{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9939263407279775,
  "eval_steps": 500,
  "global_step": 1086,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0027568382511307344,
      "grad_norm": 7.122869968414307,
      "learning_rate": 6.060606060606061e-07,
      "loss": 0.4664,
      "step": 1
    },
    {
      "epoch": 0.005513676502261469,
      "grad_norm": 7.916077136993408,
      "learning_rate": 1.2121212121212122e-06,
      "loss": 0.4771,
      "step": 2
    },
    {
      "epoch": 0.008270514753392204,
      "grad_norm": 7.081696033477783,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 0.4614,
      "step": 3
    },
    {
      "epoch": 0.011027353004522938,
      "grad_norm": 5.813568115234375,
      "learning_rate": 2.4242424242424244e-06,
      "loss": 0.4299,
      "step": 4
    },
    {
      "epoch": 0.013784191255653673,
      "grad_norm": 4.66733455657959,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 0.3728,
      "step": 5
    },
    {
      "epoch": 0.01654102950678441,
      "grad_norm": 6.349670886993408,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 0.3623,
      "step": 6
    },
    {
      "epoch": 0.01929786775791514,
      "grad_norm": 3.68088436126709,
      "learning_rate": 4.242424242424243e-06,
      "loss": 0.2288,
      "step": 7
    },
    {
      "epoch": 0.022054706009045875,
      "grad_norm": 2.508946418762207,
      "learning_rate": 4.848484848484849e-06,
      "loss": 0.1955,
      "step": 8
    },
    {
      "epoch": 0.02481154426017661,
      "grad_norm": 7.334312915802002,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 0.1915,
      "step": 9
    },
    {
      "epoch": 0.027568382511307346,
      "grad_norm": 6.7029619216918945,
      "learning_rate": 6.060606060606061e-06,
      "loss": 0.1546,
      "step": 10
    },
    {
      "epoch": 0.030325220762438078,
      "grad_norm": 4.743812561035156,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.1565,
      "step": 11
    },
    {
      "epoch": 0.03308205901356882,
      "grad_norm": 1.9065254926681519,
      "learning_rate": 7.272727272727273e-06,
      "loss": 0.1344,
      "step": 12
    },
    {
      "epoch": 0.03583889726469955,
      "grad_norm": 1.0832651853561401,
      "learning_rate": 7.87878787878788e-06,
      "loss": 0.1244,
      "step": 13
    },
    {
      "epoch": 0.03859573551583028,
      "grad_norm": 1.2000082731246948,
      "learning_rate": 8.484848484848486e-06,
      "loss": 0.1115,
      "step": 14
    },
    {
      "epoch": 0.04135257376696102,
      "grad_norm": 0.937786877155304,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.1026,
      "step": 15
    },
    {
      "epoch": 0.04410941201809175,
      "grad_norm": 1.0514339208602905,
      "learning_rate": 9.696969696969698e-06,
      "loss": 0.0976,
      "step": 16
    },
    {
      "epoch": 0.04686625026922248,
      "grad_norm": 3.448519706726074,
      "learning_rate": 1.0303030303030304e-05,
      "loss": 0.1035,
      "step": 17
    },
    {
      "epoch": 0.04962308852035322,
      "grad_norm": 1.5411934852600098,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 0.0969,
      "step": 18
    },
    {
      "epoch": 0.05237992677148395,
      "grad_norm": 0.6320972442626953,
      "learning_rate": 1.1515151515151517e-05,
      "loss": 0.0884,
      "step": 19
    },
    {
      "epoch": 0.05513676502261469,
      "grad_norm": 0.967760443687439,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 0.0851,
      "step": 20
    },
    {
      "epoch": 0.057893603273745424,
      "grad_norm": 0.5067126750946045,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 0.079,
      "step": 21
    },
    {
      "epoch": 0.060650441524876156,
      "grad_norm": 0.5082628726959229,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.0787,
      "step": 22
    },
    {
      "epoch": 0.0634072797760069,
      "grad_norm": 1.0926717519760132,
      "learning_rate": 1.3939393939393942e-05,
      "loss": 0.0751,
      "step": 23
    },
    {
      "epoch": 0.06616411802713763,
      "grad_norm": 0.43889403343200684,
      "learning_rate": 1.4545454545454546e-05,
      "loss": 0.0696,
      "step": 24
    },
    {
      "epoch": 0.06892095627826836,
      "grad_norm": 0.3727055788040161,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 0.0693,
      "step": 25
    },
    {
      "epoch": 0.0716777945293991,
      "grad_norm": 0.4280584454536438,
      "learning_rate": 1.575757575757576e-05,
      "loss": 0.0727,
      "step": 26
    },
    {
      "epoch": 0.07443463278052984,
      "grad_norm": 0.36678358912467957,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 0.0654,
      "step": 27
    },
    {
      "epoch": 0.07719147103166056,
      "grad_norm": 0.3085872530937195,
      "learning_rate": 1.6969696969696972e-05,
      "loss": 0.0614,
      "step": 28
    },
    {
      "epoch": 0.0799483092827913,
      "grad_norm": 0.2715630829334259,
      "learning_rate": 1.7575757575757576e-05,
      "loss": 0.0598,
      "step": 29
    },
    {
      "epoch": 0.08270514753392204,
      "grad_norm": 0.36059409379959106,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.0602,
      "step": 30
    },
    {
      "epoch": 0.08546198578505276,
      "grad_norm": 0.3720782995223999,
      "learning_rate": 1.8787878787878792e-05,
      "loss": 0.0602,
      "step": 31
    },
    {
      "epoch": 0.0882188240361835,
      "grad_norm": 0.24194574356079102,
      "learning_rate": 1.9393939393939395e-05,
      "loss": 0.0553,
      "step": 32
    },
    {
      "epoch": 0.09097566228731424,
      "grad_norm": 0.5713000893592834,
      "learning_rate": 2e-05,
      "loss": 0.0566,
      "step": 33
    },
    {
      "epoch": 0.09373250053844497,
      "grad_norm": 0.4913846552371979,
      "learning_rate": 1.9999955494602408e-05,
      "loss": 0.0582,
      "step": 34
    },
    {
      "epoch": 0.0964893387895757,
      "grad_norm": 0.26757943630218506,
      "learning_rate": 1.999982197880577e-05,
      "loss": 0.0544,
      "step": 35
    },
    {
      "epoch": 0.09924617704070644,
      "grad_norm": 0.31889644265174866,
      "learning_rate": 1.9999599453798523e-05,
      "loss": 0.055,
      "step": 36
    },
    {
      "epoch": 0.10200301529183717,
      "grad_norm": 0.26153671741485596,
      "learning_rate": 1.9999287921561385e-05,
      "loss": 0.0508,
      "step": 37
    },
    {
      "epoch": 0.1047598535429679,
      "grad_norm": 0.2633945345878601,
      "learning_rate": 1.9998887384867323e-05,
      "loss": 0.0506,
      "step": 38
    },
    {
      "epoch": 0.10751669179409865,
      "grad_norm": 0.22508633136749268,
      "learning_rate": 1.9998397847281548e-05,
      "loss": 0.0511,
      "step": 39
    },
    {
      "epoch": 0.11027353004522938,
      "grad_norm": 0.23420976102352142,
      "learning_rate": 1.9997819313161476e-05,
      "loss": 0.0504,
      "step": 40
    },
    {
      "epoch": 0.11303036829636011,
      "grad_norm": 0.3166053295135498,
      "learning_rate": 1.9997151787656678e-05,
      "loss": 0.0564,
      "step": 41
    },
    {
      "epoch": 0.11578720654749085,
      "grad_norm": 0.19986656308174133,
      "learning_rate": 1.9996395276708856e-05,
      "loss": 0.0512,
      "step": 42
    },
    {
      "epoch": 0.11854404479862159,
      "grad_norm": 0.274948388338089,
      "learning_rate": 1.9995549787051772e-05,
      "loss": 0.0489,
      "step": 43
    },
    {
      "epoch": 0.12130088304975231,
      "grad_norm": 0.23236143589019775,
      "learning_rate": 1.9994615326211203e-05,
      "loss": 0.0491,
      "step": 44
    },
    {
      "epoch": 0.12405772130088305,
      "grad_norm": 0.22261761128902435,
      "learning_rate": 1.9993591902504854e-05,
      "loss": 0.048,
      "step": 45
    },
    {
      "epoch": 0.1268145595520138,
      "grad_norm": 0.22965197265148163,
      "learning_rate": 1.9992479525042305e-05,
      "loss": 0.0473,
      "step": 46
    },
    {
      "epoch": 0.12957139780314453,
      "grad_norm": 0.22254930436611176,
      "learning_rate": 1.9991278203724908e-05,
      "loss": 0.0468,
      "step": 47
    },
    {
      "epoch": 0.13232823605427527,
      "grad_norm": 0.18861782550811768,
      "learning_rate": 1.9989987949245725e-05,
      "loss": 0.0446,
      "step": 48
    },
    {
      "epoch": 0.13508507430540598,
      "grad_norm": 0.2599838078022003,
      "learning_rate": 1.9988608773089413e-05,
      "loss": 0.0467,
      "step": 49
    },
    {
      "epoch": 0.13784191255653672,
      "grad_norm": 0.24845938384532928,
      "learning_rate": 1.998714068753213e-05,
      "loss": 0.0461,
      "step": 50
    },
    {
      "epoch": 0.14059875080766746,
      "grad_norm": 0.18050189316272736,
      "learning_rate": 1.9985583705641418e-05,
      "loss": 0.0459,
      "step": 51
    },
    {
      "epoch": 0.1433555890587982,
      "grad_norm": 0.2283959537744522,
      "learning_rate": 1.9983937841276103e-05,
      "loss": 0.0452,
      "step": 52
    },
    {
      "epoch": 0.14611242730992893,
      "grad_norm": 0.21824775636196136,
      "learning_rate": 1.9982203109086153e-05,
      "loss": 0.044,
      "step": 53
    },
    {
      "epoch": 0.14886926556105967,
      "grad_norm": 0.2457500696182251,
      "learning_rate": 1.998037952451255e-05,
      "loss": 0.0445,
      "step": 54
    },
    {
      "epoch": 0.15162610381219038,
      "grad_norm": 0.2404290735721588,
      "learning_rate": 1.9978467103787176e-05,
      "loss": 0.0461,
      "step": 55
    },
    {
      "epoch": 0.15438294206332112,
      "grad_norm": 0.24108456075191498,
      "learning_rate": 1.9976465863932632e-05,
      "loss": 0.044,
      "step": 56
    },
    {
      "epoch": 0.15713978031445186,
      "grad_norm": 0.19976645708084106,
      "learning_rate": 1.9974375822762117e-05,
      "loss": 0.0434,
      "step": 57
    },
    {
      "epoch": 0.1598966185655826,
      "grad_norm": 0.20052076876163483,
      "learning_rate": 1.9972196998879254e-05,
      "loss": 0.0442,
      "step": 58
    },
    {
      "epoch": 0.16265345681671334,
      "grad_norm": 0.19367817044258118,
      "learning_rate": 1.996992941167792e-05,
      "loss": 0.0436,
      "step": 59
    },
    {
      "epoch": 0.16541029506784408,
      "grad_norm": 0.21227090060710907,
      "learning_rate": 1.9967573081342103e-05,
      "loss": 0.0439,
      "step": 60
    },
    {
      "epoch": 0.1681671333189748,
      "grad_norm": 0.20237918198108673,
      "learning_rate": 1.9965128028845676e-05,
      "loss": 0.0414,
      "step": 61
    },
    {
      "epoch": 0.17092397157010553,
      "grad_norm": 0.16752171516418457,
      "learning_rate": 1.9962594275952246e-05,
      "loss": 0.0426,
      "step": 62
    },
    {
      "epoch": 0.17368080982123626,
      "grad_norm": 0.1880505383014679,
      "learning_rate": 1.9959971845214953e-05,
      "loss": 0.0414,
      "step": 63
    },
    {
      "epoch": 0.176437648072367,
      "grad_norm": 0.2021344006061554,
      "learning_rate": 1.995726075997626e-05,
      "loss": 0.0423,
      "step": 64
    },
    {
      "epoch": 0.17919448632349774,
      "grad_norm": 0.15245415270328522,
      "learning_rate": 1.9954461044367752e-05,
      "loss": 0.0413,
      "step": 65
    },
    {
      "epoch": 0.18195132457462848,
      "grad_norm": 0.17980217933654785,
      "learning_rate": 1.9951572723309918e-05,
      "loss": 0.0421,
      "step": 66
    },
    {
      "epoch": 0.18470816282575922,
      "grad_norm": 0.1832534670829773,
      "learning_rate": 1.994859582251194e-05,
      "loss": 0.0416,
      "step": 67
    },
    {
      "epoch": 0.18746500107688993,
      "grad_norm": 0.6252762079238892,
      "learning_rate": 1.9945530368471444e-05,
      "loss": 0.047,
      "step": 68
    },
    {
      "epoch": 0.19022183932802067,
      "grad_norm": 0.17044250667095184,
      "learning_rate": 1.9942376388474282e-05,
      "loss": 0.0406,
      "step": 69
    },
    {
      "epoch": 0.1929786775791514,
      "grad_norm": 0.1817871630191803,
      "learning_rate": 1.9939133910594276e-05,
      "loss": 0.042,
      "step": 70
    },
    {
      "epoch": 0.19573551583028215,
      "grad_norm": 0.1658797264099121,
      "learning_rate": 1.9935802963692988e-05,
      "loss": 0.041,
      "step": 71
    },
    {
      "epoch": 0.1984923540814129,
      "grad_norm": 0.18323002755641937,
      "learning_rate": 1.9932383577419432e-05,
      "loss": 0.0407,
      "step": 72
    },
    {
      "epoch": 0.20124919233254362,
      "grad_norm": 0.200252503156662,
      "learning_rate": 1.992887578220984e-05,
      "loss": 0.0427,
      "step": 73
    },
    {
      "epoch": 0.20400603058367434,
      "grad_norm": 0.19649459421634674,
      "learning_rate": 1.9925279609287384e-05,
      "loss": 0.0439,
      "step": 74
    },
    {
      "epoch": 0.20676286883480507,
      "grad_norm": 0.17329928278923035,
      "learning_rate": 1.9921595090661872e-05,
      "loss": 0.0415,
      "step": 75
    },
    {
      "epoch": 0.2095197070859358,
      "grad_norm": 0.4066649377346039,
      "learning_rate": 1.9917822259129508e-05,
      "loss": 0.0416,
      "step": 76
    },
    {
      "epoch": 0.21227654533706655,
      "grad_norm": 0.1922263205051422,
      "learning_rate": 1.991396114827256e-05,
      "loss": 0.041,
      "step": 77
    },
    {
      "epoch": 0.2150333835881973,
      "grad_norm": 0.1683184802532196,
      "learning_rate": 1.9910011792459086e-05,
      "loss": 0.0411,
      "step": 78
    },
    {
      "epoch": 0.21779022183932803,
      "grad_norm": 0.43947067856788635,
      "learning_rate": 1.9905974226842614e-05,
      "loss": 0.0393,
      "step": 79
    },
    {
      "epoch": 0.22054706009045877,
      "grad_norm": 0.22764648497104645,
      "learning_rate": 1.9901848487361834e-05,
      "loss": 0.0415,
      "step": 80
    },
    {
      "epoch": 0.22330389834158948,
      "grad_norm": 0.1933886855840683,
      "learning_rate": 1.989763461074029e-05,
      "loss": 0.0404,
      "step": 81
    },
    {
      "epoch": 0.22606073659272022,
      "grad_norm": 0.1945038139820099,
      "learning_rate": 1.989333263448602e-05,
      "loss": 0.0398,
      "step": 82
    },
    {
      "epoch": 0.22881757484385096,
      "grad_norm": 0.21237650513648987,
      "learning_rate": 1.9888942596891267e-05,
      "loss": 0.0406,
      "step": 83
    },
    {
      "epoch": 0.2315744130949817,
      "grad_norm": 0.19435830414295197,
      "learning_rate": 1.9884464537032103e-05,
      "loss": 0.0397,
      "step": 84
    },
    {
      "epoch": 0.23433125134611243,
      "grad_norm": 0.17015162110328674,
      "learning_rate": 1.9879898494768093e-05,
      "loss": 0.0398,
      "step": 85
    },
    {
      "epoch": 0.23708808959724317,
      "grad_norm": 0.19542446732521057,
      "learning_rate": 1.987524451074194e-05,
      "loss": 0.0386,
      "step": 86
    },
    {
      "epoch": 0.23984492784837388,
      "grad_norm": 0.23729243874549866,
      "learning_rate": 1.9870502626379127e-05,
      "loss": 0.0406,
      "step": 87
    },
    {
      "epoch": 0.24260176609950462,
      "grad_norm": 0.18620969355106354,
      "learning_rate": 1.9865672883887553e-05,
      "loss": 0.0395,
      "step": 88
    },
    {
      "epoch": 0.24535860435063536,
      "grad_norm": 0.254768043756485,
      "learning_rate": 1.9860755326257127e-05,
      "loss": 0.0405,
      "step": 89
    },
    {
      "epoch": 0.2481154426017661,
      "grad_norm": 0.2777726650238037,
      "learning_rate": 1.985574999725943e-05,
      "loss": 0.0406,
      "step": 90
    },
    {
      "epoch": 0.2508722808528968,
      "grad_norm": 0.23211295902729034,
      "learning_rate": 1.985065694144728e-05,
      "loss": 0.0389,
      "step": 91
    },
    {
      "epoch": 0.2536291191040276,
      "grad_norm": 0.22822345793247223,
      "learning_rate": 1.9845476204154387e-05,
      "loss": 0.0393,
      "step": 92
    },
    {
      "epoch": 0.2563859573551583,
      "grad_norm": 0.2358836829662323,
      "learning_rate": 1.9840207831494903e-05,
      "loss": 0.0383,
      "step": 93
    },
    {
      "epoch": 0.25914279560628906,
      "grad_norm": 0.2143673151731491,
      "learning_rate": 1.9834851870363024e-05,
      "loss": 0.0411,
      "step": 94
    },
    {
      "epoch": 0.26189963385741977,
      "grad_norm": 0.21156637370586395,
      "learning_rate": 1.9829408368432592e-05,
      "loss": 0.0393,
      "step": 95
    },
    {
      "epoch": 0.26465647210855053,
      "grad_norm": 0.22637519240379333,
      "learning_rate": 1.9823877374156647e-05,
      "loss": 0.0375,
      "step": 96
    },
    {
      "epoch": 0.26741331035968124,
      "grad_norm": 0.2607465386390686,
      "learning_rate": 1.9818258936767013e-05,
      "loss": 0.0371,
      "step": 97
    },
    {
      "epoch": 0.27017014861081196,
      "grad_norm": 0.1963801383972168,
      "learning_rate": 1.9812553106273848e-05,
      "loss": 0.0377,
      "step": 98
    },
    {
      "epoch": 0.2729269868619427,
      "grad_norm": 0.1716557741165161,
      "learning_rate": 1.98067599334652e-05,
      "loss": 0.0378,
      "step": 99
    },
    {
      "epoch": 0.27568382511307343,
      "grad_norm": 0.18233723938465118,
      "learning_rate": 1.980087946990656e-05,
      "loss": 0.0385,
      "step": 100
    },
    {
      "epoch": 0.2784406633642042,
      "grad_norm": 0.19480392336845398,
      "learning_rate": 1.9794911767940405e-05,
      "loss": 0.0377,
      "step": 101
    },
    {
      "epoch": 0.2811975016153349,
      "grad_norm": 0.21586693823337555,
      "learning_rate": 1.978885688068572e-05,
      "loss": 0.0366,
      "step": 102
    },
    {
      "epoch": 0.2839543398664656,
      "grad_norm": 0.4663015305995941,
      "learning_rate": 1.9782714862037544e-05,
      "loss": 0.0397,
      "step": 103
    },
    {
      "epoch": 0.2867111781175964,
      "grad_norm": 0.43779700994491577,
      "learning_rate": 1.977648576666647e-05,
      "loss": 0.0405,
      "step": 104
    },
    {
      "epoch": 0.2894680163687271,
      "grad_norm": 0.15922895073890686,
      "learning_rate": 1.977016965001817e-05,
      "loss": 0.0365,
      "step": 105
    },
    {
      "epoch": 0.29222485461985787,
      "grad_norm": 0.20054024457931519,
      "learning_rate": 1.9763766568312906e-05,
      "loss": 0.0397,
      "step": 106
    },
    {
      "epoch": 0.2949816928709886,
      "grad_norm": 0.23604053258895874,
      "learning_rate": 1.9757276578545013e-05,
      "loss": 0.0379,
      "step": 107
    },
    {
      "epoch": 0.29773853112211934,
      "grad_norm": 0.22856947779655457,
      "learning_rate": 1.9750699738482403e-05,
      "loss": 0.041,
      "step": 108
    },
    {
      "epoch": 0.30049536937325005,
      "grad_norm": 0.1889025717973709,
      "learning_rate": 1.974403610666606e-05,
      "loss": 0.0376,
      "step": 109
    },
    {
      "epoch": 0.30325220762438077,
      "grad_norm": 0.16721414029598236,
      "learning_rate": 1.9737285742409495e-05,
      "loss": 0.0393,
      "step": 110
    },
    {
      "epoch": 0.30600904587551153,
      "grad_norm": 0.19392457604408264,
      "learning_rate": 1.973044870579824e-05,
      "loss": 0.0382,
      "step": 111
    },
    {
      "epoch": 0.30876588412664224,
      "grad_norm": 0.17549856007099152,
      "learning_rate": 1.97235250576893e-05,
      "loss": 0.0386,
      "step": 112
    },
    {
      "epoch": 0.311522722377773,
      "grad_norm": 0.15814520418643951,
      "learning_rate": 1.971651485971062e-05,
      "loss": 0.0387,
      "step": 113
    },
    {
      "epoch": 0.3142795606289037,
      "grad_norm": 0.5654919147491455,
      "learning_rate": 1.9709418174260523e-05,
      "loss": 0.0403,
      "step": 114
    },
    {
      "epoch": 0.3170363988800345,
      "grad_norm": 0.16663803160190582,
      "learning_rate": 1.9702235064507175e-05,
      "loss": 0.0355,
      "step": 115
    },
    {
      "epoch": 0.3197932371311652,
      "grad_norm": 0.17256957292556763,
      "learning_rate": 1.9694965594388008e-05,
      "loss": 0.0379,
      "step": 116
    },
    {
      "epoch": 0.3225500753822959,
      "grad_norm": 0.23029914498329163,
      "learning_rate": 1.9687609828609156e-05,
      "loss": 0.0384,
      "step": 117
    },
    {
      "epoch": 0.3253069136334267,
      "grad_norm": 0.1971399337053299,
      "learning_rate": 1.9680167832644868e-05,
      "loss": 0.0385,
      "step": 118
    },
    {
      "epoch": 0.3280637518845574,
      "grad_norm": 0.21078315377235413,
      "learning_rate": 1.9672639672736947e-05,
      "loss": 0.0386,
      "step": 119
    },
    {
      "epoch": 0.33082059013568815,
      "grad_norm": 0.20524373650550842,
      "learning_rate": 1.966502541589414e-05,
      "loss": 0.0365,
      "step": 120
    },
    {
      "epoch": 0.33357742838681886,
      "grad_norm": 0.17085903882980347,
      "learning_rate": 1.9657325129891558e-05,
      "loss": 0.0375,
      "step": 121
    },
    {
      "epoch": 0.3363342666379496,
      "grad_norm": 0.20739462971687317,
      "learning_rate": 1.9649538883270053e-05,
      "loss": 0.0367,
      "step": 122
    },
    {
      "epoch": 0.33909110488908034,
      "grad_norm": 0.17642727494239807,
      "learning_rate": 1.9641666745335626e-05,
      "loss": 0.037,
      "step": 123
    },
    {
      "epoch": 0.34184794314021105,
      "grad_norm": 0.18456482887268066,
      "learning_rate": 1.9633708786158803e-05,
      "loss": 0.038,
      "step": 124
    },
    {
      "epoch": 0.3446047813913418,
      "grad_norm": 0.22288407385349274,
      "learning_rate": 1.962566507657402e-05,
      "loss": 0.0378,
      "step": 125
    },
    {
      "epoch": 0.34736161964247253,
      "grad_norm": 0.20137764513492584,
      "learning_rate": 1.961753568817896e-05,
      "loss": 0.0371,
      "step": 126
    },
    {
      "epoch": 0.3501184578936033,
      "grad_norm": 0.16192878782749176,
      "learning_rate": 1.9609320693333967e-05,
      "loss": 0.0375,
      "step": 127
    },
    {
      "epoch": 0.352875296144734,
      "grad_norm": 0.7257514595985413,
      "learning_rate": 1.960102016516136e-05,
      "loss": 0.0393,
      "step": 128
    },
    {
      "epoch": 0.3556321343958647,
      "grad_norm": 0.15738032758235931,
      "learning_rate": 1.9592634177544803e-05,
      "loss": 0.0374,
      "step": 129
    },
    {
      "epoch": 0.3583889726469955,
      "grad_norm": 0.15156516432762146,
      "learning_rate": 1.9584162805128636e-05,
      "loss": 0.0351,
      "step": 130
    },
    {
      "epoch": 0.3611458108981262,
      "grad_norm": 0.1443106234073639,
      "learning_rate": 1.9575606123317215e-05,
      "loss": 0.0359,
      "step": 131
    },
    {
      "epoch": 0.36390264914925696,
      "grad_norm": 0.16379106044769287,
      "learning_rate": 1.9566964208274254e-05,
      "loss": 0.0376,
      "step": 132
    },
    {
      "epoch": 0.3666594874003877,
      "grad_norm": 0.17260539531707764,
      "learning_rate": 1.9558237136922117e-05,
      "loss": 0.0382,
      "step": 133
    },
    {
      "epoch": 0.36941632565151844,
      "grad_norm": 0.1595146209001541,
      "learning_rate": 1.954942498694117e-05,
      "loss": 0.0378,
      "step": 134
    },
    {
      "epoch": 0.37217316390264915,
      "grad_norm": 0.1662372648715973,
      "learning_rate": 1.9540527836769047e-05,
      "loss": 0.036,
      "step": 135
    },
    {
      "epoch": 0.37493000215377986,
      "grad_norm": 0.17359688878059387,
      "learning_rate": 1.95315457656e-05,
      "loss": 0.0352,
      "step": 136
    },
    {
      "epoch": 0.37768684040491063,
      "grad_norm": 0.15664531290531158,
      "learning_rate": 1.9522478853384154e-05,
      "loss": 0.0372,
      "step": 137
    },
    {
      "epoch": 0.38044367865604134,
      "grad_norm": 0.14340853691101074,
      "learning_rate": 1.951332718082682e-05,
      "loss": 0.0378,
      "step": 138
    },
    {
      "epoch": 0.3832005169071721,
      "grad_norm": 0.16062387824058533,
      "learning_rate": 1.950409082938776e-05,
      "loss": 0.036,
      "step": 139
    },
    {
      "epoch": 0.3859573551583028,
      "grad_norm": 0.16279537975788116,
      "learning_rate": 1.949476988128047e-05,
      "loss": 0.0373,
      "step": 140
    },
    {
      "epoch": 0.38871419340943353,
      "grad_norm": 0.18597126007080078,
      "learning_rate": 1.9485364419471454e-05,
      "loss": 0.0367,
      "step": 141
    },
    {
      "epoch": 0.3914710316605643,
      "grad_norm": 0.6993559002876282,
      "learning_rate": 1.947587452767948e-05,
      "loss": 0.0453,
      "step": 142
    },
    {
      "epoch": 0.394227869911695,
      "grad_norm": 0.20166553556919098,
      "learning_rate": 1.946630029037482e-05,
      "loss": 0.0366,
      "step": 143
    },
    {
      "epoch": 0.3969847081628258,
      "grad_norm": 0.16907399892807007,
      "learning_rate": 1.9456641792778527e-05,
      "loss": 0.036,
      "step": 144
    },
    {
      "epoch": 0.3997415464139565,
      "grad_norm": 0.19966886937618256,
      "learning_rate": 1.9446899120861653e-05,
      "loss": 0.0348,
      "step": 145
    },
    {
      "epoch": 0.40249838466508725,
      "grad_norm": 0.25474515557289124,
      "learning_rate": 1.94370723613445e-05,
      "loss": 0.0388,
      "step": 146
    },
    {
      "epoch": 0.40525522291621796,
      "grad_norm": 0.2152235358953476,
      "learning_rate": 1.9427161601695833e-05,
      "loss": 0.0371,
      "step": 147
    },
    {
      "epoch": 0.40801206116734867,
      "grad_norm": 0.15147989988327026,
      "learning_rate": 1.941716693013211e-05,
      "loss": 0.0374,
      "step": 148
    },
    {
      "epoch": 0.41076889941847944,
      "grad_norm": 0.18094661831855774,
      "learning_rate": 1.94070884356167e-05,
      "loss": 0.036,
      "step": 149
    },
    {
      "epoch": 0.41352573766961015,
      "grad_norm": 0.20204827189445496,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.0383,
      "step": 150
    },
    {
      "epoch": 0.4162825759207409,
      "grad_norm": 0.24904154241085052,
      "learning_rate": 1.938668033731406e-05,
      "loss": 0.0368,
      "step": 151
    },
    {
      "epoch": 0.4190394141718716,
      "grad_norm": 0.15912267565727234,
      "learning_rate": 1.9376350915180935e-05,
      "loss": 0.0352,
      "step": 152
    },
    {
      "epoch": 0.4217962524230024,
      "grad_norm": 0.1520577222108841,
      "learning_rate": 1.9365938033402715e-05,
      "loss": 0.036,
      "step": 153
    },
    {
      "epoch": 0.4245530906741331,
      "grad_norm": 0.1562613546848297,
      "learning_rate": 1.9355441784665295e-05,
      "loss": 0.0367,
      "step": 154
    },
    {
      "epoch": 0.4273099289252638,
      "grad_norm": 0.13876181840896606,
      "learning_rate": 1.9344862262396612e-05,
      "loss": 0.0354,
      "step": 155
    },
    {
      "epoch": 0.4300667671763946,
      "grad_norm": 0.15534985065460205,
      "learning_rate": 1.933419956076584e-05,
      "loss": 0.0358,
      "step": 156
    },
    {
      "epoch": 0.4328236054275253,
      "grad_norm": 0.15327094495296478,
      "learning_rate": 1.932345377468253e-05,
      "loss": 0.0341,
      "step": 157
    },
    {
      "epoch": 0.43558044367865606,
      "grad_norm": 0.1699693650007248,
      "learning_rate": 1.9312624999795784e-05,
      "loss": 0.0378,
      "step": 158
    },
    {
      "epoch": 0.43833728192978677,
      "grad_norm": 0.12956377863883972,
      "learning_rate": 1.9301713332493386e-05,
      "loss": 0.0351,
      "step": 159
    },
    {
      "epoch": 0.44109412018091754,
      "grad_norm": 0.14530932903289795,
      "learning_rate": 1.929071886990095e-05,
      "loss": 0.0357,
      "step": 160
    },
    {
      "epoch": 0.44385095843204825,
      "grad_norm": 0.15227288007736206,
      "learning_rate": 1.9279641709881067e-05,
      "loss": 0.0363,
      "step": 161
    },
    {
      "epoch": 0.44660779668317896,
      "grad_norm": 0.165790393948555,
      "learning_rate": 1.926848195103242e-05,
      "loss": 0.0354,
      "step": 162
    },
    {
      "epoch": 0.4493646349343097,
      "grad_norm": 0.16768325865268707,
      "learning_rate": 1.9257239692688907e-05,
      "loss": 0.0335,
      "step": 163
    },
    {
      "epoch": 0.45212147318544044,
      "grad_norm": 0.147464320063591,
      "learning_rate": 1.9245915034918763e-05,
      "loss": 0.0339,
      "step": 164
    },
    {
      "epoch": 0.4548783114365712,
      "grad_norm": 0.1301703006029129,
      "learning_rate": 1.923450807852367e-05,
      "loss": 0.0337,
      "step": 165
    },
    {
      "epoch": 0.4576351496877019,
      "grad_norm": 0.1339288204908371,
      "learning_rate": 1.922301892503785e-05,
      "loss": 0.0353,
      "step": 166
    },
    {
      "epoch": 0.4603919879388326,
      "grad_norm": 0.16490085422992706,
      "learning_rate": 1.9211447676727174e-05,
      "loss": 0.0346,
      "step": 167
    },
    {
      "epoch": 0.4631488261899634,
      "grad_norm": 0.24428266286849976,
      "learning_rate": 1.9199794436588244e-05,
      "loss": 0.0355,
      "step": 168
    },
    {
      "epoch": 0.4659056644410941,
      "grad_norm": 0.40584465861320496,
      "learning_rate": 1.9188059308347475e-05,
      "loss": 0.0397,
      "step": 169
    },
    {
      "epoch": 0.46866250269222487,
      "grad_norm": 0.17336200177669525,
      "learning_rate": 1.9176242396460184e-05,
      "loss": 0.036,
      "step": 170
    },
    {
      "epoch": 0.4714193409433556,
      "grad_norm": 0.22676712274551392,
      "learning_rate": 1.916434380610963e-05,
      "loss": 0.0354,
      "step": 171
    },
    {
      "epoch": 0.47417617919448635,
      "grad_norm": 0.19564731419086456,
      "learning_rate": 1.9152363643206126e-05,
      "loss": 0.0341,
      "step": 172
    },
    {
      "epoch": 0.47693301744561706,
      "grad_norm": 0.16920173168182373,
      "learning_rate": 1.9140302014386044e-05,
      "loss": 0.036,
      "step": 173
    },
    {
      "epoch": 0.47968985569674777,
      "grad_norm": 0.1923132985830307,
      "learning_rate": 1.912815902701091e-05,
      "loss": 0.0334,
      "step": 174
    },
    {
      "epoch": 0.48244669394787854,
      "grad_norm": 0.18520843982696533,
      "learning_rate": 1.911593478916641e-05,
      "loss": 0.0343,
      "step": 175
    },
    {
      "epoch": 0.48520353219900925,
      "grad_norm": 0.2041483074426651,
      "learning_rate": 1.9103629409661468e-05,
      "loss": 0.0361,
      "step": 176
    },
    {
      "epoch": 0.48796037045014,
      "grad_norm": 0.17951691150665283,
      "learning_rate": 1.909124299802724e-05,
      "loss": 0.0346,
      "step": 177
    },
    {
      "epoch": 0.4907172087012707,
      "grad_norm": 0.17455315589904785,
      "learning_rate": 1.9078775664516157e-05,
      "loss": 0.0328,
      "step": 178
    },
    {
      "epoch": 0.4934740469524015,
      "grad_norm": 0.149070605635643,
      "learning_rate": 1.906622752010095e-05,
      "loss": 0.0353,
      "step": 179
    },
    {
      "epoch": 0.4962308852035322,
      "grad_norm": 0.13988004624843597,
      "learning_rate": 1.9053598676473656e-05,
      "loss": 0.0335,
      "step": 180
    },
    {
      "epoch": 0.4989877234546629,
      "grad_norm": 0.12514030933380127,
      "learning_rate": 1.904088924604461e-05,
      "loss": 0.0341,
      "step": 181
    },
    {
      "epoch": 0.5017445617057936,
      "grad_norm": 0.14698189496994019,
      "learning_rate": 1.9028099341941457e-05,
      "loss": 0.0347,
      "step": 182
    },
    {
      "epoch": 0.5045013999569244,
      "grad_norm": 0.1326190084218979,
      "learning_rate": 1.9015229078008163e-05,
      "loss": 0.0344,
      "step": 183
    },
    {
      "epoch": 0.5072582382080552,
      "grad_norm": 0.1399255394935608,
      "learning_rate": 1.900227856880396e-05,
      "loss": 0.0338,
      "step": 184
    },
    {
      "epoch": 0.5100150764591859,
      "grad_norm": 0.17403383553028107,
      "learning_rate": 1.898924792960237e-05,
      "loss": 0.034,
      "step": 185
    },
    {
      "epoch": 0.5127719147103166,
      "grad_norm": 0.14294308423995972,
      "learning_rate": 1.8976137276390145e-05,
      "loss": 0.0335,
      "step": 186
    },
    {
      "epoch": 0.5155287529614473,
      "grad_norm": 0.13353058695793152,
      "learning_rate": 1.8962946725866246e-05,
      "loss": 0.0325,
      "step": 187
    },
    {
      "epoch": 0.5182855912125781,
      "grad_norm": 0.1507033258676529,
      "learning_rate": 1.8949676395440818e-05,
      "loss": 0.0354,
      "step": 188
    },
    {
      "epoch": 0.5210424294637088,
      "grad_norm": 0.14498627185821533,
      "learning_rate": 1.8936326403234125e-05,
      "loss": 0.0327,
      "step": 189
    },
    {
      "epoch": 0.5237992677148395,
      "grad_norm": 0.15840047597885132,
      "learning_rate": 1.892289686807551e-05,
      "loss": 0.0332,
      "step": 190
    },
    {
      "epoch": 0.5265561059659702,
      "grad_norm": 0.14784608781337738,
      "learning_rate": 1.8909387909502335e-05,
      "loss": 0.0324,
      "step": 191
    },
    {
      "epoch": 0.5293129442171011,
      "grad_norm": 0.13484050333499908,
      "learning_rate": 1.8895799647758912e-05,
      "loss": 0.0336,
      "step": 192
    },
    {
      "epoch": 0.5320697824682318,
      "grad_norm": 0.17485234141349792,
      "learning_rate": 1.888213220379544e-05,
      "loss": 0.0335,
      "step": 193
    },
    {
      "epoch": 0.5348266207193625,
      "grad_norm": 0.14618845283985138,
      "learning_rate": 1.8868385699266928e-05,
      "loss": 0.0328,
      "step": 194
    },
    {
      "epoch": 0.5375834589704932,
      "grad_norm": 0.12785577774047852,
      "learning_rate": 1.8854560256532098e-05,
      "loss": 0.0335,
      "step": 195
    },
    {
      "epoch": 0.5403402972216239,
      "grad_norm": 0.16961853206157684,
      "learning_rate": 1.8840655998652324e-05,
      "loss": 0.0358,
      "step": 196
    },
    {
      "epoch": 0.5430971354727547,
      "grad_norm": 0.14937855303287506,
      "learning_rate": 1.8826673049390508e-05,
      "loss": 0.0336,
      "step": 197
    },
    {
      "epoch": 0.5458539737238854,
      "grad_norm": 0.14861802756786346,
      "learning_rate": 1.881261153320999e-05,
      "loss": 0.0346,
      "step": 198
    },
    {
      "epoch": 0.5486108119750162,
      "grad_norm": 0.13903887569904327,
      "learning_rate": 1.8798471575273445e-05,
      "loss": 0.0319,
      "step": 199
    },
    {
      "epoch": 0.5513676502261469,
      "grad_norm": 0.5592267513275146,
      "learning_rate": 1.8784253301441767e-05,
      "loss": 0.0367,
      "step": 200
    },
    {
      "epoch": 0.5541244884772776,
      "grad_norm": 0.12278741598129272,
      "learning_rate": 1.8769956838272937e-05,
      "loss": 0.0326,
      "step": 201
    },
    {
      "epoch": 0.5568813267284084,
      "grad_norm": 0.16035325825214386,
      "learning_rate": 1.8755582313020912e-05,
      "loss": 0.035,
      "step": 202
    },
    {
      "epoch": 0.5596381649795391,
      "grad_norm": 0.15898433327674866,
      "learning_rate": 1.8741129853634483e-05,
      "loss": 0.0336,
      "step": 203
    },
    {
      "epoch": 0.5623950032306698,
      "grad_norm": 0.14194312691688538,
      "learning_rate": 1.8726599588756144e-05,
      "loss": 0.0327,
      "step": 204
    },
    {
      "epoch": 0.5651518414818005,
      "grad_norm": 0.14086908102035522,
      "learning_rate": 1.8711991647720936e-05,
      "loss": 0.034,
      "step": 205
    },
    {
      "epoch": 0.5679086797329312,
      "grad_norm": 0.13377539813518524,
      "learning_rate": 1.8697306160555303e-05,
      "loss": 0.0337,
      "step": 206
    },
    {
      "epoch": 0.5706655179840621,
      "grad_norm": 0.13268251717090607,
      "learning_rate": 1.868254325797594e-05,
      "loss": 0.0329,
      "step": 207
    },
    {
      "epoch": 0.5734223562351928,
      "grad_norm": 0.13881011307239532,
      "learning_rate": 1.8667703071388607e-05,
      "loss": 0.0339,
      "step": 208
    },
    {
      "epoch": 0.5761791944863235,
      "grad_norm": 0.15615610778331757,
      "learning_rate": 1.8652785732886988e-05,
      "loss": 0.0345,
      "step": 209
    },
    {
      "epoch": 0.5789360327374542,
      "grad_norm": 0.3299935460090637,
      "learning_rate": 1.8637791375251505e-05,
      "loss": 0.0353,
      "step": 210
    },
    {
      "epoch": 0.581692870988585,
      "grad_norm": 0.13316752016544342,
      "learning_rate": 1.862272013194812e-05,
      "loss": 0.0325,
      "step": 211
    },
    {
      "epoch": 0.5844497092397157,
      "grad_norm": 0.13292838633060455,
      "learning_rate": 1.8607572137127175e-05,
      "loss": 0.0342,
      "step": 212
    },
    {
      "epoch": 0.5872065474908464,
      "grad_norm": 0.1690252423286438,
      "learning_rate": 1.859234752562217e-05,
      "loss": 0.0333,
      "step": 213
    },
    {
      "epoch": 0.5899633857419772,
      "grad_norm": 0.13950853049755096,
      "learning_rate": 1.8577046432948586e-05,
      "loss": 0.032,
      "step": 214
    },
    {
      "epoch": 0.5927202239931079,
      "grad_norm": 0.12849925458431244,
      "learning_rate": 1.8561668995302668e-05,
      "loss": 0.0323,
      "step": 215
    },
    {
      "epoch": 0.5954770622442387,
      "grad_norm": 0.22346273064613342,
      "learning_rate": 1.8546215349560204e-05,
      "loss": 0.0346,
      "step": 216
    },
    {
      "epoch": 0.5982339004953694,
      "grad_norm": 0.12624599039554596,
      "learning_rate": 1.853068563327533e-05,
      "loss": 0.0328,
      "step": 217
    },
    {
      "epoch": 0.6009907387465001,
      "grad_norm": 0.14228811860084534,
      "learning_rate": 1.851507998467929e-05,
      "loss": 0.0334,
      "step": 218
    },
    {
      "epoch": 0.6037475769976308,
      "grad_norm": 0.1500948965549469,
      "learning_rate": 1.849939854267919e-05,
      "loss": 0.0346,
      "step": 219
    },
    {
      "epoch": 0.6065044152487615,
      "grad_norm": 0.17958402633666992,
      "learning_rate": 1.8483641446856798e-05,
      "loss": 0.0341,
      "step": 220
    },
    {
      "epoch": 0.6092612534998924,
      "grad_norm": 0.15875360369682312,
      "learning_rate": 1.8467808837467277e-05,
      "loss": 0.0354,
      "step": 221
    },
    {
      "epoch": 0.6120180917510231,
      "grad_norm": 0.13992802798748016,
      "learning_rate": 1.845190085543795e-05,
      "loss": 0.0346,
      "step": 222
    },
    {
      "epoch": 0.6147749300021538,
      "grad_norm": 0.14051182568073273,
      "learning_rate": 1.843591764236702e-05,
      "loss": 0.0325,
      "step": 223
    },
    {
      "epoch": 0.6175317682532845,
      "grad_norm": 0.13775014877319336,
      "learning_rate": 1.841985934052234e-05,
      "loss": 0.0323,
      "step": 224
    },
    {
      "epoch": 0.6202886065044152,
      "grad_norm": 0.1298038810491562,
      "learning_rate": 1.840372609284013e-05,
      "loss": 0.0322,
      "step": 225
    },
    {
      "epoch": 0.623045444755546,
      "grad_norm": 0.14838284254074097,
      "learning_rate": 1.8387518042923715e-05,
      "loss": 0.0334,
      "step": 226
    },
    {
      "epoch": 0.6258022830066767,
      "grad_norm": 0.1384505033493042,
      "learning_rate": 1.8371235335042236e-05,
      "loss": 0.0332,
      "step": 227
    },
    {
      "epoch": 0.6285591212578074,
      "grad_norm": 0.1435280740261078,
      "learning_rate": 1.8354878114129368e-05,
      "loss": 0.0323,
      "step": 228
    },
    {
      "epoch": 0.6313159595089382,
      "grad_norm": 0.19593602418899536,
      "learning_rate": 1.833844652578203e-05,
      "loss": 0.033,
      "step": 229
    },
    {
      "epoch": 0.634072797760069,
      "grad_norm": 0.11905160546302795,
      "learning_rate": 1.832194071625911e-05,
      "loss": 0.0329,
      "step": 230
    },
    {
      "epoch": 0.6368296360111997,
      "grad_norm": 0.13935434818267822,
      "learning_rate": 1.8305360832480118e-05,
      "loss": 0.0326,
      "step": 231
    },
    {
      "epoch": 0.6395864742623304,
      "grad_norm": 0.13467997312545776,
      "learning_rate": 1.8288707022023926e-05,
      "loss": 0.0323,
      "step": 232
    },
    {
      "epoch": 0.6423433125134611,
      "grad_norm": 0.13442496955394745,
      "learning_rate": 1.827197943312742e-05,
      "loss": 0.0326,
      "step": 233
    },
    {
      "epoch": 0.6451001507645918,
      "grad_norm": 0.13342751562595367,
      "learning_rate": 1.82551782146842e-05,
      "loss": 0.0326,
      "step": 234
    },
    {
      "epoch": 0.6478569890157226,
      "grad_norm": 0.11170811951160431,
      "learning_rate": 1.8238303516243253e-05,
      "loss": 0.0329,
      "step": 235
    },
    {
      "epoch": 0.6506138272668534,
      "grad_norm": 0.109471395611763,
      "learning_rate": 1.8221355488007606e-05,
      "loss": 0.0312,
      "step": 236
    },
    {
      "epoch": 0.6533706655179841,
      "grad_norm": 0.10828150063753128,
      "learning_rate": 1.8204334280833005e-05,
      "loss": 0.0317,
      "step": 237
    },
    {
      "epoch": 0.6561275037691148,
      "grad_norm": 0.1234995648264885,
      "learning_rate": 1.8187240046226576e-05,
      "loss": 0.0316,
      "step": 238
    },
    {
      "epoch": 0.6588843420202455,
      "grad_norm": 0.12001900374889374,
      "learning_rate": 1.817007293634545e-05,
      "loss": 0.0318,
      "step": 239
    },
    {
      "epoch": 0.6616411802713763,
      "grad_norm": 0.12810060381889343,
      "learning_rate": 1.8152833103995443e-05,
      "loss": 0.0315,
      "step": 240
    },
    {
      "epoch": 0.664398018522507,
      "grad_norm": 0.12769843637943268,
      "learning_rate": 1.8135520702629677e-05,
      "loss": 0.0324,
      "step": 241
    },
    {
      "epoch": 0.6671548567736377,
      "grad_norm": 0.12816517055034637,
      "learning_rate": 1.8118135886347207e-05,
      "loss": 0.0311,
      "step": 242
    },
    {
      "epoch": 0.6699116950247684,
      "grad_norm": 0.1479124128818512,
      "learning_rate": 1.8100678809891668e-05,
      "loss": 0.0322,
      "step": 243
    },
    {
      "epoch": 0.6726685332758991,
      "grad_norm": 0.1499953418970108,
      "learning_rate": 1.8083149628649887e-05,
      "loss": 0.0322,
      "step": 244
    },
    {
      "epoch": 0.67542537152703,
      "grad_norm": 0.16143152117729187,
      "learning_rate": 1.8065548498650495e-05,
      "loss": 0.0327,
      "step": 245
    },
    {
      "epoch": 0.6781822097781607,
      "grad_norm": 0.11848675459623337,
      "learning_rate": 1.8047875576562556e-05,
      "loss": 0.0323,
      "step": 246
    },
    {
      "epoch": 0.6809390480292914,
      "grad_norm": 0.14054562151432037,
      "learning_rate": 1.803013101969415e-05,
      "loss": 0.0311,
      "step": 247
    },
    {
      "epoch": 0.6836958862804221,
      "grad_norm": 0.14287616312503815,
      "learning_rate": 1.801231498599099e-05,
      "loss": 0.0321,
      "step": 248
    },
    {
      "epoch": 0.6864527245315529,
      "grad_norm": 0.1472490429878235,
      "learning_rate": 1.7994427634035016e-05,
      "loss": 0.0324,
      "step": 249
    },
    {
      "epoch": 0.6892095627826836,
      "grad_norm": 0.2156575620174408,
      "learning_rate": 1.7976469123042955e-05,
      "loss": 0.0331,
      "step": 250
    },
    {
      "epoch": 0.6919664010338143,
      "grad_norm": 0.15800388157367706,
      "learning_rate": 1.7958439612864954e-05,
      "loss": 0.0319,
      "step": 251
    },
    {
      "epoch": 0.6947232392849451,
      "grad_norm": 0.13451936841011047,
      "learning_rate": 1.7940339263983112e-05,
      "loss": 0.032,
      "step": 252
    },
    {
      "epoch": 0.6974800775360758,
      "grad_norm": 0.12370197474956512,
      "learning_rate": 1.7922168237510076e-05,
      "loss": 0.0328,
      "step": 253
    },
    {
      "epoch": 0.7002369157872066,
      "grad_norm": 0.1429441124200821,
      "learning_rate": 1.7903926695187595e-05,
      "loss": 0.0318,
      "step": 254
    },
    {
      "epoch": 0.7029937540383373,
      "grad_norm": 0.13566642999649048,
      "learning_rate": 1.7885614799385086e-05,
      "loss": 0.0322,
      "step": 255
    },
    {
      "epoch": 0.705750592289468,
      "grad_norm": 0.11789460480213165,
      "learning_rate": 1.78672327130982e-05,
      "loss": 0.0316,
      "step": 256
    },
    {
      "epoch": 0.7085074305405987,
      "grad_norm": 0.14814262092113495,
      "learning_rate": 1.7848780599947334e-05,
      "loss": 0.033,
      "step": 257
    },
    {
      "epoch": 0.7112642687917294,
      "grad_norm": 0.12498864531517029,
      "learning_rate": 1.7830258624176224e-05,
      "loss": 0.0321,
      "step": 258
    },
    {
      "epoch": 0.7140211070428603,
      "grad_norm": 0.12611018121242523,
      "learning_rate": 1.7811666950650445e-05,
      "loss": 0.0318,
      "step": 259
    },
    {
      "epoch": 0.716777945293991,
      "grad_norm": 0.1379752904176712,
      "learning_rate": 1.7793005744855967e-05,
      "loss": 0.0343,
      "step": 260
    },
    {
      "epoch": 0.7195347835451217,
      "grad_norm": 0.12495708465576172,
      "learning_rate": 1.777427517289766e-05,
      "loss": 0.0337,
      "step": 261
    },
    {
      "epoch": 0.7222916217962524,
      "grad_norm": 0.11509250104427338,
      "learning_rate": 1.775547540149784e-05,
      "loss": 0.0303,
      "step": 262
    },
    {
      "epoch": 0.7250484600473831,
      "grad_norm": 0.14822109043598175,
      "learning_rate": 1.7736606597994763e-05,
      "loss": 0.0329,
      "step": 263
    },
    {
      "epoch": 0.7278052982985139,
      "grad_norm": 0.1274510771036148,
      "learning_rate": 1.7717668930341152e-05,
      "loss": 0.0321,
      "step": 264
    },
    {
      "epoch": 0.7305621365496446,
      "grad_norm": 0.13247767090797424,
      "learning_rate": 1.769866256710269e-05,
      "loss": 0.0327,
      "step": 265
    },
    {
      "epoch": 0.7333189748007753,
      "grad_norm": 0.1502748280763626,
      "learning_rate": 1.767958767745653e-05,
      "loss": 0.0336,
      "step": 266
    },
    {
      "epoch": 0.7360758130519061,
      "grad_norm": 0.11968521028757095,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.032,
      "step": 267
    },
    {
      "epoch": 0.7388326513030369,
      "grad_norm": 0.12810976803302765,
      "learning_rate": 1.7641232998698e-05,
      "loss": 0.0317,
      "step": 268
    },
    {
      "epoch": 0.7415894895541676,
      "grad_norm": 0.13857212662696838,
      "learning_rate": 1.7621953550983677e-05,
      "loss": 0.03,
      "step": 269
    },
    {
      "epoch": 0.7443463278052983,
      "grad_norm": 0.12155446410179138,
      "learning_rate": 1.7602606259654704e-05,
      "loss": 0.0317,
      "step": 270
    },
    {
      "epoch": 0.747103166056429,
      "grad_norm": 0.13759441673755646,
      "learning_rate": 1.7583191296922866e-05,
      "loss": 0.0312,
      "step": 271
    },
    {
      "epoch": 0.7498600043075597,
      "grad_norm": 0.18140603601932526,
      "learning_rate": 1.7563708835602286e-05,
      "loss": 0.0316,
      "step": 272
    },
    {
      "epoch": 0.7526168425586905,
      "grad_norm": 0.16475705802440643,
      "learning_rate": 1.7544159049107902e-05,
      "loss": 0.0306,
      "step": 273
    },
    {
      "epoch": 0.7553736808098213,
      "grad_norm": 0.13975286483764648,
      "learning_rate": 1.7524542111453923e-05,
      "loss": 0.0335,
      "step": 274
    },
    {
      "epoch": 0.758130519060952,
      "grad_norm": 0.14528636634349823,
      "learning_rate": 1.7504858197252263e-05,
      "loss": 0.032,
      "step": 275
    },
    {
      "epoch": 0.7608873573120827,
      "grad_norm": 0.15437592566013336,
      "learning_rate": 1.7485107481711014e-05,
      "loss": 0.033,
      "step": 276
    },
    {
      "epoch": 0.7636441955632134,
      "grad_norm": 0.11566643416881561,
      "learning_rate": 1.746529014063286e-05,
      "loss": 0.0305,
      "step": 277
    },
    {
      "epoch": 0.7664010338143442,
      "grad_norm": 0.1528697907924652,
      "learning_rate": 1.7445406350413533e-05,
      "loss": 0.0301,
      "step": 278
    },
    {
      "epoch": 0.7691578720654749,
      "grad_norm": 0.13452287018299103,
      "learning_rate": 1.7425456288040236e-05,
      "loss": 0.0312,
      "step": 279
    },
    {
      "epoch": 0.7719147103166056,
      "grad_norm": 0.13913044333457947,
      "learning_rate": 1.740544013109005e-05,
      "loss": 0.0328,
      "step": 280
    },
    {
      "epoch": 0.7746715485677363,
      "grad_norm": 0.11884239315986633,
      "learning_rate": 1.738535805772838e-05,
      "loss": 0.0299,
      "step": 281
    },
    {
      "epoch": 0.7774283868188671,
      "grad_norm": 0.13907060027122498,
      "learning_rate": 1.736521024670737e-05,
      "loss": 0.0326,
      "step": 282
    },
    {
      "epoch": 0.7801852250699979,
      "grad_norm": 0.13899970054626465,
      "learning_rate": 1.7344996877364282e-05,
      "loss": 0.0329,
      "step": 283
    },
    {
      "epoch": 0.7829420633211286,
      "grad_norm": 0.13733629882335663,
      "learning_rate": 1.732471812961992e-05,
      "loss": 0.0317,
      "step": 284
    },
    {
      "epoch": 0.7856989015722593,
      "grad_norm": 0.12010809779167175,
      "learning_rate": 1.7304374183977032e-05,
      "loss": 0.0315,
      "step": 285
    },
    {
      "epoch": 0.78845573982339,
      "grad_norm": 0.1484297811985016,
      "learning_rate": 1.72839652215187e-05,
      "loss": 0.0321,
      "step": 286
    },
    {
      "epoch": 0.7912125780745208,
      "grad_norm": 0.12623994052410126,
      "learning_rate": 1.7263491423906716e-05,
      "loss": 0.0319,
      "step": 287
    },
    {
      "epoch": 0.7939694163256515,
      "grad_norm": 0.11906126886606216,
      "learning_rate": 1.7242952973379983e-05,
      "loss": 0.0305,
      "step": 288
    },
    {
      "epoch": 0.7967262545767823,
      "grad_norm": 0.1313122808933258,
      "learning_rate": 1.7222350052752883e-05,
      "loss": 0.0319,
      "step": 289
    },
    {
      "epoch": 0.799483092827913,
      "grad_norm": 0.1219288781285286,
      "learning_rate": 1.720168284541365e-05,
      "loss": 0.0297,
      "step": 290
    },
    {
      "epoch": 0.8022399310790437,
      "grad_norm": 0.16674523055553436,
      "learning_rate": 1.7180951535322742e-05,
      "loss": 0.0322,
      "step": 291
    },
    {
      "epoch": 0.8049967693301745,
      "grad_norm": 0.21712636947631836,
      "learning_rate": 1.7160156307011197e-05,
      "loss": 0.0318,
      "step": 292
    },
    {
      "epoch": 0.8077536075813052,
      "grad_norm": 0.1368701457977295,
      "learning_rate": 1.7139297345578992e-05,
      "loss": 0.0302,
      "step": 293
    },
    {
      "epoch": 0.8105104458324359,
      "grad_norm": 0.1643868386745453,
      "learning_rate": 1.7118374836693407e-05,
      "loss": 0.0328,
      "step": 294
    },
    {
      "epoch": 0.8132672840835666,
      "grad_norm": 0.11599249392747879,
      "learning_rate": 1.7097388966587355e-05,
      "loss": 0.0303,
      "step": 295
    },
    {
      "epoch": 0.8160241223346973,
      "grad_norm": 0.29641032218933105,
      "learning_rate": 1.7076339922057736e-05,
      "loss": 0.034,
      "step": 296
    },
    {
      "epoch": 0.8187809605858282,
      "grad_norm": 0.1458274871110916,
      "learning_rate": 1.705522789046377e-05,
      "loss": 0.0318,
      "step": 297
    },
    {
      "epoch": 0.8215377988369589,
      "grad_norm": 0.12504877150058746,
      "learning_rate": 1.7034053059725325e-05,
      "loss": 0.0321,
      "step": 298
    },
    {
      "epoch": 0.8242946370880896,
      "grad_norm": 0.16158223152160645,
      "learning_rate": 1.7012815618321256e-05,
      "loss": 0.0326,
      "step": 299
    },
    {
      "epoch": 0.8270514753392203,
      "grad_norm": 0.12104715406894684,
      "learning_rate": 1.6991515755287715e-05,
      "loss": 0.031,
      "step": 300
    },
    {
      "epoch": 0.8298083135903511,
      "grad_norm": 0.1327955275774002,
      "learning_rate": 1.697015366021648e-05,
      "loss": 0.033,
      "step": 301
    },
    {
      "epoch": 0.8325651518414818,
      "grad_norm": 0.12627461552619934,
      "learning_rate": 1.694872952325326e-05,
      "loss": 0.0305,
      "step": 302
    },
    {
      "epoch": 0.8353219900926125,
      "grad_norm": 0.13003765046596527,
      "learning_rate": 1.6927243535095995e-05,
      "loss": 0.0313,
      "step": 303
    },
    {
      "epoch": 0.8380788283437433,
      "grad_norm": 0.15389683842658997,
      "learning_rate": 1.690569588699318e-05,
      "loss": 0.0314,
      "step": 304
    },
    {
      "epoch": 0.840835666594874,
      "grad_norm": 0.1295783370733261,
      "learning_rate": 1.6884086770742138e-05,
      "loss": 0.0311,
      "step": 305
    },
    {
      "epoch": 0.8435925048460048,
      "grad_norm": 0.14465399086475372,
      "learning_rate": 1.686241637868734e-05,
      "loss": 0.0305,
      "step": 306
    },
    {
      "epoch": 0.8463493430971355,
      "grad_norm": 0.12934014201164246,
      "learning_rate": 1.6840684903718658e-05,
      "loss": 0.0315,
      "step": 307
    },
    {
      "epoch": 0.8491061813482662,
      "grad_norm": 0.13668671250343323,
      "learning_rate": 1.681889253926969e-05,
      "loss": 0.0308,
      "step": 308
    },
    {
      "epoch": 0.8518630195993969,
      "grad_norm": 0.1645808219909668,
      "learning_rate": 1.6797039479315994e-05,
      "loss": 0.0328,
      "step": 309
    },
    {
      "epoch": 0.8546198578505276,
      "grad_norm": 0.1372358649969101,
      "learning_rate": 1.67751259183734e-05,
      "loss": 0.0309,
      "step": 310
    },
    {
      "epoch": 0.8573766961016585,
      "grad_norm": 0.151007741689682,
      "learning_rate": 1.675315205149626e-05,
      "loss": 0.033,
      "step": 311
    },
    {
      "epoch": 0.8601335343527892,
      "grad_norm": 0.18326100707054138,
      "learning_rate": 1.67311180742757e-05,
      "loss": 0.0314,
      "step": 312
    },
    {
      "epoch": 0.8628903726039199,
      "grad_norm": 0.12580075860023499,
      "learning_rate": 1.6709024182837917e-05,
      "loss": 0.0314,
      "step": 313
    },
    {
      "epoch": 0.8656472108550506,
      "grad_norm": 0.14372077584266663,
      "learning_rate": 1.6686870573842388e-05,
      "loss": 0.0298,
      "step": 314
    },
    {
      "epoch": 0.8684040491061813,
      "grad_norm": 0.1365354061126709,
      "learning_rate": 1.6664657444480145e-05,
      "loss": 0.0304,
      "step": 315
    },
    {
      "epoch": 0.8711608873573121,
      "grad_norm": 0.17227758467197418,
      "learning_rate": 1.6642384992472026e-05,
      "loss": 0.0311,
      "step": 316
    },
    {
      "epoch": 0.8739177256084428,
      "grad_norm": 0.1729755699634552,
      "learning_rate": 1.6620053416066892e-05,
      "loss": 0.0315,
      "step": 317
    },
    {
      "epoch": 0.8766745638595735,
      "grad_norm": 0.1310206949710846,
      "learning_rate": 1.6597662914039885e-05,
      "loss": 0.0313,
      "step": 318
    },
    {
      "epoch": 0.8794314021107043,
      "grad_norm": 0.15664447844028473,
      "learning_rate": 1.657521368569064e-05,
      "loss": 0.0304,
      "step": 319
    },
    {
      "epoch": 0.8821882403618351,
      "grad_norm": 0.16371680796146393,
      "learning_rate": 1.6552705930841523e-05,
      "loss": 0.0306,
      "step": 320
    },
    {
      "epoch": 0.8849450786129658,
      "grad_norm": 0.132271409034729,
      "learning_rate": 1.653013984983585e-05,
      "loss": 0.0315,
      "step": 321
    },
    {
      "epoch": 0.8877019168640965,
      "grad_norm": 0.1789659708738327,
      "learning_rate": 1.6507515643536113e-05,
      "loss": 0.0304,
      "step": 322
    },
    {
      "epoch": 0.8904587551152272,
      "grad_norm": 0.12446186691522598,
      "learning_rate": 1.6484833513322155e-05,
      "loss": 0.0297,
      "step": 323
    },
    {
      "epoch": 0.8932155933663579,
      "grad_norm": 0.13840149343013763,
      "learning_rate": 1.6462093661089432e-05,
      "loss": 0.0302,
      "step": 324
    },
    {
      "epoch": 0.8959724316174887,
      "grad_norm": 0.12458626925945282,
      "learning_rate": 1.643929628924717e-05,
      "loss": 0.0321,
      "step": 325
    },
    {
      "epoch": 0.8987292698686195,
      "grad_norm": 0.13855692744255066,
      "learning_rate": 1.6416441600716593e-05,
      "loss": 0.0305,
      "step": 326
    },
    {
      "epoch": 0.9014861081197502,
      "grad_norm": 0.1461232304573059,
      "learning_rate": 1.6393529798929103e-05,
      "loss": 0.03,
      "step": 327
    },
    {
      "epoch": 0.9042429463708809,
      "grad_norm": 0.11444980651140213,
      "learning_rate": 1.637056108782446e-05,
      "loss": 0.0292,
      "step": 328
    },
    {
      "epoch": 0.9069997846220116,
      "grad_norm": 0.1289425641298294,
      "learning_rate": 1.6347535671848998e-05,
      "loss": 0.0305,
      "step": 329
    },
    {
      "epoch": 0.9097566228731424,
      "grad_norm": 0.12594164907932281,
      "learning_rate": 1.6324453755953772e-05,
      "loss": 0.0288,
      "step": 330
    },
    {
      "epoch": 0.9125134611242731,
      "grad_norm": 0.41084402799606323,
      "learning_rate": 1.6301315545592753e-05,
      "loss": 0.0299,
      "step": 331
    },
    {
      "epoch": 0.9152702993754038,
      "grad_norm": 0.13620588183403015,
      "learning_rate": 1.627812124672099e-05,
      "loss": 0.0302,
      "step": 332
    },
    {
      "epoch": 0.9180271376265345,
      "grad_norm": 0.10916028916835785,
      "learning_rate": 1.6254871065792776e-05,
      "loss": 0.031,
      "step": 333
    },
    {
      "epoch": 0.9207839758776653,
      "grad_norm": 0.1257525086402893,
      "learning_rate": 1.623156520975983e-05,
      "loss": 0.0313,
      "step": 334
    },
    {
      "epoch": 0.9235408141287961,
      "grad_norm": 0.10521648079156876,
      "learning_rate": 1.620820388606942e-05,
      "loss": 0.0296,
      "step": 335
    },
    {
      "epoch": 0.9262976523799268,
      "grad_norm": 0.11178401112556458,
      "learning_rate": 1.618478730266255e-05,
      "loss": 0.0307,
      "step": 336
    },
    {
      "epoch": 0.9290544906310575,
      "grad_norm": 0.11989890038967133,
      "learning_rate": 1.6161315667972095e-05,
      "loss": 0.0298,
      "step": 337
    },
    {
      "epoch": 0.9318113288821882,
      "grad_norm": 0.12103229016065598,
      "learning_rate": 1.6137789190920938e-05,
      "loss": 0.0302,
      "step": 338
    },
    {
      "epoch": 0.934568167133319,
      "grad_norm": 0.10210815072059631,
      "learning_rate": 1.6114208080920125e-05,
      "loss": 0.029,
      "step": 339
    },
    {
      "epoch": 0.9373250053844497,
      "grad_norm": 0.12528973817825317,
      "learning_rate": 1.6090572547866983e-05,
      "loss": 0.0303,
      "step": 340
    },
    {
      "epoch": 0.9400818436355804,
      "grad_norm": 0.12944018840789795,
      "learning_rate": 1.606688280214328e-05,
      "loss": 0.031,
      "step": 341
    },
    {
      "epoch": 0.9428386818867112,
      "grad_norm": 0.1170840710401535,
      "learning_rate": 1.6043139054613326e-05,
      "loss": 0.0304,
      "step": 342
    },
    {
      "epoch": 0.9455955201378419,
      "grad_norm": 0.1317935287952423,
      "learning_rate": 1.60193415166221e-05,
      "loss": 0.0309,
      "step": 343
    },
    {
      "epoch": 0.9483523583889727,
      "grad_norm": 0.10986040532588959,
      "learning_rate": 1.599549039999338e-05,
      "loss": 0.0306,
      "step": 344
    },
    {
      "epoch": 0.9511091966401034,
      "grad_norm": 0.11340153217315674,
      "learning_rate": 1.5971585917027864e-05,
      "loss": 0.0293,
      "step": 345
    },
    {
      "epoch": 0.9538660348912341,
      "grad_norm": 0.1250225007534027,
      "learning_rate": 1.594762828050124e-05,
      "loss": 0.029,
      "step": 346
    },
    {
      "epoch": 0.9566228731423648,
      "grad_norm": 0.1400870531797409,
      "learning_rate": 1.5923617703662346e-05,
      "loss": 0.0313,
      "step": 347
    },
    {
      "epoch": 0.9593797113934955,
      "grad_norm": 0.10701585561037064,
      "learning_rate": 1.5899554400231233e-05,
      "loss": 0.0316,
      "step": 348
    },
    {
      "epoch": 0.9621365496446264,
      "grad_norm": 0.1186312660574913,
      "learning_rate": 1.587543858439727e-05,
      "loss": 0.0307,
      "step": 349
    },
    {
      "epoch": 0.9648933878957571,
      "grad_norm": 0.10630480200052261,
      "learning_rate": 1.585127047081727e-05,
      "loss": 0.0297,
      "step": 350
    },
    {
      "epoch": 0.9676502261468878,
      "grad_norm": 0.11205119639635086,
      "learning_rate": 1.5827050274613512e-05,
      "loss": 0.0297,
      "step": 351
    },
    {
      "epoch": 0.9704070643980185,
      "grad_norm": 0.11059113591909409,
      "learning_rate": 1.580277821137191e-05,
      "loss": 0.0296,
      "step": 352
    },
    {
      "epoch": 0.9731639026491492,
      "grad_norm": 0.13033725321292877,
      "learning_rate": 1.577845449714001e-05,
      "loss": 0.0303,
      "step": 353
    },
    {
      "epoch": 0.97592074090028,
      "grad_norm": 0.10061930865049362,
      "learning_rate": 1.5754079348425137e-05,
      "loss": 0.0296,
      "step": 354
    },
    {
      "epoch": 0.9786775791514107,
      "grad_norm": 0.11028820276260376,
      "learning_rate": 1.5729652982192428e-05,
      "loss": 0.0284,
      "step": 355
    },
    {
      "epoch": 0.9814344174025414,
      "grad_norm": 0.11424390226602554,
      "learning_rate": 1.5705175615862906e-05,
      "loss": 0.0293,
      "step": 356
    },
    {
      "epoch": 0.9841912556536722,
      "grad_norm": 0.14579032361507416,
      "learning_rate": 1.568064746731156e-05,
      "loss": 0.0299,
      "step": 357
    },
    {
      "epoch": 0.986948093904803,
      "grad_norm": 0.11156564950942993,
      "learning_rate": 1.5656068754865388e-05,
      "loss": 0.0298,
      "step": 358
    },
    {
      "epoch": 0.9897049321559337,
      "grad_norm": 0.10791637748479843,
      "learning_rate": 1.5631439697301464e-05,
      "loss": 0.03,
      "step": 359
    },
    {
      "epoch": 0.9924617704070644,
      "grad_norm": 0.11196880787611008,
      "learning_rate": 1.560676051384499e-05,
      "loss": 0.0289,
      "step": 360
    },
    {
      "epoch": 0.9952186086581951,
      "grad_norm": 0.11056679487228394,
      "learning_rate": 1.558203142416734e-05,
      "loss": 0.0285,
      "step": 361
    },
    {
      "epoch": 0.9979754469093258,
      "grad_norm": 0.1029677763581276,
      "learning_rate": 1.5557252648384103e-05,
      "loss": 0.0303,
      "step": 362
    },
    {
      "epoch": 1.0007322851604565,
      "grad_norm": 0.12087027728557587,
      "learning_rate": 1.553242440705314e-05,
      "loss": 0.0297,
      "step": 363
    },
    {
      "epoch": 1.0034891234115872,
      "grad_norm": 0.11372894048690796,
      "learning_rate": 1.5507546921172595e-05,
      "loss": 0.0251,
      "step": 364
    },
    {
      "epoch": 1.0062459616627182,
      "grad_norm": 0.11590228229761124,
      "learning_rate": 1.548262041217895e-05,
      "loss": 0.0244,
      "step": 365
    },
    {
      "epoch": 1.009002799913849,
      "grad_norm": 0.129511296749115,
      "learning_rate": 1.5457645101945046e-05,
      "loss": 0.0247,
      "step": 366
    },
    {
      "epoch": 1.0117596381649796,
      "grad_norm": 0.12303431332111359,
      "learning_rate": 1.5432621212778105e-05,
      "loss": 0.0247,
      "step": 367
    },
    {
      "epoch": 1.0145164764161103,
      "grad_norm": 0.1147053986787796,
      "learning_rate": 1.5407548967417755e-05,
      "loss": 0.0241,
      "step": 368
    },
    {
      "epoch": 1.017273314667241,
      "grad_norm": 0.117189422249794,
      "learning_rate": 1.538242858903404e-05,
      "loss": 0.0244,
      "step": 369
    },
    {
      "epoch": 1.0200301529183717,
      "grad_norm": 0.1118486151099205,
      "learning_rate": 1.5357260301225448e-05,
      "loss": 0.0249,
      "step": 370
    },
    {
      "epoch": 1.0227869911695024,
      "grad_norm": 0.4378014802932739,
      "learning_rate": 1.5332044328016916e-05,
      "loss": 0.0262,
      "step": 371
    },
    {
      "epoch": 1.0255438294206332,
      "grad_norm": 0.12007389217615128,
      "learning_rate": 1.530678089385782e-05,
      "loss": 0.0256,
      "step": 372
    },
    {
      "epoch": 1.0283006676717639,
      "grad_norm": 0.13402129709720612,
      "learning_rate": 1.5281470223619995e-05,
      "loss": 0.0254,
      "step": 373
    },
    {
      "epoch": 1.0310575059228946,
      "grad_norm": 0.1433870643377304,
      "learning_rate": 1.525611254259574e-05,
      "loss": 0.0251,
      "step": 374
    },
    {
      "epoch": 1.0338143441740255,
      "grad_norm": 0.116049624979496,
      "learning_rate": 1.5230708076495777e-05,
      "loss": 0.0246,
      "step": 375
    },
    {
      "epoch": 1.0365711824251562,
      "grad_norm": 0.10888892412185669,
      "learning_rate": 1.5205257051447291e-05,
      "loss": 0.0248,
      "step": 376
    },
    {
      "epoch": 1.039328020676287,
      "grad_norm": 0.1329452395439148,
      "learning_rate": 1.5179759693991869e-05,
      "loss": 0.0248,
      "step": 377
    },
    {
      "epoch": 1.0420848589274176,
      "grad_norm": 0.12392039597034454,
      "learning_rate": 1.5154216231083522e-05,
      "loss": 0.0255,
      "step": 378
    },
    {
      "epoch": 1.0448416971785484,
      "grad_norm": 0.11657357215881348,
      "learning_rate": 1.5128626890086647e-05,
      "loss": 0.0247,
      "step": 379
    },
    {
      "epoch": 1.047598535429679,
      "grad_norm": 0.11587107181549072,
      "learning_rate": 1.5102991898774e-05,
      "loss": 0.0256,
      "step": 380
    },
    {
      "epoch": 1.0503553736808098,
      "grad_norm": 0.5603076219558716,
      "learning_rate": 1.507731148532468e-05,
      "loss": 0.0252,
      "step": 381
    },
    {
      "epoch": 1.0531122119319405,
      "grad_norm": 0.11775552481412888,
      "learning_rate": 1.505158587832208e-05,
      "loss": 0.0246,
      "step": 382
    },
    {
      "epoch": 1.0558690501830712,
      "grad_norm": 0.10402605682611465,
      "learning_rate": 1.5025815306751888e-05,
      "loss": 0.0238,
      "step": 383
    },
    {
      "epoch": 1.058625888434202,
      "grad_norm": 0.1355026811361313,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.0262,
      "step": 384
    },
    {
      "epoch": 1.0613827266853328,
      "grad_norm": 0.11940840631723404,
      "learning_rate": 1.497414018785052e-05,
      "loss": 0.0246,
      "step": 385
    },
    {
      "epoch": 1.0641395649364636,
      "grad_norm": 0.12173695117235184,
      "learning_rate": 1.4948236100483688e-05,
      "loss": 0.025,
      "step": 386
    },
    {
      "epoch": 1.0668964031875943,
      "grad_norm": 0.11082287132740021,
      "learning_rate": 1.492228796847385e-05,
      "loss": 0.0258,
      "step": 387
    },
    {
      "epoch": 1.069653241438725,
      "grad_norm": 0.10860753059387207,
      "learning_rate": 1.4896296022787386e-05,
      "loss": 0.0244,
      "step": 388
    },
    {
      "epoch": 1.0724100796898557,
      "grad_norm": 0.10953570157289505,
      "learning_rate": 1.4870260494780679e-05,
      "loss": 0.0246,
      "step": 389
    },
    {
      "epoch": 1.0751669179409864,
      "grad_norm": 0.10230018198490143,
      "learning_rate": 1.4844181616198028e-05,
      "loss": 0.0249,
      "step": 390
    },
    {
      "epoch": 1.077923756192117,
      "grad_norm": 0.10772014409303665,
      "learning_rate": 1.4818059619169606e-05,
      "loss": 0.0247,
      "step": 391
    },
    {
      "epoch": 1.0806805944432478,
      "grad_norm": 0.11978469043970108,
      "learning_rate": 1.479189473620939e-05,
      "loss": 0.025,
      "step": 392
    },
    {
      "epoch": 1.0834374326943785,
      "grad_norm": 0.10893766582012177,
      "learning_rate": 1.4765687200213079e-05,
      "loss": 0.0239,
      "step": 393
    },
    {
      "epoch": 1.0861942709455095,
      "grad_norm": 0.1162242516875267,
      "learning_rate": 1.4739437244456039e-05,
      "loss": 0.024,
      "step": 394
    },
    {
      "epoch": 1.0889511091966402,
      "grad_norm": 0.11603809893131256,
      "learning_rate": 1.4713145102591209e-05,
      "loss": 0.0255,
      "step": 395
    },
    {
      "epoch": 1.0917079474477709,
      "grad_norm": 0.11185269802808762,
      "learning_rate": 1.4686811008647037e-05,
      "loss": 0.0244,
      "step": 396
    },
    {
      "epoch": 1.0944647856989016,
      "grad_norm": 0.11362854391336441,
      "learning_rate": 1.4660435197025391e-05,
      "loss": 0.0248,
      "step": 397
    },
    {
      "epoch": 1.0972216239500323,
      "grad_norm": 0.12096700072288513,
      "learning_rate": 1.463401790249946e-05,
      "loss": 0.0247,
      "step": 398
    },
    {
      "epoch": 1.099978462201163,
      "grad_norm": 0.12074792385101318,
      "learning_rate": 1.4607559360211688e-05,
      "loss": 0.025,
      "step": 399
    },
    {
      "epoch": 1.1027353004522937,
      "grad_norm": 0.13158197700977325,
      "learning_rate": 1.4581059805671662e-05,
      "loss": 0.026,
      "step": 400
    },
    {
      "epoch": 1.1054921387034244,
      "grad_norm": 0.1625628024339676,
      "learning_rate": 1.4554519474754025e-05,
      "loss": 0.0249,
      "step": 401
    },
    {
      "epoch": 1.1082489769545552,
      "grad_norm": 0.11231596022844315,
      "learning_rate": 1.4527938603696376e-05,
      "loss": 0.0246,
      "step": 402
    },
    {
      "epoch": 1.111005815205686,
      "grad_norm": 0.10914017260074615,
      "learning_rate": 1.4501317429097155e-05,
      "loss": 0.0247,
      "step": 403
    },
    {
      "epoch": 1.1137626534568168,
      "grad_norm": 0.11855874955654144,
      "learning_rate": 1.4474656187913558e-05,
      "loss": 0.0255,
      "step": 404
    },
    {
      "epoch": 1.1165194917079475,
      "grad_norm": 0.10382278263568878,
      "learning_rate": 1.4447955117459414e-05,
      "loss": 0.0235,
      "step": 405
    },
    {
      "epoch": 1.1192763299590782,
      "grad_norm": 0.11612187325954437,
      "learning_rate": 1.4421214455403071e-05,
      "loss": 0.0253,
      "step": 406
    },
    {
      "epoch": 1.122033168210209,
      "grad_norm": 0.11434487253427505,
      "learning_rate": 1.4394434439765295e-05,
      "loss": 0.0256,
      "step": 407
    },
    {
      "epoch": 1.1247900064613396,
      "grad_norm": 0.14260295033454895,
      "learning_rate": 1.436761530891713e-05,
      "loss": 0.0246,
      "step": 408
    },
    {
      "epoch": 1.1275468447124704,
      "grad_norm": 0.11852490156888962,
      "learning_rate": 1.4340757301577787e-05,
      "loss": 0.0242,
      "step": 409
    },
    {
      "epoch": 1.130303682963601,
      "grad_norm": 0.19492119550704956,
      "learning_rate": 1.4313860656812537e-05,
      "loss": 0.0254,
      "step": 410
    },
    {
      "epoch": 1.1330605212147318,
      "grad_norm": 0.14018341898918152,
      "learning_rate": 1.4286925614030542e-05,
      "loss": 0.0265,
      "step": 411
    },
    {
      "epoch": 1.1358173594658627,
      "grad_norm": 0.11521740257740021,
      "learning_rate": 1.425995241298277e-05,
      "loss": 0.0247,
      "step": 412
    },
    {
      "epoch": 1.1385741977169934,
      "grad_norm": 0.10554037988185883,
      "learning_rate": 1.423294129375982e-05,
      "loss": 0.0251,
      "step": 413
    },
    {
      "epoch": 1.1413310359681241,
      "grad_norm": 0.12840139865875244,
      "learning_rate": 1.4205892496789816e-05,
      "loss": 0.0248,
      "step": 414
    },
    {
      "epoch": 1.1440878742192548,
      "grad_norm": 0.11721502244472504,
      "learning_rate": 1.4178806262836252e-05,
      "loss": 0.0259,
      "step": 415
    },
    {
      "epoch": 1.1468447124703856,
      "grad_norm": 0.14833590388298035,
      "learning_rate": 1.4151682832995846e-05,
      "loss": 0.0238,
      "step": 416
    },
    {
      "epoch": 1.1496015507215163,
      "grad_norm": 0.10142626613378525,
      "learning_rate": 1.4124522448696407e-05,
      "loss": 0.0243,
      "step": 417
    },
    {
      "epoch": 1.152358388972647,
      "grad_norm": 0.13285695016384125,
      "learning_rate": 1.4097325351694674e-05,
      "loss": 0.0248,
      "step": 418
    },
    {
      "epoch": 1.1551152272237777,
      "grad_norm": 0.10552552342414856,
      "learning_rate": 1.407009178407417e-05,
      "loss": 0.0245,
      "step": 419
    },
    {
      "epoch": 1.1578720654749084,
      "grad_norm": 0.14879612624645233,
      "learning_rate": 1.404282198824305e-05,
      "loss": 0.0242,
      "step": 420
    },
    {
      "epoch": 1.160628903726039,
      "grad_norm": 0.13807976245880127,
      "learning_rate": 1.4015516206931932e-05,
      "loss": 0.0241,
      "step": 421
    },
    {
      "epoch": 1.1633857419771698,
      "grad_norm": 0.12726476788520813,
      "learning_rate": 1.3988174683191744e-05,
      "loss": 0.0244,
      "step": 422
    },
    {
      "epoch": 1.1661425802283008,
      "grad_norm": 0.11876961588859558,
      "learning_rate": 1.396079766039157e-05,
      "loss": 0.0241,
      "step": 423
    },
    {
      "epoch": 1.1688994184794315,
      "grad_norm": 0.13200323283672333,
      "learning_rate": 1.393338538221646e-05,
      "loss": 0.0254,
      "step": 424
    },
    {
      "epoch": 1.1716562567305622,
      "grad_norm": 0.1211104691028595,
      "learning_rate": 1.3905938092665283e-05,
      "loss": 0.0253,
      "step": 425
    },
    {
      "epoch": 1.1744130949816929,
      "grad_norm": 0.13063450157642365,
      "learning_rate": 1.387845603604855e-05,
      "loss": 0.0251,
      "step": 426
    },
    {
      "epoch": 1.1771699332328236,
      "grad_norm": 0.12195156514644623,
      "learning_rate": 1.385093945698623e-05,
      "loss": 0.0243,
      "step": 427
    },
    {
      "epoch": 1.1799267714839543,
      "grad_norm": 0.15483883023262024,
      "learning_rate": 1.382338860040558e-05,
      "loss": 0.0245,
      "step": 428
    },
    {
      "epoch": 1.182683609735085,
      "grad_norm": 0.11185739189386368,
      "learning_rate": 1.3795803711538966e-05,
      "loss": 0.0248,
      "step": 429
    },
    {
      "epoch": 1.1854404479862157,
      "grad_norm": 0.11978735029697418,
      "learning_rate": 1.3768185035921677e-05,
      "loss": 0.0242,
      "step": 430
    },
    {
      "epoch": 1.1881972862373464,
      "grad_norm": 0.11775299161672592,
      "learning_rate": 1.374053281938974e-05,
      "loss": 0.0245,
      "step": 431
    },
    {
      "epoch": 1.1909541244884774,
      "grad_norm": 0.10916128754615784,
      "learning_rate": 1.3712847308077737e-05,
      "loss": 0.024,
      "step": 432
    },
    {
      "epoch": 1.193710962739608,
      "grad_norm": 0.12981654703617096,
      "learning_rate": 1.3685128748416603e-05,
      "loss": 0.0251,
      "step": 433
    },
    {
      "epoch": 1.1964678009907388,
      "grad_norm": 0.1111079528927803,
      "learning_rate": 1.3657377387131443e-05,
      "loss": 0.0242,
      "step": 434
    },
    {
      "epoch": 1.1992246392418695,
      "grad_norm": 0.2875465154647827,
      "learning_rate": 1.3629593471239328e-05,
      "loss": 0.0246,
      "step": 435
    },
    {
      "epoch": 1.2019814774930002,
      "grad_norm": 0.10963843017816544,
      "learning_rate": 1.3601777248047105e-05,
      "loss": 0.0245,
      "step": 436
    },
    {
      "epoch": 1.204738315744131,
      "grad_norm": 0.15255135297775269,
      "learning_rate": 1.3573928965149188e-05,
      "loss": 0.0254,
      "step": 437
    },
    {
      "epoch": 1.2074951539952616,
      "grad_norm": 0.12013466656208038,
      "learning_rate": 1.3546048870425356e-05,
      "loss": 0.0252,
      "step": 438
    },
    {
      "epoch": 1.2102519922463923,
      "grad_norm": 0.11045455187559128,
      "learning_rate": 1.3518137212038554e-05,
      "loss": 0.0233,
      "step": 439
    },
    {
      "epoch": 1.213008830497523,
      "grad_norm": 0.11393214017152786,
      "learning_rate": 1.3490194238432665e-05,
      "loss": 0.0253,
      "step": 440
    },
    {
      "epoch": 1.215765668748654,
      "grad_norm": 0.39521655440330505,
      "learning_rate": 1.346222019833033e-05,
      "loss": 0.0243,
      "step": 441
    },
    {
      "epoch": 1.2185225069997847,
      "grad_norm": 0.11597371101379395,
      "learning_rate": 1.3434215340730692e-05,
      "loss": 0.0247,
      "step": 442
    },
    {
      "epoch": 1.2212793452509154,
      "grad_norm": 0.11905563622713089,
      "learning_rate": 1.340617991490722e-05,
      "loss": 0.0246,
      "step": 443
    },
    {
      "epoch": 1.2240361835020461,
      "grad_norm": 0.11847221851348877,
      "learning_rate": 1.3378114170405473e-05,
      "loss": 0.0244,
      "step": 444
    },
    {
      "epoch": 1.2267930217531768,
      "grad_norm": 0.11068813502788544,
      "learning_rate": 1.335001835704087e-05,
      "loss": 0.0252,
      "step": 445
    },
    {
      "epoch": 1.2295498600043075,
      "grad_norm": 0.1128387600183487,
      "learning_rate": 1.3321892724896483e-05,
      "loss": 0.025,
      "step": 446
    },
    {
      "epoch": 1.2323066982554383,
      "grad_norm": 0.10336292535066605,
      "learning_rate": 1.3293737524320798e-05,
      "loss": 0.0245,
      "step": 447
    },
    {
      "epoch": 1.235063536506569,
      "grad_norm": 0.11005783826112747,
      "learning_rate": 1.3265553005925494e-05,
      "loss": 0.0253,
      "step": 448
    },
    {
      "epoch": 1.2378203747576997,
      "grad_norm": 0.125181645154953,
      "learning_rate": 1.3237339420583213e-05,
      "loss": 0.0248,
      "step": 449
    },
    {
      "epoch": 1.2405772130088306,
      "grad_norm": 0.1266736388206482,
      "learning_rate": 1.3209097019425317e-05,
      "loss": 0.0255,
      "step": 450
    },
    {
      "epoch": 1.2433340512599613,
      "grad_norm": 0.13304491341114044,
      "learning_rate": 1.3180826053839668e-05,
      "loss": 0.0249,
      "step": 451
    },
    {
      "epoch": 1.246090889511092,
      "grad_norm": 0.09845131635665894,
      "learning_rate": 1.315252677546838e-05,
      "loss": 0.0242,
      "step": 452
    },
    {
      "epoch": 1.2488477277622227,
      "grad_norm": 0.11469008028507233,
      "learning_rate": 1.3124199436205575e-05,
      "loss": 0.0252,
      "step": 453
    },
    {
      "epoch": 1.2516045660133535,
      "grad_norm": 0.11134012043476105,
      "learning_rate": 1.309584428819516e-05,
      "loss": 0.0235,
      "step": 454
    },
    {
      "epoch": 1.2543614042644842,
      "grad_norm": 0.10330889374017715,
      "learning_rate": 1.3067461583828553e-05,
      "loss": 0.0251,
      "step": 455
    },
    {
      "epoch": 1.2571182425156149,
      "grad_norm": 0.10626427829265594,
      "learning_rate": 1.303905157574247e-05,
      "loss": 0.0232,
      "step": 456
    },
    {
      "epoch": 1.2598750807667456,
      "grad_norm": 0.10066665709018707,
      "learning_rate": 1.3010614516816652e-05,
      "loss": 0.0244,
      "step": 457
    },
    {
      "epoch": 1.2626319190178763,
      "grad_norm": 0.0952746570110321,
      "learning_rate": 1.2982150660171613e-05,
      "loss": 0.0231,
      "step": 458
    },
    {
      "epoch": 1.2653887572690072,
      "grad_norm": 0.11685788631439209,
      "learning_rate": 1.2953660259166413e-05,
      "loss": 0.024,
      "step": 459
    },
    {
      "epoch": 1.2681455955201377,
      "grad_norm": 0.10656964778900146,
      "learning_rate": 1.2925143567396374e-05,
      "loss": 0.0253,
      "step": 460
    },
    {
      "epoch": 1.2709024337712687,
      "grad_norm": 0.12450357526540756,
      "learning_rate": 1.2896600838690838e-05,
      "loss": 0.0241,
      "step": 461
    },
    {
      "epoch": 1.2736592720223994,
      "grad_norm": 0.10607194900512695,
      "learning_rate": 1.2868032327110904e-05,
      "loss": 0.0238,
      "step": 462
    },
    {
      "epoch": 1.27641611027353,
      "grad_norm": 0.12448973953723907,
      "learning_rate": 1.2839438286947163e-05,
      "loss": 0.0245,
      "step": 463
    },
    {
      "epoch": 1.2791729485246608,
      "grad_norm": 0.11434370279312134,
      "learning_rate": 1.2810818972717438e-05,
      "loss": 0.0236,
      "step": 464
    },
    {
      "epoch": 1.2819297867757915,
      "grad_norm": 0.11489138007164001,
      "learning_rate": 1.2782174639164528e-05,
      "loss": 0.0241,
      "step": 465
    },
    {
      "epoch": 1.2846866250269222,
      "grad_norm": 0.11225125193595886,
      "learning_rate": 1.2753505541253917e-05,
      "loss": 0.0238,
      "step": 466
    },
    {
      "epoch": 1.287443463278053,
      "grad_norm": 0.11676695942878723,
      "learning_rate": 1.272481193417153e-05,
      "loss": 0.0251,
      "step": 467
    },
    {
      "epoch": 1.2902003015291836,
      "grad_norm": 0.10919535160064697,
      "learning_rate": 1.269609407332144e-05,
      "loss": 0.0248,
      "step": 468
    },
    {
      "epoch": 1.2929571397803143,
      "grad_norm": 0.11963889002799988,
      "learning_rate": 1.2667352214323614e-05,
      "loss": 0.0253,
      "step": 469
    },
    {
      "epoch": 1.2957139780314453,
      "grad_norm": 0.11035304516553879,
      "learning_rate": 1.2638586613011625e-05,
      "loss": 0.0241,
      "step": 470
    },
    {
      "epoch": 1.298470816282576,
      "grad_norm": 0.1180749386548996,
      "learning_rate": 1.2609797525430374e-05,
      "loss": 0.0257,
      "step": 471
    },
    {
      "epoch": 1.3012276545337067,
      "grad_norm": 0.11271151155233383,
      "learning_rate": 1.258098520783382e-05,
      "loss": 0.024,
      "step": 472
    },
    {
      "epoch": 1.3039844927848374,
      "grad_norm": 0.10727877169847488,
      "learning_rate": 1.2552149916682695e-05,
      "loss": 0.0244,
      "step": 473
    },
    {
      "epoch": 1.3067413310359681,
      "grad_norm": 0.10205347836017609,
      "learning_rate": 1.2523291908642219e-05,
      "loss": 0.0235,
      "step": 474
    },
    {
      "epoch": 1.3094981692870988,
      "grad_norm": 0.10410701483488083,
      "learning_rate": 1.2494411440579814e-05,
      "loss": 0.0245,
      "step": 475
    },
    {
      "epoch": 1.3122550075382295,
      "grad_norm": 0.10330906510353088,
      "learning_rate": 1.2465508769562824e-05,
      "loss": 0.0246,
      "step": 476
    },
    {
      "epoch": 1.3150118457893603,
      "grad_norm": 0.10153473168611526,
      "learning_rate": 1.243658415285622e-05,
      "loss": 0.024,
      "step": 477
    },
    {
      "epoch": 1.317768684040491,
      "grad_norm": 0.13269540667533875,
      "learning_rate": 1.240763784792032e-05,
      "loss": 0.0265,
      "step": 478
    },
    {
      "epoch": 1.320525522291622,
      "grad_norm": 0.1011882796883583,
      "learning_rate": 1.2378670112408482e-05,
      "loss": 0.0244,
      "step": 479
    },
    {
      "epoch": 1.3232823605427526,
      "grad_norm": 0.10707399994134903,
      "learning_rate": 1.2349681204164823e-05,
      "loss": 0.0239,
      "step": 480
    },
    {
      "epoch": 1.3260391987938833,
      "grad_norm": 0.1126994714140892,
      "learning_rate": 1.2320671381221923e-05,
      "loss": 0.0256,
      "step": 481
    },
    {
      "epoch": 1.328796037045014,
      "grad_norm": 0.10313718020915985,
      "learning_rate": 1.229164090179852e-05,
      "loss": 0.0243,
      "step": 482
    },
    {
      "epoch": 1.3315528752961447,
      "grad_norm": 0.1096411719918251,
      "learning_rate": 1.2262590024297226e-05,
      "loss": 0.0242,
      "step": 483
    },
    {
      "epoch": 1.3343097135472755,
      "grad_norm": 0.11218464374542236,
      "learning_rate": 1.2233519007302201e-05,
      "loss": 0.0243,
      "step": 484
    },
    {
      "epoch": 1.3370665517984062,
      "grad_norm": 0.09545588493347168,
      "learning_rate": 1.2204428109576888e-05,
      "loss": 0.0235,
      "step": 485
    },
    {
      "epoch": 1.3398233900495369,
      "grad_norm": 0.10384158790111542,
      "learning_rate": 1.2175317590061676e-05,
      "loss": 0.0239,
      "step": 486
    },
    {
      "epoch": 1.3425802283006676,
      "grad_norm": 0.10288365185260773,
      "learning_rate": 1.2146187707871617e-05,
      "loss": 0.0246,
      "step": 487
    },
    {
      "epoch": 1.3453370665517985,
      "grad_norm": 0.12427128106355667,
      "learning_rate": 1.211703872229411e-05,
      "loss": 0.0247,
      "step": 488
    },
    {
      "epoch": 1.348093904802929,
      "grad_norm": 0.11715073883533478,
      "learning_rate": 1.2087870892786588e-05,
      "loss": 0.0245,
      "step": 489
    },
    {
      "epoch": 1.35085074305406,
      "grad_norm": 0.10031214356422424,
      "learning_rate": 1.2058684478974226e-05,
      "loss": 0.0237,
      "step": 490
    },
    {
      "epoch": 1.3536075813051907,
      "grad_norm": 0.13713191449642181,
      "learning_rate": 1.2029479740647613e-05,
      "loss": 0.0226,
      "step": 491
    },
    {
      "epoch": 1.3563644195563214,
      "grad_norm": 0.10992252826690674,
      "learning_rate": 1.2000256937760446e-05,
      "loss": 0.0254,
      "step": 492
    },
    {
      "epoch": 1.359121257807452,
      "grad_norm": 0.12964554131031036,
      "learning_rate": 1.1971016330427217e-05,
      "loss": 0.0249,
      "step": 493
    },
    {
      "epoch": 1.3618780960585828,
      "grad_norm": 0.10349434614181519,
      "learning_rate": 1.1941758178920899e-05,
      "loss": 0.0243,
      "step": 494
    },
    {
      "epoch": 1.3646349343097135,
      "grad_norm": 0.13970984518527985,
      "learning_rate": 1.1912482743670624e-05,
      "loss": 0.024,
      "step": 495
    },
    {
      "epoch": 1.3673917725608442,
      "grad_norm": 0.10153508186340332,
      "learning_rate": 1.188319028525937e-05,
      "loss": 0.0237,
      "step": 496
    },
    {
      "epoch": 1.3701486108119751,
      "grad_norm": 0.12652255594730377,
      "learning_rate": 1.1853881064421634e-05,
      "loss": 0.0248,
      "step": 497
    },
    {
      "epoch": 1.3729054490631056,
      "grad_norm": 0.1191074326634407,
      "learning_rate": 1.1824555342041129e-05,
      "loss": 0.0244,
      "step": 498
    },
    {
      "epoch": 1.3756622873142366,
      "grad_norm": 0.11915133893489838,
      "learning_rate": 1.1795213379148436e-05,
      "loss": 0.0232,
      "step": 499
    },
    {
      "epoch": 1.3784191255653673,
      "grad_norm": 0.11027630418539047,
      "learning_rate": 1.17658554369187e-05,
      "loss": 0.0255,
      "step": 500
    },
    {
      "epoch": 1.381175963816498,
      "grad_norm": 0.1299276500940323,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.0243,
      "step": 501
    },
    {
      "epoch": 1.3839328020676287,
      "grad_norm": 0.10648605227470398,
      "learning_rate": 1.1707092659857531e-05,
      "loss": 0.024,
      "step": 502
    },
    {
      "epoch": 1.3866896403187594,
      "grad_norm": 0.12794575095176697,
      "learning_rate": 1.1677688348078244e-05,
      "loss": 0.0241,
      "step": 503
    },
    {
      "epoch": 1.3894464785698901,
      "grad_norm": 0.11318343877792358,
      "learning_rate": 1.1648269103061567e-05,
      "loss": 0.0239,
      "step": 504
    },
    {
      "epoch": 1.3922033168210208,
      "grad_norm": 0.10124088078737259,
      "learning_rate": 1.1618835186670532e-05,
      "loss": 0.0231,
      "step": 505
    },
    {
      "epoch": 1.3949601550721518,
      "grad_norm": 0.10141489654779434,
      "learning_rate": 1.1589386860898773e-05,
      "loss": 0.0246,
      "step": 506
    },
    {
      "epoch": 1.3977169933232823,
      "grad_norm": 0.11699099838733673,
      "learning_rate": 1.155992438786818e-05,
      "loss": 0.0237,
      "step": 507
    },
    {
      "epoch": 1.4004738315744132,
      "grad_norm": 0.10175668448209763,
      "learning_rate": 1.1530448029826566e-05,
      "loss": 0.023,
      "step": 508
    },
    {
      "epoch": 1.403230669825544,
      "grad_norm": 0.12864287197589874,
      "learning_rate": 1.1500958049145342e-05,
      "loss": 0.0247,
      "step": 509
    },
    {
      "epoch": 1.4059875080766746,
      "grad_norm": 0.8590685129165649,
      "learning_rate": 1.1471454708317163e-05,
      "loss": 0.0278,
      "step": 510
    },
    {
      "epoch": 1.4087443463278053,
      "grad_norm": 0.11892825365066528,
      "learning_rate": 1.1441938269953619e-05,
      "loss": 0.0251,
      "step": 511
    },
    {
      "epoch": 1.411501184578936,
      "grad_norm": 0.12736280262470245,
      "learning_rate": 1.1412408996782871e-05,
      "loss": 0.024,
      "step": 512
    },
    {
      "epoch": 1.4142580228300667,
      "grad_norm": 0.09979959577322006,
      "learning_rate": 1.1382867151647333e-05,
      "loss": 0.0234,
      "step": 513
    },
    {
      "epoch": 1.4170148610811975,
      "grad_norm": 0.11944937705993652,
      "learning_rate": 1.1353312997501313e-05,
      "loss": 0.0241,
      "step": 514
    },
    {
      "epoch": 1.4197716993323282,
      "grad_norm": 0.09399650245904922,
      "learning_rate": 1.1323746797408687e-05,
      "loss": 0.0229,
      "step": 515
    },
    {
      "epoch": 1.4225285375834589,
      "grad_norm": 0.11699047684669495,
      "learning_rate": 1.1294168814540554e-05,
      "loss": 0.025,
      "step": 516
    },
    {
      "epoch": 1.4252853758345898,
      "grad_norm": 0.137738898396492,
      "learning_rate": 1.1264579312172896e-05,
      "loss": 0.0253,
      "step": 517
    },
    {
      "epoch": 1.4280422140857205,
      "grad_norm": 0.11010394990444183,
      "learning_rate": 1.123497855368422e-05,
      "loss": 0.0245,
      "step": 518
    },
    {
      "epoch": 1.4307990523368512,
      "grad_norm": 0.09499143809080124,
      "learning_rate": 1.1205366802553231e-05,
      "loss": 0.0232,
      "step": 519
    },
    {
      "epoch": 1.433555890587982,
      "grad_norm": 0.11812635511159897,
      "learning_rate": 1.1175744322356488e-05,
      "loss": 0.0238,
      "step": 520
    },
    {
      "epoch": 1.4363127288391127,
      "grad_norm": 0.10630928725004196,
      "learning_rate": 1.1146111376766033e-05,
      "loss": 0.0247,
      "step": 521
    },
    {
      "epoch": 1.4390695670902434,
      "grad_norm": 0.11619951575994492,
      "learning_rate": 1.1116468229547079e-05,
      "loss": 0.024,
      "step": 522
    },
    {
      "epoch": 1.441826405341374,
      "grad_norm": 0.09959083050489426,
      "learning_rate": 1.1086815144555633e-05,
      "loss": 0.0244,
      "step": 523
    },
    {
      "epoch": 1.4445832435925048,
      "grad_norm": 0.10520867258310318,
      "learning_rate": 1.105715238573616e-05,
      "loss": 0.0248,
      "step": 524
    },
    {
      "epoch": 1.4473400818436355,
      "grad_norm": 0.09931975603103638,
      "learning_rate": 1.1027480217119245e-05,
      "loss": 0.0237,
      "step": 525
    },
    {
      "epoch": 1.4500969200947664,
      "grad_norm": 0.10330688953399658,
      "learning_rate": 1.0997798902819208e-05,
      "loss": 0.0245,
      "step": 526
    },
    {
      "epoch": 1.452853758345897,
      "grad_norm": 0.09967660158872604,
      "learning_rate": 1.0968108707031792e-05,
      "loss": 0.0245,
      "step": 527
    },
    {
      "epoch": 1.4556105965970279,
      "grad_norm": 0.11198758333921432,
      "learning_rate": 1.0938409894031793e-05,
      "loss": 0.0243,
      "step": 528
    },
    {
      "epoch": 1.4583674348481586,
      "grad_norm": 0.8267385959625244,
      "learning_rate": 1.0908702728170706e-05,
      "loss": 0.0254,
      "step": 529
    },
    {
      "epoch": 1.4611242730992893,
      "grad_norm": 0.09910894185304642,
      "learning_rate": 1.0878987473874381e-05,
      "loss": 0.0238,
      "step": 530
    },
    {
      "epoch": 1.46388111135042,
      "grad_norm": 0.10041078925132751,
      "learning_rate": 1.084926439564065e-05,
      "loss": 0.0246,
      "step": 531
    },
    {
      "epoch": 1.4666379496015507,
      "grad_norm": 0.1009271889925003,
      "learning_rate": 1.0819533758037002e-05,
      "loss": 0.0241,
      "step": 532
    },
    {
      "epoch": 1.4693947878526814,
      "grad_norm": 0.10631167888641357,
      "learning_rate": 1.0789795825698206e-05,
      "loss": 0.0249,
      "step": 533
    },
    {
      "epoch": 1.4721516261038121,
      "grad_norm": 0.1118827611207962,
      "learning_rate": 1.0760050863323961e-05,
      "loss": 0.0244,
      "step": 534
    },
    {
      "epoch": 1.474908464354943,
      "grad_norm": 0.11578582972288132,
      "learning_rate": 1.0730299135676545e-05,
      "loss": 0.0239,
      "step": 535
    },
    {
      "epoch": 1.4776653026060735,
      "grad_norm": 0.102451853454113,
      "learning_rate": 1.0700540907578447e-05,
      "loss": 0.0228,
      "step": 536
    },
    {
      "epoch": 1.4804221408572045,
      "grad_norm": 0.09739474207162857,
      "learning_rate": 1.0670776443910024e-05,
      "loss": 0.0235,
      "step": 537
    },
    {
      "epoch": 1.4831789791083352,
      "grad_norm": 0.1013341024518013,
      "learning_rate": 1.0641006009607137e-05,
      "loss": 0.0241,
      "step": 538
    },
    {
      "epoch": 1.485935817359466,
      "grad_norm": 0.10632984340190887,
      "learning_rate": 1.0611229869658785e-05,
      "loss": 0.0235,
      "step": 539
    },
    {
      "epoch": 1.4886926556105966,
      "grad_norm": 0.115880087018013,
      "learning_rate": 1.0581448289104759e-05,
      "loss": 0.0238,
      "step": 540
    },
    {
      "epoch": 1.4914494938617273,
      "grad_norm": 0.1438998281955719,
      "learning_rate": 1.0551661533033274e-05,
      "loss": 0.0249,
      "step": 541
    },
    {
      "epoch": 1.494206332112858,
      "grad_norm": 0.10747671127319336,
      "learning_rate": 1.052186986657862e-05,
      "loss": 0.0239,
      "step": 542
    },
    {
      "epoch": 1.4969631703639887,
      "grad_norm": 0.11497830599546432,
      "learning_rate": 1.0492073554918782e-05,
      "loss": 0.024,
      "step": 543
    },
    {
      "epoch": 1.4997200086151197,
      "grad_norm": 0.11187911778688431,
      "learning_rate": 1.0462272863273105e-05,
      "loss": 0.0237,
      "step": 544
    },
    {
      "epoch": 1.5024768468662502,
      "grad_norm": 0.10226496309041977,
      "learning_rate": 1.0432468056899909e-05,
      "loss": 0.0226,
      "step": 545
    },
    {
      "epoch": 1.505233685117381,
      "grad_norm": 0.12337212264537811,
      "learning_rate": 1.0402659401094154e-05,
      "loss": 0.0237,
      "step": 546
    },
    {
      "epoch": 1.5079905233685116,
      "grad_norm": 0.1370334029197693,
      "learning_rate": 1.0372847161185047e-05,
      "loss": 0.0246,
      "step": 547
    },
    {
      "epoch": 1.5107473616196425,
      "grad_norm": 0.1170111671090126,
      "learning_rate": 1.0343031602533713e-05,
      "loss": 0.0232,
      "step": 548
    },
    {
      "epoch": 1.5135041998707732,
      "grad_norm": 0.10803765803575516,
      "learning_rate": 1.0313212990530804e-05,
      "loss": 0.0241,
      "step": 549
    },
    {
      "epoch": 1.516261038121904,
      "grad_norm": 0.19091671705245972,
      "learning_rate": 1.028339159059416e-05,
      "loss": 0.025,
      "step": 550
    },
    {
      "epoch": 1.5190178763730346,
      "grad_norm": 0.11063394695520401,
      "learning_rate": 1.0253567668166436e-05,
      "loss": 0.024,
      "step": 551
    },
    {
      "epoch": 1.5217747146241654,
      "grad_norm": 0.11760086566209793,
      "learning_rate": 1.0223741488712732e-05,
      "loss": 0.0239,
      "step": 552
    },
    {
      "epoch": 1.5245315528752963,
      "grad_norm": 0.10287030041217804,
      "learning_rate": 1.0193913317718245e-05,
      "loss": 0.0237,
      "step": 553
    },
    {
      "epoch": 1.5272883911264268,
      "grad_norm": 0.1050201803445816,
      "learning_rate": 1.0164083420685898e-05,
      "loss": 0.0244,
      "step": 554
    },
    {
      "epoch": 1.5300452293775577,
      "grad_norm": 0.10405981540679932,
      "learning_rate": 1.0134252063133976e-05,
      "loss": 0.0231,
      "step": 555
    },
    {
      "epoch": 1.5328020676286882,
      "grad_norm": 0.09600961208343506,
      "learning_rate": 1.0104419510593764e-05,
      "loss": 0.0222,
      "step": 556
    },
    {
      "epoch": 1.5355589058798191,
      "grad_norm": 0.2610664367675781,
      "learning_rate": 1.0074586028607184e-05,
      "loss": 0.0242,
      "step": 557
    },
    {
      "epoch": 1.5383157441309498,
      "grad_norm": 0.10342807322740555,
      "learning_rate": 1.0044751882724436e-05,
      "loss": 0.0226,
      "step": 558
    },
    {
      "epoch": 1.5410725823820806,
      "grad_norm": 0.11091192066669464,
      "learning_rate": 1.0014917338501618e-05,
      "loss": 0.0234,
      "step": 559
    },
    {
      "epoch": 1.5438294206332113,
      "grad_norm": 0.10212964564561844,
      "learning_rate": 9.985082661498384e-06,
      "loss": 0.0241,
      "step": 560
    },
    {
      "epoch": 1.546586258884342,
      "grad_norm": 0.12537163496017456,
      "learning_rate": 9.955248117275566e-06,
      "loss": 0.026,
      "step": 561
    },
    {
      "epoch": 1.549343097135473,
      "grad_norm": 0.10132145881652832,
      "learning_rate": 9.925413971392816e-06,
      "loss": 0.0237,
      "step": 562
    },
    {
      "epoch": 1.5520999353866034,
      "grad_norm": 0.11508084088563919,
      "learning_rate": 9.89558048940624e-06,
      "loss": 0.0235,
      "step": 563
    },
    {
      "epoch": 1.5548567736377343,
      "grad_norm": 0.11934718489646912,
      "learning_rate": 9.865747936866027e-06,
      "loss": 0.0234,
      "step": 564
    },
    {
      "epoch": 1.5576136118888648,
      "grad_norm": 0.10267031192779541,
      "learning_rate": 9.835916579314105e-06,
      "loss": 0.0235,
      "step": 565
    },
    {
      "epoch": 1.5603704501399958,
      "grad_norm": 0.10713180154561996,
      "learning_rate": 9.806086682281759e-06,
      "loss": 0.0235,
      "step": 566
    },
    {
      "epoch": 1.5631272883911265,
      "grad_norm": 0.11230739206075668,
      "learning_rate": 9.776258511287271e-06,
      "loss": 0.0237,
      "step": 567
    },
    {
      "epoch": 1.5658841266422572,
      "grad_norm": 0.11151719093322754,
      "learning_rate": 9.746432331833568e-06,
      "loss": 0.0232,
      "step": 568
    },
    {
      "epoch": 1.568640964893388,
      "grad_norm": 0.10779338330030441,
      "learning_rate": 9.716608409405843e-06,
      "loss": 0.0236,
      "step": 569
    },
    {
      "epoch": 1.5713978031445186,
      "grad_norm": 0.10462480783462524,
      "learning_rate": 9.6867870094692e-06,
      "loss": 0.0232,
      "step": 570
    },
    {
      "epoch": 1.5741546413956493,
      "grad_norm": 0.11873821914196014,
      "learning_rate": 9.65696839746629e-06,
      "loss": 0.0235,
      "step": 571
    },
    {
      "epoch": 1.57691147964678,
      "grad_norm": 0.10550446808338165,
      "learning_rate": 9.627152838814954e-06,
      "loss": 0.0231,
      "step": 572
    },
    {
      "epoch": 1.579668317897911,
      "grad_norm": 0.10466040670871735,
      "learning_rate": 9.597340598905851e-06,
      "loss": 0.0241,
      "step": 573
    },
    {
      "epoch": 1.5824251561490414,
      "grad_norm": 0.09853997081518173,
      "learning_rate": 9.567531943100093e-06,
      "loss": 0.0227,
      "step": 574
    },
    {
      "epoch": 1.5851819944001724,
      "grad_norm": 0.4732920825481415,
      "learning_rate": 9.537727136726899e-06,
      "loss": 0.0275,
      "step": 575
    },
    {
      "epoch": 1.587938832651303,
      "grad_norm": 0.11485108733177185,
      "learning_rate": 9.50792644508122e-06,
      "loss": 0.0229,
      "step": 576
    },
    {
      "epoch": 1.5906956709024338,
      "grad_norm": 0.10848281532526016,
      "learning_rate": 9.478130133421381e-06,
      "loss": 0.0237,
      "step": 577
    },
    {
      "epoch": 1.5934525091535645,
      "grad_norm": 0.10901174694299698,
      "learning_rate": 9.448338466966727e-06,
      "loss": 0.0233,
      "step": 578
    },
    {
      "epoch": 1.5962093474046952,
      "grad_norm": 0.10002460330724716,
      "learning_rate": 9.418551710895243e-06,
      "loss": 0.0227,
      "step": 579
    },
    {
      "epoch": 1.598966185655826,
      "grad_norm": 0.10787548869848251,
      "learning_rate": 9.388770130341217e-06,
      "loss": 0.0235,
      "step": 580
    },
    {
      "epoch": 1.6017230239069566,
      "grad_norm": 0.10888142138719559,
      "learning_rate": 9.358993990392864e-06,
      "loss": 0.0235,
      "step": 581
    },
    {
      "epoch": 1.6044798621580876,
      "grad_norm": 0.11493578553199768,
      "learning_rate": 9.329223556089976e-06,
      "loss": 0.0244,
      "step": 582
    },
    {
      "epoch": 1.607236700409218,
      "grad_norm": 0.1134362742304802,
      "learning_rate": 9.299459092421558e-06,
      "loss": 0.0251,
      "step": 583
    },
    {
      "epoch": 1.609993538660349,
      "grad_norm": 0.10327161103487015,
      "learning_rate": 9.269700864323462e-06,
      "loss": 0.024,
      "step": 584
    },
    {
      "epoch": 1.6127503769114795,
      "grad_norm": 0.10661885142326355,
      "learning_rate": 9.239949136676042e-06,
      "loss": 0.0237,
      "step": 585
    },
    {
      "epoch": 1.6155072151626104,
      "grad_norm": 0.11634720116853714,
      "learning_rate": 9.210204174301797e-06,
      "loss": 0.0239,
      "step": 586
    },
    {
      "epoch": 1.6182640534137411,
      "grad_norm": 0.10218880325555801,
      "learning_rate": 9.180466241963e-06,
      "loss": 0.0231,
      "step": 587
    },
    {
      "epoch": 1.6210208916648718,
      "grad_norm": 0.10723837465047836,
      "learning_rate": 9.15073560435935e-06,
      "loss": 0.0233,
      "step": 588
    },
    {
      "epoch": 1.6237777299160026,
      "grad_norm": 0.10045973211526871,
      "learning_rate": 9.121012526125625e-06,
      "loss": 0.0229,
      "step": 589
    },
    {
      "epoch": 1.6265345681671333,
      "grad_norm": 0.1160292997956276,
      "learning_rate": 9.091297271829296e-06,
      "loss": 0.024,
      "step": 590
    },
    {
      "epoch": 1.6292914064182642,
      "grad_norm": 0.0929265022277832,
      "learning_rate": 9.061590105968208e-06,
      "loss": 0.022,
      "step": 591
    },
    {
      "epoch": 1.6320482446693947,
      "grad_norm": 0.09764169901609421,
      "learning_rate": 9.03189129296821e-06,
      "loss": 0.0237,
      "step": 592
    },
    {
      "epoch": 1.6348050829205256,
      "grad_norm": 0.10847126692533493,
      "learning_rate": 9.002201097180796e-06,
      "loss": 0.0229,
      "step": 593
    },
    {
      "epoch": 1.637561921171656,
      "grad_norm": 0.10012809187173843,
      "learning_rate": 8.97251978288076e-06,
      "loss": 0.0237,
      "step": 594
    },
    {
      "epoch": 1.640318759422787,
      "grad_norm": 0.10542436689138412,
      "learning_rate": 8.942847614263842e-06,
      "loss": 0.0225,
      "step": 595
    },
    {
      "epoch": 1.6430755976739178,
      "grad_norm": 0.10059002041816711,
      "learning_rate": 8.91318485544437e-06,
      "loss": 0.0229,
      "step": 596
    },
    {
      "epoch": 1.6458324359250485,
      "grad_norm": 0.09918685257434845,
      "learning_rate": 8.883531770452924e-06,
      "loss": 0.023,
      "step": 597
    },
    {
      "epoch": 1.6485892741761792,
      "grad_norm": 0.7748421430587769,
      "learning_rate": 8.853888623233967e-06,
      "loss": 0.0313,
      "step": 598
    },
    {
      "epoch": 1.6513461124273099,
      "grad_norm": 0.1017061397433281,
      "learning_rate": 8.824255677643518e-06,
      "loss": 0.0237,
      "step": 599
    },
    {
      "epoch": 1.6541029506784408,
      "grad_norm": 0.10307373106479645,
      "learning_rate": 8.79463319744677e-06,
      "loss": 0.0238,
      "step": 600
    },
    {
      "epoch": 1.6568597889295713,
      "grad_norm": 0.09575623273849487,
      "learning_rate": 8.765021446315784e-06,
      "loss": 0.0227,
      "step": 601
    },
    {
      "epoch": 1.6596166271807022,
      "grad_norm": 0.10236816108226776,
      "learning_rate": 8.735420687827107e-06,
      "loss": 0.0242,
      "step": 602
    },
    {
      "epoch": 1.6623734654318327,
      "grad_norm": 0.1926388442516327,
      "learning_rate": 8.705831185459446e-06,
      "loss": 0.0269,
      "step": 603
    },
    {
      "epoch": 1.6651303036829637,
      "grad_norm": 0.10154667496681213,
      "learning_rate": 8.676253202591318e-06,
      "loss": 0.0238,
      "step": 604
    },
    {
      "epoch": 1.6678871419340944,
      "grad_norm": 0.10679133236408234,
      "learning_rate": 8.646687002498692e-06,
      "loss": 0.0235,
      "step": 605
    },
    {
      "epoch": 1.670643980185225,
      "grad_norm": 0.1166776716709137,
      "learning_rate": 8.617132848352672e-06,
      "loss": 0.0228,
      "step": 606
    },
    {
      "epoch": 1.6734008184363558,
      "grad_norm": 0.10293897241353989,
      "learning_rate": 8.58759100321713e-06,
      "loss": 0.022,
      "step": 607
    },
    {
      "epoch": 1.6761576566874865,
      "grad_norm": 0.3156561255455017,
      "learning_rate": 8.558061730046385e-06,
      "loss": 0.0259,
      "step": 608
    },
    {
      "epoch": 1.6789144949386172,
      "grad_norm": 0.10844126343727112,
      "learning_rate": 8.528545291682839e-06,
      "loss": 0.0227,
      "step": 609
    },
    {
      "epoch": 1.681671333189748,
      "grad_norm": 0.10488175600767136,
      "learning_rate": 8.499041950854665e-06,
      "loss": 0.0239,
      "step": 610
    },
    {
      "epoch": 1.6844281714408789,
      "grad_norm": 0.12386349588632584,
      "learning_rate": 8.469551970173437e-06,
      "loss": 0.0238,
      "step": 611
    },
    {
      "epoch": 1.6871850096920094,
      "grad_norm": 0.11962109059095383,
      "learning_rate": 8.440075612131823e-06,
      "loss": 0.0241,
      "step": 612
    },
    {
      "epoch": 1.6899418479431403,
      "grad_norm": 0.09642872214317322,
      "learning_rate": 8.410613139101229e-06,
      "loss": 0.0227,
      "step": 613
    },
    {
      "epoch": 1.692698686194271,
      "grad_norm": 0.10245215147733688,
      "learning_rate": 8.38116481332947e-06,
      "loss": 0.0224,
      "step": 614
    },
    {
      "epoch": 1.6954555244454017,
      "grad_norm": 0.10805162042379379,
      "learning_rate": 8.351730896938438e-06,
      "loss": 0.024,
      "step": 615
    },
    {
      "epoch": 1.6982123626965324,
      "grad_norm": 0.10161999613046646,
      "learning_rate": 8.322311651921759e-06,
      "loss": 0.0228,
      "step": 616
    },
    {
      "epoch": 1.7009692009476631,
      "grad_norm": 0.10508878529071808,
      "learning_rate": 8.292907340142472e-06,
      "loss": 0.0241,
      "step": 617
    },
    {
      "epoch": 1.7037260391987938,
      "grad_norm": 0.10359475761651993,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.0241,
      "step": 618
    },
    {
      "epoch": 1.7064828774499246,
      "grad_norm": 0.11690201610326767,
      "learning_rate": 8.2341445630813e-06,
      "loss": 0.0236,
      "step": 619
    },
    {
      "epoch": 1.7092397157010555,
      "grad_norm": 0.1021987721323967,
      "learning_rate": 8.204786620851569e-06,
      "loss": 0.0227,
      "step": 620
    },
    {
      "epoch": 1.711996553952186,
      "grad_norm": 0.10730181634426117,
      "learning_rate": 8.175444657958875e-06,
      "loss": 0.0221,
      "step": 621
    },
    {
      "epoch": 1.714753392203317,
      "grad_norm": 0.10022248327732086,
      "learning_rate": 8.146118935578367e-06,
      "loss": 0.0239,
      "step": 622
    },
    {
      "epoch": 1.7175102304544474,
      "grad_norm": 0.11165450513362885,
      "learning_rate": 8.116809714740634e-06,
      "loss": 0.0229,
      "step": 623
    },
    {
      "epoch": 1.7202670687055783,
      "grad_norm": 1.2205007076263428,
      "learning_rate": 8.087517256329376e-06,
      "loss": 0.024,
      "step": 624
    },
    {
      "epoch": 1.723023906956709,
      "grad_norm": 0.09469296783208847,
      "learning_rate": 8.058241821079106e-06,
      "loss": 0.0234,
      "step": 625
    },
    {
      "epoch": 1.7257807452078398,
      "grad_norm": 0.09778100997209549,
      "learning_rate": 8.028983669572786e-06,
      "loss": 0.0229,
      "step": 626
    },
    {
      "epoch": 1.7285375834589705,
      "grad_norm": 0.10135500133037567,
      "learning_rate": 7.999743062239557e-06,
      "loss": 0.0227,
      "step": 627
    },
    {
      "epoch": 1.7312944217101012,
      "grad_norm": 0.0966990739107132,
      "learning_rate": 7.97052025935239e-06,
      "loss": 0.0232,
      "step": 628
    },
    {
      "epoch": 1.734051259961232,
      "grad_norm": 0.09644000977277756,
      "learning_rate": 7.941315521025776e-06,
      "loss": 0.023,
      "step": 629
    },
    {
      "epoch": 1.7368080982123626,
      "grad_norm": 0.10050094872713089,
      "learning_rate": 7.912129107213417e-06,
      "loss": 0.022,
      "step": 630
    },
    {
      "epoch": 1.7395649364634935,
      "grad_norm": 0.09904894977807999,
      "learning_rate": 7.882961277705897e-06,
      "loss": 0.0224,
      "step": 631
    },
    {
      "epoch": 1.742321774714624,
      "grad_norm": 0.1084563285112381,
      "learning_rate": 7.853812292128386e-06,
      "loss": 0.0228,
      "step": 632
    },
    {
      "epoch": 1.745078612965755,
      "grad_norm": 0.11320846527814865,
      "learning_rate": 7.824682409938328e-06,
      "loss": 0.0235,
      "step": 633
    },
    {
      "epoch": 1.7478354512168857,
      "grad_norm": 0.1111724004149437,
      "learning_rate": 7.795571890423116e-06,
      "loss": 0.023,
      "step": 634
    },
    {
      "epoch": 1.7505922894680164,
      "grad_norm": 0.09731971472501755,
      "learning_rate": 7.766480992697802e-06,
      "loss": 0.0227,
      "step": 635
    },
    {
      "epoch": 1.753349127719147,
      "grad_norm": 0.11039575189352036,
      "learning_rate": 7.73740997570278e-06,
      "loss": 0.0225,
      "step": 636
    },
    {
      "epoch": 1.7561059659702778,
      "grad_norm": 0.10182961076498032,
      "learning_rate": 7.708359098201483e-06,
      "loss": 0.022,
      "step": 637
    },
    {
      "epoch": 1.7588628042214087,
      "grad_norm": 0.103904128074646,
      "learning_rate": 7.67932861877808e-06,
      "loss": 0.0227,
      "step": 638
    },
    {
      "epoch": 1.7616196424725392,
      "grad_norm": 0.12430708110332489,
      "learning_rate": 7.650318795835179e-06,
      "loss": 0.0233,
      "step": 639
    },
    {
      "epoch": 1.7643764807236701,
      "grad_norm": 0.09785618633031845,
      "learning_rate": 7.621329887591519e-06,
      "loss": 0.0236,
      "step": 640
    },
    {
      "epoch": 1.7671333189748006,
      "grad_norm": 0.10391392558813095,
      "learning_rate": 7.592362152079684e-06,
      "loss": 0.0228,
      "step": 641
    },
    {
      "epoch": 1.7698901572259316,
      "grad_norm": 0.33653295040130615,
      "learning_rate": 7.563415847143782e-06,
      "loss": 0.0281,
      "step": 642
    },
    {
      "epoch": 1.7726469954770623,
      "grad_norm": 0.12001726031303406,
      "learning_rate": 7.5344912304371785e-06,
      "loss": 0.0229,
      "step": 643
    },
    {
      "epoch": 1.775403833728193,
      "grad_norm": 0.12422462552785873,
      "learning_rate": 7.505588559420188e-06,
      "loss": 0.0238,
      "step": 644
    },
    {
      "epoch": 1.7781606719793237,
      "grad_norm": 0.09668347984552383,
      "learning_rate": 7.476708091357783e-06,
      "loss": 0.0224,
      "step": 645
    },
    {
      "epoch": 1.7809175102304544,
      "grad_norm": 0.11820239573717117,
      "learning_rate": 7.447850083317308e-06,
      "loss": 0.023,
      "step": 646
    },
    {
      "epoch": 1.7836743484815853,
      "grad_norm": 0.10456949472427368,
      "learning_rate": 7.419014792166182e-06,
      "loss": 0.0239,
      "step": 647
    },
    {
      "epoch": 1.7864311867327158,
      "grad_norm": 0.11468762159347534,
      "learning_rate": 7.39020247456963e-06,
      "loss": 0.0229,
      "step": 648
    },
    {
      "epoch": 1.7891880249838468,
      "grad_norm": 0.09829288721084595,
      "learning_rate": 7.361413386988379e-06,
      "loss": 0.0231,
      "step": 649
    },
    {
      "epoch": 1.7919448632349773,
      "grad_norm": 0.10345254838466644,
      "learning_rate": 7.332647785676388e-06,
      "loss": 0.0231,
      "step": 650
    },
    {
      "epoch": 1.7947017014861082,
      "grad_norm": 0.09892752021551132,
      "learning_rate": 7.303905926678565e-06,
      "loss": 0.0234,
      "step": 651
    },
    {
      "epoch": 1.797458539737239,
      "grad_norm": 0.09165947884321213,
      "learning_rate": 7.275188065828476e-06,
      "loss": 0.022,
      "step": 652
    },
    {
      "epoch": 1.8002153779883696,
      "grad_norm": 0.09468812495470047,
      "learning_rate": 7.246494458746085e-06,
      "loss": 0.0223,
      "step": 653
    },
    {
      "epoch": 1.8029722162395003,
      "grad_norm": 0.1075858324766159,
      "learning_rate": 7.217825360835475e-06,
      "loss": 0.0233,
      "step": 654
    },
    {
      "epoch": 1.805729054490631,
      "grad_norm": 0.10927974432706833,
      "learning_rate": 7.189181027282561e-06,
      "loss": 0.0232,
      "step": 655
    },
    {
      "epoch": 1.8084858927417617,
      "grad_norm": 0.10256339609622955,
      "learning_rate": 7.160561713052843e-06,
      "loss": 0.0237,
      "step": 656
    },
    {
      "epoch": 1.8112427309928925,
      "grad_norm": 0.11086180061101913,
      "learning_rate": 7.131967672889101e-06,
      "loss": 0.0234,
      "step": 657
    },
    {
      "epoch": 1.8139995692440234,
      "grad_norm": 0.10017542541027069,
      "learning_rate": 7.103399161309165e-06,
      "loss": 0.0225,
      "step": 658
    },
    {
      "epoch": 1.8167564074951539,
      "grad_norm": 0.10286585986614227,
      "learning_rate": 7.074856432603628e-06,
      "loss": 0.0225,
      "step": 659
    },
    {
      "epoch": 1.8195132457462848,
      "grad_norm": 0.10518784821033478,
      "learning_rate": 7.04633974083359e-06,
      "loss": 0.0234,
      "step": 660
    },
    {
      "epoch": 1.8222700839974153,
      "grad_norm": 0.10116968303918839,
      "learning_rate": 7.017849339828388e-06,
      "loss": 0.0233,
      "step": 661
    },
    {
      "epoch": 1.8250269222485462,
      "grad_norm": 0.10069679468870163,
      "learning_rate": 6.989385483183355e-06,
      "loss": 0.0234,
      "step": 662
    },
    {
      "epoch": 1.827783760499677,
      "grad_norm": 0.09656015038490295,
      "learning_rate": 6.960948424257532e-06,
      "loss": 0.0225,
      "step": 663
    },
    {
      "epoch": 1.8305405987508077,
      "grad_norm": 0.10670676827430725,
      "learning_rate": 6.9325384161714485e-06,
      "loss": 0.0231,
      "step": 664
    },
    {
      "epoch": 1.8332974370019384,
      "grad_norm": 0.12347907572984695,
      "learning_rate": 6.904155711804843e-06,
      "loss": 0.0235,
      "step": 665
    },
    {
      "epoch": 1.836054275253069,
      "grad_norm": 0.09916937351226807,
      "learning_rate": 6.8758005637944245e-06,
      "loss": 0.0229,
      "step": 666
    },
    {
      "epoch": 1.8388111135042,
      "grad_norm": 0.09796298295259476,
      "learning_rate": 6.8474732245316245e-06,
      "loss": 0.0234,
      "step": 667
    },
    {
      "epoch": 1.8415679517553305,
      "grad_norm": 0.1063319593667984,
      "learning_rate": 6.819173946160336e-06,
      "loss": 0.0232,
      "step": 668
    },
    {
      "epoch": 1.8443247900064614,
      "grad_norm": 0.09941181540489197,
      "learning_rate": 6.7909029805746855e-06,
      "loss": 0.0234,
      "step": 669
    },
    {
      "epoch": 1.847081628257592,
      "grad_norm": 0.09763844311237335,
      "learning_rate": 6.762660579416791e-06,
      "loss": 0.0237,
      "step": 670
    },
    {
      "epoch": 1.8498384665087229,
      "grad_norm": 0.094894640147686,
      "learning_rate": 6.734446994074507e-06,
      "loss": 0.0216,
      "step": 671
    },
    {
      "epoch": 1.8525953047598536,
      "grad_norm": 0.10713546723127365,
      "learning_rate": 6.706262475679205e-06,
      "loss": 0.0232,
      "step": 672
    },
    {
      "epoch": 1.8553521430109843,
      "grad_norm": 0.09470642358064651,
      "learning_rate": 6.678107275103519e-06,
      "loss": 0.0231,
      "step": 673
    },
    {
      "epoch": 1.858108981262115,
      "grad_norm": 0.09599259495735168,
      "learning_rate": 6.649981642959133e-06,
      "loss": 0.023,
      "step": 674
    },
    {
      "epoch": 1.8608658195132457,
      "grad_norm": 0.09776968508958817,
      "learning_rate": 6.62188582959453e-06,
      "loss": 0.0224,
      "step": 675
    },
    {
      "epoch": 1.8636226577643766,
      "grad_norm": 0.09774177521467209,
      "learning_rate": 6.593820085092782e-06,
      "loss": 0.0222,
      "step": 676
    },
    {
      "epoch": 1.8663794960155071,
      "grad_norm": 0.10560328513383865,
      "learning_rate": 6.565784659269314e-06,
      "loss": 0.0233,
      "step": 677
    },
    {
      "epoch": 1.869136334266638,
      "grad_norm": 0.09616672247648239,
      "learning_rate": 6.537779801669677e-06,
      "loss": 0.023,
      "step": 678
    },
    {
      "epoch": 1.8718931725177685,
      "grad_norm": 0.10657691955566406,
      "learning_rate": 6.509805761567336e-06,
      "loss": 0.0232,
      "step": 679
    },
    {
      "epoch": 1.8746500107688995,
      "grad_norm": 0.09221483767032623,
      "learning_rate": 6.481862787961448e-06,
      "loss": 0.0227,
      "step": 680
    },
    {
      "epoch": 1.8774068490200302,
      "grad_norm": 0.08830998837947845,
      "learning_rate": 6.453951129574644e-06,
      "loss": 0.0221,
      "step": 681
    },
    {
      "epoch": 1.880163687271161,
      "grad_norm": 0.08807796984910965,
      "learning_rate": 6.4260710348508115e-06,
      "loss": 0.0224,
      "step": 682
    },
    {
      "epoch": 1.8829205255222916,
      "grad_norm": 0.08681437373161316,
      "learning_rate": 6.3982227519528986e-06,
      "loss": 0.0216,
      "step": 683
    },
    {
      "epoch": 1.8856773637734223,
      "grad_norm": 0.09876042604446411,
      "learning_rate": 6.370406528760675e-06,
      "loss": 0.0228,
      "step": 684
    },
    {
      "epoch": 1.8884342020245533,
      "grad_norm": 0.10514353215694427,
      "learning_rate": 6.34262261286856e-06,
      "loss": 0.0219,
      "step": 685
    },
    {
      "epoch": 1.8911910402756837,
      "grad_norm": 0.0905836671590805,
      "learning_rate": 6.3148712515833985e-06,
      "loss": 0.022,
      "step": 686
    },
    {
      "epoch": 1.8939478785268147,
      "grad_norm": 0.09052827209234238,
      "learning_rate": 6.287152691922264e-06,
      "loss": 0.0216,
      "step": 687
    },
    {
      "epoch": 1.8967047167779452,
      "grad_norm": 0.101600281894207,
      "learning_rate": 6.259467180610262e-06,
      "loss": 0.0233,
      "step": 688
    },
    {
      "epoch": 1.899461555029076,
      "grad_norm": 0.10384192317724228,
      "learning_rate": 6.231814964078327e-06,
      "loss": 0.0233,
      "step": 689
    },
    {
      "epoch": 1.9022183932802068,
      "grad_norm": 0.1043287143111229,
      "learning_rate": 6.204196288461037e-06,
      "loss": 0.0224,
      "step": 690
    },
    {
      "epoch": 1.9049752315313375,
      "grad_norm": 0.41572827100753784,
      "learning_rate": 6.176611399594422e-06,
      "loss": 0.0241,
      "step": 691
    },
    {
      "epoch": 1.9077320697824682,
      "grad_norm": 0.0979122593998909,
      "learning_rate": 6.149060543013772e-06,
      "loss": 0.0222,
      "step": 692
    },
    {
      "epoch": 1.910488908033599,
      "grad_norm": 0.09667398780584335,
      "learning_rate": 6.121543963951453e-06,
      "loss": 0.0232,
      "step": 693
    },
    {
      "epoch": 1.9132457462847297,
      "grad_norm": 0.10258089005947113,
      "learning_rate": 6.094061907334718e-06,
      "loss": 0.023,
      "step": 694
    },
    {
      "epoch": 1.9160025845358604,
      "grad_norm": 0.09479233622550964,
      "learning_rate": 6.066614617783542e-06,
      "loss": 0.0225,
      "step": 695
    },
    {
      "epoch": 1.9187594227869913,
      "grad_norm": 0.09578879177570343,
      "learning_rate": 6.039202339608432e-06,
      "loss": 0.0226,
      "step": 696
    },
    {
      "epoch": 1.9215162610381218,
      "grad_norm": 0.09669813513755798,
      "learning_rate": 6.0118253168082555e-06,
      "loss": 0.0228,
      "step": 697
    },
    {
      "epoch": 1.9242730992892527,
      "grad_norm": 0.09967345744371414,
      "learning_rate": 5.984483793068072e-06,
      "loss": 0.023,
      "step": 698
    },
    {
      "epoch": 1.9270299375403832,
      "grad_norm": 0.09882698953151703,
      "learning_rate": 5.957178011756952e-06,
      "loss": 0.0239,
      "step": 699
    },
    {
      "epoch": 1.9297867757915141,
      "grad_norm": 0.10672547668218613,
      "learning_rate": 5.92990821592583e-06,
      "loss": 0.0235,
      "step": 700
    },
    {
      "epoch": 1.9325436140426449,
      "grad_norm": 0.10070157796144485,
      "learning_rate": 5.902674648305329e-06,
      "loss": 0.0223,
      "step": 701
    },
    {
      "epoch": 1.9353004522937756,
      "grad_norm": 0.1068674623966217,
      "learning_rate": 5.875477551303596e-06,
      "loss": 0.0236,
      "step": 702
    },
    {
      "epoch": 1.9380572905449063,
      "grad_norm": 0.1479007601737976,
      "learning_rate": 5.848317167004159e-06,
      "loss": 0.024,
      "step": 703
    },
    {
      "epoch": 1.940814128796037,
      "grad_norm": 0.09680446237325668,
      "learning_rate": 5.8211937371637525e-06,
      "loss": 0.0224,
      "step": 704
    },
    {
      "epoch": 1.943570967047168,
      "grad_norm": 0.09958844631910324,
      "learning_rate": 5.794107503210187e-06,
      "loss": 0.0228,
      "step": 705
    },
    {
      "epoch": 1.9463278052982984,
      "grad_norm": 0.10216110944747925,
      "learning_rate": 5.767058706240183e-06,
      "loss": 0.0217,
      "step": 706
    },
    {
      "epoch": 1.9490846435494293,
      "grad_norm": 0.09057790040969849,
      "learning_rate": 5.740047587017232e-06,
      "loss": 0.0225,
      "step": 707
    },
    {
      "epoch": 1.9518414818005598,
      "grad_norm": 0.10457273572683334,
      "learning_rate": 5.713074385969457e-06,
      "loss": 0.0222,
      "step": 708
    },
    {
      "epoch": 1.9545983200516908,
      "grad_norm": 0.1035657674074173,
      "learning_rate": 5.686139343187468e-06,
      "loss": 0.0229,
      "step": 709
    },
    {
      "epoch": 1.9573551583028215,
      "grad_norm": 0.1085587665438652,
      "learning_rate": 5.659242698422213e-06,
      "loss": 0.0222,
      "step": 710
    },
    {
      "epoch": 1.9601119965539522,
      "grad_norm": 0.10512305051088333,
      "learning_rate": 5.632384691082874e-06,
      "loss": 0.023,
      "step": 711
    },
    {
      "epoch": 1.962868834805083,
      "grad_norm": 0.1115739643573761,
      "learning_rate": 5.605565560234707e-06,
      "loss": 0.0228,
      "step": 712
    },
    {
      "epoch": 1.9656256730562136,
      "grad_norm": 0.11410468816757202,
      "learning_rate": 5.578785544596928e-06,
      "loss": 0.0234,
      "step": 713
    },
    {
      "epoch": 1.9683825113073445,
      "grad_norm": 0.10378382354974747,
      "learning_rate": 5.55204488254059e-06,
      "loss": 0.0242,
      "step": 714
    },
    {
      "epoch": 1.971139349558475,
      "grad_norm": 0.11169740557670593,
      "learning_rate": 5.525343812086445e-06,
      "loss": 0.0227,
      "step": 715
    },
    {
      "epoch": 1.973896187809606,
      "grad_norm": 0.11073627322912216,
      "learning_rate": 5.498682570902849e-06,
      "loss": 0.0227,
      "step": 716
    },
    {
      "epoch": 1.9766530260607365,
      "grad_norm": 0.10393860191106796,
      "learning_rate": 5.47206139630363e-06,
      "loss": 0.0231,
      "step": 717
    },
    {
      "epoch": 1.9794098643118674,
      "grad_norm": 0.09541879594326019,
      "learning_rate": 5.445480525245976e-06,
      "loss": 0.0235,
      "step": 718
    },
    {
      "epoch": 1.982166702562998,
      "grad_norm": 0.10563753545284271,
      "learning_rate": 5.418940194328344e-06,
      "loss": 0.0232,
      "step": 719
    },
    {
      "epoch": 1.9849235408141288,
      "grad_norm": 0.09059736877679825,
      "learning_rate": 5.3924406397883174e-06,
      "loss": 0.0221,
      "step": 720
    },
    {
      "epoch": 1.9876803790652595,
      "grad_norm": 0.0970776304602623,
      "learning_rate": 5.365982097500545e-06,
      "loss": 0.0219,
      "step": 721
    },
    {
      "epoch": 1.9904372173163902,
      "grad_norm": 0.10448465496301651,
      "learning_rate": 5.339564802974615e-06,
      "loss": 0.0218,
      "step": 722
    },
    {
      "epoch": 1.9931940555675212,
      "grad_norm": 0.11694996803998947,
      "learning_rate": 5.313188991352964e-06,
      "loss": 0.0223,
      "step": 723
    },
    {
      "epoch": 1.9959508938186517,
      "grad_norm": 0.10104145109653473,
      "learning_rate": 5.286854897408793e-06,
      "loss": 0.0229,
      "step": 724
    },
    {
      "epoch": 1.9987077320697826,
      "grad_norm": 0.10007411241531372,
      "learning_rate": 5.2605627555439635e-06,
      "loss": 0.0229,
      "step": 725
    },
    {
      "epoch": 2.001464570320913,
      "grad_norm": 0.10868417471647263,
      "learning_rate": 5.234312799786921e-06,
      "loss": 0.0199,
      "step": 726
    },
    {
      "epoch": 2.004221408572044,
      "grad_norm": 0.1075412854552269,
      "learning_rate": 5.208105263790611e-06,
      "loss": 0.0169,
      "step": 727
    },
    {
      "epoch": 2.0069782468231745,
      "grad_norm": 0.09774283319711685,
      "learning_rate": 5.181940380830393e-06,
      "loss": 0.0166,
      "step": 728
    },
    {
      "epoch": 2.0097350850743054,
      "grad_norm": 0.09378799796104431,
      "learning_rate": 5.155818383801976e-06,
      "loss": 0.0169,
      "step": 729
    },
    {
      "epoch": 2.0124919233254364,
      "grad_norm": 0.09300612658262253,
      "learning_rate": 5.129739505219325e-06,
      "loss": 0.0175,
      "step": 730
    },
    {
      "epoch": 2.015248761576567,
      "grad_norm": 0.10933335870504379,
      "learning_rate": 5.103703977212615e-06,
      "loss": 0.0172,
      "step": 731
    },
    {
      "epoch": 2.018005599827698,
      "grad_norm": 0.10995355993509293,
      "learning_rate": 5.077712031526153e-06,
      "loss": 0.0165,
      "step": 732
    },
    {
      "epoch": 2.0207624380788283,
      "grad_norm": 0.0987633690237999,
      "learning_rate": 5.051763899516313e-06,
      "loss": 0.0169,
      "step": 733
    },
    {
      "epoch": 2.023519276329959,
      "grad_norm": 0.10309556126594543,
      "learning_rate": 5.025859812149481e-06,
      "loss": 0.0166,
      "step": 734
    },
    {
      "epoch": 2.0262761145810897,
      "grad_norm": 0.10243550688028336,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.0176,
      "step": 735
    },
    {
      "epoch": 2.0290329528322206,
      "grad_norm": 0.10226277261972427,
      "learning_rate": 4.9741846932481154e-06,
      "loss": 0.0163,
      "step": 736
    },
    {
      "epoch": 2.031789791083351,
      "grad_norm": 0.10129394382238388,
      "learning_rate": 4.94841412167792e-06,
      "loss": 0.0167,
      "step": 737
    },
    {
      "epoch": 2.034546629334482,
      "grad_norm": 0.09485170245170593,
      "learning_rate": 4.922688514675325e-06,
      "loss": 0.0167,
      "step": 738
    },
    {
      "epoch": 2.0373034675856125,
      "grad_norm": 0.10152100771665573,
      "learning_rate": 4.8970081012260014e-06,
      "loss": 0.0169,
      "step": 739
    },
    {
      "epoch": 2.0400603058367435,
      "grad_norm": 0.10325178503990173,
      "learning_rate": 4.8713731099133576e-06,
      "loss": 0.0172,
      "step": 740
    },
    {
      "epoch": 2.0428171440878744,
      "grad_norm": 0.10417655855417252,
      "learning_rate": 4.845783768916482e-06,
      "loss": 0.0167,
      "step": 741
    },
    {
      "epoch": 2.045573982339005,
      "grad_norm": 0.0955713763833046,
      "learning_rate": 4.820240306008136e-06,
      "loss": 0.0164,
      "step": 742
    },
    {
      "epoch": 2.048330820590136,
      "grad_norm": 0.10096914321184158,
      "learning_rate": 4.794742948552716e-06,
      "loss": 0.0164,
      "step": 743
    },
    {
      "epoch": 2.0510876588412663,
      "grad_norm": 0.11180409044027328,
      "learning_rate": 4.769291923504226e-06,
      "loss": 0.0169,
      "step": 744
    },
    {
      "epoch": 2.0538444970923972,
      "grad_norm": 0.10208172351121902,
      "learning_rate": 4.743887457404268e-06,
      "loss": 0.0159,
      "step": 745
    },
    {
      "epoch": 2.0566013353435277,
      "grad_norm": 0.10103499889373779,
      "learning_rate": 4.718529776380009e-06,
      "loss": 0.0161,
      "step": 746
    },
    {
      "epoch": 2.0593581735946587,
      "grad_norm": 0.09129516035318375,
      "learning_rate": 4.693219106142186e-06,
      "loss": 0.0164,
      "step": 747
    },
    {
      "epoch": 2.062115011845789,
      "grad_norm": 0.10983917117118835,
      "learning_rate": 4.66795567198309e-06,
      "loss": 0.0165,
      "step": 748
    },
    {
      "epoch": 2.06487185009692,
      "grad_norm": 0.10270462185144424,
      "learning_rate": 4.642739698774555e-06,
      "loss": 0.0171,
      "step": 749
    },
    {
      "epoch": 2.067628688348051,
      "grad_norm": 0.10552225261926651,
      "learning_rate": 4.617571410965964e-06,
      "loss": 0.0167,
      "step": 750
    },
    {
      "epoch": 2.0703855265991815,
      "grad_norm": 0.10517138242721558,
      "learning_rate": 4.59245103258225e-06,
      "loss": 0.0164,
      "step": 751
    },
    {
      "epoch": 2.0731423648503124,
      "grad_norm": 0.09321381151676178,
      "learning_rate": 4.567378787221896e-06,
      "loss": 0.0163,
      "step": 752
    },
    {
      "epoch": 2.075899203101443,
      "grad_norm": 0.10226123034954071,
      "learning_rate": 4.542354898054953e-06,
      "loss": 0.0168,
      "step": 753
    },
    {
      "epoch": 2.078656041352574,
      "grad_norm": 0.09658444672822952,
      "learning_rate": 4.517379587821049e-06,
      "loss": 0.0161,
      "step": 754
    },
    {
      "epoch": 2.0814128796037044,
      "grad_norm": 0.08875004947185516,
      "learning_rate": 4.492453078827409e-06,
      "loss": 0.0158,
      "step": 755
    },
    {
      "epoch": 2.0841697178548353,
      "grad_norm": 0.10765478014945984,
      "learning_rate": 4.467575592946865e-06,
      "loss": 0.016,
      "step": 756
    },
    {
      "epoch": 2.086926556105966,
      "grad_norm": 0.09269033372402191,
      "learning_rate": 4.442747351615899e-06,
      "loss": 0.0156,
      "step": 757
    },
    {
      "epoch": 2.0896833943570967,
      "grad_norm": 0.099794402718544,
      "learning_rate": 4.417968575832664e-06,
      "loss": 0.0155,
      "step": 758
    },
    {
      "epoch": 2.0924402326082276,
      "grad_norm": 0.10920445621013641,
      "learning_rate": 4.393239486155011e-06,
      "loss": 0.0164,
      "step": 759
    },
    {
      "epoch": 2.095197070859358,
      "grad_norm": 0.10013429075479507,
      "learning_rate": 4.3685603026985356e-06,
      "loss": 0.0159,
      "step": 760
    },
    {
      "epoch": 2.097953909110489,
      "grad_norm": 0.1004144474864006,
      "learning_rate": 4.343931245134616e-06,
      "loss": 0.0171,
      "step": 761
    },
    {
      "epoch": 2.1007107473616196,
      "grad_norm": 0.11513813585042953,
      "learning_rate": 4.319352532688444e-06,
      "loss": 0.0167,
      "step": 762
    },
    {
      "epoch": 2.1034675856127505,
      "grad_norm": 0.10025149583816528,
      "learning_rate": 4.294824384137096e-06,
      "loss": 0.0167,
      "step": 763
    },
    {
      "epoch": 2.106224423863881,
      "grad_norm": 0.13849525153636932,
      "learning_rate": 4.270347017807575e-06,
      "loss": 0.0173,
      "step": 764
    },
    {
      "epoch": 2.108981262115012,
      "grad_norm": 0.10839790850877762,
      "learning_rate": 4.245920651574864e-06,
      "loss": 0.0164,
      "step": 765
    },
    {
      "epoch": 2.1117381003661424,
      "grad_norm": 0.09577582031488419,
      "learning_rate": 4.221545502859994e-06,
      "loss": 0.0158,
      "step": 766
    },
    {
      "epoch": 2.1144949386172733,
      "grad_norm": 0.1021181270480156,
      "learning_rate": 4.197221788628096e-06,
      "loss": 0.0163,
      "step": 767
    },
    {
      "epoch": 2.117251776868404,
      "grad_norm": 0.10877656936645508,
      "learning_rate": 4.172949725386488e-06,
      "loss": 0.0159,
      "step": 768
    },
    {
      "epoch": 2.1200086151195348,
      "grad_norm": 0.09782122820615768,
      "learning_rate": 4.148729529182736e-06,
      "loss": 0.0167,
      "step": 769
    },
    {
      "epoch": 2.1227654533706657,
      "grad_norm": 0.10682158172130585,
      "learning_rate": 4.124561415602729e-06,
      "loss": 0.0171,
      "step": 770
    },
    {
      "epoch": 2.125522291621796,
      "grad_norm": 0.10475680232048035,
      "learning_rate": 4.100445599768774e-06,
      "loss": 0.0164,
      "step": 771
    },
    {
      "epoch": 2.128279129872927,
      "grad_norm": 0.09578834474086761,
      "learning_rate": 4.0763822963376585e-06,
      "loss": 0.0158,
      "step": 772
    },
    {
      "epoch": 2.1310359681240576,
      "grad_norm": 0.09797288477420807,
      "learning_rate": 4.0523717194987634e-06,
      "loss": 0.0158,
      "step": 773
    },
    {
      "epoch": 2.1337928063751885,
      "grad_norm": 0.1016959697008133,
      "learning_rate": 4.028414082972141e-06,
      "loss": 0.0168,
      "step": 774
    },
    {
      "epoch": 2.136549644626319,
      "grad_norm": 0.09790080785751343,
      "learning_rate": 4.004509600006618e-06,
      "loss": 0.0155,
      "step": 775
    },
    {
      "epoch": 2.13930648287745,
      "grad_norm": 0.1030542254447937,
      "learning_rate": 3.980658483377903e-06,
      "loss": 0.0166,
      "step": 776
    },
    {
      "epoch": 2.142063321128581,
      "grad_norm": 0.11205972731113434,
      "learning_rate": 3.956860945386677e-06,
      "loss": 0.0158,
      "step": 777
    },
    {
      "epoch": 2.1448201593797114,
      "grad_norm": 0.09531433135271072,
      "learning_rate": 3.9331171978567206e-06,
      "loss": 0.0154,
      "step": 778
    },
    {
      "epoch": 2.1475769976308423,
      "grad_norm": 0.09850120544433594,
      "learning_rate": 3.909427452133017e-06,
      "loss": 0.0162,
      "step": 779
    },
    {
      "epoch": 2.150333835881973,
      "grad_norm": 0.11114904284477234,
      "learning_rate": 3.885791919079878e-06,
      "loss": 0.016,
      "step": 780
    },
    {
      "epoch": 2.1530906741331037,
      "grad_norm": 0.10965460538864136,
      "learning_rate": 3.862210809079061e-06,
      "loss": 0.0156,
      "step": 781
    },
    {
      "epoch": 2.155847512384234,
      "grad_norm": 0.09533203393220901,
      "learning_rate": 3.838684332027908e-06,
      "loss": 0.0163,
      "step": 782
    },
    {
      "epoch": 2.158604350635365,
      "grad_norm": 0.09494911879301071,
      "learning_rate": 3.815212697337451e-06,
      "loss": 0.0158,
      "step": 783
    },
    {
      "epoch": 2.1613611888864956,
      "grad_norm": 0.10040943324565887,
      "learning_rate": 3.7917961139305835e-06,
      "loss": 0.0162,
      "step": 784
    },
    {
      "epoch": 2.1641180271376266,
      "grad_norm": 0.09985315054655075,
      "learning_rate": 3.7684347902401753e-06,
      "loss": 0.0164,
      "step": 785
    },
    {
      "epoch": 2.166874865388757,
      "grad_norm": 0.09866054356098175,
      "learning_rate": 3.745128934207225e-06,
      "loss": 0.0161,
      "step": 786
    },
    {
      "epoch": 2.169631703639888,
      "grad_norm": 0.09665759652853012,
      "learning_rate": 3.7218787532790167e-06,
      "loss": 0.0163,
      "step": 787
    },
    {
      "epoch": 2.172388541891019,
      "grad_norm": 0.09830871224403381,
      "learning_rate": 3.6986844544072496e-06,
      "loss": 0.0165,
      "step": 788
    },
    {
      "epoch": 2.1751453801421494,
      "grad_norm": 0.10809045284986496,
      "learning_rate": 3.6755462440462288e-06,
      "loss": 0.0167,
      "step": 789
    },
    {
      "epoch": 2.1779022183932804,
      "grad_norm": 0.09549186378717422,
      "learning_rate": 3.6524643281510018e-06,
      "loss": 0.0164,
      "step": 790
    },
    {
      "epoch": 2.180659056644411,
      "grad_norm": 0.09945723414421082,
      "learning_rate": 3.6294389121755404e-06,
      "loss": 0.0159,
      "step": 791
    },
    {
      "epoch": 2.1834158948955418,
      "grad_norm": 0.09856431931257248,
      "learning_rate": 3.606470201070904e-06,
      "loss": 0.0164,
      "step": 792
    },
    {
      "epoch": 2.1861727331466723,
      "grad_norm": 0.09308286756277084,
      "learning_rate": 3.58355839928341e-06,
      "loss": 0.0155,
      "step": 793
    },
    {
      "epoch": 2.188929571397803,
      "grad_norm": 0.10592252761125565,
      "learning_rate": 3.560703710752833e-06,
      "loss": 0.0167,
      "step": 794
    },
    {
      "epoch": 2.1916864096489337,
      "grad_norm": 0.10037653893232346,
      "learning_rate": 3.5379063389105727e-06,
      "loss": 0.0167,
      "step": 795
    },
    {
      "epoch": 2.1944432479000646,
      "grad_norm": 0.1029176115989685,
      "learning_rate": 3.515166486677848e-06,
      "loss": 0.0173,
      "step": 796
    },
    {
      "epoch": 2.1972000861511956,
      "grad_norm": 0.10505726933479309,
      "learning_rate": 3.4924843564638946e-06,
      "loss": 0.0164,
      "step": 797
    },
    {
      "epoch": 2.199956924402326,
      "grad_norm": 0.12057758867740631,
      "learning_rate": 3.4698601501641517e-06,
      "loss": 0.0168,
      "step": 798
    },
    {
      "epoch": 2.202713762653457,
      "grad_norm": 0.09361850470304489,
      "learning_rate": 3.447294069158481e-06,
      "loss": 0.0155,
      "step": 799
    },
    {
      "epoch": 2.2054706009045875,
      "grad_norm": 0.10088173300027847,
      "learning_rate": 3.424786314309365e-06,
      "loss": 0.0157,
      "step": 800
    },
    {
      "epoch": 2.2082274391557184,
      "grad_norm": 0.09545209258794785,
      "learning_rate": 3.4023370859601192e-06,
      "loss": 0.016,
      "step": 801
    },
    {
      "epoch": 2.210984277406849,
      "grad_norm": 0.10133402049541473,
      "learning_rate": 3.3799465839331103e-06,
      "loss": 0.0167,
      "step": 802
    },
    {
      "epoch": 2.21374111565798,
      "grad_norm": 0.10037866234779358,
      "learning_rate": 3.3576150075279757e-06,
      "loss": 0.0166,
      "step": 803
    },
    {
      "epoch": 2.2164979539091103,
      "grad_norm": 0.13703982532024384,
      "learning_rate": 3.335342555519855e-06,
      "loss": 0.0167,
      "step": 804
    },
    {
      "epoch": 2.2192547921602412,
      "grad_norm": 0.10422453284263611,
      "learning_rate": 3.313129426157613e-06,
      "loss": 0.016,
      "step": 805
    },
    {
      "epoch": 2.222011630411372,
      "grad_norm": 0.09957727044820786,
      "learning_rate": 3.290975817162082e-06,
      "loss": 0.0156,
      "step": 806
    },
    {
      "epoch": 2.2247684686625027,
      "grad_norm": 0.0947730764746666,
      "learning_rate": 3.2688819257242963e-06,
      "loss": 0.0157,
      "step": 807
    },
    {
      "epoch": 2.2275253069136336,
      "grad_norm": 0.10239588469266891,
      "learning_rate": 3.246847948503744e-06,
      "loss": 0.016,
      "step": 808
    },
    {
      "epoch": 2.230282145164764,
      "grad_norm": 0.10617274791002274,
      "learning_rate": 3.2248740816266012e-06,
      "loss": 0.0158,
      "step": 809
    },
    {
      "epoch": 2.233038983415895,
      "grad_norm": 0.10229507088661194,
      "learning_rate": 3.2029605206840088e-06,
      "loss": 0.0162,
      "step": 810
    },
    {
      "epoch": 2.2357958216670255,
      "grad_norm": 0.09557145833969116,
      "learning_rate": 3.181107460730314e-06,
      "loss": 0.0157,
      "step": 811
    },
    {
      "epoch": 2.2385526599181564,
      "grad_norm": 0.11345598846673965,
      "learning_rate": 3.1593150962813425e-06,
      "loss": 0.0167,
      "step": 812
    },
    {
      "epoch": 2.241309498169287,
      "grad_norm": 0.11662036180496216,
      "learning_rate": 3.1375836213126653e-06,
      "loss": 0.0168,
      "step": 813
    },
    {
      "epoch": 2.244066336420418,
      "grad_norm": 0.10055527836084366,
      "learning_rate": 3.115913229257864e-06,
      "loss": 0.0165,
      "step": 814
    },
    {
      "epoch": 2.2468231746715484,
      "grad_norm": 0.10372263938188553,
      "learning_rate": 3.0943041130068243e-06,
      "loss": 0.0165,
      "step": 815
    },
    {
      "epoch": 2.2495800129226793,
      "grad_norm": 0.10061642527580261,
      "learning_rate": 3.0727564649040066e-06,
      "loss": 0.0157,
      "step": 816
    },
    {
      "epoch": 2.25233685117381,
      "grad_norm": 0.10070322453975677,
      "learning_rate": 3.0512704767467417e-06,
      "loss": 0.016,
      "step": 817
    },
    {
      "epoch": 2.2550936894249407,
      "grad_norm": 0.09765329211950302,
      "learning_rate": 3.0298463397835223e-06,
      "loss": 0.0158,
      "step": 818
    },
    {
      "epoch": 2.2578505276760716,
      "grad_norm": 0.09247851371765137,
      "learning_rate": 3.008484244712286e-06,
      "loss": 0.0161,
      "step": 819
    },
    {
      "epoch": 2.260607365927202,
      "grad_norm": 0.21034474670886993,
      "learning_rate": 2.987184381678747e-06,
      "loss": 0.0165,
      "step": 820
    },
    {
      "epoch": 2.263364204178333,
      "grad_norm": 0.1013868898153305,
      "learning_rate": 2.965946940274678e-06,
      "loss": 0.0164,
      "step": 821
    },
    {
      "epoch": 2.2661210424294636,
      "grad_norm": 0.10465855151414871,
      "learning_rate": 2.9447721095362325e-06,
      "loss": 0.0162,
      "step": 822
    },
    {
      "epoch": 2.2688778806805945,
      "grad_norm": 0.10128065943717957,
      "learning_rate": 2.9236600779422674e-06,
      "loss": 0.0162,
      "step": 823
    },
    {
      "epoch": 2.2716347189317254,
      "grad_norm": 0.1028231531381607,
      "learning_rate": 2.902611033412648e-06,
      "loss": 0.0167,
      "step": 824
    },
    {
      "epoch": 2.274391557182856,
      "grad_norm": 0.09928223490715027,
      "learning_rate": 2.8816251633065963e-06,
      "loss": 0.0163,
      "step": 825
    },
    {
      "epoch": 2.277148395433987,
      "grad_norm": 0.1016731932759285,
      "learning_rate": 2.8607026544210115e-06,
      "loss": 0.0156,
      "step": 826
    },
    {
      "epoch": 2.2799052336851173,
      "grad_norm": 0.11646661907434464,
      "learning_rate": 2.8398436929888085e-06,
      "loss": 0.0165,
      "step": 827
    },
    {
      "epoch": 2.2826620719362483,
      "grad_norm": 0.1035120040178299,
      "learning_rate": 2.819048464677261e-06,
      "loss": 0.0163,
      "step": 828
    },
    {
      "epoch": 2.2854189101873787,
      "grad_norm": 0.10418447107076645,
      "learning_rate": 2.798317154586352e-06,
      "loss": 0.016,
      "step": 829
    },
    {
      "epoch": 2.2881757484385097,
      "grad_norm": 0.095628522336483,
      "learning_rate": 2.7776499472471184e-06,
      "loss": 0.0155,
      "step": 830
    },
    {
      "epoch": 2.29093258668964,
      "grad_norm": 0.09392069280147552,
      "learning_rate": 2.7570470266200177e-06,
      "loss": 0.0155,
      "step": 831
    },
    {
      "epoch": 2.293689424940771,
      "grad_norm": 0.10431677848100662,
      "learning_rate": 2.736508576093285e-06,
      "loss": 0.0163,
      "step": 832
    },
    {
      "epoch": 2.2964462631919016,
      "grad_norm": 0.1353488713502884,
      "learning_rate": 2.716034778481301e-06,
      "loss": 0.0164,
      "step": 833
    },
    {
      "epoch": 2.2992031014430325,
      "grad_norm": 0.100525863468647,
      "learning_rate": 2.69562581602297e-06,
      "loss": 0.0161,
      "step": 834
    },
    {
      "epoch": 2.3019599396941635,
      "grad_norm": 0.0952887013554573,
      "learning_rate": 2.675281870380082e-06,
      "loss": 0.0155,
      "step": 835
    },
    {
      "epoch": 2.304716777945294,
      "grad_norm": 0.13243414461612701,
      "learning_rate": 2.65500312263572e-06,
      "loss": 0.0159,
      "step": 836
    },
    {
      "epoch": 2.307473616196425,
      "grad_norm": 0.09811341762542725,
      "learning_rate": 2.6347897532926293e-06,
      "loss": 0.0157,
      "step": 837
    },
    {
      "epoch": 2.3102304544475554,
      "grad_norm": 0.10353419929742813,
      "learning_rate": 2.6146419422716174e-06,
      "loss": 0.0157,
      "step": 838
    },
    {
      "epoch": 2.3129872926986863,
      "grad_norm": 0.09999054670333862,
      "learning_rate": 2.594559868909956e-06,
      "loss": 0.0157,
      "step": 839
    },
    {
      "epoch": 2.315744130949817,
      "grad_norm": 0.10014519840478897,
      "learning_rate": 2.5745437119597704e-06,
      "loss": 0.0162,
      "step": 840
    },
    {
      "epoch": 2.3185009692009477,
      "grad_norm": 0.09995284676551819,
      "learning_rate": 2.554593649586469e-06,
      "loss": 0.0163,
      "step": 841
    },
    {
      "epoch": 2.321257807452078,
      "grad_norm": 0.10019443184137344,
      "learning_rate": 2.5347098593671417e-06,
      "loss": 0.0161,
      "step": 842
    },
    {
      "epoch": 2.324014645703209,
      "grad_norm": 0.09603513777256012,
      "learning_rate": 2.514892518288988e-06,
      "loss": 0.0162,
      "step": 843
    },
    {
      "epoch": 2.3267714839543396,
      "grad_norm": 0.09885770082473755,
      "learning_rate": 2.49514180274774e-06,
      "loss": 0.0158,
      "step": 844
    },
    {
      "epoch": 2.3295283222054706,
      "grad_norm": 0.09956613928079605,
      "learning_rate": 2.4754578885460813e-06,
      "loss": 0.0156,
      "step": 845
    },
    {
      "epoch": 2.3322851604566015,
      "grad_norm": 0.09634565562009811,
      "learning_rate": 2.455840950892099e-06,
      "loss": 0.0158,
      "step": 846
    },
    {
      "epoch": 2.335041998707732,
      "grad_norm": 0.09536509960889816,
      "learning_rate": 2.436291164397715e-06,
      "loss": 0.0161,
      "step": 847
    },
    {
      "epoch": 2.337798836958863,
      "grad_norm": 0.09601006656885147,
      "learning_rate": 2.416808703077135e-06,
      "loss": 0.0167,
      "step": 848
    },
    {
      "epoch": 2.3405556752099934,
      "grad_norm": 0.10354162007570267,
      "learning_rate": 2.3973937403452983e-06,
      "loss": 0.0159,
      "step": 849
    },
    {
      "epoch": 2.3433125134611243,
      "grad_norm": 0.10057395696640015,
      "learning_rate": 2.3780464490163267e-06,
      "loss": 0.0163,
      "step": 850
    },
    {
      "epoch": 2.346069351712255,
      "grad_norm": 0.09877093136310577,
      "learning_rate": 2.3587670013020026e-06,
      "loss": 0.0159,
      "step": 851
    },
    {
      "epoch": 2.3488261899633858,
      "grad_norm": 0.11428674310445786,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.0162,
      "step": 852
    },
    {
      "epoch": 2.3515830282145167,
      "grad_norm": 0.10734842717647552,
      "learning_rate": 2.3204123225434714e-06,
      "loss": 0.0158,
      "step": 853
    },
    {
      "epoch": 2.354339866465647,
      "grad_norm": 0.09699169546365738,
      "learning_rate": 2.3013374328973113e-06,
      "loss": 0.0161,
      "step": 854
    },
    {
      "epoch": 2.357096704716778,
      "grad_norm": 0.09945055842399597,
      "learning_rate": 2.28233106965885e-06,
      "loss": 0.0164,
      "step": 855
    },
    {
      "epoch": 2.3598535429679086,
      "grad_norm": 0.09800931066274643,
      "learning_rate": 2.2633934020052383e-06,
      "loss": 0.0155,
      "step": 856
    },
    {
      "epoch": 2.3626103812190395,
      "grad_norm": 0.09882977604866028,
      "learning_rate": 2.2445245985021613e-06,
      "loss": 0.0154,
      "step": 857
    },
    {
      "epoch": 2.36536721947017,
      "grad_norm": 0.0957275852560997,
      "learning_rate": 2.2257248271023424e-06,
      "loss": 0.0164,
      "step": 858
    },
    {
      "epoch": 2.368124057721301,
      "grad_norm": 0.10268919914960861,
      "learning_rate": 2.206994255144036e-06,
      "loss": 0.0164,
      "step": 859
    },
    {
      "epoch": 2.3708808959724315,
      "grad_norm": 0.10079904645681381,
      "learning_rate": 2.188333049349556e-06,
      "loss": 0.0157,
      "step": 860
    },
    {
      "epoch": 2.3736377342235624,
      "grad_norm": 0.0992964580655098,
      "learning_rate": 2.1697413758237785e-06,
      "loss": 0.0158,
      "step": 861
    },
    {
      "epoch": 2.376394572474693,
      "grad_norm": 0.10243627429008484,
      "learning_rate": 2.1512194000526676e-06,
      "loss": 0.0167,
      "step": 862
    },
    {
      "epoch": 2.379151410725824,
      "grad_norm": 0.0987883061170578,
      "learning_rate": 2.1327672869018036e-06,
      "loss": 0.0163,
      "step": 863
    },
    {
      "epoch": 2.3819082489769547,
      "grad_norm": 0.10566847771406174,
      "learning_rate": 2.114385200614912e-06,
      "loss": 0.0157,
      "step": 864
    },
    {
      "epoch": 2.3846650872280852,
      "grad_norm": 0.09772182255983353,
      "learning_rate": 2.0960733048124082e-06,
      "loss": 0.0153,
      "step": 865
    },
    {
      "epoch": 2.387421925479216,
      "grad_norm": 0.11045119166374207,
      "learning_rate": 2.077831762489927e-06,
      "loss": 0.0165,
      "step": 866
    },
    {
      "epoch": 2.3901787637303467,
      "grad_norm": 0.1000387892127037,
      "learning_rate": 2.0596607360168897e-06,
      "loss": 0.0163,
      "step": 867
    },
    {
      "epoch": 2.3929356019814776,
      "grad_norm": 0.09834710508584976,
      "learning_rate": 2.0415603871350476e-06,
      "loss": 0.0162,
      "step": 868
    },
    {
      "epoch": 2.395692440232608,
      "grad_norm": 0.10450678318738937,
      "learning_rate": 2.023530876957045e-06,
      "loss": 0.0155,
      "step": 869
    },
    {
      "epoch": 2.398449278483739,
      "grad_norm": 0.11825753003358841,
      "learning_rate": 2.0055723659649907e-06,
      "loss": 0.0157,
      "step": 870
    },
    {
      "epoch": 2.40120611673487,
      "grad_norm": 0.09847236424684525,
      "learning_rate": 1.987685014009011e-06,
      "loss": 0.0158,
      "step": 871
    },
    {
      "epoch": 2.4039629549860004,
      "grad_norm": 0.09447720646858215,
      "learning_rate": 1.9698689803058523e-06,
      "loss": 0.0159,
      "step": 872
    },
    {
      "epoch": 2.406719793237131,
      "grad_norm": 0.09634242951869965,
      "learning_rate": 1.952124423437447e-06,
      "loss": 0.016,
      "step": 873
    },
    {
      "epoch": 2.409476631488262,
      "grad_norm": 0.10218273848295212,
      "learning_rate": 1.934451501349507e-06,
      "loss": 0.0153,
      "step": 874
    },
    {
      "epoch": 2.412233469739393,
      "grad_norm": 0.0962023138999939,
      "learning_rate": 1.9168503713501184e-06,
      "loss": 0.0156,
      "step": 875
    },
    {
      "epoch": 2.4149903079905233,
      "grad_norm": 0.09034065902233124,
      "learning_rate": 1.8993211901083353e-06,
      "loss": 0.0152,
      "step": 876
    },
    {
      "epoch": 2.417747146241654,
      "grad_norm": 0.1083788052201271,
      "learning_rate": 1.8818641136527959e-06,
      "loss": 0.0165,
      "step": 877
    },
    {
      "epoch": 2.4205039844927847,
      "grad_norm": 0.09179603308439255,
      "learning_rate": 1.8644792973703252e-06,
      "loss": 0.0154,
      "step": 878
    },
    {
      "epoch": 2.4232608227439156,
      "grad_norm": 0.1054045706987381,
      "learning_rate": 1.8471668960045575e-06,
      "loss": 0.0162,
      "step": 879
    },
    {
      "epoch": 2.426017660995046,
      "grad_norm": 0.10657548904418945,
      "learning_rate": 1.8299270636545518e-06,
      "loss": 0.0149,
      "step": 880
    },
    {
      "epoch": 2.428774499246177,
      "grad_norm": 0.09735960513353348,
      "learning_rate": 1.8127599537734297e-06,
      "loss": 0.0161,
      "step": 881
    },
    {
      "epoch": 2.431531337497308,
      "grad_norm": 0.11152695119380951,
      "learning_rate": 1.7956657191669969e-06,
      "loss": 0.0164,
      "step": 882
    },
    {
      "epoch": 2.4342881757484385,
      "grad_norm": 0.09965367615222931,
      "learning_rate": 1.7786445119923967e-06,
      "loss": 0.0154,
      "step": 883
    },
    {
      "epoch": 2.4370450139995694,
      "grad_norm": 0.09934116899967194,
      "learning_rate": 1.7616964837567497e-06,
      "loss": 0.0159,
      "step": 884
    },
    {
      "epoch": 2.4398018522507,
      "grad_norm": 0.10009805858135223,
      "learning_rate": 1.7448217853158e-06,
      "loss": 0.0159,
      "step": 885
    },
    {
      "epoch": 2.442558690501831,
      "grad_norm": 0.0922827422618866,
      "learning_rate": 1.7280205668725814e-06,
      "loss": 0.0156,
      "step": 886
    },
    {
      "epoch": 2.4453155287529613,
      "grad_norm": 0.1045307070016861,
      "learning_rate": 1.7112929779760768e-06,
      "loss": 0.0156,
      "step": 887
    },
    {
      "epoch": 2.4480723670040923,
      "grad_norm": 0.09331434965133667,
      "learning_rate": 1.6946391675198838e-06,
      "loss": 0.0155,
      "step": 888
    },
    {
      "epoch": 2.4508292052552227,
      "grad_norm": 0.09583646059036255,
      "learning_rate": 1.6780592837408926e-06,
      "loss": 0.0156,
      "step": 889
    },
    {
      "epoch": 2.4535860435063537,
      "grad_norm": 0.0972675010561943,
      "learning_rate": 1.6615534742179684e-06,
      "loss": 0.0158,
      "step": 890
    },
    {
      "epoch": 2.456342881757484,
      "grad_norm": 0.09544920176267624,
      "learning_rate": 1.6451218858706374e-06,
      "loss": 0.0155,
      "step": 891
    },
    {
      "epoch": 2.459099720008615,
      "grad_norm": 0.0940442830324173,
      "learning_rate": 1.6287646649577672e-06,
      "loss": 0.0154,
      "step": 892
    },
    {
      "epoch": 2.461856558259746,
      "grad_norm": 0.09687791019678116,
      "learning_rate": 1.6124819570762862e-06,
      "loss": 0.0152,
      "step": 893
    },
    {
      "epoch": 2.4646133965108765,
      "grad_norm": 0.09892424196004868,
      "learning_rate": 1.5962739071598709e-06,
      "loss": 0.0162,
      "step": 894
    },
    {
      "epoch": 2.4673702347620075,
      "grad_norm": 0.09757326543331146,
      "learning_rate": 1.5801406594776625e-06,
      "loss": 0.0154,
      "step": 895
    },
    {
      "epoch": 2.470127073013138,
      "grad_norm": 0.22701092064380646,
      "learning_rate": 1.5640823576329844e-06,
      "loss": 0.017,
      "step": 896
    },
    {
      "epoch": 2.472883911264269,
      "grad_norm": 0.094330795109272,
      "learning_rate": 1.5480991445620541e-06,
      "loss": 0.0151,
      "step": 897
    },
    {
      "epoch": 2.4756407495153994,
      "grad_norm": 0.09668730944395065,
      "learning_rate": 1.5321911625327224e-06,
      "loss": 0.0155,
      "step": 898
    },
    {
      "epoch": 2.4783975877665303,
      "grad_norm": 0.09876801818609238,
      "learning_rate": 1.5163585531432046e-06,
      "loss": 0.0157,
      "step": 899
    },
    {
      "epoch": 2.4811544260176612,
      "grad_norm": 0.1035366952419281,
      "learning_rate": 1.500601457320814e-06,
      "loss": 0.0151,
      "step": 900
    },
    {
      "epoch": 2.4839112642687917,
      "grad_norm": 0.0975731685757637,
      "learning_rate": 1.4849200153207176e-06,
      "loss": 0.0155,
      "step": 901
    },
    {
      "epoch": 2.4866681025199227,
      "grad_norm": 0.0988369807600975,
      "learning_rate": 1.4693143667246713e-06,
      "loss": 0.0157,
      "step": 902
    },
    {
      "epoch": 2.489424940771053,
      "grad_norm": 0.10393507033586502,
      "learning_rate": 1.453784650439798e-06,
      "loss": 0.0159,
      "step": 903
    },
    {
      "epoch": 2.492181779022184,
      "grad_norm": 0.09848541766405106,
      "learning_rate": 1.4383310046973365e-06,
      "loss": 0.0156,
      "step": 904
    },
    {
      "epoch": 2.4949386172733146,
      "grad_norm": 0.09526592493057251,
      "learning_rate": 1.4229535670514162e-06,
      "loss": 0.0159,
      "step": 905
    },
    {
      "epoch": 2.4976954555244455,
      "grad_norm": 0.256574809551239,
      "learning_rate": 1.407652474377832e-06,
      "loss": 0.0174,
      "step": 906
    },
    {
      "epoch": 2.500452293775576,
      "grad_norm": 0.0996571034193039,
      "learning_rate": 1.3924278628728305e-06,
      "loss": 0.0164,
      "step": 907
    },
    {
      "epoch": 2.503209132026707,
      "grad_norm": 0.10017193108797073,
      "learning_rate": 1.3772798680518828e-06,
      "loss": 0.0156,
      "step": 908
    },
    {
      "epoch": 2.5059659702778374,
      "grad_norm": 0.09400206059217453,
      "learning_rate": 1.3622086247484989e-06,
      "loss": 0.0157,
      "step": 909
    },
    {
      "epoch": 2.5087228085289683,
      "grad_norm": 0.09920066595077515,
      "learning_rate": 1.3472142671130139e-06,
      "loss": 0.0156,
      "step": 910
    },
    {
      "epoch": 2.5114796467800993,
      "grad_norm": 0.09991924464702606,
      "learning_rate": 1.3322969286113973e-06,
      "loss": 0.0162,
      "step": 911
    },
    {
      "epoch": 2.5142364850312298,
      "grad_norm": 0.09943817555904388,
      "learning_rate": 1.3174567420240647e-06,
      "loss": 0.0158,
      "step": 912
    },
    {
      "epoch": 2.5169933232823607,
      "grad_norm": 0.10019614547491074,
      "learning_rate": 1.3026938394446976e-06,
      "loss": 0.0153,
      "step": 913
    },
    {
      "epoch": 2.519750161533491,
      "grad_norm": 0.09503661096096039,
      "learning_rate": 1.2880083522790654e-06,
      "loss": 0.0158,
      "step": 914
    },
    {
      "epoch": 2.522506999784622,
      "grad_norm": 0.09582358598709106,
      "learning_rate": 1.273400411243857e-06,
      "loss": 0.0155,
      "step": 915
    },
    {
      "epoch": 2.5252638380357526,
      "grad_norm": 0.09465383738279343,
      "learning_rate": 1.2588701463655172e-06,
      "loss": 0.0154,
      "step": 916
    },
    {
      "epoch": 2.5280206762868835,
      "grad_norm": 0.10085918009281158,
      "learning_rate": 1.2444176869790925e-06,
      "loss": 0.016,
      "step": 917
    },
    {
      "epoch": 2.5307775145380145,
      "grad_norm": 0.10438365489244461,
      "learning_rate": 1.2300431617270669e-06,
      "loss": 0.0153,
      "step": 918
    },
    {
      "epoch": 2.533534352789145,
      "grad_norm": 0.1272774487733841,
      "learning_rate": 1.2157466985582367e-06,
      "loss": 0.016,
      "step": 919
    },
    {
      "epoch": 2.5362911910402755,
      "grad_norm": 0.10423656553030014,
      "learning_rate": 1.2015284247265567e-06,
      "loss": 0.0163,
      "step": 920
    },
    {
      "epoch": 2.5390480292914064,
      "grad_norm": 0.09617140144109726,
      "learning_rate": 1.1873884667900125e-06,
      "loss": 0.0157,
      "step": 921
    },
    {
      "epoch": 2.5418048675425373,
      "grad_norm": 0.09703914076089859,
      "learning_rate": 1.1733269506094957e-06,
      "loss": 0.0162,
      "step": 922
    },
    {
      "epoch": 2.544561705793668,
      "grad_norm": 0.09608243405818939,
      "learning_rate": 1.1593440013476775e-06,
      "loss": 0.0151,
      "step": 923
    },
    {
      "epoch": 2.5473185440447987,
      "grad_norm": 0.107514888048172,
      "learning_rate": 1.1454397434679022e-06,
      "loss": 0.0158,
      "step": 924
    },
    {
      "epoch": 2.5500753822959292,
      "grad_norm": 0.09657265990972519,
      "learning_rate": 1.1316143007330739e-06,
      "loss": 0.0153,
      "step": 925
    },
    {
      "epoch": 2.55283222054706,
      "grad_norm": 0.08969084173440933,
      "learning_rate": 1.1178677962045604e-06,
      "loss": 0.0154,
      "step": 926
    },
    {
      "epoch": 2.5555890587981906,
      "grad_norm": 0.09439155459403992,
      "learning_rate": 1.1042003522410882e-06,
      "loss": 0.0157,
      "step": 927
    },
    {
      "epoch": 2.5583458970493216,
      "grad_norm": 0.10195891559123993,
      "learning_rate": 1.090612090497668e-06,
      "loss": 0.0164,
      "step": 928
    },
    {
      "epoch": 2.5611027353004525,
      "grad_norm": 0.10047302395105362,
      "learning_rate": 1.077103131924493e-06,
      "loss": 0.016,
      "step": 929
    },
    {
      "epoch": 2.563859573551583,
      "grad_norm": 0.10115232318639755,
      "learning_rate": 1.0636735967658785e-06,
      "loss": 0.0158,
      "step": 930
    },
    {
      "epoch": 2.5666164118027135,
      "grad_norm": 0.09630636125802994,
      "learning_rate": 1.0503236045591857e-06,
      "loss": 0.0156,
      "step": 931
    },
    {
      "epoch": 2.5693732500538444,
      "grad_norm": 0.09693799912929535,
      "learning_rate": 1.037053274133758e-06,
      "loss": 0.0153,
      "step": 932
    },
    {
      "epoch": 2.5721300883049754,
      "grad_norm": 0.09665205329656601,
      "learning_rate": 1.0238627236098619e-06,
      "loss": 0.0151,
      "step": 933
    },
    {
      "epoch": 2.574886926556106,
      "grad_norm": 0.0942736491560936,
      "learning_rate": 1.0107520703976325e-06,
      "loss": 0.0151,
      "step": 934
    },
    {
      "epoch": 2.577643764807237,
      "grad_norm": 0.10749771445989609,
      "learning_rate": 9.977214311960404e-07,
      "loss": 0.016,
      "step": 935
    },
    {
      "epoch": 2.5804006030583673,
      "grad_norm": 0.09226176887750626,
      "learning_rate": 9.8477092199184e-07,
      "loss": 0.0152,
      "step": 936
    },
    {
      "epoch": 2.583157441309498,
      "grad_norm": 0.10179731994867325,
      "learning_rate": 9.719006580585444e-07,
      "loss": 0.0158,
      "step": 937
    },
    {
      "epoch": 2.5859142795606287,
      "grad_norm": 0.09366576373577118,
      "learning_rate": 9.591107539553945e-07,
      "loss": 0.0159,
      "step": 938
    },
    {
      "epoch": 2.5886711178117596,
      "grad_norm": 0.097577303647995,
      "learning_rate": 9.464013235263458e-07,
      "loss": 0.0157,
      "step": 939
    },
    {
      "epoch": 2.5914279560628906,
      "grad_norm": 0.10417082905769348,
      "learning_rate": 9.337724798990489e-07,
      "loss": 0.018,
      "step": 940
    },
    {
      "epoch": 2.594184794314021,
      "grad_norm": 0.1003030315041542,
      "learning_rate": 9.212243354838435e-07,
      "loss": 0.016,
      "step": 941
    },
    {
      "epoch": 2.596941632565152,
      "grad_norm": 0.09599234908819199,
      "learning_rate": 9.08757001972762e-07,
      "loss": 0.0154,
      "step": 942
    },
    {
      "epoch": 2.5996984708162825,
      "grad_norm": 0.09770214557647705,
      "learning_rate": 8.963705903385344e-07,
      "loss": 0.016,
      "step": 943
    },
    {
      "epoch": 2.6024553090674134,
      "grad_norm": 0.09537551552057266,
      "learning_rate": 8.8406521083359e-07,
      "loss": 0.0158,
      "step": 944
    },
    {
      "epoch": 2.605212147318544,
      "grad_norm": 0.09884826838970184,
      "learning_rate": 8.71840972989092e-07,
      "loss": 0.0156,
      "step": 945
    },
    {
      "epoch": 2.607968985569675,
      "grad_norm": 0.09631089121103287,
      "learning_rate": 8.596979856139553e-07,
      "loss": 0.0157,
      "step": 946
    },
    {
      "epoch": 2.6107258238208058,
      "grad_norm": 0.09276318550109863,
      "learning_rate": 8.476363567938751e-07,
      "loss": 0.0151,
      "step": 947
    },
    {
      "epoch": 2.6134826620719362,
      "grad_norm": 0.10186196863651276,
      "learning_rate": 8.356561938903707e-07,
      "loss": 0.0151,
      "step": 948
    },
    {
      "epoch": 2.6162395003230667,
      "grad_norm": 0.0975264310836792,
      "learning_rate": 8.237576035398198e-07,
      "loss": 0.0156,
      "step": 949
    },
    {
      "epoch": 2.6189963385741977,
      "grad_norm": 0.10101190954446793,
      "learning_rate": 8.119406916525252e-07,
      "loss": 0.016,
      "step": 950
    },
    {
      "epoch": 2.6217531768253286,
      "grad_norm": 0.10287696123123169,
      "learning_rate": 8.002055634117578e-07,
      "loss": 0.0157,
      "step": 951
    },
    {
      "epoch": 2.624510015076459,
      "grad_norm": 0.0948198139667511,
      "learning_rate": 7.885523232728287e-07,
      "loss": 0.0158,
      "step": 952
    },
    {
      "epoch": 2.62726685332759,
      "grad_norm": 0.10126490145921707,
      "learning_rate": 7.769810749621532e-07,
      "loss": 0.0162,
      "step": 953
    },
    {
      "epoch": 2.6300236915787205,
      "grad_norm": 0.09267105907201767,
      "learning_rate": 7.654919214763357e-07,
      "loss": 0.0155,
      "step": 954
    },
    {
      "epoch": 2.6327805298298514,
      "grad_norm": 0.10376208275556564,
      "learning_rate": 7.540849650812409e-07,
      "loss": 0.0154,
      "step": 955
    },
    {
      "epoch": 2.635537368080982,
      "grad_norm": 0.09337103366851807,
      "learning_rate": 7.427603073110967e-07,
      "loss": 0.0157,
      "step": 956
    },
    {
      "epoch": 2.638294206332113,
      "grad_norm": 0.08885804563760757,
      "learning_rate": 7.315180489675822e-07,
      "loss": 0.0153,
      "step": 957
    },
    {
      "epoch": 2.641051044583244,
      "grad_norm": 0.09852619469165802,
      "learning_rate": 7.203582901189332e-07,
      "loss": 0.0152,
      "step": 958
    },
    {
      "epoch": 2.6438078828343743,
      "grad_norm": 0.0962531641125679,
      "learning_rate": 7.092811300990521e-07,
      "loss": 0.0153,
      "step": 959
    },
    {
      "epoch": 2.6465647210855052,
      "grad_norm": 0.09466900676488876,
      "learning_rate": 6.98286667506618e-07,
      "loss": 0.0152,
      "step": 960
    },
    {
      "epoch": 2.6493215593366357,
      "grad_norm": 0.09106676280498505,
      "learning_rate": 6.87375000204219e-07,
      "loss": 0.0147,
      "step": 961
    },
    {
      "epoch": 2.6520783975877666,
      "grad_norm": 0.09509000927209854,
      "learning_rate": 6.765462253174715e-07,
      "loss": 0.0155,
      "step": 962
    },
    {
      "epoch": 2.654835235838897,
      "grad_norm": 0.09613344073295593,
      "learning_rate": 6.658004392341633e-07,
      "loss": 0.0155,
      "step": 963
    },
    {
      "epoch": 2.657592074090028,
      "grad_norm": 0.09688057750463486,
      "learning_rate": 6.551377376033896e-07,
      "loss": 0.0157,
      "step": 964
    },
    {
      "epoch": 2.660348912341159,
      "grad_norm": 0.09640353918075562,
      "learning_rate": 6.445582153347074e-07,
      "loss": 0.0159,
      "step": 965
    },
    {
      "epoch": 2.6631057505922895,
      "grad_norm": 0.10333269834518433,
      "learning_rate": 6.340619665972847e-07,
      "loss": 0.0158,
      "step": 966
    },
    {
      "epoch": 2.66586258884342,
      "grad_norm": 0.09676087647676468,
      "learning_rate": 6.236490848190657e-07,
      "loss": 0.0156,
      "step": 967
    },
    {
      "epoch": 2.668619427094551,
      "grad_norm": 0.09745439141988754,
      "learning_rate": 6.133196626859406e-07,
      "loss": 0.0163,
      "step": 968
    },
    {
      "epoch": 2.671376265345682,
      "grad_norm": 0.0986739918589592,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.0163,
      "step": 969
    },
    {
      "epoch": 2.6741331035968123,
      "grad_norm": 0.09318210184574127,
      "learning_rate": 5.929115643833005e-07,
      "loss": 0.0158,
      "step": 970
    },
    {
      "epoch": 2.6768899418479433,
      "grad_norm": 0.09318020939826965,
      "learning_rate": 5.828330698678908e-07,
      "loss": 0.0153,
      "step": 971
    },
    {
      "epoch": 2.6796467800990738,
      "grad_norm": 0.0958879366517067,
      "learning_rate": 5.728383983041696e-07,
      "loss": 0.0159,
      "step": 972
    },
    {
      "epoch": 2.6824036183502047,
      "grad_norm": 0.10100314021110535,
      "learning_rate": 5.629276386555016e-07,
      "loss": 0.0152,
      "step": 973
    },
    {
      "epoch": 2.685160456601335,
      "grad_norm": 0.0920896902680397,
      "learning_rate": 5.531008791383485e-07,
      "loss": 0.0153,
      "step": 974
    },
    {
      "epoch": 2.687917294852466,
      "grad_norm": 0.09028688073158264,
      "learning_rate": 5.43358207221476e-07,
      "loss": 0.0156,
      "step": 975
    },
    {
      "epoch": 2.690674133103597,
      "grad_norm": 0.09332931786775589,
      "learning_rate": 5.336997096251816e-07,
      "loss": 0.0151,
      "step": 976
    },
    {
      "epoch": 2.6934309713547275,
      "grad_norm": 0.10222768783569336,
      "learning_rate": 5.241254723205225e-07,
      "loss": 0.0153,
      "step": 977
    },
    {
      "epoch": 2.696187809605858,
      "grad_norm": 0.10587936639785767,
      "learning_rate": 5.146355805285452e-07,
      "loss": 0.0151,
      "step": 978
    },
    {
      "epoch": 2.698944647856989,
      "grad_norm": 0.09399544447660446,
      "learning_rate": 5.052301187195296e-07,
      "loss": 0.0156,
      "step": 979
    },
    {
      "epoch": 2.70170148610812,
      "grad_norm": 0.09084748476743698,
      "learning_rate": 4.959091706122431e-07,
      "loss": 0.0152,
      "step": 980
    },
    {
      "epoch": 2.7044583243592504,
      "grad_norm": 0.09691156446933746,
      "learning_rate": 4.866728191731829e-07,
      "loss": 0.0156,
      "step": 981
    },
    {
      "epoch": 2.7072151626103813,
      "grad_norm": 0.09536400437355042,
      "learning_rate": 4.775211466158469e-07,
      "loss": 0.0157,
      "step": 982
    },
    {
      "epoch": 2.709972000861512,
      "grad_norm": 0.0932604968547821,
      "learning_rate": 4.6845423440000315e-07,
      "loss": 0.0151,
      "step": 983
    },
    {
      "epoch": 2.7127288391126427,
      "grad_norm": 0.0943024605512619,
      "learning_rate": 4.594721632309551e-07,
      "loss": 0.0157,
      "step": 984
    },
    {
      "epoch": 2.715485677363773,
      "grad_norm": 0.11024822294712067,
      "learning_rate": 4.505750130588371e-07,
      "loss": 0.0159,
      "step": 985
    },
    {
      "epoch": 2.718242515614904,
      "grad_norm": 0.09148909151554108,
      "learning_rate": 4.4176286307788475e-07,
      "loss": 0.0152,
      "step": 986
    },
    {
      "epoch": 2.720999353866035,
      "grad_norm": 0.10189063847064972,
      "learning_rate": 4.3303579172574884e-07,
      "loss": 0.0164,
      "step": 987
    },
    {
      "epoch": 2.7237561921171656,
      "grad_norm": 0.09687193483114243,
      "learning_rate": 4.243938766827849e-07,
      "loss": 0.0156,
      "step": 988
    },
    {
      "epoch": 2.7265130303682965,
      "grad_norm": 0.0970255509018898,
      "learning_rate": 4.1583719487136575e-07,
      "loss": 0.0154,
      "step": 989
    },
    {
      "epoch": 2.729269868619427,
      "grad_norm": 0.08850499987602234,
      "learning_rate": 4.0736582245519795e-07,
      "loss": 0.0151,
      "step": 990
    },
    {
      "epoch": 2.732026706870558,
      "grad_norm": 0.10004325211048126,
      "learning_rate": 3.9897983483863866e-07,
      "loss": 0.0157,
      "step": 991
    },
    {
      "epoch": 2.7347835451216884,
      "grad_norm": 0.09604191035032272,
      "learning_rate": 3.9067930666603304e-07,
      "loss": 0.0154,
      "step": 992
    },
    {
      "epoch": 2.7375403833728194,
      "grad_norm": 0.096500463783741,
      "learning_rate": 3.824643118210403e-07,
      "loss": 0.0159,
      "step": 993
    },
    {
      "epoch": 2.7402972216239503,
      "grad_norm": 0.08940698951482773,
      "learning_rate": 3.743349234259841e-07,
      "loss": 0.0155,
      "step": 994
    },
    {
      "epoch": 2.7430540598750808,
      "grad_norm": 0.09379703551530838,
      "learning_rate": 3.662912138411967e-07,
      "loss": 0.0153,
      "step": 995
    },
    {
      "epoch": 2.7458108981262113,
      "grad_norm": 0.09963972121477127,
      "learning_rate": 3.5833325466437697e-07,
      "loss": 0.0156,
      "step": 996
    },
    {
      "epoch": 2.748567736377342,
      "grad_norm": 0.09487731754779816,
      "learning_rate": 3.5046111672995097e-07,
      "loss": 0.0158,
      "step": 997
    },
    {
      "epoch": 2.751324574628473,
      "grad_norm": 0.09154865145683289,
      "learning_rate": 3.426748701084448e-07,
      "loss": 0.0152,
      "step": 998
    },
    {
      "epoch": 2.7540814128796036,
      "grad_norm": 0.09428098797798157,
      "learning_rate": 3.349745841058605e-07,
      "loss": 0.0155,
      "step": 999
    },
    {
      "epoch": 2.7568382511307346,
      "grad_norm": 0.09710869193077087,
      "learning_rate": 3.2736032726305546e-07,
      "loss": 0.0158,
      "step": 1000
    },
    {
      "epoch": 2.759595089381865,
      "grad_norm": 0.10234350711107254,
      "learning_rate": 3.198321673551341e-07,
      "loss": 0.0159,
      "step": 1001
    },
    {
      "epoch": 2.762351927632996,
      "grad_norm": 0.09612155705690384,
      "learning_rate": 3.1239017139084725e-07,
      "loss": 0.0162,
      "step": 1002
    },
    {
      "epoch": 2.7651087658841265,
      "grad_norm": 0.10025949776172638,
      "learning_rate": 3.050344056119925e-07,
      "loss": 0.0158,
      "step": 1003
    },
    {
      "epoch": 2.7678656041352574,
      "grad_norm": 0.09752298891544342,
      "learning_rate": 2.977649354928258e-07,
      "loss": 0.0153,
      "step": 1004
    },
    {
      "epoch": 2.7706224423863883,
      "grad_norm": 0.09761834889650345,
      "learning_rate": 2.905818257394799e-07,
      "loss": 0.0155,
      "step": 1005
    },
    {
      "epoch": 2.773379280637519,
      "grad_norm": 0.09161604940891266,
      "learning_rate": 2.834851402893857e-07,
      "loss": 0.0148,
      "step": 1006
    },
    {
      "epoch": 2.7761361188886493,
      "grad_norm": 0.09183601289987564,
      "learning_rate": 2.764749423107027e-07,
      "loss": 0.0151,
      "step": 1007
    },
    {
      "epoch": 2.7788929571397802,
      "grad_norm": 0.09860436618328094,
      "learning_rate": 2.6955129420176193e-07,
      "loss": 0.0165,
      "step": 1008
    },
    {
      "epoch": 2.781649795390911,
      "grad_norm": 0.09372715651988983,
      "learning_rate": 2.627142575905062e-07,
      "loss": 0.0154,
      "step": 1009
    },
    {
      "epoch": 2.7844066336420417,
      "grad_norm": 0.10581711679697037,
      "learning_rate": 2.559638933339414e-07,
      "loss": 0.0157,
      "step": 1010
    },
    {
      "epoch": 2.7871634718931726,
      "grad_norm": 0.09646962583065033,
      "learning_rate": 2.493002615175977e-07,
      "loss": 0.0156,
      "step": 1011
    },
    {
      "epoch": 2.7899203101443035,
      "grad_norm": 0.09317374974489212,
      "learning_rate": 2.4272342145499006e-07,
      "loss": 0.0154,
      "step": 1012
    },
    {
      "epoch": 2.792677148395434,
      "grad_norm": 0.09786811470985413,
      "learning_rate": 2.3623343168709624e-07,
      "loss": 0.0157,
      "step": 1013
    },
    {
      "epoch": 2.7954339866465645,
      "grad_norm": 0.09875071793794632,
      "learning_rate": 2.2983034998182997e-07,
      "loss": 0.0157,
      "step": 1014
    },
    {
      "epoch": 2.7981908248976954,
      "grad_norm": 0.09668000042438507,
      "learning_rate": 2.235142333335316e-07,
      "loss": 0.0151,
      "step": 1015
    },
    {
      "epoch": 2.8009476631488264,
      "grad_norm": 0.09402566403150558,
      "learning_rate": 2.1728513796245855e-07,
      "loss": 0.0154,
      "step": 1016
    },
    {
      "epoch": 2.803704501399957,
      "grad_norm": 0.09588855504989624,
      "learning_rate": 2.11143119314281e-07,
      "loss": 0.0157,
      "step": 1017
    },
    {
      "epoch": 2.806461339651088,
      "grad_norm": 0.09455292671918869,
      "learning_rate": 2.0508823205959815e-07,
      "loss": 0.0154,
      "step": 1018
    },
    {
      "epoch": 2.8092181779022183,
      "grad_norm": 0.09076821804046631,
      "learning_rate": 1.991205300934429e-07,
      "loss": 0.0154,
      "step": 1019
    },
    {
      "epoch": 2.811975016153349,
      "grad_norm": 0.09466226398944855,
      "learning_rate": 1.9324006653480332e-07,
      "loss": 0.0152,
      "step": 1020
    },
    {
      "epoch": 2.8147318544044797,
      "grad_norm": 0.10528804361820221,
      "learning_rate": 1.874468937261531e-07,
      "loss": 0.0157,
      "step": 1021
    },
    {
      "epoch": 2.8174886926556106,
      "grad_norm": 0.09444960951805115,
      "learning_rate": 1.8174106323298634e-07,
      "loss": 0.0154,
      "step": 1022
    },
    {
      "epoch": 2.8202455309067416,
      "grad_norm": 0.09570340067148209,
      "learning_rate": 1.761226258433524e-07,
      "loss": 0.0157,
      "step": 1023
    },
    {
      "epoch": 2.823002369157872,
      "grad_norm": 0.09762073308229446,
      "learning_rate": 1.7059163156740943e-07,
      "loss": 0.0155,
      "step": 1024
    },
    {
      "epoch": 2.8257592074090025,
      "grad_norm": 0.09936973452568054,
      "learning_rate": 1.6514812963697723e-07,
      "loss": 0.0156,
      "step": 1025
    },
    {
      "epoch": 2.8285160456601335,
      "grad_norm": 0.09471126645803452,
      "learning_rate": 1.5979216850509848e-07,
      "loss": 0.0151,
      "step": 1026
    },
    {
      "epoch": 2.8312728839112644,
      "grad_norm": 0.09497305750846863,
      "learning_rate": 1.545237958456125e-07,
      "loss": 0.0155,
      "step": 1027
    },
    {
      "epoch": 2.834029722162395,
      "grad_norm": 0.09496881812810898,
      "learning_rate": 1.4934305855271892e-07,
      "loss": 0.0153,
      "step": 1028
    },
    {
      "epoch": 2.836786560413526,
      "grad_norm": 0.09368608891963959,
      "learning_rate": 1.4425000274057577e-07,
      "loss": 0.0157,
      "step": 1029
    },
    {
      "epoch": 2.8395433986646563,
      "grad_norm": 0.1018013209104538,
      "learning_rate": 1.3924467374287432e-07,
      "loss": 0.0157,
      "step": 1030
    },
    {
      "epoch": 2.8423002369157873,
      "grad_norm": 0.09439851343631744,
      "learning_rate": 1.343271161124493e-07,
      "loss": 0.0159,
      "step": 1031
    },
    {
      "epoch": 2.8450570751669177,
      "grad_norm": 0.10299389809370041,
      "learning_rate": 1.2949737362087156e-07,
      "loss": 0.0157,
      "step": 1032
    },
    {
      "epoch": 2.8478139134180487,
      "grad_norm": 0.10252979397773743,
      "learning_rate": 1.247554892580616e-07,
      "loss": 0.0159,
      "step": 1033
    },
    {
      "epoch": 2.8505707516691796,
      "grad_norm": 0.08925472944974899,
      "learning_rate": 1.201015052319099e-07,
      "loss": 0.0155,
      "step": 1034
    },
    {
      "epoch": 2.85332758992031,
      "grad_norm": 0.09510352462530136,
      "learning_rate": 1.1553546296789952e-07,
      "loss": 0.0159,
      "step": 1035
    },
    {
      "epoch": 2.856084428171441,
      "grad_norm": 0.09258411824703217,
      "learning_rate": 1.1105740310873414e-07,
      "loss": 0.0151,
      "step": 1036
    },
    {
      "epoch": 2.8588412664225715,
      "grad_norm": 0.09796860814094543,
      "learning_rate": 1.066673655139816e-07,
      "loss": 0.0153,
      "step": 1037
    },
    {
      "epoch": 2.8615981046737025,
      "grad_norm": 0.0932827889919281,
      "learning_rate": 1.0236538925971429e-07,
      "loss": 0.0155,
      "step": 1038
    },
    {
      "epoch": 2.864354942924833,
      "grad_norm": 0.09266915917396545,
      "learning_rate": 9.815151263816714e-08,
      "loss": 0.0153,
      "step": 1039
    },
    {
      "epoch": 2.867111781175964,
      "grad_norm": 0.09166253358125687,
      "learning_rate": 9.402577315738904e-08,
      "loss": 0.0154,
      "step": 1040
    },
    {
      "epoch": 2.869868619427095,
      "grad_norm": 0.10095134377479553,
      "learning_rate": 8.99882075409153e-08,
      "loss": 0.0161,
      "step": 1041
    },
    {
      "epoch": 2.8726254576782253,
      "grad_norm": 0.09815669804811478,
      "learning_rate": 8.603885172744131e-08,
      "loss": 0.0163,
      "step": 1042
    },
    {
      "epoch": 2.875382295929356,
      "grad_norm": 0.09382443875074387,
      "learning_rate": 8.217774087049268e-08,
      "loss": 0.016,
      "step": 1043
    },
    {
      "epoch": 2.8781391341804867,
      "grad_norm": 0.09190834313631058,
      "learning_rate": 7.840490933812783e-08,
      "loss": 0.0158,
      "step": 1044
    },
    {
      "epoch": 2.8808959724316177,
      "grad_norm": 0.09133637696504593,
      "learning_rate": 7.472039071261927e-08,
      "loss": 0.0149,
      "step": 1045
    },
    {
      "epoch": 2.883652810682748,
      "grad_norm": 0.10109441727399826,
      "learning_rate": 7.112421779015944e-08,
      "loss": 0.0152,
      "step": 1046
    },
    {
      "epoch": 2.886409648933879,
      "grad_norm": 0.09567693620920181,
      "learning_rate": 6.761642258056977e-08,
      "loss": 0.0151,
      "step": 1047
    },
    {
      "epoch": 2.8891664871850096,
      "grad_norm": 0.17240361869335175,
      "learning_rate": 6.419703630701546e-08,
      "loss": 0.0174,
      "step": 1048
    },
    {
      "epoch": 2.8919233254361405,
      "grad_norm": 0.09366879612207413,
      "learning_rate": 6.086608940572447e-08,
      "loss": 0.0157,
      "step": 1049
    },
    {
      "epoch": 2.894680163687271,
      "grad_norm": 0.09174374490976334,
      "learning_rate": 5.7623611525721155e-08,
      "loss": 0.0152,
      "step": 1050
    },
    {
      "epoch": 2.897437001938402,
      "grad_norm": 0.096317358314991,
      "learning_rate": 5.446963152855644e-08,
      "loss": 0.0156,
      "step": 1051
    },
    {
      "epoch": 2.900193840189533,
      "grad_norm": 0.09576837718486786,
      "learning_rate": 5.140417748806026e-08,
      "loss": 0.0152,
      "step": 1052
    },
    {
      "epoch": 2.9029506784406633,
      "grad_norm": 0.09430639445781708,
      "learning_rate": 4.8427276690081735e-08,
      "loss": 0.0152,
      "step": 1053
    },
    {
      "epoch": 2.905707516691794,
      "grad_norm": 0.10490168631076813,
      "learning_rate": 4.553895563225053e-08,
      "loss": 0.015,
      "step": 1054
    },
    {
      "epoch": 2.9084643549429248,
      "grad_norm": 0.0979146957397461,
      "learning_rate": 4.2739240023742526e-08,
      "loss": 0.0154,
      "step": 1055
    },
    {
      "epoch": 2.9112211931940557,
      "grad_norm": 0.10030056536197662,
      "learning_rate": 4.002815478505007e-08,
      "loss": 0.0155,
      "step": 1056
    },
    {
      "epoch": 2.913978031445186,
      "grad_norm": 0.10702443867921829,
      "learning_rate": 3.7405724047756554e-08,
      "loss": 0.0163,
      "step": 1057
    },
    {
      "epoch": 2.916734869696317,
      "grad_norm": 0.09876976162195206,
      "learning_rate": 3.487197115432883e-08,
      "loss": 0.0147,
      "step": 1058
    },
    {
      "epoch": 2.9194917079474476,
      "grad_norm": 0.09448473155498505,
      "learning_rate": 3.242691865790071e-08,
      "loss": 0.0148,
      "step": 1059
    },
    {
      "epoch": 2.9222485461985785,
      "grad_norm": 0.10175088047981262,
      "learning_rate": 3.0070588322079765e-08,
      "loss": 0.0165,
      "step": 1060
    },
    {
      "epoch": 2.925005384449709,
      "grad_norm": 0.09407947957515717,
      "learning_rate": 2.780300112074974e-08,
      "loss": 0.0154,
      "step": 1061
    },
    {
      "epoch": 2.92776222270084,
      "grad_norm": 0.09297125786542892,
      "learning_rate": 2.5624177237884017e-08,
      "loss": 0.0156,
      "step": 1062
    },
    {
      "epoch": 2.930519060951971,
      "grad_norm": 0.10525539517402649,
      "learning_rate": 2.3534136067369094e-08,
      "loss": 0.0165,
      "step": 1063
    },
    {
      "epoch": 2.9332758992031014,
      "grad_norm": 0.09468672424554825,
      "learning_rate": 2.1532896212825837e-08,
      "loss": 0.0158,
      "step": 1064
    },
    {
      "epoch": 2.9360327374542323,
      "grad_norm": 0.0892932116985321,
      "learning_rate": 1.962047548744961e-08,
      "loss": 0.0148,
      "step": 1065
    },
    {
      "epoch": 2.938789575705363,
      "grad_norm": 0.09384763240814209,
      "learning_rate": 1.7796890913850395e-08,
      "loss": 0.0154,
      "step": 1066
    },
    {
      "epoch": 2.9415464139564937,
      "grad_norm": 0.10021128505468369,
      "learning_rate": 1.606215872389738e-08,
      "loss": 0.0154,
      "step": 1067
    },
    {
      "epoch": 2.9443032522076242,
      "grad_norm": 0.09487558901309967,
      "learning_rate": 1.4416294358582383e-08,
      "loss": 0.0151,
      "step": 1068
    },
    {
      "epoch": 2.947060090458755,
      "grad_norm": 0.09665752202272415,
      "learning_rate": 1.2859312467872197e-08,
      "loss": 0.0148,
      "step": 1069
    },
    {
      "epoch": 2.949816928709886,
      "grad_norm": 0.11037097126245499,
      "learning_rate": 1.1391226910588693e-08,
      "loss": 0.0153,
      "step": 1070
    },
    {
      "epoch": 2.9525737669610166,
      "grad_norm": 0.101137176156044,
      "learning_rate": 1.0012050754277802e-08,
      "loss": 0.0153,
      "step": 1071
    },
    {
      "epoch": 2.955330605212147,
      "grad_norm": 0.09380457550287247,
      "learning_rate": 8.721796275095173e-09,
      "loss": 0.0152,
      "step": 1072
    },
    {
      "epoch": 2.958087443463278,
      "grad_norm": 0.09098217636346817,
      "learning_rate": 7.520474957699586e-09,
      "loss": 0.0149,
      "step": 1073
    },
    {
      "epoch": 2.960844281714409,
      "grad_norm": 0.0910525918006897,
      "learning_rate": 6.40809749514637e-09,
      "loss": 0.015,
      "step": 1074
    },
    {
      "epoch": 2.9636011199655394,
      "grad_norm": 0.09770431369543076,
      "learning_rate": 5.384673788797479e-09,
      "loss": 0.0151,
      "step": 1075
    },
    {
      "epoch": 2.9663579582166704,
      "grad_norm": 0.09908950328826904,
      "learning_rate": 4.450212948227117e-09,
      "loss": 0.0161,
      "step": 1076
    },
    {
      "epoch": 2.969114796467801,
      "grad_norm": 0.09452182054519653,
      "learning_rate": 3.6047232911462506e-09,
      "loss": 0.0153,
      "step": 1077
    },
    {
      "epoch": 2.971871634718932,
      "grad_norm": 0.09382626414299011,
      "learning_rate": 2.8482123433248853e-09,
      "loss": 0.0153,
      "step": 1078
    },
    {
      "epoch": 2.9746284729700623,
      "grad_norm": 0.0985400527715683,
      "learning_rate": 2.180686838527679e-09,
      "loss": 0.0158,
      "step": 1079
    },
    {
      "epoch": 2.977385311221193,
      "grad_norm": 0.09935043007135391,
      "learning_rate": 1.6021527184528761e-09,
      "loss": 0.0163,
      "step": 1080
    },
    {
      "epoch": 2.980142149472324,
      "grad_norm": 0.09754447638988495,
      "learning_rate": 1.1126151326779077e-09,
      "loss": 0.0156,
      "step": 1081
    },
    {
      "epoch": 2.9828989877234546,
      "grad_norm": 0.09899445623159409,
      "learning_rate": 7.120784386160928e-10,
      "loss": 0.0162,
      "step": 1082
    },
    {
      "epoch": 2.985655825974585,
      "grad_norm": 0.09259970486164093,
      "learning_rate": 4.005462014766703e-10,
      "loss": 0.0153,
      "step": 1083
    },
    {
      "epoch": 2.988412664225716,
      "grad_norm": 0.09933052957057953,
      "learning_rate": 1.7802119423149244e-10,
      "loss": 0.016,
      "step": 1084
    },
    {
      "epoch": 2.991169502476847,
      "grad_norm": 0.09650570899248123,
      "learning_rate": 4.450539759393024e-11,
      "loss": 0.0155,
      "step": 1085
    },
    {
      "epoch": 2.9939263407279775,
      "grad_norm": 0.0921320915222168,
      "learning_rate": 0.0,
      "loss": 0.0156,
      "step": 1086
    },
    {
      "epoch": 2.9939263407279775,
      "step": 1086,
      "total_flos": 5.826670523626553e+18,
      "train_loss": 0.02890731801551597,
      "train_runtime": 65667.5663,
      "train_samples_per_second": 8.484,
      "train_steps_per_second": 0.017
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1086,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.826670523626553e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}