{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 4002,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0007496251874062968,
      "grad_norm": 0.11279210702086269,
      "learning_rate": 4.987531172069825e-07,
      "loss": 0.0283,
      "step": 1
    },
    {
      "epoch": 0.0014992503748125937,
      "grad_norm": 0.0708025290604418,
      "learning_rate": 9.97506234413965e-07,
      "loss": 0.018,
      "step": 2
    },
    {
      "epoch": 0.0022488755622188904,
      "grad_norm": 0.08204012806077762,
      "learning_rate": 1.4962593516209476e-06,
      "loss": 0.0343,
      "step": 3
    },
    {
      "epoch": 0.0029985007496251873,
      "grad_norm": 0.1887406661276333,
      "learning_rate": 1.99501246882793e-06,
      "loss": 0.0264,
      "step": 4
    },
    {
      "epoch": 0.0037481259370314842,
      "grad_norm": 0.1702003667682934,
      "learning_rate": 2.493765586034913e-06,
      "loss": 0.0817,
      "step": 5
    },
    {
      "epoch": 0.004497751124437781,
      "grad_norm": 0.08633491698201652,
      "learning_rate": 2.992518703241895e-06,
      "loss": 0.0192,
      "step": 6
    },
    {
      "epoch": 0.005247376311844078,
      "grad_norm": 0.06668682918073623,
      "learning_rate": 3.4912718204488777e-06,
      "loss": 0.0269,
      "step": 7
    },
    {
      "epoch": 0.005997001499250375,
      "grad_norm": 0.0949644930191698,
      "learning_rate": 3.99002493765586e-06,
      "loss": 0.0214,
      "step": 8
    },
    {
      "epoch": 0.006746626686656672,
      "grad_norm": 0.05487535781427025,
      "learning_rate": 4.488778054862843e-06,
      "loss": 0.0159,
      "step": 9
    },
    {
      "epoch": 0.0074962518740629685,
      "grad_norm": 0.17217701524900897,
      "learning_rate": 4.987531172069826e-06,
      "loss": 0.0607,
      "step": 10
    },
    {
      "epoch": 0.008245877061469266,
      "grad_norm": 0.20182278003961085,
      "learning_rate": 5.486284289276808e-06,
      "loss": 0.0539,
      "step": 11
    },
    {
      "epoch": 0.008995502248875561,
      "grad_norm": 0.1312564555806688,
      "learning_rate": 5.98503740648379e-06,
      "loss": 0.0329,
      "step": 12
    },
    {
      "epoch": 0.009745127436281859,
      "grad_norm": 0.09508077532454609,
      "learning_rate": 6.483790523690773e-06,
      "loss": 0.1668,
      "step": 13
    },
    {
      "epoch": 0.010494752623688156,
      "grad_norm": 0.05902506464710472,
      "learning_rate": 6.982543640897755e-06,
      "loss": 0.0215,
      "step": 14
    },
    {
      "epoch": 0.011244377811094454,
      "grad_norm": 0.057580643912089995,
      "learning_rate": 7.481296758104739e-06,
      "loss": 0.0266,
      "step": 15
    },
    {
      "epoch": 0.01199400299850075,
      "grad_norm": 0.04716677617704238,
      "learning_rate": 7.98004987531172e-06,
      "loss": 0.0215,
      "step": 16
    },
    {
      "epoch": 0.012743628185907047,
      "grad_norm": 0.0795864015547478,
      "learning_rate": 8.478802992518704e-06,
      "loss": 0.0173,
      "step": 17
    },
    {
      "epoch": 0.013493253373313344,
      "grad_norm": 0.07671642061904087,
      "learning_rate": 8.977556109725686e-06,
      "loss": 0.0224,
      "step": 18
    },
    {
      "epoch": 0.01424287856071964,
      "grad_norm": 0.06337940392694962,
      "learning_rate": 9.476309226932668e-06,
      "loss": 0.0162,
      "step": 19
    },
    {
      "epoch": 0.014992503748125937,
      "grad_norm": 0.15247751367927734,
      "learning_rate": 9.975062344139652e-06,
      "loss": 0.0459,
      "step": 20
    },
    {
      "epoch": 0.015742128935532233,
      "grad_norm": 0.09470608329189008,
      "learning_rate": 1.0473815461346634e-05,
      "loss": 0.0076,
      "step": 21
    },
    {
      "epoch": 0.01649175412293853,
      "grad_norm": 0.06575899186194488,
      "learning_rate": 1.0972568578553616e-05,
      "loss": 0.0192,
      "step": 22
    },
    {
      "epoch": 0.017241379310344827,
      "grad_norm": 0.05557891385996695,
      "learning_rate": 1.1471321695760599e-05,
      "loss": 0.0163,
      "step": 23
    },
    {
      "epoch": 0.017991004497751123,
      "grad_norm": 0.07023715922511192,
      "learning_rate": 1.197007481296758e-05,
      "loss": 0.0299,
      "step": 24
    },
    {
      "epoch": 0.018740629685157422,
      "grad_norm": 0.05840741377060999,
      "learning_rate": 1.2468827930174564e-05,
      "loss": 0.0279,
      "step": 25
    },
    {
      "epoch": 0.019490254872563718,
      "grad_norm": 0.09879994131870222,
      "learning_rate": 1.2967581047381547e-05,
      "loss": 0.0417,
      "step": 26
    },
    {
      "epoch": 0.020239880059970013,
      "grad_norm": 0.09422607763934358,
      "learning_rate": 1.3466334164588529e-05,
      "loss": 0.0233,
      "step": 27
    },
    {
      "epoch": 0.020989505247376312,
      "grad_norm": 0.08381440041433495,
      "learning_rate": 1.396508728179551e-05,
      "loss": 0.0384,
      "step": 28
    },
    {
      "epoch": 0.021739130434782608,
      "grad_norm": 0.06866246125337262,
      "learning_rate": 1.4463840399002496e-05,
      "loss": 0.022,
      "step": 29
    },
    {
      "epoch": 0.022488755622188907,
      "grad_norm": 0.09579356805805937,
      "learning_rate": 1.4962593516209478e-05,
      "loss": 0.029,
      "step": 30
    },
    {
      "epoch": 0.023238380809595203,
      "grad_norm": 0.10676598627834114,
      "learning_rate": 1.546134663341646e-05,
      "loss": 0.0338,
      "step": 31
    },
    {
      "epoch": 0.0239880059970015,
      "grad_norm": 0.05483831376969332,
      "learning_rate": 1.596009975062344e-05,
      "loss": 0.017,
      "step": 32
    },
    {
      "epoch": 0.024737631184407798,
      "grad_norm": 0.08692197588794064,
      "learning_rate": 1.6458852867830423e-05,
      "loss": 0.0264,
      "step": 33
    },
    {
      "epoch": 0.025487256371814093,
      "grad_norm": 0.0925341267319359,
      "learning_rate": 1.695760598503741e-05,
      "loss": 0.0419,
      "step": 34
    },
    {
      "epoch": 0.02623688155922039,
      "grad_norm": 0.0671445291722069,
      "learning_rate": 1.745635910224439e-05,
      "loss": 0.0164,
      "step": 35
    },
    {
      "epoch": 0.026986506746626688,
      "grad_norm": 0.16026579884927714,
      "learning_rate": 1.7955112219451373e-05,
      "loss": 0.0221,
      "step": 36
    },
    {
      "epoch": 0.027736131934032984,
      "grad_norm": 0.18362459813157384,
      "learning_rate": 1.8453865336658355e-05,
      "loss": 0.0489,
      "step": 37
    },
    {
      "epoch": 0.02848575712143928,
      "grad_norm": 0.06518981670839212,
      "learning_rate": 1.8952618453865337e-05,
      "loss": 0.0174,
      "step": 38
    },
    {
      "epoch": 0.02923538230884558,
      "grad_norm": 0.06889355143618404,
      "learning_rate": 1.945137157107232e-05,
      "loss": 0.0191,
      "step": 39
    },
    {
      "epoch": 0.029985007496251874,
      "grad_norm": 0.21562757118442988,
      "learning_rate": 1.9950124688279304e-05,
      "loss": 0.0299,
      "step": 40
    },
    {
      "epoch": 0.03073463268365817,
      "grad_norm": 0.14245617300583052,
      "learning_rate": 2.0448877805486287e-05,
      "loss": 0.0442,
      "step": 41
    },
    {
      "epoch": 0.031484257871064465,
      "grad_norm": 0.09411252796667173,
      "learning_rate": 2.094763092269327e-05,
      "loss": 0.0319,
      "step": 42
    },
    {
      "epoch": 0.03223388305847077,
      "grad_norm": 0.08117541156848603,
      "learning_rate": 2.144638403990025e-05,
      "loss": 0.0225,
      "step": 43
    },
    {
      "epoch": 0.03298350824587706,
      "grad_norm": 0.08225245879431137,
      "learning_rate": 2.1945137157107233e-05,
      "loss": 0.0361,
      "step": 44
    },
    {
      "epoch": 0.03373313343328336,
      "grad_norm": 0.07830505167470701,
      "learning_rate": 2.2443890274314215e-05,
      "loss": 0.0315,
      "step": 45
    },
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 0.06700726880908887,
      "learning_rate": 2.2942643391521197e-05,
      "loss": 0.0153,
      "step": 46
    },
    {
      "epoch": 0.03523238380809595,
      "grad_norm": 0.4031671176539563,
      "learning_rate": 2.344139650872818e-05,
      "loss": 0.0246,
      "step": 47
    },
    {
      "epoch": 0.035982008995502246,
      "grad_norm": 0.09991732387986536,
      "learning_rate": 2.394014962593516e-05,
      "loss": 0.0346,
      "step": 48
    },
    {
      "epoch": 0.03673163418290855,
      "grad_norm": 0.07162079160721155,
      "learning_rate": 2.4438902743142143e-05,
      "loss": 0.0314,
      "step": 49
    },
    {
      "epoch": 0.037481259370314844,
      "grad_norm": 0.0592546887993547,
      "learning_rate": 2.493765586034913e-05,
      "loss": 0.0154,
      "step": 50
    },
    {
      "epoch": 0.03823088455772114,
      "grad_norm": 0.08479532758764814,
      "learning_rate": 2.5436408977556108e-05,
      "loss": 0.0511,
      "step": 51
    },
    {
      "epoch": 0.038980509745127435,
      "grad_norm": 0.10538449374560438,
      "learning_rate": 2.5935162094763093e-05,
      "loss": 0.0296,
      "step": 52
    },
    {
      "epoch": 0.03973013493253373,
      "grad_norm": 0.05703635747748183,
      "learning_rate": 2.643391521197008e-05,
      "loss": 0.0191,
      "step": 53
    },
    {
      "epoch": 0.04047976011994003,
      "grad_norm": 0.07272076563350424,
      "learning_rate": 2.6932668329177057e-05,
      "loss": 0.0289,
      "step": 54
    },
    {
      "epoch": 0.04122938530734633,
      "grad_norm": 0.06914294384159929,
      "learning_rate": 2.7431421446384043e-05,
      "loss": 0.0216,
      "step": 55
    },
    {
      "epoch": 0.041979010494752625,
      "grad_norm": 0.12458369744795132,
      "learning_rate": 2.793017456359102e-05,
      "loss": 0.0521,
      "step": 56
    },
    {
      "epoch": 0.04272863568215892,
      "grad_norm": 0.03137816907539504,
      "learning_rate": 2.8428927680798007e-05,
      "loss": 0.01,
      "step": 57
    },
    {
      "epoch": 0.043478260869565216,
      "grad_norm": 0.04739555854735861,
      "learning_rate": 2.8927680798004992e-05,
      "loss": 0.0086,
      "step": 58
    },
    {
      "epoch": 0.04422788605697151,
      "grad_norm": 0.10036103034066238,
      "learning_rate": 2.942643391521197e-05,
      "loss": 0.0878,
      "step": 59
    },
    {
      "epoch": 0.044977511244377814,
      "grad_norm": 0.06748868600841415,
      "learning_rate": 2.9925187032418957e-05,
      "loss": 0.0248,
      "step": 60
    },
    {
      "epoch": 0.04572713643178411,
      "grad_norm": 0.07502502472970331,
      "learning_rate": 3.0423940149625935e-05,
      "loss": 0.0255,
      "step": 61
    },
    {
      "epoch": 0.046476761619190406,
      "grad_norm": 0.07710892432454533,
      "learning_rate": 3.092269326683292e-05,
      "loss": 0.0263,
      "step": 62
    },
    {
      "epoch": 0.0472263868065967,
      "grad_norm": 0.10538603215724578,
      "learning_rate": 3.14214463840399e-05,
      "loss": 0.0325,
      "step": 63
    },
    {
      "epoch": 0.047976011994003,
      "grad_norm": 0.07735064418610153,
      "learning_rate": 3.192019950124688e-05,
      "loss": 0.0378,
      "step": 64
    },
    {
      "epoch": 0.04872563718140929,
      "grad_norm": 0.03295359845033927,
      "learning_rate": 3.241895261845387e-05,
      "loss": 0.0105,
      "step": 65
    },
    {
      "epoch": 0.049475262368815595,
      "grad_norm": 0.1474088544285246,
      "learning_rate": 3.2917705735660846e-05,
      "loss": 0.026,
      "step": 66
    },
    {
      "epoch": 0.05022488755622189,
      "grad_norm": 0.05921704611745389,
      "learning_rate": 3.341645885286783e-05,
      "loss": 0.0139,
      "step": 67
    },
    {
      "epoch": 0.050974512743628186,
      "grad_norm": 0.14129659505130113,
      "learning_rate": 3.391521197007482e-05,
      "loss": 0.0301,
      "step": 68
    },
    {
      "epoch": 0.05172413793103448,
      "grad_norm": 0.10626253201464791,
      "learning_rate": 3.4413965087281796e-05,
      "loss": 0.0479,
      "step": 69
    },
    {
      "epoch": 0.05247376311844078,
      "grad_norm": 0.07763289594003318,
      "learning_rate": 3.491271820448878e-05,
      "loss": 0.0106,
      "step": 70
    },
    {
      "epoch": 0.05322338830584707,
      "grad_norm": 0.0691839287861348,
      "learning_rate": 3.541147132169576e-05,
      "loss": 0.0271,
      "step": 71
    },
    {
      "epoch": 0.053973013493253376,
      "grad_norm": 0.07070200939648912,
      "learning_rate": 3.5910224438902745e-05,
      "loss": 0.0223,
      "step": 72
    },
    {
      "epoch": 0.05472263868065967,
      "grad_norm": 0.19778882861726957,
      "learning_rate": 3.640897755610973e-05,
      "loss": 0.0465,
      "step": 73
    },
    {
      "epoch": 0.05547226386806597,
      "grad_norm": 0.060353823216797546,
      "learning_rate": 3.690773067331671e-05,
      "loss": 0.0246,
      "step": 74
    },
    {
      "epoch": 0.05622188905547226,
      "grad_norm": 0.07075868858224835,
      "learning_rate": 3.7406483790523695e-05,
      "loss": 0.0209,
      "step": 75
    },
    {
      "epoch": 0.05697151424287856,
      "grad_norm": 0.08762014555675333,
      "learning_rate": 3.7905236907730674e-05,
      "loss": 0.0311,
      "step": 76
    },
    {
      "epoch": 0.05772113943028486,
      "grad_norm": 0.1326222205107144,
      "learning_rate": 3.840399002493766e-05,
      "loss": 0.0407,
      "step": 77
    },
    {
      "epoch": 0.05847076461769116,
      "grad_norm": 0.0952714563149364,
      "learning_rate": 3.890274314214464e-05,
      "loss": 0.0479,
      "step": 78
    },
    {
      "epoch": 0.05922038980509745,
      "grad_norm": 0.07813721903130615,
      "learning_rate": 3.9401496259351623e-05,
      "loss": 0.0194,
      "step": 79
    },
    {
      "epoch": 0.05997001499250375,
      "grad_norm": 0.03586574592365885,
      "learning_rate": 3.990024937655861e-05,
      "loss": 0.0113,
      "step": 80
    },
    {
      "epoch": 0.060719640179910044,
      "grad_norm": 0.05312807875353562,
      "learning_rate": 4.039900249376559e-05,
      "loss": 0.0207,
      "step": 81
    },
    {
      "epoch": 0.06146926536731634,
      "grad_norm": 0.08206370101583557,
      "learning_rate": 4.089775561097257e-05,
      "loss": 0.0213,
      "step": 82
    },
    {
      "epoch": 0.06221889055472264,
      "grad_norm": 0.057566972449176544,
      "learning_rate": 4.139650872817955e-05,
      "loss": 0.0182,
      "step": 83
    },
    {
      "epoch": 0.06296851574212893,
      "grad_norm": 0.08521467516485813,
      "learning_rate": 4.189526184538654e-05,
      "loss": 0.0241,
      "step": 84
    },
    {
      "epoch": 0.06371814092953523,
      "grad_norm": 0.03571392648309694,
      "learning_rate": 4.239401496259352e-05,
      "loss": 0.0161,
      "step": 85
    },
    {
      "epoch": 0.06446776611694154,
      "grad_norm": 0.1230559997444316,
      "learning_rate": 4.28927680798005e-05,
      "loss": 0.0351,
      "step": 86
    },
    {
      "epoch": 0.06521739130434782,
      "grad_norm": 0.12512964537023197,
      "learning_rate": 4.339152119700749e-05,
      "loss": 0.0665,
      "step": 87
    },
    {
      "epoch": 0.06596701649175413,
      "grad_norm": 0.12217119750622546,
      "learning_rate": 4.3890274314214466e-05,
      "loss": 0.0136,
      "step": 88
    },
    {
      "epoch": 0.06671664167916042,
      "grad_norm": 0.0433698380582678,
      "learning_rate": 4.438902743142145e-05,
      "loss": 0.0323,
      "step": 89
    },
    {
      "epoch": 0.06746626686656672,
      "grad_norm": 0.12633070169699792,
      "learning_rate": 4.488778054862843e-05,
      "loss": 0.0179,
      "step": 90
    },
    {
      "epoch": 0.068215892053973,
      "grad_norm": 0.04740228242710997,
      "learning_rate": 4.5386533665835415e-05,
      "loss": 0.0202,
      "step": 91
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 0.061212399209499495,
      "learning_rate": 4.5885286783042394e-05,
      "loss": 0.0256,
      "step": 92
    },
    {
      "epoch": 0.06971514242878561,
      "grad_norm": 0.05192083675767972,
      "learning_rate": 4.638403990024938e-05,
      "loss": 0.016,
      "step": 93
    },
    {
      "epoch": 0.0704647676161919,
      "grad_norm": 0.1183986051919429,
      "learning_rate": 4.688279301745636e-05,
      "loss": 0.0392,
      "step": 94
    },
    {
      "epoch": 0.0712143928035982,
      "grad_norm": 0.0841259080165031,
      "learning_rate": 4.7381546134663344e-05,
      "loss": 0.0276,
      "step": 95
    },
    {
      "epoch": 0.07196401799100449,
      "grad_norm": 0.03569960655258622,
      "learning_rate": 4.788029925187032e-05,
      "loss": 0.0095,
      "step": 96
    },
    {
      "epoch": 0.0727136431784108,
      "grad_norm": 0.11869332056036354,
      "learning_rate": 4.837905236907731e-05,
      "loss": 0.0246,
      "step": 97
    },
    {
      "epoch": 0.0734632683658171,
      "grad_norm": 0.1133992731680118,
      "learning_rate": 4.887780548628429e-05,
      "loss": 0.0455,
      "step": 98
    },
    {
      "epoch": 0.07421289355322339,
      "grad_norm": 0.05434940900593147,
      "learning_rate": 4.937655860349127e-05,
      "loss": 0.0209,
      "step": 99
    },
    {
      "epoch": 0.07496251874062969,
      "grad_norm": 0.05742963314221326,
      "learning_rate": 4.987531172069826e-05,
      "loss": 0.0179,
      "step": 100
    },
    {
      "epoch": 0.07571214392803598,
      "grad_norm": 0.07358933301031388,
      "learning_rate": 5.037406483790524e-05,
      "loss": 0.0325,
      "step": 101
    },
    {
      "epoch": 0.07646176911544228,
      "grad_norm": 0.14545582068026564,
      "learning_rate": 5.0872817955112215e-05,
      "loss": 0.0264,
      "step": 102
    },
    {
      "epoch": 0.07721139430284858,
      "grad_norm": 0.09413188759154308,
      "learning_rate": 5.13715710723192e-05,
      "loss": 0.0253,
      "step": 103
    },
    {
      "epoch": 0.07796101949025487,
      "grad_norm": 0.03963600360457223,
      "learning_rate": 5.1870324189526186e-05,
      "loss": 0.0127,
      "step": 104
    },
    {
      "epoch": 0.07871064467766117,
      "grad_norm": 0.07599629125236715,
      "learning_rate": 5.236907730673317e-05,
      "loss": 0.0251,
      "step": 105
    },
    {
      "epoch": 0.07946026986506746,
      "grad_norm": 0.04480432796665436,
      "learning_rate": 5.286783042394016e-05,
      "loss": 0.0163,
      "step": 106
    },
    {
      "epoch": 0.08020989505247376,
      "grad_norm": 0.08201871045223541,
      "learning_rate": 5.336658354114713e-05,
      "loss": 0.0251,
      "step": 107
    },
    {
      "epoch": 0.08095952023988005,
      "grad_norm": 0.13488552993210706,
      "learning_rate": 5.3865336658354115e-05,
      "loss": 0.0216,
      "step": 108
    },
    {
      "epoch": 0.08170914542728636,
      "grad_norm": 0.16982953183139035,
      "learning_rate": 5.43640897755611e-05,
      "loss": 0.0721,
      "step": 109
    },
    {
      "epoch": 0.08245877061469266,
      "grad_norm": 0.07036861440927666,
      "learning_rate": 5.4862842892768086e-05,
      "loss": 0.0195,
      "step": 110
    },
    {
      "epoch": 0.08320839580209895,
      "grad_norm": 0.10363789462535822,
      "learning_rate": 5.536159600997507e-05,
      "loss": 0.0362,
      "step": 111
    },
    {
      "epoch": 0.08395802098950525,
      "grad_norm": 0.1389694465437715,
      "learning_rate": 5.586034912718204e-05,
      "loss": 0.0402,
      "step": 112
    },
    {
      "epoch": 0.08470764617691154,
      "grad_norm": 0.037148370829441675,
      "learning_rate": 5.635910224438903e-05,
      "loss": 0.0105,
      "step": 113
    },
    {
      "epoch": 0.08545727136431784,
      "grad_norm": 0.15980843277146153,
      "learning_rate": 5.6857855361596014e-05,
      "loss": 0.0581,
      "step": 114
    },
    {
      "epoch": 0.08620689655172414,
      "grad_norm": 0.12164293405284633,
      "learning_rate": 5.7356608478803e-05,
      "loss": 0.0336,
      "step": 115
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 0.10837151443746265,
      "learning_rate": 5.7855361596009985e-05,
      "loss": 0.0503,
      "step": 116
    },
    {
      "epoch": 0.08770614692653673,
      "grad_norm": 0.1354782049348464,
      "learning_rate": 5.835411471321696e-05,
      "loss": 0.0274,
      "step": 117
    },
    {
      "epoch": 0.08845577211394302,
      "grad_norm": 0.06563788761938176,
      "learning_rate": 5.885286783042394e-05,
      "loss": 0.0272,
      "step": 118
    },
    {
      "epoch": 0.08920539730134933,
      "grad_norm": 0.07104946785041905,
      "learning_rate": 5.935162094763093e-05,
      "loss": 0.0229,
      "step": 119
    },
    {
      "epoch": 0.08995502248875563,
      "grad_norm": 0.16694741445858008,
      "learning_rate": 5.985037406483791e-05,
      "loss": 0.0372,
      "step": 120
    },
    {
      "epoch": 0.09070464767616192,
      "grad_norm": 0.07864447323446397,
      "learning_rate": 6.03491271820449e-05,
      "loss": 0.0286,
      "step": 121
    },
    {
      "epoch": 0.09145427286356822,
      "grad_norm": 0.1366701828054804,
      "learning_rate": 6.084788029925187e-05,
      "loss": 0.0231,
      "step": 122
    },
    {
      "epoch": 0.09220389805097451,
      "grad_norm": 0.1070346979738295,
      "learning_rate": 6.134663341645886e-05,
      "loss": 0.0653,
      "step": 123
    },
    {
      "epoch": 0.09295352323838081,
      "grad_norm": 0.06698803113978262,
      "learning_rate": 6.184538653366583e-05,
      "loss": 0.0292,
      "step": 124
    },
    {
      "epoch": 0.0937031484257871,
      "grad_norm": 0.07097618623732171,
      "learning_rate": 6.234413965087283e-05,
      "loss": 0.0283,
      "step": 125
    },
    {
      "epoch": 0.0944527736131934,
      "grad_norm": 0.042448624191001115,
      "learning_rate": 6.28428927680798e-05,
      "loss": 0.0086,
      "step": 126
    },
    {
      "epoch": 0.0952023988005997,
      "grad_norm": 0.08465946332106826,
      "learning_rate": 6.334164588528678e-05,
      "loss": 0.0279,
      "step": 127
    },
    {
      "epoch": 0.095952023988006,
      "grad_norm": 0.089216378819531,
      "learning_rate": 6.384039900249376e-05,
      "loss": 0.031,
      "step": 128
    },
    {
      "epoch": 0.0967016491754123,
      "grad_norm": 0.07334432224365899,
      "learning_rate": 6.433915211970076e-05,
      "loss": 0.0138,
      "step": 129
    },
    {
      "epoch": 0.09745127436281859,
      "grad_norm": 0.0647148548610915,
      "learning_rate": 6.483790523690773e-05,
      "loss": 0.009,
      "step": 130
    },
    {
      "epoch": 0.09820089955022489,
      "grad_norm": 0.058673312076161714,
      "learning_rate": 6.533665835411473e-05,
      "loss": 0.0205,
      "step": 131
    },
    {
      "epoch": 0.09895052473763119,
      "grad_norm": 0.11603287692005247,
      "learning_rate": 6.583541147132169e-05,
      "loss": 0.0437,
      "step": 132
    },
    {
      "epoch": 0.09970014992503748,
      "grad_norm": 0.21370135684699681,
      "learning_rate": 6.633416458852868e-05,
      "loss": 0.0269,
      "step": 133
    },
    {
      "epoch": 0.10044977511244378,
      "grad_norm": 0.15923649990512292,
      "learning_rate": 6.683291770573566e-05,
      "loss": 0.0316,
      "step": 134
    },
    {
      "epoch": 0.10119940029985007,
      "grad_norm": 0.03278981431531208,
      "learning_rate": 6.733167082294266e-05,
      "loss": 0.0094,
      "step": 135
    },
    {
      "epoch": 0.10194902548725637,
      "grad_norm": 0.15833124285542596,
      "learning_rate": 6.783042394014963e-05,
      "loss": 0.0437,
      "step": 136
    },
    {
      "epoch": 0.10269865067466268,
      "grad_norm": 0.12400180816439312,
      "learning_rate": 6.832917705735661e-05,
      "loss": 0.0366,
      "step": 137
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 0.07377117070595124,
      "learning_rate": 6.882793017456359e-05,
      "loss": 0.0123,
      "step": 138
    },
    {
      "epoch": 0.10419790104947527,
      "grad_norm": 0.07122232015831038,
      "learning_rate": 6.932668329177058e-05,
      "loss": 0.0142,
      "step": 139
    },
    {
      "epoch": 0.10494752623688156,
      "grad_norm": 0.0332242122691683,
      "learning_rate": 6.982543640897756e-05,
      "loss": 0.0081,
      "step": 140
    },
    {
      "epoch": 0.10569715142428786,
      "grad_norm": 0.08678818831164349,
      "learning_rate": 7.032418952618454e-05,
      "loss": 0.0201,
      "step": 141
    },
    {
      "epoch": 0.10644677661169415,
      "grad_norm": 0.10250483247896906,
      "learning_rate": 7.082294264339152e-05,
      "loss": 0.0328,
      "step": 142
    },
    {
      "epoch": 0.10719640179910045,
      "grad_norm": 0.21159615399298012,
      "learning_rate": 7.132169576059851e-05,
      "loss": 0.0515,
      "step": 143
    },
    {
      "epoch": 0.10794602698650675,
      "grad_norm": 0.10522703720494737,
      "learning_rate": 7.182044887780549e-05,
      "loss": 0.0387,
      "step": 144
    },
    {
      "epoch": 0.10869565217391304,
      "grad_norm": 0.11820893519071421,
      "learning_rate": 7.231920199501247e-05,
      "loss": 0.0257,
      "step": 145
    },
    {
      "epoch": 0.10944527736131934,
      "grad_norm": 0.13986913155043815,
      "learning_rate": 7.281795511221946e-05,
      "loss": 0.0151,
      "step": 146
    },
    {
      "epoch": 0.11019490254872563,
      "grad_norm": 0.0559111712109555,
      "learning_rate": 7.331670822942644e-05,
      "loss": 0.0118,
      "step": 147
    },
    {
      "epoch": 0.11094452773613193,
      "grad_norm": 0.0335500919451813,
      "learning_rate": 7.381546134663342e-05,
      "loss": 0.0086,
      "step": 148
    },
    {
      "epoch": 0.11169415292353824,
      "grad_norm": 0.11033605798184272,
      "learning_rate": 7.43142144638404e-05,
      "loss": 0.0478,
      "step": 149
    },
    {
      "epoch": 0.11244377811094453,
      "grad_norm": 0.214427015965264,
      "learning_rate": 7.481296758104739e-05,
      "loss": 0.0353,
      "step": 150
    },
    {
      "epoch": 0.11319340329835083,
      "grad_norm": 0.02568191244810021,
      "learning_rate": 7.531172069825437e-05,
      "loss": 0.0043,
      "step": 151
    },
    {
      "epoch": 0.11394302848575712,
      "grad_norm": 0.09246008396243098,
      "learning_rate": 7.581047381546135e-05,
      "loss": 0.0192,
      "step": 152
    },
    {
      "epoch": 0.11469265367316342,
      "grad_norm": 0.08351131161755415,
      "learning_rate": 7.630922693266833e-05,
      "loss": 0.0277,
      "step": 153
    },
    {
      "epoch": 0.11544227886056972,
      "grad_norm": 0.11694367937691498,
      "learning_rate": 7.680798004987532e-05,
      "loss": 0.0307,
      "step": 154
    },
    {
      "epoch": 0.11619190404797601,
      "grad_norm": 0.09838886537023853,
      "learning_rate": 7.73067331670823e-05,
      "loss": 0.0275,
      "step": 155
    },
    {
      "epoch": 0.11694152923538231,
      "grad_norm": 0.12195844474057815,
      "learning_rate": 7.780548628428928e-05,
      "loss": 0.0242,
      "step": 156
    },
    {
      "epoch": 0.1176911544227886,
      "grad_norm": 0.08775886644144641,
      "learning_rate": 7.830423940149625e-05,
      "loss": 0.0339,
      "step": 157
    },
    {
      "epoch": 0.1184407796101949,
      "grad_norm": 0.0960095760790364,
      "learning_rate": 7.880299251870325e-05,
      "loss": 0.0305,
      "step": 158
    },
    {
      "epoch": 0.1191904047976012,
      "grad_norm": 0.056421431695985014,
      "learning_rate": 7.930174563591023e-05,
      "loss": 0.0167,
      "step": 159
    },
    {
      "epoch": 0.1199400299850075,
      "grad_norm": 0.03780015107287049,
      "learning_rate": 7.980049875311722e-05,
      "loss": 0.0132,
      "step": 160
    },
    {
      "epoch": 0.1206896551724138,
      "grad_norm": 0.06643939369941203,
      "learning_rate": 8.029925187032418e-05,
      "loss": 0.0253,
      "step": 161
    },
    {
      "epoch": 0.12143928035982009,
      "grad_norm": 0.07834551801424494,
      "learning_rate": 8.079800498753118e-05,
      "loss": 0.0243,
      "step": 162
    },
    {
      "epoch": 0.12218890554722639,
      "grad_norm": 0.3428236638579461,
      "learning_rate": 8.129675810473815e-05,
      "loss": 0.0761,
      "step": 163
    },
    {
      "epoch": 0.12293853073463268,
      "grad_norm": 0.25562499213943013,
      "learning_rate": 8.179551122194515e-05,
      "loss": 0.0606,
      "step": 164
    },
    {
      "epoch": 0.12368815592203898,
      "grad_norm": 0.2200505423062946,
      "learning_rate": 8.229426433915212e-05,
      "loss": 0.0413,
      "step": 165
    },
    {
      "epoch": 0.12443778110944528,
      "grad_norm": 0.17323521016059398,
      "learning_rate": 8.27930174563591e-05,
      "loss": 0.065,
      "step": 166
    },
    {
      "epoch": 0.12518740629685157,
      "grad_norm": 0.12386450278132788,
      "learning_rate": 8.329177057356608e-05,
      "loss": 0.0465,
      "step": 167
    },
    {
      "epoch": 0.12593703148425786,
      "grad_norm": 0.06513383821449026,
      "learning_rate": 8.379052369077307e-05,
      "loss": 0.0235,
      "step": 168
    },
    {
      "epoch": 0.12668665667166418,
      "grad_norm": 0.0786581795256923,
      "learning_rate": 8.428927680798005e-05,
      "loss": 0.0169,
      "step": 169
    },
    {
      "epoch": 0.12743628185907047,
      "grad_norm": 0.09510857719264296,
      "learning_rate": 8.478802992518705e-05,
      "loss": 0.0328,
      "step": 170
    },
    {
      "epoch": 0.12818590704647675,
      "grad_norm": 0.06482973979218357,
      "learning_rate": 8.528678304239401e-05,
      "loss": 0.0201,
      "step": 171
    },
    {
      "epoch": 0.12893553223388307,
      "grad_norm": 0.06813364768120031,
      "learning_rate": 8.5785536159601e-05,
      "loss": 0.0223,
      "step": 172
    },
    {
      "epoch": 0.12968515742128936,
      "grad_norm": 0.11763835146175822,
      "learning_rate": 8.628428927680798e-05,
      "loss": 0.0178,
      "step": 173
    },
    {
      "epoch": 0.13043478260869565,
      "grad_norm": 0.07098697087543453,
      "learning_rate": 8.678304239401497e-05,
      "loss": 0.0186,
      "step": 174
    },
    {
      "epoch": 0.13118440779610194,
      "grad_norm": 0.06061152025510322,
      "learning_rate": 8.728179551122195e-05,
      "loss": 0.0271,
      "step": 175
    },
    {
      "epoch": 0.13193403298350825,
      "grad_norm": 0.10686088781724719,
      "learning_rate": 8.778054862842893e-05,
      "loss": 0.0354,
      "step": 176
    },
    {
      "epoch": 0.13268365817091454,
      "grad_norm": 0.04808984107806594,
      "learning_rate": 8.827930174563591e-05,
      "loss": 0.0159,
      "step": 177
    },
    {
      "epoch": 0.13343328335832083,
      "grad_norm": 0.08321421485706715,
      "learning_rate": 8.87780548628429e-05,
      "loss": 0.0113,
      "step": 178
    },
    {
      "epoch": 0.13418290854572715,
      "grad_norm": 0.06279494949837901,
      "learning_rate": 8.927680798004988e-05,
      "loss": 0.0266,
      "step": 179
    },
    {
      "epoch": 0.13493253373313344,
      "grad_norm": 0.047898422343989944,
      "learning_rate": 8.977556109725686e-05,
      "loss": 0.0126,
      "step": 180
    },
    {
      "epoch": 0.13568215892053972,
      "grad_norm": 0.07528619816418361,
      "learning_rate": 9.027431421446384e-05,
      "loss": 0.0166,
      "step": 181
    },
    {
      "epoch": 0.136431784107946,
      "grad_norm": 0.1377475203565619,
      "learning_rate": 9.077306733167083e-05,
      "loss": 0.052,
      "step": 182
    },
    {
      "epoch": 0.13718140929535233,
      "grad_norm": 0.09048176822457338,
      "learning_rate": 9.127182044887781e-05,
      "loss": 0.0285,
      "step": 183
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 0.0809521483858753,
      "learning_rate": 9.177057356608479e-05,
      "loss": 0.0141,
      "step": 184
    },
    {
      "epoch": 0.1386806596701649,
      "grad_norm": 0.09636075780960374,
      "learning_rate": 9.226932668329178e-05,
      "loss": 0.0336,
      "step": 185
    },
    {
      "epoch": 0.13943028485757122,
      "grad_norm": 0.13053296398291478,
      "learning_rate": 9.276807980049876e-05,
      "loss": 0.0397,
      "step": 186
    },
    {
      "epoch": 0.1401799100449775,
      "grad_norm": 0.07863031021793394,
      "learning_rate": 9.326683291770574e-05,
      "loss": 0.0162,
      "step": 187
    },
    {
      "epoch": 0.1409295352323838,
      "grad_norm": 0.1228536793457434,
      "learning_rate": 9.376558603491272e-05,
      "loss": 0.044,
      "step": 188
    },
    {
      "epoch": 0.14167916041979012,
      "grad_norm": 0.09209027498232235,
      "learning_rate": 9.426433915211971e-05,
      "loss": 0.0334,
      "step": 189
    },
    {
      "epoch": 0.1424287856071964,
      "grad_norm": 0.14190775672822614,
      "learning_rate": 9.476309226932669e-05,
      "loss": 0.042,
      "step": 190
    },
    {
      "epoch": 0.1431784107946027,
      "grad_norm": 0.04023312981224354,
      "learning_rate": 9.526184538653367e-05,
      "loss": 0.0142,
      "step": 191
    },
    {
      "epoch": 0.14392803598200898,
      "grad_norm": 0.09583290305937744,
      "learning_rate": 9.576059850374065e-05,
      "loss": 0.0385,
      "step": 192
    },
    {
      "epoch": 0.1446776611694153,
      "grad_norm": 0.21954528231449671,
      "learning_rate": 9.625935162094764e-05,
      "loss": 0.1279,
      "step": 193
    },
    {
      "epoch": 0.1454272863568216,
      "grad_norm": 0.059750365210102005,
      "learning_rate": 9.675810473815462e-05,
      "loss": 0.0237,
      "step": 194
    },
    {
      "epoch": 0.14617691154422788,
      "grad_norm": 0.07514131332440416,
      "learning_rate": 9.725685785536161e-05,
      "loss": 0.0283,
      "step": 195
    },
    {
      "epoch": 0.1469265367316342,
      "grad_norm": 0.29649323772724173,
      "learning_rate": 9.775561097256857e-05,
      "loss": 0.0509,
      "step": 196
    },
    {
      "epoch": 0.14767616191904048,
      "grad_norm": 0.11218042264432664,
      "learning_rate": 9.825436408977557e-05,
      "loss": 0.0416,
      "step": 197
    },
    {
      "epoch": 0.14842578710644677,
      "grad_norm": 0.09965160873863997,
      "learning_rate": 9.875311720698254e-05,
      "loss": 0.0274,
      "step": 198
    },
    {
      "epoch": 0.14917541229385306,
      "grad_norm": 0.03997333482371075,
      "learning_rate": 9.925187032418954e-05,
      "loss": 0.01,
      "step": 199
    },
    {
      "epoch": 0.14992503748125938,
      "grad_norm": 0.07793966938286397,
      "learning_rate": 9.975062344139652e-05,
      "loss": 0.0179,
      "step": 200
    },
    {
      "epoch": 0.15067466266866567,
      "grad_norm": 0.05587033266697241,
      "learning_rate": 0.0001002493765586035,
      "loss": 0.0116,
      "step": 201
    },
    {
      "epoch": 0.15142428785607195,
      "grad_norm": 0.05231986537136687,
      "learning_rate": 0.00010074812967581049,
      "loss": 0.0161,
      "step": 202
    },
    {
      "epoch": 0.15217391304347827,
      "grad_norm": 0.10417060349041267,
      "learning_rate": 0.00010124688279301747,
      "loss": 0.0364,
      "step": 203
    },
    {
      "epoch": 0.15292353823088456,
      "grad_norm": 0.05377861825830805,
      "learning_rate": 0.00010174563591022443,
      "loss": 0.0195,
      "step": 204
    },
    {
      "epoch": 0.15367316341829085,
      "grad_norm": 0.11829981219141997,
      "learning_rate": 0.00010224438902743144,
      "loss": 0.018,
      "step": 205
    },
    {
      "epoch": 0.15442278860569716,
      "grad_norm": 0.07056012168980838,
      "learning_rate": 0.0001027431421446384,
      "loss": 0.0228,
      "step": 206
    },
    {
      "epoch": 0.15517241379310345,
      "grad_norm": 0.40082059596090186,
      "learning_rate": 0.00010324189526184541,
      "loss": 0.0755,
      "step": 207
    },
    {
      "epoch": 0.15592203898050974,
      "grad_norm": 0.05881080321052975,
      "learning_rate": 0.00010374064837905237,
      "loss": 0.0126,
      "step": 208
    },
    {
      "epoch": 0.15667166416791603,
      "grad_norm": 0.04136141605240806,
      "learning_rate": 0.00010423940149625935,
      "loss": 0.0072,
      "step": 209
    },
    {
      "epoch": 0.15742128935532235,
      "grad_norm": 0.1675625733681966,
      "learning_rate": 0.00010473815461346634,
      "loss": 0.0272,
      "step": 210
    },
    {
      "epoch": 0.15817091454272864,
      "grad_norm": 0.15430875264329627,
      "learning_rate": 0.00010523690773067332,
      "loss": 0.0446,
      "step": 211
    },
    {
      "epoch": 0.15892053973013492,
      "grad_norm": 0.09830541143118073,
      "learning_rate": 0.00010573566084788031,
      "loss": 0.0199,
      "step": 212
    },
    {
      "epoch": 0.15967016491754124,
      "grad_norm": 0.10076300729558095,
      "learning_rate": 0.00010623441396508729,
      "loss": 0.0259,
      "step": 213
    },
    {
      "epoch": 0.16041979010494753,
      "grad_norm": 0.34926458260254734,
      "learning_rate": 0.00010673316708229426,
      "loss": 0.0519,
      "step": 214
    },
    {
      "epoch": 0.16116941529235382,
      "grad_norm": 0.06281413567266797,
      "learning_rate": 0.00010723192019950125,
      "loss": 0.013,
      "step": 215
    },
    {
      "epoch": 0.1619190404797601,
      "grad_norm": 0.06672214691330823,
      "learning_rate": 0.00010773067331670823,
      "loss": 0.0149,
      "step": 216
    },
    {
      "epoch": 0.16266866566716642,
      "grad_norm": 0.07641994031425416,
      "learning_rate": 0.00010822942643391522,
      "loss": 0.026,
      "step": 217
    },
    {
      "epoch": 0.1634182908545727,
      "grad_norm": 0.09389436067817795,
      "learning_rate": 0.0001087281795511222,
      "loss": 0.0326,
      "step": 218
    },
    {
      "epoch": 0.164167916041979,
      "grad_norm": 0.0648179280805623,
      "learning_rate": 0.00010922693266832918,
      "loss": 0.0191,
      "step": 219
    },
    {
      "epoch": 0.16491754122938532,
      "grad_norm": 0.0474636418457805,
      "learning_rate": 0.00010972568578553617,
      "loss": 0.0132,
      "step": 220
    },
    {
      "epoch": 0.1656671664167916,
      "grad_norm": 0.06838399746911754,
      "learning_rate": 0.00011022443890274315,
      "loss": 0.01,
      "step": 221
    },
    {
      "epoch": 0.1664167916041979,
      "grad_norm": 0.17561036272482808,
      "learning_rate": 0.00011072319201995014,
      "loss": 0.0374,
      "step": 222
    },
    {
      "epoch": 0.1671664167916042,
      "grad_norm": 0.2569925810493672,
      "learning_rate": 0.00011122194513715711,
      "loss": 0.0516,
      "step": 223
    },
    {
      "epoch": 0.1679160419790105,
      "grad_norm": 0.08357598242934498,
      "learning_rate": 0.00011172069825436409,
      "loss": 0.0291,
      "step": 224
    },
    {
      "epoch": 0.1686656671664168,
      "grad_norm": 0.05291408301494247,
      "learning_rate": 0.00011221945137157108,
      "loss": 0.0201,
      "step": 225
    },
    {
      "epoch": 0.16941529235382308,
      "grad_norm": 0.06497075638907762,
      "learning_rate": 0.00011271820448877806,
      "loss": 0.0124,
      "step": 226
    },
    {
      "epoch": 0.1701649175412294,
      "grad_norm": 0.07406198574289732,
      "learning_rate": 0.00011321695760598505,
      "loss": 0.0237,
      "step": 227
    },
    {
      "epoch": 0.17091454272863568,
      "grad_norm": 0.05439821029505069,
      "learning_rate": 0.00011371571072319203,
      "loss": 0.0211,
      "step": 228
    },
    {
      "epoch": 0.17166416791604197,
      "grad_norm": 0.08111654982385613,
      "learning_rate": 0.00011421446384039899,
      "loss": 0.0285,
      "step": 229
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 0.12398138802704543,
      "learning_rate": 0.000114713216957606,
      "loss": 0.0425,
      "step": 230
    },
    {
      "epoch": 0.17316341829085458,
      "grad_norm": 0.2334451447921542,
      "learning_rate": 0.00011521197007481296,
      "loss": 0.0763,
      "step": 231
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 0.07553003516076283,
      "learning_rate": 0.00011571072319201997,
      "loss": 0.0255,
      "step": 232
    },
    {
      "epoch": 0.17466266866566715,
      "grad_norm": 0.09362267466853937,
      "learning_rate": 0.00011620947630922693,
      "loss": 0.0178,
      "step": 233
    },
    {
      "epoch": 0.17541229385307347,
      "grad_norm": 0.12066656741285244,
      "learning_rate": 0.00011670822942643391,
      "loss": 0.0269,
      "step": 234
    },
    {
      "epoch": 0.17616191904047976,
      "grad_norm": 0.050085879729023425,
      "learning_rate": 0.0001172069825436409,
      "loss": 0.0145,
      "step": 235
    },
    {
      "epoch": 0.17691154422788605,
      "grad_norm": 0.059337800135607535,
      "learning_rate": 0.00011770573566084788,
      "loss": 0.0218,
      "step": 236
    },
    {
      "epoch": 0.17766116941529236,
      "grad_norm": 0.2599879422217277,
      "learning_rate": 0.00011820448877805488,
      "loss": 0.0584,
      "step": 237
    },
    {
      "epoch": 0.17841079460269865,
      "grad_norm": 0.0642722077940849,
      "learning_rate": 0.00011870324189526186,
      "loss": 0.0168,
      "step": 238
    },
    {
      "epoch": 0.17916041979010494,
      "grad_norm": 0.06072818847877561,
      "learning_rate": 0.00011920199501246882,
      "loss": 0.0171,
      "step": 239
    },
    {
      "epoch": 0.17991004497751126,
      "grad_norm": 0.05000902530731167,
      "learning_rate": 0.00011970074812967583,
      "loss": 0.0093,
      "step": 240
    },
    {
      "epoch": 0.18065967016491755,
      "grad_norm": 0.10030548737751505,
      "learning_rate": 0.00012019950124688279,
      "loss": 0.0258,
      "step": 241
    },
    {
      "epoch": 0.18140929535232383,
      "grad_norm": 0.12845354750733506,
      "learning_rate": 0.0001206982543640898,
      "loss": 0.0385,
      "step": 242
    },
    {
      "epoch": 0.18215892053973012,
      "grad_norm": 0.11530518879949096,
      "learning_rate": 0.00012119700748129676,
      "loss": 0.0276,
      "step": 243
    },
    {
      "epoch": 0.18290854572713644,
      "grad_norm": 0.11115377213909522,
      "learning_rate": 0.00012169576059850374,
      "loss": 0.0339,
      "step": 244
    },
    {
      "epoch": 0.18365817091454273,
      "grad_norm": 0.08437903896927137,
      "learning_rate": 0.00012219451371571073,
      "loss": 0.0147,
      "step": 245
    },
    {
      "epoch": 0.18440779610194902,
      "grad_norm": 0.08333540488761688,
      "learning_rate": 0.0001226932668329177,
      "loss": 0.0207,
      "step": 246
    },
    {
      "epoch": 0.18515742128935533,
      "grad_norm": 0.0637836007664769,
      "learning_rate": 0.0001231920199501247,
      "loss": 0.0176,
      "step": 247
    },
    {
      "epoch": 0.18590704647676162,
      "grad_norm": 0.19541845065090152,
      "learning_rate": 0.00012369077306733167,
      "loss": 0.0294,
      "step": 248
    },
    {
      "epoch": 0.1866566716641679,
      "grad_norm": 0.09604182736199869,
      "learning_rate": 0.00012418952618453865,
      "loss": 0.0281,
      "step": 249
    },
    {
      "epoch": 0.1874062968515742,
      "grad_norm": 0.1656571821759516,
      "learning_rate": 0.00012468827930174565,
      "loss": 0.023,
      "step": 250
    },
    {
      "epoch": 0.18815592203898052,
      "grad_norm": 0.07893870621876838,
      "learning_rate": 0.00012518703241895263,
      "loss": 0.0149,
      "step": 251
    },
    {
      "epoch": 0.1889055472263868,
      "grad_norm": 0.08177991639723088,
      "learning_rate": 0.0001256857855361596,
      "loss": 0.0275,
      "step": 252
    },
    {
      "epoch": 0.1896551724137931,
      "grad_norm": 0.09139205799991805,
      "learning_rate": 0.0001261845386533666,
      "loss": 0.0316,
      "step": 253
    },
    {
      "epoch": 0.1904047976011994,
      "grad_norm": 0.13893142926140026,
      "learning_rate": 0.00012668329177057357,
      "loss": 0.0169,
      "step": 254
    },
    {
      "epoch": 0.1911544227886057,
      "grad_norm": 0.12412801138084277,
      "learning_rate": 0.00012718204488778055,
      "loss": 0.039,
      "step": 255
    },
    {
      "epoch": 0.191904047976012,
      "grad_norm": 0.11645301088929443,
      "learning_rate": 0.00012768079800498753,
      "loss": 0.0249,
      "step": 256
    },
    {
      "epoch": 0.1926536731634183,
      "grad_norm": 0.08789313821943712,
      "learning_rate": 0.00012817955112219453,
      "loss": 0.0163,
      "step": 257
    },
    {
      "epoch": 0.1934032983508246,
      "grad_norm": 0.27323828643444004,
      "learning_rate": 0.0001286783042394015,
      "loss": 0.0281,
      "step": 258
    },
    {
      "epoch": 0.19415292353823088,
      "grad_norm": 0.1778613775057513,
      "learning_rate": 0.0001291770573566085,
      "loss": 0.081,
      "step": 259
    },
    {
      "epoch": 0.19490254872563717,
      "grad_norm": 0.1731020084043939,
      "learning_rate": 0.00012967581047381547,
      "loss": 0.0432,
      "step": 260
    },
    {
      "epoch": 0.1956521739130435,
      "grad_norm": 0.12374766749581116,
      "learning_rate": 0.00013017456359102245,
      "loss": 0.0497,
      "step": 261
    },
    {
      "epoch": 0.19640179910044978,
      "grad_norm": 0.12677399108624252,
      "learning_rate": 0.00013067331670822945,
      "loss": 0.0266,
      "step": 262
    },
    {
      "epoch": 0.19715142428785606,
      "grad_norm": 0.18629924644455734,
      "learning_rate": 0.0001311720698254364,
      "loss": 0.0414,
      "step": 263
    },
    {
      "epoch": 0.19790104947526238,
      "grad_norm": 0.0829328057797884,
      "learning_rate": 0.00013167082294264338,
      "loss": 0.0205,
      "step": 264
    },
    {
      "epoch": 0.19865067466266867,
      "grad_norm": 0.08620872083309367,
      "learning_rate": 0.0001321695760598504,
      "loss": 0.0173,
      "step": 265
    },
    {
      "epoch": 0.19940029985007496,
      "grad_norm": 0.09317926758121065,
      "learning_rate": 0.00013266832917705737,
      "loss": 0.0077,
      "step": 266
    },
    {
      "epoch": 0.20014992503748125,
      "grad_norm": 0.08818698775422455,
      "learning_rate": 0.00013316708229426435,
      "loss": 0.0328,
      "step": 267
    },
    {
      "epoch": 0.20089955022488756,
      "grad_norm": 0.08994483021679653,
      "learning_rate": 0.00013366583541147133,
      "loss": 0.0348,
      "step": 268
    },
    {
      "epoch": 0.20164917541229385,
      "grad_norm": 0.09591148983377801,
      "learning_rate": 0.0001341645885286783,
      "loss": 0.0236,
      "step": 269
    },
    {
      "epoch": 0.20239880059970014,
      "grad_norm": 0.11320056714033523,
      "learning_rate": 0.0001346633416458853,
      "loss": 0.0218,
      "step": 270
    },
    {
      "epoch": 0.20314842578710646,
      "grad_norm": 0.5227925892105147,
      "learning_rate": 0.00013516209476309226,
      "loss": 0.0362,
      "step": 271
    },
    {
      "epoch": 0.20389805097451275,
      "grad_norm": 0.03669513253608481,
      "learning_rate": 0.00013566084788029927,
      "loss": 0.0124,
      "step": 272
    },
    {
      "epoch": 0.20464767616191903,
      "grad_norm": 0.07688426864930054,
      "learning_rate": 0.00013615960099750625,
      "loss": 0.0256,
      "step": 273
    },
    {
      "epoch": 0.20539730134932535,
      "grad_norm": 0.08079663825268356,
      "learning_rate": 0.00013665835411471322,
      "loss": 0.0265,
      "step": 274
    },
    {
      "epoch": 0.20614692653673164,
      "grad_norm": 0.19049040538670042,
      "learning_rate": 0.0001371571072319202,
      "loss": 0.0431,
      "step": 275
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 0.19674880634163666,
      "learning_rate": 0.00013765586034912718,
      "loss": 0.0259,
      "step": 276
    },
    {
      "epoch": 0.20764617691154422,
      "grad_norm": 0.2821751560105795,
      "learning_rate": 0.0001381546134663342,
      "loss": 0.057,
      "step": 277
    },
    {
      "epoch": 0.20839580209895053,
      "grad_norm": 0.10359108265397039,
      "learning_rate": 0.00013865336658354117,
      "loss": 0.0228,
      "step": 278
    },
    {
      "epoch": 0.20914542728635682,
      "grad_norm": 0.1806972746554061,
      "learning_rate": 0.00013915211970074812,
      "loss": 0.074,
      "step": 279
    },
    {
      "epoch": 0.2098950524737631,
      "grad_norm": 0.05580356582461233,
      "learning_rate": 0.00013965087281795512,
      "loss": 0.0198,
      "step": 280
    },
    {
      "epoch": 0.21064467766116943,
      "grad_norm": 0.0852610589365401,
      "learning_rate": 0.0001401496259351621,
      "loss": 0.0268,
      "step": 281
    },
    {
      "epoch": 0.21139430284857572,
      "grad_norm": 0.1500943752318949,
      "learning_rate": 0.00014064837905236908,
      "loss": 0.0248,
      "step": 282
    },
    {
      "epoch": 0.212143928035982,
      "grad_norm": 0.34676376255278374,
      "learning_rate": 0.00014114713216957606,
      "loss": 0.0455,
      "step": 283
    },
    {
      "epoch": 0.2128935532233883,
      "grad_norm": 0.07586618072679835,
      "learning_rate": 0.00014164588528678304,
      "loss": 0.0282,
      "step": 284
    },
    {
      "epoch": 0.2136431784107946,
      "grad_norm": 0.0680131198216817,
      "learning_rate": 0.00014214463840399004,
      "loss": 0.0168,
      "step": 285
    },
    {
      "epoch": 0.2143928035982009,
      "grad_norm": 0.15412546383230158,
      "learning_rate": 0.00014264339152119702,
      "loss": 0.0262,
      "step": 286
    },
    {
      "epoch": 0.2151424287856072,
      "grad_norm": 0.32409732407737396,
      "learning_rate": 0.000143142144638404,
      "loss": 0.0431,
      "step": 287
    },
    {
      "epoch": 0.2158920539730135,
      "grad_norm": 0.05689977495195114,
      "learning_rate": 0.00014364089775561098,
      "loss": 0.0098,
      "step": 288
    },
    {
      "epoch": 0.2166416791604198,
      "grad_norm": 0.10219232043950437,
      "learning_rate": 0.00014413965087281796,
      "loss": 0.0446,
      "step": 289
    },
    {
      "epoch": 0.21739130434782608,
      "grad_norm": 0.14753874873503348,
      "learning_rate": 0.00014463840399002494,
      "loss": 0.0338,
      "step": 290
    },
    {
      "epoch": 0.2181409295352324,
      "grad_norm": 0.09688329615506609,
      "learning_rate": 0.00014513715710723192,
      "loss": 0.0317,
      "step": 291
    },
    {
      "epoch": 0.21889055472263869,
      "grad_norm": 0.37462001623399555,
      "learning_rate": 0.00014563591022443892,
      "loss": 0.0488,
      "step": 292
    },
    {
      "epoch": 0.21964017991004497,
      "grad_norm": 0.1744644075575525,
      "learning_rate": 0.0001461346633416459,
      "loss": 0.0238,
      "step": 293
    },
    {
      "epoch": 0.22038980509745126,
      "grad_norm": 0.12911989638647245,
      "learning_rate": 0.00014663341645885288,
      "loss": 0.0262,
      "step": 294
    },
    {
      "epoch": 0.22113943028485758,
      "grad_norm": 0.15596073376994574,
      "learning_rate": 0.00014713216957605986,
      "loss": 0.038,
      "step": 295
    },
    {
      "epoch": 0.22188905547226387,
      "grad_norm": 0.20990059849544168,
      "learning_rate": 0.00014763092269326684,
      "loss": 0.0497,
      "step": 296
    },
    {
      "epoch": 0.22263868065967016,
      "grad_norm": 0.185998926544591,
      "learning_rate": 0.00014812967581047384,
      "loss": 0.0438,
      "step": 297
    },
    {
      "epoch": 0.22338830584707647,
      "grad_norm": 0.1472947653395365,
      "learning_rate": 0.0001486284289276808,
      "loss": 0.0482,
      "step": 298
    },
    {
      "epoch": 0.22413793103448276,
      "grad_norm": 0.11766072318652662,
      "learning_rate": 0.00014912718204488777,
      "loss": 0.0341,
      "step": 299
    },
    {
      "epoch": 0.22488755622188905,
      "grad_norm": 0.16151375968719706,
      "learning_rate": 0.00014962593516209478,
      "loss": 0.0235,
      "step": 300
    },
    {
      "epoch": 0.22563718140929534,
      "grad_norm": 0.13978947872038194,
      "learning_rate": 0.00015012468827930176,
      "loss": 0.0393,
      "step": 301
    },
    {
      "epoch": 0.22638680659670166,
      "grad_norm": 0.21401300347047775,
      "learning_rate": 0.00015062344139650874,
      "loss": 0.0178,
      "step": 302
    },
    {
      "epoch": 0.22713643178410794,
      "grad_norm": 0.24801232446587168,
      "learning_rate": 0.00015112219451371572,
      "loss": 0.0407,
      "step": 303
    },
    {
      "epoch": 0.22788605697151423,
      "grad_norm": 0.058632535114882166,
      "learning_rate": 0.0001516209476309227,
      "loss": 0.0206,
      "step": 304
    },
    {
      "epoch": 0.22863568215892055,
      "grad_norm": 0.0840227180439461,
      "learning_rate": 0.0001521197007481297,
      "loss": 0.0287,
      "step": 305
    },
    {
      "epoch": 0.22938530734632684,
      "grad_norm": 0.08132083789297478,
      "learning_rate": 0.00015261845386533665,
      "loss": 0.026,
      "step": 306
    },
    {
      "epoch": 0.23013493253373313,
      "grad_norm": 0.10781849659948611,
      "learning_rate": 0.00015311720698254363,
      "loss": 0.0149,
      "step": 307
    },
    {
      "epoch": 0.23088455772113944,
      "grad_norm": 0.15794412633172517,
      "learning_rate": 0.00015361596009975064,
      "loss": 0.0536,
      "step": 308
    },
    {
      "epoch": 0.23163418290854573,
      "grad_norm": 0.11903463083343918,
      "learning_rate": 0.00015411471321695762,
      "loss": 0.0261,
      "step": 309
    },
    {
      "epoch": 0.23238380809595202,
      "grad_norm": 0.08270710290559068,
      "learning_rate": 0.0001546134663341646,
      "loss": 0.0177,
      "step": 310
    },
    {
      "epoch": 0.2331334332833583,
      "grad_norm": 0.12039252632311984,
      "learning_rate": 0.00015511221945137157,
      "loss": 0.0326,
      "step": 311
    },
    {
      "epoch": 0.23388305847076463,
      "grad_norm": 0.09035121559893523,
      "learning_rate": 0.00015561097256857855,
      "loss": 0.0218,
      "step": 312
    },
    {
      "epoch": 0.23463268365817092,
      "grad_norm": 0.2608168886436339,
      "learning_rate": 0.00015610972568578556,
      "loss": 0.0298,
      "step": 313
    },
    {
      "epoch": 0.2353823088455772,
      "grad_norm": 0.08183654251092078,
      "learning_rate": 0.0001566084788029925,
      "loss": 0.0149,
      "step": 314
    },
    {
      "epoch": 0.23613193403298352,
      "grad_norm": 0.19660962857791126,
      "learning_rate": 0.00015710723192019951,
      "loss": 0.0485,
      "step": 315
    },
    {
      "epoch": 0.2368815592203898,
      "grad_norm": 0.37386494103393336,
      "learning_rate": 0.0001576059850374065,
      "loss": 0.0572,
      "step": 316
    },
    {
      "epoch": 0.2376311844077961,
      "grad_norm": 0.1704419836308577,
      "learning_rate": 0.00015810473815461347,
      "loss": 0.0315,
      "step": 317
    },
    {
      "epoch": 0.2383808095952024,
      "grad_norm": 0.13091652195928158,
      "learning_rate": 0.00015860349127182045,
      "loss": 0.0388,
      "step": 318
    },
    {
      "epoch": 0.2391304347826087,
      "grad_norm": 0.14223677525604458,
      "learning_rate": 0.00015910224438902743,
      "loss": 0.0253,
      "step": 319
    },
    {
      "epoch": 0.239880059970015,
      "grad_norm": 0.17823510302750656,
      "learning_rate": 0.00015960099750623444,
      "loss": 0.0207,
      "step": 320
    },
    {
      "epoch": 0.24062968515742128,
      "grad_norm": 0.11745599847225957,
      "learning_rate": 0.00016009975062344141,
      "loss": 0.0421,
      "step": 321
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 0.2474114622826013,
      "learning_rate": 0.00016059850374064837,
      "loss": 0.0256,
      "step": 322
    },
    {
      "epoch": 0.24212893553223389,
      "grad_norm": 0.17085562013724867,
      "learning_rate": 0.00016109725685785537,
      "loss": 0.0353,
      "step": 323
    },
    {
      "epoch": 0.24287856071964017,
      "grad_norm": 0.2372708323711952,
      "learning_rate": 0.00016159600997506235,
      "loss": 0.0571,
      "step": 324
    },
    {
      "epoch": 0.2436281859070465,
      "grad_norm": 0.2519151153575236,
      "learning_rate": 0.00016209476309226933,
      "loss": 0.06,
      "step": 325
    },
    {
      "epoch": 0.24437781109445278,
      "grad_norm": 0.24246496343523052,
      "learning_rate": 0.0001625935162094763,
      "loss": 0.0518,
      "step": 326
    },
    {
      "epoch": 0.24512743628185907,
      "grad_norm": 0.10872569583987493,
      "learning_rate": 0.0001630922693266833,
      "loss": 0.02,
      "step": 327
    },
    {
      "epoch": 0.24587706146926536,
      "grad_norm": 0.25513609548593424,
      "learning_rate": 0.0001635910224438903,
      "loss": 0.0482,
      "step": 328
    },
    {
      "epoch": 0.24662668665667167,
      "grad_norm": 0.11260345413307601,
      "learning_rate": 0.00016408977556109727,
      "loss": 0.0212,
      "step": 329
    },
    {
      "epoch": 0.24737631184407796,
      "grad_norm": 0.49079259160521377,
      "learning_rate": 0.00016458852867830425,
      "loss": 0.053,
      "step": 330
    },
    {
      "epoch": 0.24812593703148425,
      "grad_norm": 0.11154822451105055,
      "learning_rate": 0.00016508728179551123,
      "loss": 0.0241,
      "step": 331
    },
    {
      "epoch": 0.24887556221889057,
      "grad_norm": 0.09571826846514553,
      "learning_rate": 0.0001655860349127182,
      "loss": 0.0252,
      "step": 332
    },
    {
      "epoch": 0.24962518740629686,
      "grad_norm": 0.0777734898245894,
      "learning_rate": 0.00016608478802992519,
      "loss": 0.027,
      "step": 333
    },
    {
      "epoch": 0.25037481259370314,
      "grad_norm": 0.14307189815218685,
      "learning_rate": 0.00016658354114713216,
      "loss": 0.0383,
      "step": 334
    },
    {
      "epoch": 0.25112443778110943,
      "grad_norm": 0.13735355901692434,
      "learning_rate": 0.00016708229426433917,
      "loss": 0.0303,
      "step": 335
    },
    {
      "epoch": 0.2518740629685157,
      "grad_norm": 0.16726996989241977,
      "learning_rate": 0.00016758104738154615,
      "loss": 0.03,
      "step": 336
    },
    {
      "epoch": 0.25262368815592207,
      "grad_norm": 0.16945012137988621,
      "learning_rate": 0.00016807980049875313,
      "loss": 0.0406,
      "step": 337
    },
    {
      "epoch": 0.25337331334332835,
      "grad_norm": 0.11530437802632285,
      "learning_rate": 0.0001685785536159601,
      "loss": 0.0225,
      "step": 338
    },
    {
      "epoch": 0.25412293853073464,
      "grad_norm": 0.09960125210392978,
      "learning_rate": 0.00016907730673316709,
      "loss": 0.0209,
      "step": 339
    },
    {
      "epoch": 0.25487256371814093,
      "grad_norm": 0.13443406892444226,
      "learning_rate": 0.0001695760598503741,
      "loss": 0.0492,
      "step": 340
    },
    {
      "epoch": 0.2556221889055472,
      "grad_norm": 0.10886582012917131,
      "learning_rate": 0.00017007481296758104,
      "loss": 0.0249,
      "step": 341
    },
    {
      "epoch": 0.2563718140929535,
      "grad_norm": 0.06084801968020566,
      "learning_rate": 0.00017057356608478802,
      "loss": 0.0144,
      "step": 342
    },
    {
      "epoch": 0.2571214392803598,
      "grad_norm": 0.4190379355567448,
      "learning_rate": 0.00017107231920199503,
      "loss": 0.0749,
      "step": 343
    },
    {
      "epoch": 0.25787106446776614,
      "grad_norm": 0.11112704571061172,
      "learning_rate": 0.000171571072319202,
      "loss": 0.0206,
      "step": 344
    },
    {
      "epoch": 0.25862068965517243,
      "grad_norm": 0.07059400346190974,
      "learning_rate": 0.00017206982543640898,
      "loss": 0.0186,
      "step": 345
    },
    {
      "epoch": 0.2593703148425787,
      "grad_norm": 0.32755669501610246,
      "learning_rate": 0.00017256857855361596,
      "loss": 0.0417,
      "step": 346
    },
    {
      "epoch": 0.260119940029985,
      "grad_norm": 0.1983041340897065,
      "learning_rate": 0.00017306733167082294,
      "loss": 0.0405,
      "step": 347
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 0.26450010843425603,
      "learning_rate": 0.00017356608478802995,
      "loss": 0.0254,
      "step": 348
    },
    {
      "epoch": 0.2616191904047976,
      "grad_norm": 0.05678132030515059,
      "learning_rate": 0.0001740648379052369,
      "loss": 0.0103,
      "step": 349
    },
    {
      "epoch": 0.2623688155922039,
      "grad_norm": 0.07668079412848387,
      "learning_rate": 0.0001745635910224439,
      "loss": 0.0162,
      "step": 350
    },
    {
      "epoch": 0.2631184407796102,
      "grad_norm": 0.22616036460660832,
      "learning_rate": 0.00017506234413965088,
      "loss": 0.0361,
      "step": 351
    },
    {
      "epoch": 0.2638680659670165,
      "grad_norm": 0.11606802219772738,
      "learning_rate": 0.00017556109725685786,
      "loss": 0.0118,
      "step": 352
    },
    {
      "epoch": 0.2646176911544228,
      "grad_norm": 0.44709462961570234,
      "learning_rate": 0.00017605985037406484,
      "loss": 0.067,
      "step": 353
    },
    {
      "epoch": 0.2653673163418291,
      "grad_norm": 0.20489244859924258,
      "learning_rate": 0.00017655860349127182,
      "loss": 0.0472,
      "step": 354
    },
    {
      "epoch": 0.2661169415292354,
      "grad_norm": 0.19112165467765932,
      "learning_rate": 0.00017705735660847883,
      "loss": 0.0503,
      "step": 355
    },
    {
      "epoch": 0.26686656671664166,
      "grad_norm": 0.09843918120704351,
      "learning_rate": 0.0001775561097256858,
      "loss": 0.0241,
      "step": 356
    },
    {
      "epoch": 0.26761619190404795,
      "grad_norm": 0.23798628300883096,
      "learning_rate": 0.00017805486284289276,
      "loss": 0.0433,
      "step": 357
    },
    {
      "epoch": 0.2683658170914543,
      "grad_norm": 0.09665589303199588,
      "learning_rate": 0.00017855361596009976,
      "loss": 0.0174,
      "step": 358
    },
    {
      "epoch": 0.2691154422788606,
      "grad_norm": 0.17403163590794157,
      "learning_rate": 0.00017905236907730674,
      "loss": 0.0327,
      "step": 359
    },
    {
      "epoch": 0.2698650674662669,
      "grad_norm": 0.13658076391511467,
      "learning_rate": 0.00017955112219451372,
      "loss": 0.0341,
      "step": 360
    },
    {
      "epoch": 0.27061469265367316,
      "grad_norm": 0.23004996818420348,
      "learning_rate": 0.0001800498753117207,
      "loss": 0.0385,
      "step": 361
    },
    {
      "epoch": 0.27136431784107945,
      "grad_norm": 0.2235475448370521,
      "learning_rate": 0.00018054862842892768,
      "loss": 0.0449,
      "step": 362
    },
    {
      "epoch": 0.27211394302848574,
      "grad_norm": 0.14802680170837254,
      "learning_rate": 0.00018104738154613468,
      "loss": 0.032,
      "step": 363
    },
    {
      "epoch": 0.272863568215892,
      "grad_norm": 0.2726885316166056,
      "learning_rate": 0.00018154613466334166,
      "loss": 0.0186,
      "step": 364
    },
    {
      "epoch": 0.27361319340329837,
      "grad_norm": 0.18956540775235434,
      "learning_rate": 0.00018204488778054864,
      "loss": 0.0293,
      "step": 365
    },
    {
      "epoch": 0.27436281859070466,
      "grad_norm": 0.15917936571093227,
      "learning_rate": 0.00018254364089775562,
      "loss": 0.0363,
      "step": 366
    },
    {
      "epoch": 0.27511244377811095,
      "grad_norm": 0.14578322218681494,
      "learning_rate": 0.0001830423940149626,
      "loss": 0.0314,
      "step": 367
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 0.10269728307226299,
      "learning_rate": 0.00018354114713216958,
      "loss": 0.0129,
      "step": 368
    },
    {
      "epoch": 0.2766116941529235,
      "grad_norm": 0.21141699577344752,
      "learning_rate": 0.00018403990024937656,
      "loss": 0.0537,
      "step": 369
    },
    {
      "epoch": 0.2773613193403298,
      "grad_norm": 0.15206079985424664,
      "learning_rate": 0.00018453865336658356,
      "loss": 0.031,
      "step": 370
    },
    {
      "epoch": 0.27811094452773616,
      "grad_norm": 0.13011557776643995,
      "learning_rate": 0.00018503740648379054,
      "loss": 0.0224,
      "step": 371
    },
    {
      "epoch": 0.27886056971514245,
      "grad_norm": 0.269149451025218,
      "learning_rate": 0.00018553615960099752,
      "loss": 0.0335,
      "step": 372
    },
    {
      "epoch": 0.27961019490254874,
      "grad_norm": 0.16896565857953327,
      "learning_rate": 0.0001860349127182045,
      "loss": 0.0304,
      "step": 373
    },
    {
      "epoch": 0.280359820089955,
      "grad_norm": 0.07219260932887692,
      "learning_rate": 0.00018653366583541148,
      "loss": 0.012,
      "step": 374
    },
    {
      "epoch": 0.2811094452773613,
      "grad_norm": 0.09981374137995903,
      "learning_rate": 0.00018703241895261848,
      "loss": 0.0201,
      "step": 375
    },
    {
      "epoch": 0.2818590704647676,
      "grad_norm": 0.11898845440731402,
      "learning_rate": 0.00018753117206982543,
      "loss": 0.0337,
      "step": 376
    },
    {
      "epoch": 0.2826086956521739,
      "grad_norm": 0.09794209512689632,
      "learning_rate": 0.0001880299251870324,
      "loss": 0.0152,
      "step": 377
    },
    {
      "epoch": 0.28335832083958024,
      "grad_norm": 0.16926227090380522,
      "learning_rate": 0.00018852867830423942,
      "loss": 0.0365,
      "step": 378
    },
    {
      "epoch": 0.2841079460269865,
      "grad_norm": 0.1336141976814021,
      "learning_rate": 0.0001890274314214464,
      "loss": 0.0183,
      "step": 379
    },
    {
      "epoch": 0.2848575712143928,
      "grad_norm": 0.17357395030789294,
      "learning_rate": 0.00018952618453865338,
      "loss": 0.0413,
      "step": 380
    },
    {
      "epoch": 0.2856071964017991,
      "grad_norm": 0.1171694991646048,
      "learning_rate": 0.00019002493765586035,
      "loss": 0.0282,
      "step": 381
    },
    {
      "epoch": 0.2863568215892054,
      "grad_norm": 0.11853021340241342,
      "learning_rate": 0.00019052369077306733,
      "loss": 0.026,
      "step": 382
    },
    {
      "epoch": 0.2871064467766117,
      "grad_norm": 0.12803104542393728,
      "learning_rate": 0.00019102244389027434,
      "loss": 0.0399,
      "step": 383
    },
    {
      "epoch": 0.28785607196401797,
      "grad_norm": 0.19010760869232313,
      "learning_rate": 0.0001915211970074813,
      "loss": 0.0583,
      "step": 384
    },
    {
      "epoch": 0.2886056971514243,
      "grad_norm": 0.1918972133083329,
      "learning_rate": 0.0001920199501246883,
      "loss": 0.0257,
      "step": 385
    },
    {
      "epoch": 0.2893553223388306,
      "grad_norm": 0.38132752615347837,
      "learning_rate": 0.00019251870324189527,
      "loss": 0.0459,
      "step": 386
    },
    {
      "epoch": 0.2901049475262369,
      "grad_norm": 0.21823951622302395,
      "learning_rate": 0.00019301745635910225,
      "loss": 0.0396,
      "step": 387
    },
    {
      "epoch": 0.2908545727136432,
      "grad_norm": 0.0934811566055218,
      "learning_rate": 0.00019351620947630923,
      "loss": 0.0213,
      "step": 388
    },
    {
      "epoch": 0.29160419790104947,
      "grad_norm": 0.09815902295103762,
      "learning_rate": 0.0001940149625935162,
      "loss": 0.0191,
      "step": 389
    },
    {
      "epoch": 0.29235382308845576,
      "grad_norm": 0.14907156908477912,
      "learning_rate": 0.00019451371571072322,
      "loss": 0.0263,
      "step": 390
    },
    {
      "epoch": 0.29310344827586204,
      "grad_norm": 0.19453601336927534,
      "learning_rate": 0.0001950124688279302,
      "loss": 0.0427,
      "step": 391
    },
    {
      "epoch": 0.2938530734632684,
      "grad_norm": 0.13879553197564173,
      "learning_rate": 0.00019551122194513715,
      "loss": 0.0448,
      "step": 392
    },
    {
      "epoch": 0.2946026986506747,
      "grad_norm": 0.16975459035194854,
      "learning_rate": 0.00019600997506234415,
      "loss": 0.0359,
      "step": 393
    },
    {
      "epoch": 0.29535232383808097,
      "grad_norm": 0.177277140657558,
      "learning_rate": 0.00019650872817955113,
      "loss": 0.0395,
      "step": 394
    },
    {
      "epoch": 0.29610194902548725,
      "grad_norm": 0.11373173330367291,
      "learning_rate": 0.0001970074812967581,
      "loss": 0.0356,
      "step": 395
    },
    {
      "epoch": 0.29685157421289354,
      "grad_norm": 0.15222992950517256,
      "learning_rate": 0.0001975062344139651,
      "loss": 0.0428,
      "step": 396
    },
    {
      "epoch": 0.29760119940029983,
      "grad_norm": 0.09436496677890617,
      "learning_rate": 0.00019800498753117207,
      "loss": 0.0159,
      "step": 397
    },
    {
      "epoch": 0.2983508245877061,
      "grad_norm": 0.1507478045803151,
      "learning_rate": 0.00019850374064837907,
      "loss": 0.0267,
      "step": 398
    },
    {
      "epoch": 0.29910044977511246,
      "grad_norm": 0.051785827436256274,
      "learning_rate": 0.00019900249376558605,
      "loss": 0.0088,
      "step": 399
    },
    {
      "epoch": 0.29985007496251875,
      "grad_norm": 0.18717721855259775,
      "learning_rate": 0.00019950124688279303,
      "loss": 0.0559,
      "step": 400
    },
    {
      "epoch": 0.30059970014992504,
      "grad_norm": 0.14425330741006745,
      "learning_rate": 0.0002,
      "loss": 0.0311,
      "step": 401
    },
    {
      "epoch": 0.30134932533733133,
      "grad_norm": 0.13207723504220212,
      "learning_rate": 0.00019999996194397014,
      "loss": 0.0215,
      "step": 402
    },
    {
      "epoch": 0.3020989505247376,
      "grad_norm": 0.059915434172776075,
      "learning_rate": 0.00019999984777590944,
      "loss": 0.0119,
      "step": 403
    },
    {
      "epoch": 0.3028485757121439,
      "grad_norm": 0.3938617437192051,
      "learning_rate": 0.00019999965749590486,
      "loss": 0.0303,
      "step": 404
    },
    {
      "epoch": 0.30359820089955025,
      "grad_norm": 0.17694572717829843,
      "learning_rate": 0.0001999993911041012,
      "loss": 0.0436,
      "step": 405
    },
    {
      "epoch": 0.30434782608695654,
      "grad_norm": 0.18667444081506102,
      "learning_rate": 0.0001999990486007012,
      "loss": 0.034,
      "step": 406
    },
    {
      "epoch": 0.30509745127436283,
      "grad_norm": 0.1792430745481953,
      "learning_rate": 0.00019999862998596558,
      "loss": 0.0574,
      "step": 407
    },
    {
      "epoch": 0.3058470764617691,
      "grad_norm": 0.15142125119749755,
      "learning_rate": 0.00019999813526021295,
      "loss": 0.0416,
      "step": 408
    },
    {
      "epoch": 0.3065967016491754,
      "grad_norm": 0.15352552956144938,
      "learning_rate": 0.00019999756442381982,
      "loss": 0.0289,
      "step": 409
    },
    {
      "epoch": 0.3073463268365817,
      "grad_norm": 0.1459033018919069,
      "learning_rate": 0.00019999691747722066,
      "loss": 0.0188,
      "step": 410
    },
    {
      "epoch": 0.308095952023988,
      "grad_norm": 0.16054037586365844,
      "learning_rate": 0.00019999619442090798,
      "loss": 0.039,
      "step": 411
    },
    {
      "epoch": 0.30884557721139433,
      "grad_norm": 0.28904351396420497,
      "learning_rate": 0.000199995395255432,
      "loss": 0.0631,
      "step": 412
    },
    {
      "epoch": 0.3095952023988006,
      "grad_norm": 0.368232762601163,
      "learning_rate": 0.000199994519981401,
      "loss": 0.0592,
      "step": 413
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 0.11645861006540552,
      "learning_rate": 0.00019999356859948123,
      "loss": 0.0216,
      "step": 414
    },
    {
      "epoch": 0.3110944527736132,
      "grad_norm": 0.13914821020028115,
      "learning_rate": 0.00019999254111039677,
      "loss": 0.0329,
      "step": 415
    },
    {
      "epoch": 0.3118440779610195,
      "grad_norm": 0.1054919349362133,
      "learning_rate": 0.00019999143751492962,
      "loss": 0.0185,
      "step": 416
    },
    {
      "epoch": 0.31259370314842577,
      "grad_norm": 0.12235988385680502,
      "learning_rate": 0.0001999902578139198,
      "loss": 0.0253,
      "step": 417
    },
    {
      "epoch": 0.31334332833583206,
      "grad_norm": 0.23220583217832894,
      "learning_rate": 0.0001999890020082652,
      "loss": 0.0966,
      "step": 418
    },
    {
      "epoch": 0.3140929535232384,
      "grad_norm": 0.15835605674777165,
      "learning_rate": 0.00019998767009892163,
      "loss": 0.0265,
      "step": 419
    },
    {
      "epoch": 0.3148425787106447,
      "grad_norm": 0.18460817393247528,
      "learning_rate": 0.00019998626208690283,
      "loss": 0.0553,
      "step": 420
    },
    {
      "epoch": 0.315592203898051,
      "grad_norm": 0.15905910975137832,
      "learning_rate": 0.00019998477797328047,
      "loss": 0.0336,
      "step": 421
    },
    {
      "epoch": 0.31634182908545727,
      "grad_norm": 0.10939651490515165,
      "learning_rate": 0.00019998321775918415,
      "loss": 0.018,
      "step": 422
    },
    {
      "epoch": 0.31709145427286356,
      "grad_norm": 0.12177572598606647,
      "learning_rate": 0.0001999815814458014,
      "loss": 0.0185,
      "step": 423
    },
    {
      "epoch": 0.31784107946026985,
      "grad_norm": 0.11831550376041258,
      "learning_rate": 0.00019997986903437759,
      "loss": 0.032,
      "step": 424
    },
    {
      "epoch": 0.31859070464767614,
      "grad_norm": 0.06514033143501995,
      "learning_rate": 0.00019997808052621608,
      "loss": 0.0128,
      "step": 425
    },
    {
      "epoch": 0.3193403298350825,
      "grad_norm": 0.19399927400695385,
      "learning_rate": 0.00019997621592267824,
      "loss": 0.043,
      "step": 426
    },
    {
      "epoch": 0.32008995502248877,
      "grad_norm": 0.11925039389623487,
      "learning_rate": 0.00019997427522518315,
      "loss": 0.0157,
      "step": 427
    },
    {
      "epoch": 0.32083958020989506,
      "grad_norm": 0.07784698086044195,
      "learning_rate": 0.00019997225843520794,
      "loss": 0.0133,
      "step": 428
    },
    {
      "epoch": 0.32158920539730135,
      "grad_norm": 0.10863162782804062,
      "learning_rate": 0.00019997016555428764,
      "loss": 0.0147,
      "step": 429
    },
    {
      "epoch": 0.32233883058470764,
      "grad_norm": 0.12068391287985351,
      "learning_rate": 0.00019996799658401519,
      "loss": 0.0197,
      "step": 430
    },
    {
      "epoch": 0.3230884557721139,
      "grad_norm": 0.23114556727231364,
      "learning_rate": 0.00019996575152604146,
      "loss": 0.0529,
      "step": 431
    },
    {
      "epoch": 0.3238380809595202,
      "grad_norm": 0.32097628810594436,
      "learning_rate": 0.00019996343038207514,
      "loss": 0.0669,
      "step": 432
    },
    {
      "epoch": 0.32458770614692656,
      "grad_norm": 0.12267343492743275,
      "learning_rate": 0.00019996103315388295,
      "loss": 0.0338,
      "step": 433
    },
    {
      "epoch": 0.32533733133433285,
      "grad_norm": 0.19841665026418323,
      "learning_rate": 0.0001999585598432895,
      "loss": 0.0261,
      "step": 434
    },
    {
      "epoch": 0.32608695652173914,
      "grad_norm": 0.19526396807296756,
      "learning_rate": 0.0001999560104521772,
      "loss": 0.0395,
      "step": 435
    },
    {
      "epoch": 0.3268365817091454,
      "grad_norm": 0.19094290559218405,
      "learning_rate": 0.00019995338498248654,
      "loss": 0.0263,
      "step": 436
    },
    {
      "epoch": 0.3275862068965517,
      "grad_norm": 0.10315023911899296,
      "learning_rate": 0.0001999506834362157,
      "loss": 0.0286,
      "step": 437
    },
    {
      "epoch": 0.328335832083958,
      "grad_norm": 0.1513368969371964,
      "learning_rate": 0.00019994790581542095,
      "loss": 0.0276,
      "step": 438
    },
    {
      "epoch": 0.32908545727136435,
      "grad_norm": 0.3896406875829373,
      "learning_rate": 0.0001999450521222164,
      "loss": 0.0438,
      "step": 439
    },
    {
      "epoch": 0.32983508245877063,
      "grad_norm": 0.18728955706645808,
      "learning_rate": 0.00019994212235877406,
      "loss": 0.0313,
      "step": 440
    },
    {
      "epoch": 0.3305847076461769,
      "grad_norm": 0.1763808296207911,
      "learning_rate": 0.00019993911652732384,
      "loss": 0.042,
      "step": 441
    },
    {
      "epoch": 0.3313343328335832,
      "grad_norm": 0.3486237934185716,
      "learning_rate": 0.00019993603463015349,
      "loss": 0.0344,
      "step": 442
    },
    {
      "epoch": 0.3320839580209895,
      "grad_norm": 0.11358739359187257,
      "learning_rate": 0.00019993287666960876,
      "loss": 0.0287,
      "step": 443
    },
    {
      "epoch": 0.3328335832083958,
      "grad_norm": 0.2771651376286589,
      "learning_rate": 0.00019992964264809316,
      "loss": 0.0539,
      "step": 444
    },
    {
      "epoch": 0.3335832083958021,
      "grad_norm": 0.15951465547214178,
      "learning_rate": 0.0001999263325680683,
      "loss": 0.0351,
      "step": 445
    },
    {
      "epoch": 0.3343328335832084,
      "grad_norm": 0.22717670366337234,
      "learning_rate": 0.00019992294643205342,
      "loss": 0.0559,
      "step": 446
    },
    {
      "epoch": 0.3350824587706147,
      "grad_norm": 0.11730729763880723,
      "learning_rate": 0.00019991948424262582,
      "loss": 0.025,
      "step": 447
    },
    {
      "epoch": 0.335832083958021,
      "grad_norm": 0.1488401866535149,
      "learning_rate": 0.00019991594600242069,
      "loss": 0.0554,
      "step": 448
    },
    {
      "epoch": 0.3365817091454273,
      "grad_norm": 0.22698278428915133,
      "learning_rate": 0.00019991233171413102,
      "loss": 0.0268,
      "step": 449
    },
    {
      "epoch": 0.3373313343328336,
      "grad_norm": 0.256587121294311,
      "learning_rate": 0.0001999086413805077,
      "loss": 0.0545,
      "step": 450
    },
    {
      "epoch": 0.33808095952023987,
      "grad_norm": 0.190566506128757,
      "learning_rate": 0.0001999048750043596,
      "loss": 0.0383,
      "step": 451
    },
    {
      "epoch": 0.33883058470764615,
      "grad_norm": 0.20383160433961658,
      "learning_rate": 0.00019990103258855327,
      "loss": 0.0373,
      "step": 452
    },
    {
      "epoch": 0.3395802098950525,
      "grad_norm": 0.05812210389585167,
      "learning_rate": 0.00019989711413601332,
      "loss": 0.0127,
      "step": 453
    },
    {
      "epoch": 0.3403298350824588,
      "grad_norm": 0.15721480837414703,
      "learning_rate": 0.00019989311964972216,
      "loss": 0.0327,
      "step": 454
    },
    {
      "epoch": 0.3410794602698651,
      "grad_norm": 0.11121198405053008,
      "learning_rate": 0.00019988904913272005,
      "loss": 0.0308,
      "step": 455
    },
    {
      "epoch": 0.34182908545727136,
      "grad_norm": 0.2792903863779934,
      "learning_rate": 0.0001998849025881052,
      "loss": 0.0208,
      "step": 456
    },
    {
      "epoch": 0.34257871064467765,
      "grad_norm": 0.32716321508579893,
      "learning_rate": 0.00019988068001903356,
      "loss": 0.0435,
      "step": 457
    },
    {
      "epoch": 0.34332833583208394,
      "grad_norm": 0.20203904519890029,
      "learning_rate": 0.00019987638142871908,
      "loss": 0.0235,
      "step": 458
    },
    {
      "epoch": 0.34407796101949023,
      "grad_norm": 0.11059483818522871,
      "learning_rate": 0.00019987200682043346,
      "loss": 0.0187,
      "step": 459
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 0.09751448421919101,
      "learning_rate": 0.0001998675561975063,
      "loss": 0.0103,
      "step": 460
    },
    {
      "epoch": 0.34557721139430286,
      "grad_norm": 0.1532154010841786,
      "learning_rate": 0.00019986302956332512,
      "loss": 0.0112,
      "step": 461
    },
    {
      "epoch": 0.34632683658170915,
      "grad_norm": 0.2078059436395786,
      "learning_rate": 0.00019985842692133517,
      "loss": 0.0379,
      "step": 462
    },
    {
      "epoch": 0.34707646176911544,
      "grad_norm": 0.24068443221087116,
      "learning_rate": 0.00019985374827503964,
      "loss": 0.044,
      "step": 463
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 0.34075693933551066,
      "learning_rate": 0.00019984899362799957,
      "loss": 0.0638,
      "step": 464
    },
    {
      "epoch": 0.348575712143928,
      "grad_norm": 0.10385908067891564,
      "learning_rate": 0.00019984416298383378,
      "loss": 0.0358,
      "step": 465
    },
    {
      "epoch": 0.3493253373313343,
      "grad_norm": 0.14788340069318578,
      "learning_rate": 0.00019983925634621893,
      "loss": 0.0203,
      "step": 466
    },
    {
      "epoch": 0.35007496251874065,
      "grad_norm": 0.5654716055817488,
      "learning_rate": 0.0001998342737188897,
      "loss": 0.0884,
      "step": 467
    },
    {
      "epoch": 0.35082458770614694,
      "grad_norm": 0.2093358211842703,
      "learning_rate": 0.00019982921510563838,
      "loss": 0.0574,
      "step": 468
    },
    {
      "epoch": 0.35157421289355323,
      "grad_norm": 0.19003237135335357,
      "learning_rate": 0.00019982408051031513,
      "loss": 0.0291,
      "step": 469
    },
    {
      "epoch": 0.3523238380809595,
      "grad_norm": 0.17192643284625606,
      "learning_rate": 0.00019981886993682813,
      "loss": 0.023,
      "step": 470
    },
    {
      "epoch": 0.3530734632683658,
      "grad_norm": 0.14062304972448308,
      "learning_rate": 0.00019981358338914317,
      "loss": 0.0367,
      "step": 471
    },
    {
      "epoch": 0.3538230884557721,
      "grad_norm": 0.10842984435009215,
      "learning_rate": 0.00019980822087128396,
      "loss": 0.0228,
      "step": 472
    },
    {
      "epoch": 0.3545727136431784,
      "grad_norm": 0.18970674109601535,
      "learning_rate": 0.00019980278238733204,
      "loss": 0.0421,
      "step": 473
    },
    {
      "epoch": 0.3553223388305847,
      "grad_norm": 0.14506946604862325,
      "learning_rate": 0.0001997972679414267,
      "loss": 0.0172,
      "step": 474
    },
    {
      "epoch": 0.356071964017991,
      "grad_norm": 0.21685436100064492,
      "learning_rate": 0.00019979167753776518,
      "loss": 0.041,
      "step": 475
    },
    {
      "epoch": 0.3568215892053973,
      "grad_norm": 0.14857894972860353,
      "learning_rate": 0.00019978601118060237,
      "loss": 0.0249,
      "step": 476
    },
    {
      "epoch": 0.3575712143928036,
      "grad_norm": 0.1216382248920481,
      "learning_rate": 0.00019978026887425117,
      "loss": 0.0229,
      "step": 477
    },
    {
      "epoch": 0.3583208395802099,
      "grad_norm": 0.2818825703680443,
      "learning_rate": 0.000199774450623082,
      "loss": 0.0415,
      "step": 478
    },
    {
      "epoch": 0.35907046476761617,
      "grad_norm": 0.23751907682089213,
      "learning_rate": 0.00019976855643152338,
      "loss": 0.053,
      "step": 479
    },
    {
      "epoch": 0.3598200899550225,
      "grad_norm": 0.1191015992983086,
      "learning_rate": 0.00019976258630406148,
      "loss": 0.0146,
      "step": 480
    },
    {
      "epoch": 0.3605697151424288,
      "grad_norm": 0.12423667768269722,
      "learning_rate": 0.00019975654024524024,
      "loss": 0.0231,
      "step": 481
    },
    {
      "epoch": 0.3613193403298351,
      "grad_norm": 0.1294080326047856,
      "learning_rate": 0.0001997504182596615,
      "loss": 0.0349,
      "step": 482
    },
    {
      "epoch": 0.3620689655172414,
      "grad_norm": 0.1642500818195986,
      "learning_rate": 0.00019974422035198478,
      "loss": 0.0448,
      "step": 483
    },
    {
      "epoch": 0.36281859070464767,
      "grad_norm": 0.13000096887910373,
      "learning_rate": 0.00019973794652692744,
      "loss": 0.0235,
      "step": 484
    },
    {
      "epoch": 0.36356821589205396,
      "grad_norm": 0.21703135831908477,
      "learning_rate": 0.00019973159678926466,
      "loss": 0.0381,
      "step": 485
    },
    {
      "epoch": 0.36431784107946025,
      "grad_norm": 0.18005988456596617,
      "learning_rate": 0.0001997251711438293,
      "loss": 0.0318,
      "step": 486
    },
    {
      "epoch": 0.3650674662668666,
      "grad_norm": 0.20888270057043598,
      "learning_rate": 0.00019971866959551208,
      "loss": 0.0473,
      "step": 487
    },
    {
      "epoch": 0.3658170914542729,
      "grad_norm": 0.4512020978781191,
      "learning_rate": 0.00019971209214926145,
      "loss": 0.064,
      "step": 488
    },
    {
      "epoch": 0.36656671664167917,
      "grad_norm": 0.14926340432556415,
      "learning_rate": 0.00019970543881008367,
      "loss": 0.0185,
      "step": 489
    },
    {
      "epoch": 0.36731634182908546,
      "grad_norm": 0.08818125993431587,
      "learning_rate": 0.0001996987095830427,
      "loss": 0.0233,
      "step": 490
    },
    {
      "epoch": 0.36806596701649175,
      "grad_norm": 0.08969986834486512,
      "learning_rate": 0.0001996919044732603,
      "loss": 0.019,
      "step": 491
    },
    {
      "epoch": 0.36881559220389803,
      "grad_norm": 0.41899818938779476,
      "learning_rate": 0.000199685023485916,
      "loss": 0.0997,
      "step": 492
    },
    {
      "epoch": 0.3695652173913043,
      "grad_norm": 0.33768765033376147,
      "learning_rate": 0.00019967806662624706,
      "loss": 0.0584,
      "step": 493
    },
    {
      "epoch": 0.37031484257871067,
      "grad_norm": 0.1901313204957896,
      "learning_rate": 0.00019967103389954843,
      "loss": 0.0335,
      "step": 494
    },
    {
      "epoch": 0.37106446776611696,
      "grad_norm": 0.19759284135729685,
      "learning_rate": 0.00019966392531117297,
      "loss": 0.0414,
      "step": 495
    },
    {
      "epoch": 0.37181409295352325,
      "grad_norm": 0.29640486472757016,
      "learning_rate": 0.0001996567408665311,
      "loss": 0.04,
      "step": 496
    },
    {
      "epoch": 0.37256371814092953,
      "grad_norm": 0.25841113695795564,
      "learning_rate": 0.00019964948057109102,
      "loss": 0.0558,
      "step": 497
    },
    {
      "epoch": 0.3733133433283358,
      "grad_norm": 0.33965462242202665,
      "learning_rate": 0.00019964214443037875,
      "loss": 0.0306,
      "step": 498
    },
    {
      "epoch": 0.3740629685157421,
      "grad_norm": 0.26841847008722564,
      "learning_rate": 0.00019963473244997797,
      "loss": 0.069,
      "step": 499
    },
    {
      "epoch": 0.3748125937031484,
      "grad_norm": 0.30309480215998835,
      "learning_rate": 0.00019962724463553006,
      "loss": 0.0459,
      "step": 500
    },
    {
      "epoch": 0.37556221889055474,
      "grad_norm": 0.14318386120653975,
      "learning_rate": 0.0001996196809927342,
      "loss": 0.0381,
      "step": 501
    },
    {
      "epoch": 0.37631184407796103,
      "grad_norm": 0.19393154852463967,
      "learning_rate": 0.00019961204152734715,
      "loss": 0.0416,
      "step": 502
    },
    {
      "epoch": 0.3770614692653673,
      "grad_norm": 0.1669666962274524,
      "learning_rate": 0.00019960432624518357,
      "loss": 0.0271,
      "step": 503
    },
    {
      "epoch": 0.3778110944527736,
      "grad_norm": 0.19198896931984977,
      "learning_rate": 0.00019959653515211567,
      "loss": 0.0189,
      "step": 504
    },
    {
      "epoch": 0.3785607196401799,
      "grad_norm": 0.27264366841448173,
      "learning_rate": 0.0001995886682540734,
      "loss": 0.0364,
      "step": 505
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 0.16942388949482184,
      "learning_rate": 0.00019958072555704441,
      "loss": 0.0346,
      "step": 506
    },
    {
      "epoch": 0.3800599700149925,
      "grad_norm": 0.413830565527178,
      "learning_rate": 0.0001995727070670741,
      "loss": 0.1878,
      "step": 507
    },
    {
      "epoch": 0.3808095952023988,
      "grad_norm": 0.16726770681064876,
      "learning_rate": 0.00019956461279026545,
      "loss": 0.0185,
      "step": 508
    },
    {
      "epoch": 0.3815592203898051,
      "grad_norm": 0.48839983148694394,
      "learning_rate": 0.00019955644273277921,
      "loss": 0.0689,
      "step": 509
    },
    {
      "epoch": 0.3823088455772114,
      "grad_norm": 0.07894912430939359,
      "learning_rate": 0.00019954819690083377,
      "loss": 0.0155,
      "step": 510
    },
    {
      "epoch": 0.3830584707646177,
      "grad_norm": 0.21924729131386728,
      "learning_rate": 0.00019953987530070522,
      "loss": 0.041,
      "step": 511
    },
    {
      "epoch": 0.383808095952024,
      "grad_norm": 0.30235520265779503,
      "learning_rate": 0.0001995314779387273,
      "loss": 0.0746,
      "step": 512
    },
    {
      "epoch": 0.38455772113943026,
      "grad_norm": 0.09074048325041456,
      "learning_rate": 0.0001995230048212914,
      "loss": 0.0219,
      "step": 513
    },
    {
      "epoch": 0.3853073463268366,
      "grad_norm": 0.1113928831644256,
      "learning_rate": 0.00019951445595484658,
      "loss": 0.0336,
      "step": 514
    },
    {
      "epoch": 0.3860569715142429,
      "grad_norm": 0.12625784415150787,
      "learning_rate": 0.00019950583134589956,
      "loss": 0.0169,
      "step": 515
    },
    {
      "epoch": 0.3868065967016492,
      "grad_norm": 0.19223761966230765,
      "learning_rate": 0.00019949713100101475,
      "loss": 0.033,
      "step": 516
    },
    {
      "epoch": 0.3875562218890555,
      "grad_norm": 0.17631590192078078,
      "learning_rate": 0.0001994883549268141,
      "loss": 0.0272,
      "step": 517
    },
    {
      "epoch": 0.38830584707646176,
      "grad_norm": 0.22486637349137442,
      "learning_rate": 0.0001994795031299773,
      "loss": 0.0198,
      "step": 518
    },
    {
      "epoch": 0.38905547226386805,
      "grad_norm": 0.12301541424429753,
      "learning_rate": 0.0001994705756172416,
      "loss": 0.016,
      "step": 519
    },
    {
      "epoch": 0.38980509745127434,
      "grad_norm": 0.19443551896854133,
      "learning_rate": 0.00019946157239540194,
      "loss": 0.0432,
      "step": 520
    },
    {
      "epoch": 0.3905547226386807,
      "grad_norm": 0.20182425703756351,
      "learning_rate": 0.00019945249347131088,
      "loss": 0.0229,
      "step": 521
    },
    {
      "epoch": 0.391304347826087,
      "grad_norm": 0.2858383123865256,
      "learning_rate": 0.00019944333885187854,
      "loss": 0.0579,
      "step": 522
    },
    {
      "epoch": 0.39205397301349326,
      "grad_norm": 0.2172640794478875,
      "learning_rate": 0.0001994341085440727,
      "loss": 0.038,
      "step": 523
    },
    {
      "epoch": 0.39280359820089955,
      "grad_norm": 0.08883145436518529,
      "learning_rate": 0.0001994248025549187,
      "loss": 0.0177,
      "step": 524
    },
    {
      "epoch": 0.39355322338830584,
      "grad_norm": 0.14015443846926087,
      "learning_rate": 0.0001994154208914996,
      "loss": 0.0231,
      "step": 525
    },
    {
      "epoch": 0.39430284857571213,
      "grad_norm": 0.10302600943071646,
      "learning_rate": 0.00019940596356095592,
      "loss": 0.0151,
      "step": 526
    },
    {
      "epoch": 0.3950524737631184,
      "grad_norm": 0.9708649007460824,
      "learning_rate": 0.00019939643057048588,
      "loss": 0.0883,
      "step": 527
    },
    {
      "epoch": 0.39580209895052476,
      "grad_norm": 0.3148212650301697,
      "learning_rate": 0.00019938682192734516,
      "loss": 0.0349,
      "step": 528
    },
    {
      "epoch": 0.39655172413793105,
      "grad_norm": 0.539851691055398,
      "learning_rate": 0.00019937713763884714,
      "loss": 0.1123,
      "step": 529
    },
    {
      "epoch": 0.39730134932533734,
      "grad_norm": 0.1175820770588898,
      "learning_rate": 0.00019936737771236273,
      "loss": 0.0178,
      "step": 530
    },
    {
      "epoch": 0.3980509745127436,
      "grad_norm": 0.8469963459052572,
      "learning_rate": 0.00019935754215532042,
      "loss": 0.0395,
      "step": 531
    },
    {
      "epoch": 0.3988005997001499,
      "grad_norm": 0.11032966949683876,
      "learning_rate": 0.00019934763097520622,
      "loss": 0.0263,
      "step": 532
    },
    {
      "epoch": 0.3995502248875562,
      "grad_norm": 0.13050708875089792,
      "learning_rate": 0.00019933764417956376,
      "loss": 0.0254,
      "step": 533
    },
    {
      "epoch": 0.4002998500749625,
      "grad_norm": 0.3062589113054434,
      "learning_rate": 0.00019932758177599423,
      "loss": 0.084,
      "step": 534
    },
    {
      "epoch": 0.40104947526236884,
      "grad_norm": 0.26243574980633094,
      "learning_rate": 0.00019931744377215625,
      "loss": 0.0536,
      "step": 535
    },
    {
      "epoch": 0.4017991004497751,
      "grad_norm": 0.33765459498988715,
      "learning_rate": 0.0001993072301757661,
      "loss": 0.0591,
      "step": 536
    },
    {
      "epoch": 0.4025487256371814,
      "grad_norm": 0.25884008527812624,
      "learning_rate": 0.0001992969409945976,
      "loss": 0.0359,
      "step": 537
    },
    {
      "epoch": 0.4032983508245877,
      "grad_norm": 0.17956799020329997,
      "learning_rate": 0.00019928657623648202,
      "loss": 0.0504,
      "step": 538
    },
    {
      "epoch": 0.404047976011994,
      "grad_norm": 0.3472847655807921,
      "learning_rate": 0.0001992761359093082,
      "loss": 0.1299,
      "step": 539
    },
    {
      "epoch": 0.4047976011994003,
      "grad_norm": 0.17649807509299384,
      "learning_rate": 0.00019926562002102244,
      "loss": 0.0271,
      "step": 540
    },
    {
      "epoch": 0.40554722638680657,
      "grad_norm": 0.23706785586528362,
      "learning_rate": 0.00019925502857962867,
      "loss": 0.0577,
      "step": 541
    },
    {
      "epoch": 0.4062968515742129,
      "grad_norm": 0.33871120855449127,
      "learning_rate": 0.00019924436159318823,
      "loss": 0.0572,
      "step": 542
    },
    {
      "epoch": 0.4070464767616192,
      "grad_norm": 0.3197595611177245,
      "learning_rate": 0.00019923361906981998,
      "loss": 0.0848,
      "step": 543
    },
    {
      "epoch": 0.4077961019490255,
      "grad_norm": 0.16797474123430559,
      "learning_rate": 0.0001992228010177003,
      "loss": 0.0343,
      "step": 544
    },
    {
      "epoch": 0.4085457271364318,
      "grad_norm": 0.2859867447131316,
      "learning_rate": 0.00019921190744506297,
      "loss": 0.0887,
      "step": 545
    },
    {
      "epoch": 0.40929535232383807,
      "grad_norm": 0.286424751553018,
      "learning_rate": 0.00019920093836019937,
      "loss": 0.046,
      "step": 546
    },
    {
      "epoch": 0.41004497751124436,
      "grad_norm": 0.26871363149433986,
      "learning_rate": 0.00019918989377145827,
      "loss": 0.0338,
      "step": 547
    },
    {
      "epoch": 0.4107946026986507,
      "grad_norm": 0.22113528855076964,
      "learning_rate": 0.00019917877368724597,
      "loss": 0.0393,
      "step": 548
    },
    {
      "epoch": 0.411544227886057,
      "grad_norm": 0.5080759113354069,
      "learning_rate": 0.00019916757811602612,
      "loss": 0.0426,
      "step": 549
    },
    {
      "epoch": 0.4122938530734633,
      "grad_norm": 0.17224155989086465,
      "learning_rate": 0.00019915630706631996,
      "loss": 0.0332,
      "step": 550
    },
    {
      "epoch": 0.41304347826086957,
      "grad_norm": 0.1413381734356381,
      "learning_rate": 0.00019914496054670612,
      "loss": 0.0207,
      "step": 551
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 0.40497374740103204,
      "learning_rate": 0.00019913353856582062,
      "loss": 0.0429,
      "step": 552
    },
    {
      "epoch": 0.41454272863568215,
      "grad_norm": 0.1827489824911213,
      "learning_rate": 0.00019912204113235702,
      "loss": 0.0433,
      "step": 553
    },
    {
      "epoch": 0.41529235382308843,
      "grad_norm": 0.13058740270083716,
      "learning_rate": 0.00019911046825506624,
      "loss": 0.0295,
      "step": 554
    },
    {
      "epoch": 0.4160419790104948,
      "grad_norm": 0.3724498056481174,
      "learning_rate": 0.00019909881994275663,
      "loss": 0.0453,
      "step": 555
    },
    {
      "epoch": 0.41679160419790107,
      "grad_norm": 0.14013096187220492,
      "learning_rate": 0.00019908709620429393,
      "loss": 0.0164,
      "step": 556
    },
    {
      "epoch": 0.41754122938530736,
      "grad_norm": 0.15261000750805073,
      "learning_rate": 0.00019907529704860137,
      "loss": 0.0398,
      "step": 557
    },
    {
      "epoch": 0.41829085457271364,
      "grad_norm": 0.19468792500213766,
      "learning_rate": 0.0001990634224846595,
      "loss": 0.028,
      "step": 558
    },
    {
      "epoch": 0.41904047976011993,
      "grad_norm": 0.20642163177246278,
      "learning_rate": 0.0001990514725215063,
      "loss": 0.0482,
      "step": 559
    },
    {
      "epoch": 0.4197901049475262,
      "grad_norm": 0.2640924448933564,
      "learning_rate": 0.00019903944716823712,
      "loss": 0.0429,
      "step": 560
    },
    {
      "epoch": 0.4205397301349325,
      "grad_norm": 0.17380599186496173,
      "learning_rate": 0.00019902734643400475,
      "loss": 0.0298,
      "step": 561
    },
    {
      "epoch": 0.42128935532233885,
      "grad_norm": 0.10058745366405412,
      "learning_rate": 0.00019901517032801927,
      "loss": 0.0234,
      "step": 562
    },
    {
      "epoch": 0.42203898050974514,
      "grad_norm": 0.09094624368245648,
      "learning_rate": 0.00019900291885954817,
      "loss": 0.0273,
      "step": 563
    },
    {
      "epoch": 0.42278860569715143,
      "grad_norm": 0.5223505243850018,
      "learning_rate": 0.0001989905920379163,
      "loss": 0.057,
      "step": 564
    },
    {
      "epoch": 0.4235382308845577,
      "grad_norm": 0.1006554752205893,
      "learning_rate": 0.00019897818987250584,
      "loss": 0.0242,
      "step": 565
    },
    {
      "epoch": 0.424287856071964,
      "grad_norm": 0.19684931467674244,
      "learning_rate": 0.00019896571237275635,
      "loss": 0.0534,
      "step": 566
    },
    {
      "epoch": 0.4250374812593703,
      "grad_norm": 0.2875002734473597,
      "learning_rate": 0.00019895315954816475,
      "loss": 0.0546,
      "step": 567
    },
    {
      "epoch": 0.4257871064467766,
      "grad_norm": 0.2722797709611978,
      "learning_rate": 0.0001989405314082852,
      "loss": 0.0451,
      "step": 568
    },
    {
      "epoch": 0.42653673163418293,
      "grad_norm": 0.33219794421491367,
      "learning_rate": 0.00019892782796272922,
      "loss": 0.0463,
      "step": 569
    },
    {
      "epoch": 0.4272863568215892,
      "grad_norm": 0.38257048354477796,
      "learning_rate": 0.0001989150492211657,
      "loss": 0.0267,
      "step": 570
    },
    {
      "epoch": 0.4280359820089955,
      "grad_norm": 0.2016531621539919,
      "learning_rate": 0.0001989021951933208,
      "loss": 0.0475,
      "step": 571
    },
    {
      "epoch": 0.4287856071964018,
      "grad_norm": 0.2735832711729896,
      "learning_rate": 0.00019888926588897802,
      "loss": 0.0208,
      "step": 572
    },
    {
      "epoch": 0.4295352323838081,
      "grad_norm": 0.2390819482301444,
      "learning_rate": 0.00019887626131797806,
      "loss": 0.0548,
      "step": 573
    },
    {
      "epoch": 0.4302848575712144,
      "grad_norm": 0.36787937181380886,
      "learning_rate": 0.000198863181490219,
      "loss": 0.0746,
      "step": 574
    },
    {
      "epoch": 0.43103448275862066,
      "grad_norm": 0.1553408786452615,
      "learning_rate": 0.00019885002641565613,
      "loss": 0.0461,
      "step": 575
    },
    {
      "epoch": 0.431784107946027,
      "grad_norm": 0.11860907854546467,
      "learning_rate": 0.00019883679610430212,
      "loss": 0.0285,
      "step": 576
    },
    {
      "epoch": 0.4325337331334333,
      "grad_norm": 0.48368138764405016,
      "learning_rate": 0.00019882349056622676,
      "loss": 0.071,
      "step": 577
    },
    {
      "epoch": 0.4332833583208396,
      "grad_norm": 0.34040734982154025,
      "learning_rate": 0.0001988101098115572,
      "loss": 0.0407,
      "step": 578
    },
    {
      "epoch": 0.4340329835082459,
      "grad_norm": 0.3794863668368257,
      "learning_rate": 0.00019879665385047777,
      "loss": 0.0867,
      "step": 579
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 0.1897465577471384,
      "learning_rate": 0.00019878312269323018,
      "loss": 0.024,
      "step": 580
    },
    {
      "epoch": 0.43553223388305845,
      "grad_norm": 0.42172080155699915,
      "learning_rate": 0.00019876951635011317,
      "loss": 0.0354,
      "step": 581
    },
    {
      "epoch": 0.4362818590704648,
      "grad_norm": 0.14200473008966963,
      "learning_rate": 0.00019875583483148286,
      "loss": 0.0301,
      "step": 582
    },
    {
      "epoch": 0.4370314842578711,
      "grad_norm": 0.30603866886887476,
      "learning_rate": 0.00019874207814775253,
      "loss": 0.07,
      "step": 583
    },
    {
      "epoch": 0.43778110944527737,
      "grad_norm": 0.29839798216202484,
      "learning_rate": 0.00019872824630939265,
      "loss": 0.0695,
      "step": 584
    },
    {
      "epoch": 0.43853073463268366,
      "grad_norm": 0.16530441756570893,
      "learning_rate": 0.00019871433932693093,
      "loss": 0.0309,
      "step": 585
    },
    {
      "epoch": 0.43928035982008995,
      "grad_norm": 0.11795224999185168,
      "learning_rate": 0.00019870035721095226,
      "loss": 0.0174,
      "step": 586
    },
    {
      "epoch": 0.44002998500749624,
      "grad_norm": 0.1263111693836232,
      "learning_rate": 0.00019868629997209873,
      "loss": 0.0147,
      "step": 587
    },
    {
      "epoch": 0.4407796101949025,
      "grad_norm": 0.885677670738033,
      "learning_rate": 0.00019867216762106958,
      "loss": 0.064,
      "step": 588
    },
    {
      "epoch": 0.44152923538230887,
      "grad_norm": 0.21259265942058897,
      "learning_rate": 0.00019865796016862124,
      "loss": 0.029,
      "step": 589
    },
    {
      "epoch": 0.44227886056971516,
      "grad_norm": 0.43037037278365686,
      "learning_rate": 0.00019864367762556731,
      "loss": 0.1131,
      "step": 590
    },
    {
      "epoch": 0.44302848575712145,
      "grad_norm": 2.717368196121025,
      "learning_rate": 0.0001986293200027785,
      "loss": 0.1806,
      "step": 591
    },
    {
      "epoch": 0.44377811094452774,
      "grad_norm": 0.1869451347832849,
      "learning_rate": 0.0001986148873111827,
      "loss": 0.0464,
      "step": 592
    },
    {
      "epoch": 0.444527736131934,
      "grad_norm": 0.11771674570347297,
      "learning_rate": 0.00019860037956176491,
      "loss": 0.0231,
      "step": 593
    },
    {
      "epoch": 0.4452773613193403,
      "grad_norm": 1.304481347752616,
      "learning_rate": 0.00019858579676556733,
      "loss": 0.1225,
      "step": 594
    },
    {
      "epoch": 0.4460269865067466,
      "grad_norm": 0.14663993783317505,
      "learning_rate": 0.00019857113893368916,
      "loss": 0.029,
      "step": 595
    },
    {
      "epoch": 0.44677661169415295,
      "grad_norm": 0.19789026692683717,
      "learning_rate": 0.00019855640607728683,
      "loss": 0.0311,
      "step": 596
    },
    {
      "epoch": 0.44752623688155924,
      "grad_norm": 0.17345157071354902,
      "learning_rate": 0.00019854159820757383,
      "loss": 0.0452,
      "step": 597
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 0.22394789265532525,
      "learning_rate": 0.0001985267153358207,
      "loss": 0.0705,
      "step": 598
    },
    {
      "epoch": 0.4490254872563718,
      "grad_norm": 0.23461119444057724,
      "learning_rate": 0.00019851175747335507,
      "loss": 0.0545,
      "step": 599
    },
    {
      "epoch": 0.4497751124437781,
      "grad_norm": 0.4304063727827009,
      "learning_rate": 0.00019849672463156175,
      "loss": 0.0616,
      "step": 600
    },
    {
      "epoch": 0.4505247376311844,
      "grad_norm": 0.16955416376624108,
      "learning_rate": 0.0001984816168218825,
      "loss": 0.0356,
      "step": 601
    },
    {
      "epoch": 0.4512743628185907,
      "grad_norm": 0.14119902116354704,
      "learning_rate": 0.00019846643405581617,
      "loss": 0.0332,
      "step": 602
    },
    {
      "epoch": 0.452023988005997,
      "grad_norm": 0.10374004297285797,
      "learning_rate": 0.00019845117634491874,
      "loss": 0.0251,
      "step": 603
    },
    {
      "epoch": 0.4527736131934033,
      "grad_norm": 0.4211702606487173,
      "learning_rate": 0.0001984358437008031,
      "loss": 0.064,
      "step": 604
    },
    {
      "epoch": 0.4535232383808096,
      "grad_norm": 0.3669747220639361,
      "learning_rate": 0.0001984204361351393,
      "loss": 0.0455,
      "step": 605
    },
    {
      "epoch": 0.4542728635682159,
      "grad_norm": 0.1945083024191305,
      "learning_rate": 0.00019840495365965432,
      "loss": 0.0279,
      "step": 606
    },
    {
      "epoch": 0.4550224887556222,
      "grad_norm": 0.21941701311044354,
      "learning_rate": 0.0001983893962861322,
      "loss": 0.03,
      "step": 607
    },
    {
      "epoch": 0.45577211394302847,
      "grad_norm": 0.29686743953828243,
      "learning_rate": 0.00019837376402641397,
      "loss": 0.0549,
      "step": 608
    },
    {
      "epoch": 0.45652173913043476,
      "grad_norm": 0.30788805888526277,
      "learning_rate": 0.00019835805689239768,
      "loss": 0.0381,
      "step": 609
    },
    {
      "epoch": 0.4572713643178411,
      "grad_norm": 0.15721381527401596,
      "learning_rate": 0.0001983422748960383,
      "loss": 0.0423,
      "step": 610
    },
    {
      "epoch": 0.4580209895052474,
      "grad_norm": 0.3045330866637826,
      "learning_rate": 0.00019832641804934793,
      "loss": 0.0356,
      "step": 611
    },
    {
      "epoch": 0.4587706146926537,
      "grad_norm": 0.5176832724837477,
      "learning_rate": 0.00019831048636439547,
      "loss": 0.0467,
      "step": 612
    },
    {
      "epoch": 0.45952023988005997,
      "grad_norm": 0.3645973754690773,
      "learning_rate": 0.00019829447985330687,
      "loss": 0.0343,
      "step": 613
    },
    {
      "epoch": 0.46026986506746626,
      "grad_norm": 0.1612943182389018,
      "learning_rate": 0.00019827839852826502,
      "loss": 0.0415,
      "step": 614
    },
    {
      "epoch": 0.46101949025487254,
      "grad_norm": 0.1792253469087533,
      "learning_rate": 0.00019826224240150975,
      "loss": 0.0186,
      "step": 615
    },
    {
      "epoch": 0.4617691154422789,
      "grad_norm": 0.7884751065266186,
      "learning_rate": 0.00019824601148533779,
      "loss": 0.0596,
      "step": 616
    },
    {
      "epoch": 0.4625187406296852,
      "grad_norm": 0.42798297314799133,
      "learning_rate": 0.00019822970579210286,
      "loss": 0.0656,
      "step": 617
    },
    {
      "epoch": 0.46326836581709147,
      "grad_norm": 0.22456024180897916,
      "learning_rate": 0.00019821332533421558,
      "loss": 0.0359,
      "step": 618
    },
    {
      "epoch": 0.46401799100449775,
      "grad_norm": 0.14179957969217744,
      "learning_rate": 0.00019819687012414338,
      "loss": 0.0194,
      "step": 619
    },
    {
      "epoch": 0.46476761619190404,
      "grad_norm": 0.19534431900414814,
      "learning_rate": 0.00019818034017441075,
      "loss": 0.0421,
      "step": 620
    },
    {
      "epoch": 0.46551724137931033,
      "grad_norm": 0.31092361297673426,
      "learning_rate": 0.00019816373549759887,
      "loss": 0.0851,
      "step": 621
    },
    {
      "epoch": 0.4662668665667166,
      "grad_norm": 0.21818167223637625,
      "learning_rate": 0.000198147056106346,
      "loss": 0.0414,
      "step": 622
    },
    {
      "epoch": 0.46701649175412296,
      "grad_norm": 0.3361005606854733,
      "learning_rate": 0.0001981303020133471,
      "loss": 0.0376,
      "step": 623
    },
    {
      "epoch": 0.46776611694152925,
      "grad_norm": 0.27917569510632445,
      "learning_rate": 0.00019811347323135415,
      "loss": 0.0328,
      "step": 624
    },
    {
      "epoch": 0.46851574212893554,
      "grad_norm": 0.2421974362279143,
      "learning_rate": 0.00019809656977317578,
      "loss": 0.0377,
      "step": 625
    },
    {
      "epoch": 0.46926536731634183,
      "grad_norm": 0.37153373285545455,
      "learning_rate": 0.00019807959165167756,
      "loss": 0.0614,
      "step": 626
    },
    {
      "epoch": 0.4700149925037481,
      "grad_norm": 0.4354760899723439,
      "learning_rate": 0.00019806253887978193,
      "loss": 0.0617,
      "step": 627
    },
    {
      "epoch": 0.4707646176911544,
      "grad_norm": 0.3409376339007273,
      "learning_rate": 0.00019804541147046812,
      "loss": 0.072,
      "step": 628
    },
    {
      "epoch": 0.4715142428785607,
      "grad_norm": 0.20518415022662134,
      "learning_rate": 0.00019802820943677217,
      "loss": 0.0353,
      "step": 629
    },
    {
      "epoch": 0.47226386806596704,
      "grad_norm": 0.48302199757179837,
      "learning_rate": 0.0001980109327917868,
      "loss": 0.0949,
      "step": 630
    },
    {
      "epoch": 0.47301349325337333,
      "grad_norm": 0.3359911080571668,
      "learning_rate": 0.00019799358154866172,
      "loss": 0.06,
      "step": 631
    },
    {
      "epoch": 0.4737631184407796,
      "grad_norm": 0.20737705335990284,
      "learning_rate": 0.00019797615572060329,
      "loss": 0.0472,
      "step": 632
    },
    {
      "epoch": 0.4745127436281859,
      "grad_norm": 0.18157374159375403,
      "learning_rate": 0.00019795865532087464,
      "loss": 0.0241,
      "step": 633
    },
    {
      "epoch": 0.4752623688155922,
      "grad_norm": 0.17515650486105908,
      "learning_rate": 0.00019794108036279572,
      "loss": 0.0382,
      "step": 634
    },
    {
      "epoch": 0.4760119940029985,
      "grad_norm": 0.17927234002552866,
      "learning_rate": 0.00019792343085974317,
      "loss": 0.0438,
      "step": 635
    },
    {
      "epoch": 0.4767616191904048,
      "grad_norm": 0.1490232907487753,
      "learning_rate": 0.00019790570682515042,
      "loss": 0.0304,
      "step": 636
    },
    {
      "epoch": 0.4775112443778111,
      "grad_norm": 0.11924600179670634,
      "learning_rate": 0.00019788790827250755,
      "loss": 0.0301,
      "step": 637
    },
    {
      "epoch": 0.4782608695652174,
      "grad_norm": 0.23746193370080762,
      "learning_rate": 0.00019787003521536142,
      "loss": 0.0605,
      "step": 638
    },
    {
      "epoch": 0.4790104947526237,
      "grad_norm": 0.17930143257936546,
      "learning_rate": 0.00019785208766731564,
      "loss": 0.0314,
      "step": 639
    },
    {
      "epoch": 0.47976011994003,
      "grad_norm": 0.2676999155407892,
      "learning_rate": 0.00019783406564203036,
      "loss": 0.0331,
      "step": 640
    },
    {
      "epoch": 0.48050974512743627,
      "grad_norm": 0.4707846057350767,
      "learning_rate": 0.00019781596915322263,
      "loss": 0.0633,
      "step": 641
    },
    {
      "epoch": 0.48125937031484256,
      "grad_norm": 0.2747406037837668,
      "learning_rate": 0.00019779779821466599,
      "loss": 0.0714,
      "step": 642
    },
    {
      "epoch": 0.48200899550224885,
      "grad_norm": 0.2882785148580262,
      "learning_rate": 0.00019777955284019068,
      "loss": 0.0729,
      "step": 643
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 0.24869504191527775,
      "learning_rate": 0.00019776123304368368,
      "loss": 0.047,
      "step": 644
    },
    {
      "epoch": 0.4835082458770615,
      "grad_norm": 0.29998707906477473,
      "learning_rate": 0.00019774283883908854,
      "loss": 0.0871,
      "step": 645
    },
    {
      "epoch": 0.48425787106446777,
      "grad_norm": 0.24329771856894528,
      "learning_rate": 0.00019772437024040556,
      "loss": 0.0399,
      "step": 646
    },
    {
      "epoch": 0.48500749625187406,
      "grad_norm": 0.14822184490418175,
      "learning_rate": 0.00019770582726169145,
      "loss": 0.0278,
      "step": 647
    },
    {
      "epoch": 0.48575712143928035,
      "grad_norm": 0.15941464998123417,
      "learning_rate": 0.0001976872099170597,
      "loss": 0.0217,
      "step": 648
    },
    {
      "epoch": 0.48650674662668664,
      "grad_norm": 0.1703461578982572,
      "learning_rate": 0.00019766851822068034,
      "loss": 0.0281,
      "step": 649
    },
    {
      "epoch": 0.487256371814093,
      "grad_norm": 0.2009908790648143,
      "learning_rate": 0.00019764975218678003,
      "loss": 0.044,
      "step": 650
    },
    {
      "epoch": 0.48800599700149927,
      "grad_norm": 0.41720382557820945,
      "learning_rate": 0.000197630911829642,
      "loss": 0.0615,
      "step": 651
    },
    {
      "epoch": 0.48875562218890556,
      "grad_norm": 0.38062572986525894,
      "learning_rate": 0.000197611997163606,
      "loss": 0.0652,
      "step": 652
    },
    {
      "epoch": 0.48950524737631185,
      "grad_norm": 0.23796291211665221,
      "learning_rate": 0.00019759300820306833,
      "loss": 0.0344,
      "step": 653
    },
    {
      "epoch": 0.49025487256371814,
      "grad_norm": 0.2940292683898967,
      "learning_rate": 0.00019757394496248195,
      "loss": 0.0278,
      "step": 654
    },
    {
      "epoch": 0.4910044977511244,
      "grad_norm": 0.12805376294248555,
      "learning_rate": 0.0001975548074563563,
      "loss": 0.0379,
      "step": 655
    },
    {
      "epoch": 0.4917541229385307,
      "grad_norm": 0.6717296161523604,
      "learning_rate": 0.00019753559569925727,
      "loss": 0.0311,
      "step": 656
    },
    {
      "epoch": 0.49250374812593706,
      "grad_norm": 0.21375850359128662,
      "learning_rate": 0.00019751630970580737,
      "loss": 0.04,
      "step": 657
    },
    {
      "epoch": 0.49325337331334335,
      "grad_norm": 1.1011544057656006,
      "learning_rate": 0.00019749694949068553,
      "loss": 0.0447,
      "step": 658
    },
    {
      "epoch": 0.49400299850074963,
      "grad_norm": 0.2344054804046405,
      "learning_rate": 0.00019747751506862723,
      "loss": 0.0398,
      "step": 659
    },
    {
      "epoch": 0.4947526236881559,
      "grad_norm": 0.2688401179053218,
      "learning_rate": 0.0001974580064544244,
      "loss": 0.0491,
      "step": 660
    },
    {
      "epoch": 0.4955022488755622,
      "grad_norm": 0.3609463716415746,
      "learning_rate": 0.0001974384236629254,
      "loss": 0.0283,
      "step": 661
    },
    {
      "epoch": 0.4962518740629685,
      "grad_norm": 0.1924430257206513,
      "learning_rate": 0.0001974187667090352,
      "loss": 0.0387,
      "step": 662
    },
    {
      "epoch": 0.4970014992503748,
      "grad_norm": 2.4858317118535243,
      "learning_rate": 0.00019739903560771506,
      "loss": 0.1203,
      "step": 663
    },
    {
      "epoch": 0.49775112443778113,
      "grad_norm": 0.15149651631112154,
      "learning_rate": 0.00019737923037398272,
      "loss": 0.0347,
      "step": 664
    },
    {
      "epoch": 0.4985007496251874,
      "grad_norm": 0.19402979633130193,
      "learning_rate": 0.00019735935102291238,
      "loss": 0.0269,
      "step": 665
    },
    {
      "epoch": 0.4992503748125937,
      "grad_norm": 0.3861280819653178,
      "learning_rate": 0.00019733939756963456,
      "loss": 0.0525,
      "step": 666
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6142151870943602,
      "learning_rate": 0.00019731937002933633,
      "loss": 0.0987,
      "step": 667
    },
    {
      "epoch": 0.5007496251874063,
      "grad_norm": 0.31598454375716306,
      "learning_rate": 0.00019729926841726098,
      "loss": 0.0529,
      "step": 668
    },
    {
      "epoch": 0.5014992503748126,
      "grad_norm": 0.5469064911673248,
      "learning_rate": 0.0001972790927487083,
      "loss": 0.0813,
      "step": 669
    },
    {
      "epoch": 0.5022488755622189,
      "grad_norm": 0.34228656676192837,
      "learning_rate": 0.00019725884303903443,
      "loss": 0.0515,
      "step": 670
    },
    {
      "epoch": 0.5029985007496252,
      "grad_norm": 0.09512824108899938,
      "learning_rate": 0.0001972385193036518,
      "loss": 0.023,
      "step": 671
    },
    {
      "epoch": 0.5037481259370314,
      "grad_norm": 1.0188850279916533,
      "learning_rate": 0.00019721812155802922,
      "loss": 0.0571,
      "step": 672
    },
    {
      "epoch": 0.5044977511244377,
      "grad_norm": 0.18675863579698465,
      "learning_rate": 0.0001971976498176919,
      "loss": 0.0377,
      "step": 673
    },
    {
      "epoch": 0.5052473763118441,
      "grad_norm": 0.3293723258020241,
      "learning_rate": 0.00019717710409822127,
      "loss": 0.0615,
      "step": 674
    },
    {
      "epoch": 0.5059970014992504,
      "grad_norm": 0.1875167046612201,
      "learning_rate": 0.00019715648441525504,
      "loss": 0.0482,
      "step": 675
    },
    {
      "epoch": 0.5067466266866567,
      "grad_norm": 0.2144866629023266,
      "learning_rate": 0.00019713579078448737,
      "loss": 0.0508,
      "step": 676
    },
    {
      "epoch": 0.507496251874063,
      "grad_norm": 0.254014517882649,
      "learning_rate": 0.00019711502322166855,
      "loss": 0.0465,
      "step": 677
    },
    {
      "epoch": 0.5082458770614693,
      "grad_norm": 0.3144605160912174,
      "learning_rate": 0.0001970941817426052,
      "loss": 0.0404,
      "step": 678
    },
    {
      "epoch": 0.5089955022488756,
      "grad_norm": 0.17268374165707537,
      "learning_rate": 0.00019707326636316024,
      "loss": 0.0403,
      "step": 679
    },
    {
      "epoch": 0.5097451274362819,
      "grad_norm": 0.3780163974603985,
      "learning_rate": 0.00019705227709925275,
      "loss": 0.1052,
      "step": 680
    },
    {
      "epoch": 0.5104947526236882,
      "grad_norm": 0.2701403020134673,
      "learning_rate": 0.0001970312139668581,
      "loss": 0.0391,
      "step": 681
    },
    {
      "epoch": 0.5112443778110944,
      "grad_norm": 0.40230538667716226,
      "learning_rate": 0.0001970100769820079,
      "loss": 0.0508,
      "step": 682
    },
    {
      "epoch": 0.5119940029985007,
      "grad_norm": 0.18935347804559302,
      "learning_rate": 0.00019698886616078992,
      "loss": 0.03,
      "step": 683
    },
    {
      "epoch": 0.512743628185907,
      "grad_norm": 0.284015406678792,
      "learning_rate": 0.00019696758151934818,
      "loss": 0.0467,
      "step": 684
    },
    {
      "epoch": 0.5134932533733133,
      "grad_norm": 0.2654446934199373,
      "learning_rate": 0.0001969462230738828,
      "loss": 0.0272,
      "step": 685
    },
    {
      "epoch": 0.5142428785607196,
      "grad_norm": 0.31418359639514515,
      "learning_rate": 0.00019692479084065022,
      "loss": 0.0559,
      "step": 686
    },
    {
      "epoch": 0.5149925037481259,
      "grad_norm": 0.17910995632397456,
      "learning_rate": 0.00019690328483596288,
      "loss": 0.023,
      "step": 687
    },
    {
      "epoch": 0.5157421289355323,
      "grad_norm": 0.3000429801531833,
      "learning_rate": 0.00019688170507618948,
      "loss": 0.0394,
      "step": 688
    },
    {
      "epoch": 0.5164917541229386,
      "grad_norm": 0.22024717592881649,
      "learning_rate": 0.00019686005157775483,
      "loss": 0.0276,
      "step": 689
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 0.19465821016826632,
      "learning_rate": 0.0001968383243571398,
      "loss": 0.0443,
      "step": 690
    },
    {
      "epoch": 0.5179910044977512,
      "grad_norm": 0.2892151529682172,
      "learning_rate": 0.00019681652343088145,
      "loss": 0.043,
      "step": 691
    },
    {
      "epoch": 0.5187406296851574,
      "grad_norm": 0.7128456619867778,
      "learning_rate": 0.00019679464881557296,
      "loss": 0.0591,
      "step": 692
    },
    {
      "epoch": 0.5194902548725637,
      "grad_norm": 0.15695747122101597,
      "learning_rate": 0.0001967727005278635,
      "loss": 0.0271,
      "step": 693
    },
    {
      "epoch": 0.52023988005997,
      "grad_norm": 0.16536616371095036,
      "learning_rate": 0.00019675067858445836,
      "loss": 0.0348,
      "step": 694
    },
    {
      "epoch": 0.5209895052473763,
      "grad_norm": 0.3521117447848854,
      "learning_rate": 0.0001967285830021189,
      "loss": 0.0587,
      "step": 695
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 0.11617823896876604,
      "learning_rate": 0.00019670641379766258,
      "loss": 0.0214,
      "step": 696
    },
    {
      "epoch": 0.5224887556221889,
      "grad_norm": 0.8403253233628167,
      "learning_rate": 0.00019668417098796278,
      "loss": 0.0485,
      "step": 697
    },
    {
      "epoch": 0.5232383808095952,
      "grad_norm": 0.3771163400001756,
      "learning_rate": 0.00019666185458994896,
      "loss": 0.0714,
      "step": 698
    },
    {
      "epoch": 0.5239880059970015,
      "grad_norm": 0.26485705204934235,
      "learning_rate": 0.00019663946462060664,
      "loss": 0.0538,
      "step": 699
    },
    {
      "epoch": 0.5247376311844077,
      "grad_norm": 0.2035383135631764,
      "learning_rate": 0.0001966170010969772,
      "loss": 0.0541,
      "step": 700
    },
    {
      "epoch": 0.525487256371814,
      "grad_norm": 0.28178178783706,
      "learning_rate": 0.00019659446403615815,
      "loss": 0.0473,
      "step": 701
    },
    {
      "epoch": 0.5262368815592204,
      "grad_norm": 0.23503639805567167,
      "learning_rate": 0.00019657185345530294,
      "loss": 0.0587,
      "step": 702
    },
    {
      "epoch": 0.5269865067466267,
      "grad_norm": 0.160425614898426,
      "learning_rate": 0.00019654916937162086,
      "loss": 0.038,
      "step": 703
    },
    {
      "epoch": 0.527736131934033,
      "grad_norm": 0.189609585963746,
      "learning_rate": 0.0001965264118023773,
      "loss": 0.0354,
      "step": 704
    },
    {
      "epoch": 0.5284857571214393,
      "grad_norm": 0.22997929354771204,
      "learning_rate": 0.00019650358076489353,
      "loss": 0.0337,
      "step": 705
    },
    {
      "epoch": 0.5292353823088456,
      "grad_norm": 0.2570696653356216,
      "learning_rate": 0.00019648067627654666,
      "loss": 0.0676,
      "step": 706
    },
    {
      "epoch": 0.5299850074962519,
      "grad_norm": 0.145032416897268,
      "learning_rate": 0.00019645769835476983,
      "loss": 0.0268,
      "step": 707
    },
    {
      "epoch": 0.5307346326836582,
      "grad_norm": 0.1375934632321734,
      "learning_rate": 0.00019643464701705194,
      "loss": 0.0342,
      "step": 708
    },
    {
      "epoch": 0.5314842578710645,
      "grad_norm": 0.1593212524417168,
      "learning_rate": 0.0001964115222809379,
      "loss": 0.0243,
      "step": 709
    },
    {
      "epoch": 0.5322338830584707,
      "grad_norm": 0.11195453186432093,
      "learning_rate": 0.0001963883241640284,
      "loss": 0.0209,
      "step": 710
    },
    {
      "epoch": 0.532983508245877,
      "grad_norm": 0.11361465549252572,
      "learning_rate": 0.00019636505268397998,
      "loss": 0.0267,
      "step": 711
    },
    {
      "epoch": 0.5337331334332833,
      "grad_norm": 0.12507931145914533,
      "learning_rate": 0.0001963417078585051,
      "loss": 0.0224,
      "step": 712
    },
    {
      "epoch": 0.5344827586206896,
      "grad_norm": 0.14794369943130184,
      "learning_rate": 0.00019631828970537196,
      "loss": 0.0417,
      "step": 713
    },
    {
      "epoch": 0.5352323838080959,
      "grad_norm": 0.31812402854885846,
      "learning_rate": 0.00019629479824240456,
      "loss": 0.0498,
      "step": 714
    },
    {
      "epoch": 0.5359820089955023,
      "grad_norm": 0.13942207901890347,
      "learning_rate": 0.00019627123348748277,
      "loss": 0.0229,
      "step": 715
    },
    {
      "epoch": 0.5367316341829086,
      "grad_norm": 0.26849290243133633,
      "learning_rate": 0.00019624759545854226,
      "loss": 0.0343,
      "step": 716
    },
    {
      "epoch": 0.5374812593703149,
      "grad_norm": 0.2824175224592763,
      "learning_rate": 0.00019622388417357434,
      "loss": 0.0497,
      "step": 717
    },
    {
      "epoch": 0.5382308845577212,
      "grad_norm": 0.26827286082010765,
      "learning_rate": 0.00019620009965062618,
      "loss": 0.056,
      "step": 718
    },
    {
      "epoch": 0.5389805097451275,
      "grad_norm": 0.2507495350520352,
      "learning_rate": 0.00019617624190780068,
      "loss": 0.0525,
      "step": 719
    },
    {
      "epoch": 0.5397301349325337,
      "grad_norm": 0.15206872543556993,
      "learning_rate": 0.00019615231096325645,
      "loss": 0.0382,
      "step": 720
    },
    {
      "epoch": 0.54047976011994,
      "grad_norm": 0.2534444723340265,
      "learning_rate": 0.00019612830683520785,
      "loss": 0.0484,
      "step": 721
    },
    {
      "epoch": 0.5412293853073463,
      "grad_norm": 0.17449911174599694,
      "learning_rate": 0.00019610422954192491,
      "loss": 0.0433,
      "step": 722
    },
    {
      "epoch": 0.5419790104947526,
      "grad_norm": 0.23763217664971512,
      "learning_rate": 0.00019608007910173334,
      "loss": 0.0243,
      "step": 723
    },
    {
      "epoch": 0.5427286356821589,
      "grad_norm": 0.39657496104464174,
      "learning_rate": 0.00019605585553301454,
      "loss": 0.0627,
      "step": 724
    },
    {
      "epoch": 0.5434782608695652,
      "grad_norm": 0.2629287185998369,
      "learning_rate": 0.00019603155885420555,
      "loss": 0.0388,
      "step": 725
    },
    {
      "epoch": 0.5442278860569715,
      "grad_norm": 0.2925833008523378,
      "learning_rate": 0.00019600718908379912,
      "loss": 0.0477,
      "step": 726
    },
    {
      "epoch": 0.5449775112443778,
      "grad_norm": 0.1835903425623574,
      "learning_rate": 0.00019598274624034353,
      "loss": 0.0488,
      "step": 727
    },
    {
      "epoch": 0.545727136431784,
      "grad_norm": 0.1309786695888962,
      "learning_rate": 0.00019595823034244275,
      "loss": 0.0283,
      "step": 728
    },
    {
      "epoch": 0.5464767616191905,
      "grad_norm": 0.21239724844772065,
      "learning_rate": 0.00019593364140875637,
      "loss": 0.0372,
      "step": 729
    },
    {
      "epoch": 0.5472263868065967,
      "grad_norm": 0.1282745709770564,
      "learning_rate": 0.0001959089794579995,
      "loss": 0.0284,
      "step": 730
    },
    {
      "epoch": 0.547976011994003,
      "grad_norm": 0.13062589463989482,
      "learning_rate": 0.00019588424450894283,
      "loss": 0.0412,
      "step": 731
    },
    {
      "epoch": 0.5487256371814093,
      "grad_norm": 0.1671330845222855,
      "learning_rate": 0.00019585943658041271,
      "loss": 0.0329,
      "step": 732
    },
    {
      "epoch": 0.5494752623688156,
      "grad_norm": 0.11756985208400414,
      "learning_rate": 0.00019583455569129092,
      "loss": 0.0184,
      "step": 733
    },
    {
      "epoch": 0.5502248875562219,
      "grad_norm": 0.18075813633689752,
      "learning_rate": 0.0001958096018605148,
      "loss": 0.0329,
      "step": 734
    },
    {
      "epoch": 0.5509745127436282,
      "grad_norm": 0.12221795958442933,
      "learning_rate": 0.00019578457510707728,
      "loss": 0.0178,
      "step": 735
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.16343977992570569,
      "learning_rate": 0.0001957594754500267,
      "loss": 0.0188,
      "step": 736
    },
    {
      "epoch": 0.5524737631184408,
      "grad_norm": 0.11873410698691594,
      "learning_rate": 0.00019573430290846692,
      "loss": 0.0215,
      "step": 737
    },
    {
      "epoch": 0.553223388305847,
      "grad_norm": 0.5353970283675913,
      "learning_rate": 0.00019570905750155733,
      "loss": 0.1006,
      "step": 738
    },
    {
      "epoch": 0.5539730134932533,
      "grad_norm": 0.11628544332223312,
      "learning_rate": 0.00019568373924851265,
      "loss": 0.0281,
      "step": 739
    },
    {
      "epoch": 0.5547226386806596,
      "grad_norm": 0.14481274986660742,
      "learning_rate": 0.0001956583481686032,
      "loss": 0.0285,
      "step": 740
    },
    {
      "epoch": 0.5554722638680659,
      "grad_norm": 0.10228835567792285,
      "learning_rate": 0.0001956328842811546,
      "loss": 0.0182,
      "step": 741
    },
    {
      "epoch": 0.5562218890554723,
      "grad_norm": 0.14598036221788213,
      "learning_rate": 0.000195607347605548,
      "loss": 0.0164,
      "step": 742
    },
    {
      "epoch": 0.5569715142428786,
      "grad_norm": 0.1337465267184873,
      "learning_rate": 0.0001955817381612198,
      "loss": 0.0226,
      "step": 743
    },
    {
      "epoch": 0.5577211394302849,
      "grad_norm": 0.17512979637507264,
      "learning_rate": 0.00019555605596766195,
      "loss": 0.0257,
      "step": 744
    },
    {
      "epoch": 0.5584707646176912,
      "grad_norm": 0.26359447087528937,
      "learning_rate": 0.00019553030104442164,
      "loss": 0.0784,
      "step": 745
    },
    {
      "epoch": 0.5592203898050975,
      "grad_norm": 0.24337056758185213,
      "learning_rate": 0.00019550447341110157,
      "loss": 0.0119,
      "step": 746
    },
    {
      "epoch": 0.5599700149925038,
      "grad_norm": 0.2291020714386086,
      "learning_rate": 0.00019547857308735954,
      "loss": 0.0386,
      "step": 747
    },
    {
      "epoch": 0.56071964017991,
      "grad_norm": 0.6099594043333565,
      "learning_rate": 0.00019545260009290894,
      "loss": 0.0686,
      "step": 748
    },
    {
      "epoch": 0.5614692653673163,
      "grad_norm": 0.23313273301357787,
      "learning_rate": 0.00019542655444751832,
      "loss": 0.0358,
      "step": 749
    },
    {
      "epoch": 0.5622188905547226,
      "grad_norm": 0.30721566706948694,
      "learning_rate": 0.0001954004361710115,
      "loss": 0.056,
      "step": 750
    },
    {
      "epoch": 0.5629685157421289,
      "grad_norm": 0.14868748862150133,
      "learning_rate": 0.00019537424528326774,
      "loss": 0.0347,
      "step": 751
    },
    {
      "epoch": 0.5637181409295352,
      "grad_norm": 0.35283868429312015,
      "learning_rate": 0.00019534798180422138,
      "loss": 0.055,
      "step": 752
    },
    {
      "epoch": 0.5644677661169415,
      "grad_norm": 0.1916124066278781,
      "learning_rate": 0.00019532164575386215,
      "loss": 0.0361,
      "step": 753
    },
    {
      "epoch": 0.5652173913043478,
      "grad_norm": 0.21612561370379293,
      "learning_rate": 0.0001952952371522349,
      "loss": 0.0249,
      "step": 754
    },
    {
      "epoch": 0.5659670164917541,
      "grad_norm": 0.2924146234215674,
      "learning_rate": 0.0001952687560194398,
      "loss": 0.0565,
      "step": 755
    },
    {
      "epoch": 0.5667166416791605,
      "grad_norm": 0.2852650400539102,
      "learning_rate": 0.0001952422023756322,
      "loss": 0.0667,
      "step": 756
    },
    {
      "epoch": 0.5674662668665668,
      "grad_norm": 0.11941722208051125,
      "learning_rate": 0.0001952155762410226,
      "loss": 0.0287,
      "step": 757
    },
    {
      "epoch": 0.568215892053973,
      "grad_norm": 0.4540888173579844,
      "learning_rate": 0.0001951888776358767,
      "loss": 0.0204,
      "step": 758
    },
    {
      "epoch": 0.5689655172413793,
      "grad_norm": 0.3373617845244113,
      "learning_rate": 0.00019516210658051535,
      "loss": 0.0682,
      "step": 759
    },
    {
      "epoch": 0.5697151424287856,
      "grad_norm": 0.22739701260998646,
      "learning_rate": 0.00019513526309531459,
      "loss": 0.0407,
      "step": 760
    },
    {
      "epoch": 0.5704647676161919,
      "grad_norm": 0.1672076747143311,
      "learning_rate": 0.0001951083472007055,
      "loss": 0.0318,
      "step": 761
    },
    {
      "epoch": 0.5712143928035982,
      "grad_norm": 0.1389020412850082,
      "learning_rate": 0.00019508135891717435,
      "loss": 0.0211,
      "step": 762
    },
    {
      "epoch": 0.5719640179910045,
      "grad_norm": 0.17191468308215127,
      "learning_rate": 0.00019505429826526246,
      "loss": 0.0453,
      "step": 763
    },
    {
      "epoch": 0.5727136431784108,
      "grad_norm": 0.17000750173740772,
      "learning_rate": 0.0001950271652655663,
      "loss": 0.0225,
      "step": 764
    },
    {
      "epoch": 0.5734632683658171,
      "grad_norm": 0.11216968441486351,
      "learning_rate": 0.0001949999599387373,
      "loss": 0.018,
      "step": 765
    },
    {
      "epoch": 0.5742128935532234,
      "grad_norm": 0.25006865233631487,
      "learning_rate": 0.00019497268230548201,
      "loss": 0.022,
      "step": 766
    },
    {
      "epoch": 0.5749625187406296,
      "grad_norm": 0.49925242688607274,
      "learning_rate": 0.00019494533238656204,
      "loss": 0.0656,
      "step": 767
    },
    {
      "epoch": 0.5757121439280359,
      "grad_norm": 0.21186420607643336,
      "learning_rate": 0.0001949179102027939,
      "loss": 0.0478,
      "step": 768
    },
    {
      "epoch": 0.5764617691154422,
      "grad_norm": 0.6140154868296298,
      "learning_rate": 0.0001948904157750492,
      "loss": 0.0392,
      "step": 769
    },
    {
      "epoch": 0.5772113943028486,
      "grad_norm": 0.09252928326152278,
      "learning_rate": 0.00019486284912425457,
      "loss": 0.0204,
      "step": 770
    },
    {
      "epoch": 0.5779610194902549,
      "grad_norm": 0.19345629880652473,
      "learning_rate": 0.0001948352102713915,
      "loss": 0.0251,
      "step": 771
    },
    {
      "epoch": 0.5787106446776612,
      "grad_norm": 0.31202632688570753,
      "learning_rate": 0.00019480749923749655,
      "loss": 0.0904,
      "step": 772
    },
    {
      "epoch": 0.5794602698650675,
      "grad_norm": 0.38726486933860926,
      "learning_rate": 0.00019477971604366106,
      "loss": 0.0521,
      "step": 773
    },
    {
      "epoch": 0.5802098950524738,
      "grad_norm": 0.2030051748098563,
      "learning_rate": 0.00019475186071103148,
      "loss": 0.0431,
      "step": 774
    },
    {
      "epoch": 0.5809595202398801,
      "grad_norm": 0.15348623359559546,
      "learning_rate": 0.00019472393326080905,
      "loss": 0.0253,
      "step": 775
    },
    {
      "epoch": 0.5817091454272864,
      "grad_norm": 0.2288982438224531,
      "learning_rate": 0.0001946959337142499,
      "loss": 0.0524,
      "step": 776
    },
    {
      "epoch": 0.5824587706146926,
      "grad_norm": 0.11109856913043137,
      "learning_rate": 0.0001946678620926651,
      "loss": 0.0269,
      "step": 777
    },
    {
      "epoch": 0.5832083958020989,
      "grad_norm": 2.07579081751696,
      "learning_rate": 0.00019463971841742056,
      "loss": 0.0681,
      "step": 778
    },
    {
      "epoch": 0.5839580209895052,
      "grad_norm": 0.31296298255106864,
      "learning_rate": 0.00019461150270993693,
      "loss": 0.045,
      "step": 779
    },
    {
      "epoch": 0.5847076461769115,
      "grad_norm": 0.22119938706955986,
      "learning_rate": 0.0001945832149916898,
      "loss": 0.0251,
      "step": 780
    },
    {
      "epoch": 0.5854572713643178,
      "grad_norm": 0.18401368936862708,
      "learning_rate": 0.00019455485528420958,
      "loss": 0.0447,
      "step": 781
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 0.34006952655746775,
      "learning_rate": 0.0001945264236090814,
      "loss": 0.0328,
      "step": 782
    },
    {
      "epoch": 0.5869565217391305,
      "grad_norm": 0.5719459819450913,
      "learning_rate": 0.00019449791998794516,
      "loss": 0.0232,
      "step": 783
    },
    {
      "epoch": 0.5877061469265368,
      "grad_norm": 0.1491340544778081,
      "learning_rate": 0.0001944693444424956,
      "loss": 0.0277,
      "step": 784
    },
    {
      "epoch": 0.5884557721139431,
      "grad_norm": 0.6464031380286603,
      "learning_rate": 0.00019444069699448214,
      "loss": 0.0444,
      "step": 785
    },
    {
      "epoch": 0.5892053973013494,
      "grad_norm": 0.23038490458838246,
      "learning_rate": 0.0001944119776657089,
      "loss": 0.0376,
      "step": 786
    },
    {
      "epoch": 0.5899550224887556,
      "grad_norm": 0.24375368417956617,
      "learning_rate": 0.00019438318647803482,
      "loss": 0.0359,
      "step": 787
    },
    {
      "epoch": 0.5907046476761619,
      "grad_norm": 0.22794673576648067,
      "learning_rate": 0.00019435432345337345,
      "loss": 0.054,
      "step": 788
    },
    {
      "epoch": 0.5914542728635682,
      "grad_norm": 1.5666537621230843,
      "learning_rate": 0.000194325388613693,
      "loss": 0.0565,
      "step": 789
    },
    {
      "epoch": 0.5922038980509745,
      "grad_norm": 0.23939494460941754,
      "learning_rate": 0.00019429638198101642,
      "loss": 0.0262,
      "step": 790
    },
    {
      "epoch": 0.5929535232383808,
      "grad_norm": 0.11512448557106804,
      "learning_rate": 0.00019426730357742123,
      "loss": 0.0333,
      "step": 791
    },
    {
      "epoch": 0.5937031484257871,
      "grad_norm": 0.08278414269358454,
      "learning_rate": 0.00019423815342503958,
      "loss": 0.0192,
      "step": 792
    },
    {
      "epoch": 0.5944527736131934,
      "grad_norm": 0.8234039282439967,
      "learning_rate": 0.00019420893154605828,
      "loss": 0.0202,
      "step": 793
    },
    {
      "epoch": 0.5952023988005997,
      "grad_norm": 0.1844774144077661,
      "learning_rate": 0.00019417963796271873,
      "loss": 0.0293,
      "step": 794
    },
    {
      "epoch": 0.595952023988006,
      "grad_norm": 0.5361123111117241,
      "learning_rate": 0.00019415027269731677,
      "loss": 0.0624,
      "step": 795
    },
    {
      "epoch": 0.5967016491754122,
      "grad_norm": 0.1558817603943623,
      "learning_rate": 0.00019412083577220304,
      "loss": 0.0299,
      "step": 796
    },
    {
      "epoch": 0.5974512743628186,
      "grad_norm": 0.25135916091871174,
      "learning_rate": 0.00019409132720978254,
      "loss": 0.0446,
      "step": 797
    },
    {
      "epoch": 0.5982008995502249,
      "grad_norm": 0.4137620063339877,
      "learning_rate": 0.00019406174703251482,
      "loss": 0.0617,
      "step": 798
    },
    {
      "epoch": 0.5989505247376312,
      "grad_norm": 0.13801847124843378,
      "learning_rate": 0.000194032095262914,
      "loss": 0.035,
      "step": 799
    },
    {
      "epoch": 0.5997001499250375,
      "grad_norm": 0.09371617366687199,
      "learning_rate": 0.0001940023719235486,
      "loss": 0.0136,
      "step": 800
    },
    {
      "epoch": 0.6004497751124438,
      "grad_norm": 0.1190313546094842,
      "learning_rate": 0.00019397257703704173,
      "loss": 0.0211,
      "step": 801
    },
    {
      "epoch": 0.6011994002998501,
      "grad_norm": 0.30906713673794106,
      "learning_rate": 0.00019394271062607085,
      "loss": 0.0345,
      "step": 802
    },
    {
      "epoch": 0.6019490254872564,
      "grad_norm": 0.25312589879783953,
      "learning_rate": 0.00019391277271336792,
      "loss": 0.0226,
      "step": 803
    },
    {
      "epoch": 0.6026986506746627,
      "grad_norm": 0.37710261116216454,
      "learning_rate": 0.0001938827633217193,
      "loss": 0.0386,
      "step": 804
    },
    {
      "epoch": 0.603448275862069,
      "grad_norm": 0.1889732863444513,
      "learning_rate": 0.00019385268247396571,
      "loss": 0.04,
      "step": 805
    },
    {
      "epoch": 0.6041979010494752,
      "grad_norm": 0.22033441208719085,
      "learning_rate": 0.0001938225301930024,
      "loss": 0.0512,
      "step": 806
    },
    {
      "epoch": 0.6049475262368815,
      "grad_norm": 0.2836356477878693,
      "learning_rate": 0.00019379230650177877,
      "loss": 0.0434,
      "step": 807
    },
    {
      "epoch": 0.6056971514242878,
      "grad_norm": 0.23341456899879873,
      "learning_rate": 0.00019376201142329882,
      "loss": 0.0452,
      "step": 808
    },
    {
      "epoch": 0.6064467766116941,
      "grad_norm": 0.25296621661074753,
      "learning_rate": 0.00019373164498062066,
      "loss": 0.0308,
      "step": 809
    },
    {
      "epoch": 0.6071964017991005,
      "grad_norm": 0.23863639164852932,
      "learning_rate": 0.00019370120719685687,
      "loss": 0.0271,
      "step": 810
    },
    {
      "epoch": 0.6079460269865068,
      "grad_norm": 0.20291122107848233,
      "learning_rate": 0.00019367069809517426,
      "loss": 0.0359,
      "step": 811
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 0.20115751012517755,
      "learning_rate": 0.00019364011769879393,
      "loss": 0.0331,
      "step": 812
    },
    {
      "epoch": 0.6094452773613194,
      "grad_norm": 0.725662505864241,
      "learning_rate": 0.00019360946603099127,
      "loss": 0.0808,
      "step": 813
    },
    {
      "epoch": 0.6101949025487257,
      "grad_norm": 0.3707632032177578,
      "learning_rate": 0.00019357874311509585,
      "loss": 0.0406,
      "step": 814
    },
    {
      "epoch": 0.610944527736132,
      "grad_norm": 0.2492238320478391,
      "learning_rate": 0.00019354794897449154,
      "loss": 0.0273,
      "step": 815
    },
    {
      "epoch": 0.6116941529235382,
      "grad_norm": 0.28949212299924676,
      "learning_rate": 0.00019351708363261641,
      "loss": 0.0741,
      "step": 816
    },
    {
      "epoch": 0.6124437781109445,
      "grad_norm": 0.16433768096408977,
      "learning_rate": 0.0001934861471129627,
      "loss": 0.0333,
      "step": 817
    },
    {
      "epoch": 0.6131934032983508,
      "grad_norm": 0.261884934270732,
      "learning_rate": 0.0001934551394390768,
      "loss": 0.0476,
      "step": 818
    },
    {
      "epoch": 0.6139430284857571,
      "grad_norm": 0.17727981078341598,
      "learning_rate": 0.00019342406063455934,
      "loss": 0.0332,
      "step": 819
    },
    {
      "epoch": 0.6146926536731634,
      "grad_norm": 0.8903617488607476,
      "learning_rate": 0.00019339291072306502,
      "loss": 0.1448,
      "step": 820
    },
    {
      "epoch": 0.6154422788605697,
      "grad_norm": 0.29030138456788634,
      "learning_rate": 0.00019336168972830265,
      "loss": 0.0523,
      "step": 821
    },
    {
      "epoch": 0.616191904047976,
      "grad_norm": 0.29500920261796953,
      "learning_rate": 0.0001933303976740352,
      "loss": 0.0559,
      "step": 822
    },
    {
      "epoch": 0.6169415292353823,
      "grad_norm": 0.19708221752070373,
      "learning_rate": 0.0001932990345840797,
      "loss": 0.0387,
      "step": 823
    },
    {
      "epoch": 0.6176911544227887,
      "grad_norm": 0.1453239404111134,
      "learning_rate": 0.00019326760048230725,
      "loss": 0.021,
      "step": 824
    },
    {
      "epoch": 0.618440779610195,
      "grad_norm": 0.16559600117050013,
      "learning_rate": 0.00019323609539264295,
      "loss": 0.0328,
      "step": 825
    },
    {
      "epoch": 0.6191904047976012,
      "grad_norm": 0.2434062543518236,
      "learning_rate": 0.00019320451933906602,
      "loss": 0.0458,
      "step": 826
    },
    {
      "epoch": 0.6199400299850075,
      "grad_norm": 0.2920576187948728,
      "learning_rate": 0.00019317287234560962,
      "loss": 0.0282,
      "step": 827
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 0.18444689764461758,
      "learning_rate": 0.00019314115443636094,
      "loss": 0.0505,
      "step": 828
    },
    {
      "epoch": 0.6214392803598201,
      "grad_norm": 0.33347763003860115,
      "learning_rate": 0.00019310936563546113,
      "loss": 0.0375,
      "step": 829
    },
    {
      "epoch": 0.6221889055472264,
      "grad_norm": 0.3151793343232866,
      "learning_rate": 0.00019307750596710528,
      "loss": 0.0688,
      "step": 830
    },
    {
      "epoch": 0.6229385307346327,
      "grad_norm": 0.17223079808853164,
      "learning_rate": 0.00019304557545554246,
      "loss": 0.0149,
      "step": 831
    },
    {
      "epoch": 0.623688155922039,
      "grad_norm": 0.16131606757257835,
      "learning_rate": 0.00019301357412507568,
      "loss": 0.0358,
      "step": 832
    },
    {
      "epoch": 0.6244377811094453,
      "grad_norm": 0.15451180093753095,
      "learning_rate": 0.00019298150200006172,
      "loss": 0.0132,
      "step": 833
    },
    {
      "epoch": 0.6251874062968515,
      "grad_norm": 0.18246287463532035,
      "learning_rate": 0.00019294935910491143,
      "loss": 0.0299,
      "step": 834
    },
    {
      "epoch": 0.6259370314842578,
      "grad_norm": 0.23038969159958683,
      "learning_rate": 0.00019291714546408936,
      "loss": 0.0299,
      "step": 835
    },
    {
      "epoch": 0.6266866566716641,
      "grad_norm": 0.15241068451072204,
      "learning_rate": 0.000192884861102114,
      "loss": 0.03,
      "step": 836
    },
    {
      "epoch": 0.6274362818590704,
      "grad_norm": 0.1772524433219351,
      "learning_rate": 0.00019285250604355767,
      "loss": 0.0291,
      "step": 837
    },
    {
      "epoch": 0.6281859070464768,
      "grad_norm": 0.2908627086925324,
      "learning_rate": 0.00019282008031304643,
      "loss": 0.0447,
      "step": 838
    },
    {
      "epoch": 0.6289355322338831,
      "grad_norm": 0.26052288609262275,
      "learning_rate": 0.0001927875839352602,
      "loss": 0.0375,
      "step": 839
    },
    {
      "epoch": 0.6296851574212894,
      "grad_norm": 0.1569085422616155,
      "learning_rate": 0.0001927550169349326,
      "loss": 0.0415,
      "step": 840
    },
    {
      "epoch": 0.6304347826086957,
      "grad_norm": 0.21810162792732946,
      "learning_rate": 0.0001927223793368511,
      "loss": 0.0292,
      "step": 841
    },
    {
      "epoch": 0.631184407796102,
      "grad_norm": 0.43588945410407454,
      "learning_rate": 0.0001926896711658568,
      "loss": 0.0402,
      "step": 842
    },
    {
      "epoch": 0.6319340329835083,
      "grad_norm": 0.18607701495773632,
      "learning_rate": 0.0001926568924468446,
      "loss": 0.0299,
      "step": 843
    },
    {
      "epoch": 0.6326836581709145,
      "grad_norm": 0.38921493550716046,
      "learning_rate": 0.00019262404320476303,
      "loss": 0.0333,
      "step": 844
    },
    {
      "epoch": 0.6334332833583208,
      "grad_norm": 0.37969401105335127,
      "learning_rate": 0.00019259112346461436,
      "loss": 0.0496,
      "step": 845
    },
    {
      "epoch": 0.6341829085457271,
      "grad_norm": 0.3216879681669849,
      "learning_rate": 0.00019255813325145448,
      "loss": 0.023,
      "step": 846
    },
    {
      "epoch": 0.6349325337331334,
      "grad_norm": 0.27089341793735805,
      "learning_rate": 0.00019252507259039285,
      "loss": 0.0459,
      "step": 847
    },
    {
      "epoch": 0.6356821589205397,
      "grad_norm": 0.19188607086964637,
      "learning_rate": 0.0001924919415065927,
      "loss": 0.0285,
      "step": 848
    },
    {
      "epoch": 0.636431784107946,
      "grad_norm": 0.4712514156925914,
      "learning_rate": 0.00019245874002527072,
      "loss": 0.0382,
      "step": 849
    },
    {
      "epoch": 0.6371814092953523,
      "grad_norm": 0.4385017620850573,
      "learning_rate": 0.00019242546817169732,
      "loss": 0.0481,
      "step": 850
    },
    {
      "epoch": 0.6379310344827587,
      "grad_norm": 0.1242568847686868,
      "learning_rate": 0.0001923921259711963,
      "loss": 0.0297,
      "step": 851
    },
    {
      "epoch": 0.638680659670165,
      "grad_norm": 0.28540644515657065,
      "learning_rate": 0.00019235871344914517,
      "loss": 0.0394,
      "step": 852
    },
    {
      "epoch": 0.6394302848575713,
      "grad_norm": 0.5744981724187342,
      "learning_rate": 0.00019232523063097487,
      "loss": 0.0742,
      "step": 853
    },
    {
      "epoch": 0.6401799100449775,
      "grad_norm": 0.24642627559782443,
      "learning_rate": 0.00019229167754216984,
      "loss": 0.0348,
      "step": 854
    },
    {
      "epoch": 0.6409295352323838,
      "grad_norm": 0.44105899266315524,
      "learning_rate": 0.00019225805420826806,
      "loss": 0.0621,
      "step": 855
    },
    {
      "epoch": 0.6416791604197901,
      "grad_norm": 0.24535853410831998,
      "learning_rate": 0.00019222436065486088,
      "loss": 0.0355,
      "step": 856
    },
    {
      "epoch": 0.6424287856071964,
      "grad_norm": 0.3648266082854025,
      "learning_rate": 0.00019219059690759322,
      "loss": 0.0378,
      "step": 857
    },
    {
      "epoch": 0.6431784107946027,
      "grad_norm": 0.23410467927646403,
      "learning_rate": 0.00019215676299216337,
      "loss": 0.0453,
      "step": 858
    },
    {
      "epoch": 0.643928035982009,
      "grad_norm": 0.5771722806268218,
      "learning_rate": 0.00019212285893432295,
      "loss": 0.0605,
      "step": 859
    },
    {
      "epoch": 0.6446776611694153,
      "grad_norm": 0.42837179995622626,
      "learning_rate": 0.0001920888847598771,
      "loss": 0.0757,
      "step": 860
    },
    {
      "epoch": 0.6454272863568216,
      "grad_norm": 0.30782392032850886,
      "learning_rate": 0.00019205484049468426,
      "loss": 0.0559,
      "step": 861
    },
    {
      "epoch": 0.6461769115442278,
      "grad_norm": 0.32187116161181983,
      "learning_rate": 0.00019202072616465618,
      "loss": 0.0325,
      "step": 862
    },
    {
      "epoch": 0.6469265367316341,
      "grad_norm": 0.3170202880484331,
      "learning_rate": 0.00019198654179575803,
      "loss": 0.0616,
      "step": 863
    },
    {
      "epoch": 0.6476761619190404,
      "grad_norm": 0.15415415536695157,
      "learning_rate": 0.00019195228741400818,
      "loss": 0.0279,
      "step": 864
    },
    {
      "epoch": 0.6484257871064468,
      "grad_norm": 0.468023792818238,
      "learning_rate": 0.00019191796304547838,
      "loss": 0.072,
      "step": 865
    },
    {
      "epoch": 0.6491754122938531,
      "grad_norm": 0.17747724891036817,
      "learning_rate": 0.00019188356871629363,
      "loss": 0.0269,
      "step": 866
    },
    {
      "epoch": 0.6499250374812594,
      "grad_norm": 0.11896332361586975,
      "learning_rate": 0.00019184910445263213,
      "loss": 0.0231,
      "step": 867
    },
    {
      "epoch": 0.6506746626686657,
      "grad_norm": 0.12177188142655841,
      "learning_rate": 0.00019181457028072537,
      "loss": 0.0324,
      "step": 868
    },
    {
      "epoch": 0.651424287856072,
      "grad_norm": 0.21226493913526695,
      "learning_rate": 0.00019177996622685799,
      "loss": 0.0303,
      "step": 869
    },
    {
      "epoch": 0.6521739130434783,
      "grad_norm": 0.35288822757454136,
      "learning_rate": 0.00019174529231736788,
      "loss": 0.0528,
      "step": 870
    },
    {
      "epoch": 0.6529235382308846,
      "grad_norm": 0.18467843757358945,
      "learning_rate": 0.00019171054857864604,
      "loss": 0.0467,
      "step": 871
    },
    {
      "epoch": 0.6536731634182908,
      "grad_norm": 0.11919292144770044,
      "learning_rate": 0.00019167573503713664,
      "loss": 0.028,
      "step": 872
    },
    {
      "epoch": 0.6544227886056971,
      "grad_norm": 0.14188580015944366,
      "learning_rate": 0.00019164085171933701,
      "loss": 0.0224,
      "step": 873
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 0.17981285086992055,
      "learning_rate": 0.00019160589865179755,
      "loss": 0.0297,
      "step": 874
    },
    {
      "epoch": 0.6559220389805097,
      "grad_norm": 0.15306582904565683,
      "learning_rate": 0.00019157087586112176,
      "loss": 0.0213,
      "step": 875
    },
    {
      "epoch": 0.656671664167916,
      "grad_norm": 0.28590338086654155,
      "learning_rate": 0.00019153578337396617,
      "loss": 0.0405,
      "step": 876
    },
    {
      "epoch": 0.6574212893553223,
      "grad_norm": 0.08011629532591379,
      "learning_rate": 0.00019150062121704047,
      "loss": 0.0122,
      "step": 877
    },
    {
      "epoch": 0.6581709145427287,
      "grad_norm": 0.09441159438236417,
      "learning_rate": 0.00019146538941710727,
      "loss": 0.018,
      "step": 878
    },
    {
      "epoch": 0.658920539730135,
      "grad_norm": 0.13011973729869275,
      "learning_rate": 0.00019143008800098218,
      "loss": 0.0201,
      "step": 879
    },
    {
      "epoch": 0.6596701649175413,
      "grad_norm": 0.9504515435787874,
      "learning_rate": 0.0001913947169955339,
      "loss": 0.1022,
      "step": 880
    },
    {
      "epoch": 0.6604197901049476,
      "grad_norm": 0.11513124353776294,
      "learning_rate": 0.00019135927642768396,
      "loss": 0.0182,
      "step": 881
    },
    {
      "epoch": 0.6611694152923538,
      "grad_norm": 0.18336727207182985,
      "learning_rate": 0.00019132376632440695,
      "loss": 0.0348,
      "step": 882
    },
    {
      "epoch": 0.6619190404797601,
      "grad_norm": 0.2576296296304189,
      "learning_rate": 0.0001912881867127304,
      "loss": 0.052,
      "step": 883
    },
    {
      "epoch": 0.6626686656671664,
      "grad_norm": 0.17875011992570874,
      "learning_rate": 0.00019125253761973453,
      "loss": 0.0382,
      "step": 884
    },
    {
      "epoch": 0.6634182908545727,
      "grad_norm": 0.3514697728255144,
      "learning_rate": 0.00019121681907255272,
      "loss": 0.065,
      "step": 885
    },
    {
      "epoch": 0.664167916041979,
      "grad_norm": 0.49314124227628053,
      "learning_rate": 0.0001911810310983711,
      "loss": 0.0901,
      "step": 886
    },
    {
      "epoch": 0.6649175412293853,
      "grad_norm": 0.30243395137434004,
      "learning_rate": 0.00019114517372442857,
      "loss": 0.06,
      "step": 887
    },
    {
      "epoch": 0.6656671664167916,
      "grad_norm": 0.2127563337244439,
      "learning_rate": 0.00019110924697801694,
      "loss": 0.0319,
      "step": 888
    },
    {
      "epoch": 0.6664167916041979,
      "grad_norm": 0.14128878156969202,
      "learning_rate": 0.00019107325088648077,
      "loss": 0.0269,
      "step": 889
    },
    {
      "epoch": 0.6671664167916042,
      "grad_norm": 0.14278769406678316,
      "learning_rate": 0.00019103718547721748,
      "loss": 0.0515,
      "step": 890
    },
    {
      "epoch": 0.6679160419790104,
      "grad_norm": 0.09835609392274643,
      "learning_rate": 0.00019100105077767717,
      "loss": 0.0209,
      "step": 891
    },
    {
      "epoch": 0.6686656671664168,
      "grad_norm": 0.4225039625143984,
      "learning_rate": 0.00019096484681536274,
      "loss": 0.0626,
      "step": 892
    },
    {
      "epoch": 0.6694152923538231,
      "grad_norm": 0.1979603681699738,
      "learning_rate": 0.00019092857361782968,
      "loss": 0.048,
      "step": 893
    },
    {
      "epoch": 0.6701649175412294,
      "grad_norm": 0.22017581022730506,
      "learning_rate": 0.00019089223121268633,
      "loss": 0.0433,
      "step": 894
    },
    {
      "epoch": 0.6709145427286357,
      "grad_norm": 0.23746282110411276,
      "learning_rate": 0.00019085581962759366,
      "loss": 0.0394,
      "step": 895
    },
    {
      "epoch": 0.671664167916042,
      "grad_norm": 0.22877488071584384,
      "learning_rate": 0.00019081933889026522,
      "loss": 0.0689,
      "step": 896
    },
    {
      "epoch": 0.6724137931034483,
      "grad_norm": 0.17291962523254184,
      "learning_rate": 0.00019078278902846734,
      "loss": 0.0386,
      "step": 897
    },
    {
      "epoch": 0.6731634182908546,
      "grad_norm": 0.11386075637226863,
      "learning_rate": 0.00019074617007001877,
      "loss": 0.0241,
      "step": 898
    },
    {
      "epoch": 0.6739130434782609,
      "grad_norm": 0.36286240601431663,
      "learning_rate": 0.00019070948204279103,
      "loss": 0.0587,
      "step": 899
    },
    {
      "epoch": 0.6746626686656672,
      "grad_norm": 0.15708001038991823,
      "learning_rate": 0.00019067272497470808,
      "loss": 0.0412,
      "step": 900
    },
    {
      "epoch": 0.6754122938530734,
      "grad_norm": 0.2063576324808673,
      "learning_rate": 0.0001906358988937465,
      "loss": 0.0307,
      "step": 901
    },
    {
      "epoch": 0.6761619190404797,
      "grad_norm": 0.5738564722564278,
      "learning_rate": 0.0001905990038279354,
      "loss": 0.0964,
      "step": 902
    },
    {
      "epoch": 0.676911544227886,
      "grad_norm": 0.39568723739967954,
      "learning_rate": 0.00019056203980535638,
      "loss": 0.0671,
      "step": 903
    },
    {
      "epoch": 0.6776611694152923,
      "grad_norm": 0.23745034772582727,
      "learning_rate": 0.00019052500685414348,
      "loss": 0.0448,
      "step": 904
    },
    {
      "epoch": 0.6784107946026986,
      "grad_norm": 0.2186670327549479,
      "learning_rate": 0.00019048790500248325,
      "loss": 0.0382,
      "step": 905
    },
    {
      "epoch": 0.679160419790105,
      "grad_norm": 0.23083494649569086,
      "learning_rate": 0.00019045073427861469,
      "loss": 0.0534,
      "step": 906
    },
    {
      "epoch": 0.6799100449775113,
      "grad_norm": 0.20107179690681276,
      "learning_rate": 0.00019041349471082922,
      "loss": 0.036,
      "step": 907
    },
    {
      "epoch": 0.6806596701649176,
      "grad_norm": 0.14110579856132485,
      "learning_rate": 0.0001903761863274706,
      "loss": 0.0267,
      "step": 908
    },
    {
      "epoch": 0.6814092953523239,
      "grad_norm": 0.11737441028479874,
      "learning_rate": 0.000190338809156935,
      "loss": 0.013,
      "step": 909
    },
    {
      "epoch": 0.6821589205397302,
      "grad_norm": 0.3354065360433217,
      "learning_rate": 0.00019030136322767104,
      "loss": 0.0503,
      "step": 910
    },
    {
      "epoch": 0.6829085457271364,
      "grad_norm": 0.2933379028903616,
      "learning_rate": 0.0001902638485681795,
      "loss": 0.0484,
      "step": 911
    },
    {
      "epoch": 0.6836581709145427,
      "grad_norm": 0.2772432648087348,
      "learning_rate": 0.00019022626520701357,
      "loss": 0.0501,
      "step": 912
    },
    {
      "epoch": 0.684407796101949,
      "grad_norm": 0.06103394009376496,
      "learning_rate": 0.00019018861317277876,
      "loss": 0.014,
      "step": 913
    },
    {
      "epoch": 0.6851574212893553,
      "grad_norm": 0.30436820029229084,
      "learning_rate": 0.00019015089249413278,
      "loss": 0.0227,
      "step": 914
    },
    {
      "epoch": 0.6859070464767616,
      "grad_norm": 0.21316900529347577,
      "learning_rate": 0.00019011310319978566,
      "loss": 0.0301,
      "step": 915
    },
    {
      "epoch": 0.6866566716641679,
      "grad_norm": 0.15624463828972068,
      "learning_rate": 0.00019007524531849954,
      "loss": 0.0175,
      "step": 916
    },
    {
      "epoch": 0.6874062968515742,
      "grad_norm": 0.21724546668213623,
      "learning_rate": 0.00019003731887908888,
      "loss": 0.0419,
      "step": 917
    },
    {
      "epoch": 0.6881559220389805,
      "grad_norm": 0.17497708434454778,
      "learning_rate": 0.00018999932391042026,
      "loss": 0.0494,
      "step": 918
    },
    {
      "epoch": 0.6889055472263869,
      "grad_norm": 0.28296123614292606,
      "learning_rate": 0.00018996126044141243,
      "loss": 0.0542,
      "step": 919
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 0.3138450154926571,
      "learning_rate": 0.00018992312850103633,
      "loss": 0.0451,
      "step": 920
    },
    {
      "epoch": 0.6904047976011994,
      "grad_norm": 0.23985575635803402,
      "learning_rate": 0.00018988492811831487,
      "loss": 0.0425,
      "step": 921
    },
    {
      "epoch": 0.6911544227886057,
      "grad_norm": 0.28213195649994594,
      "learning_rate": 0.00018984665932232324,
      "loss": 0.0432,
      "step": 922
    },
    {
      "epoch": 0.691904047976012,
      "grad_norm": 0.17556584000730519,
      "learning_rate": 0.00018980832214218854,
      "loss": 0.0417,
      "step": 923
    },
    {
      "epoch": 0.6926536731634183,
      "grad_norm": 0.2153080576912875,
      "learning_rate": 0.00018976991660709,
      "loss": 0.0248,
      "step": 924
    },
    {
      "epoch": 0.6934032983508246,
      "grad_norm": 0.21085999350466364,
      "learning_rate": 0.0001897314427462589,
      "loss": 0.039,
      "step": 925
    },
    {
      "epoch": 0.6941529235382309,
      "grad_norm": 0.23685975191705796,
      "learning_rate": 0.00018969290058897849,
      "loss": 0.0416,
      "step": 926
    },
    {
      "epoch": 0.6949025487256372,
      "grad_norm": 0.2349320329159903,
      "learning_rate": 0.00018965429016458397,
      "loss": 0.0485,
      "step": 927
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.2003843750715276,
      "learning_rate": 0.00018961561150246254,
      "loss": 0.0248,
      "step": 928
    },
    {
      "epoch": 0.6964017991004497,
      "grad_norm": 0.17207195243686713,
      "learning_rate": 0.00018957686463205332,
      "loss": 0.0393,
      "step": 929
    },
    {
      "epoch": 0.697151424287856,
      "grad_norm": 0.10409802702518689,
      "learning_rate": 0.00018953804958284732,
      "loss": 0.0143,
      "step": 930
    },
    {
      "epoch": 0.6979010494752623,
      "grad_norm": 0.2834877094860702,
      "learning_rate": 0.00018949916638438753,
      "loss": 0.0423,
      "step": 931
    },
    {
      "epoch": 0.6986506746626686,
      "grad_norm": 0.16706698433341532,
      "learning_rate": 0.00018946021506626874,
      "loss": 0.0311,
      "step": 932
    },
    {
      "epoch": 0.699400299850075,
      "grad_norm": 0.27550981150523285,
      "learning_rate": 0.0001894211956581376,
      "loss": 0.0485,
      "step": 933
    },
    {
      "epoch": 0.7001499250374813,
      "grad_norm": 0.5822939921119175,
      "learning_rate": 0.00018938210818969254,
      "loss": 0.0735,
      "step": 934
    },
    {
      "epoch": 0.7008995502248876,
      "grad_norm": 0.15172545323797987,
      "learning_rate": 0.00018934295269068391,
      "loss": 0.0313,
      "step": 935
    },
    {
      "epoch": 0.7016491754122939,
      "grad_norm": 0.19155910351663208,
      "learning_rate": 0.00018930372919091372,
      "loss": 0.036,
      "step": 936
    },
    {
      "epoch": 0.7023988005997002,
      "grad_norm": 0.1920009030641335,
      "learning_rate": 0.0001892644377202358,
      "loss": 0.0303,
      "step": 937
    },
    {
      "epoch": 0.7031484257871065,
      "grad_norm": 0.2315620580933706,
      "learning_rate": 0.0001892250783085557,
      "loss": 0.0564,
      "step": 938
    },
    {
      "epoch": 0.7038980509745127,
      "grad_norm": 0.2277782259514648,
      "learning_rate": 0.00018918565098583064,
      "loss": 0.0482,
      "step": 939
    },
    {
      "epoch": 0.704647676161919,
      "grad_norm": 0.1290013927843416,
      "learning_rate": 0.00018914615578206965,
      "loss": 0.0374,
      "step": 940
    },
    {
      "epoch": 0.7053973013493253,
      "grad_norm": 0.3096012036471806,
      "learning_rate": 0.00018910659272733325,
      "loss": 0.0425,
      "step": 941
    },
    {
      "epoch": 0.7061469265367316,
      "grad_norm": 0.0962987698303328,
      "learning_rate": 0.00018906696185173373,
      "loss": 0.0122,
      "step": 942
    },
    {
      "epoch": 0.7068965517241379,
      "grad_norm": 0.6081649862961295,
      "learning_rate": 0.000189027263185435,
      "loss": 0.0342,
      "step": 943
    },
    {
      "epoch": 0.7076461769115442,
      "grad_norm": 0.34556925423240414,
      "learning_rate": 0.0001889874967586525,
      "loss": 0.035,
      "step": 944
    },
    {
      "epoch": 0.7083958020989505,
      "grad_norm": 0.25383191175507674,
      "learning_rate": 0.00018894766260165328,
      "loss": 0.0301,
      "step": 945
    },
    {
      "epoch": 0.7091454272863568,
      "grad_norm": 0.20690331301143616,
      "learning_rate": 0.00018890776074475594,
      "loss": 0.0309,
      "step": 946
    },
    {
      "epoch": 0.7098950524737632,
      "grad_norm": 0.11773090179399598,
      "learning_rate": 0.00018886779121833063,
      "loss": 0.0306,
      "step": 947
    },
    {
      "epoch": 0.7106446776611695,
      "grad_norm": 0.5402748502999926,
      "learning_rate": 0.0001888277540527989,
      "loss": 0.0838,
      "step": 948
    },
    {
      "epoch": 0.7113943028485757,
      "grad_norm": 0.13353170834904762,
      "learning_rate": 0.00018878764927863397,
      "loss": 0.0243,
      "step": 949
    },
    {
      "epoch": 0.712143928035982,
      "grad_norm": 0.30817237528562064,
      "learning_rate": 0.00018874747692636032,
      "loss": 0.0443,
      "step": 950
    },
    {
      "epoch": 0.7128935532233883,
      "grad_norm": 0.31287844891395794,
      "learning_rate": 0.00018870723702655402,
      "loss": 0.0427,
      "step": 951
    },
    {
      "epoch": 0.7136431784107946,
      "grad_norm": 0.13038883844419452,
      "learning_rate": 0.00018866692960984246,
      "loss": 0.0267,
      "step": 952
    },
    {
      "epoch": 0.7143928035982009,
      "grad_norm": 0.1564690576279143,
      "learning_rate": 0.0001886265547069044,
      "loss": 0.0321,
      "step": 953
    },
    {
      "epoch": 0.7151424287856072,
      "grad_norm": 0.18995784247308192,
      "learning_rate": 0.0001885861123484701,
      "loss": 0.0403,
      "step": 954
    },
    {
      "epoch": 0.7158920539730135,
      "grad_norm": 0.1130067211033811,
      "learning_rate": 0.000188545602565321,
      "loss": 0.0195,
      "step": 955
    },
    {
      "epoch": 0.7166416791604198,
      "grad_norm": 0.18772400417992988,
      "learning_rate": 0.00018850502538828997,
      "loss": 0.0264,
      "step": 956
    },
    {
      "epoch": 0.717391304347826,
      "grad_norm": 0.3597706845751724,
      "learning_rate": 0.0001884643808482611,
      "loss": 0.0346,
      "step": 957
    },
    {
      "epoch": 0.7181409295352323,
      "grad_norm": 0.15381224969583301,
      "learning_rate": 0.0001884236689761698,
      "loss": 0.0204,
      "step": 958
    },
    {
      "epoch": 0.7188905547226386,
      "grad_norm": 0.16678496756982067,
      "learning_rate": 0.00018838288980300275,
      "loss": 0.0238,
      "step": 959
    },
    {
      "epoch": 0.719640179910045,
      "grad_norm": 0.11629607248863856,
      "learning_rate": 0.00018834204335979777,
      "loss": 0.0164,
      "step": 960
    },
    {
      "epoch": 0.7203898050974513,
      "grad_norm": 0.12667186231323446,
      "learning_rate": 0.00018830112967764396,
      "loss": 0.0238,
      "step": 961
    },
    {
      "epoch": 0.7211394302848576,
      "grad_norm": 0.22995623152007172,
      "learning_rate": 0.00018826014878768157,
      "loss": 0.0164,
      "step": 962
    },
    {
      "epoch": 0.7218890554722639,
      "grad_norm": 0.2358255161333237,
      "learning_rate": 0.00018821910072110197,
      "loss": 0.0331,
      "step": 963
    },
    {
      "epoch": 0.7226386806596702,
      "grad_norm": 0.3438825305304082,
      "learning_rate": 0.00018817798550914772,
      "loss": 0.0322,
      "step": 964
    },
    {
      "epoch": 0.7233883058470765,
      "grad_norm": 0.20848681012820786,
      "learning_rate": 0.00018813680318311243,
      "loss": 0.0454,
      "step": 965
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 0.15272896327846594,
      "learning_rate": 0.00018809555377434078,
      "loss": 0.0335,
      "step": 966
    },
    {
      "epoch": 0.724887556221889,
      "grad_norm": 0.23046040770462634,
      "learning_rate": 0.00018805423731422868,
      "loss": 0.029,
      "step": 967
    },
    {
      "epoch": 0.7256371814092953,
      "grad_norm": 0.29532647572282444,
      "learning_rate": 0.00018801285383422277,
      "loss": 0.0653,
      "step": 968
    },
    {
      "epoch": 0.7263868065967016,
      "grad_norm": 0.25954608236551174,
      "learning_rate": 0.00018797140336582096,
      "loss": 0.0299,
      "step": 969
    },
    {
      "epoch": 0.7271364317841079,
      "grad_norm": 0.15951338875544732,
      "learning_rate": 0.00018792988594057208,
      "loss": 0.0158,
      "step": 970
    },
    {
      "epoch": 0.7278860569715142,
      "grad_norm": 0.19570079875891358,
      "learning_rate": 0.00018788830159007582,
      "loss": 0.0209,
      "step": 971
    },
    {
      "epoch": 0.7286356821589205,
      "grad_norm": 0.27173924703759006,
      "learning_rate": 0.00018784665034598294,
      "loss": 0.0526,
      "step": 972
    },
    {
      "epoch": 0.7293853073463268,
      "grad_norm": 0.20541123116423304,
      "learning_rate": 0.00018780493223999507,
      "loss": 0.0541,
      "step": 973
    },
    {
      "epoch": 0.7301349325337332,
      "grad_norm": 0.08279968415297809,
      "learning_rate": 0.0001877631473038647,
      "loss": 0.014,
      "step": 974
    },
    {
      "epoch": 0.7308845577211395,
      "grad_norm": 0.261041791157805,
      "learning_rate": 0.00018772129556939517,
      "loss": 0.0339,
      "step": 975
    },
    {
      "epoch": 0.7316341829085458,
      "grad_norm": 0.20121706293886343,
      "learning_rate": 0.0001876793770684407,
      "loss": 0.0383,
      "step": 976
    },
    {
      "epoch": 0.732383808095952,
      "grad_norm": 0.22576321461521023,
      "learning_rate": 0.00018763739183290642,
      "loss": 0.0471,
      "step": 977
    },
    {
      "epoch": 0.7331334332833583,
      "grad_norm": 0.20509942958433353,
      "learning_rate": 0.00018759533989474807,
      "loss": 0.0559,
      "step": 978
    },
    {
      "epoch": 0.7338830584707646,
      "grad_norm": 0.13069410558257796,
      "learning_rate": 0.00018755322128597226,
      "loss": 0.0237,
      "step": 979
    },
    {
      "epoch": 0.7346326836581709,
      "grad_norm": 0.4352376492807258,
      "learning_rate": 0.00018751103603863634,
      "loss": 0.0483,
      "step": 980
    },
    {
      "epoch": 0.7353823088455772,
      "grad_norm": 0.4586041127351121,
      "learning_rate": 0.00018746878418484837,
      "loss": 0.0367,
      "step": 981
    },
    {
      "epoch": 0.7361319340329835,
      "grad_norm": 0.20081207301221854,
      "learning_rate": 0.0001874264657567671,
      "loss": 0.0373,
      "step": 982
    },
    {
      "epoch": 0.7368815592203898,
      "grad_norm": 0.17972854685564316,
      "learning_rate": 0.00018738408078660196,
      "loss": 0.0499,
      "step": 983
    },
    {
      "epoch": 0.7376311844077961,
      "grad_norm": 0.5123799767559905,
      "learning_rate": 0.00018734162930661303,
      "loss": 0.044,
      "step": 984
    },
    {
      "epoch": 0.7383808095952024,
      "grad_norm": 0.15387818818143484,
      "learning_rate": 0.00018729911134911099,
      "loss": 0.0247,
      "step": 985
    },
    {
      "epoch": 0.7391304347826086,
      "grad_norm": 0.15234012077579065,
      "learning_rate": 0.00018725652694645716,
      "loss": 0.0369,
      "step": 986
    },
    {
      "epoch": 0.739880059970015,
      "grad_norm": 0.0840260959966149,
      "learning_rate": 0.00018721387613106338,
      "loss": 0.0158,
      "step": 987
    },
    {
      "epoch": 0.7406296851574213,
      "grad_norm": 0.28193234623834884,
      "learning_rate": 0.00018717115893539208,
      "loss": 0.0595,
      "step": 988
    },
    {
      "epoch": 0.7413793103448276,
      "grad_norm": 0.2901103028757637,
      "learning_rate": 0.0001871283753919562,
      "loss": 0.0271,
      "step": 989
    },
    {
      "epoch": 0.7421289355322339,
      "grad_norm": 0.20023279045247946,
      "learning_rate": 0.00018708552553331918,
      "loss": 0.0284,
      "step": 990
    },
    {
      "epoch": 0.7428785607196402,
      "grad_norm": 0.16813357710639712,
      "learning_rate": 0.0001870426093920949,
      "loss": 0.0282,
      "step": 991
    },
    {
      "epoch": 0.7436281859070465,
      "grad_norm": 0.5817170450684567,
      "learning_rate": 0.00018699962700094772,
      "loss": 0.0557,
      "step": 992
    },
    {
      "epoch": 0.7443778110944528,
      "grad_norm": 0.23910035514534853,
      "learning_rate": 0.00018695657839259245,
      "loss": 0.0361,
      "step": 993
    },
    {
      "epoch": 0.7451274362818591,
      "grad_norm": 0.10757725591523912,
      "learning_rate": 0.00018691346359979427,
      "loss": 0.0148,
      "step": 994
    },
    {
      "epoch": 0.7458770614692654,
      "grad_norm": 0.2389873118034019,
      "learning_rate": 0.00018687028265536873,
      "loss": 0.0269,
      "step": 995
    },
    {
      "epoch": 0.7466266866566716,
      "grad_norm": 0.09890578148764179,
      "learning_rate": 0.00018682703559218174,
      "loss": 0.028,
      "step": 996
    },
    {
      "epoch": 0.7473763118440779,
      "grad_norm": 0.423335856348218,
      "learning_rate": 0.00018678372244314954,
      "loss": 0.0511,
      "step": 997
    },
    {
      "epoch": 0.7481259370314842,
      "grad_norm": 0.1664160505747008,
      "learning_rate": 0.00018674034324123865,
      "loss": 0.0178,
      "step": 998
    },
    {
      "epoch": 0.7488755622188905,
      "grad_norm": 0.30443104002016347,
      "learning_rate": 0.00018669689801946584,
      "loss": 0.0306,
      "step": 999
    },
    {
      "epoch": 0.7496251874062968,
      "grad_norm": 0.16915223787278677,
      "learning_rate": 0.0001866533868108982,
      "loss": 0.0202,
      "step": 1000
    },
    {
      "epoch": 0.7503748125937032,
      "grad_norm": 0.21806633940083398,
      "learning_rate": 0.000186609809648653,
      "loss": 0.0391,
      "step": 1001
    },
    {
      "epoch": 0.7511244377811095,
      "grad_norm": 0.18073732919950314,
      "learning_rate": 0.00018656616656589775,
      "loss": 0.0341,
      "step": 1002
    },
    {
      "epoch": 0.7518740629685158,
      "grad_norm": 0.32621454820715445,
      "learning_rate": 0.00018652245759585002,
      "loss": 0.0354,
      "step": 1003
    },
    {
      "epoch": 0.7526236881559221,
      "grad_norm": 0.17220315147771398,
      "learning_rate": 0.0001864786827717777,
      "loss": 0.0388,
      "step": 1004
    },
    {
      "epoch": 0.7533733133433284,
      "grad_norm": 0.19048101156044403,
      "learning_rate": 0.0001864348421269986,
      "loss": 0.0243,
      "step": 1005
    },
    {
      "epoch": 0.7541229385307346,
      "grad_norm": 0.1467422816576331,
      "learning_rate": 0.00018639093569488088,
      "loss": 0.0219,
      "step": 1006
    },
    {
      "epoch": 0.7548725637181409,
      "grad_norm": 0.17892458855046647,
      "learning_rate": 0.0001863469635088425,
      "loss": 0.0234,
      "step": 1007
    },
    {
      "epoch": 0.7556221889055472,
      "grad_norm": 1.2435338434659133,
      "learning_rate": 0.00018630292560235166,
      "loss": 0.109,
      "step": 1008
    },
    {
      "epoch": 0.7563718140929535,
      "grad_norm": 0.08933758885347165,
      "learning_rate": 0.00018625882200892652,
      "loss": 0.01,
      "step": 1009
    },
    {
      "epoch": 0.7571214392803598,
      "grad_norm": 0.15900071145370545,
      "learning_rate": 0.0001862146527621352,
      "loss": 0.0249,
      "step": 1010
    },
    {
      "epoch": 0.7578710644677661,
      "grad_norm": 0.22387635530307326,
      "learning_rate": 0.00018617041789559586,
      "loss": 0.0503,
      "step": 1011
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 0.2773550375385691,
      "learning_rate": 0.0001861261174429765,
      "loss": 0.0689,
      "step": 1012
    },
    {
      "epoch": 0.7593703148425787,
      "grad_norm": 0.13482706212842185,
      "learning_rate": 0.00018608175143799524,
      "loss": 0.0199,
      "step": 1013
    },
    {
      "epoch": 0.760119940029985,
      "grad_norm": 0.14016837389353834,
      "learning_rate": 0.0001860373199144198,
      "loss": 0.025,
      "step": 1014
    },
    {
      "epoch": 0.7608695652173914,
      "grad_norm": 0.15921551130271686,
      "learning_rate": 0.00018599282290606808,
      "loss": 0.0411,
      "step": 1015
    },
    {
      "epoch": 0.7616191904047976,
      "grad_norm": 0.18076247696963452,
      "learning_rate": 0.00018594826044680756,
      "loss": 0.042,
      "step": 1016
    },
    {
      "epoch": 0.7623688155922039,
      "grad_norm": 0.3393069856609448,
      "learning_rate": 0.00018590363257055565,
      "loss": 0.0518,
      "step": 1017
    },
    {
      "epoch": 0.7631184407796102,
      "grad_norm": 0.8923903150439438,
      "learning_rate": 0.0001858589393112796,
      "loss": 0.0684,
      "step": 1018
    },
    {
      "epoch": 0.7638680659670165,
      "grad_norm": 0.3761190414620271,
      "learning_rate": 0.00018581418070299637,
      "loss": 0.048,
      "step": 1019
    },
    {
      "epoch": 0.7646176911544228,
      "grad_norm": 0.18620916906321824,
      "learning_rate": 0.0001857693567797726,
      "loss": 0.057,
      "step": 1020
    },
    {
      "epoch": 0.7653673163418291,
      "grad_norm": 0.14050003177244677,
      "learning_rate": 0.00018572446757572476,
      "loss": 0.0281,
      "step": 1021
    },
    {
      "epoch": 0.7661169415292354,
      "grad_norm": 0.15336984527633807,
      "learning_rate": 0.0001856795131250189,
      "loss": 0.0468,
      "step": 1022
    },
    {
      "epoch": 0.7668665667166417,
      "grad_norm": 0.19002269472919547,
      "learning_rate": 0.00018563449346187084,
      "loss": 0.0209,
      "step": 1023
    },
    {
      "epoch": 0.767616191904048,
      "grad_norm": 0.3336582335944263,
      "learning_rate": 0.00018558940862054593,
      "loss": 0.0559,
      "step": 1024
    },
    {
      "epoch": 0.7683658170914542,
      "grad_norm": 0.19568831389842506,
      "learning_rate": 0.00018554425863535917,
      "loss": 0.0555,
      "step": 1025
    },
    {
      "epoch": 0.7691154422788605,
      "grad_norm": 0.3240431823925636,
      "learning_rate": 0.00018549904354067516,
      "loss": 0.0585,
      "step": 1026
    },
    {
      "epoch": 0.7698650674662668,
      "grad_norm": 0.10923976048976274,
      "learning_rate": 0.00018545376337090802,
      "loss": 0.017,
      "step": 1027
    },
    {
      "epoch": 0.7706146926536732,
      "grad_norm": 0.23005253382946222,
      "learning_rate": 0.0001854084181605214,
      "loss": 0.0421,
      "step": 1028
    },
    {
      "epoch": 0.7713643178410795,
      "grad_norm": 0.1420842462842462,
      "learning_rate": 0.00018536300794402854,
      "loss": 0.027,
      "step": 1029
    },
    {
      "epoch": 0.7721139430284858,
      "grad_norm": 0.2659412690343279,
      "learning_rate": 0.00018531753275599205,
      "loss": 0.0691,
      "step": 1030
    },
    {
      "epoch": 0.7728635682158921,
      "grad_norm": 0.19542245079333642,
      "learning_rate": 0.00018527199263102402,
      "loss": 0.0294,
      "step": 1031
    },
    {
      "epoch": 0.7736131934032984,
      "grad_norm": 0.2289918980003493,
      "learning_rate": 0.00018522638760378604,
      "loss": 0.0212,
      "step": 1032
    },
    {
      "epoch": 0.7743628185907047,
      "grad_norm": 0.2237723071198558,
      "learning_rate": 0.00018518071770898896,
      "loss": 0.0282,
      "step": 1033
    },
    {
      "epoch": 0.775112443778111,
      "grad_norm": 0.16713847253625994,
      "learning_rate": 0.00018513498298139308,
      "loss": 0.0431,
      "step": 1034
    },
    {
      "epoch": 0.7758620689655172,
      "grad_norm": 0.4190531497604175,
      "learning_rate": 0.00018508918345580812,
      "loss": 0.0977,
      "step": 1035
    },
    {
      "epoch": 0.7766116941529235,
      "grad_norm": 0.11647871866815543,
      "learning_rate": 0.00018504331916709295,
      "loss": 0.0288,
      "step": 1036
    },
    {
      "epoch": 0.7773613193403298,
      "grad_norm": 0.12692899870309066,
      "learning_rate": 0.00018499739015015592,
      "loss": 0.018,
      "step": 1037
    },
    {
      "epoch": 0.7781109445277361,
      "grad_norm": 0.25096730088442054,
      "learning_rate": 0.00018495139643995447,
      "loss": 0.0411,
      "step": 1038
    },
    {
      "epoch": 0.7788605697151424,
      "grad_norm": 0.12678759482102697,
      "learning_rate": 0.0001849053380714954,
      "loss": 0.0268,
      "step": 1039
    },
    {
      "epoch": 0.7796101949025487,
      "grad_norm": 0.21993853166525845,
      "learning_rate": 0.00018485921507983464,
      "loss": 0.0331,
      "step": 1040
    },
    {
      "epoch": 0.780359820089955,
      "grad_norm": 0.13735175115597095,
      "learning_rate": 0.00018481302750007742,
      "loss": 0.0208,
      "step": 1041
    },
    {
      "epoch": 0.7811094452773614,
      "grad_norm": 0.27636509366101925,
      "learning_rate": 0.000184766775367378,
      "loss": 0.0401,
      "step": 1042
    },
    {
      "epoch": 0.7818590704647677,
      "grad_norm": 0.08300196710365276,
      "learning_rate": 0.00018472045871693985,
      "loss": 0.0155,
      "step": 1043
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 0.10884016679085827,
      "learning_rate": 0.00018467407758401553,
      "loss": 0.022,
      "step": 1044
    },
    {
      "epoch": 0.7833583208395802,
      "grad_norm": 0.1414092080989368,
      "learning_rate": 0.00018462763200390668,
      "loss": 0.0176,
      "step": 1045
    },
    {
      "epoch": 0.7841079460269865,
      "grad_norm": 0.1592806396396988,
      "learning_rate": 0.00018458112201196397,
      "loss": 0.029,
      "step": 1046
    },
    {
      "epoch": 0.7848575712143928,
      "grad_norm": 0.3496710914903041,
      "learning_rate": 0.00018453454764358714,
      "loss": 0.0515,
      "step": 1047
    },
    {
      "epoch": 0.7856071964017991,
      "grad_norm": 0.19305827437982326,
      "learning_rate": 0.0001844879089342249,
      "loss": 0.0236,
      "step": 1048
    },
    {
      "epoch": 0.7863568215892054,
      "grad_norm": 0.2784510112379799,
      "learning_rate": 0.00018444120591937487,
      "loss": 0.0524,
      "step": 1049
    },
    {
      "epoch": 0.7871064467766117,
      "grad_norm": 0.16340562242653447,
      "learning_rate": 0.00018439443863458376,
      "loss": 0.036,
      "step": 1050
    },
    {
      "epoch": 0.787856071964018,
      "grad_norm": 0.09076231538746275,
      "learning_rate": 0.00018434760711544707,
      "loss": 0.0205,
      "step": 1051
    },
    {
      "epoch": 0.7886056971514243,
      "grad_norm": 0.10274725254355516,
      "learning_rate": 0.00018430071139760923,
      "loss": 0.019,
      "step": 1052
    },
    {
      "epoch": 0.7893553223388305,
      "grad_norm": 0.07424607061274292,
      "learning_rate": 0.00018425375151676356,
      "loss": 0.0124,
      "step": 1053
    },
    {
      "epoch": 0.7901049475262368,
      "grad_norm": 0.1478021118244442,
      "learning_rate": 0.0001842067275086522,
      "loss": 0.0315,
      "step": 1054
    },
    {
      "epoch": 0.7908545727136432,
      "grad_norm": 0.34576822503473104,
      "learning_rate": 0.00018415963940906601,
      "loss": 0.0386,
      "step": 1055
    },
    {
      "epoch": 0.7916041979010495,
      "grad_norm": 0.32702451129931603,
      "learning_rate": 0.0001841124872538448,
      "loss": 0.0823,
      "step": 1056
    },
    {
      "epoch": 0.7923538230884558,
      "grad_norm": 0.14429418533170335,
      "learning_rate": 0.00018406527107887706,
      "loss": 0.0228,
      "step": 1057
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 0.24367175933694912,
      "learning_rate": 0.0001840179909200999,
      "loss": 0.0299,
      "step": 1058
    },
    {
      "epoch": 0.7938530734632684,
      "grad_norm": 0.6099095348127705,
      "learning_rate": 0.00018397064681349932,
      "loss": 0.0532,
      "step": 1059
    },
    {
      "epoch": 0.7946026986506747,
      "grad_norm": 0.1197860468696163,
      "learning_rate": 0.00018392323879510983,
      "loss": 0.0232,
      "step": 1060
    },
    {
      "epoch": 0.795352323838081,
      "grad_norm": 0.12722737968329859,
      "learning_rate": 0.00018387576690101465,
      "loss": 0.0201,
      "step": 1061
    },
    {
      "epoch": 0.7961019490254873,
      "grad_norm": 0.14544820117937343,
      "learning_rate": 0.00018382823116734568,
      "loss": 0.0506,
      "step": 1062
    },
    {
      "epoch": 0.7968515742128935,
      "grad_norm": 0.20201164576960434,
      "learning_rate": 0.00018378063163028327,
      "loss": 0.0333,
      "step": 1063
    },
    {
      "epoch": 0.7976011994002998,
      "grad_norm": 0.19983512293761604,
      "learning_rate": 0.00018373296832605647,
      "loss": 0.0316,
      "step": 1064
    },
    {
      "epoch": 0.7983508245877061,
      "grad_norm": 0.2025587075303726,
      "learning_rate": 0.00018368524129094277,
      "loss": 0.0404,
      "step": 1065
    },
    {
      "epoch": 0.7991004497751124,
      "grad_norm": 0.21875036758596228,
      "learning_rate": 0.00018363745056126818,
      "loss": 0.0409,
      "step": 1066
    },
    {
      "epoch": 0.7998500749625187,
      "grad_norm": 0.08751800766519974,
      "learning_rate": 0.00018358959617340722,
      "loss": 0.0134,
      "step": 1067
    },
    {
      "epoch": 0.800599700149925,
      "grad_norm": 0.17751840607101965,
      "learning_rate": 0.00018354167816378293,
      "loss": 0.0298,
      "step": 1068
    },
    {
      "epoch": 0.8013493253373314,
      "grad_norm": 0.13844585273110852,
      "learning_rate": 0.00018349369656886654,
      "loss": 0.0225,
      "step": 1069
    },
    {
      "epoch": 0.8020989505247377,
      "grad_norm": 0.14779825985182682,
      "learning_rate": 0.00018344565142517798,
      "loss": 0.0172,
      "step": 1070
    },
    {
      "epoch": 0.802848575712144,
      "grad_norm": 0.23329649714840747,
      "learning_rate": 0.00018339754276928532,
      "loss": 0.0166,
      "step": 1071
    },
    {
      "epoch": 0.8035982008995503,
      "grad_norm": 0.27924723804645607,
      "learning_rate": 0.00018334937063780506,
      "loss": 0.0455,
      "step": 1072
    },
    {
      "epoch": 0.8043478260869565,
      "grad_norm": 0.3767242325154973,
      "learning_rate": 0.000183301135067402,
      "loss": 0.0504,
      "step": 1073
    },
    {
      "epoch": 0.8050974512743628,
      "grad_norm": 0.23456829764371648,
      "learning_rate": 0.00018325283609478925,
      "loss": 0.0295,
      "step": 1074
    },
    {
      "epoch": 0.8058470764617691,
      "grad_norm": 0.0953486660630768,
      "learning_rate": 0.0001832044737567281,
      "loss": 0.0142,
      "step": 1075
    },
    {
      "epoch": 0.8065967016491754,
      "grad_norm": 0.328669019767791,
      "learning_rate": 0.0001831560480900282,
      "loss": 0.046,
      "step": 1076
    },
    {
      "epoch": 0.8073463268365817,
      "grad_norm": 0.13689582912930398,
      "learning_rate": 0.00018310755913154725,
      "loss": 0.0246,
      "step": 1077
    },
    {
      "epoch": 0.808095952023988,
      "grad_norm": 0.1414467884002127,
      "learning_rate": 0.0001830590069181912,
      "loss": 0.0223,
      "step": 1078
    },
    {
      "epoch": 0.8088455772113943,
      "grad_norm": 0.4175305732858486,
      "learning_rate": 0.0001830103914869142,
      "loss": 0.038,
      "step": 1079
    },
    {
      "epoch": 0.8095952023988006,
      "grad_norm": 0.31226288028632215,
      "learning_rate": 0.00018296171287471844,
      "loss": 0.0302,
      "step": 1080
    },
    {
      "epoch": 0.8103448275862069,
      "grad_norm": 0.326138105636252,
      "learning_rate": 0.00018291297111865416,
      "loss": 0.0307,
      "step": 1081
    },
    {
      "epoch": 0.8110944527736131,
      "grad_norm": 0.19855722071274357,
      "learning_rate": 0.00018286416625581974,
      "loss": 0.0333,
      "step": 1082
    },
    {
      "epoch": 0.8118440779610195,
      "grad_norm": 0.19808218582591067,
      "learning_rate": 0.00018281529832336158,
      "loss": 0.0128,
      "step": 1083
    },
    {
      "epoch": 0.8125937031484258,
      "grad_norm": 0.18000093916311433,
      "learning_rate": 0.0001827663673584741,
      "loss": 0.0375,
      "step": 1084
    },
    {
      "epoch": 0.8133433283358321,
      "grad_norm": 0.34396375971888243,
      "learning_rate": 0.0001827173733983996,
      "loss": 0.0665,
      "step": 1085
    },
    {
      "epoch": 0.8140929535232384,
      "grad_norm": 0.2262246608766294,
      "learning_rate": 0.0001826683164804284,
      "loss": 0.0308,
      "step": 1086
    },
    {
      "epoch": 0.8148425787106447,
      "grad_norm": 0.2962731621576649,
      "learning_rate": 0.0001826191966418988,
      "loss": 0.0696,
      "step": 1087
    },
    {
      "epoch": 0.815592203898051,
      "grad_norm": 0.20833559391810397,
      "learning_rate": 0.00018257001392019685,
      "loss": 0.0313,
      "step": 1088
    },
    {
      "epoch": 0.8163418290854573,
      "grad_norm": 0.15919323978485042,
      "learning_rate": 0.00018252076835275656,
      "loss": 0.0308,
      "step": 1089
    },
    {
      "epoch": 0.8170914542728636,
      "grad_norm": 0.16911598179212334,
      "learning_rate": 0.00018247145997705975,
      "loss": 0.0176,
      "step": 1090
    },
    {
      "epoch": 0.8178410794602698,
      "grad_norm": 0.19861291408359866,
      "learning_rate": 0.000182422088830636,
      "loss": 0.0255,
      "step": 1091
    },
    {
      "epoch": 0.8185907046476761,
      "grad_norm": 0.17452275737517106,
      "learning_rate": 0.0001823726549510628,
      "loss": 0.0324,
      "step": 1092
    },
    {
      "epoch": 0.8193403298350824,
      "grad_norm": 0.09352058354683734,
      "learning_rate": 0.00018232315837596517,
      "loss": 0.0103,
      "step": 1093
    },
    {
      "epoch": 0.8200899550224887,
      "grad_norm": 0.32286404722292333,
      "learning_rate": 0.00018227359914301606,
      "loss": 0.0449,
      "step": 1094
    },
    {
      "epoch": 0.820839580209895,
      "grad_norm": 0.1281920640041683,
      "learning_rate": 0.00018222397728993604,
      "loss": 0.0223,
      "step": 1095
    },
    {
      "epoch": 0.8215892053973014,
      "grad_norm": 0.11089662157992478,
      "learning_rate": 0.0001821742928544932,
      "loss": 0.0187,
      "step": 1096
    },
    {
      "epoch": 0.8223388305847077,
      "grad_norm": 0.287036679017573,
      "learning_rate": 0.0001821245458745035,
      "loss": 0.042,
      "step": 1097
    },
    {
      "epoch": 0.823088455772114,
      "grad_norm": 0.19024568124342148,
      "learning_rate": 0.00018207473638783038,
      "loss": 0.0227,
      "step": 1098
    },
    {
      "epoch": 0.8238380809595203,
      "grad_norm": 0.1254493943051864,
      "learning_rate": 0.00018202486443238486,
      "loss": 0.035,
      "step": 1099
    },
    {
      "epoch": 0.8245877061469266,
      "grad_norm": 0.2299310002082326,
      "learning_rate": 0.00018197493004612548,
      "loss": 0.0409,
      "step": 1100
    },
    {
      "epoch": 0.8253373313343328,
      "grad_norm": 0.20942758491518929,
      "learning_rate": 0.00018192493326705836,
      "loss": 0.0719,
      "step": 1101
    },
    {
      "epoch": 0.8260869565217391,
      "grad_norm": 0.2169193782357245,
      "learning_rate": 0.00018187487413323707,
      "loss": 0.0522,
      "step": 1102
    },
    {
      "epoch": 0.8268365817091454,
      "grad_norm": 0.10248145734203475,
      "learning_rate": 0.00018182475268276263,
      "loss": 0.0252,
      "step": 1103
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 0.14669484702742797,
      "learning_rate": 0.00018177456895378353,
      "loss": 0.0326,
      "step": 1104
    },
    {
      "epoch": 0.828335832083958,
      "grad_norm": 0.29384935141714214,
      "learning_rate": 0.00018172432298449565,
      "loss": 0.0334,
      "step": 1105
    },
    {
      "epoch": 0.8290854572713643,
      "grad_norm": 0.3435307574051933,
      "learning_rate": 0.00018167401481314222,
      "loss": 0.0447,
      "step": 1106
    },
    {
      "epoch": 0.8298350824587706,
      "grad_norm": 0.10926290483056297,
      "learning_rate": 0.0001816236444780138,
      "loss": 0.0254,
      "step": 1107
    },
    {
      "epoch": 0.8305847076461769,
      "grad_norm": 0.3592195940240025,
      "learning_rate": 0.0001815732120174483,
      "loss": 0.0432,
      "step": 1108
    },
    {
      "epoch": 0.8313343328335832,
      "grad_norm": 0.2745019580357286,
      "learning_rate": 0.00018152271746983093,
      "loss": 0.0317,
      "step": 1109
    },
    {
      "epoch": 0.8320839580209896,
      "grad_norm": 0.3719022292991417,
      "learning_rate": 0.00018147216087359414,
      "loss": 0.0554,
      "step": 1110
    },
    {
      "epoch": 0.8328335832083958,
      "grad_norm": 0.14829998786574425,
      "learning_rate": 0.00018142154226721754,
      "loss": 0.0267,
      "step": 1111
    },
    {
      "epoch": 0.8335832083958021,
      "grad_norm": 0.3984907269507194,
      "learning_rate": 0.00018137086168922806,
      "loss": 0.0392,
      "step": 1112
    },
    {
      "epoch": 0.8343328335832084,
      "grad_norm": 0.2696505935588497,
      "learning_rate": 0.00018132011917819968,
      "loss": 0.029,
      "step": 1113
    },
    {
      "epoch": 0.8350824587706147,
      "grad_norm": 0.16595550885495156,
      "learning_rate": 0.0001812693147727536,
      "loss": 0.0366,
      "step": 1114
    },
    {
      "epoch": 0.835832083958021,
      "grad_norm": 0.11780495839903224,
      "learning_rate": 0.00018121844851155805,
      "loss": 0.0326,
      "step": 1115
    },
    {
      "epoch": 0.8365817091454273,
      "grad_norm": 0.30894718763924667,
      "learning_rate": 0.00018116752043332847,
      "loss": 0.0225,
      "step": 1116
    },
    {
      "epoch": 0.8373313343328336,
      "grad_norm": 0.2601722543581295,
      "learning_rate": 0.0001811165305768272,
      "loss": 0.0298,
      "step": 1117
    },
    {
      "epoch": 0.8380809595202399,
      "grad_norm": 0.1612494703238606,
      "learning_rate": 0.00018106547898086374,
      "loss": 0.0213,
      "step": 1118
    },
    {
      "epoch": 0.8388305847076462,
      "grad_norm": 0.18569002752088673,
      "learning_rate": 0.0001810143656842944,
      "loss": 0.0218,
      "step": 1119
    },
    {
      "epoch": 0.8395802098950524,
      "grad_norm": 0.19615867709287715,
      "learning_rate": 0.00018096319072602267,
      "loss": 0.0239,
      "step": 1120
    },
    {
      "epoch": 0.8403298350824587,
      "grad_norm": 0.11600381929740078,
      "learning_rate": 0.00018091195414499884,
      "loss": 0.0125,
      "step": 1121
    },
    {
      "epoch": 0.841079460269865,
      "grad_norm": 0.2012290615387593,
      "learning_rate": 0.0001808606559802201,
      "loss": 0.0356,
      "step": 1122
    },
    {
      "epoch": 0.8418290854572713,
      "grad_norm": 0.5241588035079642,
      "learning_rate": 0.0001808092962707306,
      "loss": 0.0557,
      "step": 1123
    },
    {
      "epoch": 0.8425787106446777,
      "grad_norm": 0.16468248800615976,
      "learning_rate": 0.00018075787505562115,
      "loss": 0.0174,
      "step": 1124
    },
    {
      "epoch": 0.843328335832084,
      "grad_norm": 0.17673252568720604,
      "learning_rate": 0.00018070639237402958,
      "loss": 0.0185,
      "step": 1125
    },
    {
      "epoch": 0.8440779610194903,
      "grad_norm": 0.15194993456361727,
      "learning_rate": 0.00018065484826514046,
      "loss": 0.0267,
      "step": 1126
    },
    {
      "epoch": 0.8448275862068966,
      "grad_norm": 0.44198164798151507,
      "learning_rate": 0.00018060324276818497,
      "loss": 0.0432,
      "step": 1127
    },
    {
      "epoch": 0.8455772113943029,
      "grad_norm": 0.11663777926382558,
      "learning_rate": 0.0001805515759224412,
      "loss": 0.0142,
      "step": 1128
    },
    {
      "epoch": 0.8463268365817092,
      "grad_norm": 0.21518598232120772,
      "learning_rate": 0.00018049984776723384,
      "loss": 0.0492,
      "step": 1129
    },
    {
      "epoch": 0.8470764617691154,
      "grad_norm": 0.4390773125412656,
      "learning_rate": 0.00018044805834193422,
      "loss": 0.0214,
      "step": 1130
    },
    {
      "epoch": 0.8478260869565217,
      "grad_norm": 0.2342293522852248,
      "learning_rate": 0.00018039620768596035,
      "loss": 0.0327,
      "step": 1131
    },
    {
      "epoch": 0.848575712143928,
      "grad_norm": 0.28434849206147805,
      "learning_rate": 0.00018034429583877682,
      "loss": 0.0468,
      "step": 1132
    },
    {
      "epoch": 0.8493253373313343,
      "grad_norm": 0.20845198697998343,
      "learning_rate": 0.00018029232283989485,
      "loss": 0.0512,
      "step": 1133
    },
    {
      "epoch": 0.8500749625187406,
      "grad_norm": 0.13380972738039215,
      "learning_rate": 0.0001802402887288721,
      "loss": 0.0347,
      "step": 1134
    },
    {
      "epoch": 0.8508245877061469,
      "grad_norm": 0.21020059893001336,
      "learning_rate": 0.00018018819354531288,
      "loss": 0.0324,
      "step": 1135
    },
    {
      "epoch": 0.8515742128935532,
      "grad_norm": 0.10604539154052423,
      "learning_rate": 0.00018013603732886782,
      "loss": 0.0229,
      "step": 1136
    },
    {
      "epoch": 0.8523238380809596,
      "grad_norm": 0.1316986405730497,
      "learning_rate": 0.00018008382011923418,
      "loss": 0.0289,
      "step": 1137
    },
    {
      "epoch": 0.8530734632683659,
      "grad_norm": 0.24477833124890808,
      "learning_rate": 0.00018003154195615547,
      "loss": 0.0377,
      "step": 1138
    },
    {
      "epoch": 0.8538230884557722,
      "grad_norm": 0.5055205643247289,
      "learning_rate": 0.00017997920287942174,
      "loss": 0.0709,
      "step": 1139
    },
    {
      "epoch": 0.8545727136431784,
      "grad_norm": 0.10917773052009153,
      "learning_rate": 0.0001799268029288693,
      "loss": 0.0134,
      "step": 1140
    },
    {
      "epoch": 0.8553223388305847,
      "grad_norm": 0.11921356496141748,
      "learning_rate": 0.00017987434214438084,
      "loss": 0.0171,
      "step": 1141
    },
    {
      "epoch": 0.856071964017991,
      "grad_norm": 0.09046442975402154,
      "learning_rate": 0.00017982182056588535,
      "loss": 0.0164,
      "step": 1142
    },
    {
      "epoch": 0.8568215892053973,
      "grad_norm": 0.7470385217785291,
      "learning_rate": 0.0001797692382333581,
      "loss": 0.0853,
      "step": 1143
    },
    {
      "epoch": 0.8575712143928036,
      "grad_norm": 0.3587711861829344,
      "learning_rate": 0.00017971659518682055,
      "loss": 0.069,
      "step": 1144
    },
    {
      "epoch": 0.8583208395802099,
      "grad_norm": 0.13478931125708365,
      "learning_rate": 0.00017966389146634043,
      "loss": 0.0267,
      "step": 1145
    },
    {
      "epoch": 0.8590704647676162,
      "grad_norm": 0.31830295239537176,
      "learning_rate": 0.00017961112711203166,
      "loss": 0.0338,
      "step": 1146
    },
    {
      "epoch": 0.8598200899550225,
      "grad_norm": 0.3826627534808154,
      "learning_rate": 0.0001795583021640542,
      "loss": 0.0386,
      "step": 1147
    },
    {
      "epoch": 0.8605697151424287,
      "grad_norm": 0.34627413661240874,
      "learning_rate": 0.00017950541666261427,
      "loss": 0.026,
      "step": 1148
    },
    {
      "epoch": 0.861319340329835,
      "grad_norm": 0.16387687839960238,
      "learning_rate": 0.0001794524706479641,
      "loss": 0.0315,
      "step": 1149
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 0.1641245742855337,
      "learning_rate": 0.000179399464160402,
      "loss": 0.0311,
      "step": 1150
    },
    {
      "epoch": 0.8628185907046477,
      "grad_norm": 0.07336979330054261,
      "learning_rate": 0.00017934639724027224,
      "loss": 0.0084,
      "step": 1151
    },
    {
      "epoch": 0.863568215892054,
      "grad_norm": 0.12978420808613153,
      "learning_rate": 0.0001792932699279652,
      "loss": 0.0331,
      "step": 1152
    },
    {
      "epoch": 0.8643178410794603,
      "grad_norm": 0.14286036892050807,
      "learning_rate": 0.00017924008226391717,
      "loss": 0.0317,
      "step": 1153
    },
    {
      "epoch": 0.8650674662668666,
      "grad_norm": 0.32962370915972483,
      "learning_rate": 0.00017918683428861035,
      "loss": 0.0382,
      "step": 1154
    },
    {
      "epoch": 0.8658170914542729,
      "grad_norm": 0.13110929982565675,
      "learning_rate": 0.0001791335260425729,
      "loss": 0.0223,
      "step": 1155
    },
    {
      "epoch": 0.8665667166416792,
      "grad_norm": 0.2966167078785812,
      "learning_rate": 0.00017908015756637884,
      "loss": 0.0402,
      "step": 1156
    },
    {
      "epoch": 0.8673163418290855,
      "grad_norm": 0.19862644988985317,
      "learning_rate": 0.00017902672890064797,
      "loss": 0.026,
      "step": 1157
    },
    {
      "epoch": 0.8680659670164917,
      "grad_norm": 0.19852351054120773,
      "learning_rate": 0.00017897324008604598,
      "loss": 0.0343,
      "step": 1158
    },
    {
      "epoch": 0.868815592203898,
      "grad_norm": 0.1875769862756106,
      "learning_rate": 0.0001789196911632843,
      "loss": 0.0231,
      "step": 1159
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 0.1897377931741348,
      "learning_rate": 0.0001788660821731201,
      "loss": 0.0286,
      "step": 1160
    },
    {
      "epoch": 0.8703148425787106,
      "grad_norm": 0.14919402045830019,
      "learning_rate": 0.0001788124131563563,
      "loss": 0.0234,
      "step": 1161
    },
    {
      "epoch": 0.8710644677661169,
      "grad_norm": 0.42801423004173667,
      "learning_rate": 0.0001787586841538415,
      "loss": 0.0477,
      "step": 1162
    },
    {
      "epoch": 0.8718140929535232,
      "grad_norm": 0.17922410832990773,
      "learning_rate": 0.00017870489520646992,
      "loss": 0.0339,
      "step": 1163
    },
    {
      "epoch": 0.8725637181409296,
      "grad_norm": 0.22900265727699864,
      "learning_rate": 0.00017865104635518151,
      "loss": 0.0542,
      "step": 1164
    },
    {
      "epoch": 0.8733133433283359,
      "grad_norm": 0.1694529110436179,
      "learning_rate": 0.00017859713764096166,
      "loss": 0.024,
      "step": 1165
    },
    {
      "epoch": 0.8740629685157422,
      "grad_norm": 0.3762369755923995,
      "learning_rate": 0.00017854316910484146,
      "loss": 0.034,
      "step": 1166
    },
    {
      "epoch": 0.8748125937031485,
      "grad_norm": 0.31175400181963114,
      "learning_rate": 0.00017848914078789742,
      "loss": 0.038,
      "step": 1167
    },
    {
      "epoch": 0.8755622188905547,
      "grad_norm": 0.3913855756756641,
      "learning_rate": 0.00017843505273125164,
      "loss": 0.0691,
      "step": 1168
    },
    {
      "epoch": 0.876311844077961,
      "grad_norm": 0.14525941244746013,
      "learning_rate": 0.00017838090497607166,
      "loss": 0.0239,
      "step": 1169
    },
    {
      "epoch": 0.8770614692653673,
      "grad_norm": 0.34454983943420076,
      "learning_rate": 0.00017832669756357044,
      "loss": 0.0374,
      "step": 1170
    },
    {
      "epoch": 0.8778110944527736,
      "grad_norm": 0.15201977953343096,
      "learning_rate": 0.00017827243053500632,
      "loss": 0.0215,
      "step": 1171
    },
    {
      "epoch": 0.8785607196401799,
      "grad_norm": 0.11462754614358202,
      "learning_rate": 0.00017821810393168312,
      "loss": 0.0186,
      "step": 1172
    },
    {
      "epoch": 0.8793103448275862,
      "grad_norm": 0.48064716216895115,
      "learning_rate": 0.00017816371779494992,
      "loss": 0.0497,
      "step": 1173
    },
    {
      "epoch": 0.8800599700149925,
      "grad_norm": 0.1236272833087385,
      "learning_rate": 0.00017810927216620111,
      "loss": 0.0196,
      "step": 1174
    },
    {
      "epoch": 0.8808095952023988,
      "grad_norm": 0.1709420115474228,
      "learning_rate": 0.00017805476708687639,
      "loss": 0.0222,
      "step": 1175
    },
    {
      "epoch": 0.881559220389805,
      "grad_norm": 0.2602585822615709,
      "learning_rate": 0.00017800020259846068,
      "loss": 0.0397,
      "step": 1176
    },
    {
      "epoch": 0.8823088455772113,
      "grad_norm": 0.19903973260321675,
      "learning_rate": 0.00017794557874248413,
      "loss": 0.0259,
      "step": 1177
    },
    {
      "epoch": 0.8830584707646177,
      "grad_norm": 0.40046335200360744,
      "learning_rate": 0.00017789089556052214,
      "loss": 0.0891,
      "step": 1178
    },
    {
      "epoch": 0.883808095952024,
      "grad_norm": 0.7922737046828083,
      "learning_rate": 0.00017783615309419518,
      "loss": 0.1839,
      "step": 1179
    },
    {
      "epoch": 0.8845577211394303,
      "grad_norm": 0.4024668107918972,
      "learning_rate": 0.00017778135138516883,
      "loss": 0.0494,
      "step": 1180
    },
    {
      "epoch": 0.8853073463268366,
      "grad_norm": 0.19193099715331405,
      "learning_rate": 0.00017772649047515383,
      "loss": 0.0535,
      "step": 1181
    },
    {
      "epoch": 0.8860569715142429,
      "grad_norm": 0.22104073938037833,
      "learning_rate": 0.00017767157040590595,
      "loss": 0.0285,
      "step": 1182
    },
    {
      "epoch": 0.8868065967016492,
      "grad_norm": 0.22765987584867606,
      "learning_rate": 0.000177616591219226,
      "loss": 0.0277,
      "step": 1183
    },
    {
      "epoch": 0.8875562218890555,
      "grad_norm": 0.2934152221276665,
      "learning_rate": 0.0001775615529569597,
      "loss": 0.0361,
      "step": 1184
    },
    {
      "epoch": 0.8883058470764618,
      "grad_norm": 0.28250721417463565,
      "learning_rate": 0.0001775064556609979,
      "loss": 0.0306,
      "step": 1185
    },
    {
      "epoch": 0.889055472263868,
      "grad_norm": 0.49569143870479365,
      "learning_rate": 0.00017745129937327622,
      "loss": 0.066,
      "step": 1186
    },
    {
      "epoch": 0.8898050974512743,
      "grad_norm": 0.24149842890659284,
      "learning_rate": 0.0001773960841357753,
      "loss": 0.0533,
      "step": 1187
    },
    {
      "epoch": 0.8905547226386806,
      "grad_norm": 0.23085852188961126,
      "learning_rate": 0.00017734080999052054,
      "loss": 0.0299,
      "step": 1188
    },
    {
      "epoch": 0.8913043478260869,
      "grad_norm": 0.21446160288818875,
      "learning_rate": 0.00017728547697958228,
      "loss": 0.0398,
      "step": 1189
    },
    {
      "epoch": 0.8920539730134932,
      "grad_norm": 0.16362181582236918,
      "learning_rate": 0.00017723008514507557,
      "loss": 0.0336,
      "step": 1190
    },
    {
      "epoch": 0.8928035982008995,
      "grad_norm": 0.22547577091360518,
      "learning_rate": 0.00017717463452916033,
      "loss": 0.0168,
      "step": 1191
    },
    {
      "epoch": 0.8935532233883059,
      "grad_norm": 0.41468130480243326,
      "learning_rate": 0.00017711912517404108,
      "loss": 0.033,
      "step": 1192
    },
    {
      "epoch": 0.8943028485757122,
      "grad_norm": 0.1367013504712274,
      "learning_rate": 0.00017706355712196722,
      "loss": 0.0227,
      "step": 1193
    },
    {
      "epoch": 0.8950524737631185,
      "grad_norm": 0.30838009069288647,
      "learning_rate": 0.00017700793041523273,
      "loss": 0.0256,
      "step": 1194
    },
    {
      "epoch": 0.8958020989505248,
      "grad_norm": 0.3643600538498598,
      "learning_rate": 0.00017695224509617615,
      "loss": 0.0264,
      "step": 1195
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 0.1773585712190774,
      "learning_rate": 0.00017689650120718083,
      "loss": 0.0206,
      "step": 1196
    },
    {
      "epoch": 0.8973013493253373,
      "grad_norm": 0.09466427230844304,
      "learning_rate": 0.00017684069879067457,
      "loss": 0.0164,
      "step": 1197
    },
    {
      "epoch": 0.8980509745127436,
      "grad_norm": 0.2478374046281075,
      "learning_rate": 0.00017678483788912972,
      "loss": 0.0345,
      "step": 1198
    },
    {
      "epoch": 0.8988005997001499,
      "grad_norm": 0.1835506836680278,
      "learning_rate": 0.00017672891854506314,
      "loss": 0.0314,
      "step": 1199
    },
    {
      "epoch": 0.8995502248875562,
      "grad_norm": 0.7432904340156298,
      "learning_rate": 0.00017667294080103623,
      "loss": 0.1117,
      "step": 1200
    },
    {
      "epoch": 0.9002998500749625,
      "grad_norm": 0.19874364295361727,
      "learning_rate": 0.00017661690469965478,
      "loss": 0.0422,
      "step": 1201
    },
    {
      "epoch": 0.9010494752623688,
      "grad_norm": 0.2525657860864409,
      "learning_rate": 0.00017656081028356906,
      "loss": 0.0285,
      "step": 1202
    },
    {
      "epoch": 0.9017991004497751,
      "grad_norm": 0.15560729781092428,
      "learning_rate": 0.00017650465759547364,
      "loss": 0.0271,
      "step": 1203
    },
    {
      "epoch": 0.9025487256371814,
      "grad_norm": 0.20738627856114195,
      "learning_rate": 0.0001764484466781075,
      "loss": 0.0411,
      "step": 1204
    },
    {
      "epoch": 0.9032983508245878,
      "grad_norm": 0.25912924094622075,
      "learning_rate": 0.00017639217757425398,
      "loss": 0.0519,
      "step": 1205
    },
    {
      "epoch": 0.904047976011994,
      "grad_norm": 0.20049698324538573,
      "learning_rate": 0.0001763358503267406,
      "loss": 0.0285,
      "step": 1206
    },
    {
      "epoch": 0.9047976011994003,
      "grad_norm": 0.07636996700298185,
      "learning_rate": 0.00017627946497843916,
      "loss": 0.0129,
      "step": 1207
    },
    {
      "epoch": 0.9055472263868066,
      "grad_norm": 0.5409845972277662,
      "learning_rate": 0.00017622302157226574,
      "loss": 0.0385,
      "step": 1208
    },
    {
      "epoch": 0.9062968515742129,
      "grad_norm": 0.1377468501829297,
      "learning_rate": 0.0001761665201511806,
      "loss": 0.0305,
      "step": 1209
    },
    {
      "epoch": 0.9070464767616192,
      "grad_norm": 0.14571322925380692,
      "learning_rate": 0.00017610996075818813,
      "loss": 0.0291,
      "step": 1210
    },
    {
      "epoch": 0.9077961019490255,
      "grad_norm": 0.17281825498288494,
      "learning_rate": 0.00017605334343633684,
      "loss": 0.0288,
      "step": 1211
    },
    {
      "epoch": 0.9085457271364318,
      "grad_norm": 0.15628256876301608,
      "learning_rate": 0.00017599666822871933,
      "loss": 0.02,
      "step": 1212
    },
    {
      "epoch": 0.9092953523238381,
      "grad_norm": 0.1626084608026224,
      "learning_rate": 0.0001759399351784723,
      "loss": 0.0203,
      "step": 1213
    },
    {
      "epoch": 0.9100449775112444,
      "grad_norm": 0.37674788969646644,
      "learning_rate": 0.00017588314432877638,
      "loss": 0.0397,
      "step": 1214
    },
    {
      "epoch": 0.9107946026986506,
      "grad_norm": 0.1760643627933756,
      "learning_rate": 0.00017582629572285633,
      "loss": 0.0284,
      "step": 1215
    },
    {
      "epoch": 0.9115442278860569,
      "grad_norm": 0.26890020727090713,
      "learning_rate": 0.00017576938940398076,
      "loss": 0.0317,
      "step": 1216
    },
    {
      "epoch": 0.9122938530734632,
      "grad_norm": 0.1377644628031859,
      "learning_rate": 0.00017571242541546223,
      "loss": 0.0073,
      "step": 1217
    },
    {
      "epoch": 0.9130434782608695,
      "grad_norm": 0.1248429742984794,
      "learning_rate": 0.0001756554038006572,
      "loss": 0.0175,
      "step": 1218
    },
    {
      "epoch": 0.9137931034482759,
      "grad_norm": 0.12377973795358645,
      "learning_rate": 0.00017559832460296604,
      "loss": 0.0073,
      "step": 1219
    },
    {
      "epoch": 0.9145427286356822,
      "grad_norm": 0.16832600619943616,
      "learning_rate": 0.0001755411878658329,
      "loss": 0.0355,
      "step": 1220
    },
    {
      "epoch": 0.9152923538230885,
      "grad_norm": 0.05515583642707854,
      "learning_rate": 0.00017548399363274565,
      "loss": 0.0049,
      "step": 1221
    },
    {
      "epoch": 0.9160419790104948,
      "grad_norm": 0.17631658341381273,
      "learning_rate": 0.00017542674194723608,
      "loss": 0.0353,
      "step": 1222
    },
    {
      "epoch": 0.9167916041979011,
      "grad_norm": 0.3291715350286628,
      "learning_rate": 0.0001753694328528796,
      "loss": 0.0463,
      "step": 1223
    },
    {
      "epoch": 0.9175412293853074,
      "grad_norm": 0.38547473671666316,
      "learning_rate": 0.00017531206639329533,
      "loss": 0.0348,
      "step": 1224
    },
    {
      "epoch": 0.9182908545727136,
      "grad_norm": 0.14435028574913475,
      "learning_rate": 0.0001752546426121461,
      "loss": 0.0158,
      "step": 1225
    },
    {
      "epoch": 0.9190404797601199,
      "grad_norm": 0.47189193230826404,
      "learning_rate": 0.00017519716155313828,
      "loss": 0.0408,
      "step": 1226
    },
    {
      "epoch": 0.9197901049475262,
      "grad_norm": 0.12512750733662412,
      "learning_rate": 0.00017513962326002193,
      "loss": 0.0164,
      "step": 1227
    },
    {
      "epoch": 0.9205397301349325,
      "grad_norm": 0.2270342675936685,
      "learning_rate": 0.0001750820277765906,
      "loss": 0.0451,
      "step": 1228
    },
    {
      "epoch": 0.9212893553223388,
      "grad_norm": 0.13846789875210425,
      "learning_rate": 0.00017502437514668143,
      "loss": 0.0241,
      "step": 1229
    },
    {
      "epoch": 0.9220389805097451,
      "grad_norm": 0.475479072807183,
      "learning_rate": 0.00017496666541417502,
      "loss": 0.0733,
      "step": 1230
    },
    {
      "epoch": 0.9227886056971514,
      "grad_norm": 0.18052319968526903,
      "learning_rate": 0.00017490889862299542,
      "loss": 0.0255,
      "step": 1231
    },
    {
      "epoch": 0.9235382308845578,
      "grad_norm": 0.19929368687111212,
      "learning_rate": 0.00017485107481711012,
      "loss": 0.0577,
      "step": 1232
    },
    {
      "epoch": 0.9242878560719641,
      "grad_norm": 0.42547844329682477,
      "learning_rate": 0.00017479319404053003,
      "loss": 0.0437,
      "step": 1233
    },
    {
      "epoch": 0.9250374812593704,
      "grad_norm": 0.18707097210297646,
      "learning_rate": 0.00017473525633730937,
      "loss": 0.0315,
      "step": 1234
    },
    {
      "epoch": 0.9257871064467766,
      "grad_norm": 0.5106290605728168,
      "learning_rate": 0.00017467726175154574,
      "loss": 0.0551,
      "step": 1235
    },
    {
      "epoch": 0.9265367316341829,
      "grad_norm": 0.5377982295452615,
      "learning_rate": 0.00017461921032738004,
      "loss": 0.0546,
      "step": 1236
    },
    {
      "epoch": 0.9272863568215892,
      "grad_norm": 0.18458832162076752,
      "learning_rate": 0.00017456110210899639,
      "loss": 0.0275,
      "step": 1237
    },
    {
      "epoch": 0.9280359820089955,
      "grad_norm": 0.29982500992143923,
      "learning_rate": 0.00017450293714062212,
      "loss": 0.041,
      "step": 1238
    },
    {
      "epoch": 0.9287856071964018,
      "grad_norm": 0.14029784308266852,
      "learning_rate": 0.00017444471546652782,
      "loss": 0.03,
      "step": 1239
    },
    {
      "epoch": 0.9295352323838081,
      "grad_norm": 0.1792643339212138,
      "learning_rate": 0.00017438643713102716,
      "loss": 0.043,
      "step": 1240
    },
    {
      "epoch": 0.9302848575712144,
      "grad_norm": 0.1536906615959804,
      "learning_rate": 0.000174328102178477,
      "loss": 0.0302,
      "step": 1241
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 0.11341392647383206,
      "learning_rate": 0.00017426971065327736,
      "loss": 0.0152,
      "step": 1242
    },
    {
      "epoch": 0.931784107946027,
      "grad_norm": 0.13502410688401909,
      "learning_rate": 0.0001742112625998711,
      "loss": 0.025,
      "step": 1243
    },
    {
      "epoch": 0.9325337331334332,
      "grad_norm": 0.1615541948769028,
      "learning_rate": 0.00017415275806274431,
      "loss": 0.0237,
      "step": 1244
    },
    {
      "epoch": 0.9332833583208395,
      "grad_norm": 0.17981864857011048,
      "learning_rate": 0.000174094197086426,
      "loss": 0.0506,
      "step": 1245
    },
    {
      "epoch": 0.9340329835082459,
      "grad_norm": 0.24992613967096078,
      "learning_rate": 0.0001740355797154881,
      "loss": 0.0409,
      "step": 1246
    },
    {
      "epoch": 0.9347826086956522,
      "grad_norm": 0.1892497454856021,
      "learning_rate": 0.00017397690599454553,
      "loss": 0.0415,
      "step": 1247
    },
    {
      "epoch": 0.9355322338830585,
      "grad_norm": 0.12789737367676784,
      "learning_rate": 0.00017391817596825604,
      "loss": 0.015,
      "step": 1248
    },
    {
      "epoch": 0.9362818590704648,
      "grad_norm": 0.2617865941401239,
      "learning_rate": 0.0001738593896813203,
      "loss": 0.044,
      "step": 1249
    },
    {
      "epoch": 0.9370314842578711,
      "grad_norm": 0.21519189097963973,
      "learning_rate": 0.00017380054717848172,
      "loss": 0.0262,
      "step": 1250
    },
    {
      "epoch": 0.9377811094452774,
      "grad_norm": 0.34850546167695395,
      "learning_rate": 0.0001737416485045266,
      "loss": 0.0831,
      "step": 1251
    },
    {
      "epoch": 0.9385307346326837,
      "grad_norm": 0.20079517269559174,
      "learning_rate": 0.00017368269370428388,
      "loss": 0.0394,
      "step": 1252
    },
    {
      "epoch": 0.93928035982009,
      "grad_norm": 0.11554891447368328,
      "learning_rate": 0.0001736236828226253,
      "loss": 0.0246,
      "step": 1253
    },
    {
      "epoch": 0.9400299850074962,
      "grad_norm": 0.22652384701168027,
      "learning_rate": 0.00017356461590446523,
      "loss": 0.0339,
      "step": 1254
    },
    {
      "epoch": 0.9407796101949025,
      "grad_norm": 0.231228100996135,
      "learning_rate": 0.00017350549299476072,
      "loss": 0.0541,
      "step": 1255
    },
    {
      "epoch": 0.9415292353823088,
      "grad_norm": 0.1731235239961964,
      "learning_rate": 0.0001734463141385115,
      "loss": 0.0231,
      "step": 1256
    },
    {
      "epoch": 0.9422788605697151,
      "grad_norm": 0.15592401116422774,
      "learning_rate": 0.00017338707938075972,
      "loss": 0.0329,
      "step": 1257
    },
    {
      "epoch": 0.9430284857571214,
      "grad_norm": 0.24034812804004377,
      "learning_rate": 0.00017332778876659027,
      "loss": 0.0316,
      "step": 1258
    },
    {
      "epoch": 0.9437781109445277,
      "grad_norm": 0.13463299904907405,
      "learning_rate": 0.00017326844234113038,
      "loss": 0.019,
      "step": 1259
    },
    {
      "epoch": 0.9445277361319341,
      "grad_norm": 0.11207675048093935,
      "learning_rate": 0.00017320904014954985,
      "loss": 0.0187,
      "step": 1260
    },
    {
      "epoch": 0.9452773613193404,
      "grad_norm": 0.323716163461882,
      "learning_rate": 0.00017314958223706094,
      "loss": 0.0279,
      "step": 1261
    },
    {
      "epoch": 0.9460269865067467,
      "grad_norm": 0.13799052831708086,
      "learning_rate": 0.00017309006864891827,
      "loss": 0.0194,
      "step": 1262
    },
    {
      "epoch": 0.946776611694153,
      "grad_norm": 0.21693844938529427,
      "learning_rate": 0.00017303049943041888,
      "loss": 0.0251,
      "step": 1263
    },
    {
      "epoch": 0.9475262368815592,
      "grad_norm": 0.135271477910859,
      "learning_rate": 0.00017297087462690215,
      "loss": 0.0183,
      "step": 1264
    },
    {
      "epoch": 0.9482758620689655,
      "grad_norm": 0.14842531923446786,
      "learning_rate": 0.00017291119428374966,
      "loss": 0.0149,
      "step": 1265
    },
    {
      "epoch": 0.9490254872563718,
      "grad_norm": 0.2968947241483329,
      "learning_rate": 0.00017285145844638542,
      "loss": 0.0402,
      "step": 1266
    },
    {
      "epoch": 0.9497751124437781,
      "grad_norm": 0.20604469545460485,
      "learning_rate": 0.00017279166716027559,
      "loss": 0.0287,
      "step": 1267
    },
    {
      "epoch": 0.9505247376311844,
      "grad_norm": 0.20275501855488534,
      "learning_rate": 0.00017273182047092855,
      "loss": 0.0118,
      "step": 1268
    },
    {
      "epoch": 0.9512743628185907,
      "grad_norm": 0.3031556099834936,
      "learning_rate": 0.00017267191842389482,
      "loss": 0.0535,
      "step": 1269
    },
    {
      "epoch": 0.952023988005997,
      "grad_norm": 0.1446261371898781,
      "learning_rate": 0.00017261196106476713,
      "loss": 0.0241,
      "step": 1270
    },
    {
      "epoch": 0.9527736131934033,
      "grad_norm": 0.4708395724862073,
      "learning_rate": 0.0001725519484391802,
      "loss": 0.0488,
      "step": 1271
    },
    {
      "epoch": 0.9535232383808095,
      "grad_norm": 0.2111295614057366,
      "learning_rate": 0.00017249188059281098,
      "loss": 0.0246,
      "step": 1272
    },
    {
      "epoch": 0.954272863568216,
      "grad_norm": 0.15133666839742155,
      "learning_rate": 0.0001724317575713782,
      "loss": 0.0279,
      "step": 1273
    },
    {
      "epoch": 0.9550224887556222,
      "grad_norm": 0.2522194496469408,
      "learning_rate": 0.0001723715794206429,
      "loss": 0.0621,
      "step": 1274
    },
    {
      "epoch": 0.9557721139430285,
      "grad_norm": 0.2112780955153208,
      "learning_rate": 0.00017231134618640774,
      "loss": 0.0475,
      "step": 1275
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 0.2880812439470075,
      "learning_rate": 0.00017225105791451756,
      "loss": 0.0386,
      "step": 1276
    },
    {
      "epoch": 0.9572713643178411,
      "grad_norm": 0.2982729221462802,
      "learning_rate": 0.00017219071465085904,
      "loss": 0.0377,
      "step": 1277
    },
    {
      "epoch": 0.9580209895052474,
      "grad_norm": 0.2542609303536709,
      "learning_rate": 0.00017213031644136063,
      "loss": 0.0449,
      "step": 1278
    },
    {
      "epoch": 0.9587706146926537,
      "grad_norm": 0.09945597804821016,
      "learning_rate": 0.00017206986333199266,
      "loss": 0.018,
      "step": 1279
    },
    {
      "epoch": 0.95952023988006,
      "grad_norm": 0.19852384643166057,
      "learning_rate": 0.00017200935536876722,
      "loss": 0.0308,
      "step": 1280
    },
    {
      "epoch": 0.9602698650674663,
      "grad_norm": 0.31496987120458725,
      "learning_rate": 0.00017194879259773815,
      "loss": 0.0279,
      "step": 1281
    },
    {
      "epoch": 0.9610194902548725,
      "grad_norm": 0.2644898077162191,
      "learning_rate": 0.00017188817506500113,
      "loss": 0.0227,
      "step": 1282
    },
    {
      "epoch": 0.9617691154422788,
      "grad_norm": 0.11459251760842831,
      "learning_rate": 0.00017182750281669328,
      "loss": 0.0245,
      "step": 1283
    },
    {
      "epoch": 0.9625187406296851,
      "grad_norm": 0.2852291865642535,
      "learning_rate": 0.00017176677589899358,
      "loss": 0.0434,
      "step": 1284
    },
    {
      "epoch": 0.9632683658170914,
      "grad_norm": 0.12222861445621339,
      "learning_rate": 0.00017170599435812253,
      "loss": 0.0188,
      "step": 1285
    },
    {
      "epoch": 0.9640179910044977,
      "grad_norm": 0.22805218335960845,
      "learning_rate": 0.0001716451582403422,
      "loss": 0.0394,
      "step": 1286
    },
    {
      "epoch": 0.9647676161919041,
      "grad_norm": 0.21051893238370895,
      "learning_rate": 0.0001715842675919562,
      "loss": 0.0372,
      "step": 1287
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 0.1634771437962494,
      "learning_rate": 0.00017152332245930967,
      "loss": 0.0259,
      "step": 1288
    },
    {
      "epoch": 0.9662668665667167,
      "grad_norm": 0.39264183246076684,
      "learning_rate": 0.00017146232288878918,
      "loss": 0.0498,
      "step": 1289
    },
    {
      "epoch": 0.967016491754123,
      "grad_norm": 0.07351653247369035,
      "learning_rate": 0.00017140126892682282,
      "loss": 0.0115,
      "step": 1290
    },
    {
      "epoch": 0.9677661169415293,
      "grad_norm": 0.1389315827033655,
      "learning_rate": 0.00017134016061987994,
      "loss": 0.0333,
      "step": 1291
    },
    {
      "epoch": 0.9685157421289355,
      "grad_norm": 0.07247032170621293,
      "learning_rate": 0.00017127899801447143,
      "loss": 0.008,
      "step": 1292
    },
    {
      "epoch": 0.9692653673163418,
      "grad_norm": 0.375106496168661,
      "learning_rate": 0.00017121778115714928,
      "loss": 0.0364,
      "step": 1293
    },
    {
      "epoch": 0.9700149925037481,
      "grad_norm": 0.11378529206241661,
      "learning_rate": 0.00017115651009450702,
      "loss": 0.0172,
      "step": 1294
    },
    {
      "epoch": 0.9707646176911544,
      "grad_norm": 0.14501111657727805,
      "learning_rate": 0.00017109518487317925,
      "loss": 0.0168,
      "step": 1295
    },
    {
      "epoch": 0.9715142428785607,
      "grad_norm": 0.14111436249266363,
      "learning_rate": 0.0001710338055398419,
      "loss": 0.0138,
      "step": 1296
    },
    {
      "epoch": 0.972263868065967,
      "grad_norm": 0.15700514015924621,
      "learning_rate": 0.000170972372141212,
      "loss": 0.0368,
      "step": 1297
    },
    {
      "epoch": 0.9730134932533733,
      "grad_norm": 0.09296493562932354,
      "learning_rate": 0.0001709108847240478,
      "loss": 0.0185,
      "step": 1298
    },
    {
      "epoch": 0.9737631184407796,
      "grad_norm": 0.1852179762312589,
      "learning_rate": 0.00017084934333514866,
      "loss": 0.0309,
      "step": 1299
    },
    {
      "epoch": 0.974512743628186,
      "grad_norm": 0.1591504875482633,
      "learning_rate": 0.00017078774802135494,
      "loss": 0.0428,
      "step": 1300
    },
    {
      "epoch": 0.9752623688155923,
      "grad_norm": 0.22413860967262877,
      "learning_rate": 0.00017072609882954817,
      "loss": 0.0583,
      "step": 1301
    },
    {
      "epoch": 0.9760119940029985,
      "grad_norm": 0.10757982979112977,
      "learning_rate": 0.00017066439580665078,
      "loss": 0.0171,
      "step": 1302
    },
    {
      "epoch": 0.9767616191904048,
      "grad_norm": 0.04943454155924863,
      "learning_rate": 0.00017060263899962622,
      "loss": 0.0053,
      "step": 1303
    },
    {
      "epoch": 0.9775112443778111,
      "grad_norm": 0.22475189797552722,
      "learning_rate": 0.00017054082845547886,
      "loss": 0.0248,
      "step": 1304
    },
    {
      "epoch": 0.9782608695652174,
      "grad_norm": 0.07570818993749279,
      "learning_rate": 0.000170478964221254,
      "loss": 0.0129,
      "step": 1305
    },
    {
      "epoch": 0.9790104947526237,
      "grad_norm": 0.10998906191991285,
      "learning_rate": 0.00017041704634403776,
      "loss": 0.0182,
      "step": 1306
    },
    {
      "epoch": 0.97976011994003,
      "grad_norm": 0.16522886925498625,
      "learning_rate": 0.00017035507487095714,
      "loss": 0.0357,
      "step": 1307
    },
    {
      "epoch": 0.9805097451274363,
      "grad_norm": 0.1231331490493064,
      "learning_rate": 0.0001702930498491799,
      "loss": 0.017,
      "step": 1308
    },
    {
      "epoch": 0.9812593703148426,
      "grad_norm": 0.17243252970067333,
      "learning_rate": 0.0001702309713259145,
      "loss": 0.0368,
      "step": 1309
    },
    {
      "epoch": 0.9820089955022488,
      "grad_norm": 0.1995148535367195,
      "learning_rate": 0.00017016883934841025,
      "loss": 0.0167,
      "step": 1310
    },
    {
      "epoch": 0.9827586206896551,
      "grad_norm": 0.22039155231203994,
      "learning_rate": 0.00017010665396395704,
      "loss": 0.0394,
      "step": 1311
    },
    {
      "epoch": 0.9835082458770614,
      "grad_norm": 0.20335893313531456,
      "learning_rate": 0.0001700444152198855,
      "loss": 0.0574,
      "step": 1312
    },
    {
      "epoch": 0.9842578710644677,
      "grad_norm": 0.40000935163353163,
      "learning_rate": 0.00016998212316356677,
      "loss": 0.0873,
      "step": 1313
    },
    {
      "epoch": 0.9850074962518741,
      "grad_norm": 0.2006157626929846,
      "learning_rate": 0.00016991977784241262,
      "loss": 0.0294,
      "step": 1314
    },
    {
      "epoch": 0.9857571214392804,
      "grad_norm": 0.3823245678319009,
      "learning_rate": 0.00016985737930387537,
      "loss": 0.0553,
      "step": 1315
    },
    {
      "epoch": 0.9865067466266867,
      "grad_norm": 0.16980459817111984,
      "learning_rate": 0.00016979492759544785,
      "loss": 0.0329,
      "step": 1316
    },
    {
      "epoch": 0.987256371814093,
      "grad_norm": 0.19056633968766426,
      "learning_rate": 0.00016973242276466333,
      "loss": 0.0375,
      "step": 1317
    },
    {
      "epoch": 0.9880059970014993,
      "grad_norm": 0.09073012938358295,
      "learning_rate": 0.00016966986485909548,
      "loss": 0.0188,
      "step": 1318
    },
    {
      "epoch": 0.9887556221889056,
      "grad_norm": 0.32659909705802087,
      "learning_rate": 0.0001696072539263585,
      "loss": 0.0364,
      "step": 1319
    },
    {
      "epoch": 0.9895052473763118,
      "grad_norm": 0.284602610836338,
      "learning_rate": 0.0001695445900141068,
      "loss": 0.0635,
      "step": 1320
    },
    {
      "epoch": 0.9902548725637181,
      "grad_norm": 0.20725592747179133,
      "learning_rate": 0.00016948187317003516,
      "loss": 0.0542,
      "step": 1321
    },
    {
      "epoch": 0.9910044977511244,
      "grad_norm": 0.20036234880761852,
      "learning_rate": 0.0001694191034418787,
      "loss": 0.0351,
      "step": 1322
    },
    {
      "epoch": 0.9917541229385307,
      "grad_norm": 0.25074475113235223,
      "learning_rate": 0.00016935628087741273,
      "loss": 0.0259,
      "step": 1323
    },
    {
      "epoch": 0.992503748125937,
      "grad_norm": 0.19747618705563696,
      "learning_rate": 0.00016929340552445282,
      "loss": 0.0343,
      "step": 1324
    },
    {
      "epoch": 0.9932533733133433,
      "grad_norm": 0.14414285780381583,
      "learning_rate": 0.00016923047743085467,
      "loss": 0.0223,
      "step": 1325
    },
    {
      "epoch": 0.9940029985007496,
      "grad_norm": 0.2723679181126613,
      "learning_rate": 0.00016916749664451416,
      "loss": 0.0294,
      "step": 1326
    },
    {
      "epoch": 0.9947526236881559,
      "grad_norm": 0.21451817983901414,
      "learning_rate": 0.0001691044632133673,
      "loss": 0.0484,
      "step": 1327
    },
    {
      "epoch": 0.9955022488755623,
      "grad_norm": 0.15711266739216587,
      "learning_rate": 0.00016904137718539005,
      "loss": 0.0298,
      "step": 1328
    },
    {
      "epoch": 0.9962518740629686,
      "grad_norm": 0.1379956124539861,
      "learning_rate": 0.00016897823860859856,
      "loss": 0.0144,
      "step": 1329
    },
    {
      "epoch": 0.9970014992503748,
      "grad_norm": 0.21097619659308514,
      "learning_rate": 0.00016891504753104887,
      "loss": 0.0264,
      "step": 1330
    },
    {
      "epoch": 0.9977511244377811,
      "grad_norm": 0.10797556881915736,
      "learning_rate": 0.00016885180400083702,
      "loss": 0.0168,
      "step": 1331
    },
    {
      "epoch": 0.9985007496251874,
      "grad_norm": 0.15790769385517053,
      "learning_rate": 0.00016878850806609897,
      "loss": 0.0274,
      "step": 1332
    },
    {
      "epoch": 0.9992503748125937,
      "grad_norm": 0.14995259627727942,
      "learning_rate": 0.00016872515977501055,
      "loss": 0.0338,
      "step": 1333
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.23881055263244724,
      "learning_rate": 0.0001686617591757874,
      "loss": 0.0331,
      "step": 1334
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.03587876632809639,
      "eval_runtime": 1844.658,
      "eval_samples_per_second": 5.625,
      "eval_steps_per_second": 0.703,
      "step": 1334
    },
    {
      "epoch": 1.0007496251874064,
      "grad_norm": 0.08185628595380386,
      "learning_rate": 0.00016859830631668513,
      "loss": 0.0095,
      "step": 1335
    },
    {
      "epoch": 1.0014992503748126,
      "grad_norm": 0.17881441136974335,
      "learning_rate": 0.0001685348012459989,
      "loss": 0.0363,
      "step": 1336
    },
    {
      "epoch": 1.002248875562219,
      "grad_norm": 0.13731889201420747,
      "learning_rate": 0.00016847124401206384,
      "loss": 0.0159,
      "step": 1337
    },
    {
      "epoch": 1.0029985007496252,
      "grad_norm": 0.17747397404715173,
      "learning_rate": 0.00016840763466325457,
      "loss": 0.029,
      "step": 1338
    },
    {
      "epoch": 1.0037481259370316,
      "grad_norm": 0.2158747869092809,
      "learning_rate": 0.00016834397324798554,
      "loss": 0.0296,
      "step": 1339
    },
    {
      "epoch": 1.0044977511244377,
      "grad_norm": 0.1762430900702256,
      "learning_rate": 0.00016828025981471074,
      "loss": 0.0246,
      "step": 1340
    },
    {
      "epoch": 1.0052473763118441,
      "grad_norm": 0.2791849833350819,
      "learning_rate": 0.00016821649441192379,
      "loss": 0.022,
      "step": 1341
    },
    {
      "epoch": 1.0059970014992503,
      "grad_norm": 0.21450121217367965,
      "learning_rate": 0.00016815267708815784,
      "loss": 0.038,
      "step": 1342
    },
    {
      "epoch": 1.0067466266866567,
      "grad_norm": 0.10815207403591505,
      "learning_rate": 0.0001680888078919856,
      "loss": 0.0172,
      "step": 1343
    },
    {
      "epoch": 1.0074962518740629,
      "grad_norm": 0.07090930239731541,
      "learning_rate": 0.00016802488687201917,
      "loss": 0.0103,
      "step": 1344
    },
    {
      "epoch": 1.0082458770614693,
      "grad_norm": 0.2098963791136399,
      "learning_rate": 0.0001679609140769102,
      "loss": 0.0361,
      "step": 1345
    },
    {
      "epoch": 1.0089955022488755,
      "grad_norm": 0.09496829420002569,
      "learning_rate": 0.00016789688955534966,
      "loss": 0.014,
      "step": 1346
    },
    {
      "epoch": 1.0097451274362819,
      "grad_norm": 0.5919358618067819,
      "learning_rate": 0.000167832813356068,
      "loss": 0.0555,
      "step": 1347
    },
    {
      "epoch": 1.0104947526236883,
      "grad_norm": 0.28060088970632135,
      "learning_rate": 0.00016776868552783488,
      "loss": 0.0309,
      "step": 1348
    },
    {
      "epoch": 1.0112443778110944,
      "grad_norm": 0.09028108389815097,
      "learning_rate": 0.00016770450611945935,
      "loss": 0.0158,
      "step": 1349
    },
    {
      "epoch": 1.0119940029985008,
      "grad_norm": 0.12178089448441314,
      "learning_rate": 0.00016764027517978961,
      "loss": 0.0207,
      "step": 1350
    },
    {
      "epoch": 1.012743628185907,
      "grad_norm": 0.3504490855417358,
      "learning_rate": 0.00016757599275771324,
      "loss": 0.0585,
      "step": 1351
    },
    {
      "epoch": 1.0134932533733134,
      "grad_norm": 0.15135554030374876,
      "learning_rate": 0.00016751165890215686,
      "loss": 0.0265,
      "step": 1352
    },
    {
      "epoch": 1.0142428785607196,
      "grad_norm": 0.1888391656646971,
      "learning_rate": 0.00016744727366208633,
      "loss": 0.0173,
      "step": 1353
    },
    {
      "epoch": 1.014992503748126,
      "grad_norm": 0.24059619446312613,
      "learning_rate": 0.00016738283708650652,
      "loss": 0.0111,
      "step": 1354
    },
    {
      "epoch": 1.0157421289355322,
      "grad_norm": 0.15515112646263388,
      "learning_rate": 0.0001673183492244615,
      "loss": 0.0183,
      "step": 1355
    },
    {
      "epoch": 1.0164917541229386,
      "grad_norm": 0.05867960140949975,
      "learning_rate": 0.00016725381012503427,
      "loss": 0.0084,
      "step": 1356
    },
    {
      "epoch": 1.0172413793103448,
      "grad_norm": 0.15773163802954165,
      "learning_rate": 0.0001671892198373469,
      "loss": 0.0253,
      "step": 1357
    },
    {
      "epoch": 1.0179910044977512,
      "grad_norm": 0.21478798582104225,
      "learning_rate": 0.00016712457841056038,
      "loss": 0.0236,
      "step": 1358
    },
    {
      "epoch": 1.0187406296851573,
      "grad_norm": 0.22090328678852036,
      "learning_rate": 0.0001670598858938746,
      "loss": 0.0173,
      "step": 1359
    },
    {
      "epoch": 1.0194902548725637,
      "grad_norm": 0.1499121573611819,
      "learning_rate": 0.0001669951423365284,
      "loss": 0.0255,
      "step": 1360
    },
    {
      "epoch": 1.02023988005997,
      "grad_norm": 0.1831800231984105,
      "learning_rate": 0.00016693034778779943,
      "loss": 0.0172,
      "step": 1361
    },
    {
      "epoch": 1.0209895052473763,
      "grad_norm": 0.1492805212650543,
      "learning_rate": 0.00016686550229700413,
      "loss": 0.013,
      "step": 1362
    },
    {
      "epoch": 1.0217391304347827,
      "grad_norm": 0.34305168405576464,
      "learning_rate": 0.00016680060591349775,
      "loss": 0.0373,
      "step": 1363
    },
    {
      "epoch": 1.0224887556221889,
      "grad_norm": 0.13975032211968622,
      "learning_rate": 0.00016673565868667432,
      "loss": 0.0161,
      "step": 1364
    },
    {
      "epoch": 1.0232383808095953,
      "grad_norm": 0.3064630006414311,
      "learning_rate": 0.00016667066066596643,
      "loss": 0.0496,
      "step": 1365
    },
    {
      "epoch": 1.0239880059970015,
      "grad_norm": 0.0826424133342666,
      "learning_rate": 0.00016660561190084546,
      "loss": 0.0125,
      "step": 1366
    },
    {
      "epoch": 1.0247376311844079,
      "grad_norm": 0.1382847007841654,
      "learning_rate": 0.00016654051244082137,
      "loss": 0.0135,
      "step": 1367
    },
    {
      "epoch": 1.025487256371814,
      "grad_norm": 0.14488556357441107,
      "learning_rate": 0.00016647536233544265,
      "loss": 0.0159,
      "step": 1368
    },
    {
      "epoch": 1.0262368815592204,
      "grad_norm": 0.2429422011210165,
      "learning_rate": 0.00016641016163429644,
      "loss": 0.0261,
      "step": 1369
    },
    {
      "epoch": 1.0269865067466266,
      "grad_norm": 0.09274489305019605,
      "learning_rate": 0.0001663449103870083,
      "loss": 0.0127,
      "step": 1370
    },
    {
      "epoch": 1.027736131934033,
      "grad_norm": 0.13706007234801748,
      "learning_rate": 0.00016627960864324233,
      "loss": 0.0186,
      "step": 1371
    },
    {
      "epoch": 1.0284857571214392,
      "grad_norm": 0.1667184859979947,
      "learning_rate": 0.00016621425645270099,
      "loss": 0.0469,
      "step": 1372
    },
    {
      "epoch": 1.0292353823088456,
      "grad_norm": 0.13711825824543697,
      "learning_rate": 0.00016614885386512525,
      "loss": 0.017,
      "step": 1373
    },
    {
      "epoch": 1.0299850074962518,
      "grad_norm": 0.15682613258083772,
      "learning_rate": 0.00016608340093029428,
      "loss": 0.0153,
      "step": 1374
    },
    {
      "epoch": 1.0307346326836582,
      "grad_norm": 0.09438959849052585,
      "learning_rate": 0.0001660178976980257,
      "loss": 0.0099,
      "step": 1375
    },
    {
      "epoch": 1.0314842578710646,
      "grad_norm": 0.15132109568049978,
      "learning_rate": 0.0001659523442181754,
      "loss": 0.0197,
      "step": 1376
    },
    {
      "epoch": 1.0322338830584707,
      "grad_norm": 0.08767583362202569,
      "learning_rate": 0.00016588674054063743,
      "loss": 0.011,
      "step": 1377
    },
    {
      "epoch": 1.0329835082458771,
      "grad_norm": 0.21411442763454988,
      "learning_rate": 0.0001658210867153441,
      "loss": 0.0137,
      "step": 1378
    },
    {
      "epoch": 1.0337331334332833,
      "grad_norm": 0.1454566222783013,
      "learning_rate": 0.00016575538279226594,
      "loss": 0.0146,
      "step": 1379
    },
    {
      "epoch": 1.0344827586206897,
      "grad_norm": 0.13381699143135503,
      "learning_rate": 0.00016568962882141156,
      "loss": 0.0219,
      "step": 1380
    },
    {
      "epoch": 1.035232383808096,
      "grad_norm": 0.2426147235827075,
      "learning_rate": 0.00016562382485282758,
      "loss": 0.0184,
      "step": 1381
    },
    {
      "epoch": 1.0359820089955023,
      "grad_norm": 0.6098398132822853,
      "learning_rate": 0.00016555797093659882,
      "loss": 0.0134,
      "step": 1382
    },
    {
      "epoch": 1.0367316341829085,
      "grad_norm": 0.18837425350082296,
      "learning_rate": 0.00016549206712284807,
      "loss": 0.0155,
      "step": 1383
    },
    {
      "epoch": 1.0374812593703149,
      "grad_norm": 0.16473229178818735,
      "learning_rate": 0.000165426113461736,
      "loss": 0.0257,
      "step": 1384
    },
    {
      "epoch": 1.038230884557721,
      "grad_norm": 0.23267668027297075,
      "learning_rate": 0.00016536011000346137,
      "loss": 0.0318,
      "step": 1385
    },
    {
      "epoch": 1.0389805097451275,
      "grad_norm": 0.16538388110134583,
      "learning_rate": 0.00016529405679826075,
      "loss": 0.029,
      "step": 1386
    },
    {
      "epoch": 1.0397301349325336,
      "grad_norm": 0.2169175282271481,
      "learning_rate": 0.00016522795389640858,
      "loss": 0.0146,
      "step": 1387
    },
    {
      "epoch": 1.04047976011994,
      "grad_norm": 0.1584369960804152,
      "learning_rate": 0.00016516180134821718,
      "loss": 0.0293,
      "step": 1388
    },
    {
      "epoch": 1.0412293853073464,
      "grad_norm": 0.22755520337638765,
      "learning_rate": 0.0001650955992040366,
      "loss": 0.0219,
      "step": 1389
    },
    {
      "epoch": 1.0419790104947526,
      "grad_norm": 0.13611778505409552,
      "learning_rate": 0.00016502934751425467,
      "loss": 0.0307,
      "step": 1390
    },
    {
      "epoch": 1.042728635682159,
      "grad_norm": 0.3217127284596519,
      "learning_rate": 0.00016496304632929686,
      "loss": 0.014,
      "step": 1391
    },
    {
      "epoch": 1.0434782608695652,
      "grad_norm": 0.16275994278824357,
      "learning_rate": 0.00016489669569962637,
      "loss": 0.0177,
      "step": 1392
    },
    {
      "epoch": 1.0442278860569716,
      "grad_norm": 0.36007736364563725,
      "learning_rate": 0.00016483029567574412,
      "loss": 0.0148,
      "step": 1393
    },
    {
      "epoch": 1.0449775112443778,
      "grad_norm": 0.31673298912243997,
      "learning_rate": 0.00016476384630818847,
      "loss": 0.0311,
      "step": 1394
    },
    {
      "epoch": 1.0457271364317842,
      "grad_norm": 0.24465021912040835,
      "learning_rate": 0.0001646973476475354,
      "loss": 0.0322,
      "step": 1395
    },
    {
      "epoch": 1.0464767616191903,
      "grad_norm": 0.13745202157374345,
      "learning_rate": 0.00016463079974439842,
      "loss": 0.0138,
      "step": 1396
    },
    {
      "epoch": 1.0472263868065967,
      "grad_norm": 0.22563440257592166,
      "learning_rate": 0.0001645642026494285,
      "loss": 0.0246,
      "step": 1397
    },
    {
      "epoch": 1.047976011994003,
      "grad_norm": 0.17875194448102819,
      "learning_rate": 0.00016449755641331407,
      "loss": 0.022,
      "step": 1398
    },
    {
      "epoch": 1.0487256371814093,
      "grad_norm": 0.2036285436358285,
      "learning_rate": 0.00016443086108678098,
      "loss": 0.021,
      "step": 1399
    },
    {
      "epoch": 1.0494752623688155,
      "grad_norm": 0.20156890156581853,
      "learning_rate": 0.00016436411672059238,
      "loss": 0.0109,
      "step": 1400
    },
    {
      "epoch": 1.050224887556222,
      "grad_norm": 0.395941922751957,
      "learning_rate": 0.0001642973233655488,
      "loss": 0.03,
      "step": 1401
    },
    {
      "epoch": 1.050974512743628,
      "grad_norm": 0.061274010065021596,
      "learning_rate": 0.00016423048107248802,
      "loss": 0.0055,
      "step": 1402
    },
    {
      "epoch": 1.0517241379310345,
      "grad_norm": 0.19642591911550913,
      "learning_rate": 0.00016416358989228508,
      "loss": 0.0243,
      "step": 1403
    },
    {
      "epoch": 1.0524737631184409,
      "grad_norm": 0.17905794054557944,
      "learning_rate": 0.00016409664987585232,
      "loss": 0.0256,
      "step": 1404
    },
    {
      "epoch": 1.053223388305847,
      "grad_norm": 0.08570887447141709,
      "learning_rate": 0.00016402966107413903,
      "loss": 0.0068,
      "step": 1405
    },
    {
      "epoch": 1.0539730134932535,
      "grad_norm": 0.25819303796441384,
      "learning_rate": 0.00016396262353813188,
      "loss": 0.0452,
      "step": 1406
    },
    {
      "epoch": 1.0547226386806596,
      "grad_norm": 0.14319320724859413,
      "learning_rate": 0.00016389553731885445,
      "loss": 0.0307,
      "step": 1407
    },
    {
      "epoch": 1.055472263868066,
      "grad_norm": 0.08861014982546456,
      "learning_rate": 0.0001638284024673675,
      "loss": 0.02,
      "step": 1408
    },
    {
      "epoch": 1.0562218890554722,
      "grad_norm": 0.42806895391787697,
      "learning_rate": 0.00016376121903476865,
      "loss": 0.0354,
      "step": 1409
    },
    {
      "epoch": 1.0569715142428786,
      "grad_norm": 0.05845392704546912,
      "learning_rate": 0.0001636939870721927,
      "loss": 0.0047,
      "step": 1410
    },
    {
      "epoch": 1.0577211394302848,
      "grad_norm": 0.22523256553737497,
      "learning_rate": 0.00016362670663081126,
      "loss": 0.0256,
      "step": 1411
    },
    {
      "epoch": 1.0584707646176912,
      "grad_norm": 0.2390347688368189,
      "learning_rate": 0.0001635593777618328,
      "loss": 0.0269,
      "step": 1412
    },
    {
      "epoch": 1.0592203898050974,
      "grad_norm": 0.23149230743504765,
      "learning_rate": 0.0001634920005165028,
      "loss": 0.0291,
      "step": 1413
    },
    {
      "epoch": 1.0599700149925038,
      "grad_norm": 0.13249901049153695,
      "learning_rate": 0.00016342457494610338,
      "loss": 0.0228,
      "step": 1414
    },
    {
      "epoch": 1.06071964017991,
      "grad_norm": 0.18082496024917116,
      "learning_rate": 0.0001633571011019536,
      "loss": 0.0366,
      "step": 1415
    },
    {
      "epoch": 1.0614692653673163,
      "grad_norm": 0.2076705932655202,
      "learning_rate": 0.00016328957903540917,
      "loss": 0.1393,
      "step": 1416
    },
    {
      "epoch": 1.0622188905547227,
      "grad_norm": 0.08378805485028017,
      "learning_rate": 0.00016322200879786248,
      "loss": 0.0143,
      "step": 1417
    },
    {
      "epoch": 1.062968515742129,
      "grad_norm": 0.2566888867792056,
      "learning_rate": 0.0001631543904407427,
      "loss": 0.0427,
      "step": 1418
    },
    {
      "epoch": 1.0637181409295353,
      "grad_norm": 0.18958504372246548,
      "learning_rate": 0.0001630867240155155,
      "loss": 0.0197,
      "step": 1419
    },
    {
      "epoch": 1.0644677661169415,
      "grad_norm": 0.134466245166497,
      "learning_rate": 0.00016301900957368321,
      "loss": 0.0197,
      "step": 1420
    },
    {
      "epoch": 1.065217391304348,
      "grad_norm": 0.23636192887764287,
      "learning_rate": 0.00016295124716678473,
      "loss": 0.0307,
      "step": 1421
    },
    {
      "epoch": 1.065967016491754,
      "grad_norm": 0.3062339725051187,
      "learning_rate": 0.00016288343684639535,
      "loss": 0.018,
      "step": 1422
    },
    {
      "epoch": 1.0667166416791605,
      "grad_norm": 0.22431620780731476,
      "learning_rate": 0.00016281557866412692,
      "loss": 0.0234,
      "step": 1423
    },
    {
      "epoch": 1.0674662668665666,
      "grad_norm": 0.4321605036607824,
      "learning_rate": 0.00016274767267162777,
      "loss": 0.0139,
      "step": 1424
    },
    {
      "epoch": 1.068215892053973,
      "grad_norm": 0.23333187413650924,
      "learning_rate": 0.00016267971892058243,
      "loss": 0.0429,
      "step": 1425
    },
    {
      "epoch": 1.0689655172413792,
      "grad_norm": 0.19561907405199422,
      "learning_rate": 0.000162611717462712,
      "loss": 0.0249,
      "step": 1426
    },
    {
      "epoch": 1.0697151424287856,
      "grad_norm": 0.11527450520671127,
      "learning_rate": 0.00016254366834977374,
      "loss": 0.0068,
      "step": 1427
    },
    {
      "epoch": 1.0704647676161918,
      "grad_norm": 0.17629341766686163,
      "learning_rate": 0.00016247557163356127,
      "loss": 0.0128,
      "step": 1428
    },
    {
      "epoch": 1.0712143928035982,
      "grad_norm": 0.3968345991829092,
      "learning_rate": 0.00016240742736590438,
      "loss": 0.0218,
      "step": 1429
    },
    {
      "epoch": 1.0719640179910046,
      "grad_norm": 0.16395160947758386,
      "learning_rate": 0.00016233923559866902,
      "loss": 0.0215,
      "step": 1430
    },
    {
      "epoch": 1.0727136431784108,
      "grad_norm": 0.14224442019132602,
      "learning_rate": 0.00016227099638375746,
      "loss": 0.009,
      "step": 1431
    },
    {
      "epoch": 1.0734632683658172,
      "grad_norm": 0.26354074290357404,
      "learning_rate": 0.0001622027097731079,
      "loss": 0.041,
      "step": 1432
    },
    {
      "epoch": 1.0742128935532234,
      "grad_norm": 0.20767258252642307,
      "learning_rate": 0.00016213437581869472,
      "loss": 0.0202,
      "step": 1433
    },
    {
      "epoch": 1.0749625187406298,
      "grad_norm": 0.09750442000754214,
      "learning_rate": 0.00016206599457252826,
      "loss": 0.0127,
      "step": 1434
    },
    {
      "epoch": 1.075712143928036,
      "grad_norm": 0.1912404856886841,
      "learning_rate": 0.0001619975660866549,
      "loss": 0.0155,
      "step": 1435
    },
    {
      "epoch": 1.0764617691154423,
      "grad_norm": 0.18752152354183707,
      "learning_rate": 0.000161929090413157,
      "loss": 0.0214,
      "step": 1436
    },
    {
      "epoch": 1.0772113943028485,
      "grad_norm": 0.13332532955034693,
      "learning_rate": 0.00016186056760415278,
      "loss": 0.0221,
      "step": 1437
    },
    {
      "epoch": 1.077961019490255,
      "grad_norm": 0.1908859896689113,
      "learning_rate": 0.0001617919977117964,
      "loss": 0.0195,
      "step": 1438
    },
    {
      "epoch": 1.078710644677661,
      "grad_norm": 0.35561503622368495,
      "learning_rate": 0.0001617233807882778,
      "loss": 0.0903,
      "step": 1439
    },
    {
      "epoch": 1.0794602698650675,
      "grad_norm": 0.1191563810615422,
      "learning_rate": 0.00016165471688582269,
      "loss": 0.0144,
      "step": 1440
    },
    {
      "epoch": 1.0802098950524737,
      "grad_norm": 0.10310384566107852,
      "learning_rate": 0.00016158600605669263,
      "loss": 0.0155,
      "step": 1441
    },
    {
      "epoch": 1.08095952023988,
      "grad_norm": 0.16808318985944135,
      "learning_rate": 0.00016151724835318482,
      "loss": 0.0298,
      "step": 1442
    },
    {
      "epoch": 1.0817091454272862,
      "grad_norm": 0.2928806936676149,
      "learning_rate": 0.0001614484438276322,
      "loss": 0.0274,
      "step": 1443
    },
    {
      "epoch": 1.0824587706146926,
      "grad_norm": 0.3014557599632698,
      "learning_rate": 0.00016137959253240328,
      "loss": 0.0612,
      "step": 1444
    },
    {
      "epoch": 1.083208395802099,
      "grad_norm": 0.2780565413334579,
      "learning_rate": 0.0001613106945199022,
      "loss": 0.009,
      "step": 1445
    },
    {
      "epoch": 1.0839580209895052,
      "grad_norm": 0.18495490207774667,
      "learning_rate": 0.0001612417498425687,
      "loss": 0.0177,
      "step": 1446
    },
    {
      "epoch": 1.0847076461769116,
      "grad_norm": 0.24486514912871213,
      "learning_rate": 0.0001611727585528779,
      "loss": 0.0478,
      "step": 1447
    },
    {
      "epoch": 1.0854572713643178,
      "grad_norm": 0.11918873624620073,
      "learning_rate": 0.0001611037207033406,
      "loss": 0.0175,
      "step": 1448
    },
    {
      "epoch": 1.0862068965517242,
      "grad_norm": 0.1785435197671633,
      "learning_rate": 0.00016103463634650284,
      "loss": 0.0135,
      "step": 1449
    },
    {
      "epoch": 1.0869565217391304,
      "grad_norm": 0.12194464031268934,
      "learning_rate": 0.0001609655055349462,
      "loss": 0.0181,
      "step": 1450
    },
    {
      "epoch": 1.0877061469265368,
      "grad_norm": 0.09997546100461988,
      "learning_rate": 0.00016089632832128756,
      "loss": 0.0169,
      "step": 1451
    },
    {
      "epoch": 1.088455772113943,
      "grad_norm": 0.14404140334250182,
      "learning_rate": 0.00016082710475817912,
      "loss": 0.0117,
      "step": 1452
    },
    {
      "epoch": 1.0892053973013494,
      "grad_norm": 0.22062567051623222,
      "learning_rate": 0.00016075783489830834,
      "loss": 0.034,
      "step": 1453
    },
    {
      "epoch": 1.0899550224887555,
      "grad_norm": 0.13431333548606741,
      "learning_rate": 0.00016068851879439793,
      "loss": 0.0172,
      "step": 1454
    },
    {
      "epoch": 1.090704647676162,
      "grad_norm": 0.10841985313441686,
      "learning_rate": 0.00016061915649920585,
      "loss": 0.0136,
      "step": 1455
    },
    {
      "epoch": 1.0914542728635683,
      "grad_norm": 0.15014258139931097,
      "learning_rate": 0.00016054974806552514,
      "loss": 0.018,
      "step": 1456
    },
    {
      "epoch": 1.0922038980509745,
      "grad_norm": 0.14319312568552878,
      "learning_rate": 0.00016048029354618398,
      "loss": 0.0166,
      "step": 1457
    },
    {
      "epoch": 1.092953523238381,
      "grad_norm": 0.25933770441264264,
      "learning_rate": 0.0001604107929940457,
      "loss": 0.043,
      "step": 1458
    },
    {
      "epoch": 1.093703148425787,
      "grad_norm": 0.08961664023952531,
      "learning_rate": 0.00016034124646200848,
      "loss": 0.013,
      "step": 1459
    },
    {
      "epoch": 1.0944527736131935,
      "grad_norm": 0.29012788304698733,
      "learning_rate": 0.00016027165400300572,
      "loss": 0.028,
      "step": 1460
    },
    {
      "epoch": 1.0952023988005997,
      "grad_norm": 0.15912351000061195,
      "learning_rate": 0.00016020201567000565,
      "loss": 0.014,
      "step": 1461
    },
    {
      "epoch": 1.095952023988006,
      "grad_norm": 0.16601612022448584,
      "learning_rate": 0.00016013233151601142,
      "loss": 0.0198,
      "step": 1462
    },
    {
      "epoch": 1.0967016491754122,
      "grad_norm": 0.12808278580064933,
      "learning_rate": 0.00016006260159406112,
      "loss": 0.0136,
      "step": 1463
    },
    {
      "epoch": 1.0974512743628186,
      "grad_norm": 0.6327112833083882,
      "learning_rate": 0.00015999282595722758,
      "loss": 0.0392,
      "step": 1464
    },
    {
      "epoch": 1.0982008995502248,
      "grad_norm": 0.08472043991318816,
      "learning_rate": 0.00015992300465861848,
      "loss": 0.0114,
      "step": 1465
    },
    {
      "epoch": 1.0989505247376312,
      "grad_norm": 0.1909500285972242,
      "learning_rate": 0.00015985313775137628,
      "loss": 0.0168,
      "step": 1466
    },
    {
      "epoch": 1.0997001499250374,
      "grad_norm": 0.19173488400546776,
      "learning_rate": 0.00015978322528867808,
      "loss": 0.033,
      "step": 1467
    },
    {
      "epoch": 1.1004497751124438,
      "grad_norm": 0.07865480318728033,
      "learning_rate": 0.00015971326732373573,
      "loss": 0.0106,
      "step": 1468
    },
    {
      "epoch": 1.10119940029985,
      "grad_norm": 0.3808152008530821,
      "learning_rate": 0.00015964326390979566,
      "loss": 0.0431,
      "step": 1469
    },
    {
      "epoch": 1.1019490254872564,
      "grad_norm": 0.12153952973021516,
      "learning_rate": 0.00015957321510013894,
      "loss": 0.0157,
      "step": 1470
    },
    {
      "epoch": 1.1026986506746628,
      "grad_norm": 0.18323525625666828,
      "learning_rate": 0.0001595031209480811,
      "loss": 0.0288,
      "step": 1471
    },
    {
      "epoch": 1.103448275862069,
      "grad_norm": 0.21675767974438637,
      "learning_rate": 0.0001594329815069723,
      "loss": 0.0203,
      "step": 1472
    },
    {
      "epoch": 1.1041979010494753,
      "grad_norm": 0.14831560448490197,
      "learning_rate": 0.0001593627968301971,
      "loss": 0.0257,
      "step": 1473
    },
    {
      "epoch": 1.1049475262368815,
      "grad_norm": 0.11944549610257972,
      "learning_rate": 0.0001592925669711745,
      "loss": 0.0089,
      "step": 1474
    },
    {
      "epoch": 1.105697151424288,
      "grad_norm": 0.1388075886641613,
      "learning_rate": 0.00015922229198335786,
      "loss": 0.0156,
      "step": 1475
    },
    {
      "epoch": 1.106446776611694,
      "grad_norm": 0.1396408694801958,
      "learning_rate": 0.000159151971920235,
      "loss": 0.0216,
      "step": 1476
    },
    {
      "epoch": 1.1071964017991005,
      "grad_norm": 0.20531653159393498,
      "learning_rate": 0.00015908160683532788,
      "loss": 0.0282,
      "step": 1477
    },
    {
      "epoch": 1.1079460269865067,
      "grad_norm": 0.07127518191582545,
      "learning_rate": 0.00015901119678219286,
      "loss": 0.0103,
      "step": 1478
    },
    {
      "epoch": 1.108695652173913,
      "grad_norm": 0.25309598215724977,
      "learning_rate": 0.0001589407418144205,
      "loss": 0.0265,
      "step": 1479
    },
    {
      "epoch": 1.1094452773613193,
      "grad_norm": 0.08942874508347794,
      "learning_rate": 0.00015887024198563552,
      "loss": 0.01,
      "step": 1480
    },
    {
      "epoch": 1.1101949025487257,
      "grad_norm": 0.23687972616077854,
      "learning_rate": 0.00015879969734949675,
      "loss": 0.011,
      "step": 1481
    },
    {
      "epoch": 1.1109445277361318,
      "grad_norm": 0.4708176787292712,
      "learning_rate": 0.00015872910795969718,
      "loss": 0.0453,
      "step": 1482
    },
    {
      "epoch": 1.1116941529235382,
      "grad_norm": 0.1814159239414209,
      "learning_rate": 0.00015865847386996386,
      "loss": 0.0213,
      "step": 1483
    },
    {
      "epoch": 1.1124437781109444,
      "grad_norm": 0.2884094234411925,
      "learning_rate": 0.00015858779513405784,
      "loss": 0.085,
      "step": 1484
    },
    {
      "epoch": 1.1131934032983508,
      "grad_norm": 0.1495251782404274,
      "learning_rate": 0.0001585170718057742,
      "loss": 0.0362,
      "step": 1485
    },
    {
      "epoch": 1.1139430284857572,
      "grad_norm": 0.22512374680856442,
      "learning_rate": 0.00015844630393894188,
      "loss": 0.0314,
      "step": 1486
    },
    {
      "epoch": 1.1146926536731634,
      "grad_norm": 0.09988941306056143,
      "learning_rate": 0.00015837549158742378,
      "loss": 0.0147,
      "step": 1487
    },
    {
      "epoch": 1.1154422788605698,
      "grad_norm": 0.0901363000558977,
      "learning_rate": 0.00015830463480511663,
      "loss": 0.0131,
      "step": 1488
    },
    {
      "epoch": 1.116191904047976,
      "grad_norm": 0.16402213224909407,
      "learning_rate": 0.000158233733645951,
      "loss": 0.0293,
      "step": 1489
    },
    {
      "epoch": 1.1169415292353824,
      "grad_norm": 0.13463191653600406,
      "learning_rate": 0.0001581627881638912,
      "loss": 0.0081,
      "step": 1490
    },
    {
      "epoch": 1.1176911544227885,
      "grad_norm": 0.15770222986745397,
      "learning_rate": 0.0001580917984129353,
      "loss": 0.0224,
      "step": 1491
    },
    {
      "epoch": 1.118440779610195,
      "grad_norm": 0.16417954231468038,
      "learning_rate": 0.0001580207644471151,
      "loss": 0.0274,
      "step": 1492
    },
    {
      "epoch": 1.1191904047976011,
      "grad_norm": 0.07265795493103328,
      "learning_rate": 0.000157949686320496,
      "loss": 0.0089,
      "step": 1493
    },
    {
      "epoch": 1.1199400299850075,
      "grad_norm": 0.13153007530375596,
      "learning_rate": 0.00015787856408717697,
      "loss": 0.0151,
      "step": 1494
    },
    {
      "epoch": 1.1206896551724137,
      "grad_norm": 0.212497179793201,
      "learning_rate": 0.00015780739780129068,
      "loss": 0.0182,
      "step": 1495
    },
    {
      "epoch": 1.12143928035982,
      "grad_norm": 0.18185328948368945,
      "learning_rate": 0.0001577361875170032,
      "loss": 0.0334,
      "step": 1496
    },
    {
      "epoch": 1.1221889055472265,
      "grad_norm": 0.1384683028189125,
      "learning_rate": 0.0001576649332885142,
      "loss": 0.0122,
      "step": 1497
    },
    {
      "epoch": 1.1229385307346327,
      "grad_norm": 0.13502693122215248,
      "learning_rate": 0.0001575936351700567,
      "loss": 0.0146,
      "step": 1498
    },
    {
      "epoch": 1.123688155922039,
      "grad_norm": 0.2647766754878005,
      "learning_rate": 0.00015752229321589717,
      "loss": 0.0338,
      "step": 1499
    },
    {
      "epoch": 1.1244377811094453,
      "grad_norm": 0.16050841014726616,
      "learning_rate": 0.00015745090748033546,
      "loss": 0.0197,
      "step": 1500
    },
    {
      "epoch": 1.1251874062968517,
      "grad_norm": 0.21556781551128099,
      "learning_rate": 0.0001573794780177047,
      "loss": 0.0196,
      "step": 1501
    },
    {
      "epoch": 1.1259370314842578,
      "grad_norm": 0.10147478324129008,
      "learning_rate": 0.00015730800488237135,
      "loss": 0.013,
      "step": 1502
    },
    {
      "epoch": 1.1266866566716642,
      "grad_norm": 0.17908072704859868,
      "learning_rate": 0.00015723648812873507,
      "loss": 0.0208,
      "step": 1503
    },
    {
      "epoch": 1.1274362818590704,
      "grad_norm": 0.1427235790172978,
      "learning_rate": 0.00015716492781122875,
      "loss": 0.0334,
      "step": 1504
    },
    {
      "epoch": 1.1281859070464768,
      "grad_norm": 0.09601915395576731,
      "learning_rate": 0.00015709332398431843,
      "loss": 0.0139,
      "step": 1505
    },
    {
      "epoch": 1.128935532233883,
      "grad_norm": 0.2653400088799801,
      "learning_rate": 0.0001570216767025032,
      "loss": 0.0296,
      "step": 1506
    },
    {
      "epoch": 1.1296851574212894,
      "grad_norm": 0.22056456709614147,
      "learning_rate": 0.0001569499860203153,
      "loss": 0.0343,
      "step": 1507
    },
    {
      "epoch": 1.1304347826086956,
      "grad_norm": 0.16808986480694899,
      "learning_rate": 0.00015687825199232,
      "loss": 0.0204,
      "step": 1508
    },
    {
      "epoch": 1.131184407796102,
      "grad_norm": 0.17447894186209695,
      "learning_rate": 0.00015680647467311557,
      "loss": 0.0234,
      "step": 1509
    },
    {
      "epoch": 1.1319340329835081,
      "grad_norm": 0.12815018624473096,
      "learning_rate": 0.00015673465411733322,
      "loss": 0.0227,
      "step": 1510
    },
    {
      "epoch": 1.1326836581709145,
      "grad_norm": 0.09165413310209258,
      "learning_rate": 0.00015666279037963697,
      "loss": 0.0111,
      "step": 1511
    },
    {
      "epoch": 1.133433283358321,
      "grad_norm": 0.07284436146532805,
      "learning_rate": 0.00015659088351472388,
      "loss": 0.0153,
      "step": 1512
    },
    {
      "epoch": 1.1341829085457271,
      "grad_norm": 0.1039038555643137,
      "learning_rate": 0.00015651893357732368,
      "loss": 0.0192,
      "step": 1513
    },
    {
      "epoch": 1.1349325337331335,
      "grad_norm": 0.12641273866798364,
      "learning_rate": 0.00015644694062219897,
      "loss": 0.0212,
      "step": 1514
    },
    {
      "epoch": 1.1356821589205397,
      "grad_norm": 0.1303955647396933,
      "learning_rate": 0.00015637490470414513,
      "loss": 0.0271,
      "step": 1515
    },
    {
      "epoch": 1.136431784107946,
      "grad_norm": 0.14201348367234287,
      "learning_rate": 0.00015630282587799008,
      "loss": 0.0315,
      "step": 1516
    },
    {
      "epoch": 1.1371814092953523,
      "grad_norm": 0.11012632959351705,
      "learning_rate": 0.00015623070419859455,
      "loss": 0.0103,
      "step": 1517
    },
    {
      "epoch": 1.1379310344827587,
      "grad_norm": 0.12460339857679491,
      "learning_rate": 0.00015615853972085185,
      "loss": 0.0172,
      "step": 1518
    },
    {
      "epoch": 1.1386806596701649,
      "grad_norm": 0.22870490943904723,
      "learning_rate": 0.00015608633249968783,
      "loss": 0.0318,
      "step": 1519
    },
    {
      "epoch": 1.1394302848575713,
      "grad_norm": 0.22798883978071888,
      "learning_rate": 0.0001560140825900609,
      "loss": 0.0598,
      "step": 1520
    },
    {
      "epoch": 1.1401799100449774,
      "grad_norm": 0.07875878857169033,
      "learning_rate": 0.00015594179004696192,
      "loss": 0.0087,
      "step": 1521
    },
    {
      "epoch": 1.1409295352323838,
      "grad_norm": 0.11520020041983646,
      "learning_rate": 0.0001558694549254143,
      "loss": 0.0096,
      "step": 1522
    },
    {
      "epoch": 1.1416791604197902,
      "grad_norm": 0.4275082260936782,
      "learning_rate": 0.00015579707728047377,
      "loss": 0.0275,
      "step": 1523
    },
    {
      "epoch": 1.1424287856071964,
      "grad_norm": 0.13461389601112744,
      "learning_rate": 0.0001557246571672284,
      "loss": 0.0128,
      "step": 1524
    },
    {
      "epoch": 1.1431784107946026,
      "grad_norm": 0.2116201093302502,
      "learning_rate": 0.00015565219464079867,
      "loss": 0.016,
      "step": 1525
    },
    {
      "epoch": 1.143928035982009,
      "grad_norm": 0.12566130628845476,
      "learning_rate": 0.0001555796897563373,
      "loss": 0.0196,
      "step": 1526
    },
    {
      "epoch": 1.1446776611694154,
      "grad_norm": 0.19729834701794738,
      "learning_rate": 0.00015550714256902924,
      "loss": 0.0229,
      "step": 1527
    },
    {
      "epoch": 1.1454272863568216,
      "grad_norm": 0.1943109518312495,
      "learning_rate": 0.00015543455313409167,
      "loss": 0.0286,
      "step": 1528
    },
    {
      "epoch": 1.146176911544228,
      "grad_norm": 0.15684836904901786,
      "learning_rate": 0.00015536192150677387,
      "loss": 0.0266,
      "step": 1529
    },
    {
      "epoch": 1.1469265367316341,
      "grad_norm": 0.19744182244166947,
      "learning_rate": 0.00015528924774235728,
      "loss": 0.0388,
      "step": 1530
    },
    {
      "epoch": 1.1476761619190405,
      "grad_norm": 0.1428512558705438,
      "learning_rate": 0.00015521653189615542,
      "loss": 0.0109,
      "step": 1531
    },
    {
      "epoch": 1.1484257871064467,
      "grad_norm": 0.09805806018139068,
      "learning_rate": 0.00015514377402351377,
      "loss": 0.0078,
      "step": 1532
    },
    {
      "epoch": 1.1491754122938531,
      "grad_norm": 0.1727329923961472,
      "learning_rate": 0.00015507097417980992,
      "loss": 0.0259,
      "step": 1533
    },
    {
      "epoch": 1.1499250374812593,
      "grad_norm": 0.0781432848720514,
      "learning_rate": 0.00015499813242045326,
      "loss": 0.0088,
      "step": 1534
    },
    {
      "epoch": 1.1506746626686657,
      "grad_norm": 0.1793998560755001,
      "learning_rate": 0.0001549252488008852,
      "loss": 0.0187,
      "step": 1535
    },
    {
      "epoch": 1.1514242878560719,
      "grad_norm": 0.20464042558859094,
      "learning_rate": 0.00015485232337657893,
      "loss": 0.0252,
      "step": 1536
    },
    {
      "epoch": 1.1521739130434783,
      "grad_norm": 0.15466608218391978,
      "learning_rate": 0.00015477935620303952,
      "loss": 0.0119,
      "step": 1537
    },
    {
      "epoch": 1.1529235382308847,
      "grad_norm": 0.14546232820077265,
      "learning_rate": 0.0001547063473358038,
      "loss": 0.0221,
      "step": 1538
    },
    {
      "epoch": 1.1536731634182908,
      "grad_norm": 0.20254604263054207,
      "learning_rate": 0.00015463329683044027,
      "loss": 0.0203,
      "step": 1539
    },
    {
      "epoch": 1.1544227886056972,
      "grad_norm": 0.11962821907793976,
      "learning_rate": 0.0001545602047425492,
      "loss": 0.0171,
      "step": 1540
    },
    {
      "epoch": 1.1551724137931034,
      "grad_norm": 0.17742009738804226,
      "learning_rate": 0.0001544870711277625,
      "loss": 0.0388,
      "step": 1541
    },
    {
      "epoch": 1.1559220389805098,
      "grad_norm": 0.13547950643645995,
      "learning_rate": 0.00015441389604174365,
      "loss": 0.0224,
      "step": 1542
    },
    {
      "epoch": 1.156671664167916,
      "grad_norm": 0.13957811910369117,
      "learning_rate": 0.00015434067954018773,
      "loss": 0.0181,
      "step": 1543
    },
    {
      "epoch": 1.1574212893553224,
      "grad_norm": 0.10593824435236912,
      "learning_rate": 0.00015426742167882131,
      "loss": 0.0192,
      "step": 1544
    },
    {
      "epoch": 1.1581709145427286,
      "grad_norm": 0.1923422077623307,
      "learning_rate": 0.0001541941225134025,
      "loss": 0.0223,
      "step": 1545
    },
    {
      "epoch": 1.158920539730135,
      "grad_norm": 0.09584418842068963,
      "learning_rate": 0.00015412078209972076,
      "loss": 0.01,
      "step": 1546
    },
    {
      "epoch": 1.1596701649175412,
      "grad_norm": 0.1928952002851109,
      "learning_rate": 0.00015404740049359696,
      "loss": 0.0392,
      "step": 1547
    },
    {
      "epoch": 1.1604197901049476,
      "grad_norm": 0.07336743595400493,
      "learning_rate": 0.00015397397775088347,
      "loss": 0.0151,
      "step": 1548
    },
    {
      "epoch": 1.1611694152923537,
      "grad_norm": 0.08828692920585819,
      "learning_rate": 0.0001539005139274637,
      "loss": 0.0124,
      "step": 1549
    },
    {
      "epoch": 1.1619190404797601,
      "grad_norm": 0.13783225187379142,
      "learning_rate": 0.0001538270090792526,
      "loss": 0.0238,
      "step": 1550
    },
    {
      "epoch": 1.1626686656671663,
      "grad_norm": 0.10249729690468168,
      "learning_rate": 0.00015375346326219617,
      "loss": 0.0081,
      "step": 1551
    },
    {
      "epoch": 1.1634182908545727,
      "grad_norm": 0.20748713541551198,
      "learning_rate": 0.00015367987653227164,
      "loss": 0.0179,
      "step": 1552
    },
    {
      "epoch": 1.1641679160419791,
      "grad_norm": 0.10779303756805979,
      "learning_rate": 0.00015360624894548744,
      "loss": 0.0193,
      "step": 1553
    },
    {
      "epoch": 1.1649175412293853,
      "grad_norm": 0.11649926171215857,
      "learning_rate": 0.00015353258055788297,
      "loss": 0.0119,
      "step": 1554
    },
    {
      "epoch": 1.1656671664167917,
      "grad_norm": 0.08919536536879394,
      "learning_rate": 0.0001534588714255288,
      "loss": 0.0113,
      "step": 1555
    },
    {
      "epoch": 1.1664167916041979,
      "grad_norm": 0.16624570100412386,
      "learning_rate": 0.00015338512160452647,
      "loss": 0.0192,
      "step": 1556
    },
    {
      "epoch": 1.1671664167916043,
      "grad_norm": 0.16784360601652096,
      "learning_rate": 0.00015331133115100847,
      "loss": 0.0172,
      "step": 1557
    },
    {
      "epoch": 1.1679160419790104,
      "grad_norm": 0.07319159417719012,
      "learning_rate": 0.00015323750012113827,
      "loss": 0.0169,
      "step": 1558
    },
    {
      "epoch": 1.1686656671664168,
      "grad_norm": 0.18373874199472398,
      "learning_rate": 0.00015316362857111012,
      "loss": 0.0216,
      "step": 1559
    },
    {
      "epoch": 1.169415292353823,
      "grad_norm": 0.06673105844962858,
      "learning_rate": 0.00015308971655714925,
      "loss": 0.0066,
      "step": 1560
    },
    {
      "epoch": 1.1701649175412294,
      "grad_norm": 0.21392038301444416,
      "learning_rate": 0.00015301576413551154,
      "loss": 0.0242,
      "step": 1561
    },
    {
      "epoch": 1.1709145427286356,
      "grad_norm": 0.18719939485446224,
      "learning_rate": 0.00015294177136248379,
      "loss": 0.0205,
      "step": 1562
    },
    {
      "epoch": 1.171664167916042,
      "grad_norm": 0.14217515743768686,
      "learning_rate": 0.00015286773829438334,
      "loss": 0.0176,
      "step": 1563
    },
    {
      "epoch": 1.1724137931034484,
      "grad_norm": 0.2329284232310269,
      "learning_rate": 0.00015279366498755837,
      "loss": 0.0314,
      "step": 1564
    },
    {
      "epoch": 1.1731634182908546,
      "grad_norm": 0.1225692888740097,
      "learning_rate": 0.0001527195514983875,
      "loss": 0.0139,
      "step": 1565
    },
    {
      "epoch": 1.1739130434782608,
      "grad_norm": 0.3449262128498415,
      "learning_rate": 0.00015264539788328012,
      "loss": 0.0237,
      "step": 1566
    },
    {
      "epoch": 1.1746626686656672,
      "grad_norm": 0.13106031341373675,
      "learning_rate": 0.00015257120419867603,
      "loss": 0.0076,
      "step": 1567
    },
    {
      "epoch": 1.1754122938530736,
      "grad_norm": 0.09537967707527366,
      "learning_rate": 0.0001524969705010456,
      "loss": 0.0089,
      "step": 1568
    },
    {
      "epoch": 1.1761619190404797,
      "grad_norm": 0.3558565450829918,
      "learning_rate": 0.00015242269684688956,
      "loss": 0.0376,
      "step": 1569
    },
    {
      "epoch": 1.1769115442278861,
      "grad_norm": 0.3121634965388302,
      "learning_rate": 0.00015234838329273922,
      "loss": 0.0205,
      "step": 1570
    },
    {
      "epoch": 1.1776611694152923,
      "grad_norm": 0.1452208720901486,
      "learning_rate": 0.00015227402989515608,
      "loss": 0.0185,
      "step": 1571
    },
    {
      "epoch": 1.1784107946026987,
      "grad_norm": 0.14467715052341956,
      "learning_rate": 0.00015219963671073204,
      "loss": 0.0212,
      "step": 1572
    },
    {
      "epoch": 1.1791604197901049,
      "grad_norm": 0.12022759396311145,
      "learning_rate": 0.00015212520379608932,
      "loss": 0.0211,
      "step": 1573
    },
    {
      "epoch": 1.1799100449775113,
      "grad_norm": 0.1459844281580083,
      "learning_rate": 0.00015205073120788036,
      "loss": 0.0128,
      "step": 1574
    },
    {
      "epoch": 1.1806596701649175,
      "grad_norm": 0.22936341003304037,
      "learning_rate": 0.0001519762190027877,
      "loss": 0.027,
      "step": 1575
    },
    {
      "epoch": 1.1814092953523239,
      "grad_norm": 0.19513599358000275,
      "learning_rate": 0.00015190166723752423,
      "loss": 0.0267,
      "step": 1576
    },
    {
      "epoch": 1.18215892053973,
      "grad_norm": 0.1167567522117949,
      "learning_rate": 0.00015182707596883275,
      "loss": 0.0206,
      "step": 1577
    },
    {
      "epoch": 1.1829085457271364,
      "grad_norm": 0.22445548930184916,
      "learning_rate": 0.00015175244525348624,
      "loss": 0.0355,
      "step": 1578
    },
    {
      "epoch": 1.1836581709145428,
      "grad_norm": 0.08670964454479742,
      "learning_rate": 0.00015167777514828767,
      "loss": 0.0129,
      "step": 1579
    },
    {
      "epoch": 1.184407796101949,
      "grad_norm": 0.14385266850244458,
      "learning_rate": 0.00015160306571006995,
      "loss": 0.0177,
      "step": 1580
    },
    {
      "epoch": 1.1851574212893554,
      "grad_norm": 0.08759140642886128,
      "learning_rate": 0.00015152831699569605,
      "loss": 0.0081,
      "step": 1581
    },
    {
      "epoch": 1.1859070464767616,
      "grad_norm": 0.1316505612026678,
      "learning_rate": 0.00015145352906205872,
      "loss": 0.0161,
      "step": 1582
    },
    {
      "epoch": 1.186656671664168,
      "grad_norm": 0.11305442190492103,
      "learning_rate": 0.0001513787019660806,
      "loss": 0.0168,
      "step": 1583
    },
    {
      "epoch": 1.1874062968515742,
      "grad_norm": 0.1521461905461006,
      "learning_rate": 0.00015130383576471415,
      "loss": 0.0136,
      "step": 1584
    },
    {
      "epoch": 1.1881559220389806,
      "grad_norm": 0.18712535443586656,
      "learning_rate": 0.00015122893051494152,
      "loss": 0.0273,
      "step": 1585
    },
    {
      "epoch": 1.1889055472263867,
      "grad_norm": 0.21720392213142015,
      "learning_rate": 0.0001511539862737747,
      "loss": 0.0252,
      "step": 1586
    },
    {
      "epoch": 1.1896551724137931,
      "grad_norm": 0.09435652810077971,
      "learning_rate": 0.00015107900309825528,
      "loss": 0.0124,
      "step": 1587
    },
    {
      "epoch": 1.1904047976011993,
      "grad_norm": 0.8196844795700977,
      "learning_rate": 0.0001510039810454545,
      "loss": 0.06,
      "step": 1588
    },
    {
      "epoch": 1.1911544227886057,
      "grad_norm": 0.13157054198911564,
      "learning_rate": 0.00015092892017247317,
      "loss": 0.0139,
      "step": 1589
    },
    {
      "epoch": 1.191904047976012,
      "grad_norm": 0.09674622860826092,
      "learning_rate": 0.0001508538205364417,
      "loss": 0.0119,
      "step": 1590
    },
    {
      "epoch": 1.1926536731634183,
      "grad_norm": 0.22671538004519848,
      "learning_rate": 0.00015077868219451993,
      "loss": 0.0248,
      "step": 1591
    },
    {
      "epoch": 1.1934032983508245,
      "grad_norm": 0.6288248427499471,
      "learning_rate": 0.0001507035052038972,
      "loss": 0.0329,
      "step": 1592
    },
    {
      "epoch": 1.1941529235382309,
      "grad_norm": 0.10526177711938337,
      "learning_rate": 0.00015062828962179232,
      "loss": 0.0149,
      "step": 1593
    },
    {
      "epoch": 1.1949025487256373,
      "grad_norm": 0.11834890522369187,
      "learning_rate": 0.00015055303550545336,
      "loss": 0.0138,
      "step": 1594
    },
    {
      "epoch": 1.1956521739130435,
      "grad_norm": 0.144173189523043,
      "learning_rate": 0.0001504777429121578,
      "loss": 0.0105,
      "step": 1595
    },
    {
      "epoch": 1.1964017991004499,
      "grad_norm": 0.13345656446876325,
      "learning_rate": 0.0001504024118992124,
      "loss": 0.0118,
      "step": 1596
    },
    {
      "epoch": 1.197151424287856,
      "grad_norm": 0.16313411751578458,
      "learning_rate": 0.00015032704252395315,
      "loss": 0.021,
      "step": 1597
    },
    {
      "epoch": 1.1979010494752624,
      "grad_norm": 0.17781430400215803,
      "learning_rate": 0.0001502516348437452,
      "loss": 0.0107,
      "step": 1598
    },
    {
      "epoch": 1.1986506746626686,
      "grad_norm": 0.30242153760052526,
      "learning_rate": 0.0001501761889159829,
      "loss": 0.0211,
      "step": 1599
    },
    {
      "epoch": 1.199400299850075,
      "grad_norm": 0.1716985732561441,
      "learning_rate": 0.0001501007047980897,
      "loss": 0.0193,
      "step": 1600
    },
    {
      "epoch": 1.2001499250374812,
      "grad_norm": 0.16363815472205612,
      "learning_rate": 0.00015002518254751817,
      "loss": 0.0158,
      "step": 1601
    },
    {
      "epoch": 1.2008995502248876,
      "grad_norm": 0.18657535077120233,
      "learning_rate": 0.00014994962222174976,
      "loss": 0.0137,
      "step": 1602
    },
    {
      "epoch": 1.2016491754122938,
      "grad_norm": 0.14295222527249,
      "learning_rate": 0.00014987402387829502,
      "loss": 0.0094,
      "step": 1603
    },
    {
      "epoch": 1.2023988005997002,
      "grad_norm": 0.13886556519199952,
      "learning_rate": 0.00014979838757469343,
      "loss": 0.014,
      "step": 1604
    },
    {
      "epoch": 1.2031484257871066,
      "grad_norm": 0.2803664879457345,
      "learning_rate": 0.00014972271336851332,
      "loss": 0.0216,
      "step": 1605
    },
    {
      "epoch": 1.2038980509745127,
      "grad_norm": 0.2360500791249835,
      "learning_rate": 0.0001496470013173519,
      "loss": 0.0193,
      "step": 1606
    },
    {
      "epoch": 1.204647676161919,
      "grad_norm": 0.1433081534684736,
      "learning_rate": 0.00014957125147883516,
      "loss": 0.02,
      "step": 1607
    },
    {
      "epoch": 1.2053973013493253,
      "grad_norm": 0.07375451718256198,
      "learning_rate": 0.00014949546391061785,
      "loss": 0.0215,
      "step": 1608
    },
    {
      "epoch": 1.2061469265367317,
      "grad_norm": 0.22419918033335512,
      "learning_rate": 0.0001494196386703835,
      "loss": 0.0308,
      "step": 1609
    },
    {
      "epoch": 1.206896551724138,
      "grad_norm": 0.17684058082971194,
      "learning_rate": 0.00014934377581584424,
      "loss": 0.0168,
      "step": 1610
    },
    {
      "epoch": 1.2076461769115443,
      "grad_norm": 0.3677656468075877,
      "learning_rate": 0.00014926787540474082,
      "loss": 0.0384,
      "step": 1611
    },
    {
      "epoch": 1.2083958020989505,
      "grad_norm": 0.18620787001438915,
      "learning_rate": 0.00014919193749484265,
      "loss": 0.0289,
      "step": 1612
    },
    {
      "epoch": 1.2091454272863569,
      "grad_norm": 0.16725832597887177,
      "learning_rate": 0.00014911596214394756,
      "loss": 0.0212,
      "step": 1613
    },
    {
      "epoch": 1.209895052473763,
      "grad_norm": 0.24480319746883517,
      "learning_rate": 0.00014903994940988207,
      "loss": 0.0285,
      "step": 1614
    },
    {
      "epoch": 1.2106446776611695,
      "grad_norm": 0.17387780125153765,
      "learning_rate": 0.00014896389935050092,
      "loss": 0.0202,
      "step": 1615
    },
    {
      "epoch": 1.2113943028485756,
      "grad_norm": 0.06439508057734938,
      "learning_rate": 0.00014888781202368744,
      "loss": 0.0103,
      "step": 1616
    },
    {
      "epoch": 1.212143928035982,
      "grad_norm": 0.635346044877298,
      "learning_rate": 0.00014881168748735328,
      "loss": 0.0779,
      "step": 1617
    },
    {
      "epoch": 1.2128935532233882,
      "grad_norm": 0.23203401133628196,
      "learning_rate": 0.00014873552579943838,
      "loss": 0.027,
      "step": 1618
    },
    {
      "epoch": 1.2136431784107946,
      "grad_norm": 0.21435827140817057,
      "learning_rate": 0.00014865932701791092,
      "loss": 0.0124,
      "step": 1619
    },
    {
      "epoch": 1.214392803598201,
      "grad_norm": 0.15654195261076978,
      "learning_rate": 0.00014858309120076738,
      "loss": 0.0253,
      "step": 1620
    },
    {
      "epoch": 1.2151424287856072,
      "grad_norm": 0.10613248035654502,
      "learning_rate": 0.0001485068184060325,
      "loss": 0.0161,
      "step": 1621
    },
    {
      "epoch": 1.2158920539730136,
      "grad_norm": 0.17812431129324593,
      "learning_rate": 0.00014843050869175895,
      "loss": 0.0228,
      "step": 1622
    },
    {
      "epoch": 1.2166416791604198,
      "grad_norm": 0.13830600870492107,
      "learning_rate": 0.0001483541621160277,
      "loss": 0.0115,
      "step": 1623
    },
    {
      "epoch": 1.2173913043478262,
      "grad_norm": 0.25717512053655084,
      "learning_rate": 0.0001482777787369477,
      "loss": 0.0286,
      "step": 1624
    },
    {
      "epoch": 1.2181409295352323,
      "grad_norm": 0.19518494420243698,
      "learning_rate": 0.00014820135861265586,
      "loss": 0.0224,
      "step": 1625
    },
    {
      "epoch": 1.2188905547226387,
      "grad_norm": 0.19758206936074088,
      "learning_rate": 0.00014812490180131715,
      "loss": 0.0319,
      "step": 1626
    },
    {
      "epoch": 1.219640179910045,
      "grad_norm": 0.4488302974160454,
      "learning_rate": 0.00014804840836112445,
      "loss": 0.0288,
      "step": 1627
    },
    {
      "epoch": 1.2203898050974513,
      "grad_norm": 0.1997144958727106,
      "learning_rate": 0.0001479718783502984,
      "loss": 0.0165,
      "step": 1628
    },
    {
      "epoch": 1.2211394302848575,
      "grad_norm": 0.2055089164972565,
      "learning_rate": 0.00014789531182708766,
      "loss": 0.0244,
      "step": 1629
    },
    {
      "epoch": 1.221889055472264,
      "grad_norm": 0.2692431065482803,
      "learning_rate": 0.0001478187088497686,
      "loss": 0.0228,
      "step": 1630
    },
    {
      "epoch": 1.22263868065967,
      "grad_norm": 0.20209950356916023,
      "learning_rate": 0.0001477420694766452,
      "loss": 0.0351,
      "step": 1631
    },
    {
      "epoch": 1.2233883058470765,
      "grad_norm": 0.19805813440782882,
      "learning_rate": 0.0001476653937660494,
      "loss": 0.0181,
      "step": 1632
    },
    {
      "epoch": 1.2241379310344827,
      "grad_norm": 0.15315490197525022,
      "learning_rate": 0.0001475886817763406,
      "loss": 0.0213,
      "step": 1633
    },
    {
      "epoch": 1.224887556221889,
      "grad_norm": 0.407888110454625,
      "learning_rate": 0.0001475119335659059,
      "loss": 0.0372,
      "step": 1634
    },
    {
      "epoch": 1.2256371814092955,
      "grad_norm": 0.1625357469940467,
      "learning_rate": 0.0001474351491931599,
      "loss": 0.0258,
      "step": 1635
    },
    {
      "epoch": 1.2263868065967016,
      "grad_norm": 0.15337239039044595,
      "learning_rate": 0.0001473583287165448,
      "loss": 0.0188,
      "step": 1636
    },
    {
      "epoch": 1.227136431784108,
      "grad_norm": 0.18287395919285426,
      "learning_rate": 0.00014728147219453025,
      "loss": 0.0246,
      "step": 1637
    },
    {
      "epoch": 1.2278860569715142,
      "grad_norm": 0.322448187413205,
      "learning_rate": 0.00014720457968561335,
      "loss": 0.033,
      "step": 1638
    },
    {
      "epoch": 1.2286356821589206,
      "grad_norm": 0.3009810363920835,
      "learning_rate": 0.0001471276512483185,
      "loss": 0.0282,
      "step": 1639
    },
    {
      "epoch": 1.2293853073463268,
      "grad_norm": 0.07444066072515641,
      "learning_rate": 0.00014705068694119758,
      "loss": 0.01,
      "step": 1640
    },
    {
      "epoch": 1.2301349325337332,
      "grad_norm": 0.1344181162415075,
      "learning_rate": 0.0001469736868228297,
      "loss": 0.0219,
      "step": 1641
    },
    {
      "epoch": 1.2308845577211394,
      "grad_norm": 0.09155453389791111,
      "learning_rate": 0.00014689665095182127,
      "loss": 0.0103,
      "step": 1642
    },
    {
      "epoch": 1.2316341829085458,
      "grad_norm": 0.26088856461980525,
      "learning_rate": 0.00014681957938680578,
      "loss": 0.0247,
      "step": 1643
    },
    {
      "epoch": 1.232383808095952,
      "grad_norm": 0.19509461351710658,
      "learning_rate": 0.00014674247218644405,
      "loss": 0.0141,
      "step": 1644
    },
    {
      "epoch": 1.2331334332833583,
      "grad_norm": 0.4607207890686461,
      "learning_rate": 0.00014666532940942398,
      "loss": 0.0246,
      "step": 1645
    },
    {
      "epoch": 1.2338830584707647,
      "grad_norm": 0.2869915343589999,
      "learning_rate": 0.0001465881511144605,
      "loss": 0.0348,
      "step": 1646
    },
    {
      "epoch": 1.234632683658171,
      "grad_norm": 0.14409772847201524,
      "learning_rate": 0.0001465109373602956,
      "loss": 0.0253,
      "step": 1647
    },
    {
      "epoch": 1.235382308845577,
      "grad_norm": 0.11869113344240399,
      "learning_rate": 0.00014643368820569825,
      "loss": 0.0149,
      "step": 1648
    },
    {
      "epoch": 1.2361319340329835,
      "grad_norm": 0.1286931104098736,
      "learning_rate": 0.0001463564037094644,
      "loss": 0.0108,
      "step": 1649
    },
    {
      "epoch": 1.23688155922039,
      "grad_norm": 0.11931274021012966,
      "learning_rate": 0.00014627908393041682,
      "loss": 0.0141,
      "step": 1650
    },
    {
      "epoch": 1.237631184407796,
      "grad_norm": 0.08123893228448557,
      "learning_rate": 0.00014620172892740524,
      "loss": 0.0083,
      "step": 1651
    },
    {
      "epoch": 1.2383808095952025,
      "grad_norm": 0.3291940283809328,
      "learning_rate": 0.00014612433875930611,
      "loss": 0.0284,
      "step": 1652
    },
    {
      "epoch": 1.2391304347826086,
      "grad_norm": 0.08529523042895133,
      "learning_rate": 0.0001460469134850227,
      "loss": 0.0125,
      "step": 1653
    },
    {
      "epoch": 1.239880059970015,
      "grad_norm": 0.16550478244779057,
      "learning_rate": 0.000145969453163485,
      "loss": 0.0157,
      "step": 1654
    },
    {
      "epoch": 1.2406296851574212,
      "grad_norm": 0.09919390179136478,
      "learning_rate": 0.0001458919578536496,
      "loss": 0.0164,
      "step": 1655
    },
    {
      "epoch": 1.2413793103448276,
      "grad_norm": 0.36606333145203,
      "learning_rate": 0.00014581442761449985,
      "loss": 0.0295,
      "step": 1656
    },
    {
      "epoch": 1.2421289355322338,
      "grad_norm": 0.28225111197774455,
      "learning_rate": 0.00014573686250504554,
      "loss": 0.0353,
      "step": 1657
    },
    {
      "epoch": 1.2428785607196402,
      "grad_norm": 0.13363727729556232,
      "learning_rate": 0.00014565926258432312,
      "loss": 0.0134,
      "step": 1658
    },
    {
      "epoch": 1.2436281859070464,
      "grad_norm": 0.09140998955860476,
      "learning_rate": 0.00014558162791139547,
      "loss": 0.013,
      "step": 1659
    },
    {
      "epoch": 1.2443778110944528,
      "grad_norm": 0.3050733321128361,
      "learning_rate": 0.00014550395854535196,
      "loss": 0.0606,
      "step": 1660
    },
    {
      "epoch": 1.2451274362818592,
      "grad_norm": 0.17081104749370116,
      "learning_rate": 0.00014542625454530832,
      "loss": 0.0332,
      "step": 1661
    },
    {
      "epoch": 1.2458770614692654,
      "grad_norm": 0.2857070156163614,
      "learning_rate": 0.00014534851597040665,
      "loss": 0.0389,
      "step": 1662
    },
    {
      "epoch": 1.2466266866566718,
      "grad_norm": 0.1122991717428358,
      "learning_rate": 0.0001452707428798154,
      "loss": 0.0129,
      "step": 1663
    },
    {
      "epoch": 1.247376311844078,
      "grad_norm": 0.1440697543098693,
      "learning_rate": 0.00014519293533272928,
      "loss": 0.0205,
      "step": 1664
    },
    {
      "epoch": 1.2481259370314843,
      "grad_norm": 0.16650823684741128,
      "learning_rate": 0.00014511509338836922,
      "loss": 0.0243,
      "step": 1665
    },
    {
      "epoch": 1.2488755622188905,
      "grad_norm": 0.08828955807548386,
      "learning_rate": 0.0001450372171059823,
      "loss": 0.019,
      "step": 1666
    },
    {
      "epoch": 1.249625187406297,
      "grad_norm": 0.14752096738070544,
      "learning_rate": 0.0001449593065448418,
      "loss": 0.0233,
      "step": 1667
    },
    {
      "epoch": 1.250374812593703,
      "grad_norm": 0.34921232162258087,
      "learning_rate": 0.00014488136176424703,
      "loss": 0.0379,
      "step": 1668
    },
    {
      "epoch": 1.2511244377811095,
      "grad_norm": 0.11129278503195328,
      "learning_rate": 0.00014480338282352336,
      "loss": 0.0267,
      "step": 1669
    },
    {
      "epoch": 1.2518740629685157,
      "grad_norm": 0.14544728814047606,
      "learning_rate": 0.00014472536978202218,
      "loss": 0.0165,
      "step": 1670
    },
    {
      "epoch": 1.252623688155922,
      "grad_norm": 0.22625628725911506,
      "learning_rate": 0.00014464732269912082,
      "loss": 0.0221,
      "step": 1671
    },
    {
      "epoch": 1.2533733133433285,
      "grad_norm": 0.13860636776343901,
      "learning_rate": 0.00014456924163422255,
      "loss": 0.0243,
      "step": 1672
    },
    {
      "epoch": 1.2541229385307346,
      "grad_norm": 0.1101369473658894,
      "learning_rate": 0.0001444911266467564,
      "loss": 0.0096,
      "step": 1673
    },
    {
      "epoch": 1.2548725637181408,
      "grad_norm": 0.2876119675897652,
      "learning_rate": 0.0001444129777961774,
      "loss": 0.0202,
      "step": 1674
    },
    {
      "epoch": 1.2556221889055472,
      "grad_norm": 0.29583502202052947,
      "learning_rate": 0.00014433479514196616,
      "loss": 0.0822,
      "step": 1675
    },
    {
      "epoch": 1.2563718140929536,
      "grad_norm": 0.21487573028514387,
      "learning_rate": 0.00014425657874362913,
      "loss": 0.0348,
      "step": 1676
    },
    {
      "epoch": 1.2571214392803598,
      "grad_norm": 0.11451811504867611,
      "learning_rate": 0.00014417832866069847,
      "loss": 0.0225,
      "step": 1677
    },
    {
      "epoch": 1.2578710644677662,
      "grad_norm": 0.2173650139609283,
      "learning_rate": 0.00014410004495273186,
      "loss": 0.0299,
      "step": 1678
    },
    {
      "epoch": 1.2586206896551724,
      "grad_norm": 0.3899048511072257,
      "learning_rate": 0.0001440217276793127,
      "loss": 0.0206,
      "step": 1679
    },
    {
      "epoch": 1.2593703148425788,
      "grad_norm": 0.3059178365171424,
      "learning_rate": 0.00014394337690004985,
      "loss": 0.0387,
      "step": 1680
    },
    {
      "epoch": 1.260119940029985,
      "grad_norm": 0.09967613478175827,
      "learning_rate": 0.00014386499267457774,
      "loss": 0.0157,
      "step": 1681
    },
    {
      "epoch": 1.2608695652173914,
      "grad_norm": 0.1172357014637228,
      "learning_rate": 0.00014378657506255616,
      "loss": 0.0136,
      "step": 1682
    },
    {
      "epoch": 1.2616191904047975,
      "grad_norm": 0.18697067489098376,
      "learning_rate": 0.00014370812412367042,
      "loss": 0.0378,
      "step": 1683
    },
    {
      "epoch": 1.262368815592204,
      "grad_norm": 0.19464374859211744,
      "learning_rate": 0.00014362963991763114,
      "loss": 0.0323,
      "step": 1684
    },
    {
      "epoch": 1.26311844077961,
      "grad_norm": 0.10611168216251465,
      "learning_rate": 0.0001435511225041742,
      "loss": 0.0149,
      "step": 1685
    },
    {
      "epoch": 1.2638680659670165,
      "grad_norm": 0.2880623504224047,
      "learning_rate": 0.00014347257194306093,
      "loss": 0.0464,
      "step": 1686
    },
    {
      "epoch": 1.264617691154423,
      "grad_norm": 0.15171315173537525,
      "learning_rate": 0.00014339398829407767,
      "loss": 0.0175,
      "step": 1687
    },
    {
      "epoch": 1.265367316341829,
      "grad_norm": 0.2888464808325791,
      "learning_rate": 0.0001433153716170361,
      "loss": 0.0406,
      "step": 1688
    },
    {
      "epoch": 1.2661169415292353,
      "grad_norm": 0.10076514229124206,
      "learning_rate": 0.00014323672197177304,
      "loss": 0.0251,
      "step": 1689
    },
    {
      "epoch": 1.2668665667166417,
      "grad_norm": 0.20768788736228558,
      "learning_rate": 0.00014315803941815028,
      "loss": 0.0293,
      "step": 1690
    },
    {
      "epoch": 1.267616191904048,
      "grad_norm": 0.20206611512917563,
      "learning_rate": 0.00014307932401605477,
      "loss": 0.0286,
      "step": 1691
    },
    {
      "epoch": 1.2683658170914542,
      "grad_norm": 0.15658400789301777,
      "learning_rate": 0.00014300057582539844,
      "loss": 0.0214,
      "step": 1692
    },
    {
      "epoch": 1.2691154422788606,
      "grad_norm": 0.14329880535204859,
      "learning_rate": 0.00014292179490611812,
      "loss": 0.0147,
      "step": 1693
    },
    {
      "epoch": 1.2698650674662668,
      "grad_norm": 0.12107833682851883,
      "learning_rate": 0.00014284298131817563,
      "loss": 0.02,
      "step": 1694
    },
    {
      "epoch": 1.2706146926536732,
      "grad_norm": 0.17914505564164096,
      "learning_rate": 0.00014276413512155756,
      "loss": 0.02,
      "step": 1695
    },
    {
      "epoch": 1.2713643178410794,
      "grad_norm": 0.21049302813792745,
      "learning_rate": 0.00014268525637627541,
      "loss": 0.0409,
      "step": 1696
    },
    {
      "epoch": 1.2721139430284858,
      "grad_norm": 0.08102354032890279,
      "learning_rate": 0.00014260634514236544,
      "loss": 0.0101,
      "step": 1697
    },
    {
      "epoch": 1.272863568215892,
      "grad_norm": 0.1820823122811385,
      "learning_rate": 0.00014252740147988857,
      "loss": 0.018,
      "step": 1698
    },
    {
      "epoch": 1.2736131934032984,
      "grad_norm": 0.12064252019808996,
      "learning_rate": 0.00014244842544893047,
      "loss": 0.0158,
      "step": 1699
    },
    {
      "epoch": 1.2743628185907045,
      "grad_norm": 0.18327142948153743,
      "learning_rate": 0.00014236941710960143,
      "loss": 0.0133,
      "step": 1700
    },
    {
      "epoch": 1.275112443778111,
      "grad_norm": 0.12017554093437577,
      "learning_rate": 0.00014229037652203628,
      "loss": 0.0142,
      "step": 1701
    },
    {
      "epoch": 1.2758620689655173,
      "grad_norm": 0.2064898698033883,
      "learning_rate": 0.00014221130374639454,
      "loss": 0.02,
      "step": 1702
    },
    {
      "epoch": 1.2766116941529235,
      "grad_norm": 0.11760686399607434,
      "learning_rate": 0.00014213219884286004,
      "loss": 0.007,
      "step": 1703
    },
    {
      "epoch": 1.2773613193403297,
      "grad_norm": 0.16875461561682745,
      "learning_rate": 0.00014205306187164117,
      "loss": 0.0222,
      "step": 1704
    },
    {
      "epoch": 1.278110944527736,
      "grad_norm": 0.21819466446133043,
      "learning_rate": 0.0001419738928929707,
      "loss": 0.0132,
      "step": 1705
    },
    {
      "epoch": 1.2788605697151425,
      "grad_norm": 0.07003267858663324,
      "learning_rate": 0.00014189469196710584,
      "loss": 0.0079,
      "step": 1706
    },
    {
      "epoch": 1.2796101949025487,
      "grad_norm": 0.23700881089716225,
      "learning_rate": 0.00014181545915432795,
      "loss": 0.0392,
      "step": 1707
    },
    {
      "epoch": 1.280359820089955,
      "grad_norm": 0.29435932277262994,
      "learning_rate": 0.0001417361945149428,
      "loss": 0.021,
      "step": 1708
    },
    {
      "epoch": 1.2811094452773613,
      "grad_norm": 0.14827007937512016,
      "learning_rate": 0.00014165689810928035,
      "loss": 0.0189,
      "step": 1709
    },
    {
      "epoch": 1.2818590704647677,
      "grad_norm": 0.11574931571363065,
      "learning_rate": 0.0001415775699976947,
      "loss": 0.0118,
      "step": 1710
    },
    {
      "epoch": 1.2826086956521738,
      "grad_norm": 0.2424262114701182,
      "learning_rate": 0.0001414982102405641,
      "loss": 0.0214,
      "step": 1711
    },
    {
      "epoch": 1.2833583208395802,
      "grad_norm": 0.1395124838749881,
      "learning_rate": 0.00014141881889829095,
      "loss": 0.0189,
      "step": 1712
    },
    {
      "epoch": 1.2841079460269866,
      "grad_norm": 0.2711886619263315,
      "learning_rate": 0.0001413393960313016,
      "loss": 0.0215,
      "step": 1713
    },
    {
      "epoch": 1.2848575712143928,
      "grad_norm": 0.4334505201331919,
      "learning_rate": 0.00014125994170004644,
      "loss": 0.0278,
      "step": 1714
    },
    {
      "epoch": 1.285607196401799,
      "grad_norm": 0.1254629293847402,
      "learning_rate": 0.0001411804559649998,
      "loss": 0.0117,
      "step": 1715
    },
    {
      "epoch": 1.2863568215892054,
      "grad_norm": 0.12718887367154608,
      "learning_rate": 0.0001411009388866599,
      "loss": 0.0171,
      "step": 1716
    },
    {
      "epoch": 1.2871064467766118,
      "grad_norm": 0.19073545908028275,
      "learning_rate": 0.00014102139052554882,
      "loss": 0.0176,
      "step": 1717
    },
    {
      "epoch": 1.287856071964018,
      "grad_norm": 0.13287567380141152,
      "learning_rate": 0.00014094181094221247,
      "loss": 0.0158,
      "step": 1718
    },
    {
      "epoch": 1.2886056971514244,
      "grad_norm": 0.0876281346085455,
      "learning_rate": 0.00014086220019722052,
      "loss": 0.0092,
      "step": 1719
    },
    {
      "epoch": 1.2893553223388305,
      "grad_norm": 0.1728777781891654,
      "learning_rate": 0.00014078255835116635,
      "loss": 0.0443,
      "step": 1720
    },
    {
      "epoch": 1.290104947526237,
      "grad_norm": 0.10517570102792818,
      "learning_rate": 0.00014070288546466696,
      "loss": 0.0135,
      "step": 1721
    },
    {
      "epoch": 1.2908545727136431,
      "grad_norm": 0.2599468156608288,
      "learning_rate": 0.00014062318159836304,
      "loss": 0.0437,
      "step": 1722
    },
    {
      "epoch": 1.2916041979010495,
      "grad_norm": 0.30636437280383555,
      "learning_rate": 0.00014054344681291888,
      "loss": 0.0445,
      "step": 1723
    },
    {
      "epoch": 1.2923538230884557,
      "grad_norm": 0.2589336307064375,
      "learning_rate": 0.00014046368116902227,
      "loss": 0.0236,
      "step": 1724
    },
    {
      "epoch": 1.293103448275862,
      "grad_norm": 0.18574128506800772,
      "learning_rate": 0.00014038388472738445,
      "loss": 0.0168,
      "step": 1725
    },
    {
      "epoch": 1.2938530734632683,
      "grad_norm": 0.07816126620288197,
      "learning_rate": 0.00014030405754874014,
      "loss": 0.0118,
      "step": 1726
    },
    {
      "epoch": 1.2946026986506747,
      "grad_norm": 0.10180035843931197,
      "learning_rate": 0.00014022419969384747,
      "loss": 0.0089,
      "step": 1727
    },
    {
      "epoch": 1.295352323838081,
      "grad_norm": 0.0925964533291629,
      "learning_rate": 0.0001401443112234879,
      "loss": 0.0086,
      "step": 1728
    },
    {
      "epoch": 1.2961019490254873,
      "grad_norm": 0.20432854716273183,
      "learning_rate": 0.0001400643921984662,
      "loss": 0.0226,
      "step": 1729
    },
    {
      "epoch": 1.2968515742128934,
      "grad_norm": 0.19414344538018496,
      "learning_rate": 0.00013998444267961031,
      "loss": 0.0201,
      "step": 1730
    },
    {
      "epoch": 1.2976011994002998,
      "grad_norm": 0.4773435279947028,
      "learning_rate": 0.00013990446272777157,
      "loss": 0.0371,
      "step": 1731
    },
    {
      "epoch": 1.2983508245877062,
      "grad_norm": 0.19809194521611023,
      "learning_rate": 0.00013982445240382426,
      "loss": 0.0243,
      "step": 1732
    },
    {
      "epoch": 1.2991004497751124,
      "grad_norm": 0.30608293614416343,
      "learning_rate": 0.00013974441176866598,
      "loss": 0.024,
      "step": 1733
    },
    {
      "epoch": 1.2998500749625188,
      "grad_norm": 0.16507027944955746,
      "learning_rate": 0.0001396643408832172,
      "loss": 0.0288,
      "step": 1734
    },
    {
      "epoch": 1.300599700149925,
      "grad_norm": 0.0654535038808333,
      "learning_rate": 0.00013958423980842163,
      "loss": 0.0055,
      "step": 1735
    },
    {
      "epoch": 1.3013493253373314,
      "grad_norm": 0.21694312244015637,
      "learning_rate": 0.00013950410860524577,
      "loss": 0.0329,
      "step": 1736
    },
    {
      "epoch": 1.3020989505247376,
      "grad_norm": 0.3783138781837035,
      "learning_rate": 0.00013942394733467916,
      "loss": 0.0341,
      "step": 1737
    },
    {
      "epoch": 1.302848575712144,
      "grad_norm": 0.14809260941669292,
      "learning_rate": 0.0001393437560577342,
      "loss": 0.0325,
      "step": 1738
    },
    {
      "epoch": 1.3035982008995504,
      "grad_norm": 0.1703374135823408,
      "learning_rate": 0.00013926353483544615,
      "loss": 0.0245,
      "step": 1739
    },
    {
      "epoch": 1.3043478260869565,
      "grad_norm": 0.1380282229997763,
      "learning_rate": 0.00013918328372887294,
      "loss": 0.0133,
      "step": 1740
    },
    {
      "epoch": 1.3050974512743627,
      "grad_norm": 0.06564048171598381,
      "learning_rate": 0.00013910300279909544,
      "loss": 0.0099,
      "step": 1741
    },
    {
      "epoch": 1.3058470764617691,
      "grad_norm": 0.11204069567707248,
      "learning_rate": 0.0001390226921072171,
      "loss": 0.0084,
      "step": 1742
    },
    {
      "epoch": 1.3065967016491755,
      "grad_norm": 0.10595645979145439,
      "learning_rate": 0.00013894235171436399,
      "loss": 0.0088,
      "step": 1743
    },
    {
      "epoch": 1.3073463268365817,
      "grad_norm": 0.2804525087464718,
      "learning_rate": 0.00013886198168168486,
      "loss": 0.0143,
      "step": 1744
    },
    {
      "epoch": 1.3080959520239879,
      "grad_norm": 0.15181289771519554,
      "learning_rate": 0.00013878158207035102,
      "loss": 0.0163,
      "step": 1745
    },
    {
      "epoch": 1.3088455772113943,
      "grad_norm": 0.10709342533170021,
      "learning_rate": 0.00013870115294155623,
      "loss": 0.0145,
      "step": 1746
    },
    {
      "epoch": 1.3095952023988007,
      "grad_norm": 0.16300325070420085,
      "learning_rate": 0.0001386206943565168,
      "loss": 0.0254,
      "step": 1747
    },
    {
      "epoch": 1.3103448275862069,
      "grad_norm": 0.07927816829120087,
      "learning_rate": 0.0001385402063764714,
      "loss": 0.0127,
      "step": 1748
    },
    {
      "epoch": 1.3110944527736133,
      "grad_norm": 0.14735840705021575,
      "learning_rate": 0.0001384596890626811,
      "loss": 0.0203,
      "step": 1749
    },
    {
      "epoch": 1.3118440779610194,
      "grad_norm": 0.2586943019227501,
      "learning_rate": 0.00013837914247642928,
      "loss": 0.0518,
      "step": 1750
    },
    {
      "epoch": 1.3125937031484258,
      "grad_norm": 0.12290073980900268,
      "learning_rate": 0.00013829856667902157,
      "loss": 0.014,
      "step": 1751
    },
    {
      "epoch": 1.313343328335832,
      "grad_norm": 0.1621805016410538,
      "learning_rate": 0.00013821796173178595,
      "loss": 0.0307,
      "step": 1752
    },
    {
      "epoch": 1.3140929535232384,
      "grad_norm": 0.21411790431804426,
      "learning_rate": 0.00013813732769607238,
      "loss": 0.0263,
      "step": 1753
    },
    {
      "epoch": 1.3148425787106448,
      "grad_norm": 0.11902217132590387,
      "learning_rate": 0.00013805666463325325,
      "loss": 0.0133,
      "step": 1754
    },
    {
      "epoch": 1.315592203898051,
      "grad_norm": 0.1449363778880504,
      "learning_rate": 0.00013797597260472273,
      "loss": 0.0291,
      "step": 1755
    },
    {
      "epoch": 1.3163418290854572,
      "grad_norm": 0.09905236252703252,
      "learning_rate": 0.00013789525167189726,
      "loss": 0.0139,
      "step": 1756
    },
    {
      "epoch": 1.3170914542728636,
      "grad_norm": 0.3075729201212794,
      "learning_rate": 0.00013781450189621517,
      "loss": 0.0285,
      "step": 1757
    },
    {
      "epoch": 1.31784107946027,
      "grad_norm": 0.35471879427394654,
      "learning_rate": 0.00013773372333913682,
      "loss": 0.0549,
      "step": 1758
    },
    {
      "epoch": 1.3185907046476761,
      "grad_norm": 0.12116073666563233,
      "learning_rate": 0.0001376529160621444,
      "loss": 0.0228,
      "step": 1759
    },
    {
      "epoch": 1.3193403298350825,
      "grad_norm": 0.21253060476515195,
      "learning_rate": 0.000137572080126742,
      "loss": 0.018,
      "step": 1760
    },
    {
      "epoch": 1.3200899550224887,
      "grad_norm": 0.09814476061104814,
      "learning_rate": 0.0001374912155944555,
      "loss": 0.0208,
      "step": 1761
    },
    {
      "epoch": 1.3208395802098951,
      "grad_norm": 0.1093144771005278,
      "learning_rate": 0.00013741032252683263,
      "loss": 0.015,
      "step": 1762
    },
    {
      "epoch": 1.3215892053973013,
      "grad_norm": 0.14958978266296255,
      "learning_rate": 0.00013732940098544268,
      "loss": 0.0166,
      "step": 1763
    },
    {
      "epoch": 1.3223388305847077,
      "grad_norm": 0.16548122115072086,
      "learning_rate": 0.00013724845103187678,
      "loss": 0.0161,
      "step": 1764
    },
    {
      "epoch": 1.3230884557721139,
      "grad_norm": 0.11831041233749966,
      "learning_rate": 0.00013716747272774754,
      "loss": 0.0199,
      "step": 1765
    },
    {
      "epoch": 1.3238380809595203,
      "grad_norm": 0.11784371645916646,
      "learning_rate": 0.00013708646613468925,
      "loss": 0.0233,
      "step": 1766
    },
    {
      "epoch": 1.3245877061469264,
      "grad_norm": 0.18173569247378074,
      "learning_rate": 0.0001370054313143577,
      "loss": 0.018,
      "step": 1767
    },
    {
      "epoch": 1.3253373313343328,
      "grad_norm": 0.7794282666624248,
      "learning_rate": 0.00013692436832843014,
      "loss": 0.0424,
      "step": 1768
    },
    {
      "epoch": 1.3260869565217392,
      "grad_norm": 0.18519517663898552,
      "learning_rate": 0.00013684327723860527,
      "loss": 0.0296,
      "step": 1769
    },
    {
      "epoch": 1.3268365817091454,
      "grad_norm": 0.18195460378044268,
      "learning_rate": 0.00013676215810660324,
      "loss": 0.0203,
      "step": 1770
    },
    {
      "epoch": 1.3275862068965516,
      "grad_norm": 0.19484493296286087,
      "learning_rate": 0.00013668101099416547,
      "loss": 0.0137,
      "step": 1771
    },
    {
      "epoch": 1.328335832083958,
      "grad_norm": 0.3283993250061566,
      "learning_rate": 0.00013659983596305466,
      "loss": 0.0288,
      "step": 1772
    },
    {
      "epoch": 1.3290854572713644,
      "grad_norm": 0.4043023801075964,
      "learning_rate": 0.0001365186330750548,
      "loss": 0.0611,
      "step": 1773
    },
    {
      "epoch": 1.3298350824587706,
      "grad_norm": 0.15411743278139997,
      "learning_rate": 0.0001364374023919712,
      "loss": 0.0119,
      "step": 1774
    },
    {
      "epoch": 1.330584707646177,
      "grad_norm": 0.19170968954122777,
      "learning_rate": 0.00013635614397563002,
      "loss": 0.0232,
      "step": 1775
    },
    {
      "epoch": 1.3313343328335832,
      "grad_norm": 0.17839369834152638,
      "learning_rate": 0.00013627485788787885,
      "loss": 0.0106,
      "step": 1776
    },
    {
      "epoch": 1.3320839580209896,
      "grad_norm": 0.25787303687719876,
      "learning_rate": 0.00013619354419058618,
      "loss": 0.0241,
      "step": 1777
    },
    {
      "epoch": 1.3328335832083957,
      "grad_norm": 0.47991060517567485,
      "learning_rate": 0.00013611220294564152,
      "loss": 0.0625,
      "step": 1778
    },
    {
      "epoch": 1.3335832083958021,
      "grad_norm": 0.2458517936158923,
      "learning_rate": 0.00013603083421495534,
      "loss": 0.0399,
      "step": 1779
    },
    {
      "epoch": 1.3343328335832085,
      "grad_norm": 0.05758846150700338,
      "learning_rate": 0.0001359494380604591,
      "loss": 0.0078,
      "step": 1780
    },
    {
      "epoch": 1.3350824587706147,
      "grad_norm": 0.12083746471620882,
      "learning_rate": 0.0001358680145441051,
      "loss": 0.0195,
      "step": 1781
    },
    {
      "epoch": 1.3358320839580209,
      "grad_norm": 0.36133090387357214,
      "learning_rate": 0.0001357865637278664,
      "loss": 0.0393,
      "step": 1782
    },
    {
      "epoch": 1.3365817091454273,
      "grad_norm": 0.18830531508318982,
      "learning_rate": 0.00013570508567373697,
      "loss": 0.021,
      "step": 1783
    },
    {
      "epoch": 1.3373313343328337,
      "grad_norm": 0.18900867761081533,
      "learning_rate": 0.00013562358044373135,
      "loss": 0.0195,
      "step": 1784
    },
    {
      "epoch": 1.3380809595202399,
      "grad_norm": 0.16864151254310344,
      "learning_rate": 0.00013554204809988489,
      "loss": 0.0148,
      "step": 1785
    },
    {
      "epoch": 1.338830584707646,
      "grad_norm": 0.1821482393097731,
      "learning_rate": 0.00013546048870425356,
      "loss": 0.0315,
      "step": 1786
    },
    {
      "epoch": 1.3395802098950524,
      "grad_norm": 0.22791765279336143,
      "learning_rate": 0.00013537890231891389,
      "loss": 0.0264,
      "step": 1787
    },
    {
      "epoch": 1.3403298350824588,
      "grad_norm": 0.11951864033785457,
      "learning_rate": 0.00013529728900596292,
      "loss": 0.0186,
      "step": 1788
    },
    {
      "epoch": 1.341079460269865,
      "grad_norm": 0.2607319327329998,
      "learning_rate": 0.00013521564882751824,
      "loss": 0.0306,
      "step": 1789
    },
    {
      "epoch": 1.3418290854572714,
      "grad_norm": 0.13017071114069304,
      "learning_rate": 0.0001351339818457179,
      "loss": 0.047,
      "step": 1790
    },
    {
      "epoch": 1.3425787106446776,
      "grad_norm": 0.1450659367892797,
      "learning_rate": 0.00013505228812272027,
      "loss": 0.028,
      "step": 1791
    },
    {
      "epoch": 1.343328335832084,
      "grad_norm": 0.14477929492194744,
      "learning_rate": 0.00013497056772070418,
      "loss": 0.0228,
      "step": 1792
    },
    {
      "epoch": 1.3440779610194902,
      "grad_norm": 0.22976772294738085,
      "learning_rate": 0.0001348888207018687,
      "loss": 0.0261,
      "step": 1793
    },
    {
      "epoch": 1.3448275862068966,
      "grad_norm": 0.18993895374099645,
      "learning_rate": 0.0001348070471284331,
      "loss": 0.0272,
      "step": 1794
    },
    {
      "epoch": 1.345577211394303,
      "grad_norm": 0.18607634443900709,
      "learning_rate": 0.00013472524706263704,
      "loss": 0.0164,
      "step": 1795
    },
    {
      "epoch": 1.3463268365817092,
      "grad_norm": 0.21608895921140736,
      "learning_rate": 0.00013464342056674014,
      "loss": 0.0266,
      "step": 1796
    },
    {
      "epoch": 1.3470764617691153,
      "grad_norm": 0.13419548434843545,
      "learning_rate": 0.00013456156770302232,
      "loss": 0.0182,
      "step": 1797
    },
    {
      "epoch": 1.3478260869565217,
      "grad_norm": 0.09113490437408855,
      "learning_rate": 0.00013447968853378339,
      "loss": 0.0164,
      "step": 1798
    },
    {
      "epoch": 1.3485757121439281,
      "grad_norm": 0.1880407479342471,
      "learning_rate": 0.00013439778312134335,
      "loss": 0.0304,
      "step": 1799
    },
    {
      "epoch": 1.3493253373313343,
      "grad_norm": 0.20552205705509324,
      "learning_rate": 0.00013431585152804204,
      "loss": 0.0174,
      "step": 1800
    },
    {
      "epoch": 1.3500749625187407,
      "grad_norm": 0.30216264696825906,
      "learning_rate": 0.00013423389381623933,
      "loss": 0.0389,
      "step": 1801
    },
    {
      "epoch": 1.3508245877061469,
      "grad_norm": 0.1466753916995075,
      "learning_rate": 0.00013415191004831486,
      "loss": 0.0214,
      "step": 1802
    },
    {
      "epoch": 1.3515742128935533,
      "grad_norm": 0.13263747900098585,
      "learning_rate": 0.00013406990028666825,
      "loss": 0.0249,
      "step": 1803
    },
    {
      "epoch": 1.3523238380809595,
      "grad_norm": 0.3684032218976006,
      "learning_rate": 0.0001339878645937187,
      "loss": 0.0269,
      "step": 1804
    },
    {
      "epoch": 1.3530734632683659,
      "grad_norm": 0.07399337775573261,
      "learning_rate": 0.0001339058030319054,
      "loss": 0.0119,
      "step": 1805
    },
    {
      "epoch": 1.353823088455772,
      "grad_norm": 0.2539273759373133,
      "learning_rate": 0.000133823715663687,
      "loss": 0.0347,
      "step": 1806
    },
    {
      "epoch": 1.3545727136431784,
      "grad_norm": 0.10545031650173455,
      "learning_rate": 0.00013374160255154197,
      "loss": 0.0188,
      "step": 1807
    },
    {
      "epoch": 1.3553223388305846,
      "grad_norm": 0.058772091983560384,
      "learning_rate": 0.00013365946375796817,
      "loss": 0.0065,
      "step": 1808
    },
    {
      "epoch": 1.356071964017991,
      "grad_norm": 0.11314109561850787,
      "learning_rate": 0.00013357729934548325,
      "loss": 0.0119,
      "step": 1809
    },
    {
      "epoch": 1.3568215892053974,
      "grad_norm": 0.06928897914135622,
      "learning_rate": 0.00013349510937662415,
      "loss": 0.0062,
      "step": 1810
    },
    {
      "epoch": 1.3575712143928036,
      "grad_norm": 0.11684449276906832,
      "learning_rate": 0.0001334128939139474,
      "loss": 0.0134,
      "step": 1811
    },
    {
      "epoch": 1.3583208395802098,
      "grad_norm": 0.23702047079708088,
      "learning_rate": 0.00013333065302002887,
      "loss": 0.0206,
      "step": 1812
    },
    {
      "epoch": 1.3590704647676162,
      "grad_norm": 0.2643609076687036,
      "learning_rate": 0.0001332483867574638,
      "loss": 0.0116,
      "step": 1813
    },
    {
      "epoch": 1.3598200899550226,
      "grad_norm": 0.13595700377100592,
      "learning_rate": 0.0001331660951888667,
      "loss": 0.0142,
      "step": 1814
    },
    {
      "epoch": 1.3605697151424287,
      "grad_norm": 0.13346544162743765,
      "learning_rate": 0.00013308377837687143,
      "loss": 0.0254,
      "step": 1815
    },
    {
      "epoch": 1.3613193403298351,
      "grad_norm": 0.1695190423319145,
      "learning_rate": 0.00013300143638413098,
      "loss": 0.0213,
      "step": 1816
    },
    {
      "epoch": 1.3620689655172413,
      "grad_norm": 0.10765569160056693,
      "learning_rate": 0.00013291906927331753,
      "loss": 0.01,
      "step": 1817
    },
    {
      "epoch": 1.3628185907046477,
      "grad_norm": 0.1354854839518553,
      "learning_rate": 0.00013283667710712243,
      "loss": 0.0066,
      "step": 1818
    },
    {
      "epoch": 1.363568215892054,
      "grad_norm": 0.22292484082791938,
      "learning_rate": 0.00013275425994825605,
      "loss": 0.0151,
      "step": 1819
    },
    {
      "epoch": 1.3643178410794603,
      "grad_norm": 0.14596732401139165,
      "learning_rate": 0.0001326718178594477,
      "loss": 0.0204,
      "step": 1820
    },
    {
      "epoch": 1.3650674662668667,
      "grad_norm": 0.17866809132707354,
      "learning_rate": 0.00013258935090344585,
      "loss": 0.0363,
      "step": 1821
    },
    {
      "epoch": 1.3658170914542729,
      "grad_norm": 0.17175637013243722,
      "learning_rate": 0.00013250685914301775,
      "loss": 0.0153,
      "step": 1822
    },
    {
      "epoch": 1.366566716641679,
      "grad_norm": 0.160782474163614,
      "learning_rate": 0.0001324243426409496,
      "loss": 0.0265,
      "step": 1823
    },
    {
      "epoch": 1.3673163418290855,
      "grad_norm": 0.16591399321605485,
      "learning_rate": 0.00013234180146004637,
      "loss": 0.0201,
      "step": 1824
    },
    {
      "epoch": 1.3680659670164919,
      "grad_norm": 0.30566822867483184,
      "learning_rate": 0.00013225923566313193,
      "loss": 0.0361,
      "step": 1825
    },
    {
      "epoch": 1.368815592203898,
      "grad_norm": 0.10134118807960453,
      "learning_rate": 0.00013217664531304875,
      "loss": 0.0153,
      "step": 1826
    },
    {
      "epoch": 1.3695652173913042,
      "grad_norm": 0.2673027205910499,
      "learning_rate": 0.00013209403047265803,
      "loss": 0.0313,
      "step": 1827
    },
    {
      "epoch": 1.3703148425787106,
      "grad_norm": 0.18874733815150369,
      "learning_rate": 0.00013201139120483968,
      "loss": 0.0241,
      "step": 1828
    },
    {
      "epoch": 1.371064467766117,
      "grad_norm": 0.11031481892631355,
      "learning_rate": 0.00013192872757249212,
      "loss": 0.0146,
      "step": 1829
    },
    {
      "epoch": 1.3718140929535232,
      "grad_norm": 0.18598840361478705,
      "learning_rate": 0.00013184603963853234,
      "loss": 0.0237,
      "step": 1830
    },
    {
      "epoch": 1.3725637181409296,
      "grad_norm": 0.04790301796130572,
      "learning_rate": 0.00013176332746589586,
      "loss": 0.008,
      "step": 1831
    },
    {
      "epoch": 1.3733133433283358,
      "grad_norm": 0.14298108928099823,
      "learning_rate": 0.00013168059111753658,
      "loss": 0.0205,
      "step": 1832
    },
    {
      "epoch": 1.3740629685157422,
      "grad_norm": 0.06819818800878674,
      "learning_rate": 0.00013159783065642684,
      "loss": 0.0051,
      "step": 1833
    },
    {
      "epoch": 1.3748125937031483,
      "grad_norm": 0.17992706431322583,
      "learning_rate": 0.00013151504614555734,
      "loss": 0.0183,
      "step": 1834
    },
    {
      "epoch": 1.3755622188905547,
      "grad_norm": 0.04716179303045061,
      "learning_rate": 0.0001314322376479371,
      "loss": 0.0056,
      "step": 1835
    },
    {
      "epoch": 1.3763118440779611,
      "grad_norm": 0.06108436289594824,
      "learning_rate": 0.0001313494052265934,
      "loss": 0.0075,
      "step": 1836
    },
    {
      "epoch": 1.3770614692653673,
      "grad_norm": 0.1114869532383892,
      "learning_rate": 0.0001312665489445716,
      "loss": 0.0163,
      "step": 1837
    },
    {
      "epoch": 1.3778110944527735,
      "grad_norm": 0.14439215967136948,
      "learning_rate": 0.0001311836688649354,
      "loss": 0.0242,
      "step": 1838
    },
    {
      "epoch": 1.37856071964018,
      "grad_norm": 0.13410528402626273,
      "learning_rate": 0.0001311007650507665,
      "loss": 0.0219,
      "step": 1839
    },
    {
      "epoch": 1.3793103448275863,
      "grad_norm": 0.10440850471031544,
      "learning_rate": 0.00013101783756516472,
      "loss": 0.014,
      "step": 1840
    },
    {
      "epoch": 1.3800599700149925,
      "grad_norm": 0.09242749157688925,
      "learning_rate": 0.0001309348864712479,
      "loss": 0.0126,
      "step": 1841
    },
    {
      "epoch": 1.3808095952023989,
      "grad_norm": 0.0887009346031432,
      "learning_rate": 0.00013085191183215177,
      "loss": 0.0209,
      "step": 1842
    },
    {
      "epoch": 1.381559220389805,
      "grad_norm": 0.2960216409085203,
      "learning_rate": 0.0001307689137110301,
      "loss": 0.0266,
      "step": 1843
    },
    {
      "epoch": 1.3823088455772115,
      "grad_norm": 0.068323210842311,
      "learning_rate": 0.00013068589217105441,
      "loss": 0.0055,
      "step": 1844
    },
    {
      "epoch": 1.3830584707646176,
      "grad_norm": 0.08638662473406213,
      "learning_rate": 0.00013060284727541414,
      "loss": 0.007,
      "step": 1845
    },
    {
      "epoch": 1.383808095952024,
      "grad_norm": 0.1702916586839961,
      "learning_rate": 0.00013051977908731644,
      "loss": 0.0218,
      "step": 1846
    },
    {
      "epoch": 1.3845577211394302,
      "grad_norm": 0.18436823215505085,
      "learning_rate": 0.00013043668766998627,
      "loss": 0.0125,
      "step": 1847
    },
    {
      "epoch": 1.3853073463268366,
      "grad_norm": 0.44702716210172605,
      "learning_rate": 0.0001303535730866662,
      "loss": 0.0215,
      "step": 1848
    },
    {
      "epoch": 1.3860569715142428,
      "grad_norm": 0.10089046147113483,
      "learning_rate": 0.0001302704354006164,
      "loss": 0.0197,
      "step": 1849
    },
    {
      "epoch": 1.3868065967016492,
      "grad_norm": 0.26863373480847147,
      "learning_rate": 0.00013018727467511474,
      "loss": 0.031,
      "step": 1850
    },
    {
      "epoch": 1.3875562218890556,
      "grad_norm": 0.22774278816662305,
      "learning_rate": 0.00013010409097345655,
      "loss": 0.0602,
      "step": 1851
    },
    {
      "epoch": 1.3883058470764618,
      "grad_norm": 0.14560522069571033,
      "learning_rate": 0.00013002088435895464,
      "loss": 0.0178,
      "step": 1852
    },
    {
      "epoch": 1.389055472263868,
      "grad_norm": 0.12305291457680514,
      "learning_rate": 0.00012993765489493928,
      "loss": 0.0216,
      "step": 1853
    },
    {
      "epoch": 1.3898050974512743,
      "grad_norm": 0.15409826230237533,
      "learning_rate": 0.00012985440264475813,
      "loss": 0.019,
      "step": 1854
    },
    {
      "epoch": 1.3905547226386807,
      "grad_norm": 0.21164672474158286,
      "learning_rate": 0.0001297711276717762,
      "loss": 0.0117,
      "step": 1855
    },
    {
      "epoch": 1.391304347826087,
      "grad_norm": 0.1355214980132497,
      "learning_rate": 0.00012968783003937577,
      "loss": 0.0219,
      "step": 1856
    },
    {
      "epoch": 1.3920539730134933,
      "grad_norm": 0.12625414203875335,
      "learning_rate": 0.00012960450981095643,
      "loss": 0.0169,
      "step": 1857
    },
    {
      "epoch": 1.3928035982008995,
      "grad_norm": 0.3024182672541223,
      "learning_rate": 0.00012952116704993482,
      "loss": 0.0168,
      "step": 1858
    },
    {
      "epoch": 1.393553223388306,
      "grad_norm": 0.11275880823677788,
      "learning_rate": 0.00012943780181974497,
      "loss": 0.0139,
      "step": 1859
    },
    {
      "epoch": 1.394302848575712,
      "grad_norm": 0.20070703000689094,
      "learning_rate": 0.0001293544141838378,
      "loss": 0.0212,
      "step": 1860
    },
    {
      "epoch": 1.3950524737631185,
      "grad_norm": 0.09003839234478617,
      "learning_rate": 0.00012927100420568132,
      "loss": 0.0151,
      "step": 1861
    },
    {
      "epoch": 1.3958020989505249,
      "grad_norm": 0.12825471757454604,
      "learning_rate": 0.00012918757194876066,
      "loss": 0.0154,
      "step": 1862
    },
    {
      "epoch": 1.396551724137931,
      "grad_norm": 0.08699576227833801,
      "learning_rate": 0.00012910411747657778,
      "loss": 0.0113,
      "step": 1863
    },
    {
      "epoch": 1.3973013493253372,
      "grad_norm": 0.21761113805906224,
      "learning_rate": 0.0001290206408526516,
      "loss": 0.0301,
      "step": 1864
    },
    {
      "epoch": 1.3980509745127436,
      "grad_norm": 0.1548273942024453,
      "learning_rate": 0.00012893714214051792,
      "loss": 0.014,
      "step": 1865
    },
    {
      "epoch": 1.39880059970015,
      "grad_norm": 0.09941037024754006,
      "learning_rate": 0.00012885362140372934,
      "loss": 0.0136,
      "step": 1866
    },
    {
      "epoch": 1.3995502248875562,
      "grad_norm": 0.18751823083887997,
      "learning_rate": 0.0001287700787058552,
      "loss": 0.0204,
      "step": 1867
    },
    {
      "epoch": 1.4002998500749624,
      "grad_norm": 0.15499241538245642,
      "learning_rate": 0.00012868651411048154,
      "loss": 0.0141,
      "step": 1868
    },
    {
      "epoch": 1.4010494752623688,
      "grad_norm": 0.12843498545392468,
      "learning_rate": 0.0001286029276812111,
      "loss": 0.0179,
      "step": 1869
    },
    {
      "epoch": 1.4017991004497752,
      "grad_norm": 0.4731933208232578,
      "learning_rate": 0.00012851931948166327,
      "loss": 0.0661,
      "step": 1870
    },
    {
      "epoch": 1.4025487256371814,
      "grad_norm": 0.14952964839476857,
      "learning_rate": 0.00012843568957547394,
      "loss": 0.0211,
      "step": 1871
    },
    {
      "epoch": 1.4032983508245878,
      "grad_norm": 0.17276714217465589,
      "learning_rate": 0.00012835203802629555,
      "loss": 0.0135,
      "step": 1872
    },
    {
      "epoch": 1.404047976011994,
      "grad_norm": 0.46872245156442344,
      "learning_rate": 0.00012826836489779708,
      "loss": 0.0497,
      "step": 1873
    },
    {
      "epoch": 1.4047976011994003,
      "grad_norm": 0.08363564829601658,
      "learning_rate": 0.00012818467025366376,
      "loss": 0.016,
      "step": 1874
    },
    {
      "epoch": 1.4055472263868065,
      "grad_norm": 0.10232915132284348,
      "learning_rate": 0.00012810095415759744,
      "loss": 0.0128,
      "step": 1875
    },
    {
      "epoch": 1.406296851574213,
      "grad_norm": 0.061288942568132866,
      "learning_rate": 0.00012801721667331605,
      "loss": 0.0089,
      "step": 1876
    },
    {
      "epoch": 1.4070464767616193,
      "grad_norm": 0.13128930961987118,
      "learning_rate": 0.000127933457864554,
      "loss": 0.0215,
      "step": 1877
    },
    {
      "epoch": 1.4077961019490255,
      "grad_norm": 0.11242897724452197,
      "learning_rate": 0.00012784967779506178,
      "loss": 0.0204,
      "step": 1878
    },
    {
      "epoch": 1.4085457271364317,
      "grad_norm": 0.10983058504415504,
      "learning_rate": 0.00012776587652860613,
      "loss": 0.0302,
      "step": 1879
    },
    {
      "epoch": 1.409295352323838,
      "grad_norm": 0.045545185026683276,
      "learning_rate": 0.00012768205412897,
      "loss": 0.0059,
      "step": 1880
    },
    {
      "epoch": 1.4100449775112445,
      "grad_norm": 0.15299570276557586,
      "learning_rate": 0.00012759821065995224,
      "loss": 0.0323,
      "step": 1881
    },
    {
      "epoch": 1.4107946026986506,
      "grad_norm": 0.08608560440679149,
      "learning_rate": 0.00012751434618536787,
      "loss": 0.0108,
      "step": 1882
    },
    {
      "epoch": 1.411544227886057,
      "grad_norm": 0.27803913808993475,
      "learning_rate": 0.00012743046076904793,
      "loss": 0.0201,
      "step": 1883
    },
    {
      "epoch": 1.4122938530734632,
      "grad_norm": 0.10612971680946036,
      "learning_rate": 0.0001273465544748393,
      "loss": 0.0099,
      "step": 1884
    },
    {
      "epoch": 1.4130434782608696,
      "grad_norm": 0.12436357529377723,
      "learning_rate": 0.00012726262736660474,
      "loss": 0.0284,
      "step": 1885
    },
    {
      "epoch": 1.4137931034482758,
      "grad_norm": 0.11607890022114893,
      "learning_rate": 0.00012717867950822297,
      "loss": 0.0101,
      "step": 1886
    },
    {
      "epoch": 1.4145427286356822,
      "grad_norm": 0.14770329705497223,
      "learning_rate": 0.00012709471096358842,
      "loss": 0.0315,
      "step": 1887
    },
    {
      "epoch": 1.4152923538230884,
      "grad_norm": 0.18604699056734458,
      "learning_rate": 0.00012701072179661127,
      "loss": 0.0293,
      "step": 1888
    },
    {
      "epoch": 1.4160419790104948,
      "grad_norm": 0.18470673903506774,
      "learning_rate": 0.0001269267120712174,
      "loss": 0.024,
      "step": 1889
    },
    {
      "epoch": 1.416791604197901,
      "grad_norm": 0.060328972475384775,
      "learning_rate": 0.00012684268185134834,
      "loss": 0.0059,
      "step": 1890
    },
    {
      "epoch": 1.4175412293853074,
      "grad_norm": 0.173380371633488,
      "learning_rate": 0.00012675863120096122,
      "loss": 0.0339,
      "step": 1891
    },
    {
      "epoch": 1.4182908545727138,
      "grad_norm": 0.11015965133252213,
      "learning_rate": 0.00012667456018402874,
      "loss": 0.0131,
      "step": 1892
    },
    {
      "epoch": 1.41904047976012,
      "grad_norm": 0.09097324818575385,
      "learning_rate": 0.00012659046886453904,
      "loss": 0.0116,
      "step": 1893
    },
    {
      "epoch": 1.419790104947526,
      "grad_norm": 0.1388229356745322,
      "learning_rate": 0.0001265063573064958,
      "loss": 0.013,
      "step": 1894
    },
    {
      "epoch": 1.4205397301349325,
      "grad_norm": 0.12052532519735364,
      "learning_rate": 0.00012642222557391808,
      "loss": 0.0134,
      "step": 1895
    },
    {
      "epoch": 1.421289355322339,
      "grad_norm": 0.15500408517715691,
      "learning_rate": 0.0001263380737308402,
      "loss": 0.0338,
      "step": 1896
    },
    {
      "epoch": 1.422038980509745,
      "grad_norm": 0.11227641668568701,
      "learning_rate": 0.0001262539018413119,
      "loss": 0.0159,
      "step": 1897
    },
    {
      "epoch": 1.4227886056971515,
      "grad_norm": 0.1628839753085024,
      "learning_rate": 0.00012616970996939814,
      "loss": 0.0173,
      "step": 1898
    },
    {
      "epoch": 1.4235382308845577,
      "grad_norm": 0.315608669835069,
      "learning_rate": 0.00012608549817917913,
      "loss": 0.0568,
      "step": 1899
    },
    {
      "epoch": 1.424287856071964,
      "grad_norm": 0.09860125372264825,
      "learning_rate": 0.00012600126653475012,
      "loss": 0.009,
      "step": 1900
    },
    {
      "epoch": 1.4250374812593702,
      "grad_norm": 0.13144368707140067,
      "learning_rate": 0.0001259170151002216,
      "loss": 0.0077,
      "step": 1901
    },
    {
      "epoch": 1.4257871064467766,
      "grad_norm": 0.0982665549847179,
      "learning_rate": 0.00012583274393971906,
      "loss": 0.0089,
      "step": 1902
    },
    {
      "epoch": 1.426536731634183,
      "grad_norm": 0.1386778146043919,
      "learning_rate": 0.000125748453117383,
      "loss": 0.011,
      "step": 1903
    },
    {
      "epoch": 1.4272863568215892,
      "grad_norm": 0.07388630267828614,
      "learning_rate": 0.0001256641426973689,
      "loss": 0.0117,
      "step": 1904
    },
    {
      "epoch": 1.4280359820089954,
      "grad_norm": 0.18284401715096338,
      "learning_rate": 0.0001255798127438472,
      "loss": 0.0084,
      "step": 1905
    },
    {
      "epoch": 1.4287856071964018,
      "grad_norm": 0.10679066875328382,
      "learning_rate": 0.0001254954633210031,
      "loss": 0.0132,
      "step": 1906
    },
    {
      "epoch": 1.4295352323838082,
      "grad_norm": 0.19195698414866189,
      "learning_rate": 0.00012541109449303676,
      "loss": 0.0216,
      "step": 1907
    },
    {
      "epoch": 1.4302848575712144,
      "grad_norm": 0.3666573212530311,
      "learning_rate": 0.00012532670632416293,
      "loss": 0.0326,
      "step": 1908
    },
    {
      "epoch": 1.4310344827586206,
      "grad_norm": 0.11101667561738876,
      "learning_rate": 0.00012524229887861133,
      "loss": 0.0171,
      "step": 1909
    },
    {
      "epoch": 1.431784107946027,
      "grad_norm": 0.08791025320937725,
      "learning_rate": 0.00012515787222062608,
      "loss": 0.0095,
      "step": 1910
    },
    {
      "epoch": 1.4325337331334334,
      "grad_norm": 0.1342867251099514,
      "learning_rate": 0.0001250734264144661,
      "loss": 0.0188,
      "step": 1911
    },
    {
      "epoch": 1.4332833583208395,
      "grad_norm": 0.09608017121352272,
      "learning_rate": 0.00012498896152440483,
      "loss": 0.0075,
      "step": 1912
    },
    {
      "epoch": 1.434032983508246,
      "grad_norm": 0.13020892593954997,
      "learning_rate": 0.00012490447761473022,
      "loss": 0.0083,
      "step": 1913
    },
    {
      "epoch": 1.434782608695652,
      "grad_norm": 0.09629403477170517,
      "learning_rate": 0.0001248199747497447,
      "loss": 0.0103,
      "step": 1914
    },
    {
      "epoch": 1.4355322338830585,
      "grad_norm": 0.13425262075259645,
      "learning_rate": 0.0001247354529937652,
      "loss": 0.0199,
      "step": 1915
    },
    {
      "epoch": 1.4362818590704647,
      "grad_norm": 0.08860400439856671,
      "learning_rate": 0.0001246509124111229,
      "loss": 0.0134,
      "step": 1916
    },
    {
      "epoch": 1.437031484257871,
      "grad_norm": 0.3389225893719704,
      "learning_rate": 0.00012456635306616344,
      "loss": 0.0534,
      "step": 1917
    },
    {
      "epoch": 1.4377811094452775,
      "grad_norm": 0.1685711891815974,
      "learning_rate": 0.00012448177502324663,
      "loss": 0.0216,
      "step": 1918
    },
    {
      "epoch": 1.4385307346326837,
      "grad_norm": 0.2572067238711098,
      "learning_rate": 0.0001243971783467466,
      "loss": 0.0159,
      "step": 1919
    },
    {
      "epoch": 1.4392803598200898,
      "grad_norm": 0.13640220342782158,
      "learning_rate": 0.00012431256310105158,
      "loss": 0.0183,
      "step": 1920
    },
    {
      "epoch": 1.4400299850074962,
      "grad_norm": 0.09959171259587224,
      "learning_rate": 0.000124227929350564,
      "loss": 0.0136,
      "step": 1921
    },
    {
      "epoch": 1.4407796101949026,
      "grad_norm": 0.22152896373552985,
      "learning_rate": 0.0001241432771597004,
      "loss": 0.0262,
      "step": 1922
    },
    {
      "epoch": 1.4415292353823088,
      "grad_norm": 0.15349376103343082,
      "learning_rate": 0.00012405860659289123,
      "loss": 0.0092,
      "step": 1923
    },
    {
      "epoch": 1.4422788605697152,
      "grad_norm": 0.14915283299222099,
      "learning_rate": 0.00012397391771458105,
      "loss": 0.0077,
      "step": 1924
    },
    {
      "epoch": 1.4430284857571214,
      "grad_norm": 0.12541301370117197,
      "learning_rate": 0.00012388921058922829,
      "loss": 0.0181,
      "step": 1925
    },
    {
      "epoch": 1.4437781109445278,
      "grad_norm": 0.1414621996215833,
      "learning_rate": 0.00012380448528130525,
      "loss": 0.0251,
      "step": 1926
    },
    {
      "epoch": 1.444527736131934,
      "grad_norm": 0.09985535961077165,
      "learning_rate": 0.00012371974185529817,
      "loss": 0.0087,
      "step": 1927
    },
    {
      "epoch": 1.4452773613193404,
      "grad_norm": 0.20252240412543931,
      "learning_rate": 0.000123634980375707,
      "loss": 0.0314,
      "step": 1928
    },
    {
      "epoch": 1.4460269865067465,
      "grad_norm": 0.07083834632533098,
      "learning_rate": 0.00012355020090704544,
      "loss": 0.0094,
      "step": 1929
    },
    {
      "epoch": 1.446776611694153,
      "grad_norm": 0.0988355632047036,
      "learning_rate": 0.0001234654035138409,
      "loss": 0.0153,
      "step": 1930
    },
    {
      "epoch": 1.4475262368815591,
      "grad_norm": 0.07249163434224347,
      "learning_rate": 0.00012338058826063438,
      "loss": 0.0104,
      "step": 1931
    },
    {
      "epoch": 1.4482758620689655,
      "grad_norm": 0.09806488951161967,
      "learning_rate": 0.00012329575521198054,
      "loss": 0.0138,
      "step": 1932
    },
    {
      "epoch": 1.449025487256372,
      "grad_norm": 0.13624110709087367,
      "learning_rate": 0.00012321090443244762,
      "loss": 0.0246,
      "step": 1933
    },
    {
      "epoch": 1.449775112443778,
      "grad_norm": 0.1179493925747529,
      "learning_rate": 0.0001231260359866172,
      "loss": 0.0054,
      "step": 1934
    },
    {
      "epoch": 1.4505247376311843,
      "grad_norm": 0.09174066897570911,
      "learning_rate": 0.00012304114993908448,
      "loss": 0.0189,
      "step": 1935
    },
    {
      "epoch": 1.4512743628185907,
      "grad_norm": 0.14148182215863198,
      "learning_rate": 0.00012295624635445796,
      "loss": 0.0266,
      "step": 1936
    },
    {
      "epoch": 1.452023988005997,
      "grad_norm": 0.2045256699403224,
      "learning_rate": 0.00012287132529735945,
      "loss": 0.0195,
      "step": 1937
    },
    {
      "epoch": 1.4527736131934033,
      "grad_norm": 0.25985288542875007,
      "learning_rate": 0.0001227863868324242,
      "loss": 0.0242,
      "step": 1938
    },
    {
      "epoch": 1.4535232383808097,
      "grad_norm": 0.21860590313539296,
      "learning_rate": 0.00012270143102430055,
      "loss": 0.0396,
      "step": 1939
    },
    {
      "epoch": 1.4542728635682158,
      "grad_norm": 0.07925976743388066,
      "learning_rate": 0.0001226164579376502,
      "loss": 0.0083,
      "step": 1940
    },
    {
      "epoch": 1.4550224887556222,
      "grad_norm": 0.24388208826017888,
      "learning_rate": 0.00012253146763714783,
      "loss": 0.0148,
      "step": 1941
    },
    {
      "epoch": 1.4557721139430284,
      "grad_norm": 0.12812885050817915,
      "learning_rate": 0.00012244646018748134,
      "loss": 0.016,
      "step": 1942
    },
    {
      "epoch": 1.4565217391304348,
      "grad_norm": 0.12975353474198595,
      "learning_rate": 0.00012236143565335164,
      "loss": 0.0158,
      "step": 1943
    },
    {
      "epoch": 1.4572713643178412,
      "grad_norm": 0.21293324073380665,
      "learning_rate": 0.00012227639409947272,
      "loss": 0.0349,
      "step": 1944
    },
    {
      "epoch": 1.4580209895052474,
      "grad_norm": 0.223606496350551,
      "learning_rate": 0.0001221913355905714,
      "loss": 0.021,
      "step": 1945
    },
    {
      "epoch": 1.4587706146926536,
      "grad_norm": 0.40452905855669036,
      "learning_rate": 0.00012210626019138748,
      "loss": 0.0524,
      "step": 1946
    },
    {
      "epoch": 1.45952023988006,
      "grad_norm": 0.20036469604349294,
      "learning_rate": 0.00012202116796667356,
      "loss": 0.0191,
      "step": 1947
    },
    {
      "epoch": 1.4602698650674664,
      "grad_norm": 0.20366839982219276,
      "learning_rate": 0.00012193605898119514,
      "loss": 0.0296,
      "step": 1948
    },
    {
      "epoch": 1.4610194902548725,
      "grad_norm": 0.18687874497395973,
      "learning_rate": 0.00012185093329973038,
      "loss": 0.0203,
      "step": 1949
    },
    {
      "epoch": 1.461769115442279,
      "grad_norm": 0.09124165683037777,
      "learning_rate": 0.00012176579098707023,
      "loss": 0.0118,
      "step": 1950
    },
    {
      "epoch": 1.4625187406296851,
      "grad_norm": 0.2532954888220123,
      "learning_rate": 0.0001216806321080182,
      "loss": 0.0285,
      "step": 1951
    },
    {
      "epoch": 1.4632683658170915,
      "grad_norm": 0.1361058594029782,
      "learning_rate": 0.00012159545672739054,
      "loss": 0.0146,
      "step": 1952
    },
    {
      "epoch": 1.4640179910044977,
      "grad_norm": 0.10342119836800451,
      "learning_rate": 0.00012151026491001591,
      "loss": 0.0073,
      "step": 1953
    },
    {
      "epoch": 1.464767616191904,
      "grad_norm": 0.16486238138863205,
      "learning_rate": 0.00012142505672073564,
      "loss": 0.0193,
      "step": 1954
    },
    {
      "epoch": 1.4655172413793103,
      "grad_norm": 0.13536447693810422,
      "learning_rate": 0.00012133983222440335,
      "loss": 0.0491,
      "step": 1955
    },
    {
      "epoch": 1.4662668665667167,
      "grad_norm": 0.23330516206788135,
      "learning_rate": 0.00012125459148588523,
      "loss": 0.0301,
      "step": 1956
    },
    {
      "epoch": 1.4670164917541229,
      "grad_norm": 0.15292897747369386,
      "learning_rate": 0.00012116933457005972,
      "loss": 0.0167,
      "step": 1957
    },
    {
      "epoch": 1.4677661169415293,
      "grad_norm": 0.1901558613459153,
      "learning_rate": 0.00012108406154181766,
      "loss": 0.0093,
      "step": 1958
    },
    {
      "epoch": 1.4685157421289357,
      "grad_norm": 0.2138954820034095,
      "learning_rate": 0.00012099877246606206,
      "loss": 0.0241,
      "step": 1959
    },
    {
      "epoch": 1.4692653673163418,
      "grad_norm": 0.07976981594385227,
      "learning_rate": 0.00012091346740770821,
      "loss": 0.0156,
      "step": 1960
    },
    {
      "epoch": 1.470014992503748,
      "grad_norm": 0.08969290312785852,
      "learning_rate": 0.00012082814643168357,
      "loss": 0.0141,
      "step": 1961
    },
    {
      "epoch": 1.4707646176911544,
      "grad_norm": 0.08385247469572182,
      "learning_rate": 0.00012074280960292763,
      "loss": 0.0082,
      "step": 1962
    },
    {
      "epoch": 1.4715142428785608,
      "grad_norm": 0.11224097369553376,
      "learning_rate": 0.00012065745698639208,
      "loss": 0.0077,
      "step": 1963
    },
    {
      "epoch": 1.472263868065967,
      "grad_norm": 0.6110527097464731,
      "learning_rate": 0.00012057208864704051,
      "loss": 0.0154,
      "step": 1964
    },
    {
      "epoch": 1.4730134932533734,
      "grad_norm": 0.20503801470472588,
      "learning_rate": 0.0001204867046498485,
      "loss": 0.0238,
      "step": 1965
    },
    {
      "epoch": 1.4737631184407796,
      "grad_norm": 0.058020772911770324,
      "learning_rate": 0.00012040130505980365,
      "loss": 0.0052,
      "step": 1966
    },
    {
      "epoch": 1.474512743628186,
      "grad_norm": 0.2052620451143929,
      "learning_rate": 0.00012031588994190526,
      "loss": 0.0168,
      "step": 1967
    },
    {
      "epoch": 1.4752623688155921,
      "grad_norm": 0.09603284968761991,
      "learning_rate": 0.00012023045936116459,
      "loss": 0.0098,
      "step": 1968
    },
    {
      "epoch": 1.4760119940029985,
      "grad_norm": 0.1805164431825691,
      "learning_rate": 0.00012014501338260462,
      "loss": 0.007,
      "step": 1969
    },
    {
      "epoch": 1.4767616191904047,
      "grad_norm": 0.18701411878470012,
      "learning_rate": 0.00012005955207126,
      "loss": 0.0204,
      "step": 1970
    },
    {
      "epoch": 1.4775112443778111,
      "grad_norm": 0.19871909726227946,
      "learning_rate": 0.00011997407549217717,
      "loss": 0.0299,
      "step": 1971
    },
    {
      "epoch": 1.4782608695652173,
      "grad_norm": 0.09369152034670397,
      "learning_rate": 0.000119888583710414,
      "loss": 0.0134,
      "step": 1972
    },
    {
      "epoch": 1.4790104947526237,
      "grad_norm": 0.2156334124260664,
      "learning_rate": 0.00011980307679104015,
      "loss": 0.0197,
      "step": 1973
    },
    {
      "epoch": 1.47976011994003,
      "grad_norm": 0.09856555218171896,
      "learning_rate": 0.00011971755479913665,
      "loss": 0.0096,
      "step": 1974
    },
    {
      "epoch": 1.4805097451274363,
      "grad_norm": 0.320903279814656,
      "learning_rate": 0.00011963201779979604,
      "loss": 0.02,
      "step": 1975
    },
    {
      "epoch": 1.4812593703148424,
      "grad_norm": 0.07879126478261443,
      "learning_rate": 0.00011954646585812233,
      "loss": 0.0081,
      "step": 1976
    },
    {
      "epoch": 1.4820089955022488,
      "grad_norm": 0.17444491435700865,
      "learning_rate": 0.00011946089903923081,
      "loss": 0.0274,
      "step": 1977
    },
    {
      "epoch": 1.4827586206896552,
      "grad_norm": 0.12417897909896483,
      "learning_rate": 0.00011937531740824819,
      "loss": 0.017,
      "step": 1978
    },
    {
      "epoch": 1.4835082458770614,
      "grad_norm": 0.30630006591297376,
      "learning_rate": 0.00011928972103031242,
      "loss": 0.0366,
      "step": 1979
    },
    {
      "epoch": 1.4842578710644678,
      "grad_norm": 0.2002393018537466,
      "learning_rate": 0.00011920410997057265,
      "loss": 0.0079,
      "step": 1980
    },
    {
      "epoch": 1.485007496251874,
      "grad_norm": 0.26995257634941094,
      "learning_rate": 0.00011911848429418923,
      "loss": 0.018,
      "step": 1981
    },
    {
      "epoch": 1.4857571214392804,
      "grad_norm": 0.23129376115177372,
      "learning_rate": 0.00011903284406633361,
      "loss": 0.0159,
      "step": 1982
    },
    {
      "epoch": 1.4865067466266866,
      "grad_norm": 0.21531353753419102,
      "learning_rate": 0.00011894718935218836,
      "loss": 0.021,
      "step": 1983
    },
    {
      "epoch": 1.487256371814093,
      "grad_norm": 0.1347198207422198,
      "learning_rate": 0.000118861520216947,
      "loss": 0.0263,
      "step": 1984
    },
    {
      "epoch": 1.4880059970014994,
      "grad_norm": 0.10419003906180573,
      "learning_rate": 0.00011877583672581411,
      "loss": 0.0132,
      "step": 1985
    },
    {
      "epoch": 1.4887556221889056,
      "grad_norm": 0.1425071654540163,
      "learning_rate": 0.00011869013894400516,
      "loss": 0.0105,
      "step": 1986
    },
    {
      "epoch": 1.4895052473763117,
      "grad_norm": 0.09482127929336295,
      "learning_rate": 0.00011860442693674647,
      "loss": 0.0074,
      "step": 1987
    },
    {
      "epoch": 1.4902548725637181,
      "grad_norm": 0.1545877408511147,
      "learning_rate": 0.0001185187007692752,
      "loss": 0.0198,
      "step": 1988
    },
    {
      "epoch": 1.4910044977511245,
      "grad_norm": 0.1406951822773715,
      "learning_rate": 0.00011843296050683937,
      "loss": 0.0091,
      "step": 1989
    },
    {
      "epoch": 1.4917541229385307,
      "grad_norm": 0.1092127925059156,
      "learning_rate": 0.00011834720621469762,
      "loss": 0.0187,
      "step": 1990
    },
    {
      "epoch": 1.4925037481259371,
      "grad_norm": 0.08263137479052818,
      "learning_rate": 0.00011826143795811929,
      "loss": 0.0128,
      "step": 1991
    },
    {
      "epoch": 1.4932533733133433,
      "grad_norm": 0.18058529469223594,
      "learning_rate": 0.00011817565580238442,
      "loss": 0.0174,
      "step": 1992
    },
    {
      "epoch": 1.4940029985007497,
      "grad_norm": 0.2926184202629552,
      "learning_rate": 0.00011808985981278353,
      "loss": 0.0417,
      "step": 1993
    },
    {
      "epoch": 1.4947526236881559,
      "grad_norm": 0.15063107867199743,
      "learning_rate": 0.00011800405005461772,
      "loss": 0.025,
      "step": 1994
    },
    {
      "epoch": 1.4955022488755623,
      "grad_norm": 0.09316966552874384,
      "learning_rate": 0.00011791822659319858,
      "loss": 0.0184,
      "step": 1995
    },
    {
      "epoch": 1.4962518740629684,
      "grad_norm": 0.3583120976917404,
      "learning_rate": 0.0001178323894938481,
      "loss": 0.0192,
      "step": 1996
    },
    {
      "epoch": 1.4970014992503748,
      "grad_norm": 0.06682455310812929,
      "learning_rate": 0.00011774653882189867,
      "loss": 0.0106,
      "step": 1997
    },
    {
      "epoch": 1.497751124437781,
      "grad_norm": 0.11924451335046508,
      "learning_rate": 0.00011766067464269301,
      "loss": 0.0184,
      "step": 1998
    },
    {
      "epoch": 1.4985007496251874,
      "grad_norm": 0.21903498673009728,
      "learning_rate": 0.00011757479702158412,
      "loss": 0.0362,
      "step": 1999
    },
    {
      "epoch": 1.4992503748125938,
      "grad_norm": 0.14552346714109463,
      "learning_rate": 0.00011748890602393521,
      "loss": 0.0131,
      "step": 2000
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.38878473530349916,
      "learning_rate": 0.00011740300171511969,
      "loss": 0.0343,
      "step": 2001
    },
    {
      "epoch": 1.5007496251874062,
      "grad_norm": 0.1819908689214082,
      "learning_rate": 0.0001173170841605211,
      "loss": 0.0098,
      "step": 2002
    },
    {
      "epoch": 1.5014992503748126,
      "grad_norm": 0.22899646815252692,
      "learning_rate": 0.00011723115342553306,
      "loss": 0.0215,
      "step": 2003
    },
    {
      "epoch": 1.502248875562219,
      "grad_norm": 0.11740484135767436,
      "learning_rate": 0.00011714520957555924,
      "loss": 0.0113,
      "step": 2004
    },
    {
      "epoch": 1.5029985007496252,
      "grad_norm": 0.3885868302596495,
      "learning_rate": 0.00011705925267601326,
      "loss": 0.043,
      "step": 2005
    },
    {
      "epoch": 1.5037481259370313,
      "grad_norm": 0.29570525506634887,
      "learning_rate": 0.00011697328279231868,
      "loss": 0.0335,
      "step": 2006
    },
    {
      "epoch": 1.5044977511244377,
      "grad_norm": 0.17289926376387163,
      "learning_rate": 0.00011688729998990897,
      "loss": 0.0305,
      "step": 2007
    },
    {
      "epoch": 1.5052473763118441,
      "grad_norm": 0.5384751241597092,
      "learning_rate": 0.0001168013043342274,
      "loss": 0.0403,
      "step": 2008
    },
    {
      "epoch": 1.5059970014992503,
      "grad_norm": 0.35965339130185314,
      "learning_rate": 0.00011671529589072702,
      "loss": 0.0202,
      "step": 2009
    },
    {
      "epoch": 1.5067466266866567,
      "grad_norm": 0.3715537347174127,
      "learning_rate": 0.00011662927472487064,
      "loss": 0.0589,
      "step": 2010
    },
    {
      "epoch": 1.507496251874063,
      "grad_norm": 0.1289465699723788,
      "learning_rate": 0.00011654324090213073,
      "loss": 0.0247,
      "step": 2011
    },
    {
      "epoch": 1.5082458770614693,
      "grad_norm": 0.2841186121236052,
      "learning_rate": 0.00011645719448798943,
      "loss": 0.0191,
      "step": 2012
    },
    {
      "epoch": 1.5089955022488755,
      "grad_norm": 0.08209758171226106,
      "learning_rate": 0.00011637113554793846,
      "loss": 0.0065,
      "step": 2013
    },
    {
      "epoch": 1.5097451274362819,
      "grad_norm": 0.13132648526713056,
      "learning_rate": 0.000116285064147479,
      "loss": 0.0125,
      "step": 2014
    },
    {
      "epoch": 1.5104947526236883,
      "grad_norm": 0.1583770195328219,
      "learning_rate": 0.00011619898035212174,
      "loss": 0.0265,
      "step": 2015
    },
    {
      "epoch": 1.5112443778110944,
      "grad_norm": 0.35884442000176264,
      "learning_rate": 0.00011611288422738691,
      "loss": 0.0298,
      "step": 2016
    },
    {
      "epoch": 1.5119940029985006,
      "grad_norm": 0.11422568057412678,
      "learning_rate": 0.00011602677583880401,
      "loss": 0.017,
      "step": 2017
    },
    {
      "epoch": 1.512743628185907,
      "grad_norm": 0.2073809987153342,
      "learning_rate": 0.00011594065525191188,
      "loss": 0.0237,
      "step": 2018
    },
    {
      "epoch": 1.5134932533733134,
      "grad_norm": 0.18272649079364311,
      "learning_rate": 0.00011585452253225872,
      "loss": 0.0252,
      "step": 2019
    },
    {
      "epoch": 1.5142428785607196,
      "grad_norm": 0.26491460784372994,
      "learning_rate": 0.0001157683777454019,
      "loss": 0.0433,
      "step": 2020
    },
    {
      "epoch": 1.5149925037481258,
      "grad_norm": 0.1556835860082349,
      "learning_rate": 0.00011568222095690797,
      "loss": 0.0176,
      "step": 2021
    },
    {
      "epoch": 1.5157421289355324,
      "grad_norm": 0.08848124098762954,
      "learning_rate": 0.00011559605223235268,
      "loss": 0.0124,
      "step": 2022
    },
    {
      "epoch": 1.5164917541229386,
      "grad_norm": 0.12709504996224588,
      "learning_rate": 0.00011550987163732075,
      "loss": 0.0195,
      "step": 2023
    },
    {
      "epoch": 1.5172413793103448,
      "grad_norm": 0.1158807293852502,
      "learning_rate": 0.00011542367923740606,
      "loss": 0.0206,
      "step": 2024
    },
    {
      "epoch": 1.5179910044977512,
      "grad_norm": 0.2924531296368726,
      "learning_rate": 0.00011533747509821145,
      "loss": 0.0287,
      "step": 2025
    },
    {
      "epoch": 1.5187406296851576,
      "grad_norm": 0.10536500142368622,
      "learning_rate": 0.00011525125928534859,
      "loss": 0.007,
      "step": 2026
    },
    {
      "epoch": 1.5194902548725637,
      "grad_norm": 0.12511524507356026,
      "learning_rate": 0.00011516503186443814,
      "loss": 0.0173,
      "step": 2027
    },
    {
      "epoch": 1.52023988005997,
      "grad_norm": 0.2148503368956103,
      "learning_rate": 0.0001150787929011096,
      "loss": 0.0236,
      "step": 2028
    },
    {
      "epoch": 1.5209895052473763,
      "grad_norm": 0.15375775401892824,
      "learning_rate": 0.00011499254246100118,
      "loss": 0.0152,
      "step": 2029
    },
    {
      "epoch": 1.5217391304347827,
      "grad_norm": 0.23824273067912238,
      "learning_rate": 0.0001149062806097599,
      "loss": 0.0492,
      "step": 2030
    },
    {
      "epoch": 1.5224887556221889,
      "grad_norm": 0.23191671523223392,
      "learning_rate": 0.0001148200074130414,
      "loss": 0.0339,
      "step": 2031
    },
    {
      "epoch": 1.523238380809595,
      "grad_norm": 0.13047509312978461,
      "learning_rate": 0.00011473372293650999,
      "loss": 0.0169,
      "step": 2032
    },
    {
      "epoch": 1.5239880059970015,
      "grad_norm": 0.14189321877652575,
      "learning_rate": 0.00011464742724583859,
      "loss": 0.012,
      "step": 2033
    },
    {
      "epoch": 1.5247376311844079,
      "grad_norm": 0.07837067512463858,
      "learning_rate": 0.00011456112040670859,
      "loss": 0.0104,
      "step": 2034
    },
    {
      "epoch": 1.525487256371814,
      "grad_norm": 0.10328103959029392,
      "learning_rate": 0.00011447480248480996,
      "loss": 0.007,
      "step": 2035
    },
    {
      "epoch": 1.5262368815592204,
      "grad_norm": 0.172214593174865,
      "learning_rate": 0.00011438847354584097,
      "loss": 0.0121,
      "step": 2036
    },
    {
      "epoch": 1.5269865067466268,
      "grad_norm": 0.08022427081753603,
      "learning_rate": 0.00011430213365550842,
      "loss": 0.0109,
      "step": 2037
    },
    {
      "epoch": 1.527736131934033,
      "grad_norm": 0.14878089897566166,
      "learning_rate": 0.00011421578287952735,
      "loss": 0.0059,
      "step": 2038
    },
    {
      "epoch": 1.5284857571214392,
      "grad_norm": 0.11671209020576799,
      "learning_rate": 0.0001141294212836211,
      "loss": 0.0148,
      "step": 2039
    },
    {
      "epoch": 1.5292353823088456,
      "grad_norm": 0.08369220048037276,
      "learning_rate": 0.00011404304893352132,
      "loss": 0.01,
      "step": 2040
    },
    {
      "epoch": 1.529985007496252,
      "grad_norm": 0.15577552287134389,
      "learning_rate": 0.0001139566658949677,
      "loss": 0.0144,
      "step": 2041
    },
    {
      "epoch": 1.5307346326836582,
      "grad_norm": 0.12315101821922537,
      "learning_rate": 0.00011387027223370821,
      "loss": 0.0072,
      "step": 2042
    },
    {
      "epoch": 1.5314842578710643,
      "grad_norm": 0.29388303557281226,
      "learning_rate": 0.00011378386801549883,
      "loss": 0.0131,
      "step": 2043
    },
    {
      "epoch": 1.5322338830584707,
      "grad_norm": 0.13980800324774958,
      "learning_rate": 0.0001136974533061036,
      "loss": 0.0174,
      "step": 2044
    },
    {
      "epoch": 1.5329835082458771,
      "grad_norm": 0.06204640754946424,
      "learning_rate": 0.00011361102817129451,
      "loss": 0.0076,
      "step": 2045
    },
    {
      "epoch": 1.5337331334332833,
      "grad_norm": 0.12680010891087348,
      "learning_rate": 0.00011352459267685152,
      "loss": 0.0158,
      "step": 2046
    },
    {
      "epoch": 1.5344827586206895,
      "grad_norm": 0.18932679871696698,
      "learning_rate": 0.00011343814688856246,
      "loss": 0.0263,
      "step": 2047
    },
    {
      "epoch": 1.535232383808096,
      "grad_norm": 0.11530877706564625,
      "learning_rate": 0.00011335169087222304,
      "loss": 0.0185,
      "step": 2048
    },
    {
      "epoch": 1.5359820089955023,
      "grad_norm": 0.2879999540228137,
      "learning_rate": 0.00011326522469363666,
      "loss": 0.0399,
      "step": 2049
    },
    {
      "epoch": 1.5367316341829085,
      "grad_norm": 0.10544309045823727,
      "learning_rate": 0.00011317874841861454,
      "loss": 0.009,
      "step": 2050
    },
    {
      "epoch": 1.5374812593703149,
      "grad_norm": 0.1720851235854106,
      "learning_rate": 0.00011309226211297552,
      "loss": 0.0136,
      "step": 2051
    },
    {
      "epoch": 1.5382308845577213,
      "grad_norm": 0.061021940481680546,
      "learning_rate": 0.00011300576584254617,
      "loss": 0.0091,
      "step": 2052
    },
    {
      "epoch": 1.5389805097451275,
      "grad_norm": 0.24495019966265658,
      "learning_rate": 0.00011291925967316055,
      "loss": 0.029,
      "step": 2053
    },
    {
      "epoch": 1.5397301349325336,
      "grad_norm": 0.19048454116953006,
      "learning_rate": 0.00011283274367066028,
      "loss": 0.0176,
      "step": 2054
    },
    {
      "epoch": 1.54047976011994,
      "grad_norm": 0.1966686629949861,
      "learning_rate": 0.00011274621790089445,
      "loss": 0.0203,
      "step": 2055
    },
    {
      "epoch": 1.5412293853073464,
      "grad_norm": 0.10997093120263526,
      "learning_rate": 0.00011265968242971967,
      "loss": 0.0097,
      "step": 2056
    },
    {
      "epoch": 1.5419790104947526,
      "grad_norm": 0.17625970195223611,
      "learning_rate": 0.00011257313732299984,
      "loss": 0.0122,
      "step": 2057
    },
    {
      "epoch": 1.5427286356821588,
      "grad_norm": 0.1733777878808708,
      "learning_rate": 0.00011248658264660617,
      "loss": 0.022,
      "step": 2058
    },
    {
      "epoch": 1.5434782608695652,
      "grad_norm": 0.12541753676579082,
      "learning_rate": 0.00011240001846641727,
      "loss": 0.016,
      "step": 2059
    },
    {
      "epoch": 1.5442278860569716,
      "grad_norm": 0.13712378929986008,
      "learning_rate": 0.00011231344484831894,
      "loss": 0.0157,
      "step": 2060
    },
    {
      "epoch": 1.5449775112443778,
      "grad_norm": 0.13682059079314143,
      "learning_rate": 0.00011222686185820409,
      "loss": 0.0215,
      "step": 2061
    },
    {
      "epoch": 1.545727136431784,
      "grad_norm": 0.13012752394248844,
      "learning_rate": 0.00011214026956197281,
      "loss": 0.0261,
      "step": 2062
    },
    {
      "epoch": 1.5464767616191906,
      "grad_norm": 0.11619923535577313,
      "learning_rate": 0.0001120536680255323,
      "loss": 0.0186,
      "step": 2063
    },
    {
      "epoch": 1.5472263868065967,
      "grad_norm": 0.10045403290997133,
      "learning_rate": 0.00011196705731479682,
      "loss": 0.0108,
      "step": 2064
    },
    {
      "epoch": 1.547976011994003,
      "grad_norm": 0.1341996957759823,
      "learning_rate": 0.00011188043749568751,
      "loss": 0.0125,
      "step": 2065
    },
    {
      "epoch": 1.5487256371814093,
      "grad_norm": 0.3351069282470784,
      "learning_rate": 0.00011179380863413249,
      "loss": 0.0262,
      "step": 2066
    },
    {
      "epoch": 1.5494752623688157,
      "grad_norm": 0.09092680298844832,
      "learning_rate": 0.00011170717079606683,
      "loss": 0.0151,
      "step": 2067
    },
    {
      "epoch": 1.550224887556222,
      "grad_norm": 0.07557937002901241,
      "learning_rate": 0.00011162052404743231,
      "loss": 0.0059,
      "step": 2068
    },
    {
      "epoch": 1.550974512743628,
      "grad_norm": 0.09023632236008271,
      "learning_rate": 0.00011153386845417757,
      "loss": 0.0105,
      "step": 2069
    },
    {
      "epoch": 1.5517241379310345,
      "grad_norm": 0.0890917938114594,
      "learning_rate": 0.00011144720408225799,
      "loss": 0.0124,
      "step": 2070
    },
    {
      "epoch": 1.5524737631184409,
      "grad_norm": 0.23400085026304457,
      "learning_rate": 0.00011136053099763561,
      "loss": 0.0213,
      "step": 2071
    },
    {
      "epoch": 1.553223388305847,
      "grad_norm": 0.21856781003492168,
      "learning_rate": 0.0001112738492662791,
      "loss": 0.0206,
      "step": 2072
    },
    {
      "epoch": 1.5539730134932532,
      "grad_norm": 0.08859642750860837,
      "learning_rate": 0.00011118715895416369,
      "loss": 0.0134,
      "step": 2073
    },
    {
      "epoch": 1.5547226386806596,
      "grad_norm": 0.22023807470902312,
      "learning_rate": 0.00011110046012727115,
      "loss": 0.0167,
      "step": 2074
    },
    {
      "epoch": 1.555472263868066,
      "grad_norm": 0.15376324747434672,
      "learning_rate": 0.00011101375285158977,
      "loss": 0.0185,
      "step": 2075
    },
    {
      "epoch": 1.5562218890554722,
      "grad_norm": 0.14143650908884217,
      "learning_rate": 0.00011092703719311423,
      "loss": 0.0247,
      "step": 2076
    },
    {
      "epoch": 1.5569715142428786,
      "grad_norm": 0.14443931346792963,
      "learning_rate": 0.00011084031321784565,
      "loss": 0.0078,
      "step": 2077
    },
    {
      "epoch": 1.557721139430285,
      "grad_norm": 0.21577033946330054,
      "learning_rate": 0.00011075358099179137,
      "loss": 0.0296,
      "step": 2078
    },
    {
      "epoch": 1.5584707646176912,
      "grad_norm": 0.14822292735668258,
      "learning_rate": 0.0001106668405809651,
      "loss": 0.0103,
      "step": 2079
    },
    {
      "epoch": 1.5592203898050974,
      "grad_norm": 0.280728591961263,
      "learning_rate": 0.00011058009205138675,
      "loss": 0.0325,
      "step": 2080
    },
    {
      "epoch": 1.5599700149925038,
      "grad_norm": 0.13964380316407649,
      "learning_rate": 0.00011049333546908238,
      "loss": 0.0168,
      "step": 2081
    },
    {
      "epoch": 1.5607196401799102,
      "grad_norm": 0.2665676112900603,
      "learning_rate": 0.00011040657090008429,
      "loss": 0.0342,
      "step": 2082
    },
    {
      "epoch": 1.5614692653673163,
      "grad_norm": 0.18879808411881546,
      "learning_rate": 0.00011031979841043074,
      "loss": 0.0191,
      "step": 2083
    },
    {
      "epoch": 1.5622188905547225,
      "grad_norm": 0.17893200215311023,
      "learning_rate": 0.00011023301806616604,
      "loss": 0.0593,
      "step": 2084
    },
    {
      "epoch": 1.562968515742129,
      "grad_norm": 0.10130024416265714,
      "learning_rate": 0.00011014622993334051,
      "loss": 0.0173,
      "step": 2085
    },
    {
      "epoch": 1.5637181409295353,
      "grad_norm": 0.11583261119153629,
      "learning_rate": 0.0001100594340780104,
      "loss": 0.0051,
      "step": 2086
    },
    {
      "epoch": 1.5644677661169415,
      "grad_norm": 0.20382548872544762,
      "learning_rate": 0.00010997263056623781,
      "loss": 0.0207,
      "step": 2087
    },
    {
      "epoch": 1.5652173913043477,
      "grad_norm": 0.26119103145261624,
      "learning_rate": 0.00010988581946409067,
      "loss": 0.0259,
      "step": 2088
    },
    {
      "epoch": 1.565967016491754,
      "grad_norm": 0.26387441495902386,
      "learning_rate": 0.00010979900083764274,
      "loss": 0.0295,
      "step": 2089
    },
    {
      "epoch": 1.5667166416791605,
      "grad_norm": 0.12167374287075801,
      "learning_rate": 0.00010971217475297341,
      "loss": 0.0135,
      "step": 2090
    },
    {
      "epoch": 1.5674662668665666,
      "grad_norm": 0.09404259679501854,
      "learning_rate": 0.00010962534127616784,
      "loss": 0.0143,
      "step": 2091
    },
    {
      "epoch": 1.568215892053973,
      "grad_norm": 0.11618029483945659,
      "learning_rate": 0.00010953850047331677,
      "loss": 0.0241,
      "step": 2092
    },
    {
      "epoch": 1.5689655172413794,
      "grad_norm": 0.12694899413054597,
      "learning_rate": 0.00010945165241051649,
      "loss": 0.0066,
      "step": 2093
    },
    {
      "epoch": 1.5697151424287856,
      "grad_norm": 0.13936338138363769,
      "learning_rate": 0.0001093647971538689,
      "loss": 0.0138,
      "step": 2094
    },
    {
      "epoch": 1.5704647676161918,
      "grad_norm": 0.1873410269184778,
      "learning_rate": 0.00010927793476948132,
      "loss": 0.0101,
      "step": 2095
    },
    {
      "epoch": 1.5712143928035982,
      "grad_norm": 0.3206581121869378,
      "learning_rate": 0.00010919106532346646,
      "loss": 0.0808,
      "step": 2096
    },
    {
      "epoch": 1.5719640179910046,
      "grad_norm": 0.10184226493788878,
      "learning_rate": 0.00010910418888194247,
      "loss": 0.0088,
      "step": 2097
    },
    {
      "epoch": 1.5727136431784108,
      "grad_norm": 0.144140262326729,
      "learning_rate": 0.0001090173055110328,
      "loss": 0.0183,
      "step": 2098
    },
    {
      "epoch": 1.573463268365817,
      "grad_norm": 0.16734098531032404,
      "learning_rate": 0.00010893041527686616,
      "loss": 0.0114,
      "step": 2099
    },
    {
      "epoch": 1.5742128935532234,
      "grad_norm": 0.20313748731878348,
      "learning_rate": 0.00010884351824557653,
      "loss": 0.0201,
      "step": 2100
    },
    {
      "epoch": 1.5749625187406298,
      "grad_norm": 0.18552738843936079,
      "learning_rate": 0.00010875661448330299,
      "loss": 0.0189,
      "step": 2101
    },
    {
      "epoch": 1.575712143928036,
      "grad_norm": 0.2181205184419006,
      "learning_rate": 0.00010866970405618979,
      "loss": 0.0146,
      "step": 2102
    },
    {
      "epoch": 1.5764617691154421,
      "grad_norm": 0.197999347547449,
      "learning_rate": 0.0001085827870303863,
      "loss": 0.0131,
      "step": 2103
    },
    {
      "epoch": 1.5772113943028487,
      "grad_norm": 0.18070888463567594,
      "learning_rate": 0.00010849586347204676,
      "loss": 0.0239,
      "step": 2104
    },
    {
      "epoch": 1.577961019490255,
      "grad_norm": 0.11994394448856549,
      "learning_rate": 0.00010840893344733056,
      "loss": 0.0149,
      "step": 2105
    },
    {
      "epoch": 1.578710644677661,
      "grad_norm": 0.22206888863262822,
      "learning_rate": 0.00010832199702240193,
      "loss": 0.0216,
      "step": 2106
    },
    {
      "epoch": 1.5794602698650675,
      "grad_norm": 0.09966607612949638,
      "learning_rate": 0.00010823505426342993,
      "loss": 0.0154,
      "step": 2107
    },
    {
      "epoch": 1.580209895052474,
      "grad_norm": 0.20815952559053078,
      "learning_rate": 0.00010814810523658851,
      "loss": 0.0334,
      "step": 2108
    },
    {
      "epoch": 1.58095952023988,
      "grad_norm": 0.13817033872523685,
      "learning_rate": 0.00010806115000805639,
      "loss": 0.0179,
      "step": 2109
    },
    {
      "epoch": 1.5817091454272862,
      "grad_norm": 0.20216668988737746,
      "learning_rate": 0.00010797418864401693,
      "loss": 0.0255,
      "step": 2110
    },
    {
      "epoch": 1.5824587706146926,
      "grad_norm": 0.11483519803723446,
      "learning_rate": 0.00010788722121065822,
      "loss": 0.0143,
      "step": 2111
    },
    {
      "epoch": 1.583208395802099,
      "grad_norm": 0.16669136560946193,
      "learning_rate": 0.00010780024777417304,
      "loss": 0.0224,
      "step": 2112
    },
    {
      "epoch": 1.5839580209895052,
      "grad_norm": 0.20514344820145913,
      "learning_rate": 0.00010771326840075857,
      "loss": 0.0237,
      "step": 2113
    },
    {
      "epoch": 1.5847076461769114,
      "grad_norm": 0.22128861461740099,
      "learning_rate": 0.00010762628315661667,
      "loss": 0.0208,
      "step": 2114
    },
    {
      "epoch": 1.5854572713643178,
      "grad_norm": 0.22105747964468944,
      "learning_rate": 0.00010753929210795358,
      "loss": 0.0126,
      "step": 2115
    },
    {
      "epoch": 1.5862068965517242,
      "grad_norm": 0.2316847935712145,
      "learning_rate": 0.00010745229532097995,
      "loss": 0.0133,
      "step": 2116
    },
    {
      "epoch": 1.5869565217391304,
      "grad_norm": 0.17387233501627947,
      "learning_rate": 0.00010736529286191086,
      "loss": 0.0149,
      "step": 2117
    },
    {
      "epoch": 1.5877061469265368,
      "grad_norm": 0.1357298657618581,
      "learning_rate": 0.00010727828479696569,
      "loss": 0.0167,
      "step": 2118
    },
    {
      "epoch": 1.5884557721139432,
      "grad_norm": 0.1105206812339558,
      "learning_rate": 0.00010719127119236803,
      "loss": 0.0127,
      "step": 2119
    },
    {
      "epoch": 1.5892053973013494,
      "grad_norm": 0.15783026288917643,
      "learning_rate": 0.00010710425211434574,
      "loss": 0.0259,
      "step": 2120
    },
    {
      "epoch": 1.5899550224887555,
      "grad_norm": 0.11771204673005133,
      "learning_rate": 0.00010701722762913082,
      "loss": 0.0112,
      "step": 2121
    },
    {
      "epoch": 1.590704647676162,
      "grad_norm": 0.32453207336541906,
      "learning_rate": 0.00010693019780295943,
      "loss": 0.0277,
      "step": 2122
    },
    {
      "epoch": 1.5914542728635683,
      "grad_norm": 0.1966072178788373,
      "learning_rate": 0.00010684316270207173,
      "loss": 0.037,
      "step": 2123
    },
    {
      "epoch": 1.5922038980509745,
      "grad_norm": 0.09304717367685852,
      "learning_rate": 0.00010675612239271195,
      "loss": 0.0118,
      "step": 2124
    },
    {
      "epoch": 1.5929535232383807,
      "grad_norm": 0.20236023984052592,
      "learning_rate": 0.00010666907694112826,
      "loss": 0.0219,
      "step": 2125
    },
    {
      "epoch": 1.593703148425787,
      "grad_norm": 0.11436132069365781,
      "learning_rate": 0.00010658202641357275,
      "loss": 0.0096,
      "step": 2126
    },
    {
      "epoch": 1.5944527736131935,
      "grad_norm": 0.08452129352834276,
      "learning_rate": 0.00010649497087630137,
      "loss": 0.0084,
      "step": 2127
    },
    {
      "epoch": 1.5952023988005997,
      "grad_norm": 0.22035349865028075,
      "learning_rate": 0.00010640791039557386,
      "loss": 0.048,
      "step": 2128
    },
    {
      "epoch": 1.5959520239880058,
      "grad_norm": 0.11498711500927268,
      "learning_rate": 0.00010632084503765376,
      "loss": 0.02,
      "step": 2129
    },
    {
      "epoch": 1.5967016491754122,
      "grad_norm": 0.0839174086862166,
      "learning_rate": 0.0001062337748688083,
      "loss": 0.0066,
      "step": 2130
    },
    {
      "epoch": 1.5974512743628186,
      "grad_norm": 0.12497149674304994,
      "learning_rate": 0.00010614669995530842,
      "loss": 0.0058,
      "step": 2131
    },
    {
      "epoch": 1.5982008995502248,
      "grad_norm": 0.10287884248033055,
      "learning_rate": 0.00010605962036342856,
      "loss": 0.0119,
      "step": 2132
    },
    {
      "epoch": 1.5989505247376312,
      "grad_norm": 0.1796680641405242,
      "learning_rate": 0.00010597253615944684,
      "loss": 0.0113,
      "step": 2133
    },
    {
      "epoch": 1.5997001499250376,
      "grad_norm": 0.19508523661443805,
      "learning_rate": 0.00010588544740964483,
      "loss": 0.0237,
      "step": 2134
    },
    {
      "epoch": 1.6004497751124438,
      "grad_norm": 0.19838721596349307,
      "learning_rate": 0.00010579835418030759,
      "loss": 0.0318,
      "step": 2135
    },
    {
      "epoch": 1.60119940029985,
      "grad_norm": 0.2255229237940063,
      "learning_rate": 0.00010571125653772351,
      "loss": 0.0278,
      "step": 2136
    },
    {
      "epoch": 1.6019490254872564,
      "grad_norm": 0.1940287143164911,
      "learning_rate": 0.00010562415454818446,
      "loss": 0.039,
      "step": 2137
    },
    {
      "epoch": 1.6026986506746628,
      "grad_norm": 0.19664254017654553,
      "learning_rate": 0.00010553704827798553,
      "loss": 0.0238,
      "step": 2138
    },
    {
      "epoch": 1.603448275862069,
      "grad_norm": 0.3490322734081589,
      "learning_rate": 0.00010544993779342511,
      "loss": 0.0129,
      "step": 2139
    },
    {
      "epoch": 1.6041979010494751,
      "grad_norm": 0.11696677496354138,
      "learning_rate": 0.00010536282316080479,
      "loss": 0.0138,
      "step": 2140
    },
    {
      "epoch": 1.6049475262368815,
      "grad_norm": 0.11630342379823932,
      "learning_rate": 0.00010527570444642929,
      "loss": 0.0073,
      "step": 2141
    },
    {
      "epoch": 1.605697151424288,
      "grad_norm": 0.16193962839259002,
      "learning_rate": 0.00010518858171660648,
      "loss": 0.0243,
      "step": 2142
    },
    {
      "epoch": 1.606446776611694,
      "grad_norm": 0.14777930178574433,
      "learning_rate": 0.00010510145503764726,
      "loss": 0.0151,
      "step": 2143
    },
    {
      "epoch": 1.6071964017991005,
      "grad_norm": 0.08578678039922254,
      "learning_rate": 0.00010501432447586552,
      "loss": 0.008,
      "step": 2144
    },
    {
      "epoch": 1.607946026986507,
      "grad_norm": 0.12769657648998697,
      "learning_rate": 0.00010492719009757815,
      "loss": 0.015,
      "step": 2145
    },
    {
      "epoch": 1.608695652173913,
      "grad_norm": 0.1485501355032563,
      "learning_rate": 0.00010484005196910489,
      "loss": 0.0248,
      "step": 2146
    },
    {
      "epoch": 1.6094452773613193,
      "grad_norm": 0.19054280205114993,
      "learning_rate": 0.00010475291015676838,
      "loss": 0.02,
      "step": 2147
    },
    {
      "epoch": 1.6101949025487257,
      "grad_norm": 0.22448232935165505,
      "learning_rate": 0.00010466576472689405,
      "loss": 0.0339,
      "step": 2148
    },
    {
      "epoch": 1.610944527736132,
      "grad_norm": 0.17100326139227798,
      "learning_rate": 0.0001045786157458101,
      "loss": 0.0211,
      "step": 2149
    },
    {
      "epoch": 1.6116941529235382,
      "grad_norm": 0.3515785672403247,
      "learning_rate": 0.00010449146327984737,
      "loss": 0.0297,
      "step": 2150
    },
    {
      "epoch": 1.6124437781109444,
      "grad_norm": 0.1375851914687004,
      "learning_rate": 0.00010440430739533945,
      "loss": 0.012,
      "step": 2151
    },
    {
      "epoch": 1.6131934032983508,
      "grad_norm": 0.17245457771112946,
      "learning_rate": 0.00010431714815862244,
      "loss": 0.0129,
      "step": 2152
    },
    {
      "epoch": 1.6139430284857572,
      "grad_norm": 0.1973816822681929,
      "learning_rate": 0.00010422998563603505,
      "loss": 0.0322,
      "step": 2153
    },
    {
      "epoch": 1.6146926536731634,
      "grad_norm": 0.06788410784408126,
      "learning_rate": 0.00010414281989391846,
      "loss": 0.0051,
      "step": 2154
    },
    {
      "epoch": 1.6154422788605696,
      "grad_norm": 0.09242706243543868,
      "learning_rate": 0.00010405565099861632,
      "loss": 0.0164,
      "step": 2155
    },
    {
      "epoch": 1.616191904047976,
      "grad_norm": 0.19222715958385686,
      "learning_rate": 0.00010396847901647468,
      "loss": 0.0152,
      "step": 2156
    },
    {
      "epoch": 1.6169415292353824,
      "grad_norm": 0.14608118270870263,
      "learning_rate": 0.00010388130401384191,
      "loss": 0.021,
      "step": 2157
    },
    {
      "epoch": 1.6176911544227885,
      "grad_norm": 0.10021220905764734,
      "learning_rate": 0.0001037941260570687,
      "loss": 0.0083,
      "step": 2158
    },
    {
      "epoch": 1.618440779610195,
      "grad_norm": 0.3437202444059941,
      "learning_rate": 0.00010370694521250801,
      "loss": 0.0263,
      "step": 2159
    },
    {
      "epoch": 1.6191904047976013,
      "grad_norm": 0.10922401081992254,
      "learning_rate": 0.00010361976154651493,
      "loss": 0.0135,
      "step": 2160
    },
    {
      "epoch": 1.6199400299850075,
      "grad_norm": 0.1275763182671347,
      "learning_rate": 0.00010353257512544683,
      "loss": 0.0192,
      "step": 2161
    },
    {
      "epoch": 1.6206896551724137,
      "grad_norm": 0.2856664323507451,
      "learning_rate": 0.00010344538601566301,
      "loss": 0.02,
      "step": 2162
    },
    {
      "epoch": 1.62143928035982,
      "grad_norm": 0.3372086035734109,
      "learning_rate": 0.0001033581942835249,
      "loss": 0.0556,
      "step": 2163
    },
    {
      "epoch": 1.6221889055472265,
      "grad_norm": 0.10753985912490954,
      "learning_rate": 0.000103270999995396,
      "loss": 0.0148,
      "step": 2164
    },
    {
      "epoch": 1.6229385307346327,
      "grad_norm": 0.05309486665403327,
      "learning_rate": 0.00010318380321764163,
      "loss": 0.0073,
      "step": 2165
    },
    {
      "epoch": 1.6236881559220389,
      "grad_norm": 0.17602802641704984,
      "learning_rate": 0.00010309660401662902,
      "loss": 0.0227,
      "step": 2166
    },
    {
      "epoch": 1.6244377811094453,
      "grad_norm": 0.1569913385165896,
      "learning_rate": 0.00010300940245872734,
      "loss": 0.0166,
      "step": 2167
    },
    {
      "epoch": 1.6251874062968517,
      "grad_norm": 0.12704934685718333,
      "learning_rate": 0.00010292219861030746,
      "loss": 0.0097,
      "step": 2168
    },
    {
      "epoch": 1.6259370314842578,
      "grad_norm": 0.09420389848758143,
      "learning_rate": 0.000102834992537742,
      "loss": 0.0085,
      "step": 2169
    },
    {
      "epoch": 1.626686656671664,
      "grad_norm": 0.14393386909154185,
      "learning_rate": 0.00010274778430740534,
      "loss": 0.0196,
      "step": 2170
    },
    {
      "epoch": 1.6274362818590704,
      "grad_norm": 0.09500661896692436,
      "learning_rate": 0.00010266057398567345,
      "loss": 0.0155,
      "step": 2171
    },
    {
      "epoch": 1.6281859070464768,
      "grad_norm": 0.08039664476375386,
      "learning_rate": 0.00010257336163892388,
      "loss": 0.0058,
      "step": 2172
    },
    {
      "epoch": 1.628935532233883,
      "grad_norm": 0.16527855875803454,
      "learning_rate": 0.00010248614733353576,
      "loss": 0.0109,
      "step": 2173
    },
    {
      "epoch": 1.6296851574212894,
      "grad_norm": 0.15900030028556839,
      "learning_rate": 0.0001023989311358897,
      "loss": 0.0184,
      "step": 2174
    },
    {
      "epoch": 1.6304347826086958,
      "grad_norm": 0.18865051802381824,
      "learning_rate": 0.00010231171311236776,
      "loss": 0.0162,
      "step": 2175
    },
    {
      "epoch": 1.631184407796102,
      "grad_norm": 0.22266468034711098,
      "learning_rate": 0.00010222449332935334,
      "loss": 0.0211,
      "step": 2176
    },
    {
      "epoch": 1.6319340329835081,
      "grad_norm": 0.12644029888344444,
      "learning_rate": 0.00010213727185323123,
      "loss": 0.0146,
      "step": 2177
    },
    {
      "epoch": 1.6326836581709145,
      "grad_norm": 0.08269922841842478,
      "learning_rate": 0.00010205004875038749,
      "loss": 0.0034,
      "step": 2178
    },
    {
      "epoch": 1.633433283358321,
      "grad_norm": 0.13467295910773985,
      "learning_rate": 0.00010196282408720943,
      "loss": 0.0164,
      "step": 2179
    },
    {
      "epoch": 1.6341829085457271,
      "grad_norm": 0.09751608794364057,
      "learning_rate": 0.00010187559793008553,
      "loss": 0.0089,
      "step": 2180
    },
    {
      "epoch": 1.6349325337331333,
      "grad_norm": 0.10433575276139269,
      "learning_rate": 0.0001017883703454054,
      "loss": 0.0065,
      "step": 2181
    },
    {
      "epoch": 1.6356821589205397,
      "grad_norm": 0.15904032420506176,
      "learning_rate": 0.00010170114139955975,
      "loss": 0.0158,
      "step": 2182
    },
    {
      "epoch": 1.636431784107946,
      "grad_norm": 0.3355373475253385,
      "learning_rate": 0.00010161391115894036,
      "loss": 0.0492,
      "step": 2183
    },
    {
      "epoch": 1.6371814092953523,
      "grad_norm": 0.07955988970300694,
      "learning_rate": 0.00010152667968993994,
      "loss": 0.0081,
      "step": 2184
    },
    {
      "epoch": 1.6379310344827587,
      "grad_norm": 0.10382669378097661,
      "learning_rate": 0.00010143944705895217,
      "loss": 0.0133,
      "step": 2185
    },
    {
      "epoch": 1.638680659670165,
      "grad_norm": 0.13676157868630912,
      "learning_rate": 0.00010135221333237154,
      "loss": 0.0119,
      "step": 2186
    },
    {
      "epoch": 1.6394302848575713,
      "grad_norm": 0.09265941037356405,
      "learning_rate": 0.00010126497857659356,
      "loss": 0.0107,
      "step": 2187
    },
    {
      "epoch": 1.6401799100449774,
      "grad_norm": 0.19844386592103025,
      "learning_rate": 0.00010117774285801432,
      "loss": 0.0189,
      "step": 2188
    },
    {
      "epoch": 1.6409295352323838,
      "grad_norm": 0.09822194199584616,
      "learning_rate": 0.00010109050624303071,
      "loss": 0.0153,
      "step": 2189
    },
    {
      "epoch": 1.6416791604197902,
      "grad_norm": 0.1049195490202721,
      "learning_rate": 0.00010100326879804034,
      "loss": 0.0057,
      "step": 2190
    },
    {
      "epoch": 1.6424287856071964,
      "grad_norm": 0.10430346317698513,
      "learning_rate": 0.00010091603058944143,
      "loss": 0.0116,
      "step": 2191
    },
    {
      "epoch": 1.6431784107946026,
      "grad_norm": 0.23977722911803956,
      "learning_rate": 0.00010082879168363277,
      "loss": 0.0253,
      "step": 2192
    },
    {
      "epoch": 1.643928035982009,
      "grad_norm": 0.19249755711105562,
      "learning_rate": 0.00010074155214701369,
      "loss": 0.0241,
      "step": 2193
    },
    {
      "epoch": 1.6446776611694154,
      "grad_norm": 0.13018140823584648,
      "learning_rate": 0.00010065431204598398,
      "loss": 0.0181,
      "step": 2194
    },
    {
      "epoch": 1.6454272863568216,
      "grad_norm": 0.1914619455012303,
      "learning_rate": 0.0001005670714469439,
      "loss": 0.0165,
      "step": 2195
    },
    {
      "epoch": 1.6461769115442277,
      "grad_norm": 0.1382501986835231,
      "learning_rate": 0.00010047983041629408,
      "loss": 0.0127,
      "step": 2196
    },
    {
      "epoch": 1.6469265367316341,
      "grad_norm": 0.19493616103935438,
      "learning_rate": 0.00010039258902043542,
      "loss": 0.0189,
      "step": 2197
    },
    {
      "epoch": 1.6476761619190405,
      "grad_norm": 0.21639137368174913,
      "learning_rate": 0.00010030534732576919,
      "loss": 0.0326,
      "step": 2198
    },
    {
      "epoch": 1.6484257871064467,
      "grad_norm": 0.08163997103958853,
      "learning_rate": 0.0001002181053986968,
      "loss": 0.0091,
      "step": 2199
    },
    {
      "epoch": 1.6491754122938531,
      "grad_norm": 0.10984390492478067,
      "learning_rate": 0.00010013086330561994,
      "loss": 0.0173,
      "step": 2200
    },
    {
      "epoch": 1.6499250374812595,
      "grad_norm": 0.11543924109680062,
      "learning_rate": 0.0001000436211129403,
      "loss": 0.0199,
      "step": 2201
    },
    {
      "epoch": 1.6506746626686657,
      "grad_norm": 0.12631791559353478,
      "learning_rate": 9.995637888705971e-05,
      "loss": 0.0069,
      "step": 2202
    },
    {
      "epoch": 1.6514242878560719,
      "grad_norm": 0.1760866474337531,
      "learning_rate": 9.986913669438006e-05,
      "loss": 0.0124,
      "step": 2203
    },
    {
      "epoch": 1.6521739130434783,
      "grad_norm": 0.1275355517822058,
      "learning_rate": 9.978189460130318e-05,
      "loss": 0.0287,
      "step": 2204
    },
    {
      "epoch": 1.6529235382308847,
      "grad_norm": 0.07944793817543527,
      "learning_rate": 9.969465267423082e-05,
      "loss": 0.01,
      "step": 2205
    },
    {
      "epoch": 1.6536731634182908,
      "grad_norm": 0.11128776003101262,
      "learning_rate": 9.960741097956459e-05,
      "loss": 0.0089,
      "step": 2206
    },
    {
      "epoch": 1.654422788605697,
      "grad_norm": 0.13148517577463179,
      "learning_rate": 9.952016958370595e-05,
      "loss": 0.0151,
      "step": 2207
    },
    {
      "epoch": 1.6551724137931034,
      "grad_norm": 0.19695789121035406,
      "learning_rate": 9.943292855305612e-05,
      "loss": 0.022,
      "step": 2208
    },
    {
      "epoch": 1.6559220389805098,
      "grad_norm": 0.08508472187420428,
      "learning_rate": 9.934568795401603e-05,
      "loss": 0.0105,
      "step": 2209
    },
    {
      "epoch": 1.656671664167916,
      "grad_norm": 0.15133628187895673,
      "learning_rate": 9.925844785298632e-05,
      "loss": 0.0341,
      "step": 2210
    },
    {
      "epoch": 1.6574212893553222,
      "grad_norm": 0.1922398060665263,
      "learning_rate": 9.917120831636725e-05,
      "loss": 0.0102,
      "step": 2211
    },
    {
      "epoch": 1.6581709145427288,
      "grad_norm": 0.0938351190848418,
      "learning_rate": 9.908396941055858e-05,
      "loss": 0.0126,
      "step": 2212
    },
    {
      "epoch": 1.658920539730135,
      "grad_norm": 0.15477205535520777,
      "learning_rate": 9.899673120195968e-05,
      "loss": 0.0166,
      "step": 2213
    },
    {
      "epoch": 1.6596701649175412,
      "grad_norm": 0.09216439861864109,
      "learning_rate": 9.890949375696932e-05,
      "loss": 0.0075,
      "step": 2214
    },
    {
      "epoch": 1.6604197901049476,
      "grad_norm": 0.16799827639624468,
      "learning_rate": 9.882225714198572e-05,
      "loss": 0.0104,
      "step": 2215
    },
    {
      "epoch": 1.661169415292354,
      "grad_norm": 0.08444261604782366,
      "learning_rate": 9.873502142340647e-05,
      "loss": 0.0054,
      "step": 2216
    },
    {
      "epoch": 1.6619190404797601,
      "grad_norm": 0.0954091400877419,
      "learning_rate": 9.864778666762848e-05,
      "loss": 0.0131,
      "step": 2217
    },
    {
      "epoch": 1.6626686656671663,
      "grad_norm": 0.14137959599399608,
      "learning_rate": 9.856055294104787e-05,
      "loss": 0.0136,
      "step": 2218
    },
    {
      "epoch": 1.6634182908545727,
      "grad_norm": 0.19674432327170557,
      "learning_rate": 9.84733203100601e-05,
      "loss": 0.022,
      "step": 2219
    },
    {
      "epoch": 1.6641679160419791,
      "grad_norm": 0.19241725608754048,
      "learning_rate": 9.838608884105967e-05,
      "loss": 0.0196,
      "step": 2220
    },
    {
      "epoch": 1.6649175412293853,
      "grad_norm": 0.0741158166637822,
      "learning_rate": 9.829885860044028e-05,
      "loss": 0.0099,
      "step": 2221
    },
    {
      "epoch": 1.6656671664167915,
      "grad_norm": 0.08107170814118782,
      "learning_rate": 9.821162965459464e-05,
      "loss": 0.0082,
      "step": 2222
    },
    {
      "epoch": 1.6664167916041979,
      "grad_norm": 0.45306028725278275,
      "learning_rate": 9.812440206991452e-05,
      "loss": 0.0748,
      "step": 2223
    },
    {
      "epoch": 1.6671664167916043,
      "grad_norm": 0.2149625837226016,
      "learning_rate": 9.803717591279061e-05,
      "loss": 0.0167,
      "step": 2224
    },
    {
      "epoch": 1.6679160419790104,
      "grad_norm": 0.17153411211542954,
      "learning_rate": 9.794995124961255e-05,
      "loss": 0.0256,
      "step": 2225
    },
    {
      "epoch": 1.6686656671664168,
      "grad_norm": 0.0827609854771755,
      "learning_rate": 9.786272814676877e-05,
      "loss": 0.0116,
      "step": 2226
    },
    {
      "epoch": 1.6694152923538232,
      "grad_norm": 0.1885871305873978,
      "learning_rate": 9.777550667064666e-05,
      "loss": 0.0264,
      "step": 2227
    },
    {
      "epoch": 1.6701649175412294,
      "grad_norm": 0.17937268516669583,
      "learning_rate": 9.768828688763225e-05,
      "loss": 0.0283,
      "step": 2228
    },
    {
      "epoch": 1.6709145427286356,
      "grad_norm": 0.060729515568319854,
      "learning_rate": 9.76010688641103e-05,
      "loss": 0.0041,
      "step": 2229
    },
    {
      "epoch": 1.671664167916042,
      "grad_norm": 0.13836789615313236,
      "learning_rate": 9.751385266646425e-05,
      "loss": 0.0191,
      "step": 2230
    },
    {
      "epoch": 1.6724137931034484,
      "grad_norm": 0.13593051507269474,
      "learning_rate": 9.742663836107616e-05,
      "loss": 0.015,
      "step": 2231
    },
    {
      "epoch": 1.6731634182908546,
      "grad_norm": 0.1999216010464126,
      "learning_rate": 9.733942601432657e-05,
      "loss": 0.0174,
      "step": 2232
    },
    {
      "epoch": 1.6739130434782608,
      "grad_norm": 0.06516165629180859,
      "learning_rate": 9.725221569259468e-05,
      "loss": 0.0115,
      "step": 2233
    },
    {
      "epoch": 1.6746626686656672,
      "grad_norm": 0.11151068671306183,
      "learning_rate": 9.716500746225802e-05,
      "loss": 0.0112,
      "step": 2234
    },
    {
      "epoch": 1.6754122938530736,
      "grad_norm": 0.09629089815406607,
      "learning_rate": 9.707780138969258e-05,
      "loss": 0.0093,
      "step": 2235
    },
    {
      "epoch": 1.6761619190404797,
      "grad_norm": 0.14162644878631606,
      "learning_rate": 9.699059754127269e-05,
      "loss": 0.0073,
      "step": 2236
    },
    {
      "epoch": 1.676911544227886,
      "grad_norm": 0.11520040224766206,
      "learning_rate": 9.690339598337099e-05,
      "loss": 0.0139,
      "step": 2237
    },
    {
      "epoch": 1.6776611694152923,
      "grad_norm": 0.1335076882114969,
      "learning_rate": 9.681619678235841e-05,
      "loss": 0.0179,
      "step": 2238
    },
    {
      "epoch": 1.6784107946026987,
      "grad_norm": 0.07575372094517098,
      "learning_rate": 9.672900000460403e-05,
      "loss": 0.0062,
      "step": 2239
    },
    {
      "epoch": 1.6791604197901049,
      "grad_norm": 0.13863362510064225,
      "learning_rate": 9.66418057164751e-05,
      "loss": 0.0101,
      "step": 2240
    },
    {
      "epoch": 1.6799100449775113,
      "grad_norm": 0.17980979209392906,
      "learning_rate": 9.655461398433703e-05,
      "loss": 0.0194,
      "step": 2241
    },
    {
      "epoch": 1.6806596701649177,
      "grad_norm": 0.0735726093429126,
      "learning_rate": 9.646742487455319e-05,
      "loss": 0.006,
      "step": 2242
    },
    {
      "epoch": 1.6814092953523239,
      "grad_norm": 0.10716385387797484,
      "learning_rate": 9.638023845348508e-05,
      "loss": 0.007,
      "step": 2243
    },
    {
      "epoch": 1.68215892053973,
      "grad_norm": 0.106194164771032,
      "learning_rate": 9.629305478749203e-05,
      "loss": 0.0093,
      "step": 2244
    },
    {
      "epoch": 1.6829085457271364,
      "grad_norm": 0.07602179907137255,
      "learning_rate": 9.620587394293133e-05,
      "loss": 0.0167,
      "step": 2245
    },
    {
      "epoch": 1.6836581709145428,
      "grad_norm": 0.10184174507913761,
      "learning_rate": 9.611869598615814e-05,
      "loss": 0.0259,
      "step": 2246
    },
    {
      "epoch": 1.684407796101949,
      "grad_norm": 0.10190431935445256,
      "learning_rate": 9.603152098352537e-05,
      "loss": 0.0122,
      "step": 2247
    },
    {
      "epoch": 1.6851574212893552,
      "grad_norm": 0.14810433391255484,
      "learning_rate": 9.594434900138372e-05,
      "loss": 0.0131,
      "step": 2248
    },
    {
      "epoch": 1.6859070464767616,
      "grad_norm": 0.1372986326124049,
      "learning_rate": 9.585718010608158e-05,
      "loss": 0.0242,
      "step": 2249
    },
    {
      "epoch": 1.686656671664168,
      "grad_norm": 0.22024790492225385,
      "learning_rate": 9.577001436396496e-05,
      "loss": 0.02,
      "step": 2250
    },
    {
      "epoch": 1.6874062968515742,
      "grad_norm": 0.23818210366138526,
      "learning_rate": 9.568285184137756e-05,
      "loss": 0.0339,
      "step": 2251
    },
    {
      "epoch": 1.6881559220389803,
      "grad_norm": 0.24880745595783435,
      "learning_rate": 9.559569260466056e-05,
      "loss": 0.0203,
      "step": 2252
    },
    {
      "epoch": 1.688905547226387,
      "grad_norm": 0.08934150225370609,
      "learning_rate": 9.550853672015264e-05,
      "loss": 0.0057,
      "step": 2253
    },
    {
      "epoch": 1.6896551724137931,
      "grad_norm": 0.14124142644205057,
      "learning_rate": 9.542138425418994e-05,
      "loss": 0.0243,
      "step": 2254
    },
    {
      "epoch": 1.6904047976011993,
      "grad_norm": 0.128834478994259,
      "learning_rate": 9.533423527310595e-05,
      "loss": 0.0231,
      "step": 2255
    },
    {
      "epoch": 1.6911544227886057,
      "grad_norm": 0.10252762235808675,
      "learning_rate": 9.524708984323164e-05,
      "loss": 0.0094,
      "step": 2256
    },
    {
      "epoch": 1.6919040479760121,
      "grad_norm": 0.11919608393359798,
      "learning_rate": 9.515994803089514e-05,
      "loss": 0.011,
      "step": 2257
    },
    {
      "epoch": 1.6926536731634183,
      "grad_norm": 0.10028302535446303,
      "learning_rate": 9.507280990242189e-05,
      "loss": 0.0065,
      "step": 2258
    },
    {
      "epoch": 1.6934032983508245,
      "grad_norm": 0.15029212251390084,
      "learning_rate": 9.498567552413451e-05,
      "loss": 0.0202,
      "step": 2259
    },
    {
      "epoch": 1.6941529235382309,
      "grad_norm": 0.14739855354036463,
      "learning_rate": 9.489854496235278e-05,
      "loss": 0.0137,
      "step": 2260
    },
    {
      "epoch": 1.6949025487256373,
      "grad_norm": 0.13250478800352208,
      "learning_rate": 9.481141828339354e-05,
      "loss": 0.0166,
      "step": 2261
    },
    {
      "epoch": 1.6956521739130435,
      "grad_norm": 0.19895184445701702,
      "learning_rate": 9.472429555357074e-05,
      "loss": 0.0253,
      "step": 2262
    },
    {
      "epoch": 1.6964017991004496,
      "grad_norm": 0.20379249682360126,
      "learning_rate": 9.463717683919525e-05,
      "loss": 0.0142,
      "step": 2263
    },
    {
      "epoch": 1.697151424287856,
      "grad_norm": 0.22588524145582545,
      "learning_rate": 9.455006220657491e-05,
      "loss": 0.0142,
      "step": 2264
    },
    {
      "epoch": 1.6979010494752624,
      "grad_norm": 0.2183364170199202,
      "learning_rate": 9.446295172201447e-05,
      "loss": 0.0294,
      "step": 2265
    },
    {
      "epoch": 1.6986506746626686,
      "grad_norm": 0.10529773015330877,
      "learning_rate": 9.437584545181556e-05,
      "loss": 0.0117,
      "step": 2266
    },
    {
      "epoch": 1.699400299850075,
      "grad_norm": 0.13228476642079293,
      "learning_rate": 9.428874346227651e-05,
      "loss": 0.0105,
      "step": 2267
    },
    {
      "epoch": 1.7001499250374814,
      "grad_norm": 0.11116500642428082,
      "learning_rate": 9.420164581969246e-05,
      "loss": 0.0192,
      "step": 2268
    },
    {
      "epoch": 1.7008995502248876,
      "grad_norm": 0.20191971260760297,
      "learning_rate": 9.41145525903552e-05,
      "loss": 0.0306,
      "step": 2269
    },
    {
      "epoch": 1.7016491754122938,
      "grad_norm": 0.1733740741546028,
      "learning_rate": 9.402746384055319e-05,
      "loss": 0.0148,
      "step": 2270
    },
    {
      "epoch": 1.7023988005997002,
      "grad_norm": 0.1985233647138693,
      "learning_rate": 9.394037963657147e-05,
      "loss": 0.027,
      "step": 2271
    },
    {
      "epoch": 1.7031484257871066,
      "grad_norm": 0.15488629938664647,
      "learning_rate": 9.385330004469165e-05,
      "loss": 0.0137,
      "step": 2272
    },
    {
      "epoch": 1.7038980509745127,
      "grad_norm": 0.12550295707673448,
      "learning_rate": 9.376622513119173e-05,
      "loss": 0.0213,
      "step": 2273
    },
    {
      "epoch": 1.704647676161919,
      "grad_norm": 0.0970176046056629,
      "learning_rate": 9.36791549623463e-05,
      "loss": 0.0112,
      "step": 2274
    },
    {
      "epoch": 1.7053973013493253,
      "grad_norm": 0.09282452066698203,
      "learning_rate": 9.359208960442614e-05,
      "loss": 0.0136,
      "step": 2275
    },
    {
      "epoch": 1.7061469265367317,
      "grad_norm": 0.16370503644010123,
      "learning_rate": 9.350502912369864e-05,
      "loss": 0.0135,
      "step": 2276
    },
    {
      "epoch": 1.706896551724138,
      "grad_norm": 0.3022753723913286,
      "learning_rate": 9.341797358642726e-05,
      "loss": 0.039,
      "step": 2277
    },
    {
      "epoch": 1.707646176911544,
      "grad_norm": 0.07035860921333878,
      "learning_rate": 9.333092305887175e-05,
      "loss": 0.0058,
      "step": 2278
    },
    {
      "epoch": 1.7083958020989505,
      "grad_norm": 0.1348057402631347,
      "learning_rate": 9.324387760728804e-05,
      "loss": 0.0137,
      "step": 2279
    },
    {
      "epoch": 1.7091454272863569,
      "grad_norm": 0.20890376229362348,
      "learning_rate": 9.315683729792827e-05,
      "loss": 0.0211,
      "step": 2280
    },
    {
      "epoch": 1.709895052473763,
      "grad_norm": 0.11310850996722045,
      "learning_rate": 9.306980219704059e-05,
      "loss": 0.0199,
      "step": 2281
    },
    {
      "epoch": 1.7106446776611695,
      "grad_norm": 0.24161359492379683,
      "learning_rate": 9.298277237086919e-05,
      "loss": 0.0112,
      "step": 2282
    },
    {
      "epoch": 1.7113943028485759,
      "grad_norm": 0.07882883973262984,
      "learning_rate": 9.28957478856543e-05,
      "loss": 0.0077,
      "step": 2283
    },
    {
      "epoch": 1.712143928035982,
      "grad_norm": 0.08496704552979792,
      "learning_rate": 9.280872880763201e-05,
      "loss": 0.0061,
      "step": 2284
    },
    {
      "epoch": 1.7128935532233882,
      "grad_norm": 0.14163766893656093,
      "learning_rate": 9.272171520303434e-05,
      "loss": 0.0125,
      "step": 2285
    },
    {
      "epoch": 1.7136431784107946,
      "grad_norm": 0.32724263022641503,
      "learning_rate": 9.263470713808915e-05,
      "loss": 0.0188,
      "step": 2286
    },
    {
      "epoch": 1.714392803598201,
      "grad_norm": 0.10439175468241432,
      "learning_rate": 9.254770467902007e-05,
      "loss": 0.0078,
      "step": 2287
    },
    {
      "epoch": 1.7151424287856072,
      "grad_norm": 0.14755758211112238,
      "learning_rate": 9.246070789204643e-05,
      "loss": 0.0357,
      "step": 2288
    },
    {
      "epoch": 1.7158920539730134,
      "grad_norm": 0.16791677288964893,
      "learning_rate": 9.237371684338334e-05,
      "loss": 0.0216,
      "step": 2289
    },
    {
      "epoch": 1.7166416791604198,
      "grad_norm": 0.1295912107491882,
      "learning_rate": 9.228673159924144e-05,
      "loss": 0.0174,
      "step": 2290
    },
    {
      "epoch": 1.7173913043478262,
      "grad_norm": 0.2166942636210779,
      "learning_rate": 9.2199752225827e-05,
      "loss": 0.0219,
      "step": 2291
    },
    {
      "epoch": 1.7181409295352323,
      "grad_norm": 0.19356004105232413,
      "learning_rate": 9.211277878934179e-05,
      "loss": 0.0161,
      "step": 2292
    },
    {
      "epoch": 1.7188905547226385,
      "grad_norm": 0.1116573751433848,
      "learning_rate": 9.202581135598312e-05,
      "loss": 0.011,
      "step": 2293
    },
    {
      "epoch": 1.7196401799100451,
      "grad_norm": 0.09744720227055026,
      "learning_rate": 9.193884999194366e-05,
      "loss": 0.0151,
      "step": 2294
    },
    {
      "epoch": 1.7203898050974513,
      "grad_norm": 0.10492655393579968,
      "learning_rate": 9.185189476341151e-05,
      "loss": 0.0149,
      "step": 2295
    },
    {
      "epoch": 1.7211394302848575,
      "grad_norm": 0.1727812476879939,
      "learning_rate": 9.176494573657011e-05,
      "loss": 0.017,
      "step": 2296
    },
    {
      "epoch": 1.721889055472264,
      "grad_norm": 0.17214475886272773,
      "learning_rate": 9.167800297759812e-05,
      "loss": 0.014,
      "step": 2297
    },
    {
      "epoch": 1.7226386806596703,
      "grad_norm": 0.10015290651528999,
      "learning_rate": 9.159106655266946e-05,
      "loss": 0.0094,
      "step": 2298
    },
    {
      "epoch": 1.7233883058470765,
      "grad_norm": 0.12022968701022108,
      "learning_rate": 9.150413652795325e-05,
      "loss": 0.0107,
      "step": 2299
    },
    {
      "epoch": 1.7241379310344827,
      "grad_norm": 0.20433992444486745,
      "learning_rate": 9.141721296961373e-05,
      "loss": 0.0282,
      "step": 2300
    },
    {
      "epoch": 1.724887556221889,
      "grad_norm": 0.1251335600249641,
      "learning_rate": 9.133029594381022e-05,
      "loss": 0.0055,
      "step": 2301
    },
    {
      "epoch": 1.7256371814092955,
      "grad_norm": 0.12307454879878728,
      "learning_rate": 9.124338551669702e-05,
      "loss": 0.0072,
      "step": 2302
    },
    {
      "epoch": 1.7263868065967016,
      "grad_norm": 0.10974319222209682,
      "learning_rate": 9.115648175442348e-05,
      "loss": 0.0069,
      "step": 2303
    },
    {
      "epoch": 1.7271364317841078,
      "grad_norm": 0.06147589682688045,
      "learning_rate": 9.106958472313385e-05,
      "loss": 0.0074,
      "step": 2304
    },
    {
      "epoch": 1.7278860569715142,
      "grad_norm": 0.1692104250013026,
      "learning_rate": 9.098269448896722e-05,
      "loss": 0.0212,
      "step": 2305
    },
    {
      "epoch": 1.7286356821589206,
      "grad_norm": 0.08275924919315929,
      "learning_rate": 9.089581111805756e-05,
      "loss": 0.0091,
      "step": 2306
    },
    {
      "epoch": 1.7293853073463268,
      "grad_norm": 0.3002478715846399,
      "learning_rate": 9.080893467653357e-05,
      "loss": 0.0253,
      "step": 2307
    },
    {
      "epoch": 1.7301349325337332,
      "grad_norm": 0.04807159341900259,
      "learning_rate": 9.072206523051872e-05,
      "loss": 0.0015,
      "step": 2308
    },
    {
      "epoch": 1.7308845577211396,
      "grad_norm": 0.1257871319990705,
      "learning_rate": 9.063520284613112e-05,
      "loss": 0.0092,
      "step": 2309
    },
    {
      "epoch": 1.7316341829085458,
      "grad_norm": 0.24012971821535473,
      "learning_rate": 9.054834758948354e-05,
      "loss": 0.0121,
      "step": 2310
    },
    {
      "epoch": 1.732383808095952,
      "grad_norm": 0.06868813800563243,
      "learning_rate": 9.046149952668326e-05,
      "loss": 0.006,
      "step": 2311
    },
    {
      "epoch": 1.7331334332833583,
      "grad_norm": 0.2466917223603374,
      "learning_rate": 9.037465872383218e-05,
      "loss": 0.0254,
      "step": 2312
    },
    {
      "epoch": 1.7338830584707647,
      "grad_norm": 0.09160274133796535,
      "learning_rate": 9.028782524702661e-05,
      "loss": 0.0119,
      "step": 2313
    },
    {
      "epoch": 1.734632683658171,
      "grad_norm": 0.1231184165811145,
      "learning_rate": 9.020099916235729e-05,
      "loss": 0.016,
      "step": 2314
    },
    {
      "epoch": 1.735382308845577,
      "grad_norm": 0.16531511126818504,
      "learning_rate": 9.011418053590934e-05,
      "loss": 0.007,
      "step": 2315
    },
    {
      "epoch": 1.7361319340329835,
      "grad_norm": 0.06487807881747268,
      "learning_rate": 9.002736943376222e-05,
      "loss": 0.0111,
      "step": 2316
    },
    {
      "epoch": 1.73688155922039,
      "grad_norm": 0.1349339609143589,
      "learning_rate": 8.994056592198963e-05,
      "loss": 0.0111,
      "step": 2317
    },
    {
      "epoch": 1.737631184407796,
      "grad_norm": 0.052501414741449384,
      "learning_rate": 8.985377006665952e-05,
      "loss": 0.0048,
      "step": 2318
    },
    {
      "epoch": 1.7383808095952022,
      "grad_norm": 0.1993518878703383,
      "learning_rate": 8.9766981933834e-05,
      "loss": 0.0224,
      "step": 2319
    },
    {
      "epoch": 1.7391304347826086,
      "grad_norm": 0.1527207076809983,
      "learning_rate": 8.968020158956932e-05,
      "loss": 0.009,
      "step": 2320
    },
    {
      "epoch": 1.739880059970015,
      "grad_norm": 0.31428872722225654,
      "learning_rate": 8.959342909991572e-05,
      "loss": 0.0907,
      "step": 2321
    },
    {
      "epoch": 1.7406296851574212,
      "grad_norm": 0.042976893923091876,
      "learning_rate": 8.950666453091763e-05,
      "loss": 0.0038,
      "step": 2322
    },
    {
      "epoch": 1.7413793103448276,
      "grad_norm": 0.2108541705440652,
      "learning_rate": 8.941990794861329e-05,
      "loss": 0.0221,
      "step": 2323
    },
    {
      "epoch": 1.742128935532234,
      "grad_norm": 0.09913233384673842,
      "learning_rate": 8.933315941903493e-05,
      "loss": 0.0177,
      "step": 2324
    },
    {
      "epoch": 1.7428785607196402,
      "grad_norm": 0.1689140975435586,
      "learning_rate": 8.924641900820864e-05,
      "loss": 0.0121,
      "step": 2325
    },
    {
      "epoch": 1.7436281859070464,
      "grad_norm": 0.11147509449054176,
      "learning_rate": 8.915968678215436e-05,
      "loss": 0.015,
      "step": 2326
    },
    {
      "epoch": 1.7443778110944528,
      "grad_norm": 0.13181839721952637,
      "learning_rate": 8.907296280688575e-05,
      "loss": 0.0137,
      "step": 2327
    },
    {
      "epoch": 1.7451274362818592,
      "grad_norm": 0.08804738058468466,
      "learning_rate": 8.898624714841024e-05,
      "loss": 0.005,
      "step": 2328
    },
    {
      "epoch": 1.7458770614692654,
      "grad_norm": 0.07071846345027209,
      "learning_rate": 8.889953987272887e-05,
      "loss": 0.0043,
      "step": 2329
    },
    {
      "epoch": 1.7466266866566715,
      "grad_norm": 0.07836457908704735,
      "learning_rate": 8.881284104583633e-05,
      "loss": 0.0064,
      "step": 2330
    },
    {
      "epoch": 1.747376311844078,
      "grad_norm": 0.24495018065737204,
      "learning_rate": 8.872615073372092e-05,
      "loss": 0.0145,
      "step": 2331
    },
    {
      "epoch": 1.7481259370314843,
      "grad_norm": 0.13956385729137918,
      "learning_rate": 8.86394690023644e-05,
      "loss": 0.01,
      "step": 2332
    },
    {
      "epoch": 1.7488755622188905,
      "grad_norm": 0.24149279564904225,
      "learning_rate": 8.855279591774202e-05,
      "loss": 0.0265,
      "step": 2333
    },
    {
      "epoch": 1.7496251874062967,
      "grad_norm": 0.0974355064338702,
      "learning_rate": 8.846613154582246e-05,
      "loss": 0.0106,
      "step": 2334
    },
    {
      "epoch": 1.7503748125937033,
      "grad_norm": 0.19091651414304264,
      "learning_rate": 8.837947595256771e-05,
      "loss": 0.0162,
      "step": 2335
    },
    {
      "epoch": 1.7511244377811095,
      "grad_norm": 0.13646927135675968,
      "learning_rate": 8.82928292039332e-05,
      "loss": 0.0133,
      "step": 2336
    },
    {
      "epoch": 1.7518740629685157,
      "grad_norm": 0.10580595673508876,
      "learning_rate": 8.820619136586752e-05,
      "loss": 0.015,
      "step": 2337
    },
    {
      "epoch": 1.752623688155922,
      "grad_norm": 0.16799968327604245,
      "learning_rate": 8.811956250431253e-05,
      "loss": 0.0113,
      "step": 2338
    },
    {
      "epoch": 1.7533733133433285,
      "grad_norm": 0.10343603191332432,
      "learning_rate": 8.803294268520321e-05,
      "loss": 0.0035,
      "step": 2339
    },
    {
      "epoch": 1.7541229385307346,
      "grad_norm": 0.12820836792714202,
      "learning_rate": 8.79463319744677e-05,
      "loss": 0.0199,
      "step": 2340
    },
    {
      "epoch": 1.7548725637181408,
      "grad_norm": 0.12156490486890247,
      "learning_rate": 8.785973043802723e-05,
      "loss": 0.0122,
      "step": 2341
    },
    {
      "epoch": 1.7556221889055472,
      "grad_norm": 0.08880129633365962,
      "learning_rate": 8.777313814179598e-05,
      "loss": 0.0103,
      "step": 2342
    },
    {
      "epoch": 1.7563718140929536,
      "grad_norm": 0.1361385886586821,
      "learning_rate": 8.768655515168111e-05,
      "loss": 0.0073,
      "step": 2343
    },
    {
      "epoch": 1.7571214392803598,
      "grad_norm": 0.08717390078439331,
      "learning_rate": 8.759998153358274e-05,
      "loss": 0.011,
      "step": 2344
    },
    {
      "epoch": 1.757871064467766,
      "grad_norm": 0.13739874298614746,
      "learning_rate": 8.751341735339386e-05,
      "loss": 0.026,
      "step": 2345
    },
    {
      "epoch": 1.7586206896551724,
      "grad_norm": 0.1362043887687259,
      "learning_rate": 8.742686267700021e-05,
      "loss": 0.0146,
      "step": 2346
    },
    {
      "epoch": 1.7593703148425788,
      "grad_norm": 0.16907979495658385,
      "learning_rate": 8.734031757028034e-05,
      "loss": 0.0168,
      "step": 2347
    },
    {
      "epoch": 1.760119940029985,
      "grad_norm": 0.12248994769638612,
      "learning_rate": 8.725378209910553e-05,
      "loss": 0.0182,
      "step": 2348
    },
    {
      "epoch": 1.7608695652173914,
      "grad_norm": 0.1228893815046215,
      "learning_rate": 8.716725632933973e-05,
      "loss": 0.0143,
      "step": 2349
    },
    {
      "epoch": 1.7616191904047978,
      "grad_norm": 0.19568888006797103,
      "learning_rate": 8.708074032683946e-05,
      "loss": 0.0199,
      "step": 2350
    },
    {
      "epoch": 1.762368815592204,
      "grad_norm": 0.40680375508994865,
      "learning_rate": 8.699423415745383e-05,
      "loss": 0.0362,
      "step": 2351
    },
    {
      "epoch": 1.76311844077961,
      "grad_norm": 0.10472777978935661,
      "learning_rate": 8.690773788702447e-05,
      "loss": 0.0107,
      "step": 2352
    },
    {
      "epoch": 1.7638680659670165,
      "grad_norm": 0.13838529500657282,
      "learning_rate": 8.682125158138547e-05,
      "loss": 0.0143,
      "step": 2353
    },
    {
      "epoch": 1.764617691154423,
      "grad_norm": 0.05052992508658696,
      "learning_rate": 8.673477530636335e-05,
      "loss": 0.0025,
      "step": 2354
    },
    {
      "epoch": 1.765367316341829,
      "grad_norm": 0.21295167297071438,
      "learning_rate": 8.6648309127777e-05,
      "loss": 0.0132,
      "step": 2355
    },
    {
      "epoch": 1.7661169415292353,
      "grad_norm": 0.08171822983671556,
      "learning_rate": 8.656185311143756e-05,
      "loss": 0.0088,
      "step": 2356
    },
    {
      "epoch": 1.7668665667166417,
      "grad_norm": 0.10411541551669401,
      "learning_rate": 8.647540732314852e-05,
      "loss": 0.0168,
      "step": 2357
    },
    {
      "epoch": 1.767616191904048,
      "grad_norm": 0.08343662762648035,
      "learning_rate": 8.638897182870551e-05,
      "loss": 0.0095,
      "step": 2358
    },
    {
      "epoch": 1.7683658170914542,
      "grad_norm": 0.4010883192599926,
      "learning_rate": 8.630254669389643e-05,
      "loss": 0.0507,
      "step": 2359
    },
    {
      "epoch": 1.7691154422788604,
      "grad_norm": 0.144194524741157,
      "learning_rate": 8.621613198450118e-05,
      "loss": 0.0248,
      "step": 2360
    },
    {
      "epoch": 1.7698650674662668,
      "grad_norm": 0.12433647392529536,
      "learning_rate": 8.61297277662918e-05,
      "loss": 0.0278,
      "step": 2361
    },
    {
      "epoch": 1.7706146926536732,
      "grad_norm": 0.1929365072777267,
      "learning_rate": 8.604333410503231e-05,
      "loss": 0.0393,
      "step": 2362
    },
    {
      "epoch": 1.7713643178410794,
      "grad_norm": 0.1684109117453673,
      "learning_rate": 8.595695106647872e-05,
      "loss": 0.0185,
      "step": 2363
    },
    {
      "epoch": 1.7721139430284858,
      "grad_norm": 0.07411413587451561,
      "learning_rate": 8.587057871637891e-05,
      "loss": 0.0089,
      "step": 2364
    },
    {
      "epoch": 1.7728635682158922,
      "grad_norm": 0.10625695114970526,
      "learning_rate": 8.578421712047269e-05,
      "loss": 0.0092,
      "step": 2365
    },
    {
      "epoch": 1.7736131934032984,
      "grad_norm": 0.12012490456270324,
      "learning_rate": 8.569786634449162e-05,
      "loss": 0.0156,
      "step": 2366
    },
    {
      "epoch": 1.7743628185907045,
      "grad_norm": 0.12171823516493953,
      "learning_rate": 8.561152645415907e-05,
      "loss": 0.0109,
      "step": 2367
    },
    {
      "epoch": 1.775112443778111,
      "grad_norm": 0.1400595657322927,
      "learning_rate": 8.552519751519008e-05,
      "loss": 0.0229,
      "step": 2368
    },
    {
      "epoch": 1.7758620689655173,
      "grad_norm": 0.16177276746735106,
      "learning_rate": 8.543887959329144e-05,
      "loss": 0.0219,
      "step": 2369
    },
    {
      "epoch": 1.7766116941529235,
      "grad_norm": 0.15222405621160293,
      "learning_rate": 8.535257275416145e-05,
      "loss": 0.0158,
      "step": 2370
    },
    {
      "epoch": 1.7773613193403297,
      "grad_norm": 0.05362026788424701,
      "learning_rate": 8.526627706349002e-05,
      "loss": 0.0056,
      "step": 2371
    },
    {
      "epoch": 1.778110944527736,
      "grad_norm": 0.17846732790097433,
      "learning_rate": 8.517999258695861e-05,
      "loss": 0.0109,
      "step": 2372
    },
    {
      "epoch": 1.7788605697151425,
      "grad_norm": 0.09478484403879284,
      "learning_rate": 8.509371939024011e-05,
      "loss": 0.0171,
      "step": 2373
    },
    {
      "epoch": 1.7796101949025487,
      "grad_norm": 0.09896839439022605,
      "learning_rate": 8.500745753899883e-05,
      "loss": 0.0093,
      "step": 2374
    },
    {
      "epoch": 1.7803598200899549,
      "grad_norm": 0.14864187174565247,
      "learning_rate": 8.492120709889041e-05,
      "loss": 0.0159,
      "step": 2375
    },
    {
      "epoch": 1.7811094452773615,
      "grad_norm": 0.20883578340056738,
      "learning_rate": 8.483496813556187e-05,
      "loss": 0.02,
      "step": 2376
    },
    {
      "epoch": 1.7818590704647677,
      "grad_norm": 0.06932679928893044,
      "learning_rate": 8.474874071465144e-05,
      "loss": 0.008,
      "step": 2377
    },
    {
      "epoch": 1.7826086956521738,
      "grad_norm": 0.113622304438288,
      "learning_rate": 8.466252490178859e-05,
      "loss": 0.0097,
      "step": 2378
    },
    {
      "epoch": 1.7833583208395802,
      "grad_norm": 0.10037254661684161,
      "learning_rate": 8.457632076259395e-05,
      "loss": 0.0092,
      "step": 2379
    },
    {
      "epoch": 1.7841079460269866,
      "grad_norm": 0.0823820071462789,
      "learning_rate": 8.449012836267928e-05,
      "loss": 0.0131,
      "step": 2380
    },
    {
      "epoch": 1.7848575712143928,
      "grad_norm": 0.13449999499783072,
      "learning_rate": 8.440394776764735e-05,
      "loss": 0.0114,
      "step": 2381
    },
    {
      "epoch": 1.785607196401799,
      "grad_norm": 0.2303829733454292,
      "learning_rate": 8.431777904309204e-05,
      "loss": 0.0263,
      "step": 2382
    },
    {
      "epoch": 1.7863568215892054,
      "grad_norm": 0.10160743650504148,
      "learning_rate": 8.423162225459812e-05,
      "loss": 0.0088,
      "step": 2383
    },
    {
      "epoch": 1.7871064467766118,
      "grad_norm": 0.09810569209049344,
      "learning_rate": 8.414547746774129e-05,
      "loss": 0.0076,
      "step": 2384
    },
    {
      "epoch": 1.787856071964018,
      "grad_norm": 0.10763901521364005,
      "learning_rate": 8.405934474808813e-05,
      "loss": 0.0048,
      "step": 2385
    },
    {
      "epoch": 1.7886056971514241,
      "grad_norm": 0.08881457612672823,
      "learning_rate": 8.397322416119602e-05,
      "loss": 0.009,
      "step": 2386
    },
    {
      "epoch": 1.7893553223388305,
      "grad_norm": 0.1489624361101976,
      "learning_rate": 8.388711577261311e-05,
      "loss": 0.0202,
      "step": 2387
    },
    {
      "epoch": 1.790104947526237,
      "grad_norm": 0.10090392870455397,
      "learning_rate": 8.380101964787827e-05,
      "loss": 0.0112,
      "step": 2388
    },
    {
      "epoch": 1.7908545727136431,
      "grad_norm": 0.19129451114334656,
      "learning_rate": 8.371493585252105e-05,
      "loss": 0.0196,
      "step": 2389
    },
    {
      "epoch": 1.7916041979010495,
      "grad_norm": 0.19691572762188736,
      "learning_rate": 8.362886445206159e-05,
      "loss": 0.0199,
      "step": 2390
    },
    {
      "epoch": 1.792353823088456,
      "grad_norm": 0.15452745623652986,
      "learning_rate": 8.354280551201058e-05,
      "loss": 0.0157,
      "step": 2391
    },
    {
      "epoch": 1.793103448275862,
      "grad_norm": 0.27046459835526604,
      "learning_rate": 8.345675909786927e-05,
      "loss": 0.0201,
      "step": 2392
    },
    {
      "epoch": 1.7938530734632683,
      "grad_norm": 0.13222942919639982,
      "learning_rate": 8.337072527512939e-05,
      "loss": 0.031,
      "step": 2393
    },
    {
      "epoch": 1.7946026986506747,
      "grad_norm": 0.1387837427042574,
      "learning_rate": 8.328470410927303e-05,
      "loss": 0.0271,
      "step": 2394
    },
    {
      "epoch": 1.795352323838081,
      "grad_norm": 0.09030559122487354,
      "learning_rate": 8.319869566577261e-05,
      "loss": 0.0118,
      "step": 2395
    },
    {
      "epoch": 1.7961019490254873,
      "grad_norm": 0.11471102933534097,
      "learning_rate": 8.311270001009103e-05,
      "loss": 0.0082,
      "step": 2396
    },
    {
      "epoch": 1.7968515742128934,
      "grad_norm": 0.24614017032481977,
      "learning_rate": 8.302671720768132e-05,
      "loss": 0.022,
      "step": 2397
    },
    {
      "epoch": 1.7976011994002998,
      "grad_norm": 0.07106249804454486,
      "learning_rate": 8.294074732398674e-05,
      "loss": 0.0084,
      "step": 2398
    },
    {
      "epoch": 1.7983508245877062,
      "grad_norm": 0.27858692463728674,
      "learning_rate": 8.285479042444076e-05,
      "loss": 0.028,
      "step": 2399
    },
    {
      "epoch": 1.7991004497751124,
      "grad_norm": 0.08144062567952201,
      "learning_rate": 8.276884657446695e-05,
      "loss": 0.0123,
      "step": 2400
    },
    {
      "epoch": 1.7998500749625186,
      "grad_norm": 0.1507172031836174,
      "learning_rate": 8.268291583947891e-05,
      "loss": 0.0153,
      "step": 2401
    },
    {
      "epoch": 1.800599700149925,
      "grad_norm": 0.10025116654911155,
      "learning_rate": 8.259699828488033e-05,
      "loss": 0.0087,
      "step": 2402
    },
    {
      "epoch": 1.8013493253373314,
      "grad_norm": 0.2954824278349512,
      "learning_rate": 8.251109397606483e-05,
      "loss": 0.0285,
      "step": 2403
    },
    {
      "epoch": 1.8020989505247376,
      "grad_norm": 0.11698150619357166,
      "learning_rate": 8.242520297841592e-05,
      "loss": 0.0131,
      "step": 2404
    },
    {
      "epoch": 1.802848575712144,
      "grad_norm": 0.16459252010828768,
      "learning_rate": 8.2339325357307e-05,
      "loss": 0.0197,
      "step": 2405
    },
    {
      "epoch": 1.8035982008995504,
      "grad_norm": 0.15037083494560705,
      "learning_rate": 8.225346117810134e-05,
      "loss": 0.017,
      "step": 2406
    },
    {
      "epoch": 1.8043478260869565,
      "grad_norm": 0.1285506317397058,
      "learning_rate": 8.216761050615191e-05,
      "loss": 0.0063,
      "step": 2407
    },
    {
      "epoch": 1.8050974512743627,
      "grad_norm": 0.11691442119619251,
      "learning_rate": 8.208177340680144e-05,
      "loss": 0.0162,
      "step": 2408
    },
    {
      "epoch": 1.8058470764617691,
      "grad_norm": 0.28019089269121017,
      "learning_rate": 8.19959499453823e-05,
      "loss": 0.0437,
      "step": 2409
    },
    {
      "epoch": 1.8065967016491755,
      "grad_norm": 0.156066382708508,
      "learning_rate": 8.19101401872165e-05,
      "loss": 0.0116,
      "step": 2410
    },
    {
      "epoch": 1.8073463268365817,
      "grad_norm": 0.07752708295458294,
      "learning_rate": 8.182434419761561e-05,
      "loss": 0.0068,
      "step": 2411
    },
    {
      "epoch": 1.8080959520239879,
      "grad_norm": 0.11088280544682265,
      "learning_rate": 8.173856204188072e-05,
      "loss": 0.0112,
      "step": 2412
    },
    {
      "epoch": 1.8088455772113943,
      "grad_norm": 0.06605750093101989,
      "learning_rate": 8.165279378530242e-05,
      "loss": 0.0068,
      "step": 2413
    },
    {
      "epoch": 1.8095952023988007,
      "grad_norm": 0.22384537369273771,
      "learning_rate": 8.156703949316064e-05,
      "loss": 0.0183,
      "step": 2414
    },
    {
      "epoch": 1.8103448275862069,
      "grad_norm": 0.17784538402035052,
      "learning_rate": 8.148129923072482e-05,
      "loss": 0.0116,
      "step": 2415
    },
    {
      "epoch": 1.811094452773613,
      "grad_norm": 0.24431438066392663,
      "learning_rate": 8.139557306325358e-05,
      "loss": 0.0218,
      "step": 2416
    },
    {
      "epoch": 1.8118440779610197,
      "grad_norm": 0.11453677059618239,
      "learning_rate": 8.13098610559949e-05,
      "loss": 0.0238,
      "step": 2417
    },
    {
      "epoch": 1.8125937031484258,
      "grad_norm": 0.1020464200148881,
      "learning_rate": 8.122416327418594e-05,
      "loss": 0.0086,
      "step": 2418
    },
    {
      "epoch": 1.813343328335832,
      "grad_norm": 0.2358627914517762,
      "learning_rate": 8.113847978305304e-05,
      "loss": 0.0299,
      "step": 2419
    },
    {
      "epoch": 1.8140929535232384,
      "grad_norm": 0.14323494877777182,
      "learning_rate": 8.105281064781165e-05,
      "loss": 0.0148,
      "step": 2420
    },
    {
      "epoch": 1.8148425787106448,
      "grad_norm": 0.18538828017344644,
      "learning_rate": 8.096715593366637e-05,
      "loss": 0.0243,
      "step": 2421
    },
    {
      "epoch": 1.815592203898051,
      "grad_norm": 0.26309294444457143,
      "learning_rate": 8.088151570581076e-05,
      "loss": 0.0344,
      "step": 2422
    },
    {
      "epoch": 1.8163418290854572,
      "grad_norm": 0.14817054423968734,
      "learning_rate": 8.079589002942733e-05,
      "loss": 0.0146,
      "step": 2423
    },
    {
      "epoch": 1.8170914542728636,
      "grad_norm": 0.06615204111385492,
      "learning_rate": 8.071027896968757e-05,
      "loss": 0.007,
      "step": 2424
    },
    {
      "epoch": 1.81784107946027,
      "grad_norm": 0.2143277633690874,
      "learning_rate": 8.06246825917518e-05,
      "loss": 0.0333,
      "step": 2425
    },
    {
      "epoch": 1.8185907046476761,
      "grad_norm": 0.20622070273292714,
      "learning_rate": 8.05391009607692e-05,
      "loss": 0.0245,
      "step": 2426
    },
    {
      "epoch": 1.8193403298350823,
      "grad_norm": 0.36423970885157425,
      "learning_rate": 8.045353414187771e-05,
      "loss": 0.0443,
      "step": 2427
    },
    {
      "epoch": 1.8200899550224887,
      "grad_norm": 0.09663320469898391,
      "learning_rate": 8.036798220020397e-05,
      "loss": 0.0198,
      "step": 2428
    },
    {
      "epoch": 1.8208395802098951,
      "grad_norm": 0.11993671368104146,
      "learning_rate": 8.028244520086337e-05,
      "loss": 0.0139,
      "step": 2429
    },
    {
      "epoch": 1.8215892053973013,
      "grad_norm": 0.23012223953056263,
      "learning_rate": 8.019692320895986e-05,
      "loss": 0.0212,
      "step": 2430
    },
    {
      "epoch": 1.8223388305847077,
      "grad_norm": 0.3594473594904879,
      "learning_rate": 8.0111416289586e-05,
      "loss": 0.0235,
      "step": 2431
    },
    {
      "epoch": 1.823088455772114,
      "grad_norm": 0.16619943656526007,
      "learning_rate": 8.002592450782287e-05,
      "loss": 0.0169,
      "step": 2432
    },
    {
      "epoch": 1.8238380809595203,
      "grad_norm": 0.10821109610505597,
      "learning_rate": 7.994044792874001e-05,
      "loss": 0.0109,
      "step": 2433
    },
    {
      "epoch": 1.8245877061469264,
      "grad_norm": 0.10944259492215419,
      "learning_rate": 7.98549866173954e-05,
      "loss": 0.0115,
      "step": 2434
    },
    {
      "epoch": 1.8253373313343328,
      "grad_norm": 0.16858836245980474,
      "learning_rate": 7.976954063883542e-05,
      "loss": 0.0105,
      "step": 2435
    },
    {
      "epoch": 1.8260869565217392,
      "grad_norm": 0.12882911211010897,
      "learning_rate": 7.968411005809476e-05,
      "loss": 0.01,
      "step": 2436
    },
    {
      "epoch": 1.8268365817091454,
      "grad_norm": 0.1869655169698048,
      "learning_rate": 7.959869494019638e-05,
      "loss": 0.0165,
      "step": 2437
    },
    {
      "epoch": 1.8275862068965516,
      "grad_norm": 0.15709456935311567,
      "learning_rate": 7.951329535015152e-05,
      "loss": 0.0202,
      "step": 2438
    },
    {
      "epoch": 1.828335832083958,
      "grad_norm": 0.26854060632952176,
      "learning_rate": 7.942791135295954e-05,
      "loss": 0.0137,
      "step": 2439
    },
    {
      "epoch": 1.8290854572713644,
      "grad_norm": 0.16090283940108951,
      "learning_rate": 7.934254301360797e-05,
      "loss": 0.0097,
      "step": 2440
    },
    {
      "epoch": 1.8298350824587706,
      "grad_norm": 0.0699283924536712,
      "learning_rate": 7.92571903970724e-05,
      "loss": 0.0099,
      "step": 2441
    },
    {
      "epoch": 1.8305847076461768,
      "grad_norm": 0.13273853251021484,
      "learning_rate": 7.917185356831648e-05,
      "loss": 0.0194,
      "step": 2442
    },
    {
      "epoch": 1.8313343328335832,
      "grad_norm": 0.08065492868490784,
      "learning_rate": 7.908653259229182e-05,
      "loss": 0.0092,
      "step": 2443
    },
    {
      "epoch": 1.8320839580209896,
      "grad_norm": 0.22776619342338636,
      "learning_rate": 7.900122753393794e-05,
      "loss": 0.0284,
      "step": 2444
    },
    {
      "epoch": 1.8328335832083957,
      "grad_norm": 0.07395140232369679,
      "learning_rate": 7.891593845818235e-05,
      "loss": 0.0136,
      "step": 2445
    },
    {
      "epoch": 1.8335832083958021,
      "grad_norm": 0.1562336933540751,
      "learning_rate": 7.883066542994026e-05,
      "loss": 0.0221,
      "step": 2446
    },
    {
      "epoch": 1.8343328335832085,
      "grad_norm": 0.2278781011842574,
      "learning_rate": 7.874540851411477e-05,
      "loss": 0.0341,
      "step": 2447
    },
    {
      "epoch": 1.8350824587706147,
      "grad_norm": 0.3267357831045852,
      "learning_rate": 7.866016777559665e-05,
      "loss": 0.0387,
      "step": 2448
    },
    {
      "epoch": 1.8358320839580209,
      "grad_norm": 0.09800556584697193,
      "learning_rate": 7.857494327926439e-05,
      "loss": 0.0121,
      "step": 2449
    },
    {
      "epoch": 1.8365817091454273,
      "grad_norm": 0.2104509100294957,
      "learning_rate": 7.84897350899841e-05,
      "loss": 0.0164,
      "step": 2450
    },
    {
      "epoch": 1.8373313343328337,
      "grad_norm": 0.14104718753315004,
      "learning_rate": 7.840454327260948e-05,
      "loss": 0.0171,
      "step": 2451
    },
    {
      "epoch": 1.8380809595202399,
      "grad_norm": 0.15792235266223165,
      "learning_rate": 7.83193678919818e-05,
      "loss": 0.0164,
      "step": 2452
    },
    {
      "epoch": 1.838830584707646,
      "grad_norm": 0.1687305523880618,
      "learning_rate": 7.823420901292979e-05,
      "loss": 0.0173,
      "step": 2453
    },
    {
      "epoch": 1.8395802098950524,
      "grad_norm": 0.16484720609451164,
      "learning_rate": 7.814906670026963e-05,
      "loss": 0.0136,
      "step": 2454
    },
    {
      "epoch": 1.8403298350824588,
      "grad_norm": 0.15187479733924325,
      "learning_rate": 7.806394101880489e-05,
      "loss": 0.0132,
      "step": 2455
    },
    {
      "epoch": 1.841079460269865,
      "grad_norm": 0.10826429549268864,
      "learning_rate": 7.797883203332645e-05,
      "loss": 0.0074,
      "step": 2456
    },
    {
      "epoch": 1.8418290854572712,
      "grad_norm": 0.08732079300226159,
      "learning_rate": 7.789373980861256e-05,
      "loss": 0.0107,
      "step": 2457
    },
    {
      "epoch": 1.8425787106446778,
      "grad_norm": 0.11564816944872554,
      "learning_rate": 7.780866440942862e-05,
      "loss": 0.0133,
      "step": 2458
    },
    {
      "epoch": 1.843328335832084,
      "grad_norm": 0.1478322613189044,
      "learning_rate": 7.77236059005273e-05,
      "loss": 0.0126,
      "step": 2459
    },
    {
      "epoch": 1.8440779610194902,
      "grad_norm": 0.12555558944479056,
      "learning_rate": 7.763856434664837e-05,
      "loss": 0.0106,
      "step": 2460
    },
    {
      "epoch": 1.8448275862068966,
      "grad_norm": 0.16809980334871072,
      "learning_rate": 7.755353981251867e-05,
      "loss": 0.0201,
      "step": 2461
    },
    {
      "epoch": 1.845577211394303,
      "grad_norm": 0.15373833992286495,
      "learning_rate": 7.74685323628522e-05,
      "loss": 0.0194,
      "step": 2462
    },
    {
      "epoch": 1.8463268365817092,
      "grad_norm": 0.105119977223022,
      "learning_rate": 7.738354206234984e-05,
      "loss": 0.0101,
      "step": 2463
    },
    {
      "epoch": 1.8470764617691153,
      "grad_norm": 0.19700610621960898,
      "learning_rate": 7.729856897569947e-05,
      "loss": 0.0187,
      "step": 2464
    },
    {
      "epoch": 1.8478260869565217,
      "grad_norm": 0.1627518862213529,
      "learning_rate": 7.721361316757584e-05,
      "loss": 0.0209,
      "step": 2465
    },
    {
      "epoch": 1.8485757121439281,
      "grad_norm": 0.07826135196893862,
      "learning_rate": 7.712867470264058e-05,
      "loss": 0.0082,
      "step": 2466
    },
    {
      "epoch": 1.8493253373313343,
      "grad_norm": 0.11640801846060447,
      "learning_rate": 7.704375364554209e-05,
      "loss": 0.0092,
      "step": 2467
    },
    {
      "epoch": 1.8500749625187405,
      "grad_norm": 0.1629061980346892,
      "learning_rate": 7.695885006091552e-05,
      "loss": 0.0226,
      "step": 2468
    },
    {
      "epoch": 1.8508245877061469,
      "grad_norm": 0.07211534313473594,
      "learning_rate": 7.68739640133828e-05,
      "loss": 0.006,
      "step": 2469
    },
    {
      "epoch": 1.8515742128935533,
      "grad_norm": 0.0693554167568505,
      "learning_rate": 7.678909556755239e-05,
      "loss": 0.0092,
      "step": 2470
    },
    {
      "epoch": 1.8523238380809595,
      "grad_norm": 0.11865432701123758,
      "learning_rate": 7.670424478801945e-05,
      "loss": 0.0169,
      "step": 2471
    },
    {
      "epoch": 1.8530734632683659,
      "grad_norm": 0.08334899698876731,
      "learning_rate": 7.661941173936564e-05,
      "loss": 0.0073,
      "step": 2472
    },
    {
      "epoch": 1.8538230884557723,
      "grad_norm": 0.24857862856866222,
      "learning_rate": 7.653459648615915e-05,
      "loss": 0.0199,
      "step": 2473
    },
    {
      "epoch": 1.8545727136431784,
      "grad_norm": 0.2482687204364232,
      "learning_rate": 7.644979909295458e-05,
      "loss": 0.0208,
      "step": 2474
    },
    {
      "epoch": 1.8553223388305846,
      "grad_norm": 0.19686906475653315,
      "learning_rate": 7.636501962429301e-05,
      "loss": 0.0159,
      "step": 2475
    },
    {
      "epoch": 1.856071964017991,
      "grad_norm": 0.10028826224834073,
      "learning_rate": 7.628025814470183e-05,
      "loss": 0.0139,
      "step": 2476
    },
    {
      "epoch": 1.8568215892053974,
      "grad_norm": 0.07742702749673577,
      "learning_rate": 7.619551471869476e-05,
      "loss": 0.0053,
      "step": 2477
    },
    {
      "epoch": 1.8575712143928036,
      "grad_norm": 0.14772903844749294,
      "learning_rate": 7.611078941077174e-05,
      "loss": 0.0192,
      "step": 2478
    },
    {
      "epoch": 1.8583208395802098,
      "grad_norm": 1.4667137683554423,
      "learning_rate": 7.602608228541898e-05,
      "loss": 0.0407,
      "step": 2479
    },
    {
      "epoch": 1.8590704647676162,
      "grad_norm": 0.12049533052266952,
      "learning_rate": 7.594139340710878e-05,
      "loss": 0.0098,
      "step": 2480
    },
    {
      "epoch": 1.8598200899550226,
      "grad_norm": 0.10478542644073217,
      "learning_rate": 7.585672284029962e-05,
      "loss": 0.0209,
      "step": 2481
    },
    {
      "epoch": 1.8605697151424287,
      "grad_norm": 0.16305850275521108,
      "learning_rate": 7.577207064943599e-05,
      "loss": 0.0201,
      "step": 2482
    },
    {
      "epoch": 1.861319340329835,
      "grad_norm": 0.12384962164874974,
      "learning_rate": 7.568743689894844e-05,
      "loss": 0.0101,
      "step": 2483
    },
    {
      "epoch": 1.8620689655172413,
      "grad_norm": 0.16888590528757494,
      "learning_rate": 7.560282165325342e-05,
      "loss": 0.0139,
      "step": 2484
    },
    {
      "epoch": 1.8628185907046477,
      "grad_norm": 0.09058292457085314,
      "learning_rate": 7.551822497675339e-05,
      "loss": 0.008,
      "step": 2485
    },
    {
      "epoch": 1.863568215892054,
      "grad_norm": 0.1359901168221532,
      "learning_rate": 7.54336469338366e-05,
      "loss": 0.0132,
      "step": 2486
    },
    {
      "epoch": 1.8643178410794603,
      "grad_norm": 0.0636307909796474,
      "learning_rate": 7.534908758887714e-05,
      "loss": 0.0045,
      "step": 2487
    },
    {
      "epoch": 1.8650674662668667,
      "grad_norm": 0.11762048285293739,
      "learning_rate": 7.526454700623484e-05,
      "loss": 0.0132,
      "step": 2488
    },
    {
      "epoch": 1.8658170914542729,
      "grad_norm": 0.14584312270419342,
      "learning_rate": 7.518002525025533e-05,
      "loss": 0.0111,
      "step": 2489
    },
    {
      "epoch": 1.866566716641679,
      "grad_norm": 0.057156129657237674,
      "learning_rate": 7.509552238526983e-05,
      "loss": 0.0052,
      "step": 2490
    },
    {
      "epoch": 1.8673163418290855,
      "grad_norm": 0.12061801169326321,
      "learning_rate": 7.501103847559523e-05,
      "loss": 0.0088,
      "step": 2491
    },
    {
      "epoch": 1.8680659670164919,
      "grad_norm": 0.3208573344470076,
      "learning_rate": 7.49265735855339e-05,
      "loss": 0.0237,
      "step": 2492
    },
    {
      "epoch": 1.868815592203898,
      "grad_norm": 0.08171237944606213,
      "learning_rate": 7.484212777937391e-05,
      "loss": 0.005,
      "step": 2493
    },
    {
      "epoch": 1.8695652173913042,
      "grad_norm": 0.10375078006941929,
      "learning_rate": 7.475770112138866e-05,
      "loss": 0.0075,
      "step": 2494
    },
    {
      "epoch": 1.8703148425787106,
      "grad_norm": 0.09544820985988885,
      "learning_rate": 7.467329367583705e-05,
      "loss": 0.01,
      "step": 2495
    },
    {
      "epoch": 1.871064467766117,
      "grad_norm": 0.05263185304535703,
      "learning_rate": 7.458890550696327e-05,
      "loss": 0.0067,
      "step": 2496
    },
    {
      "epoch": 1.8718140929535232,
      "grad_norm": 0.20191098821286693,
      "learning_rate": 7.45045366789969e-05,
      "loss": 0.0273,
      "step": 2497
    },
    {
      "epoch": 1.8725637181409296,
      "grad_norm": 0.1398162193094364,
      "learning_rate": 7.442018725615281e-05,
      "loss": 0.0097,
      "step": 2498
    },
    {
      "epoch": 1.873313343328336,
      "grad_norm": 0.07068993934881566,
      "learning_rate": 7.43358573026311e-05,
      "loss": 0.0063,
      "step": 2499
    },
    {
      "epoch": 1.8740629685157422,
      "grad_norm": 0.22625443784283683,
      "learning_rate": 7.425154688261702e-05,
      "loss": 0.0266,
      "step": 2500
    },
    {
      "epoch": 1.8748125937031483,
      "grad_norm": 0.12500420397821987,
      "learning_rate": 7.416725606028098e-05,
      "loss": 0.0075,
      "step": 2501
    },
    {
      "epoch": 1.8755622188905547,
      "grad_norm": 0.23427172980645034,
      "learning_rate": 7.408298489977843e-05,
      "loss": 0.0249,
      "step": 2502
    },
    {
      "epoch": 1.8763118440779611,
      "grad_norm": 0.11898534192386219,
      "learning_rate": 7.39987334652499e-05,
      "loss": 0.0076,
      "step": 2503
    },
    {
      "epoch": 1.8770614692653673,
      "grad_norm": 0.08373734719913158,
      "learning_rate": 7.39145018208209e-05,
      "loss": 0.0074,
      "step": 2504
    },
    {
      "epoch": 1.8778110944527735,
      "grad_norm": 0.09218030773502157,
      "learning_rate": 7.383029003060187e-05,
      "loss": 0.0055,
      "step": 2505
    },
    {
      "epoch": 1.87856071964018,
      "grad_norm": 0.16710726389458688,
      "learning_rate": 7.374609815868811e-05,
      "loss": 0.0113,
      "step": 2506
    },
    {
      "epoch": 1.8793103448275863,
      "grad_norm": 0.5572105217760861,
      "learning_rate": 7.366192626915981e-05,
      "loss": 0.0298,
      "step": 2507
    },
    {
      "epoch": 1.8800599700149925,
      "grad_norm": 0.14773496446583861,
      "learning_rate": 7.357777442608195e-05,
      "loss": 0.0178,
      "step": 2508
    },
    {
      "epoch": 1.8808095952023987,
      "grad_norm": 0.10878118667541402,
      "learning_rate": 7.34936426935042e-05,
      "loss": 0.011,
      "step": 2509
    },
    {
      "epoch": 1.881559220389805,
      "grad_norm": 0.3147812075243767,
      "learning_rate": 7.340953113546099e-05,
      "loss": 0.0275,
      "step": 2510
    },
    {
      "epoch": 1.8823088455772115,
      "grad_norm": 0.0634699160517354,
      "learning_rate": 7.332543981597131e-05,
      "loss": 0.0057,
      "step": 2511
    },
    {
      "epoch": 1.8830584707646176,
      "grad_norm": 0.1037282284127577,
      "learning_rate": 7.324136879903882e-05,
      "loss": 0.0059,
      "step": 2512
    },
    {
      "epoch": 1.883808095952024,
      "grad_norm": 0.28102259056009204,
      "learning_rate": 7.315731814865172e-05,
      "loss": 0.0418,
      "step": 2513
    },
    {
      "epoch": 1.8845577211394304,
      "grad_norm": 0.18082904773371083,
      "learning_rate": 7.307328792878265e-05,
      "loss": 0.0109,
      "step": 2514
    },
    {
      "epoch": 1.8853073463268366,
      "grad_norm": 0.27733764852475407,
      "learning_rate": 7.298927820338878e-05,
      "loss": 0.0129,
      "step": 2515
    },
    {
      "epoch": 1.8860569715142428,
      "grad_norm": 0.051804527098103306,
      "learning_rate": 7.290528903641158e-05,
      "loss": 0.0055,
      "step": 2516
    },
    {
      "epoch": 1.8868065967016492,
      "grad_norm": 0.1645664449774772,
      "learning_rate": 7.282132049177702e-05,
      "loss": 0.0125,
      "step": 2517
    },
    {
      "epoch": 1.8875562218890556,
      "grad_norm": 0.11093437917334256,
      "learning_rate": 7.273737263339526e-05,
      "loss": 0.0128,
      "step": 2518
    },
    {
      "epoch": 1.8883058470764618,
      "grad_norm": 0.1369863753526021,
      "learning_rate": 7.265344552516073e-05,
      "loss": 0.0171,
      "step": 2519
    },
    {
      "epoch": 1.889055472263868,
      "grad_norm": 0.1396122038028002,
      "learning_rate": 7.256953923095209e-05,
      "loss": 0.0133,
      "step": 2520
    },
    {
      "epoch": 1.8898050974512743,
      "grad_norm": 0.16666444163694127,
      "learning_rate": 7.248565381463211e-05,
      "loss": 0.0241,
      "step": 2521
    },
    {
      "epoch": 1.8905547226386807,
      "grad_norm": 0.09667613590644915,
      "learning_rate": 7.240178934004778e-05,
      "loss": 0.0084,
      "step": 2522
    },
    {
      "epoch": 1.891304347826087,
      "grad_norm": 0.28358846469437893,
      "learning_rate": 7.231794587103004e-05,
      "loss": 0.0534,
      "step": 2523
    },
    {
      "epoch": 1.892053973013493,
      "grad_norm": 0.2908858163773263,
      "learning_rate": 7.223412347139386e-05,
      "loss": 0.0159,
      "step": 2524
    },
    {
      "epoch": 1.8928035982008995,
      "grad_norm": 0.24064697503110768,
      "learning_rate": 7.215032220493825e-05,
      "loss": 0.014,
      "step": 2525
    },
    {
      "epoch": 1.893553223388306,
      "grad_norm": 0.3016866461196789,
      "learning_rate": 7.206654213544603e-05,
      "loss": 0.0119,
      "step": 2526
    },
    {
      "epoch": 1.894302848575712,
      "grad_norm": 0.2621412679543309,
      "learning_rate": 7.198278332668397e-05,
      "loss": 0.0357,
      "step": 2527
    },
    {
      "epoch": 1.8950524737631185,
      "grad_norm": 0.14120466768101003,
      "learning_rate": 7.189904584240258e-05,
      "loss": 0.0234,
      "step": 2528
    },
    {
      "epoch": 1.8958020989505249,
      "grad_norm": 0.09072777647343415,
      "learning_rate": 7.181532974633625e-05,
      "loss": 0.0103,
      "step": 2529
    },
    {
      "epoch": 1.896551724137931,
      "grad_norm": 0.14239900493075416,
      "learning_rate": 7.173163510220295e-05,
      "loss": 0.032,
      "step": 2530
    },
    {
      "epoch": 1.8973013493253372,
      "grad_norm": 0.19534998045802685,
      "learning_rate": 7.164796197370444e-05,
      "loss": 0.0221,
      "step": 2531
    },
    {
      "epoch": 1.8980509745127436,
      "grad_norm": 0.11225916161140852,
      "learning_rate": 7.156431042452609e-05,
      "loss": 0.0153,
      "step": 2532
    },
    {
      "epoch": 1.89880059970015,
      "grad_norm": 0.18199528832938758,
      "learning_rate": 7.148068051833676e-05,
      "loss": 0.034,
      "step": 2533
    },
    {
      "epoch": 1.8995502248875562,
      "grad_norm": 0.13537685599687774,
      "learning_rate": 7.139707231878893e-05,
      "loss": 0.0115,
      "step": 2534
    },
    {
      "epoch": 1.9002998500749624,
      "grad_norm": 0.07626680894580906,
      "learning_rate": 7.131348588951851e-05,
      "loss": 0.0101,
      "step": 2535
    },
    {
      "epoch": 1.9010494752623688,
      "grad_norm": 0.11597308993208941,
      "learning_rate": 7.122992129414485e-05,
      "loss": 0.0151,
      "step": 2536
    },
    {
      "epoch": 1.9017991004497752,
      "grad_norm": 0.09926554517236055,
      "learning_rate": 7.11463785962707e-05,
      "loss": 0.0091,
      "step": 2537
    },
    {
      "epoch": 1.9025487256371814,
      "grad_norm": 0.5994144403129555,
      "learning_rate": 7.10628578594821e-05,
      "loss": 0.0256,
      "step": 2538
    },
    {
      "epoch": 1.9032983508245878,
      "grad_norm": 0.21192680149026147,
      "learning_rate": 7.097935914734843e-05,
      "loss": 0.0247,
      "step": 2539
    },
    {
      "epoch": 1.9040479760119942,
      "grad_norm": 0.13979130816746754,
      "learning_rate": 7.089588252342222e-05,
      "loss": 0.0131,
      "step": 2540
    },
    {
      "epoch": 1.9047976011994003,
      "grad_norm": 0.1522540153218576,
      "learning_rate": 7.081242805123935e-05,
      "loss": 0.0138,
      "step": 2541
    },
    {
      "epoch": 1.9055472263868065,
      "grad_norm": 0.3475730065259247,
      "learning_rate": 7.072899579431868e-05,
      "loss": 0.0238,
      "step": 2542
    },
    {
      "epoch": 1.906296851574213,
      "grad_norm": 0.1584546299657851,
      "learning_rate": 7.064558581616223e-05,
      "loss": 0.0176,
      "step": 2543
    },
    {
      "epoch": 1.9070464767616193,
      "grad_norm": 0.12628040865730267,
      "learning_rate": 7.056219818025504e-05,
      "loss": 0.0204,
      "step": 2544
    },
    {
      "epoch": 1.9077961019490255,
      "grad_norm": 0.158931488712341,
      "learning_rate": 7.047883295006516e-05,
      "loss": 0.0138,
      "step": 2545
    },
    {
      "epoch": 1.9085457271364317,
      "grad_norm": 0.140558609819867,
      "learning_rate": 7.03954901890436e-05,
      "loss": 0.0122,
      "step": 2546
    },
    {
      "epoch": 1.909295352323838,
      "grad_norm": 0.07932710076810945,
      "learning_rate": 7.031216996062426e-05,
      "loss": 0.0134,
      "step": 2547
    },
    {
      "epoch": 1.9100449775112445,
      "grad_norm": 0.08371999708812815,
      "learning_rate": 7.022887232822384e-05,
      "loss": 0.0104,
      "step": 2548
    },
    {
      "epoch": 1.9107946026986506,
      "grad_norm": 0.13582280816137274,
      "learning_rate": 7.01455973552419e-05,
      "loss": 0.0083,
      "step": 2549
    },
    {
      "epoch": 1.9115442278860568,
      "grad_norm": 0.10624807259301532,
      "learning_rate": 7.006234510506076e-05,
      "loss": 0.0119,
      "step": 2550
    },
    {
      "epoch": 1.9122938530734632,
      "grad_norm": 0.15543251325160595,
      "learning_rate": 6.997911564104541e-05,
      "loss": 0.0152,
      "step": 2551
    },
    {
      "epoch": 1.9130434782608696,
      "grad_norm": 0.0921239950798916,
      "learning_rate": 6.989590902654349e-05,
      "loss": 0.0088,
      "step": 2552
    },
    {
      "epoch": 1.9137931034482758,
      "grad_norm": 0.41388948225918165,
      "learning_rate": 6.981272532488529e-05,
      "loss": 0.0568,
      "step": 2553
    },
    {
      "epoch": 1.9145427286356822,
      "grad_norm": 0.100566670663976,
      "learning_rate": 6.97295645993836e-05,
      "loss": 0.0108,
      "step": 2554
    },
    {
      "epoch": 1.9152923538230886,
      "grad_norm": 0.09049484776525325,
      "learning_rate": 6.964642691333383e-05,
      "loss": 0.014,
      "step": 2555
    },
    {
      "epoch": 1.9160419790104948,
      "grad_norm": 0.07059096508827815,
      "learning_rate": 6.956331233001374e-05,
      "loss": 0.0056,
      "step": 2556
    },
    {
      "epoch": 1.916791604197901,
      "grad_norm": 0.3242025573619945,
      "learning_rate": 6.948022091268356e-05,
      "loss": 0.0338,
      "step": 2557
    },
    {
      "epoch": 1.9175412293853074,
      "grad_norm": 0.1967536374678093,
      "learning_rate": 6.93971527245859e-05,
      "loss": 0.0181,
      "step": 2558
    },
    {
      "epoch": 1.9182908545727138,
      "grad_norm": 0.09964807966099669,
      "learning_rate": 6.931410782894562e-05,
      "loss": 0.0055,
      "step": 2559
    },
    {
      "epoch": 1.91904047976012,
      "grad_norm": 0.1056229679621337,
      "learning_rate": 6.923108628896995e-05,
      "loss": 0.0109,
      "step": 2560
    },
    {
      "epoch": 1.919790104947526,
      "grad_norm": 0.2675761383101984,
      "learning_rate": 6.914808816784826e-05,
      "loss": 0.0175,
      "step": 2561
    },
    {
      "epoch": 1.9205397301349325,
      "grad_norm": 0.07748755906739309,
      "learning_rate": 6.906511352875216e-05,
      "loss": 0.0079,
      "step": 2562
    },
    {
      "epoch": 1.921289355322339,
      "grad_norm": 0.07310728412254099,
      "learning_rate": 6.89821624348353e-05,
      "loss": 0.011,
      "step": 2563
    },
    {
      "epoch": 1.922038980509745,
      "grad_norm": 0.0957840682527892,
      "learning_rate": 6.889923494923352e-05,
      "loss": 0.0084,
      "step": 2564
    },
    {
      "epoch": 1.9227886056971513,
      "grad_norm": 0.36935961501338727,
      "learning_rate": 6.881633113506463e-05,
      "loss": 0.0422,
      "step": 2565
    },
    {
      "epoch": 1.9235382308845579,
      "grad_norm": 0.16293408389417657,
      "learning_rate": 6.873345105542842e-05,
      "loss": 0.0141,
      "step": 2566
    },
    {
      "epoch": 1.924287856071964,
      "grad_norm": 0.13888112998811752,
      "learning_rate": 6.865059477340662e-05,
      "loss": 0.0138,
      "step": 2567
    },
    {
      "epoch": 1.9250374812593702,
      "grad_norm": 0.289847503646193,
      "learning_rate": 6.856776235206288e-05,
      "loss": 0.0334,
      "step": 2568
    },
    {
      "epoch": 1.9257871064467766,
      "grad_norm": 0.12933008451268538,
      "learning_rate": 6.848495385444265e-05,
      "loss": 0.01,
      "step": 2569
    },
    {
      "epoch": 1.926536731634183,
      "grad_norm": 0.16191506189682173,
      "learning_rate": 6.840216934357318e-05,
      "loss": 0.018,
      "step": 2570
    },
    {
      "epoch": 1.9272863568215892,
      "grad_norm": 0.18943025035089628,
      "learning_rate": 6.831940888246344e-05,
      "loss": 0.0138,
      "step": 2571
    },
    {
      "epoch": 1.9280359820089954,
      "grad_norm": 0.1061248577111503,
      "learning_rate": 6.823667253410417e-05,
      "loss": 0.0143,
      "step": 2572
    },
    {
      "epoch": 1.9287856071964018,
      "grad_norm": 0.12445565669107847,
      "learning_rate": 6.815396036146767e-05,
      "loss": 0.0078,
      "step": 2573
    },
    {
      "epoch": 1.9295352323838082,
      "grad_norm": 0.110562071079399,
      "learning_rate": 6.80712724275079e-05,
      "loss": 0.0091,
      "step": 2574
    },
    {
      "epoch": 1.9302848575712144,
      "grad_norm": 0.11468291069499537,
      "learning_rate": 6.798860879516035e-05,
      "loss": 0.0141,
      "step": 2575
    },
    {
      "epoch": 1.9310344827586206,
      "grad_norm": 0.0934202521942271,
      "learning_rate": 6.790596952734199e-05,
      "loss": 0.0099,
      "step": 2576
    },
    {
      "epoch": 1.931784107946027,
      "grad_norm": 0.0929281480526714,
      "learning_rate": 6.782335468695127e-05,
      "loss": 0.0117,
      "step": 2577
    },
    {
      "epoch": 1.9325337331334334,
      "grad_norm": 0.10176325279678822,
      "learning_rate": 6.774076433686809e-05,
      "loss": 0.0103,
      "step": 2578
    },
    {
      "epoch": 1.9332833583208395,
      "grad_norm": 0.21709643686481206,
      "learning_rate": 6.765819853995364e-05,
      "loss": 0.0403,
      "step": 2579
    },
    {
      "epoch": 1.934032983508246,
      "grad_norm": 0.17589843106110198,
      "learning_rate": 6.757565735905043e-05,
      "loss": 0.0217,
      "step": 2580
    },
    {
      "epoch": 1.9347826086956523,
      "grad_norm": 0.11434728905622538,
      "learning_rate": 6.749314085698229e-05,
      "loss": 0.0223,
      "step": 2581
    },
    {
      "epoch": 1.9355322338830585,
      "grad_norm": 0.14487603652395825,
      "learning_rate": 6.74106490965542e-05,
      "loss": 0.0083,
      "step": 2582
    },
    {
      "epoch": 1.9362818590704647,
      "grad_norm": 0.1289344714288651,
      "learning_rate": 6.732818214055233e-05,
      "loss": 0.0103,
      "step": 2583
    },
    {
      "epoch": 1.937031484257871,
      "grad_norm": 0.11890488272970408,
      "learning_rate": 6.7245740051744e-05,
      "loss": 0.0125,
      "step": 2584
    },
    {
      "epoch": 1.9377811094452775,
      "grad_norm": 0.09133697176455517,
      "learning_rate": 6.716332289287759e-05,
      "loss": 0.0145,
      "step": 2585
    },
    {
      "epoch": 1.9385307346326837,
      "grad_norm": 0.109331379422392,
      "learning_rate": 6.70809307266825e-05,
      "loss": 0.0168,
      "step": 2586
    },
    {
      "epoch": 1.9392803598200898,
      "grad_norm": 0.20016189865941686,
      "learning_rate": 6.699856361586905e-05,
      "loss": 0.0332,
      "step": 2587
    },
    {
      "epoch": 1.9400299850074962,
      "grad_norm": 0.18856090528988828,
      "learning_rate": 6.69162216231286e-05,
      "loss": 0.0121,
      "step": 2588
    },
    {
      "epoch": 1.9407796101949026,
      "grad_norm": 0.09007171352149351,
      "learning_rate": 6.68339048111333e-05,
      "loss": 0.0126,
      "step": 2589
    },
    {
      "epoch": 1.9415292353823088,
      "grad_norm": 0.10575179313755036,
      "learning_rate": 6.675161324253623e-05,
      "loss": 0.0096,
      "step": 2590
    },
    {
      "epoch": 1.942278860569715,
      "grad_norm": 0.11672959400915874,
      "learning_rate": 6.666934697997113e-05,
      "loss": 0.016,
      "step": 2591
    },
    {
      "epoch": 1.9430284857571214,
      "grad_norm": 0.1559846824352955,
      "learning_rate": 6.658710608605259e-05,
      "loss": 0.0165,
      "step": 2592
    },
    {
      "epoch": 1.9437781109445278,
      "grad_norm": 0.16962939613420358,
      "learning_rate": 6.650489062337584e-05,
      "loss": 0.029,
      "step": 2593
    },
    {
      "epoch": 1.944527736131934,
      "grad_norm": 0.12100652827555615,
      "learning_rate": 6.642270065451677e-05,
      "loss": 0.0035,
      "step": 2594
    },
    {
      "epoch": 1.9452773613193404,
      "grad_norm": 0.22173551476179107,
      "learning_rate": 6.634053624203182e-05,
      "loss": 0.02,
      "step": 2595
    },
    {
      "epoch": 1.9460269865067468,
      "grad_norm": 0.16418019953890273,
      "learning_rate": 6.625839744845807e-05,
      "loss": 0.0092,
      "step": 2596
    },
    {
      "epoch": 1.946776611694153,
      "grad_norm": 0.08129102733379626,
      "learning_rate": 6.6176284336313e-05,
      "loss": 0.0058,
      "step": 2597
    },
    {
      "epoch": 1.9475262368815591,
      "grad_norm": 0.1808712429174813,
      "learning_rate": 6.609419696809462e-05,
      "loss": 0.0113,
      "step": 2598
    },
    {
      "epoch": 1.9482758620689655,
      "grad_norm": 0.11860297035413558,
      "learning_rate": 6.601213540628131e-05,
      "loss": 0.0092,
      "step": 2599
    },
    {
      "epoch": 1.949025487256372,
      "grad_norm": 0.13540127438080782,
      "learning_rate": 6.593009971333179e-05,
      "loss": 0.0101,
      "step": 2600
    },
    {
      "epoch": 1.949775112443778,
      "grad_norm": 0.10447994790470146,
      "learning_rate": 6.584808995168515e-05,
      "loss": 0.0084,
      "step": 2601
    },
    {
      "epoch": 1.9505247376311843,
      "grad_norm": 0.15497645159707168,
      "learning_rate": 6.576610618376071e-05,
      "loss": 0.0117,
      "step": 2602
    },
    {
      "epoch": 1.9512743628185907,
      "grad_norm": 0.09902340572842126,
      "learning_rate": 6.568414847195798e-05,
      "loss": 0.0057,
      "step": 2603
    },
    {
      "epoch": 1.952023988005997,
      "grad_norm": 0.1239254369290511,
      "learning_rate": 6.560221687865669e-05,
      "loss": 0.0131,
      "step": 2604
    },
    {
      "epoch": 1.9527736131934033,
      "grad_norm": 0.19577055242023142,
      "learning_rate": 6.552031146621664e-05,
      "loss": 0.0149,
      "step": 2605
    },
    {
      "epoch": 1.9535232383808094,
      "grad_norm": 0.14953347555603855,
      "learning_rate": 6.543843229697772e-05,
      "loss": 0.0175,
      "step": 2606
    },
    {
      "epoch": 1.954272863568216,
      "grad_norm": 0.16228867246023873,
      "learning_rate": 6.535657943325988e-05,
      "loss": 0.0149,
      "step": 2607
    },
    {
      "epoch": 1.9550224887556222,
      "grad_norm": 0.18842441467737261,
      "learning_rate": 6.527475293736301e-05,
      "loss": 0.0153,
      "step": 2608
    },
    {
      "epoch": 1.9557721139430284,
      "grad_norm": 0.11570043501747544,
      "learning_rate": 6.519295287156692e-05,
      "loss": 0.0073,
      "step": 2609
    },
    {
      "epoch": 1.9565217391304348,
      "grad_norm": 0.10125016943360128,
      "learning_rate": 6.511117929813135e-05,
      "loss": 0.0113,
      "step": 2610
    },
    {
      "epoch": 1.9572713643178412,
      "grad_norm": 0.27885175542005797,
      "learning_rate": 6.502943227929586e-05,
      "loss": 0.024,
      "step": 2611
    },
    {
      "epoch": 1.9580209895052474,
      "grad_norm": 0.12116488059211396,
      "learning_rate": 6.494771187727975e-05,
      "loss": 0.0265,
      "step": 2612
    },
    {
      "epoch": 1.9587706146926536,
      "grad_norm": 0.23356339007737725,
      "learning_rate": 6.486601815428213e-05,
      "loss": 0.0214,
      "step": 2613
    },
    {
      "epoch": 1.95952023988006,
      "grad_norm": 0.16281117081626795,
      "learning_rate": 6.478435117248175e-05,
      "loss": 0.0282,
      "step": 2614
    },
    {
      "epoch": 1.9602698650674664,
      "grad_norm": 0.14826397127572882,
      "learning_rate": 6.470271099403709e-05,
      "loss": 0.0172,
      "step": 2615
    },
    {
      "epoch": 1.9610194902548725,
      "grad_norm": 0.16639910228693838,
      "learning_rate": 6.462109768108612e-05,
      "loss": 0.0155,
      "step": 2616
    },
    {
      "epoch": 1.9617691154422787,
      "grad_norm": 0.12878213613918643,
      "learning_rate": 6.453951129574644e-05,
      "loss": 0.0047,
      "step": 2617
    },
    {
      "epoch": 1.9625187406296851,
      "grad_norm": 0.23620891755437637,
      "learning_rate": 6.44579519001151e-05,
      "loss": 0.0191,
      "step": 2618
    },
    {
      "epoch": 1.9632683658170915,
      "grad_norm": 0.47003799940530394,
      "learning_rate": 6.437641955626867e-05,
      "loss": 0.0346,
      "step": 2619
    },
    {
      "epoch": 1.9640179910044977,
      "grad_norm": 0.10834439872357306,
      "learning_rate": 6.429491432626306e-05,
      "loss": 0.0074,
      "step": 2620
    },
    {
      "epoch": 1.964767616191904,
      "grad_norm": 0.30263593519837156,
      "learning_rate": 6.42134362721336e-05,
      "loss": 0.0345,
      "step": 2621
    },
    {
      "epoch": 1.9655172413793105,
      "grad_norm": 0.09886349728154153,
      "learning_rate": 6.413198545589493e-05,
      "loss": 0.0086,
      "step": 2622
    },
    {
      "epoch": 1.9662668665667167,
      "grad_norm": 0.11214795676054878,
      "learning_rate": 6.405056193954092e-05,
      "loss": 0.0159,
      "step": 2623
    },
    {
      "epoch": 1.9670164917541229,
      "grad_norm": 0.08743795758245536,
      "learning_rate": 6.396916578504467e-05,
      "loss": 0.0087,
      "step": 2624
    },
    {
      "epoch": 1.9677661169415293,
      "grad_norm": 0.1203291476144883,
      "learning_rate": 6.388779705435852e-05,
      "loss": 0.0124,
      "step": 2625
    },
    {
      "epoch": 1.9685157421289357,
      "grad_norm": 0.13694478120673487,
      "learning_rate": 6.380645580941385e-05,
      "loss": 0.0144,
      "step": 2626
    },
    {
      "epoch": 1.9692653673163418,
      "grad_norm": 0.20963426631790033,
      "learning_rate": 6.372514211212116e-05,
      "loss": 0.0204,
      "step": 2627
    },
    {
      "epoch": 1.970014992503748,
      "grad_norm": 0.10474781082841776,
      "learning_rate": 6.364385602437e-05,
      "loss": 0.0113,
      "step": 2628
    },
    {
      "epoch": 1.9707646176911544,
      "grad_norm": 0.12605094387952967,
      "learning_rate": 6.356259760802886e-05,
      "loss": 0.0206,
      "step": 2629
    },
    {
      "epoch": 1.9715142428785608,
      "grad_norm": 0.10242238045281239,
      "learning_rate": 6.348136692494519e-05,
      "loss": 0.0084,
      "step": 2630
    },
    {
      "epoch": 1.972263868065967,
      "grad_norm": 0.4949245664636189,
      "learning_rate": 6.340016403694537e-05,
      "loss": 0.0146,
      "step": 2631
    },
    {
      "epoch": 1.9730134932533732,
      "grad_norm": 0.24511036686641371,
      "learning_rate": 6.331898900583458e-05,
      "loss": 0.022,
      "step": 2632
    },
    {
      "epoch": 1.9737631184407796,
      "grad_norm": 0.07186690641381413,
      "learning_rate": 6.323784189339677e-05,
      "loss": 0.0064,
      "step": 2633
    },
    {
      "epoch": 1.974512743628186,
      "grad_norm": 0.19864117764803166,
      "learning_rate": 6.315672276139474e-05,
      "loss": 0.0375,
      "step": 2634
    },
    {
      "epoch": 1.9752623688155921,
      "grad_norm": 0.27687594522339515,
      "learning_rate": 6.307563167156988e-05,
      "loss": 0.0272,
      "step": 2635
    },
    {
      "epoch": 1.9760119940029985,
      "grad_norm": 0.1775523484598218,
      "learning_rate": 6.299456868564235e-05,
      "loss": 0.0188,
      "step": 2636
    },
    {
      "epoch": 1.976761619190405,
      "grad_norm": 0.12004662288287847,
      "learning_rate": 6.291353386531074e-05,
      "loss": 0.0099,
      "step": 2637
    },
    {
      "epoch": 1.9775112443778111,
      "grad_norm": 0.3073542949957989,
      "learning_rate": 6.283252727225245e-05,
      "loss": 0.0338,
      "step": 2638
    },
    {
      "epoch": 1.9782608695652173,
      "grad_norm": 0.21760402679314944,
      "learning_rate": 6.275154896812323e-05,
      "loss": 0.0111,
      "step": 2639
    },
    {
      "epoch": 1.9790104947526237,
      "grad_norm": 0.04176693977010684,
      "learning_rate": 6.26705990145573e-05,
      "loss": 0.0038,
      "step": 2640
    },
    {
      "epoch": 1.97976011994003,
      "grad_norm": 0.08343985178592077,
      "learning_rate": 6.258967747316738e-05,
      "loss": 0.0042,
      "step": 2641
    },
    {
      "epoch": 1.9805097451274363,
      "grad_norm": 0.09999060085646638,
      "learning_rate": 6.250878440554448e-05,
      "loss": 0.0141,
      "step": 2642
    },
    {
      "epoch": 1.9812593703148424,
      "grad_norm": 0.096873305567595,
      "learning_rate": 6.2427919873258e-05,
      "loss": 0.0073,
      "step": 2643
    },
    {
      "epoch": 1.9820089955022488,
      "grad_norm": 0.11376993253947391,
      "learning_rate": 6.234708393785563e-05,
      "loss": 0.0085,
      "step": 2644
    },
    {
      "epoch": 1.9827586206896552,
      "grad_norm": 0.13578217581490346,
      "learning_rate": 6.22662766608632e-05,
      "loss": 0.0277,
      "step": 2645
    },
    {
      "epoch": 1.9835082458770614,
      "grad_norm": 0.10521727242358125,
      "learning_rate": 6.218549810378485e-05,
      "loss": 0.0215,
      "step": 2646
    },
    {
      "epoch": 1.9842578710644676,
      "grad_norm": 0.13965904247520539,
      "learning_rate": 6.210474832810276e-05,
      "loss": 0.012,
      "step": 2647
    },
    {
      "epoch": 1.9850074962518742,
      "grad_norm": 0.11152054461686513,
      "learning_rate": 6.202402739527729e-05,
      "loss": 0.0198,
      "step": 2648
    },
    {
      "epoch": 1.9857571214392804,
      "grad_norm": 0.1477238847811376,
      "learning_rate": 6.19433353667468e-05,
      "loss": 0.0209,
      "step": 2649
    },
    {
      "epoch": 1.9865067466266866,
      "grad_norm": 0.19783507828736474,
      "learning_rate": 6.186267230392762e-05,
      "loss": 0.0241,
      "step": 2650
    },
    {
      "epoch": 1.987256371814093,
      "grad_norm": 0.13984137791498613,
      "learning_rate": 6.178203826821409e-05,
      "loss": 0.0167,
      "step": 2651
    },
    {
      "epoch": 1.9880059970014994,
      "grad_norm": 0.1739092071743353,
      "learning_rate": 6.170143332097843e-05,
      "loss": 0.0264,
      "step": 2652
    },
    {
      "epoch": 1.9887556221889056,
      "grad_norm": 0.080112857244923,
      "learning_rate": 6.162085752357076e-05,
      "loss": 0.0105,
      "step": 2653
    },
    {
      "epoch": 1.9895052473763117,
      "grad_norm": 0.1773606487440933,
      "learning_rate": 6.154031093731894e-05,
      "loss": 0.0189,
      "step": 2654
    },
    {
      "epoch": 1.9902548725637181,
      "grad_norm": 0.08855210818751634,
      "learning_rate": 6.145979362352862e-05,
      "loss": 0.0138,
      "step": 2655
    },
    {
      "epoch": 1.9910044977511245,
      "grad_norm": 0.09639909177775023,
      "learning_rate": 6.137930564348322e-05,
      "loss": 0.0125,
      "step": 2656
    },
    {
      "epoch": 1.9917541229385307,
      "grad_norm": 0.08980465027172353,
      "learning_rate": 6.12988470584438e-05,
      "loss": 0.0117,
      "step": 2657
    },
    {
      "epoch": 1.992503748125937,
      "grad_norm": 0.1926546990612494,
      "learning_rate": 6.121841792964901e-05,
      "loss": 0.0122,
      "step": 2658
    },
    {
      "epoch": 1.9932533733133433,
      "grad_norm": 0.15440555314144272,
      "learning_rate": 6.113801831831518e-05,
      "loss": 0.0156,
      "step": 2659
    },
    {
      "epoch": 1.9940029985007497,
      "grad_norm": 0.09339135907497718,
      "learning_rate": 6.105764828563607e-05,
      "loss": 0.0069,
      "step": 2660
    },
    {
      "epoch": 1.9947526236881559,
      "grad_norm": 0.2918423018977411,
      "learning_rate": 6.097730789278292e-05,
      "loss": 0.0532,
      "step": 2661
    },
    {
      "epoch": 1.9955022488755623,
      "grad_norm": 0.10195907150794488,
      "learning_rate": 6.089699720090455e-05,
      "loss": 0.0067,
      "step": 2662
    },
    {
      "epoch": 1.9962518740629687,
      "grad_norm": 0.1235237078063373,
      "learning_rate": 6.081671627112704e-05,
      "loss": 0.0094,
      "step": 2663
    },
    {
      "epoch": 1.9970014992503748,
      "grad_norm": 0.07032145735999436,
      "learning_rate": 6.073646516455387e-05,
      "loss": 0.006,
      "step": 2664
    },
    {
      "epoch": 1.997751124437781,
      "grad_norm": 0.05541573305685681,
      "learning_rate": 6.065624394226579e-05,
      "loss": 0.0059,
      "step": 2665
    },
    {
      "epoch": 1.9985007496251874,
      "grad_norm": 0.10269356709036573,
      "learning_rate": 6.057605266532084e-05,
      "loss": 0.018,
      "step": 2666
    },
    {
      "epoch": 1.9992503748125938,
      "grad_norm": 0.21964472646986308,
      "learning_rate": 6.049589139475424e-05,
      "loss": 0.0181,
      "step": 2667
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.11976827179422106,
      "learning_rate": 6.04157601915784e-05,
      "loss": 0.028,
      "step": 2668
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.029895687475800514,
      "eval_runtime": 1872.329,
      "eval_samples_per_second": 5.542,
      "eval_steps_per_second": 0.693,
      "step": 2668
    },
    {
      "epoch": 2.000749625187406,
      "grad_norm": 0.12920673369246413,
      "learning_rate": 6.0335659116782825e-05,
      "loss": 0.0149,
      "step": 2669
    },
    {
      "epoch": 2.001499250374813,
      "grad_norm": 0.0920728899931142,
      "learning_rate": 6.0255588231334056e-05,
      "loss": 0.0103,
      "step": 2670
    },
    {
      "epoch": 2.002248875562219,
      "grad_norm": 0.08511625814268858,
      "learning_rate": 6.017554759617575e-05,
      "loss": 0.0081,
      "step": 2671
    },
    {
      "epoch": 2.002998500749625,
      "grad_norm": 0.05220608860817644,
      "learning_rate": 6.009553727222846e-05,
      "loss": 0.0049,
      "step": 2672
    },
    {
      "epoch": 2.0037481259370313,
      "grad_norm": 0.4497006749186715,
      "learning_rate": 6.0015557320389695e-05,
      "loss": 0.0193,
      "step": 2673
    },
    {
      "epoch": 2.004497751124438,
      "grad_norm": 0.04101615202066704,
      "learning_rate": 5.993560780153384e-05,
      "loss": 0.0024,
      "step": 2674
    },
    {
      "epoch": 2.005247376311844,
      "grad_norm": 0.08383419790926522,
      "learning_rate": 5.985568877651211e-05,
      "loss": 0.0089,
      "step": 2675
    },
    {
      "epoch": 2.0059970014992503,
      "grad_norm": 0.1879415339577029,
      "learning_rate": 5.977580030615254e-05,
      "loss": 0.0078,
      "step": 2676
    },
    {
      "epoch": 2.0067466266866565,
      "grad_norm": 0.031140431853141123,
      "learning_rate": 5.969594245125988e-05,
      "loss": 0.0024,
      "step": 2677
    },
    {
      "epoch": 2.007496251874063,
      "grad_norm": 0.08823839659496005,
      "learning_rate": 5.961611527261559e-05,
      "loss": 0.0118,
      "step": 2678
    },
    {
      "epoch": 2.0082458770614693,
      "grad_norm": 0.07251362090044315,
      "learning_rate": 5.953631883097777e-05,
      "loss": 0.0063,
      "step": 2679
    },
    {
      "epoch": 2.0089955022488755,
      "grad_norm": 0.06889138777796779,
      "learning_rate": 5.9456553187081146e-05,
      "loss": 0.0063,
      "step": 2680
    },
    {
      "epoch": 2.0097451274362816,
      "grad_norm": 0.17612485680220627,
      "learning_rate": 5.937681840163698e-05,
      "loss": 0.0164,
      "step": 2681
    },
    {
      "epoch": 2.0104947526236883,
      "grad_norm": 0.18739782304747712,
      "learning_rate": 5.92971145353331e-05,
      "loss": 0.0118,
      "step": 2682
    },
    {
      "epoch": 2.0112443778110944,
      "grad_norm": 0.14708308312536258,
      "learning_rate": 5.9217441648833714e-05,
      "loss": 0.0115,
      "step": 2683
    },
    {
      "epoch": 2.0119940029985006,
      "grad_norm": 0.15935890992827711,
      "learning_rate": 5.913779980277951e-05,
      "loss": 0.0159,
      "step": 2684
    },
    {
      "epoch": 2.0127436281859072,
      "grad_norm": 0.02942228231752664,
      "learning_rate": 5.905818905778753e-05,
      "loss": 0.0019,
      "step": 2685
    },
    {
      "epoch": 2.0134932533733134,
      "grad_norm": 0.11449594933651004,
      "learning_rate": 5.8978609474451176e-05,
      "loss": 0.0137,
      "step": 2686
    },
    {
      "epoch": 2.0142428785607196,
      "grad_norm": 0.16924864902098877,
      "learning_rate": 5.8899061113340117e-05,
      "loss": 0.0257,
      "step": 2687
    },
    {
      "epoch": 2.0149925037481258,
      "grad_norm": 0.11744959118368047,
      "learning_rate": 5.881954403500021e-05,
      "loss": 0.0245,
      "step": 2688
    },
    {
      "epoch": 2.0157421289355324,
      "grad_norm": 0.08773876902030525,
      "learning_rate": 5.874005829995357e-05,
      "loss": 0.002,
      "step": 2689
    },
    {
      "epoch": 2.0164917541229386,
      "grad_norm": 0.08362034791556955,
      "learning_rate": 5.8660603968698416e-05,
      "loss": 0.0069,
      "step": 2690
    },
    {
      "epoch": 2.0172413793103448,
      "grad_norm": 0.1398009462913696,
      "learning_rate": 5.858118110170906e-05,
      "loss": 0.0143,
      "step": 2691
    },
    {
      "epoch": 2.017991004497751,
      "grad_norm": 0.15455916642541287,
      "learning_rate": 5.850178975943591e-05,
      "loss": 0.0062,
      "step": 2692
    },
    {
      "epoch": 2.0187406296851576,
      "grad_norm": 0.06891922749265855,
      "learning_rate": 5.842243000230532e-05,
      "loss": 0.0045,
      "step": 2693
    },
    {
      "epoch": 2.0194902548725637,
      "grad_norm": 0.17892536666544281,
      "learning_rate": 5.83431018907197e-05,
      "loss": 0.0194,
      "step": 2694
    },
    {
      "epoch": 2.02023988005997,
      "grad_norm": 0.11063179206160058,
      "learning_rate": 5.8263805485057235e-05,
      "loss": 0.0083,
      "step": 2695
    },
    {
      "epoch": 2.0209895052473765,
      "grad_norm": 0.09425654621063746,
      "learning_rate": 5.8184540845672085e-05,
      "loss": 0.0043,
      "step": 2696
    },
    {
      "epoch": 2.0217391304347827,
      "grad_norm": 0.09750334625327865,
      "learning_rate": 5.8105308032894216e-05,
      "loss": 0.0079,
      "step": 2697
    },
    {
      "epoch": 2.022488755622189,
      "grad_norm": 0.0794308388984354,
      "learning_rate": 5.802610710702927e-05,
      "loss": 0.0046,
      "step": 2698
    },
    {
      "epoch": 2.023238380809595,
      "grad_norm": 0.09362722963585723,
      "learning_rate": 5.794693812835883e-05,
      "loss": 0.0049,
      "step": 2699
    },
    {
      "epoch": 2.0239880059970017,
      "grad_norm": 0.06944043013400002,
      "learning_rate": 5.7867801157139965e-05,
      "loss": 0.0044,
      "step": 2700
    },
    {
      "epoch": 2.024737631184408,
      "grad_norm": 0.049746400026874464,
      "learning_rate": 5.778869625360546e-05,
      "loss": 0.0052,
      "step": 2701
    },
    {
      "epoch": 2.025487256371814,
      "grad_norm": 0.3734118178269151,
      "learning_rate": 5.7709623477963694e-05,
      "loss": 0.0091,
      "step": 2702
    },
    {
      "epoch": 2.02623688155922,
      "grad_norm": 0.4872267575568091,
      "learning_rate": 5.763058289039859e-05,
      "loss": 0.0376,
      "step": 2703
    },
    {
      "epoch": 2.026986506746627,
      "grad_norm": 0.057540021071794865,
      "learning_rate": 5.7551574551069554e-05,
      "loss": 0.0033,
      "step": 2704
    },
    {
      "epoch": 2.027736131934033,
      "grad_norm": 0.15688083528482955,
      "learning_rate": 5.7472598520111464e-05,
      "loss": 0.0102,
      "step": 2705
    },
    {
      "epoch": 2.028485757121439,
      "grad_norm": 0.129700793724637,
      "learning_rate": 5.73936548576346e-05,
      "loss": 0.0118,
      "step": 2706
    },
    {
      "epoch": 2.0292353823088454,
      "grad_norm": 0.09155531612100912,
      "learning_rate": 5.731474362372462e-05,
      "loss": 0.0088,
      "step": 2707
    },
    {
      "epoch": 2.029985007496252,
      "grad_norm": 0.0830634850988361,
      "learning_rate": 5.723586487844248e-05,
      "loss": 0.005,
      "step": 2708
    },
    {
      "epoch": 2.030734632683658,
      "grad_norm": 0.18270164608031197,
      "learning_rate": 5.7157018681824434e-05,
      "loss": 0.0041,
      "step": 2709
    },
    {
      "epoch": 2.0314842578710643,
      "grad_norm": 0.1053575328747016,
      "learning_rate": 5.7078205093881885e-05,
      "loss": 0.0087,
      "step": 2710
    },
    {
      "epoch": 2.032233883058471,
      "grad_norm": 0.04528738014190908,
      "learning_rate": 5.699942417460158e-05,
      "loss": 0.0055,
      "step": 2711
    },
    {
      "epoch": 2.032983508245877,
      "grad_norm": 0.16142423316077267,
      "learning_rate": 5.692067598394523e-05,
      "loss": 0.0065,
      "step": 2712
    },
    {
      "epoch": 2.0337331334332833,
      "grad_norm": 0.1840512166065806,
      "learning_rate": 5.684196058184972e-05,
      "loss": 0.0058,
      "step": 2713
    },
    {
      "epoch": 2.0344827586206895,
      "grad_norm": 0.09386316725998878,
      "learning_rate": 5.6763278028226986e-05,
      "loss": 0.0041,
      "step": 2714
    },
    {
      "epoch": 2.035232383808096,
      "grad_norm": 0.2704026073642783,
      "learning_rate": 5.66846283829639e-05,
      "loss": 0.0142,
      "step": 2715
    },
    {
      "epoch": 2.0359820089955023,
      "grad_norm": 0.053751714799146394,
      "learning_rate": 5.660601170592236e-05,
      "loss": 0.0027,
      "step": 2716
    },
    {
      "epoch": 2.0367316341829085,
      "grad_norm": 0.1665132193521529,
      "learning_rate": 5.6527428056939113e-05,
      "loss": 0.0058,
      "step": 2717
    },
    {
      "epoch": 2.0374812593703147,
      "grad_norm": 0.0837882288789223,
      "learning_rate": 5.644887749582581e-05,
      "loss": 0.0056,
      "step": 2718
    },
    {
      "epoch": 2.0382308845577213,
      "grad_norm": 0.17344687620216664,
      "learning_rate": 5.6370360082368915e-05,
      "loss": 0.012,
      "step": 2719
    },
    {
      "epoch": 2.0389805097451275,
      "grad_norm": 0.08545712755787346,
      "learning_rate": 5.62918758763296e-05,
      "loss": 0.0033,
      "step": 2720
    },
    {
      "epoch": 2.0397301349325336,
      "grad_norm": 0.15846203084186525,
      "learning_rate": 5.621342493744381e-05,
      "loss": 0.009,
      "step": 2721
    },
    {
      "epoch": 2.04047976011994,
      "grad_norm": 0.13787943275064699,
      "learning_rate": 5.613500732542225e-05,
      "loss": 0.0058,
      "step": 2722
    },
    {
      "epoch": 2.0412293853073464,
      "grad_norm": 0.1739206890447071,
      "learning_rate": 5.605662309995012e-05,
      "loss": 0.0102,
      "step": 2723
    },
    {
      "epoch": 2.0419790104947526,
      "grad_norm": 0.14476344648170317,
      "learning_rate": 5.5978272320687286e-05,
      "loss": 0.0146,
      "step": 2724
    },
    {
      "epoch": 2.042728635682159,
      "grad_norm": 0.08895660251404246,
      "learning_rate": 5.589995504726814e-05,
      "loss": 0.0059,
      "step": 2725
    },
    {
      "epoch": 2.0434782608695654,
      "grad_norm": 0.034946758693803905,
      "learning_rate": 5.582167133930156e-05,
      "loss": 0.0016,
      "step": 2726
    },
    {
      "epoch": 2.0442278860569716,
      "grad_norm": 0.05293482370990212,
      "learning_rate": 5.574342125637089e-05,
      "loss": 0.0027,
      "step": 2727
    },
    {
      "epoch": 2.0449775112443778,
      "grad_norm": 0.1539671338465271,
      "learning_rate": 5.566520485803388e-05,
      "loss": 0.021,
      "step": 2728
    },
    {
      "epoch": 2.045727136431784,
      "grad_norm": 0.259356096848114,
      "learning_rate": 5.558702220382266e-05,
      "loss": 0.0131,
      "step": 2729
    },
    {
      "epoch": 2.0464767616191906,
      "grad_norm": 0.047567614571502104,
      "learning_rate": 5.550887335324363e-05,
      "loss": 0.0026,
      "step": 2730
    },
    {
      "epoch": 2.0472263868065967,
      "grad_norm": 0.14037541136100798,
      "learning_rate": 5.5430758365777514e-05,
      "loss": 0.0079,
      "step": 2731
    },
    {
      "epoch": 2.047976011994003,
      "grad_norm": 0.11024060952702228,
      "learning_rate": 5.535267730087922e-05,
      "loss": 0.0079,
      "step": 2732
    },
    {
      "epoch": 2.048725637181409,
      "grad_norm": 0.1868952649017366,
      "learning_rate": 5.527463021797786e-05,
      "loss": 0.0151,
      "step": 2733
    },
    {
      "epoch": 2.0494752623688157,
      "grad_norm": 0.0685871767217177,
      "learning_rate": 5.519661717647664e-05,
      "loss": 0.004,
      "step": 2734
    },
    {
      "epoch": 2.050224887556222,
      "grad_norm": 0.36893532451261046,
      "learning_rate": 5.511863823575298e-05,
      "loss": 0.03,
      "step": 2735
    },
    {
      "epoch": 2.050974512743628,
      "grad_norm": 0.1024390603249343,
      "learning_rate": 5.5040693455158206e-05,
      "loss": 0.0064,
      "step": 2736
    },
    {
      "epoch": 2.0517241379310347,
      "grad_norm": 0.06003674834006714,
      "learning_rate": 5.496278289401771e-05,
      "loss": 0.0048,
      "step": 2737
    },
    {
      "epoch": 2.052473763118441,
      "grad_norm": 0.0676091350613358,
      "learning_rate": 5.488490661163079e-05,
      "loss": 0.0039,
      "step": 2738
    },
    {
      "epoch": 2.053223388305847,
      "grad_norm": 0.06811684148720669,
      "learning_rate": 5.480706466727074e-05,
      "loss": 0.004,
      "step": 2739
    },
    {
      "epoch": 2.0539730134932532,
      "grad_norm": 0.22457808229890616,
      "learning_rate": 5.472925712018462e-05,
      "loss": 0.023,
      "step": 2740
    },
    {
      "epoch": 2.05472263868066,
      "grad_norm": 0.14011710230532487,
      "learning_rate": 5.465148402959339e-05,
      "loss": 0.0165,
      "step": 2741
    },
    {
      "epoch": 2.055472263868066,
      "grad_norm": 0.09396544964614165,
      "learning_rate": 5.457374545469174e-05,
      "loss": 0.0068,
      "step": 2742
    },
    {
      "epoch": 2.056221889055472,
      "grad_norm": 0.20625852590985544,
      "learning_rate": 5.4496041454648086e-05,
      "loss": 0.0085,
      "step": 2743
    },
    {
      "epoch": 2.0569715142428784,
      "grad_norm": 0.20411992021066738,
      "learning_rate": 5.441837208860456e-05,
      "loss": 0.0088,
      "step": 2744
    },
    {
      "epoch": 2.057721139430285,
      "grad_norm": 0.051337110665052886,
      "learning_rate": 5.4340737415676866e-05,
      "loss": 0.0042,
      "step": 2745
    },
    {
      "epoch": 2.058470764617691,
      "grad_norm": 0.24525192857059822,
      "learning_rate": 5.426313749495447e-05,
      "loss": 0.0079,
      "step": 2746
    },
    {
      "epoch": 2.0592203898050974,
      "grad_norm": 0.15175784875891044,
      "learning_rate": 5.418557238550016e-05,
      "loss": 0.0051,
      "step": 2747
    },
    {
      "epoch": 2.0599700149925035,
      "grad_norm": 0.04180425435431888,
      "learning_rate": 5.410804214635039e-05,
      "loss": 0.0022,
      "step": 2748
    },
    {
      "epoch": 2.06071964017991,
      "grad_norm": 0.15208975614765013,
      "learning_rate": 5.403054683651502e-05,
      "loss": 0.0077,
      "step": 2749
    },
    {
      "epoch": 2.0614692653673163,
      "grad_norm": 0.0745353317998331,
      "learning_rate": 5.39530865149773e-05,
      "loss": 0.0045,
      "step": 2750
    },
    {
      "epoch": 2.0622188905547225,
      "grad_norm": 0.0857172018999341,
      "learning_rate": 5.387566124069391e-05,
      "loss": 0.005,
      "step": 2751
    },
    {
      "epoch": 2.062968515742129,
      "grad_norm": 0.07316934882364763,
      "learning_rate": 5.3798271072594786e-05,
      "loss": 0.0047,
      "step": 2752
    },
    {
      "epoch": 2.0637181409295353,
      "grad_norm": 0.141839675930256,
      "learning_rate": 5.372091606958321e-05,
      "loss": 0.0148,
      "step": 2753
    },
    {
      "epoch": 2.0644677661169415,
      "grad_norm": 0.1318488328244515,
      "learning_rate": 5.364359629053566e-05,
      "loss": 0.0162,
      "step": 2754
    },
    {
      "epoch": 2.0652173913043477,
      "grad_norm": 0.2125988213178984,
      "learning_rate": 5.3566311794301784e-05,
      "loss": 0.0229,
      "step": 2755
    },
    {
      "epoch": 2.0659670164917543,
      "grad_norm": 0.11935852353236064,
      "learning_rate": 5.3489062639704434e-05,
      "loss": 0.0033,
      "step": 2756
    },
    {
      "epoch": 2.0667166416791605,
      "grad_norm": 0.045018097123316676,
      "learning_rate": 5.341184888553954e-05,
      "loss": 0.002,
      "step": 2757
    },
    {
      "epoch": 2.0674662668665666,
      "grad_norm": 0.17893936598833707,
      "learning_rate": 5.333467059057602e-05,
      "loss": 0.0158,
      "step": 2758
    },
    {
      "epoch": 2.068215892053973,
      "grad_norm": 0.1267157060412366,
      "learning_rate": 5.325752781355593e-05,
      "loss": 0.0067,
      "step": 2759
    },
    {
      "epoch": 2.0689655172413794,
      "grad_norm": 0.11327073400568807,
      "learning_rate": 5.3180420613194226e-05,
      "loss": 0.0086,
      "step": 2760
    },
    {
      "epoch": 2.0697151424287856,
      "grad_norm": 0.026122915286548982,
      "learning_rate": 5.310334904817878e-05,
      "loss": 0.002,
      "step": 2761
    },
    {
      "epoch": 2.070464767616192,
      "grad_norm": 0.0624022067515344,
      "learning_rate": 5.3026313177170303e-05,
      "loss": 0.0019,
      "step": 2762
    },
    {
      "epoch": 2.071214392803598,
      "grad_norm": 0.07979267478998299,
      "learning_rate": 5.294931305880243e-05,
      "loss": 0.0075,
      "step": 2763
    },
    {
      "epoch": 2.0719640179910046,
      "grad_norm": 0.34235288461531843,
      "learning_rate": 5.287234875168152e-05,
      "loss": 0.0096,
      "step": 2764
    },
    {
      "epoch": 2.072713643178411,
      "grad_norm": 0.053751697993312225,
      "learning_rate": 5.27954203143867e-05,
      "loss": 0.0036,
      "step": 2765
    },
    {
      "epoch": 2.073463268365817,
      "grad_norm": 0.09879108309949992,
      "learning_rate": 5.271852780546977e-05,
      "loss": 0.006,
      "step": 2766
    },
    {
      "epoch": 2.0742128935532236,
      "grad_norm": 0.11817567253208243,
      "learning_rate": 5.264167128345523e-05,
      "loss": 0.0175,
      "step": 2767
    },
    {
      "epoch": 2.0749625187406298,
      "grad_norm": 0.07336750448769226,
      "learning_rate": 5.256485080684011e-05,
      "loss": 0.005,
      "step": 2768
    },
    {
      "epoch": 2.075712143928036,
      "grad_norm": 0.07420326577156641,
      "learning_rate": 5.2488066434094116e-05,
      "loss": 0.0062,
      "step": 2769
    },
    {
      "epoch": 2.076461769115442,
      "grad_norm": 0.11458244964011671,
      "learning_rate": 5.241131822365941e-05,
      "loss": 0.0123,
      "step": 2770
    },
    {
      "epoch": 2.0772113943028487,
      "grad_norm": 0.08138043870500519,
      "learning_rate": 5.233460623395061e-05,
      "loss": 0.0051,
      "step": 2771
    },
    {
      "epoch": 2.077961019490255,
      "grad_norm": 0.06648218497103654,
      "learning_rate": 5.225793052335479e-05,
      "loss": 0.0056,
      "step": 2772
    },
    {
      "epoch": 2.078710644677661,
      "grad_norm": 0.10901113040322437,
      "learning_rate": 5.218129115023145e-05,
      "loss": 0.0049,
      "step": 2773
    },
    {
      "epoch": 2.0794602698650673,
      "grad_norm": 0.15590821022125223,
      "learning_rate": 5.210468817291234e-05,
      "loss": 0.0135,
      "step": 2774
    },
    {
      "epoch": 2.080209895052474,
      "grad_norm": 0.15411055638884816,
      "learning_rate": 5.2028121649701614e-05,
      "loss": 0.0066,
      "step": 2775
    },
    {
      "epoch": 2.08095952023988,
      "grad_norm": 0.1844008895299288,
      "learning_rate": 5.1951591638875605e-05,
      "loss": 0.0211,
      "step": 2776
    },
    {
      "epoch": 2.0817091454272862,
      "grad_norm": 0.17972940945558835,
      "learning_rate": 5.187509819868288e-05,
      "loss": 0.0118,
      "step": 2777
    },
    {
      "epoch": 2.082458770614693,
      "grad_norm": 0.1818504662636259,
      "learning_rate": 5.1798641387344184e-05,
      "loss": 0.013,
      "step": 2778
    },
    {
      "epoch": 2.083208395802099,
      "grad_norm": 0.13366106028225347,
      "learning_rate": 5.172222126305235e-05,
      "loss": 0.0139,
      "step": 2779
    },
    {
      "epoch": 2.0839580209895052,
      "grad_norm": 0.047564458389697646,
      "learning_rate": 5.164583788397234e-05,
      "loss": 0.0023,
      "step": 2780
    },
    {
      "epoch": 2.0847076461769114,
      "grad_norm": 0.08583461517386021,
      "learning_rate": 5.1569491308241094e-05,
      "loss": 0.0077,
      "step": 2781
    },
    {
      "epoch": 2.085457271364318,
      "grad_norm": 0.08232377535332387,
      "learning_rate": 5.149318159396752e-05,
      "loss": 0.0052,
      "step": 2782
    },
    {
      "epoch": 2.086206896551724,
      "grad_norm": 0.12920830793795776,
      "learning_rate": 5.14169087992326e-05,
      "loss": 0.0116,
      "step": 2783
    },
    {
      "epoch": 2.0869565217391304,
      "grad_norm": 0.28023458386351713,
      "learning_rate": 5.134067298208909e-05,
      "loss": 0.0283,
      "step": 2784
    },
    {
      "epoch": 2.0877061469265366,
      "grad_norm": 0.11795585858524446,
      "learning_rate": 5.126447420056165e-05,
      "loss": 0.0048,
      "step": 2785
    },
    {
      "epoch": 2.088455772113943,
      "grad_norm": 0.19569779614562738,
      "learning_rate": 5.118831251264673e-05,
      "loss": 0.0164,
      "step": 2786
    },
    {
      "epoch": 2.0892053973013494,
      "grad_norm": 0.1391438714541468,
      "learning_rate": 5.111218797631256e-05,
      "loss": 0.0037,
      "step": 2787
    },
    {
      "epoch": 2.0899550224887555,
      "grad_norm": 0.12083354676952428,
      "learning_rate": 5.103610064949911e-05,
      "loss": 0.0238,
      "step": 2788
    },
    {
      "epoch": 2.0907046476761617,
      "grad_norm": 0.11756022743007565,
      "learning_rate": 5.096005059011799e-05,
      "loss": 0.0073,
      "step": 2789
    },
    {
      "epoch": 2.0914542728635683,
      "grad_norm": 0.05654885803442007,
      "learning_rate": 5.088403785605247e-05,
      "loss": 0.0028,
      "step": 2790
    },
    {
      "epoch": 2.0922038980509745,
      "grad_norm": 0.07698581805694782,
      "learning_rate": 5.080806250515737e-05,
      "loss": 0.007,
      "step": 2791
    },
    {
      "epoch": 2.0929535232383807,
      "grad_norm": 0.2584924571449496,
      "learning_rate": 5.073212459525918e-05,
      "loss": 0.019,
      "step": 2792
    },
    {
      "epoch": 2.0937031484257873,
      "grad_norm": 0.19263855328707216,
      "learning_rate": 5.065622418415577e-05,
      "loss": 0.0177,
      "step": 2793
    },
    {
      "epoch": 2.0944527736131935,
      "grad_norm": 0.19491448696329708,
      "learning_rate": 5.058036132961649e-05,
      "loss": 0.0228,
      "step": 2794
    },
    {
      "epoch": 2.0952023988005997,
      "grad_norm": 0.07619742031197467,
      "learning_rate": 5.0504536089382124e-05,
      "loss": 0.0056,
      "step": 2795
    },
    {
      "epoch": 2.095952023988006,
      "grad_norm": 0.22193470194375173,
      "learning_rate": 5.042874852116485e-05,
      "loss": 0.0124,
      "step": 2796
    },
    {
      "epoch": 2.0967016491754125,
      "grad_norm": 0.14331149167371612,
      "learning_rate": 5.035299868264811e-05,
      "loss": 0.0065,
      "step": 2797
    },
    {
      "epoch": 2.0974512743628186,
      "grad_norm": 0.172637392773366,
      "learning_rate": 5.02772866314867e-05,
      "loss": 0.0142,
      "step": 2798
    },
    {
      "epoch": 2.098200899550225,
      "grad_norm": 0.12579988048753785,
      "learning_rate": 5.0201612425306597e-05,
      "loss": 0.0058,
      "step": 2799
    },
    {
      "epoch": 2.098950524737631,
      "grad_norm": 0.10993273901263476,
      "learning_rate": 5.012597612170502e-05,
      "loss": 0.0023,
      "step": 2800
    },
    {
      "epoch": 2.0997001499250376,
      "grad_norm": 0.065006704120964,
      "learning_rate": 5.005037777825029e-05,
      "loss": 0.0054,
      "step": 2801
    },
    {
      "epoch": 2.100449775112444,
      "grad_norm": 0.08488518929085496,
      "learning_rate": 4.9974817452481885e-05,
      "loss": 0.0094,
      "step": 2802
    },
    {
      "epoch": 2.10119940029985,
      "grad_norm": 0.06996280682513732,
      "learning_rate": 4.989929520191031e-05,
      "loss": 0.0044,
      "step": 2803
    },
    {
      "epoch": 2.101949025487256,
      "grad_norm": 0.10614255374552094,
      "learning_rate": 4.982381108401712e-05,
      "loss": 0.0051,
      "step": 2804
    },
    {
      "epoch": 2.1026986506746628,
      "grad_norm": 0.08931334364433949,
      "learning_rate": 4.974836515625484e-05,
      "loss": 0.0097,
      "step": 2805
    },
    {
      "epoch": 2.103448275862069,
      "grad_norm": 0.09602657010448278,
      "learning_rate": 4.967295747604685e-05,
      "loss": 0.0045,
      "step": 2806
    },
    {
      "epoch": 2.104197901049475,
      "grad_norm": 0.07385611806369675,
      "learning_rate": 4.9597588100787585e-05,
      "loss": 0.0031,
      "step": 2807
    },
    {
      "epoch": 2.1049475262368817,
      "grad_norm": 0.03093310014099065,
      "learning_rate": 4.9522257087842196e-05,
      "loss": 0.0011,
      "step": 2808
    },
    {
      "epoch": 2.105697151424288,
      "grad_norm": 0.11316962936701583,
      "learning_rate": 4.9446964494546655e-05,
      "loss": 0.0081,
      "step": 2809
    },
    {
      "epoch": 2.106446776611694,
      "grad_norm": 0.05960739161768324,
      "learning_rate": 4.93717103782077e-05,
      "loss": 0.002,
      "step": 2810
    },
    {
      "epoch": 2.1071964017991003,
      "grad_norm": 0.13704128945073163,
      "learning_rate": 4.929649479610282e-05,
      "loss": 0.0098,
      "step": 2811
    },
    {
      "epoch": 2.107946026986507,
      "grad_norm": 0.17973157057976616,
      "learning_rate": 4.9221317805480115e-05,
      "loss": 0.0141,
      "step": 2812
    },
    {
      "epoch": 2.108695652173913,
      "grad_norm": 0.05729033581567592,
      "learning_rate": 4.914617946355835e-05,
      "loss": 0.0036,
      "step": 2813
    },
    {
      "epoch": 2.1094452773613193,
      "grad_norm": 0.04930374438074895,
      "learning_rate": 4.907107982752684e-05,
      "loss": 0.0021,
      "step": 2814
    },
    {
      "epoch": 2.1101949025487254,
      "grad_norm": 0.10265639378615501,
      "learning_rate": 4.899601895454551e-05,
      "loss": 0.0116,
      "step": 2815
    },
    {
      "epoch": 2.110944527736132,
      "grad_norm": 0.17308963975557312,
      "learning_rate": 4.892099690174472e-05,
      "loss": 0.0151,
      "step": 2816
    },
    {
      "epoch": 2.1116941529235382,
      "grad_norm": 0.14137961791206421,
      "learning_rate": 4.8846013726225295e-05,
      "loss": 0.0126,
      "step": 2817
    },
    {
      "epoch": 2.1124437781109444,
      "grad_norm": 0.19629272020786975,
      "learning_rate": 4.8771069485058486e-05,
      "loss": 0.0043,
      "step": 2818
    },
    {
      "epoch": 2.113193403298351,
      "grad_norm": 0.1248330816683146,
      "learning_rate": 4.869616423528588e-05,
      "loss": 0.0061,
      "step": 2819
    },
    {
      "epoch": 2.113943028485757,
      "grad_norm": 0.07729755271577264,
      "learning_rate": 4.86212980339194e-05,
      "loss": 0.0042,
      "step": 2820
    },
    {
      "epoch": 2.1146926536731634,
      "grad_norm": 0.1215731772373695,
      "learning_rate": 4.854647093794129e-05,
      "loss": 0.016,
      "step": 2821
    },
    {
      "epoch": 2.1154422788605696,
      "grad_norm": 0.3168985126036469,
      "learning_rate": 4.847168300430397e-05,
      "loss": 0.0081,
      "step": 2822
    },
    {
      "epoch": 2.116191904047976,
      "grad_norm": 0.18347302630220838,
      "learning_rate": 4.8396934289930064e-05,
      "loss": 0.0163,
      "step": 2823
    },
    {
      "epoch": 2.1169415292353824,
      "grad_norm": 0.09802171455989667,
      "learning_rate": 4.8322224851712386e-05,
      "loss": 0.0037,
      "step": 2824
    },
    {
      "epoch": 2.1176911544227885,
      "grad_norm": 0.11466768689466496,
      "learning_rate": 4.82475547465138e-05,
      "loss": 0.0107,
      "step": 2825
    },
    {
      "epoch": 2.1184407796101947,
      "grad_norm": 0.15766024712357615,
      "learning_rate": 4.817292403116729e-05,
      "loss": 0.0073,
      "step": 2826
    },
    {
      "epoch": 2.1191904047976013,
      "grad_norm": 0.08553525297480681,
      "learning_rate": 4.8098332762475804e-05,
      "loss": 0.0057,
      "step": 2827
    },
    {
      "epoch": 2.1199400299850075,
      "grad_norm": 0.08917342986018453,
      "learning_rate": 4.802378099721232e-05,
      "loss": 0.0054,
      "step": 2828
    },
    {
      "epoch": 2.1206896551724137,
      "grad_norm": 0.017895313173530512,
      "learning_rate": 4.79492687921197e-05,
      "loss": 0.0005,
      "step": 2829
    },
    {
      "epoch": 2.12143928035982,
      "grad_norm": 0.15167899697433201,
      "learning_rate": 4.787479620391068e-05,
      "loss": 0.0068,
      "step": 2830
    },
    {
      "epoch": 2.1221889055472265,
      "grad_norm": 0.14085737808607107,
      "learning_rate": 4.780036328926797e-05,
      "loss": 0.0081,
      "step": 2831
    },
    {
      "epoch": 2.1229385307346327,
      "grad_norm": 0.2396864168220026,
      "learning_rate": 4.772597010484395e-05,
      "loss": 0.0185,
      "step": 2832
    },
    {
      "epoch": 2.123688155922039,
      "grad_norm": 0.08151691227139982,
      "learning_rate": 4.7651616707260815e-05,
      "loss": 0.005,
      "step": 2833
    },
    {
      "epoch": 2.1244377811094455,
      "grad_norm": 0.07783032987855823,
      "learning_rate": 4.757730315311045e-05,
      "loss": 0.0023,
      "step": 2834
    },
    {
      "epoch": 2.1251874062968517,
      "grad_norm": 0.07764939687149099,
      "learning_rate": 4.750302949895444e-05,
      "loss": 0.0033,
      "step": 2835
    },
    {
      "epoch": 2.125937031484258,
      "grad_norm": 0.12788204639640396,
      "learning_rate": 4.7428795801324e-05,
      "loss": 0.0073,
      "step": 2836
    },
    {
      "epoch": 2.126686656671664,
      "grad_norm": 0.14298556685133668,
      "learning_rate": 4.73546021167199e-05,
      "loss": 0.0091,
      "step": 2837
    },
    {
      "epoch": 2.1274362818590706,
      "grad_norm": 0.2196406994123255,
      "learning_rate": 4.7280448501612485e-05,
      "loss": 0.0046,
      "step": 2838
    },
    {
      "epoch": 2.128185907046477,
      "grad_norm": 0.07390369972391603,
      "learning_rate": 4.720633501244165e-05,
      "loss": 0.0024,
      "step": 2839
    },
    {
      "epoch": 2.128935532233883,
      "grad_norm": 0.21273204391802142,
      "learning_rate": 4.713226170561665e-05,
      "loss": 0.0181,
      "step": 2840
    },
    {
      "epoch": 2.129685157421289,
      "grad_norm": 0.07017065388286854,
      "learning_rate": 4.705822863751622e-05,
      "loss": 0.0032,
      "step": 2841
    },
    {
      "epoch": 2.130434782608696,
      "grad_norm": 0.08423142817528334,
      "learning_rate": 4.698423586448846e-05,
      "loss": 0.0039,
      "step": 2842
    },
    {
      "epoch": 2.131184407796102,
      "grad_norm": 0.07577123486070463,
      "learning_rate": 4.691028344285078e-05,
      "loss": 0.0062,
      "step": 2843
    },
    {
      "epoch": 2.131934032983508,
      "grad_norm": 0.10405964473337807,
      "learning_rate": 4.68363714288899e-05,
      "loss": 0.0057,
      "step": 2844
    },
    {
      "epoch": 2.1326836581709143,
      "grad_norm": 0.06725575601961063,
      "learning_rate": 4.6762499878861764e-05,
      "loss": 0.002,
      "step": 2845
    },
    {
      "epoch": 2.133433283358321,
      "grad_norm": 0.11374990049739897,
      "learning_rate": 4.668866884899154e-05,
      "loss": 0.0121,
      "step": 2846
    },
    {
      "epoch": 2.134182908545727,
      "grad_norm": 0.028375524722939163,
      "learning_rate": 4.661487839547355e-05,
      "loss": 0.0022,
      "step": 2847
    },
    {
      "epoch": 2.1349325337331333,
      "grad_norm": 0.1701170626738846,
      "learning_rate": 4.654112857447121e-05,
      "loss": 0.0086,
      "step": 2848
    },
    {
      "epoch": 2.13568215892054,
      "grad_norm": 0.04823068839835657,
      "learning_rate": 4.6467419442117046e-05,
      "loss": 0.0023,
      "step": 2849
    },
    {
      "epoch": 2.136431784107946,
      "grad_norm": 0.24689421196932598,
      "learning_rate": 4.639375105451259e-05,
      "loss": 0.0219,
      "step": 2850
    },
    {
      "epoch": 2.1371814092953523,
      "grad_norm": 0.054759916362402856,
      "learning_rate": 4.632012346772837e-05,
      "loss": 0.0017,
      "step": 2851
    },
    {
      "epoch": 2.1379310344827585,
      "grad_norm": 0.043825477313645465,
      "learning_rate": 4.624653673780386e-05,
      "loss": 0.0016,
      "step": 2852
    },
    {
      "epoch": 2.138680659670165,
      "grad_norm": 0.10507495229072013,
      "learning_rate": 4.617299092074744e-05,
      "loss": 0.0077,
      "step": 2853
    },
    {
      "epoch": 2.1394302848575713,
      "grad_norm": 0.14357076370983768,
      "learning_rate": 4.6099486072536334e-05,
      "loss": 0.0073,
      "step": 2854
    },
    {
      "epoch": 2.1401799100449774,
      "grad_norm": 0.1592531766218126,
      "learning_rate": 4.602602224911656e-05,
      "loss": 0.0254,
      "step": 2855
    },
    {
      "epoch": 2.1409295352323836,
      "grad_norm": 0.02049681399116887,
      "learning_rate": 4.595259950640304e-05,
      "loss": 0.0012,
      "step": 2856
    },
    {
      "epoch": 2.1416791604197902,
      "grad_norm": 0.07794412146800708,
      "learning_rate": 4.587921790027927e-05,
      "loss": 0.006,
      "step": 2857
    },
    {
      "epoch": 2.1424287856071964,
      "grad_norm": 0.09195761790497897,
      "learning_rate": 4.580587748659752e-05,
      "loss": 0.0111,
      "step": 2858
    },
    {
      "epoch": 2.1431784107946026,
      "grad_norm": 0.16849596417897122,
      "learning_rate": 4.57325783211787e-05,
      "loss": 0.0044,
      "step": 2859
    },
    {
      "epoch": 2.143928035982009,
      "grad_norm": 0.054270439033802784,
      "learning_rate": 4.5659320459812296e-05,
      "loss": 0.0026,
      "step": 2860
    },
    {
      "epoch": 2.1446776611694154,
      "grad_norm": 0.13695792493236258,
      "learning_rate": 4.5586103958256344e-05,
      "loss": 0.0118,
      "step": 2861
    },
    {
      "epoch": 2.1454272863568216,
      "grad_norm": 0.058343514676726244,
      "learning_rate": 4.55129288722375e-05,
      "loss": 0.0027,
      "step": 2862
    },
    {
      "epoch": 2.1461769115442277,
      "grad_norm": 0.06574267821808014,
      "learning_rate": 4.5439795257450804e-05,
      "loss": 0.0039,
      "step": 2863
    },
    {
      "epoch": 2.1469265367316344,
      "grad_norm": 0.10332596227148909,
      "learning_rate": 4.536670316955974e-05,
      "loss": 0.0033,
      "step": 2864
    },
    {
      "epoch": 2.1476761619190405,
      "grad_norm": 0.16550320030944354,
      "learning_rate": 4.5293652664196215e-05,
      "loss": 0.0083,
      "step": 2865
    },
    {
      "epoch": 2.1484257871064467,
      "grad_norm": 0.1822057557002658,
      "learning_rate": 4.522064379696048e-05,
      "loss": 0.0071,
      "step": 2866
    },
    {
      "epoch": 2.149175412293853,
      "grad_norm": 0.17039104462095303,
      "learning_rate": 4.5147676623421076e-05,
      "loss": 0.0081,
      "step": 2867
    },
    {
      "epoch": 2.1499250374812595,
      "grad_norm": 0.1184742262621423,
      "learning_rate": 4.507475119911482e-05,
      "loss": 0.0045,
      "step": 2868
    },
    {
      "epoch": 2.1506746626686657,
      "grad_norm": 0.06221806454687913,
      "learning_rate": 4.500186757954675e-05,
      "loss": 0.0021,
      "step": 2869
    },
    {
      "epoch": 2.151424287856072,
      "grad_norm": 0.08515968083332011,
      "learning_rate": 4.4929025820190105e-05,
      "loss": 0.0063,
      "step": 2870
    },
    {
      "epoch": 2.1521739130434785,
      "grad_norm": 0.3366899734636794,
      "learning_rate": 4.485622597648624e-05,
      "loss": 0.0339,
      "step": 2871
    },
    {
      "epoch": 2.1529235382308847,
      "grad_norm": 0.2936218445138808,
      "learning_rate": 4.4783468103844615e-05,
      "loss": 0.0214,
      "step": 2872
    },
    {
      "epoch": 2.153673163418291,
      "grad_norm": 0.026388492126442124,
      "learning_rate": 4.471075225764275e-05,
      "loss": 0.0013,
      "step": 2873
    },
    {
      "epoch": 2.154422788605697,
      "grad_norm": 0.12018222361567683,
      "learning_rate": 4.4638078493226176e-05,
      "loss": 0.0086,
      "step": 2874
    },
    {
      "epoch": 2.1551724137931036,
      "grad_norm": 0.08220028929729363,
      "learning_rate": 4.4565446865908376e-05,
      "loss": 0.0053,
      "step": 2875
    },
    {
      "epoch": 2.15592203898051,
      "grad_norm": 0.08399578333992191,
      "learning_rate": 4.44928574309708e-05,
      "loss": 0.0059,
      "step": 2876
    },
    {
      "epoch": 2.156671664167916,
      "grad_norm": 0.1404142689687109,
      "learning_rate": 4.442031024366274e-05,
      "loss": 0.0115,
      "step": 2877
    },
    {
      "epoch": 2.157421289355322,
      "grad_norm": 0.054640401967214404,
      "learning_rate": 4.434780535920138e-05,
      "loss": 0.0026,
      "step": 2878
    },
    {
      "epoch": 2.158170914542729,
      "grad_norm": 0.10806349724783954,
      "learning_rate": 4.427534283277162e-05,
      "loss": 0.0052,
      "step": 2879
    },
    {
      "epoch": 2.158920539730135,
      "grad_norm": 0.043779809246776935,
      "learning_rate": 4.420292271952626e-05,
      "loss": 0.0033,
      "step": 2880
    },
    {
      "epoch": 2.159670164917541,
      "grad_norm": 0.07964633918529888,
      "learning_rate": 4.413054507458571e-05,
      "loss": 0.0047,
      "step": 2881
    },
    {
      "epoch": 2.1604197901049473,
      "grad_norm": 0.18538411473729013,
      "learning_rate": 4.4058209953038086e-05,
      "loss": 0.0158,
      "step": 2882
    },
    {
      "epoch": 2.161169415292354,
      "grad_norm": 0.4188366014235997,
      "learning_rate": 4.398591740993914e-05,
      "loss": 0.0078,
      "step": 2883
    },
    {
      "epoch": 2.16191904047976,
      "grad_norm": 0.10735652365189656,
      "learning_rate": 4.391366750031217e-05,
      "loss": 0.006,
      "step": 2884
    },
    {
      "epoch": 2.1626686656671663,
      "grad_norm": 0.07968789225277143,
      "learning_rate": 4.384146027914815e-05,
      "loss": 0.0049,
      "step": 2885
    },
    {
      "epoch": 2.1634182908545725,
      "grad_norm": 0.04463803516349911,
      "learning_rate": 4.376929580140544e-05,
      "loss": 0.0024,
      "step": 2886
    },
    {
      "epoch": 2.164167916041979,
      "grad_norm": 0.07875465351103227,
      "learning_rate": 4.3697174122009934e-05,
      "loss": 0.0038,
      "step": 2887
    },
    {
      "epoch": 2.1649175412293853,
      "grad_norm": 0.06776221559374107,
      "learning_rate": 4.362509529585489e-05,
      "loss": 0.003,
      "step": 2888
    },
    {
      "epoch": 2.1656671664167915,
      "grad_norm": 0.08488959892000786,
      "learning_rate": 4.355305937780102e-05,
      "loss": 0.0052,
      "step": 2889
    },
    {
      "epoch": 2.166416791604198,
      "grad_norm": 0.08370426819698125,
      "learning_rate": 4.348106642267633e-05,
      "loss": 0.0037,
      "step": 2890
    },
    {
      "epoch": 2.1671664167916043,
      "grad_norm": 0.07325818139028772,
      "learning_rate": 4.3409116485276134e-05,
      "loss": 0.0062,
      "step": 2891
    },
    {
      "epoch": 2.1679160419790104,
      "grad_norm": 0.12924771341879,
      "learning_rate": 4.333720962036304e-05,
      "loss": 0.0072,
      "step": 2892
    },
    {
      "epoch": 2.1686656671664166,
      "grad_norm": 0.12552100078424228,
      "learning_rate": 4.326534588266681e-05,
      "loss": 0.0102,
      "step": 2893
    },
    {
      "epoch": 2.1694152923538232,
      "grad_norm": 0.13763354141825093,
      "learning_rate": 4.3193525326884435e-05,
      "loss": 0.0066,
      "step": 2894
    },
    {
      "epoch": 2.1701649175412294,
      "grad_norm": 0.05131955834253436,
      "learning_rate": 4.312174800768001e-05,
      "loss": 0.002,
      "step": 2895
    },
    {
      "epoch": 2.1709145427286356,
      "grad_norm": 0.18141305745303182,
      "learning_rate": 4.305001397968474e-05,
      "loss": 0.0045,
      "step": 2896
    },
    {
      "epoch": 2.1716641679160418,
      "grad_norm": 0.09518754190642088,
      "learning_rate": 4.297832329749687e-05,
      "loss": 0.0078,
      "step": 2897
    },
    {
      "epoch": 2.1724137931034484,
      "grad_norm": 0.1094915327192126,
      "learning_rate": 4.2906676015681644e-05,
      "loss": 0.0062,
      "step": 2898
    },
    {
      "epoch": 2.1731634182908546,
      "grad_norm": 0.14326768073583626,
      "learning_rate": 4.283507218877128e-05,
      "loss": 0.0095,
      "step": 2899
    },
    {
      "epoch": 2.1739130434782608,
      "grad_norm": 0.12477399089755312,
      "learning_rate": 4.276351187126496e-05,
      "loss": 0.0089,
      "step": 2900
    },
    {
      "epoch": 2.1746626686656674,
      "grad_norm": 0.11015881276147756,
      "learning_rate": 4.269199511762868e-05,
      "loss": 0.0036,
      "step": 2901
    },
    {
      "epoch": 2.1754122938530736,
      "grad_norm": 0.14941368767564808,
      "learning_rate": 4.2620521982295334e-05,
      "loss": 0.0118,
      "step": 2902
    },
    {
      "epoch": 2.1761619190404797,
      "grad_norm": 0.09648693740069292,
      "learning_rate": 4.2549092519664546e-05,
      "loss": 0.0041,
      "step": 2903
    },
    {
      "epoch": 2.176911544227886,
      "grad_norm": 0.11454174987068287,
      "learning_rate": 4.247770678410284e-05,
      "loss": 0.0043,
      "step": 2904
    },
    {
      "epoch": 2.1776611694152925,
      "grad_norm": 0.25554248116246986,
      "learning_rate": 4.240636482994332e-05,
      "loss": 0.0111,
      "step": 2905
    },
    {
      "epoch": 2.1784107946026987,
      "grad_norm": 0.20926240916107086,
      "learning_rate": 4.233506671148583e-05,
      "loss": 0.0059,
      "step": 2906
    },
    {
      "epoch": 2.179160419790105,
      "grad_norm": 0.04413399814127862,
      "learning_rate": 4.2263812482996814e-05,
      "loss": 0.0028,
      "step": 2907
    },
    {
      "epoch": 2.179910044977511,
      "grad_norm": 0.09739179400173857,
      "learning_rate": 4.219260219870932e-05,
      "loss": 0.0095,
      "step": 2908
    },
    {
      "epoch": 2.1806596701649177,
      "grad_norm": 0.1956946447085361,
      "learning_rate": 4.212143591282303e-05,
      "loss": 0.0194,
      "step": 2909
    },
    {
      "epoch": 2.181409295352324,
      "grad_norm": 0.09475992658654644,
      "learning_rate": 4.2050313679504015e-05,
      "loss": 0.0054,
      "step": 2910
    },
    {
      "epoch": 2.18215892053973,
      "grad_norm": 0.03654156738299712,
      "learning_rate": 4.19792355528849e-05,
      "loss": 0.0025,
      "step": 2911
    },
    {
      "epoch": 2.1829085457271367,
      "grad_norm": 0.13210150734611542,
      "learning_rate": 4.190820158706469e-05,
      "loss": 0.0047,
      "step": 2912
    },
    {
      "epoch": 2.183658170914543,
      "grad_norm": 0.1602288514619366,
      "learning_rate": 4.1837211836108814e-05,
      "loss": 0.0104,
      "step": 2913
    },
    {
      "epoch": 2.184407796101949,
      "grad_norm": 0.10919867016817089,
      "learning_rate": 4.176626635404902e-05,
      "loss": 0.0285,
      "step": 2914
    },
    {
      "epoch": 2.185157421289355,
      "grad_norm": 0.09016451941471065,
      "learning_rate": 4.169536519488339e-05,
      "loss": 0.0097,
      "step": 2915
    },
    {
      "epoch": 2.185907046476762,
      "grad_norm": 0.20616458903558743,
      "learning_rate": 4.162450841257623e-05,
      "loss": 0.0049,
      "step": 2916
    },
    {
      "epoch": 2.186656671664168,
      "grad_norm": 0.12993953040956022,
      "learning_rate": 4.1553696061058144e-05,
      "loss": 0.0074,
      "step": 2917
    },
    {
      "epoch": 2.187406296851574,
      "grad_norm": 0.06387145155896923,
      "learning_rate": 4.1482928194225826e-05,
      "loss": 0.0034,
      "step": 2918
    },
    {
      "epoch": 2.1881559220389803,
      "grad_norm": 0.1412799170073803,
      "learning_rate": 4.141220486594217e-05,
      "loss": 0.0109,
      "step": 2919
    },
    {
      "epoch": 2.188905547226387,
      "grad_norm": 0.08496139202783914,
      "learning_rate": 4.134152613003618e-05,
      "loss": 0.0063,
      "step": 2920
    },
    {
      "epoch": 2.189655172413793,
      "grad_norm": 0.07256124080205822,
      "learning_rate": 4.127089204030287e-05,
      "loss": 0.0061,
      "step": 2921
    },
    {
      "epoch": 2.1904047976011993,
      "grad_norm": 0.05522989278346765,
      "learning_rate": 4.120030265050331e-05,
      "loss": 0.0037,
      "step": 2922
    },
    {
      "epoch": 2.1911544227886055,
      "grad_norm": 0.1584495856412796,
      "learning_rate": 4.112975801436454e-05,
      "loss": 0.0083,
      "step": 2923
    },
    {
      "epoch": 2.191904047976012,
      "grad_norm": 0.19621265325676193,
      "learning_rate": 4.105925818557953e-05,
      "loss": 0.0306,
      "step": 2924
    },
    {
      "epoch": 2.1926536731634183,
      "grad_norm": 0.1490287113856066,
      "learning_rate": 4.098880321780716e-05,
      "loss": 0.01,
      "step": 2925
    },
    {
      "epoch": 2.1934032983508245,
      "grad_norm": 0.16054903709881607,
      "learning_rate": 4.0918393164672155e-05,
      "loss": 0.0262,
      "step": 2926
    },
    {
      "epoch": 2.1941529235382307,
      "grad_norm": 0.06522683759255349,
      "learning_rate": 4.0848028079765014e-05,
      "loss": 0.0022,
      "step": 2927
    },
    {
      "epoch": 2.1949025487256373,
      "grad_norm": 0.07793807891903581,
      "learning_rate": 4.0777708016642133e-05,
      "loss": 0.0025,
      "step": 2928
    },
    {
      "epoch": 2.1956521739130435,
      "grad_norm": 0.0671421190639738,
      "learning_rate": 4.070743302882551e-05,
      "loss": 0.0047,
      "step": 2929
    },
    {
      "epoch": 2.1964017991004496,
      "grad_norm": 0.09863003282295409,
      "learning_rate": 4.063720316980292e-05,
      "loss": 0.0112,
      "step": 2930
    },
    {
      "epoch": 2.1971514242878563,
      "grad_norm": 0.033368571987164586,
      "learning_rate": 4.0567018493027686e-05,
      "loss": 0.0016,
      "step": 2931
    },
    {
      "epoch": 2.1979010494752624,
      "grad_norm": 0.11283284783196637,
      "learning_rate": 4.049687905191889e-05,
      "loss": 0.0034,
      "step": 2932
    },
    {
      "epoch": 2.1986506746626686,
      "grad_norm": 0.15030626032340028,
      "learning_rate": 4.042678489986107e-05,
      "loss": 0.0073,
      "step": 2933
    },
    {
      "epoch": 2.199400299850075,
      "grad_norm": 0.08288939135255141,
      "learning_rate": 4.035673609020434e-05,
      "loss": 0.0065,
      "step": 2934
    },
    {
      "epoch": 2.2001499250374814,
      "grad_norm": 0.06272651785014262,
      "learning_rate": 4.028673267626427e-05,
      "loss": 0.0038,
      "step": 2935
    },
    {
      "epoch": 2.2008995502248876,
      "grad_norm": 0.04537491332847573,
      "learning_rate": 4.0216774711321925e-05,
      "loss": 0.0032,
      "step": 2936
    },
    {
      "epoch": 2.2016491754122938,
      "grad_norm": 0.1259908642818158,
      "learning_rate": 4.014686224862374e-05,
      "loss": 0.0023,
      "step": 2937
    },
    {
      "epoch": 2.2023988005997,
      "grad_norm": 0.1471177203320294,
      "learning_rate": 4.007699534138153e-05,
      "loss": 0.0113,
      "step": 2938
    },
    {
      "epoch": 2.2031484257871066,
      "grad_norm": 0.10461123602031132,
      "learning_rate": 4.000717404277244e-05,
      "loss": 0.0041,
      "step": 2939
    },
    {
      "epoch": 2.2038980509745127,
      "grad_norm": 0.0730772652802554,
      "learning_rate": 3.9937398405938896e-05,
      "loss": 0.002,
      "step": 2940
    },
    {
      "epoch": 2.204647676161919,
      "grad_norm": 0.053078479449533464,
      "learning_rate": 3.9867668483988585e-05,
      "loss": 0.0022,
      "step": 2941
    },
    {
      "epoch": 2.2053973013493255,
      "grad_norm": 0.0834540587187214,
      "learning_rate": 3.9797984329994363e-05,
      "loss": 0.0062,
      "step": 2942
    },
    {
      "epoch": 2.2061469265367317,
      "grad_norm": 0.07044363379314998,
      "learning_rate": 3.9728345996994296e-05,
      "loss": 0.0052,
      "step": 2943
    },
    {
      "epoch": 2.206896551724138,
      "grad_norm": 0.12451614573261646,
      "learning_rate": 3.9658753537991545e-05,
      "loss": 0.0104,
      "step": 2944
    },
    {
      "epoch": 2.207646176911544,
      "grad_norm": 0.06448779690377877,
      "learning_rate": 3.958920700595435e-05,
      "loss": 0.004,
      "step": 2945
    },
    {
      "epoch": 2.2083958020989507,
      "grad_norm": 0.03599955797498942,
      "learning_rate": 3.951970645381604e-05,
      "loss": 0.0025,
      "step": 2946
    },
    {
      "epoch": 2.209145427286357,
      "grad_norm": 0.06325898149816306,
      "learning_rate": 3.94502519344749e-05,
      "loss": 0.0051,
      "step": 2947
    },
    {
      "epoch": 2.209895052473763,
      "grad_norm": 0.14544048825696881,
      "learning_rate": 3.9380843500794195e-05,
      "loss": 0.0065,
      "step": 2948
    },
    {
      "epoch": 2.2106446776611692,
      "grad_norm": 0.06951395805907995,
      "learning_rate": 3.931148120560211e-05,
      "loss": 0.0043,
      "step": 2949
    },
    {
      "epoch": 2.211394302848576,
      "grad_norm": 0.04920109718547303,
      "learning_rate": 3.9242165101691685e-05,
      "loss": 0.0014,
      "step": 2950
    },
    {
      "epoch": 2.212143928035982,
      "grad_norm": 0.19249354074046146,
      "learning_rate": 3.9172895241820905e-05,
      "loss": 0.0105,
      "step": 2951
    },
    {
      "epoch": 2.212893553223388,
      "grad_norm": 0.06689161024484175,
      "learning_rate": 3.910367167871246e-05,
      "loss": 0.0061,
      "step": 2952
    },
    {
      "epoch": 2.213643178410795,
      "grad_norm": 0.1350737337552095,
      "learning_rate": 3.9034494465053815e-05,
      "loss": 0.0057,
      "step": 2953
    },
    {
      "epoch": 2.214392803598201,
      "grad_norm": 0.24292345106490407,
      "learning_rate": 3.8965363653497144e-05,
      "loss": 0.0042,
      "step": 2954
    },
    {
      "epoch": 2.215142428785607,
      "grad_norm": 0.026553507259357522,
      "learning_rate": 3.889627929665941e-05,
      "loss": 0.0007,
      "step": 2955
    },
    {
      "epoch": 2.2158920539730134,
      "grad_norm": 0.07556591664485746,
      "learning_rate": 3.882724144712209e-05,
      "loss": 0.0052,
      "step": 2956
    },
    {
      "epoch": 2.21664167916042,
      "grad_norm": 0.12006927039685981,
      "learning_rate": 3.875825015743132e-05,
      "loss": 0.0069,
      "step": 2957
    },
    {
      "epoch": 2.217391304347826,
      "grad_norm": 0.04691662091207386,
      "learning_rate": 3.868930548009779e-05,
      "loss": 0.0026,
      "step": 2958
    },
    {
      "epoch": 2.2181409295352323,
      "grad_norm": 0.13750868464230268,
      "learning_rate": 3.862040746759672e-05,
      "loss": 0.0042,
      "step": 2959
    },
    {
      "epoch": 2.2188905547226385,
      "grad_norm": 0.09332905233243183,
      "learning_rate": 3.855155617236781e-05,
      "loss": 0.0051,
      "step": 2960
    },
    {
      "epoch": 2.219640179910045,
      "grad_norm": 0.0354143689169886,
      "learning_rate": 3.8482751646815185e-05,
      "loss": 0.0021,
      "step": 2961
    },
    {
      "epoch": 2.2203898050974513,
      "grad_norm": 0.06158795473163416,
      "learning_rate": 3.841399394330739e-05,
      "loss": 0.0039,
      "step": 2962
    },
    {
      "epoch": 2.2211394302848575,
      "grad_norm": 0.1746590159335278,
      "learning_rate": 3.834528311417733e-05,
      "loss": 0.015,
      "step": 2963
    },
    {
      "epoch": 2.2218890554722637,
      "grad_norm": 0.1431609012747353,
      "learning_rate": 3.8276619211722245e-05,
      "loss": 0.0049,
      "step": 2964
    },
    {
      "epoch": 2.2226386806596703,
      "grad_norm": 0.04710948736106486,
      "learning_rate": 3.820800228820363e-05,
      "loss": 0.0033,
      "step": 2965
    },
    {
      "epoch": 2.2233883058470765,
      "grad_norm": 0.10769067240238436,
      "learning_rate": 3.813943239584723e-05,
      "loss": 0.0086,
      "step": 2966
    },
    {
      "epoch": 2.2241379310344827,
      "grad_norm": 0.1112212265090816,
      "learning_rate": 3.8070909586843026e-05,
      "loss": 0.0024,
      "step": 2967
    },
    {
      "epoch": 2.224887556221889,
      "grad_norm": 0.06666088121431606,
      "learning_rate": 3.8002433913345135e-05,
      "loss": 0.0042,
      "step": 2968
    },
    {
      "epoch": 2.2256371814092955,
      "grad_norm": 0.14479957975949176,
      "learning_rate": 3.79340054274718e-05,
      "loss": 0.006,
      "step": 2969
    },
    {
      "epoch": 2.2263868065967016,
      "grad_norm": 0.09796350711148064,
      "learning_rate": 3.786562418130535e-05,
      "loss": 0.0058,
      "step": 2970
    },
    {
      "epoch": 2.227136431784108,
      "grad_norm": 0.10417840290808125,
      "learning_rate": 3.7797290226892145e-05,
      "loss": 0.0046,
      "step": 2971
    },
    {
      "epoch": 2.2278860569715144,
      "grad_norm": 0.14637983584801909,
      "learning_rate": 3.7729003616242584e-05,
      "loss": 0.0069,
      "step": 2972
    },
    {
      "epoch": 2.2286356821589206,
      "grad_norm": 0.0596059585180575,
      "learning_rate": 3.766076440133097e-05,
      "loss": 0.0046,
      "step": 2973
    },
    {
      "epoch": 2.229385307346327,
      "grad_norm": 0.1298320643372351,
      "learning_rate": 3.7592572634095654e-05,
      "loss": 0.012,
      "step": 2974
    },
    {
      "epoch": 2.230134932533733,
      "grad_norm": 0.06279833117448319,
      "learning_rate": 3.752442836643876e-05,
      "loss": 0.0025,
      "step": 2975
    },
    {
      "epoch": 2.2308845577211396,
      "grad_norm": 0.053545212737444454,
      "learning_rate": 3.745633165022627e-05,
      "loss": 0.0041,
      "step": 2976
    },
    {
      "epoch": 2.2316341829085458,
      "grad_norm": 0.03920380410713475,
      "learning_rate": 3.738828253728799e-05,
      "loss": 0.0039,
      "step": 2977
    },
    {
      "epoch": 2.232383808095952,
      "grad_norm": 0.12025123048502838,
      "learning_rate": 3.732028107941755e-05,
      "loss": 0.0105,
      "step": 2978
    },
    {
      "epoch": 2.233133433283358,
      "grad_norm": 0.07880294740994104,
      "learning_rate": 3.725232732837225e-05,
      "loss": 0.0041,
      "step": 2979
    },
    {
      "epoch": 2.2338830584707647,
      "grad_norm": 0.10338096755245214,
      "learning_rate": 3.7184421335873067e-05,
      "loss": 0.0069,
      "step": 2980
    },
    {
      "epoch": 2.234632683658171,
      "grad_norm": 0.11514938715289685,
      "learning_rate": 3.7116563153604656e-05,
      "loss": 0.006,
      "step": 2981
    },
    {
      "epoch": 2.235382308845577,
      "grad_norm": 0.17877610807644287,
      "learning_rate": 3.70487528332153e-05,
      "loss": 0.0079,
      "step": 2982
    },
    {
      "epoch": 2.2361319340329837,
      "grad_norm": 0.06067208500268504,
      "learning_rate": 3.698099042631679e-05,
      "loss": 0.0018,
      "step": 2983
    },
    {
      "epoch": 2.23688155922039,
      "grad_norm": 0.3582779971925068,
      "learning_rate": 3.691327598448451e-05,
      "loss": 0.018,
      "step": 2984
    },
    {
      "epoch": 2.237631184407796,
      "grad_norm": 0.08483678159039827,
      "learning_rate": 3.684560955925732e-05,
      "loss": 0.0035,
      "step": 2985
    },
    {
      "epoch": 2.2383808095952022,
      "grad_norm": 0.13213318981736838,
      "learning_rate": 3.677799120213754e-05,
      "loss": 0.006,
      "step": 2986
    },
    {
      "epoch": 2.239130434782609,
      "grad_norm": 0.205558251722483,
      "learning_rate": 3.6710420964590876e-05,
      "loss": 0.0137,
      "step": 2987
    },
    {
      "epoch": 2.239880059970015,
      "grad_norm": 0.06881382245027329,
      "learning_rate": 3.664289889804643e-05,
      "loss": 0.0075,
      "step": 2988
    },
    {
      "epoch": 2.2406296851574212,
      "grad_norm": 0.04868954415719796,
      "learning_rate": 3.6575425053896636e-05,
      "loss": 0.0029,
      "step": 2989
    },
    {
      "epoch": 2.2413793103448274,
      "grad_norm": 0.10665721197763921,
      "learning_rate": 3.6507999483497235e-05,
      "loss": 0.0097,
      "step": 2990
    },
    {
      "epoch": 2.242128935532234,
      "grad_norm": 0.06542037858851574,
      "learning_rate": 3.644062223816721e-05,
      "loss": 0.003,
      "step": 2991
    },
    {
      "epoch": 2.24287856071964,
      "grad_norm": 0.07231973807330334,
      "learning_rate": 3.637329336918878e-05,
      "loss": 0.0036,
      "step": 2992
    },
    {
      "epoch": 2.2436281859070464,
      "grad_norm": 0.16776365912815602,
      "learning_rate": 3.630601292780733e-05,
      "loss": 0.0084,
      "step": 2993
    },
    {
      "epoch": 2.244377811094453,
      "grad_norm": 0.23586581493040168,
      "learning_rate": 3.6238780965231375e-05,
      "loss": 0.0069,
      "step": 2994
    },
    {
      "epoch": 2.245127436281859,
      "grad_norm": 0.1033982312441333,
      "learning_rate": 3.6171597532632575e-05,
      "loss": 0.0126,
      "step": 2995
    },
    {
      "epoch": 2.2458770614692654,
      "grad_norm": 0.171232132167371,
      "learning_rate": 3.61044626811456e-05,
      "loss": 0.0102,
      "step": 2996
    },
    {
      "epoch": 2.2466266866566715,
      "grad_norm": 0.09715820745250381,
      "learning_rate": 3.603737646186814e-05,
      "loss": 0.0035,
      "step": 2997
    },
    {
      "epoch": 2.247376311844078,
      "grad_norm": 0.10968137200157604,
      "learning_rate": 3.597033892586098e-05,
      "loss": 0.003,
      "step": 2998
    },
    {
      "epoch": 2.2481259370314843,
      "grad_norm": 0.05217799438191155,
      "learning_rate": 3.590335012414772e-05,
      "loss": 0.0059,
      "step": 2999
    },
    {
      "epoch": 2.2488755622188905,
      "grad_norm": 0.05733050677905598,
      "learning_rate": 3.5836410107714925e-05,
      "loss": 0.0032,
      "step": 3000
    },
    {
      "epoch": 2.2496251874062967,
      "grad_norm": 0.1021056862324412,
      "learning_rate": 3.5769518927511966e-05,
      "loss": 0.0032,
      "step": 3001
    },
    {
      "epoch": 2.2503748125937033,
      "grad_norm": 0.07310938180823594,
      "learning_rate": 3.5702676634451205e-05,
      "loss": 0.0053,
      "step": 3002
    },
    {
      "epoch": 2.2511244377811095,
      "grad_norm": 0.0818866352771629,
      "learning_rate": 3.563588327940762e-05,
      "loss": 0.0055,
      "step": 3003
    },
    {
      "epoch": 2.2518740629685157,
      "grad_norm": 0.01913471780865569,
      "learning_rate": 3.5569138913219026e-05,
      "loss": 0.0008,
      "step": 3004
    },
    {
      "epoch": 2.2526236881559223,
      "grad_norm": 0.055861302921822106,
      "learning_rate": 3.550244358668592e-05,
      "loss": 0.0028,
      "step": 3005
    },
    {
      "epoch": 2.2533733133433285,
      "grad_norm": 0.08024208404821104,
      "learning_rate": 3.543579735057151e-05,
      "loss": 0.0018,
      "step": 3006
    },
    {
      "epoch": 2.2541229385307346,
      "grad_norm": 0.13793260963967538,
      "learning_rate": 3.53692002556016e-05,
      "loss": 0.0049,
      "step": 3007
    },
    {
      "epoch": 2.254872563718141,
      "grad_norm": 0.12180580380490126,
      "learning_rate": 3.5302652352464625e-05,
      "loss": 0.0092,
      "step": 3008
    },
    {
      "epoch": 2.255622188905547,
      "grad_norm": 0.24832447046922468,
      "learning_rate": 3.5236153691811556e-05,
      "loss": 0.011,
      "step": 3009
    },
    {
      "epoch": 2.2563718140929536,
      "grad_norm": 0.10710552637315611,
      "learning_rate": 3.5169704324255895e-05,
      "loss": 0.0041,
      "step": 3010
    },
    {
      "epoch": 2.25712143928036,
      "grad_norm": 0.023998445286078023,
      "learning_rate": 3.510330430037363e-05,
      "loss": 0.0005,
      "step": 3011
    },
    {
      "epoch": 2.257871064467766,
      "grad_norm": 0.08477613453589629,
      "learning_rate": 3.503695367070318e-05,
      "loss": 0.0036,
      "step": 3012
    },
    {
      "epoch": 2.2586206896551726,
      "grad_norm": 0.11309357409035045,
      "learning_rate": 3.497065248574537e-05,
      "loss": 0.0074,
      "step": 3013
    },
    {
      "epoch": 2.2593703148425788,
      "grad_norm": 0.2413432662111237,
      "learning_rate": 3.490440079596341e-05,
      "loss": 0.011,
      "step": 3014
    },
    {
      "epoch": 2.260119940029985,
      "grad_norm": 0.15438899388396532,
      "learning_rate": 3.483819865178283e-05,
      "loss": 0.0051,
      "step": 3015
    },
    {
      "epoch": 2.260869565217391,
      "grad_norm": 0.11248704273274097,
      "learning_rate": 3.477204610359143e-05,
      "loss": 0.0097,
      "step": 3016
    },
    {
      "epoch": 2.2616191904047978,
      "grad_norm": 0.10270473751697841,
      "learning_rate": 3.4705943201739274e-05,
      "loss": 0.0037,
      "step": 3017
    },
    {
      "epoch": 2.262368815592204,
      "grad_norm": 0.06663120310158174,
      "learning_rate": 3.4639889996538664e-05,
      "loss": 0.0044,
      "step": 3018
    },
    {
      "epoch": 2.26311844077961,
      "grad_norm": 0.04856854864541843,
      "learning_rate": 3.457388653826405e-05,
      "loss": 0.0022,
      "step": 3019
    },
    {
      "epoch": 2.2638680659670163,
      "grad_norm": 0.048466365705767143,
      "learning_rate": 3.450793287715196e-05,
      "loss": 0.0048,
      "step": 3020
    },
    {
      "epoch": 2.264617691154423,
      "grad_norm": 0.09503234848236053,
      "learning_rate": 3.4442029063401195e-05,
      "loss": 0.0073,
      "step": 3021
    },
    {
      "epoch": 2.265367316341829,
      "grad_norm": 0.16817200927825984,
      "learning_rate": 3.437617514717244e-05,
      "loss": 0.015,
      "step": 3022
    },
    {
      "epoch": 2.2661169415292353,
      "grad_norm": 0.02779505203286547,
      "learning_rate": 3.431037117858848e-05,
      "loss": 0.0013,
      "step": 3023
    },
    {
      "epoch": 2.266866566716642,
      "grad_norm": 0.10431680116886312,
      "learning_rate": 3.424461720773403e-05,
      "loss": 0.0037,
      "step": 3024
    },
    {
      "epoch": 2.267616191904048,
      "grad_norm": 0.15647341956223487,
      "learning_rate": 3.4178913284655865e-05,
      "loss": 0.0132,
      "step": 3025
    },
    {
      "epoch": 2.2683658170914542,
      "grad_norm": 0.4506494810308986,
      "learning_rate": 3.411325945936257e-05,
      "loss": 0.0342,
      "step": 3026
    },
    {
      "epoch": 2.2691154422788604,
      "grad_norm": 0.05291685737207425,
      "learning_rate": 3.40476557818246e-05,
      "loss": 0.0039,
      "step": 3027
    },
    {
      "epoch": 2.269865067466267,
      "grad_norm": 0.0529791034376625,
      "learning_rate": 3.398210230197429e-05,
      "loss": 0.0066,
      "step": 3028
    },
    {
      "epoch": 2.270614692653673,
      "grad_norm": 0.266638755312817,
      "learning_rate": 3.391659906970573e-05,
      "loss": 0.0235,
      "step": 3029
    },
    {
      "epoch": 2.2713643178410794,
      "grad_norm": 0.09890484403926711,
      "learning_rate": 3.3851146134874775e-05,
      "loss": 0.0066,
      "step": 3030
    },
    {
      "epoch": 2.2721139430284856,
      "grad_norm": 0.0848537069501797,
      "learning_rate": 3.378574354729901e-05,
      "loss": 0.0088,
      "step": 3031
    },
    {
      "epoch": 2.272863568215892,
      "grad_norm": 0.05974783788622155,
      "learning_rate": 3.37203913567577e-05,
      "loss": 0.003,
      "step": 3032
    },
    {
      "epoch": 2.2736131934032984,
      "grad_norm": 0.24716829048847536,
      "learning_rate": 3.365508961299172e-05,
      "loss": 0.0097,
      "step": 3033
    },
    {
      "epoch": 2.2743628185907045,
      "grad_norm": 0.148063506124021,
      "learning_rate": 3.3589838365703594e-05,
      "loss": 0.0087,
      "step": 3034
    },
    {
      "epoch": 2.275112443778111,
      "grad_norm": 0.08608244559058832,
      "learning_rate": 3.352463766455738e-05,
      "loss": 0.0153,
      "step": 3035
    },
    {
      "epoch": 2.2758620689655173,
      "grad_norm": 0.08896344091548676,
      "learning_rate": 3.345948755917867e-05,
      "loss": 0.0033,
      "step": 3036
    },
    {
      "epoch": 2.2766116941529235,
      "grad_norm": 0.1066313799481121,
      "learning_rate": 3.339438809915456e-05,
      "loss": 0.0053,
      "step": 3037
    },
    {
      "epoch": 2.2773613193403297,
      "grad_norm": 0.06683865574093746,
      "learning_rate": 3.3329339334033595e-05,
      "loss": 0.0029,
      "step": 3038
    },
    {
      "epoch": 2.2781109445277363,
      "grad_norm": 0.08087808202357481,
      "learning_rate": 3.326434131332571e-05,
      "loss": 0.0034,
      "step": 3039
    },
    {
      "epoch": 2.2788605697151425,
      "grad_norm": 0.01588263533994879,
      "learning_rate": 3.319939408650225e-05,
      "loss": 0.0011,
      "step": 3040
    },
    {
      "epoch": 2.2796101949025487,
      "grad_norm": 0.12522455341806843,
      "learning_rate": 3.31344977029959e-05,
      "loss": 0.004,
      "step": 3041
    },
    {
      "epoch": 2.280359820089955,
      "grad_norm": 0.1123280419849201,
      "learning_rate": 3.306965221220062e-05,
      "loss": 0.0074,
      "step": 3042
    },
    {
      "epoch": 2.2811094452773615,
      "grad_norm": 0.2228488028010439,
      "learning_rate": 3.300485766347161e-05,
      "loss": 0.0068,
      "step": 3043
    },
    {
      "epoch": 2.2818590704647677,
      "grad_norm": 0.20103985308072542,
      "learning_rate": 3.294011410612541e-05,
      "loss": 0.0126,
      "step": 3044
    },
    {
      "epoch": 2.282608695652174,
      "grad_norm": 0.13115501711622973,
      "learning_rate": 3.2875421589439634e-05,
      "loss": 0.0042,
      "step": 3045
    },
    {
      "epoch": 2.2833583208395805,
      "grad_norm": 0.07938313674178223,
      "learning_rate": 3.281078016265311e-05,
      "loss": 0.0058,
      "step": 3046
    },
    {
      "epoch": 2.2841079460269866,
      "grad_norm": 0.22928488253082546,
      "learning_rate": 3.274618987496574e-05,
      "loss": 0.0142,
      "step": 3047
    },
    {
      "epoch": 2.284857571214393,
      "grad_norm": 0.040375034631480115,
      "learning_rate": 3.268165077553849e-05,
      "loss": 0.0022,
      "step": 3048
    },
    {
      "epoch": 2.285607196401799,
      "grad_norm": 0.07016817070990827,
      "learning_rate": 3.261716291349347e-05,
      "loss": 0.0032,
      "step": 3049
    },
    {
      "epoch": 2.286356821589205,
      "grad_norm": 0.055459174216538384,
      "learning_rate": 3.2552726337913686e-05,
      "loss": 0.003,
      "step": 3050
    },
    {
      "epoch": 2.287106446776612,
      "grad_norm": 0.2100936606920281,
      "learning_rate": 3.248834109784313e-05,
      "loss": 0.007,
      "step": 3051
    },
    {
      "epoch": 2.287856071964018,
      "grad_norm": 0.23237588883682203,
      "learning_rate": 3.242400724228677e-05,
      "loss": 0.027,
      "step": 3052
    },
    {
      "epoch": 2.288605697151424,
      "grad_norm": 0.09144615215224357,
      "learning_rate": 3.235972482021039e-05,
      "loss": 0.006,
      "step": 3053
    },
    {
      "epoch": 2.2893553223388308,
      "grad_norm": 0.08436516257363523,
      "learning_rate": 3.2295493880540687e-05,
      "loss": 0.0049,
      "step": 3054
    },
    {
      "epoch": 2.290104947526237,
      "grad_norm": 0.09340848215971319,
      "learning_rate": 3.223131447216514e-05,
      "loss": 0.0018,
      "step": 3055
    },
    {
      "epoch": 2.290854572713643,
      "grad_norm": 0.14138316361171396,
      "learning_rate": 3.2167186643932025e-05,
      "loss": 0.0101,
      "step": 3056
    },
    {
      "epoch": 2.2916041979010493,
      "grad_norm": 0.0780038451097566,
      "learning_rate": 3.2103110444650354e-05,
      "loss": 0.0041,
      "step": 3057
    },
    {
      "epoch": 2.292353823088456,
      "grad_norm": 0.13771768007587726,
      "learning_rate": 3.203908592308984e-05,
      "loss": 0.0052,
      "step": 3058
    },
    {
      "epoch": 2.293103448275862,
      "grad_norm": 0.11314310653938886,
      "learning_rate": 3.197511312798087e-05,
      "loss": 0.0031,
      "step": 3059
    },
    {
      "epoch": 2.2938530734632683,
      "grad_norm": 0.06740799151396357,
      "learning_rate": 3.191119210801444e-05,
      "loss": 0.0019,
      "step": 3060
    },
    {
      "epoch": 2.2946026986506745,
      "grad_norm": 0.08458018097254301,
      "learning_rate": 3.184732291184218e-05,
      "loss": 0.0044,
      "step": 3061
    },
    {
      "epoch": 2.295352323838081,
      "grad_norm": 0.08296125916158942,
      "learning_rate": 3.178350558807624e-05,
      "loss": 0.0054,
      "step": 3062
    },
    {
      "epoch": 2.2961019490254873,
      "grad_norm": 0.08461209266922365,
      "learning_rate": 3.171974018528928e-05,
      "loss": 0.0044,
      "step": 3063
    },
    {
      "epoch": 2.2968515742128934,
      "grad_norm": 0.31138688440724316,
      "learning_rate": 3.1656026752014486e-05,
      "loss": 0.0161,
      "step": 3064
    },
    {
      "epoch": 2.2976011994003,
      "grad_norm": 0.1700687652529863,
      "learning_rate": 3.159236533674547e-05,
      "loss": 0.0112,
      "step": 3065
    },
    {
      "epoch": 2.2983508245877062,
      "grad_norm": 0.12293576424028,
      "learning_rate": 3.1528755987936186e-05,
      "loss": 0.0078,
      "step": 3066
    },
    {
      "epoch": 2.2991004497751124,
      "grad_norm": 0.25224580670874097,
      "learning_rate": 3.14651987540011e-05,
      "loss": 0.0231,
      "step": 3067
    },
    {
      "epoch": 2.2998500749625186,
      "grad_norm": 0.12752668587526034,
      "learning_rate": 3.1401693683314904e-05,
      "loss": 0.0089,
      "step": 3068
    },
    {
      "epoch": 2.300599700149925,
      "grad_norm": 0.07308291934037676,
      "learning_rate": 3.1338240824212606e-05,
      "loss": 0.004,
      "step": 3069
    },
    {
      "epoch": 2.3013493253373314,
      "grad_norm": 0.1442663761239963,
      "learning_rate": 3.1274840224989485e-05,
      "loss": 0.008,
      "step": 3070
    },
    {
      "epoch": 2.3020989505247376,
      "grad_norm": 0.097064895836265,
      "learning_rate": 3.121149193390106e-05,
      "loss": 0.0031,
      "step": 3071
    },
    {
      "epoch": 2.3028485757121437,
      "grad_norm": 0.10366709486171628,
      "learning_rate": 3.1148195999162966e-05,
      "loss": 0.0038,
      "step": 3072
    },
    {
      "epoch": 2.3035982008995504,
      "grad_norm": 0.10820343477738625,
      "learning_rate": 3.108495246895111e-05,
      "loss": 0.0099,
      "step": 3073
    },
    {
      "epoch": 2.3043478260869565,
      "grad_norm": 0.052683725617420046,
      "learning_rate": 3.1021761391401424e-05,
      "loss": 0.0022,
      "step": 3074
    },
    {
      "epoch": 2.3050974512743627,
      "grad_norm": 0.04856462072758565,
      "learning_rate": 3.095862281460994e-05,
      "loss": 0.004,
      "step": 3075
    },
    {
      "epoch": 2.3058470764617693,
      "grad_norm": 0.10917084487952686,
      "learning_rate": 3.0895536786632716e-05,
      "loss": 0.0053,
      "step": 3076
    },
    {
      "epoch": 2.3065967016491755,
      "grad_norm": 0.09271317437410774,
      "learning_rate": 3.083250335548583e-05,
      "loss": 0.0063,
      "step": 3077
    },
    {
      "epoch": 2.3073463268365817,
      "grad_norm": 0.1014603016537938,
      "learning_rate": 3.0769522569145335e-05,
      "loss": 0.0051,
      "step": 3078
    },
    {
      "epoch": 2.308095952023988,
      "grad_norm": 0.182073711613753,
      "learning_rate": 3.070659447554719e-05,
      "loss": 0.0134,
      "step": 3079
    },
    {
      "epoch": 2.3088455772113945,
      "grad_norm": 0.051493830991612595,
      "learning_rate": 3.0643719122587275e-05,
      "loss": 0.0031,
      "step": 3080
    },
    {
      "epoch": 2.3095952023988007,
      "grad_norm": 0.06813819702008223,
      "learning_rate": 3.058089655812132e-05,
      "loss": 0.0022,
      "step": 3081
    },
    {
      "epoch": 2.310344827586207,
      "grad_norm": 0.1497102741665549,
      "learning_rate": 3.051812682996487e-05,
      "loss": 0.0113,
      "step": 3082
    },
    {
      "epoch": 2.311094452773613,
      "grad_norm": 0.13331343034849766,
      "learning_rate": 3.0455409985893236e-05,
      "loss": 0.0074,
      "step": 3083
    },
    {
      "epoch": 2.3118440779610197,
      "grad_norm": 0.11990281948287221,
      "learning_rate": 3.0392746073641533e-05,
      "loss": 0.0034,
      "step": 3084
    },
    {
      "epoch": 2.312593703148426,
      "grad_norm": 0.24197839414958924,
      "learning_rate": 3.0330135140904526e-05,
      "loss": 0.0166,
      "step": 3085
    },
    {
      "epoch": 2.313343328335832,
      "grad_norm": 0.054404535737172195,
      "learning_rate": 3.026757723533671e-05,
      "loss": 0.0033,
      "step": 3086
    },
    {
      "epoch": 2.3140929535232386,
      "grad_norm": 0.08049491433420755,
      "learning_rate": 3.0205072404552182e-05,
      "loss": 0.0032,
      "step": 3087
    },
    {
      "epoch": 2.314842578710645,
      "grad_norm": 0.16292162964705675,
      "learning_rate": 3.0142620696124656e-05,
      "loss": 0.0055,
      "step": 3088
    },
    {
      "epoch": 2.315592203898051,
      "grad_norm": 0.07344569034686321,
      "learning_rate": 3.0080222157587424e-05,
      "loss": 0.0023,
      "step": 3089
    },
    {
      "epoch": 2.316341829085457,
      "grad_norm": 0.1092571190290509,
      "learning_rate": 3.0017876836433245e-05,
      "loss": 0.0025,
      "step": 3090
    },
    {
      "epoch": 2.3170914542728633,
      "grad_norm": 0.04795213189239423,
      "learning_rate": 2.995558478011452e-05,
      "loss": 0.0014,
      "step": 3091
    },
    {
      "epoch": 2.31784107946027,
      "grad_norm": 0.0501643876826235,
      "learning_rate": 2.989334603604297e-05,
      "loss": 0.0025,
      "step": 3092
    },
    {
      "epoch": 2.318590704647676,
      "grad_norm": 0.04523174534856307,
      "learning_rate": 2.9831160651589775e-05,
      "loss": 0.0023,
      "step": 3093
    },
    {
      "epoch": 2.3193403298350823,
      "grad_norm": 0.02758127641822319,
      "learning_rate": 2.9769028674085532e-05,
      "loss": 0.0009,
      "step": 3094
    },
    {
      "epoch": 2.320089955022489,
      "grad_norm": 0.1421280114887876,
      "learning_rate": 2.9706950150820144e-05,
      "loss": 0.0074,
      "step": 3095
    },
    {
      "epoch": 2.320839580209895,
      "grad_norm": 0.1258121786875046,
      "learning_rate": 2.9644925129042857e-05,
      "loss": 0.0078,
      "step": 3096
    },
    {
      "epoch": 2.3215892053973013,
      "grad_norm": 0.09869709212905185,
      "learning_rate": 2.958295365596222e-05,
      "loss": 0.0063,
      "step": 3097
    },
    {
      "epoch": 2.3223388305847075,
      "grad_norm": 0.044815860770597356,
      "learning_rate": 2.9521035778745988e-05,
      "loss": 0.0039,
      "step": 3098
    },
    {
      "epoch": 2.323088455772114,
      "grad_norm": 0.07516192279941192,
      "learning_rate": 2.9459171544521125e-05,
      "loss": 0.0058,
      "step": 3099
    },
    {
      "epoch": 2.3238380809595203,
      "grad_norm": 0.11244585004434644,
      "learning_rate": 2.939736100037378e-05,
      "loss": 0.0101,
      "step": 3100
    },
    {
      "epoch": 2.3245877061469264,
      "grad_norm": 0.18701938835082377,
      "learning_rate": 2.933560419334922e-05,
      "loss": 0.0122,
      "step": 3101
    },
    {
      "epoch": 2.3253373313343326,
      "grad_norm": 0.10305684565093204,
      "learning_rate": 2.9273901170451835e-05,
      "loss": 0.0024,
      "step": 3102
    },
    {
      "epoch": 2.3260869565217392,
      "grad_norm": 0.0993002558158889,
      "learning_rate": 2.9212251978645056e-05,
      "loss": 0.0043,
      "step": 3103
    },
    {
      "epoch": 2.3268365817091454,
      "grad_norm": 0.10134589099502897,
      "learning_rate": 2.915065666485136e-05,
      "loss": 0.0046,
      "step": 3104
    },
    {
      "epoch": 2.3275862068965516,
      "grad_norm": 0.09878875420161998,
      "learning_rate": 2.9089115275952218e-05,
      "loss": 0.002,
      "step": 3105
    },
    {
      "epoch": 2.3283358320839582,
      "grad_norm": 0.11019541110942332,
      "learning_rate": 2.902762785878802e-05,
      "loss": 0.0057,
      "step": 3106
    },
    {
      "epoch": 2.3290854572713644,
      "grad_norm": 0.11157700778282026,
      "learning_rate": 2.8966194460158137e-05,
      "loss": 0.004,
      "step": 3107
    },
    {
      "epoch": 2.3298350824587706,
      "grad_norm": 0.40933062048626645,
      "learning_rate": 2.8904815126820774e-05,
      "loss": 0.0129,
      "step": 3108
    },
    {
      "epoch": 2.3305847076461768,
      "grad_norm": 0.07305006067035855,
      "learning_rate": 2.8843489905493016e-05,
      "loss": 0.0052,
      "step": 3109
    },
    {
      "epoch": 2.3313343328335834,
      "grad_norm": 0.0680123486769411,
      "learning_rate": 2.878221884285074e-05,
      "loss": 0.0079,
      "step": 3110
    },
    {
      "epoch": 2.3320839580209896,
      "grad_norm": 0.06019169149018181,
      "learning_rate": 2.872100198552863e-05,
      "loss": 0.0034,
      "step": 3111
    },
    {
      "epoch": 2.3328335832083957,
      "grad_norm": 0.056783640717217754,
      "learning_rate": 2.8659839380120092e-05,
      "loss": 0.0032,
      "step": 3112
    },
    {
      "epoch": 2.333583208395802,
      "grad_norm": 0.11551220701800663,
      "learning_rate": 2.8598731073177198e-05,
      "loss": 0.0086,
      "step": 3113
    },
    {
      "epoch": 2.3343328335832085,
      "grad_norm": 0.06999471091101042,
      "learning_rate": 2.8537677111210825e-05,
      "loss": 0.0036,
      "step": 3114
    },
    {
      "epoch": 2.3350824587706147,
      "grad_norm": 0.04765342441440993,
      "learning_rate": 2.8476677540690355e-05,
      "loss": 0.0015,
      "step": 3115
    },
    {
      "epoch": 2.335832083958021,
      "grad_norm": 0.05319190193486766,
      "learning_rate": 2.841573240804383e-05,
      "loss": 0.0017,
      "step": 3116
    },
    {
      "epoch": 2.3365817091454275,
      "grad_norm": 0.1813902628820559,
      "learning_rate": 2.8354841759657823e-05,
      "loss": 0.0181,
      "step": 3117
    },
    {
      "epoch": 2.3373313343328337,
      "grad_norm": 0.12581938917539412,
      "learning_rate": 2.8294005641877486e-05,
      "loss": 0.0019,
      "step": 3118
    },
    {
      "epoch": 2.33808095952024,
      "grad_norm": 0.14448099642639844,
      "learning_rate": 2.8233224101006428e-05,
      "loss": 0.0129,
      "step": 3119
    },
    {
      "epoch": 2.338830584707646,
      "grad_norm": 0.13954587924358886,
      "learning_rate": 2.8172497183306712e-05,
      "loss": 0.0058,
      "step": 3120
    },
    {
      "epoch": 2.3395802098950527,
      "grad_norm": 0.14411637591598964,
      "learning_rate": 2.8111824934998887e-05,
      "loss": 0.0043,
      "step": 3121
    },
    {
      "epoch": 2.340329835082459,
      "grad_norm": 0.06198186086028406,
      "learning_rate": 2.8051207402261838e-05,
      "loss": 0.007,
      "step": 3122
    },
    {
      "epoch": 2.341079460269865,
      "grad_norm": 0.14413877090723234,
      "learning_rate": 2.799064463123281e-05,
      "loss": 0.008,
      "step": 3123
    },
    {
      "epoch": 2.341829085457271,
      "grad_norm": 0.09347129089982951,
      "learning_rate": 2.7930136668007378e-05,
      "loss": 0.0043,
      "step": 3124
    },
    {
      "epoch": 2.342578710644678,
      "grad_norm": 0.13541759433461203,
      "learning_rate": 2.7869683558639393e-05,
      "loss": 0.0061,
      "step": 3125
    },
    {
      "epoch": 2.343328335832084,
      "grad_norm": 0.19373063801056187,
      "learning_rate": 2.7809285349140978e-05,
      "loss": 0.02,
      "step": 3126
    },
    {
      "epoch": 2.34407796101949,
      "grad_norm": 0.0766996044013267,
      "learning_rate": 2.7748942085482432e-05,
      "loss": 0.0045,
      "step": 3127
    },
    {
      "epoch": 2.344827586206897,
      "grad_norm": 0.21287550164640445,
      "learning_rate": 2.768865381359228e-05,
      "loss": 0.0146,
      "step": 3128
    },
    {
      "epoch": 2.345577211394303,
      "grad_norm": 0.1467023997059412,
      "learning_rate": 2.762842057935714e-05,
      "loss": 0.011,
      "step": 3129
    },
    {
      "epoch": 2.346326836581709,
      "grad_norm": 0.07213302935373363,
      "learning_rate": 2.7568242428621795e-05,
      "loss": 0.0076,
      "step": 3130
    },
    {
      "epoch": 2.3470764617691153,
      "grad_norm": 0.11244197680293319,
      "learning_rate": 2.750811940718906e-05,
      "loss": 0.0078,
      "step": 3131
    },
    {
      "epoch": 2.3478260869565215,
      "grad_norm": 0.13515852196537875,
      "learning_rate": 2.7448051560819798e-05,
      "loss": 0.0113,
      "step": 3132
    },
    {
      "epoch": 2.348575712143928,
      "grad_norm": 0.08773623719969813,
      "learning_rate": 2.7388038935232906e-05,
      "loss": 0.005,
      "step": 3133
    },
    {
      "epoch": 2.3493253373313343,
      "grad_norm": 0.12461713634230395,
      "learning_rate": 2.7328081576105212e-05,
      "loss": 0.0061,
      "step": 3134
    },
    {
      "epoch": 2.3500749625187405,
      "grad_norm": 0.5424887360890198,
      "learning_rate": 2.7268179529071504e-05,
      "loss": 0.0212,
      "step": 3135
    },
    {
      "epoch": 2.350824587706147,
      "grad_norm": 0.0903500759016297,
      "learning_rate": 2.7208332839724428e-05,
      "loss": 0.0042,
      "step": 3136
    },
    {
      "epoch": 2.3515742128935533,
      "grad_norm": 0.18532535793380175,
      "learning_rate": 2.7148541553614593e-05,
      "loss": 0.008,
      "step": 3137
    },
    {
      "epoch": 2.3523238380809595,
      "grad_norm": 0.07390377641173043,
      "learning_rate": 2.7088805716250353e-05,
      "loss": 0.0047,
      "step": 3138
    },
    {
      "epoch": 2.3530734632683656,
      "grad_norm": 0.14290629394801793,
      "learning_rate": 2.7029125373097885e-05,
      "loss": 0.0034,
      "step": 3139
    },
    {
      "epoch": 2.3538230884557723,
      "grad_norm": 0.154780779006307,
      "learning_rate": 2.696950056958112e-05,
      "loss": 0.013,
      "step": 3140
    },
    {
      "epoch": 2.3545727136431784,
      "grad_norm": 0.06244049624897275,
      "learning_rate": 2.690993135108173e-05,
      "loss": 0.0102,
      "step": 3141
    },
    {
      "epoch": 2.3553223388305846,
      "grad_norm": 0.06907429271018437,
      "learning_rate": 2.6850417762939085e-05,
      "loss": 0.0022,
      "step": 3142
    },
    {
      "epoch": 2.356071964017991,
      "grad_norm": 0.04446260531409627,
      "learning_rate": 2.6790959850450182e-05,
      "loss": 0.0026,
      "step": 3143
    },
    {
      "epoch": 2.3568215892053974,
      "grad_norm": 0.07036016666397808,
      "learning_rate": 2.6731557658869666e-05,
      "loss": 0.0051,
      "step": 3144
    },
    {
      "epoch": 2.3575712143928036,
      "grad_norm": 0.06282767585097133,
      "learning_rate": 2.667221123340975e-05,
      "loss": 0.0027,
      "step": 3145
    },
    {
      "epoch": 2.3583208395802098,
      "grad_norm": 0.043149230951043435,
      "learning_rate": 2.6612920619240267e-05,
      "loss": 0.0006,
      "step": 3146
    },
    {
      "epoch": 2.3590704647676164,
      "grad_norm": 0.07576831760682239,
      "learning_rate": 2.655368586148852e-05,
      "loss": 0.0024,
      "step": 3147
    },
    {
      "epoch": 2.3598200899550226,
      "grad_norm": 0.08108836245646409,
      "learning_rate": 2.6494507005239266e-05,
      "loss": 0.0062,
      "step": 3148
    },
    {
      "epoch": 2.3605697151424287,
      "grad_norm": 0.11713349844058288,
      "learning_rate": 2.6435384095534787e-05,
      "loss": 0.0057,
      "step": 3149
    },
    {
      "epoch": 2.361319340329835,
      "grad_norm": 0.08772970147787831,
      "learning_rate": 2.6376317177374733e-05,
      "loss": 0.008,
      "step": 3150
    },
    {
      "epoch": 2.3620689655172415,
      "grad_norm": 0.14803691236224617,
      "learning_rate": 2.631730629571614e-05,
      "loss": 0.0069,
      "step": 3151
    },
    {
      "epoch": 2.3628185907046477,
      "grad_norm": 0.1144611596997879,
      "learning_rate": 2.6258351495473422e-05,
      "loss": 0.0083,
      "step": 3152
    },
    {
      "epoch": 2.363568215892054,
      "grad_norm": 0.06841431105601573,
      "learning_rate": 2.6199452821518276e-05,
      "loss": 0.0053,
      "step": 3153
    },
    {
      "epoch": 2.36431784107946,
      "grad_norm": 0.16048168050678102,
      "learning_rate": 2.6140610318679713e-05,
      "loss": 0.0079,
      "step": 3154
    },
    {
      "epoch": 2.3650674662668667,
      "grad_norm": 0.1513435464155828,
      "learning_rate": 2.6081824031743973e-05,
      "loss": 0.0072,
      "step": 3155
    },
    {
      "epoch": 2.365817091454273,
      "grad_norm": 0.20449265633671435,
      "learning_rate": 2.6023094005454508e-05,
      "loss": 0.0075,
      "step": 3156
    },
    {
      "epoch": 2.366566716641679,
      "grad_norm": 0.18888809594549003,
      "learning_rate": 2.5964420284511936e-05,
      "loss": 0.0078,
      "step": 3157
    },
    {
      "epoch": 2.3673163418290857,
      "grad_norm": 0.08862512881682469,
      "learning_rate": 2.590580291357404e-05,
      "loss": 0.0038,
      "step": 3158
    },
    {
      "epoch": 2.368065967016492,
      "grad_norm": 0.1560274818938899,
      "learning_rate": 2.5847241937255696e-05,
      "loss": 0.0037,
      "step": 3159
    },
    {
      "epoch": 2.368815592203898,
      "grad_norm": 0.09133329862797442,
      "learning_rate": 2.5788737400128915e-05,
      "loss": 0.0043,
      "step": 3160
    },
    {
      "epoch": 2.369565217391304,
      "grad_norm": 0.03709155544184032,
      "learning_rate": 2.573028934672267e-05,
      "loss": 0.0034,
      "step": 3161
    },
    {
      "epoch": 2.370314842578711,
      "grad_norm": 0.05432304607523067,
      "learning_rate": 2.5671897821522994e-05,
      "loss": 0.0022,
      "step": 3162
    },
    {
      "epoch": 2.371064467766117,
      "grad_norm": 0.19186007553771886,
      "learning_rate": 2.5613562868972863e-05,
      "loss": 0.0032,
      "step": 3163
    },
    {
      "epoch": 2.371814092953523,
      "grad_norm": 0.10272902414474801,
      "learning_rate": 2.5555284533472225e-05,
      "loss": 0.0052,
      "step": 3164
    },
    {
      "epoch": 2.3725637181409294,
      "grad_norm": 0.33886922406643966,
      "learning_rate": 2.5497062859377907e-05,
      "loss": 0.0214,
      "step": 3165
    },
    {
      "epoch": 2.373313343328336,
      "grad_norm": 0.09329404194299042,
      "learning_rate": 2.5438897891003644e-05,
      "loss": 0.0058,
      "step": 3166
    },
    {
      "epoch": 2.374062968515742,
      "grad_norm": 0.027590342633528796,
      "learning_rate": 2.5380789672619976e-05,
      "loss": 0.0005,
      "step": 3167
    },
    {
      "epoch": 2.3748125937031483,
      "grad_norm": 0.10410756246755437,
      "learning_rate": 2.532273824845426e-05,
      "loss": 0.0028,
      "step": 3168
    },
    {
      "epoch": 2.375562218890555,
      "grad_norm": 0.05486602305277263,
      "learning_rate": 2.5264743662690626e-05,
      "loss": 0.0033,
      "step": 3169
    },
    {
      "epoch": 2.376311844077961,
      "grad_norm": 0.22503256238636352,
      "learning_rate": 2.5206805959469982e-05,
      "loss": 0.0051,
      "step": 3170
    },
    {
      "epoch": 2.3770614692653673,
      "grad_norm": 0.06505037277743252,
      "learning_rate": 2.514892518288988e-05,
      "loss": 0.0032,
      "step": 3171
    },
    {
      "epoch": 2.3778110944527735,
      "grad_norm": 0.11335554596112911,
      "learning_rate": 2.509110137700459e-05,
      "loss": 0.0045,
      "step": 3172
    },
    {
      "epoch": 2.3785607196401797,
      "grad_norm": 0.09207603652791845,
      "learning_rate": 2.5033334585824987e-05,
      "loss": 0.0068,
      "step": 3173
    },
    {
      "epoch": 2.3793103448275863,
      "grad_norm": 0.08269320542812943,
      "learning_rate": 2.497562485331857e-05,
      "loss": 0.0024,
      "step": 3174
    },
    {
      "epoch": 2.3800599700149925,
      "grad_norm": 0.0921472923368709,
      "learning_rate": 2.4917972223409403e-05,
      "loss": 0.003,
      "step": 3175
    },
    {
      "epoch": 2.3808095952023987,
      "grad_norm": 0.04366359037083029,
      "learning_rate": 2.4860376739978087e-05,
      "loss": 0.0019,
      "step": 3176
    },
    {
      "epoch": 2.3815592203898053,
      "grad_norm": 0.03400033401219638,
      "learning_rate": 2.480283844686174e-05,
      "loss": 0.0015,
      "step": 3177
    },
    {
      "epoch": 2.3823088455772115,
      "grad_norm": 0.09706802506791543,
      "learning_rate": 2.474535738785394e-05,
      "loss": 0.0071,
      "step": 3178
    },
    {
      "epoch": 2.3830584707646176,
      "grad_norm": 0.05951984507419919,
      "learning_rate": 2.4687933606704694e-05,
      "loss": 0.0023,
      "step": 3179
    },
    {
      "epoch": 2.383808095952024,
      "grad_norm": 0.043996798867113274,
      "learning_rate": 2.463056714712043e-05,
      "loss": 0.0018,
      "step": 3180
    },
    {
      "epoch": 2.3845577211394304,
      "grad_norm": 0.03593451814711463,
      "learning_rate": 2.4573258052763948e-05,
      "loss": 0.001,
      "step": 3181
    },
    {
      "epoch": 2.3853073463268366,
      "grad_norm": 0.14919848701160657,
      "learning_rate": 2.4516006367254384e-05,
      "loss": 0.0068,
      "step": 3182
    },
    {
      "epoch": 2.386056971514243,
      "grad_norm": 0.19455939856494417,
      "learning_rate": 2.4458812134167132e-05,
      "loss": 0.0132,
      "step": 3183
    },
    {
      "epoch": 2.386806596701649,
      "grad_norm": 0.07319318949194768,
      "learning_rate": 2.4401675397033964e-05,
      "loss": 0.005,
      "step": 3184
    },
    {
      "epoch": 2.3875562218890556,
      "grad_norm": 0.04140646732606931,
      "learning_rate": 2.4344596199342793e-05,
      "loss": 0.0011,
      "step": 3185
    },
    {
      "epoch": 2.3883058470764618,
      "grad_norm": 0.20088390351180652,
      "learning_rate": 2.4287574584537786e-05,
      "loss": 0.0094,
      "step": 3186
    },
    {
      "epoch": 2.389055472263868,
      "grad_norm": 0.15192548332497946,
      "learning_rate": 2.4230610596019265e-05,
      "loss": 0.0107,
      "step": 3187
    },
    {
      "epoch": 2.3898050974512746,
      "grad_norm": 0.11208152247157709,
      "learning_rate": 2.4173704277143695e-05,
      "loss": 0.0063,
      "step": 3188
    },
    {
      "epoch": 2.3905547226386807,
      "grad_norm": 0.08848833128247399,
      "learning_rate": 2.4116855671223625e-05,
      "loss": 0.0051,
      "step": 3189
    },
    {
      "epoch": 2.391304347826087,
      "grad_norm": 0.05282982327077367,
      "learning_rate": 2.406006482152774e-05,
      "loss": 0.003,
      "step": 3190
    },
    {
      "epoch": 2.392053973013493,
      "grad_norm": 0.08195828669539026,
      "learning_rate": 2.400333177128069e-05,
      "loss": 0.008,
      "step": 3191
    },
    {
      "epoch": 2.3928035982008997,
      "grad_norm": 0.04534374907943794,
      "learning_rate": 2.3946656563663196e-05,
      "loss": 0.002,
      "step": 3192
    },
    {
      "epoch": 2.393553223388306,
      "grad_norm": 0.08263871987216773,
      "learning_rate": 2.3890039241811867e-05,
      "loss": 0.0032,
      "step": 3193
    },
    {
      "epoch": 2.394302848575712,
      "grad_norm": 0.08472564331894322,
      "learning_rate": 2.3833479848819395e-05,
      "loss": 0.002,
      "step": 3194
    },
    {
      "epoch": 2.3950524737631183,
      "grad_norm": 0.08397816672253118,
      "learning_rate": 2.3776978427734263e-05,
      "loss": 0.0023,
      "step": 3195
    },
    {
      "epoch": 2.395802098950525,
      "grad_norm": 0.08482986650966184,
      "learning_rate": 2.3720535021560865e-05,
      "loss": 0.0061,
      "step": 3196
    },
    {
      "epoch": 2.396551724137931,
      "grad_norm": 0.12808246072318563,
      "learning_rate": 2.366414967325944e-05,
      "loss": 0.012,
      "step": 3197
    },
    {
      "epoch": 2.3973013493253372,
      "grad_norm": 0.024415326203775196,
      "learning_rate": 2.360782242574604e-05,
      "loss": 0.0007,
      "step": 3198
    },
    {
      "epoch": 2.398050974512744,
      "grad_norm": 0.06455213192453027,
      "learning_rate": 2.3551553321892494e-05,
      "loss": 0.0089,
      "step": 3199
    },
    {
      "epoch": 2.39880059970015,
      "grad_norm": 0.039121680575436184,
      "learning_rate": 2.3495342404526364e-05,
      "loss": 0.0009,
      "step": 3200
    },
    {
      "epoch": 2.399550224887556,
      "grad_norm": 0.11826139251554775,
      "learning_rate": 2.3439189716430954e-05,
      "loss": 0.0071,
      "step": 3201
    },
    {
      "epoch": 2.4002998500749624,
      "grad_norm": 0.20739755707080024,
      "learning_rate": 2.3383095300345224e-05,
      "loss": 0.0072,
      "step": 3202
    },
    {
      "epoch": 2.401049475262369,
      "grad_norm": 0.06532052332726593,
      "learning_rate": 2.3327059198963798e-05,
      "loss": 0.002,
      "step": 3203
    },
    {
      "epoch": 2.401799100449775,
      "grad_norm": 0.22750847274796626,
      "learning_rate": 2.3271081454936883e-05,
      "loss": 0.0104,
      "step": 3204
    },
    {
      "epoch": 2.4025487256371814,
      "grad_norm": 0.017377477297441345,
      "learning_rate": 2.3215162110870315e-05,
      "loss": 0.0004,
      "step": 3205
    },
    {
      "epoch": 2.4032983508245875,
      "grad_norm": 0.1778878319909849,
      "learning_rate": 2.3159301209325422e-05,
      "loss": 0.0047,
      "step": 3206
    },
    {
      "epoch": 2.404047976011994,
      "grad_norm": 0.100210020891019,
      "learning_rate": 2.3103498792819157e-05,
      "loss": 0.0041,
      "step": 3207
    },
    {
      "epoch": 2.4047976011994003,
      "grad_norm": 0.15348021525655878,
      "learning_rate": 2.3047754903823837e-05,
      "loss": 0.007,
      "step": 3208
    },
    {
      "epoch": 2.4055472263868065,
      "grad_norm": 0.09173669806021993,
      "learning_rate": 2.2992069584767306e-05,
      "loss": 0.0044,
      "step": 3209
    },
    {
      "epoch": 2.406296851574213,
      "grad_norm": 0.1449428363068518,
      "learning_rate": 2.2936442878032782e-05,
      "loss": 0.0103,
      "step": 3210
    },
    {
      "epoch": 2.4070464767616193,
      "grad_norm": 0.1931471716433352,
      "learning_rate": 2.288087482595892e-05,
      "loss": 0.0056,
      "step": 3211
    },
    {
      "epoch": 2.4077961019490255,
      "grad_norm": 0.18221181616530532,
      "learning_rate": 2.282536547083971e-05,
      "loss": 0.0088,
      "step": 3212
    },
    {
      "epoch": 2.4085457271364317,
      "grad_norm": 0.07535885371074759,
      "learning_rate": 2.2769914854924447e-05,
      "loss": 0.004,
      "step": 3213
    },
    {
      "epoch": 2.409295352323838,
      "grad_norm": 0.2867869116874561,
      "learning_rate": 2.2714523020417743e-05,
      "loss": 0.0045,
      "step": 3214
    },
    {
      "epoch": 2.4100449775112445,
      "grad_norm": 0.10463185396513008,
      "learning_rate": 2.2659190009479482e-05,
      "loss": 0.0144,
      "step": 3215
    },
    {
      "epoch": 2.4107946026986506,
      "grad_norm": 0.08266492494464286,
      "learning_rate": 2.2603915864224734e-05,
      "loss": 0.0048,
      "step": 3216
    },
    {
      "epoch": 2.411544227886057,
      "grad_norm": 0.13151923674905358,
      "learning_rate": 2.2548700626723774e-05,
      "loss": 0.0066,
      "step": 3217
    },
    {
      "epoch": 2.4122938530734634,
      "grad_norm": 0.1498302704791928,
      "learning_rate": 2.2493544339002116e-05,
      "loss": 0.0102,
      "step": 3218
    },
    {
      "epoch": 2.4130434782608696,
      "grad_norm": 0.0545047426124523,
      "learning_rate": 2.24384470430403e-05,
      "loss": 0.0033,
      "step": 3219
    },
    {
      "epoch": 2.413793103448276,
      "grad_norm": 0.13852701257857522,
      "learning_rate": 2.238340878077404e-05,
      "loss": 0.0091,
      "step": 3220
    },
    {
      "epoch": 2.414542728635682,
      "grad_norm": 0.21838529938682086,
      "learning_rate": 2.232842959409407e-05,
      "loss": 0.0092,
      "step": 3221
    },
    {
      "epoch": 2.4152923538230886,
      "grad_norm": 0.14462270899799032,
      "learning_rate": 2.2273509524846192e-05,
      "loss": 0.0077,
      "step": 3222
    },
    {
      "epoch": 2.4160419790104948,
      "grad_norm": 0.09223350982849962,
      "learning_rate": 2.2218648614831193e-05,
      "loss": 0.0088,
      "step": 3223
    },
    {
      "epoch": 2.416791604197901,
      "grad_norm": 0.11345239267831092,
      "learning_rate": 2.2163846905804852e-05,
      "loss": 0.0031,
      "step": 3224
    },
    {
      "epoch": 2.417541229385307,
      "grad_norm": 0.06562607715456643,
      "learning_rate": 2.2109104439477867e-05,
      "loss": 0.0033,
      "step": 3225
    },
    {
      "epoch": 2.4182908545727138,
      "grad_norm": 0.1258976567530301,
      "learning_rate": 2.205442125751588e-05,
      "loss": 0.0039,
      "step": 3226
    },
    {
      "epoch": 2.41904047976012,
      "grad_norm": 0.0592420034040759,
      "learning_rate": 2.1999797401539358e-05,
      "loss": 0.0036,
      "step": 3227
    },
    {
      "epoch": 2.419790104947526,
      "grad_norm": 0.062410574467906405,
      "learning_rate": 2.1945232913123647e-05,
      "loss": 0.0026,
      "step": 3228
    },
    {
      "epoch": 2.4205397301349327,
      "grad_norm": 0.16072667320626324,
      "learning_rate": 2.1890727833798895e-05,
      "loss": 0.0051,
      "step": 3229
    },
    {
      "epoch": 2.421289355322339,
      "grad_norm": 0.06435646077977134,
      "learning_rate": 2.1836282205050085e-05,
      "loss": 0.0044,
      "step": 3230
    },
    {
      "epoch": 2.422038980509745,
      "grad_norm": 0.10547962395023756,
      "learning_rate": 2.1781896068316875e-05,
      "loss": 0.0043,
      "step": 3231
    },
    {
      "epoch": 2.4227886056971513,
      "grad_norm": 0.34184567189195214,
      "learning_rate": 2.172756946499368e-05,
      "loss": 0.0205,
      "step": 3232
    },
    {
      "epoch": 2.423538230884558,
      "grad_norm": 0.01943317922835859,
      "learning_rate": 2.167330243642959e-05,
      "loss": 0.0006,
      "step": 3233
    },
    {
      "epoch": 2.424287856071964,
      "grad_norm": 0.20086372463230104,
      "learning_rate": 2.161909502392836e-05,
      "loss": 0.011,
      "step": 3234
    },
    {
      "epoch": 2.4250374812593702,
      "grad_norm": 0.07101192234715667,
      "learning_rate": 2.156494726874838e-05,
      "loss": 0.0083,
      "step": 3235
    },
    {
      "epoch": 2.4257871064467764,
      "grad_norm": 0.04049036261930732,
      "learning_rate": 2.151085921210261e-05,
      "loss": 0.0047,
      "step": 3236
    },
    {
      "epoch": 2.426536731634183,
      "grad_norm": 0.16307343023420023,
      "learning_rate": 2.1456830895158574e-05,
      "loss": 0.0059,
      "step": 3237
    },
    {
      "epoch": 2.427286356821589,
      "grad_norm": 0.12506840491379043,
      "learning_rate": 2.1402862359038368e-05,
      "loss": 0.004,
      "step": 3238
    },
    {
      "epoch": 2.4280359820089954,
      "grad_norm": 0.0885584667949349,
      "learning_rate": 2.1348953644818525e-05,
      "loss": 0.0073,
      "step": 3239
    },
    {
      "epoch": 2.428785607196402,
      "grad_norm": 0.14648244588484946,
      "learning_rate": 2.1295104793530097e-05,
      "loss": 0.0062,
      "step": 3240
    },
    {
      "epoch": 2.429535232383808,
      "grad_norm": 0.07240228865975842,
      "learning_rate": 2.1241315846158505e-05,
      "loss": 0.0055,
      "step": 3241
    },
    {
      "epoch": 2.4302848575712144,
      "grad_norm": 0.03639532646793308,
      "learning_rate": 2.1187586843643713e-05,
      "loss": 0.0017,
      "step": 3242
    },
    {
      "epoch": 2.4310344827586206,
      "grad_norm": 0.10077748270325737,
      "learning_rate": 2.1133917826879913e-05,
      "loss": 0.0131,
      "step": 3243
    },
    {
      "epoch": 2.431784107946027,
      "grad_norm": 0.26074454882474535,
      "learning_rate": 2.1080308836715724e-05,
      "loss": 0.0147,
      "step": 3244
    },
    {
      "epoch": 2.4325337331334334,
      "grad_norm": 0.092146972275415,
      "learning_rate": 2.102675991395403e-05,
      "loss": 0.0027,
      "step": 3245
    },
    {
      "epoch": 2.4332833583208395,
      "grad_norm": 0.13457985066514191,
      "learning_rate": 2.0973271099352034e-05,
      "loss": 0.0055,
      "step": 3246
    },
    {
      "epoch": 2.4340329835082457,
      "grad_norm": 0.09271671929768435,
      "learning_rate": 2.0919842433621174e-05,
      "loss": 0.0088,
      "step": 3247
    },
    {
      "epoch": 2.4347826086956523,
      "grad_norm": 0.0579896275974002,
      "learning_rate": 2.086647395742709e-05,
      "loss": 0.0025,
      "step": 3248
    },
    {
      "epoch": 2.4355322338830585,
      "grad_norm": 0.03481173805902309,
      "learning_rate": 2.0813165711389647e-05,
      "loss": 0.002,
      "step": 3249
    },
    {
      "epoch": 2.4362818590704647,
      "grad_norm": 0.11311610359304168,
      "learning_rate": 2.0759917736082857e-05,
      "loss": 0.0038,
      "step": 3250
    },
    {
      "epoch": 2.4370314842578713,
      "grad_norm": 0.13988931045307879,
      "learning_rate": 2.070673007203483e-05,
      "loss": 0.0028,
      "step": 3251
    },
    {
      "epoch": 2.4377811094452775,
      "grad_norm": 0.14957193004353075,
      "learning_rate": 2.0653602759727774e-05,
      "loss": 0.0061,
      "step": 3252
    },
    {
      "epoch": 2.4385307346326837,
      "grad_norm": 0.04143554274829354,
      "learning_rate": 2.0600535839598034e-05,
      "loss": 0.0036,
      "step": 3253
    },
    {
      "epoch": 2.43928035982009,
      "grad_norm": 0.06687076357394921,
      "learning_rate": 2.054752935203591e-05,
      "loss": 0.0017,
      "step": 3254
    },
    {
      "epoch": 2.440029985007496,
      "grad_norm": 0.05813849507678315,
      "learning_rate": 2.049458333738573e-05,
      "loss": 0.0053,
      "step": 3255
    },
    {
      "epoch": 2.4407796101949026,
      "grad_norm": 0.161437410408533,
      "learning_rate": 2.0441697835945794e-05,
      "loss": 0.0106,
      "step": 3256
    },
    {
      "epoch": 2.441529235382309,
      "grad_norm": 0.07793383157440188,
      "learning_rate": 2.0388872887968356e-05,
      "loss": 0.0033,
      "step": 3257
    },
    {
      "epoch": 2.442278860569715,
      "grad_norm": 0.3149123134517686,
      "learning_rate": 2.033610853365957e-05,
      "loss": 0.0522,
      "step": 3258
    },
    {
      "epoch": 2.4430284857571216,
      "grad_norm": 0.01664652143123375,
      "learning_rate": 2.0283404813179463e-05,
      "loss": 0.0006,
      "step": 3259
    },
    {
      "epoch": 2.443778110944528,
      "grad_norm": 0.08198414275140468,
      "learning_rate": 2.0230761766641927e-05,
      "loss": 0.0073,
      "step": 3260
    },
    {
      "epoch": 2.444527736131934,
      "grad_norm": 0.0754390799513741,
      "learning_rate": 2.0178179434114674e-05,
      "loss": 0.014,
      "step": 3261
    },
    {
      "epoch": 2.44527736131934,
      "grad_norm": 0.10862792424697457,
      "learning_rate": 2.0125657855619197e-05,
      "loss": 0.0055,
      "step": 3262
    },
    {
      "epoch": 2.4460269865067468,
      "grad_norm": 0.21754292192596295,
      "learning_rate": 2.0073197071130745e-05,
      "loss": 0.0088,
      "step": 3263
    },
    {
      "epoch": 2.446776611694153,
      "grad_norm": 0.1918952617560053,
      "learning_rate": 2.0020797120578306e-05,
      "loss": 0.0283,
      "step": 3264
    },
    {
      "epoch": 2.447526236881559,
      "grad_norm": 0.043382904168564676,
      "learning_rate": 1.996845804384454e-05,
      "loss": 0.0026,
      "step": 3265
    },
    {
      "epoch": 2.4482758620689653,
      "grad_norm": 0.19778157747190214,
      "learning_rate": 1.9916179880765838e-05,
      "loss": 0.0177,
      "step": 3266
    },
    {
      "epoch": 2.449025487256372,
      "grad_norm": 0.18805089046965645,
      "learning_rate": 1.9863962671132184e-05,
      "loss": 0.0072,
      "step": 3267
    },
    {
      "epoch": 2.449775112443778,
      "grad_norm": 0.18755316996292778,
      "learning_rate": 1.9811806454687142e-05,
      "loss": 0.018,
      "step": 3268
    },
    {
      "epoch": 2.4505247376311843,
      "grad_norm": 0.06614250670744387,
      "learning_rate": 1.97597112711279e-05,
      "loss": 0.0033,
      "step": 3269
    },
    {
      "epoch": 2.451274362818591,
      "grad_norm": 0.11396886184625207,
      "learning_rate": 1.9707677160105176e-05,
      "loss": 0.0091,
      "step": 3270
    },
    {
      "epoch": 2.452023988005997,
      "grad_norm": 0.1191653191852702,
      "learning_rate": 1.9655704161223197e-05,
      "loss": 0.0091,
      "step": 3271
    },
    {
      "epoch": 2.4527736131934033,
      "grad_norm": 0.04978826362799082,
      "learning_rate": 1.9603792314039693e-05,
      "loss": 0.002,
      "step": 3272
    },
    {
      "epoch": 2.4535232383808094,
      "grad_norm": 0.06209037623530401,
      "learning_rate": 1.955194165806582e-05,
      "loss": 0.0026,
      "step": 3273
    },
    {
      "epoch": 2.454272863568216,
      "grad_norm": 0.15311403649462496,
      "learning_rate": 1.95001522327662e-05,
      "loss": 0.0138,
      "step": 3274
    },
    {
      "epoch": 2.4550224887556222,
      "grad_norm": 0.08934123417692379,
      "learning_rate": 1.9448424077558823e-05,
      "loss": 0.0036,
      "step": 3275
    },
    {
      "epoch": 2.4557721139430284,
      "grad_norm": 0.06125059469568494,
      "learning_rate": 1.939675723181502e-05,
      "loss": 0.004,
      "step": 3276
    },
    {
      "epoch": 2.4565217391304346,
      "grad_norm": 0.08209081530940565,
      "learning_rate": 1.934515173485957e-05,
      "loss": 0.0054,
      "step": 3277
    },
    {
      "epoch": 2.457271364317841,
      "grad_norm": 0.11537540780526452,
      "learning_rate": 1.9293607625970423e-05,
      "loss": 0.0107,
      "step": 3278
    },
    {
      "epoch": 2.4580209895052474,
      "grad_norm": 0.13870867629619593,
      "learning_rate": 1.924212494437887e-05,
      "loss": 0.0098,
      "step": 3279
    },
    {
      "epoch": 2.4587706146926536,
      "grad_norm": 0.02878266136949601,
      "learning_rate": 1.9190703729269456e-05,
      "loss": 0.0009,
      "step": 3280
    },
    {
      "epoch": 2.45952023988006,
      "grad_norm": 0.1279207625734458,
      "learning_rate": 1.9139344019779915e-05,
      "loss": 0.0077,
      "step": 3281
    },
    {
      "epoch": 2.4602698650674664,
      "grad_norm": 0.07684664298238444,
      "learning_rate": 1.908804585500117e-05,
      "loss": 0.003,
      "step": 3282
    },
    {
      "epoch": 2.4610194902548725,
      "grad_norm": 0.0543302880388553,
      "learning_rate": 1.9036809273977328e-05,
      "loss": 0.0018,
      "step": 3283
    },
    {
      "epoch": 2.4617691154422787,
      "grad_norm": 0.10888747269891666,
      "learning_rate": 1.89856343157056e-05,
      "loss": 0.0047,
      "step": 3284
    },
    {
      "epoch": 2.4625187406296853,
      "grad_norm": 0.2646476257190446,
      "learning_rate": 1.8934521019136308e-05,
      "loss": 0.012,
      "step": 3285
    },
    {
      "epoch": 2.4632683658170915,
      "grad_norm": 0.17184286978028573,
      "learning_rate": 1.8883469423172817e-05,
      "loss": 0.0125,
      "step": 3286
    },
    {
      "epoch": 2.4640179910044977,
      "grad_norm": 0.1390931631406034,
      "learning_rate": 1.883247956667157e-05,
      "loss": 0.0049,
      "step": 3287
    },
    {
      "epoch": 2.464767616191904,
      "grad_norm": 0.0714921482628974,
      "learning_rate": 1.8781551488441972e-05,
      "loss": 0.003,
      "step": 3288
    },
    {
      "epoch": 2.4655172413793105,
      "grad_norm": 0.133046984565039,
      "learning_rate": 1.8730685227246457e-05,
      "loss": 0.006,
      "step": 3289
    },
    {
      "epoch": 2.4662668665667167,
      "grad_norm": 0.11209588746179831,
      "learning_rate": 1.8679880821800343e-05,
      "loss": 0.0073,
      "step": 3290
    },
    {
      "epoch": 2.467016491754123,
      "grad_norm": 0.06819768532382646,
      "learning_rate": 1.8629138310771965e-05,
      "loss": 0.0022,
      "step": 3291
    },
    {
      "epoch": 2.4677661169415295,
      "grad_norm": 0.054661698085546696,
      "learning_rate": 1.8578457732782474e-05,
      "loss": 0.0013,
      "step": 3292
    },
    {
      "epoch": 2.4685157421289357,
      "grad_norm": 0.1085360087024495,
      "learning_rate": 1.8527839126405887e-05,
      "loss": 0.0034,
      "step": 3293
    },
    {
      "epoch": 2.469265367316342,
      "grad_norm": 0.049756364784295085,
      "learning_rate": 1.8477282530169083e-05,
      "loss": 0.0024,
      "step": 3294
    },
    {
      "epoch": 2.470014992503748,
      "grad_norm": 0.0366405441766717,
      "learning_rate": 1.842678798255172e-05,
      "loss": 0.0024,
      "step": 3295
    },
    {
      "epoch": 2.470764617691154,
      "grad_norm": 0.09309741368973933,
      "learning_rate": 1.837635552198623e-05,
      "loss": 0.0095,
      "step": 3296
    },
    {
      "epoch": 2.471514242878561,
      "grad_norm": 0.15714420420691394,
      "learning_rate": 1.832598518685782e-05,
      "loss": 0.0118,
      "step": 3297
    },
    {
      "epoch": 2.472263868065967,
      "grad_norm": 0.2070285577468285,
      "learning_rate": 1.8275677015504378e-05,
      "loss": 0.0086,
      "step": 3298
    },
    {
      "epoch": 2.473013493253373,
      "grad_norm": 0.14826770507822962,
      "learning_rate": 1.8225431046216467e-05,
      "loss": 0.0123,
      "step": 3299
    },
    {
      "epoch": 2.47376311844078,
      "grad_norm": 0.1449443151390573,
      "learning_rate": 1.8175247317237366e-05,
      "loss": 0.0071,
      "step": 3300
    },
    {
      "epoch": 2.474512743628186,
      "grad_norm": 0.08054687812669925,
      "learning_rate": 1.812512586676294e-05,
      "loss": 0.0009,
      "step": 3301
    },
    {
      "epoch": 2.475262368815592,
      "grad_norm": 0.019756970121557204,
      "learning_rate": 1.8075066732941647e-05,
      "loss": 0.0008,
      "step": 3302
    },
    {
      "epoch": 2.4760119940029983,
      "grad_norm": 0.19447683391193743,
      "learning_rate": 1.802506995387452e-05,
      "loss": 0.0058,
      "step": 3303
    },
    {
      "epoch": 2.476761619190405,
      "grad_norm": 0.04533386155965311,
      "learning_rate": 1.797513556761514e-05,
      "loss": 0.002,
      "step": 3304
    },
    {
      "epoch": 2.477511244377811,
      "grad_norm": 0.05840662803628417,
      "learning_rate": 1.7925263612169608e-05,
      "loss": 0.0019,
      "step": 3305
    },
    {
      "epoch": 2.4782608695652173,
      "grad_norm": 0.07904173424928156,
      "learning_rate": 1.7875454125496482e-05,
      "loss": 0.0034,
      "step": 3306
    },
    {
      "epoch": 2.4790104947526235,
      "grad_norm": 0.07591255699380649,
      "learning_rate": 1.78257071455068e-05,
      "loss": 0.0049,
      "step": 3307
    },
    {
      "epoch": 2.47976011994003,
      "grad_norm": 0.15415615413389439,
      "learning_rate": 1.777602271006401e-05,
      "loss": 0.0042,
      "step": 3308
    },
    {
      "epoch": 2.4805097451274363,
      "grad_norm": 0.10602648104833119,
      "learning_rate": 1.7726400856983948e-05,
      "loss": 0.0039,
      "step": 3309
    },
    {
      "epoch": 2.4812593703148424,
      "grad_norm": 0.1428032240200061,
      "learning_rate": 1.7676841624034846e-05,
      "loss": 0.0023,
      "step": 3310
    },
    {
      "epoch": 2.482008995502249,
      "grad_norm": 0.1057352560848211,
      "learning_rate": 1.7627345048937237e-05,
      "loss": 0.0106,
      "step": 3311
    },
    {
      "epoch": 2.4827586206896552,
      "grad_norm": 0.21025082098498768,
      "learning_rate": 1.7577911169364004e-05,
      "loss": 0.012,
      "step": 3312
    },
    {
      "epoch": 2.4835082458770614,
      "grad_norm": 0.2070421759988733,
      "learning_rate": 1.7528540022940288e-05,
      "loss": 0.0108,
      "step": 3313
    },
    {
      "epoch": 2.4842578710644676,
      "grad_norm": 0.3783571009031663,
      "learning_rate": 1.7479231647243444e-05,
      "loss": 0.0092,
      "step": 3314
    },
    {
      "epoch": 2.4850074962518742,
      "grad_norm": 0.10719782576136593,
      "learning_rate": 1.742998607980315e-05,
      "loss": 0.0041,
      "step": 3315
    },
    {
      "epoch": 2.4857571214392804,
      "grad_norm": 0.06422424759686289,
      "learning_rate": 1.7380803358101206e-05,
      "loss": 0.0031,
      "step": 3316
    },
    {
      "epoch": 2.4865067466266866,
      "grad_norm": 0.07860804952533651,
      "learning_rate": 1.7331683519571594e-05,
      "loss": 0.0078,
      "step": 3317
    },
    {
      "epoch": 2.4872563718140928,
      "grad_norm": 0.035314739290592416,
      "learning_rate": 1.7282626601600415e-05,
      "loss": 0.0021,
      "step": 3318
    },
    {
      "epoch": 2.4880059970014994,
      "grad_norm": 0.15164882963503368,
      "learning_rate": 1.7233632641525934e-05,
      "loss": 0.0046,
      "step": 3319
    },
    {
      "epoch": 2.4887556221889056,
      "grad_norm": 0.11015905813111565,
      "learning_rate": 1.7184701676638427e-05,
      "loss": 0.0065,
      "step": 3320
    },
    {
      "epoch": 2.4895052473763117,
      "grad_norm": 0.12187436113335419,
      "learning_rate": 1.7135833744180274e-05,
      "loss": 0.0097,
      "step": 3321
    },
    {
      "epoch": 2.4902548725637184,
      "grad_norm": 0.06096731943620705,
      "learning_rate": 1.7087028881345846e-05,
      "loss": 0.0058,
      "step": 3322
    },
    {
      "epoch": 2.4910044977511245,
      "grad_norm": 0.11027908003154516,
      "learning_rate": 1.7038287125281584e-05,
      "loss": 0.0171,
      "step": 3323
    },
    {
      "epoch": 2.4917541229385307,
      "grad_norm": 0.1564353026772974,
      "learning_rate": 1.6989608513085786e-05,
      "loss": 0.0073,
      "step": 3324
    },
    {
      "epoch": 2.492503748125937,
      "grad_norm": 0.10037155018203812,
      "learning_rate": 1.6940993081808787e-05,
      "loss": 0.0104,
      "step": 3325
    },
    {
      "epoch": 2.4932533733133435,
      "grad_norm": 0.14119281576355544,
      "learning_rate": 1.6892440868452764e-05,
      "loss": 0.006,
      "step": 3326
    },
    {
      "epoch": 2.4940029985007497,
      "grad_norm": 0.20832412941132497,
      "learning_rate": 1.684395190997182e-05,
      "loss": 0.0125,
      "step": 3327
    },
    {
      "epoch": 2.494752623688156,
      "grad_norm": 0.13363735419048947,
      "learning_rate": 1.6795526243271908e-05,
      "loss": 0.0037,
      "step": 3328
    },
    {
      "epoch": 2.495502248875562,
      "grad_norm": 0.23693648411466714,
      "learning_rate": 1.6747163905210782e-05,
      "loss": 0.0259,
      "step": 3329
    },
    {
      "epoch": 2.4962518740629687,
      "grad_norm": 0.05846092967251272,
      "learning_rate": 1.6698864932598012e-05,
      "loss": 0.0049,
      "step": 3330
    },
    {
      "epoch": 2.497001499250375,
      "grad_norm": 0.0954056777302459,
      "learning_rate": 1.665062936219497e-05,
      "loss": 0.0054,
      "step": 3331
    },
    {
      "epoch": 2.497751124437781,
      "grad_norm": 0.13874395577918525,
      "learning_rate": 1.6602457230714707e-05,
      "loss": 0.0054,
      "step": 3332
    },
    {
      "epoch": 2.4985007496251876,
      "grad_norm": 0.09704989563203428,
      "learning_rate": 1.6554348574822043e-05,
      "loss": 0.0074,
      "step": 3333
    },
    {
      "epoch": 2.499250374812594,
      "grad_norm": 0.10633359164911542,
      "learning_rate": 1.6506303431133465e-05,
      "loss": 0.0048,
      "step": 3334
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.0803813036357924,
      "learning_rate": 1.6458321836217118e-05,
      "loss": 0.0033,
      "step": 3335
    },
    {
      "epoch": 2.500749625187406,
      "grad_norm": 0.0678889800093358,
      "learning_rate": 1.641040382659279e-05,
      "loss": 0.0063,
      "step": 3336
    },
    {
      "epoch": 2.5014992503748124,
      "grad_norm": 0.11564895659247448,
      "learning_rate": 1.6362549438731857e-05,
      "loss": 0.011,
      "step": 3337
    },
    {
      "epoch": 2.502248875562219,
      "grad_norm": 0.10691469240997556,
      "learning_rate": 1.6314758709057254e-05,
      "loss": 0.0022,
      "step": 3338
    },
    {
      "epoch": 2.502998500749625,
      "grad_norm": 0.04258097389584946,
      "learning_rate": 1.6267031673943543e-05,
      "loss": 0.0029,
      "step": 3339
    },
    {
      "epoch": 2.5037481259370313,
      "grad_norm": 0.2022447545861962,
      "learning_rate": 1.6219368369716737e-05,
      "loss": 0.01,
      "step": 3340
    },
    {
      "epoch": 2.504497751124438,
      "grad_norm": 0.08254344073917239,
      "learning_rate": 1.617176883265433e-05,
      "loss": 0.013,
      "step": 3341
    },
    {
      "epoch": 2.505247376311844,
      "grad_norm": 0.24658516758530183,
      "learning_rate": 1.6124233098985352e-05,
      "loss": 0.0183,
      "step": 3342
    },
    {
      "epoch": 2.5059970014992503,
      "grad_norm": 0.08495882689774024,
      "learning_rate": 1.60767612048902e-05,
      "loss": 0.0053,
      "step": 3343
    },
    {
      "epoch": 2.506746626686657,
      "grad_norm": 0.07942922256090312,
      "learning_rate": 1.6029353186500708e-05,
      "loss": 0.0042,
      "step": 3344
    },
    {
      "epoch": 2.507496251874063,
      "grad_norm": 0.07118141791709931,
      "learning_rate": 1.5982009079900118e-05,
      "loss": 0.0066,
      "step": 3345
    },
    {
      "epoch": 2.5082458770614693,
      "grad_norm": 0.12139061932172453,
      "learning_rate": 1.5934728921122955e-05,
      "loss": 0.0036,
      "step": 3346
    },
    {
      "epoch": 2.5089955022488755,
      "grad_norm": 0.06628093070288814,
      "learning_rate": 1.5887512746155185e-05,
      "loss": 0.0031,
      "step": 3347
    },
    {
      "epoch": 2.5097451274362816,
      "grad_norm": 0.10456872395325517,
      "learning_rate": 1.5840360590933988e-05,
      "loss": 0.0059,
      "step": 3348
    },
    {
      "epoch": 2.5104947526236883,
      "grad_norm": 0.04529821859284247,
      "learning_rate": 1.579327249134783e-05,
      "loss": 0.0025,
      "step": 3349
    },
    {
      "epoch": 2.5112443778110944,
      "grad_norm": 0.08226942305606083,
      "learning_rate": 1.574624848323646e-05,
      "loss": 0.0065,
      "step": 3350
    },
    {
      "epoch": 2.5119940029985006,
      "grad_norm": 0.004788370896557784,
      "learning_rate": 1.5699288602390782e-05,
      "loss": 0.0001,
      "step": 3351
    },
    {
      "epoch": 2.5127436281859072,
      "grad_norm": 0.04635220007278972,
      "learning_rate": 1.5652392884552947e-05,
      "loss": 0.0029,
      "step": 3352
    },
    {
      "epoch": 2.5134932533733134,
      "grad_norm": 0.10648525699498171,
      "learning_rate": 1.5605561365416256e-05,
      "loss": 0.0077,
      "step": 3353
    },
    {
      "epoch": 2.5142428785607196,
      "grad_norm": 0.12715298764016186,
      "learning_rate": 1.5558794080625127e-05,
      "loss": 0.0082,
      "step": 3354
    },
    {
      "epoch": 2.5149925037481258,
      "grad_norm": 0.06798855324869771,
      "learning_rate": 1.5512091065775126e-05,
      "loss": 0.0076,
      "step": 3355
    },
    {
      "epoch": 2.5157421289355324,
      "grad_norm": 0.07077620336304406,
      "learning_rate": 1.546545235641287e-05,
      "loss": 0.0033,
      "step": 3356
    },
    {
      "epoch": 2.5164917541229386,
      "grad_norm": 0.06352248995169558,
      "learning_rate": 1.5418877988036028e-05,
      "loss": 0.0046,
      "step": 3357
    },
    {
      "epoch": 2.5172413793103448,
      "grad_norm": 0.13447931451764636,
      "learning_rate": 1.537236799609333e-05,
      "loss": 0.0048,
      "step": 3358
    },
    {
      "epoch": 2.517991004497751,
      "grad_norm": 0.23300269027514214,
      "learning_rate": 1.5325922415984482e-05,
      "loss": 0.0192,
      "step": 3359
    },
    {
      "epoch": 2.5187406296851576,
      "grad_norm": 0.10911871613118115,
      "learning_rate": 1.5279541283060163e-05,
      "loss": 0.0046,
      "step": 3360
    },
    {
      "epoch": 2.5194902548725637,
      "grad_norm": 0.07059905883441622,
      "learning_rate": 1.5233224632622023e-05,
      "loss": 0.0059,
      "step": 3361
    },
    {
      "epoch": 2.52023988005997,
      "grad_norm": 0.12077716301718372,
      "learning_rate": 1.5186972499922592e-05,
      "loss": 0.0054,
      "step": 3362
    },
    {
      "epoch": 2.5209895052473765,
      "grad_norm": 0.12174224416791476,
      "learning_rate": 1.5140784920165362e-05,
      "loss": 0.0061,
      "step": 3363
    },
    {
      "epoch": 2.5217391304347827,
      "grad_norm": 0.07569846123809262,
      "learning_rate": 1.5094661928504628e-05,
      "loss": 0.0049,
      "step": 3364
    },
    {
      "epoch": 2.522488755622189,
      "grad_norm": 0.11639308956168937,
      "learning_rate": 1.5048603560045549e-05,
      "loss": 0.0039,
      "step": 3365
    },
    {
      "epoch": 2.523238380809595,
      "grad_norm": 0.10689395952382459,
      "learning_rate": 1.5002609849844106e-05,
      "loss": 0.0132,
      "step": 3366
    },
    {
      "epoch": 2.5239880059970012,
      "grad_norm": 0.045761003984327045,
      "learning_rate": 1.4956680832907056e-05,
      "loss": 0.0013,
      "step": 3367
    },
    {
      "epoch": 2.524737631184408,
      "grad_norm": 0.11936305725887401,
      "learning_rate": 1.4910816544191908e-05,
      "loss": 0.0048,
      "step": 3368
    },
    {
      "epoch": 2.525487256371814,
      "grad_norm": 0.041702788660834045,
      "learning_rate": 1.4865017018606909e-05,
      "loss": 0.0006,
      "step": 3369
    },
    {
      "epoch": 2.52623688155922,
      "grad_norm": 0.20843594068598312,
      "learning_rate": 1.481928229101106e-05,
      "loss": 0.0093,
      "step": 3370
    },
    {
      "epoch": 2.526986506746627,
      "grad_norm": 0.06508216640159331,
      "learning_rate": 1.4773612396213988e-05,
      "loss": 0.003,
      "step": 3371
    },
    {
      "epoch": 2.527736131934033,
      "grad_norm": 0.10586528511790065,
      "learning_rate": 1.4728007368975983e-05,
      "loss": 0.004,
      "step": 3372
    },
    {
      "epoch": 2.528485757121439,
      "grad_norm": 0.05561954050455667,
      "learning_rate": 1.468246724400797e-05,
      "loss": 0.0024,
      "step": 3373
    },
    {
      "epoch": 2.529235382308846,
      "grad_norm": 0.18058892377561644,
      "learning_rate": 1.4636992055971477e-05,
      "loss": 0.0041,
      "step": 3374
    },
    {
      "epoch": 2.529985007496252,
      "grad_norm": 0.09340388762629254,
      "learning_rate": 1.4591581839478607e-05,
      "loss": 0.0038,
      "step": 3375
    },
    {
      "epoch": 2.530734632683658,
      "grad_norm": 0.09641022352823704,
      "learning_rate": 1.4546236629092025e-05,
      "loss": 0.0112,
      "step": 3376
    },
    {
      "epoch": 2.5314842578710643,
      "grad_norm": 0.017482766975639216,
      "learning_rate": 1.4500956459324876e-05,
      "loss": 0.0009,
      "step": 3377
    },
    {
      "epoch": 2.5322338830584705,
      "grad_norm": 0.006491921022898667,
      "learning_rate": 1.4455741364640862e-05,
      "loss": 0.0001,
      "step": 3378
    },
    {
      "epoch": 2.532983508245877,
      "grad_norm": 0.4206000354963242,
      "learning_rate": 1.4410591379454097e-05,
      "loss": 0.0258,
      "step": 3379
    },
    {
      "epoch": 2.5337331334332833,
      "grad_norm": 0.04486177965619197,
      "learning_rate": 1.436550653812918e-05,
      "loss": 0.0017,
      "step": 3380
    },
    {
      "epoch": 2.5344827586206895,
      "grad_norm": 0.0518983797022331,
      "learning_rate": 1.4320486874981099e-05,
      "loss": 0.0042,
      "step": 3381
    },
    {
      "epoch": 2.535232383808096,
      "grad_norm": 0.14914624735948265,
      "learning_rate": 1.4275532424275263e-05,
      "loss": 0.0115,
      "step": 3382
    },
    {
      "epoch": 2.5359820089955023,
      "grad_norm": 0.06403302361142668,
      "learning_rate": 1.4230643220227413e-05,
      "loss": 0.0014,
      "step": 3383
    },
    {
      "epoch": 2.5367316341829085,
      "grad_norm": 0.05932830774052173,
      "learning_rate": 1.4185819297003666e-05,
      "loss": 0.0059,
      "step": 3384
    },
    {
      "epoch": 2.537481259370315,
      "grad_norm": 0.08205924145282562,
      "learning_rate": 1.414106068872042e-05,
      "loss": 0.0055,
      "step": 3385
    },
    {
      "epoch": 2.5382308845577213,
      "grad_norm": 0.24183435329526498,
      "learning_rate": 1.4096367429444346e-05,
      "loss": 0.0072,
      "step": 3386
    },
    {
      "epoch": 2.5389805097451275,
      "grad_norm": 0.16890306225531435,
      "learning_rate": 1.4051739553192467e-05,
      "loss": 0.0042,
      "step": 3387
    },
    {
      "epoch": 2.5397301349325336,
      "grad_norm": 0.17121384666776437,
      "learning_rate": 1.4007177093931945e-05,
      "loss": 0.0065,
      "step": 3388
    },
    {
      "epoch": 2.54047976011994,
      "grad_norm": 0.06053968873643824,
      "learning_rate": 1.3962680085580194e-05,
      "loss": 0.0037,
      "step": 3389
    },
    {
      "epoch": 2.5412293853073464,
      "grad_norm": 0.07343884227867872,
      "learning_rate": 1.3918248562004788e-05,
      "loss": 0.0038,
      "step": 3390
    },
    {
      "epoch": 2.5419790104947526,
      "grad_norm": 0.10825304746636069,
      "learning_rate": 1.3873882557023488e-05,
      "loss": 0.0036,
      "step": 3391
    },
    {
      "epoch": 2.542728635682159,
      "grad_norm": 0.10489755580343259,
      "learning_rate": 1.3829582104404149e-05,
      "loss": 0.0047,
      "step": 3392
    },
    {
      "epoch": 2.5434782608695654,
      "grad_norm": 0.07540453424961847,
      "learning_rate": 1.3785347237864799e-05,
      "loss": 0.0029,
      "step": 3393
    },
    {
      "epoch": 2.5442278860569716,
      "grad_norm": 0.07260094751569643,
      "learning_rate": 1.3741177991073484e-05,
      "loss": 0.0035,
      "step": 3394
    },
    {
      "epoch": 2.5449775112443778,
      "grad_norm": 0.03176271883763127,
      "learning_rate": 1.3697074397648335e-05,
      "loss": 0.0014,
      "step": 3395
    },
    {
      "epoch": 2.545727136431784,
      "grad_norm": 0.08957438106312125,
      "learning_rate": 1.36530364911575e-05,
      "loss": 0.0101,
      "step": 3396
    },
    {
      "epoch": 2.5464767616191906,
      "grad_norm": 0.06852095264691668,
      "learning_rate": 1.3609064305119135e-05,
      "loss": 0.0051,
      "step": 3397
    },
    {
      "epoch": 2.5472263868065967,
      "grad_norm": 0.18697086935588128,
      "learning_rate": 1.3565157873001377e-05,
      "loss": 0.012,
      "step": 3398
    },
    {
      "epoch": 2.547976011994003,
      "grad_norm": 0.019448086688255033,
      "learning_rate": 1.3521317228222318e-05,
      "loss": 0.0009,
      "step": 3399
    },
    {
      "epoch": 2.548725637181409,
      "grad_norm": 0.16603021849626445,
      "learning_rate": 1.347754240414998e-05,
      "loss": 0.0091,
      "step": 3400
    },
    {
      "epoch": 2.5494752623688157,
      "grad_norm": 0.06125424646624753,
      "learning_rate": 1.3433833434102272e-05,
      "loss": 0.0028,
      "step": 3401
    },
    {
      "epoch": 2.550224887556222,
      "grad_norm": 0.03927286803929153,
      "learning_rate": 1.339019035134701e-05,
      "loss": 0.0016,
      "step": 3402
    },
    {
      "epoch": 2.550974512743628,
      "grad_norm": 0.08078465371000328,
      "learning_rate": 1.3346613189101819e-05,
      "loss": 0.0023,
      "step": 3403
    },
    {
      "epoch": 2.5517241379310347,
      "grad_norm": 0.18222397947614588,
      "learning_rate": 1.3303101980534184e-05,
      "loss": 0.0075,
      "step": 3404
    },
    {
      "epoch": 2.552473763118441,
      "grad_norm": 0.09719479831833734,
      "learning_rate": 1.3259656758761397e-05,
      "loss": 0.0076,
      "step": 3405
    },
    {
      "epoch": 2.553223388305847,
      "grad_norm": 0.11960845259668604,
      "learning_rate": 1.3216277556850487e-05,
      "loss": 0.0062,
      "step": 3406
    },
    {
      "epoch": 2.5539730134932532,
      "grad_norm": 0.04845046202394285,
      "learning_rate": 1.3172964407818277e-05,
      "loss": 0.0021,
      "step": 3407
    },
    {
      "epoch": 2.5547226386806594,
      "grad_norm": 0.08315262578770383,
      "learning_rate": 1.3129717344631288e-05,
      "loss": 0.0049,
      "step": 3408
    },
    {
      "epoch": 2.555472263868066,
      "grad_norm": 0.14214322041507318,
      "learning_rate": 1.3086536400205752e-05,
      "loss": 0.0186,
      "step": 3409
    },
    {
      "epoch": 2.556221889055472,
      "grad_norm": 0.09052640614593274,
      "learning_rate": 1.3043421607407557e-05,
      "loss": 0.0045,
      "step": 3410
    },
    {
      "epoch": 2.5569715142428784,
      "grad_norm": 0.02934896328124305,
      "learning_rate": 1.300037299905229e-05,
      "loss": 0.0008,
      "step": 3411
    },
    {
      "epoch": 2.557721139430285,
      "grad_norm": 0.04848408509157833,
      "learning_rate": 1.295739060790514e-05,
      "loss": 0.0039,
      "step": 3412
    },
    {
      "epoch": 2.558470764617691,
      "grad_norm": 0.02128701888303665,
      "learning_rate": 1.291447446668086e-05,
      "loss": 0.0006,
      "step": 3413
    },
    {
      "epoch": 2.5592203898050974,
      "grad_norm": 0.09413095292963528,
      "learning_rate": 1.287162460804382e-05,
      "loss": 0.0044,
      "step": 3414
    },
    {
      "epoch": 2.559970014992504,
      "grad_norm": 0.17174537693955969,
      "learning_rate": 1.2828841064607911e-05,
      "loss": 0.0081,
      "step": 3415
    },
    {
      "epoch": 2.56071964017991,
      "grad_norm": 0.07794231023141801,
      "learning_rate": 1.2786123868936617e-05,
      "loss": 0.003,
      "step": 3416
    },
    {
      "epoch": 2.5614692653673163,
      "grad_norm": 0.030253277823394865,
      "learning_rate": 1.2743473053542842e-05,
      "loss": 0.0021,
      "step": 3417
    },
    {
      "epoch": 2.5622188905547225,
      "grad_norm": 0.07176729695739768,
      "learning_rate": 1.2700888650889008e-05,
      "loss": 0.005,
      "step": 3418
    },
    {
      "epoch": 2.5629685157421287,
      "grad_norm": 0.07416769918242676,
      "learning_rate": 1.265837069338699e-05,
      "loss": 0.0019,
      "step": 3419
    },
    {
      "epoch": 2.5637181409295353,
      "grad_norm": 0.06427223920828888,
      "learning_rate": 1.2615919213398053e-05,
      "loss": 0.0026,
      "step": 3420
    },
    {
      "epoch": 2.5644677661169415,
      "grad_norm": 0.19369364107159953,
      "learning_rate": 1.2573534243232921e-05,
      "loss": 0.0036,
      "step": 3421
    },
    {
      "epoch": 2.5652173913043477,
      "grad_norm": 0.04976782879154124,
      "learning_rate": 1.2531215815151653e-05,
      "loss": 0.0017,
      "step": 3422
    },
    {
      "epoch": 2.5659670164917543,
      "grad_norm": 0.06578714098838653,
      "learning_rate": 1.2488963961363676e-05,
      "loss": 0.0059,
      "step": 3423
    },
    {
      "epoch": 2.5667166416791605,
      "grad_norm": 0.23465623124533136,
      "learning_rate": 1.2446778714027762e-05,
      "loss": 0.0102,
      "step": 3424
    },
    {
      "epoch": 2.5674662668665666,
      "grad_norm": 0.05855814761088096,
      "learning_rate": 1.2404660105251952e-05,
      "loss": 0.0027,
      "step": 3425
    },
    {
      "epoch": 2.5682158920539733,
      "grad_norm": 0.1571853358091456,
      "learning_rate": 1.2362608167093593e-05,
      "loss": 0.0027,
      "step": 3426
    },
    {
      "epoch": 2.5689655172413794,
      "grad_norm": 0.08157672990923209,
      "learning_rate": 1.2320622931559289e-05,
      "loss": 0.0035,
      "step": 3427
    },
    {
      "epoch": 2.5697151424287856,
      "grad_norm": 0.20591070104559273,
      "learning_rate": 1.227870443060486e-05,
      "loss": 0.016,
      "step": 3428
    },
    {
      "epoch": 2.570464767616192,
      "grad_norm": 0.22646797574654348,
      "learning_rate": 1.2236852696135347e-05,
      "loss": 0.0148,
      "step": 3429
    },
    {
      "epoch": 2.571214392803598,
      "grad_norm": 0.12966875286555696,
      "learning_rate": 1.2195067760004953e-05,
      "loss": 0.0111,
      "step": 3430
    },
    {
      "epoch": 2.5719640179910046,
      "grad_norm": 0.15851610153186296,
      "learning_rate": 1.2153349654017066e-05,
      "loss": 0.0149,
      "step": 3431
    },
    {
      "epoch": 2.572713643178411,
      "grad_norm": 0.04538796725577889,
      "learning_rate": 1.21116984099242e-05,
      "loss": 0.001,
      "step": 3432
    },
    {
      "epoch": 2.573463268365817,
      "grad_norm": 0.2040980296949387,
      "learning_rate": 1.2070114059427962e-05,
      "loss": 0.0174,
      "step": 3433
    },
    {
      "epoch": 2.5742128935532236,
      "grad_norm": 0.2319504801025156,
      "learning_rate": 1.2028596634179035e-05,
      "loss": 0.0101,
      "step": 3434
    },
    {
      "epoch": 2.5749625187406298,
      "grad_norm": 0.07389525246705174,
      "learning_rate": 1.1987146165777251e-05,
      "loss": 0.0034,
      "step": 3435
    },
    {
      "epoch": 2.575712143928036,
      "grad_norm": 0.04947547613037489,
      "learning_rate": 1.1945762685771367e-05,
      "loss": 0.0023,
      "step": 3436
    },
    {
      "epoch": 2.576461769115442,
      "grad_norm": 0.041088136619714506,
      "learning_rate": 1.1904446225659216e-05,
      "loss": 0.0038,
      "step": 3437
    },
    {
      "epoch": 2.5772113943028487,
      "grad_norm": 0.06362798204591215,
      "learning_rate": 1.1863196816887611e-05,
      "loss": 0.0057,
      "step": 3438
    },
    {
      "epoch": 2.577961019490255,
      "grad_norm": 0.0712690589017783,
      "learning_rate": 1.1822014490852296e-05,
      "loss": 0.0033,
      "step": 3439
    },
    {
      "epoch": 2.578710644677661,
      "grad_norm": 0.06630960067042307,
      "learning_rate": 1.1780899278898027e-05,
      "loss": 0.0007,
      "step": 3440
    },
    {
      "epoch": 2.5794602698650673,
      "grad_norm": 0.032333683634671456,
      "learning_rate": 1.173985121231843e-05,
      "loss": 0.0016,
      "step": 3441
    },
    {
      "epoch": 2.580209895052474,
      "grad_norm": 0.06634983990847639,
      "learning_rate": 1.1698870322356025e-05,
      "loss": 0.0031,
      "step": 3442
    },
    {
      "epoch": 2.58095952023988,
      "grad_norm": 0.08736825173573612,
      "learning_rate": 1.1657956640202217e-05,
      "loss": 0.0062,
      "step": 3443
    },
    {
      "epoch": 2.5817091454272862,
      "grad_norm": 0.0852109705429149,
      "learning_rate": 1.161711019699725e-05,
      "loss": 0.0074,
      "step": 3444
    },
    {
      "epoch": 2.582458770614693,
      "grad_norm": 0.06128883036691688,
      "learning_rate": 1.1576331023830189e-05,
      "loss": 0.0015,
      "step": 3445
    },
    {
      "epoch": 2.583208395802099,
      "grad_norm": 0.08392962163284858,
      "learning_rate": 1.1535619151738907e-05,
      "loss": 0.0042,
      "step": 3446
    },
    {
      "epoch": 2.5839580209895052,
      "grad_norm": 0.12943974098761474,
      "learning_rate": 1.1494974611710052e-05,
      "loss": 0.0065,
      "step": 3447
    },
    {
      "epoch": 2.5847076461769114,
      "grad_norm": 0.11254286919409974,
      "learning_rate": 1.1454397434679021e-05,
      "loss": 0.0058,
      "step": 3448
    },
    {
      "epoch": 2.5854572713643176,
      "grad_norm": 0.11485343113759286,
      "learning_rate": 1.141388765152992e-05,
      "loss": 0.0029,
      "step": 3449
    },
    {
      "epoch": 2.586206896551724,
      "grad_norm": 0.13626055755279318,
      "learning_rate": 1.1373445293095609e-05,
      "loss": 0.0181,
      "step": 3450
    },
    {
      "epoch": 2.5869565217391304,
      "grad_norm": 0.0631198928363645,
      "learning_rate": 1.1333070390157575e-05,
      "loss": 0.0037,
      "step": 3451
    },
    {
      "epoch": 2.5877061469265366,
      "grad_norm": 0.10345333813257934,
      "learning_rate": 1.1292762973446003e-05,
      "loss": 0.0067,
      "step": 3452
    },
    {
      "epoch": 2.588455772113943,
      "grad_norm": 0.04474885996453497,
      "learning_rate": 1.1252523073639686e-05,
      "loss": 0.0026,
      "step": 3453
    },
    {
      "epoch": 2.5892053973013494,
      "grad_norm": 0.2113317755158727,
      "learning_rate": 1.1212350721366061e-05,
      "loss": 0.0035,
      "step": 3454
    },
    {
      "epoch": 2.5899550224887555,
      "grad_norm": 0.4604011176462181,
      "learning_rate": 1.117224594720111e-05,
      "loss": 0.0249,
      "step": 3455
    },
    {
      "epoch": 2.590704647676162,
      "grad_norm": 0.08213204327434842,
      "learning_rate": 1.1132208781669417e-05,
      "loss": 0.002,
      "step": 3456
    },
    {
      "epoch": 2.5914542728635683,
      "grad_norm": 0.07210620882306176,
      "learning_rate": 1.1092239255244085e-05,
      "loss": 0.0047,
      "step": 3457
    },
    {
      "epoch": 2.5922038980509745,
      "grad_norm": 0.11264038850928215,
      "learning_rate": 1.1052337398346724e-05,
      "loss": 0.007,
      "step": 3458
    },
    {
      "epoch": 2.5929535232383807,
      "grad_norm": 0.03105684997056953,
      "learning_rate": 1.1012503241347505e-05,
      "loss": 0.001,
      "step": 3459
    },
    {
      "epoch": 2.593703148425787,
      "grad_norm": 0.12112129065732546,
      "learning_rate": 1.097273681456501e-05,
      "loss": 0.0074,
      "step": 3460
    },
    {
      "epoch": 2.5944527736131935,
      "grad_norm": 0.11672808096353475,
      "learning_rate": 1.0933038148266273e-05,
      "loss": 0.006,
      "step": 3461
    },
    {
      "epoch": 2.5952023988005997,
      "grad_norm": 0.04661222628633944,
      "learning_rate": 1.0893407272666755e-05,
      "loss": 0.0023,
      "step": 3462
    },
    {
      "epoch": 2.595952023988006,
      "grad_norm": 0.04318285411314425,
      "learning_rate": 1.0853844217930364e-05,
      "loss": 0.0022,
      "step": 3463
    },
    {
      "epoch": 2.5967016491754125,
      "grad_norm": 0.15642717583329105,
      "learning_rate": 1.0814349014169345e-05,
      "loss": 0.0083,
      "step": 3464
    },
    {
      "epoch": 2.5974512743628186,
      "grad_norm": 0.1978976408685472,
      "learning_rate": 1.0774921691444318e-05,
      "loss": 0.0174,
      "step": 3465
    },
    {
      "epoch": 2.598200899550225,
      "grad_norm": 0.0659862221373999,
      "learning_rate": 1.07355622797642e-05,
      "loss": 0.003,
      "step": 3466
    },
    {
      "epoch": 2.5989505247376314,
      "grad_norm": 0.09980610197149213,
      "learning_rate": 1.0696270809086284e-05,
      "loss": 0.0091,
      "step": 3467
    },
    {
      "epoch": 2.5997001499250376,
      "grad_norm": 0.07549789682372308,
      "learning_rate": 1.0657047309316093e-05,
      "loss": 0.0043,
      "step": 3468
    },
    {
      "epoch": 2.600449775112444,
      "grad_norm": 0.09400871513513802,
      "learning_rate": 1.0617891810307456e-05,
      "loss": 0.0016,
      "step": 3469
    },
    {
      "epoch": 2.60119940029985,
      "grad_norm": 0.07409113774820665,
      "learning_rate": 1.0578804341862425e-05,
      "loss": 0.003,
      "step": 3470
    },
    {
      "epoch": 2.601949025487256,
      "grad_norm": 0.11373640219809003,
      "learning_rate": 1.0539784933731267e-05,
      "loss": 0.0039,
      "step": 3471
    },
    {
      "epoch": 2.6026986506746628,
      "grad_norm": 0.09964998752157113,
      "learning_rate": 1.0500833615612482e-05,
      "loss": 0.0072,
      "step": 3472
    },
    {
      "epoch": 2.603448275862069,
      "grad_norm": 0.030168470320950023,
      "learning_rate": 1.0461950417152699e-05,
      "loss": 0.001,
      "step": 3473
    },
    {
      "epoch": 2.604197901049475,
      "grad_norm": 0.06410917881047125,
      "learning_rate": 1.0423135367946724e-05,
      "loss": 0.0017,
      "step": 3474
    },
    {
      "epoch": 2.6049475262368817,
      "grad_norm": 0.10069157005767379,
      "learning_rate": 1.0384388497537501e-05,
      "loss": 0.0076,
      "step": 3475
    },
    {
      "epoch": 2.605697151424288,
      "grad_norm": 0.10216851017070903,
      "learning_rate": 1.034570983541605e-05,
      "loss": 0.0052,
      "step": 3476
    },
    {
      "epoch": 2.606446776611694,
      "grad_norm": 0.05589678656898697,
      "learning_rate": 1.0307099411021525e-05,
      "loss": 0.0031,
      "step": 3477
    },
    {
      "epoch": 2.6071964017991007,
      "grad_norm": 0.03938018053569718,
      "learning_rate": 1.0268557253741107e-05,
      "loss": 0.0027,
      "step": 3478
    },
    {
      "epoch": 2.607946026986507,
      "grad_norm": 0.09995308721093468,
      "learning_rate": 1.0230083392910006e-05,
      "loss": 0.0114,
      "step": 3479
    },
    {
      "epoch": 2.608695652173913,
      "grad_norm": 0.09674423576429927,
      "learning_rate": 1.0191677857811499e-05,
      "loss": 0.0048,
      "step": 3480
    },
    {
      "epoch": 2.6094452773613193,
      "grad_norm": 0.06263119023620396,
      "learning_rate": 1.0153340677676781e-05,
      "loss": 0.0055,
      "step": 3481
    },
    {
      "epoch": 2.6101949025487254,
      "grad_norm": 0.09036050664429081,
      "learning_rate": 1.0115071881685134e-05,
      "loss": 0.0047,
      "step": 3482
    },
    {
      "epoch": 2.610944527736132,
      "grad_norm": 0.11983979358718444,
      "learning_rate": 1.0076871498963691e-05,
      "loss": 0.0043,
      "step": 3483
    },
    {
      "epoch": 2.6116941529235382,
      "grad_norm": 0.045515194585739716,
      "learning_rate": 1.0038739558587562e-05,
      "loss": 0.0036,
      "step": 3484
    },
    {
      "epoch": 2.6124437781109444,
      "grad_norm": 0.08646768960077883,
      "learning_rate": 1.000067608957973e-05,
      "loss": 0.0035,
      "step": 3485
    },
    {
      "epoch": 2.613193403298351,
      "grad_norm": 0.33876013907706143,
      "learning_rate": 9.96268112091111e-06,
      "loss": 0.013,
      "step": 3486
    },
    {
      "epoch": 2.613943028485757,
      "grad_norm": 0.04606817630370628,
      "learning_rate": 9.924754681500459e-06,
      "loss": 0.0031,
      "step": 3487
    },
    {
      "epoch": 2.6146926536731634,
      "grad_norm": 0.1435363255689042,
      "learning_rate": 9.88689680021435e-06,
      "loss": 0.0035,
      "step": 3488
    },
    {
      "epoch": 2.6154422788605696,
      "grad_norm": 0.09134904091137633,
      "learning_rate": 9.849107505867205e-06,
      "loss": 0.0032,
      "step": 3489
    },
    {
      "epoch": 2.6161919040479757,
      "grad_norm": 0.056743738916346556,
      "learning_rate": 9.811386827221247e-06,
      "loss": 0.002,
      "step": 3490
    },
    {
      "epoch": 2.6169415292353824,
      "grad_norm": 0.04951286923208861,
      "learning_rate": 9.773734792986434e-06,
      "loss": 0.0023,
      "step": 3491
    },
    {
      "epoch": 2.6176911544227885,
      "grad_norm": 0.09584100347231342,
      "learning_rate": 9.736151431820528e-06,
      "loss": 0.0028,
      "step": 3492
    },
    {
      "epoch": 2.6184407796101947,
      "grad_norm": 0.12609103636803198,
      "learning_rate": 9.698636772328996e-06,
      "loss": 0.005,
      "step": 3493
    },
    {
      "epoch": 2.6191904047976013,
      "grad_norm": 0.2714084765268302,
      "learning_rate": 9.661190843065004e-06,
      "loss": 0.1229,
      "step": 3494
    },
    {
      "epoch": 2.6199400299850075,
      "grad_norm": 0.06185219904845644,
      "learning_rate": 9.623813672529435e-06,
      "loss": 0.0019,
      "step": 3495
    },
    {
      "epoch": 2.6206896551724137,
      "grad_norm": 0.2980290546335867,
      "learning_rate": 9.586505289170811e-06,
      "loss": 0.0156,
      "step": 3496
    },
    {
      "epoch": 2.6214392803598203,
      "grad_norm": 0.09600426776503927,
      "learning_rate": 9.549265721385326e-06,
      "loss": 0.0047,
      "step": 3497
    },
    {
      "epoch": 2.6221889055472265,
      "grad_norm": 0.07846840703663327,
      "learning_rate": 9.512094997516763e-06,
      "loss": 0.0033,
      "step": 3498
    },
    {
      "epoch": 2.6229385307346327,
      "grad_norm": 0.09452717454419779,
      "learning_rate": 9.474993145856548e-06,
      "loss": 0.0078,
      "step": 3499
    },
    {
      "epoch": 2.623688155922039,
      "grad_norm": 0.021453022700013084,
      "learning_rate": 9.437960194643647e-06,
      "loss": 0.0009,
      "step": 3500
    },
    {
      "epoch": 2.624437781109445,
      "grad_norm": 0.08020110491333328,
      "learning_rate": 9.400996172064602e-06,
      "loss": 0.0036,
      "step": 3501
    },
    {
      "epoch": 2.6251874062968517,
      "grad_norm": 0.07715720284674087,
      "learning_rate": 9.364101106253509e-06,
      "loss": 0.0027,
      "step": 3502
    },
    {
      "epoch": 2.625937031484258,
      "grad_norm": 0.06303706939805435,
      "learning_rate": 9.327275025291948e-06,
      "loss": 0.0054,
      "step": 3503
    },
    {
      "epoch": 2.626686656671664,
      "grad_norm": 0.06986054148862804,
      "learning_rate": 9.290517957208988e-06,
      "loss": 0.0025,
      "step": 3504
    },
    {
      "epoch": 2.6274362818590706,
      "grad_norm": 0.2573435063825212,
      "learning_rate": 9.25382992998124e-06,
      "loss": 0.0096,
      "step": 3505
    },
    {
      "epoch": 2.628185907046477,
      "grad_norm": 0.07567213887903118,
      "learning_rate": 9.217210971532685e-06,
      "loss": 0.0058,
      "step": 3506
    },
    {
      "epoch": 2.628935532233883,
      "grad_norm": 0.22383937647866028,
      "learning_rate": 9.180661109734779e-06,
      "loss": 0.0134,
      "step": 3507
    },
    {
      "epoch": 2.6296851574212896,
      "grad_norm": 0.06378477883975421,
      "learning_rate": 9.144180372406341e-06,
      "loss": 0.0035,
      "step": 3508
    },
    {
      "epoch": 2.630434782608696,
      "grad_norm": 0.06176310498690345,
      "learning_rate": 9.107768787313664e-06,
      "loss": 0.0052,
      "step": 3509
    },
    {
      "epoch": 2.631184407796102,
      "grad_norm": 0.06780408252735382,
      "learning_rate": 9.071426382170334e-06,
      "loss": 0.0052,
      "step": 3510
    },
    {
      "epoch": 2.631934032983508,
      "grad_norm": 0.0754928588054794,
      "learning_rate": 9.03515318463729e-06,
      "loss": 0.007,
      "step": 3511
    },
    {
      "epoch": 2.6326836581709143,
      "grad_norm": 0.2771991765187472,
      "learning_rate": 8.998949222322828e-06,
      "loss": 0.0067,
      "step": 3512
    },
    {
      "epoch": 2.633433283358321,
      "grad_norm": 0.16520195859466377,
      "learning_rate": 8.962814522782514e-06,
      "loss": 0.0138,
      "step": 3513
    },
    {
      "epoch": 2.634182908545727,
      "grad_norm": 0.18707832243954112,
      "learning_rate": 8.926749113519228e-06,
      "loss": 0.0142,
      "step": 3514
    },
    {
      "epoch": 2.6349325337331333,
      "grad_norm": 0.12112654939224787,
      "learning_rate": 8.890753021983078e-06,
      "loss": 0.0062,
      "step": 3515
    },
    {
      "epoch": 2.63568215892054,
      "grad_norm": 0.09972947468133485,
      "learning_rate": 8.854826275571449e-06,
      "loss": 0.0047,
      "step": 3516
    },
    {
      "epoch": 2.636431784107946,
      "grad_norm": 0.05652703877587957,
      "learning_rate": 8.81896890162892e-06,
      "loss": 0.0032,
      "step": 3517
    },
    {
      "epoch": 2.6371814092953523,
      "grad_norm": 0.09734359585560508,
      "learning_rate": 8.783180927447265e-06,
      "loss": 0.0042,
      "step": 3518
    },
    {
      "epoch": 2.637931034482759,
      "grad_norm": 0.19651972601960688,
      "learning_rate": 8.74746238026547e-06,
      "loss": 0.0075,
      "step": 3519
    },
    {
      "epoch": 2.638680659670165,
      "grad_norm": 0.11666043079140781,
      "learning_rate": 8.711813287269644e-06,
      "loss": 0.0045,
      "step": 3520
    },
    {
      "epoch": 2.6394302848575713,
      "grad_norm": 0.10923156443553575,
      "learning_rate": 8.676233675593037e-06,
      "loss": 0.0056,
      "step": 3521
    },
    {
      "epoch": 2.6401799100449774,
      "grad_norm": 0.04615933797824494,
      "learning_rate": 8.640723572316045e-06,
      "loss": 0.0018,
      "step": 3522
    },
    {
      "epoch": 2.6409295352323836,
      "grad_norm": 0.12348983813140332,
      "learning_rate": 8.605283004466136e-06,
      "loss": 0.011,
      "step": 3523
    },
    {
      "epoch": 2.6416791604197902,
      "grad_norm": 0.1323678653865999,
      "learning_rate": 8.569911999017832e-06,
      "loss": 0.0048,
      "step": 3524
    },
    {
      "epoch": 2.6424287856071964,
      "grad_norm": 0.08003696554324499,
      "learning_rate": 8.534610582892765e-06,
      "loss": 0.0034,
      "step": 3525
    },
    {
      "epoch": 2.6431784107946026,
      "grad_norm": 0.1904035214537228,
      "learning_rate": 8.499378782959555e-06,
      "loss": 0.0103,
      "step": 3526
    },
    {
      "epoch": 2.643928035982009,
      "grad_norm": 0.15545593523005102,
      "learning_rate": 8.46421662603385e-06,
      "loss": 0.0193,
      "step": 3527
    },
    {
      "epoch": 2.6446776611694154,
      "grad_norm": 0.07256795232832776,
      "learning_rate": 8.429124138878274e-06,
      "loss": 0.004,
      "step": 3528
    },
    {
      "epoch": 2.6454272863568216,
      "grad_norm": 0.04997265242737797,
      "learning_rate": 8.394101348202477e-06,
      "loss": 0.0026,
      "step": 3529
    },
    {
      "epoch": 2.6461769115442277,
      "grad_norm": 0.09839523717548251,
      "learning_rate": 8.359148280663021e-06,
      "loss": 0.0021,
      "step": 3530
    },
    {
      "epoch": 2.646926536731634,
      "grad_norm": 0.2272177074499261,
      "learning_rate": 8.324264962863392e-06,
      "loss": 0.0375,
      "step": 3531
    },
    {
      "epoch": 2.6476761619190405,
      "grad_norm": 0.05316402736053975,
      "learning_rate": 8.289451421353978e-06,
      "loss": 0.0031,
      "step": 3532
    },
    {
      "epoch": 2.6484257871064467,
      "grad_norm": 0.09666760826592079,
      "learning_rate": 8.254707682632134e-06,
      "loss": 0.0044,
      "step": 3533
    },
    {
      "epoch": 2.649175412293853,
      "grad_norm": 0.08548886953032514,
      "learning_rate": 8.220033773142023e-06,
      "loss": 0.0031,
      "step": 3534
    },
    {
      "epoch": 2.6499250374812595,
      "grad_norm": 0.055433019245836296,
      "learning_rate": 8.18542971927464e-06,
      "loss": 0.0037,
      "step": 3535
    },
    {
      "epoch": 2.6506746626686657,
      "grad_norm": 0.06702626082211281,
      "learning_rate": 8.150895547367876e-06,
      "loss": 0.0028,
      "step": 3536
    },
    {
      "epoch": 2.651424287856072,
      "grad_norm": 0.08834920550221038,
      "learning_rate": 8.116431283706383e-06,
      "loss": 0.0056,
      "step": 3537
    },
    {
      "epoch": 2.6521739130434785,
      "grad_norm": 0.07939187196088908,
      "learning_rate": 8.082036954521621e-06,
      "loss": 0.0082,
      "step": 3538
    },
    {
      "epoch": 2.6529235382308847,
      "grad_norm": 0.0693663151265394,
      "learning_rate": 8.04771258599184e-06,
      "loss": 0.0036,
      "step": 3539
    },
    {
      "epoch": 2.653673163418291,
      "grad_norm": 0.0489002340147583,
      "learning_rate": 8.013458204242007e-06,
      "loss": 0.0027,
      "step": 3540
    },
    {
      "epoch": 2.654422788605697,
      "grad_norm": 0.06117963549083394,
      "learning_rate": 7.979273835343836e-06,
      "loss": 0.0029,
      "step": 3541
    },
    {
      "epoch": 2.655172413793103,
      "grad_norm": 0.17285850209634904,
      "learning_rate": 7.94515950531577e-06,
      "loss": 0.0111,
      "step": 3542
    },
    {
      "epoch": 2.65592203898051,
      "grad_norm": 0.07674916468104553,
      "learning_rate": 7.911115240122913e-06,
      "loss": 0.0084,
      "step": 3543
    },
    {
      "epoch": 2.656671664167916,
      "grad_norm": 0.06939015226317331,
      "learning_rate": 7.877141065677063e-06,
      "loss": 0.003,
      "step": 3544
    },
    {
      "epoch": 2.657421289355322,
      "grad_norm": 0.03697970239963438,
      "learning_rate": 7.843237007836657e-06,
      "loss": 0.0019,
      "step": 3545
    },
    {
      "epoch": 2.658170914542729,
      "grad_norm": 0.10161008931335178,
      "learning_rate": 7.809403092406787e-06,
      "loss": 0.0039,
      "step": 3546
    },
    {
      "epoch": 2.658920539730135,
      "grad_norm": 0.05432572136277282,
      "learning_rate": 7.77563934513913e-06,
      "loss": 0.0054,
      "step": 3547
    },
    {
      "epoch": 2.659670164917541,
      "grad_norm": 0.09434216477553212,
      "learning_rate": 7.741945791731975e-06,
      "loss": 0.0043,
      "step": 3548
    },
    {
      "epoch": 2.660419790104948,
      "grad_norm": 0.13438109952231186,
      "learning_rate": 7.708322457830175e-06,
      "loss": 0.0149,
      "step": 3549
    },
    {
      "epoch": 2.661169415292354,
      "grad_norm": 0.044515473296177166,
      "learning_rate": 7.674769369025148e-06,
      "loss": 0.0017,
      "step": 3550
    },
    {
      "epoch": 2.66191904047976,
      "grad_norm": 0.07980519638723763,
      "learning_rate": 7.64128655085482e-06,
      "loss": 0.0027,
      "step": 3551
    },
    {
      "epoch": 2.6626686656671663,
      "grad_norm": 0.09115039728168225,
      "learning_rate": 7.607874028803685e-06,
      "loss": 0.0049,
      "step": 3552
    },
    {
      "epoch": 2.6634182908545725,
      "grad_norm": 0.11713526622844393,
      "learning_rate": 7.5745318283027e-06,
      "loss": 0.0041,
      "step": 3553
    },
    {
      "epoch": 2.664167916041979,
      "grad_norm": 0.10597277800023548,
      "learning_rate": 7.541259974729276e-06,
      "loss": 0.0083,
      "step": 3554
    },
    {
      "epoch": 2.6649175412293853,
      "grad_norm": 0.09684578884196773,
      "learning_rate": 7.5080584934073105e-06,
      "loss": 0.0058,
      "step": 3555
    },
    {
      "epoch": 2.6656671664167915,
      "grad_norm": 0.04135982192021682,
      "learning_rate": 7.4749274096071575e-06,
      "loss": 0.002,
      "step": 3556
    },
    {
      "epoch": 2.666416791604198,
      "grad_norm": 0.06499780333835134,
      "learning_rate": 7.441866748545545e-06,
      "loss": 0.0034,
      "step": 3557
    },
    {
      "epoch": 2.6671664167916043,
      "grad_norm": 0.06255583464347211,
      "learning_rate": 7.408876535385645e-06,
      "loss": 0.0078,
      "step": 3558
    },
    {
      "epoch": 2.6679160419790104,
      "grad_norm": 0.11011116172772083,
      "learning_rate": 7.37595679523696e-06,
      "loss": 0.0042,
      "step": 3559
    },
    {
      "epoch": 2.668665667166417,
      "grad_norm": 0.0722927183692007,
      "learning_rate": 7.343107553155404e-06,
      "loss": 0.0065,
      "step": 3560
    },
    {
      "epoch": 2.6694152923538232,
      "grad_norm": 0.0678410468157105,
      "learning_rate": 7.31032883414321e-06,
      "loss": 0.0066,
      "step": 3561
    },
    {
      "epoch": 2.6701649175412294,
      "grad_norm": 0.058365571482256254,
      "learning_rate": 7.277620663148921e-06,
      "loss": 0.0025,
      "step": 3562
    },
    {
      "epoch": 2.6709145427286356,
      "grad_norm": 0.08197814793200882,
      "learning_rate": 7.244983065067412e-06,
      "loss": 0.0029,
      "step": 3563
    },
    {
      "epoch": 2.6716641679160418,
      "grad_norm": 0.18926405575286565,
      "learning_rate": 7.2124160647398354e-06,
      "loss": 0.0097,
      "step": 3564
    },
    {
      "epoch": 2.6724137931034484,
      "grad_norm": 0.06395203764121683,
      "learning_rate": 7.179919686953585e-06,
      "loss": 0.0042,
      "step": 3565
    },
    {
      "epoch": 2.6731634182908546,
      "grad_norm": 0.07331177755241726,
      "learning_rate": 7.147493956442353e-06,
      "loss": 0.0032,
      "step": 3566
    },
    {
      "epoch": 2.6739130434782608,
      "grad_norm": 0.4873608362749581,
      "learning_rate": 7.115138897886009e-06,
      "loss": 0.0245,
      "step": 3567
    },
    {
      "epoch": 2.6746626686656674,
      "grad_norm": 0.17374373823700442,
      "learning_rate": 7.082854535910655e-06,
      "loss": 0.0061,
      "step": 3568
    },
    {
      "epoch": 2.6754122938530736,
      "grad_norm": 0.08369814478199004,
      "learning_rate": 7.050640895088589e-06,
      "loss": 0.0067,
      "step": 3569
    },
    {
      "epoch": 2.6761619190404797,
      "grad_norm": 0.1308839941432331,
      "learning_rate": 7.01849799993829e-06,
      "loss": 0.0051,
      "step": 3570
    },
    {
      "epoch": 2.676911544227886,
      "grad_norm": 0.12482607476795445,
      "learning_rate": 6.986425874924352e-06,
      "loss": 0.0076,
      "step": 3571
    },
    {
      "epoch": 2.677661169415292,
      "grad_norm": 0.0895789380819289,
      "learning_rate": 6.954424544457549e-06,
      "loss": 0.0037,
      "step": 3572
    },
    {
      "epoch": 2.6784107946026987,
      "grad_norm": 0.23901158055522503,
      "learning_rate": 6.922494032894744e-06,
      "loss": 0.0092,
      "step": 3573
    },
    {
      "epoch": 2.679160419790105,
      "grad_norm": 0.05298578704840303,
      "learning_rate": 6.8906343645388945e-06,
      "loss": 0.0008,
      "step": 3574
    },
    {
      "epoch": 2.679910044977511,
      "grad_norm": 0.09438117470957384,
      "learning_rate": 6.858845563639083e-06,
      "loss": 0.0053,
      "step": 3575
    },
    {
      "epoch": 2.6806596701649177,
      "grad_norm": 0.08681760628142808,
      "learning_rate": 6.8271276543903975e-06,
      "loss": 0.0022,
      "step": 3576
    },
    {
      "epoch": 2.681409295352324,
      "grad_norm": 0.06474070177166225,
      "learning_rate": 6.7954806609339946e-06,
      "loss": 0.006,
      "step": 3577
    },
    {
      "epoch": 2.68215892053973,
      "grad_norm": 0.07631249932565169,
      "learning_rate": 6.763904607357063e-06,
      "loss": 0.0034,
      "step": 3578
    },
    {
      "epoch": 2.6829085457271367,
      "grad_norm": 0.15131390870827074,
      "learning_rate": 6.732399517692778e-06,
      "loss": 0.0086,
      "step": 3579
    },
    {
      "epoch": 2.683658170914543,
      "grad_norm": 0.12343656877439753,
      "learning_rate": 6.700965415920302e-06,
      "loss": 0.0072,
      "step": 3580
    },
    {
      "epoch": 2.684407796101949,
      "grad_norm": 0.02283778423621663,
      "learning_rate": 6.669602325964796e-06,
      "loss": 0.0015,
      "step": 3581
    },
    {
      "epoch": 2.685157421289355,
      "grad_norm": 0.09042680601212649,
      "learning_rate": 6.6383102716973564e-06,
      "loss": 0.0058,
      "step": 3582
    },
    {
      "epoch": 2.6859070464767614,
      "grad_norm": 0.10654143512759034,
      "learning_rate": 6.607089276934997e-06,
      "loss": 0.0034,
      "step": 3583
    },
    {
      "epoch": 2.686656671664168,
      "grad_norm": 0.008937713892268067,
      "learning_rate": 6.575939365440664e-06,
      "loss": 0.0003,
      "step": 3584
    },
    {
      "epoch": 2.687406296851574,
      "grad_norm": 0.08699528259742419,
      "learning_rate": 6.544860560923205e-06,
      "loss": 0.0117,
      "step": 3585
    },
    {
      "epoch": 2.6881559220389803,
      "grad_norm": 0.04413936298947285,
      "learning_rate": 6.51385288703732e-06,
      "loss": 0.0044,
      "step": 3586
    },
    {
      "epoch": 2.688905547226387,
      "grad_norm": 0.40494077939008205,
      "learning_rate": 6.482916367383596e-06,
      "loss": 0.0185,
      "step": 3587
    },
    {
      "epoch": 2.689655172413793,
      "grad_norm": 0.025317057448168645,
      "learning_rate": 6.4520510255084655e-06,
      "loss": 0.001,
      "step": 3588
    },
    {
      "epoch": 2.6904047976011993,
      "grad_norm": 0.1281661479079944,
      "learning_rate": 6.42125688490417e-06,
      "loss": 0.0047,
      "step": 3589
    },
    {
      "epoch": 2.691154422788606,
      "grad_norm": 0.14593951083085305,
      "learning_rate": 6.390533969008761e-06,
      "loss": 0.01,
      "step": 3590
    },
    {
      "epoch": 2.691904047976012,
      "grad_norm": 0.1504086206960067,
      "learning_rate": 6.359882301206077e-06,
      "loss": 0.0062,
      "step": 3591
    },
    {
      "epoch": 2.6926536731634183,
      "grad_norm": 0.03789441124058746,
      "learning_rate": 6.329301904825746e-06,
      "loss": 0.0013,
      "step": 3592
    },
    {
      "epoch": 2.6934032983508245,
      "grad_norm": 0.12908249338770855,
      "learning_rate": 6.298792803143138e-06,
      "loss": 0.0108,
      "step": 3593
    },
    {
      "epoch": 2.6941529235382307,
      "grad_norm": 0.2053287462164705,
      "learning_rate": 6.268355019379346e-06,
      "loss": 0.0196,
      "step": 3594
    },
    {
      "epoch": 2.6949025487256373,
      "grad_norm": 0.20678108268250486,
      "learning_rate": 6.237988576701204e-06,
      "loss": 0.0054,
      "step": 3595
    },
    {
      "epoch": 2.6956521739130435,
      "grad_norm": 0.0540984851276832,
      "learning_rate": 6.207693498221235e-06,
      "loss": 0.0038,
      "step": 3596
    },
    {
      "epoch": 2.6964017991004496,
      "grad_norm": 0.09401306330490736,
      "learning_rate": 6.177469806997627e-06,
      "loss": 0.0025,
      "step": 3597
    },
    {
      "epoch": 2.6971514242878563,
      "grad_norm": 0.04383015446253963,
      "learning_rate": 6.147317526034291e-06,
      "loss": 0.001,
      "step": 3598
    },
    {
      "epoch": 2.6979010494752624,
      "grad_norm": 0.024997535985464992,
      "learning_rate": 6.1172366782807354e-06,
      "loss": 0.0009,
      "step": 3599
    },
    {
      "epoch": 2.6986506746626686,
      "grad_norm": 0.0469201721239403,
      "learning_rate": 6.087227286632103e-06,
      "loss": 0.0014,
      "step": 3600
    },
    {
      "epoch": 2.6994002998500752,
      "grad_norm": 0.051395596885820925,
      "learning_rate": 6.0572893739291756e-06,
      "loss": 0.0025,
      "step": 3601
    },
    {
      "epoch": 2.7001499250374814,
      "grad_norm": 0.045749901122774626,
      "learning_rate": 6.027422962958296e-06,
      "loss": 0.0016,
      "step": 3602
    },
    {
      "epoch": 2.7008995502248876,
      "grad_norm": 0.12403820171571336,
      "learning_rate": 5.9976280764514116e-06,
      "loss": 0.0098,
      "step": 3603
    },
    {
      "epoch": 2.7016491754122938,
      "grad_norm": 0.07200283473771944,
      "learning_rate": 5.967904737086016e-06,
      "loss": 0.0049,
      "step": 3604
    },
    {
      "epoch": 2.7023988005997,
      "grad_norm": 0.07568877846311735,
      "learning_rate": 5.938252967485192e-06,
      "loss": 0.0035,
      "step": 3605
    },
    {
      "epoch": 2.7031484257871066,
      "grad_norm": 0.08252827999998712,
      "learning_rate": 5.908672790217474e-06,
      "loss": 0.0077,
      "step": 3606
    },
    {
      "epoch": 2.7038980509745127,
      "grad_norm": 0.17930247189997617,
      "learning_rate": 5.879164227796963e-06,
      "loss": 0.0153,
      "step": 3607
    },
    {
      "epoch": 2.704647676161919,
      "grad_norm": 0.045819957196655735,
      "learning_rate": 5.849727302683217e-06,
      "loss": 0.0028,
      "step": 3608
    },
    {
      "epoch": 2.7053973013493255,
      "grad_norm": 0.1472539901727524,
      "learning_rate": 5.820362037281302e-06,
      "loss": 0.004,
      "step": 3609
    },
    {
      "epoch": 2.7061469265367317,
      "grad_norm": 0.0571665163793095,
      "learning_rate": 5.791068453941728e-06,
      "loss": 0.0021,
      "step": 3610
    },
    {
      "epoch": 2.706896551724138,
      "grad_norm": 0.12266217583120254,
      "learning_rate": 5.761846574960428e-06,
      "loss": 0.0067,
      "step": 3611
    },
    {
      "epoch": 2.707646176911544,
      "grad_norm": 0.15635234506738496,
      "learning_rate": 5.732696422578787e-06,
      "loss": 0.0089,
      "step": 3612
    },
    {
      "epoch": 2.7083958020989503,
      "grad_norm": 0.24646222619048166,
      "learning_rate": 5.70361801898358e-06,
      "loss": 0.007,
      "step": 3613
    },
    {
      "epoch": 2.709145427286357,
      "grad_norm": 0.09446261697690682,
      "learning_rate": 5.674611386306994e-06,
      "loss": 0.0055,
      "step": 3614
    },
    {
      "epoch": 2.709895052473763,
      "grad_norm": 0.05438453674274307,
      "learning_rate": 5.645676546626555e-06,
      "loss": 0.0021,
      "step": 3615
    },
    {
      "epoch": 2.7106446776611692,
      "grad_norm": 0.08994894529395658,
      "learning_rate": 5.616813521965181e-06,
      "loss": 0.0055,
      "step": 3616
    },
    {
      "epoch": 2.711394302848576,
      "grad_norm": 0.17303568276679598,
      "learning_rate": 5.58802233429111e-06,
      "loss": 0.0095,
      "step": 3617
    },
    {
      "epoch": 2.712143928035982,
      "grad_norm": 0.17064533406947632,
      "learning_rate": 5.5593030055179014e-06,
      "loss": 0.0048,
      "step": 3618
    },
    {
      "epoch": 2.712893553223388,
      "grad_norm": 0.0578184933228903,
      "learning_rate": 5.5306555575044385e-06,
      "loss": 0.0041,
      "step": 3619
    },
    {
      "epoch": 2.713643178410795,
      "grad_norm": 0.07165755854478759,
      "learning_rate": 5.502080012054867e-06,
      "loss": 0.0026,
      "step": 3620
    },
    {
      "epoch": 2.714392803598201,
      "grad_norm": 0.08253336755974179,
      "learning_rate": 5.473576390918622e-06,
      "loss": 0.0042,
      "step": 3621
    },
    {
      "epoch": 2.715142428785607,
      "grad_norm": 0.38558400392740255,
      "learning_rate": 5.4451447157904286e-06,
      "loss": 0.0095,
      "step": 3622
    },
    {
      "epoch": 2.7158920539730134,
      "grad_norm": 0.11407609591189503,
      "learning_rate": 5.416785008310199e-06,
      "loss": 0.0084,
      "step": 3623
    },
    {
      "epoch": 2.7166416791604195,
      "grad_norm": 0.11103306738736951,
      "learning_rate": 5.388497290063099e-06,
      "loss": 0.0024,
      "step": 3624
    },
    {
      "epoch": 2.717391304347826,
      "grad_norm": 0.06833840067269992,
      "learning_rate": 5.3602815825794735e-06,
      "loss": 0.0036,
      "step": 3625
    },
    {
      "epoch": 2.7181409295352323,
      "grad_norm": 0.0714532486964811,
      "learning_rate": 5.3321379073349e-06,
      "loss": 0.004,
      "step": 3626
    },
    {
      "epoch": 2.7188905547226385,
      "grad_norm": 0.18719531064096853,
      "learning_rate": 5.304066285750109e-06,
      "loss": 0.0218,
      "step": 3627
    },
    {
      "epoch": 2.719640179910045,
      "grad_norm": 0.1206736823435074,
      "learning_rate": 5.276066739190955e-06,
      "loss": 0.0095,
      "step": 3628
    },
    {
      "epoch": 2.7203898050974513,
      "grad_norm": 0.13394479538875573,
      "learning_rate": 5.248139288968523e-06,
      "loss": 0.0065,
      "step": 3629
    },
    {
      "epoch": 2.7211394302848575,
      "grad_norm": 0.07118915867328775,
      "learning_rate": 5.220283956338934e-06,
      "loss": 0.004,
      "step": 3630
    },
    {
      "epoch": 2.721889055472264,
      "grad_norm": 0.07518397660658808,
      "learning_rate": 5.192500762503472e-06,
      "loss": 0.0021,
      "step": 3631
    },
    {
      "epoch": 2.7226386806596703,
      "grad_norm": 0.13623715597075606,
      "learning_rate": 5.164789728608488e-06,
      "loss": 0.0068,
      "step": 3632
    },
    {
      "epoch": 2.7233883058470765,
      "grad_norm": 0.08135641564886333,
      "learning_rate": 5.137150875745433e-06,
      "loss": 0.0047,
      "step": 3633
    },
    {
      "epoch": 2.7241379310344827,
      "grad_norm": 0.2905574517744519,
      "learning_rate": 5.109584224950792e-06,
      "loss": 0.0129,
      "step": 3634
    },
    {
      "epoch": 2.724887556221889,
      "grad_norm": 0.03198727740319656,
      "learning_rate": 5.082089797206124e-06,
      "loss": 0.0014,
      "step": 3635
    },
    {
      "epoch": 2.7256371814092955,
      "grad_norm": 0.0738532460460911,
      "learning_rate": 5.054667613437991e-06,
      "loss": 0.0105,
      "step": 3636
    },
    {
      "epoch": 2.7263868065967016,
      "grad_norm": 0.15003634894034606,
      "learning_rate": 5.027317694517996e-06,
      "loss": 0.005,
      "step": 3637
    },
    {
      "epoch": 2.727136431784108,
      "grad_norm": 0.2861318864965912,
      "learning_rate": 5.0000400612627114e-06,
      "loss": 0.022,
      "step": 3638
    },
    {
      "epoch": 2.7278860569715144,
      "grad_norm": 0.0830609886318904,
      "learning_rate": 4.972834734433718e-06,
      "loss": 0.01,
      "step": 3639
    },
    {
      "epoch": 2.7286356821589206,
      "grad_norm": 0.3801169987148443,
      "learning_rate": 4.94570173473754e-06,
      "loss": 0.0061,
      "step": 3640
    },
    {
      "epoch": 2.729385307346327,
      "grad_norm": 0.056034639807724315,
      "learning_rate": 4.91864108282567e-06,
      "loss": 0.0036,
      "step": 3641
    },
    {
      "epoch": 2.7301349325337334,
      "grad_norm": 0.08424484116620803,
      "learning_rate": 4.8916527992945325e-06,
      "loss": 0.0033,
      "step": 3642
    },
    {
      "epoch": 2.7308845577211396,
      "grad_norm": 0.09459805879195689,
      "learning_rate": 4.8647369046854515e-06,
      "loss": 0.003,
      "step": 3643
    },
    {
      "epoch": 2.7316341829085458,
      "grad_norm": 0.035342901239077684,
      "learning_rate": 4.837893419484663e-06,
      "loss": 0.001,
      "step": 3644
    },
    {
      "epoch": 2.732383808095952,
      "grad_norm": 0.024040876961602176,
      "learning_rate": 4.811122364123322e-06,
      "loss": 0.001,
      "step": 3645
    },
    {
      "epoch": 2.733133433283358,
      "grad_norm": 0.09592235037490139,
      "learning_rate": 4.78442375897743e-06,
      "loss": 0.0119,
      "step": 3646
    },
    {
      "epoch": 2.7338830584707647,
      "grad_norm": 0.07500277826770634,
      "learning_rate": 4.757797624367821e-06,
      "loss": 0.0027,
      "step": 3647
    },
    {
      "epoch": 2.734632683658171,
      "grad_norm": 0.022403652497716163,
      "learning_rate": 4.731243980560207e-06,
      "loss": 0.0006,
      "step": 3648
    },
    {
      "epoch": 2.735382308845577,
      "grad_norm": 0.0308911192677184,
      "learning_rate": 4.7047628477651095e-06,
      "loss": 0.0009,
      "step": 3649
    },
    {
      "epoch": 2.7361319340329837,
      "grad_norm": 0.04449662376820901,
      "learning_rate": 4.678354246137873e-06,
      "loss": 0.003,
      "step": 3650
    },
    {
      "epoch": 2.73688155922039,
      "grad_norm": 0.1476368913508825,
      "learning_rate": 4.652018195778629e-06,
      "loss": 0.007,
      "step": 3651
    },
    {
      "epoch": 2.737631184407796,
      "grad_norm": 0.11712744485200172,
      "learning_rate": 4.625754716732256e-06,
      "loss": 0.0074,
      "step": 3652
    },
    {
      "epoch": 2.7383808095952022,
      "grad_norm": 0.08772348977368045,
      "learning_rate": 4.5995638289884735e-06,
      "loss": 0.0064,
      "step": 3653
    },
    {
      "epoch": 2.7391304347826084,
      "grad_norm": 0.10833839381144214,
      "learning_rate": 4.573445552481692e-06,
      "loss": 0.0088,
      "step": 3654
    },
    {
      "epoch": 2.739880059970015,
      "grad_norm": 0.11699172964342061,
      "learning_rate": 4.547399907091055e-06,
      "loss": 0.005,
      "step": 3655
    },
    {
      "epoch": 2.7406296851574212,
      "grad_norm": 0.048607452332993795,
      "learning_rate": 4.521426912640458e-06,
      "loss": 0.0012,
      "step": 3656
    },
    {
      "epoch": 2.7413793103448274,
      "grad_norm": 0.1476414516790435,
      "learning_rate": 4.495526588898458e-06,
      "loss": 0.0135,
      "step": 3657
    },
    {
      "epoch": 2.742128935532234,
      "grad_norm": 0.08161599938984887,
      "learning_rate": 4.469698955578349e-06,
      "loss": 0.0031,
      "step": 3658
    },
    {
      "epoch": 2.74287856071964,
      "grad_norm": 0.08333476285401598,
      "learning_rate": 4.443944032338066e-06,
      "loss": 0.0052,
      "step": 3659
    },
    {
      "epoch": 2.7436281859070464,
      "grad_norm": 0.029269837281302242,
      "learning_rate": 4.4182618387802134e-06,
      "loss": 0.001,
      "step": 3660
    },
    {
      "epoch": 2.744377811094453,
      "grad_norm": 0.1003170085626985,
      "learning_rate": 4.392652394452035e-06,
      "loss": 0.0112,
      "step": 3661
    },
    {
      "epoch": 2.745127436281859,
      "grad_norm": 0.11014654771097936,
      "learning_rate": 4.367115718845405e-06,
      "loss": 0.0106,
      "step": 3662
    },
    {
      "epoch": 2.7458770614692654,
      "grad_norm": 0.1790138493348603,
      "learning_rate": 4.341651831396809e-06,
      "loss": 0.0093,
      "step": 3663
    },
    {
      "epoch": 2.7466266866566715,
      "grad_norm": 0.16014756329328614,
      "learning_rate": 4.316260751487356e-06,
      "loss": 0.0114,
      "step": 3664
    },
    {
      "epoch": 2.7473763118440777,
      "grad_norm": 0.08790151629116845,
      "learning_rate": 4.290942498442696e-06,
      "loss": 0.0065,
      "step": 3665
    },
    {
      "epoch": 2.7481259370314843,
      "grad_norm": 0.10020052895833542,
      "learning_rate": 4.265697091533083e-06,
      "loss": 0.007,
      "step": 3666
    },
    {
      "epoch": 2.7488755622188905,
      "grad_norm": 0.07804375760777828,
      "learning_rate": 4.240524549973301e-06,
      "loss": 0.0042,
      "step": 3667
    },
    {
      "epoch": 2.7496251874062967,
      "grad_norm": 0.11710400068698333,
      "learning_rate": 4.215424892922726e-06,
      "loss": 0.0033,
      "step": 3668
    },
    {
      "epoch": 2.7503748125937033,
      "grad_norm": 0.09659294495448248,
      "learning_rate": 4.190398139485196e-06,
      "loss": 0.0061,
      "step": 3669
    },
    {
      "epoch": 2.7511244377811095,
      "grad_norm": 0.10322465262354132,
      "learning_rate": 4.165444308709099e-06,
      "loss": 0.0025,
      "step": 3670
    },
    {
      "epoch": 2.7518740629685157,
      "grad_norm": 0.1150037801522404,
      "learning_rate": 4.1405634195873066e-06,
      "loss": 0.0058,
      "step": 3671
    },
    {
      "epoch": 2.7526236881559223,
      "grad_norm": 0.11439789106912057,
      "learning_rate": 4.115755491057171e-06,
      "loss": 0.0052,
      "step": 3672
    },
    {
      "epoch": 2.7533733133433285,
      "grad_norm": 0.09507224179197715,
      "learning_rate": 4.091020542000534e-06,
      "loss": 0.0114,
      "step": 3673
    },
    {
      "epoch": 2.7541229385307346,
      "grad_norm": 0.1418154647433098,
      "learning_rate": 4.0663585912436465e-06,
      "loss": 0.0091,
      "step": 3674
    },
    {
      "epoch": 2.754872563718141,
      "grad_norm": 0.0519004217806522,
      "learning_rate": 4.041769657557259e-06,
      "loss": 0.0043,
      "step": 3675
    },
    {
      "epoch": 2.755622188905547,
      "grad_norm": 0.1642216620545012,
      "learning_rate": 4.017253759656481e-06,
      "loss": 0.0113,
      "step": 3676
    },
    {
      "epoch": 2.7563718140929536,
      "grad_norm": 0.14499900898011348,
      "learning_rate": 3.992810916200895e-06,
      "loss": 0.0061,
      "step": 3677
    },
    {
      "epoch": 2.75712143928036,
      "grad_norm": 0.0981948755530292,
      "learning_rate": 3.968441145794455e-06,
      "loss": 0.0085,
      "step": 3678
    },
    {
      "epoch": 2.757871064467766,
      "grad_norm": 0.06275409857769353,
      "learning_rate": 3.944144466985477e-06,
      "loss": 0.0028,
      "step": 3679
    },
    {
      "epoch": 2.7586206896551726,
      "grad_norm": 0.06965249096923765,
      "learning_rate": 3.919920898266671e-06,
      "loss": 0.0042,
      "step": 3680
    },
    {
      "epoch": 2.7593703148425788,
      "grad_norm": 0.07969732276284872,
      "learning_rate": 3.895770458075087e-06,
      "loss": 0.0044,
      "step": 3681
    },
    {
      "epoch": 2.760119940029985,
      "grad_norm": 0.03777464572367483,
      "learning_rate": 3.871693164792145e-06,
      "loss": 0.0019,
      "step": 3682
    },
    {
      "epoch": 2.7608695652173916,
      "grad_norm": 0.1295711523465296,
      "learning_rate": 3.84768903674354e-06,
      "loss": 0.0078,
      "step": 3683
    },
    {
      "epoch": 2.7616191904047978,
      "grad_norm": 0.06514779876174903,
      "learning_rate": 3.823758092199325e-06,
      "loss": 0.0054,
      "step": 3684
    },
    {
      "epoch": 2.762368815592204,
      "grad_norm": 0.05547738547510124,
      "learning_rate": 3.7999003493738415e-06,
      "loss": 0.0031,
      "step": 3685
    },
    {
      "epoch": 2.76311844077961,
      "grad_norm": 0.059133497938647625,
      "learning_rate": 3.776115826425686e-06,
      "loss": 0.0018,
      "step": 3686
    },
    {
      "epoch": 2.7638680659670163,
      "grad_norm": 0.034869095838589936,
      "learning_rate": 3.752404541457766e-06,
      "loss": 0.0021,
      "step": 3687
    },
    {
      "epoch": 2.764617691154423,
      "grad_norm": 0.16037864663324225,
      "learning_rate": 3.728766512517223e-06,
      "loss": 0.0039,
      "step": 3688
    },
    {
      "epoch": 2.765367316341829,
      "grad_norm": 0.05774969577903951,
      "learning_rate": 3.7052017575954557e-06,
      "loss": 0.0028,
      "step": 3689
    },
    {
      "epoch": 2.7661169415292353,
      "grad_norm": 0.07922232786939425,
      "learning_rate": 3.6817102946280602e-06,
      "loss": 0.0018,
      "step": 3690
    },
    {
      "epoch": 2.766866566716642,
      "grad_norm": 0.16312725330182026,
      "learning_rate": 3.6582921414949124e-06,
      "loss": 0.0063,
      "step": 3691
    },
    {
      "epoch": 2.767616191904048,
      "grad_norm": 0.19983176773659292,
      "learning_rate": 3.6349473160200207e-06,
      "loss": 0.0115,
      "step": 3692
    },
    {
      "epoch": 2.7683658170914542,
      "grad_norm": 0.03632042805227702,
      "learning_rate": 3.6116758359716173e-06,
      "loss": 0.0009,
      "step": 3693
    },
    {
      "epoch": 2.7691154422788604,
      "grad_norm": 0.8993147264773352,
      "learning_rate": 3.588477719062111e-06,
      "loss": 0.0065,
      "step": 3694
    },
    {
      "epoch": 2.7698650674662666,
      "grad_norm": 0.10458974932419897,
      "learning_rate": 3.565352982948067e-06,
      "loss": 0.0079,
      "step": 3695
    },
    {
      "epoch": 2.770614692653673,
      "grad_norm": 0.05507296047559786,
      "learning_rate": 3.542301645230206e-06,
      "loss": 0.0021,
      "step": 3696
    },
    {
      "epoch": 2.7713643178410794,
      "grad_norm": 0.08673372948135946,
      "learning_rate": 3.5193237234533495e-06,
      "loss": 0.0073,
      "step": 3697
    },
    {
      "epoch": 2.7721139430284856,
      "grad_norm": 0.1187522679105704,
      "learning_rate": 3.4964192351064963e-06,
      "loss": 0.0067,
      "step": 3698
    },
    {
      "epoch": 2.772863568215892,
      "grad_norm": 0.09235170094818261,
      "learning_rate": 3.4735881976227013e-06,
      "loss": 0.0068,
      "step": 3699
    },
    {
      "epoch": 2.7736131934032984,
      "grad_norm": 0.104110530083035,
      "learning_rate": 3.4508306283791423e-06,
      "loss": 0.003,
      "step": 3700
    },
    {
      "epoch": 2.7743628185907045,
      "grad_norm": 0.10901153145720517,
      "learning_rate": 3.428146544697086e-06,
      "loss": 0.0058,
      "step": 3701
    },
    {
      "epoch": 2.775112443778111,
      "grad_norm": 0.15846183030533673,
      "learning_rate": 3.405535963841844e-06,
      "loss": 0.0067,
      "step": 3702
    },
    {
      "epoch": 2.7758620689655173,
      "grad_norm": 0.26506094090563465,
      "learning_rate": 3.382998903022816e-06,
      "loss": 0.0102,
      "step": 3703
    },
    {
      "epoch": 2.7766116941529235,
      "grad_norm": 0.0913707250463421,
      "learning_rate": 3.3605353793933926e-06,
      "loss": 0.0016,
      "step": 3704
    },
    {
      "epoch": 2.7773613193403297,
      "grad_norm": 0.1774842417966673,
      "learning_rate": 3.3381454100510414e-06,
      "loss": 0.0142,
      "step": 3705
    },
    {
      "epoch": 2.778110944527736,
      "grad_norm": 0.1706681027415529,
      "learning_rate": 3.3158290120372304e-06,
      "loss": 0.0091,
      "step": 3706
    },
    {
      "epoch": 2.7788605697151425,
      "grad_norm": 0.11213132655813458,
      "learning_rate": 3.2935862023374175e-06,
      "loss": 0.0041,
      "step": 3707
    },
    {
      "epoch": 2.7796101949025487,
      "grad_norm": 0.12547887711062505,
      "learning_rate": 3.2714169978810827e-06,
      "loss": 0.0022,
      "step": 3708
    },
    {
      "epoch": 2.780359820089955,
      "grad_norm": 0.025223296745068364,
      "learning_rate": 3.2493214155416507e-06,
      "loss": 0.0007,
      "step": 3709
    },
    {
      "epoch": 2.7811094452773615,
      "grad_norm": 0.055322515411058336,
      "learning_rate": 3.2272994721365245e-06,
      "loss": 0.003,
      "step": 3710
    },
    {
      "epoch": 2.7818590704647677,
      "grad_norm": 0.17480386434490955,
      "learning_rate": 3.205351184427063e-06,
      "loss": 0.0072,
      "step": 3711
    },
    {
      "epoch": 2.782608695652174,
      "grad_norm": 0.12545903845757225,
      "learning_rate": 3.1834765691185597e-06,
      "loss": 0.0029,
      "step": 3712
    },
    {
      "epoch": 2.7833583208395805,
      "grad_norm": 0.07361961469995401,
      "learning_rate": 3.161675642860229e-06,
      "loss": 0.003,
      "step": 3713
    },
    {
      "epoch": 2.7841079460269866,
      "grad_norm": 0.1633375249165271,
      "learning_rate": 3.139948422245198e-06,
      "loss": 0.0068,
      "step": 3714
    },
    {
      "epoch": 2.784857571214393,
      "grad_norm": 0.1566962819235762,
      "learning_rate": 3.1182949238105273e-06,
      "loss": 0.0093,
      "step": 3715
    },
    {
      "epoch": 2.785607196401799,
      "grad_norm": 0.23511158067147805,
      "learning_rate": 3.096715164037123e-06,
      "loss": 0.0119,
      "step": 3716
    },
    {
      "epoch": 2.786356821589205,
      "grad_norm": 0.09863644320338857,
      "learning_rate": 3.07520915934979e-06,
      "loss": 0.0037,
      "step": 3717
    },
    {
      "epoch": 2.787106446776612,
      "grad_norm": 0.07351010386178865,
      "learning_rate": 3.053776926117191e-06,
      "loss": 0.0026,
      "step": 3718
    },
    {
      "epoch": 2.787856071964018,
      "grad_norm": 0.036135799952357044,
      "learning_rate": 3.032418480651833e-06,
      "loss": 0.0008,
      "step": 3719
    },
    {
      "epoch": 2.788605697151424,
      "grad_norm": 0.11281415219568655,
      "learning_rate": 3.0111338392100785e-06,
      "loss": 0.0078,
      "step": 3720
    },
    {
      "epoch": 2.7893553223388308,
      "grad_norm": 0.07179961099252788,
      "learning_rate": 2.9899230179921135e-06,
      "loss": 0.0054,
      "step": 3721
    },
    {
      "epoch": 2.790104947526237,
      "grad_norm": 0.1429941007376049,
      "learning_rate": 2.9687860331419016e-06,
      "loss": 0.0011,
      "step": 3722
    },
    {
      "epoch": 2.790854572713643,
      "grad_norm": 0.09215340792878511,
      "learning_rate": 2.947722900747274e-06,
      "loss": 0.0059,
      "step": 3723
    },
    {
      "epoch": 2.7916041979010497,
      "grad_norm": 0.12098246407629774,
      "learning_rate": 2.926733636839785e-06,
      "loss": 0.0134,
      "step": 3724
    },
    {
      "epoch": 2.792353823088456,
      "grad_norm": 0.017250456202789255,
      "learning_rate": 2.905818257394799e-06,
      "loss": 0.0005,
      "step": 3725
    },
    {
      "epoch": 2.793103448275862,
      "grad_norm": 0.14333256620093657,
      "learning_rate": 2.8849767783314607e-06,
      "loss": 0.006,
      "step": 3726
    },
    {
      "epoch": 2.7938530734632683,
      "grad_norm": 0.08138190706949305,
      "learning_rate": 2.8642092155126367e-06,
      "loss": 0.0056,
      "step": 3727
    },
    {
      "epoch": 2.7946026986506745,
      "grad_norm": 0.1542464999221198,
      "learning_rate": 2.843515584744949e-06,
      "loss": 0.0101,
      "step": 3728
    },
    {
      "epoch": 2.795352323838081,
      "grad_norm": 0.2655814663047673,
      "learning_rate": 2.822895901778744e-06,
      "loss": 0.0156,
      "step": 3729
    },
    {
      "epoch": 2.7961019490254873,
      "grad_norm": 0.07464939867101461,
      "learning_rate": 2.8023501823080887e-06,
      "loss": 0.0015,
      "step": 3730
    },
    {
      "epoch": 2.7968515742128934,
      "grad_norm": 0.03688259668135321,
      "learning_rate": 2.7818784419707646e-06,
      "loss": 0.0019,
      "step": 3731
    },
    {
      "epoch": 2.7976011994003,
      "grad_norm": 0.05642397865060001,
      "learning_rate": 2.761480696348218e-06,
      "loss": 0.004,
      "step": 3732
    },
    {
      "epoch": 2.7983508245877062,
      "grad_norm": 0.12829403525319455,
      "learning_rate": 2.7411569609655962e-06,
      "loss": 0.0048,
      "step": 3733
    },
    {
      "epoch": 2.7991004497751124,
      "grad_norm": 0.020893617177272582,
      "learning_rate": 2.7209072512917154e-06,
      "loss": 0.0009,
      "step": 3734
    },
    {
      "epoch": 2.7998500749625186,
      "grad_norm": 0.05167962072763174,
      "learning_rate": 2.7007315827390467e-06,
      "loss": 0.0016,
      "step": 3735
    },
    {
      "epoch": 2.8005997001499248,
      "grad_norm": 0.11663704090659255,
      "learning_rate": 2.680629970663717e-06,
      "loss": 0.0047,
      "step": 3736
    },
    {
      "epoch": 2.8013493253373314,
      "grad_norm": 0.17861922868467878,
      "learning_rate": 2.660602430365444e-06,
      "loss": 0.0096,
      "step": 3737
    },
    {
      "epoch": 2.8020989505247376,
      "grad_norm": 0.09685293415007153,
      "learning_rate": 2.6406489770876453e-06,
      "loss": 0.006,
      "step": 3738
    },
    {
      "epoch": 2.8028485757121437,
      "grad_norm": 0.08273466032580473,
      "learning_rate": 2.620769626017283e-06,
      "loss": 0.004,
      "step": 3739
    },
    {
      "epoch": 2.8035982008995504,
      "grad_norm": 0.09701803775699593,
      "learning_rate": 2.6009643922849438e-06,
      "loss": 0.0035,
      "step": 3740
    },
    {
      "epoch": 2.8043478260869565,
      "grad_norm": 0.08971236519901336,
      "learning_rate": 2.5812332909647907e-06,
      "loss": 0.0084,
      "step": 3741
    },
    {
      "epoch": 2.8050974512743627,
      "grad_norm": 0.05662903915501467,
      "learning_rate": 2.5615763370745895e-06,
      "loss": 0.0042,
      "step": 3742
    },
    {
      "epoch": 2.8058470764617693,
      "grad_norm": 0.05845551070544907,
      "learning_rate": 2.5419935455756383e-06,
      "loss": 0.0015,
      "step": 3743
    },
    {
      "epoch": 2.8065967016491755,
      "grad_norm": 0.04267221242974165,
      "learning_rate": 2.522484931372804e-06,
      "loss": 0.0026,
      "step": 3744
    },
    {
      "epoch": 2.8073463268365817,
      "grad_norm": 0.2170617038162767,
      "learning_rate": 2.503050509314486e-06,
      "loss": 0.0096,
      "step": 3745
    },
    {
      "epoch": 2.808095952023988,
      "grad_norm": 0.10786811045959256,
      "learning_rate": 2.483690294192653e-06,
      "loss": 0.0069,
      "step": 3746
    },
    {
      "epoch": 2.808845577211394,
      "grad_norm": 0.10031047771448215,
      "learning_rate": 2.46440430074274e-06,
      "loss": 0.0065,
      "step": 3747
    },
    {
      "epoch": 2.8095952023988007,
      "grad_norm": 0.521257586527663,
      "learning_rate": 2.445192543643715e-06,
      "loss": 0.0243,
      "step": 3748
    },
    {
      "epoch": 2.810344827586207,
      "grad_norm": 0.04209398645898599,
      "learning_rate": 2.426055037518027e-06,
      "loss": 0.0023,
      "step": 3749
    },
    {
      "epoch": 2.811094452773613,
      "grad_norm": 0.11360610892491614,
      "learning_rate": 2.406991796931668e-06,
      "loss": 0.003,
      "step": 3750
    },
    {
      "epoch": 2.8118440779610197,
      "grad_norm": 0.06493169758031436,
      "learning_rate": 2.388002836394043e-06,
      "loss": 0.0031,
      "step": 3751
    },
    {
      "epoch": 2.812593703148426,
      "grad_norm": 0.06315294635620365,
      "learning_rate": 2.3690881703580247e-06,
      "loss": 0.002,
      "step": 3752
    },
    {
      "epoch": 2.813343328335832,
      "grad_norm": 0.08553548216051235,
      "learning_rate": 2.3502478132199745e-06,
      "loss": 0.0049,
      "step": 3753
    },
    {
      "epoch": 2.8140929535232386,
      "grad_norm": 0.018008874414034905,
      "learning_rate": 2.3314817793196665e-06,
      "loss": 0.0005,
      "step": 3754
    },
    {
      "epoch": 2.814842578710645,
      "grad_norm": 0.08817547281035387,
      "learning_rate": 2.3127900829403306e-06,
      "loss": 0.0052,
      "step": 3755
    },
    {
      "epoch": 2.815592203898051,
      "grad_norm": 0.10272056274155184,
      "learning_rate": 2.2941727383085754e-06,
      "loss": 0.0109,
      "step": 3756
    },
    {
      "epoch": 2.816341829085457,
      "grad_norm": 0.05751309501790076,
      "learning_rate": 2.2756297595944777e-06,
      "loss": 0.0051,
      "step": 3757
    },
    {
      "epoch": 2.8170914542728633,
      "grad_norm": 0.07757600986659895,
      "learning_rate": 2.257161160911447e-06,
      "loss": 0.0048,
      "step": 3758
    },
    {
      "epoch": 2.81784107946027,
      "grad_norm": 0.03507914686943822,
      "learning_rate": 2.2387669563163406e-06,
      "loss": 0.002,
      "step": 3759
    },
    {
      "epoch": 2.818590704647676,
      "grad_norm": 0.06233321008617979,
      "learning_rate": 2.2204471598093358e-06,
      "loss": 0.0055,
      "step": 3760
    },
    {
      "epoch": 2.8193403298350823,
      "grad_norm": 0.08369995481880844,
      "learning_rate": 2.202201785334046e-06,
      "loss": 0.0049,
      "step": 3761
    },
    {
      "epoch": 2.820089955022489,
      "grad_norm": 0.3486093717381726,
      "learning_rate": 2.1840308467773853e-06,
      "loss": 0.0208,
      "step": 3762
    },
    {
      "epoch": 2.820839580209895,
      "grad_norm": 0.18296843989128325,
      "learning_rate": 2.1659343579696235e-06,
      "loss": 0.0078,
      "step": 3763
    },
    {
      "epoch": 2.8215892053973013,
      "grad_norm": 0.04568970976539647,
      "learning_rate": 2.1479123326843764e-06,
      "loss": 0.0011,
      "step": 3764
    },
    {
      "epoch": 2.822338830584708,
      "grad_norm": 0.04835958607273264,
      "learning_rate": 2.129964784638572e-06,
      "loss": 0.004,
      "step": 3765
    },
    {
      "epoch": 2.823088455772114,
      "grad_norm": 0.0728104278943814,
      "learning_rate": 2.112091727492471e-06,
      "loss": 0.004,
      "step": 3766
    },
    {
      "epoch": 2.8238380809595203,
      "grad_norm": 0.13355061657936787,
      "learning_rate": 2.094293174849604e-06,
      "loss": 0.005,
      "step": 3767
    },
    {
      "epoch": 2.8245877061469264,
      "grad_norm": 0.04349484855573775,
      "learning_rate": 2.0765691402568454e-06,
      "loss": 0.002,
      "step": 3768
    },
    {
      "epoch": 2.8253373313343326,
      "grad_norm": 0.1122534941982581,
      "learning_rate": 2.058919637204304e-06,
      "loss": 0.0134,
      "step": 3769
    },
    {
      "epoch": 2.8260869565217392,
      "grad_norm": 0.1021653774283419,
      "learning_rate": 2.041344679125379e-06,
      "loss": 0.011,
      "step": 3770
    },
    {
      "epoch": 2.8268365817091454,
      "grad_norm": 0.08401879274385336,
      "learning_rate": 2.023844279396736e-06,
      "loss": 0.0066,
      "step": 3771
    },
    {
      "epoch": 2.8275862068965516,
      "grad_norm": 0.08403663673675378,
      "learning_rate": 2.0064184513382987e-06,
      "loss": 0.0056,
      "step": 3772
    },
    {
      "epoch": 2.8283358320839582,
      "grad_norm": 0.08761283834204707,
      "learning_rate": 1.989067208213202e-06,
      "loss": 0.0043,
      "step": 3773
    },
    {
      "epoch": 2.8290854572713644,
      "grad_norm": 0.048936710908496085,
      "learning_rate": 1.971790563227871e-06,
      "loss": 0.0022,
      "step": 3774
    },
    {
      "epoch": 2.8298350824587706,
      "grad_norm": 0.1903573056337051,
      "learning_rate": 1.9545885295318754e-06,
      "loss": 0.0083,
      "step": 3775
    },
    {
      "epoch": 2.8305847076461768,
      "grad_norm": 0.10385777442522533,
      "learning_rate": 1.937461120218065e-06,
      "loss": 0.0042,
      "step": 3776
    },
    {
      "epoch": 2.831334332833583,
      "grad_norm": 0.08232033379572473,
      "learning_rate": 1.9204083483224553e-06,
      "loss": 0.0045,
      "step": 3777
    },
    {
      "epoch": 2.8320839580209896,
      "grad_norm": 0.0866681413818064,
      "learning_rate": 1.9034302268242654e-06,
      "loss": 0.0089,
      "step": 3778
    },
    {
      "epoch": 2.8328335832083957,
      "grad_norm": 0.15012606367667924,
      "learning_rate": 1.88652676864588e-06,
      "loss": 0.0053,
      "step": 3779
    },
    {
      "epoch": 2.833583208395802,
      "grad_norm": 0.09515478494613794,
      "learning_rate": 1.8696979866528851e-06,
      "loss": 0.0055,
      "step": 3780
    },
    {
      "epoch": 2.8343328335832085,
      "grad_norm": 0.05510011211789895,
      "learning_rate": 1.8529438936540021e-06,
      "loss": 0.0025,
      "step": 3781
    },
    {
      "epoch": 2.8350824587706147,
      "grad_norm": 0.05029481154099812,
      "learning_rate": 1.8362645024011305e-06,
      "loss": 0.0045,
      "step": 3782
    },
    {
      "epoch": 2.835832083958021,
      "grad_norm": 0.06596027338712288,
      "learning_rate": 1.8196598255892816e-06,
      "loss": 0.0036,
      "step": 3783
    },
    {
      "epoch": 2.8365817091454275,
      "grad_norm": 0.07545314831200027,
      "learning_rate": 1.8031298758566129e-06,
      "loss": 0.0037,
      "step": 3784
    },
    {
      "epoch": 2.8373313343328337,
      "grad_norm": 0.21628185122520407,
      "learning_rate": 1.786674665784438e-06,
      "loss": 0.0166,
      "step": 3785
    },
    {
      "epoch": 2.83808095952024,
      "grad_norm": 0.11500979436017138,
      "learning_rate": 1.7702942078971384e-06,
      "loss": 0.0075,
      "step": 3786
    },
    {
      "epoch": 2.838830584707646,
      "grad_norm": 0.20084826385099694,
      "learning_rate": 1.7539885146622192e-06,
      "loss": 0.0054,
      "step": 3787
    },
    {
      "epoch": 2.839580209895052,
      "grad_norm": 0.08656276104804662,
      "learning_rate": 1.737757598490275e-06,
      "loss": 0.0036,
      "step": 3788
    },
    {
      "epoch": 2.840329835082459,
      "grad_norm": 0.20638457703316246,
      "learning_rate": 1.7216014717350015e-06,
      "loss": 0.0046,
      "step": 3789
    },
    {
      "epoch": 2.841079460269865,
      "grad_norm": 0.1817487754895775,
      "learning_rate": 1.7055201466931515e-06,
      "loss": 0.0033,
      "step": 3790
    },
    {
      "epoch": 2.841829085457271,
      "grad_norm": 0.06294821656981923,
      "learning_rate": 1.6895136356045449e-06,
      "loss": 0.0021,
      "step": 3791
    },
    {
      "epoch": 2.842578710644678,
      "grad_norm": 0.0595355858491971,
      "learning_rate": 1.673581950652081e-06,
      "loss": 0.0065,
      "step": 3792
    },
    {
      "epoch": 2.843328335832084,
      "grad_norm": 0.07798007068273854,
      "learning_rate": 1.6577251039616936e-06,
      "loss": 0.0011,
      "step": 3793
    },
    {
      "epoch": 2.84407796101949,
      "grad_norm": 0.057368905011597304,
      "learning_rate": 1.6419431076023506e-06,
      "loss": 0.0024,
      "step": 3794
    },
    {
      "epoch": 2.844827586206897,
      "grad_norm": 0.09375199404270616,
      "learning_rate": 1.6262359735860544e-06,
      "loss": 0.007,
      "step": 3795
    },
    {
      "epoch": 2.845577211394303,
      "grad_norm": 0.08821695567906807,
      "learning_rate": 1.6106037138678199e-06,
      "loss": 0.003,
      "step": 3796
    },
    {
      "epoch": 2.846326836581709,
      "grad_norm": 0.11360534842385173,
      "learning_rate": 1.595046340345685e-06,
      "loss": 0.0055,
      "step": 3797
    },
    {
      "epoch": 2.8470764617691153,
      "grad_norm": 0.12330877431872257,
      "learning_rate": 1.5795638648607003e-06,
      "loss": 0.0032,
      "step": 3798
    },
    {
      "epoch": 2.8478260869565215,
      "grad_norm": 0.06392015519757543,
      "learning_rate": 1.5641562991968949e-06,
      "loss": 0.0027,
      "step": 3799
    },
    {
      "epoch": 2.848575712143928,
      "grad_norm": 0.3197110206802998,
      "learning_rate": 1.5488236550812663e-06,
      "loss": 0.021,
      "step": 3800
    },
    {
      "epoch": 2.8493253373313343,
      "grad_norm": 0.06995177459050711,
      "learning_rate": 1.5335659441838347e-06,
      "loss": 0.0048,
      "step": 3801
    },
    {
      "epoch": 2.8500749625187405,
      "grad_norm": 0.0483401155538583,
      "learning_rate": 1.5183831781175217e-06,
      "loss": 0.0033,
      "step": 3802
    },
    {
      "epoch": 2.850824587706147,
      "grad_norm": 0.19698717564288212,
      "learning_rate": 1.5032753684382727e-06,
      "loss": 0.0051,
      "step": 3803
    },
    {
      "epoch": 2.8515742128935533,
      "grad_norm": 0.13355720806571342,
      "learning_rate": 1.4882425266449452e-06,
      "loss": 0.0048,
      "step": 3804
    },
    {
      "epoch": 2.8523238380809595,
      "grad_norm": 0.06441256905321165,
      "learning_rate": 1.4732846641793419e-06,
      "loss": 0.0032,
      "step": 3805
    },
    {
      "epoch": 2.853073463268366,
      "grad_norm": 0.0877528657305153,
      "learning_rate": 1.4584017924262005e-06,
      "loss": 0.0052,
      "step": 3806
    },
    {
      "epoch": 2.8538230884557723,
      "grad_norm": 0.05421868798784275,
      "learning_rate": 1.4435939227131713e-06,
      "loss": 0.0038,
      "step": 3807
    },
    {
      "epoch": 2.8545727136431784,
      "grad_norm": 0.0690574032640628,
      "learning_rate": 1.4288610663108382e-06,
      "loss": 0.0021,
      "step": 3808
    },
    {
      "epoch": 2.8553223388305846,
      "grad_norm": 0.12791567353295813,
      "learning_rate": 1.4142032344326983e-06,
      "loss": 0.0083,
      "step": 3809
    },
    {
      "epoch": 2.856071964017991,
      "grad_norm": 0.053996680372903,
      "learning_rate": 1.3996204382351051e-06,
      "loss": 0.0024,
      "step": 3810
    },
    {
      "epoch": 2.8568215892053974,
      "grad_norm": 0.12263371388253382,
      "learning_rate": 1.385112688817325e-06,
      "loss": 0.005,
      "step": 3811
    },
    {
      "epoch": 2.8575712143928036,
      "grad_norm": 0.10742409395354456,
      "learning_rate": 1.3706799972215246e-06,
      "loss": 0.0041,
      "step": 3812
    },
    {
      "epoch": 2.8583208395802098,
      "grad_norm": 0.11053414179643328,
      "learning_rate": 1.3563223744327058e-06,
      "loss": 0.0091,
      "step": 3813
    },
    {
      "epoch": 2.8590704647676164,
      "grad_norm": 0.23767852215258636,
      "learning_rate": 1.3420398313787607e-06,
      "loss": 0.0074,
      "step": 3814
    },
    {
      "epoch": 2.8598200899550226,
      "grad_norm": 0.23954200853942614,
      "learning_rate": 1.327832378930427e-06,
      "loss": 0.0084,
      "step": 3815
    },
    {
      "epoch": 2.8605697151424287,
      "grad_norm": 0.11676655596471934,
      "learning_rate": 1.3137000279012767e-06,
      "loss": 0.0076,
      "step": 3816
    },
    {
      "epoch": 2.861319340329835,
      "grad_norm": 0.08462846856681903,
      "learning_rate": 1.2996427890477503e-06,
      "loss": 0.0038,
      "step": 3817
    },
    {
      "epoch": 2.862068965517241,
      "grad_norm": 0.11947819513361728,
      "learning_rate": 1.285660673069089e-06,
      "loss": 0.0063,
      "step": 3818
    },
    {
      "epoch": 2.8628185907046477,
      "grad_norm": 0.06530871712869708,
      "learning_rate": 1.271753690607369e-06,
      "loss": 0.0027,
      "step": 3819
    },
    {
      "epoch": 2.863568215892054,
      "grad_norm": 0.2695348273709761,
      "learning_rate": 1.25792185224749e-06,
      "loss": 0.0082,
      "step": 3820
    },
    {
      "epoch": 2.86431784107946,
      "grad_norm": 0.03476298207593885,
      "learning_rate": 1.2441651685171417e-06,
      "loss": 0.0014,
      "step": 3821
    },
    {
      "epoch": 2.8650674662668667,
      "grad_norm": 0.3226581869020099,
      "learning_rate": 1.2304836498868265e-06,
      "loss": 0.0689,
      "step": 3822
    },
    {
      "epoch": 2.865817091454273,
      "grad_norm": 0.07761082439778805,
      "learning_rate": 1.2168773067698259e-06,
      "loss": 0.0045,
      "step": 3823
    },
    {
      "epoch": 2.866566716641679,
      "grad_norm": 0.3412718354202291,
      "learning_rate": 1.2033461495222108e-06,
      "loss": 0.0112,
      "step": 3824
    },
    {
      "epoch": 2.8673163418290857,
      "grad_norm": 0.16766941008019637,
      "learning_rate": 1.1898901884428104e-06,
      "loss": 0.0054,
      "step": 3825
    },
    {
      "epoch": 2.868065967016492,
      "grad_norm": 0.12532044546710538,
      "learning_rate": 1.1765094337732542e-06,
      "loss": 0.0123,
      "step": 3826
    },
    {
      "epoch": 2.868815592203898,
      "grad_norm": 0.10200359226472538,
      "learning_rate": 1.1632038956979064e-06,
      "loss": 0.0049,
      "step": 3827
    },
    {
      "epoch": 2.869565217391304,
      "grad_norm": 0.09724262866667624,
      "learning_rate": 1.1499735843438665e-06,
      "loss": 0.0024,
      "step": 3828
    },
    {
      "epoch": 2.8703148425787104,
      "grad_norm": 0.10813501029772407,
      "learning_rate": 1.1368185097810236e-06,
      "loss": 0.0059,
      "step": 3829
    },
    {
      "epoch": 2.871064467766117,
      "grad_norm": 0.06405598123253393,
      "learning_rate": 1.1237386820219465e-06,
      "loss": 0.0052,
      "step": 3830
    },
    {
      "epoch": 2.871814092953523,
      "grad_norm": 0.11886189362782622,
      "learning_rate": 1.1107341110219938e-06,
      "loss": 0.0061,
      "step": 3831
    },
    {
      "epoch": 2.8725637181409294,
      "grad_norm": 0.04310115337204529,
      "learning_rate": 1.0978048066791925e-06,
      "loss": 0.0017,
      "step": 3832
    },
    {
      "epoch": 2.873313343328336,
      "grad_norm": 0.21923193277890474,
      "learning_rate": 1.084950778834304e-06,
      "loss": 0.0186,
      "step": 3833
    },
    {
      "epoch": 2.874062968515742,
      "grad_norm": 0.15533190052223408,
      "learning_rate": 1.0721720372707911e-06,
      "loss": 0.0053,
      "step": 3834
    },
    {
      "epoch": 2.8748125937031483,
      "grad_norm": 0.16032577319751884,
      "learning_rate": 1.0594685917148294e-06,
      "loss": 0.0094,
      "step": 3835
    },
    {
      "epoch": 2.875562218890555,
      "grad_norm": 0.10724245151178886,
      "learning_rate": 1.0468404518352738e-06,
      "loss": 0.0084,
      "step": 3836
    },
    {
      "epoch": 2.876311844077961,
      "grad_norm": 0.06086983438719523,
      "learning_rate": 1.0342876272436464e-06,
      "loss": 0.0038,
      "step": 3837
    },
    {
      "epoch": 2.8770614692653673,
      "grad_norm": 0.09034829380911473,
      "learning_rate": 1.0218101274941604e-06,
      "loss": 0.0041,
      "step": 3838
    },
    {
      "epoch": 2.8778110944527735,
      "grad_norm": 0.06363614232706609,
      "learning_rate": 1.0094079620837194e-06,
      "loss": 0.0013,
      "step": 3839
    },
    {
      "epoch": 2.8785607196401797,
      "grad_norm": 0.04423131224523309,
      "learning_rate": 9.970811404518389e-07,
      "loss": 0.0029,
      "step": 3840
    },
    {
      "epoch": 2.8793103448275863,
      "grad_norm": 0.08304277803334834,
      "learning_rate": 9.848296719807359e-07,
      "loss": 0.0045,
      "step": 3841
    },
    {
      "epoch": 2.8800599700149925,
      "grad_norm": 0.028576192852339508,
      "learning_rate": 9.726535659952519e-07,
      "loss": 0.0007,
      "step": 3842
    },
    {
      "epoch": 2.8808095952023987,
      "grad_norm": 0.15636058161463368,
      "learning_rate": 9.60552831762873e-07,
      "loss": 0.0095,
      "step": 3843
    },
    {
      "epoch": 2.8815592203898053,
      "grad_norm": 0.03345494562713568,
      "learning_rate": 9.485274784937104e-07,
      "loss": 0.0019,
      "step": 3844
    },
    {
      "epoch": 2.8823088455772115,
      "grad_norm": 0.07728964927781995,
      "learning_rate": 9.365775153405088e-07,
      "loss": 0.007,
      "step": 3845
    },
    {
      "epoch": 2.8830584707646176,
      "grad_norm": 0.05631770312584182,
      "learning_rate": 9.247029513986482e-07,
      "loss": 0.0025,
      "step": 3846
    },
    {
      "epoch": 2.8838080959520243,
      "grad_norm": 0.06305904315543834,
      "learning_rate": 9.129037957060771e-07,
      "loss": 0.0009,
      "step": 3847
    },
    {
      "epoch": 2.8845577211394304,
      "grad_norm": 0.02943176703260356,
      "learning_rate": 9.011800572434003e-07,
      "loss": 0.0035,
      "step": 3848
    },
    {
      "epoch": 2.8853073463268366,
      "grad_norm": 0.12886854986809387,
      "learning_rate": 8.895317449337803e-07,
      "loss": 0.012,
      "step": 3849
    },
    {
      "epoch": 2.886056971514243,
      "grad_norm": 0.0639795246401034,
      "learning_rate": 8.779588676429917e-07,
      "loss": 0.0041,
      "step": 3850
    },
    {
      "epoch": 2.886806596701649,
      "grad_norm": 0.1330508162503493,
      "learning_rate": 8.664614341793886e-07,
      "loss": 0.0047,
      "step": 3851
    },
    {
      "epoch": 2.8875562218890556,
      "grad_norm": 0.0864366190069689,
      "learning_rate": 8.550394532939154e-07,
      "loss": 0.0018,
      "step": 3852
    },
    {
      "epoch": 2.8883058470764618,
      "grad_norm": 0.0501558109337864,
      "learning_rate": 8.436929336800515e-07,
      "loss": 0.0041,
      "step": 3853
    },
    {
      "epoch": 2.889055472263868,
      "grad_norm": 0.03194449960152962,
      "learning_rate": 8.324218839738996e-07,
      "loss": 0.0015,
      "step": 3854
    },
    {
      "epoch": 2.8898050974512746,
      "grad_norm": 0.04313150198455092,
      "learning_rate": 8.212263127540642e-07,
      "loss": 0.002,
      "step": 3855
    },
    {
      "epoch": 2.8905547226386807,
      "grad_norm": 0.15162586669753475,
      "learning_rate": 8.101062285417405e-07,
      "loss": 0.0135,
      "step": 3856
    },
    {
      "epoch": 2.891304347826087,
      "grad_norm": 0.059281772264859076,
      "learning_rate": 7.990616398006356e-07,
      "loss": 0.0034,
      "step": 3857
    },
    {
      "epoch": 2.892053973013493,
      "grad_norm": 0.07432819629295129,
      "learning_rate": 7.880925549370366e-07,
      "loss": 0.0041,
      "step": 3858
    },
    {
      "epoch": 2.8928035982008993,
      "grad_norm": 0.036534280314377575,
      "learning_rate": 7.771989822997206e-07,
      "loss": 0.0006,
      "step": 3859
    },
    {
      "epoch": 2.893553223388306,
      "grad_norm": 0.15373465674637074,
      "learning_rate": 7.663809301800106e-07,
      "loss": 0.0044,
      "step": 3860
    },
    {
      "epoch": 2.894302848575712,
      "grad_norm": 0.037159260867517294,
      "learning_rate": 7.55638406811765e-07,
      "loss": 0.0018,
      "step": 3861
    },
    {
      "epoch": 2.8950524737631183,
      "grad_norm": 0.05407922317555088,
      "learning_rate": 7.449714203713321e-07,
      "loss": 0.0034,
      "step": 3862
    },
    {
      "epoch": 2.895802098950525,
      "grad_norm": 0.13115050264454448,
      "learning_rate": 7.343799789775618e-07,
      "loss": 0.0107,
      "step": 3863
    },
    {
      "epoch": 2.896551724137931,
      "grad_norm": 0.07134410667091412,
      "learning_rate": 7.238640906918282e-07,
      "loss": 0.0073,
      "step": 3864
    },
    {
      "epoch": 2.8973013493253372,
      "grad_norm": 0.07090129947559054,
      "learning_rate": 7.134237635180063e-07,
      "loss": 0.0031,
      "step": 3865
    },
    {
      "epoch": 2.898050974512744,
      "grad_norm": 0.05134175014094741,
      "learning_rate": 7.030590054024178e-07,
      "loss": 0.0044,
      "step": 3866
    },
    {
      "epoch": 2.89880059970015,
      "grad_norm": 0.1938597397603204,
      "learning_rate": 6.927698242339076e-07,
      "loss": 0.0147,
      "step": 3867
    },
    {
      "epoch": 2.899550224887556,
      "grad_norm": 0.041853566186652764,
      "learning_rate": 6.825562278437781e-07,
      "loss": 0.0025,
      "step": 3868
    },
    {
      "epoch": 2.9002998500749624,
      "grad_norm": 0.064640348356761,
      "learning_rate": 6.724182240058108e-07,
      "loss": 0.0046,
      "step": 3869
    },
    {
      "epoch": 2.9010494752623686,
      "grad_norm": 0.21030206970660664,
      "learning_rate": 6.623558204362446e-07,
      "loss": 0.0092,
      "step": 3870
    },
    {
      "epoch": 2.901799100449775,
      "grad_norm": 0.048488401349587804,
      "learning_rate": 6.523690247937864e-07,
      "loss": 0.0023,
      "step": 3871
    },
    {
      "epoch": 2.9025487256371814,
      "grad_norm": 0.11700840312090334,
      "learning_rate": 6.424578446796003e-07,
      "loss": 0.0049,
      "step": 3872
    },
    {
      "epoch": 2.9032983508245875,
      "grad_norm": 0.09652391917869022,
      "learning_rate": 6.326222876372745e-07,
      "loss": 0.0044,
      "step": 3873
    },
    {
      "epoch": 2.904047976011994,
      "grad_norm": 0.06097657800986373,
      "learning_rate": 6.22862361152865e-07,
      "loss": 0.0067,
      "step": 3874
    },
    {
      "epoch": 2.9047976011994003,
      "grad_norm": 0.13290948988113532,
      "learning_rate": 6.13178072654852e-07,
      "loss": 0.0062,
      "step": 3875
    },
    {
      "epoch": 2.9055472263868065,
      "grad_norm": 0.03547007821767269,
      "learning_rate": 6.035694295141391e-07,
      "loss": 0.0014,
      "step": 3876
    },
    {
      "epoch": 2.906296851574213,
      "grad_norm": 0.1449129728244748,
      "learning_rate": 5.940364390440656e-07,
      "loss": 0.0081,
      "step": 3877
    },
    {
      "epoch": 2.9070464767616193,
      "grad_norm": 0.05884229221265448,
      "learning_rate": 5.845791085003937e-07,
      "loss": 0.0011,
      "step": 3878
    },
    {
      "epoch": 2.9077961019490255,
      "grad_norm": 0.048490012177309055,
      "learning_rate": 5.75197445081288e-07,
      "loss": 0.0026,
      "step": 3879
    },
    {
      "epoch": 2.9085457271364317,
      "grad_norm": 0.12293404610823899,
      "learning_rate": 5.658914559273143e-07,
      "loss": 0.0044,
      "step": 3880
    },
    {
      "epoch": 2.909295352323838,
      "grad_norm": 0.11175197743283243,
      "learning_rate": 5.566611481214734e-07,
      "loss": 0.0036,
      "step": 3881
    },
    {
      "epoch": 2.9100449775112445,
      "grad_norm": 0.027393754256709764,
      "learning_rate": 5.475065286891346e-07,
      "loss": 0.0004,
      "step": 3882
    },
    {
      "epoch": 2.9107946026986506,
      "grad_norm": 0.08192982198985065,
      "learning_rate": 5.384276045980574e-07,
      "loss": 0.0048,
      "step": 3883
    },
    {
      "epoch": 2.911544227886057,
      "grad_norm": 0.05348463886759356,
      "learning_rate": 5.294243827584145e-07,
      "loss": 0.003,
      "step": 3884
    },
    {
      "epoch": 2.9122938530734634,
      "grad_norm": 0.11884429591828721,
      "learning_rate": 5.204968700227242e-07,
      "loss": 0.0169,
      "step": 3885
    },
    {
      "epoch": 2.9130434782608696,
      "grad_norm": 0.1553683804200957,
      "learning_rate": 5.116450731859179e-07,
      "loss": 0.003,
      "step": 3886
    },
    {
      "epoch": 2.913793103448276,
      "grad_norm": 0.26089796779801405,
      "learning_rate": 5.02868998985273e-07,
      "loss": 0.019,
      "step": 3887
    },
    {
      "epoch": 2.9145427286356824,
      "grad_norm": 0.07710209310944306,
      "learning_rate": 4.941686541004464e-07,
      "loss": 0.0052,
      "step": 3888
    },
    {
      "epoch": 2.9152923538230886,
      "grad_norm": 0.09064721229356534,
      "learning_rate": 4.85544045153441e-07,
      "loss": 0.0055,
      "step": 3889
    },
    {
      "epoch": 2.9160419790104948,
      "grad_norm": 0.04485489321328558,
      "learning_rate": 4.76995178708628e-07,
      "loss": 0.0011,
      "step": 3890
    },
    {
      "epoch": 2.916791604197901,
      "grad_norm": 0.10881058038757435,
      "learning_rate": 4.685220612727248e-07,
      "loss": 0.0092,
      "step": 3891
    },
    {
      "epoch": 2.917541229385307,
      "grad_norm": 0.11097857641116497,
      "learning_rate": 4.6012469929479496e-07,
      "loss": 0.0015,
      "step": 3892
    },
    {
      "epoch": 2.9182908545727138,
      "grad_norm": 0.08700092245246938,
      "learning_rate": 4.5180309916623697e-07,
      "loss": 0.0028,
      "step": 3893
    },
    {
      "epoch": 2.91904047976012,
      "grad_norm": 0.08473276032542719,
      "learning_rate": 4.435572672208066e-07,
      "loss": 0.0078,
      "step": 3894
    },
    {
      "epoch": 2.919790104947526,
      "grad_norm": 0.10730176722413753,
      "learning_rate": 4.3538720973457235e-07,
      "loss": 0.0024,
      "step": 3895
    },
    {
      "epoch": 2.9205397301349327,
      "grad_norm": 0.03297186138157908,
      "learning_rate": 4.2729293292592677e-07,
      "loss": 0.0006,
      "step": 3896
    },
    {
      "epoch": 2.921289355322339,
      "grad_norm": 0.06178056449219258,
      "learning_rate": 4.1927444295559725e-07,
      "loss": 0.0037,
      "step": 3897
    },
    {
      "epoch": 2.922038980509745,
      "grad_norm": 0.08020392353543454,
      "learning_rate": 4.113317459266242e-07,
      "loss": 0.009,
      "step": 3898
    },
    {
      "epoch": 2.9227886056971513,
      "grad_norm": 0.20249707480901977,
      "learning_rate": 4.0346484788434945e-07,
      "loss": 0.0109,
      "step": 3899
    },
    {
      "epoch": 2.923538230884558,
      "grad_norm": 0.03934935977733298,
      "learning_rate": 3.95673754816428e-07,
      "loss": 0.0025,
      "step": 3900
    },
    {
      "epoch": 2.924287856071964,
      "grad_norm": 0.08764117867567876,
      "learning_rate": 3.8795847265282735e-07,
      "loss": 0.0058,
      "step": 3901
    },
    {
      "epoch": 2.9250374812593702,
      "grad_norm": 0.22565307935971812,
      "learning_rate": 3.8031900726581694e-07,
      "loss": 0.0107,
      "step": 3902
    },
    {
      "epoch": 2.9257871064467764,
      "grad_norm": 0.08387411944908658,
      "learning_rate": 3.727553644699455e-07,
      "loss": 0.0032,
      "step": 3903
    },
    {
      "epoch": 2.926536731634183,
      "grad_norm": 0.08318669900705351,
      "learning_rate": 3.652675500220415e-07,
      "loss": 0.0045,
      "step": 3904
    },
    {
      "epoch": 2.927286356821589,
      "grad_norm": 0.06422323549076467,
      "learning_rate": 3.578555696212571e-07,
      "loss": 0.0033,
      "step": 3905
    },
    {
      "epoch": 2.9280359820089954,
      "grad_norm": 0.04610187902934583,
      "learning_rate": 3.505194289089908e-07,
      "loss": 0.0036,
      "step": 3906
    },
    {
      "epoch": 2.928785607196402,
      "grad_norm": 0.07675639671383808,
      "learning_rate": 3.432591334689317e-07,
      "loss": 0.0073,
      "step": 3907
    },
    {
      "epoch": 2.929535232383808,
      "grad_norm": 0.05767654466937584,
      "learning_rate": 3.360746888270372e-07,
      "loss": 0.0012,
      "step": 3908
    },
    {
      "epoch": 2.9302848575712144,
      "grad_norm": 0.17132906497501388,
      "learning_rate": 3.289661004515554e-07,
      "loss": 0.0085,
      "step": 3909
    },
    {
      "epoch": 2.9310344827586206,
      "grad_norm": 0.027438192464950117,
      "learning_rate": 3.219333737529473e-07,
      "loss": 0.0008,
      "step": 3910
    },
    {
      "epoch": 2.9317841079460267,
      "grad_norm": 0.054228259408073395,
      "learning_rate": 3.1497651408399776e-07,
      "loss": 0.0016,
      "step": 3911
    },
    {
      "epoch": 2.9325337331334334,
      "grad_norm": 0.08815393033743125,
      "learning_rate": 3.080955267396934e-07,
      "loss": 0.0143,
      "step": 3912
    },
    {
      "epoch": 2.9332833583208395,
      "grad_norm": 0.20265750857265044,
      "learning_rate": 3.0129041695730054e-07,
      "loss": 0.0212,
      "step": 3913
    },
    {
      "epoch": 2.9340329835082457,
      "grad_norm": 0.12156823056329677,
      "learning_rate": 2.945611899163425e-07,
      "loss": 0.003,
      "step": 3914
    },
    {
      "epoch": 2.9347826086956523,
      "grad_norm": 0.018979792274225624,
      "learning_rate": 2.8790785073855574e-07,
      "loss": 0.0008,
      "step": 3915
    },
    {
      "epoch": 2.9355322338830585,
      "grad_norm": 0.1942467859990612,
      "learning_rate": 2.813304044879339e-07,
      "loss": 0.0114,
      "step": 3916
    },
    {
      "epoch": 2.9362818590704647,
      "grad_norm": 0.12520849384259206,
      "learning_rate": 2.7482885617071686e-07,
      "loss": 0.0049,
      "step": 3917
    },
    {
      "epoch": 2.9370314842578713,
      "grad_norm": 0.2781515226567302,
      "learning_rate": 2.6840321073536843e-07,
      "loss": 0.0124,
      "step": 3918
    },
    {
      "epoch": 2.9377811094452775,
      "grad_norm": 0.06083224080379495,
      "learning_rate": 2.6205347307256546e-07,
      "loss": 0.0027,
      "step": 3919
    },
    {
      "epoch": 2.9385307346326837,
      "grad_norm": 0.16590619218430297,
      "learning_rate": 2.557796480152308e-07,
      "loss": 0.0064,
      "step": 3920
    },
    {
      "epoch": 2.93928035982009,
      "grad_norm": 0.09392875545963016,
      "learning_rate": 2.495817403385114e-07,
      "loss": 0.0079,
      "step": 3921
    },
    {
      "epoch": 2.940029985007496,
      "grad_norm": 0.1475733909454278,
      "learning_rate": 2.4345975475975615e-07,
      "loss": 0.0091,
      "step": 3922
    },
    {
      "epoch": 2.9407796101949026,
      "grad_norm": 0.25972195245536994,
      "learning_rate": 2.3741369593852647e-07,
      "loss": 0.0126,
      "step": 3923
    },
    {
      "epoch": 2.941529235382309,
      "grad_norm": 0.06200435894219216,
      "learning_rate": 2.314435684766081e-07,
      "loss": 0.0082,
      "step": 3924
    },
    {
      "epoch": 2.942278860569715,
      "grad_norm": 0.1400935741912931,
      "learning_rate": 2.2554937691798839e-07,
      "loss": 0.0102,
      "step": 3925
    },
    {
      "epoch": 2.9430284857571216,
      "grad_norm": 0.07050883912900091,
      "learning_rate": 2.1973112574885656e-07,
      "loss": 0.0032,
      "step": 3926
    },
    {
      "epoch": 2.943778110944528,
      "grad_norm": 0.09879521943653216,
      "learning_rate": 2.1398881939761474e-07,
      "loss": 0.0047,
      "step": 3927
    },
    {
      "epoch": 2.944527736131934,
      "grad_norm": 0.09017841555169903,
      "learning_rate": 2.0832246223483344e-07,
      "loss": 0.0045,
      "step": 3928
    },
    {
      "epoch": 2.9452773613193406,
      "grad_norm": 0.0503549413885895,
      "learning_rate": 2.0273205857330723e-07,
      "loss": 0.003,
      "step": 3929
    },
    {
      "epoch": 2.9460269865067468,
      "grad_norm": 0.11413117402719496,
      "learning_rate": 1.972176126679881e-07,
      "loss": 0.0062,
      "step": 3930
    },
    {
      "epoch": 2.946776611694153,
      "grad_norm": 0.04264351749436684,
      "learning_rate": 1.9177912871606307e-07,
      "loss": 0.002,
      "step": 3931
    },
    {
      "epoch": 2.947526236881559,
      "grad_norm": 0.10185445906343903,
      "learning_rate": 1.8641661085685436e-07,
      "loss": 0.0061,
      "step": 3932
    },
    {
      "epoch": 2.9482758620689653,
      "grad_norm": 0.22741461874925956,
      "learning_rate": 1.8113006317188596e-07,
      "loss": 0.0182,
      "step": 3933
    },
    {
      "epoch": 2.949025487256372,
      "grad_norm": 0.02685321960312984,
      "learning_rate": 1.759194896848615e-07,
      "loss": 0.0015,
      "step": 3934
    },
    {
      "epoch": 2.949775112443778,
      "grad_norm": 0.04166166861207708,
      "learning_rate": 1.7078489436165301e-07,
      "loss": 0.001,
      "step": 3935
    },
    {
      "epoch": 2.9505247376311843,
      "grad_norm": 0.12658296999295252,
      "learning_rate": 1.657262811103122e-07,
      "loss": 0.0043,
      "step": 3936
    },
    {
      "epoch": 2.951274362818591,
      "grad_norm": 0.07800235121237264,
      "learning_rate": 1.6074365378105915e-07,
      "loss": 0.0028,
      "step": 3937
    },
    {
      "epoch": 2.952023988005997,
      "grad_norm": 0.1575521316690672,
      "learning_rate": 1.5583701616626034e-07,
      "loss": 0.0053,
      "step": 3938
    },
    {
      "epoch": 2.9527736131934033,
      "grad_norm": 0.049391341850405314,
      "learning_rate": 1.5100637200046176e-07,
      "loss": 0.0015,
      "step": 3939
    },
    {
      "epoch": 2.9535232383808094,
      "grad_norm": 0.08463145717344953,
      "learning_rate": 1.4625172496036676e-07,
      "loss": 0.0101,
      "step": 3940
    },
    {
      "epoch": 2.954272863568216,
      "grad_norm": 0.0913568967277725,
      "learning_rate": 1.4157307866484726e-07,
      "loss": 0.005,
      "step": 3941
    },
    {
      "epoch": 2.9550224887556222,
      "grad_norm": 0.16458575578486678,
      "learning_rate": 1.3697043667489918e-07,
      "loss": 0.0041,
      "step": 3942
    },
    {
      "epoch": 2.9557721139430284,
      "grad_norm": 0.03985907742532866,
      "learning_rate": 1.3244380249369804e-07,
      "loss": 0.001,
      "step": 3943
    },
    {
      "epoch": 2.9565217391304346,
      "grad_norm": 0.11386553888333875,
      "learning_rate": 1.2799317956655454e-07,
      "loss": 0.0065,
      "step": 3944
    },
    {
      "epoch": 2.957271364317841,
      "grad_norm": 0.09145340692489545,
      "learning_rate": 1.236185712809368e-07,
      "loss": 0.0078,
      "step": 3945
    },
    {
      "epoch": 2.9580209895052474,
      "grad_norm": 0.1222934590256073,
      "learning_rate": 1.1931998096644804e-07,
      "loss": 0.0055,
      "step": 3946
    },
    {
      "epoch": 2.9587706146926536,
      "grad_norm": 0.1625788169609916,
      "learning_rate": 1.150974118948156e-07,
      "loss": 0.0077,
      "step": 3947
    },
    {
      "epoch": 2.95952023988006,
      "grad_norm": 0.10754275067318649,
      "learning_rate": 1.1095086727994641e-07,
      "loss": 0.0054,
      "step": 3948
    },
    {
      "epoch": 2.9602698650674664,
      "grad_norm": 0.07812351741966694,
      "learning_rate": 1.0688035027786036e-07,
      "loss": 0.0046,
      "step": 3949
    },
    {
      "epoch": 2.9610194902548725,
      "grad_norm": 0.06017618913652728,
      "learning_rate": 1.0288586398670142e-07,
      "loss": 0.0019,
      "step": 3950
    },
    {
      "epoch": 2.9617691154422787,
      "grad_norm": 0.06594705186407322,
      "learning_rate": 9.89674114467487e-08,
      "loss": 0.0062,
      "step": 3951
    },
    {
      "epoch": 2.962518740629685,
      "grad_norm": 0.05508768796277397,
      "learning_rate": 9.512499564042764e-08,
      "loss": 0.0027,
      "step": 3952
    },
    {
      "epoch": 2.9632683658170915,
      "grad_norm": 0.07035356274760843,
      "learning_rate": 9.135861949228774e-08,
      "loss": 0.0029,
      "step": 3953
    },
    {
      "epoch": 2.9640179910044977,
      "grad_norm": 0.09665084820902284,
      "learning_rate": 8.766828586898035e-08,
      "loss": 0.0035,
      "step": 3954
    },
    {
      "epoch": 2.964767616191904,
      "grad_norm": 0.15054682362176386,
      "learning_rate": 8.40539975793031e-08,
      "loss": 0.006,
      "step": 3955
    },
    {
      "epoch": 2.9655172413793105,
      "grad_norm": 0.05185536960868925,
      "learning_rate": 8.051575737416661e-08,
      "loss": 0.0669,
      "step": 3956
    },
    {
      "epoch": 2.9662668665667167,
      "grad_norm": 0.04477626103777263,
      "learning_rate": 7.705356794659446e-08,
      "loss": 0.0017,
      "step": 3957
    },
    {
      "epoch": 2.967016491754123,
      "grad_norm": 0.21173344383560222,
      "learning_rate": 7.36674319317232e-08,
      "loss": 0.0099,
      "step": 3958
    },
    {
      "epoch": 2.9677661169415295,
      "grad_norm": 0.08864354866618206,
      "learning_rate": 7.035735190682458e-08,
      "loss": 0.0067,
      "step": 3959
    },
    {
      "epoch": 2.9685157421289357,
      "grad_norm": 0.08168212547347438,
      "learning_rate": 6.712333039126107e-08,
      "loss": 0.0033,
      "step": 3960
    },
    {
      "epoch": 2.969265367316342,
      "grad_norm": 0.1056606139859158,
      "learning_rate": 6.396536984650814e-08,
      "loss": 0.0167,
      "step": 3961
    },
    {
      "epoch": 2.970014992503748,
      "grad_norm": 0.23985455411275883,
      "learning_rate": 6.088347267616534e-08,
      "loss": 0.0083,
      "step": 3962
    },
    {
      "epoch": 2.970764617691154,
      "grad_norm": 0.0962979369431135,
      "learning_rate": 5.787764122592298e-08,
      "loss": 0.0018,
      "step": 3963
    },
    {
      "epoch": 2.971514242878561,
      "grad_norm": 0.07750924360925224,
      "learning_rate": 5.4947877783584344e-08,
      "loss": 0.0057,
      "step": 3964
    },
    {
      "epoch": 2.972263868065967,
      "grad_norm": 0.10843891872380376,
      "learning_rate": 5.209418457904347e-08,
      "loss": 0.0051,
      "step": 3965
    },
    {
      "epoch": 2.973013493253373,
      "grad_norm": 0.0920430540669819,
      "learning_rate": 4.931656378431848e-08,
      "loss": 0.0067,
      "step": 3966
    },
    {
      "epoch": 2.97376311844078,
      "grad_norm": 0.09960702743932079,
      "learning_rate": 4.661501751349606e-08,
      "loss": 0.0057,
      "step": 3967
    },
    {
      "epoch": 2.974512743628186,
      "grad_norm": 0.03518596057286467,
      "learning_rate": 4.3989547822798074e-08,
      "loss": 0.002,
      "step": 3968
    },
    {
      "epoch": 2.975262368815592,
      "grad_norm": 0.1287565629497624,
      "learning_rate": 4.144015671051493e-08,
      "loss": 0.0029,
      "step": 3969
    },
    {
      "epoch": 2.9760119940029988,
      "grad_norm": 0.07846377737079575,
      "learning_rate": 3.896684611705004e-08,
      "loss": 0.006,
      "step": 3970
    },
    {
      "epoch": 2.976761619190405,
      "grad_norm": 0.14591446042608083,
      "learning_rate": 3.656961792486424e-08,
      "loss": 0.003,
      "step": 3971
    },
    {
      "epoch": 2.977511244377811,
      "grad_norm": 0.17905020957307904,
      "learning_rate": 3.424847395856467e-08,
      "loss": 0.005,
      "step": 3972
    },
    {
      "epoch": 2.9782608695652173,
      "grad_norm": 0.01906684909524204,
      "learning_rate": 3.2003415984815934e-08,
      "loss": 0.0009,
      "step": 3973
    },
    {
      "epoch": 2.9790104947526235,
      "grad_norm": 0.11418258519719206,
      "learning_rate": 2.9834445712362266e-08,
      "loss": 0.0036,
      "step": 3974
    },
    {
      "epoch": 2.97976011994003,
      "grad_norm": 0.11876471382011124,
      "learning_rate": 2.7741564792072018e-08,
      "loss": 0.0068,
      "step": 3975
    },
    {
      "epoch": 2.9805097451274363,
      "grad_norm": 0.08256987244801461,
      "learning_rate": 2.5724774816870965e-08,
      "loss": 0.0052,
      "step": 3976
    },
    {
      "epoch": 2.9812593703148424,
      "grad_norm": 0.08847358453985199,
      "learning_rate": 2.3784077321775678e-08,
      "loss": 0.0094,
      "step": 3977
    },
    {
      "epoch": 2.982008995502249,
      "grad_norm": 0.0710505712846508,
      "learning_rate": 2.1919473783904575e-08,
      "loss": 0.0073,
      "step": 3978
    },
    {
      "epoch": 2.9827586206896552,
      "grad_norm": 0.05908748373326097,
      "learning_rate": 2.013096562242245e-08,
      "loss": 0.0013,
      "step": 3979
    },
    {
      "epoch": 2.9835082458770614,
      "grad_norm": 0.03487128992436554,
      "learning_rate": 1.8418554198629257e-08,
      "loss": 0.0011,
      "step": 3980
    },
    {
      "epoch": 2.9842578710644676,
      "grad_norm": 0.03919515081336705,
      "learning_rate": 1.6782240815849116e-08,
      "loss": 0.0018,
      "step": 3981
    },
    {
      "epoch": 2.9850074962518742,
      "grad_norm": 0.032663649258333316,
      "learning_rate": 1.5222026719530213e-08,
      "loss": 0.0017,
      "step": 3982
    },
    {
      "epoch": 2.9857571214392804,
      "grad_norm": 0.26096980807808,
      "learning_rate": 1.3737913097178201e-08,
      "loss": 0.0058,
      "step": 3983
    },
    {
      "epoch": 2.9865067466266866,
      "grad_norm": 0.12084288234059795,
      "learning_rate": 1.2329901078378392e-08,
      "loss": 0.0024,
      "step": 3984
    },
    {
      "epoch": 2.9872563718140928,
      "grad_norm": 0.06595470019370689,
      "learning_rate": 1.0997991734806868e-08,
      "loss": 0.0048,
      "step": 3985
    },
    {
      "epoch": 2.9880059970014994,
      "grad_norm": 0.1666299653323181,
      "learning_rate": 9.742186080208271e-09,
      "loss": 0.0096,
      "step": 3986
    },
    {
      "epoch": 2.9887556221889056,
      "grad_norm": 0.19747301065827289,
      "learning_rate": 8.56248507039581e-09,
      "loss": 0.0186,
      "step": 3987
    },
    {
      "epoch": 2.9895052473763117,
      "grad_norm": 0.076916580436097,
      "learning_rate": 7.458889603262353e-09,
      "loss": 0.006,
      "step": 3988
    },
    {
      "epoch": 2.9902548725637184,
      "grad_norm": 0.06781979930201935,
      "learning_rate": 6.4314005187804394e-09,
      "loss": 0.0021,
      "step": 3989
    },
    {
      "epoch": 2.9910044977511245,
      "grad_norm": 0.06140423539445794,
      "learning_rate": 5.480018598991166e-09,
      "loss": 0.0019,
      "step": 3990
    },
    {
      "epoch": 2.9917541229385307,
      "grad_norm": 0.1661028082846302,
      "learning_rate": 4.604744568015296e-09,
      "loss": 0.0034,
      "step": 3991
    },
    {
      "epoch": 2.992503748125937,
      "grad_norm": 0.04500281969970609,
      "learning_rate": 3.805579092042155e-09,
      "loss": 0.003,
      "step": 3992
    },
    {
      "epoch": 2.993253373313343,
      "grad_norm": 0.07088654187160519,
      "learning_rate": 3.0825227793296328e-09,
      "loss": 0.0037,
      "step": 3993
    },
    {
      "epoch": 2.9940029985007497,
      "grad_norm": 0.15553196796955818,
      "learning_rate": 2.43557618020418e-09,
      "loss": 0.0053,
      "step": 3994
    },
    {
      "epoch": 2.994752623688156,
      "grad_norm": 0.14671990188627287,
      "learning_rate": 1.864739787083014e-09,
      "loss": 0.0077,
      "step": 3995
    },
    {
      "epoch": 2.995502248875562,
      "grad_norm": 0.04418675692717229,
      "learning_rate": 1.3700140344408141e-09,
      "loss": 0.0022,
      "step": 3996
    },
    {
      "epoch": 2.9962518740629687,
      "grad_norm": 0.1705912775123118,
      "learning_rate": 9.513992988097188e-10,
      "loss": 0.016,
      "step": 3997
    },
    {
      "epoch": 2.997001499250375,
      "grad_norm": 0.1381779105328255,
      "learning_rate": 6.088958988237359e-10,
      "loss": 0.0071,
      "step": 3998
    },
    {
      "epoch": 2.997751124437781,
      "grad_norm": 0.04728810879882926,
      "learning_rate": 3.425040951521297e-10,
      "loss": 0.0027,
      "step": 3999
    },
    {
      "epoch": 2.9985007496251876,
      "grad_norm": 0.08131220348567887,
      "learning_rate": 1.5222409056603327e-10,
      "loss": 0.0013,
      "step": 4000
    },
    {
      "epoch": 2.999250374812594,
      "grad_norm": 0.08993400957198339,
      "learning_rate": 3.805602988293799e-11,
      "loss": 0.0045,
      "step": 4001
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.04477219697998604,
      "learning_rate": 0.0,
      "loss": 0.0009,
      "step": 4002
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.03329204395413399,
      "eval_runtime": 1896.306,
      "eval_samples_per_second": 5.472,
      "eval_steps_per_second": 0.684,
      "step": 4002
    },
    {
      "epoch": 3.0,
      "step": 4002,
      "total_flos": 3989157755650048.0,
      "train_loss": 0.020737408151027925,
      "train_runtime": 46333.6393,
      "train_samples_per_second": 1.382,
      "train_steps_per_second": 0.086
    }
  ],
  "logging_steps": 1,
  "max_steps": 4002,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3989157755650048.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}