.9385898407884761,
      "grad_norm": 0.10074444860219955,
      "learning_rate": 0.00011520217729393467,
      "loss": 0.019390100613236427,
      "step": 4952
    },
    {
      "epoch": 0.9389689158453374,
      "grad_norm": 0.34859681129455566,
      "learning_rate": 0.00011512441679626751,
      "loss": 0.040207307785749435,
      "step": 4954
    },
    {
      "epoch": 0.9393479909021987,
      "grad_norm": 0.09601709246635437,
      "learning_rate": 0.00011504665629860031,
      "loss": 0.022962331771850586,
      "step": 4956
    },
    {
      "epoch": 0.9397270659590599,
      "grad_norm": 0.502491295337677,
      "learning_rate": 0.00011496889580093313,
      "loss": 0.04376460611820221,
      "step": 4958
    },
    {
      "epoch": 0.9401061410159212,
      "grad_norm": 0.3469422459602356,
      "learning_rate": 0.00011489113530326594,
      "loss": 0.028952213004231453,
      "step": 4960
    },
    {
      "epoch": 0.9404852160727825,
      "grad_norm": 0.12607219815254211,
      "learning_rate": 0.00011481337480559878,
      "loss": 0.028032666072249413,
      "step": 4962
    },
    {
      "epoch": 0.9408642911296436,
      "grad_norm": 0.244042307138443,
      "learning_rate": 0.00011473561430793158,
      "loss": 0.019227268174290657,
      "step": 4964
    },
    {
      "epoch": 0.941243366186505,
      "grad_norm": 0.09190215170383453,
      "learning_rate": 0.00011465785381026439,
      "loss": 0.01900441385805607,
      "step": 4966
    },
    {
      "epoch": 0.9416224412433661,
      "grad_norm": 0.08453391492366791,
      "learning_rate": 0.00011458009331259721,
      "loss": 0.017019430175423622,
      "step": 4968
    },
    {
      "epoch": 0.9420015163002274,
      "grad_norm": 0.27463841438293457,
      "learning_rate": 0.00011450233281493002,
      "loss": 0.020001022145152092,
      "step": 4970
    },
    {
      "epoch": 0.9423805913570887,
      "grad_norm": 0.3555813431739807,
      "learning_rate": 0.00011442457231726285,
      "loss": 0.02939087152481079,
      "step": 4972
    },
    {
      "epoch": 0.9427596664139499,
      "grad_norm": 0.08473888039588928,
      "learning_rate": 0.00011434681181959566,
      "loss": 0.013947902247309685,
      "step": 4974
    },
    {
      "epoch": 0.9431387414708112,
      "grad_norm": 0.09320195019245148,
      "learning_rate": 0.00011426905132192846,
      "loss": 0.016414828598499298,
      "step": 4976
    },
    {
      "epoch": 0.9435178165276725,
      "grad_norm": 0.08953038603067398,
      "learning_rate": 0.00011419129082426127,
      "loss": 0.027495810762047768,
      "step": 4978
    },
    {
      "epoch": 0.9438968915845337,
      "grad_norm": 0.23818883299827576,
      "learning_rate": 0.00011411353032659409,
      "loss": 0.0298098586499691,
      "step": 4980
    },
    {
      "epoch": 0.944275966641395,
      "grad_norm": 0.4726382791996002,
      "learning_rate": 0.00011403576982892693,
      "loss": 0.04117533564567566,
      "step": 4982
    },
    {
      "epoch": 0.9446550416982562,
      "grad_norm": 0.3019396960735321,
      "learning_rate": 0.00011395800933125973,
      "loss": 0.02670246921479702,
      "step": 4984
    },
    {
      "epoch": 0.9450341167551175,
      "grad_norm": 0.46869465708732605,
      "learning_rate": 0.00011388024883359254,
      "loss": 0.029262367635965347,
      "step": 4986
    },
    {
      "epoch": 0.9454131918119788,
      "grad_norm": 0.0874653160572052,
      "learning_rate": 0.00011380248833592535,
      "loss": 0.012979386374354362,
      "step": 4988
    },
    {
      "epoch": 0.94579226686884,
      "grad_norm": 0.3521135151386261,
      "learning_rate": 0.00011372472783825817,
      "loss": 0.02521304041147232,
      "step": 4990
    },
    {
      "epoch": 0.9461713419257013,
      "grad_norm": 0.3190361261367798,
      "learning_rate": 0.00011364696734059099,
      "loss": 0.02484937198460102,
      "step": 4992
    },
    {
      "epoch": 0.9465504169825626,
      "grad_norm": 0.3855615258216858,
      "learning_rate": 0.00011356920684292381,
      "loss": 0.028783123940229416,
      "step": 4994
    },
    {
      "epoch": 0.9469294920394238,
      "grad_norm": 0.0885850116610527,
      "learning_rate": 0.00011349144634525661,
      "loss": 0.017389824613928795,
      "step": 4996
    },
    {
      "epoch": 0.9473085670962851,
      "grad_norm": 0.08134520798921585,
      "learning_rate": 0.00011341368584758942,
      "loss": 0.01997191645205021,
      "step": 4998
    },
    {
      "epoch": 0.9476876421531463,
      "grad_norm": 0.13023529946804047,
      "learning_rate": 0.00011333592534992223,
      "loss": 0.025187915191054344,
      "step": 5000
    },
    {
      "epoch": 0.9480667172100076,
      "grad_norm": 0.10457002371549606,
      "learning_rate": 0.00011325816485225506,
      "loss": 0.03111450746655464,
      "step": 5002
    },
    {
      "epoch": 0.9484457922668689,
      "grad_norm": 0.27922841906547546,
      "learning_rate": 0.00011318040435458788,
      "loss": 0.020082568749785423,
      "step": 5004
    },
    {
      "epoch": 0.9488248673237301,
      "grad_norm": 0.15552905201911926,
      "learning_rate": 0.00011310264385692069,
      "loss": 0.01844649761915207,
      "step": 5006
    },
    {
      "epoch": 0.9492039423805914,
      "grad_norm": 0.20978215336799622,
      "learning_rate": 0.0001130248833592535,
      "loss": 0.015090817585587502,
      "step": 5008
    },
    {
      "epoch": 0.9495830174374527,
      "grad_norm": 0.12382128089666367,
      "learning_rate": 0.00011294712286158633,
      "loss": 0.014536709524691105,
      "step": 5010
    },
    {
      "epoch": 0.9499620924943138,
      "grad_norm": 0.5698447823524475,
      "learning_rate": 0.00011286936236391914,
      "loss": 0.024188527837395668,
      "step": 5012
    },
    {
      "epoch": 0.9503411675511751,
      "grad_norm": 0.28963324427604675,
      "learning_rate": 0.00011279160186625194,
      "loss": 0.02130458876490593,
      "step": 5014
    },
    {
      "epoch": 0.9507202426080363,
      "grad_norm": 0.2185649424791336,
      "learning_rate": 0.00011271384136858476,
      "loss": 0.020127512514591217,
      "step": 5016
    },
    {
      "epoch": 0.9510993176648976,
      "grad_norm": 0.0951029434800148,
      "learning_rate": 0.00011263608087091757,
      "loss": 0.013359088450670242,
      "step": 5018
    },
    {
      "epoch": 0.9514783927217589,
      "grad_norm": 0.2716871500015259,
      "learning_rate": 0.0001125583203732504,
      "loss": 0.01759992726147175,
      "step": 5020
    },
    {
      "epoch": 0.9518574677786201,
      "grad_norm": 0.09681299328804016,
      "learning_rate": 0.00011248055987558321,
      "loss": 0.029949363321065903,
      "step": 5022
    },
    {
      "epoch": 0.9522365428354814,
      "grad_norm": 0.08749725669622421,
      "learning_rate": 0.00011240279937791602,
      "loss": 0.012676817364990711,
      "step": 5024
    },
    {
      "epoch": 0.9526156178923427,
      "grad_norm": 0.3324546813964844,
      "learning_rate": 0.00011232503888024884,
      "loss": 0.03872239589691162,
      "step": 5026
    },
    {
      "epoch": 0.9529946929492039,
      "grad_norm": 0.2964637875556946,
      "learning_rate": 0.00011224727838258165,
      "loss": 0.01702718809247017,
      "step": 5028
    },
    {
      "epoch": 0.9533737680060652,
      "grad_norm": 0.23935917019844055,
      "learning_rate": 0.00011216951788491448,
      "loss": 0.015202660113573074,
      "step": 5030
    },
    {
      "epoch": 0.9537528430629265,
      "grad_norm": 0.371750146150589,
      "learning_rate": 0.00011209175738724729,
      "loss": 0.021788623183965683,
      "step": 5032
    },
    {
      "epoch": 0.9541319181197877,
      "grad_norm": 0.31683415174484253,
      "learning_rate": 0.0001120139968895801,
      "loss": 0.03777461498975754,
      "step": 5034
    },
    {
      "epoch": 0.954510993176649,
      "grad_norm": 0.43718481063842773,
      "learning_rate": 0.0001119362363919129,
      "loss": 0.02754024788737297,
      "step": 5036
    },
    {
      "epoch": 0.9548900682335102,
      "grad_norm": 0.2776760160923004,
      "learning_rate": 0.00011185847589424572,
      "loss": 0.02827516384422779,
      "step": 5038
    },
    {
      "epoch": 0.9552691432903715,
      "grad_norm": 0.2924278974533081,
      "learning_rate": 0.00011178071539657856,
      "loss": 0.021006669849157333,
      "step": 5040
    },
    {
      "epoch": 0.9556482183472328,
      "grad_norm": 0.3307605981826782,
      "learning_rate": 0.00011170295489891136,
      "loss": 0.02098381333053112,
      "step": 5042
    },
    {
      "epoch": 0.956027293404094,
      "grad_norm": 0.0682869702577591,
      "learning_rate": 0.00011162519440124417,
      "loss": 0.029896492138504982,
      "step": 5044
    },
    {
      "epoch": 0.9564063684609553,
      "grad_norm": 0.2653925120830536,
      "learning_rate": 0.00011154743390357698,
      "loss": 0.01940985396504402,
      "step": 5046
    },
    {
      "epoch": 0.9567854435178166,
      "grad_norm": 0.36381885409355164,
      "learning_rate": 0.00011146967340590981,
      "loss": 0.02614188753068447,
      "step": 5048
    },
    {
      "epoch": 0.9571645185746778,
      "grad_norm": 0.09573698788881302,
      "learning_rate": 0.00011139191290824263,
      "loss": 0.02058625966310501,
      "step": 5050
    },
    {
      "epoch": 0.9575435936315391,
      "grad_norm": 0.3030441403388977,
      "learning_rate": 0.00011131415241057544,
      "loss": 0.021759964525699615,
      "step": 5052
    },
    {
      "epoch": 0.9579226686884003,
      "grad_norm": 0.40273770689964294,
      "learning_rate": 0.00011123639191290824,
      "loss": 0.03727290406823158,
      "step": 5054
    },
    {
      "epoch": 0.9583017437452616,
      "grad_norm": 0.25416821241378784,
      "learning_rate": 0.00011115863141524105,
      "loss": 0.024773750454187393,
      "step": 5056
    },
    {
      "epoch": 0.9586808188021229,
      "grad_norm": 0.10385920852422714,
      "learning_rate": 0.00011108087091757389,
      "loss": 0.014366071671247482,
      "step": 5058
    },
    {
      "epoch": 0.959059893858984,
      "grad_norm": 0.33972644805908203,
      "learning_rate": 0.00011100311041990669,
      "loss": 0.02037942409515381,
      "step": 5060
    },
    {
      "epoch": 0.9594389689158453,
      "grad_norm": 0.505351185798645,
      "learning_rate": 0.00011092534992223951,
      "loss": 0.05049866810441017,
      "step": 5062
    },
    {
      "epoch": 0.9598180439727066,
      "grad_norm": 0.09672936797142029,
      "learning_rate": 0.00011084758942457232,
      "loss": 0.012613728642463684,
      "step": 5064
    },
    {
      "epoch": 0.9601971190295678,
      "grad_norm": 0.32539182901382446,
      "learning_rate": 0.00011076982892690513,
      "loss": 0.028203649446368217,
      "step": 5066
    },
    {
      "epoch": 0.9605761940864291,
      "grad_norm": 0.3812316656112671,
      "learning_rate": 0.00011069206842923796,
      "loss": 0.03362613543868065,
      "step": 5068
    },
    {
      "epoch": 0.9609552691432903,
      "grad_norm": 0.37011921405792236,
      "learning_rate": 0.00011061430793157077,
      "loss": 0.037951547652482986,
      "step": 5070
    },
    {
      "epoch": 0.9613343442001516,
      "grad_norm": 0.08185549825429916,
      "learning_rate": 0.00011053654743390359,
      "loss": 0.024817049503326416,
      "step": 5072
    },
    {
      "epoch": 0.9617134192570129,
      "grad_norm": 0.09768442064523697,
      "learning_rate": 0.0001104587869362364,
      "loss": 0.020126938819885254,
      "step": 5074
    },
    {
      "epoch": 0.9620924943138741,
      "grad_norm": 0.2025967240333557,
      "learning_rate": 0.0001103810264385692,
      "loss": 0.03388335555791855,
      "step": 5076
    },
    {
      "epoch": 0.9624715693707354,
      "grad_norm": 0.21754279732704163,
      "learning_rate": 0.00011030326594090204,
      "loss": 0.023501262068748474,
      "step": 5078
    },
    {
      "epoch": 0.9628506444275967,
      "grad_norm": 0.42746129631996155,
      "learning_rate": 0.00011022550544323484,
      "loss": 0.03563623130321503,
      "step": 5080
    },
    {
      "epoch": 0.9632297194844579,
      "grad_norm": 0.11843869835138321,
      "learning_rate": 0.00011014774494556765,
      "loss": 0.014318597503006458,
      "step": 5082
    },
    {
      "epoch": 0.9636087945413192,
      "grad_norm": 0.3887012302875519,
      "learning_rate": 0.00011006998444790047,
      "loss": 0.042609501630067825,
      "step": 5084
    },
    {
      "epoch": 0.9639878695981804,
      "grad_norm": 0.3736778497695923,
      "learning_rate": 0.0001099922239502333,
      "loss": 0.023569846525788307,
      "step": 5086
    },
    {
      "epoch": 0.9643669446550417,
      "grad_norm": 0.35726284980773926,
      "learning_rate": 0.00010991446345256611,
      "loss": 0.0241637472063303,
      "step": 5088
    },
    {
      "epoch": 0.964746019711903,
      "grad_norm": 0.31790998578071594,
      "learning_rate": 0.00010983670295489892,
      "loss": 0.022250963374972343,
      "step": 5090
    },
    {
      "epoch": 0.9651250947687642,
      "grad_norm": 0.3782654404640198,
      "learning_rate": 0.00010975894245723172,
      "loss": 0.03798791021108627,
      "step": 5092
    },
    {
      "epoch": 0.9655041698256255,
      "grad_norm": 0.38757380843162537,
      "learning_rate": 0.00010968118195956454,
      "loss": 0.034112073481082916,
      "step": 5094
    },
    {
      "epoch": 0.9658832448824868,
      "grad_norm": 0.34900638461112976,
      "learning_rate": 0.00010960342146189737,
      "loss": 0.04849625751376152,
      "step": 5096
    },
    {
      "epoch": 0.966262319939348,
      "grad_norm": 0.16582651436328888,
      "learning_rate": 0.00010952566096423019,
      "loss": 0.02440313622355461,
      "step": 5098
    },
    {
      "epoch": 0.9666413949962093,
      "grad_norm": 0.07859539985656738,
      "learning_rate": 0.00010944790046656299,
      "loss": 0.011746378615498543,
      "step": 5100
    },
    {
      "epoch": 0.9670204700530705,
      "grad_norm": 0.25690212845802307,
      "learning_rate": 0.0001093701399688958,
      "loss": 0.02371961437165737,
      "step": 5102
    },
    {
      "epoch": 0.9673995451099318,
      "grad_norm": 0.27259546518325806,
      "learning_rate": 0.0001092923794712286,
      "loss": 0.02430342324078083,
      "step": 5104
    },
    {
      "epoch": 0.9677786201667931,
      "grad_norm": 0.10571956634521484,
      "learning_rate": 0.00010921461897356144,
      "loss": 0.01694103144109249,
      "step": 5106
    },
    {
      "epoch": 0.9681576952236542,
      "grad_norm": 0.07066389173269272,
      "learning_rate": 0.00010913685847589426,
      "loss": 0.022757116705179214,
      "step": 5108
    },
    {
      "epoch": 0.9685367702805155,
      "grad_norm": 0.10699370503425598,
      "learning_rate": 0.00010905909797822707,
      "loss": 0.014784898608922958,
      "step": 5110
    },
    {
      "epoch": 0.9689158453373768,
      "grad_norm": 0.2840251326560974,
      "learning_rate": 0.00010898133748055987,
      "loss": 0.029725532978773117,
      "step": 5112
    },
    {
      "epoch": 0.969294920394238,
      "grad_norm": 0.07909297198057175,
      "learning_rate": 0.00010890357698289268,
      "loss": 0.011602689512073994,
      "step": 5114
    },
    {
      "epoch": 0.9696739954510993,
      "grad_norm": 0.42402368783950806,
      "learning_rate": 0.00010882581648522552,
      "loss": 0.053045064210891724,
      "step": 5116
    },
    {
      "epoch": 0.9700530705079605,
      "grad_norm": 0.12848863005638123,
      "learning_rate": 0.00010874805598755832,
      "loss": 0.015862883999943733,
      "step": 5118
    },
    {
      "epoch": 0.9704321455648218,
      "grad_norm": 0.0823836401104927,
      "learning_rate": 0.00010867029548989114,
      "loss": 0.012148543260991573,
      "step": 5120
    },
    {
      "epoch": 0.9708112206216831,
      "grad_norm": 0.08886410295963287,
      "learning_rate": 0.00010859253499222395,
      "loss": 0.01745183952152729,
      "step": 5122
    },
    {
      "epoch": 0.9711902956785443,
      "grad_norm": 0.35072648525238037,
      "learning_rate": 0.00010851477449455678,
      "loss": 0.03437145799398422,
      "step": 5124
    },
    {
      "epoch": 0.9715693707354056,
      "grad_norm": 0.20363713800907135,
      "learning_rate": 0.00010843701399688959,
      "loss": 0.01808706484735012,
      "step": 5126
    },
    {
      "epoch": 0.9719484457922669,
      "grad_norm": 0.41580313444137573,
      "learning_rate": 0.0001083592534992224,
      "loss": 0.03759952634572983,
      "step": 5128
    },
    {
      "epoch": 0.9723275208491281,
      "grad_norm": 0.0798427015542984,
      "learning_rate": 0.00010828149300155522,
      "loss": 0.011847133748233318,
      "step": 5130
    },
    {
      "epoch": 0.9727065959059894,
      "grad_norm": 0.3626061677932739,
      "learning_rate": 0.00010820373250388802,
      "loss": 0.039734434336423874,
      "step": 5132
    },
    {
      "epoch": 0.9730856709628506,
      "grad_norm": 0.4354677200317383,
      "learning_rate": 0.00010812597200622086,
      "loss": 0.04169399291276932,
      "step": 5134
    },
    {
      "epoch": 0.9734647460197119,
      "grad_norm": 0.37803223729133606,
      "learning_rate": 0.00010804821150855367,
      "loss": 0.032371945679187775,
      "step": 5136
    },
    {
      "epoch": 0.9738438210765732,
      "grad_norm": 0.1055411621928215,
      "learning_rate": 0.00010797045101088647,
      "loss": 0.02618596702814102,
      "step": 5138
    },
    {
      "epoch": 0.9742228961334344,
      "grad_norm": 0.30195894837379456,
      "learning_rate": 0.00010789269051321928,
      "loss": 0.030563097447156906,
      "step": 5140
    },
    {
      "epoch": 0.9746019711902957,
      "grad_norm": 0.09725116938352585,
      "learning_rate": 0.0001078149300155521,
      "loss": 0.017334576696157455,
      "step": 5142
    },
    {
      "epoch": 0.974981046247157,
      "grad_norm": 0.23547662794589996,
      "learning_rate": 0.00010773716951788493,
      "loss": 0.015158798545598984,
      "step": 5144
    },
    {
      "epoch": 0.9753601213040182,
      "grad_norm": 0.11475508660078049,
      "learning_rate": 0.00010765940902021774,
      "loss": 0.017478909343481064,
      "step": 5146
    },
    {
      "epoch": 0.9757391963608795,
      "grad_norm": 0.2629486918449402,
      "learning_rate": 0.00010758164852255055,
      "loss": 0.024684175848960876,
      "step": 5148
    },
    {
      "epoch": 0.9761182714177407,
      "grad_norm": 0.37361711263656616,
      "learning_rate": 0.00010750388802488335,
      "loss": 0.023809321224689484,
      "step": 5150
    },
    {
      "epoch": 0.976497346474602,
      "grad_norm": 0.09375901520252228,
      "learning_rate": 0.00010742612752721617,
      "loss": 0.017935287207365036,
      "step": 5152
    },
    {
      "epoch": 0.9768764215314633,
      "grad_norm": 0.08857747167348862,
      "learning_rate": 0.000107348367029549,
      "loss": 0.02030780538916588,
      "step": 5154
    },
    {
      "epoch": 0.9772554965883244,
      "grad_norm": 0.12364040315151215,
      "learning_rate": 0.00010727060653188182,
      "loss": 0.035045333206653595,
      "step": 5156
    },
    {
      "epoch": 0.9776345716451857,
      "grad_norm": 0.11439073830842972,
      "learning_rate": 0.00010719284603421462,
      "loss": 0.02571970969438553,
      "step": 5158
    },
    {
      "epoch": 0.978013646702047,
      "grad_norm": 0.35007616877555847,
      "learning_rate": 0.00010711508553654743,
      "loss": 0.028213854879140854,
      "step": 5160
    },
    {
      "epoch": 0.9783927217589082,
      "grad_norm": 0.26746973395347595,
      "learning_rate": 0.00010703732503888024,
      "loss": 0.021159980446100235,
      "step": 5162
    },
    {
      "epoch": 0.9787717968157695,
      "grad_norm": 0.2779791057109833,
      "learning_rate": 0.00010695956454121307,
      "loss": 0.020681127905845642,
      "step": 5164
    },
    {
      "epoch": 0.9791508718726308,
      "grad_norm": 0.3885350823402405,
      "learning_rate": 0.00010688180404354589,
      "loss": 0.04036242514848709,
      "step": 5166
    },
    {
      "epoch": 0.979529946929492,
      "grad_norm": 0.09971494227647781,
      "learning_rate": 0.0001068040435458787,
      "loss": 0.02629981003701687,
      "step": 5168
    },
    {
      "epoch": 0.9799090219863533,
      "grad_norm": 0.32070523500442505,
      "learning_rate": 0.0001067262830482115,
      "loss": 0.02343890629708767,
      "step": 5170
    },
    {
      "epoch": 0.9802880970432145,
      "grad_norm": 0.11242836713790894,
      "learning_rate": 0.00010664852255054434,
      "loss": 0.01875661127269268,
      "step": 5172
    },
    {
      "epoch": 0.9806671721000758,
      "grad_norm": 0.17431242763996124,
      "learning_rate": 0.00010657076205287715,
      "loss": 0.013347710482776165,
      "step": 5174
    },
    {
      "epoch": 0.9810462471569371,
      "grad_norm": 0.4102349877357483,
      "learning_rate": 0.00010649300155520997,
      "loss": 0.04208451509475708,
      "step": 5176
    },
    {
      "epoch": 0.9814253222137983,
      "grad_norm": 0.10470772534608841,
      "learning_rate": 0.00010641524105754277,
      "loss": 0.019380759447813034,
      "step": 5178
    },
    {
      "epoch": 0.9818043972706596,
      "grad_norm": 0.44045424461364746,
      "learning_rate": 0.00010633748055987558,
      "loss": 0.03446047380566597,
      "step": 5180
    },
    {
      "epoch": 0.9821834723275209,
      "grad_norm": 0.24936836957931519,
      "learning_rate": 0.00010625972006220841,
      "loss": 0.0385277196764946,
      "step": 5182
    },
    {
      "epoch": 0.9825625473843821,
      "grad_norm": 0.33368444442749023,
      "learning_rate": 0.00010618195956454122,
      "loss": 0.030955299735069275,
      "step": 5184
    },
    {
      "epoch": 0.9829416224412434,
      "grad_norm": 0.31608888506889343,
      "learning_rate": 0.00010610419906687403,
      "loss": 0.04258919879794121,
      "step": 5186
    },
    {
      "epoch": 0.9833206974981046,
      "grad_norm": 0.42257624864578247,
      "learning_rate": 0.00010602643856920685,
      "loss": 0.041522618383169174,
      "step": 5188
    },
    {
      "epoch": 0.9836997725549659,
      "grad_norm": 0.30644699931144714,
      "learning_rate": 0.00010594867807153965,
      "loss": 0.028005465865135193,
      "step": 5190
    },
    {
      "epoch": 0.9840788476118272,
      "grad_norm": 0.3286208212375641,
      "learning_rate": 0.00010587091757387249,
      "loss": 0.036263126879930496,
      "step": 5192
    },
    {
      "epoch": 0.9844579226686884,
      "grad_norm": 0.2783266007900238,
      "learning_rate": 0.0001057931570762053,
      "loss": 0.03122895024716854,
      "step": 5194
    },
    {
      "epoch": 0.9848369977255497,
      "grad_norm": 0.2576582431793213,
      "learning_rate": 0.0001057153965785381,
      "loss": 0.02093229442834854,
      "step": 5196
    },
    {
      "epoch": 0.985216072782411,
      "grad_norm": 0.36161574721336365,
      "learning_rate": 0.00010563763608087092,
      "loss": 0.02427045814692974,
      "step": 5198
    },
    {
      "epoch": 0.9855951478392722,
      "grad_norm": 0.34615209698677063,
      "learning_rate": 0.00010555987558320373,
      "loss": 0.03776323050260544,
      "step": 5200
    },
    {
      "epoch": 0.9859742228961335,
      "grad_norm": 0.20531627535820007,
      "learning_rate": 0.00010548211508553656,
      "loss": 0.013983424752950668,
      "step": 5202
    },
    {
      "epoch": 0.9863532979529946,
      "grad_norm": 0.08513618260622025,
      "learning_rate": 0.00010540435458786937,
      "loss": 0.012115513905882835,
      "step": 5204
    },
    {
      "epoch": 0.986732373009856,
      "grad_norm": 0.2940901517868042,
      "learning_rate": 0.00010532659409020218,
      "loss": 0.017184380441904068,
      "step": 5206
    },
    {
      "epoch": 0.9871114480667172,
      "grad_norm": 0.38676923513412476,
      "learning_rate": 0.00010524883359253498,
      "loss": 0.034978121519088745,
      "step": 5208
    },
    {
      "epoch": 0.9874905231235784,
      "grad_norm": 0.08466393500566483,
      "learning_rate": 0.00010517107309486782,
      "loss": 0.011992692947387695,
      "step": 5210
    },
    {
      "epoch": 0.9878695981804397,
      "grad_norm": 0.11037158966064453,
      "learning_rate": 0.00010509331259720064,
      "loss": 0.026138486340641975,
      "step": 5212
    },
    {
      "epoch": 0.988248673237301,
      "grad_norm": 0.255740761756897,
      "learning_rate": 0.00010501555209953345,
      "loss": 0.02986755035817623,
      "step": 5214
    },
    {
      "epoch": 0.9886277482941622,
      "grad_norm": 0.4149527847766876,
      "learning_rate": 0.00010493779160186625,
      "loss": 0.03047347068786621,
      "step": 5216
    },
    {
      "epoch": 0.9890068233510235,
      "grad_norm": 0.31271860003471375,
      "learning_rate": 0.00010486003110419906,
      "loss": 0.01891297847032547,
      "step": 5218
    },
    {
      "epoch": 0.9893858984078847,
      "grad_norm": 0.08388005942106247,
      "learning_rate": 0.0001047822706065319,
      "loss": 0.012107453308999538,
      "step": 5220
    },
    {
      "epoch": 0.989764973464746,
      "grad_norm": 0.19855596125125885,
      "learning_rate": 0.0001047045101088647,
      "loss": 0.017232481390237808,
      "step": 5222
    },
    {
      "epoch": 0.9901440485216073,
      "grad_norm": 0.411757230758667,
      "learning_rate": 0.00010462674961119752,
      "loss": 0.03168807551264763,
      "step": 5224
    },
    {
      "epoch": 0.9905231235784685,
      "grad_norm": 0.24057811498641968,
      "learning_rate": 0.00010454898911353033,
      "loss": 0.01917405053973198,
      "step": 5226
    },
    {
      "epoch": 0.9909021986353298,
      "grad_norm": 0.27030786871910095,
      "learning_rate": 0.00010447122861586313,
      "loss": 0.027280142530798912,
      "step": 5228
    },
    {
      "epoch": 0.9912812736921911,
      "grad_norm": 0.3338140845298767,
      "learning_rate": 0.00010439346811819597,
      "loss": 0.024369923397898674,
      "step": 5230
    },
    {
      "epoch": 0.9916603487490523,
      "grad_norm": 0.39840272068977356,
      "learning_rate": 0.00010431570762052878,
      "loss": 0.033725086599588394,
      "step": 5232
    },
    {
      "epoch": 0.9920394238059136,
      "grad_norm": 0.3738676607608795,
      "learning_rate": 0.0001042379471228616,
      "loss": 0.027815235778689384,
      "step": 5234
    },
    {
      "epoch": 0.9924184988627748,
      "grad_norm": 0.3020363450050354,
      "learning_rate": 0.0001041601866251944,
      "loss": 0.021251026540994644,
      "step": 5236
    },
    {
      "epoch": 0.9927975739196361,
      "grad_norm": 0.1378682404756546,
      "learning_rate": 0.00010408242612752721,
      "loss": 0.019685346633195877,
      "step": 5238
    },
    {
      "epoch": 0.9931766489764974,
      "grad_norm": 0.4341296851634979,
      "learning_rate": 0.00010400466562986004,
      "loss": 0.032224178314208984,
      "step": 5240
    },
    {
      "epoch": 0.9935557240333586,
      "grad_norm": 0.10482027381658554,
      "learning_rate": 0.00010392690513219285,
      "loss": 0.018327394500374794,
      "step": 5242
    },
    {
      "epoch": 0.9939347990902199,
      "grad_norm": 0.43545499444007874,
      "learning_rate": 0.00010384914463452566,
      "loss": 0.028097335249185562,
      "step": 5244
    },
    {
      "epoch": 0.9943138741470812,
      "grad_norm": 0.40478992462158203,
      "learning_rate": 0.00010377138413685848,
      "loss": 0.033817604184150696,
      "step": 5246
    },
    {
      "epoch": 0.9946929492039424,
      "grad_norm": 0.2815457880496979,
      "learning_rate": 0.00010369362363919131,
      "loss": 0.01797349750995636,
      "step": 5248
    },
    {
      "epoch": 0.9950720242608037,
      "grad_norm": 0.34078946709632874,
      "learning_rate": 0.00010361586314152412,
      "loss": 0.03693264722824097,
      "step": 5250
    },
    {
      "epoch": 0.9954510993176648,
      "grad_norm": 0.23021046817302704,
      "learning_rate": 0.00010353810264385693,
      "loss": 0.026245003566145897,
      "step": 5252
    },
    {
      "epoch": 0.9958301743745261,
      "grad_norm": 0.3742973804473877,
      "learning_rate": 0.00010346034214618973,
      "loss": 0.03267374634742737,
      "step": 5254
    },
    {
      "epoch": 0.9962092494313874,
      "grad_norm": 0.35034871101379395,
      "learning_rate": 0.00010338258164852255,
      "loss": 0.03121718019247055,
      "step": 5256
    },
    {
      "epoch": 0.9965883244882486,
      "grad_norm": 0.3997228741645813,
      "learning_rate": 0.00010330482115085537,
      "loss": 0.029377296566963196,
      "step": 5258
    },
    {
      "epoch": 0.9969673995451099,
      "grad_norm": 0.09984969347715378,
      "learning_rate": 0.0001032270606531882,
      "loss": 0.011166119016706944,
      "step": 5260
    },
    {
      "epoch": 0.9973464746019712,
      "grad_norm": 0.06718039512634277,
      "learning_rate": 0.000103149300155521,
      "loss": 0.010532928630709648,
      "step": 5262
    },
    {
      "epoch": 0.9977255496588324,
      "grad_norm": 0.41275325417518616,
      "learning_rate": 0.00010307153965785381,
      "loss": 0.03394073247909546,
      "step": 5264
    },
    {
      "epoch": 0.9981046247156937,
      "grad_norm": 0.18102896213531494,
      "learning_rate": 0.00010299377916018661,
      "loss": 0.014103037305176258,
      "step": 5266
    },
    {
      "epoch": 0.9984836997725549,
      "grad_norm": 0.4602293074131012,
      "learning_rate": 0.00010291601866251945,
      "loss": 0.04648048058152199,
      "step": 5268
    },
    {
      "epoch": 0.9988627748294162,
      "grad_norm": 0.19581001996994019,
      "learning_rate": 0.00010283825816485227,
      "loss": 0.014690292067825794,
      "step": 5270
    },
    {
      "epoch": 0.9992418498862775,
      "grad_norm": 0.11498107016086578,
      "learning_rate": 0.00010276049766718508,
      "loss": 0.019092947244644165,
      "step": 5272
    },
    {
      "epoch": 0.9996209249431387,
      "grad_norm": 0.07375180721282959,
      "learning_rate": 0.00010268273716951788,
      "loss": 0.01560200098901987,
      "step": 5274
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.10705119371414185,
      "learning_rate": 0.00010260497667185069,
      "loss": 0.02429817058146,
      "step": 5276
    },
    {
      "epoch": 1.0003790750568613,
      "grad_norm": 0.33051198720932007,
      "learning_rate": 0.00010252721617418352,
      "loss": 0.030120331794023514,
      "step": 5278
    },
    {
      "epoch": 1.0007581501137226,
      "grad_norm": 0.10626332461833954,
      "learning_rate": 0.00010244945567651633,
      "loss": 0.013555437326431274,
      "step": 5280
    },
    {
      "epoch": 1.0011372251705837,
      "grad_norm": 0.18706682324409485,
      "learning_rate": 0.00010237169517884915,
      "loss": 0.016583802178502083,
      "step": 5282
    },
    {
      "epoch": 1.001516300227445,
      "grad_norm": 0.0997309684753418,
      "learning_rate": 0.00010229393468118196,
      "loss": 0.0184945035725832,
      "step": 5284
    },
    {
      "epoch": 1.0018953752843063,
      "grad_norm": 0.18488168716430664,
      "learning_rate": 0.00010221617418351479,
      "loss": 0.01575258933007717,
      "step": 5286
    },
    {
      "epoch": 1.0022744503411676,
      "grad_norm": 0.25104668736457825,
      "learning_rate": 0.0001021384136858476,
      "loss": 0.01852145791053772,
      "step": 5288
    },
    {
      "epoch": 1.0026535253980289,
      "grad_norm": 0.1297694742679596,
      "learning_rate": 0.0001020606531881804,
      "loss": 0.016968706622719765,
      "step": 5290
    },
    {
      "epoch": 1.0030326004548902,
      "grad_norm": 0.08537404239177704,
      "learning_rate": 0.00010198289269051323,
      "loss": 0.015101819299161434,
      "step": 5292
    },
    {
      "epoch": 1.0034116755117513,
      "grad_norm": 0.16213220357894897,
      "learning_rate": 0.00010190513219284603,
      "loss": 0.016047077253460884,
      "step": 5294
    },
    {
      "epoch": 1.0037907505686126,
      "grad_norm": 0.21308700740337372,
      "learning_rate": 0.00010182737169517887,
      "loss": 0.02581379935145378,
      "step": 5296
    },
    {
      "epoch": 1.0041698256254739,
      "grad_norm": 0.0958734080195427,
      "learning_rate": 0.00010174961119751167,
      "loss": 0.01994365267455578,
      "step": 5298
    },
    {
      "epoch": 1.0045489006823352,
      "grad_norm": 0.20222355425357819,
      "learning_rate": 0.00010167185069984448,
      "loss": 0.015549154952168465,
      "step": 5300
    },
    {
      "epoch": 1.0049279757391965,
      "grad_norm": 0.11312001198530197,
      "learning_rate": 0.0001015940902021773,
      "loss": 0.016151903197169304,
      "step": 5302
    },
    {
      "epoch": 1.0053070507960575,
      "grad_norm": 0.09201933443546295,
      "learning_rate": 0.00010151632970451011,
      "loss": 0.017569473013281822,
      "step": 5304
    },
    {
      "epoch": 1.0056861258529188,
      "grad_norm": 0.17264491319656372,
      "learning_rate": 0.00010143856920684294,
      "loss": 0.014760839752852917,
      "step": 5306
    },
    {
      "epoch": 1.0060652009097801,
      "grad_norm": 0.3580473065376282,
      "learning_rate": 0.00010136080870917575,
      "loss": 0.024774931371212006,
      "step": 5308
    },
    {
      "epoch": 1.0064442759666414,
      "grad_norm": 0.35445472598075867,
      "learning_rate": 0.00010128304821150856,
      "loss": 0.019793810322880745,
      "step": 5310
    },
    {
      "epoch": 1.0068233510235027,
      "grad_norm": 0.08993738889694214,
      "learning_rate": 0.00010120528771384136,
      "loss": 0.01496351882815361,
      "step": 5312
    },
    {
      "epoch": 1.0072024260803638,
      "grad_norm": 0.0969192236661911,
      "learning_rate": 0.00010112752721617418,
      "loss": 0.012909802608191967,
      "step": 5314
    },
    {
      "epoch": 1.007581501137225,
      "grad_norm": 0.1239321306347847,
      "learning_rate": 0.00010104976671850702,
      "loss": 0.01799040101468563,
      "step": 5316
    },
    {
      "epoch": 1.0079605761940864,
      "grad_norm": 0.2804524898529053,
      "learning_rate": 0.00010097200622083982,
      "loss": 0.021945593878626823,
      "step": 5318
    },
    {
      "epoch": 1.0083396512509477,
      "grad_norm": 0.0983278676867485,
      "learning_rate": 0.00010089424572317263,
      "loss": 0.01840817555785179,
      "step": 5320
    },
    {
      "epoch": 1.008718726307809,
      "grad_norm": 0.13340875506401062,
      "learning_rate": 0.00010081648522550544,
      "loss": 0.013309381902217865,
      "step": 5322
    },
    {
      "epoch": 1.0090978013646703,
      "grad_norm": 0.24816803634166718,
      "learning_rate": 0.00010073872472783827,
      "loss": 0.014539385214447975,
      "step": 5324
    },
    {
      "epoch": 1.0094768764215314,
      "grad_norm": 0.3742637038230896,
      "learning_rate": 0.00010066096423017108,
      "loss": 0.027009813115000725,
      "step": 5326
    },
    {
      "epoch": 1.0098559514783927,
      "grad_norm": 0.36516112089157104,
      "learning_rate": 0.0001005832037325039,
      "loss": 0.022133056074380875,
      "step": 5328
    },
    {
      "epoch": 1.010235026535254,
      "grad_norm": 0.09398195147514343,
      "learning_rate": 0.0001005054432348367,
      "loss": 0.011019312776625156,
      "step": 5330
    },
    {
      "epoch": 1.0106141015921153,
      "grad_norm": 0.0935821458697319,
      "learning_rate": 0.00010042768273716951,
      "loss": 0.012162554077804089,
      "step": 5332
    },
    {
      "epoch": 1.0109931766489766,
      "grad_norm": 0.10209573805332184,
      "learning_rate": 0.00010034992223950235,
      "loss": 0.013778790831565857,
      "step": 5334
    },
    {
      "epoch": 1.0113722517058377,
      "grad_norm": 0.07805859297513962,
      "learning_rate": 0.00010027216174183515,
      "loss": 0.012275461107492447,
      "step": 5336
    },
    {
      "epoch": 1.011751326762699,
      "grad_norm": 0.1806955486536026,
      "learning_rate": 0.00010019440124416797,
      "loss": 0.013442925177514553,
      "step": 5338
    },
    {
      "epoch": 1.0121304018195603,
      "grad_norm": 0.3016287386417389,
      "learning_rate": 0.00010011664074650078,
      "loss": 0.021802477538585663,
      "step": 5340
    },
    {
      "epoch": 1.0125094768764216,
      "grad_norm": 0.08333360403776169,
      "learning_rate": 0.00010003888024883359,
      "loss": 0.024443859234452248,
      "step": 5342
    },
    {
      "epoch": 1.0128885519332829,
      "grad_norm": 0.11549496650695801,
      "learning_rate": 9.996111975116641e-05,
      "loss": 0.018155548721551895,
      "step": 5344
    },
    {
      "epoch": 1.013267626990144,
      "grad_norm": 0.20001931488513947,
      "learning_rate": 9.988335925349923e-05,
      "loss": 0.022533593699336052,
      "step": 5346
    },
    {
      "epoch": 1.0136467020470052,
      "grad_norm": 0.12689585983753204,
      "learning_rate": 9.980559875583204e-05,
      "loss": 0.02106563188135624,
      "step": 5348
    },
    {
      "epoch": 1.0140257771038665,
      "grad_norm": 0.09621083736419678,
      "learning_rate": 9.972783825816486e-05,
      "loss": 0.012871823273599148,
      "step": 5350
    },
    {
      "epoch": 1.0144048521607278,
      "grad_norm": 0.3152386546134949,
      "learning_rate": 9.965007776049768e-05,
      "loss": 0.02544749155640602,
      "step": 5352
    },
    {
      "epoch": 1.0147839272175891,
      "grad_norm": 0.09882980585098267,
      "learning_rate": 9.957231726283048e-05,
      "loss": 0.016934579238295555,
      "step": 5354
    },
    {
      "epoch": 1.0151630022744504,
      "grad_norm": 0.23803703486919403,
      "learning_rate": 9.94945567651633e-05,
      "loss": 0.018785245716571808,
      "step": 5356
    },
    {
      "epoch": 1.0155420773313115,
      "grad_norm": 0.20138081908226013,
      "learning_rate": 9.941679626749611e-05,
      "loss": 0.013376863673329353,
      "step": 5358
    },
    {
      "epoch": 1.0159211523881728,
      "grad_norm": 0.3336670994758606,
      "learning_rate": 9.933903576982893e-05,
      "loss": 0.018662698566913605,
      "step": 5360
    },
    {
      "epoch": 1.0163002274450341,
      "grad_norm": 0.3891942799091339,
      "learning_rate": 9.926127527216175e-05,
      "loss": 0.03248418867588043,
      "step": 5362
    },
    {
      "epoch": 1.0166793025018954,
      "grad_norm": 0.10404258221387863,
      "learning_rate": 9.918351477449456e-05,
      "loss": 0.01542374212294817,
      "step": 5364
    },
    {
      "epoch": 1.0170583775587567,
      "grad_norm": 0.17720355093479156,
      "learning_rate": 9.910575427682738e-05,
      "loss": 0.013366783037781715,
      "step": 5366
    },
    {
      "epoch": 1.0174374526156178,
      "grad_norm": 0.08770980685949326,
      "learning_rate": 9.902799377916019e-05,
      "loss": 0.0130698774009943,
      "step": 5368
    },
    {
      "epoch": 1.017816527672479,
      "grad_norm": 0.11840029805898666,
      "learning_rate": 9.8950233281493e-05,
      "loss": 0.014842617325484753,
      "step": 5370
    },
    {
      "epoch": 1.0181956027293404,
      "grad_norm": 0.33520159125328064,
      "learning_rate": 9.887247278382583e-05,
      "loss": 0.017210688441991806,
      "step": 5372
    },
    {
      "epoch": 1.0185746777862017,
      "grad_norm": 0.1011263057589531,
      "learning_rate": 9.879471228615865e-05,
      "loss": 0.01558513194322586,
      "step": 5374
    },
    {
      "epoch": 1.018953752843063,
      "grad_norm": 0.0866982489824295,
      "learning_rate": 9.871695178849145e-05,
      "loss": 0.011960496194660664,
      "step": 5376
    },
    {
      "epoch": 1.019332827899924,
      "grad_norm": 0.3699359595775604,
      "learning_rate": 9.863919129082426e-05,
      "loss": 0.027118414640426636,
      "step": 5378
    },
    {
      "epoch": 1.0197119029567854,
      "grad_norm": 0.22528378665447235,
      "learning_rate": 9.856143079315708e-05,
      "loss": 0.01553428266197443,
      "step": 5380
    },
    {
      "epoch": 1.0200909780136467,
      "grad_norm": 0.13846533000469208,
      "learning_rate": 9.848367029548989e-05,
      "loss": 0.012231133878231049,
      "step": 5382
    },
    {
      "epoch": 1.020470053070508,
      "grad_norm": 0.11530452221632004,
      "learning_rate": 9.840590979782271e-05,
      "loss": 0.015216686762869358,
      "step": 5384
    },
    {
      "epoch": 1.0208491281273693,
      "grad_norm": 0.3502960801124573,
      "learning_rate": 9.832814930015553e-05,
      "loss": 0.027004987001419067,
      "step": 5386
    },
    {
      "epoch": 1.0212282031842306,
      "grad_norm": 0.19905081391334534,
      "learning_rate": 9.825038880248834e-05,
      "loss": 0.02327483519911766,
      "step": 5388
    },
    {
      "epoch": 1.0216072782410917,
      "grad_norm": 0.11131252348423004,
      "learning_rate": 9.817262830482116e-05,
      "loss": 0.022564949467778206,
      "step": 5390
    },
    {
      "epoch": 1.021986353297953,
      "grad_norm": 0.3013923168182373,
      "learning_rate": 9.809486780715396e-05,
      "loss": 0.021052202209830284,
      "step": 5392
    },
    {
      "epoch": 1.0223654283548143,
      "grad_norm": 0.08708786964416504,
      "learning_rate": 9.801710730948678e-05,
      "loss": 0.011533827520906925,
      "step": 5394
    },
    {
      "epoch": 1.0227445034116756,
      "grad_norm": 0.1306350827217102,
      "learning_rate": 9.79393468118196e-05,
      "loss": 0.016448969021439552,
      "step": 5396
    },
    {
      "epoch": 1.0231235784685369,
      "grad_norm": 0.09804105013608932,
      "learning_rate": 9.786158631415242e-05,
      "loss": 0.01637275144457817,
      "step": 5398
    },
    {
      "epoch": 1.023502653525398,
      "grad_norm": 0.3306742012500763,
      "learning_rate": 9.778382581648523e-05,
      "loss": 0.019354969263076782,
      "step": 5400
    },
    {
      "epoch": 1.0238817285822592,
      "grad_norm": 0.20605741441249847,
      "learning_rate": 9.770606531881804e-05,
      "loss": 0.014065349474549294,
      "step": 5402
    },
    {
      "epoch": 1.0242608036391205,
      "grad_norm": 0.2740415036678314,
      "learning_rate": 9.762830482115086e-05,
      "loss": 0.023741167038679123,
      "step": 5404
    },
    {
      "epoch": 1.0246398786959818,
      "grad_norm": 0.10125033557415009,
      "learning_rate": 9.755054432348368e-05,
      "loss": 0.014917959459125996,
      "step": 5406
    },
    {
      "epoch": 1.0250189537528431,
      "grad_norm": 0.20026294887065887,
      "learning_rate": 9.74727838258165e-05,
      "loss": 0.01805431954562664,
      "step": 5408
    },
    {
      "epoch": 1.0253980288097044,
      "grad_norm": 0.35624954104423523,
      "learning_rate": 9.73950233281493e-05,
      "loss": 0.026193764060735703,
      "step": 5410
    },
    {
      "epoch": 1.0257771038665655,
      "grad_norm": 0.0984259620308876,
      "learning_rate": 9.731726283048211e-05,
      "loss": 0.019722580909729004,
      "step": 5412
    },
    {
      "epoch": 1.0261561789234268,
      "grad_norm": 0.3694266676902771,
      "learning_rate": 9.723950233281493e-05,
      "loss": 0.020705504342913628,
      "step": 5414
    },
    {
      "epoch": 1.026535253980288,
      "grad_norm": 0.1777893602848053,
      "learning_rate": 9.716174183514774e-05,
      "loss": 0.015016200952231884,
      "step": 5416
    },
    {
      "epoch": 1.0269143290371494,
      "grad_norm": 0.12566958367824554,
      "learning_rate": 9.708398133748056e-05,
      "loss": 0.01235666312277317,
      "step": 5418
    },
    {
      "epoch": 1.0272934040940107,
      "grad_norm": 0.23879769444465637,
      "learning_rate": 9.700622083981338e-05,
      "loss": 0.014049690216779709,
      "step": 5420
    },
    {
      "epoch": 1.0276724791508718,
      "grad_norm": 0.11572455614805222,
      "learning_rate": 9.69284603421462e-05,
      "loss": 0.011628411710262299,
      "step": 5422
    },
    {
      "epoch": 1.028051554207733,
      "grad_norm": 0.23180386424064636,
      "learning_rate": 9.685069984447901e-05,
      "loss": 0.015305621549487114,
      "step": 5424
    },
    {
      "epoch": 1.0284306292645944,
      "grad_norm": 0.07746459543704987,
      "learning_rate": 9.677293934681182e-05,
      "loss": 0.013398375362157822,
      "step": 5426
    },
    {
      "epoch": 1.0288097043214557,
      "grad_norm": 0.2743913531303406,
      "learning_rate": 9.669517884914464e-05,
      "loss": 0.019679972901940346,
      "step": 5428
    },
    {
      "epoch": 1.029188779378317,
      "grad_norm": 0.1735534518957138,
      "learning_rate": 9.661741835147746e-05,
      "loss": 0.014094888232648373,
      "step": 5430
    },
    {
      "epoch": 1.029567854435178,
      "grad_norm": 0.0873703807592392,
      "learning_rate": 9.653965785381028e-05,
      "loss": 0.020682698115706444,
      "step": 5432
    },
    {
      "epoch": 1.0299469294920394,
      "grad_norm": 0.08689716458320618,
      "learning_rate": 9.646189735614308e-05,
      "loss": 0.01247484888881445,
      "step": 5434
    },
    {
      "epoch": 1.0303260045489007,
      "grad_norm": 0.11352961510419846,
      "learning_rate": 9.63841368584759e-05,
      "loss": 0.020537693053483963,
      "step": 5436
    },
    {
      "epoch": 1.030705079605762,
      "grad_norm": 0.10766389966011047,
      "learning_rate": 9.630637636080871e-05,
      "loss": 0.014769620262086391,
      "step": 5438
    },
    {
      "epoch": 1.0310841546626233,
      "grad_norm": 0.2848151624202728,
      "learning_rate": 9.622861586314152e-05,
      "loss": 0.022091306746006012,
      "step": 5440
    },
    {
      "epoch": 1.0314632297194846,
      "grad_norm": 0.3077424466609955,
      "learning_rate": 9.615085536547435e-05,
      "loss": 0.013704158365726471,
      "step": 5442
    },
    {
      "epoch": 1.0318423047763456,
      "grad_norm": 0.29164478182792664,
      "learning_rate": 9.607309486780716e-05,
      "loss": 0.033443912863731384,
      "step": 5444
    },
    {
      "epoch": 1.032221379833207,
      "grad_norm": 0.3787733018398285,
      "learning_rate": 9.599533437013998e-05,
      "loss": 0.030136143788695335,
      "step": 5446
    },
    {
      "epoch": 1.0326004548900682,
      "grad_norm": 0.29052504897117615,
      "learning_rate": 9.591757387247279e-05,
      "loss": 0.02400844544172287,
      "step": 5448
    },
    {
      "epoch": 1.0329795299469295,
      "grad_norm": 0.19567856192588806,
      "learning_rate": 9.58398133748056e-05,
      "loss": 0.014231200329959393,
      "step": 5450
    },
    {
      "epoch": 1.0333586050037908,
      "grad_norm": 0.27820464968681335,
      "learning_rate": 9.576205287713841e-05,
      "loss": 0.023639511317014694,
      "step": 5452
    },
    {
      "epoch": 1.033737680060652,
      "grad_norm": 0.1483076512813568,
      "learning_rate": 9.568429237947123e-05,
      "loss": 0.01967611163854599,
      "step": 5454
    },
    {
      "epoch": 1.0341167551175132,
      "grad_norm": 0.11223980784416199,
      "learning_rate": 9.560653188180405e-05,
      "loss": 0.013122877106070518,
      "step": 5456
    },
    {
      "epoch": 1.0344958301743745,
      "grad_norm": 0.11724534630775452,
      "learning_rate": 9.552877138413686e-05,
      "loss": 0.01832497864961624,
      "step": 5458
    },
    {
      "epoch": 1.0348749052312358,
      "grad_norm": 0.1079336404800415,
      "learning_rate": 9.545101088646968e-05,
      "loss": 0.01173208374530077,
      "step": 5460
    },
    {
      "epoch": 1.0352539802880971,
      "grad_norm": 0.2572896480560303,
      "learning_rate": 9.537325038880249e-05,
      "loss": 0.016547702252864838,
      "step": 5462
    },
    {
      "epoch": 1.0356330553449582,
      "grad_norm": 0.3308436870574951,
      "learning_rate": 9.529548989113531e-05,
      "loss": 0.021486764773726463,
      "step": 5464
    },
    {
      "epoch": 1.0360121304018195,
      "grad_norm": 0.08586468547582626,
      "learning_rate": 9.521772939346813e-05,
      "loss": 0.020046599209308624,
      "step": 5466
    },
    {
      "epoch": 1.0363912054586808,
      "grad_norm": 0.37184253334999084,
      "learning_rate": 9.513996889580094e-05,
      "loss": 0.02139345556497574,
      "step": 5468
    },
    {
      "epoch": 1.036770280515542,
      "grad_norm": 0.289531946182251,
      "learning_rate": 9.506220839813376e-05,
      "loss": 0.0177882369607687,
      "step": 5470
    },
    {
      "epoch": 1.0371493555724034,
      "grad_norm": 0.1272176206111908,
      "learning_rate": 9.498444790046656e-05,
      "loss": 0.010725134052336216,
      "step": 5472
    },
    {
      "epoch": 1.0375284306292647,
      "grad_norm": 0.10471192002296448,
      "learning_rate": 9.490668740279938e-05,
      "loss": 0.014618871733546257,
      "step": 5474
    },
    {
      "epoch": 1.0379075056861258,
      "grad_norm": 0.2377527356147766,
      "learning_rate": 9.48289269051322e-05,
      "loss": 0.013824587687849998,
      "step": 5476
    },
    {
      "epoch": 1.038286580742987,
      "grad_norm": 0.21112316846847534,
      "learning_rate": 9.475116640746501e-05,
      "loss": 0.012940947897732258,
      "step": 5478
    },
    {
      "epoch": 1.0386656557998484,
      "grad_norm": 0.12029413878917694,
      "learning_rate": 9.467340590979783e-05,
      "loss": 0.016009533777832985,
      "step": 5480
    },
    {
      "epoch": 1.0390447308567097,
      "grad_norm": 0.3502540588378906,
      "learning_rate": 9.459564541213064e-05,
      "loss": 0.02502923458814621,
      "step": 5482
    },
    {
      "epoch": 1.039423805913571,
      "grad_norm": 0.20816302299499512,
      "learning_rate": 9.451788491446346e-05,
      "loss": 0.014543949626386166,
      "step": 5484
    },
    {
      "epoch": 1.039802880970432,
      "grad_norm": 0.11230815201997757,
      "learning_rate": 9.444012441679627e-05,
      "loss": 0.019979426637291908,
      "step": 5486
    },
    {
      "epoch": 1.0401819560272934,
      "grad_norm": 0.26090866327285767,
      "learning_rate": 9.436236391912909e-05,
      "loss": 0.0244913250207901,
      "step": 5488
    },
    {
      "epoch": 1.0405610310841547,
      "grad_norm": 0.08324436098337173,
      "learning_rate": 9.428460342146191e-05,
      "loss": 0.011487127281725407,
      "step": 5490
    },
    {
      "epoch": 1.040940106141016,
      "grad_norm": 0.16599220037460327,
      "learning_rate": 9.420684292379471e-05,
      "loss": 0.012748071923851967,
      "step": 5492
    },
    {
      "epoch": 1.0413191811978773,
      "grad_norm": 0.26409879326820374,
      "learning_rate": 9.412908242612753e-05,
      "loss": 0.01587301306426525,
      "step": 5494
    },
    {
      "epoch": 1.0416982562547386,
      "grad_norm": 0.10566183179616928,
      "learning_rate": 9.405132192846034e-05,
      "loss": 0.020893199369311333,
      "step": 5496
    },
    {
      "epoch": 1.0420773313115996,
      "grad_norm": 0.10533969849348068,
      "learning_rate": 9.397356143079316e-05,
      "loss": 0.015594685450196266,
      "step": 5498
    },
    {
      "epoch": 1.042456406368461,
      "grad_norm": 0.33053237199783325,
      "learning_rate": 9.389580093312598e-05,
      "loss": 0.02229117974638939,
      "step": 5500
    },
    {
      "epoch": 1.0428354814253222,
      "grad_norm": 0.27690982818603516,
      "learning_rate": 9.381804043545879e-05,
      "loss": 0.017757993191480637,
      "step": 5502
    },
    {
      "epoch": 1.0432145564821835,
      "grad_norm": 0.27177488803863525,
      "learning_rate": 9.374027993779161e-05,
      "loss": 0.018003610894083977,
      "step": 5504
    },
    {
      "epoch": 1.0435936315390448,
      "grad_norm": 0.22330114245414734,
      "learning_rate": 9.366251944012442e-05,
      "loss": 0.01792118325829506,
      "step": 5506
    },
    {
      "epoch": 1.043972706595906,
      "grad_norm": 0.1874147653579712,
      "learning_rate": 9.358475894245724e-05,
      "loss": 0.01321326568722725,
      "step": 5508
    },
    {
      "epoch": 1.0443517816527672,
      "grad_norm": 0.33031201362609863,
      "learning_rate": 9.350699844479004e-05,
      "loss": 0.02490287274122238,
      "step": 5510
    },
    {
      "epoch": 1.0447308567096285,
      "grad_norm": 0.3028697669506073,
      "learning_rate": 9.342923794712286e-05,
      "loss": 0.027465010061860085,
      "step": 5512
    },
    {
      "epoch": 1.0451099317664898,
      "grad_norm": 0.24987469613552094,
      "learning_rate": 9.335147744945568e-05,
      "loss": 0.017022470012307167,
      "step": 5514
    },
    {
      "epoch": 1.045489006823351,
      "grad_norm": 0.14513574540615082,
      "learning_rate": 9.327371695178849e-05,
      "loss": 0.012787571176886559,
      "step": 5516
    },
    {
      "epoch": 1.0458680818802122,
      "grad_norm": 0.08557494729757309,
      "learning_rate": 9.319595645412131e-05,
      "loss": 0.012233566492795944,
      "step": 5518
    },
    {
      "epoch": 1.0462471569370735,
      "grad_norm": 0.2851097285747528,
      "learning_rate": 9.311819595645412e-05,
      "loss": 0.024606890976428986,
      "step": 5520
    },
    {
      "epoch": 1.0466262319939348,
      "grad_norm": 0.09511065483093262,
      "learning_rate": 9.304043545878694e-05,
      "loss": 0.012623230926692486,
      "step": 5522
    },
    {
      "epoch": 1.047005307050796,
      "grad_norm": 0.2221929430961609,
      "learning_rate": 9.296267496111976e-05,
      "loss": 0.014235375449061394,
      "step": 5524
    },
    {
      "epoch": 1.0473843821076574,
      "grad_norm": 0.15092605352401733,
      "learning_rate": 9.288491446345257e-05,
      "loss": 0.013536419719457626,
      "step": 5526
    },
    {
      "epoch": 1.0477634571645185,
      "grad_norm": 0.07350442558526993,
      "learning_rate": 9.280715396578539e-05,
      "loss": 0.010856824927031994,
      "step": 5528
    },
    {
      "epoch": 1.0481425322213798,
      "grad_norm": 0.09930683672428131,
      "learning_rate": 9.27293934681182e-05,
      "loss": 0.015812747180461884,
      "step": 5530
    },
    {
      "epoch": 1.048521607278241,
      "grad_norm": 0.24629749357700348,
      "learning_rate": 9.265163297045101e-05,
      "loss": 0.01567652076482773,
      "step": 5532
    },
    {
      "epoch": 1.0489006823351024,
      "grad_norm": 0.12615136802196503,
      "learning_rate": 9.257387247278383e-05,
      "loss": 0.021667836233973503,
      "step": 5534
    },
    {
      "epoch": 1.0492797573919637,
      "grad_norm": 0.06946789473295212,
      "learning_rate": 9.249611197511666e-05,
      "loss": 0.010988696478307247,
      "step": 5536
    },
    {
      "epoch": 1.049658832448825,
      "grad_norm": 0.08234026283025742,
      "learning_rate": 9.241835147744946e-05,
      "loss": 0.016432665288448334,
      "step": 5538
    },
    {
      "epoch": 1.050037907505686,
      "grad_norm": 0.1445610672235489,
      "learning_rate": 9.234059097978227e-05,
      "loss": 0.012220725417137146,
      "step": 5540
    },
    {
      "epoch": 1.0504169825625473,
      "grad_norm": 0.09376371651887894,
      "learning_rate": 9.226283048211509e-05,
      "loss": 0.01386516634374857,
      "step": 5542
    },
    {
      "epoch": 1.0507960576194086,
      "grad_norm": 0.36080285906791687,
      "learning_rate": 9.21850699844479e-05,
      "loss": 0.025408508256077766,
      "step": 5544
    },
    {
      "epoch": 1.05117513267627,
      "grad_norm": 0.3667946755886078,
      "learning_rate": 9.210730948678073e-05,
      "loss": 0.024016626179218292,
      "step": 5546
    },
    {
      "epoch": 1.0515542077331312,
      "grad_norm": 0.2839130461215973,
      "learning_rate": 9.202954898911354e-05,
      "loss": 0.016453834250569344,
      "step": 5548
    },
    {
      "epoch": 1.0519332827899923,
      "grad_norm": 0.17669439315795898,
      "learning_rate": 9.195178849144634e-05,
      "loss": 0.02130028046667576,
      "step": 5550
    },
    {
      "epoch": 1.0523123578468536,
      "grad_norm": 0.28049254417419434,
      "learning_rate": 9.187402799377916e-05,
      "loss": 0.015424232929944992,
      "step": 5552
    },
    {
      "epoch": 1.052691432903715,
      "grad_norm": 0.2567935585975647,
      "learning_rate": 9.179626749611197e-05,
      "loss": 0.013911749236285686,
      "step": 5554
    },
    {
      "epoch": 1.0530705079605762,
      "grad_norm": 0.10608581453561783,
      "learning_rate": 9.171850699844479e-05,
      "loss": 0.01324798446148634,
      "step": 5556
    },
    {
      "epoch": 1.0534495830174375,
      "grad_norm": 0.1462164670228958,
      "learning_rate": 9.164074650077761e-05,
      "loss": 0.017259564250707626,
      "step": 5558
    },
    {
      "epoch": 1.0538286580742988,
      "grad_norm": 0.09063100814819336,
      "learning_rate": 9.156298600311043e-05,
      "loss": 0.014989602379500866,
      "step": 5560
    },
    {
      "epoch": 1.05420773313116,
      "grad_norm": 0.09291625022888184,
      "learning_rate": 9.148522550544324e-05,
      "loss": 0.012588946148753166,
      "step": 5562
    },
    {
      "epoch": 1.0545868081880212,
      "grad_norm": 0.26962485909461975,
      "learning_rate": 9.140746500777605e-05,
      "loss": 0.01813768967986107,
      "step": 5564
    },
    {
      "epoch": 1.0549658832448825,
      "grad_norm": 0.36603930592536926,
      "learning_rate": 9.132970451010887e-05,
      "loss": 0.02741197869181633,
      "step": 5566
    },
    {
      "epoch": 1.0553449583017438,
      "grad_norm": 0.10970902442932129,
      "learning_rate": 9.125194401244169e-05,
      "loss": 0.018837464973330498,
      "step": 5568
    },
    {
      "epoch": 1.055724033358605,
      "grad_norm": 0.22937697172164917,
      "learning_rate": 9.117418351477451e-05,
      "loss": 0.01587367244064808,
      "step": 5570
    },
    {
      "epoch": 1.0561031084154662,
      "grad_norm": 0.24442341923713684,
      "learning_rate": 9.109642301710731e-05,
      "loss": 0.02114756777882576,
      "step": 5572
    },
    {
      "epoch": 1.0564821834723275,
      "grad_norm": 0.11328095197677612,
      "learning_rate": 9.101866251944012e-05,
      "loss": 0.014450639486312866,
      "step": 5574
    },
    {
      "epoch": 1.0568612585291888,
      "grad_norm": 0.08662250638008118,
      "learning_rate": 9.094090202177294e-05,
      "loss": 0.011926093138754368,
      "step": 5576
    },
    {
      "epoch": 1.05724033358605,
      "grad_norm": 0.08298711478710175,
      "learning_rate": 9.086314152410575e-05,
      "loss": 0.013538673520088196,
      "step": 5578
    },
    {
      "epoch": 1.0576194086429114,
      "grad_norm": 0.28007972240448,
      "learning_rate": 9.078538102643858e-05,
      "loss": 0.013216827996075153,
      "step": 5580
    },
    {
      "epoch": 1.0579984836997725,
      "grad_norm": 0.09504323452711105,
      "learning_rate": 9.070762052877139e-05,
      "loss": 0.014583908021450043,
      "step": 5582
    },
    {
      "epoch": 1.0583775587566338,
      "grad_norm": 0.20934613049030304,
      "learning_rate": 9.062986003110421e-05,
      "loss": 0.0171902347356081,
      "step": 5584
    },
    {
      "epoch": 1.058756633813495,
      "grad_norm": 0.13098935782909393,
      "learning_rate": 9.055209953343702e-05,
      "loss": 0.019684726372361183,
      "step": 5586
    },
    {
      "epoch": 1.0591357088703564,
      "grad_norm": 0.2234562486410141,
      "learning_rate": 9.047433903576982e-05,
      "loss": 0.01745675876736641,
      "step": 5588
    },
    {
      "epoch": 1.0595147839272177,
      "grad_norm": 0.22425991296768188,
      "learning_rate": 9.039657853810264e-05,
      "loss": 0.021594751626253128,
      "step": 5590
    },
    {
      "epoch": 1.059893858984079,
      "grad_norm": 0.2771625220775604,
      "learning_rate": 9.031881804043546e-05,
      "loss": 0.023501386865973473,
      "step": 5592
    },
    {
      "epoch": 1.06027293404094,
      "grad_norm": 0.07900496572256088,
      "learning_rate": 9.024105754276829e-05,
      "loss": 0.011560531333088875,
      "step": 5594
    },
    {
      "epoch": 1.0606520090978013,
      "grad_norm": 0.16145704686641693,
      "learning_rate": 9.016329704510109e-05,
      "loss": 0.011470927856862545,
      "step": 5596
    },
    {
      "epoch": 1.0610310841546626,
      "grad_norm": 0.0972648486495018,
      "learning_rate": 9.008553654743391e-05,
      "loss": 0.01652497425675392,
      "step": 5598
    },
    {
      "epoch": 1.061410159211524,
      "grad_norm": 0.39455512166023254,
      "learning_rate": 9.000777604976672e-05,
      "loss": 0.026594551280140877,
      "step": 5600
    },
    {
      "epoch": 1.0617892342683852,
      "grad_norm": 0.2995770573616028,
      "learning_rate": 8.993001555209954e-05,
      "loss": 0.02687237039208412,
      "step": 5602
    },
    {
      "epoch": 1.0621683093252463,
      "grad_norm": 0.24454453587532043,
      "learning_rate": 8.985225505443236e-05,
      "loss": 0.025352636352181435,
      "step": 5604
    },
    {
      "epoch": 1.0625473843821076,
      "grad_norm": 0.12194573879241943,
      "learning_rate": 8.977449455676517e-05,
      "loss": 0.015328306704759598,
      "step": 5606
    },
    {
      "epoch": 1.062926459438969,
      "grad_norm": 0.16996414959430695,
      "learning_rate": 8.969673405909799e-05,
      "loss": 0.014738869853317738,
      "step": 5608
    },
    {
      "epoch": 1.0633055344958302,
      "grad_norm": 0.08595800399780273,
      "learning_rate": 8.96189735614308e-05,
      "loss": 0.01463007926940918,
      "step": 5610
    },
    {
      "epoch": 1.0636846095526915,
      "grad_norm": 0.09674971550703049,
      "learning_rate": 8.95412130637636e-05,
      "loss": 0.015909288078546524,
      "step": 5612
    },
    {
      "epoch": 1.0640636846095526,
      "grad_norm": 0.16176722943782806,
      "learning_rate": 8.946345256609642e-05,
      "loss": 0.014849198050796986,
      "step": 5614
    },
    {
      "epoch": 1.0644427596664139,
      "grad_norm": 0.2863903343677521,
      "learning_rate": 8.938569206842924e-05,
      "loss": 0.01900758408010006,
      "step": 5616
    },
    {
      "epoch": 1.0648218347232752,
      "grad_norm": 0.25859057903289795,
      "learning_rate": 8.930793157076206e-05,
      "loss": 0.01631295122206211,
      "step": 5618
    },
    {
      "epoch": 1.0652009097801365,
      "grad_norm": 0.07881476730108261,
      "learning_rate": 8.923017107309487e-05,
      "loss": 0.0124965226277709,
      "step": 5620
    },
    {
      "epoch": 1.0655799848369978,
      "grad_norm": 0.3638054430484772,
      "learning_rate": 8.915241057542769e-05,
      "loss": 0.021760553121566772,
      "step": 5622
    },
    {
      "epoch": 1.065959059893859,
      "grad_norm": 0.11748757213354111,
      "learning_rate": 8.90746500777605e-05,
      "loss": 0.01874411851167679,
      "step": 5624
    },
    {
      "epoch": 1.0663381349507202,
      "grad_norm": 0.3776737451553345,
      "learning_rate": 8.899688958009332e-05,
      "loss": 0.02652384154498577,
      "step": 5626
    },
    {
      "epoch": 1.0667172100075815,
      "grad_norm": 0.08601853996515274,
      "learning_rate": 8.891912908242614e-05,
      "loss": 0.01525803841650486,
      "step": 5628
    },
    {
      "epoch": 1.0670962850644428,
      "grad_norm": 0.18284256756305695,
      "learning_rate": 8.884136858475894e-05,
      "loss": 0.01991979591548443,
      "step": 5630
    },
    {
      "epoch": 1.067475360121304,
      "grad_norm": 0.13986484706401825,
      "learning_rate": 8.876360808709177e-05,
      "loss": 0.013703204691410065,
      "step": 5632
    },
    {
      "epoch": 1.0678544351781654,
      "grad_norm": 0.09087768197059631,
      "learning_rate": 8.868584758942457e-05,
      "loss": 0.014943884685635567,
      "step": 5634
    },
    {
      "epoch": 1.0682335102350264,
      "grad_norm": 0.07668139040470123,
      "learning_rate": 8.860808709175739e-05,
      "loss": 0.015191950835287571,
      "step": 5636
    },
    {
      "epoch": 1.0686125852918877,
      "grad_norm": 0.28429585695266724,
      "learning_rate": 8.853032659409021e-05,
      "loss": 0.019228115677833557,
      "step": 5638
    },
    {
      "epoch": 1.068991660348749,
      "grad_norm": 0.10750018060207367,
      "learning_rate": 8.845256609642302e-05,
      "loss": 0.011063637211918831,
      "step": 5640
    },
    {
      "epoch": 1.0693707354056103,
      "grad_norm": 0.3658938705921173,
      "learning_rate": 8.837480559875584e-05,
      "loss": 0.03071574494242668,
      "step": 5642
    },
    {
      "epoch": 1.0697498104624716,
      "grad_norm": 0.31764641404151917,
      "learning_rate": 8.829704510108865e-05,
      "loss": 0.020462537184357643,
      "step": 5644
    },
    {
      "epoch": 1.070128885519333,
      "grad_norm": 0.19069477915763855,
      "learning_rate": 8.821928460342147e-05,
      "loss": 0.02677733264863491,
      "step": 5646
    },
    {
      "epoch": 1.070507960576194,
      "grad_norm": 0.09844266623258591,
      "learning_rate": 8.814152410575427e-05,
      "loss": 0.023318342864513397,
      "step": 5648
    },
    {
      "epoch": 1.0708870356330553,
      "grad_norm": 0.26815587282180786,
      "learning_rate": 8.80637636080871e-05,
      "loss": 0.015158780850470066,
      "step": 5650
    },
    {
      "epoch": 1.0712661106899166,
      "grad_norm": 0.11953488737344742,
      "learning_rate": 8.798600311041992e-05,
      "loss": 0.012699148617684841,
      "step": 5652
    },
    {
      "epoch": 1.071645185746778,
      "grad_norm": 0.11880351603031158,
      "learning_rate": 8.790824261275272e-05,
      "loss": 0.01382971927523613,
      "step": 5654
    },
    {
      "epoch": 1.0720242608036392,
      "grad_norm": 0.09941986948251724,
      "learning_rate": 8.783048211508554e-05,
      "loss": 0.012791860848665237,
      "step": 5656
    },
    {
      "epoch": 1.0724033358605003,
      "grad_norm": 0.07729235291481018,
      "learning_rate": 8.775272161741835e-05,
      "loss": 0.014019137248396873,
      "step": 5658
    },
    {
      "epoch": 1.0727824109173616,
      "grad_norm": 0.3590768277645111,
      "learning_rate": 8.767496111975117e-05,
      "loss": 0.018812593072652817,
      "step": 5660
    },
    {
      "epoch": 1.073161485974223,
      "grad_norm": 0.26709893345832825,
      "learning_rate": 8.759720062208399e-05,
      "loss": 0.028603069484233856,
      "step": 5662
    },
    {
      "epoch": 1.0735405610310842,
      "grad_norm": 0.12679076194763184,
      "learning_rate": 8.75194401244168e-05,
      "loss": 0.028213374316692352,
      "step": 5664
    },
    {
      "epoch": 1.0739196360879455,
      "grad_norm": 0.10379723459482193,
      "learning_rate": 8.744167962674962e-05,
      "loss": 0.012240038253366947,
      "step": 5666
    },
    {
      "epoch": 1.0742987111448066,
      "grad_norm": 0.2030516117811203,
      "learning_rate": 8.736391912908242e-05,
      "loss": 0.01815926842391491,
      "step": 5668
    },
    {
      "epoch": 1.0746777862016679,
      "grad_norm": 0.24651598930358887,
      "learning_rate": 8.728615863141525e-05,
      "loss": 0.02112535387277603,
      "step": 5670
    },
    {
      "epoch": 1.0750568612585292,
      "grad_norm": 0.09186781197786331,
      "learning_rate": 8.720839813374807e-05,
      "loss": 0.013233995996415615,
      "step": 5672
    },
    {
      "epoch": 1.0754359363153905,
      "grad_norm": 0.19957216084003448,
      "learning_rate": 8.713063763608087e-05,
      "loss": 0.01360487099736929,
      "step": 5674
    },
    {
      "epoch": 1.0758150113722518,
      "grad_norm": 0.24281565845012665,
      "learning_rate": 8.705287713841369e-05,
      "loss": 0.015483858063817024,
      "step": 5676
    },
    {
      "epoch": 1.0761940864291129,
      "grad_norm": 0.24232594668865204,
      "learning_rate": 8.69751166407465e-05,
      "loss": 0.0226869098842144,
      "step": 5678
    },
    {
      "epoch": 1.0765731614859742,
      "grad_norm": 0.09807395935058594,
      "learning_rate": 8.689735614307932e-05,
      "loss": 0.022414470091462135,
      "step": 5680
    },
    {
      "epoch": 1.0769522365428355,
      "grad_norm": 0.08173399418592453,
      "learning_rate": 8.681959564541213e-05,
      "loss": 0.014133214950561523,
      "step": 5682
    },
    {
      "epoch": 1.0773313115996968,
      "grad_norm": 0.09478939324617386,
      "learning_rate": 8.674183514774495e-05,
      "loss": 0.023296570405364037,
      "step": 5684
    },
    {
      "epoch": 1.077710386656558,
      "grad_norm": 0.19065457582473755,
      "learning_rate": 8.666407465007777e-05,
      "loss": 0.012735094875097275,
      "step": 5686
    },
    {
      "epoch": 1.0780894617134194,
      "grad_norm": 0.2998434901237488,
      "learning_rate": 8.658631415241057e-05,
      "loss": 0.02203092724084854,
      "step": 5688
    },
    {
      "epoch": 1.0784685367702804,
      "grad_norm": 0.07624565809965134,
      "learning_rate": 8.65085536547434e-05,
      "loss": 0.012862994335591793,
      "step": 5690
    },
    {
      "epoch": 1.0788476118271417,
      "grad_norm": 0.08264656364917755,
      "learning_rate": 8.64307931570762e-05,
      "loss": 0.01169533934444189,
      "step": 5692
    },
    {
      "epoch": 1.079226686884003,
      "grad_norm": 0.17196571826934814,
      "learning_rate": 8.635303265940902e-05,
      "loss": 0.018366556614637375,
      "step": 5694
    },
    {
      "epoch": 1.0796057619408643,
      "grad_norm": 0.0819835290312767,
      "learning_rate": 8.627527216174184e-05,
      "loss": 0.015907062217593193,
      "step": 5696
    },
    {
      "epoch": 1.0799848369977256,
      "grad_norm": 0.07037928700447083,
      "learning_rate": 8.619751166407466e-05,
      "loss": 0.01525009423494339,
      "step": 5698
    },
    {
      "epoch": 1.0803639120545867,
      "grad_norm": 0.25962358713150024,
      "learning_rate": 8.611975116640747e-05,
      "loss": 0.015932902693748474,
      "step": 5700
    },
    {
      "epoch": 1.080742987111448,
      "grad_norm": 0.36224254965782166,
      "learning_rate": 8.604199066874028e-05,
      "loss": 0.017955029383301735,
      "step": 5702
    },
    {
      "epoch": 1.0811220621683093,
      "grad_norm": 0.14493413269519806,
      "learning_rate": 8.59642301710731e-05,
      "loss": 0.019746791571378708,
      "step": 5704
    },
    {
      "epoch": 1.0815011372251706,
      "grad_norm": 0.33278873562812805,
      "learning_rate": 8.588646967340592e-05,
      "loss": 0.02275605872273445,
      "step": 5706
    },
    {
      "epoch": 1.081880212282032,
      "grad_norm": 0.39658960700035095,
      "learning_rate": 8.580870917573874e-05,
      "loss": 0.02626257762312889,
      "step": 5708
    },
    {
      "epoch": 1.0822592873388932,
      "grad_norm": 0.15875042974948883,
      "learning_rate": 8.573094867807155e-05,
      "loss": 0.02075258083641529,
      "step": 5710
    },
    {
      "epoch": 1.0826383623957543,
      "grad_norm": 0.3698574900627136,
      "learning_rate": 8.565318818040435e-05,
      "loss": 0.026206880807876587,
      "step": 5712
    },
    {
      "epoch": 1.0830174374526156,
      "grad_norm": 0.3777303099632263,
      "learning_rate": 8.557542768273717e-05,
      "loss": 0.02901102975010872,
      "step": 5714
    },
    {
      "epoch": 1.0833965125094769,
      "grad_norm": 0.24047930538654327,
      "learning_rate": 8.549766718506998e-05,
      "loss": 0.021798506379127502,
      "step": 5716
    },
    {
      "epoch": 1.0837755875663382,
      "grad_norm": 0.419861376285553,
      "learning_rate": 8.54199066874028e-05,
      "loss": 0.025126246735453606,
      "step": 5718
    },
    {
      "epoch": 1.0841546626231995,
      "grad_norm": 0.08875644952058792,
      "learning_rate": 8.534214618973562e-05,
      "loss": 0.012020062655210495,
      "step": 5720
    },
    {
      "epoch": 1.0845337376800606,
      "grad_norm": 0.3239932954311371,
      "learning_rate": 8.526438569206844e-05,
      "loss": 0.018052207306027412,
      "step": 5722
    },
    {
      "epoch": 1.0849128127369219,
      "grad_norm": 0.4464094340801239,
      "learning_rate": 8.518662519440125e-05,
      "loss": 0.031130433082580566,
      "step": 5724
    },
    {
      "epoch": 1.0852918877937832,
      "grad_norm": 0.17924684286117554,
      "learning_rate": 8.510886469673405e-05,
      "loss": 0.015666278079152107,
      "step": 5726
    },
    {
      "epoch": 1.0856709628506445,
      "grad_norm": 0.4768761694431305,
      "learning_rate": 8.503110419906688e-05,
      "loss": 0.029683366417884827,
      "step": 5728
    },
    {
      "epoch": 1.0860500379075058,
      "grad_norm": 0.29031267762184143,
      "learning_rate": 8.49533437013997e-05,
      "loss": 0.014711176976561546,
      "step": 5730
    },
    {
      "epoch": 1.086429112964367,
      "grad_norm": 0.1637934446334839,
      "learning_rate": 8.487558320373252e-05,
      "loss": 0.012694234028458595,
      "step": 5732
    },
    {
      "epoch": 1.0868081880212281,
      "grad_norm": 0.27724334597587585,
      "learning_rate": 8.479782270606532e-05,
      "loss": 0.014718604274094105,
      "step": 5734
    },
    {
      "epoch": 1.0871872630780894,
      "grad_norm": 0.3029767572879791,
      "learning_rate": 8.472006220839814e-05,
      "loss": 0.02459993213415146,
      "step": 5736
    },
    {
      "epoch": 1.0875663381349507,
      "grad_norm": 0.31528815627098083,
      "learning_rate": 8.464230171073095e-05,
      "loss": 0.02520427107810974,
      "step": 5738
    },
    {
      "epoch": 1.087945413191812,
      "grad_norm": 0.17295125126838684,
      "learning_rate": 8.456454121306376e-05,
      "loss": 0.016923250630497932,
      "step": 5740
    },
    {
      "epoch": 1.0883244882486733,
      "grad_norm": 0.2863968312740326,
      "learning_rate": 8.448678071539659e-05,
      "loss": 0.018927082419395447,
      "step": 5742
    },
    {
      "epoch": 1.0887035633055344,
      "grad_norm": 0.08854573965072632,
      "learning_rate": 8.44090202177294e-05,
      "loss": 0.013010168448090553,
      "step": 5744
    },
    {
      "epoch": 1.0890826383623957,
      "grad_norm": 0.11597151309251785,
      "learning_rate": 8.433125972006222e-05,
      "loss": 0.013820935040712357,
      "step": 5746
    },
    {
      "epoch": 1.089461713419257,
      "grad_norm": 0.09737052768468857,
      "learning_rate": 8.425349922239503e-05,
      "loss": 0.018258919939398766,
      "step": 5748
    },
    {
      "epoch": 1.0898407884761183,
      "grad_norm": 0.09696222841739655,
      "learning_rate": 8.417573872472783e-05,
      "loss": 0.019805891439318657,
      "step": 5750
    },
    {
      "epoch": 1.0902198635329796,
      "grad_norm": 0.1096850261092186,
      "learning_rate": 8.409797822706065e-05,
      "loss": 0.017899850383400917,
      "step": 5752
    },
    {
      "epoch": 1.0905989385898407,
      "grad_norm": 0.09397299587726593,
      "learning_rate": 8.402021772939347e-05,
      "loss": 0.013327115215361118,
      "step": 5754
    },
    {
      "epoch": 1.090978013646702,
      "grad_norm": 0.17233581840991974,
      "learning_rate": 8.39424572317263e-05,
      "loss": 0.017026245594024658,
      "step": 5756
    },
    {
      "epoch": 1.0913570887035633,
      "grad_norm": 0.17801782488822937,
      "learning_rate": 8.38646967340591e-05,
      "loss": 0.01929633505642414,
      "step": 5758
    },
    {
      "epoch": 1.0917361637604246,
      "grad_norm": 0.3331056237220764,
      "learning_rate": 8.378693623639192e-05,
      "loss": 0.028658220544457436,
      "step": 5760
    },
    {
      "epoch": 1.092115238817286,
      "grad_norm": 0.3403438627719879,
      "learning_rate": 8.370917573872473e-05,
      "loss": 0.025052160024642944,
      "step": 5762
    },
    {
      "epoch": 1.092494313874147,
      "grad_norm": 0.33564233779907227,
      "learning_rate": 8.363141524105755e-05,
      "loss": 0.02248547412455082,
      "step": 5764
    },
    {
      "epoch": 1.0928733889310083,
      "grad_norm": 0.07938660681247711,
      "learning_rate": 8.355365474339037e-05,
      "loss": 0.014379815198481083,
      "step": 5766
    },
    {
      "epoch": 1.0932524639878696,
      "grad_norm": 0.27713218331336975,
      "learning_rate": 8.347589424572318e-05,
      "loss": 0.016115285456180573,
      "step": 5768
    },
    {
      "epoch": 1.0936315390447309,
      "grad_norm": 0.2369307279586792,
      "learning_rate": 8.3398133748056e-05,
      "loss": 0.013057480566203594,
      "step": 5770
    },
    {
      "epoch": 1.0940106141015922,
      "grad_norm": 0.2165757268667221,
      "learning_rate": 8.33203732503888e-05,
      "loss": 0.014548258855938911,
      "step": 5772
    },
    {
      "epoch": 1.0943896891584535,
      "grad_norm": 0.2569042444229126,
      "learning_rate": 8.324261275272161e-05,
      "loss": 0.014959360472857952,
      "step": 5774
    },
    {
      "epoch": 1.0947687642153145,
      "grad_norm": 0.0855637863278389,
      "learning_rate": 8.316485225505444e-05,
      "loss": 0.013696418143808842,
      "step": 5776
    },
    {
      "epoch": 1.0951478392721758,
      "grad_norm": 0.2387666553258896,
      "learning_rate": 8.308709175738725e-05,
      "loss": 0.01638496108353138,
      "step": 5778
    },
    {
      "epoch": 1.0955269143290371,
      "grad_norm": 0.0936400443315506,
      "learning_rate": 8.300933125972007e-05,
      "loss": 0.02115480788052082,
      "step": 5780
    },
    {
      "epoch": 1.0959059893858984,
      "grad_norm": 0.09790290892124176,
      "learning_rate": 8.293157076205288e-05,
      "loss": 0.011410954408347607,
      "step": 5782
    },
    {
      "epoch": 1.0962850644427597,
      "grad_norm": 0.08248036354780197,
      "learning_rate": 8.28538102643857e-05,
      "loss": 0.01293619628995657,
      "step": 5784
    },
    {
      "epoch": 1.0966641394996208,
      "grad_norm": 0.31526246666908264,
      "learning_rate": 8.27760497667185e-05,
      "loss": 0.02083897776901722,
      "step": 5786
    },
    {
      "epoch": 1.0970432145564821,
      "grad_norm": 0.11733312904834747,
      "learning_rate": 8.269828926905133e-05,
      "loss": 0.0206212867051363,
      "step": 5788
    },
    {
      "epoch": 1.0974222896133434,
      "grad_norm": 0.09043586999177933,
      "learning_rate": 8.262052877138415e-05,
      "loss": 0.011903727427124977,
      "step": 5790
    },
    {
      "epoch": 1.0978013646702047,
      "grad_norm": 0.24839575588703156,
      "learning_rate": 8.254276827371695e-05,
      "loss": 0.018771057948470116,
      "step": 5792
    },
    {
      "epoch": 1.098180439727066,
      "grad_norm": 0.39068803191185,
      "learning_rate": 8.246500777604977e-05,
      "loss": 0.030858132988214493,
      "step": 5794
    },
    {
      "epoch": 1.0985595147839273,
      "grad_norm": 0.10113295167684555,
      "learning_rate": 8.238724727838258e-05,
      "loss": 0.012172200717031956,
      "step": 5796
    },
    {
      "epoch": 1.0989385898407884,
      "grad_norm": 0.2863364517688751,
      "learning_rate": 8.23094867807154e-05,
      "loss": 0.022480154410004616,
      "step": 5798
    },
    {
      "epoch": 1.0993176648976497,
      "grad_norm": 0.3685663044452667,
      "learning_rate": 8.223172628304822e-05,
      "loss": 0.02151469886302948,
      "step": 5800
    },
    {
      "epoch": 1.099696739954511,
      "grad_norm": 0.18069666624069214,
      "learning_rate": 8.215396578538103e-05,
      "loss": 0.015996679663658142,
      "step": 5802
    },
    {
      "epoch": 1.1000758150113723,
      "grad_norm": 0.08902040123939514,
      "learning_rate": 8.207620528771385e-05,
      "loss": 0.017495570704340935,
      "step": 5804
    },
    {
      "epoch": 1.1004548900682336,
      "grad_norm": 0.3270830810070038,
      "learning_rate": 8.199844479004666e-05,
      "loss": 0.021676335483789444,
      "step": 5806
    },
    {
      "epoch": 1.1008339651250947,
      "grad_norm": 0.10134711861610413,
      "learning_rate": 8.192068429237948e-05,
      "loss": 0.019135287031531334,
      "step": 5808
    },
    {
      "epoch": 1.101213040181956,
      "grad_norm": 0.14354181289672852,
      "learning_rate": 8.184292379471228e-05,
      "loss": 0.012987041845917702,
      "step": 5810
    },
    {
      "epoch": 1.1015921152388173,
      "grad_norm": 0.441080778837204,
      "learning_rate": 8.17651632970451e-05,
      "loss": 0.020535020157694817,
      "step": 5812
    },
    {
      "epoch": 1.1019711902956786,
      "grad_norm": 0.449971467256546,
      "learning_rate": 8.168740279937792e-05,
      "loss": 0.02026291936635971,
      "step": 5814
    },
    {
      "epoch": 1.1023502653525399,
      "grad_norm": 0.09690073877573013,
      "learning_rate": 8.160964230171073e-05,
      "loss": 0.011836235411465168,
      "step": 5816
    },
    {
      "epoch": 1.102729340409401,
      "grad_norm": 0.1294454038143158,
      "learning_rate": 8.153188180404355e-05,
      "loss": 0.01695035584270954,
      "step": 5818
    },
    {
      "epoch": 1.1031084154662623,
      "grad_norm": 0.1286335438489914,
      "learning_rate": 8.145412130637636e-05,
      "loss": 0.016248637810349464,
      "step": 5820
    },
    {
      "epoch": 1.1034874905231236,
      "grad_norm": 0.18130916357040405,
      "learning_rate": 8.137636080870918e-05,
      "loss": 0.01780240423977375,
      "step": 5822
    },
    {
      "epoch": 1.1038665655799849,
      "grad_norm": 0.0836791917681694,
      "learning_rate": 8.1298600311042e-05,
      "loss": 0.012022938579320908,
      "step": 5824
    },
    {
      "epoch": 1.1042456406368462,
      "grad_norm": 0.209860160946846,
      "learning_rate": 8.12208398133748e-05,
      "loss": 0.02594844438135624,
      "step": 5826
    },
    {
      "epoch": 1.1046247156937072,
      "grad_norm": 0.09290865808725357,
      "learning_rate": 8.114307931570763e-05,
      "loss": 0.012218613177537918,
      "step": 5828
    },
    {
      "epoch": 1.1050037907505685,
      "grad_norm": 0.18204183876514435,
      "learning_rate": 8.106531881804043e-05,
      "loss": 0.01978760026395321,
      "step": 5830
    },
    {
      "epoch": 1.1053828658074298,
      "grad_norm": 0.10663839429616928,
      "learning_rate": 8.098755832037325e-05,
      "loss": 0.01560982409864664,
      "step": 5832
    },
    {
      "epoch": 1.1057619408642911,
      "grad_norm": 0.281179279088974,
      "learning_rate": 8.090979782270607e-05,
      "loss": 0.01992403157055378,
      "step": 5834
    },
    {
      "epoch": 1.1061410159211524,
      "grad_norm": 0.15661217272281647,
      "learning_rate": 8.08320373250389e-05,
      "loss": 0.017737368121743202,
      "step": 5836
    },
    {
      "epoch": 1.1065200909780137,
      "grad_norm": 0.29639148712158203,
      "learning_rate": 8.07542768273717e-05,
      "loss": 0.025359662249684334,
      "step": 5838
    },
    {
      "epoch": 1.1068991660348748,
      "grad_norm": 0.08383413404226303,
      "learning_rate": 8.067651632970451e-05,
      "loss": 0.01903291419148445,
      "step": 5840
    },
    {
      "epoch": 1.1072782410917361,
      "grad_norm": 0.12550689280033112,
      "learning_rate": 8.059875583203733e-05,
      "loss": 0.010732043534517288,
      "step": 5842
    },
    {
      "epoch": 1.1076573161485974,
      "grad_norm": 0.0918327048420906,
      "learning_rate": 8.052099533437014e-05,
      "loss": 0.01378362812101841,
      "step": 5844
    },
    {
      "epoch": 1.1080363912054587,
      "grad_norm": 0.246381476521492,
      "learning_rate": 8.044323483670297e-05,
      "loss": 0.014900433830916882,
      "step": 5846
    },
    {
      "epoch": 1.10841546626232,
      "grad_norm": 0.28843265771865845,
      "learning_rate": 8.036547433903578e-05,
      "loss": 0.03249835595488548,
      "step": 5848
    },
    {
      "epoch": 1.108794541319181,
      "grad_norm": 0.3216630220413208,
      "learning_rate": 8.028771384136858e-05,
      "loss": 0.0164847020059824,
      "step": 5850
    },
    {
      "epoch": 1.1091736163760424,
      "grad_norm": 0.08473149687051773,
      "learning_rate": 8.02099533437014e-05,
      "loss": 0.010447846725583076,
      "step": 5852
    },
    {
      "epoch": 1.1095526914329037,
      "grad_norm": 0.09316760301589966,
      "learning_rate": 8.013219284603421e-05,
      "loss": 0.016835616901516914,
      "step": 5854
    },
    {
      "epoch": 1.109931766489765,
      "grad_norm": 0.33693942427635193,
      "learning_rate": 8.005443234836703e-05,
      "loss": 0.022310450673103333,
      "step": 5856
    },
    {
      "epoch": 1.1103108415466263,
      "grad_norm": 0.08451282232999802,
      "learning_rate": 7.997667185069985e-05,
      "loss": 0.01807580143213272,
      "step": 5858
    },
    {
      "epoch": 1.1106899166034876,
      "grad_norm": 0.3141365647315979,
      "learning_rate": 7.989891135303267e-05,
      "loss": 0.017764201387763023,
      "step": 5860
    },
    {
      "epoch": 1.1110689916603487,
      "grad_norm": 0.29069650173187256,
      "learning_rate": 7.982115085536548e-05,
      "loss": 0.026148255914449692,
      "step": 5862
    },
    {
      "epoch": 1.11144806671721,
      "grad_norm": 0.0885644182562828,
      "learning_rate": 7.974339035769829e-05,
      "loss": 0.017061254009604454,
      "step": 5864
    },
    {
      "epoch": 1.1118271417740713,
      "grad_norm": 0.09672714024782181,
      "learning_rate": 7.96656298600311e-05,
      "loss": 0.010077008977532387,
      "step": 5866
    },
    {
      "epoch": 1.1122062168309326,
      "grad_norm": 0.08317916840314865,
      "learning_rate": 7.958786936236393e-05,
      "loss": 0.012880380265414715,
      "step": 5868
    },
    {
      "epoch": 1.1125852918877939,
      "grad_norm": 0.2507750391960144,
      "learning_rate": 7.951010886469675e-05,
      "loss": 0.014273663982748985,
      "step": 5870
    },
    {
      "epoch": 1.112964366944655,
      "grad_norm": 0.22323991358280182,
      "learning_rate": 7.943234836702955e-05,
      "loss": 0.02411886490881443,
      "step": 5872
    },
    {
      "epoch": 1.1133434420015162,
      "grad_norm": 0.13752122223377228,
      "learning_rate": 7.935458786936236e-05,
      "loss": 0.013590113259851933,
      "step": 5874
    },
    {
      "epoch": 1.1137225170583775,
      "grad_norm": 0.3435041606426239,
      "learning_rate": 7.927682737169518e-05,
      "loss": 0.023126445710659027,
      "step": 5876
    },
    {
      "epoch": 1.1141015921152388,
      "grad_norm": 0.29189440608024597,
      "learning_rate": 7.919906687402799e-05,
      "loss": 0.023394959047436714,
      "step": 5878
    },
    {
      "epoch": 1.1144806671721001,
      "grad_norm": 0.11274904012680054,
      "learning_rate": 7.912130637636082e-05,
      "loss": 0.022115958854556084,
      "step": 5880
    },
    {
      "epoch": 1.1148597422289614,
      "grad_norm": 0.29991522431373596,
      "learning_rate": 7.904354587869363e-05,
      "loss": 0.014874645508825779,
      "step": 5882
    },
    {
      "epoch": 1.1152388172858225,
      "grad_norm": 0.29410305619239807,
      "learning_rate": 7.896578538102645e-05,
      "loss": 0.01877550780773163,
      "step": 5884
    },
    {
      "epoch": 1.1156178923426838,
      "grad_norm": 0.3042997419834137,
      "learning_rate": 7.888802488335926e-05,
      "loss": 0.026075389236211777,
      "step": 5886
    },
    {
      "epoch": 1.1159969673995451,
      "grad_norm": 0.3463083803653717,
      "learning_rate": 7.881026438569206e-05,
      "loss": 0.02162707969546318,
      "step": 5888
    },
    {
      "epoch": 1.1163760424564064,
      "grad_norm": 0.13696040213108063,
      "learning_rate": 7.873250388802488e-05,
      "loss": 0.021822959184646606,
      "step": 5890
    },
    {
      "epoch": 1.1167551175132677,
      "grad_norm": 0.31922027468681335,
      "learning_rate": 7.86547433903577e-05,
      "loss": 0.018942086026072502,
      "step": 5892
    },
    {
      "epoch": 1.1171341925701288,
      "grad_norm": 0.10976080596446991,
      "learning_rate": 7.857698289269052e-05,
      "loss": 0.011979905888438225,
      "step": 5894
    },
    {
      "epoch": 1.11751326762699,
      "grad_norm": 0.3139943778514862,
      "learning_rate": 7.849922239502333e-05,
      "loss": 0.02255493588745594,
      "step": 5896
    },
    {
      "epoch": 1.1178923426838514,
      "grad_norm": 0.32466214895248413,
      "learning_rate": 7.842146189735615e-05,
      "loss": 0.024965740740299225,
      "step": 5898
    },
    {
      "epoch": 1.1182714177407127,
      "grad_norm": 0.2829338610172272,
      "learning_rate": 7.834370139968896e-05,
      "loss": 0.02249298244714737,
      "step": 5900
    },
    {
      "epoch": 1.118650492797574,
      "grad_norm": 0.1750907301902771,
      "learning_rate": 7.826594090202178e-05,
      "loss": 0.014219935983419418,
      "step": 5902
    },
    {
      "epoch": 1.119029567854435,
      "grad_norm": 0.2808343172073364,
      "learning_rate": 7.81881804043546e-05,
      "loss": 0.01669846847653389,
      "step": 5904
    },
    {
      "epoch": 1.1194086429112964,
      "grad_norm": 0.2411644458770752,
      "learning_rate": 7.81104199066874e-05,
      "loss": 0.01517188735306263,
      "step": 5906
    },
    {
      "epoch": 1.1197877179681577,
      "grad_norm": 0.3306577503681183,
      "learning_rate": 7.803265940902023e-05,
      "loss": 0.018166590481996536,
      "step": 5908
    },
    {
      "epoch": 1.120166793025019,
      "grad_norm": 0.08697523176670074,
      "learning_rate": 7.795489891135303e-05,
      "loss": 0.013091147877275944,
      "step": 5910
    },
    {
      "epoch": 1.1205458680818803,
      "grad_norm": 0.2084471732378006,
      "learning_rate": 7.787713841368584e-05,
      "loss": 0.013921152800321579,
      "step": 5912
    },
    {
      "epoch": 1.1209249431387414,
      "grad_norm": 0.14051729440689087,
      "learning_rate": 7.779937791601866e-05,
      "loss": 0.01249717641621828,
      "step": 5914
    },
    {
      "epoch": 1.1213040181956027,
      "grad_norm": 0.11519020050764084,
      "learning_rate": 7.772161741835148e-05,
      "loss": 0.021594563499093056,
      "step": 5916
    },
    {
      "epoch": 1.121683093252464,
      "grad_norm": 0.1640649437904358,
      "learning_rate": 7.76438569206843e-05,
      "loss": 0.017050568014383316,
      "step": 5918
    },
    {
      "epoch": 1.1220621683093253,
      "grad_norm": 0.3656827509403229,
      "learning_rate": 7.756609642301711e-05,
      "loss": 0.021385280415415764,
      "step": 5920
    },
    {
      "epoch": 1.1224412433661866,
      "grad_norm": 0.14820504188537598,
      "learning_rate": 7.748833592534993e-05,
      "loss": 0.011868583038449287,
      "step": 5922
    },
    {
      "epoch": 1.1228203184230479,
      "grad_norm": 0.0989629328250885,
      "learning_rate": 7.741057542768274e-05,
      "loss": 0.0171363465487957,
      "step": 5924
    },
    {
      "epoch": 1.123199393479909,
      "grad_norm": 0.2806685268878937,
      "learning_rate": 7.733281493001556e-05,
      "loss": 0.023123467341065407,
      "step": 5926
    },
    {
      "epoch": 1.1235784685367702,
      "grad_norm": 0.09536155313253403,
      "learning_rate": 7.725505443234838e-05,
      "loss": 0.012058693915605545,
      "step": 5928
    },
    {
      "epoch": 1.1239575435936315,
      "grad_norm": 0.07417938113212585,
      "learning_rate": 7.717729393468118e-05,
      "loss": 0.016839507967233658,
      "step": 5930
    },
    {
      "epoch": 1.1243366186504928,
      "grad_norm": 0.08716905862092972,
      "learning_rate": 7.7099533437014e-05,
      "loss": 0.015193622559309006,
      "step": 5932
    },
    {
      "epoch": 1.1247156937073541,
      "grad_norm": 0.17144733667373657,
      "learning_rate": 7.702177293934681e-05,
      "loss": 0.013569163158535957,
      "step": 5934
    },
    {
      "epoch": 1.1250947687642152,
      "grad_norm": 0.26437944173812866,
      "learning_rate": 7.694401244167963e-05,
      "loss": 0.021261556074023247,
      "step": 5936
    },
    {
      "epoch": 1.1254738438210765,
      "grad_norm": 0.19503434002399445,
      "learning_rate": 7.686625194401245e-05,
      "loss": 0.012760329991579056,
      "step": 5938
    },
    {
      "epoch": 1.1258529188779378,
      "grad_norm": 0.10461946576833725,
      "learning_rate": 7.678849144634526e-05,
      "loss": 0.013322890736162663,
      "step": 5940
    },
    {
      "epoch": 1.1262319939347991,
      "grad_norm": 0.24806125462055206,
      "learning_rate": 7.671073094867808e-05,
      "loss": 0.0271754190325737,
      "step": 5942
    },
    {
      "epoch": 1.1266110689916604,
      "grad_norm": 0.11852440983057022,
      "learning_rate": 7.663297045101089e-05,
      "loss": 0.014081060886383057,
      "step": 5944
    },
    {
      "epoch": 1.1269901440485217,
      "grad_norm": 0.10843423008918762,
      "learning_rate": 7.65552099533437e-05,
      "loss": 0.019838295876979828,
      "step": 5946
    },
    {
      "epoch": 1.1273692191053828,
      "grad_norm": 0.08755998313426971,
      "learning_rate": 7.647744945567651e-05,
      "loss": 0.012935087084770203,
      "step": 5948
    },
    {
      "epoch": 1.127748294162244,
      "grad_norm": 0.11810675263404846,
      "learning_rate": 7.639968895800933e-05,
      "loss": 0.01388112735003233,
      "step": 5950
    },
    {
      "epoch": 1.1281273692191054,
      "grad_norm": 0.10192784667015076,
      "learning_rate": 7.632192846034215e-05,
      "loss": 0.012969860807061195,
      "step": 5952
    },
    {
      "epoch": 1.1285064442759667,
      "grad_norm": 0.08634641766548157,
      "learning_rate": 7.624416796267496e-05,
      "loss": 0.011562301777303219,
      "step": 5954
    },
    {
      "epoch": 1.128885519332828,
      "grad_norm": 0.31848040223121643,
      "learning_rate": 7.616640746500778e-05,
      "loss": 0.017598634585738182,
      "step": 5956
    },
    {
      "epoch": 1.129264594389689,
      "grad_norm": 0.5531744360923767,
      "learning_rate": 7.608864696734059e-05,
      "loss": 0.028178183361887932,
      "step": 5958
    },
    {
      "epoch": 1.1296436694465504,
      "grad_norm": 0.3339092433452606,
      "learning_rate": 7.601088646967341e-05,
      "loss": 0.02169998176395893,
      "step": 5960
    },
    {
      "epoch": 1.1300227445034117,
      "grad_norm": 0.2622300088405609,
      "learning_rate": 7.593312597200623e-05,
      "loss": 0.022876230999827385,
      "step": 5962
    },
    {
      "epoch": 1.130401819560273,
      "grad_norm": 0.3659355938434601,
      "learning_rate": 7.585536547433904e-05,
      "loss": 0.024263296276330948,
      "step": 5964
    },
    {
      "epoch": 1.1307808946171343,
      "grad_norm": 0.11715426295995712,
      "learning_rate": 7.577760497667186e-05,
      "loss": 0.013687221333384514,
      "step": 5966
    },
    {
      "epoch": 1.1311599696739956,
      "grad_norm": 0.18992897868156433,
      "learning_rate": 7.569984447900466e-05,
      "loss": 0.016122210770845413,
      "step": 5968
    },
    {
      "epoch": 1.1315390447308566,
      "grad_norm": 0.4144405722618103,
      "learning_rate": 7.562208398133748e-05,
      "loss": 0.02797575108706951,
      "step": 5970
    },
    {
      "epoch": 1.131918119787718,
      "grad_norm": 0.25550538301467896,
      "learning_rate": 7.55443234836703e-05,
      "loss": 0.013542693108320236,
      "step": 5972
    },
    {
      "epoch": 1.1322971948445792,
      "grad_norm": 0.0990130603313446,
      "learning_rate": 7.546656298600311e-05,
      "loss": 0.02330790087580681,
      "step": 5974
    },
    {
      "epoch": 1.1326762699014405,
      "grad_norm": 0.276313453912735,
      "learning_rate": 7.538880248833593e-05,
      "loss": 0.027730701491236687,
      "step": 5976
    },
    {
      "epoch": 1.1330553449583016,
      "grad_norm": 0.22229041159152985,
      "learning_rate": 7.531104199066874e-05,
      "loss": 0.013657876290380955,
      "step": 5978
    },
    {
      "epoch": 1.133434420015163,
      "grad_norm": 0.08066880702972412,
      "learning_rate": 7.523328149300156e-05,
      "loss": 0.011136041954159737,
      "step": 5980
    },
    {
      "epoch": 1.1338134950720242,
      "grad_norm": 0.41624292731285095,
      "learning_rate": 7.515552099533437e-05,
      "loss": 0.027583541348576546,
      "step": 5982
    },
    {
      "epoch": 1.1341925701288855,
      "grad_norm": 0.10815082490444183,
      "learning_rate": 7.50777604976672e-05,
      "loss": 0.01688445918262005,
      "step": 5984
    },
    {
      "epoch": 1.1345716451857468,
      "grad_norm": 0.09794727712869644,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.0188006404787302,
      "step": 5986
    },
    {
      "epoch": 1.1349507202426081,
      "grad_norm": 0.1291080117225647,
      "learning_rate": 7.492223950233281e-05,
      "loss": 0.02004930004477501,
      "step": 5988
    },
    {
      "epoch": 1.1353297952994692,
      "grad_norm": 0.12054335325956345,
      "learning_rate": 7.484447900466563e-05,
      "loss": 0.012025636620819569,
      "step": 5990
    },
    {
      "epoch": 1.1357088703563305,
      "grad_norm": 0.07303113490343094,
      "learning_rate": 7.476671850699844e-05,
      "loss": 0.015582721680402756,
      "step": 5992
    },
    {
      "epoch": 1.1360879454131918,
      "grad_norm": 0.12745502591133118,
      "learning_rate": 7.468895800933126e-05,
      "loss": 0.019310347735881805,
      "step": 5994
    },
    {
      "epoch": 1.136467020470053,
      "grad_norm": 0.09071961045265198,
      "learning_rate": 7.461119751166408e-05,
      "loss": 0.01801878586411476,
      "step": 5996
    },
    {
      "epoch": 1.1368460955269144,
      "grad_norm": 0.10573235154151917,
      "learning_rate": 7.45334370139969e-05,
      "loss": 0.014423402957618237,
      "step": 5998
    },
    {
      "epoch": 1.1372251705837755,
      "grad_norm": 0.08523380011320114,
      "learning_rate": 7.445567651632971e-05,
      "loss": 0.014407815411686897,
      "step": 6000
    },
    {
      "epoch": 1.1376042456406368,
      "grad_norm": 0.08498004078865051,
      "learning_rate": 7.437791601866252e-05,
      "loss": 0.011835074983537197,
      "step": 6002
    },
    {
      "epoch": 1.137983320697498,
      "grad_norm": 0.09691985696554184,
      "learning_rate": 7.430015552099534e-05,
      "loss": 0.01804182678461075,
      "step": 6004
    },
    {
      "epoch": 1.1383623957543594,
      "grad_norm": 0.1424732804298401,
      "learning_rate": 7.422239502332816e-05,
      "loss": 0.01438089832663536,
      "step": 6006
    },
    {
      "epoch": 1.1387414708112207,
      "grad_norm": 0.29119226336479187,
      "learning_rate": 7.414463452566098e-05,
      "loss": 0.016658861190080643,
      "step": 6008
    },
    {
      "epoch": 1.139120545868082,
      "grad_norm": 0.24298155307769775,
      "learning_rate": 7.406687402799378e-05,
      "loss": 0.019658979028463364,
      "step": 6010
    },
    {
      "epoch": 1.139499620924943,
      "grad_norm": 0.10806798189878464,
      "learning_rate": 7.398911353032659e-05,
      "loss": 0.015886414796113968,
      "step": 6012
    },
    {
      "epoch": 1.1398786959818044,
      "grad_norm": 0.10863487422466278,
      "learning_rate": 7.391135303265941e-05,
      "loss": 0.015157831832766533,
      "step": 6014
    },
    {
      "epoch": 1.1402577710386657,
      "grad_norm": 0.08597935736179352,
      "learning_rate": 7.383359253499222e-05,
      "loss": 0.0164040494710207,
      "step": 6016
    },
    {
      "epoch": 1.140636846095527,
      "grad_norm": 0.07637980580329895,
      "learning_rate": 7.375583203732504e-05,
      "loss": 0.018783142790198326,
      "step": 6018
    },
    {
      "epoch": 1.1410159211523883,
      "grad_norm": 0.17562580108642578,
      "learning_rate": 7.367807153965786e-05,
      "loss": 0.013129929080605507,
      "step": 6020
    },
    {
      "epoch": 1.1413949962092493,
      "grad_norm": 0.28238147497177124,
      "learning_rate": 7.360031104199068e-05,
      "loss": 0.02804134227335453,
      "step": 6022
    },
    {
      "epoch": 1.1417740712661106,
      "grad_norm": 0.3503873348236084,
      "learning_rate": 7.352255054432349e-05,
      "loss": 0.022464249283075333,
      "step": 6024
    },
    {
      "epoch": 1.142153146322972,
      "grad_norm": 0.3430752754211426,
      "learning_rate": 7.34447900466563e-05,
      "loss": 0.022631986066699028,
      "step": 6026
    },
    {
      "epoch": 1.1425322213798332,
      "grad_norm": 0.38262423872947693,
      "learning_rate": 7.336702954898911e-05,
      "loss": 0.0197780579328537,
      "step": 6028
    },
    {
      "epoch": 1.1429112964366945,
      "grad_norm": 0.15496809780597687,
      "learning_rate": 7.328926905132193e-05,
      "loss": 0.014225631020963192,
      "step": 6030
    },
    {
      "epoch": 1.1432903714935558,
      "grad_norm": 0.33121517300605774,
      "learning_rate": 7.321150855365476e-05,
      "loss": 0.016147147864103317,
      "step": 6032
    },
    {
      "epoch": 1.143669446550417,
      "grad_norm": 0.304617315530777,
      "learning_rate": 7.313374805598756e-05,
      "loss": 0.02152477204799652,
      "step": 6034
    },
    {
      "epoch": 1.1440485216072782,
      "grad_norm": 0.10520640015602112,
      "learning_rate": 7.305598755832037e-05,
      "loss": 0.014464126899838448,
      "step": 6036
    },
    {
      "epoch": 1.1444275966641395,
      "grad_norm": 0.3736661672592163,
      "learning_rate": 7.297822706065319e-05,
      "loss": 0.032950203865766525,
      "step": 6038
    },
    {
      "epoch": 1.1448066717210008,
      "grad_norm": 0.3648495674133301,
      "learning_rate": 7.2900466562986e-05,
      "loss": 0.02177327498793602,
      "step": 6040
    },
    {
      "epoch": 1.1451857467778621,
      "grad_norm": 0.11490315943956375,
      "learning_rate": 7.282270606531883e-05,
      "loss": 0.012765134684741497,
      "step": 6042
    },
    {
      "epoch": 1.1455648218347232,
      "grad_norm": 0.09887164831161499,
      "learning_rate": 7.274494556765164e-05,
      "loss": 0.017072562128305435,
      "step": 6044
    },
    {
      "epoch": 1.1459438968915845,
      "grad_norm": 0.17076286673545837,
      "learning_rate": 7.266718506998446e-05,
      "loss": 0.013494539074599743,
      "step": 6046
    },
    {
      "epoch": 1.1463229719484458,
      "grad_norm": 0.3896145820617676,
      "learning_rate": 7.258942457231726e-05,
      "loss": 0.01707424782216549,
      "step": 6048
    },
    {
      "epoch": 1.146702047005307,
      "grad_norm": 0.08882646262645721,
      "learning_rate": 7.251166407465007e-05,
      "loss": 0.01343767810612917,
      "step": 6050
    },
    {
      "epoch": 1.1470811220621684,
      "grad_norm": 0.10774257034063339,
      "learning_rate": 7.243390357698289e-05,
      "loss": 0.019343968480825424,
      "step": 6052
    },
    {
      "epoch": 1.1474601971190297,
      "grad_norm": 0.3759993612766266,
      "learning_rate": 7.235614307931571e-05,
      "loss": 0.022322654724121094,
      "step": 6054
    },
    {
      "epoch": 1.1478392721758908,
      "grad_norm": 0.4366273581981659,
      "learning_rate": 7.227838258164853e-05,
      "loss": 0.024223417043685913,
      "step": 6056
    },
    {
      "epoch": 1.148218347232752,
      "grad_norm": 0.09469181299209595,
      "learning_rate": 7.220062208398134e-05,
      "loss": 0.013021920807659626,
      "step": 6058
    },
    {
      "epoch": 1.1485974222896134,
      "grad_norm": 0.08875107020139694,
      "learning_rate": 7.212286158631416e-05,
      "loss": 0.0166095532476902,
      "step": 6060
    },
    {
      "epoch": 1.1489764973464747,
      "grad_norm": 0.24370919167995453,
      "learning_rate": 7.204510108864697e-05,
      "loss": 0.01908363401889801,
      "step": 6062
    },
    {
      "epoch": 1.1493555724033357,
      "grad_norm": 0.09592143446207047,
      "learning_rate": 7.196734059097979e-05,
      "loss": 0.017263270914554596,
      "step": 6064
    },
    {
      "epoch": 1.149734647460197,
      "grad_norm": 0.10783033072948456,
      "learning_rate": 7.188958009331261e-05,
      "loss": 0.01905996911227703,
      "step": 6066
    },
    {
      "epoch": 1.1501137225170583,
      "grad_norm": 0.2635158896446228,
      "learning_rate": 7.181181959564541e-05,
      "loss": 0.014958742074668407,
      "step": 6068
    },
    {
      "epoch": 1.1504927975739196,
      "grad_norm": 0.0916750505566597,
      "learning_rate": 7.173405909797823e-05,
      "loss": 0.01573101244866848,
      "step": 6070
    },
    {
      "epoch": 1.150871872630781,
      "grad_norm": 0.2692079544067383,
      "learning_rate": 7.165629860031104e-05,
      "loss": 0.015621587634086609,
      "step": 6072
    },
    {
      "epoch": 1.1512509476876422,
      "grad_norm": 0.24781404435634613,
      "learning_rate": 7.157853810264385e-05,
      "loss": 0.02032293565571308,
      "step": 6074
    },
    {
      "epoch": 1.1516300227445033,
      "grad_norm": 0.08097449690103531,
      "learning_rate": 7.150077760497668e-05,
      "loss": 0.00977319572120905,
      "step": 6076
    },
    {
      "epoch": 1.1520090978013646,
      "grad_norm": 0.2793129086494446,
      "learning_rate": 7.142301710730949e-05,
      "loss": 0.019040485844016075,
      "step": 6078
    },
    {
      "epoch": 1.152388172858226,
      "grad_norm": 0.40270695090293884,
      "learning_rate": 7.134525660964231e-05,
      "loss": 0.02991372160613537,
      "step": 6080
    },
    {
      "epoch": 1.1527672479150872,
      "grad_norm": 0.11576355248689651,
      "learning_rate": 7.126749611197512e-05,
      "loss": 0.013147321529686451,
      "step": 6082
    },
    {
      "epoch": 1.1531463229719485,
      "grad_norm": 0.3070548474788666,
      "learning_rate": 7.118973561430794e-05,
      "loss": 0.019320223480463028,
      "step": 6084
    },
    {
      "epoch": 1.1535253980288096,
      "grad_norm": 0.08873246610164642,
      "learning_rate": 7.111197511664074e-05,
      "loss": 0.011559290811419487,
      "step": 6086
    },
    {
      "epoch": 1.153904473085671,
      "grad_norm": 0.21330858767032623,
      "learning_rate": 7.103421461897356e-05,
      "loss": 0.011970633640885353,
      "step": 6088
    },
    {
      "epoch": 1.1542835481425322,
      "grad_norm": 0.07951776683330536,
      "learning_rate": 7.095645412130639e-05,
      "loss": 0.010415111668407917,
      "step": 6090
    },
    {
      "epoch": 1.1546626231993935,
      "grad_norm": 0.23528195917606354,
      "learning_rate": 7.087869362363919e-05,
      "loss": 0.016636112704873085,
      "step": 6092
    },
    {
      "epoch": 1.1550416982562548,
      "grad_norm": 0.1448570340871811,
      "learning_rate": 7.080093312597201e-05,
      "loss": 0.014971854165196419,
      "step": 6094
    },
    {
      "epoch": 1.155420773313116,
      "grad_norm": 0.32740965485572815,
      "learning_rate": 7.072317262830482e-05,
      "loss": 0.021763106808066368,
      "step": 6096
    },
    {
      "epoch": 1.1557998483699772,
      "grad_norm": 0.09230723977088928,
      "learning_rate": 7.064541213063764e-05,
      "loss": 0.015059573575854301,
      "step": 6098
    },
    {
      "epoch": 1.1561789234268385,
      "grad_norm": 0.24218222498893738,
      "learning_rate": 7.056765163297046e-05,
      "loss": 0.015857473015785217,
      "step": 6100
    },
    {
      "epoch": 1.1565579984836998,
      "grad_norm": 0.3034389913082123,
      "learning_rate": 7.048989113530327e-05,
      "loss": 0.02647155523300171,
      "step": 6102
    },
    {
      "epoch": 1.156937073540561,
      "grad_norm": 0.26081550121307373,
      "learning_rate": 7.041213063763609e-05,
      "loss": 0.025291020050644875,
      "step": 6104
    },
    {
      "epoch": 1.1573161485974224,
      "grad_norm": 0.10060757398605347,
      "learning_rate": 7.03343701399689e-05,
      "loss": 0.016856051981449127,
      "step": 6106
    },
    {
      "epoch": 1.1576952236542835,
      "grad_norm": 0.34306684136390686,
      "learning_rate": 7.025660964230171e-05,
      "loss": 0.020363014191389084,
      "step": 6108
    },
    {
      "epoch": 1.1580742987111448,
      "grad_norm": 0.08239816129207611,
      "learning_rate": 7.017884914463454e-05,
      "loss": 0.014310736209154129,
      "step": 6110
    },
    {
      "epoch": 1.158453373768006,
      "grad_norm": 0.30939024686813354,
      "learning_rate": 7.010108864696734e-05,
      "loss": 0.01967807300388813,
      "step": 6112
    },
    {
      "epoch": 1.1588324488248674,
      "grad_norm": 0.32195213437080383,
      "learning_rate": 7.002332814930016e-05,
      "loss": 0.017636068165302277,
      "step": 6114
    },
    {
      "epoch": 1.1592115238817287,
      "grad_norm": 0.08740177005529404,
      "learning_rate": 6.994556765163297e-05,
      "loss": 0.01728089712560177,
      "step": 6116
    },
    {
      "epoch": 1.15959059893859,
      "grad_norm": 0.46835246682167053,
      "learning_rate": 6.986780715396579e-05,
      "loss": 0.03416753560304642,
      "step": 6118
    },
    {
      "epoch": 1.159969673995451,
      "grad_norm": 0.12026500701904297,
      "learning_rate": 6.97900466562986e-05,
      "loss": 0.012061400339007378,
      "step": 6120
    },
    {
      "epoch": 1.1603487490523123,
      "grad_norm": 0.1103426143527031,
      "learning_rate": 6.971228615863142e-05,
      "loss": 0.014273455366492271,
      "step": 6122
    },
    {
      "epoch": 1.1607278241091736,
      "grad_norm": 0.3227853775024414,
      "learning_rate": 6.963452566096424e-05,
      "loss": 0.022407107055187225,
      "step": 6124
    },
    {
      "epoch": 1.161106899166035,
      "grad_norm": 0.08699873834848404,
      "learning_rate": 6.955676516329704e-05,
      "loss": 0.016929369419813156,
      "step": 6126
    },
    {
      "epoch": 1.161485974222896,
      "grad_norm": 0.07840058207511902,
      "learning_rate": 6.947900466562987e-05,
      "loss": 0.014335932210087776,
      "step": 6128
    },
    {
      "epoch": 1.1618650492797573,
      "grad_norm": 0.08430197834968567,
      "learning_rate": 6.940124416796267e-05,
      "loss": 0.014459691010415554,
      "step": 6130
    },
    {
      "epoch": 1.1622441243366186,
      "grad_norm": 0.07906344532966614,
      "learning_rate": 6.932348367029549e-05,
      "loss": 0.015130841173231602,
      "step": 6132
    },
    {
      "epoch": 1.16262319939348,
      "grad_norm": 0.2846616208553314,
      "learning_rate": 6.924572317262831e-05,
      "loss": 0.022143462672829628,
      "step": 6134
    },
    {
      "epoch": 1.1630022744503412,
      "grad_norm": 0.08427365124225616,
      "learning_rate": 6.916796267496112e-05,
      "loss": 0.011680971831083298,
      "step": 6136
    },
    {
      "epoch": 1.1633813495072025,
      "grad_norm": 0.10361579805612564,
      "learning_rate": 6.909020217729394e-05,
      "loss": 0.011824713088572025,
      "step": 6138
    },
    {
      "epoch": 1.1637604245640636,
      "grad_norm": 0.32525697350502014,
      "learning_rate": 6.901244167962675e-05,
      "loss": 0.0210051778703928,
      "step": 6140
    },
    {
      "epoch": 1.164139499620925,
      "grad_norm": 0.28249648213386536,
      "learning_rate": 6.893468118195957e-05,
      "loss": 0.019904866814613342,
      "step": 6142
    },
    {
      "epoch": 1.1645185746777862,
      "grad_norm": 0.24465201795101166,
      "learning_rate": 6.885692068429237e-05,
      "loss": 0.013208193704485893,
      "step": 6144
    },
    {
      "epoch": 1.1648976497346475,
      "grad_norm": 0.10095037519931793,
      "learning_rate": 6.877916018662521e-05,
      "loss": 0.013491987250745296,
      "step": 6146
    },
    {
      "epoch": 1.1652767247915088,
      "grad_norm": 0.1419881284236908,
      "learning_rate": 6.870139968895802e-05,
      "loss": 0.022133175283670425,
      "step": 6148
    },
    {
      "epoch": 1.1656557998483699,
      "grad_norm": 0.18097567558288574,
      "learning_rate": 6.862363919129082e-05,
      "loss": 0.017917422577738762,
      "step": 6150
    },
    {
      "epoch": 1.1660348749052312,
      "grad_norm": 0.25109925866127014,
      "learning_rate": 6.854587869362364e-05,
      "loss": 0.03031531162559986,
      "step": 6152
    },
    {
      "epoch": 1.1664139499620925,
      "grad_norm": 0.09025678038597107,
      "learning_rate": 6.846811819595645e-05,
      "loss": 0.017782287672162056,
      "step": 6154
    },
    {
      "epoch": 1.1667930250189538,
      "grad_norm": 0.28924447298049927,
      "learning_rate": 6.839035769828927e-05,
      "loss": 0.0192362442612648,
      "step": 6156
    },
    {
      "epoch": 1.167172100075815,
      "grad_norm": 0.3045401871204376,
      "learning_rate": 6.831259720062209e-05,
      "loss": 0.0212149228900671,
      "step": 6158
    },
    {
      "epoch": 1.1675511751326764,
      "grad_norm": 0.100053571164608,
      "learning_rate": 6.823483670295491e-05,
      "loss": 0.012786848470568657,
      "step": 6160
    },
    {
      "epoch": 1.1679302501895374,
      "grad_norm": 0.14004528522491455,
      "learning_rate": 6.815707620528772e-05,
      "loss": 0.02379642054438591,
      "step": 6162
    },
    {
      "epoch": 1.1683093252463987,
      "grad_norm": 0.12865012884140015,
      "learning_rate": 6.807931570762052e-05,
      "loss": 0.01278962753713131,
      "step": 6164
    },
    {
      "epoch": 1.16868840030326,
      "grad_norm": 0.33248791098594666,
      "learning_rate": 6.800155520995334e-05,
      "loss": 0.025575414299964905,
      "step": 6166
    },
    {
      "epoch": 1.1690674753601213,
      "grad_norm": 0.1061965823173523,
      "learning_rate": 6.792379471228617e-05,
      "loss": 0.016359196975827217,
      "step": 6168
    },
    {
      "epoch": 1.1694465504169826,
      "grad_norm": 0.122673399746418,
      "learning_rate": 6.784603421461899e-05,
      "loss": 0.021134160459041595,
      "step": 6170
    },
    {
      "epoch": 1.1698256254738437,
      "grad_norm": 0.09815099090337753,
      "learning_rate": 6.776827371695179e-05,
      "loss": 0.013807849027216434,
      "step": 6172
    },
    {
      "epoch": 1.170204700530705,
      "grad_norm": 0.30289512872695923,
      "learning_rate": 6.76905132192846e-05,
      "loss": 0.026420969516038895,
      "step": 6174
    },
    {
      "epoch": 1.1705837755875663,
      "grad_norm": 0.10664902627468109,
      "learning_rate": 6.761275272161742e-05,
      "loss": 0.016474712640047073,
      "step": 6176
    },
    {
      "epoch": 1.1709628506444276,
      "grad_norm": 0.4093254506587982,
      "learning_rate": 6.753499222395023e-05,
      "loss": 0.02508496306836605,
      "step": 6178
    },
    {
      "epoch": 1.171341925701289,
      "grad_norm": 0.3154144287109375,
      "learning_rate": 6.745723172628306e-05,
      "loss": 0.018629197031259537,
      "step": 6180
    },
    {
      "epoch": 1.1717210007581502,
      "grad_norm": 0.09396287798881531,
      "learning_rate": 6.737947122861587e-05,
      "loss": 0.02017476223409176,
      "step": 6182
    },
    {
      "epoch": 1.1721000758150113,
      "grad_norm": 0.08739585429430008,
      "learning_rate": 6.730171073094869e-05,
      "loss": 0.018216239288449287,
      "step": 6184
    },
    {
      "epoch": 1.1724791508718726,
      "grad_norm": 0.20080429315567017,
      "learning_rate": 6.72239502332815e-05,
      "loss": 0.014797346666455269,
      "step": 6186
    },
    {
      "epoch": 1.172858225928734,
      "grad_norm": 0.15990562736988068,
      "learning_rate": 6.71461897356143e-05,
      "loss": 0.021103637292981148,
      "step": 6188
    },
    {
      "epoch": 1.1732373009855952,
      "grad_norm": 0.09193393588066101,
      "learning_rate": 6.706842923794712e-05,
      "loss": 0.018062151968479156,
      "step": 6190
    },
    {
      "epoch": 1.1736163760424565,
      "grad_norm": 0.0894952267408371,
      "learning_rate": 6.699066874027994e-05,
      "loss": 0.015553604811429977,
      "step": 6192
    },
    {
      "epoch": 1.1739954510993176,
      "grad_norm": 0.3102535307407379,
      "learning_rate": 6.691290824261276e-05,
      "loss": 0.02227085642516613,
      "step": 6194
    },
    {
      "epoch": 1.1743745261561789,
      "grad_norm": 0.25297102332115173,
      "learning_rate": 6.683514774494557e-05,
      "loss": 0.013667339459061623,
      "step": 6196
    },
    {
      "epoch": 1.1747536012130402,
      "grad_norm": 0.13222219049930573,
      "learning_rate": 6.675738724727839e-05,
      "loss": 0.02370486781001091,
      "step": 6198
    },
    {
      "epoch": 1.1751326762699015,
      "grad_norm": 0.32107651233673096,
      "learning_rate": 6.66796267496112e-05,
      "loss": 0.017578233033418655,
      "step": 6200
    },
    {
      "epoch": 1.1755117513267628,
      "grad_norm": 0.2621693015098572,
      "learning_rate": 6.660186625194402e-05,
      "loss": 0.016837647184729576,
      "step": 6202
    },
    {
      "epoch": 1.175890826383624,
      "grad_norm": 0.15447331964969635,
      "learning_rate": 6.652410575427684e-05,
      "loss": 0.01589328609406948,
      "step": 6204
    },
    {
      "epoch": 1.1762699014404852,
      "grad_norm": 0.4025198817253113,
      "learning_rate": 6.644634525660965e-05,
      "loss": 0.02194148302078247,
      "step": 6206
    },
    {
      "epoch": 1.1766489764973465,
      "grad_norm": 0.09745101630687714,
      "learning_rate": 6.636858475894247e-05,
      "loss": 0.021654406562447548,
      "step": 6208
    },
    {
      "epoch": 1.1770280515542078,
      "grad_norm": 0.2940644323825836,
      "learning_rate": 6.629082426127527e-05,
      "loss": 0.018386004492640495,
      "step": 6210
    },
    {
      "epoch": 1.177407126611069,
      "grad_norm": 0.09198614954948425,
      "learning_rate": 6.621306376360808e-05,
      "loss": 0.016615258529782295,
      "step": 6212
    },
    {
      "epoch": 1.1777862016679301,
      "grad_norm": 0.38985899090766907,
      "learning_rate": 6.61353032659409e-05,
      "loss": 0.027549980208277702,
      "step": 6214
    },
    {
      "epoch": 1.1781652767247914,
      "grad_norm": 0.21416887640953064,
      "learning_rate": 6.605754276827372e-05,
      "loss": 0.02272080071270466,
      "step": 6216
    },
    {
      "epoch": 1.1785443517816527,
      "grad_norm": 0.25762128829956055,
      "learning_rate": 6.597978227060654e-05,
      "loss": 0.019619232043623924,
      "step": 6218
    },
    {
      "epoch": 1.178923426838514,
      "grad_norm": 0.2827235460281372,
      "learning_rate": 6.590202177293935e-05,
      "loss": 0.018924199044704437,
      "step": 6220
    },
    {
      "epoch": 1.1793025018953753,
      "grad_norm": 0.0874418169260025,
      "learning_rate": 6.582426127527217e-05,
      "loss": 0.015506381168961525,
      "step": 6222
    },
    {
      "epoch": 1.1796815769522366,
      "grad_norm": 0.11131356656551361,
      "learning_rate": 6.574650077760498e-05,
      "loss": 0.014559132046997547,
      "step": 6224
    },
    {
      "epoch": 1.1800606520090977,
      "grad_norm": 0.33788564801216125,
      "learning_rate": 6.56687402799378e-05,
      "loss": 0.017166875302791595,
      "step": 6226
    },
    {
      "epoch": 1.180439727065959,
      "grad_norm": 0.317414790391922,
      "learning_rate": 6.559097978227062e-05,
      "loss": 0.023378800600767136,
      "step": 6228
    },
    {
      "epoch": 1.1808188021228203,
      "grad_norm": 0.10574235022068024,
      "learning_rate": 6.551321928460342e-05,
      "loss": 0.015482356771826744,
      "step": 6230
    },
    {
      "epoch": 1.1811978771796816,
      "grad_norm": 0.2890948951244354,
      "learning_rate": 6.543545878693624e-05,
      "loss": 0.02399097941815853,
      "step": 6232
    },
    {
      "epoch": 1.181576952236543,
      "grad_norm": 0.36053696274757385,
      "learning_rate": 6.535769828926905e-05,
      "loss": 0.022724226117134094,
      "step": 6234
    },
    {
      "epoch": 1.181956027293404,
      "grad_norm": 0.27962803840637207,
      "learning_rate": 6.527993779160187e-05,
      "loss": 0.01816137135028839,
      "step": 6236
    },
    {
      "epoch": 1.1823351023502653,
      "grad_norm": 0.10292515158653259,
      "learning_rate": 6.520217729393469e-05,
      "loss": 0.013841737061738968,
      "step": 6238
    },
    {
      "epoch": 1.1827141774071266,
      "grad_norm": 0.10013864934444427,
      "learning_rate": 6.51244167962675e-05,
      "loss": 0.011838647536933422,
      "step": 6240
    },
    {
      "epoch": 1.1830932524639879,
      "grad_norm": 0.08559630066156387,
      "learning_rate": 6.504665629860032e-05,
      "loss": 0.01368190348148346,
      "step": 6242
    },
    {
      "epoch": 1.1834723275208492,
      "grad_norm": 0.10152211040258408,
      "learning_rate": 6.496889580093313e-05,
      "loss": 0.017754876986145973,
      "step": 6244
    },
    {
      "epoch": 1.1838514025777105,
      "grad_norm": 0.10052108019590378,
      "learning_rate": 6.489113530326595e-05,
      "loss": 0.018012039363384247,
      "step": 6246
    },
    {
      "epoch": 1.1842304776345716,
      "grad_norm": 0.11313782632350922,
      "learning_rate": 6.481337480559875e-05,
      "loss": 0.014665824361145496,
      "step": 6248
    },
    {
      "epoch": 1.1846095526914329,
      "grad_norm": 0.33427271246910095,
      "learning_rate": 6.473561430793157e-05,
      "loss": 0.022343818098306656,
      "step": 6250
    },
    {
      "epoch": 1.1849886277482942,
      "grad_norm": 0.3421596586704254,
      "learning_rate": 6.46578538102644e-05,
      "loss": 0.019889164716005325,
      "step": 6252
    },
    {
      "epoch": 1.1853677028051555,
      "grad_norm": 0.12839621305465698,
      "learning_rate": 6.45800933125972e-05,
      "loss": 0.013550760224461555,
      "step": 6254
    },
    {
      "epoch": 1.1857467778620168,
      "grad_norm": 0.20503336191177368,
      "learning_rate": 6.450233281493002e-05,
      "loss": 0.017368679866194725,
      "step": 6256
    },
    {
      "epoch": 1.1861258529188778,
      "grad_norm": 0.27107954025268555,
      "learning_rate": 6.442457231726283e-05,
      "loss": 0.017702987417578697,
      "step": 6258
    },
    {
      "epoch": 1.1865049279757391,
      "grad_norm": 0.4800977408885956,
      "learning_rate": 6.434681181959565e-05,
      "loss": 0.027037009596824646,
      "step": 6260
    },
    {
      "epoch": 1.1868840030326004,
      "grad_norm": 0.16141583025455475,
      "learning_rate": 6.426905132192847e-05,
      "loss": 0.013395111076533794,
      "step": 6262
    },
    {
      "epoch": 1.1872630780894617,
      "grad_norm": 0.37441790103912354,
      "learning_rate": 6.419129082426128e-05,
      "loss": 0.021007031202316284,
      "step": 6264
    },
    {
      "epoch": 1.187642153146323,
      "grad_norm": 0.3202742636203766,
      "learning_rate": 6.41135303265941e-05,
      "loss": 0.017335787415504456,
      "step": 6266
    },
    {
      "epoch": 1.1880212282031843,
      "grad_norm": 0.08456775546073914,
      "learning_rate": 6.40357698289269e-05,
      "loss": 0.01644972898066044,
      "step": 6268
    },
    {
      "epoch": 1.1884003032600454,
      "grad_norm": 0.4996781349182129,
      "learning_rate": 6.395800933125972e-05,
      "loss": 0.02171907387673855,
      "step": 6270
    },
    {
      "epoch": 1.1887793783169067,
      "grad_norm": 0.1247066780924797,
      "learning_rate": 6.388024883359254e-05,
      "loss": 0.012071989476680756,
      "step": 6272
    },
    {
      "epoch": 1.189158453373768,
      "grad_norm": 0.09341282397508621,
      "learning_rate": 6.380248833592535e-05,
      "loss": 0.015423881821334362,
      "step": 6274
    },
    {
      "epoch": 1.1895375284306293,
      "grad_norm": 0.13203339278697968,
      "learning_rate": 6.372472783825817e-05,
      "loss": 0.012934763915836811,
      "step": 6276
    },
    {
      "epoch": 1.1899166034874904,
      "grad_norm": 0.2382124662399292,
      "learning_rate": 6.364696734059098e-05,
      "loss": 0.01318735908716917,
      "step": 6278
    },
    {
      "epoch": 1.1902956785443517,
      "grad_norm": 0.24593418836593628,
      "learning_rate": 6.35692068429238e-05,
      "loss": 0.017223291099071503,
      "step": 6280
    },
    {
      "epoch": 1.190674753601213,
      "grad_norm": 0.10933253169059753,
      "learning_rate": 6.34914463452566e-05,
      "loss": 0.012047995813190937,
      "step": 6282
    },
    {
      "epoch": 1.1910538286580743,
      "grad_norm": 0.12932097911834717,
      "learning_rate": 6.341368584758944e-05,
      "loss": 0.018925370648503304,
      "step": 6284
    },
    {
      "epoch": 1.1914329037149356,
      "grad_norm": 0.20290254056453705,
      "learning_rate": 6.333592534992225e-05,
      "loss": 0.02220279537141323,
      "step": 6286
    },
    {
      "epoch": 1.191811978771797,
      "grad_norm": 0.19692783057689667,
      "learning_rate": 6.325816485225505e-05,
      "loss": 0.014992056414484978,
      "step": 6288
    },
    {
      "epoch": 1.1921910538286582,
      "grad_norm": 0.10829626768827438,
      "learning_rate": 6.318040435458787e-05,
      "loss": 0.013848591595888138,
      "step": 6290
    },
    {
      "epoch": 1.1925701288855193,
      "grad_norm": 0.13487961888313293,
      "learning_rate": 6.310264385692068e-05,
      "loss": 0.024419836699962616,
      "step": 6292
    },
    {
      "epoch": 1.1929492039423806,
      "grad_norm": 0.19823376834392548,
      "learning_rate": 6.30248833592535e-05,
      "loss": 0.014793352223932743,
      "step": 6294
    },
    {
      "epoch": 1.1933282789992419,
      "grad_norm": 0.3482336401939392,
      "learning_rate": 6.294712286158632e-05,
      "loss": 0.02469705417752266,
      "step": 6296
    },
    {
      "epoch": 1.1937073540561032,
      "grad_norm": 0.07492935657501221,
      "learning_rate": 6.286936236391914e-05,
      "loss": 0.012258573435246944,
      "step": 6298
    },
    {
      "epoch": 1.1940864291129643,
      "grad_norm": 0.18498635292053223,
      "learning_rate": 6.279160186625195e-05,
      "loss": 0.01414254680275917,
      "step": 6300
    },
    {
      "epoch": 1.1944655041698256,
      "grad_norm": 0.08765418082475662,
      "learning_rate": 6.271384136858476e-05,
      "loss": 0.020850781351327896,
      "step": 6302
    },
    {
      "epoch": 1.1948445792266869,
      "grad_norm": 0.2457093745470047,
      "learning_rate": 6.263608087091758e-05,
      "loss": 0.020235678181052208,
      "step": 6304
    },
    {
      "epoch": 1.1952236542835482,
      "grad_norm": 0.24851682782173157,
      "learning_rate": 6.25583203732504e-05,
      "loss": 0.015360687859356403,
      "step": 6306
    },
    {
      "epoch": 1.1956027293404095,
      "grad_norm": 0.3284883201122284,
      "learning_rate": 6.248055987558322e-05,
      "loss": 0.022546816617250443,
      "step": 6308
    },
    {
      "epoch": 1.1959818043972708,
      "grad_norm": 0.2615007162094116,
      "learning_rate": 6.240279937791602e-05,
      "loss": 0.019091807305812836,
      "step": 6310
    },
    {
      "epoch": 1.1963608794541318,
      "grad_norm": 0.10462827235460281,
      "learning_rate": 6.232503888024883e-05,
      "loss": 0.015337320975959301,
      "step": 6312
    },
    {
      "epoch": 1.1967399545109931,
      "grad_norm": 0.19275625050067902,
      "learning_rate": 6.224727838258165e-05,
      "loss": 0.013794252648949623,
      "step": 6314
    },
    {
      "epoch": 1.1971190295678544,
      "grad_norm": 0.10002799332141876,
      "learning_rate": 6.216951788491446e-05,
      "loss": 0.01496365386992693,
      "step": 6316
    },
    {
      "epoch": 1.1974981046247157,
      "grad_norm": 0.24961066246032715,
      "learning_rate": 6.209175738724728e-05,
      "loss": 0.029249556362628937,
      "step": 6318
    },
    {
      "epoch": 1.197877179681577,
      "grad_norm": 0.08128700405359268,
      "learning_rate": 6.20139968895801e-05,
      "loss": 0.01563321426510811,
      "step": 6320
    },
    {
      "epoch": 1.198256254738438,
      "grad_norm": 0.40598535537719727,
      "learning_rate": 6.193623639191292e-05,
      "loss": 0.026375314220786095,
      "step": 6322
    },
    {
      "epoch": 1.1986353297952994,
      "grad_norm": 0.1306745707988739,
      "learning_rate": 6.185847589424573e-05,
      "loss": 0.01411717850714922,
      "step": 6324
    },
    {
      "epoch": 1.1990144048521607,
      "grad_norm": 0.09453743696212769,
      "learning_rate": 6.178071539657853e-05,
      "loss": 0.01518023107200861,
      "step": 6326
    },
    {
      "epoch": 1.199393479909022,
      "grad_norm": 0.23321940004825592,
      "learning_rate": 6.170295489891135e-05,
      "loss": 0.014367001131176949,
      "step": 6328
    },
    {
      "epoch": 1.1997725549658833,
      "grad_norm": 0.10031359642744064,
      "learning_rate": 6.162519440124417e-05,
      "loss": 0.012865809723734856,
      "step": 6330
    },
    {
      "epoch": 1.2001516300227446,
      "grad_norm": 0.09343080967664719,
      "learning_rate": 6.1547433903577e-05,
      "loss": 0.019952382892370224,
      "step": 6332
    },
    {
      "epoch": 1.2005307050796057,
      "grad_norm": 0.10685739666223526,
      "learning_rate": 6.14696734059098e-05,
      "loss": 0.016512319445610046,
      "step": 6334
    },
    {
      "epoch": 1.200909780136467,
      "grad_norm": 0.24465948343276978,
      "learning_rate": 6.139191290824261e-05,
      "loss": 0.020065752789378166,
      "step": 6336
    },
    {
      "epoch": 1.2012888551933283,
      "grad_norm": 0.15719056129455566,
      "learning_rate": 6.131415241057543e-05,
      "loss": 0.012033293955028057,
      "step": 6338
    },
    {
      "epoch": 1.2016679302501896,
      "grad_norm": 0.09356866776943207,
      "learning_rate": 6.123639191290825e-05,
      "loss": 0.013715642504394054,
      "step": 6340
    },
    {
      "epoch": 1.2020470053070509,
      "grad_norm": 0.09491132944822311,
      "learning_rate": 6.115863141524107e-05,
      "loss": 0.012714879587292671,
      "step": 6342
    },
    {
      "epoch": 1.202426080363912,
      "grad_norm": 0.4373061954975128,
      "learning_rate": 6.108087091757388e-05,
      "loss": 0.02359389141201973,
      "step": 6344
    },
    {
      "epoch": 1.2028051554207733,
      "grad_norm": 0.24160747230052948,
      "learning_rate": 6.1003110419906696e-05,
      "loss": 0.017881298437714577,
      "step": 6346
    },
    {
      "epoch": 1.2031842304776346,
      "grad_norm": 0.3179434835910797,
      "learning_rate": 6.09253499222395e-05,
      "loss": 0.020769044756889343,
      "step": 6348
    },
    {
      "epoch": 1.2035633055344959,
      "grad_norm": 0.324907511472702,
      "learning_rate": 6.084758942457232e-05,
      "loss": 0.02618054486811161,
      "step": 6350
    },
    {
      "epoch": 1.2039423805913572,
      "grad_norm": 0.2706593871116638,
      "learning_rate": 6.076982892690514e-05,
      "loss": 0.01771911419928074,
      "step": 6352
    },
    {
      "epoch": 1.2043214556482185,
      "grad_norm": 0.22512932121753693,
      "learning_rate": 6.0692068429237944e-05,
      "loss": 0.015653816983103752,
      "step": 6354
    },
    {
      "epoch": 1.2047005307050795,
      "grad_norm": 0.1180492639541626,
      "learning_rate": 6.061430793157077e-05,
      "loss": 0.013422629795968533,
      "step": 6356
    },
    {
      "epoch": 1.2050796057619408,
      "grad_norm": 0.27374210953712463,
      "learning_rate": 6.053654743390358e-05,
      "loss": 0.022156978026032448,
      "step": 6358
    },
    {
      "epoch": 1.2054586808188021,
      "grad_norm": 0.08576232939958572,
      "learning_rate": 6.04587869362364e-05,
      "loss": 0.01446045096963644,
      "step": 6360
    },
    {
      "epoch": 1.2058377558756634,
      "grad_norm": 0.3667774498462677,
      "learning_rate": 6.038102643856921e-05,
      "loss": 0.024921854957938194,
      "step": 6362
    },
    {
      "epoch": 1.2062168309325245,
      "grad_norm": 0.09681126475334167,
      "learning_rate": 6.030326594090202e-05,
      "loss": 0.02211802639067173,
      "step": 6364
    },
    {
      "epoch": 1.2065959059893858,
      "grad_norm": 0.09238740056753159,
      "learning_rate": 6.022550544323484e-05,
      "loss": 0.011126796714961529,
      "step": 6366
    },
    {
      "epoch": 1.2069749810462471,
      "grad_norm": 0.09851730614900589,
      "learning_rate": 6.0147744945567654e-05,
      "loss": 0.018866345286369324,
      "step": 6368
    },
    {
      "epoch": 1.2073540561031084,
      "grad_norm": 0.20304760336875916,
      "learning_rate": 6.0069984447900474e-05,
      "loss": 0.019454065710306168,
      "step": 6370
    },
    {
      "epoch": 1.2077331311599697,
      "grad_norm": 0.09282153099775314,
      "learning_rate": 5.999222395023328e-05,
      "loss": 0.01072689052671194,
      "step": 6372
    },
    {
      "epoch": 1.208112206216831,
      "grad_norm": 0.275784432888031,
      "learning_rate": 5.9914463452566095e-05,
      "loss": 0.027874311432242393,
      "step": 6374
    },
    {
      "epoch": 1.208491281273692,
      "grad_norm": 0.3889502286911011,
      "learning_rate": 5.9836702954898915e-05,
      "loss": 0.021810902282595634,
      "step": 6376
    },
    {
      "epoch": 1.2088703563305534,
      "grad_norm": 0.1077721044421196,
      "learning_rate": 5.975894245723173e-05,
      "loss": 0.014013019390404224,
      "step": 6378
    },
    {
      "epoch": 1.2092494313874147,
      "grad_norm": 0.2149156928062439,
      "learning_rate": 5.968118195956455e-05,
      "loss": 0.014053000137209892,
      "step": 6380
    },
    {
      "epoch": 1.209628506444276,
      "grad_norm": 0.09361483156681061,
      "learning_rate": 5.9603421461897356e-05,
      "loss": 0.018578100949525833,
      "step": 6382
    },
    {
      "epoch": 1.2100075815011373,
      "grad_norm": 0.09079177677631378,
      "learning_rate": 5.9525660964230176e-05,
      "loss": 0.013101140037178993,
      "step": 6384
    },
    {
      "epoch": 1.2103866565579984,
      "grad_norm": 0.2940463721752167,
      "learning_rate": 5.944790046656299e-05,
      "loss": 0.027358219027519226,
      "step": 6386
    },
    {
      "epoch": 1.2107657316148597,
      "grad_norm": 0.10005920380353928,
      "learning_rate": 5.93701399688958e-05,
      "loss": 0.013695024885237217,
      "step": 6388
    },
    {
      "epoch": 1.211144806671721,
      "grad_norm": 0.3209916353225708,
      "learning_rate": 5.929237947122862e-05,
      "loss": 0.028121644631028175,
      "step": 6390
    },
    {
      "epoch": 1.2115238817285823,
      "grad_norm": 0.3602473735809326,
      "learning_rate": 5.921461897356143e-05,
      "loss": 0.01819569244980812,
      "step": 6392
    },
    {
      "epoch": 1.2119029567854436,
      "grad_norm": 0.10206498950719833,
      "learning_rate": 5.913685847589425e-05,
      "loss": 0.011470306664705276,
      "step": 6394
    },
    {
      "epoch": 1.2122820318423049,
      "grad_norm": 0.1099764034152031,
      "learning_rate": 5.9059097978227065e-05,
      "loss": 0.012633945792913437,
      "step": 6396
    },
    {
      "epoch": 1.212661106899166,
      "grad_norm": 0.10611998289823532,
      "learning_rate": 5.8981337480559886e-05,
      "loss": 0.014772419817745686,
      "step": 6398
    },
    {
      "epoch": 1.2130401819560273,
      "grad_norm": 0.08990549296140671,
      "learning_rate": 5.890357698289269e-05,
      "loss": 0.012792189605534077,
      "step": 6400
    },
    {
      "epoch": 1.2134192570128886,
      "grad_norm": 0.11494190245866776,
      "learning_rate": 5.8825816485225506e-05,
      "loss": 0.01730509102344513,
      "step": 6402
    },
    {
      "epoch": 1.2137983320697499,
      "grad_norm": 0.09110274165868759,
      "learning_rate": 5.8748055987558327e-05,
      "loss": 0.015411511063575745,
      "step": 6404
    },
    {
      "epoch": 1.2141774071266112,
      "grad_norm": 0.32092010974884033,
      "learning_rate": 5.8670295489891133e-05,
      "loss": 0.017549801617860794,
      "step": 6406
    },
    {
      "epoch": 1.2145564821834722,
      "grad_norm": 0.45817455649375916,
      "learning_rate": 5.859253499222396e-05,
      "loss": 0.035252463072538376,
      "step": 6408
    },
    {
      "epoch": 1.2149355572403335,
      "grad_norm": 0.24788841605186462,
      "learning_rate": 5.851477449455677e-05,
      "loss": 0.015718894079327583,
      "step": 6410
    },
    {
      "epoch": 1.2153146322971948,
      "grad_norm": 0.16123180091381073,
      "learning_rate": 5.843701399688958e-05,
      "loss": 0.020789436995983124,
      "step": 6412
    },
    {
      "epoch": 1.2156937073540561,
      "grad_norm": 0.296629399061203,
      "learning_rate": 5.83592534992224e-05,
      "loss": 0.017592623829841614,
      "step": 6414
    },
    {
      "epoch": 1.2160727824109174,
      "grad_norm": 0.09330003708600998,
      "learning_rate": 5.828149300155521e-05,
      "loss": 0.011924791149795055,
      "step": 6416
    },
    {
      "epoch": 1.2164518574677787,
      "grad_norm": 0.29089468717575073,
      "learning_rate": 5.820373250388803e-05,
      "loss": 0.015712810680270195,
      "step": 6418
    },
    {
      "epoch": 1.2168309325246398,
      "grad_norm": 0.2281501144170761,
      "learning_rate": 5.812597200622084e-05,
      "loss": 0.01563522405922413,
      "step": 6420
    },
    {
      "epoch": 1.217210007581501,
      "grad_norm": 0.39973726868629456,
      "learning_rate": 5.804821150855366e-05,
      "loss": 0.022105716168880463,
      "step": 6422
    },
    {
      "epoch": 1.2175890826383624,
      "grad_norm": 0.09277337789535522,
      "learning_rate": 5.797045101088647e-05,
      "loss": 0.018118394538760185,
      "step": 6424
    },
    {
      "epoch": 1.2179681576952237,
      "grad_norm": 0.173731729388237,
      "learning_rate": 5.7892690513219284e-05,
      "loss": 0.02131512202322483,
      "step": 6426
    },
    {
      "epoch": 1.218347232752085,
      "grad_norm": 0.247223362326622,
      "learning_rate": 5.7814930015552104e-05,
      "loss": 0.01940269209444523,
      "step": 6428
    },
    {
      "epoch": 1.218726307808946,
      "grad_norm": 0.14708009362220764,
      "learning_rate": 5.773716951788492e-05,
      "loss": 0.012067603878676891,
      "step": 6430
    },
    {
      "epoch": 1.2191053828658074,
      "grad_norm": 0.13432088494300842,
      "learning_rate": 5.765940902021774e-05,
      "loss": 0.011675133369863033,
      "step": 6432
    },
    {
      "epoch": 1.2194844579226687,
      "grad_norm": 0.4383014142513275,
      "learning_rate": 5.7581648522550545e-05,
      "loss": 0.027306340634822845,
      "step": 6434
    },
    {
      "epoch": 1.21986353297953,
      "grad_norm": 0.09641839563846588,
      "learning_rate": 5.750388802488336e-05,
      "loss": 0.013324783183634281,
      "step": 6436
    },
    {
      "epoch": 1.2202426080363913,
      "grad_norm": 0.2715272009372711,
      "learning_rate": 5.742612752721618e-05,
      "loss": 0.0249321386218071,
      "step": 6438
    },
    {
      "epoch": 1.2206216830932526,
      "grad_norm": 0.28936654329299927,
      "learning_rate": 5.7348367029548986e-05,
      "loss": 0.029160384088754654,
      "step": 6440
    },
    {
      "epoch": 1.2210007581501137,
      "grad_norm": 0.13795986771583557,
      "learning_rate": 5.7270606531881806e-05,
      "loss": 0.01763402484357357,
      "step": 6442
    },
    {
      "epoch": 1.221379833206975,
      "grad_norm": 0.12503860890865326,
      "learning_rate": 5.719284603421462e-05,
      "loss": 0.016572151333093643,
      "step": 6444
    },
    {
      "epoch": 1.2217589082638363,
      "grad_norm": 0.07718665897846222,
      "learning_rate": 5.711508553654744e-05,
      "loss": 0.01631186716258526,
      "step": 6446
    },
    {
      "epoch": 1.2221379833206976,
      "grad_norm": 0.3193527162075043,
      "learning_rate": 5.7037325038880254e-05,
      "loss": 0.016860434785485268,
      "step": 6448
    },
    {
      "epoch": 1.2225170583775586,
      "grad_norm": 0.21418415009975433,
      "learning_rate": 5.695956454121306e-05,
      "loss": 0.01524271909147501,
      "step": 6450
    },
    {
      "epoch": 1.22289613343442,
      "grad_norm": 0.08740539103746414,
      "learning_rate": 5.688180404354588e-05,
      "loss": 0.011215193197131157,
      "step": 6452
    },
    {
      "epoch": 1.2232752084912812,
      "grad_norm": 0.3365560472011566,
      "learning_rate": 5.6804043545878695e-05,
      "loss": 0.021854843944311142,
      "step": 6454
    },
    {
      "epoch": 1.2236542835481425,
      "grad_norm": 0.12064898014068604,
      "learning_rate": 5.6726283048211516e-05,
      "loss": 0.019420860335230827,
      "step": 6456
    },
    {
      "epoch": 1.2240333586050038,
      "grad_norm": 0.2717916965484619,
      "learning_rate": 5.664852255054432e-05,
      "loss": 0.01950657181441784,
      "step": 6458
    },
    {
      "epoch": 1.2244124336618651,
      "grad_norm": 0.321755975484848,
      "learning_rate": 5.657076205287715e-05,
      "loss": 0.023398905992507935,
      "step": 6460
    },
    {
      "epoch": 1.2247915087187262,
      "grad_norm": 0.100419782102108,
      "learning_rate": 5.649300155520996e-05,
      "loss": 0.024872347712516785,
      "step": 6462
    },
    {
      "epoch": 1.2251705837755875,
      "grad_norm": 0.0806308165192604,
      "learning_rate": 5.641524105754277e-05,
      "loss": 0.014432406052947044,
      "step": 6464
    },
    {
      "epoch": 1.2255496588324488,
      "grad_norm": 0.3552096486091614,
      "learning_rate": 5.633748055987559e-05,
      "loss": 0.025575721636414528,
      "step": 6466
    },
    {
      "epoch": 1.2259287338893101,
      "grad_norm": 0.29471221566200256,
      "learning_rate": 5.62597200622084e-05,
      "loss": 0.02662278525531292,
      "step": 6468
    },
    {
      "epoch": 1.2263078089461714,
      "grad_norm": 0.30473628640174866,
      "learning_rate": 5.618195956454122e-05,
      "loss": 0.020489228889346123,
      "step": 6470
    },
    {
      "epoch": 1.2266868840030325,
      "grad_norm": 0.32313084602355957,
      "learning_rate": 5.610419906687403e-05,
      "loss": 0.020656652748584747,
      "step": 6472
    },
    {
      "epoch": 1.2270659590598938,
      "grad_norm": 0.3460982143878937,
      "learning_rate": 5.602643856920684e-05,
      "loss": 0.018190693110227585,
      "step": 6474
    },
    {
      "epoch": 1.227445034116755,
      "grad_norm": 0.34067460894584656,
      "learning_rate": 5.594867807153966e-05,
      "loss": 0.01648704893887043,
      "step": 6476
    },
    {
      "epoch": 1.2278241091736164,
      "grad_norm": 0.267316073179245,
      "learning_rate": 5.587091757387247e-05,
      "loss": 0.0229484923183918,
      "step": 6478
    },
    {
      "epoch": 1.2282031842304777,
      "grad_norm": 0.3420708179473877,
      "learning_rate": 5.579315707620529e-05,
      "loss": 0.019424576312303543,
      "step": 6480
    },
    {
      "epoch": 1.228582259287339,
      "grad_norm": 0.38786572217941284,
      "learning_rate": 5.571539657853811e-05,
      "loss": 0.02583460882306099,
      "step": 6482
    },
    {
      "epoch": 1.2289613343442,
      "grad_norm": 0.08479564636945724,
      "learning_rate": 5.563763608087093e-05,
      "loss": 0.012250595726072788,
      "step": 6484
    },
    {
      "epoch": 1.2293404094010614,
      "grad_norm": 0.2685691714286804,
      "learning_rate": 5.5559875583203734e-05,
      "loss": 0.030878402292728424,
      "step": 6486
    },
    {
      "epoch": 1.2297194844579227,
      "grad_norm": 0.24452455341815948,
      "learning_rate": 5.548211508553655e-05,
      "loss": 0.019374549388885498,
      "step": 6488
    },
    {
      "epoch": 1.230098559514784,
      "grad_norm": 0.2897247076034546,
      "learning_rate": 5.540435458786937e-05,
      "loss": 0.01659223437309265,
      "step": 6490
    },
    {
      "epoch": 1.2304776345716453,
      "grad_norm": 0.10419486463069916,
      "learning_rate": 5.5326594090202175e-05,
      "loss": 0.012231425382196903,
      "step": 6492
    },
    {
      "epoch": 1.2308567096285064,
      "grad_norm": 0.07435934245586395,
      "learning_rate": 5.5248833592534996e-05,
      "loss": 0.016316495835781097,
      "step": 6494
    },
    {
      "epoch": 1.2312357846853677,
      "grad_norm": 0.39297083020210266,
      "learning_rate": 5.517107309486781e-05,
      "loss": 0.027547281235456467,
      "step": 6496
    },
    {
      "epoch": 1.231614859742229,
      "grad_norm": 0.10686418414115906,
      "learning_rate": 5.5093312597200616e-05,
      "loss": 0.014828301966190338,
      "step": 6498
    },
    {
      "epoch": 1.2319939347990903,
      "grad_norm": 0.09654749184846878,
      "learning_rate": 5.501555209953344e-05,
      "loss": 0.011750442907214165,
      "step": 6500
    },
    {
      "epoch": 1.2323730098559516,
      "grad_norm": 0.38048189878463745,
      "learning_rate": 5.493779160186625e-05,
      "loss": 0.024039376527071,
      "step": 6502
    },
    {
      "epoch": 1.2327520849128129,
      "grad_norm": 0.10946713387966156,
      "learning_rate": 5.486003110419907e-05,
      "loss": 0.01461044605821371,
      "step": 6504
    },
    {
      "epoch": 1.233131159969674,
      "grad_norm": 0.19076725840568542,
      "learning_rate": 5.4782270606531884e-05,
      "loss": 0.013751516118645668,
      "step": 6506
    },
    {
      "epoch": 1.2335102350265352,
      "grad_norm": 0.3625965416431427,
      "learning_rate": 5.4704510108864705e-05,
      "loss": 0.029328277334570885,
      "step": 6508
    },
    {
      "epoch": 1.2338893100833965,
      "grad_norm": 0.440005898475647,
      "learning_rate": 5.462674961119751e-05,
      "loss": 0.031225059181451797,
      "step": 6510
    },
    {
      "epoch": 1.2342683851402578,
      "grad_norm": 0.28962263464927673,
      "learning_rate": 5.4548989113530325e-05,
      "loss": 0.016894403845071793,
      "step": 6512
    },
    {
      "epoch": 1.234647460197119,
      "grad_norm": 0.29713261127471924,
      "learning_rate": 5.4471228615863146e-05,
      "loss": 0.02009391598403454,
      "step": 6514
    },
    {
      "epoch": 1.2350265352539802,
      "grad_norm": 0.10183078795671463,
      "learning_rate": 5.439346811819596e-05,
      "loss": 0.014307976700365543,
      "step": 6516
    },
    {
      "epoch": 1.2354056103108415,
      "grad_norm": 0.08961620926856995,
      "learning_rate": 5.431570762052878e-05,
      "loss": 0.01172261405736208,
      "step": 6518
    },
    {
      "epoch": 1.2357846853677028,
      "grad_norm": 0.13395291566848755,
      "learning_rate": 5.423794712286159e-05,
      "loss": 0.017483608797192574,
      "step": 6520
    },
    {
      "epoch": 1.236163760424564,
      "grad_norm": 0.09740249812602997,
      "learning_rate": 5.416018662519441e-05,
      "loss": 0.012895406223833561,
      "step": 6522
    },
    {
      "epoch": 1.2365428354814254,
      "grad_norm": 0.24238796532154083,
      "learning_rate": 5.408242612752722e-05,
      "loss": 0.01684867963194847,
      "step": 6524
    },
    {
      "epoch": 1.2369219105382865,
      "grad_norm": 0.17494115233421326,
      "learning_rate": 5.400466562986003e-05,
      "loss": 0.018205558881163597,
      "step": 6526
    },
    {
      "epoch": 1.2373009855951478,
      "grad_norm": 0.3416663110256195,
      "learning_rate": 5.392690513219285e-05,
      "loss": 0.018688833341002464,
      "step": 6528
    },
    {
      "epoch": 1.237680060652009,
      "grad_norm": 0.2225375771522522,
      "learning_rate": 5.384914463452566e-05,
      "loss": 0.014129049144685268,
      "step": 6530
    },
    {
      "epoch": 1.2380591357088704,
      "grad_norm": 0.08367101848125458,
      "learning_rate": 5.377138413685848e-05,
      "loss": 0.013678235933184624,
      "step": 6532
    },
    {
      "epoch": 1.2384382107657317,
      "grad_norm": 0.13579320907592773,
      "learning_rate": 5.3693623639191296e-05,
      "loss": 0.013210979290306568,
      "step": 6534
    },
    {
      "epoch": 1.2388172858225928,
      "grad_norm": 0.09422765672206879,
      "learning_rate": 5.36158631415241e-05,
      "loss": 0.01392685528844595,
      "step": 6536
    },
    {
      "epoch": 1.239196360879454,
      "grad_norm": 0.12507417798042297,
      "learning_rate": 5.353810264385692e-05,
      "loss": 0.013763990253210068,
      "step": 6538
    },
    {
      "epoch": 1.2395754359363154,
      "grad_norm": 0.289298951625824,
      "learning_rate": 5.346034214618974e-05,
      "loss": 0.020742595195770264,
      "step": 6540
    },
    {
      "epoch": 1.2399545109931767,
      "grad_norm": 0.3300250172615051,
      "learning_rate": 5.338258164852256e-05,
      "loss": 0.019002187997102737,
      "step": 6542
    },
    {
      "epoch": 1.240333586050038,
      "grad_norm": 0.24260738492012024,
      "learning_rate": 5.3304821150855364e-05,
      "loss": 0.021560028195381165,
      "step": 6544
    },
    {
      "epoch": 1.2407126611068993,
      "grad_norm": 0.18846189975738525,
      "learning_rate": 5.3227060653188185e-05,
      "loss": 0.01685902290046215,
      "step": 6546
    },
    {
      "epoch": 1.2410917361637603,
      "grad_norm": 0.08016867190599442,
      "learning_rate": 5.3149300155521e-05,
      "loss": 0.01269586756825447,
      "step": 6548
    },
    {
      "epoch": 1.2414708112206216,
      "grad_norm": 0.17521196603775024,
      "learning_rate": 5.3071539657853805e-05,
      "loss": 0.0205913744866848,
      "step": 6550
    },
    {
      "epoch": 1.241849886277483,
      "grad_norm": 0.1193728968501091,
      "learning_rate": 5.299377916018663e-05,
      "loss": 0.01610746420919895,
      "step": 6552
    },
    {
      "epoch": 1.2422289613343442,
      "grad_norm": 0.2872998118400574,
      "learning_rate": 5.291601866251944e-05,
      "loss": 0.018124330788850784,
      "step": 6554
    },
    {
      "epoch": 1.2426080363912055,
      "grad_norm": 0.4651348888874054,
      "learning_rate": 5.283825816485226e-05,
      "loss": 0.03555655851960182,
      "step": 6556
    },
    {
      "epoch": 1.2429871114480666,
      "grad_norm": 0.09471651911735535,
      "learning_rate": 5.276049766718507e-05,
      "loss": 0.017541514709591866,
      "step": 6558
    },
    {
      "epoch": 1.243366186504928,
      "grad_norm": 0.33059611916542053,
      "learning_rate": 5.2682737169517894e-05,
      "loss": 0.020963238552212715,
      "step": 6560
    },
    {
      "epoch": 1.2437452615617892,
      "grad_norm": 0.1076115220785141,
      "learning_rate": 5.26049766718507e-05,
      "loss": 0.02418288215994835,
      "step": 6562
    },
    {
      "epoch": 1.2441243366186505,
      "grad_norm": 0.38628795742988586,
      "learning_rate": 5.2527216174183514e-05,
      "loss": 0.022477880120277405,
      "step": 6564
    },
    {
      "epoch": 1.2445034116755118,
      "grad_norm": 0.11156751960515976,
      "learning_rate": 5.2449455676516335e-05,
      "loss": 0.020462535321712494,
      "step": 6566
    },
    {
      "epoch": 1.2448824867323731,
      "grad_norm": 0.1272927075624466,
      "learning_rate": 5.237169517884914e-05,
      "loss": 0.013446606695652008,
      "step": 6568
    },
    {
      "epoch": 1.2452615617892342,
      "grad_norm": 0.2627268135547638,
      "learning_rate": 5.229393468118197e-05,
      "loss": 0.01399513054639101,
      "step": 6570
    },
    {
      "epoch": 1.2456406368460955,
      "grad_norm": 0.1969190388917923,
      "learning_rate": 5.2216174183514776e-05,
      "loss": 0.02188953198492527,
      "step": 6572
    },
    {
      "epoch": 1.2460197119029568,
      "grad_norm": 0.13006211817264557,
      "learning_rate": 5.213841368584759e-05,
      "loss": 0.012145333923399448,
      "step": 6574
    },
    {
      "epoch": 1.246398786959818,
      "grad_norm": 0.22001232206821442,
      "learning_rate": 5.206065318818041e-05,
      "loss": 0.01480875350534916,
      "step": 6576
    },
    {
      "epoch": 1.2467778620166794,
      "grad_norm": 0.2832320034503937,
      "learning_rate": 5.198289269051322e-05,
      "loss": 0.01605958119034767,
      "step": 6578
    },
    {
      "epoch": 1.2471569370735405,
      "grad_norm": 0.1864720582962036,
      "learning_rate": 5.190513219284604e-05,
      "loss": 0.019302930682897568,
      "step": 6580
    },
    {
      "epoch": 1.2475360121304018,
      "grad_norm": 0.2723497748374939,
      "learning_rate": 5.182737169517885e-05,
      "loss": 0.01329825259745121,
      "step": 6582
    },
    {
      "epoch": 1.247915087187263,
      "grad_norm": 0.18247267603874207,
      "learning_rate": 5.174961119751167e-05,
      "loss": 0.01591726951301098,
      "step": 6584
    },
    {
      "epoch": 1.2482941622441244,
      "grad_norm": 0.312033087015152,
      "learning_rate": 5.1671850699844485e-05,
      "loss": 0.024608323350548744,
      "step": 6586
    },
    {
      "epoch": 1.2486732373009857,
      "grad_norm": 0.09229253977537155,
      "learning_rate": 5.159409020217729e-05,
      "loss": 0.011467719450592995,
      "step": 6588
    },
    {
      "epoch": 1.249052312357847,
      "grad_norm": 0.3218768835067749,
      "learning_rate": 5.151632970451011e-05,
      "loss": 0.019417330622673035,
      "step": 6590
    },
    {
      "epoch": 1.249431387414708,
      "grad_norm": 0.3051908314228058,
      "learning_rate": 5.1438569206842926e-05,
      "loss": 0.022750254720449448,
      "step": 6592
    },
    {
      "epoch": 1.2498104624715694,
      "grad_norm": 0.09359011054039001,
      "learning_rate": 5.1360808709175746e-05,
      "loss": 0.011321380734443665,
      "step": 6594
    },
    {
      "epoch": 1.2501895375284306,
      "grad_norm": 0.26745906472206116,
      "learning_rate": 5.128304821150855e-05,
      "loss": 0.02015247382223606,
      "step": 6596
    },
    {
      "epoch": 1.250568612585292,
      "grad_norm": 0.1047593206167221,
      "learning_rate": 5.120528771384137e-05,
      "loss": 0.01467406377196312,
      "step": 6598
    },
    {
      "epoch": 1.250947687642153,
      "grad_norm": 0.35953718423843384,
      "learning_rate": 5.112752721617419e-05,
      "loss": 0.023523611947894096,
      "step": 6600
    },
    {
      "epoch": 1.2513267626990143,
      "grad_norm": 0.09729735553264618,
      "learning_rate": 5.1049766718506994e-05,
      "loss": 0.012449660338461399,
      "step": 6602
    },
    {
      "epoch": 1.2517058377558756,
      "grad_norm": 0.2535962462425232,
      "learning_rate": 5.097200622083982e-05,
      "loss": 0.01382282841950655,
      "step": 6604
    },
    {
      "epoch": 1.252084912812737,
      "grad_norm": 0.11706600338220596,
      "learning_rate": 5.089424572317263e-05,
      "loss": 0.01383461058139801,
      "step": 6606
    },
    {
      "epoch": 1.2524639878695982,
      "grad_norm": 0.23716135323047638,
      "learning_rate": 5.081648522550545e-05,
      "loss": 0.015631891787052155,
      "step": 6608
    },
    {
      "epoch": 1.2528430629264595,
      "grad_norm": 0.21195638179779053,
      "learning_rate": 5.073872472783826e-05,
      "loss": 0.018170760944485664,
      "step": 6610
    },
    {
      "epoch": 1.2532221379833208,
      "grad_norm": 0.09606947004795074,
      "learning_rate": 5.066096423017107e-05,
      "loss": 0.021436259150505066,
      "step": 6612
    },
    {
      "epoch": 1.253601213040182,
      "grad_norm": 0.26288899779319763,
      "learning_rate": 5.058320373250389e-05,
      "loss": 0.01664581522345543,
      "step": 6614
    },
    {
      "epoch": 1.2539802880970432,
      "grad_norm": 0.2365722507238388,
      "learning_rate": 5.0505443234836703e-05,
      "loss": 0.0146090779453516,
      "step": 6616
    },
    {
      "epoch": 1.2543593631539045,
      "grad_norm": 0.2123231589794159,
      "learning_rate": 5.0427682737169524e-05,
      "loss": 0.014006976038217545,
      "step": 6618
    },
    {
      "epoch": 1.2547384382107658,
      "grad_norm": 0.10192007571458817,
      "learning_rate": 5.034992223950233e-05,
      "loss": 0.012327651493251324,
      "step": 6620
    },
    {
      "epoch": 1.2551175132676269,
      "grad_norm": 0.1307271122932434,
      "learning_rate": 5.027216174183516e-05,
      "loss": 0.015406850725412369,
      "step": 6622
    },
    {
      "epoch": 1.2554965883244882,
      "grad_norm": 0.2606370449066162,
      "learning_rate": 5.0194401244167965e-05,
      "loss": 0.014216038398444653,
      "step": 6624
    },
    {
      "epoch": 1.2558756633813495,
      "grad_norm": 0.25376439094543457,
      "learning_rate": 5.011664074650078e-05,
      "loss": 0.01531671080738306,
      "step": 6626
    },
    {
      "epoch": 1.2562547384382108,
      "grad_norm": 0.3731122612953186,
      "learning_rate": 5.00388802488336e-05,
      "loss": 0.016485391184687614,
      "step": 6628
    },
    {
      "epoch": 1.256633813495072,
      "grad_norm": 0.0916827842593193,
      "learning_rate": 4.9961119751166406e-05,
      "loss": 0.01403854414820671,
      "step": 6630
    },
    {
      "epoch": 1.2570128885519334,
      "grad_norm": 0.24796703457832336,
      "learning_rate": 4.9883359253499226e-05,
      "loss": 0.01485227420926094,
      "step": 6632
    },
    {
      "epoch": 1.2573919636087945,
      "grad_norm": 0.5141931176185608,
      "learning_rate": 4.980559875583204e-05,
      "loss": 0.039041001349687576,
      "step": 6634
    },
    {
      "epoch": 1.2577710386656558,
      "grad_norm": 0.3276703953742981,
      "learning_rate": 4.9727838258164854e-05,
      "loss": 0.022407447919249535,
      "step": 6636
    },
    {
      "epoch": 1.258150113722517,
      "grad_norm": 0.23539462685585022,
      "learning_rate": 4.9650077760497674e-05,
      "loss": 0.01732475310564041,
      "step": 6638
    },
    {
      "epoch": 1.2585291887793784,
      "grad_norm": 0.22159849107265472,
      "learning_rate": 4.957231726283049e-05,
      "loss": 0.014804009348154068,
      "step": 6640
    },
    {
      "epoch": 1.2589082638362394,
      "grad_norm": 0.09710431843996048,
      "learning_rate": 4.9494556765163295e-05,
      "loss": 0.019312605261802673,
      "step": 6642
    },
    {
      "epoch": 1.2592873388931007,
      "grad_norm": 0.15917278826236725,
      "learning_rate": 4.9416796267496115e-05,
      "loss": 0.012938236817717552,
      "step": 6644
    },
    {
      "epoch": 1.259666413949962,
      "grad_norm": 0.26350200176239014,
      "learning_rate": 4.933903576982893e-05,
      "loss": 0.02917845919728279,
      "step": 6646
    },
    {
      "epoch": 1.2600454890068233,
      "grad_norm": 0.11109036207199097,
      "learning_rate": 4.926127527216174e-05,
      "loss": 0.02264559268951416,
      "step": 6648
    },
    {
      "epoch": 1.2604245640636846,
      "grad_norm": 0.22474487125873566,
      "learning_rate": 4.918351477449456e-05,
      "loss": 0.01307457871735096,
      "step": 6650
    },
    {
      "epoch": 1.260803639120546,
      "grad_norm": 0.3275551497936249,
      "learning_rate": 4.9105754276827376e-05,
      "loss": 0.018205109983682632,
      "step": 6652
    },
    {
      "epoch": 1.2611827141774072,
      "grad_norm": 0.3408803641796112,
      "learning_rate": 4.902799377916018e-05,
      "loss": 0.01938813365995884,
      "step": 6654
    },
    {
      "epoch": 1.2615617892342683,
      "grad_norm": 0.3668753206729889,
      "learning_rate": 4.8950233281493004e-05,
      "loss": 0.03307902440428734,
      "step": 6656
    },
    {
      "epoch": 1.2619408642911296,
      "grad_norm": 0.18397437036037445,
      "learning_rate": 4.887247278382582e-05,
      "loss": 0.014377855695784092,
      "step": 6658
    },
    {
      "epoch": 1.262319939347991,
      "grad_norm": 0.10215871036052704,
      "learning_rate": 4.879471228615863e-05,
      "loss": 0.012707054615020752,
      "step": 6660
    },
    {
      "epoch": 1.2626990144048522,
      "grad_norm": 0.09178727865219116,
      "learning_rate": 4.871695178849145e-05,
      "loss": 0.013776799663901329,
      "step": 6662
    },
    {
      "epoch": 1.2630780894617133,
      "grad_norm": 0.3062099516391754,
      "learning_rate": 4.8639191290824265e-05,
      "loss": 0.018404070287942886,
      "step": 6664
    },
    {
      "epoch": 1.2634571645185746,
      "grad_norm": 0.08078951388597488,
      "learning_rate": 4.856143079315708e-05,
      "loss": 0.01113127637654543,
      "step": 6666
    },
    {
      "epoch": 1.263836239575436,
      "grad_norm": 0.3056524097919464,
      "learning_rate": 4.848367029548989e-05,
      "loss": 0.017948653548955917,
      "step": 6668
    },
    {
      "epoch": 1.2642153146322972,
      "grad_norm": 0.10581252723932266,
      "learning_rate": 4.8405909797822706e-05,
      "loss": 0.012365492060780525,
      "step": 6670
    },
    {
      "epoch": 1.2645943896891585,
      "grad_norm": 0.09511284530162811,
      "learning_rate": 4.832814930015552e-05,
      "loss": 0.012179193086922169,
      "step": 6672
    },
    {
      "epoch": 1.2649734647460198,
      "grad_norm": 0.3311889171600342,
      "learning_rate": 4.825038880248834e-05,
      "loss": 0.024393953382968903,
      "step": 6674
    },
    {
      "epoch": 1.265352539802881,
      "grad_norm": 0.23333054780960083,
      "learning_rate": 4.8172628304821154e-05,
      "loss": 0.019175935536623,
      "step": 6676
    },
    {
      "epoch": 1.2657316148597422,
      "grad_norm": 0.311240553855896,
      "learning_rate": 4.809486780715397e-05,
      "loss": 0.017175737768411636,
      "step": 6678
    },
    {
      "epoch": 1.2661106899166035,
      "grad_norm": 0.0827278271317482,
      "learning_rate": 4.801710730948678e-05,
      "loss": 0.015156029723584652,
      "step": 6680
    },
    {
      "epoch": 1.2664897649734648,
      "grad_norm": 0.18476302921772003,
      "learning_rate": 4.7939346811819595e-05,
      "loss": 0.0188934113830328,
      "step": 6682
    },
    {
      "epoch": 1.266868840030326,
      "grad_norm": 0.28275632858276367,
      "learning_rate": 4.7861586314152415e-05,
      "loss": 0.0191994346678257,
      "step": 6684
    },
    {
      "epoch": 1.2672479150871871,
      "grad_norm": 0.34957581758499146,
      "learning_rate": 4.778382581648523e-05,
      "loss": 0.02295362763106823,
      "step": 6686
    },
    {
      "epoch": 1.2676269901440484,
      "grad_norm": 0.15441444516181946,
      "learning_rate": 4.770606531881804e-05,
      "loss": 0.01271742582321167,
      "step": 6688
    },
    {
      "epoch": 1.2680060652009097,
      "grad_norm": 0.10465138405561447,
      "learning_rate": 4.762830482115086e-05,
      "loss": 0.014482580125331879,
      "step": 6690
    },
    {
      "epoch": 1.268385140257771,
      "grad_norm": 0.24350960552692413,
      "learning_rate": 4.755054432348367e-05,
      "loss": 0.016538016498088837,
      "step": 6692
    },
    {
      "epoch": 1.2687642153146323,
      "grad_norm": 0.3602878749370575,
      "learning_rate": 4.7472783825816484e-05,
      "loss": 0.024795861914753914,
      "step": 6694
    },
    {
      "epoch": 1.2691432903714936,
      "grad_norm": 0.28054341673851013,
      "learning_rate": 4.7395023328149304e-05,
      "loss": 0.015329709276556969,
      "step": 6696
    },
    {
      "epoch": 1.269522365428355,
      "grad_norm": 0.29563161730766296,
      "learning_rate": 4.731726283048212e-05,
      "loss": 0.022254301235079765,
      "step": 6698
    },
    {
      "epoch": 1.269901440485216,
      "grad_norm": 0.2159825712442398,
      "learning_rate": 4.723950233281493e-05,
      "loss": 0.013102126307785511,
      "step": 6700
    },
    {
      "epoch": 1.2702805155420773,
      "grad_norm": 0.08612479269504547,
      "learning_rate": 4.716174183514775e-05,
      "loss": 0.011899834498763084,
      "step": 6702
    },
    {
      "epoch": 1.2706595905989386,
      "grad_norm": 0.17244678735733032,
      "learning_rate": 4.708398133748056e-05,
      "loss": 0.015321638435125351,
      "step": 6704
    },
    {
      "epoch": 1.2710386656558,
      "grad_norm": 0.3476986885070801,
      "learning_rate": 4.700622083981337e-05,
      "loss": 0.017650511115789413,
      "step": 6706
    },
    {
      "epoch": 1.271417740712661,
      "grad_norm": 0.0778338611125946,
      "learning_rate": 4.692846034214619e-05,
      "loss": 0.010668348520994186,
      "step": 6708
    },
    {
      "epoch": 1.2717968157695223,
      "grad_norm": 0.2727532684803009,
      "learning_rate": 4.6850699844479007e-05,
      "loss": 0.020377732813358307,
      "step": 6710
    },
    {
      "epoch": 1.2721758908263836,
      "grad_norm": 0.35075896978378296,
      "learning_rate": 4.677293934681182e-05,
      "loss": 0.021613994613289833,
      "step": 6712
    },
    {
      "epoch": 1.272554965883245,
      "grad_norm": 0.387128084897995,
      "learning_rate": 4.669517884914464e-05,
      "loss": 0.020127300173044205,
      "step": 6714
    },
    {
      "epoch": 1.2729340409401062,
      "grad_norm": 0.31301769614219666,
      "learning_rate": 4.6617418351477454e-05,
      "loss": 0.022949695587158203,
      "step": 6716
    },
    {
      "epoch": 1.2733131159969675,
      "grad_norm": 0.1773374378681183,
      "learning_rate": 4.653965785381027e-05,
      "loss": 0.015780769288539886,
      "step": 6718
    },
    {
      "epoch": 1.2736921910538286,
      "grad_norm": 0.087679423391819,
      "learning_rate": 4.646189735614308e-05,
      "loss": 0.01161986030638218,
      "step": 6720
    },
    {
      "epoch": 1.2740712661106899,
      "grad_norm": 0.11965058743953705,
      "learning_rate": 4.6384136858475895e-05,
      "loss": 0.02199510484933853,
      "step": 6722
    },
    {
      "epoch": 1.2744503411675512,
      "grad_norm": 0.17691513895988464,
      "learning_rate": 4.630637636080871e-05,
      "loss": 0.02219136245548725,
      "step": 6724
    },
    {
      "epoch": 1.2748294162244125,
      "grad_norm": 0.22453930974006653,
      "learning_rate": 4.622861586314153e-05,
      "loss": 0.013672828674316406,
      "step": 6726
    },
    {
      "epoch": 1.2752084912812736,
      "grad_norm": 0.09213010221719742,
      "learning_rate": 4.615085536547434e-05,
      "loss": 0.01646639220416546,
      "step": 6728
    },
    {
      "epoch": 1.2755875663381349,
      "grad_norm": 0.09987153112888336,
      "learning_rate": 4.607309486780716e-05,
      "loss": 0.012295449152588844,
      "step": 6730
    },
    {
      "epoch": 1.2759666413949962,
      "grad_norm": 0.23739810287952423,
      "learning_rate": 4.599533437013997e-05,
      "loss": 0.015027303248643875,
      "step": 6732
    },
    {
      "epoch": 1.2763457164518575,
      "grad_norm": 0.4335266947746277,
      "learning_rate": 4.5917573872472784e-05,
      "loss": 0.024937963113188744,
      "step": 6734
    },
    {
      "epoch": 1.2767247915087188,
      "grad_norm": 0.34860900044441223,
      "learning_rate": 4.5839813374805604e-05,
      "loss": 0.024580104276537895,
      "step": 6736
    },
    {
      "epoch": 1.27710386656558,
      "grad_norm": 0.11101323366165161,
      "learning_rate": 4.576205287713842e-05,
      "loss": 0.02507062815129757,
      "step": 6738
    },
    {
      "epoch": 1.2774829416224414,
      "grad_norm": 0.38399022817611694,
      "learning_rate": 4.568429237947123e-05,
      "loss": 0.024449428543448448,
      "step": 6740
    },
    {
      "epoch": 1.2778620166793024,
      "grad_norm": 0.09834415465593338,
      "learning_rate": 4.5606531881804045e-05,
      "loss": 0.018300877884030342,
      "step": 6742
    },
    {
      "epoch": 1.2782410917361637,
      "grad_norm": 0.10193822532892227,
      "learning_rate": 4.552877138413686e-05,
      "loss": 0.020390169695019722,
      "step": 6744
    },
    {
      "epoch": 1.278620166793025,
      "grad_norm": 0.34750619530677795,
      "learning_rate": 4.545101088646967e-05,
      "loss": 0.023547369986772537,
      "step": 6746
    },
    {
      "epoch": 1.2789992418498863,
      "grad_norm": 0.07722742855548859,
      "learning_rate": 4.537325038880249e-05,
      "loss": 0.012720857746899128,
      "step": 6748
    },
    {
      "epoch": 1.2793783169067474,
      "grad_norm": 0.2590624988079071,
      "learning_rate": 4.529548989113531e-05,
      "loss": 0.016674669459462166,
      "step": 6750
    },
    {
      "epoch": 1.2797573919636087,
      "grad_norm": 0.07350701093673706,
      "learning_rate": 4.521772939346812e-05,
      "loss": 0.012425791472196579,
      "step": 6752
    },
    {
      "epoch": 1.28013646702047,
      "grad_norm": 0.10172899812459946,
      "learning_rate": 4.5139968895800934e-05,
      "loss": 0.014347730204463005,
      "step": 6754
    },
    {
      "epoch": 1.2805155420773313,
      "grad_norm": 0.0887758806347847,
      "learning_rate": 4.506220839813375e-05,
      "loss": 0.019634416326880455,
      "step": 6756
    },
    {
      "epoch": 1.2808946171341926,
      "grad_norm": 0.08215755969285965,
      "learning_rate": 4.498444790046656e-05,
      "loss": 0.014701257459819317,
      "step": 6758
    },
    {
      "epoch": 1.281273692191054,
      "grad_norm": 0.16507884860038757,
      "learning_rate": 4.490668740279938e-05,
      "loss": 0.012882157228887081,
      "step": 6760
    },
    {
      "epoch": 1.2816527672479152,
      "grad_norm": 0.09147443622350693,
      "learning_rate": 4.4828926905132196e-05,
      "loss": 0.011846274137496948,
      "step": 6762
    },
    {
      "epoch": 1.2820318423047763,
      "grad_norm": 0.2506336569786072,
      "learning_rate": 4.475116640746501e-05,
      "loss": 0.022739043459296227,
      "step": 6764
    },
    {
      "epoch": 1.2824109173616376,
      "grad_norm": 0.26085183024406433,
      "learning_rate": 4.467340590979783e-05,
      "loss": 0.015728430822491646,
      "step": 6766
    },
    {
      "epoch": 1.282789992418499,
      "grad_norm": 0.10744857788085938,
      "learning_rate": 4.4595645412130637e-05,
      "loss": 0.01750854402780533,
      "step": 6768
    },
    {
      "epoch": 1.2831690674753602,
      "grad_norm": 0.1616607904434204,
      "learning_rate": 4.451788491446345e-05,
      "loss": 0.01640707440674305,
      "step": 6770
    },
    {
      "epoch": 1.2835481425322213,
      "grad_norm": 0.09759391099214554,
      "learning_rate": 4.444012441679627e-05,
      "loss": 0.013546339236199856,
      "step": 6772
    },
    {
      "epoch": 1.2839272175890826,
      "grad_norm": 0.10964616388082504,
      "learning_rate": 4.4362363919129084e-05,
      "loss": 0.01540332566946745,
      "step": 6774
    },
    {
      "epoch": 1.2843062926459439,
      "grad_norm": 0.20608565211296082,
      "learning_rate": 4.42846034214619e-05,
      "loss": 0.02409704029560089,
      "step": 6776
    },
    {
      "epoch": 1.2846853677028052,
      "grad_norm": 0.10301036387681961,
      "learning_rate": 4.420684292379472e-05,
      "loss": 0.015897484496235847,
      "step": 6778
    },
    {
      "epoch": 1.2850644427596665,
      "grad_norm": 0.43813371658325195,
      "learning_rate": 4.4129082426127525e-05,
      "loss": 0.027665840461850166,
      "step": 6780
    },
    {
      "epoch": 1.2854435178165278,
      "grad_norm": 0.3235606253147125,
      "learning_rate": 4.4051321928460346e-05,
      "loss": 0.020190022885799408,
      "step": 6782
    },
    {
      "epoch": 1.285822592873389,
      "grad_norm": 0.41405990719795227,
      "learning_rate": 4.397356143079316e-05,
      "loss": 0.02171286754310131,
      "step": 6784
    },
    {
      "epoch": 1.2862016679302501,
      "grad_norm": 0.32820627093315125,
      "learning_rate": 4.389580093312597e-05,
      "loss": 0.026646966114640236,
      "step": 6786
    },
    {
      "epoch": 1.2865807429871114,
      "grad_norm": 0.09782728552818298,
      "learning_rate": 4.3818040435458794e-05,
      "loss": 0.013289806433022022,
      "step": 6788
    },
    {
      "epoch": 1.2869598180439727,
      "grad_norm": 0.21477089822292328,
      "learning_rate": 4.374027993779161e-05,
      "loss": 0.014561587013304234,
      "step": 6790
    },
    {
      "epoch": 1.287338893100834,
      "grad_norm": 0.09810450673103333,
      "learning_rate": 4.3662519440124414e-05,
      "loss": 0.017292913049459457,
      "step": 6792
    },
    {
      "epoch": 1.2877179681576951,
      "grad_norm": 0.20321901142597198,
      "learning_rate": 4.3584758942457235e-05,
      "loss": 0.022135024890303612,
      "step": 6794
    },
    {
      "epoch": 1.2880970432145564,
      "grad_norm": 0.09849094599485397,
      "learning_rate": 4.350699844479005e-05,
      "loss": 0.012210364453494549,
      "step": 6796
    },
    {
      "epoch": 1.2884761182714177,
      "grad_norm": 0.11535486578941345,
      "learning_rate": 4.342923794712286e-05,
      "loss": 0.016022782772779465,
      "step": 6798
    },
    {
      "epoch": 1.288855193328279,
      "grad_norm": 0.12025796622037888,
      "learning_rate": 4.335147744945568e-05,
      "loss": 0.019703133031725883,
      "step": 6800
    },
    {
      "epoch": 1.2892342683851403,
      "grad_norm": 0.34028759598731995,
      "learning_rate": 4.3273716951788496e-05,
      "loss": 0.020543821156024933,
      "step": 6802
    },
    {
      "epoch": 1.2896133434420016,
      "grad_norm": 0.20886346697807312,
      "learning_rate": 4.31959564541213e-05,
      "loss": 0.0210945475846529,
      "step": 6804
    },
    {
      "epoch": 1.2899924184988627,
      "grad_norm": 0.39384135603904724,
      "learning_rate": 4.311819595645412e-05,
      "loss": 0.028534753248095512,
      "step": 6806
    },
    {
      "epoch": 1.290371493555724,
      "grad_norm": 0.39229118824005127,
      "learning_rate": 4.304043545878694e-05,
      "loss": 0.02078673616051674,
      "step": 6808
    },
    {
      "epoch": 1.2907505686125853,
      "grad_norm": 0.1050083264708519,
      "learning_rate": 4.296267496111975e-05,
      "loss": 0.021589849144220352,
      "step": 6810
    },
    {
      "epoch": 1.2911296436694466,
      "grad_norm": 0.18046949803829193,
      "learning_rate": 4.288491446345257e-05,
      "loss": 0.0223689042031765,
      "step": 6812
    },
    {
      "epoch": 1.2915087187263077,
      "grad_norm": 0.10601748526096344,
      "learning_rate": 4.2807153965785385e-05,
      "loss": 0.013571534305810928,
      "step": 6814
    },
    {
      "epoch": 1.291887793783169,
      "grad_norm": 0.32527756690979004,
      "learning_rate": 4.27293934681182e-05,
      "loss": 0.021951282396912575,
      "step": 6816
    },
    {
      "epoch": 1.2922668688400303,
      "grad_norm": 0.10407596826553345,
      "learning_rate": 4.265163297045101e-05,
      "loss": 0.013377000577747822,
      "step": 6818
    },
    {
      "epoch": 1.2926459438968916,
      "grad_norm": 0.20858433842658997,
      "learning_rate": 4.2573872472783826e-05,
      "loss": 0.01592605747282505,
      "step": 6820
    },
    {
      "epoch": 1.2930250189537529,
      "grad_norm": 0.10741230100393295,
      "learning_rate": 4.249611197511664e-05,
      "loss": 0.020901354029774666,
      "step": 6822
    },
    {
      "epoch": 1.2934040940106142,
      "grad_norm": 0.1507345736026764,
      "learning_rate": 4.241835147744946e-05,
      "loss": 0.01506338082253933,
      "step": 6824
    },
    {
      "epoch": 1.2937831690674755,
      "grad_norm": 0.28346773982048035,
      "learning_rate": 4.2340590979782273e-05,
      "loss": 0.01618008315563202,
      "step": 6826
    },
    {
      "epoch": 1.2941622441243366,
      "grad_norm": 0.21118156611919403,
      "learning_rate": 4.226283048211509e-05,
      "loss": 0.016221055760979652,
      "step": 6828
    },
    {
      "epoch": 1.2945413191811979,
      "grad_norm": 0.30092984437942505,
      "learning_rate": 4.21850699844479e-05,
      "loss": 0.02111215889453888,
      "step": 6830
    },
    {
      "epoch": 1.2949203942380592,
      "grad_norm": 0.08399129658937454,
      "learning_rate": 4.2107309486780714e-05,
      "loss": 0.011370539665222168,
      "step": 6832
    },
    {
      "epoch": 1.2952994692949205,
      "grad_norm": 0.27485203742980957,
      "learning_rate": 4.2029548989113535e-05,
      "loss": 0.026429632678627968,
      "step": 6834
    },
    {
      "epoch": 1.2956785443517815,
      "grad_norm": 0.2538629472255707,
      "learning_rate": 4.195178849144635e-05,
      "loss": 0.023872114717960358,
      "step": 6836
    },
    {
      "epoch": 1.2960576194086428,
      "grad_norm": 0.2075246423482895,
      "learning_rate": 4.187402799377916e-05,
      "loss": 0.01573754847049713,
      "step": 6838
    },
    {
      "epoch": 1.2964366944655041,
      "grad_norm": 0.08776737749576569,
      "learning_rate": 4.179626749611198e-05,
      "loss": 0.015847910195589066,
      "step": 6840
    },
    {
      "epoch": 1.2968157695223654,
      "grad_norm": 0.09900802373886108,
      "learning_rate": 4.171850699844479e-05,
      "loss": 0.019263645634055138,
      "step": 6842
    },
    {
      "epoch": 1.2971948445792267,
      "grad_norm": 0.3745097219944,
      "learning_rate": 4.16407465007776e-05,
      "loss": 0.024431614205241203,
      "step": 6844
    },
    {
      "epoch": 1.297573919636088,
      "grad_norm": 0.2726362347602844,
      "learning_rate": 4.1562986003110424e-05,
      "loss": 0.014333815313875675,
      "step": 6846
    },
    {
      "epoch": 1.2979529946929493,
      "grad_norm": 0.1101895272731781,
      "learning_rate": 4.148522550544324e-05,
      "loss": 0.01421152800321579,
      "step": 6848
    },
    {
      "epoch": 1.2983320697498104,
      "grad_norm": 0.10010267049074173,
      "learning_rate": 4.140746500777605e-05,
      "loss": 0.014291657134890556,
      "step": 6850
    },
    {
      "epoch": 1.2987111448066717,
      "grad_norm": 0.32296523451805115,
      "learning_rate": 4.132970451010887e-05,
      "loss": 0.014632634818553925,
      "step": 6852
    },
    {
      "epoch": 1.299090219863533,
      "grad_norm": 0.37054917216300964,
      "learning_rate": 4.125194401244168e-05,
      "loss": 0.022140607237815857,
      "step": 6854
    },
    {
      "epoch": 1.2994692949203943,
      "grad_norm": 0.6272478699684143,
      "learning_rate": 4.117418351477449e-05,
      "loss": 0.041496556252241135,
      "step": 6856
    },
    {
      "epoch": 1.2998483699772554,
      "grad_norm": 0.09090086072683334,
      "learning_rate": 4.109642301710731e-05,
      "loss": 0.013401562348008156,
      "step": 6858
    },
    {
      "epoch": 1.3002274450341167,
      "grad_norm": 0.14654211699962616,
      "learning_rate": 4.1018662519440126e-05,
      "loss": 0.013612187467515469,
      "step": 6860
    },
    {
      "epoch": 1.300606520090978,
      "grad_norm": 0.20941613614559174,
      "learning_rate": 4.094090202177294e-05,
      "loss": 0.01755112037062645,
      "step": 6862
    },
    {
      "epoch": 1.3009855951478393,
      "grad_norm": 0.26958155632019043,
      "learning_rate": 4.086314152410576e-05,
      "loss": 0.014772050082683563,
      "step": 6864
    },
    {
      "epoch": 1.3013646702047006,
      "grad_norm": 0.2256636619567871,
      "learning_rate": 4.0785381026438574e-05,
      "loss": 0.013732654973864555,
      "step": 6866
    },
    {
      "epoch": 1.301743745261562,
      "grad_norm": 0.09198566526174545,
      "learning_rate": 4.070762052877139e-05,
      "loss": 0.014553564600646496,
      "step": 6868
    },
    {
      "epoch": 1.302122820318423,
      "grad_norm": 0.3314821720123291,
      "learning_rate": 4.06298600311042e-05,
      "loss": 0.01775992661714554,
      "step": 6870
    },
    {
      "epoch": 1.3025018953752843,
      "grad_norm": 0.34394434094429016,
      "learning_rate": 4.0552099533437015e-05,
      "loss": 0.01943797804415226,
      "step": 6872
    },
    {
      "epoch": 1.3028809704321456,
      "grad_norm": 0.41157716512680054,
      "learning_rate": 4.047433903576983e-05,
      "loss": 0.02323996275663376,
      "step": 6874
    },
    {
      "epoch": 1.3032600454890069,
      "grad_norm": 0.09714145213365555,
      "learning_rate": 4.039657853810265e-05,
      "loss": 0.010909459553658962,
      "step": 6876
    },
    {
      "epoch": 1.303639120545868,
      "grad_norm": 0.17910443246364594,
      "learning_rate": 4.031881804043546e-05,
      "loss": 0.021100562065839767,
      "step": 6878
    },
    {
      "epoch": 1.3040181956027292,
      "grad_norm": 0.1646917164325714,
      "learning_rate": 4.0241057542768276e-05,
      "loss": 0.01382403913885355,
      "step": 6880
    },
    {
      "epoch": 1.3043972706595905,
      "grad_norm": 0.2795945703983307,
      "learning_rate": 4.016329704510109e-05,
      "loss": 0.015612194314599037,
      "step": 6882
    },
    {
      "epoch": 1.3047763457164518,
      "grad_norm": 0.09421509504318237,
      "learning_rate": 4.0085536547433904e-05,
      "loss": 0.012735328637063503,
      "step": 6884
    },
    {
      "epoch": 1.3051554207733131,
      "grad_norm": 0.09597474336624146,
      "learning_rate": 4.0007776049766724e-05,
      "loss": 0.013107165694236755,
      "step": 6886
    },
    {
      "epoch": 1.3055344958301744,
      "grad_norm": 0.07783867418766022,
      "learning_rate": 3.993001555209954e-05,
      "loss": 0.013469018042087555,
      "step": 6888
    },
    {
      "epoch": 1.3059135708870357,
      "grad_norm": 0.38964638113975525,
      "learning_rate": 3.985225505443235e-05,
      "loss": 0.01867057755589485,
      "step": 6890
    },
    {
      "epoch": 1.3062926459438968,
      "grad_norm": 0.10615058988332748,
      "learning_rate": 3.9774494556765165e-05,
      "loss": 0.017947575077414513,
      "step": 6892
    },
    {
      "epoch": 1.3066717210007581,
      "grad_norm": 0.1293010264635086,
      "learning_rate": 3.969673405909798e-05,
      "loss": 0.019810372963547707,
      "step": 6894
    },
    {
      "epoch": 1.3070507960576194,
      "grad_norm": 0.08842366188764572,
      "learning_rate": 3.961897356143079e-05,
      "loss": 0.01699863374233246,
      "step": 6896
    },
    {
      "epoch": 1.3074298711144807,
      "grad_norm": 0.09426766633987427,
      "learning_rate": 3.954121306376361e-05,
      "loss": 0.01810978725552559,
      "step": 6898
    },
    {
      "epoch": 1.3078089461713418,
      "grad_norm": 0.36551859974861145,
      "learning_rate": 3.9463452566096426e-05,
      "loss": 0.021492891013622284,
      "step": 6900
    },
    {
      "epoch": 1.308188021228203,
      "grad_norm": 0.10801100730895996,
      "learning_rate": 3.938569206842924e-05,
      "loss": 0.016153592616319656,
      "step": 6902
    },
    {
      "epoch": 1.3085670962850644,
      "grad_norm": 0.31147295236587524,
      "learning_rate": 3.9307931570762054e-05,
      "loss": 0.030154196545481682,
      "step": 6904
    },
    {
      "epoch": 1.3089461713419257,
      "grad_norm": 0.31920257210731506,
      "learning_rate": 3.923017107309487e-05,
      "loss": 0.01779846101999283,
      "step": 6906
    },
    {
      "epoch": 1.309325246398787,
      "grad_norm": 0.11293549090623856,
      "learning_rate": 3.915241057542768e-05,
      "loss": 0.020235126838088036,
      "step": 6908
    },
    {
      "epoch": 1.3097043214556483,
      "grad_norm": 0.33704468607902527,
      "learning_rate": 3.90746500777605e-05,
      "loss": 0.02211638167500496,
      "step": 6910
    },
    {
      "epoch": 1.3100833965125096,
      "grad_norm": 0.2916768193244934,
      "learning_rate": 3.8996889580093315e-05,
      "loss": 0.02685336396098137,
      "step": 6912
    },
    {
      "epoch": 1.3104624715693707,
      "grad_norm": 0.4087522625923157,
      "learning_rate": 3.891912908242613e-05,
      "loss": 0.023640993982553482,
      "step": 6914
    },
    {
      "epoch": 1.310841546626232,
      "grad_norm": 0.10911471396684647,
      "learning_rate": 3.884136858475895e-05,
      "loss": 0.01569041982293129,
      "step": 6916
    },
    {
      "epoch": 1.3112206216830933,
      "grad_norm": 0.2386486828327179,
      "learning_rate": 3.8763608087091756e-05,
      "loss": 0.019946472719311714,
      "step": 6918
    },
    {
      "epoch": 1.3115996967399546,
      "grad_norm": 0.10693595558404922,
      "learning_rate": 3.868584758942457e-05,
      "loss": 0.0229493398219347,
      "step": 6920
    },
    {
      "epoch": 1.3119787717968157,
      "grad_norm": 0.12572655081748962,
      "learning_rate": 3.860808709175739e-05,
      "loss": 0.015784071758389473,
      "step": 6922
    },
    {
      "epoch": 1.312357846853677,
      "grad_norm": 0.1539125144481659,
      "learning_rate": 3.8530326594090204e-05,
      "loss": 0.02637699991464615,
      "step": 6924
    },
    {
      "epoch": 1.3127369219105383,
      "grad_norm": 0.11130647361278534,
      "learning_rate": 3.845256609642302e-05,
      "loss": 0.01860138401389122,
      "step": 6926
    },
    {
      "epoch": 1.3131159969673996,
      "grad_norm": 0.35557618737220764,
      "learning_rate": 3.837480559875584e-05,
      "loss": 0.01833251118659973,
      "step": 6928
    },
    {
      "epoch": 1.3134950720242609,
      "grad_norm": 0.11704309284687042,
      "learning_rate": 3.8297045101088645e-05,
      "loss": 0.022770410403609276,
      "step": 6930
    },
    {
      "epoch": 1.3138741470811222,
      "grad_norm": 0.27550020813941956,
      "learning_rate": 3.8219284603421465e-05,
      "loss": 0.014041858725249767,
      "step": 6932
    },
    {
      "epoch": 1.3142532221379835,
      "grad_norm": 0.33896660804748535,
      "learning_rate": 3.814152410575428e-05,
      "loss": 0.019035013392567635,
      "step": 6934
    },
    {
      "epoch": 1.3146322971948445,
      "grad_norm": 0.38832980394363403,
      "learning_rate": 3.806376360808709e-05,
      "loss": 0.024245427921414375,
      "step": 6936
    },
    {
      "epoch": 1.3150113722517058,
      "grad_norm": 0.14632023870944977,
      "learning_rate": 3.798600311041991e-05,
      "loss": 0.012641670182347298,
      "step": 6938
    },
    {
      "epoch": 1.3153904473085671,
      "grad_norm": 0.11400631815195084,
      "learning_rate": 3.790824261275273e-05,
      "loss": 0.012243100441992283,
      "step": 6940
    },
    {
      "epoch": 1.3157695223654284,
      "grad_norm": 0.37765541672706604,
      "learning_rate": 3.7830482115085534e-05,
      "loss": 0.02998371608555317,
      "step": 6942
    },
    {
      "epoch": 1.3161485974222895,
      "grad_norm": 0.35802650451660156,
      "learning_rate": 3.7752721617418354e-05,
      "loss": 0.02168341912329197,
      "step": 6944
    },
    {
      "epoch": 1.3165276724791508,
      "grad_norm": 0.13223543763160706,
      "learning_rate": 3.767496111975117e-05,
      "loss": 0.018708763644099236,
      "step": 6946
    },
    {
      "epoch": 1.3169067475360121,
      "grad_norm": 0.08670308440923691,
      "learning_rate": 3.759720062208398e-05,
      "loss": 0.012345380149781704,
      "step": 6948
    },
    {
      "epoch": 1.3172858225928734,
      "grad_norm": 0.2220458835363388,
      "learning_rate": 3.75194401244168e-05,
      "loss": 0.013585429638624191,
      "step": 6950
    },
    {
      "epoch": 1.3176648976497347,
      "grad_norm": 0.4585461914539337,
      "learning_rate": 3.7441679626749615e-05,
      "loss": 0.03344937413930893,
      "step": 6952
    },
    {
      "epoch": 1.318043972706596,
      "grad_norm": 0.16000327467918396,
      "learning_rate": 3.736391912908242e-05,
      "loss": 0.013883688487112522,
      "step": 6954
    },
    {
      "epoch": 1.318423047763457,
      "grad_norm": 0.10516282916069031,
      "learning_rate": 3.728615863141524e-05,
      "loss": 0.013020012527704239,
      "step": 6956
    },
    {
      "epoch": 1.3188021228203184,
      "grad_norm": 0.1124725490808487,
      "learning_rate": 3.7208398133748056e-05,
      "loss": 0.015605375170707703,
      "step": 6958
    },
    {
      "epoch": 1.3191811978771797,
      "grad_norm": 0.3581644594669342,
      "learning_rate": 3.713063763608087e-05,
      "loss": 0.02876441925764084,
      "step": 6960
    },
    {
      "epoch": 1.319560272934041,
      "grad_norm": 0.32597485184669495,
      "learning_rate": 3.705287713841369e-05,
      "loss": 0.01969079114496708,
      "step": 6962
    },
    {
      "epoch": 1.319939347990902,
      "grad_norm": 0.10909604281187057,
      "learning_rate": 3.6975116640746504e-05,
      "loss": 0.015889795497059822,
      "step": 6964
    },
    {
      "epoch": 1.3203184230477634,
      "grad_norm": 0.0963759645819664,
      "learning_rate": 3.689735614307932e-05,
      "loss": 0.013149050995707512,
      "step": 6966
    },
    {
      "epoch": 1.3206974981046247,
      "grad_norm": 0.38263270258903503,
      "learning_rate": 3.681959564541213e-05,
      "loss": 0.02536228485405445,
      "step": 6968
    },
    {
      "epoch": 1.321076573161486,
      "grad_norm": 0.08327317237854004,
      "learning_rate": 3.6741835147744945e-05,
      "loss": 0.015229670330882072,
      "step": 6970
    },
    {
      "epoch": 1.3214556482183473,
      "grad_norm": 0.14364852011203766,
      "learning_rate": 3.666407465007776e-05,
      "loss": 0.011799897998571396,
      "step": 6972
    },
    {
      "epoch": 1.3218347232752086,
      "grad_norm": 0.2855004668235779,
      "learning_rate": 3.658631415241058e-05,
      "loss": 0.02193605527281761,
      "step": 6974
    },
    {
      "epoch": 1.3222137983320699,
      "grad_norm": 0.11189418286085129,
      "learning_rate": 3.650855365474339e-05,
      "loss": 0.018991177901625633,
      "step": 6976
    },
    {
      "epoch": 1.322592873388931,
      "grad_norm": 0.16199317574501038,
      "learning_rate": 3.6430793157076207e-05,
      "loss": 0.015011243522167206,
      "step": 6978
    },
    {
      "epoch": 1.3229719484457922,
      "grad_norm": 0.20758134126663208,
      "learning_rate": 3.635303265940902e-05,
      "loss": 0.017250578850507736,
      "step": 6980
    },
    {
      "epoch": 1.3233510235026535,
      "grad_norm": 0.09454892575740814,
      "learning_rate": 3.6275272161741834e-05,
      "loss": 0.011724129319190979,
      "step": 6982
    },
    {
      "epoch": 1.3237300985595148,
      "grad_norm": 0.315833181142807,
      "learning_rate": 3.6197511664074654e-05,
      "loss": 0.017134737223386765,
      "step": 6984
    },
    {
      "epoch": 1.324109173616376,
      "grad_norm": 0.3686119019985199,
      "learning_rate": 3.611975116640747e-05,
      "loss": 0.021528353914618492,
      "step": 6986
    },
    {
      "epoch": 1.3244882486732372,
      "grad_norm": 0.1430811583995819,
      "learning_rate": 3.604199066874028e-05,
      "loss": 0.012262069620192051,
      "step": 6988
    },
    {
      "epoch": 1.3248673237300985,
      "grad_norm": 0.3133852183818817,
      "learning_rate": 3.59642301710731e-05,
      "loss": 0.024227503687143326,
      "step": 6990
    },
    {
      "epoch": 1.3252463987869598,
      "grad_norm": 0.0873170867562294,
      "learning_rate": 3.588646967340591e-05,
      "loss": 0.014746197499334812,
      "step": 6992
    },
    {
      "epoch": 1.3256254738438211,
      "grad_norm": 0.21638639271259308,
      "learning_rate": 3.580870917573872e-05,
      "loss": 0.021136701107025146,
      "step": 6994
    },
    {
      "epoch": 1.3260045489006824,
      "grad_norm": 0.10831960290670395,
      "learning_rate": 3.573094867807154e-05,
      "loss": 0.013485527597367764,
      "step": 6996
    },
    {
      "epoch": 1.3263836239575437,
      "grad_norm": 0.09888295084238052,
      "learning_rate": 3.565318818040436e-05,
      "loss": 0.011069085448980331,
      "step": 6998
    },
    {
      "epoch": 1.3267626990144048,
      "grad_norm": 0.08116347342729568,
      "learning_rate": 3.557542768273717e-05,
      "loss": 0.014500055462121964,
      "step": 7000
    },
    {
      "epoch": 1.327141774071266,
      "grad_norm": 0.11749434471130371,
      "learning_rate": 3.549766718506999e-05,
      "loss": 0.016835741698741913,
      "step": 7002
    },
    {
      "epoch": 1.3275208491281274,
      "grad_norm": 0.4057128429412842,
      "learning_rate": 3.54199066874028e-05,
      "loss": 0.021245207637548447,
      "step": 7004
    },
    {
      "epoch": 1.3278999241849887,
      "grad_norm": 0.13641363382339478,
      "learning_rate": 3.534214618973561e-05,
      "loss": 0.014275741763412952,
      "step": 7006
    },
    {
      "epoch": 1.3282789992418498,
      "grad_norm": 0.3333810865879059,
      "learning_rate": 3.526438569206843e-05,
      "loss": 0.025892168283462524,
      "step": 7008
    },
    {
      "epoch": 1.328658074298711,
      "grad_norm": 0.07960397005081177,
      "learning_rate": 3.5186625194401246e-05,
      "loss": 0.011674914509057999,
      "step": 7010
    },
    {
      "epoch": 1.3290371493555724,
      "grad_norm": 0.09702818840742111,
      "learning_rate": 3.510886469673406e-05,
      "loss": 0.013889622874557972,
      "step": 7012
    },
    {
      "epoch": 1.3294162244124337,
      "grad_norm": 0.07394696027040482,
      "learning_rate": 3.503110419906688e-05,
      "loss": 0.012732177972793579,
      "step": 7014
    },
    {
      "epoch": 1.329795299469295,
      "grad_norm": 0.327348530292511,
      "learning_rate": 3.495334370139969e-05,
      "loss": 0.016279764473438263,
      "step": 7016
    },
    {
      "epoch": 1.3301743745261563,
      "grad_norm": 0.14588497579097748,
      "learning_rate": 3.487558320373251e-05,
      "loss": 0.011810185387730598,
      "step": 7018
    },
    {
      "epoch": 1.3305534495830174,
      "grad_norm": 0.11049766838550568,
      "learning_rate": 3.479782270606532e-05,
      "loss": 0.01777423731982708,
      "step": 7020
    },
    {
      "epoch": 1.3309325246398787,
      "grad_norm": 0.0907900258898735,
      "learning_rate": 3.4720062208398134e-05,
      "loss": 0.017259856685996056,
      "step": 7022
    },
    {
      "epoch": 1.33131159969674,
      "grad_norm": 0.07304060459136963,
      "learning_rate": 3.464230171073095e-05,
      "loss": 0.015668010339140892,
      "step": 7024
    },
    {
      "epoch": 1.3316906747536013,
      "grad_norm": 0.33725494146347046,
      "learning_rate": 3.456454121306377e-05,
      "loss": 0.025901885703206062,
      "step": 7026
    },
    {
      "epoch": 1.3320697498104623,
      "grad_norm": 0.07682058960199356,
      "learning_rate": 3.448678071539658e-05,
      "loss": 0.012554884888231754,
      "step": 7028
    },
    {
      "epoch": 1.3324488248673236,
      "grad_norm": 0.26720473170280457,
      "learning_rate": 3.4409020217729396e-05,
      "loss": 0.021070856600999832,
      "step": 7030
    },
    {
      "epoch": 1.332827899924185,
      "grad_norm": 0.2541983425617218,
      "learning_rate": 3.433125972006221e-05,
      "loss": 0.02624576911330223,
      "step": 7032
    },
    {
      "epoch": 1.3332069749810462,
      "grad_norm": 0.21937701106071472,
      "learning_rate": 3.425349922239502e-05,
      "loss": 0.017293356359004974,
      "step": 7034
    },
    {
      "epoch": 1.3335860500379075,
      "grad_norm": 0.27133750915527344,
      "learning_rate": 3.4175738724727843e-05,
      "loss": 0.014832057058811188,
      "step": 7036
    },
    {
      "epoch": 1.3339651250947688,
      "grad_norm": 0.3076864778995514,
      "learning_rate": 3.409797822706066e-05,
      "loss": 0.01604297384619713,
      "step": 7038
    },
    {
      "epoch": 1.3343442001516301,
      "grad_norm": 0.3630676865577698,
      "learning_rate": 3.402021772939347e-05,
      "loss": 0.025126056745648384,
      "step": 7040
    },
    {
      "epoch": 1.3347232752084912,
      "grad_norm": 0.420365571975708,
      "learning_rate": 3.3942457231726284e-05,
      "loss": 0.025438036769628525,
      "step": 7042
    },
    {
      "epoch": 1.3351023502653525,
      "grad_norm": 0.24395908415317535,
      "learning_rate": 3.38646967340591e-05,
      "loss": 0.014806739054620266,
      "step": 7044
    },
    {
      "epoch": 1.3354814253222138,
      "grad_norm": 0.22370609641075134,
      "learning_rate": 3.378693623639191e-05,
      "loss": 0.0218973308801651,
      "step": 7046
    },
    {
      "epoch": 1.335860500379075,
      "grad_norm": 0.44449561834335327,
      "learning_rate": 3.370917573872473e-05,
      "loss": 0.030073126778006554,
      "step": 7048
    },
    {
      "epoch": 1.3362395754359362,
      "grad_norm": 0.35569289326667786,
      "learning_rate": 3.3631415241057546e-05,
      "loss": 0.024607883766293526,
      "step": 7050
    },
    {
      "epoch": 1.3366186504927975,
      "grad_norm": 0.33939072489738464,
      "learning_rate": 3.355365474339036e-05,
      "loss": 0.020132776349782944,
      "step": 7052
    },
    {
      "epoch": 1.3369977255496588,
      "grad_norm": 0.38996806740760803,
      "learning_rate": 3.347589424572317e-05,
      "loss": 0.027632543817162514,
      "step": 7054
    },
    {
      "epoch": 1.33737680060652,
      "grad_norm": 0.4831487834453583,
      "learning_rate": 3.339813374805599e-05,
      "loss": 0.027082111686468124,
      "step": 7056
    },
    {
      "epoch": 1.3377558756633814,
      "grad_norm": 0.3057292103767395,
      "learning_rate": 3.33203732503888e-05,
      "loss": 0.02139776013791561,
      "step": 7058
    },
    {
      "epoch": 1.3381349507202427,
      "grad_norm": 0.3969627916812897,
      "learning_rate": 3.324261275272162e-05,
      "loss": 0.029444023966789246,
      "step": 7060
    },
    {
      "epoch": 1.338514025777104,
      "grad_norm": 0.11597791314125061,
      "learning_rate": 3.3164852255054435e-05,
      "loss": 0.011832162737846375,
      "step": 7062
    },
    {
      "epoch": 1.338893100833965,
      "grad_norm": 0.3019917607307434,
      "learning_rate": 3.308709175738725e-05,
      "loss": 0.01781550422310829,
      "step": 7064
    },
    {
      "epoch": 1.3392721758908264,
      "grad_norm": 0.2391713708639145,
      "learning_rate": 3.300933125972006e-05,
      "loss": 0.01916477084159851,
      "step": 7066
    },
    {
      "epoch": 1.3396512509476877,
      "grad_norm": 0.14139635860919952,
      "learning_rate": 3.2931570762052876e-05,
      "loss": 0.020075906068086624,
      "step": 7068
    },
    {
      "epoch": 1.340030326004549,
      "grad_norm": 0.16595587134361267,
      "learning_rate": 3.2853810264385696e-05,
      "loss": 0.015470298007130623,
      "step": 7070
    },
    {
      "epoch": 1.34040940106141,
      "grad_norm": 0.11122491210699081,
      "learning_rate": 3.277604976671851e-05,
      "loss": 0.013395694084465504,
      "step": 7072
    },
    {
      "epoch": 1.3407884761182713,
      "grad_norm": 0.14108067750930786,
      "learning_rate": 3.269828926905132e-05,
      "loss": 0.016507592052221298,
      "step": 7074
    },
    {
      "epoch": 1.3411675511751326,
      "grad_norm": 0.19386368989944458,
      "learning_rate": 3.262052877138414e-05,
      "loss": 0.0199641864746809,
      "step": 7076
    },
    {
      "epoch": 1.341546626231994,
      "grad_norm": 0.08635647594928741,
      "learning_rate": 3.254276827371696e-05,
      "loss": 0.012204709462821484,
      "step": 7078
    },
    {
      "epoch": 1.3419257012888552,
      "grad_norm": 0.10346367955207825,
      "learning_rate": 3.2465007776049764e-05,
      "loss": 0.013885973021388054,
      "step": 7080
    },
    {
      "epoch": 1.3423047763457165,
      "grad_norm": 0.3591643273830414,
      "learning_rate": 3.2387247278382585e-05,
      "loss": 0.018571853637695312,
      "step": 7082
    },
    {
      "epoch": 1.3426838514025778,
      "grad_norm": 0.1295822709798813,
      "learning_rate": 3.23094867807154e-05,
      "loss": 0.011774566024541855,
      "step": 7084
    },
    {
      "epoch": 1.343062926459439,
      "grad_norm": 0.23650424182415009,
      "learning_rate": 3.223172628304821e-05,
      "loss": 0.020735550671815872,
      "step": 7086
    },
    {
      "epoch": 1.3434420015163002,
      "grad_norm": 0.09456319361925125,
      "learning_rate": 3.215396578538103e-05,
      "loss": 0.01797882840037346,
      "step": 7088
    },
    {
      "epoch": 1.3438210765731615,
      "grad_norm": 0.07849165797233582,
      "learning_rate": 3.2076205287713846e-05,
      "loss": 0.015110492706298828,
      "step": 7090
    },
    {
      "epoch": 1.3442001516300228,
      "grad_norm": 0.274565190076828,
      "learning_rate": 3.199844479004665e-05,
      "loss": 0.017977287992835045,
      "step": 7092
    },
    {
      "epoch": 1.344579226686884,
      "grad_norm": 0.11070484668016434,
      "learning_rate": 3.1920684292379474e-05,
      "loss": 0.017915986478328705,
      "step": 7094
    },
    {
      "epoch": 1.3449583017437452,
      "grad_norm": 0.09465125948190689,
      "learning_rate": 3.184292379471229e-05,
      "loss": 0.013767309486865997,
      "step": 7096
    },
    {
      "epoch": 1.3453373768006065,
      "grad_norm": 0.0962444469332695,
      "learning_rate": 3.17651632970451e-05,
      "loss": 0.013191688805818558,
      "step": 7098
    },
    {
      "epoch": 1.3457164518574678,
      "grad_norm": 0.10409560054540634,
      "learning_rate": 3.168740279937792e-05,
      "loss": 0.022574730217456818,
      "step": 7100
    },
    {
      "epoch": 1.346095526914329,
      "grad_norm": 0.08370233327150345,
      "learning_rate": 3.1609642301710735e-05,
      "loss": 0.013929038308560848,
      "step": 7102
    },
    {
      "epoch": 1.3464746019711904,
      "grad_norm": 0.3775716722011566,
      "learning_rate": 3.153188180404354e-05,
      "loss": 0.020938239991664886,
      "step": 7104
    },
    {
      "epoch": 1.3468536770280515,
      "grad_norm": 0.3808495104312897,
      "learning_rate": 3.145412130637636e-05,
      "loss": 0.022203030064702034,
      "step": 7106
    },
    {
      "epoch": 1.3472327520849128,
      "grad_norm": 0.11695746332406998,
      "learning_rate": 3.1376360808709176e-05,
      "loss": 0.012756990268826485,
      "step": 7108
    },
    {
      "epoch": 1.347611827141774,
      "grad_norm": 0.2800600230693817,
      "learning_rate": 3.129860031104199e-05,
      "loss": 0.016774091869592667,
      "step": 7110
    },
    {
      "epoch": 1.3479909021986354,
      "grad_norm": 0.12105324864387512,
      "learning_rate": 3.122083981337481e-05,
      "loss": 0.018964381888508797,
      "step": 7112
    },
    {
      "epoch": 1.3483699772554965,
      "grad_norm": 0.35131004452705383,
      "learning_rate": 3.1143079315707624e-05,
      "loss": 0.02847745642066002,
      "step": 7114
    },
    {
      "epoch": 1.3487490523123578,
      "grad_norm": 0.31992653012275696,
      "learning_rate": 3.106531881804044e-05,
      "loss": 0.021496880799531937,
      "step": 7116
    },
    {
      "epoch": 1.349128127369219,
      "grad_norm": 0.2591596841812134,
      "learning_rate": 3.098755832037325e-05,
      "loss": 0.02007909305393696,
      "step": 7118
    },
    {
      "epoch": 1.3495072024260804,
      "grad_norm": 0.4958329200744629,
      "learning_rate": 3.0909797822706065e-05,
      "loss": 0.034775376319885254,
      "step": 7120
    },
    {
      "epoch": 1.3498862774829417,
      "grad_norm": 0.17406632006168365,
      "learning_rate": 3.083203732503888e-05,
      "loss": 0.014697134494781494,
      "step": 7122
    },
    {
      "epoch": 1.350265352539803,
      "grad_norm": 0.13312114775180817,
      "learning_rate": 3.07542768273717e-05,
      "loss": 0.013727381825447083,
      "step": 7124
    }
  ],
  "logging_steps": 2,
  "max_steps": 7914,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2375,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.0687963760020685e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}