{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.1000107828337287,
  "eval_steps": 500,
  "global_step": 1855,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.391416864351952e-05,
      "grad_norm": 53.75010299682617,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 2.5864,
      "step": 1
    },
    {
      "epoch": 0.00010782833728703904,
      "grad_norm": 45.00067138671875,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 2.3757,
      "step": 2
    },
    {
      "epoch": 0.00016174250593055855,
      "grad_norm": 51.22366714477539,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 2.4653,
      "step": 3
    },
    {
      "epoch": 0.00021565667457407807,
      "grad_norm": 62.225242614746094,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 2.5819,
      "step": 4
    },
    {
      "epoch": 0.0002695708432175976,
      "grad_norm": 54.67008590698242,
      "learning_rate": 5.000000000000001e-07,
      "loss": 2.6368,
      "step": 5
    },
    {
      "epoch": 0.0003234850118611171,
      "grad_norm": 51.261009216308594,
      "learning_rate": 6.000000000000001e-07,
      "loss": 2.3245,
      "step": 6
    },
    {
      "epoch": 0.0003773991805046366,
      "grad_norm": 53.58714294433594,
      "learning_rate": 7.000000000000001e-07,
      "loss": 2.7622,
      "step": 7
    },
    {
      "epoch": 0.00043131334914815614,
      "grad_norm": 41.32997131347656,
      "learning_rate": 8.000000000000001e-07,
      "loss": 2.6444,
      "step": 8
    },
    {
      "epoch": 0.00048522751779167566,
      "grad_norm": 33.232242584228516,
      "learning_rate": 9.000000000000001e-07,
      "loss": 2.1475,
      "step": 9
    },
    {
      "epoch": 0.0005391416864351952,
      "grad_norm": 34.1890983581543,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 2.7256,
      "step": 10
    },
    {
      "epoch": 0.0005930558550787146,
      "grad_norm": 19.263437271118164,
      "learning_rate": 1.1e-06,
      "loss": 2.4132,
      "step": 11
    },
    {
      "epoch": 0.0006469700237222342,
      "grad_norm": 15.612638473510742,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 2.0422,
      "step": 12
    },
    {
      "epoch": 0.0007008841923657537,
      "grad_norm": 13.81751537322998,
      "learning_rate": 1.3e-06,
      "loss": 1.9663,
      "step": 13
    },
    {
      "epoch": 0.0007547983610092732,
      "grad_norm": 16.390897750854492,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 2.1135,
      "step": 14
    },
    {
      "epoch": 0.0008087125296527927,
      "grad_norm": 21.830646514892578,
      "learning_rate": 1.5e-06,
      "loss": 2.217,
      "step": 15
    },
    {
      "epoch": 0.0008626266982963123,
      "grad_norm": 18.630046844482422,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 2.1612,
      "step": 16
    },
    {
      "epoch": 0.0009165408669398317,
      "grad_norm": 12.403571128845215,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 1.9358,
      "step": 17
    },
    {
      "epoch": 0.0009704550355833513,
      "grad_norm": 7.713366508483887,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 1.8522,
      "step": 18
    },
    {
      "epoch": 0.001024369204226871,
      "grad_norm": 7.731616973876953,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 1.7984,
      "step": 19
    },
    {
      "epoch": 0.0010782833728703904,
      "grad_norm": 7.5799174308776855,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.701,
      "step": 20
    },
    {
      "epoch": 0.0011321975415139098,
      "grad_norm": 5.5428080558776855,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 1.624,
      "step": 21
    },
    {
      "epoch": 0.0011861117101574293,
      "grad_norm": 5.851474285125732,
      "learning_rate": 2.2e-06,
      "loss": 1.8064,
      "step": 22
    },
    {
      "epoch": 0.001240025878800949,
      "grad_norm": 5.243111610412598,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 1.7246,
      "step": 23
    },
    {
      "epoch": 0.0012939400474444684,
      "grad_norm": 4.835971832275391,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 1.763,
      "step": 24
    },
    {
      "epoch": 0.0013478542160879879,
      "grad_norm": 4.127845287322998,
      "learning_rate": 2.5e-06,
      "loss": 1.5869,
      "step": 25
    },
    {
      "epoch": 0.0014017683847315074,
      "grad_norm": 3.7648322582244873,
      "learning_rate": 2.6e-06,
      "loss": 1.5599,
      "step": 26
    },
    {
      "epoch": 0.001455682553375027,
      "grad_norm": 3.5424962043762207,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 1.4703,
      "step": 27
    },
    {
      "epoch": 0.0015095967220185465,
      "grad_norm": 3.3707985877990723,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 1.354,
      "step": 28
    },
    {
      "epoch": 0.001563510890662066,
      "grad_norm": 4.71254825592041,
      "learning_rate": 2.9e-06,
      "loss": 1.8162,
      "step": 29
    },
    {
      "epoch": 0.0016174250593055854,
      "grad_norm": 3.7660300731658936,
      "learning_rate": 3e-06,
      "loss": 1.5951,
      "step": 30
    },
    {
      "epoch": 0.001671339227949105,
      "grad_norm": 3.4810571670532227,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 1.5183,
      "step": 31
    },
    {
      "epoch": 0.0017252533965926246,
      "grad_norm": 3.672693967819214,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 1.6374,
      "step": 32
    },
    {
      "epoch": 0.001779167565236144,
      "grad_norm": 3.3589682579040527,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 1.4371,
      "step": 33
    },
    {
      "epoch": 0.0018330817338796635,
      "grad_norm": 3.6365807056427,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 1.595,
      "step": 34
    },
    {
      "epoch": 0.0018869959025231832,
      "grad_norm": 3.6467039585113525,
      "learning_rate": 3.5e-06,
      "loss": 1.5714,
      "step": 35
    },
    {
      "epoch": 0.0019409100711667026,
      "grad_norm": 3.4684648513793945,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 1.4897,
      "step": 36
    },
    {
      "epoch": 0.001994824239810222,
      "grad_norm": 3.70845627784729,
      "learning_rate": 3.7e-06,
      "loss": 1.5954,
      "step": 37
    },
    {
      "epoch": 0.002048738408453742,
      "grad_norm": 3.1803395748138428,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 1.3976,
      "step": 38
    },
    {
      "epoch": 0.002102652577097261,
      "grad_norm": 2.851703405380249,
      "learning_rate": 3.900000000000001e-06,
      "loss": 1.1894,
      "step": 39
    },
    {
      "epoch": 0.0021565667457407807,
      "grad_norm": 2.832003593444824,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.353,
      "step": 40
    },
    {
      "epoch": 0.0022104809143843004,
      "grad_norm": 3.397498607635498,
      "learning_rate": 4.1e-06,
      "loss": 1.4541,
      "step": 41
    },
    {
      "epoch": 0.0022643950830278196,
      "grad_norm": 3.4537954330444336,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 1.4475,
      "step": 42
    },
    {
      "epoch": 0.0023183092516713393,
      "grad_norm": 3.1131632328033447,
      "learning_rate": 4.3e-06,
      "loss": 1.2707,
      "step": 43
    },
    {
      "epoch": 0.0023722234203148586,
      "grad_norm": 3.0421881675720215,
      "learning_rate": 4.4e-06,
      "loss": 1.3418,
      "step": 44
    },
    {
      "epoch": 0.0024261375889583782,
      "grad_norm": 3.528514862060547,
      "learning_rate": 4.5e-06,
      "loss": 1.4432,
      "step": 45
    },
    {
      "epoch": 0.002480051757601898,
      "grad_norm": 3.6783225536346436,
      "learning_rate": 4.600000000000001e-06,
      "loss": 1.4863,
      "step": 46
    },
    {
      "epoch": 0.002533965926245417,
      "grad_norm": 2.9829189777374268,
      "learning_rate": 4.7e-06,
      "loss": 1.2856,
      "step": 47
    },
    {
      "epoch": 0.002587880094888937,
      "grad_norm": 3.4480350017547607,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.4129,
      "step": 48
    },
    {
      "epoch": 0.0026417942635324565,
      "grad_norm": 3.4247214794158936,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 1.3467,
      "step": 49
    },
    {
      "epoch": 0.0026957084321759758,
      "grad_norm": 3.5268948078155518,
      "learning_rate": 5e-06,
      "loss": 1.4795,
      "step": 50
    },
    {
      "epoch": 0.0027496226008194955,
      "grad_norm": 3.3228304386138916,
      "learning_rate": 5.1e-06,
      "loss": 1.461,
      "step": 51
    },
    {
      "epoch": 0.0028035367694630147,
      "grad_norm": 3.365630865097046,
      "learning_rate": 5.2e-06,
      "loss": 1.2947,
      "step": 52
    },
    {
      "epoch": 0.0028574509381065344,
      "grad_norm": 3.4889328479766846,
      "learning_rate": 5.300000000000001e-06,
      "loss": 1.432,
      "step": 53
    },
    {
      "epoch": 0.002911365106750054,
      "grad_norm": 3.5767273902893066,
      "learning_rate": 5.400000000000001e-06,
      "loss": 1.3773,
      "step": 54
    },
    {
      "epoch": 0.0029652792753935733,
      "grad_norm": 3.499298095703125,
      "learning_rate": 5.500000000000001e-06,
      "loss": 1.4132,
      "step": 55
    },
    {
      "epoch": 0.003019193444037093,
      "grad_norm": 3.6990244388580322,
      "learning_rate": 5.600000000000001e-06,
      "loss": 1.4595,
      "step": 56
    },
    {
      "epoch": 0.0030731076126806127,
      "grad_norm": 3.0908327102661133,
      "learning_rate": 5.7e-06,
      "loss": 1.1873,
      "step": 57
    },
    {
      "epoch": 0.003127021781324132,
      "grad_norm": 3.149425745010376,
      "learning_rate": 5.8e-06,
      "loss": 1.3306,
      "step": 58
    },
    {
      "epoch": 0.0031809359499676516,
      "grad_norm": 3.193023204803467,
      "learning_rate": 5.9e-06,
      "loss": 1.3326,
      "step": 59
    },
    {
      "epoch": 0.003234850118611171,
      "grad_norm": 3.610344409942627,
      "learning_rate": 6e-06,
      "loss": 1.4527,
      "step": 60
    },
    {
      "epoch": 0.0032887642872546905,
      "grad_norm": 2.9877095222473145,
      "learning_rate": 6.1e-06,
      "loss": 1.2029,
      "step": 61
    },
    {
      "epoch": 0.00334267845589821,
      "grad_norm": 3.0241923332214355,
      "learning_rate": 6.200000000000001e-06,
      "loss": 1.3413,
      "step": 62
    },
    {
      "epoch": 0.0033965926245417295,
      "grad_norm": 3.212700366973877,
      "learning_rate": 6.300000000000001e-06,
      "loss": 1.3471,
      "step": 63
    },
    {
      "epoch": 0.003450506793185249,
      "grad_norm": 2.7138960361480713,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 1.0885,
      "step": 64
    },
    {
      "epoch": 0.0035044209618287684,
      "grad_norm": 2.5690340995788574,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 1.1168,
      "step": 65
    },
    {
      "epoch": 0.003558335130472288,
      "grad_norm": 3.0344784259796143,
      "learning_rate": 6.600000000000001e-06,
      "loss": 1.2828,
      "step": 66
    },
    {
      "epoch": 0.0036122492991158077,
      "grad_norm": 3.0589816570281982,
      "learning_rate": 6.700000000000001e-06,
      "loss": 1.2604,
      "step": 67
    },
    {
      "epoch": 0.003666163467759327,
      "grad_norm": 2.676417112350464,
      "learning_rate": 6.800000000000001e-06,
      "loss": 1.1679,
      "step": 68
    },
    {
      "epoch": 0.0037200776364028467,
      "grad_norm": 2.6590960025787354,
      "learning_rate": 6.9e-06,
      "loss": 1.2283,
      "step": 69
    },
    {
      "epoch": 0.0037739918050463664,
      "grad_norm": 2.6973354816436768,
      "learning_rate": 7e-06,
      "loss": 1.2028,
      "step": 70
    },
    {
      "epoch": 0.0038279059736898856,
      "grad_norm": 2.7046608924865723,
      "learning_rate": 7.100000000000001e-06,
      "loss": 1.2629,
      "step": 71
    },
    {
      "epoch": 0.0038818201423334053,
      "grad_norm": 2.2172696590423584,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 1.1367,
      "step": 72
    },
    {
      "epoch": 0.0039357343109769245,
      "grad_norm": 2.6138789653778076,
      "learning_rate": 7.3e-06,
      "loss": 1.3167,
      "step": 73
    },
    {
      "epoch": 0.003989648479620444,
      "grad_norm": 2.2926838397979736,
      "learning_rate": 7.4e-06,
      "loss": 1.2909,
      "step": 74
    },
    {
      "epoch": 0.004043562648263964,
      "grad_norm": 2.0647220611572266,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.2054,
      "step": 75
    },
    {
      "epoch": 0.004097476816907484,
      "grad_norm": 2.1190452575683594,
      "learning_rate": 7.600000000000001e-06,
      "loss": 1.1497,
      "step": 76
    },
    {
      "epoch": 0.004151390985551002,
      "grad_norm": 1.9973243474960327,
      "learning_rate": 7.7e-06,
      "loss": 1.1997,
      "step": 77
    },
    {
      "epoch": 0.004205305154194522,
      "grad_norm": 2.11751651763916,
      "learning_rate": 7.800000000000002e-06,
      "loss": 1.2181,
      "step": 78
    },
    {
      "epoch": 0.004259219322838042,
      "grad_norm": 1.8975950479507446,
      "learning_rate": 7.9e-06,
      "loss": 1.1582,
      "step": 79
    },
    {
      "epoch": 0.004313133491481561,
      "grad_norm": 1.8368147611618042,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.1389,
      "step": 80
    },
    {
      "epoch": 0.004367047660125081,
      "grad_norm": 1.7472988367080688,
      "learning_rate": 8.1e-06,
      "loss": 1.0959,
      "step": 81
    },
    {
      "epoch": 0.004420961828768601,
      "grad_norm": 1.7325443029403687,
      "learning_rate": 8.2e-06,
      "loss": 1.1847,
      "step": 82
    },
    {
      "epoch": 0.00447487599741212,
      "grad_norm": 1.6171561479568481,
      "learning_rate": 8.3e-06,
      "loss": 0.9834,
      "step": 83
    },
    {
      "epoch": 0.004528790166055639,
      "grad_norm": 1.6583327054977417,
      "learning_rate": 8.400000000000001e-06,
      "loss": 1.0413,
      "step": 84
    },
    {
      "epoch": 0.004582704334699159,
      "grad_norm": 1.8914967775344849,
      "learning_rate": 8.5e-06,
      "loss": 1.2413,
      "step": 85
    },
    {
      "epoch": 0.004636618503342679,
      "grad_norm": 1.6018317937850952,
      "learning_rate": 8.6e-06,
      "loss": 1.0577,
      "step": 86
    },
    {
      "epoch": 0.004690532671986198,
      "grad_norm": 1.9170053005218506,
      "learning_rate": 8.700000000000001e-06,
      "loss": 1.2463,
      "step": 87
    },
    {
      "epoch": 0.004744446840629717,
      "grad_norm": 1.666536569595337,
      "learning_rate": 8.8e-06,
      "loss": 1.0532,
      "step": 88
    },
    {
      "epoch": 0.004798361009273237,
      "grad_norm": 1.660115361213684,
      "learning_rate": 8.900000000000001e-06,
      "loss": 1.0514,
      "step": 89
    },
    {
      "epoch": 0.0048522751779167565,
      "grad_norm": 1.8667477369308472,
      "learning_rate": 9e-06,
      "loss": 1.2039,
      "step": 90
    },
    {
      "epoch": 0.004906189346560276,
      "grad_norm": 1.9490039348602295,
      "learning_rate": 9.100000000000001e-06,
      "loss": 1.1804,
      "step": 91
    },
    {
      "epoch": 0.004960103515203796,
      "grad_norm": 1.8415377140045166,
      "learning_rate": 9.200000000000002e-06,
      "loss": 1.1435,
      "step": 92
    },
    {
      "epoch": 0.005014017683847315,
      "grad_norm": 1.8571438789367676,
      "learning_rate": 9.3e-06,
      "loss": 1.0974,
      "step": 93
    },
    {
      "epoch": 0.005067931852490834,
      "grad_norm": 1.8480113744735718,
      "learning_rate": 9.4e-06,
      "loss": 1.149,
      "step": 94
    },
    {
      "epoch": 0.005121846021134354,
      "grad_norm": 2.003490447998047,
      "learning_rate": 9.5e-06,
      "loss": 1.1954,
      "step": 95
    },
    {
      "epoch": 0.005175760189777874,
      "grad_norm": 1.8002668619155884,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.9953,
      "step": 96
    },
    {
      "epoch": 0.005229674358421393,
      "grad_norm": 1.9040817022323608,
      "learning_rate": 9.7e-06,
      "loss": 1.1195,
      "step": 97
    },
    {
      "epoch": 0.005283588527064913,
      "grad_norm": 1.8311433792114258,
      "learning_rate": 9.800000000000001e-06,
      "loss": 1.083,
      "step": 98
    },
    {
      "epoch": 0.005337502695708432,
      "grad_norm": 1.9509624242782593,
      "learning_rate": 9.9e-06,
      "loss": 1.176,
      "step": 99
    },
    {
      "epoch": 0.0053914168643519516,
      "grad_norm": 2.0624589920043945,
      "learning_rate": 1e-05,
      "loss": 1.119,
      "step": 100
    },
    {
      "epoch": 0.005445331032995471,
      "grad_norm": 1.9618796110153198,
      "learning_rate": 9.999999995505339e-06,
      "loss": 1.1371,
      "step": 101
    },
    {
      "epoch": 0.005499245201638991,
      "grad_norm": 1.946245551109314,
      "learning_rate": 9.999999982021349e-06,
      "loss": 0.9736,
      "step": 102
    },
    {
      "epoch": 0.005553159370282511,
      "grad_norm": 1.9871301651000977,
      "learning_rate": 9.999999959548035e-06,
      "loss": 1.1077,
      "step": 103
    },
    {
      "epoch": 0.005607073538926029,
      "grad_norm": 1.86216402053833,
      "learning_rate": 9.999999928085396e-06,
      "loss": 1.0882,
      "step": 104
    },
    {
      "epoch": 0.005660987707569549,
      "grad_norm": 1.8447723388671875,
      "learning_rate": 9.999999887633432e-06,
      "loss": 1.0344,
      "step": 105
    },
    {
      "epoch": 0.005714901876213069,
      "grad_norm": 1.8345638513565063,
      "learning_rate": 9.99999983819214e-06,
      "loss": 1.1077,
      "step": 106
    },
    {
      "epoch": 0.0057688160448565885,
      "grad_norm": 1.8410178422927856,
      "learning_rate": 9.999999779761524e-06,
      "loss": 1.0824,
      "step": 107
    },
    {
      "epoch": 0.005822730213500108,
      "grad_norm": 1.5881969928741455,
      "learning_rate": 9.999999712341583e-06,
      "loss": 0.9439,
      "step": 108
    },
    {
      "epoch": 0.005876644382143627,
      "grad_norm": 1.6704047918319702,
      "learning_rate": 9.999999635932316e-06,
      "loss": 1.033,
      "step": 109
    },
    {
      "epoch": 0.005930558550787147,
      "grad_norm": 1.792449712753296,
      "learning_rate": 9.999999550533726e-06,
      "loss": 1.0279,
      "step": 110
    },
    {
      "epoch": 0.005984472719430666,
      "grad_norm": 1.6515668630599976,
      "learning_rate": 9.999999456145809e-06,
      "loss": 1.0301,
      "step": 111
    },
    {
      "epoch": 0.006038386888074186,
      "grad_norm": 1.8541395664215088,
      "learning_rate": 9.999999352768568e-06,
      "loss": 1.1057,
      "step": 112
    },
    {
      "epoch": 0.006092301056717706,
      "grad_norm": 1.6490236520767212,
      "learning_rate": 9.999999240402002e-06,
      "loss": 1.0523,
      "step": 113
    },
    {
      "epoch": 0.006146215225361225,
      "grad_norm": 1.655333161354065,
      "learning_rate": 9.999999119046113e-06,
      "loss": 1.0448,
      "step": 114
    },
    {
      "epoch": 0.006200129394004744,
      "grad_norm": 1.5721609592437744,
      "learning_rate": 9.999998988700899e-06,
      "loss": 0.9883,
      "step": 115
    },
    {
      "epoch": 0.006254043562648264,
      "grad_norm": 1.6411349773406982,
      "learning_rate": 9.99999884936636e-06,
      "loss": 1.0255,
      "step": 116
    },
    {
      "epoch": 0.0063079577312917835,
      "grad_norm": 1.6399502754211426,
      "learning_rate": 9.999998701042501e-06,
      "loss": 1.0146,
      "step": 117
    },
    {
      "epoch": 0.006361871899935303,
      "grad_norm": 1.615026831626892,
      "learning_rate": 9.999998543729316e-06,
      "loss": 1.0022,
      "step": 118
    },
    {
      "epoch": 0.006415786068578823,
      "grad_norm": 1.4867664575576782,
      "learning_rate": 9.99999837742681e-06,
      "loss": 1.0164,
      "step": 119
    },
    {
      "epoch": 0.006469700237222342,
      "grad_norm": 1.540153980255127,
      "learning_rate": 9.999998202134979e-06,
      "loss": 0.989,
      "step": 120
    },
    {
      "epoch": 0.006523614405865861,
      "grad_norm": 1.5535691976547241,
      "learning_rate": 9.999998017853825e-06,
      "loss": 0.9942,
      "step": 121
    },
    {
      "epoch": 0.006577528574509381,
      "grad_norm": 1.4892929792404175,
      "learning_rate": 9.999997824583351e-06,
      "loss": 1.0537,
      "step": 122
    },
    {
      "epoch": 0.006631442743152901,
      "grad_norm": 1.4674094915390015,
      "learning_rate": 9.999997622323554e-06,
      "loss": 1.0239,
      "step": 123
    },
    {
      "epoch": 0.00668535691179642,
      "grad_norm": 1.394027590751648,
      "learning_rate": 9.999997411074436e-06,
      "loss": 0.9781,
      "step": 124
    },
    {
      "epoch": 0.006739271080439939,
      "grad_norm": 1.372728705406189,
      "learning_rate": 9.999997190835999e-06,
      "loss": 1.0433,
      "step": 125
    },
    {
      "epoch": 0.006793185249083459,
      "grad_norm": 1.2535908222198486,
      "learning_rate": 9.999996961608238e-06,
      "loss": 0.958,
      "step": 126
    },
    {
      "epoch": 0.006847099417726979,
      "grad_norm": 1.337633490562439,
      "learning_rate": 9.999996723391158e-06,
      "loss": 1.0213,
      "step": 127
    },
    {
      "epoch": 0.006901013586370498,
      "grad_norm": 1.3640319108963013,
      "learning_rate": 9.999996476184759e-06,
      "loss": 1.0432,
      "step": 128
    },
    {
      "epoch": 0.006954927755014018,
      "grad_norm": 1.2663391828536987,
      "learning_rate": 9.99999621998904e-06,
      "loss": 1.0154,
      "step": 129
    },
    {
      "epoch": 0.007008841923657537,
      "grad_norm": 1.450737476348877,
      "learning_rate": 9.999995954804004e-06,
      "loss": 1.0074,
      "step": 130
    },
    {
      "epoch": 0.0070627560923010565,
      "grad_norm": 1.2757987976074219,
      "learning_rate": 9.999995680629649e-06,
      "loss": 0.9996,
      "step": 131
    },
    {
      "epoch": 0.007116670260944576,
      "grad_norm": 1.3978132009506226,
      "learning_rate": 9.999995397465974e-06,
      "loss": 1.04,
      "step": 132
    },
    {
      "epoch": 0.007170584429588096,
      "grad_norm": 1.3167297840118408,
      "learning_rate": 9.999995105312982e-06,
      "loss": 1.0069,
      "step": 133
    },
    {
      "epoch": 0.0072244985982316155,
      "grad_norm": 1.1626744270324707,
      "learning_rate": 9.999994804170674e-06,
      "loss": 0.9722,
      "step": 134
    },
    {
      "epoch": 0.007278412766875135,
      "grad_norm": 1.354797601699829,
      "learning_rate": 9.99999449403905e-06,
      "loss": 0.9019,
      "step": 135
    },
    {
      "epoch": 0.007332326935518654,
      "grad_norm": 1.2605732679367065,
      "learning_rate": 9.99999417491811e-06,
      "loss": 1.0038,
      "step": 136
    },
    {
      "epoch": 0.007386241104162174,
      "grad_norm": 1.3804657459259033,
      "learning_rate": 9.999993846807855e-06,
      "loss": 1.0139,
      "step": 137
    },
    {
      "epoch": 0.007440155272805693,
      "grad_norm": 1.3001742362976074,
      "learning_rate": 9.999993509708286e-06,
      "loss": 1.1436,
      "step": 138
    },
    {
      "epoch": 0.007494069441449213,
      "grad_norm": 1.2776422500610352,
      "learning_rate": 9.999993163619401e-06,
      "loss": 0.9792,
      "step": 139
    },
    {
      "epoch": 0.007547983610092733,
      "grad_norm": 1.2149187326431274,
      "learning_rate": 9.999992808541204e-06,
      "loss": 0.963,
      "step": 140
    },
    {
      "epoch": 0.0076018977787362515,
      "grad_norm": 1.341806173324585,
      "learning_rate": 9.999992444473694e-06,
      "loss": 0.9639,
      "step": 141
    },
    {
      "epoch": 0.007655811947379771,
      "grad_norm": 1.2565757036209106,
      "learning_rate": 9.999992071416874e-06,
      "loss": 0.9193,
      "step": 142
    },
    {
      "epoch": 0.007709726116023291,
      "grad_norm": 1.3059918880462646,
      "learning_rate": 9.99999168937074e-06,
      "loss": 0.9632,
      "step": 143
    },
    {
      "epoch": 0.0077636402846668106,
      "grad_norm": 1.1719332933425903,
      "learning_rate": 9.999991298335295e-06,
      "loss": 0.9687,
      "step": 144
    },
    {
      "epoch": 0.00781755445331033,
      "grad_norm": 1.125950813293457,
      "learning_rate": 9.999990898310542e-06,
      "loss": 0.968,
      "step": 145
    },
    {
      "epoch": 0.007871468621953849,
      "grad_norm": 1.2400416135787964,
      "learning_rate": 9.999990489296478e-06,
      "loss": 0.972,
      "step": 146
    },
    {
      "epoch": 0.007925382790597369,
      "grad_norm": 1.172117829322815,
      "learning_rate": 9.999990071293106e-06,
      "loss": 0.9243,
      "step": 147
    },
    {
      "epoch": 0.007979296959240888,
      "grad_norm": 1.240317463874817,
      "learning_rate": 9.999989644300427e-06,
      "loss": 1.0655,
      "step": 148
    },
    {
      "epoch": 0.008033211127884408,
      "grad_norm": 1.1535708904266357,
      "learning_rate": 9.999989208318438e-06,
      "loss": 0.9871,
      "step": 149
    },
    {
      "epoch": 0.008087125296527928,
      "grad_norm": 1.2711198329925537,
      "learning_rate": 9.999988763347145e-06,
      "loss": 1.0307,
      "step": 150
    },
    {
      "epoch": 0.008141039465171447,
      "grad_norm": 1.2345954179763794,
      "learning_rate": 9.999988309386548e-06,
      "loss": 1.1343,
      "step": 151
    },
    {
      "epoch": 0.008194953633814967,
      "grad_norm": 1.2489601373672485,
      "learning_rate": 9.999987846436645e-06,
      "loss": 1.0303,
      "step": 152
    },
    {
      "epoch": 0.008248867802458487,
      "grad_norm": 1.264240026473999,
      "learning_rate": 9.999987374497439e-06,
      "loss": 0.9562,
      "step": 153
    },
    {
      "epoch": 0.008302781971102005,
      "grad_norm": 1.2613575458526611,
      "learning_rate": 9.99998689356893e-06,
      "loss": 0.954,
      "step": 154
    },
    {
      "epoch": 0.008356696139745524,
      "grad_norm": 1.2091072797775269,
      "learning_rate": 9.999986403651116e-06,
      "loss": 1.0734,
      "step": 155
    },
    {
      "epoch": 0.008410610308389044,
      "grad_norm": 1.18421471118927,
      "learning_rate": 9.999985904744002e-06,
      "loss": 0.9167,
      "step": 156
    },
    {
      "epoch": 0.008464524477032564,
      "grad_norm": 1.0399659872055054,
      "learning_rate": 9.99998539684759e-06,
      "loss": 0.9068,
      "step": 157
    },
    {
      "epoch": 0.008518438645676083,
      "grad_norm": 1.1292288303375244,
      "learning_rate": 9.999984879961877e-06,
      "loss": 1.0027,
      "step": 158
    },
    {
      "epoch": 0.008572352814319603,
      "grad_norm": 1.2592105865478516,
      "learning_rate": 9.999984354086867e-06,
      "loss": 1.0794,
      "step": 159
    },
    {
      "epoch": 0.008626266982963123,
      "grad_norm": 1.1646504402160645,
      "learning_rate": 9.999983819222558e-06,
      "loss": 1.0468,
      "step": 160
    },
    {
      "epoch": 0.008680181151606643,
      "grad_norm": 1.156711220741272,
      "learning_rate": 9.999983275368952e-06,
      "loss": 0.9053,
      "step": 161
    },
    {
      "epoch": 0.008734095320250162,
      "grad_norm": 1.1169341802597046,
      "learning_rate": 9.999982722526051e-06,
      "loss": 0.97,
      "step": 162
    },
    {
      "epoch": 0.008788009488893682,
      "grad_norm": 1.3474149703979492,
      "learning_rate": 9.999982160693856e-06,
      "loss": 1.0221,
      "step": 163
    },
    {
      "epoch": 0.008841923657537202,
      "grad_norm": 1.2021468877792358,
      "learning_rate": 9.999981589872368e-06,
      "loss": 0.9303,
      "step": 164
    },
    {
      "epoch": 0.00889583782618072,
      "grad_norm": 1.0625534057617188,
      "learning_rate": 9.999981010061586e-06,
      "loss": 0.8765,
      "step": 165
    },
    {
      "epoch": 0.00894975199482424,
      "grad_norm": 1.2688498497009277,
      "learning_rate": 9.999980421261512e-06,
      "loss": 1.0163,
      "step": 166
    },
    {
      "epoch": 0.009003666163467759,
      "grad_norm": 1.122948408126831,
      "learning_rate": 9.999979823472148e-06,
      "loss": 0.9953,
      "step": 167
    },
    {
      "epoch": 0.009057580332111279,
      "grad_norm": 1.1817872524261475,
      "learning_rate": 9.999979216693495e-06,
      "loss": 1.0774,
      "step": 168
    },
    {
      "epoch": 0.009111494500754798,
      "grad_norm": 1.1483280658721924,
      "learning_rate": 9.999978600925553e-06,
      "loss": 1.0105,
      "step": 169
    },
    {
      "epoch": 0.009165408669398318,
      "grad_norm": 1.4039335250854492,
      "learning_rate": 9.999977976168325e-06,
      "loss": 0.944,
      "step": 170
    },
    {
      "epoch": 0.009219322838041838,
      "grad_norm": 1.1459723711013794,
      "learning_rate": 9.999977342421812e-06,
      "loss": 0.9208,
      "step": 171
    },
    {
      "epoch": 0.009273237006685357,
      "grad_norm": 1.0897774696350098,
      "learning_rate": 9.999976699686011e-06,
      "loss": 0.8719,
      "step": 172
    },
    {
      "epoch": 0.009327151175328877,
      "grad_norm": 1.206467866897583,
      "learning_rate": 9.999976047960928e-06,
      "loss": 1.0645,
      "step": 173
    },
    {
      "epoch": 0.009381065343972397,
      "grad_norm": 1.004550814628601,
      "learning_rate": 9.999975387246563e-06,
      "loss": 0.9317,
      "step": 174
    },
    {
      "epoch": 0.009434979512615916,
      "grad_norm": 1.2359992265701294,
      "learning_rate": 9.999974717542916e-06,
      "loss": 1.1136,
      "step": 175
    },
    {
      "epoch": 0.009488893681259434,
      "grad_norm": 1.1922352313995361,
      "learning_rate": 9.999974038849989e-06,
      "loss": 1.0307,
      "step": 176
    },
    {
      "epoch": 0.009542807849902954,
      "grad_norm": 1.1597613096237183,
      "learning_rate": 9.999973351167782e-06,
      "loss": 1.0275,
      "step": 177
    },
    {
      "epoch": 0.009596722018546474,
      "grad_norm": 1.172133445739746,
      "learning_rate": 9.999972654496298e-06,
      "loss": 0.9269,
      "step": 178
    },
    {
      "epoch": 0.009650636187189993,
      "grad_norm": 1.1879733800888062,
      "learning_rate": 9.999971948835538e-06,
      "loss": 0.9547,
      "step": 179
    },
    {
      "epoch": 0.009704550355833513,
      "grad_norm": 1.0029833316802979,
      "learning_rate": 9.999971234185502e-06,
      "loss": 0.8994,
      "step": 180
    },
    {
      "epoch": 0.009758464524477033,
      "grad_norm": 1.0769891738891602,
      "learning_rate": 9.999970510546194e-06,
      "loss": 0.9107,
      "step": 181
    },
    {
      "epoch": 0.009812378693120552,
      "grad_norm": 1.3288064002990723,
      "learning_rate": 9.99996977791761e-06,
      "loss": 1.0116,
      "step": 182
    },
    {
      "epoch": 0.009866292861764072,
      "grad_norm": 1.142452597618103,
      "learning_rate": 9.999969036299757e-06,
      "loss": 0.9367,
      "step": 183
    },
    {
      "epoch": 0.009920207030407592,
      "grad_norm": 1.2458518743515015,
      "learning_rate": 9.999968285692632e-06,
      "loss": 1.1398,
      "step": 184
    },
    {
      "epoch": 0.009974121199051111,
      "grad_norm": 1.3373422622680664,
      "learning_rate": 9.99996752609624e-06,
      "loss": 0.959,
      "step": 185
    },
    {
      "epoch": 0.01002803536769463,
      "grad_norm": 1.2288920879364014,
      "learning_rate": 9.99996675751058e-06,
      "loss": 0.9908,
      "step": 186
    },
    {
      "epoch": 0.010081949536338149,
      "grad_norm": 1.1954001188278198,
      "learning_rate": 9.999965979935656e-06,
      "loss": 0.9332,
      "step": 187
    },
    {
      "epoch": 0.010135863704981669,
      "grad_norm": 1.171021819114685,
      "learning_rate": 9.999965193371466e-06,
      "loss": 0.9119,
      "step": 188
    },
    {
      "epoch": 0.010189777873625188,
      "grad_norm": 1.025169014930725,
      "learning_rate": 9.999964397818013e-06,
      "loss": 0.784,
      "step": 189
    },
    {
      "epoch": 0.010243692042268708,
      "grad_norm": 1.1340326070785522,
      "learning_rate": 9.999963593275298e-06,
      "loss": 1.0036,
      "step": 190
    },
    {
      "epoch": 0.010297606210912228,
      "grad_norm": 1.0302847623825073,
      "learning_rate": 9.999962779743324e-06,
      "loss": 0.8293,
      "step": 191
    },
    {
      "epoch": 0.010351520379555747,
      "grad_norm": 1.2410109043121338,
      "learning_rate": 9.99996195722209e-06,
      "loss": 0.9507,
      "step": 192
    },
    {
      "epoch": 0.010405434548199267,
      "grad_norm": 1.2054308652877808,
      "learning_rate": 9.9999611257116e-06,
      "loss": 0.9356,
      "step": 193
    },
    {
      "epoch": 0.010459348716842787,
      "grad_norm": 1.2046679258346558,
      "learning_rate": 9.999960285211853e-06,
      "loss": 1.0638,
      "step": 194
    },
    {
      "epoch": 0.010513262885486306,
      "grad_norm": 1.4594306945800781,
      "learning_rate": 9.999959435722852e-06,
      "loss": 0.9624,
      "step": 195
    },
    {
      "epoch": 0.010567177054129826,
      "grad_norm": 1.0909247398376465,
      "learning_rate": 9.999958577244598e-06,
      "loss": 0.9503,
      "step": 196
    },
    {
      "epoch": 0.010621091222773344,
      "grad_norm": 1.1524754762649536,
      "learning_rate": 9.999957709777094e-06,
      "loss": 0.8954,
      "step": 197
    },
    {
      "epoch": 0.010675005391416864,
      "grad_norm": 1.4128906726837158,
      "learning_rate": 9.99995683332034e-06,
      "loss": 0.8903,
      "step": 198
    },
    {
      "epoch": 0.010728919560060383,
      "grad_norm": 1.1304652690887451,
      "learning_rate": 9.999955947874338e-06,
      "loss": 0.9247,
      "step": 199
    },
    {
      "epoch": 0.010782833728703903,
      "grad_norm": 1.2978957891464233,
      "learning_rate": 9.99995505343909e-06,
      "loss": 0.9473,
      "step": 200
    },
    {
      "epoch": 0.010836747897347423,
      "grad_norm": 1.0742554664611816,
      "learning_rate": 9.999954150014595e-06,
      "loss": 0.9626,
      "step": 201
    },
    {
      "epoch": 0.010890662065990942,
      "grad_norm": 1.0707745552062988,
      "learning_rate": 9.999953237600859e-06,
      "loss": 0.8721,
      "step": 202
    },
    {
      "epoch": 0.010944576234634462,
      "grad_norm": 1.17974853515625,
      "learning_rate": 9.99995231619788e-06,
      "loss": 1.0059,
      "step": 203
    },
    {
      "epoch": 0.010998490403277982,
      "grad_norm": 1.0108370780944824,
      "learning_rate": 9.999951385805662e-06,
      "loss": 0.9527,
      "step": 204
    },
    {
      "epoch": 0.011052404571921502,
      "grad_norm": 0.9983445405960083,
      "learning_rate": 9.999950446424204e-06,
      "loss": 0.7626,
      "step": 205
    },
    {
      "epoch": 0.011106318740565021,
      "grad_norm": 1.0860002040863037,
      "learning_rate": 9.99994949805351e-06,
      "loss": 0.9591,
      "step": 206
    },
    {
      "epoch": 0.01116023290920854,
      "grad_norm": 1.0447322130203247,
      "learning_rate": 9.999948540693584e-06,
      "loss": 0.9861,
      "step": 207
    },
    {
      "epoch": 0.011214147077852059,
      "grad_norm": 1.2582998275756836,
      "learning_rate": 9.999947574344423e-06,
      "loss": 0.8949,
      "step": 208
    },
    {
      "epoch": 0.011268061246495579,
      "grad_norm": 1.1507002115249634,
      "learning_rate": 9.99994659900603e-06,
      "loss": 0.918,
      "step": 209
    },
    {
      "epoch": 0.011321975415139098,
      "grad_norm": 1.135169267654419,
      "learning_rate": 9.999945614678408e-06,
      "loss": 0.9891,
      "step": 210
    },
    {
      "epoch": 0.011375889583782618,
      "grad_norm": 1.1746275424957275,
      "learning_rate": 9.999944621361558e-06,
      "loss": 1.0186,
      "step": 211
    },
    {
      "epoch": 0.011429803752426138,
      "grad_norm": 1.1137248277664185,
      "learning_rate": 9.999943619055483e-06,
      "loss": 0.9584,
      "step": 212
    },
    {
      "epoch": 0.011483717921069657,
      "grad_norm": 1.336651086807251,
      "learning_rate": 9.999942607760182e-06,
      "loss": 1.091,
      "step": 213
    },
    {
      "epoch": 0.011537632089713177,
      "grad_norm": 1.1966856718063354,
      "learning_rate": 9.999941587475658e-06,
      "loss": 0.9761,
      "step": 214
    },
    {
      "epoch": 0.011591546258356697,
      "grad_norm": 1.0843144655227661,
      "learning_rate": 9.999940558201915e-06,
      "loss": 0.8917,
      "step": 215
    },
    {
      "epoch": 0.011645460427000216,
      "grad_norm": 1.2089293003082275,
      "learning_rate": 9.999939519938953e-06,
      "loss": 0.9704,
      "step": 216
    },
    {
      "epoch": 0.011699374595643736,
      "grad_norm": 1.2409982681274414,
      "learning_rate": 9.999938472686775e-06,
      "loss": 0.9949,
      "step": 217
    },
    {
      "epoch": 0.011753288764287254,
      "grad_norm": 1.1310094594955444,
      "learning_rate": 9.99993741644538e-06,
      "loss": 0.9666,
      "step": 218
    },
    {
      "epoch": 0.011807202932930774,
      "grad_norm": 1.120510220527649,
      "learning_rate": 9.999936351214772e-06,
      "loss": 0.8844,
      "step": 219
    },
    {
      "epoch": 0.011861117101574293,
      "grad_norm": 1.0931518077850342,
      "learning_rate": 9.999935276994954e-06,
      "loss": 0.9647,
      "step": 220
    },
    {
      "epoch": 0.011915031270217813,
      "grad_norm": 1.2821122407913208,
      "learning_rate": 9.999934193785926e-06,
      "loss": 1.0533,
      "step": 221
    },
    {
      "epoch": 0.011968945438861333,
      "grad_norm": 1.183580756187439,
      "learning_rate": 9.999933101587691e-06,
      "loss": 0.9196,
      "step": 222
    },
    {
      "epoch": 0.012022859607504852,
      "grad_norm": 1.045825719833374,
      "learning_rate": 9.99993200040025e-06,
      "loss": 0.8953,
      "step": 223
    },
    {
      "epoch": 0.012076773776148372,
      "grad_norm": 1.0963969230651855,
      "learning_rate": 9.999930890223605e-06,
      "loss": 0.9723,
      "step": 224
    },
    {
      "epoch": 0.012130687944791892,
      "grad_norm": 1.0356731414794922,
      "learning_rate": 9.999929771057761e-06,
      "loss": 1.0215,
      "step": 225
    },
    {
      "epoch": 0.012184602113435411,
      "grad_norm": 1.112277626991272,
      "learning_rate": 9.999928642902717e-06,
      "loss": 0.9886,
      "step": 226
    },
    {
      "epoch": 0.012238516282078931,
      "grad_norm": 0.9969072937965393,
      "learning_rate": 9.999927505758475e-06,
      "loss": 0.8601,
      "step": 227
    },
    {
      "epoch": 0.01229243045072245,
      "grad_norm": 1.123781442642212,
      "learning_rate": 9.999926359625036e-06,
      "loss": 0.9894,
      "step": 228
    },
    {
      "epoch": 0.012346344619365969,
      "grad_norm": 1.2122100591659546,
      "learning_rate": 9.999925204502406e-06,
      "loss": 1.0783,
      "step": 229
    },
    {
      "epoch": 0.012400258788009488,
      "grad_norm": 1.1256672143936157,
      "learning_rate": 9.999924040390584e-06,
      "loss": 0.9116,
      "step": 230
    },
    {
      "epoch": 0.012454172956653008,
      "grad_norm": 1.0646952390670776,
      "learning_rate": 9.999922867289573e-06,
      "loss": 0.8993,
      "step": 231
    },
    {
      "epoch": 0.012508087125296528,
      "grad_norm": 1.194676399230957,
      "learning_rate": 9.999921685199376e-06,
      "loss": 1.0377,
      "step": 232
    },
    {
      "epoch": 0.012562001293940047,
      "grad_norm": 1.0519152879714966,
      "learning_rate": 9.999920494119992e-06,
      "loss": 0.8283,
      "step": 233
    },
    {
      "epoch": 0.012615915462583567,
      "grad_norm": 1.243249773979187,
      "learning_rate": 9.999919294051427e-06,
      "loss": 0.9741,
      "step": 234
    },
    {
      "epoch": 0.012669829631227087,
      "grad_norm": 1.1071687936782837,
      "learning_rate": 9.999918084993681e-06,
      "loss": 1.0402,
      "step": 235
    },
    {
      "epoch": 0.012723743799870606,
      "grad_norm": 1.1224809885025024,
      "learning_rate": 9.999916866946757e-06,
      "loss": 0.8793,
      "step": 236
    },
    {
      "epoch": 0.012777657968514126,
      "grad_norm": 1.0458532571792603,
      "learning_rate": 9.999915639910656e-06,
      "loss": 0.9855,
      "step": 237
    },
    {
      "epoch": 0.012831572137157646,
      "grad_norm": 1.0610811710357666,
      "learning_rate": 9.999914403885383e-06,
      "loss": 0.8092,
      "step": 238
    },
    {
      "epoch": 0.012885486305801164,
      "grad_norm": 1.2818992137908936,
      "learning_rate": 9.999913158870936e-06,
      "loss": 1.0101,
      "step": 239
    },
    {
      "epoch": 0.012939400474444683,
      "grad_norm": 1.110400915145874,
      "learning_rate": 9.999911904867319e-06,
      "loss": 0.9782,
      "step": 240
    },
    {
      "epoch": 0.012993314643088203,
      "grad_norm": 1.3290835618972778,
      "learning_rate": 9.999910641874537e-06,
      "loss": 1.0683,
      "step": 241
    },
    {
      "epoch": 0.013047228811731723,
      "grad_norm": 1.1448980569839478,
      "learning_rate": 9.999909369892588e-06,
      "loss": 0.9223,
      "step": 242
    },
    {
      "epoch": 0.013101142980375242,
      "grad_norm": 1.1710877418518066,
      "learning_rate": 9.999908088921477e-06,
      "loss": 0.8022,
      "step": 243
    },
    {
      "epoch": 0.013155057149018762,
      "grad_norm": 1.1242793798446655,
      "learning_rate": 9.999906798961207e-06,
      "loss": 0.9238,
      "step": 244
    },
    {
      "epoch": 0.013208971317662282,
      "grad_norm": 1.0338802337646484,
      "learning_rate": 9.999905500011778e-06,
      "loss": 0.8386,
      "step": 245
    },
    {
      "epoch": 0.013262885486305801,
      "grad_norm": 1.0910224914550781,
      "learning_rate": 9.999904192073193e-06,
      "loss": 0.937,
      "step": 246
    },
    {
      "epoch": 0.013316799654949321,
      "grad_norm": 1.297788143157959,
      "learning_rate": 9.999902875145453e-06,
      "loss": 0.9054,
      "step": 247
    },
    {
      "epoch": 0.01337071382359284,
      "grad_norm": 1.1317543983459473,
      "learning_rate": 9.999901549228564e-06,
      "loss": 0.9418,
      "step": 248
    },
    {
      "epoch": 0.01342462799223636,
      "grad_norm": 1.0944132804870605,
      "learning_rate": 9.999900214322526e-06,
      "loss": 0.9445,
      "step": 249
    },
    {
      "epoch": 0.013478542160879878,
      "grad_norm": 1.4942843914031982,
      "learning_rate": 9.999898870427342e-06,
      "loss": 0.8956,
      "step": 250
    },
    {
      "epoch": 0.013532456329523398,
      "grad_norm": 1.0630019903182983,
      "learning_rate": 9.999897517543013e-06,
      "loss": 0.8381,
      "step": 251
    },
    {
      "epoch": 0.013586370498166918,
      "grad_norm": 1.65073561668396,
      "learning_rate": 9.999896155669544e-06,
      "loss": 1.0148,
      "step": 252
    },
    {
      "epoch": 0.013640284666810438,
      "grad_norm": 1.035731315612793,
      "learning_rate": 9.999894784806936e-06,
      "loss": 0.8092,
      "step": 253
    },
    {
      "epoch": 0.013694198835453957,
      "grad_norm": 1.308863639831543,
      "learning_rate": 9.99989340495519e-06,
      "loss": 0.9742,
      "step": 254
    },
    {
      "epoch": 0.013748113004097477,
      "grad_norm": 1.1512938737869263,
      "learning_rate": 9.999892016114313e-06,
      "loss": 0.8747,
      "step": 255
    },
    {
      "epoch": 0.013802027172740997,
      "grad_norm": 0.9977009296417236,
      "learning_rate": 9.9998906182843e-06,
      "loss": 0.8183,
      "step": 256
    },
    {
      "epoch": 0.013855941341384516,
      "grad_norm": 1.2228175401687622,
      "learning_rate": 9.99988921146516e-06,
      "loss": 0.9917,
      "step": 257
    },
    {
      "epoch": 0.013909855510028036,
      "grad_norm": 1.0753847360610962,
      "learning_rate": 9.999887795656896e-06,
      "loss": 1.0063,
      "step": 258
    },
    {
      "epoch": 0.013963769678671556,
      "grad_norm": 1.0010429620742798,
      "learning_rate": 9.999886370859506e-06,
      "loss": 0.9315,
      "step": 259
    },
    {
      "epoch": 0.014017683847315074,
      "grad_norm": 1.2038911581039429,
      "learning_rate": 9.999884937072995e-06,
      "loss": 0.8764,
      "step": 260
    },
    {
      "epoch": 0.014071598015958593,
      "grad_norm": 1.1268917322158813,
      "learning_rate": 9.999883494297365e-06,
      "loss": 1.0059,
      "step": 261
    },
    {
      "epoch": 0.014125512184602113,
      "grad_norm": 1.1053709983825684,
      "learning_rate": 9.999882042532619e-06,
      "loss": 0.8866,
      "step": 262
    },
    {
      "epoch": 0.014179426353245633,
      "grad_norm": 1.091145396232605,
      "learning_rate": 9.999880581778758e-06,
      "loss": 1.0415,
      "step": 263
    },
    {
      "epoch": 0.014233340521889152,
      "grad_norm": 1.0019958019256592,
      "learning_rate": 9.999879112035786e-06,
      "loss": 0.8177,
      "step": 264
    },
    {
      "epoch": 0.014287254690532672,
      "grad_norm": 1.1044156551361084,
      "learning_rate": 9.999877633303708e-06,
      "loss": 0.9508,
      "step": 265
    },
    {
      "epoch": 0.014341168859176192,
      "grad_norm": 0.9750218391418457,
      "learning_rate": 9.999876145582524e-06,
      "loss": 0.8501,
      "step": 266
    },
    {
      "epoch": 0.014395083027819711,
      "grad_norm": 1.4015804529190063,
      "learning_rate": 9.999874648872235e-06,
      "loss": 0.9491,
      "step": 267
    },
    {
      "epoch": 0.014448997196463231,
      "grad_norm": 1.066422939300537,
      "learning_rate": 9.999873143172848e-06,
      "loss": 1.0104,
      "step": 268
    },
    {
      "epoch": 0.01450291136510675,
      "grad_norm": 1.1133167743682861,
      "learning_rate": 9.99987162848436e-06,
      "loss": 1.0142,
      "step": 269
    },
    {
      "epoch": 0.01455682553375027,
      "grad_norm": 1.1259140968322754,
      "learning_rate": 9.999870104806782e-06,
      "loss": 0.9803,
      "step": 270
    },
    {
      "epoch": 0.014610739702393788,
      "grad_norm": 1.0813393592834473,
      "learning_rate": 9.999868572140108e-06,
      "loss": 0.8728,
      "step": 271
    },
    {
      "epoch": 0.014664653871037308,
      "grad_norm": 0.9939939379692078,
      "learning_rate": 9.999867030484347e-06,
      "loss": 0.8826,
      "step": 272
    },
    {
      "epoch": 0.014718568039680828,
      "grad_norm": 1.0081939697265625,
      "learning_rate": 9.999865479839499e-06,
      "loss": 0.8682,
      "step": 273
    },
    {
      "epoch": 0.014772482208324347,
      "grad_norm": 1.0190658569335938,
      "learning_rate": 9.999863920205567e-06,
      "loss": 0.9094,
      "step": 274
    },
    {
      "epoch": 0.014826396376967867,
      "grad_norm": 1.0702111721038818,
      "learning_rate": 9.999862351582553e-06,
      "loss": 0.9244,
      "step": 275
    },
    {
      "epoch": 0.014880310545611387,
      "grad_norm": 1.0891972780227661,
      "learning_rate": 9.999860773970461e-06,
      "loss": 1.0318,
      "step": 276
    },
    {
      "epoch": 0.014934224714254906,
      "grad_norm": 0.9788139462471008,
      "learning_rate": 9.999859187369294e-06,
      "loss": 0.8779,
      "step": 277
    },
    {
      "epoch": 0.014988138882898426,
      "grad_norm": 1.0678125619888306,
      "learning_rate": 9.999857591779055e-06,
      "loss": 0.8962,
      "step": 278
    },
    {
      "epoch": 0.015042053051541946,
      "grad_norm": 0.9882293343544006,
      "learning_rate": 9.999855987199747e-06,
      "loss": 0.9082,
      "step": 279
    },
    {
      "epoch": 0.015095967220185465,
      "grad_norm": 0.9987571835517883,
      "learning_rate": 9.999854373631371e-06,
      "loss": 0.9708,
      "step": 280
    },
    {
      "epoch": 0.015149881388828985,
      "grad_norm": 1.0238722562789917,
      "learning_rate": 9.99985275107393e-06,
      "loss": 0.9461,
      "step": 281
    },
    {
      "epoch": 0.015203795557472503,
      "grad_norm": 0.9628013372421265,
      "learning_rate": 9.999851119527431e-06,
      "loss": 0.9412,
      "step": 282
    },
    {
      "epoch": 0.015257709726116023,
      "grad_norm": 1.0021862983703613,
      "learning_rate": 9.999849478991873e-06,
      "loss": 0.8461,
      "step": 283
    },
    {
      "epoch": 0.015311623894759542,
      "grad_norm": 0.9776142239570618,
      "learning_rate": 9.99984782946726e-06,
      "loss": 0.962,
      "step": 284
    },
    {
      "epoch": 0.015365538063403062,
      "grad_norm": 1.0114799737930298,
      "learning_rate": 9.999846170953593e-06,
      "loss": 0.8732,
      "step": 285
    },
    {
      "epoch": 0.015419452232046582,
      "grad_norm": 0.9860401749610901,
      "learning_rate": 9.999844503450879e-06,
      "loss": 0.8204,
      "step": 286
    },
    {
      "epoch": 0.015473366400690101,
      "grad_norm": 1.0743263959884644,
      "learning_rate": 9.999842826959119e-06,
      "loss": 0.9445,
      "step": 287
    },
    {
      "epoch": 0.015527280569333621,
      "grad_norm": 1.0456606149673462,
      "learning_rate": 9.999841141478315e-06,
      "loss": 0.8869,
      "step": 288
    },
    {
      "epoch": 0.01558119473797714,
      "grad_norm": 1.0299748182296753,
      "learning_rate": 9.99983944700847e-06,
      "loss": 0.9543,
      "step": 289
    },
    {
      "epoch": 0.01563510890662066,
      "grad_norm": 1.0176036357879639,
      "learning_rate": 9.99983774354959e-06,
      "loss": 0.9672,
      "step": 290
    },
    {
      "epoch": 0.01568902307526418,
      "grad_norm": 1.0023303031921387,
      "learning_rate": 9.999836031101675e-06,
      "loss": 0.9417,
      "step": 291
    },
    {
      "epoch": 0.015742937243907698,
      "grad_norm": 0.9801005721092224,
      "learning_rate": 9.99983430966473e-06,
      "loss": 0.9376,
      "step": 292
    },
    {
      "epoch": 0.01579685141255122,
      "grad_norm": 1.002906322479248,
      "learning_rate": 9.999832579238756e-06,
      "loss": 0.8973,
      "step": 293
    },
    {
      "epoch": 0.015850765581194737,
      "grad_norm": 1.0014845132827759,
      "learning_rate": 9.999830839823759e-06,
      "loss": 0.9583,
      "step": 294
    },
    {
      "epoch": 0.01590467974983826,
      "grad_norm": 1.0173449516296387,
      "learning_rate": 9.999829091419739e-06,
      "loss": 0.9006,
      "step": 295
    },
    {
      "epoch": 0.015958593918481777,
      "grad_norm": 0.9779545664787292,
      "learning_rate": 9.999827334026702e-06,
      "loss": 0.9342,
      "step": 296
    },
    {
      "epoch": 0.016012508087125298,
      "grad_norm": 0.9800315499305725,
      "learning_rate": 9.999825567644648e-06,
      "loss": 0.7948,
      "step": 297
    },
    {
      "epoch": 0.016066422255768816,
      "grad_norm": 0.9628249406814575,
      "learning_rate": 9.999823792273583e-06,
      "loss": 0.8415,
      "step": 298
    },
    {
      "epoch": 0.016120336424412334,
      "grad_norm": 1.1227449178695679,
      "learning_rate": 9.99982200791351e-06,
      "loss": 0.9646,
      "step": 299
    },
    {
      "epoch": 0.016174250593055856,
      "grad_norm": 1.1018567085266113,
      "learning_rate": 9.99982021456443e-06,
      "loss": 0.8647,
      "step": 300
    },
    {
      "epoch": 0.016228164761699373,
      "grad_norm": 1.1017298698425293,
      "learning_rate": 9.999818412226347e-06,
      "loss": 0.8708,
      "step": 301
    },
    {
      "epoch": 0.016282078930342895,
      "grad_norm": 1.084594488143921,
      "learning_rate": 9.999816600899267e-06,
      "loss": 0.9765,
      "step": 302
    },
    {
      "epoch": 0.016335993098986413,
      "grad_norm": 1.3735941648483276,
      "learning_rate": 9.99981478058319e-06,
      "loss": 1.0253,
      "step": 303
    },
    {
      "epoch": 0.016389907267629934,
      "grad_norm": 1.1644489765167236,
      "learning_rate": 9.999812951278119e-06,
      "loss": 0.8519,
      "step": 304
    },
    {
      "epoch": 0.016443821436273452,
      "grad_norm": 1.0079474449157715,
      "learning_rate": 9.99981111298406e-06,
      "loss": 0.9422,
      "step": 305
    },
    {
      "epoch": 0.016497735604916974,
      "grad_norm": 1.0046736001968384,
      "learning_rate": 9.999809265701015e-06,
      "loss": 0.7766,
      "step": 306
    },
    {
      "epoch": 0.01655164977356049,
      "grad_norm": 1.0312374830245972,
      "learning_rate": 9.999807409428987e-06,
      "loss": 0.8844,
      "step": 307
    },
    {
      "epoch": 0.01660556394220401,
      "grad_norm": 1.0419421195983887,
      "learning_rate": 9.99980554416798e-06,
      "loss": 0.8902,
      "step": 308
    },
    {
      "epoch": 0.01665947811084753,
      "grad_norm": 1.2056832313537598,
      "learning_rate": 9.999803669917996e-06,
      "loss": 0.9842,
      "step": 309
    },
    {
      "epoch": 0.01671339227949105,
      "grad_norm": 0.9645346403121948,
      "learning_rate": 9.999801786679039e-06,
      "loss": 0.7837,
      "step": 310
    },
    {
      "epoch": 0.01676730644813457,
      "grad_norm": 1.0259841680526733,
      "learning_rate": 9.999799894451115e-06,
      "loss": 0.8927,
      "step": 311
    },
    {
      "epoch": 0.016821220616778088,
      "grad_norm": 0.9932212233543396,
      "learning_rate": 9.999797993234224e-06,
      "loss": 0.815,
      "step": 312
    },
    {
      "epoch": 0.01687513478542161,
      "grad_norm": 1.0666078329086304,
      "learning_rate": 9.99979608302837e-06,
      "loss": 0.8245,
      "step": 313
    },
    {
      "epoch": 0.016929048954065128,
      "grad_norm": 0.9566568732261658,
      "learning_rate": 9.999794163833557e-06,
      "loss": 0.851,
      "step": 314
    },
    {
      "epoch": 0.01698296312270865,
      "grad_norm": 1.0056332349777222,
      "learning_rate": 9.999792235649789e-06,
      "loss": 0.8704,
      "step": 315
    },
    {
      "epoch": 0.017036877291352167,
      "grad_norm": 1.036537528038025,
      "learning_rate": 9.999790298477068e-06,
      "loss": 0.9512,
      "step": 316
    },
    {
      "epoch": 0.01709079145999569,
      "grad_norm": 1.1026023626327515,
      "learning_rate": 9.9997883523154e-06,
      "loss": 1.0007,
      "step": 317
    },
    {
      "epoch": 0.017144705628639206,
      "grad_norm": 1.006659984588623,
      "learning_rate": 9.999786397164786e-06,
      "loss": 0.8992,
      "step": 318
    },
    {
      "epoch": 0.017198619797282724,
      "grad_norm": 1.0100573301315308,
      "learning_rate": 9.99978443302523e-06,
      "loss": 0.9545,
      "step": 319
    },
    {
      "epoch": 0.017252533965926246,
      "grad_norm": 1.000086784362793,
      "learning_rate": 9.999782459896735e-06,
      "loss": 0.8732,
      "step": 320
    },
    {
      "epoch": 0.017306448134569764,
      "grad_norm": 1.2039650678634644,
      "learning_rate": 9.999780477779306e-06,
      "loss": 0.9881,
      "step": 321
    },
    {
      "epoch": 0.017360362303213285,
      "grad_norm": 1.0316474437713623,
      "learning_rate": 9.999778486672948e-06,
      "loss": 0.8686,
      "step": 322
    },
    {
      "epoch": 0.017414276471856803,
      "grad_norm": 1.1697666645050049,
      "learning_rate": 9.999776486577661e-06,
      "loss": 0.9185,
      "step": 323
    },
    {
      "epoch": 0.017468190640500324,
      "grad_norm": 0.9523053169250488,
      "learning_rate": 9.999774477493451e-06,
      "loss": 0.858,
      "step": 324
    },
    {
      "epoch": 0.017522104809143842,
      "grad_norm": 0.9660015106201172,
      "learning_rate": 9.999772459420319e-06,
      "loss": 0.9964,
      "step": 325
    },
    {
      "epoch": 0.017576018977787364,
      "grad_norm": 0.971128523349762,
      "learning_rate": 9.999770432358271e-06,
      "loss": 0.8999,
      "step": 326
    },
    {
      "epoch": 0.01762993314643088,
      "grad_norm": 1.221969485282898,
      "learning_rate": 9.999768396307312e-06,
      "loss": 0.8628,
      "step": 327
    },
    {
      "epoch": 0.017683847315074403,
      "grad_norm": 1.0868507623672485,
      "learning_rate": 9.999766351267442e-06,
      "loss": 1.0732,
      "step": 328
    },
    {
      "epoch": 0.01773776148371792,
      "grad_norm": 0.9527992606163025,
      "learning_rate": 9.999764297238666e-06,
      "loss": 0.8221,
      "step": 329
    },
    {
      "epoch": 0.01779167565236144,
      "grad_norm": 0.9969122409820557,
      "learning_rate": 9.99976223422099e-06,
      "loss": 0.9234,
      "step": 330
    },
    {
      "epoch": 0.01784558982100496,
      "grad_norm": 0.9291784763336182,
      "learning_rate": 9.999760162214415e-06,
      "loss": 0.7839,
      "step": 331
    },
    {
      "epoch": 0.01789950398964848,
      "grad_norm": 0.9766960144042969,
      "learning_rate": 9.999758081218944e-06,
      "loss": 0.7929,
      "step": 332
    },
    {
      "epoch": 0.017953418158292,
      "grad_norm": 0.9536904692649841,
      "learning_rate": 9.999755991234585e-06,
      "loss": 0.9136,
      "step": 333
    },
    {
      "epoch": 0.018007332326935518,
      "grad_norm": 1.0325372219085693,
      "learning_rate": 9.999753892261337e-06,
      "loss": 0.8367,
      "step": 334
    },
    {
      "epoch": 0.01806124649557904,
      "grad_norm": 0.9486141800880432,
      "learning_rate": 9.999751784299207e-06,
      "loss": 0.8802,
      "step": 335
    },
    {
      "epoch": 0.018115160664222557,
      "grad_norm": 0.9880577921867371,
      "learning_rate": 9.999749667348198e-06,
      "loss": 0.8597,
      "step": 336
    },
    {
      "epoch": 0.01816907483286608,
      "grad_norm": 1.043199896812439,
      "learning_rate": 9.999747541408312e-06,
      "loss": 0.9142,
      "step": 337
    },
    {
      "epoch": 0.018222989001509596,
      "grad_norm": 1.0606465339660645,
      "learning_rate": 9.999745406479554e-06,
      "loss": 0.9876,
      "step": 338
    },
    {
      "epoch": 0.018276903170153118,
      "grad_norm": 1.139449954032898,
      "learning_rate": 9.999743262561929e-06,
      "loss": 0.7773,
      "step": 339
    },
    {
      "epoch": 0.018330817338796636,
      "grad_norm": 1.1416115760803223,
      "learning_rate": 9.99974110965544e-06,
      "loss": 0.9566,
      "step": 340
    },
    {
      "epoch": 0.018384731507440154,
      "grad_norm": 1.0145153999328613,
      "learning_rate": 9.99973894776009e-06,
      "loss": 0.9543,
      "step": 341
    },
    {
      "epoch": 0.018438645676083675,
      "grad_norm": 0.950528621673584,
      "learning_rate": 9.999736776875885e-06,
      "loss": 0.8007,
      "step": 342
    },
    {
      "epoch": 0.018492559844727193,
      "grad_norm": 0.9080097079277039,
      "learning_rate": 9.999734597002826e-06,
      "loss": 0.8273,
      "step": 343
    },
    {
      "epoch": 0.018546474013370715,
      "grad_norm": 1.0038888454437256,
      "learning_rate": 9.99973240814092e-06,
      "loss": 0.9394,
      "step": 344
    },
    {
      "epoch": 0.018600388182014232,
      "grad_norm": 1.05253267288208,
      "learning_rate": 9.999730210290168e-06,
      "loss": 0.9485,
      "step": 345
    },
    {
      "epoch": 0.018654302350657754,
      "grad_norm": 0.9396592974662781,
      "learning_rate": 9.999728003450577e-06,
      "loss": 0.8943,
      "step": 346
    },
    {
      "epoch": 0.018708216519301272,
      "grad_norm": 1.149387240409851,
      "learning_rate": 9.999725787622148e-06,
      "loss": 0.8566,
      "step": 347
    },
    {
      "epoch": 0.018762130687944793,
      "grad_norm": 1.1573290824890137,
      "learning_rate": 9.999723562804887e-06,
      "loss": 0.9641,
      "step": 348
    },
    {
      "epoch": 0.01881604485658831,
      "grad_norm": 1.0217385292053223,
      "learning_rate": 9.999721328998797e-06,
      "loss": 0.9555,
      "step": 349
    },
    {
      "epoch": 0.018869959025231833,
      "grad_norm": 1.034690499305725,
      "learning_rate": 9.999719086203884e-06,
      "loss": 0.9407,
      "step": 350
    },
    {
      "epoch": 0.01892387319387535,
      "grad_norm": 0.9819002151489258,
      "learning_rate": 9.999716834420148e-06,
      "loss": 0.9104,
      "step": 351
    },
    {
      "epoch": 0.01897778736251887,
      "grad_norm": 1.0459688901901245,
      "learning_rate": 9.999714573647597e-06,
      "loss": 0.9296,
      "step": 352
    },
    {
      "epoch": 0.01903170153116239,
      "grad_norm": 0.9575183391571045,
      "learning_rate": 9.999712303886232e-06,
      "loss": 0.8517,
      "step": 353
    },
    {
      "epoch": 0.019085615699805908,
      "grad_norm": 1.0018881559371948,
      "learning_rate": 9.99971002513606e-06,
      "loss": 0.9208,
      "step": 354
    },
    {
      "epoch": 0.01913952986844943,
      "grad_norm": 1.0291972160339355,
      "learning_rate": 9.999707737397085e-06,
      "loss": 0.8765,
      "step": 355
    },
    {
      "epoch": 0.019193444037092947,
      "grad_norm": 1.0081498622894287,
      "learning_rate": 9.999705440669306e-06,
      "loss": 0.9204,
      "step": 356
    },
    {
      "epoch": 0.01924735820573647,
      "grad_norm": 0.956950843334198,
      "learning_rate": 9.999703134952733e-06,
      "loss": 0.8058,
      "step": 357
    },
    {
      "epoch": 0.019301272374379987,
      "grad_norm": 1.1130229234695435,
      "learning_rate": 9.999700820247369e-06,
      "loss": 0.8202,
      "step": 358
    },
    {
      "epoch": 0.019355186543023508,
      "grad_norm": 1.047211766242981,
      "learning_rate": 9.999698496553216e-06,
      "loss": 0.9357,
      "step": 359
    },
    {
      "epoch": 0.019409100711667026,
      "grad_norm": 1.0225415229797363,
      "learning_rate": 9.99969616387028e-06,
      "loss": 0.8306,
      "step": 360
    },
    {
      "epoch": 0.019463014880310544,
      "grad_norm": 1.060727596282959,
      "learning_rate": 9.999693822198564e-06,
      "loss": 0.9178,
      "step": 361
    },
    {
      "epoch": 0.019516929048954065,
      "grad_norm": 1.0743412971496582,
      "learning_rate": 9.999691471538074e-06,
      "loss": 0.8761,
      "step": 362
    },
    {
      "epoch": 0.019570843217597583,
      "grad_norm": 1.2229491472244263,
      "learning_rate": 9.99968911188881e-06,
      "loss": 1.0738,
      "step": 363
    },
    {
      "epoch": 0.019624757386241105,
      "grad_norm": 0.9889073967933655,
      "learning_rate": 9.999686743250783e-06,
      "loss": 0.9458,
      "step": 364
    },
    {
      "epoch": 0.019678671554884623,
      "grad_norm": 1.0398520231246948,
      "learning_rate": 9.999684365623992e-06,
      "loss": 0.9096,
      "step": 365
    },
    {
      "epoch": 0.019732585723528144,
      "grad_norm": 1.0613081455230713,
      "learning_rate": 9.999681979008442e-06,
      "loss": 0.9312,
      "step": 366
    },
    {
      "epoch": 0.019786499892171662,
      "grad_norm": 0.946211040019989,
      "learning_rate": 9.99967958340414e-06,
      "loss": 0.9208,
      "step": 367
    },
    {
      "epoch": 0.019840414060815183,
      "grad_norm": 1.1298933029174805,
      "learning_rate": 9.999677178811087e-06,
      "loss": 0.9378,
      "step": 368
    },
    {
      "epoch": 0.0198943282294587,
      "grad_norm": 1.1042351722717285,
      "learning_rate": 9.999674765229288e-06,
      "loss": 0.9487,
      "step": 369
    },
    {
      "epoch": 0.019948242398102223,
      "grad_norm": 1.0717188119888306,
      "learning_rate": 9.999672342658751e-06,
      "loss": 0.939,
      "step": 370
    },
    {
      "epoch": 0.02000215656674574,
      "grad_norm": 1.0936871767044067,
      "learning_rate": 9.999669911099474e-06,
      "loss": 1.1361,
      "step": 371
    },
    {
      "epoch": 0.02005607073538926,
      "grad_norm": 1.0650005340576172,
      "learning_rate": 9.999667470551466e-06,
      "loss": 0.9709,
      "step": 372
    },
    {
      "epoch": 0.02010998490403278,
      "grad_norm": 1.0154083967208862,
      "learning_rate": 9.999665021014731e-06,
      "loss": 0.9422,
      "step": 373
    },
    {
      "epoch": 0.020163899072676298,
      "grad_norm": 1.1382607221603394,
      "learning_rate": 9.999662562489272e-06,
      "loss": 0.984,
      "step": 374
    },
    {
      "epoch": 0.02021781324131982,
      "grad_norm": 0.9372896552085876,
      "learning_rate": 9.999660094975095e-06,
      "loss": 0.9857,
      "step": 375
    },
    {
      "epoch": 0.020271727409963337,
      "grad_norm": 1.1777011156082153,
      "learning_rate": 9.999657618472203e-06,
      "loss": 0.9731,
      "step": 376
    },
    {
      "epoch": 0.02032564157860686,
      "grad_norm": 0.9054237604141235,
      "learning_rate": 9.9996551329806e-06,
      "loss": 0.9104,
      "step": 377
    },
    {
      "epoch": 0.020379555747250377,
      "grad_norm": 0.9255661964416504,
      "learning_rate": 9.999652638500292e-06,
      "loss": 0.8632,
      "step": 378
    },
    {
      "epoch": 0.020433469915893898,
      "grad_norm": 0.9440998435020447,
      "learning_rate": 9.999650135031282e-06,
      "loss": 0.8945,
      "step": 379
    },
    {
      "epoch": 0.020487384084537416,
      "grad_norm": 0.9822732210159302,
      "learning_rate": 9.999647622573577e-06,
      "loss": 0.8874,
      "step": 380
    },
    {
      "epoch": 0.020541298253180938,
      "grad_norm": 1.1294387578964233,
      "learning_rate": 9.999645101127179e-06,
      "loss": 0.9892,
      "step": 381
    },
    {
      "epoch": 0.020595212421824455,
      "grad_norm": 1.0458290576934814,
      "learning_rate": 9.999642570692094e-06,
      "loss": 0.9163,
      "step": 382
    },
    {
      "epoch": 0.020649126590467973,
      "grad_norm": 0.8124557733535767,
      "learning_rate": 9.999640031268326e-06,
      "loss": 0.6927,
      "step": 383
    },
    {
      "epoch": 0.020703040759111495,
      "grad_norm": 1.1053259372711182,
      "learning_rate": 9.999637482855878e-06,
      "loss": 0.8651,
      "step": 384
    },
    {
      "epoch": 0.020756954927755013,
      "grad_norm": 1.1280632019042969,
      "learning_rate": 9.999634925454757e-06,
      "loss": 0.9708,
      "step": 385
    },
    {
      "epoch": 0.020810869096398534,
      "grad_norm": 0.9916180372238159,
      "learning_rate": 9.999632359064965e-06,
      "loss": 0.9081,
      "step": 386
    },
    {
      "epoch": 0.020864783265042052,
      "grad_norm": 1.0430771112442017,
      "learning_rate": 9.99962978368651e-06,
      "loss": 0.9837,
      "step": 387
    },
    {
      "epoch": 0.020918697433685574,
      "grad_norm": 1.031343698501587,
      "learning_rate": 9.999627199319398e-06,
      "loss": 0.9156,
      "step": 388
    },
    {
      "epoch": 0.02097261160232909,
      "grad_norm": 1.0157191753387451,
      "learning_rate": 9.999624605963627e-06,
      "loss": 0.9379,
      "step": 389
    },
    {
      "epoch": 0.021026525770972613,
      "grad_norm": 0.9524544477462769,
      "learning_rate": 9.999622003619204e-06,
      "loss": 0.8448,
      "step": 390
    },
    {
      "epoch": 0.02108043993961613,
      "grad_norm": 1.091670036315918,
      "learning_rate": 9.999619392286137e-06,
      "loss": 0.9794,
      "step": 391
    },
    {
      "epoch": 0.021134354108259652,
      "grad_norm": 1.0502233505249023,
      "learning_rate": 9.999616771964429e-06,
      "loss": 1.0047,
      "step": 392
    },
    {
      "epoch": 0.02118826827690317,
      "grad_norm": 1.2087476253509521,
      "learning_rate": 9.999614142654084e-06,
      "loss": 0.8964,
      "step": 393
    },
    {
      "epoch": 0.021242182445546688,
      "grad_norm": 1.0264590978622437,
      "learning_rate": 9.999611504355106e-06,
      "loss": 0.8608,
      "step": 394
    },
    {
      "epoch": 0.02129609661419021,
      "grad_norm": 0.9883281588554382,
      "learning_rate": 9.999608857067503e-06,
      "loss": 0.9109,
      "step": 395
    },
    {
      "epoch": 0.021350010782833728,
      "grad_norm": 0.9913623332977295,
      "learning_rate": 9.999606200791276e-06,
      "loss": 0.8993,
      "step": 396
    },
    {
      "epoch": 0.02140392495147725,
      "grad_norm": 1.019178867340088,
      "learning_rate": 9.999603535526432e-06,
      "loss": 0.9115,
      "step": 397
    },
    {
      "epoch": 0.021457839120120767,
      "grad_norm": 0.9756026864051819,
      "learning_rate": 9.999600861272974e-06,
      "loss": 0.834,
      "step": 398
    },
    {
      "epoch": 0.02151175328876429,
      "grad_norm": 0.9956341981887817,
      "learning_rate": 9.999598178030909e-06,
      "loss": 0.8756,
      "step": 399
    },
    {
      "epoch": 0.021565667457407806,
      "grad_norm": 1.0267717838287354,
      "learning_rate": 9.999595485800239e-06,
      "loss": 0.9427,
      "step": 400
    },
    {
      "epoch": 0.021619581626051328,
      "grad_norm": 1.061139464378357,
      "learning_rate": 9.999592784580974e-06,
      "loss": 0.9835,
      "step": 401
    },
    {
      "epoch": 0.021673495794694846,
      "grad_norm": 0.9970353245735168,
      "learning_rate": 9.999590074373114e-06,
      "loss": 0.8946,
      "step": 402
    },
    {
      "epoch": 0.021727409963338367,
      "grad_norm": 1.056242823600769,
      "learning_rate": 9.999587355176664e-06,
      "loss": 0.9076,
      "step": 403
    },
    {
      "epoch": 0.021781324131981885,
      "grad_norm": 1.0285427570343018,
      "learning_rate": 9.999584626991632e-06,
      "loss": 0.8506,
      "step": 404
    },
    {
      "epoch": 0.021835238300625403,
      "grad_norm": 1.0026901960372925,
      "learning_rate": 9.99958188981802e-06,
      "loss": 0.8457,
      "step": 405
    },
    {
      "epoch": 0.021889152469268924,
      "grad_norm": 0.8921003341674805,
      "learning_rate": 9.999579143655833e-06,
      "loss": 0.8215,
      "step": 406
    },
    {
      "epoch": 0.021943066637912442,
      "grad_norm": 1.2816855907440186,
      "learning_rate": 9.99957638850508e-06,
      "loss": 0.8779,
      "step": 407
    },
    {
      "epoch": 0.021996980806555964,
      "grad_norm": 1.4713681936264038,
      "learning_rate": 9.99957362436576e-06,
      "loss": 0.8581,
      "step": 408
    },
    {
      "epoch": 0.02205089497519948,
      "grad_norm": 1.0117568969726562,
      "learning_rate": 9.999570851237883e-06,
      "loss": 0.8865,
      "step": 409
    },
    {
      "epoch": 0.022104809143843003,
      "grad_norm": 0.9530962705612183,
      "learning_rate": 9.99956806912145e-06,
      "loss": 0.8888,
      "step": 410
    },
    {
      "epoch": 0.02215872331248652,
      "grad_norm": 0.865692675113678,
      "learning_rate": 9.99956527801647e-06,
      "loss": 0.8075,
      "step": 411
    },
    {
      "epoch": 0.022212637481130042,
      "grad_norm": 0.9613220691680908,
      "learning_rate": 9.999562477922944e-06,
      "loss": 0.9289,
      "step": 412
    },
    {
      "epoch": 0.02226655164977356,
      "grad_norm": 0.9419745802879333,
      "learning_rate": 9.99955966884088e-06,
      "loss": 0.8758,
      "step": 413
    },
    {
      "epoch": 0.02232046581841708,
      "grad_norm": 1.0120573043823242,
      "learning_rate": 9.999556850770282e-06,
      "loss": 0.9014,
      "step": 414
    },
    {
      "epoch": 0.0223743799870606,
      "grad_norm": 0.9833963513374329,
      "learning_rate": 9.999554023711155e-06,
      "loss": 0.9354,
      "step": 415
    },
    {
      "epoch": 0.022428294155704118,
      "grad_norm": 0.9058681130409241,
      "learning_rate": 9.999551187663505e-06,
      "loss": 0.9201,
      "step": 416
    },
    {
      "epoch": 0.02248220832434764,
      "grad_norm": 1.0103633403778076,
      "learning_rate": 9.999548342627334e-06,
      "loss": 0.9023,
      "step": 417
    },
    {
      "epoch": 0.022536122492991157,
      "grad_norm": 0.8671039342880249,
      "learning_rate": 9.99954548860265e-06,
      "loss": 0.7263,
      "step": 418
    },
    {
      "epoch": 0.02259003666163468,
      "grad_norm": 1.0967090129852295,
      "learning_rate": 9.999542625589461e-06,
      "loss": 1.0616,
      "step": 419
    },
    {
      "epoch": 0.022643950830278196,
      "grad_norm": 0.9032139778137207,
      "learning_rate": 9.999539753587764e-06,
      "loss": 0.782,
      "step": 420
    },
    {
      "epoch": 0.022697864998921718,
      "grad_norm": 0.9532387256622314,
      "learning_rate": 9.99953687259757e-06,
      "loss": 0.9628,
      "step": 421
    },
    {
      "epoch": 0.022751779167565236,
      "grad_norm": 0.9732246994972229,
      "learning_rate": 9.999533982618885e-06,
      "loss": 0.8682,
      "step": 422
    },
    {
      "epoch": 0.022805693336208757,
      "grad_norm": 0.9160019159317017,
      "learning_rate": 9.99953108365171e-06,
      "loss": 0.9051,
      "step": 423
    },
    {
      "epoch": 0.022859607504852275,
      "grad_norm": 1.0100488662719727,
      "learning_rate": 9.999528175696054e-06,
      "loss": 0.9836,
      "step": 424
    },
    {
      "epoch": 0.022913521673495793,
      "grad_norm": 1.0130014419555664,
      "learning_rate": 9.99952525875192e-06,
      "loss": 0.8653,
      "step": 425
    },
    {
      "epoch": 0.022967435842139314,
      "grad_norm": 0.9726247787475586,
      "learning_rate": 9.999522332819313e-06,
      "loss": 0.8761,
      "step": 426
    },
    {
      "epoch": 0.023021350010782832,
      "grad_norm": 0.9457972049713135,
      "learning_rate": 9.99951939789824e-06,
      "loss": 0.8792,
      "step": 427
    },
    {
      "epoch": 0.023075264179426354,
      "grad_norm": 1.083130121231079,
      "learning_rate": 9.999516453988706e-06,
      "loss": 0.9035,
      "step": 428
    },
    {
      "epoch": 0.023129178348069872,
      "grad_norm": 0.9195771217346191,
      "learning_rate": 9.999513501090714e-06,
      "loss": 0.8586,
      "step": 429
    },
    {
      "epoch": 0.023183092516713393,
      "grad_norm": 0.983346700668335,
      "learning_rate": 9.999510539204273e-06,
      "loss": 0.8335,
      "step": 430
    },
    {
      "epoch": 0.02323700668535691,
      "grad_norm": 1.0524029731750488,
      "learning_rate": 9.999507568329386e-06,
      "loss": 0.838,
      "step": 431
    },
    {
      "epoch": 0.023290920854000433,
      "grad_norm": 1.0267860889434814,
      "learning_rate": 9.999504588466058e-06,
      "loss": 0.9345,
      "step": 432
    },
    {
      "epoch": 0.02334483502264395,
      "grad_norm": 1.025707483291626,
      "learning_rate": 9.999501599614294e-06,
      "loss": 0.9042,
      "step": 433
    },
    {
      "epoch": 0.023398749191287472,
      "grad_norm": 0.9739174842834473,
      "learning_rate": 9.999498601774101e-06,
      "loss": 0.7433,
      "step": 434
    },
    {
      "epoch": 0.02345266335993099,
      "grad_norm": 0.9468310475349426,
      "learning_rate": 9.999495594945486e-06,
      "loss": 0.8447,
      "step": 435
    },
    {
      "epoch": 0.023506577528574508,
      "grad_norm": 0.9820529818534851,
      "learning_rate": 9.99949257912845e-06,
      "loss": 0.8842,
      "step": 436
    },
    {
      "epoch": 0.02356049169721803,
      "grad_norm": 0.998515784740448,
      "learning_rate": 9.999489554323e-06,
      "loss": 0.9226,
      "step": 437
    },
    {
      "epoch": 0.023614405865861547,
      "grad_norm": 0.9819791316986084,
      "learning_rate": 9.999486520529144e-06,
      "loss": 0.8559,
      "step": 438
    },
    {
      "epoch": 0.02366832003450507,
      "grad_norm": 0.9468326568603516,
      "learning_rate": 9.999483477746884e-06,
      "loss": 0.8064,
      "step": 439
    },
    {
      "epoch": 0.023722234203148587,
      "grad_norm": 1.0087614059448242,
      "learning_rate": 9.999480425976229e-06,
      "loss": 0.9232,
      "step": 440
    },
    {
      "epoch": 0.023776148371792108,
      "grad_norm": 0.9446098208427429,
      "learning_rate": 9.99947736521718e-06,
      "loss": 0.8511,
      "step": 441
    },
    {
      "epoch": 0.023830062540435626,
      "grad_norm": 1.0966850519180298,
      "learning_rate": 9.999474295469746e-06,
      "loss": 0.9929,
      "step": 442
    },
    {
      "epoch": 0.023883976709079147,
      "grad_norm": 0.8858770728111267,
      "learning_rate": 9.99947121673393e-06,
      "loss": 0.8492,
      "step": 443
    },
    {
      "epoch": 0.023937890877722665,
      "grad_norm": 1.083717703819275,
      "learning_rate": 9.999468129009742e-06,
      "loss": 0.9948,
      "step": 444
    },
    {
      "epoch": 0.023991805046366187,
      "grad_norm": 1.0251178741455078,
      "learning_rate": 9.999465032297184e-06,
      "loss": 0.8769,
      "step": 445
    },
    {
      "epoch": 0.024045719215009705,
      "grad_norm": 0.9331875443458557,
      "learning_rate": 9.999461926596261e-06,
      "loss": 0.8663,
      "step": 446
    },
    {
      "epoch": 0.024099633383653223,
      "grad_norm": 0.8941493034362793,
      "learning_rate": 9.999458811906979e-06,
      "loss": 0.8172,
      "step": 447
    },
    {
      "epoch": 0.024153547552296744,
      "grad_norm": 0.9978699684143066,
      "learning_rate": 9.999455688229347e-06,
      "loss": 0.9303,
      "step": 448
    },
    {
      "epoch": 0.024207461720940262,
      "grad_norm": 0.8835211992263794,
      "learning_rate": 9.999452555563366e-06,
      "loss": 0.8921,
      "step": 449
    },
    {
      "epoch": 0.024261375889583783,
      "grad_norm": 0.9061810970306396,
      "learning_rate": 9.999449413909043e-06,
      "loss": 0.8201,
      "step": 450
    },
    {
      "epoch": 0.0243152900582273,
      "grad_norm": 1.0061571598052979,
      "learning_rate": 9.999446263266385e-06,
      "loss": 0.8506,
      "step": 451
    },
    {
      "epoch": 0.024369204226870823,
      "grad_norm": 0.9286402463912964,
      "learning_rate": 9.999443103635398e-06,
      "loss": 0.8532,
      "step": 452
    },
    {
      "epoch": 0.02442311839551434,
      "grad_norm": 1.0919772386550903,
      "learning_rate": 9.999439935016087e-06,
      "loss": 0.9466,
      "step": 453
    },
    {
      "epoch": 0.024477032564157862,
      "grad_norm": 1.0552513599395752,
      "learning_rate": 9.999436757408453e-06,
      "loss": 0.8406,
      "step": 454
    },
    {
      "epoch": 0.02453094673280138,
      "grad_norm": 0.9604331851005554,
      "learning_rate": 9.999433570812511e-06,
      "loss": 0.8928,
      "step": 455
    },
    {
      "epoch": 0.0245848609014449,
      "grad_norm": 1.0126323699951172,
      "learning_rate": 9.999430375228259e-06,
      "loss": 0.924,
      "step": 456
    },
    {
      "epoch": 0.02463877507008842,
      "grad_norm": 1.0540791749954224,
      "learning_rate": 9.999427170655707e-06,
      "loss": 0.9656,
      "step": 457
    },
    {
      "epoch": 0.024692689238731937,
      "grad_norm": 0.8622417449951172,
      "learning_rate": 9.999423957094857e-06,
      "loss": 0.7428,
      "step": 458
    },
    {
      "epoch": 0.02474660340737546,
      "grad_norm": 1.106581211090088,
      "learning_rate": 9.999420734545719e-06,
      "loss": 0.9258,
      "step": 459
    },
    {
      "epoch": 0.024800517576018977,
      "grad_norm": 0.990807294845581,
      "learning_rate": 9.999417503008296e-06,
      "loss": 0.9083,
      "step": 460
    },
    {
      "epoch": 0.024854431744662498,
      "grad_norm": 0.9302589893341064,
      "learning_rate": 9.999414262482594e-06,
      "loss": 0.8654,
      "step": 461
    },
    {
      "epoch": 0.024908345913306016,
      "grad_norm": 1.0218255519866943,
      "learning_rate": 9.999411012968621e-06,
      "loss": 0.8996,
      "step": 462
    },
    {
      "epoch": 0.024962260081949537,
      "grad_norm": 0.976108193397522,
      "learning_rate": 9.99940775446638e-06,
      "loss": 0.9423,
      "step": 463
    },
    {
      "epoch": 0.025016174250593055,
      "grad_norm": 1.1027617454528809,
      "learning_rate": 9.99940448697588e-06,
      "loss": 1.0407,
      "step": 464
    },
    {
      "epoch": 0.025070088419236577,
      "grad_norm": 1.0148764848709106,
      "learning_rate": 9.999401210497122e-06,
      "loss": 0.9418,
      "step": 465
    },
    {
      "epoch": 0.025124002587880095,
      "grad_norm": 1.0120681524276733,
      "learning_rate": 9.999397925030116e-06,
      "loss": 0.92,
      "step": 466
    },
    {
      "epoch": 0.025177916756523613,
      "grad_norm": 1.1855127811431885,
      "learning_rate": 9.999394630574868e-06,
      "loss": 0.9285,
      "step": 467
    },
    {
      "epoch": 0.025231830925167134,
      "grad_norm": 1.8014320135116577,
      "learning_rate": 9.999391327131383e-06,
      "loss": 0.979,
      "step": 468
    },
    {
      "epoch": 0.025285745093810652,
      "grad_norm": 1.1568403244018555,
      "learning_rate": 9.999388014699664e-06,
      "loss": 0.9574,
      "step": 469
    },
    {
      "epoch": 0.025339659262454173,
      "grad_norm": 1.2544865608215332,
      "learning_rate": 9.99938469327972e-06,
      "loss": 0.8356,
      "step": 470
    },
    {
      "epoch": 0.02539357343109769,
      "grad_norm": 1.8647997379302979,
      "learning_rate": 9.99938136287156e-06,
      "loss": 0.9181,
      "step": 471
    },
    {
      "epoch": 0.025447487599741213,
      "grad_norm": 0.9942222237586975,
      "learning_rate": 9.999378023475184e-06,
      "loss": 0.9297,
      "step": 472
    },
    {
      "epoch": 0.02550140176838473,
      "grad_norm": 0.9839766621589661,
      "learning_rate": 9.9993746750906e-06,
      "loss": 0.9181,
      "step": 473
    },
    {
      "epoch": 0.025555315937028252,
      "grad_norm": 0.9353258609771729,
      "learning_rate": 9.999371317717817e-06,
      "loss": 0.8789,
      "step": 474
    },
    {
      "epoch": 0.02560923010567177,
      "grad_norm": 0.9256170988082886,
      "learning_rate": 9.999367951356838e-06,
      "loss": 0.8725,
      "step": 475
    },
    {
      "epoch": 0.02566314427431529,
      "grad_norm": 1.1102124452590942,
      "learning_rate": 9.999364576007669e-06,
      "loss": 0.9818,
      "step": 476
    },
    {
      "epoch": 0.02571705844295881,
      "grad_norm": 1.04171884059906,
      "learning_rate": 9.999361191670316e-06,
      "loss": 0.9275,
      "step": 477
    },
    {
      "epoch": 0.025770972611602327,
      "grad_norm": 0.9670290350914001,
      "learning_rate": 9.999357798344787e-06,
      "loss": 0.8919,
      "step": 478
    },
    {
      "epoch": 0.02582488678024585,
      "grad_norm": 1.0543723106384277,
      "learning_rate": 9.999354396031085e-06,
      "loss": 0.9356,
      "step": 479
    },
    {
      "epoch": 0.025878800948889367,
      "grad_norm": 1.1368457078933716,
      "learning_rate": 9.99935098472922e-06,
      "loss": 0.9387,
      "step": 480
    },
    {
      "epoch": 0.025932715117532888,
      "grad_norm": 1.0627872943878174,
      "learning_rate": 9.999347564439196e-06,
      "loss": 1.0047,
      "step": 481
    },
    {
      "epoch": 0.025986629286176406,
      "grad_norm": 0.9553730487823486,
      "learning_rate": 9.999344135161018e-06,
      "loss": 0.8845,
      "step": 482
    },
    {
      "epoch": 0.026040543454819928,
      "grad_norm": 0.9605830907821655,
      "learning_rate": 9.999340696894694e-06,
      "loss": 0.8816,
      "step": 483
    },
    {
      "epoch": 0.026094457623463446,
      "grad_norm": 1.0464140176773071,
      "learning_rate": 9.999337249640232e-06,
      "loss": 0.9344,
      "step": 484
    },
    {
      "epoch": 0.026148371792106967,
      "grad_norm": 1.0667988061904907,
      "learning_rate": 9.999333793397635e-06,
      "loss": 0.8834,
      "step": 485
    },
    {
      "epoch": 0.026202285960750485,
      "grad_norm": 0.8996486663818359,
      "learning_rate": 9.999330328166908e-06,
      "loss": 0.8247,
      "step": 486
    },
    {
      "epoch": 0.026256200129394006,
      "grad_norm": 1.0483838319778442,
      "learning_rate": 9.99932685394806e-06,
      "loss": 0.9414,
      "step": 487
    },
    {
      "epoch": 0.026310114298037524,
      "grad_norm": 1.2089953422546387,
      "learning_rate": 9.999323370741097e-06,
      "loss": 1.0913,
      "step": 488
    },
    {
      "epoch": 0.026364028466681042,
      "grad_norm": 1.074291467666626,
      "learning_rate": 9.999319878546025e-06,
      "loss": 0.8882,
      "step": 489
    },
    {
      "epoch": 0.026417942635324564,
      "grad_norm": 1.0076494216918945,
      "learning_rate": 9.99931637736285e-06,
      "loss": 0.8393,
      "step": 490
    },
    {
      "epoch": 0.02647185680396808,
      "grad_norm": 1.2263407707214355,
      "learning_rate": 9.99931286719158e-06,
      "loss": 0.955,
      "step": 491
    },
    {
      "epoch": 0.026525770972611603,
      "grad_norm": 0.9093664884567261,
      "learning_rate": 9.999309348032218e-06,
      "loss": 0.8366,
      "step": 492
    },
    {
      "epoch": 0.02657968514125512,
      "grad_norm": 1.0704407691955566,
      "learning_rate": 9.999305819884772e-06,
      "loss": 0.981,
      "step": 493
    },
    {
      "epoch": 0.026633599309898642,
      "grad_norm": 1.2105270624160767,
      "learning_rate": 9.999302282749249e-06,
      "loss": 0.8896,
      "step": 494
    },
    {
      "epoch": 0.02668751347854216,
      "grad_norm": 1.0142449140548706,
      "learning_rate": 9.999298736625654e-06,
      "loss": 0.8627,
      "step": 495
    },
    {
      "epoch": 0.02674142764718568,
      "grad_norm": 1.0887057781219482,
      "learning_rate": 9.999295181513994e-06,
      "loss": 0.8884,
      "step": 496
    },
    {
      "epoch": 0.0267953418158292,
      "grad_norm": 0.9958952069282532,
      "learning_rate": 9.999291617414277e-06,
      "loss": 0.7768,
      "step": 497
    },
    {
      "epoch": 0.02684925598447272,
      "grad_norm": 0.8576722741127014,
      "learning_rate": 9.999288044326508e-06,
      "loss": 0.715,
      "step": 498
    },
    {
      "epoch": 0.02690317015311624,
      "grad_norm": 1.058148741722107,
      "learning_rate": 9.999284462250691e-06,
      "loss": 0.8693,
      "step": 499
    },
    {
      "epoch": 0.026957084321759757,
      "grad_norm": 0.9429569244384766,
      "learning_rate": 9.999280871186837e-06,
      "loss": 0.8883,
      "step": 500
    },
    {
      "epoch": 0.02701099849040328,
      "grad_norm": 0.9450993537902832,
      "learning_rate": 9.999277271134948e-06,
      "loss": 0.9376,
      "step": 501
    },
    {
      "epoch": 0.027064912659046796,
      "grad_norm": 1.0307891368865967,
      "learning_rate": 9.999273662095035e-06,
      "loss": 0.9098,
      "step": 502
    },
    {
      "epoch": 0.027118826827690318,
      "grad_norm": 0.9515891671180725,
      "learning_rate": 9.999270044067101e-06,
      "loss": 0.8854,
      "step": 503
    },
    {
      "epoch": 0.027172740996333836,
      "grad_norm": 1.1173255443572998,
      "learning_rate": 9.999266417051154e-06,
      "loss": 0.7977,
      "step": 504
    },
    {
      "epoch": 0.027226655164977357,
      "grad_norm": 1.028194785118103,
      "learning_rate": 9.9992627810472e-06,
      "loss": 0.9585,
      "step": 505
    },
    {
      "epoch": 0.027280569333620875,
      "grad_norm": 1.0855528116226196,
      "learning_rate": 9.999259136055245e-06,
      "loss": 0.9807,
      "step": 506
    },
    {
      "epoch": 0.027334483502264396,
      "grad_norm": 1.1148236989974976,
      "learning_rate": 9.999255482075298e-06,
      "loss": 0.9672,
      "step": 507
    },
    {
      "epoch": 0.027388397670907914,
      "grad_norm": 0.9697713255882263,
      "learning_rate": 9.999251819107364e-06,
      "loss": 0.9073,
      "step": 508
    },
    {
      "epoch": 0.027442311839551436,
      "grad_norm": 0.9802384972572327,
      "learning_rate": 9.999248147151448e-06,
      "loss": 0.8704,
      "step": 509
    },
    {
      "epoch": 0.027496226008194954,
      "grad_norm": 0.963330090045929,
      "learning_rate": 9.999244466207559e-06,
      "loss": 0.9312,
      "step": 510
    },
    {
      "epoch": 0.02755014017683847,
      "grad_norm": 0.8776309490203857,
      "learning_rate": 9.999240776275703e-06,
      "loss": 0.8068,
      "step": 511
    },
    {
      "epoch": 0.027604054345481993,
      "grad_norm": 1.1159353256225586,
      "learning_rate": 9.999237077355886e-06,
      "loss": 0.8164,
      "step": 512
    },
    {
      "epoch": 0.02765796851412551,
      "grad_norm": 1.004232406616211,
      "learning_rate": 9.999233369448115e-06,
      "loss": 0.8666,
      "step": 513
    },
    {
      "epoch": 0.027711882682769032,
      "grad_norm": 1.0300110578536987,
      "learning_rate": 9.999229652552395e-06,
      "loss": 0.8774,
      "step": 514
    },
    {
      "epoch": 0.02776579685141255,
      "grad_norm": 0.8823155164718628,
      "learning_rate": 9.999225926668736e-06,
      "loss": 0.7579,
      "step": 515
    },
    {
      "epoch": 0.027819711020056072,
      "grad_norm": 0.938956618309021,
      "learning_rate": 9.999222191797144e-06,
      "loss": 0.8749,
      "step": 516
    },
    {
      "epoch": 0.02787362518869959,
      "grad_norm": 0.9111800789833069,
      "learning_rate": 9.999218447937624e-06,
      "loss": 0.8915,
      "step": 517
    },
    {
      "epoch": 0.02792753935734311,
      "grad_norm": 0.971813440322876,
      "learning_rate": 9.999214695090182e-06,
      "loss": 0.9038,
      "step": 518
    },
    {
      "epoch": 0.02798145352598663,
      "grad_norm": 0.9159868359565735,
      "learning_rate": 9.999210933254828e-06,
      "loss": 0.8726,
      "step": 519
    },
    {
      "epoch": 0.028035367694630147,
      "grad_norm": 1.0223439931869507,
      "learning_rate": 9.999207162431566e-06,
      "loss": 0.8738,
      "step": 520
    },
    {
      "epoch": 0.02808928186327367,
      "grad_norm": 0.9844004511833191,
      "learning_rate": 9.999203382620404e-06,
      "loss": 0.8815,
      "step": 521
    },
    {
      "epoch": 0.028143196031917186,
      "grad_norm": 1.1636719703674316,
      "learning_rate": 9.99919959382135e-06,
      "loss": 0.8781,
      "step": 522
    },
    {
      "epoch": 0.028197110200560708,
      "grad_norm": 0.9637702703475952,
      "learning_rate": 9.999195796034407e-06,
      "loss": 0.8491,
      "step": 523
    },
    {
      "epoch": 0.028251024369204226,
      "grad_norm": 0.975931704044342,
      "learning_rate": 9.999191989259584e-06,
      "loss": 0.9983,
      "step": 524
    },
    {
      "epoch": 0.028304938537847747,
      "grad_norm": 0.9855527877807617,
      "learning_rate": 9.999188173496889e-06,
      "loss": 0.9587,
      "step": 525
    },
    {
      "epoch": 0.028358852706491265,
      "grad_norm": 0.9925652742385864,
      "learning_rate": 9.99918434874633e-06,
      "loss": 0.8408,
      "step": 526
    },
    {
      "epoch": 0.028412766875134787,
      "grad_norm": 0.9272180795669556,
      "learning_rate": 9.999180515007908e-06,
      "loss": 0.8267,
      "step": 527
    },
    {
      "epoch": 0.028466681043778305,
      "grad_norm": 1.161076307296753,
      "learning_rate": 9.999176672281636e-06,
      "loss": 0.9282,
      "step": 528
    },
    {
      "epoch": 0.028520595212421826,
      "grad_norm": 0.8953909277915955,
      "learning_rate": 9.99917282056752e-06,
      "loss": 0.8078,
      "step": 529
    },
    {
      "epoch": 0.028574509381065344,
      "grad_norm": 0.9194382429122925,
      "learning_rate": 9.999168959865562e-06,
      "loss": 0.8385,
      "step": 530
    },
    {
      "epoch": 0.028628423549708862,
      "grad_norm": 1.0351816415786743,
      "learning_rate": 9.999165090175775e-06,
      "loss": 0.8155,
      "step": 531
    },
    {
      "epoch": 0.028682337718352383,
      "grad_norm": 0.9233224391937256,
      "learning_rate": 9.999161211498163e-06,
      "loss": 0.8825,
      "step": 532
    },
    {
      "epoch": 0.0287362518869959,
      "grad_norm": 1.0415356159210205,
      "learning_rate": 9.999157323832732e-06,
      "loss": 0.7844,
      "step": 533
    },
    {
      "epoch": 0.028790166055639423,
      "grad_norm": 1.0329923629760742,
      "learning_rate": 9.999153427179492e-06,
      "loss": 0.893,
      "step": 534
    },
    {
      "epoch": 0.02884408022428294,
      "grad_norm": 1.237291932106018,
      "learning_rate": 9.999149521538448e-06,
      "loss": 0.9786,
      "step": 535
    },
    {
      "epoch": 0.028897994392926462,
      "grad_norm": 0.9952654242515564,
      "learning_rate": 9.999145606909607e-06,
      "loss": 0.9262,
      "step": 536
    },
    {
      "epoch": 0.02895190856156998,
      "grad_norm": 1.016533374786377,
      "learning_rate": 9.999141683292977e-06,
      "loss": 0.9854,
      "step": 537
    },
    {
      "epoch": 0.0290058227302135,
      "grad_norm": 1.0334454774856567,
      "learning_rate": 9.999137750688564e-06,
      "loss": 0.8928,
      "step": 538
    },
    {
      "epoch": 0.02905973689885702,
      "grad_norm": 0.941662609577179,
      "learning_rate": 9.999133809096374e-06,
      "loss": 0.8698,
      "step": 539
    },
    {
      "epoch": 0.02911365106750054,
      "grad_norm": 0.9454428553581238,
      "learning_rate": 9.999129858516418e-06,
      "loss": 0.9261,
      "step": 540
    },
    {
      "epoch": 0.02916756523614406,
      "grad_norm": 1.0921217203140259,
      "learning_rate": 9.9991258989487e-06,
      "loss": 0.9163,
      "step": 541
    },
    {
      "epoch": 0.029221479404787577,
      "grad_norm": 0.8999170064926147,
      "learning_rate": 9.999121930393227e-06,
      "loss": 0.883,
      "step": 542
    },
    {
      "epoch": 0.029275393573431098,
      "grad_norm": 0.9732702970504761,
      "learning_rate": 9.999117952850009e-06,
      "loss": 0.9168,
      "step": 543
    },
    {
      "epoch": 0.029329307742074616,
      "grad_norm": 1.00196373462677,
      "learning_rate": 9.99911396631905e-06,
      "loss": 0.826,
      "step": 544
    },
    {
      "epoch": 0.029383221910718137,
      "grad_norm": 0.9776156544685364,
      "learning_rate": 9.999109970800358e-06,
      "loss": 0.8176,
      "step": 545
    },
    {
      "epoch": 0.029437136079361655,
      "grad_norm": 1.0503387451171875,
      "learning_rate": 9.99910596629394e-06,
      "loss": 0.8617,
      "step": 546
    },
    {
      "epoch": 0.029491050248005177,
      "grad_norm": 0.9195687174797058,
      "learning_rate": 9.999101952799805e-06,
      "loss": 0.8224,
      "step": 547
    },
    {
      "epoch": 0.029544964416648695,
      "grad_norm": 0.8746809959411621,
      "learning_rate": 9.999097930317959e-06,
      "loss": 0.8407,
      "step": 548
    },
    {
      "epoch": 0.029598878585292216,
      "grad_norm": 0.9035898447036743,
      "learning_rate": 9.999093898848407e-06,
      "loss": 0.8344,
      "step": 549
    },
    {
      "epoch": 0.029652792753935734,
      "grad_norm": 0.8764795064926147,
      "learning_rate": 9.99908985839116e-06,
      "loss": 0.8323,
      "step": 550
    },
    {
      "epoch": 0.029706706922579255,
      "grad_norm": 0.9654614329338074,
      "learning_rate": 9.999085808946224e-06,
      "loss": 0.8696,
      "step": 551
    },
    {
      "epoch": 0.029760621091222773,
      "grad_norm": 1.1295796632766724,
      "learning_rate": 9.999081750513606e-06,
      "loss": 0.9608,
      "step": 552
    },
    {
      "epoch": 0.02981453525986629,
      "grad_norm": 0.9591107368469238,
      "learning_rate": 9.999077683093313e-06,
      "loss": 0.8762,
      "step": 553
    },
    {
      "epoch": 0.029868449428509813,
      "grad_norm": 0.8287899494171143,
      "learning_rate": 9.999073606685353e-06,
      "loss": 0.7265,
      "step": 554
    },
    {
      "epoch": 0.02992236359715333,
      "grad_norm": 0.9429282546043396,
      "learning_rate": 9.99906952128973e-06,
      "loss": 0.8835,
      "step": 555
    },
    {
      "epoch": 0.029976277765796852,
      "grad_norm": 0.9617370963096619,
      "learning_rate": 9.999065426906459e-06,
      "loss": 0.9138,
      "step": 556
    },
    {
      "epoch": 0.03003019193444037,
      "grad_norm": 1.2346372604370117,
      "learning_rate": 9.999061323535538e-06,
      "loss": 0.831,
      "step": 557
    },
    {
      "epoch": 0.03008410610308389,
      "grad_norm": 1.2413623332977295,
      "learning_rate": 9.999057211176982e-06,
      "loss": 1.0211,
      "step": 558
    },
    {
      "epoch": 0.03013802027172741,
      "grad_norm": 0.98906010389328,
      "learning_rate": 9.999053089830794e-06,
      "loss": 0.7821,
      "step": 559
    },
    {
      "epoch": 0.03019193444037093,
      "grad_norm": 0.96706622838974,
      "learning_rate": 9.999048959496983e-06,
      "loss": 0.8593,
      "step": 560
    },
    {
      "epoch": 0.03024584860901445,
      "grad_norm": 0.9400071501731873,
      "learning_rate": 9.999044820175556e-06,
      "loss": 0.8731,
      "step": 561
    },
    {
      "epoch": 0.03029976277765797,
      "grad_norm": 1.1276499032974243,
      "learning_rate": 9.999040671866522e-06,
      "loss": 0.86,
      "step": 562
    },
    {
      "epoch": 0.030353676946301488,
      "grad_norm": 0.8859087228775024,
      "learning_rate": 9.999036514569885e-06,
      "loss": 0.8274,
      "step": 563
    },
    {
      "epoch": 0.030407591114945006,
      "grad_norm": 1.1617575883865356,
      "learning_rate": 9.999032348285656e-06,
      "loss": 1.0519,
      "step": 564
    },
    {
      "epoch": 0.030461505283588527,
      "grad_norm": 0.9717594385147095,
      "learning_rate": 9.99902817301384e-06,
      "loss": 0.9276,
      "step": 565
    },
    {
      "epoch": 0.030515419452232045,
      "grad_norm": 1.000722050666809,
      "learning_rate": 9.999023988754446e-06,
      "loss": 0.8714,
      "step": 566
    },
    {
      "epoch": 0.030569333620875567,
      "grad_norm": 1.1744625568389893,
      "learning_rate": 9.999019795507481e-06,
      "loss": 1.0087,
      "step": 567
    },
    {
      "epoch": 0.030623247789519085,
      "grad_norm": 1.0199978351593018,
      "learning_rate": 9.999015593272953e-06,
      "loss": 0.8537,
      "step": 568
    },
    {
      "epoch": 0.030677161958162606,
      "grad_norm": 0.9232216477394104,
      "learning_rate": 9.999011382050869e-06,
      "loss": 0.8488,
      "step": 569
    },
    {
      "epoch": 0.030731076126806124,
      "grad_norm": 0.9905959367752075,
      "learning_rate": 9.99900716184124e-06,
      "loss": 0.9048,
      "step": 570
    },
    {
      "epoch": 0.030784990295449646,
      "grad_norm": 0.9921644330024719,
      "learning_rate": 9.999002932644066e-06,
      "loss": 0.9294,
      "step": 571
    },
    {
      "epoch": 0.030838904464093164,
      "grad_norm": 1.1583740711212158,
      "learning_rate": 9.99899869445936e-06,
      "loss": 0.727,
      "step": 572
    },
    {
      "epoch": 0.03089281863273668,
      "grad_norm": 0.906736433506012,
      "learning_rate": 9.998994447287127e-06,
      "loss": 0.7889,
      "step": 573
    },
    {
      "epoch": 0.030946732801380203,
      "grad_norm": 0.9060770869255066,
      "learning_rate": 9.998990191127379e-06,
      "loss": 0.8493,
      "step": 574
    },
    {
      "epoch": 0.03100064697002372,
      "grad_norm": 0.9094041585922241,
      "learning_rate": 9.99898592598012e-06,
      "loss": 0.8604,
      "step": 575
    },
    {
      "epoch": 0.031054561138667242,
      "grad_norm": 1.0964977741241455,
      "learning_rate": 9.998981651845358e-06,
      "loss": 0.8481,
      "step": 576
    },
    {
      "epoch": 0.03110847530731076,
      "grad_norm": 0.9509627223014832,
      "learning_rate": 9.998977368723102e-06,
      "loss": 0.8601,
      "step": 577
    },
    {
      "epoch": 0.03116238947595428,
      "grad_norm": 1.0108642578125,
      "learning_rate": 9.998973076613359e-06,
      "loss": 0.9076,
      "step": 578
    },
    {
      "epoch": 0.0312163036445978,
      "grad_norm": 1.0268129110336304,
      "learning_rate": 9.998968775516136e-06,
      "loss": 0.8273,
      "step": 579
    },
    {
      "epoch": 0.03127021781324132,
      "grad_norm": 0.968941867351532,
      "learning_rate": 9.99896446543144e-06,
      "loss": 0.8859,
      "step": 580
    },
    {
      "epoch": 0.03132413198188484,
      "grad_norm": 0.936779260635376,
      "learning_rate": 9.998960146359283e-06,
      "loss": 0.8589,
      "step": 581
    },
    {
      "epoch": 0.03137804615052836,
      "grad_norm": 0.9675167202949524,
      "learning_rate": 9.998955818299667e-06,
      "loss": 0.973,
      "step": 582
    },
    {
      "epoch": 0.03143196031917188,
      "grad_norm": 0.9475553035736084,
      "learning_rate": 9.998951481252604e-06,
      "loss": 0.8936,
      "step": 583
    },
    {
      "epoch": 0.031485874487815396,
      "grad_norm": 0.9130968451499939,
      "learning_rate": 9.9989471352181e-06,
      "loss": 0.7668,
      "step": 584
    },
    {
      "epoch": 0.031539788656458914,
      "grad_norm": 0.8890071511268616,
      "learning_rate": 9.998942780196164e-06,
      "loss": 0.8971,
      "step": 585
    },
    {
      "epoch": 0.03159370282510244,
      "grad_norm": 0.9298738837242126,
      "learning_rate": 9.998938416186803e-06,
      "loss": 0.9313,
      "step": 586
    },
    {
      "epoch": 0.03164761699374596,
      "grad_norm": 1.0683361291885376,
      "learning_rate": 9.998934043190025e-06,
      "loss": 0.9018,
      "step": 587
    },
    {
      "epoch": 0.031701531162389475,
      "grad_norm": 0.939253568649292,
      "learning_rate": 9.99892966120584e-06,
      "loss": 0.9119,
      "step": 588
    },
    {
      "epoch": 0.03175544533103299,
      "grad_norm": 0.9245349764823914,
      "learning_rate": 9.99892527023425e-06,
      "loss": 0.9258,
      "step": 589
    },
    {
      "epoch": 0.03180935949967652,
      "grad_norm": 0.9318797588348389,
      "learning_rate": 9.998920870275267e-06,
      "loss": 0.9557,
      "step": 590
    },
    {
      "epoch": 0.031863273668320036,
      "grad_norm": 0.8909592628479004,
      "learning_rate": 9.998916461328899e-06,
      "loss": 0.8122,
      "step": 591
    },
    {
      "epoch": 0.031917187836963554,
      "grad_norm": 1.0637080669403076,
      "learning_rate": 9.998912043395154e-06,
      "loss": 0.9517,
      "step": 592
    },
    {
      "epoch": 0.03197110200560707,
      "grad_norm": 0.881934642791748,
      "learning_rate": 9.99890761647404e-06,
      "loss": 0.8729,
      "step": 593
    },
    {
      "epoch": 0.032025016174250596,
      "grad_norm": 0.8882094025611877,
      "learning_rate": 9.998903180565562e-06,
      "loss": 0.7943,
      "step": 594
    },
    {
      "epoch": 0.032078930342894114,
      "grad_norm": 0.965085506439209,
      "learning_rate": 9.99889873566973e-06,
      "loss": 0.8894,
      "step": 595
    },
    {
      "epoch": 0.03213284451153763,
      "grad_norm": 0.9679432511329651,
      "learning_rate": 9.998894281786556e-06,
      "loss": 0.854,
      "step": 596
    },
    {
      "epoch": 0.03218675868018115,
      "grad_norm": 1.4454354047775269,
      "learning_rate": 9.998889818916043e-06,
      "loss": 0.9944,
      "step": 597
    },
    {
      "epoch": 0.03224067284882467,
      "grad_norm": 0.9369311928749084,
      "learning_rate": 9.998885347058198e-06,
      "loss": 0.8699,
      "step": 598
    },
    {
      "epoch": 0.03229458701746819,
      "grad_norm": 0.9014303088188171,
      "learning_rate": 9.998880866213033e-06,
      "loss": 0.8735,
      "step": 599
    },
    {
      "epoch": 0.03234850118611171,
      "grad_norm": 0.989251971244812,
      "learning_rate": 9.998876376380555e-06,
      "loss": 0.8872,
      "step": 600
    },
    {
      "epoch": 0.03240241535475523,
      "grad_norm": 1.0256885290145874,
      "learning_rate": 9.99887187756077e-06,
      "loss": 0.8787,
      "step": 601
    },
    {
      "epoch": 0.03245632952339875,
      "grad_norm": 0.9560148119926453,
      "learning_rate": 9.998867369753688e-06,
      "loss": 0.8301,
      "step": 602
    },
    {
      "epoch": 0.03251024369204227,
      "grad_norm": 1.044754147529602,
      "learning_rate": 9.998862852959316e-06,
      "loss": 0.9286,
      "step": 603
    },
    {
      "epoch": 0.03256415786068579,
      "grad_norm": 0.8769629597663879,
      "learning_rate": 9.998858327177665e-06,
      "loss": 0.7927,
      "step": 604
    },
    {
      "epoch": 0.03261807202932931,
      "grad_norm": 0.9217430949211121,
      "learning_rate": 9.99885379240874e-06,
      "loss": 0.8327,
      "step": 605
    },
    {
      "epoch": 0.032671986197972826,
      "grad_norm": 0.8202590942382812,
      "learning_rate": 9.99884924865255e-06,
      "loss": 0.7269,
      "step": 606
    },
    {
      "epoch": 0.032725900366616344,
      "grad_norm": 0.9598796367645264,
      "learning_rate": 9.998844695909102e-06,
      "loss": 0.9329,
      "step": 607
    },
    {
      "epoch": 0.03277981453525987,
      "grad_norm": 1.1016643047332764,
      "learning_rate": 9.998840134178407e-06,
      "loss": 0.9836,
      "step": 608
    },
    {
      "epoch": 0.032833728703903386,
      "grad_norm": 0.9639281630516052,
      "learning_rate": 9.998835563460471e-06,
      "loss": 0.8475,
      "step": 609
    },
    {
      "epoch": 0.032887642872546904,
      "grad_norm": 0.9266204833984375,
      "learning_rate": 9.998830983755304e-06,
      "loss": 0.7307,
      "step": 610
    },
    {
      "epoch": 0.03294155704119042,
      "grad_norm": 0.9282877445220947,
      "learning_rate": 9.99882639506291e-06,
      "loss": 0.8163,
      "step": 611
    },
    {
      "epoch": 0.03299547120983395,
      "grad_norm": 0.8939738869667053,
      "learning_rate": 9.998821797383302e-06,
      "loss": 0.6902,
      "step": 612
    },
    {
      "epoch": 0.033049385378477465,
      "grad_norm": 0.9041041731834412,
      "learning_rate": 9.998817190716488e-06,
      "loss": 0.8735,
      "step": 613
    },
    {
      "epoch": 0.03310329954712098,
      "grad_norm": 0.9973318576812744,
      "learning_rate": 9.998812575062473e-06,
      "loss": 0.9017,
      "step": 614
    },
    {
      "epoch": 0.0331572137157645,
      "grad_norm": 1.0416412353515625,
      "learning_rate": 9.998807950421268e-06,
      "loss": 0.9293,
      "step": 615
    },
    {
      "epoch": 0.03321112788440802,
      "grad_norm": 0.8686584234237671,
      "learning_rate": 9.998803316792882e-06,
      "loss": 0.8585,
      "step": 616
    },
    {
      "epoch": 0.033265042053051544,
      "grad_norm": 0.9907833337783813,
      "learning_rate": 9.998798674177319e-06,
      "loss": 0.9264,
      "step": 617
    },
    {
      "epoch": 0.03331895622169506,
      "grad_norm": 0.9927001595497131,
      "learning_rate": 9.998794022574592e-06,
      "loss": 0.895,
      "step": 618
    },
    {
      "epoch": 0.03337287039033858,
      "grad_norm": 0.9314623475074768,
      "learning_rate": 9.998789361984707e-06,
      "loss": 0.8353,
      "step": 619
    },
    {
      "epoch": 0.0334267845589821,
      "grad_norm": 0.9768248796463013,
      "learning_rate": 9.998784692407673e-06,
      "loss": 0.8917,
      "step": 620
    },
    {
      "epoch": 0.03348069872762562,
      "grad_norm": 0.9487942457199097,
      "learning_rate": 9.998780013843498e-06,
      "loss": 0.9022,
      "step": 621
    },
    {
      "epoch": 0.03353461289626914,
      "grad_norm": 1.0376895666122437,
      "learning_rate": 9.99877532629219e-06,
      "loss": 0.7692,
      "step": 622
    },
    {
      "epoch": 0.03358852706491266,
      "grad_norm": 1.021345853805542,
      "learning_rate": 9.99877062975376e-06,
      "loss": 1.0386,
      "step": 623
    },
    {
      "epoch": 0.033642441233556176,
      "grad_norm": 0.9979421496391296,
      "learning_rate": 9.998765924228214e-06,
      "loss": 0.9209,
      "step": 624
    },
    {
      "epoch": 0.0336963554021997,
      "grad_norm": 0.8552166819572449,
      "learning_rate": 9.998761209715559e-06,
      "loss": 0.8765,
      "step": 625
    },
    {
      "epoch": 0.03375026957084322,
      "grad_norm": 0.9737898707389832,
      "learning_rate": 9.998756486215809e-06,
      "loss": 0.7459,
      "step": 626
    },
    {
      "epoch": 0.03380418373948674,
      "grad_norm": 1.1067259311676025,
      "learning_rate": 9.998751753728967e-06,
      "loss": 0.8582,
      "step": 627
    },
    {
      "epoch": 0.033858097908130255,
      "grad_norm": 1.0689613819122314,
      "learning_rate": 9.998747012255044e-06,
      "loss": 0.8523,
      "step": 628
    },
    {
      "epoch": 0.03391201207677377,
      "grad_norm": 1.1880419254302979,
      "learning_rate": 9.998742261794048e-06,
      "loss": 0.9085,
      "step": 629
    },
    {
      "epoch": 0.0339659262454173,
      "grad_norm": 0.9569217562675476,
      "learning_rate": 9.998737502345987e-06,
      "loss": 0.9112,
      "step": 630
    },
    {
      "epoch": 0.034019840414060816,
      "grad_norm": 0.9955928921699524,
      "learning_rate": 9.99873273391087e-06,
      "loss": 0.9166,
      "step": 631
    },
    {
      "epoch": 0.034073754582704334,
      "grad_norm": 0.8906963467597961,
      "learning_rate": 9.998727956488708e-06,
      "loss": 0.882,
      "step": 632
    },
    {
      "epoch": 0.03412766875134785,
      "grad_norm": 0.9241589307785034,
      "learning_rate": 9.998723170079506e-06,
      "loss": 0.8488,
      "step": 633
    },
    {
      "epoch": 0.03418158291999138,
      "grad_norm": 0.9666005969047546,
      "learning_rate": 9.998718374683271e-06,
      "loss": 0.8432,
      "step": 634
    },
    {
      "epoch": 0.034235497088634895,
      "grad_norm": 0.9036918878555298,
      "learning_rate": 9.998713570300018e-06,
      "loss": 0.7979,
      "step": 635
    },
    {
      "epoch": 0.03428941125727841,
      "grad_norm": 0.8946508765220642,
      "learning_rate": 9.998708756929751e-06,
      "loss": 0.8854,
      "step": 636
    },
    {
      "epoch": 0.03434332542592193,
      "grad_norm": 1.0300164222717285,
      "learning_rate": 9.99870393457248e-06,
      "loss": 0.9116,
      "step": 637
    },
    {
      "epoch": 0.03439723959456545,
      "grad_norm": 1.0635035037994385,
      "learning_rate": 9.998699103228214e-06,
      "loss": 0.9138,
      "step": 638
    },
    {
      "epoch": 0.03445115376320897,
      "grad_norm": 1.0362621545791626,
      "learning_rate": 9.998694262896962e-06,
      "loss": 1.0177,
      "step": 639
    },
    {
      "epoch": 0.03450506793185249,
      "grad_norm": 0.9081454873085022,
      "learning_rate": 9.99868941357873e-06,
      "loss": 0.7802,
      "step": 640
    },
    {
      "epoch": 0.03455898210049601,
      "grad_norm": 0.9943915605545044,
      "learning_rate": 9.998684555273529e-06,
      "loss": 0.9356,
      "step": 641
    },
    {
      "epoch": 0.03461289626913953,
      "grad_norm": 0.9647786021232605,
      "learning_rate": 9.998679687981367e-06,
      "loss": 0.741,
      "step": 642
    },
    {
      "epoch": 0.03466681043778305,
      "grad_norm": 0.9655315279960632,
      "learning_rate": 9.998674811702255e-06,
      "loss": 0.8644,
      "step": 643
    },
    {
      "epoch": 0.03472072460642657,
      "grad_norm": 0.9162091612815857,
      "learning_rate": 9.998669926436197e-06,
      "loss": 0.8383,
      "step": 644
    },
    {
      "epoch": 0.03477463877507009,
      "grad_norm": 0.9509754776954651,
      "learning_rate": 9.998665032183207e-06,
      "loss": 0.8066,
      "step": 645
    },
    {
      "epoch": 0.034828552943713606,
      "grad_norm": 1.0545740127563477,
      "learning_rate": 9.998660128943292e-06,
      "loss": 0.8455,
      "step": 646
    },
    {
      "epoch": 0.03488246711235713,
      "grad_norm": 1.0928760766983032,
      "learning_rate": 9.998655216716458e-06,
      "loss": 0.8708,
      "step": 647
    },
    {
      "epoch": 0.03493638128100065,
      "grad_norm": 0.9743762016296387,
      "learning_rate": 9.998650295502717e-06,
      "loss": 0.878,
      "step": 648
    },
    {
      "epoch": 0.03499029544964417,
      "grad_norm": 1.016741156578064,
      "learning_rate": 9.998645365302077e-06,
      "loss": 0.867,
      "step": 649
    },
    {
      "epoch": 0.035044209618287685,
      "grad_norm": 1.125252366065979,
      "learning_rate": 9.998640426114548e-06,
      "loss": 0.9443,
      "step": 650
    },
    {
      "epoch": 0.0350981237869312,
      "grad_norm": 0.9555762410163879,
      "learning_rate": 9.998635477940135e-06,
      "loss": 0.8353,
      "step": 651
    },
    {
      "epoch": 0.03515203795557473,
      "grad_norm": 0.930173397064209,
      "learning_rate": 9.998630520778851e-06,
      "loss": 0.8383,
      "step": 652
    },
    {
      "epoch": 0.035205952124218245,
      "grad_norm": 1.1592127084732056,
      "learning_rate": 9.998625554630704e-06,
      "loss": 0.9708,
      "step": 653
    },
    {
      "epoch": 0.03525986629286176,
      "grad_norm": 0.9333894848823547,
      "learning_rate": 9.998620579495701e-06,
      "loss": 0.9055,
      "step": 654
    },
    {
      "epoch": 0.03531378046150528,
      "grad_norm": 0.9495646357536316,
      "learning_rate": 9.998615595373853e-06,
      "loss": 0.7993,
      "step": 655
    },
    {
      "epoch": 0.035367694630148806,
      "grad_norm": 1.0919233560562134,
      "learning_rate": 9.99861060226517e-06,
      "loss": 0.8852,
      "step": 656
    },
    {
      "epoch": 0.035421608798792324,
      "grad_norm": 0.907940685749054,
      "learning_rate": 9.998605600169657e-06,
      "loss": 0.8294,
      "step": 657
    },
    {
      "epoch": 0.03547552296743584,
      "grad_norm": 1.0423756837844849,
      "learning_rate": 9.998600589087328e-06,
      "loss": 0.8758,
      "step": 658
    },
    {
      "epoch": 0.03552943713607936,
      "grad_norm": 1.0387269258499146,
      "learning_rate": 9.998595569018186e-06,
      "loss": 0.9099,
      "step": 659
    },
    {
      "epoch": 0.03558335130472288,
      "grad_norm": 0.9186104536056519,
      "learning_rate": 9.998590539962245e-06,
      "loss": 0.9025,
      "step": 660
    },
    {
      "epoch": 0.0356372654733664,
      "grad_norm": 1.0173289775848389,
      "learning_rate": 9.998585501919514e-06,
      "loss": 0.8468,
      "step": 661
    },
    {
      "epoch": 0.03569117964200992,
      "grad_norm": 0.9579570889472961,
      "learning_rate": 9.998580454889996e-06,
      "loss": 0.8542,
      "step": 662
    },
    {
      "epoch": 0.03574509381065344,
      "grad_norm": 1.093515396118164,
      "learning_rate": 9.99857539887371e-06,
      "loss": 0.8932,
      "step": 663
    },
    {
      "epoch": 0.03579900797929696,
      "grad_norm": 1.0651243925094604,
      "learning_rate": 9.998570333870656e-06,
      "loss": 0.8822,
      "step": 664
    },
    {
      "epoch": 0.03585292214794048,
      "grad_norm": 0.973278284072876,
      "learning_rate": 9.998565259880845e-06,
      "loss": 0.8724,
      "step": 665
    },
    {
      "epoch": 0.035906836316584,
      "grad_norm": 0.961321234703064,
      "learning_rate": 9.998560176904291e-06,
      "loss": 0.947,
      "step": 666
    },
    {
      "epoch": 0.03596075048522752,
      "grad_norm": 1.0216654539108276,
      "learning_rate": 9.998555084940999e-06,
      "loss": 0.8528,
      "step": 667
    },
    {
      "epoch": 0.036014664653871035,
      "grad_norm": 0.9917817711830139,
      "learning_rate": 9.99854998399098e-06,
      "loss": 0.8608,
      "step": 668
    },
    {
      "epoch": 0.03606857882251455,
      "grad_norm": 1.0164326429367065,
      "learning_rate": 9.998544874054243e-06,
      "loss": 0.8752,
      "step": 669
    },
    {
      "epoch": 0.03612249299115808,
      "grad_norm": 0.9181317687034607,
      "learning_rate": 9.998539755130793e-06,
      "loss": 0.8032,
      "step": 670
    },
    {
      "epoch": 0.036176407159801596,
      "grad_norm": 1.0100011825561523,
      "learning_rate": 9.998534627220646e-06,
      "loss": 0.9205,
      "step": 671
    },
    {
      "epoch": 0.036230321328445114,
      "grad_norm": 0.9306463599205017,
      "learning_rate": 9.998529490323807e-06,
      "loss": 0.8209,
      "step": 672
    },
    {
      "epoch": 0.03628423549708863,
      "grad_norm": 1.8988754749298096,
      "learning_rate": 9.998524344440286e-06,
      "loss": 0.8455,
      "step": 673
    },
    {
      "epoch": 0.03633814966573216,
      "grad_norm": 0.9742317795753479,
      "learning_rate": 9.998519189570091e-06,
      "loss": 0.8733,
      "step": 674
    },
    {
      "epoch": 0.036392063834375675,
      "grad_norm": 0.9334224462509155,
      "learning_rate": 9.998514025713234e-06,
      "loss": 0.8761,
      "step": 675
    },
    {
      "epoch": 0.03644597800301919,
      "grad_norm": 0.9729838371276855,
      "learning_rate": 9.998508852869724e-06,
      "loss": 0.8916,
      "step": 676
    },
    {
      "epoch": 0.03649989217166271,
      "grad_norm": 0.9721505641937256,
      "learning_rate": 9.998503671039568e-06,
      "loss": 0.8735,
      "step": 677
    },
    {
      "epoch": 0.036553806340306236,
      "grad_norm": 0.9600850939750671,
      "learning_rate": 9.998498480222775e-06,
      "loss": 0.9157,
      "step": 678
    },
    {
      "epoch": 0.036607720508949754,
      "grad_norm": 0.9010732173919678,
      "learning_rate": 9.998493280419358e-06,
      "loss": 0.9215,
      "step": 679
    },
    {
      "epoch": 0.03666163467759327,
      "grad_norm": 0.8708087801933289,
      "learning_rate": 9.998488071629324e-06,
      "loss": 0.7218,
      "step": 680
    },
    {
      "epoch": 0.03671554884623679,
      "grad_norm": 0.9739180207252502,
      "learning_rate": 9.998482853852682e-06,
      "loss": 0.8845,
      "step": 681
    },
    {
      "epoch": 0.03676946301488031,
      "grad_norm": 0.9823595881462097,
      "learning_rate": 9.998477627089443e-06,
      "loss": 0.896,
      "step": 682
    },
    {
      "epoch": 0.03682337718352383,
      "grad_norm": 0.9629859328269958,
      "learning_rate": 9.998472391339612e-06,
      "loss": 0.8636,
      "step": 683
    },
    {
      "epoch": 0.03687729135216735,
      "grad_norm": 0.8644251823425293,
      "learning_rate": 9.998467146603206e-06,
      "loss": 0.9124,
      "step": 684
    },
    {
      "epoch": 0.03693120552081087,
      "grad_norm": 0.8987632989883423,
      "learning_rate": 9.99846189288023e-06,
      "loss": 0.801,
      "step": 685
    },
    {
      "epoch": 0.036985119689454386,
      "grad_norm": 0.9017630219459534,
      "learning_rate": 9.99845663017069e-06,
      "loss": 0.8675,
      "step": 686
    },
    {
      "epoch": 0.03703903385809791,
      "grad_norm": 0.8905850648880005,
      "learning_rate": 9.998451358474603e-06,
      "loss": 0.8512,
      "step": 687
    },
    {
      "epoch": 0.03709294802674143,
      "grad_norm": 0.9807800650596619,
      "learning_rate": 9.998446077791972e-06,
      "loss": 0.9258,
      "step": 688
    },
    {
      "epoch": 0.03714686219538495,
      "grad_norm": 0.8916336894035339,
      "learning_rate": 9.99844078812281e-06,
      "loss": 0.8236,
      "step": 689
    },
    {
      "epoch": 0.037200776364028465,
      "grad_norm": 0.9330187439918518,
      "learning_rate": 9.998435489467126e-06,
      "loss": 0.7812,
      "step": 690
    },
    {
      "epoch": 0.03725469053267198,
      "grad_norm": 0.9859142899513245,
      "learning_rate": 9.99843018182493e-06,
      "loss": 0.8699,
      "step": 691
    },
    {
      "epoch": 0.03730860470131551,
      "grad_norm": 0.9277002215385437,
      "learning_rate": 9.998424865196228e-06,
      "loss": 0.9276,
      "step": 692
    },
    {
      "epoch": 0.037362518869959026,
      "grad_norm": 0.9764281511306763,
      "learning_rate": 9.998419539581034e-06,
      "loss": 0.9482,
      "step": 693
    },
    {
      "epoch": 0.037416433038602544,
      "grad_norm": 1.0108616352081299,
      "learning_rate": 9.998414204979357e-06,
      "loss": 0.8582,
      "step": 694
    },
    {
      "epoch": 0.03747034720724606,
      "grad_norm": 1.2767362594604492,
      "learning_rate": 9.998408861391202e-06,
      "loss": 0.7833,
      "step": 695
    },
    {
      "epoch": 0.03752426137588959,
      "grad_norm": 0.8874560594558716,
      "learning_rate": 9.998403508816585e-06,
      "loss": 0.8935,
      "step": 696
    },
    {
      "epoch": 0.037578175544533104,
      "grad_norm": 0.8549458980560303,
      "learning_rate": 9.998398147255511e-06,
      "loss": 0.7747,
      "step": 697
    },
    {
      "epoch": 0.03763208971317662,
      "grad_norm": 0.9971988201141357,
      "learning_rate": 9.998392776707993e-06,
      "loss": 0.753,
      "step": 698
    },
    {
      "epoch": 0.03768600388182014,
      "grad_norm": 0.9822113513946533,
      "learning_rate": 9.998387397174037e-06,
      "loss": 0.9121,
      "step": 699
    },
    {
      "epoch": 0.037739918050463665,
      "grad_norm": 0.996151864528656,
      "learning_rate": 9.998382008653656e-06,
      "loss": 0.9356,
      "step": 700
    },
    {
      "epoch": 0.03779383221910718,
      "grad_norm": 1.7505156993865967,
      "learning_rate": 9.998376611146857e-06,
      "loss": 0.8351,
      "step": 701
    },
    {
      "epoch": 0.0378477463877507,
      "grad_norm": 1.070356011390686,
      "learning_rate": 9.998371204653651e-06,
      "loss": 0.9153,
      "step": 702
    },
    {
      "epoch": 0.03790166055639422,
      "grad_norm": 0.9383741617202759,
      "learning_rate": 9.998365789174048e-06,
      "loss": 0.8904,
      "step": 703
    },
    {
      "epoch": 0.03795557472503774,
      "grad_norm": 0.8444882035255432,
      "learning_rate": 9.998360364708058e-06,
      "loss": 0.8243,
      "step": 704
    },
    {
      "epoch": 0.03800948889368126,
      "grad_norm": 1.0012257099151611,
      "learning_rate": 9.99835493125569e-06,
      "loss": 0.9439,
      "step": 705
    },
    {
      "epoch": 0.03806340306232478,
      "grad_norm": 0.9745193719863892,
      "learning_rate": 9.998349488816954e-06,
      "loss": 0.8667,
      "step": 706
    },
    {
      "epoch": 0.0381173172309683,
      "grad_norm": 0.8363852500915527,
      "learning_rate": 9.998344037391859e-06,
      "loss": 0.8082,
      "step": 707
    },
    {
      "epoch": 0.038171231399611816,
      "grad_norm": 0.9389918446540833,
      "learning_rate": 9.998338576980417e-06,
      "loss": 0.8113,
      "step": 708
    },
    {
      "epoch": 0.03822514556825534,
      "grad_norm": 0.9216110110282898,
      "learning_rate": 9.998333107582635e-06,
      "loss": 0.8179,
      "step": 709
    },
    {
      "epoch": 0.03827905973689886,
      "grad_norm": 1.0292471647262573,
      "learning_rate": 9.998327629198526e-06,
      "loss": 0.8605,
      "step": 710
    },
    {
      "epoch": 0.03833297390554238,
      "grad_norm": 0.9812708497047424,
      "learning_rate": 9.998322141828097e-06,
      "loss": 0.9279,
      "step": 711
    },
    {
      "epoch": 0.038386888074185894,
      "grad_norm": 0.8186620473861694,
      "learning_rate": 9.998316645471358e-06,
      "loss": 0.7877,
      "step": 712
    },
    {
      "epoch": 0.03844080224282941,
      "grad_norm": 1.034134864807129,
      "learning_rate": 9.99831114012832e-06,
      "loss": 0.9867,
      "step": 713
    },
    {
      "epoch": 0.03849471641147294,
      "grad_norm": 1.1604938507080078,
      "learning_rate": 9.998305625798993e-06,
      "loss": 0.9134,
      "step": 714
    },
    {
      "epoch": 0.038548630580116455,
      "grad_norm": 0.8452483415603638,
      "learning_rate": 9.998300102483388e-06,
      "loss": 0.8732,
      "step": 715
    },
    {
      "epoch": 0.03860254474875997,
      "grad_norm": 0.8881269693374634,
      "learning_rate": 9.998294570181512e-06,
      "loss": 0.847,
      "step": 716
    },
    {
      "epoch": 0.03865645891740349,
      "grad_norm": 0.8822013735771179,
      "learning_rate": 9.998289028893375e-06,
      "loss": 0.8404,
      "step": 717
    },
    {
      "epoch": 0.038710373086047016,
      "grad_norm": 1.0011916160583496,
      "learning_rate": 9.998283478618991e-06,
      "loss": 0.8133,
      "step": 718
    },
    {
      "epoch": 0.038764287254690534,
      "grad_norm": 1.0004018545150757,
      "learning_rate": 9.998277919358367e-06,
      "loss": 0.9556,
      "step": 719
    },
    {
      "epoch": 0.03881820142333405,
      "grad_norm": 0.8176954984664917,
      "learning_rate": 9.998272351111513e-06,
      "loss": 0.7977,
      "step": 720
    },
    {
      "epoch": 0.03887211559197757,
      "grad_norm": 0.9160690307617188,
      "learning_rate": 9.99826677387844e-06,
      "loss": 0.9239,
      "step": 721
    },
    {
      "epoch": 0.03892602976062109,
      "grad_norm": 1.2158405780792236,
      "learning_rate": 9.998261187659157e-06,
      "loss": 0.9023,
      "step": 722
    },
    {
      "epoch": 0.03897994392926461,
      "grad_norm": 0.9564448595046997,
      "learning_rate": 9.998255592453674e-06,
      "loss": 0.8585,
      "step": 723
    },
    {
      "epoch": 0.03903385809790813,
      "grad_norm": 0.8902252316474915,
      "learning_rate": 9.998249988262002e-06,
      "loss": 0.8388,
      "step": 724
    },
    {
      "epoch": 0.03908777226655165,
      "grad_norm": 0.8738620281219482,
      "learning_rate": 9.998244375084152e-06,
      "loss": 0.9545,
      "step": 725
    },
    {
      "epoch": 0.03914168643519517,
      "grad_norm": 0.9670735001564026,
      "learning_rate": 9.99823875292013e-06,
      "loss": 0.8335,
      "step": 726
    },
    {
      "epoch": 0.03919560060383869,
      "grad_norm": 0.8719429969787598,
      "learning_rate": 9.998233121769952e-06,
      "loss": 0.8546,
      "step": 727
    },
    {
      "epoch": 0.03924951477248221,
      "grad_norm": 1.318429708480835,
      "learning_rate": 9.998227481633622e-06,
      "loss": 1.0658,
      "step": 728
    },
    {
      "epoch": 0.03930342894112573,
      "grad_norm": 0.962630569934845,
      "learning_rate": 9.998221832511155e-06,
      "loss": 0.9049,
      "step": 729
    },
    {
      "epoch": 0.039357343109769245,
      "grad_norm": 0.9639857411384583,
      "learning_rate": 9.998216174402558e-06,
      "loss": 0.9114,
      "step": 730
    },
    {
      "epoch": 0.03941125727841277,
      "grad_norm": 1.1621571779251099,
      "learning_rate": 9.998210507307843e-06,
      "loss": 0.8776,
      "step": 731
    },
    {
      "epoch": 0.03946517144705629,
      "grad_norm": 1.170089840888977,
      "learning_rate": 9.998204831227019e-06,
      "loss": 0.9928,
      "step": 732
    },
    {
      "epoch": 0.039519085615699806,
      "grad_norm": 0.8257297873497009,
      "learning_rate": 9.998199146160098e-06,
      "loss": 0.7885,
      "step": 733
    },
    {
      "epoch": 0.039572999784343324,
      "grad_norm": 0.8887513279914856,
      "learning_rate": 9.998193452107088e-06,
      "loss": 0.8389,
      "step": 734
    },
    {
      "epoch": 0.03962691395298684,
      "grad_norm": 0.9321185350418091,
      "learning_rate": 9.998187749068001e-06,
      "loss": 0.9083,
      "step": 735
    },
    {
      "epoch": 0.03968082812163037,
      "grad_norm": 0.9926772713661194,
      "learning_rate": 9.998182037042847e-06,
      "loss": 0.9102,
      "step": 736
    },
    {
      "epoch": 0.039734742290273885,
      "grad_norm": 1.0760009288787842,
      "learning_rate": 9.998176316031634e-06,
      "loss": 0.7781,
      "step": 737
    },
    {
      "epoch": 0.0397886564589174,
      "grad_norm": 1.0998133420944214,
      "learning_rate": 9.998170586034376e-06,
      "loss": 0.9725,
      "step": 738
    },
    {
      "epoch": 0.03984257062756092,
      "grad_norm": 0.9367475509643555,
      "learning_rate": 9.99816484705108e-06,
      "loss": 0.8277,
      "step": 739
    },
    {
      "epoch": 0.039896484796204446,
      "grad_norm": 0.942954957485199,
      "learning_rate": 9.998159099081758e-06,
      "loss": 0.8542,
      "step": 740
    },
    {
      "epoch": 0.039950398964847963,
      "grad_norm": 0.9841166138648987,
      "learning_rate": 9.998153342126421e-06,
      "loss": 0.9179,
      "step": 741
    },
    {
      "epoch": 0.04000431313349148,
      "grad_norm": 0.9215245246887207,
      "learning_rate": 9.998147576185077e-06,
      "loss": 0.8899,
      "step": 742
    },
    {
      "epoch": 0.040058227302135,
      "grad_norm": 1.0368192195892334,
      "learning_rate": 9.998141801257739e-06,
      "loss": 0.9828,
      "step": 743
    },
    {
      "epoch": 0.04011214147077852,
      "grad_norm": 0.9696660041809082,
      "learning_rate": 9.998136017344416e-06,
      "loss": 0.9431,
      "step": 744
    },
    {
      "epoch": 0.04016605563942204,
      "grad_norm": 1.111257791519165,
      "learning_rate": 9.998130224445117e-06,
      "loss": 0.9666,
      "step": 745
    },
    {
      "epoch": 0.04021996980806556,
      "grad_norm": 0.9260644316673279,
      "learning_rate": 9.998124422559856e-06,
      "loss": 0.8941,
      "step": 746
    },
    {
      "epoch": 0.04027388397670908,
      "grad_norm": 0.8622020483016968,
      "learning_rate": 9.99811861168864e-06,
      "loss": 0.8148,
      "step": 747
    },
    {
      "epoch": 0.040327798145352596,
      "grad_norm": 0.8767471313476562,
      "learning_rate": 9.998112791831483e-06,
      "loss": 0.7093,
      "step": 748
    },
    {
      "epoch": 0.04038171231399612,
      "grad_norm": 0.902917206287384,
      "learning_rate": 9.998106962988391e-06,
      "loss": 0.7677,
      "step": 749
    },
    {
      "epoch": 0.04043562648263964,
      "grad_norm": 1.351694941520691,
      "learning_rate": 9.998101125159377e-06,
      "loss": 1.0382,
      "step": 750
    },
    {
      "epoch": 0.04048954065128316,
      "grad_norm": 0.8547930121421814,
      "learning_rate": 9.998095278344452e-06,
      "loss": 0.7974,
      "step": 751
    },
    {
      "epoch": 0.040543454819926675,
      "grad_norm": 0.941149115562439,
      "learning_rate": 9.998089422543626e-06,
      "loss": 0.8518,
      "step": 752
    },
    {
      "epoch": 0.0405973689885702,
      "grad_norm": 0.8671521544456482,
      "learning_rate": 9.998083557756908e-06,
      "loss": 0.8049,
      "step": 753
    },
    {
      "epoch": 0.04065128315721372,
      "grad_norm": 0.9877942800521851,
      "learning_rate": 9.998077683984311e-06,
      "loss": 0.8874,
      "step": 754
    },
    {
      "epoch": 0.040705197325857236,
      "grad_norm": 1.2130393981933594,
      "learning_rate": 9.998071801225843e-06,
      "loss": 0.9794,
      "step": 755
    },
    {
      "epoch": 0.040759111494500753,
      "grad_norm": 0.9422823786735535,
      "learning_rate": 9.998065909481518e-06,
      "loss": 0.899,
      "step": 756
    },
    {
      "epoch": 0.04081302566314427,
      "grad_norm": 0.9770492911338806,
      "learning_rate": 9.998060008751343e-06,
      "loss": 0.8434,
      "step": 757
    },
    {
      "epoch": 0.040866939831787796,
      "grad_norm": 0.9227531552314758,
      "learning_rate": 9.998054099035332e-06,
      "loss": 0.8797,
      "step": 758
    },
    {
      "epoch": 0.040920854000431314,
      "grad_norm": 1.0452102422714233,
      "learning_rate": 9.998048180333492e-06,
      "loss": 0.8702,
      "step": 759
    },
    {
      "epoch": 0.04097476816907483,
      "grad_norm": 1.034125566482544,
      "learning_rate": 9.998042252645837e-06,
      "loss": 0.9041,
      "step": 760
    },
    {
      "epoch": 0.04102868233771835,
      "grad_norm": 0.886029064655304,
      "learning_rate": 9.998036315972375e-06,
      "loss": 0.7805,
      "step": 761
    },
    {
      "epoch": 0.041082596506361875,
      "grad_norm": 0.9845888614654541,
      "learning_rate": 9.998030370313116e-06,
      "loss": 0.9836,
      "step": 762
    },
    {
      "epoch": 0.04113651067500539,
      "grad_norm": 0.9223973155021667,
      "learning_rate": 9.998024415668075e-06,
      "loss": 0.768,
      "step": 763
    },
    {
      "epoch": 0.04119042484364891,
      "grad_norm": 1.0607362985610962,
      "learning_rate": 9.99801845203726e-06,
      "loss": 0.865,
      "step": 764
    },
    {
      "epoch": 0.04124433901229243,
      "grad_norm": 0.9620907306671143,
      "learning_rate": 9.998012479420683e-06,
      "loss": 0.7645,
      "step": 765
    },
    {
      "epoch": 0.04129825318093595,
      "grad_norm": 0.9490310549736023,
      "learning_rate": 9.99800649781835e-06,
      "loss": 0.9124,
      "step": 766
    },
    {
      "epoch": 0.04135216734957947,
      "grad_norm": 0.9684557914733887,
      "learning_rate": 9.99800050723028e-06,
      "loss": 0.876,
      "step": 767
    },
    {
      "epoch": 0.04140608151822299,
      "grad_norm": 0.9633080959320068,
      "learning_rate": 9.997994507656476e-06,
      "loss": 0.8976,
      "step": 768
    },
    {
      "epoch": 0.04145999568686651,
      "grad_norm": 0.9495208263397217,
      "learning_rate": 9.997988499096953e-06,
      "loss": 0.9049,
      "step": 769
    },
    {
      "epoch": 0.041513909855510026,
      "grad_norm": 1.0614326000213623,
      "learning_rate": 9.997982481551721e-06,
      "loss": 0.905,
      "step": 770
    },
    {
      "epoch": 0.04156782402415355,
      "grad_norm": 0.820672869682312,
      "learning_rate": 9.99797645502079e-06,
      "loss": 0.8306,
      "step": 771
    },
    {
      "epoch": 0.04162173819279707,
      "grad_norm": 0.9719771146774292,
      "learning_rate": 9.997970419504171e-06,
      "loss": 0.828,
      "step": 772
    },
    {
      "epoch": 0.041675652361440586,
      "grad_norm": 0.893326997756958,
      "learning_rate": 9.997964375001875e-06,
      "loss": 0.8416,
      "step": 773
    },
    {
      "epoch": 0.041729566530084104,
      "grad_norm": 0.858121395111084,
      "learning_rate": 9.997958321513915e-06,
      "loss": 0.8779,
      "step": 774
    },
    {
      "epoch": 0.04178348069872762,
      "grad_norm": 0.9703636765480042,
      "learning_rate": 9.997952259040297e-06,
      "loss": 0.8623,
      "step": 775
    },
    {
      "epoch": 0.04183739486737115,
      "grad_norm": 0.9626398086547852,
      "learning_rate": 9.997946187581039e-06,
      "loss": 0.8309,
      "step": 776
    },
    {
      "epoch": 0.041891309036014665,
      "grad_norm": 0.9132344722747803,
      "learning_rate": 9.997940107136143e-06,
      "loss": 0.8798,
      "step": 777
    },
    {
      "epoch": 0.04194522320465818,
      "grad_norm": 0.9608821272850037,
      "learning_rate": 9.997934017705629e-06,
      "loss": 0.8764,
      "step": 778
    },
    {
      "epoch": 0.0419991373733017,
      "grad_norm": 1.0852513313293457,
      "learning_rate": 9.997927919289501e-06,
      "loss": 0.8908,
      "step": 779
    },
    {
      "epoch": 0.042053051541945226,
      "grad_norm": 0.9690573215484619,
      "learning_rate": 9.997921811887774e-06,
      "loss": 0.8556,
      "step": 780
    },
    {
      "epoch": 0.042106965710588744,
      "grad_norm": 0.9107050895690918,
      "learning_rate": 9.997915695500458e-06,
      "loss": 0.9249,
      "step": 781
    },
    {
      "epoch": 0.04216087987923226,
      "grad_norm": 1.029974102973938,
      "learning_rate": 9.997909570127564e-06,
      "loss": 0.8369,
      "step": 782
    },
    {
      "epoch": 0.04221479404787578,
      "grad_norm": 0.8179258704185486,
      "learning_rate": 9.997903435769101e-06,
      "loss": 0.7729,
      "step": 783
    },
    {
      "epoch": 0.042268708216519305,
      "grad_norm": 1.0664961338043213,
      "learning_rate": 9.997897292425082e-06,
      "loss": 0.8815,
      "step": 784
    },
    {
      "epoch": 0.04232262238516282,
      "grad_norm": 0.9794465899467468,
      "learning_rate": 9.997891140095519e-06,
      "loss": 0.9244,
      "step": 785
    },
    {
      "epoch": 0.04237653655380634,
      "grad_norm": 0.875953197479248,
      "learning_rate": 9.99788497878042e-06,
      "loss": 0.9191,
      "step": 786
    },
    {
      "epoch": 0.04243045072244986,
      "grad_norm": 0.9880902767181396,
      "learning_rate": 9.9978788084798e-06,
      "loss": 0.8639,
      "step": 787
    },
    {
      "epoch": 0.042484364891093376,
      "grad_norm": 1.0391566753387451,
      "learning_rate": 9.997872629193666e-06,
      "loss": 0.9943,
      "step": 788
    },
    {
      "epoch": 0.0425382790597369,
      "grad_norm": 0.9321290850639343,
      "learning_rate": 9.997866440922033e-06,
      "loss": 0.7809,
      "step": 789
    },
    {
      "epoch": 0.04259219322838042,
      "grad_norm": 0.8898556232452393,
      "learning_rate": 9.99786024366491e-06,
      "loss": 0.9353,
      "step": 790
    },
    {
      "epoch": 0.04264610739702394,
      "grad_norm": 1.1177983283996582,
      "learning_rate": 9.997854037422306e-06,
      "loss": 0.8157,
      "step": 791
    },
    {
      "epoch": 0.042700021565667455,
      "grad_norm": 0.8821296691894531,
      "learning_rate": 9.997847822194236e-06,
      "loss": 0.8729,
      "step": 792
    },
    {
      "epoch": 0.04275393573431098,
      "grad_norm": 0.8545325398445129,
      "learning_rate": 9.997841597980709e-06,
      "loss": 0.8415,
      "step": 793
    },
    {
      "epoch": 0.0428078499029545,
      "grad_norm": 0.9313606023788452,
      "learning_rate": 9.997835364781739e-06,
      "loss": 0.8411,
      "step": 794
    },
    {
      "epoch": 0.042861764071598016,
      "grad_norm": 0.9587781429290771,
      "learning_rate": 9.997829122597332e-06,
      "loss": 0.8086,
      "step": 795
    },
    {
      "epoch": 0.042915678240241534,
      "grad_norm": 0.9708360433578491,
      "learning_rate": 9.997822871427504e-06,
      "loss": 0.8715,
      "step": 796
    },
    {
      "epoch": 0.04296959240888505,
      "grad_norm": 0.8868080973625183,
      "learning_rate": 9.997816611272265e-06,
      "loss": 0.8549,
      "step": 797
    },
    {
      "epoch": 0.04302350657752858,
      "grad_norm": 0.9147778153419495,
      "learning_rate": 9.997810342131624e-06,
      "loss": 0.7854,
      "step": 798
    },
    {
      "epoch": 0.043077420746172095,
      "grad_norm": 0.9853960275650024,
      "learning_rate": 9.997804064005596e-06,
      "loss": 0.8243,
      "step": 799
    },
    {
      "epoch": 0.04313133491481561,
      "grad_norm": 1.0076130628585815,
      "learning_rate": 9.997797776894189e-06,
      "loss": 0.9077,
      "step": 800
    },
    {
      "epoch": 0.04318524908345913,
      "grad_norm": 0.9694076776504517,
      "learning_rate": 9.997791480797417e-06,
      "loss": 0.8767,
      "step": 801
    },
    {
      "epoch": 0.043239163252102655,
      "grad_norm": 1.114001750946045,
      "learning_rate": 9.99778517571529e-06,
      "loss": 0.8211,
      "step": 802
    },
    {
      "epoch": 0.04329307742074617,
      "grad_norm": 0.9701128005981445,
      "learning_rate": 9.997778861647817e-06,
      "loss": 0.9084,
      "step": 803
    },
    {
      "epoch": 0.04334699158938969,
      "grad_norm": 0.868299126625061,
      "learning_rate": 9.997772538595015e-06,
      "loss": 0.7556,
      "step": 804
    },
    {
      "epoch": 0.04340090575803321,
      "grad_norm": 0.9160446524620056,
      "learning_rate": 9.997766206556888e-06,
      "loss": 0.821,
      "step": 805
    },
    {
      "epoch": 0.043454819926676734,
      "grad_norm": 0.934198260307312,
      "learning_rate": 9.997759865533454e-06,
      "loss": 0.9113,
      "step": 806
    },
    {
      "epoch": 0.04350873409532025,
      "grad_norm": 0.8949079513549805,
      "learning_rate": 9.997753515524722e-06,
      "loss": 0.7821,
      "step": 807
    },
    {
      "epoch": 0.04356264826396377,
      "grad_norm": 0.9035944938659668,
      "learning_rate": 9.997747156530702e-06,
      "loss": 0.8233,
      "step": 808
    },
    {
      "epoch": 0.04361656243260729,
      "grad_norm": 0.9681552052497864,
      "learning_rate": 9.99774078855141e-06,
      "loss": 0.9241,
      "step": 809
    },
    {
      "epoch": 0.043670476601250806,
      "grad_norm": 0.906092643737793,
      "learning_rate": 9.99773441158685e-06,
      "loss": 0.8948,
      "step": 810
    },
    {
      "epoch": 0.04372439076989433,
      "grad_norm": 0.9229143261909485,
      "learning_rate": 9.997728025637039e-06,
      "loss": 0.8897,
      "step": 811
    },
    {
      "epoch": 0.04377830493853785,
      "grad_norm": 0.9263061881065369,
      "learning_rate": 9.997721630701986e-06,
      "loss": 0.7923,
      "step": 812
    },
    {
      "epoch": 0.04383221910718137,
      "grad_norm": 0.8474372029304504,
      "learning_rate": 9.997715226781706e-06,
      "loss": 0.796,
      "step": 813
    },
    {
      "epoch": 0.043886133275824885,
      "grad_norm": 0.9960548877716064,
      "learning_rate": 9.997708813876206e-06,
      "loss": 0.9166,
      "step": 814
    },
    {
      "epoch": 0.04394004744446841,
      "grad_norm": 0.9843032956123352,
      "learning_rate": 9.997702391985499e-06,
      "loss": 0.9354,
      "step": 815
    },
    {
      "epoch": 0.04399396161311193,
      "grad_norm": 0.9313154220581055,
      "learning_rate": 9.997695961109599e-06,
      "loss": 0.8972,
      "step": 816
    },
    {
      "epoch": 0.044047875781755445,
      "grad_norm": 0.8846973180770874,
      "learning_rate": 9.997689521248515e-06,
      "loss": 0.8599,
      "step": 817
    },
    {
      "epoch": 0.04410178995039896,
      "grad_norm": 0.8113641738891602,
      "learning_rate": 9.99768307240226e-06,
      "loss": 0.8509,
      "step": 818
    },
    {
      "epoch": 0.04415570411904248,
      "grad_norm": 1.0659984350204468,
      "learning_rate": 9.997676614570844e-06,
      "loss": 0.938,
      "step": 819
    },
    {
      "epoch": 0.044209618287686006,
      "grad_norm": 0.9183745384216309,
      "learning_rate": 9.99767014775428e-06,
      "loss": 0.8761,
      "step": 820
    },
    {
      "epoch": 0.044263532456329524,
      "grad_norm": 0.87090003490448,
      "learning_rate": 9.997663671952578e-06,
      "loss": 0.8535,
      "step": 821
    },
    {
      "epoch": 0.04431744662497304,
      "grad_norm": 0.9857214093208313,
      "learning_rate": 9.997657187165753e-06,
      "loss": 0.9434,
      "step": 822
    },
    {
      "epoch": 0.04437136079361656,
      "grad_norm": 1.0443209409713745,
      "learning_rate": 9.997650693393812e-06,
      "loss": 0.8994,
      "step": 823
    },
    {
      "epoch": 0.044425274962260085,
      "grad_norm": 0.8348391652107239,
      "learning_rate": 9.99764419063677e-06,
      "loss": 0.8383,
      "step": 824
    },
    {
      "epoch": 0.0444791891309036,
      "grad_norm": 1.2708821296691895,
      "learning_rate": 9.997637678894639e-06,
      "loss": 0.8733,
      "step": 825
    },
    {
      "epoch": 0.04453310329954712,
      "grad_norm": 0.9863126277923584,
      "learning_rate": 9.997631158167428e-06,
      "loss": 0.9364,
      "step": 826
    },
    {
      "epoch": 0.04458701746819064,
      "grad_norm": 1.0223352909088135,
      "learning_rate": 9.99762462845515e-06,
      "loss": 0.9139,
      "step": 827
    },
    {
      "epoch": 0.04464093163683416,
      "grad_norm": 0.8559738397598267,
      "learning_rate": 9.997618089757818e-06,
      "loss": 0.7461,
      "step": 828
    },
    {
      "epoch": 0.04469484580547768,
      "grad_norm": 0.9347368478775024,
      "learning_rate": 9.997611542075442e-06,
      "loss": 0.9275,
      "step": 829
    },
    {
      "epoch": 0.0447487599741212,
      "grad_norm": 1.0208019018173218,
      "learning_rate": 9.997604985408036e-06,
      "loss": 0.8338,
      "step": 830
    },
    {
      "epoch": 0.04480267414276472,
      "grad_norm": 0.9792174100875854,
      "learning_rate": 9.997598419755607e-06,
      "loss": 0.9437,
      "step": 831
    },
    {
      "epoch": 0.044856588311408235,
      "grad_norm": 0.851665198802948,
      "learning_rate": 9.997591845118173e-06,
      "loss": 0.8008,
      "step": 832
    },
    {
      "epoch": 0.04491050248005176,
      "grad_norm": 0.9315025806427002,
      "learning_rate": 9.997585261495742e-06,
      "loss": 0.8389,
      "step": 833
    },
    {
      "epoch": 0.04496441664869528,
      "grad_norm": 0.9658921360969543,
      "learning_rate": 9.997578668888326e-06,
      "loss": 0.9252,
      "step": 834
    },
    {
      "epoch": 0.045018330817338796,
      "grad_norm": 0.8989397287368774,
      "learning_rate": 9.997572067295938e-06,
      "loss": 0.8648,
      "step": 835
    },
    {
      "epoch": 0.045072244985982314,
      "grad_norm": 0.8874988555908203,
      "learning_rate": 9.99756545671859e-06,
      "loss": 0.7801,
      "step": 836
    },
    {
      "epoch": 0.04512615915462584,
      "grad_norm": 0.9186223745346069,
      "learning_rate": 9.997558837156293e-06,
      "loss": 0.767,
      "step": 837
    },
    {
      "epoch": 0.04518007332326936,
      "grad_norm": 1.163044810295105,
      "learning_rate": 9.997552208609059e-06,
      "loss": 0.8938,
      "step": 838
    },
    {
      "epoch": 0.045233987491912875,
      "grad_norm": 0.8315468430519104,
      "learning_rate": 9.997545571076901e-06,
      "loss": 0.725,
      "step": 839
    },
    {
      "epoch": 0.04528790166055639,
      "grad_norm": 1.0088660717010498,
      "learning_rate": 9.99753892455983e-06,
      "loss": 0.8533,
      "step": 840
    },
    {
      "epoch": 0.04534181582919991,
      "grad_norm": 0.9268692135810852,
      "learning_rate": 9.997532269057857e-06,
      "loss": 0.8739,
      "step": 841
    },
    {
      "epoch": 0.045395729997843436,
      "grad_norm": 1.0793242454528809,
      "learning_rate": 9.997525604570995e-06,
      "loss": 0.9605,
      "step": 842
    },
    {
      "epoch": 0.045449644166486954,
      "grad_norm": 1.101798176765442,
      "learning_rate": 9.997518931099258e-06,
      "loss": 0.9525,
      "step": 843
    },
    {
      "epoch": 0.04550355833513047,
      "grad_norm": 0.9046466946601868,
      "learning_rate": 9.997512248642654e-06,
      "loss": 0.8853,
      "step": 844
    },
    {
      "epoch": 0.04555747250377399,
      "grad_norm": 0.9629097580909729,
      "learning_rate": 9.997505557201198e-06,
      "loss": 0.8882,
      "step": 845
    },
    {
      "epoch": 0.045611386672417514,
      "grad_norm": 1.1880977153778076,
      "learning_rate": 9.997498856774898e-06,
      "loss": 0.8812,
      "step": 846
    },
    {
      "epoch": 0.04566530084106103,
      "grad_norm": 0.8678451180458069,
      "learning_rate": 9.997492147363772e-06,
      "loss": 0.887,
      "step": 847
    },
    {
      "epoch": 0.04571921500970455,
      "grad_norm": 1.3359739780426025,
      "learning_rate": 9.99748542896783e-06,
      "loss": 0.8141,
      "step": 848
    },
    {
      "epoch": 0.04577312917834807,
      "grad_norm": 0.9263296127319336,
      "learning_rate": 9.99747870158708e-06,
      "loss": 0.9357,
      "step": 849
    },
    {
      "epoch": 0.045827043346991586,
      "grad_norm": 0.9199776649475098,
      "learning_rate": 9.997471965221541e-06,
      "loss": 0.8352,
      "step": 850
    },
    {
      "epoch": 0.04588095751563511,
      "grad_norm": 0.8880730867385864,
      "learning_rate": 9.997465219871218e-06,
      "loss": 0.7802,
      "step": 851
    },
    {
      "epoch": 0.04593487168427863,
      "grad_norm": 0.8561250567436218,
      "learning_rate": 9.99745846553613e-06,
      "loss": 0.7987,
      "step": 852
    },
    {
      "epoch": 0.04598878585292215,
      "grad_norm": 0.8975661396980286,
      "learning_rate": 9.997451702216283e-06,
      "loss": 0.8325,
      "step": 853
    },
    {
      "epoch": 0.046042700021565665,
      "grad_norm": 0.9350215196609497,
      "learning_rate": 9.997444929911693e-06,
      "loss": 0.7708,
      "step": 854
    },
    {
      "epoch": 0.04609661419020919,
      "grad_norm": 1.0229014158248901,
      "learning_rate": 9.99743814862237e-06,
      "loss": 0.9643,
      "step": 855
    },
    {
      "epoch": 0.04615052835885271,
      "grad_norm": 0.9249217510223389,
      "learning_rate": 9.997431358348329e-06,
      "loss": 0.8411,
      "step": 856
    },
    {
      "epoch": 0.046204442527496226,
      "grad_norm": 0.9823042154312134,
      "learning_rate": 9.99742455908958e-06,
      "loss": 0.9406,
      "step": 857
    },
    {
      "epoch": 0.046258356696139744,
      "grad_norm": 1.2525794506072998,
      "learning_rate": 9.997417750846134e-06,
      "loss": 0.8507,
      "step": 858
    },
    {
      "epoch": 0.04631227086478327,
      "grad_norm": 0.9583309888839722,
      "learning_rate": 9.997410933618006e-06,
      "loss": 0.8504,
      "step": 859
    },
    {
      "epoch": 0.046366185033426786,
      "grad_norm": 0.9264401793479919,
      "learning_rate": 9.997404107405207e-06,
      "loss": 0.8595,
      "step": 860
    },
    {
      "epoch": 0.046420099202070304,
      "grad_norm": 0.9833316206932068,
      "learning_rate": 9.99739727220775e-06,
      "loss": 0.9025,
      "step": 861
    },
    {
      "epoch": 0.04647401337071382,
      "grad_norm": 1.0220664739608765,
      "learning_rate": 9.997390428025645e-06,
      "loss": 0.8671,
      "step": 862
    },
    {
      "epoch": 0.04652792753935734,
      "grad_norm": 1.0774664878845215,
      "learning_rate": 9.997383574858908e-06,
      "loss": 0.8463,
      "step": 863
    },
    {
      "epoch": 0.046581841708000865,
      "grad_norm": 0.8821879029273987,
      "learning_rate": 9.997376712707547e-06,
      "loss": 0.7565,
      "step": 864
    },
    {
      "epoch": 0.04663575587664438,
      "grad_norm": 0.9233925938606262,
      "learning_rate": 9.997369841571577e-06,
      "loss": 0.9151,
      "step": 865
    },
    {
      "epoch": 0.0466896700452879,
      "grad_norm": 1.0006109476089478,
      "learning_rate": 9.997362961451015e-06,
      "loss": 0.8339,
      "step": 866
    },
    {
      "epoch": 0.04674358421393142,
      "grad_norm": 0.865035891532898,
      "learning_rate": 9.997356072345863e-06,
      "loss": 0.8997,
      "step": 867
    },
    {
      "epoch": 0.046797498382574944,
      "grad_norm": 1.0450654029846191,
      "learning_rate": 9.99734917425614e-06,
      "loss": 0.7966,
      "step": 868
    },
    {
      "epoch": 0.04685141255121846,
      "grad_norm": 0.8878824710845947,
      "learning_rate": 9.997342267181857e-06,
      "loss": 0.831,
      "step": 869
    },
    {
      "epoch": 0.04690532671986198,
      "grad_norm": 1.0056546926498413,
      "learning_rate": 9.997335351123028e-06,
      "loss": 0.8178,
      "step": 870
    },
    {
      "epoch": 0.0469592408885055,
      "grad_norm": 1.0531659126281738,
      "learning_rate": 9.997328426079661e-06,
      "loss": 0.7773,
      "step": 871
    },
    {
      "epoch": 0.047013155057149016,
      "grad_norm": 0.911021888256073,
      "learning_rate": 9.997321492051775e-06,
      "loss": 0.9001,
      "step": 872
    },
    {
      "epoch": 0.04706706922579254,
      "grad_norm": 0.920103132724762,
      "learning_rate": 9.997314549039379e-06,
      "loss": 0.7222,
      "step": 873
    },
    {
      "epoch": 0.04712098339443606,
      "grad_norm": 0.9449265599250793,
      "learning_rate": 9.997307597042483e-06,
      "loss": 0.9197,
      "step": 874
    },
    {
      "epoch": 0.047174897563079576,
      "grad_norm": 1.013066291809082,
      "learning_rate": 9.997300636061103e-06,
      "loss": 0.8854,
      "step": 875
    },
    {
      "epoch": 0.047228811731723094,
      "grad_norm": 0.8990256786346436,
      "learning_rate": 9.99729366609525e-06,
      "loss": 0.81,
      "step": 876
    },
    {
      "epoch": 0.04728272590036662,
      "grad_norm": 1.0211769342422485,
      "learning_rate": 9.997286687144938e-06,
      "loss": 0.8335,
      "step": 877
    },
    {
      "epoch": 0.04733664006901014,
      "grad_norm": 1.14606773853302,
      "learning_rate": 9.997279699210178e-06,
      "loss": 1.0956,
      "step": 878
    },
    {
      "epoch": 0.047390554237653655,
      "grad_norm": 0.982725977897644,
      "learning_rate": 9.997272702290981e-06,
      "loss": 0.8289,
      "step": 879
    },
    {
      "epoch": 0.04744446840629717,
      "grad_norm": 0.8667361736297607,
      "learning_rate": 9.997265696387364e-06,
      "loss": 0.8056,
      "step": 880
    },
    {
      "epoch": 0.04749838257494069,
      "grad_norm": 0.9029837250709534,
      "learning_rate": 9.997258681499338e-06,
      "loss": 0.8461,
      "step": 881
    },
    {
      "epoch": 0.047552296743584216,
      "grad_norm": 0.8767060041427612,
      "learning_rate": 9.997251657626915e-06,
      "loss": 0.8162,
      "step": 882
    },
    {
      "epoch": 0.047606210912227734,
      "grad_norm": 1.4750713109970093,
      "learning_rate": 9.997244624770104e-06,
      "loss": 0.8677,
      "step": 883
    },
    {
      "epoch": 0.04766012508087125,
      "grad_norm": 1.001286506652832,
      "learning_rate": 9.997237582928924e-06,
      "loss": 0.7673,
      "step": 884
    },
    {
      "epoch": 0.04771403924951477,
      "grad_norm": 0.9560269713401794,
      "learning_rate": 9.997230532103384e-06,
      "loss": 0.8597,
      "step": 885
    },
    {
      "epoch": 0.047767953418158295,
      "grad_norm": 0.834237277507782,
      "learning_rate": 9.997223472293499e-06,
      "loss": 0.7629,
      "step": 886
    },
    {
      "epoch": 0.04782186758680181,
      "grad_norm": 0.9642406702041626,
      "learning_rate": 9.997216403499278e-06,
      "loss": 0.83,
      "step": 887
    },
    {
      "epoch": 0.04787578175544533,
      "grad_norm": 1.2931480407714844,
      "learning_rate": 9.997209325720736e-06,
      "loss": 1.0333,
      "step": 888
    },
    {
      "epoch": 0.04792969592408885,
      "grad_norm": 0.8024531602859497,
      "learning_rate": 9.997202238957886e-06,
      "loss": 0.7166,
      "step": 889
    },
    {
      "epoch": 0.04798361009273237,
      "grad_norm": 0.9585899710655212,
      "learning_rate": 9.997195143210741e-06,
      "loss": 0.8099,
      "step": 890
    },
    {
      "epoch": 0.04803752426137589,
      "grad_norm": 0.9917063117027283,
      "learning_rate": 9.997188038479313e-06,
      "loss": 0.8486,
      "step": 891
    },
    {
      "epoch": 0.04809143843001941,
      "grad_norm": 1.6290080547332764,
      "learning_rate": 9.997180924763616e-06,
      "loss": 0.863,
      "step": 892
    },
    {
      "epoch": 0.04814535259866293,
      "grad_norm": 0.9488585591316223,
      "learning_rate": 9.99717380206366e-06,
      "loss": 0.8277,
      "step": 893
    },
    {
      "epoch": 0.048199266767306445,
      "grad_norm": 1.0710817575454712,
      "learning_rate": 9.997166670379459e-06,
      "loss": 0.8898,
      "step": 894
    },
    {
      "epoch": 0.04825318093594997,
      "grad_norm": 0.9916248917579651,
      "learning_rate": 9.997159529711026e-06,
      "loss": 0.9144,
      "step": 895
    },
    {
      "epoch": 0.04830709510459349,
      "grad_norm": 1.0074565410614014,
      "learning_rate": 9.997152380058378e-06,
      "loss": 0.8391,
      "step": 896
    },
    {
      "epoch": 0.048361009273237006,
      "grad_norm": 1.0258312225341797,
      "learning_rate": 9.99714522142152e-06,
      "loss": 0.973,
      "step": 897
    },
    {
      "epoch": 0.048414923441880524,
      "grad_norm": 0.9497826099395752,
      "learning_rate": 9.99713805380047e-06,
      "loss": 0.9221,
      "step": 898
    },
    {
      "epoch": 0.04846883761052405,
      "grad_norm": 0.9103115200996399,
      "learning_rate": 9.99713087719524e-06,
      "loss": 0.7942,
      "step": 899
    },
    {
      "epoch": 0.04852275177916757,
      "grad_norm": 0.9810470938682556,
      "learning_rate": 9.997123691605843e-06,
      "loss": 0.8673,
      "step": 900
    },
    {
      "epoch": 0.048576665947811085,
      "grad_norm": 1.0422937870025635,
      "learning_rate": 9.997116497032291e-06,
      "loss": 0.9263,
      "step": 901
    },
    {
      "epoch": 0.0486305801164546,
      "grad_norm": 0.8522017002105713,
      "learning_rate": 9.997109293474596e-06,
      "loss": 0.8296,
      "step": 902
    },
    {
      "epoch": 0.04868449428509812,
      "grad_norm": 0.818270742893219,
      "learning_rate": 9.997102080932775e-06,
      "loss": 0.7898,
      "step": 903
    },
    {
      "epoch": 0.048738408453741645,
      "grad_norm": 0.9286766648292542,
      "learning_rate": 9.997094859406838e-06,
      "loss": 0.8751,
      "step": 904
    },
    {
      "epoch": 0.04879232262238516,
      "grad_norm": 1.0779087543487549,
      "learning_rate": 9.997087628896797e-06,
      "loss": 0.8377,
      "step": 905
    },
    {
      "epoch": 0.04884623679102868,
      "grad_norm": 0.8711867928504944,
      "learning_rate": 9.997080389402667e-06,
      "loss": 0.8547,
      "step": 906
    },
    {
      "epoch": 0.0489001509596722,
      "grad_norm": 0.8919721245765686,
      "learning_rate": 9.99707314092446e-06,
      "loss": 0.8178,
      "step": 907
    },
    {
      "epoch": 0.048954065128315724,
      "grad_norm": 0.9084917306900024,
      "learning_rate": 9.997065883462192e-06,
      "loss": 0.8618,
      "step": 908
    },
    {
      "epoch": 0.04900797929695924,
      "grad_norm": 0.869216799736023,
      "learning_rate": 9.997058617015871e-06,
      "loss": 0.8636,
      "step": 909
    },
    {
      "epoch": 0.04906189346560276,
      "grad_norm": 0.9376553893089294,
      "learning_rate": 9.997051341585513e-06,
      "loss": 0.8986,
      "step": 910
    },
    {
      "epoch": 0.04911580763424628,
      "grad_norm": 0.9041107892990112,
      "learning_rate": 9.99704405717113e-06,
      "loss": 0.817,
      "step": 911
    },
    {
      "epoch": 0.0491697218028898,
      "grad_norm": 0.9530431628227234,
      "learning_rate": 9.997036763772737e-06,
      "loss": 0.9464,
      "step": 912
    },
    {
      "epoch": 0.04922363597153332,
      "grad_norm": 0.9601117968559265,
      "learning_rate": 9.997029461390344e-06,
      "loss": 0.9014,
      "step": 913
    },
    {
      "epoch": 0.04927755014017684,
      "grad_norm": 0.9162781834602356,
      "learning_rate": 9.997022150023968e-06,
      "loss": 0.8851,
      "step": 914
    },
    {
      "epoch": 0.04933146430882036,
      "grad_norm": 0.9514605402946472,
      "learning_rate": 9.99701482967362e-06,
      "loss": 0.8975,
      "step": 915
    },
    {
      "epoch": 0.049385378477463875,
      "grad_norm": 0.897203803062439,
      "learning_rate": 9.997007500339313e-06,
      "loss": 0.8371,
      "step": 916
    },
    {
      "epoch": 0.0494392926461074,
      "grad_norm": 0.9372673630714417,
      "learning_rate": 9.99700016202106e-06,
      "loss": 0.9432,
      "step": 917
    },
    {
      "epoch": 0.04949320681475092,
      "grad_norm": 0.8993443846702576,
      "learning_rate": 9.996992814718875e-06,
      "loss": 0.8528,
      "step": 918
    },
    {
      "epoch": 0.049547120983394435,
      "grad_norm": 0.9300720691680908,
      "learning_rate": 9.996985458432771e-06,
      "loss": 0.873,
      "step": 919
    },
    {
      "epoch": 0.04960103515203795,
      "grad_norm": 0.9311426281929016,
      "learning_rate": 9.996978093162761e-06,
      "loss": 0.9092,
      "step": 920
    },
    {
      "epoch": 0.04965494932068148,
      "grad_norm": 0.9244507551193237,
      "learning_rate": 9.996970718908859e-06,
      "loss": 0.764,
      "step": 921
    },
    {
      "epoch": 0.049708863489324996,
      "grad_norm": 0.915512204170227,
      "learning_rate": 9.996963335671074e-06,
      "loss": 0.8328,
      "step": 922
    },
    {
      "epoch": 0.049762777657968514,
      "grad_norm": 0.889994740486145,
      "learning_rate": 9.996955943449426e-06,
      "loss": 0.8491,
      "step": 923
    },
    {
      "epoch": 0.04981669182661203,
      "grad_norm": 0.8676478266716003,
      "learning_rate": 9.996948542243925e-06,
      "loss": 0.7677,
      "step": 924
    },
    {
      "epoch": 0.04987060599525555,
      "grad_norm": 0.9795013070106506,
      "learning_rate": 9.996941132054586e-06,
      "loss": 0.9279,
      "step": 925
    },
    {
      "epoch": 0.049924520163899075,
      "grad_norm": 0.940078854560852,
      "learning_rate": 9.996933712881419e-06,
      "loss": 0.8685,
      "step": 926
    },
    {
      "epoch": 0.04997843433254259,
      "grad_norm": 0.9440926313400269,
      "learning_rate": 9.996926284724437e-06,
      "loss": 0.9634,
      "step": 927
    },
    {
      "epoch": 0.05003234850118611,
      "grad_norm": 0.9120537638664246,
      "learning_rate": 9.99691884758366e-06,
      "loss": 0.7656,
      "step": 928
    },
    {
      "epoch": 0.05008626266982963,
      "grad_norm": 1.1514596939086914,
      "learning_rate": 9.996911401459093e-06,
      "loss": 0.864,
      "step": 929
    },
    {
      "epoch": 0.050140176838473154,
      "grad_norm": 0.8924434185028076,
      "learning_rate": 9.996903946350756e-06,
      "loss": 0.877,
      "step": 930
    },
    {
      "epoch": 0.05019409100711667,
      "grad_norm": 0.9884456992149353,
      "learning_rate": 9.996896482258657e-06,
      "loss": 0.94,
      "step": 931
    },
    {
      "epoch": 0.05024800517576019,
      "grad_norm": 0.9282665252685547,
      "learning_rate": 9.996889009182814e-06,
      "loss": 0.8443,
      "step": 932
    },
    {
      "epoch": 0.05030191934440371,
      "grad_norm": 1.1029064655303955,
      "learning_rate": 9.996881527123237e-06,
      "loss": 0.9168,
      "step": 933
    },
    {
      "epoch": 0.050355833513047225,
      "grad_norm": 0.839625358581543,
      "learning_rate": 9.996874036079942e-06,
      "loss": 0.8261,
      "step": 934
    },
    {
      "epoch": 0.05040974768169075,
      "grad_norm": 0.8612869381904602,
      "learning_rate": 9.996866536052942e-06,
      "loss": 0.8197,
      "step": 935
    },
    {
      "epoch": 0.05046366185033427,
      "grad_norm": 0.9483891129493713,
      "learning_rate": 9.996859027042249e-06,
      "loss": 0.8374,
      "step": 936
    },
    {
      "epoch": 0.050517576018977786,
      "grad_norm": 0.9374566674232483,
      "learning_rate": 9.996851509047877e-06,
      "loss": 0.8884,
      "step": 937
    },
    {
      "epoch": 0.050571490187621304,
      "grad_norm": 0.9164647459983826,
      "learning_rate": 9.99684398206984e-06,
      "loss": 0.8419,
      "step": 938
    },
    {
      "epoch": 0.05062540435626483,
      "grad_norm": 1.0109184980392456,
      "learning_rate": 9.996836446108153e-06,
      "loss": 0.8912,
      "step": 939
    },
    {
      "epoch": 0.05067931852490835,
      "grad_norm": 0.8549674153327942,
      "learning_rate": 9.996828901162825e-06,
      "loss": 0.8043,
      "step": 940
    },
    {
      "epoch": 0.050733232693551865,
      "grad_norm": 0.9618684649467468,
      "learning_rate": 9.996821347233875e-06,
      "loss": 0.8246,
      "step": 941
    },
    {
      "epoch": 0.05078714686219538,
      "grad_norm": 0.9777100682258606,
      "learning_rate": 9.996813784321314e-06,
      "loss": 0.887,
      "step": 942
    },
    {
      "epoch": 0.05084106103083891,
      "grad_norm": 0.8675182461738586,
      "learning_rate": 9.996806212425157e-06,
      "loss": 0.7584,
      "step": 943
    },
    {
      "epoch": 0.050894975199482426,
      "grad_norm": 0.9174523949623108,
      "learning_rate": 9.996798631545414e-06,
      "loss": 0.8911,
      "step": 944
    },
    {
      "epoch": 0.050948889368125944,
      "grad_norm": 0.9269078373908997,
      "learning_rate": 9.996791041682101e-06,
      "loss": 0.8049,
      "step": 945
    },
    {
      "epoch": 0.05100280353676946,
      "grad_norm": 0.8447721600532532,
      "learning_rate": 9.996783442835233e-06,
      "loss": 0.7781,
      "step": 946
    },
    {
      "epoch": 0.05105671770541298,
      "grad_norm": 0.9178231954574585,
      "learning_rate": 9.99677583500482e-06,
      "loss": 0.8107,
      "step": 947
    },
    {
      "epoch": 0.051110631874056504,
      "grad_norm": 0.8741039633750916,
      "learning_rate": 9.996768218190879e-06,
      "loss": 0.9278,
      "step": 948
    },
    {
      "epoch": 0.05116454604270002,
      "grad_norm": 0.7997228503227234,
      "learning_rate": 9.996760592393425e-06,
      "loss": 0.7706,
      "step": 949
    },
    {
      "epoch": 0.05121846021134354,
      "grad_norm": 1.003300428390503,
      "learning_rate": 9.996752957612468e-06,
      "loss": 0.8464,
      "step": 950
    },
    {
      "epoch": 0.05127237437998706,
      "grad_norm": 0.9237748980522156,
      "learning_rate": 9.996745313848021e-06,
      "loss": 0.9088,
      "step": 951
    },
    {
      "epoch": 0.05132628854863058,
      "grad_norm": 0.8565654754638672,
      "learning_rate": 9.996737661100103e-06,
      "loss": 0.8208,
      "step": 952
    },
    {
      "epoch": 0.0513802027172741,
      "grad_norm": 1.0590770244598389,
      "learning_rate": 9.996729999368722e-06,
      "loss": 0.9272,
      "step": 953
    },
    {
      "epoch": 0.05143411688591762,
      "grad_norm": 0.8888198733329773,
      "learning_rate": 9.996722328653897e-06,
      "loss": 0.8264,
      "step": 954
    },
    {
      "epoch": 0.05148803105456114,
      "grad_norm": 0.9211130142211914,
      "learning_rate": 9.996714648955636e-06,
      "loss": 0.8807,
      "step": 955
    },
    {
      "epoch": 0.051541945223204655,
      "grad_norm": 1.0241321325302124,
      "learning_rate": 9.996706960273958e-06,
      "loss": 0.7638,
      "step": 956
    },
    {
      "epoch": 0.05159585939184818,
      "grad_norm": 0.903762698173523,
      "learning_rate": 9.996699262608875e-06,
      "loss": 0.8583,
      "step": 957
    },
    {
      "epoch": 0.0516497735604917,
      "grad_norm": 0.9271189570426941,
      "learning_rate": 9.9966915559604e-06,
      "loss": 0.8341,
      "step": 958
    },
    {
      "epoch": 0.051703687729135216,
      "grad_norm": 0.865260899066925,
      "learning_rate": 9.996683840328546e-06,
      "loss": 0.9136,
      "step": 959
    },
    {
      "epoch": 0.051757601897778734,
      "grad_norm": 0.8903625011444092,
      "learning_rate": 9.996676115713332e-06,
      "loss": 0.8706,
      "step": 960
    },
    {
      "epoch": 0.05181151606642226,
      "grad_norm": 0.9228227138519287,
      "learning_rate": 9.996668382114765e-06,
      "loss": 0.8825,
      "step": 961
    },
    {
      "epoch": 0.051865430235065776,
      "grad_norm": 0.9146421551704407,
      "learning_rate": 9.996660639532863e-06,
      "loss": 0.8347,
      "step": 962
    },
    {
      "epoch": 0.051919344403709294,
      "grad_norm": 0.9010991454124451,
      "learning_rate": 9.99665288796764e-06,
      "loss": 0.8016,
      "step": 963
    },
    {
      "epoch": 0.05197325857235281,
      "grad_norm": 0.8763105869293213,
      "learning_rate": 9.996645127419107e-06,
      "loss": 0.8651,
      "step": 964
    },
    {
      "epoch": 0.05202717274099634,
      "grad_norm": 0.9506256580352783,
      "learning_rate": 9.996637357887281e-06,
      "loss": 0.9429,
      "step": 965
    },
    {
      "epoch": 0.052081086909639855,
      "grad_norm": 0.9484269022941589,
      "learning_rate": 9.996629579372175e-06,
      "loss": 0.855,
      "step": 966
    },
    {
      "epoch": 0.05213500107828337,
      "grad_norm": 0.8970646262168884,
      "learning_rate": 9.996621791873804e-06,
      "loss": 0.8611,
      "step": 967
    },
    {
      "epoch": 0.05218891524692689,
      "grad_norm": 0.8925203680992126,
      "learning_rate": 9.99661399539218e-06,
      "loss": 0.8206,
      "step": 968
    },
    {
      "epoch": 0.05224282941557041,
      "grad_norm": 1.069669246673584,
      "learning_rate": 9.996606189927318e-06,
      "loss": 0.876,
      "step": 969
    },
    {
      "epoch": 0.052296743584213934,
      "grad_norm": 0.8456307649612427,
      "learning_rate": 9.996598375479232e-06,
      "loss": 0.7514,
      "step": 970
    },
    {
      "epoch": 0.05235065775285745,
      "grad_norm": 0.9182801246643066,
      "learning_rate": 9.996590552047936e-06,
      "loss": 0.8915,
      "step": 971
    },
    {
      "epoch": 0.05240457192150097,
      "grad_norm": 0.7616676688194275,
      "learning_rate": 9.996582719633445e-06,
      "loss": 0.7106,
      "step": 972
    },
    {
      "epoch": 0.05245848609014449,
      "grad_norm": 0.8873127102851868,
      "learning_rate": 9.99657487823577e-06,
      "loss": 0.9171,
      "step": 973
    },
    {
      "epoch": 0.05251240025878801,
      "grad_norm": 0.9724618792533875,
      "learning_rate": 9.996567027854929e-06,
      "loss": 0.9765,
      "step": 974
    },
    {
      "epoch": 0.05256631442743153,
      "grad_norm": 0.9106513857841492,
      "learning_rate": 9.996559168490933e-06,
      "loss": 0.8332,
      "step": 975
    },
    {
      "epoch": 0.05262022859607505,
      "grad_norm": 0.8551159501075745,
      "learning_rate": 9.996551300143798e-06,
      "loss": 0.8128,
      "step": 976
    },
    {
      "epoch": 0.052674142764718566,
      "grad_norm": 0.9829822182655334,
      "learning_rate": 9.996543422813539e-06,
      "loss": 0.9088,
      "step": 977
    },
    {
      "epoch": 0.052728056933362084,
      "grad_norm": 0.8281888961791992,
      "learning_rate": 9.996535536500166e-06,
      "loss": 0.8338,
      "step": 978
    },
    {
      "epoch": 0.05278197110200561,
      "grad_norm": 0.951319694519043,
      "learning_rate": 9.9965276412037e-06,
      "loss": 0.9359,
      "step": 979
    },
    {
      "epoch": 0.05283588527064913,
      "grad_norm": 0.841390073299408,
      "learning_rate": 9.996519736924148e-06,
      "loss": 0.7952,
      "step": 980
    },
    {
      "epoch": 0.052889799439292645,
      "grad_norm": 0.8847686648368835,
      "learning_rate": 9.996511823661528e-06,
      "loss": 0.8435,
      "step": 981
    },
    {
      "epoch": 0.05294371360793616,
      "grad_norm": 0.9261316061019897,
      "learning_rate": 9.996503901415855e-06,
      "loss": 0.8646,
      "step": 982
    },
    {
      "epoch": 0.05299762777657969,
      "grad_norm": 0.9366586804389954,
      "learning_rate": 9.99649597018714e-06,
      "loss": 0.8586,
      "step": 983
    },
    {
      "epoch": 0.053051541945223206,
      "grad_norm": 0.8916764259338379,
      "learning_rate": 9.9964880299754e-06,
      "loss": 0.8215,
      "step": 984
    },
    {
      "epoch": 0.053105456113866724,
      "grad_norm": 0.9496534466743469,
      "learning_rate": 9.996480080780648e-06,
      "loss": 0.7984,
      "step": 985
    },
    {
      "epoch": 0.05315937028251024,
      "grad_norm": 0.9736526608467102,
      "learning_rate": 9.9964721226029e-06,
      "loss": 0.7881,
      "step": 986
    },
    {
      "epoch": 0.05321328445115376,
      "grad_norm": 0.9533856511116028,
      "learning_rate": 9.996464155442167e-06,
      "loss": 0.9855,
      "step": 987
    },
    {
      "epoch": 0.053267198619797285,
      "grad_norm": 0.9656437039375305,
      "learning_rate": 9.996456179298467e-06,
      "loss": 0.9571,
      "step": 988
    },
    {
      "epoch": 0.0533211127884408,
      "grad_norm": 0.8887313008308411,
      "learning_rate": 9.996448194171813e-06,
      "loss": 0.9381,
      "step": 989
    },
    {
      "epoch": 0.05337502695708432,
      "grad_norm": 1.0181535482406616,
      "learning_rate": 9.996440200062217e-06,
      "loss": 0.8834,
      "step": 990
    },
    {
      "epoch": 0.05342894112572784,
      "grad_norm": 0.9083503484725952,
      "learning_rate": 9.996432196969696e-06,
      "loss": 0.9733,
      "step": 991
    },
    {
      "epoch": 0.05348285529437136,
      "grad_norm": 0.9051093459129333,
      "learning_rate": 9.996424184894264e-06,
      "loss": 0.8531,
      "step": 992
    },
    {
      "epoch": 0.05353676946301488,
      "grad_norm": 1.0264357328414917,
      "learning_rate": 9.996416163835935e-06,
      "loss": 0.9212,
      "step": 993
    },
    {
      "epoch": 0.0535906836316584,
      "grad_norm": 1.0350812673568726,
      "learning_rate": 9.996408133794726e-06,
      "loss": 0.7843,
      "step": 994
    },
    {
      "epoch": 0.05364459780030192,
      "grad_norm": 0.9610341787338257,
      "learning_rate": 9.996400094770647e-06,
      "loss": 0.8561,
      "step": 995
    },
    {
      "epoch": 0.05369851196894544,
      "grad_norm": 0.8123961687088013,
      "learning_rate": 9.996392046763714e-06,
      "loss": 0.8296,
      "step": 996
    },
    {
      "epoch": 0.05375242613758896,
      "grad_norm": 0.9337920546531677,
      "learning_rate": 9.996383989773942e-06,
      "loss": 0.8525,
      "step": 997
    },
    {
      "epoch": 0.05380634030623248,
      "grad_norm": 1.1319444179534912,
      "learning_rate": 9.996375923801347e-06,
      "loss": 0.9127,
      "step": 998
    },
    {
      "epoch": 0.053860254474875996,
      "grad_norm": 0.8506798148155212,
      "learning_rate": 9.996367848845941e-06,
      "loss": 0.884,
      "step": 999
    },
    {
      "epoch": 0.053914168643519514,
      "grad_norm": 0.8248615860939026,
      "learning_rate": 9.996359764907739e-06,
      "loss": 0.7579,
      "step": 1000
    },
    {
      "epoch": 0.05396808281216304,
      "grad_norm": 0.9258946180343628,
      "learning_rate": 9.996351671986756e-06,
      "loss": 0.8632,
      "step": 1001
    },
    {
      "epoch": 0.05402199698080656,
      "grad_norm": 0.8891279101371765,
      "learning_rate": 9.996343570083006e-06,
      "loss": 0.8758,
      "step": 1002
    },
    {
      "epoch": 0.054075911149450075,
      "grad_norm": 0.9592086672782898,
      "learning_rate": 9.996335459196505e-06,
      "loss": 0.8962,
      "step": 1003
    },
    {
      "epoch": 0.05412982531809359,
      "grad_norm": 0.8937798738479614,
      "learning_rate": 9.996327339327267e-06,
      "loss": 0.8434,
      "step": 1004
    },
    {
      "epoch": 0.05418373948673712,
      "grad_norm": 0.9602083563804626,
      "learning_rate": 9.996319210475307e-06,
      "loss": 0.9692,
      "step": 1005
    },
    {
      "epoch": 0.054237653655380635,
      "grad_norm": 0.870637834072113,
      "learning_rate": 9.996311072640637e-06,
      "loss": 0.9146,
      "step": 1006
    },
    {
      "epoch": 0.05429156782402415,
      "grad_norm": 0.9330273866653442,
      "learning_rate": 9.996302925823276e-06,
      "loss": 0.8584,
      "step": 1007
    },
    {
      "epoch": 0.05434548199266767,
      "grad_norm": 0.8185963034629822,
      "learning_rate": 9.996294770023234e-06,
      "loss": 0.7854,
      "step": 1008
    },
    {
      "epoch": 0.05439939616131119,
      "grad_norm": 0.8727489113807678,
      "learning_rate": 9.996286605240528e-06,
      "loss": 0.7388,
      "step": 1009
    },
    {
      "epoch": 0.054453310329954714,
      "grad_norm": 1.0858477354049683,
      "learning_rate": 9.996278431475172e-06,
      "loss": 0.9201,
      "step": 1010
    },
    {
      "epoch": 0.05450722449859823,
      "grad_norm": 0.9749255776405334,
      "learning_rate": 9.996270248727184e-06,
      "loss": 0.9041,
      "step": 1011
    },
    {
      "epoch": 0.05456113866724175,
      "grad_norm": 0.9460576176643372,
      "learning_rate": 9.996262056996575e-06,
      "loss": 0.8553,
      "step": 1012
    },
    {
      "epoch": 0.05461505283588527,
      "grad_norm": 0.9379808306694031,
      "learning_rate": 9.99625385628336e-06,
      "loss": 0.9253,
      "step": 1013
    },
    {
      "epoch": 0.05466896700452879,
      "grad_norm": 0.8154170513153076,
      "learning_rate": 9.996245646587553e-06,
      "loss": 0.8703,
      "step": 1014
    },
    {
      "epoch": 0.05472288117317231,
      "grad_norm": 0.9122161269187927,
      "learning_rate": 9.996237427909172e-06,
      "loss": 0.7734,
      "step": 1015
    },
    {
      "epoch": 0.05477679534181583,
      "grad_norm": 0.9049486517906189,
      "learning_rate": 9.996229200248228e-06,
      "loss": 0.8991,
      "step": 1016
    },
    {
      "epoch": 0.05483070951045935,
      "grad_norm": 0.9244295358657837,
      "learning_rate": 9.996220963604741e-06,
      "loss": 0.8514,
      "step": 1017
    },
    {
      "epoch": 0.05488462367910287,
      "grad_norm": 0.9817934036254883,
      "learning_rate": 9.99621271797872e-06,
      "loss": 0.8641,
      "step": 1018
    },
    {
      "epoch": 0.05493853784774639,
      "grad_norm": 0.9253972768783569,
      "learning_rate": 9.996204463370182e-06,
      "loss": 0.9199,
      "step": 1019
    },
    {
      "epoch": 0.05499245201638991,
      "grad_norm": 0.9114319682121277,
      "learning_rate": 9.996196199779145e-06,
      "loss": 0.8063,
      "step": 1020
    },
    {
      "epoch": 0.055046366185033425,
      "grad_norm": 0.9643195867538452,
      "learning_rate": 9.996187927205619e-06,
      "loss": 0.9668,
      "step": 1021
    },
    {
      "epoch": 0.05510028035367694,
      "grad_norm": 0.8127598166465759,
      "learning_rate": 9.996179645649622e-06,
      "loss": 0.764,
      "step": 1022
    },
    {
      "epoch": 0.05515419452232047,
      "grad_norm": 0.8728108406066895,
      "learning_rate": 9.996171355111167e-06,
      "loss": 0.7703,
      "step": 1023
    },
    {
      "epoch": 0.055208108690963986,
      "grad_norm": 0.8554317355155945,
      "learning_rate": 9.996163055590269e-06,
      "loss": 0.8266,
      "step": 1024
    },
    {
      "epoch": 0.055262022859607504,
      "grad_norm": 0.7951076030731201,
      "learning_rate": 9.996154747086946e-06,
      "loss": 0.7601,
      "step": 1025
    },
    {
      "epoch": 0.05531593702825102,
      "grad_norm": 0.8916927576065063,
      "learning_rate": 9.996146429601208e-06,
      "loss": 0.8936,
      "step": 1026
    },
    {
      "epoch": 0.05536985119689455,
      "grad_norm": 1.0242576599121094,
      "learning_rate": 9.996138103133075e-06,
      "loss": 0.8868,
      "step": 1027
    },
    {
      "epoch": 0.055423765365538065,
      "grad_norm": 0.9273019433021545,
      "learning_rate": 9.996129767682557e-06,
      "loss": 0.8622,
      "step": 1028
    },
    {
      "epoch": 0.05547767953418158,
      "grad_norm": 0.9547039866447449,
      "learning_rate": 9.996121423249673e-06,
      "loss": 0.7814,
      "step": 1029
    },
    {
      "epoch": 0.0555315937028251,
      "grad_norm": 0.8750621676445007,
      "learning_rate": 9.996113069834437e-06,
      "loss": 0.7717,
      "step": 1030
    },
    {
      "epoch": 0.05558550787146862,
      "grad_norm": 0.9547988176345825,
      "learning_rate": 9.996104707436862e-06,
      "loss": 0.8877,
      "step": 1031
    },
    {
      "epoch": 0.055639422040112144,
      "grad_norm": 0.8856480717658997,
      "learning_rate": 9.996096336056966e-06,
      "loss": 0.7927,
      "step": 1032
    },
    {
      "epoch": 0.05569333620875566,
      "grad_norm": 0.8311342000961304,
      "learning_rate": 9.99608795569476e-06,
      "loss": 0.7847,
      "step": 1033
    },
    {
      "epoch": 0.05574725037739918,
      "grad_norm": 1.0720731019973755,
      "learning_rate": 9.996079566350266e-06,
      "loss": 0.9243,
      "step": 1034
    },
    {
      "epoch": 0.0558011645460427,
      "grad_norm": 0.9498684406280518,
      "learning_rate": 9.996071168023491e-06,
      "loss": 0.8605,
      "step": 1035
    },
    {
      "epoch": 0.05585507871468622,
      "grad_norm": 0.9043952822685242,
      "learning_rate": 9.996062760714456e-06,
      "loss": 0.8488,
      "step": 1036
    },
    {
      "epoch": 0.05590899288332974,
      "grad_norm": 0.8051116466522217,
      "learning_rate": 9.996054344423173e-06,
      "loss": 0.8275,
      "step": 1037
    },
    {
      "epoch": 0.05596290705197326,
      "grad_norm": 0.857120156288147,
      "learning_rate": 9.996045919149658e-06,
      "loss": 0.8837,
      "step": 1038
    },
    {
      "epoch": 0.056016821220616776,
      "grad_norm": 0.8810911774635315,
      "learning_rate": 9.996037484893926e-06,
      "loss": 0.8179,
      "step": 1039
    },
    {
      "epoch": 0.056070735389260294,
      "grad_norm": 0.8783093690872192,
      "learning_rate": 9.996029041655994e-06,
      "loss": 0.7734,
      "step": 1040
    },
    {
      "epoch": 0.05612464955790382,
      "grad_norm": 0.9281952977180481,
      "learning_rate": 9.996020589435874e-06,
      "loss": 0.8747,
      "step": 1041
    },
    {
      "epoch": 0.05617856372654734,
      "grad_norm": 0.8307299613952637,
      "learning_rate": 9.996012128233583e-06,
      "loss": 0.8055,
      "step": 1042
    },
    {
      "epoch": 0.056232477895190855,
      "grad_norm": 0.9520873427391052,
      "learning_rate": 9.996003658049136e-06,
      "loss": 0.8181,
      "step": 1043
    },
    {
      "epoch": 0.05628639206383437,
      "grad_norm": 0.8753806948661804,
      "learning_rate": 9.995995178882549e-06,
      "loss": 0.808,
      "step": 1044
    },
    {
      "epoch": 0.0563403062324779,
      "grad_norm": 1.067691683769226,
      "learning_rate": 9.995986690733836e-06,
      "loss": 0.8048,
      "step": 1045
    },
    {
      "epoch": 0.056394220401121416,
      "grad_norm": 0.8575261235237122,
      "learning_rate": 9.995978193603013e-06,
      "loss": 0.9231,
      "step": 1046
    },
    {
      "epoch": 0.056448134569764934,
      "grad_norm": 0.9857104420661926,
      "learning_rate": 9.995969687490096e-06,
      "loss": 0.8883,
      "step": 1047
    },
    {
      "epoch": 0.05650204873840845,
      "grad_norm": 0.9203484654426575,
      "learning_rate": 9.995961172395098e-06,
      "loss": 0.7634,
      "step": 1048
    },
    {
      "epoch": 0.056555962907051976,
      "grad_norm": 0.8741904497146606,
      "learning_rate": 9.995952648318036e-06,
      "loss": 0.8061,
      "step": 1049
    },
    {
      "epoch": 0.056609877075695494,
      "grad_norm": 0.9495588541030884,
      "learning_rate": 9.995944115258925e-06,
      "loss": 0.8922,
      "step": 1050
    },
    {
      "epoch": 0.05666379124433901,
      "grad_norm": 0.9306020140647888,
      "learning_rate": 9.99593557321778e-06,
      "loss": 0.8454,
      "step": 1051
    },
    {
      "epoch": 0.05671770541298253,
      "grad_norm": 0.9457784295082092,
      "learning_rate": 9.995927022194615e-06,
      "loss": 0.8701,
      "step": 1052
    },
    {
      "epoch": 0.05677161958162605,
      "grad_norm": 0.88719242811203,
      "learning_rate": 9.99591846218945e-06,
      "loss": 0.8416,
      "step": 1053
    },
    {
      "epoch": 0.05682553375026957,
      "grad_norm": 0.8740848302841187,
      "learning_rate": 9.995909893202296e-06,
      "loss": 0.7962,
      "step": 1054
    },
    {
      "epoch": 0.05687944791891309,
      "grad_norm": 1.0149377584457397,
      "learning_rate": 9.99590131523317e-06,
      "loss": 0.8352,
      "step": 1055
    },
    {
      "epoch": 0.05693336208755661,
      "grad_norm": 0.9014917016029358,
      "learning_rate": 9.995892728282088e-06,
      "loss": 0.9244,
      "step": 1056
    },
    {
      "epoch": 0.05698727625620013,
      "grad_norm": 0.9351898431777954,
      "learning_rate": 9.995884132349062e-06,
      "loss": 0.865,
      "step": 1057
    },
    {
      "epoch": 0.05704119042484365,
      "grad_norm": 0.8656749129295349,
      "learning_rate": 9.995875527434113e-06,
      "loss": 0.8836,
      "step": 1058
    },
    {
      "epoch": 0.05709510459348717,
      "grad_norm": 0.9120789170265198,
      "learning_rate": 9.995866913537254e-06,
      "loss": 0.8772,
      "step": 1059
    },
    {
      "epoch": 0.05714901876213069,
      "grad_norm": 1.0019149780273438,
      "learning_rate": 9.995858290658497e-06,
      "loss": 0.9338,
      "step": 1060
    },
    {
      "epoch": 0.057202932930774206,
      "grad_norm": 0.8492977023124695,
      "learning_rate": 9.995849658797863e-06,
      "loss": 0.742,
      "step": 1061
    },
    {
      "epoch": 0.057256847099417724,
      "grad_norm": 1.000607967376709,
      "learning_rate": 9.995841017955363e-06,
      "loss": 0.8498,
      "step": 1062
    },
    {
      "epoch": 0.05731076126806125,
      "grad_norm": 1.0268487930297852,
      "learning_rate": 9.995832368131016e-06,
      "loss": 0.8937,
      "step": 1063
    },
    {
      "epoch": 0.057364675436704766,
      "grad_norm": 0.9388830661773682,
      "learning_rate": 9.995823709324836e-06,
      "loss": 0.877,
      "step": 1064
    },
    {
      "epoch": 0.057418589605348284,
      "grad_norm": 0.9747199416160583,
      "learning_rate": 9.99581504153684e-06,
      "loss": 0.8436,
      "step": 1065
    },
    {
      "epoch": 0.0574725037739918,
      "grad_norm": 0.9125073552131653,
      "learning_rate": 9.99580636476704e-06,
      "loss": 0.8853,
      "step": 1066
    },
    {
      "epoch": 0.05752641794263533,
      "grad_norm": 0.8910282254219055,
      "learning_rate": 9.995797679015455e-06,
      "loss": 0.8566,
      "step": 1067
    },
    {
      "epoch": 0.057580332111278845,
      "grad_norm": 0.8546010255813599,
      "learning_rate": 9.995788984282101e-06,
      "loss": 0.8209,
      "step": 1068
    },
    {
      "epoch": 0.05763424627992236,
      "grad_norm": 0.9205883145332336,
      "learning_rate": 9.99578028056699e-06,
      "loss": 0.7814,
      "step": 1069
    },
    {
      "epoch": 0.05768816044856588,
      "grad_norm": 0.9627780914306641,
      "learning_rate": 9.995771567870142e-06,
      "loss": 0.8686,
      "step": 1070
    },
    {
      "epoch": 0.057742074617209406,
      "grad_norm": 0.9917465448379517,
      "learning_rate": 9.995762846191569e-06,
      "loss": 0.9672,
      "step": 1071
    },
    {
      "epoch": 0.057795988785852924,
      "grad_norm": 0.9396706223487854,
      "learning_rate": 9.995754115531288e-06,
      "loss": 0.8631,
      "step": 1072
    },
    {
      "epoch": 0.05784990295449644,
      "grad_norm": 0.8310922980308533,
      "learning_rate": 9.995745375889317e-06,
      "loss": 0.8637,
      "step": 1073
    },
    {
      "epoch": 0.05790381712313996,
      "grad_norm": 0.9085954427719116,
      "learning_rate": 9.995736627265667e-06,
      "loss": 0.8821,
      "step": 1074
    },
    {
      "epoch": 0.05795773129178348,
      "grad_norm": 0.8529816269874573,
      "learning_rate": 9.995727869660357e-06,
      "loss": 0.8426,
      "step": 1075
    },
    {
      "epoch": 0.058011645460427,
      "grad_norm": 0.8288499116897583,
      "learning_rate": 9.995719103073403e-06,
      "loss": 0.8415,
      "step": 1076
    },
    {
      "epoch": 0.05806555962907052,
      "grad_norm": 0.9105609059333801,
      "learning_rate": 9.995710327504819e-06,
      "loss": 0.7683,
      "step": 1077
    },
    {
      "epoch": 0.05811947379771404,
      "grad_norm": 0.9578274488449097,
      "learning_rate": 9.995701542954622e-06,
      "loss": 0.8796,
      "step": 1078
    },
    {
      "epoch": 0.058173387966357556,
      "grad_norm": 0.8542460799217224,
      "learning_rate": 9.995692749422827e-06,
      "loss": 0.8363,
      "step": 1079
    },
    {
      "epoch": 0.05822730213500108,
      "grad_norm": 0.8723183274269104,
      "learning_rate": 9.99568394690945e-06,
      "loss": 0.8434,
      "step": 1080
    },
    {
      "epoch": 0.0582812163036446,
      "grad_norm": 0.9157887697219849,
      "learning_rate": 9.995675135414507e-06,
      "loss": 0.6532,
      "step": 1081
    },
    {
      "epoch": 0.05833513047228812,
      "grad_norm": 0.9055691361427307,
      "learning_rate": 9.995666314938014e-06,
      "loss": 0.8762,
      "step": 1082
    },
    {
      "epoch": 0.058389044640931635,
      "grad_norm": 0.8224693536758423,
      "learning_rate": 9.995657485479987e-06,
      "loss": 0.7976,
      "step": 1083
    },
    {
      "epoch": 0.05844295880957515,
      "grad_norm": 0.925414502620697,
      "learning_rate": 9.995648647040441e-06,
      "loss": 0.8673,
      "step": 1084
    },
    {
      "epoch": 0.05849687297821868,
      "grad_norm": 0.9194141626358032,
      "learning_rate": 9.995639799619395e-06,
      "loss": 0.7916,
      "step": 1085
    },
    {
      "epoch": 0.058550787146862196,
      "grad_norm": 1.08795166015625,
      "learning_rate": 9.995630943216859e-06,
      "loss": 0.9135,
      "step": 1086
    },
    {
      "epoch": 0.058604701315505714,
      "grad_norm": 0.9648925065994263,
      "learning_rate": 9.995622077832854e-06,
      "loss": 0.8442,
      "step": 1087
    },
    {
      "epoch": 0.05865861548414923,
      "grad_norm": 1.0012339353561401,
      "learning_rate": 9.995613203467394e-06,
      "loss": 0.9543,
      "step": 1088
    },
    {
      "epoch": 0.05871252965279276,
      "grad_norm": 0.9333881735801697,
      "learning_rate": 9.995604320120496e-06,
      "loss": 0.9267,
      "step": 1089
    },
    {
      "epoch": 0.058766443821436275,
      "grad_norm": 0.8566498160362244,
      "learning_rate": 9.995595427792173e-06,
      "loss": 0.8539,
      "step": 1090
    },
    {
      "epoch": 0.05882035799007979,
      "grad_norm": 0.8766364455223083,
      "learning_rate": 9.995586526482446e-06,
      "loss": 0.9293,
      "step": 1091
    },
    {
      "epoch": 0.05887427215872331,
      "grad_norm": 0.9181047677993774,
      "learning_rate": 9.995577616191326e-06,
      "loss": 0.8333,
      "step": 1092
    },
    {
      "epoch": 0.05892818632736683,
      "grad_norm": 0.8831031918525696,
      "learning_rate": 9.995568696918833e-06,
      "loss": 0.8016,
      "step": 1093
    },
    {
      "epoch": 0.05898210049601035,
      "grad_norm": 0.8618754148483276,
      "learning_rate": 9.99555976866498e-06,
      "loss": 0.8988,
      "step": 1094
    },
    {
      "epoch": 0.05903601466465387,
      "grad_norm": 0.9083183407783508,
      "learning_rate": 9.995550831429785e-06,
      "loss": 0.8626,
      "step": 1095
    },
    {
      "epoch": 0.05908992883329739,
      "grad_norm": 0.8423884510993958,
      "learning_rate": 9.995541885213262e-06,
      "loss": 0.9121,
      "step": 1096
    },
    {
      "epoch": 0.05914384300194091,
      "grad_norm": 0.7747607827186584,
      "learning_rate": 9.99553293001543e-06,
      "loss": 0.8087,
      "step": 1097
    },
    {
      "epoch": 0.05919775717058443,
      "grad_norm": 0.8828368186950684,
      "learning_rate": 9.995523965836302e-06,
      "loss": 0.8284,
      "step": 1098
    },
    {
      "epoch": 0.05925167133922795,
      "grad_norm": 0.9448524713516235,
      "learning_rate": 9.995514992675896e-06,
      "loss": 0.9565,
      "step": 1099
    },
    {
      "epoch": 0.05930558550787147,
      "grad_norm": 0.8967006206512451,
      "learning_rate": 9.99550601053423e-06,
      "loss": 0.8412,
      "step": 1100
    },
    {
      "epoch": 0.059359499676514986,
      "grad_norm": 0.9394551515579224,
      "learning_rate": 9.995497019411315e-06,
      "loss": 0.929,
      "step": 1101
    },
    {
      "epoch": 0.05941341384515851,
      "grad_norm": 0.9002842903137207,
      "learning_rate": 9.995488019307172e-06,
      "loss": 0.734,
      "step": 1102
    },
    {
      "epoch": 0.05946732801380203,
      "grad_norm": 1.3590562343597412,
      "learning_rate": 9.995479010221816e-06,
      "loss": 0.8843,
      "step": 1103
    },
    {
      "epoch": 0.05952124218244555,
      "grad_norm": 1.041528582572937,
      "learning_rate": 9.99546999215526e-06,
      "loss": 0.9001,
      "step": 1104
    },
    {
      "epoch": 0.059575156351089065,
      "grad_norm": 0.9846720099449158,
      "learning_rate": 9.995460965107524e-06,
      "loss": 0.8174,
      "step": 1105
    },
    {
      "epoch": 0.05962907051973258,
      "grad_norm": 0.9171685576438904,
      "learning_rate": 9.995451929078624e-06,
      "loss": 0.8756,
      "step": 1106
    },
    {
      "epoch": 0.05968298468837611,
      "grad_norm": 0.9155516028404236,
      "learning_rate": 9.995442884068574e-06,
      "loss": 0.7327,
      "step": 1107
    },
    {
      "epoch": 0.059736898857019625,
      "grad_norm": 0.8734007477760315,
      "learning_rate": 9.99543383007739e-06,
      "loss": 0.8385,
      "step": 1108
    },
    {
      "epoch": 0.05979081302566314,
      "grad_norm": 0.8580977320671082,
      "learning_rate": 9.99542476710509e-06,
      "loss": 0.885,
      "step": 1109
    },
    {
      "epoch": 0.05984472719430666,
      "grad_norm": 0.8499299883842468,
      "learning_rate": 9.995415695151692e-06,
      "loss": 0.8323,
      "step": 1110
    },
    {
      "epoch": 0.059898641362950186,
      "grad_norm": 0.8348694443702698,
      "learning_rate": 9.99540661421721e-06,
      "loss": 0.7947,
      "step": 1111
    },
    {
      "epoch": 0.059952555531593704,
      "grad_norm": 0.8865199685096741,
      "learning_rate": 9.99539752430166e-06,
      "loss": 0.9363,
      "step": 1112
    },
    {
      "epoch": 0.06000646970023722,
      "grad_norm": 0.9492315649986267,
      "learning_rate": 9.995388425405059e-06,
      "loss": 0.913,
      "step": 1113
    },
    {
      "epoch": 0.06006038386888074,
      "grad_norm": 0.938252329826355,
      "learning_rate": 9.995379317527422e-06,
      "loss": 0.861,
      "step": 1114
    },
    {
      "epoch": 0.06011429803752426,
      "grad_norm": 1.2601032257080078,
      "learning_rate": 9.995370200668768e-06,
      "loss": 0.9435,
      "step": 1115
    },
    {
      "epoch": 0.06016821220616778,
      "grad_norm": 0.915830671787262,
      "learning_rate": 9.995361074829112e-06,
      "loss": 0.9372,
      "step": 1116
    },
    {
      "epoch": 0.0602221263748113,
      "grad_norm": 1.4548465013504028,
      "learning_rate": 9.995351940008473e-06,
      "loss": 0.9055,
      "step": 1117
    },
    {
      "epoch": 0.06027604054345482,
      "grad_norm": 0.9090906381607056,
      "learning_rate": 9.995342796206861e-06,
      "loss": 0.8849,
      "step": 1118
    },
    {
      "epoch": 0.06032995471209834,
      "grad_norm": 0.9860616326332092,
      "learning_rate": 9.995333643424298e-06,
      "loss": 0.8304,
      "step": 1119
    },
    {
      "epoch": 0.06038386888074186,
      "grad_norm": 0.8320879340171814,
      "learning_rate": 9.9953244816608e-06,
      "loss": 0.8432,
      "step": 1120
    },
    {
      "epoch": 0.06043778304938538,
      "grad_norm": 0.8633564114570618,
      "learning_rate": 9.995315310916381e-06,
      "loss": 0.7461,
      "step": 1121
    },
    {
      "epoch": 0.0604916972180289,
      "grad_norm": 0.881287693977356,
      "learning_rate": 9.995306131191059e-06,
      "loss": 0.8512,
      "step": 1122
    },
    {
      "epoch": 0.060545611386672415,
      "grad_norm": 0.8888201713562012,
      "learning_rate": 9.99529694248485e-06,
      "loss": 0.8416,
      "step": 1123
    },
    {
      "epoch": 0.06059952555531594,
      "grad_norm": 0.8073605895042419,
      "learning_rate": 9.99528774479777e-06,
      "loss": 0.8369,
      "step": 1124
    },
    {
      "epoch": 0.06065343972395946,
      "grad_norm": 0.9260549545288086,
      "learning_rate": 9.995278538129837e-06,
      "loss": 0.8548,
      "step": 1125
    },
    {
      "epoch": 0.060707353892602976,
      "grad_norm": 0.9169156551361084,
      "learning_rate": 9.99526932248107e-06,
      "loss": 0.9149,
      "step": 1126
    },
    {
      "epoch": 0.060761268061246494,
      "grad_norm": 0.8481706380844116,
      "learning_rate": 9.995260097851478e-06,
      "loss": 0.8591,
      "step": 1127
    },
    {
      "epoch": 0.06081518222989001,
      "grad_norm": 0.8934486508369446,
      "learning_rate": 9.995250864241085e-06,
      "loss": 0.9322,
      "step": 1128
    },
    {
      "epoch": 0.06086909639853354,
      "grad_norm": 0.947390615940094,
      "learning_rate": 9.995241621649902e-06,
      "loss": 1.0015,
      "step": 1129
    },
    {
      "epoch": 0.060923010567177055,
      "grad_norm": 0.9185096025466919,
      "learning_rate": 9.995232370077949e-06,
      "loss": 0.9293,
      "step": 1130
    },
    {
      "epoch": 0.06097692473582057,
      "grad_norm": 0.9517882466316223,
      "learning_rate": 9.995223109525245e-06,
      "loss": 0.8673,
      "step": 1131
    },
    {
      "epoch": 0.06103083890446409,
      "grad_norm": 1.065699815750122,
      "learning_rate": 9.9952138399918e-06,
      "loss": 0.9144,
      "step": 1132
    },
    {
      "epoch": 0.061084753073107616,
      "grad_norm": 0.9048404693603516,
      "learning_rate": 9.995204561477635e-06,
      "loss": 0.7773,
      "step": 1133
    },
    {
      "epoch": 0.061138667241751134,
      "grad_norm": 1.104457139968872,
      "learning_rate": 9.995195273982768e-06,
      "loss": 0.8847,
      "step": 1134
    },
    {
      "epoch": 0.06119258141039465,
      "grad_norm": 0.9009587168693542,
      "learning_rate": 9.995185977507212e-06,
      "loss": 0.8118,
      "step": 1135
    },
    {
      "epoch": 0.06124649557903817,
      "grad_norm": 1.0740209817886353,
      "learning_rate": 9.995176672050983e-06,
      "loss": 0.9173,
      "step": 1136
    },
    {
      "epoch": 0.06130040974768169,
      "grad_norm": 0.9820743203163147,
      "learning_rate": 9.995167357614104e-06,
      "loss": 0.8555,
      "step": 1137
    },
    {
      "epoch": 0.06135432391632521,
      "grad_norm": 0.9250825047492981,
      "learning_rate": 9.995158034196586e-06,
      "loss": 0.8771,
      "step": 1138
    },
    {
      "epoch": 0.06140823808496873,
      "grad_norm": 0.8952597379684448,
      "learning_rate": 9.995148701798447e-06,
      "loss": 0.8598,
      "step": 1139
    },
    {
      "epoch": 0.06146215225361225,
      "grad_norm": 0.8485212922096252,
      "learning_rate": 9.995139360419706e-06,
      "loss": 0.8557,
      "step": 1140
    },
    {
      "epoch": 0.061516066422255766,
      "grad_norm": 0.9676715731620789,
      "learning_rate": 9.995130010060377e-06,
      "loss": 0.7748,
      "step": 1141
    },
    {
      "epoch": 0.06156998059089929,
      "grad_norm": 0.7896347045898438,
      "learning_rate": 9.995120650720478e-06,
      "loss": 0.6183,
      "step": 1142
    },
    {
      "epoch": 0.06162389475954281,
      "grad_norm": 0.8746615052223206,
      "learning_rate": 9.995111282400024e-06,
      "loss": 0.8321,
      "step": 1143
    },
    {
      "epoch": 0.06167780892818633,
      "grad_norm": 0.9029875993728638,
      "learning_rate": 9.995101905099036e-06,
      "loss": 0.8686,
      "step": 1144
    },
    {
      "epoch": 0.061731723096829845,
      "grad_norm": 0.9529547095298767,
      "learning_rate": 9.995092518817528e-06,
      "loss": 0.8878,
      "step": 1145
    },
    {
      "epoch": 0.06178563726547336,
      "grad_norm": 0.8280455470085144,
      "learning_rate": 9.995083123555517e-06,
      "loss": 0.8232,
      "step": 1146
    },
    {
      "epoch": 0.06183955143411689,
      "grad_norm": 0.908881664276123,
      "learning_rate": 9.995073719313021e-06,
      "loss": 0.8387,
      "step": 1147
    },
    {
      "epoch": 0.061893465602760406,
      "grad_norm": 0.9137653708457947,
      "learning_rate": 9.995064306090055e-06,
      "loss": 0.8943,
      "step": 1148
    },
    {
      "epoch": 0.061947379771403924,
      "grad_norm": 0.863861620426178,
      "learning_rate": 9.995054883886639e-06,
      "loss": 0.7435,
      "step": 1149
    },
    {
      "epoch": 0.06200129394004744,
      "grad_norm": 0.8534915447235107,
      "learning_rate": 9.995045452702786e-06,
      "loss": 0.941,
      "step": 1150
    },
    {
      "epoch": 0.06205520810869097,
      "grad_norm": 0.9469791650772095,
      "learning_rate": 9.995036012538515e-06,
      "loss": 0.9137,
      "step": 1151
    },
    {
      "epoch": 0.062109122277334484,
      "grad_norm": 0.9044890999794006,
      "learning_rate": 9.995026563393844e-06,
      "loss": 0.9117,
      "step": 1152
    },
    {
      "epoch": 0.062163036445978,
      "grad_norm": 0.989772379398346,
      "learning_rate": 9.995017105268789e-06,
      "loss": 0.8306,
      "step": 1153
    },
    {
      "epoch": 0.06221695061462152,
      "grad_norm": 0.8586496114730835,
      "learning_rate": 9.995007638163365e-06,
      "loss": 0.8012,
      "step": 1154
    },
    {
      "epoch": 0.062270864783265045,
      "grad_norm": 0.9221116304397583,
      "learning_rate": 9.994998162077594e-06,
      "loss": 0.7935,
      "step": 1155
    },
    {
      "epoch": 0.06232477895190856,
      "grad_norm": 0.9453061819076538,
      "learning_rate": 9.994988677011489e-06,
      "loss": 0.8257,
      "step": 1156
    },
    {
      "epoch": 0.06237869312055208,
      "grad_norm": 0.8065335154533386,
      "learning_rate": 9.994979182965065e-06,
      "loss": 0.86,
      "step": 1157
    },
    {
      "epoch": 0.0624326072891956,
      "grad_norm": 0.9597793817520142,
      "learning_rate": 9.994969679938346e-06,
      "loss": 0.862,
      "step": 1158
    },
    {
      "epoch": 0.06248652145783912,
      "grad_norm": 0.9118353128433228,
      "learning_rate": 9.994960167931342e-06,
      "loss": 0.8925,
      "step": 1159
    },
    {
      "epoch": 0.06254043562648263,
      "grad_norm": 1.0216273069381714,
      "learning_rate": 9.994950646944077e-06,
      "loss": 0.7078,
      "step": 1160
    },
    {
      "epoch": 0.06259434979512615,
      "grad_norm": 0.960182785987854,
      "learning_rate": 9.994941116976562e-06,
      "loss": 0.8936,
      "step": 1161
    },
    {
      "epoch": 0.06264826396376968,
      "grad_norm": 0.9551856517791748,
      "learning_rate": 9.994931578028817e-06,
      "loss": 0.8053,
      "step": 1162
    },
    {
      "epoch": 0.0627021781324132,
      "grad_norm": 0.9419867992401123,
      "learning_rate": 9.994922030100857e-06,
      "loss": 0.8333,
      "step": 1163
    },
    {
      "epoch": 0.06275609230105672,
      "grad_norm": 0.9780306816101074,
      "learning_rate": 9.994912473192702e-06,
      "loss": 0.88,
      "step": 1164
    },
    {
      "epoch": 0.06281000646970024,
      "grad_norm": 0.9320577383041382,
      "learning_rate": 9.99490290730437e-06,
      "loss": 0.8859,
      "step": 1165
    },
    {
      "epoch": 0.06286392063834376,
      "grad_norm": 0.7692422270774841,
      "learning_rate": 9.994893332435874e-06,
      "loss": 0.8093,
      "step": 1166
    },
    {
      "epoch": 0.06291783480698727,
      "grad_norm": 1.0622048377990723,
      "learning_rate": 9.994883748587234e-06,
      "loss": 0.8959,
      "step": 1167
    },
    {
      "epoch": 0.06297174897563079,
      "grad_norm": 0.9598555564880371,
      "learning_rate": 9.994874155758467e-06,
      "loss": 0.8153,
      "step": 1168
    },
    {
      "epoch": 0.06302566314427431,
      "grad_norm": 0.9207014441490173,
      "learning_rate": 9.994864553949591e-06,
      "loss": 0.9383,
      "step": 1169
    },
    {
      "epoch": 0.06307957731291783,
      "grad_norm": 1.0074093341827393,
      "learning_rate": 9.99485494316062e-06,
      "loss": 0.9999,
      "step": 1170
    },
    {
      "epoch": 0.06313349148156136,
      "grad_norm": 0.8454248905181885,
      "learning_rate": 9.994845323391575e-06,
      "loss": 0.7946,
      "step": 1171
    },
    {
      "epoch": 0.06318740565020488,
      "grad_norm": 0.847578763961792,
      "learning_rate": 9.99483569464247e-06,
      "loss": 0.7144,
      "step": 1172
    },
    {
      "epoch": 0.0632413198188484,
      "grad_norm": 0.9083126187324524,
      "learning_rate": 9.994826056913325e-06,
      "loss": 0.774,
      "step": 1173
    },
    {
      "epoch": 0.06329523398749191,
      "grad_norm": 0.8995345830917358,
      "learning_rate": 9.994816410204158e-06,
      "loss": 0.8995,
      "step": 1174
    },
    {
      "epoch": 0.06334914815613543,
      "grad_norm": 1.0547746419906616,
      "learning_rate": 9.994806754514983e-06,
      "loss": 0.8142,
      "step": 1175
    },
    {
      "epoch": 0.06340306232477895,
      "grad_norm": 0.946854829788208,
      "learning_rate": 9.99479708984582e-06,
      "loss": 0.8639,
      "step": 1176
    },
    {
      "epoch": 0.06345697649342247,
      "grad_norm": 0.8746247291564941,
      "learning_rate": 9.994787416196683e-06,
      "loss": 0.8601,
      "step": 1177
    },
    {
      "epoch": 0.06351089066206599,
      "grad_norm": 0.9075024127960205,
      "learning_rate": 9.994777733567595e-06,
      "loss": 0.7969,
      "step": 1178
    },
    {
      "epoch": 0.0635648048307095,
      "grad_norm": 0.9435486197471619,
      "learning_rate": 9.994768041958569e-06,
      "loss": 0.8199,
      "step": 1179
    },
    {
      "epoch": 0.06361871899935304,
      "grad_norm": 0.8597564697265625,
      "learning_rate": 9.994758341369624e-06,
      "loss": 0.8791,
      "step": 1180
    },
    {
      "epoch": 0.06367263316799655,
      "grad_norm": 0.7960480451583862,
      "learning_rate": 9.994748631800777e-06,
      "loss": 0.8035,
      "step": 1181
    },
    {
      "epoch": 0.06372654733664007,
      "grad_norm": 1.1984984874725342,
      "learning_rate": 9.994738913252045e-06,
      "loss": 0.7372,
      "step": 1182
    },
    {
      "epoch": 0.06378046150528359,
      "grad_norm": 0.8532997369766235,
      "learning_rate": 9.994729185723446e-06,
      "loss": 0.9094,
      "step": 1183
    },
    {
      "epoch": 0.06383437567392711,
      "grad_norm": 0.8327267169952393,
      "learning_rate": 9.994719449214999e-06,
      "loss": 0.809,
      "step": 1184
    },
    {
      "epoch": 0.06388828984257063,
      "grad_norm": 0.9086306691169739,
      "learning_rate": 9.99470970372672e-06,
      "loss": 0.8278,
      "step": 1185
    },
    {
      "epoch": 0.06394220401121414,
      "grad_norm": 0.8422104716300964,
      "learning_rate": 9.994699949258626e-06,
      "loss": 0.7754,
      "step": 1186
    },
    {
      "epoch": 0.06399611817985766,
      "grad_norm": 1.0434929132461548,
      "learning_rate": 9.994690185810733e-06,
      "loss": 0.908,
      "step": 1187
    },
    {
      "epoch": 0.06405003234850119,
      "grad_norm": 1.1625720262527466,
      "learning_rate": 9.994680413383064e-06,
      "loss": 0.8814,
      "step": 1188
    },
    {
      "epoch": 0.06410394651714471,
      "grad_norm": 0.9940767288208008,
      "learning_rate": 9.994670631975631e-06,
      "loss": 0.7846,
      "step": 1189
    },
    {
      "epoch": 0.06415786068578823,
      "grad_norm": 0.8356907963752747,
      "learning_rate": 9.994660841588457e-06,
      "loss": 0.798,
      "step": 1190
    },
    {
      "epoch": 0.06421177485443175,
      "grad_norm": 0.830348014831543,
      "learning_rate": 9.994651042221552e-06,
      "loss": 0.7875,
      "step": 1191
    },
    {
      "epoch": 0.06426568902307526,
      "grad_norm": 1.1060880422592163,
      "learning_rate": 9.994641233874943e-06,
      "loss": 0.8893,
      "step": 1192
    },
    {
      "epoch": 0.06431960319171878,
      "grad_norm": 0.9319590926170349,
      "learning_rate": 9.994631416548637e-06,
      "loss": 0.791,
      "step": 1193
    },
    {
      "epoch": 0.0643735173603623,
      "grad_norm": 0.8345780968666077,
      "learning_rate": 9.994621590242661e-06,
      "loss": 0.8213,
      "step": 1194
    },
    {
      "epoch": 0.06442743152900582,
      "grad_norm": 0.9848359227180481,
      "learning_rate": 9.99461175495703e-06,
      "loss": 0.735,
      "step": 1195
    },
    {
      "epoch": 0.06448134569764934,
      "grad_norm": 0.9134055972099304,
      "learning_rate": 9.994601910691758e-06,
      "loss": 0.8415,
      "step": 1196
    },
    {
      "epoch": 0.06453525986629287,
      "grad_norm": 0.8084586262702942,
      "learning_rate": 9.994592057446866e-06,
      "loss": 0.8702,
      "step": 1197
    },
    {
      "epoch": 0.06458917403493639,
      "grad_norm": 0.9168767333030701,
      "learning_rate": 9.994582195222371e-06,
      "loss": 0.8921,
      "step": 1198
    },
    {
      "epoch": 0.0646430882035799,
      "grad_norm": 0.8380446434020996,
      "learning_rate": 9.994572324018292e-06,
      "loss": 0.7705,
      "step": 1199
    },
    {
      "epoch": 0.06469700237222342,
      "grad_norm": 0.8120049238204956,
      "learning_rate": 9.994562443834646e-06,
      "loss": 0.7576,
      "step": 1200
    },
    {
      "epoch": 0.06475091654086694,
      "grad_norm": 0.9559764266014099,
      "learning_rate": 9.994552554671448e-06,
      "loss": 0.8427,
      "step": 1201
    },
    {
      "epoch": 0.06480483070951046,
      "grad_norm": 0.9473673105239868,
      "learning_rate": 9.99454265652872e-06,
      "loss": 0.9988,
      "step": 1202
    },
    {
      "epoch": 0.06485874487815398,
      "grad_norm": 1.0704870223999023,
      "learning_rate": 9.994532749406477e-06,
      "loss": 0.9499,
      "step": 1203
    },
    {
      "epoch": 0.0649126590467975,
      "grad_norm": 0.9905646443367004,
      "learning_rate": 9.994522833304738e-06,
      "loss": 0.8801,
      "step": 1204
    },
    {
      "epoch": 0.06496657321544101,
      "grad_norm": 1.194190502166748,
      "learning_rate": 9.99451290822352e-06,
      "loss": 0.9051,
      "step": 1205
    },
    {
      "epoch": 0.06502048738408454,
      "grad_norm": 0.8571314811706543,
      "learning_rate": 9.994502974162843e-06,
      "loss": 0.8131,
      "step": 1206
    },
    {
      "epoch": 0.06507440155272806,
      "grad_norm": 0.9769417643547058,
      "learning_rate": 9.994493031122721e-06,
      "loss": 0.8524,
      "step": 1207
    },
    {
      "epoch": 0.06512831572137158,
      "grad_norm": 0.8106759786605835,
      "learning_rate": 9.994483079103176e-06,
      "loss": 0.8142,
      "step": 1208
    },
    {
      "epoch": 0.0651822298900151,
      "grad_norm": 0.8817846775054932,
      "learning_rate": 9.994473118104223e-06,
      "loss": 0.9076,
      "step": 1209
    },
    {
      "epoch": 0.06523614405865862,
      "grad_norm": 0.8271930813789368,
      "learning_rate": 9.994463148125882e-06,
      "loss": 0.7914,
      "step": 1210
    },
    {
      "epoch": 0.06529005822730213,
      "grad_norm": 0.9060614705085754,
      "learning_rate": 9.994453169168169e-06,
      "loss": 0.8375,
      "step": 1211
    },
    {
      "epoch": 0.06534397239594565,
      "grad_norm": 0.880614697933197,
      "learning_rate": 9.994443181231103e-06,
      "loss": 0.7751,
      "step": 1212
    },
    {
      "epoch": 0.06539788656458917,
      "grad_norm": 0.9420819282531738,
      "learning_rate": 9.994433184314702e-06,
      "loss": 0.8532,
      "step": 1213
    },
    {
      "epoch": 0.06545180073323269,
      "grad_norm": 0.8587054014205933,
      "learning_rate": 9.994423178418984e-06,
      "loss": 0.8804,
      "step": 1214
    },
    {
      "epoch": 0.06550571490187622,
      "grad_norm": 0.9624550938606262,
      "learning_rate": 9.994413163543965e-06,
      "loss": 0.9782,
      "step": 1215
    },
    {
      "epoch": 0.06555962907051974,
      "grad_norm": 0.9458224773406982,
      "learning_rate": 9.994403139689665e-06,
      "loss": 0.8274,
      "step": 1216
    },
    {
      "epoch": 0.06561354323916326,
      "grad_norm": 1.0417940616607666,
      "learning_rate": 9.994393106856104e-06,
      "loss": 0.9065,
      "step": 1217
    },
    {
      "epoch": 0.06566745740780677,
      "grad_norm": 1.0225417613983154,
      "learning_rate": 9.994383065043296e-06,
      "loss": 0.8642,
      "step": 1218
    },
    {
      "epoch": 0.06572137157645029,
      "grad_norm": 0.9015594720840454,
      "learning_rate": 9.994373014251261e-06,
      "loss": 0.8775,
      "step": 1219
    },
    {
      "epoch": 0.06577528574509381,
      "grad_norm": 0.8473883271217346,
      "learning_rate": 9.994362954480018e-06,
      "loss": 0.8566,
      "step": 1220
    },
    {
      "epoch": 0.06582919991373733,
      "grad_norm": 0.8571242690086365,
      "learning_rate": 9.994352885729584e-06,
      "loss": 0.8502,
      "step": 1221
    },
    {
      "epoch": 0.06588311408238084,
      "grad_norm": 0.8793268799781799,
      "learning_rate": 9.994342807999977e-06,
      "loss": 0.9062,
      "step": 1222
    },
    {
      "epoch": 0.06593702825102436,
      "grad_norm": 0.8866230249404907,
      "learning_rate": 9.994332721291214e-06,
      "loss": 0.9026,
      "step": 1223
    },
    {
      "epoch": 0.0659909424196679,
      "grad_norm": 0.9135996103286743,
      "learning_rate": 9.994322625603314e-06,
      "loss": 0.8558,
      "step": 1224
    },
    {
      "epoch": 0.06604485658831141,
      "grad_norm": 0.9904530048370361,
      "learning_rate": 9.994312520936297e-06,
      "loss": 0.8823,
      "step": 1225
    },
    {
      "epoch": 0.06609877075695493,
      "grad_norm": 0.8590260148048401,
      "learning_rate": 9.99430240729018e-06,
      "loss": 0.8344,
      "step": 1226
    },
    {
      "epoch": 0.06615268492559845,
      "grad_norm": 1.1669397354125977,
      "learning_rate": 9.99429228466498e-06,
      "loss": 0.9459,
      "step": 1227
    },
    {
      "epoch": 0.06620659909424197,
      "grad_norm": 0.9290857315063477,
      "learning_rate": 9.994282153060715e-06,
      "loss": 0.8723,
      "step": 1228
    },
    {
      "epoch": 0.06626051326288548,
      "grad_norm": 0.9619696140289307,
      "learning_rate": 9.994272012477405e-06,
      "loss": 0.8986,
      "step": 1229
    },
    {
      "epoch": 0.066314427431529,
      "grad_norm": 0.8312071561813354,
      "learning_rate": 9.994261862915068e-06,
      "loss": 0.7291,
      "step": 1230
    },
    {
      "epoch": 0.06636834160017252,
      "grad_norm": 1.0099300146102905,
      "learning_rate": 9.994251704373721e-06,
      "loss": 0.8725,
      "step": 1231
    },
    {
      "epoch": 0.06642225576881604,
      "grad_norm": 0.8522336483001709,
      "learning_rate": 9.994241536853384e-06,
      "loss": 0.8656,
      "step": 1232
    },
    {
      "epoch": 0.06647616993745957,
      "grad_norm": 0.919360339641571,
      "learning_rate": 9.994231360354074e-06,
      "loss": 0.8854,
      "step": 1233
    },
    {
      "epoch": 0.06653008410610309,
      "grad_norm": 0.8002495169639587,
      "learning_rate": 9.994221174875809e-06,
      "loss": 0.7879,
      "step": 1234
    },
    {
      "epoch": 0.0665839982747466,
      "grad_norm": 0.9539757370948792,
      "learning_rate": 9.994210980418607e-06,
      "loss": 0.9027,
      "step": 1235
    },
    {
      "epoch": 0.06663791244339012,
      "grad_norm": 0.9222649335861206,
      "learning_rate": 9.99420077698249e-06,
      "loss": 0.7611,
      "step": 1236
    },
    {
      "epoch": 0.06669182661203364,
      "grad_norm": 0.8629900813102722,
      "learning_rate": 9.994190564567472e-06,
      "loss": 0.8122,
      "step": 1237
    },
    {
      "epoch": 0.06674574078067716,
      "grad_norm": 0.8339203000068665,
      "learning_rate": 9.994180343173574e-06,
      "loss": 0.7873,
      "step": 1238
    },
    {
      "epoch": 0.06679965494932068,
      "grad_norm": 0.8844656348228455,
      "learning_rate": 9.994170112800812e-06,
      "loss": 0.8176,
      "step": 1239
    },
    {
      "epoch": 0.0668535691179642,
      "grad_norm": 1.0024579763412476,
      "learning_rate": 9.994159873449206e-06,
      "loss": 0.844,
      "step": 1240
    },
    {
      "epoch": 0.06690748328660773,
      "grad_norm": 0.8317261338233948,
      "learning_rate": 9.994149625118774e-06,
      "loss": 0.9103,
      "step": 1241
    },
    {
      "epoch": 0.06696139745525125,
      "grad_norm": 0.8915300965309143,
      "learning_rate": 9.994139367809534e-06,
      "loss": 0.9084,
      "step": 1242
    },
    {
      "epoch": 0.06701531162389476,
      "grad_norm": 0.9270803332328796,
      "learning_rate": 9.994129101521506e-06,
      "loss": 0.7634,
      "step": 1243
    },
    {
      "epoch": 0.06706922579253828,
      "grad_norm": 0.9891652464866638,
      "learning_rate": 9.994118826254708e-06,
      "loss": 0.9776,
      "step": 1244
    },
    {
      "epoch": 0.0671231399611818,
      "grad_norm": 0.7778229713439941,
      "learning_rate": 9.994108542009156e-06,
      "loss": 0.7481,
      "step": 1245
    },
    {
      "epoch": 0.06717705412982532,
      "grad_norm": 0.8451201319694519,
      "learning_rate": 9.994098248784872e-06,
      "loss": 0.8012,
      "step": 1246
    },
    {
      "epoch": 0.06723096829846884,
      "grad_norm": 0.8115825057029724,
      "learning_rate": 9.994087946581873e-06,
      "loss": 0.874,
      "step": 1247
    },
    {
      "epoch": 0.06728488246711235,
      "grad_norm": 0.815934419631958,
      "learning_rate": 9.994077635400175e-06,
      "loss": 0.8114,
      "step": 1248
    },
    {
      "epoch": 0.06733879663575587,
      "grad_norm": 1.1179388761520386,
      "learning_rate": 9.9940673152398e-06,
      "loss": 0.9078,
      "step": 1249
    },
    {
      "epoch": 0.0673927108043994,
      "grad_norm": 0.9235454201698303,
      "learning_rate": 9.994056986100767e-06,
      "loss": 0.7511,
      "step": 1250
    },
    {
      "epoch": 0.06744662497304292,
      "grad_norm": 0.8568270206451416,
      "learning_rate": 9.994046647983093e-06,
      "loss": 0.7805,
      "step": 1251
    },
    {
      "epoch": 0.06750053914168644,
      "grad_norm": 1.1337388753890991,
      "learning_rate": 9.994036300886796e-06,
      "loss": 0.8835,
      "step": 1252
    },
    {
      "epoch": 0.06755445331032996,
      "grad_norm": 0.9154239892959595,
      "learning_rate": 9.994025944811896e-06,
      "loss": 0.8804,
      "step": 1253
    },
    {
      "epoch": 0.06760836747897347,
      "grad_norm": 0.8301606774330139,
      "learning_rate": 9.99401557975841e-06,
      "loss": 0.7905,
      "step": 1254
    },
    {
      "epoch": 0.06766228164761699,
      "grad_norm": 0.9907017350196838,
      "learning_rate": 9.994005205726358e-06,
      "loss": 0.9091,
      "step": 1255
    },
    {
      "epoch": 0.06771619581626051,
      "grad_norm": 0.8883876204490662,
      "learning_rate": 9.993994822715758e-06,
      "loss": 0.8815,
      "step": 1256
    },
    {
      "epoch": 0.06777010998490403,
      "grad_norm": 0.9746614098548889,
      "learning_rate": 9.993984430726627e-06,
      "loss": 0.7897,
      "step": 1257
    },
    {
      "epoch": 0.06782402415354755,
      "grad_norm": 0.9773344993591309,
      "learning_rate": 9.993974029758988e-06,
      "loss": 0.8499,
      "step": 1258
    },
    {
      "epoch": 0.06787793832219108,
      "grad_norm": 0.9552164077758789,
      "learning_rate": 9.993963619812856e-06,
      "loss": 0.711,
      "step": 1259
    },
    {
      "epoch": 0.0679318524908346,
      "grad_norm": 0.9146968126296997,
      "learning_rate": 9.993953200888252e-06,
      "loss": 0.9016,
      "step": 1260
    },
    {
      "epoch": 0.06798576665947811,
      "grad_norm": 0.924244225025177,
      "learning_rate": 9.993942772985192e-06,
      "loss": 0.7534,
      "step": 1261
    },
    {
      "epoch": 0.06803968082812163,
      "grad_norm": 1.2963265180587769,
      "learning_rate": 9.993932336103699e-06,
      "loss": 0.9409,
      "step": 1262
    },
    {
      "epoch": 0.06809359499676515,
      "grad_norm": 0.7954462766647339,
      "learning_rate": 9.993921890243788e-06,
      "loss": 0.7669,
      "step": 1263
    },
    {
      "epoch": 0.06814750916540867,
      "grad_norm": 0.9115849137306213,
      "learning_rate": 9.993911435405478e-06,
      "loss": 0.7567,
      "step": 1264
    },
    {
      "epoch": 0.06820142333405219,
      "grad_norm": 1.0030237436294556,
      "learning_rate": 9.99390097158879e-06,
      "loss": 0.8952,
      "step": 1265
    },
    {
      "epoch": 0.0682553375026957,
      "grad_norm": 0.8897690773010254,
      "learning_rate": 9.993890498793742e-06,
      "loss": 0.7993,
      "step": 1266
    },
    {
      "epoch": 0.06830925167133922,
      "grad_norm": 0.9283807277679443,
      "learning_rate": 9.993880017020349e-06,
      "loss": 0.8808,
      "step": 1267
    },
    {
      "epoch": 0.06836316583998275,
      "grad_norm": 0.848922848701477,
      "learning_rate": 9.993869526268637e-06,
      "loss": 0.7979,
      "step": 1268
    },
    {
      "epoch": 0.06841708000862627,
      "grad_norm": 0.8896105289459229,
      "learning_rate": 9.993859026538618e-06,
      "loss": 0.8886,
      "step": 1269
    },
    {
      "epoch": 0.06847099417726979,
      "grad_norm": 0.8602685928344727,
      "learning_rate": 9.993848517830318e-06,
      "loss": 0.8209,
      "step": 1270
    },
    {
      "epoch": 0.06852490834591331,
      "grad_norm": 0.9300077557563782,
      "learning_rate": 9.99383800014375e-06,
      "loss": 0.9261,
      "step": 1271
    },
    {
      "epoch": 0.06857882251455683,
      "grad_norm": 0.8691270351409912,
      "learning_rate": 9.993827473478934e-06,
      "loss": 0.9217,
      "step": 1272
    },
    {
      "epoch": 0.06863273668320034,
      "grad_norm": 0.7943814992904663,
      "learning_rate": 9.99381693783589e-06,
      "loss": 0.8557,
      "step": 1273
    },
    {
      "epoch": 0.06868665085184386,
      "grad_norm": 0.9060125946998596,
      "learning_rate": 9.993806393214638e-06,
      "loss": 0.8314,
      "step": 1274
    },
    {
      "epoch": 0.06874056502048738,
      "grad_norm": 0.8014434576034546,
      "learning_rate": 9.993795839615194e-06,
      "loss": 0.8047,
      "step": 1275
    },
    {
      "epoch": 0.0687944791891309,
      "grad_norm": 1.0498815774917603,
      "learning_rate": 9.993785277037578e-06,
      "loss": 0.7125,
      "step": 1276
    },
    {
      "epoch": 0.06884839335777443,
      "grad_norm": 0.8868438005447388,
      "learning_rate": 9.993774705481812e-06,
      "loss": 0.8594,
      "step": 1277
    },
    {
      "epoch": 0.06890230752641795,
      "grad_norm": 0.8213896155357361,
      "learning_rate": 9.993764124947911e-06,
      "loss": 0.7995,
      "step": 1278
    },
    {
      "epoch": 0.06895622169506146,
      "grad_norm": 0.9007741212844849,
      "learning_rate": 9.993753535435895e-06,
      "loss": 0.8982,
      "step": 1279
    },
    {
      "epoch": 0.06901013586370498,
      "grad_norm": 0.8377478122711182,
      "learning_rate": 9.993742936945785e-06,
      "loss": 0.7387,
      "step": 1280
    },
    {
      "epoch": 0.0690640500323485,
      "grad_norm": 0.8009492754936218,
      "learning_rate": 9.993732329477598e-06,
      "loss": 0.8079,
      "step": 1281
    },
    {
      "epoch": 0.06911796420099202,
      "grad_norm": 0.8478789925575256,
      "learning_rate": 9.993721713031354e-06,
      "loss": 0.8682,
      "step": 1282
    },
    {
      "epoch": 0.06917187836963554,
      "grad_norm": 0.7498561143875122,
      "learning_rate": 9.993711087607072e-06,
      "loss": 0.8107,
      "step": 1283
    },
    {
      "epoch": 0.06922579253827905,
      "grad_norm": 0.8972634077072144,
      "learning_rate": 9.99370045320477e-06,
      "loss": 0.8494,
      "step": 1284
    },
    {
      "epoch": 0.06927970670692257,
      "grad_norm": 0.942449152469635,
      "learning_rate": 9.99368980982447e-06,
      "loss": 0.8487,
      "step": 1285
    },
    {
      "epoch": 0.0693336208755661,
      "grad_norm": 0.8752795457839966,
      "learning_rate": 9.993679157466188e-06,
      "loss": 0.8859,
      "step": 1286
    },
    {
      "epoch": 0.06938753504420962,
      "grad_norm": 0.8289507031440735,
      "learning_rate": 9.993668496129945e-06,
      "loss": 0.8726,
      "step": 1287
    },
    {
      "epoch": 0.06944144921285314,
      "grad_norm": 0.9452151656150818,
      "learning_rate": 9.993657825815759e-06,
      "loss": 0.9266,
      "step": 1288
    },
    {
      "epoch": 0.06949536338149666,
      "grad_norm": 0.8697348237037659,
      "learning_rate": 9.993647146523651e-06,
      "loss": 0.8946,
      "step": 1289
    },
    {
      "epoch": 0.06954927755014018,
      "grad_norm": 0.8712061643600464,
      "learning_rate": 9.993636458253637e-06,
      "loss": 0.8551,
      "step": 1290
    },
    {
      "epoch": 0.0696031917187837,
      "grad_norm": 0.9295617938041687,
      "learning_rate": 9.993625761005739e-06,
      "loss": 0.8963,
      "step": 1291
    },
    {
      "epoch": 0.06965710588742721,
      "grad_norm": 0.9441055059432983,
      "learning_rate": 9.993615054779975e-06,
      "loss": 0.9567,
      "step": 1292
    },
    {
      "epoch": 0.06971102005607073,
      "grad_norm": 0.8742032051086426,
      "learning_rate": 9.993604339576365e-06,
      "loss": 0.8341,
      "step": 1293
    },
    {
      "epoch": 0.06976493422471426,
      "grad_norm": 0.8596220016479492,
      "learning_rate": 9.993593615394928e-06,
      "loss": 0.8576,
      "step": 1294
    },
    {
      "epoch": 0.06981884839335778,
      "grad_norm": 0.8011770844459534,
      "learning_rate": 9.993582882235682e-06,
      "loss": 0.7317,
      "step": 1295
    },
    {
      "epoch": 0.0698727625620013,
      "grad_norm": 0.8578245043754578,
      "learning_rate": 9.993572140098648e-06,
      "loss": 0.8853,
      "step": 1296
    },
    {
      "epoch": 0.06992667673064482,
      "grad_norm": 1.1155178546905518,
      "learning_rate": 9.993561388983845e-06,
      "loss": 0.8199,
      "step": 1297
    },
    {
      "epoch": 0.06998059089928833,
      "grad_norm": 1.035699486732483,
      "learning_rate": 9.993550628891293e-06,
      "loss": 0.9498,
      "step": 1298
    },
    {
      "epoch": 0.07003450506793185,
      "grad_norm": 0.8635748028755188,
      "learning_rate": 9.99353985982101e-06,
      "loss": 0.8741,
      "step": 1299
    },
    {
      "epoch": 0.07008841923657537,
      "grad_norm": 0.8650850653648376,
      "learning_rate": 9.993529081773016e-06,
      "loss": 0.7337,
      "step": 1300
    },
    {
      "epoch": 0.07014233340521889,
      "grad_norm": 0.8334539532661438,
      "learning_rate": 9.99351829474733e-06,
      "loss": 0.8927,
      "step": 1301
    },
    {
      "epoch": 0.0701962475738624,
      "grad_norm": 0.9150926470756531,
      "learning_rate": 9.993507498743971e-06,
      "loss": 0.8464,
      "step": 1302
    },
    {
      "epoch": 0.07025016174250594,
      "grad_norm": 0.8916522860527039,
      "learning_rate": 9.993496693762958e-06,
      "loss": 0.7899,
      "step": 1303
    },
    {
      "epoch": 0.07030407591114946,
      "grad_norm": 1.0224976539611816,
      "learning_rate": 9.993485879804314e-06,
      "loss": 0.8256,
      "step": 1304
    },
    {
      "epoch": 0.07035799007979297,
      "grad_norm": 0.921816885471344,
      "learning_rate": 9.993475056868054e-06,
      "loss": 0.7944,
      "step": 1305
    },
    {
      "epoch": 0.07041190424843649,
      "grad_norm": 0.8775705099105835,
      "learning_rate": 9.9934642249542e-06,
      "loss": 0.9098,
      "step": 1306
    },
    {
      "epoch": 0.07046581841708001,
      "grad_norm": 0.9802567362785339,
      "learning_rate": 9.99345338406277e-06,
      "loss": 0.9756,
      "step": 1307
    },
    {
      "epoch": 0.07051973258572353,
      "grad_norm": 0.9785491228103638,
      "learning_rate": 9.993442534193786e-06,
      "loss": 1.0017,
      "step": 1308
    },
    {
      "epoch": 0.07057364675436704,
      "grad_norm": 0.8796840906143188,
      "learning_rate": 9.993431675347265e-06,
      "loss": 0.7202,
      "step": 1309
    },
    {
      "epoch": 0.07062756092301056,
      "grad_norm": 0.878099799156189,
      "learning_rate": 9.993420807523227e-06,
      "loss": 0.8655,
      "step": 1310
    },
    {
      "epoch": 0.07068147509165408,
      "grad_norm": 0.8361509442329407,
      "learning_rate": 9.99340993072169e-06,
      "loss": 0.8522,
      "step": 1311
    },
    {
      "epoch": 0.07073538926029761,
      "grad_norm": 0.8556873798370361,
      "learning_rate": 9.99339904494268e-06,
      "loss": 0.8603,
      "step": 1312
    },
    {
      "epoch": 0.07078930342894113,
      "grad_norm": 0.8434461355209351,
      "learning_rate": 9.993388150186208e-06,
      "loss": 0.8571,
      "step": 1313
    },
    {
      "epoch": 0.07084321759758465,
      "grad_norm": 0.8545907139778137,
      "learning_rate": 9.9933772464523e-06,
      "loss": 0.8145,
      "step": 1314
    },
    {
      "epoch": 0.07089713176622817,
      "grad_norm": 0.9502561092376709,
      "learning_rate": 9.993366333740971e-06,
      "loss": 0.8068,
      "step": 1315
    },
    {
      "epoch": 0.07095104593487168,
      "grad_norm": 0.848628580570221,
      "learning_rate": 9.993355412052244e-06,
      "loss": 0.8793,
      "step": 1316
    },
    {
      "epoch": 0.0710049601035152,
      "grad_norm": 0.9699797630310059,
      "learning_rate": 9.993344481386137e-06,
      "loss": 0.9904,
      "step": 1317
    },
    {
      "epoch": 0.07105887427215872,
      "grad_norm": 0.8888396620750427,
      "learning_rate": 9.993333541742671e-06,
      "loss": 0.8363,
      "step": 1318
    },
    {
      "epoch": 0.07111278844080224,
      "grad_norm": 0.8805423974990845,
      "learning_rate": 9.993322593121863e-06,
      "loss": 0.8905,
      "step": 1319
    },
    {
      "epoch": 0.07116670260944576,
      "grad_norm": 0.8875272274017334,
      "learning_rate": 9.993311635523736e-06,
      "loss": 0.7717,
      "step": 1320
    },
    {
      "epoch": 0.07122061677808929,
      "grad_norm": 0.8853299617767334,
      "learning_rate": 9.993300668948308e-06,
      "loss": 0.9077,
      "step": 1321
    },
    {
      "epoch": 0.0712745309467328,
      "grad_norm": 0.8847644329071045,
      "learning_rate": 9.993289693395599e-06,
      "loss": 0.8362,
      "step": 1322
    },
    {
      "epoch": 0.07132844511537632,
      "grad_norm": 0.9531683325767517,
      "learning_rate": 9.993278708865629e-06,
      "loss": 0.8848,
      "step": 1323
    },
    {
      "epoch": 0.07138235928401984,
      "grad_norm": 0.8573325276374817,
      "learning_rate": 9.993267715358414e-06,
      "loss": 0.8367,
      "step": 1324
    },
    {
      "epoch": 0.07143627345266336,
      "grad_norm": 0.8920298218727112,
      "learning_rate": 9.99325671287398e-06,
      "loss": 0.8838,
      "step": 1325
    },
    {
      "epoch": 0.07149018762130688,
      "grad_norm": 0.8472782969474792,
      "learning_rate": 9.993245701412343e-06,
      "loss": 0.8313,
      "step": 1326
    },
    {
      "epoch": 0.0715441017899504,
      "grad_norm": 1.047664761543274,
      "learning_rate": 9.993234680973525e-06,
      "loss": 0.8663,
      "step": 1327
    },
    {
      "epoch": 0.07159801595859391,
      "grad_norm": 0.9395570158958435,
      "learning_rate": 9.993223651557542e-06,
      "loss": 0.7703,
      "step": 1328
    },
    {
      "epoch": 0.07165193012723743,
      "grad_norm": 0.9125472903251648,
      "learning_rate": 9.993212613164419e-06,
      "loss": 0.9335,
      "step": 1329
    },
    {
      "epoch": 0.07170584429588096,
      "grad_norm": 0.9043323397636414,
      "learning_rate": 9.993201565794172e-06,
      "loss": 0.9185,
      "step": 1330
    },
    {
      "epoch": 0.07175975846452448,
      "grad_norm": 0.8764339089393616,
      "learning_rate": 9.993190509446821e-06,
      "loss": 0.8807,
      "step": 1331
    },
    {
      "epoch": 0.071813672633168,
      "grad_norm": 0.9123268723487854,
      "learning_rate": 9.99317944412239e-06,
      "loss": 0.8134,
      "step": 1332
    },
    {
      "epoch": 0.07186758680181152,
      "grad_norm": 0.9625567197799683,
      "learning_rate": 9.993168369820892e-06,
      "loss": 0.8132,
      "step": 1333
    },
    {
      "epoch": 0.07192150097045504,
      "grad_norm": 0.880536675453186,
      "learning_rate": 9.993157286542352e-06,
      "loss": 0.8107,
      "step": 1334
    },
    {
      "epoch": 0.07197541513909855,
      "grad_norm": 0.9165224432945251,
      "learning_rate": 9.99314619428679e-06,
      "loss": 0.8376,
      "step": 1335
    },
    {
      "epoch": 0.07202932930774207,
      "grad_norm": 0.8278066515922546,
      "learning_rate": 9.993135093054223e-06,
      "loss": 0.8075,
      "step": 1336
    },
    {
      "epoch": 0.07208324347638559,
      "grad_norm": 0.9237795472145081,
      "learning_rate": 9.993123982844674e-06,
      "loss": 0.7838,
      "step": 1337
    },
    {
      "epoch": 0.0721371576450291,
      "grad_norm": 0.8200939297676086,
      "learning_rate": 9.993112863658161e-06,
      "loss": 0.8475,
      "step": 1338
    },
    {
      "epoch": 0.07219107181367264,
      "grad_norm": 0.8505958318710327,
      "learning_rate": 9.993101735494704e-06,
      "loss": 0.7891,
      "step": 1339
    },
    {
      "epoch": 0.07224498598231616,
      "grad_norm": 0.8407264351844788,
      "learning_rate": 9.993090598354323e-06,
      "loss": 0.8128,
      "step": 1340
    },
    {
      "epoch": 0.07229890015095967,
      "grad_norm": 0.8039887547492981,
      "learning_rate": 9.993079452237038e-06,
      "loss": 0.8504,
      "step": 1341
    },
    {
      "epoch": 0.07235281431960319,
      "grad_norm": 0.7590643167495728,
      "learning_rate": 9.993068297142871e-06,
      "loss": 0.7402,
      "step": 1342
    },
    {
      "epoch": 0.07240672848824671,
      "grad_norm": 0.7866249680519104,
      "learning_rate": 9.993057133071842e-06,
      "loss": 0.7076,
      "step": 1343
    },
    {
      "epoch": 0.07246064265689023,
      "grad_norm": 0.9846029281616211,
      "learning_rate": 9.993045960023967e-06,
      "loss": 0.9179,
      "step": 1344
    },
    {
      "epoch": 0.07251455682553375,
      "grad_norm": 0.8918319940567017,
      "learning_rate": 9.99303477799927e-06,
      "loss": 0.8087,
      "step": 1345
    },
    {
      "epoch": 0.07256847099417726,
      "grad_norm": 0.8407700061798096,
      "learning_rate": 9.99302358699777e-06,
      "loss": 0.7272,
      "step": 1346
    },
    {
      "epoch": 0.0726223851628208,
      "grad_norm": 0.9637326598167419,
      "learning_rate": 9.993012387019486e-06,
      "loss": 0.8613,
      "step": 1347
    },
    {
      "epoch": 0.07267629933146431,
      "grad_norm": 0.8362317681312561,
      "learning_rate": 9.99300117806444e-06,
      "loss": 0.917,
      "step": 1348
    },
    {
      "epoch": 0.07273021350010783,
      "grad_norm": 0.8584982752799988,
      "learning_rate": 9.992989960132651e-06,
      "loss": 0.8857,
      "step": 1349
    },
    {
      "epoch": 0.07278412766875135,
      "grad_norm": 0.8341198563575745,
      "learning_rate": 9.992978733224139e-06,
      "loss": 0.802,
      "step": 1350
    },
    {
      "epoch": 0.07283804183739487,
      "grad_norm": 1.6860167980194092,
      "learning_rate": 9.992967497338926e-06,
      "loss": 0.8789,
      "step": 1351
    },
    {
      "epoch": 0.07289195600603839,
      "grad_norm": 0.8399189114570618,
      "learning_rate": 9.99295625247703e-06,
      "loss": 0.6338,
      "step": 1352
    },
    {
      "epoch": 0.0729458701746819,
      "grad_norm": 0.9616976976394653,
      "learning_rate": 9.992944998638473e-06,
      "loss": 0.9735,
      "step": 1353
    },
    {
      "epoch": 0.07299978434332542,
      "grad_norm": 0.8592861890792847,
      "learning_rate": 9.992933735823272e-06,
      "loss": 0.8159,
      "step": 1354
    },
    {
      "epoch": 0.07305369851196894,
      "grad_norm": 0.8448725342750549,
      "learning_rate": 9.992922464031451e-06,
      "loss": 0.7942,
      "step": 1355
    },
    {
      "epoch": 0.07310761268061247,
      "grad_norm": 0.8015927672386169,
      "learning_rate": 9.99291118326303e-06,
      "loss": 0.7429,
      "step": 1356
    },
    {
      "epoch": 0.07316152684925599,
      "grad_norm": 0.8255912065505981,
      "learning_rate": 9.992899893518025e-06,
      "loss": 0.8532,
      "step": 1357
    },
    {
      "epoch": 0.07321544101789951,
      "grad_norm": 0.8764085173606873,
      "learning_rate": 9.992888594796462e-06,
      "loss": 0.7989,
      "step": 1358
    },
    {
      "epoch": 0.07326935518654303,
      "grad_norm": 0.8405522704124451,
      "learning_rate": 9.992877287098357e-06,
      "loss": 0.8709,
      "step": 1359
    },
    {
      "epoch": 0.07332326935518654,
      "grad_norm": 0.8657836318016052,
      "learning_rate": 9.992865970423733e-06,
      "loss": 0.8236,
      "step": 1360
    },
    {
      "epoch": 0.07337718352383006,
      "grad_norm": 0.8817959427833557,
      "learning_rate": 9.992854644772609e-06,
      "loss": 0.902,
      "step": 1361
    },
    {
      "epoch": 0.07343109769247358,
      "grad_norm": 0.8290701508522034,
      "learning_rate": 9.992843310145006e-06,
      "loss": 0.8454,
      "step": 1362
    },
    {
      "epoch": 0.0734850118611171,
      "grad_norm": 0.9637642502784729,
      "learning_rate": 9.992831966540946e-06,
      "loss": 0.9414,
      "step": 1363
    },
    {
      "epoch": 0.07353892602976062,
      "grad_norm": 0.9220197200775146,
      "learning_rate": 9.992820613960446e-06,
      "loss": 0.9827,
      "step": 1364
    },
    {
      "epoch": 0.07359284019840415,
      "grad_norm": 0.9008362889289856,
      "learning_rate": 9.992809252403526e-06,
      "loss": 0.8388,
      "step": 1365
    },
    {
      "epoch": 0.07364675436704766,
      "grad_norm": 0.9517331123352051,
      "learning_rate": 9.992797881870212e-06,
      "loss": 0.8758,
      "step": 1366
    },
    {
      "epoch": 0.07370066853569118,
      "grad_norm": 0.7811571359634399,
      "learning_rate": 9.992786502360517e-06,
      "loss": 0.6984,
      "step": 1367
    },
    {
      "epoch": 0.0737545827043347,
      "grad_norm": 0.9887184500694275,
      "learning_rate": 9.992775113874466e-06,
      "loss": 0.7832,
      "step": 1368
    },
    {
      "epoch": 0.07380849687297822,
      "grad_norm": 1.025869607925415,
      "learning_rate": 9.99276371641208e-06,
      "loss": 0.8417,
      "step": 1369
    },
    {
      "epoch": 0.07386241104162174,
      "grad_norm": 0.8479165434837341,
      "learning_rate": 9.99275230997338e-06,
      "loss": 0.7862,
      "step": 1370
    },
    {
      "epoch": 0.07391632521026525,
      "grad_norm": 0.9213555455207825,
      "learning_rate": 9.992740894558381e-06,
      "loss": 0.915,
      "step": 1371
    },
    {
      "epoch": 0.07397023937890877,
      "grad_norm": 0.832306444644928,
      "learning_rate": 9.992729470167109e-06,
      "loss": 0.7566,
      "step": 1372
    },
    {
      "epoch": 0.07402415354755229,
      "grad_norm": 1.0360348224639893,
      "learning_rate": 9.992718036799583e-06,
      "loss": 0.9096,
      "step": 1373
    },
    {
      "epoch": 0.07407806771619582,
      "grad_norm": 0.8898483514785767,
      "learning_rate": 9.992706594455823e-06,
      "loss": 0.8738,
      "step": 1374
    },
    {
      "epoch": 0.07413198188483934,
      "grad_norm": 0.8813758492469788,
      "learning_rate": 9.992695143135849e-06,
      "loss": 0.8736,
      "step": 1375
    },
    {
      "epoch": 0.07418589605348286,
      "grad_norm": 1.1480571031570435,
      "learning_rate": 9.992683682839683e-06,
      "loss": 0.915,
      "step": 1376
    },
    {
      "epoch": 0.07423981022212638,
      "grad_norm": 0.8588376641273499,
      "learning_rate": 9.992672213567345e-06,
      "loss": 0.8295,
      "step": 1377
    },
    {
      "epoch": 0.0742937243907699,
      "grad_norm": 0.8729918599128723,
      "learning_rate": 9.992660735318858e-06,
      "loss": 0.9058,
      "step": 1378
    },
    {
      "epoch": 0.07434763855941341,
      "grad_norm": 0.7953224778175354,
      "learning_rate": 9.992649248094236e-06,
      "loss": 0.7857,
      "step": 1379
    },
    {
      "epoch": 0.07440155272805693,
      "grad_norm": 0.8485717177391052,
      "learning_rate": 9.992637751893508e-06,
      "loss": 0.7641,
      "step": 1380
    },
    {
      "epoch": 0.07445546689670045,
      "grad_norm": 0.8630878329277039,
      "learning_rate": 9.99262624671669e-06,
      "loss": 0.8624,
      "step": 1381
    },
    {
      "epoch": 0.07450938106534397,
      "grad_norm": 0.8655185103416443,
      "learning_rate": 9.992614732563802e-06,
      "loss": 0.8428,
      "step": 1382
    },
    {
      "epoch": 0.0745632952339875,
      "grad_norm": 0.7875732779502869,
      "learning_rate": 9.992603209434868e-06,
      "loss": 0.7272,
      "step": 1383
    },
    {
      "epoch": 0.07461720940263102,
      "grad_norm": 0.875879168510437,
      "learning_rate": 9.992591677329905e-06,
      "loss": 0.8539,
      "step": 1384
    },
    {
      "epoch": 0.07467112357127453,
      "grad_norm": 0.8618319034576416,
      "learning_rate": 9.992580136248934e-06,
      "loss": 0.879,
      "step": 1385
    },
    {
      "epoch": 0.07472503773991805,
      "grad_norm": 0.8695591688156128,
      "learning_rate": 9.992568586191981e-06,
      "loss": 0.8477,
      "step": 1386
    },
    {
      "epoch": 0.07477895190856157,
      "grad_norm": 0.8539825677871704,
      "learning_rate": 9.992557027159062e-06,
      "loss": 0.7347,
      "step": 1387
    },
    {
      "epoch": 0.07483286607720509,
      "grad_norm": 0.9625217914581299,
      "learning_rate": 9.992545459150197e-06,
      "loss": 0.8561,
      "step": 1388
    },
    {
      "epoch": 0.0748867802458486,
      "grad_norm": 0.9862298369407654,
      "learning_rate": 9.992533882165409e-06,
      "loss": 0.9583,
      "step": 1389
    },
    {
      "epoch": 0.07494069441449212,
      "grad_norm": 0.8217719793319702,
      "learning_rate": 9.99252229620472e-06,
      "loss": 0.7995,
      "step": 1390
    },
    {
      "epoch": 0.07499460858313564,
      "grad_norm": 0.8668621182441711,
      "learning_rate": 9.992510701268147e-06,
      "loss": 0.8484,
      "step": 1391
    },
    {
      "epoch": 0.07504852275177917,
      "grad_norm": 0.8549453616142273,
      "learning_rate": 9.992499097355716e-06,
      "loss": 0.8552,
      "step": 1392
    },
    {
      "epoch": 0.07510243692042269,
      "grad_norm": 0.8262618184089661,
      "learning_rate": 9.992487484467444e-06,
      "loss": 0.7054,
      "step": 1393
    },
    {
      "epoch": 0.07515635108906621,
      "grad_norm": 0.8524961471557617,
      "learning_rate": 9.992475862603352e-06,
      "loss": 0.8231,
      "step": 1394
    },
    {
      "epoch": 0.07521026525770973,
      "grad_norm": 0.7805570363998413,
      "learning_rate": 9.99246423176346e-06,
      "loss": 0.7778,
      "step": 1395
    },
    {
      "epoch": 0.07526417942635324,
      "grad_norm": 0.950484037399292,
      "learning_rate": 9.992452591947794e-06,
      "loss": 0.8662,
      "step": 1396
    },
    {
      "epoch": 0.07531809359499676,
      "grad_norm": 0.8746458888053894,
      "learning_rate": 9.99244094315637e-06,
      "loss": 0.7854,
      "step": 1397
    },
    {
      "epoch": 0.07537200776364028,
      "grad_norm": 0.9450538754463196,
      "learning_rate": 9.992429285389212e-06,
      "loss": 0.954,
      "step": 1398
    },
    {
      "epoch": 0.0754259219322838,
      "grad_norm": 0.9048300385475159,
      "learning_rate": 9.992417618646337e-06,
      "loss": 0.8915,
      "step": 1399
    },
    {
      "epoch": 0.07547983610092733,
      "grad_norm": 0.8735381364822388,
      "learning_rate": 9.99240594292777e-06,
      "loss": 0.8391,
      "step": 1400
    },
    {
      "epoch": 0.07553375026957085,
      "grad_norm": 1.0980675220489502,
      "learning_rate": 9.99239425823353e-06,
      "loss": 0.8892,
      "step": 1401
    },
    {
      "epoch": 0.07558766443821437,
      "grad_norm": 0.9016425013542175,
      "learning_rate": 9.992382564563638e-06,
      "loss": 0.8192,
      "step": 1402
    },
    {
      "epoch": 0.07564157860685788,
      "grad_norm": 0.801419198513031,
      "learning_rate": 9.992370861918117e-06,
      "loss": 0.7914,
      "step": 1403
    },
    {
      "epoch": 0.0756954927755014,
      "grad_norm": 0.9043407440185547,
      "learning_rate": 9.992359150296985e-06,
      "loss": 0.8767,
      "step": 1404
    },
    {
      "epoch": 0.07574940694414492,
      "grad_norm": 0.9703086018562317,
      "learning_rate": 9.992347429700266e-06,
      "loss": 0.9173,
      "step": 1405
    },
    {
      "epoch": 0.07580332111278844,
      "grad_norm": 0.8154104351997375,
      "learning_rate": 9.992335700127978e-06,
      "loss": 0.8453,
      "step": 1406
    },
    {
      "epoch": 0.07585723528143196,
      "grad_norm": 0.8551482558250427,
      "learning_rate": 9.992323961580146e-06,
      "loss": 0.9132,
      "step": 1407
    },
    {
      "epoch": 0.07591114945007547,
      "grad_norm": 0.9425063729286194,
      "learning_rate": 9.992312214056785e-06,
      "loss": 0.8171,
      "step": 1408
    },
    {
      "epoch": 0.075965063618719,
      "grad_norm": 0.8958794474601746,
      "learning_rate": 9.992300457557922e-06,
      "loss": 0.7983,
      "step": 1409
    },
    {
      "epoch": 0.07601897778736252,
      "grad_norm": 0.873874843120575,
      "learning_rate": 9.992288692083579e-06,
      "loss": 0.798,
      "step": 1410
    },
    {
      "epoch": 0.07607289195600604,
      "grad_norm": 0.7951189279556274,
      "learning_rate": 9.99227691763377e-06,
      "loss": 0.8671,
      "step": 1411
    },
    {
      "epoch": 0.07612680612464956,
      "grad_norm": 0.8073802590370178,
      "learning_rate": 9.992265134208522e-06,
      "loss": 0.8214,
      "step": 1412
    },
    {
      "epoch": 0.07618072029329308,
      "grad_norm": 0.918222188949585,
      "learning_rate": 9.992253341807854e-06,
      "loss": 0.807,
      "step": 1413
    },
    {
      "epoch": 0.0762346344619366,
      "grad_norm": 0.834381103515625,
      "learning_rate": 9.992241540431789e-06,
      "loss": 0.8737,
      "step": 1414
    },
    {
      "epoch": 0.07628854863058011,
      "grad_norm": 0.808437168598175,
      "learning_rate": 9.992229730080347e-06,
      "loss": 0.7982,
      "step": 1415
    },
    {
      "epoch": 0.07634246279922363,
      "grad_norm": 0.7868708968162537,
      "learning_rate": 9.992217910753547e-06,
      "loss": 0.7071,
      "step": 1416
    },
    {
      "epoch": 0.07639637696786715,
      "grad_norm": 0.8445919156074524,
      "learning_rate": 9.992206082451416e-06,
      "loss": 0.8353,
      "step": 1417
    },
    {
      "epoch": 0.07645029113651068,
      "grad_norm": 0.8283419609069824,
      "learning_rate": 9.992194245173969e-06,
      "loss": 0.867,
      "step": 1418
    },
    {
      "epoch": 0.0765042053051542,
      "grad_norm": 0.8390635251998901,
      "learning_rate": 9.99218239892123e-06,
      "loss": 0.822,
      "step": 1419
    },
    {
      "epoch": 0.07655811947379772,
      "grad_norm": 0.9037001132965088,
      "learning_rate": 9.992170543693222e-06,
      "loss": 0.8759,
      "step": 1420
    },
    {
      "epoch": 0.07661203364244124,
      "grad_norm": 0.9708169102668762,
      "learning_rate": 9.992158679489965e-06,
      "loss": 0.875,
      "step": 1421
    },
    {
      "epoch": 0.07666594781108475,
      "grad_norm": 0.8712205290794373,
      "learning_rate": 9.992146806311479e-06,
      "loss": 0.8711,
      "step": 1422
    },
    {
      "epoch": 0.07671986197972827,
      "grad_norm": 0.953936755657196,
      "learning_rate": 9.992134924157786e-06,
      "loss": 0.8117,
      "step": 1423
    },
    {
      "epoch": 0.07677377614837179,
      "grad_norm": 1.3178669214248657,
      "learning_rate": 9.992123033028908e-06,
      "loss": 0.8932,
      "step": 1424
    },
    {
      "epoch": 0.0768276903170153,
      "grad_norm": 0.8657799959182739,
      "learning_rate": 9.992111132924867e-06,
      "loss": 0.8429,
      "step": 1425
    },
    {
      "epoch": 0.07688160448565882,
      "grad_norm": 0.8979378938674927,
      "learning_rate": 9.992099223845681e-06,
      "loss": 0.9165,
      "step": 1426
    },
    {
      "epoch": 0.07693551865430236,
      "grad_norm": 0.797493040561676,
      "learning_rate": 9.992087305791376e-06,
      "loss": 0.8139,
      "step": 1427
    },
    {
      "epoch": 0.07698943282294587,
      "grad_norm": 0.9762497544288635,
      "learning_rate": 9.99207537876197e-06,
      "loss": 0.8006,
      "step": 1428
    },
    {
      "epoch": 0.07704334699158939,
      "grad_norm": 0.9322238564491272,
      "learning_rate": 9.992063442757487e-06,
      "loss": 0.8708,
      "step": 1429
    },
    {
      "epoch": 0.07709726116023291,
      "grad_norm": 0.9208402037620544,
      "learning_rate": 9.992051497777947e-06,
      "loss": 0.9137,
      "step": 1430
    },
    {
      "epoch": 0.07715117532887643,
      "grad_norm": 0.9262849688529968,
      "learning_rate": 9.99203954382337e-06,
      "loss": 0.8043,
      "step": 1431
    },
    {
      "epoch": 0.07720508949751995,
      "grad_norm": 1.0556507110595703,
      "learning_rate": 9.992027580893781e-06,
      "loss": 0.8321,
      "step": 1432
    },
    {
      "epoch": 0.07725900366616346,
      "grad_norm": 1.0503417253494263,
      "learning_rate": 9.9920156089892e-06,
      "loss": 0.8875,
      "step": 1433
    },
    {
      "epoch": 0.07731291783480698,
      "grad_norm": 0.8772387504577637,
      "learning_rate": 9.992003628109647e-06,
      "loss": 0.7407,
      "step": 1434
    },
    {
      "epoch": 0.0773668320034505,
      "grad_norm": 0.942286491394043,
      "learning_rate": 9.991991638255146e-06,
      "loss": 0.8493,
      "step": 1435
    },
    {
      "epoch": 0.07742074617209403,
      "grad_norm": 0.8584794998168945,
      "learning_rate": 9.991979639425717e-06,
      "loss": 0.8003,
      "step": 1436
    },
    {
      "epoch": 0.07747466034073755,
      "grad_norm": 0.8247780203819275,
      "learning_rate": 9.99196763162138e-06,
      "loss": 0.9156,
      "step": 1437
    },
    {
      "epoch": 0.07752857450938107,
      "grad_norm": 0.859018862247467,
      "learning_rate": 9.99195561484216e-06,
      "loss": 0.8255,
      "step": 1438
    },
    {
      "epoch": 0.07758248867802459,
      "grad_norm": 0.9073282480239868,
      "learning_rate": 9.991943589088078e-06,
      "loss": 0.903,
      "step": 1439
    },
    {
      "epoch": 0.0776364028466681,
      "grad_norm": 0.9324385523796082,
      "learning_rate": 9.991931554359154e-06,
      "loss": 0.8618,
      "step": 1440
    },
    {
      "epoch": 0.07769031701531162,
      "grad_norm": 0.8038938045501709,
      "learning_rate": 9.991919510655409e-06,
      "loss": 0.7545,
      "step": 1441
    },
    {
      "epoch": 0.07774423118395514,
      "grad_norm": 0.7999526858329773,
      "learning_rate": 9.991907457976866e-06,
      "loss": 0.6804,
      "step": 1442
    },
    {
      "epoch": 0.07779814535259866,
      "grad_norm": 1.0165048837661743,
      "learning_rate": 9.991895396323548e-06,
      "loss": 0.7664,
      "step": 1443
    },
    {
      "epoch": 0.07785205952124218,
      "grad_norm": 0.9513073563575745,
      "learning_rate": 9.991883325695475e-06,
      "loss": 0.8115,
      "step": 1444
    },
    {
      "epoch": 0.07790597368988571,
      "grad_norm": 1.0391769409179688,
      "learning_rate": 9.991871246092669e-06,
      "loss": 0.9197,
      "step": 1445
    },
    {
      "epoch": 0.07795988785852923,
      "grad_norm": 0.8990768194198608,
      "learning_rate": 9.991859157515151e-06,
      "loss": 0.9507,
      "step": 1446
    },
    {
      "epoch": 0.07801380202717274,
      "grad_norm": 0.9990912079811096,
      "learning_rate": 9.991847059962945e-06,
      "loss": 0.7951,
      "step": 1447
    },
    {
      "epoch": 0.07806771619581626,
      "grad_norm": 1.0030032396316528,
      "learning_rate": 9.99183495343607e-06,
      "loss": 0.7237,
      "step": 1448
    },
    {
      "epoch": 0.07812163036445978,
      "grad_norm": 0.889561116695404,
      "learning_rate": 9.991822837934551e-06,
      "loss": 0.9061,
      "step": 1449
    },
    {
      "epoch": 0.0781755445331033,
      "grad_norm": 0.8766982555389404,
      "learning_rate": 9.991810713458405e-06,
      "loss": 0.7952,
      "step": 1450
    },
    {
      "epoch": 0.07822945870174682,
      "grad_norm": 0.9144406914710999,
      "learning_rate": 9.991798580007658e-06,
      "loss": 0.9235,
      "step": 1451
    },
    {
      "epoch": 0.07828337287039033,
      "grad_norm": 0.895516037940979,
      "learning_rate": 9.99178643758233e-06,
      "loss": 0.9469,
      "step": 1452
    },
    {
      "epoch": 0.07833728703903386,
      "grad_norm": 0.8802943229675293,
      "learning_rate": 9.991774286182443e-06,
      "loss": 0.8548,
      "step": 1453
    },
    {
      "epoch": 0.07839120120767738,
      "grad_norm": 1.2773913145065308,
      "learning_rate": 9.99176212580802e-06,
      "loss": 0.794,
      "step": 1454
    },
    {
      "epoch": 0.0784451153763209,
      "grad_norm": 0.9501168131828308,
      "learning_rate": 9.99174995645908e-06,
      "loss": 0.8711,
      "step": 1455
    },
    {
      "epoch": 0.07849902954496442,
      "grad_norm": 0.9047390222549438,
      "learning_rate": 9.991737778135649e-06,
      "loss": 0.8419,
      "step": 1456
    },
    {
      "epoch": 0.07855294371360794,
      "grad_norm": 0.9492837190628052,
      "learning_rate": 9.991725590837747e-06,
      "loss": 0.9832,
      "step": 1457
    },
    {
      "epoch": 0.07860685788225145,
      "grad_norm": 0.9585106372833252,
      "learning_rate": 9.991713394565394e-06,
      "loss": 0.8393,
      "step": 1458
    },
    {
      "epoch": 0.07866077205089497,
      "grad_norm": 0.9568297266960144,
      "learning_rate": 9.991701189318615e-06,
      "loss": 0.8711,
      "step": 1459
    },
    {
      "epoch": 0.07871468621953849,
      "grad_norm": 0.9201347231864929,
      "learning_rate": 9.991688975097429e-06,
      "loss": 0.7947,
      "step": 1460
    },
    {
      "epoch": 0.07876860038818201,
      "grad_norm": 0.8375768661499023,
      "learning_rate": 9.99167675190186e-06,
      "loss": 0.8051,
      "step": 1461
    },
    {
      "epoch": 0.07882251455682554,
      "grad_norm": 0.8397765755653381,
      "learning_rate": 9.99166451973193e-06,
      "loss": 0.7727,
      "step": 1462
    },
    {
      "epoch": 0.07887642872546906,
      "grad_norm": 0.8697947859764099,
      "learning_rate": 9.99165227858766e-06,
      "loss": 0.8171,
      "step": 1463
    },
    {
      "epoch": 0.07893034289411258,
      "grad_norm": 0.8894750475883484,
      "learning_rate": 9.991640028469073e-06,
      "loss": 0.8773,
      "step": 1464
    },
    {
      "epoch": 0.0789842570627561,
      "grad_norm": 0.8817871809005737,
      "learning_rate": 9.991627769376189e-06,
      "loss": 0.8983,
      "step": 1465
    },
    {
      "epoch": 0.07903817123139961,
      "grad_norm": 0.9241123795509338,
      "learning_rate": 9.99161550130903e-06,
      "loss": 0.8967,
      "step": 1466
    },
    {
      "epoch": 0.07909208540004313,
      "grad_norm": 0.852982223033905,
      "learning_rate": 9.991603224267623e-06,
      "loss": 0.9054,
      "step": 1467
    },
    {
      "epoch": 0.07914599956868665,
      "grad_norm": 0.7719098925590515,
      "learning_rate": 9.991590938251986e-06,
      "loss": 0.7845,
      "step": 1468
    },
    {
      "epoch": 0.07919991373733017,
      "grad_norm": 0.8700329661369324,
      "learning_rate": 9.99157864326214e-06,
      "loss": 0.9664,
      "step": 1469
    },
    {
      "epoch": 0.07925382790597368,
      "grad_norm": 0.880553126335144,
      "learning_rate": 9.991566339298112e-06,
      "loss": 0.8803,
      "step": 1470
    },
    {
      "epoch": 0.07930774207461722,
      "grad_norm": 0.9425762295722961,
      "learning_rate": 9.991554026359918e-06,
      "loss": 0.8259,
      "step": 1471
    },
    {
      "epoch": 0.07936165624326073,
      "grad_norm": 0.8611294031143188,
      "learning_rate": 9.991541704447585e-06,
      "loss": 0.8693,
      "step": 1472
    },
    {
      "epoch": 0.07941557041190425,
      "grad_norm": 0.856023907661438,
      "learning_rate": 9.99152937356113e-06,
      "loss": 0.7073,
      "step": 1473
    },
    {
      "epoch": 0.07946948458054777,
      "grad_norm": 0.7763693332672119,
      "learning_rate": 9.991517033700582e-06,
      "loss": 0.6815,
      "step": 1474
    },
    {
      "epoch": 0.07952339874919129,
      "grad_norm": 0.8417321443557739,
      "learning_rate": 9.991504684865959e-06,
      "loss": 0.8239,
      "step": 1475
    },
    {
      "epoch": 0.0795773129178348,
      "grad_norm": 0.9151323437690735,
      "learning_rate": 9.991492327057282e-06,
      "loss": 0.8327,
      "step": 1476
    },
    {
      "epoch": 0.07963122708647832,
      "grad_norm": 0.8285405039787292,
      "learning_rate": 9.991479960274576e-06,
      "loss": 0.8623,
      "step": 1477
    },
    {
      "epoch": 0.07968514125512184,
      "grad_norm": 0.8204792141914368,
      "learning_rate": 9.991467584517863e-06,
      "loss": 0.8494,
      "step": 1478
    },
    {
      "epoch": 0.07973905542376536,
      "grad_norm": 0.8516230583190918,
      "learning_rate": 9.991455199787164e-06,
      "loss": 0.8219,
      "step": 1479
    },
    {
      "epoch": 0.07979296959240889,
      "grad_norm": 0.9418333172798157,
      "learning_rate": 9.991442806082501e-06,
      "loss": 0.9293,
      "step": 1480
    },
    {
      "epoch": 0.07984688376105241,
      "grad_norm": 0.8852763175964355,
      "learning_rate": 9.991430403403898e-06,
      "loss": 0.8124,
      "step": 1481
    },
    {
      "epoch": 0.07990079792969593,
      "grad_norm": 0.8435791730880737,
      "learning_rate": 9.991417991751376e-06,
      "loss": 0.8634,
      "step": 1482
    },
    {
      "epoch": 0.07995471209833944,
      "grad_norm": 0.7795083522796631,
      "learning_rate": 9.991405571124957e-06,
      "loss": 0.802,
      "step": 1483
    },
    {
      "epoch": 0.08000862626698296,
      "grad_norm": 0.8102303743362427,
      "learning_rate": 9.991393141524663e-06,
      "loss": 0.7492,
      "step": 1484
    },
    {
      "epoch": 0.08006254043562648,
      "grad_norm": 0.8433593511581421,
      "learning_rate": 9.99138070295052e-06,
      "loss": 0.7926,
      "step": 1485
    },
    {
      "epoch": 0.08011645460427,
      "grad_norm": 0.8992267847061157,
      "learning_rate": 9.991368255402546e-06,
      "loss": 0.7859,
      "step": 1486
    },
    {
      "epoch": 0.08017036877291352,
      "grad_norm": 0.8748059868812561,
      "learning_rate": 9.991355798880765e-06,
      "loss": 0.8245,
      "step": 1487
    },
    {
      "epoch": 0.08022428294155703,
      "grad_norm": 0.8456832766532898,
      "learning_rate": 9.9913433333852e-06,
      "loss": 0.9009,
      "step": 1488
    },
    {
      "epoch": 0.08027819711020057,
      "grad_norm": 0.8582474589347839,
      "learning_rate": 9.991330858915873e-06,
      "loss": 0.7607,
      "step": 1489
    },
    {
      "epoch": 0.08033211127884408,
      "grad_norm": 0.8157060146331787,
      "learning_rate": 9.991318375472807e-06,
      "loss": 0.8426,
      "step": 1490
    },
    {
      "epoch": 0.0803860254474876,
      "grad_norm": 0.7474784851074219,
      "learning_rate": 9.991305883056021e-06,
      "loss": 0.8014,
      "step": 1491
    },
    {
      "epoch": 0.08043993961613112,
      "grad_norm": 0.8432475924491882,
      "learning_rate": 9.991293381665543e-06,
      "loss": 0.8254,
      "step": 1492
    },
    {
      "epoch": 0.08049385378477464,
      "grad_norm": 0.8733057379722595,
      "learning_rate": 9.991280871301392e-06,
      "loss": 0.8694,
      "step": 1493
    },
    {
      "epoch": 0.08054776795341816,
      "grad_norm": 0.8694074153900146,
      "learning_rate": 9.991268351963592e-06,
      "loss": 0.7306,
      "step": 1494
    },
    {
      "epoch": 0.08060168212206167,
      "grad_norm": 0.8981258869171143,
      "learning_rate": 9.991255823652162e-06,
      "loss": 0.7821,
      "step": 1495
    },
    {
      "epoch": 0.08065559629070519,
      "grad_norm": 0.9740719795227051,
      "learning_rate": 9.99124328636713e-06,
      "loss": 0.7678,
      "step": 1496
    },
    {
      "epoch": 0.08070951045934871,
      "grad_norm": 0.8847763538360596,
      "learning_rate": 9.991230740108515e-06,
      "loss": 0.73,
      "step": 1497
    },
    {
      "epoch": 0.08076342462799224,
      "grad_norm": 0.8909339308738708,
      "learning_rate": 9.99121818487634e-06,
      "loss": 0.7713,
      "step": 1498
    },
    {
      "epoch": 0.08081733879663576,
      "grad_norm": 0.8183975219726562,
      "learning_rate": 9.991205620670626e-06,
      "loss": 0.8234,
      "step": 1499
    },
    {
      "epoch": 0.08087125296527928,
      "grad_norm": 1.241355299949646,
      "learning_rate": 9.991193047491399e-06,
      "loss": 0.8135,
      "step": 1500
    },
    {
      "epoch": 0.0809251671339228,
      "grad_norm": 0.9039500951766968,
      "learning_rate": 9.991180465338682e-06,
      "loss": 0.8642,
      "step": 1501
    },
    {
      "epoch": 0.08097908130256631,
      "grad_norm": 1.1762068271636963,
      "learning_rate": 9.991167874212493e-06,
      "loss": 0.7892,
      "step": 1502
    },
    {
      "epoch": 0.08103299547120983,
      "grad_norm": 0.8402833938598633,
      "learning_rate": 9.991155274112857e-06,
      "loss": 0.9054,
      "step": 1503
    },
    {
      "epoch": 0.08108690963985335,
      "grad_norm": 0.9271976351737976,
      "learning_rate": 9.991142665039799e-06,
      "loss": 0.8902,
      "step": 1504
    },
    {
      "epoch": 0.08114082380849687,
      "grad_norm": 0.9105845093727112,
      "learning_rate": 9.991130046993337e-06,
      "loss": 0.8522,
      "step": 1505
    },
    {
      "epoch": 0.0811947379771404,
      "grad_norm": 0.8248290419578552,
      "learning_rate": 9.991117419973499e-06,
      "loss": 0.882,
      "step": 1506
    },
    {
      "epoch": 0.08124865214578392,
      "grad_norm": 1.0726820230484009,
      "learning_rate": 9.991104783980305e-06,
      "loss": 0.8001,
      "step": 1507
    },
    {
      "epoch": 0.08130256631442744,
      "grad_norm": 1.296281337738037,
      "learning_rate": 9.991092139013776e-06,
      "loss": 1.0022,
      "step": 1508
    },
    {
      "epoch": 0.08135648048307095,
      "grad_norm": 1.7287628650665283,
      "learning_rate": 9.991079485073938e-06,
      "loss": 0.914,
      "step": 1509
    },
    {
      "epoch": 0.08141039465171447,
      "grad_norm": 0.8731694221496582,
      "learning_rate": 9.991066822160813e-06,
      "loss": 0.8672,
      "step": 1510
    },
    {
      "epoch": 0.08146430882035799,
      "grad_norm": 0.875747799873352,
      "learning_rate": 9.99105415027442e-06,
      "loss": 0.8044,
      "step": 1511
    },
    {
      "epoch": 0.08151822298900151,
      "grad_norm": 0.9055120348930359,
      "learning_rate": 9.991041469414787e-06,
      "loss": 0.8312,
      "step": 1512
    },
    {
      "epoch": 0.08157213715764502,
      "grad_norm": 0.8849499821662903,
      "learning_rate": 9.991028779581935e-06,
      "loss": 0.889,
      "step": 1513
    },
    {
      "epoch": 0.08162605132628854,
      "grad_norm": 0.9549855589866638,
      "learning_rate": 9.991016080775884e-06,
      "loss": 0.8929,
      "step": 1514
    },
    {
      "epoch": 0.08167996549493207,
      "grad_norm": 0.8395527005195618,
      "learning_rate": 9.991003372996662e-06,
      "loss": 0.6774,
      "step": 1515
    },
    {
      "epoch": 0.08173387966357559,
      "grad_norm": 0.7791672945022583,
      "learning_rate": 9.990990656244287e-06,
      "loss": 0.7178,
      "step": 1516
    },
    {
      "epoch": 0.08178779383221911,
      "grad_norm": 0.91841721534729,
      "learning_rate": 9.990977930518785e-06,
      "loss": 0.8372,
      "step": 1517
    },
    {
      "epoch": 0.08184170800086263,
      "grad_norm": 0.923937976360321,
      "learning_rate": 9.990965195820178e-06,
      "loss": 0.8467,
      "step": 1518
    },
    {
      "epoch": 0.08189562216950615,
      "grad_norm": 0.9804415106773376,
      "learning_rate": 9.990952452148488e-06,
      "loss": 0.9281,
      "step": 1519
    },
    {
      "epoch": 0.08194953633814966,
      "grad_norm": 0.9396255016326904,
      "learning_rate": 9.99093969950374e-06,
      "loss": 0.8606,
      "step": 1520
    },
    {
      "epoch": 0.08200345050679318,
      "grad_norm": 0.8492118120193481,
      "learning_rate": 9.990926937885953e-06,
      "loss": 0.8253,
      "step": 1521
    },
    {
      "epoch": 0.0820573646754367,
      "grad_norm": 0.8482204079627991,
      "learning_rate": 9.990914167295154e-06,
      "loss": 0.7361,
      "step": 1522
    },
    {
      "epoch": 0.08211127884408022,
      "grad_norm": 1.1302778720855713,
      "learning_rate": 9.990901387731365e-06,
      "loss": 0.7511,
      "step": 1523
    },
    {
      "epoch": 0.08216519301272375,
      "grad_norm": 0.9285756945610046,
      "learning_rate": 9.990888599194607e-06,
      "loss": 0.8329,
      "step": 1524
    },
    {
      "epoch": 0.08221910718136727,
      "grad_norm": 0.8932104110717773,
      "learning_rate": 9.990875801684905e-06,
      "loss": 0.8146,
      "step": 1525
    },
    {
      "epoch": 0.08227302135001079,
      "grad_norm": 0.8232647180557251,
      "learning_rate": 9.990862995202282e-06,
      "loss": 0.763,
      "step": 1526
    },
    {
      "epoch": 0.0823269355186543,
      "grad_norm": 0.8582163453102112,
      "learning_rate": 9.990850179746759e-06,
      "loss": 0.7675,
      "step": 1527
    },
    {
      "epoch": 0.08238084968729782,
      "grad_norm": 0.9890977144241333,
      "learning_rate": 9.990837355318362e-06,
      "loss": 0.8438,
      "step": 1528
    },
    {
      "epoch": 0.08243476385594134,
      "grad_norm": 0.9228235483169556,
      "learning_rate": 9.990824521917113e-06,
      "loss": 0.9324,
      "step": 1529
    },
    {
      "epoch": 0.08248867802458486,
      "grad_norm": 0.8286252617835999,
      "learning_rate": 9.990811679543033e-06,
      "loss": 0.872,
      "step": 1530
    },
    {
      "epoch": 0.08254259219322838,
      "grad_norm": 0.8546530604362488,
      "learning_rate": 9.990798828196146e-06,
      "loss": 0.7256,
      "step": 1531
    },
    {
      "epoch": 0.0825965063618719,
      "grad_norm": 0.8240640759468079,
      "learning_rate": 9.990785967876478e-06,
      "loss": 0.8083,
      "step": 1532
    },
    {
      "epoch": 0.08265042053051543,
      "grad_norm": 0.8650565147399902,
      "learning_rate": 9.99077309858405e-06,
      "loss": 0.8274,
      "step": 1533
    },
    {
      "epoch": 0.08270433469915894,
      "grad_norm": 0.7865849137306213,
      "learning_rate": 9.990760220318884e-06,
      "loss": 0.7978,
      "step": 1534
    },
    {
      "epoch": 0.08275824886780246,
      "grad_norm": 0.8567995429039001,
      "learning_rate": 9.990747333081005e-06,
      "loss": 0.8172,
      "step": 1535
    },
    {
      "epoch": 0.08281216303644598,
      "grad_norm": 0.8242521286010742,
      "learning_rate": 9.990734436870435e-06,
      "loss": 0.8045,
      "step": 1536
    },
    {
      "epoch": 0.0828660772050895,
      "grad_norm": 0.801266074180603,
      "learning_rate": 9.990721531687197e-06,
      "loss": 0.8312,
      "step": 1537
    },
    {
      "epoch": 0.08291999137373302,
      "grad_norm": 0.8027862906455994,
      "learning_rate": 9.990708617531314e-06,
      "loss": 0.7227,
      "step": 1538
    },
    {
      "epoch": 0.08297390554237653,
      "grad_norm": 1.0332401990890503,
      "learning_rate": 9.990695694402811e-06,
      "loss": 0.9091,
      "step": 1539
    },
    {
      "epoch": 0.08302781971102005,
      "grad_norm": 0.8537373542785645,
      "learning_rate": 9.99068276230171e-06,
      "loss": 0.7573,
      "step": 1540
    },
    {
      "epoch": 0.08308173387966357,
      "grad_norm": 0.8734087944030762,
      "learning_rate": 9.990669821228037e-06,
      "loss": 0.901,
      "step": 1541
    },
    {
      "epoch": 0.0831356480483071,
      "grad_norm": 0.8546577095985413,
      "learning_rate": 9.99065687118181e-06,
      "loss": 0.8294,
      "step": 1542
    },
    {
      "epoch": 0.08318956221695062,
      "grad_norm": 0.9555438756942749,
      "learning_rate": 9.990643912163055e-06,
      "loss": 0.83,
      "step": 1543
    },
    {
      "epoch": 0.08324347638559414,
      "grad_norm": 0.8778670430183411,
      "learning_rate": 9.990630944171798e-06,
      "loss": 0.8694,
      "step": 1544
    },
    {
      "epoch": 0.08329739055423765,
      "grad_norm": 0.973791241645813,
      "learning_rate": 9.990617967208058e-06,
      "loss": 0.8348,
      "step": 1545
    },
    {
      "epoch": 0.08335130472288117,
      "grad_norm": 0.7933714389801025,
      "learning_rate": 9.990604981271858e-06,
      "loss": 0.8208,
      "step": 1546
    },
    {
      "epoch": 0.08340521889152469,
      "grad_norm": 0.9328469634056091,
      "learning_rate": 9.990591986363226e-06,
      "loss": 0.8188,
      "step": 1547
    },
    {
      "epoch": 0.08345913306016821,
      "grad_norm": 0.8217103481292725,
      "learning_rate": 9.990578982482183e-06,
      "loss": 0.7948,
      "step": 1548
    },
    {
      "epoch": 0.08351304722881173,
      "grad_norm": 0.8556894659996033,
      "learning_rate": 9.990565969628749e-06,
      "loss": 0.8129,
      "step": 1549
    },
    {
      "epoch": 0.08356696139745524,
      "grad_norm": 0.901633083820343,
      "learning_rate": 9.990552947802954e-06,
      "loss": 0.9025,
      "step": 1550
    },
    {
      "epoch": 0.08362087556609878,
      "grad_norm": 0.9021494388580322,
      "learning_rate": 9.990539917004815e-06,
      "loss": 0.8882,
      "step": 1551
    },
    {
      "epoch": 0.0836747897347423,
      "grad_norm": 0.8187722563743591,
      "learning_rate": 9.990526877234359e-06,
      "loss": 0.7385,
      "step": 1552
    },
    {
      "epoch": 0.08372870390338581,
      "grad_norm": 0.9237630367279053,
      "learning_rate": 9.990513828491609e-06,
      "loss": 0.851,
      "step": 1553
    },
    {
      "epoch": 0.08378261807202933,
      "grad_norm": 1.1868582963943481,
      "learning_rate": 9.990500770776589e-06,
      "loss": 0.7701,
      "step": 1554
    },
    {
      "epoch": 0.08383653224067285,
      "grad_norm": 0.9831421971321106,
      "learning_rate": 9.990487704089322e-06,
      "loss": 0.836,
      "step": 1555
    },
    {
      "epoch": 0.08389044640931637,
      "grad_norm": 0.9255663752555847,
      "learning_rate": 9.99047462842983e-06,
      "loss": 0.7916,
      "step": 1556
    },
    {
      "epoch": 0.08394436057795988,
      "grad_norm": 1.0069084167480469,
      "learning_rate": 9.990461543798137e-06,
      "loss": 0.8652,
      "step": 1557
    },
    {
      "epoch": 0.0839982747466034,
      "grad_norm": 0.943044900894165,
      "learning_rate": 9.990448450194267e-06,
      "loss": 0.9511,
      "step": 1558
    },
    {
      "epoch": 0.08405218891524693,
      "grad_norm": 0.9996150135993958,
      "learning_rate": 9.990435347618246e-06,
      "loss": 0.8751,
      "step": 1559
    },
    {
      "epoch": 0.08410610308389045,
      "grad_norm": 0.9531681537628174,
      "learning_rate": 9.990422236070094e-06,
      "loss": 0.8988,
      "step": 1560
    },
    {
      "epoch": 0.08416001725253397,
      "grad_norm": 0.9504678249359131,
      "learning_rate": 9.990409115549837e-06,
      "loss": 0.808,
      "step": 1561
    },
    {
      "epoch": 0.08421393142117749,
      "grad_norm": 0.9796282052993774,
      "learning_rate": 9.990395986057496e-06,
      "loss": 0.778,
      "step": 1562
    },
    {
      "epoch": 0.084267845589821,
      "grad_norm": 0.8871618509292603,
      "learning_rate": 9.990382847593096e-06,
      "loss": 0.8945,
      "step": 1563
    },
    {
      "epoch": 0.08432175975846452,
      "grad_norm": 0.8253110647201538,
      "learning_rate": 9.990369700156662e-06,
      "loss": 0.8206,
      "step": 1564
    },
    {
      "epoch": 0.08437567392710804,
      "grad_norm": 0.8799824118614197,
      "learning_rate": 9.990356543748216e-06,
      "loss": 0.7665,
      "step": 1565
    },
    {
      "epoch": 0.08442958809575156,
      "grad_norm": 0.8275637626647949,
      "learning_rate": 9.990343378367782e-06,
      "loss": 0.8468,
      "step": 1566
    },
    {
      "epoch": 0.08448350226439508,
      "grad_norm": 1.0431691408157349,
      "learning_rate": 9.990330204015382e-06,
      "loss": 0.8539,
      "step": 1567
    },
    {
      "epoch": 0.08453741643303861,
      "grad_norm": 1.298999547958374,
      "learning_rate": 9.990317020691043e-06,
      "loss": 0.8989,
      "step": 1568
    },
    {
      "epoch": 0.08459133060168213,
      "grad_norm": 0.865868866443634,
      "learning_rate": 9.990303828394787e-06,
      "loss": 0.8296,
      "step": 1569
    },
    {
      "epoch": 0.08464524477032564,
      "grad_norm": 0.9162652492523193,
      "learning_rate": 9.990290627126637e-06,
      "loss": 0.8617,
      "step": 1570
    },
    {
      "epoch": 0.08469915893896916,
      "grad_norm": 0.9753283858299255,
      "learning_rate": 9.990277416886618e-06,
      "loss": 0.8082,
      "step": 1571
    },
    {
      "epoch": 0.08475307310761268,
      "grad_norm": 0.9561176300048828,
      "learning_rate": 9.990264197674754e-06,
      "loss": 0.8678,
      "step": 1572
    },
    {
      "epoch": 0.0848069872762562,
      "grad_norm": 0.833341658115387,
      "learning_rate": 9.990250969491067e-06,
      "loss": 0.8164,
      "step": 1573
    },
    {
      "epoch": 0.08486090144489972,
      "grad_norm": 0.9928603172302246,
      "learning_rate": 9.990237732335581e-06,
      "loss": 0.6889,
      "step": 1574
    },
    {
      "epoch": 0.08491481561354323,
      "grad_norm": 1.0163367986679077,
      "learning_rate": 9.990224486208322e-06,
      "loss": 0.8278,
      "step": 1575
    },
    {
      "epoch": 0.08496872978218675,
      "grad_norm": 0.9905970096588135,
      "learning_rate": 9.990211231109312e-06,
      "loss": 0.8094,
      "step": 1576
    },
    {
      "epoch": 0.08502264395083028,
      "grad_norm": 0.9112648963928223,
      "learning_rate": 9.990197967038574e-06,
      "loss": 0.8782,
      "step": 1577
    },
    {
      "epoch": 0.0850765581194738,
      "grad_norm": 1.1176974773406982,
      "learning_rate": 9.990184693996136e-06,
      "loss": 0.8826,
      "step": 1578
    },
    {
      "epoch": 0.08513047228811732,
      "grad_norm": 0.7696222066879272,
      "learning_rate": 9.990171411982016e-06,
      "loss": 0.8025,
      "step": 1579
    },
    {
      "epoch": 0.08518438645676084,
      "grad_norm": 0.9288634061813354,
      "learning_rate": 9.990158120996242e-06,
      "loss": 0.8777,
      "step": 1580
    },
    {
      "epoch": 0.08523830062540436,
      "grad_norm": 0.9235022068023682,
      "learning_rate": 9.990144821038839e-06,
      "loss": 0.9339,
      "step": 1581
    },
    {
      "epoch": 0.08529221479404787,
      "grad_norm": 0.9124205708503723,
      "learning_rate": 9.990131512109826e-06,
      "loss": 0.8368,
      "step": 1582
    },
    {
      "epoch": 0.08534612896269139,
      "grad_norm": 0.8409048914909363,
      "learning_rate": 9.990118194209229e-06,
      "loss": 0.7772,
      "step": 1583
    },
    {
      "epoch": 0.08540004313133491,
      "grad_norm": 0.8279136419296265,
      "learning_rate": 9.990104867337074e-06,
      "loss": 0.738,
      "step": 1584
    },
    {
      "epoch": 0.08545395729997843,
      "grad_norm": 0.8895745873451233,
      "learning_rate": 9.990091531493382e-06,
      "loss": 0.7669,
      "step": 1585
    },
    {
      "epoch": 0.08550787146862196,
      "grad_norm": 0.9280734062194824,
      "learning_rate": 9.99007818667818e-06,
      "loss": 0.9052,
      "step": 1586
    },
    {
      "epoch": 0.08556178563726548,
      "grad_norm": 0.7676610350608826,
      "learning_rate": 9.990064832891491e-06,
      "loss": 0.807,
      "step": 1587
    },
    {
      "epoch": 0.085615699805909,
      "grad_norm": 0.9035676121711731,
      "learning_rate": 9.990051470133337e-06,
      "loss": 0.8848,
      "step": 1588
    },
    {
      "epoch": 0.08566961397455251,
      "grad_norm": 1.0960334539413452,
      "learning_rate": 9.990038098403742e-06,
      "loss": 0.8279,
      "step": 1589
    },
    {
      "epoch": 0.08572352814319603,
      "grad_norm": 0.87922203540802,
      "learning_rate": 9.990024717702736e-06,
      "loss": 0.8325,
      "step": 1590
    },
    {
      "epoch": 0.08577744231183955,
      "grad_norm": 0.922815203666687,
      "learning_rate": 9.990011328030335e-06,
      "loss": 0.881,
      "step": 1591
    },
    {
      "epoch": 0.08583135648048307,
      "grad_norm": 0.9880780577659607,
      "learning_rate": 9.989997929386567e-06,
      "loss": 0.7506,
      "step": 1592
    },
    {
      "epoch": 0.08588527064912659,
      "grad_norm": 0.8827483057975769,
      "learning_rate": 9.989984521771456e-06,
      "loss": 0.8961,
      "step": 1593
    },
    {
      "epoch": 0.0859391848177701,
      "grad_norm": 0.8395072817802429,
      "learning_rate": 9.989971105185026e-06,
      "loss": 0.8564,
      "step": 1594
    },
    {
      "epoch": 0.08599309898641364,
      "grad_norm": 0.8731534481048584,
      "learning_rate": 9.989957679627302e-06,
      "loss": 0.8209,
      "step": 1595
    },
    {
      "epoch": 0.08604701315505715,
      "grad_norm": 0.7969424724578857,
      "learning_rate": 9.989944245098305e-06,
      "loss": 0.8031,
      "step": 1596
    },
    {
      "epoch": 0.08610092732370067,
      "grad_norm": 0.8420547246932983,
      "learning_rate": 9.989930801598062e-06,
      "loss": 0.8027,
      "step": 1597
    },
    {
      "epoch": 0.08615484149234419,
      "grad_norm": 0.7900253534317017,
      "learning_rate": 9.989917349126597e-06,
      "loss": 0.8246,
      "step": 1598
    },
    {
      "epoch": 0.08620875566098771,
      "grad_norm": 0.8860716819763184,
      "learning_rate": 9.989903887683934e-06,
      "loss": 0.7846,
      "step": 1599
    },
    {
      "epoch": 0.08626266982963122,
      "grad_norm": 0.907744288444519,
      "learning_rate": 9.989890417270097e-06,
      "loss": 0.7813,
      "step": 1600
    },
    {
      "epoch": 0.08631658399827474,
      "grad_norm": 0.764076828956604,
      "learning_rate": 9.989876937885108e-06,
      "loss": 0.7953,
      "step": 1601
    },
    {
      "epoch": 0.08637049816691826,
      "grad_norm": 1.0143790245056152,
      "learning_rate": 9.989863449528994e-06,
      "loss": 0.8854,
      "step": 1602
    },
    {
      "epoch": 0.08642441233556178,
      "grad_norm": 0.8605815172195435,
      "learning_rate": 9.989849952201779e-06,
      "loss": 0.9289,
      "step": 1603
    },
    {
      "epoch": 0.08647832650420531,
      "grad_norm": 0.8897641897201538,
      "learning_rate": 9.989836445903487e-06,
      "loss": 0.8659,
      "step": 1604
    },
    {
      "epoch": 0.08653224067284883,
      "grad_norm": 0.8893518447875977,
      "learning_rate": 9.989822930634141e-06,
      "loss": 0.8724,
      "step": 1605
    },
    {
      "epoch": 0.08658615484149235,
      "grad_norm": 0.8152129054069519,
      "learning_rate": 9.989809406393767e-06,
      "loss": 0.8321,
      "step": 1606
    },
    {
      "epoch": 0.08664006901013586,
      "grad_norm": 0.8394732475280762,
      "learning_rate": 9.98979587318239e-06,
      "loss": 0.8074,
      "step": 1607
    },
    {
      "epoch": 0.08669398317877938,
      "grad_norm": 0.8038346767425537,
      "learning_rate": 9.989782331000031e-06,
      "loss": 0.8132,
      "step": 1608
    },
    {
      "epoch": 0.0867478973474229,
      "grad_norm": 0.8574134111404419,
      "learning_rate": 9.989768779846717e-06,
      "loss": 0.8191,
      "step": 1609
    },
    {
      "epoch": 0.08680181151606642,
      "grad_norm": 1.0049889087677002,
      "learning_rate": 9.989755219722472e-06,
      "loss": 0.8771,
      "step": 1610
    },
    {
      "epoch": 0.08685572568470994,
      "grad_norm": 0.9765112996101379,
      "learning_rate": 9.989741650627319e-06,
      "loss": 0.839,
      "step": 1611
    },
    {
      "epoch": 0.08690963985335347,
      "grad_norm": 0.9430082440376282,
      "learning_rate": 9.989728072561284e-06,
      "loss": 1.0316,
      "step": 1612
    },
    {
      "epoch": 0.08696355402199699,
      "grad_norm": 0.841590404510498,
      "learning_rate": 9.989714485524391e-06,
      "loss": 0.8727,
      "step": 1613
    },
    {
      "epoch": 0.0870174681906405,
      "grad_norm": 0.9475975632667542,
      "learning_rate": 9.989700889516664e-06,
      "loss": 0.8131,
      "step": 1614
    },
    {
      "epoch": 0.08707138235928402,
      "grad_norm": 0.8059530258178711,
      "learning_rate": 9.98968728453813e-06,
      "loss": 0.8297,
      "step": 1615
    },
    {
      "epoch": 0.08712529652792754,
      "grad_norm": 0.8513601422309875,
      "learning_rate": 9.989673670588808e-06,
      "loss": 0.8016,
      "step": 1616
    },
    {
      "epoch": 0.08717921069657106,
      "grad_norm": 0.8434658646583557,
      "learning_rate": 9.989660047668728e-06,
      "loss": 0.866,
      "step": 1617
    },
    {
      "epoch": 0.08723312486521458,
      "grad_norm": 0.9081484079360962,
      "learning_rate": 9.989646415777912e-06,
      "loss": 0.816,
      "step": 1618
    },
    {
      "epoch": 0.0872870390338581,
      "grad_norm": 0.7941877841949463,
      "learning_rate": 9.989632774916385e-06,
      "loss": 0.7191,
      "step": 1619
    },
    {
      "epoch": 0.08734095320250161,
      "grad_norm": 0.8800172209739685,
      "learning_rate": 9.98961912508417e-06,
      "loss": 0.8135,
      "step": 1620
    },
    {
      "epoch": 0.08739486737114514,
      "grad_norm": 0.7940575480461121,
      "learning_rate": 9.989605466281292e-06,
      "loss": 0.8124,
      "step": 1621
    },
    {
      "epoch": 0.08744878153978866,
      "grad_norm": 0.9570618271827698,
      "learning_rate": 9.989591798507779e-06,
      "loss": 0.9043,
      "step": 1622
    },
    {
      "epoch": 0.08750269570843218,
      "grad_norm": 0.8635395169258118,
      "learning_rate": 9.98957812176365e-06,
      "loss": 0.835,
      "step": 1623
    },
    {
      "epoch": 0.0875566098770757,
      "grad_norm": 0.8289955258369446,
      "learning_rate": 9.989564436048932e-06,
      "loss": 0.8265,
      "step": 1624
    },
    {
      "epoch": 0.08761052404571922,
      "grad_norm": 0.9519028663635254,
      "learning_rate": 9.989550741363654e-06,
      "loss": 0.8127,
      "step": 1625
    },
    {
      "epoch": 0.08766443821436273,
      "grad_norm": 0.9611422419548035,
      "learning_rate": 9.989537037707834e-06,
      "loss": 0.8422,
      "step": 1626
    },
    {
      "epoch": 0.08771835238300625,
      "grad_norm": 0.8824746608734131,
      "learning_rate": 9.9895233250815e-06,
      "loss": 0.8669,
      "step": 1627
    },
    {
      "epoch": 0.08777226655164977,
      "grad_norm": 0.8402838706970215,
      "learning_rate": 9.989509603484676e-06,
      "loss": 0.8072,
      "step": 1628
    },
    {
      "epoch": 0.08782618072029329,
      "grad_norm": 0.7537099719047546,
      "learning_rate": 9.989495872917386e-06,
      "loss": 0.7127,
      "step": 1629
    },
    {
      "epoch": 0.08788009488893682,
      "grad_norm": 0.78285151720047,
      "learning_rate": 9.989482133379656e-06,
      "loss": 0.819,
      "step": 1630
    },
    {
      "epoch": 0.08793400905758034,
      "grad_norm": 0.9339445233345032,
      "learning_rate": 9.98946838487151e-06,
      "loss": 0.8694,
      "step": 1631
    },
    {
      "epoch": 0.08798792322622385,
      "grad_norm": 0.8022040128707886,
      "learning_rate": 9.989454627392973e-06,
      "loss": 0.7601,
      "step": 1632
    },
    {
      "epoch": 0.08804183739486737,
      "grad_norm": 0.8593827486038208,
      "learning_rate": 9.98944086094407e-06,
      "loss": 0.8536,
      "step": 1633
    },
    {
      "epoch": 0.08809575156351089,
      "grad_norm": 0.8415039777755737,
      "learning_rate": 9.989427085524824e-06,
      "loss": 0.9027,
      "step": 1634
    },
    {
      "epoch": 0.08814966573215441,
      "grad_norm": 0.9551103711128235,
      "learning_rate": 9.989413301135263e-06,
      "loss": 0.8063,
      "step": 1635
    },
    {
      "epoch": 0.08820357990079793,
      "grad_norm": 0.8554351925849915,
      "learning_rate": 9.989399507775407e-06,
      "loss": 0.7694,
      "step": 1636
    },
    {
      "epoch": 0.08825749406944144,
      "grad_norm": 0.8688547015190125,
      "learning_rate": 9.989385705445285e-06,
      "loss": 0.8862,
      "step": 1637
    },
    {
      "epoch": 0.08831140823808496,
      "grad_norm": 0.816558837890625,
      "learning_rate": 9.98937189414492e-06,
      "loss": 0.7302,
      "step": 1638
    },
    {
      "epoch": 0.0883653224067285,
      "grad_norm": 0.8164445757865906,
      "learning_rate": 9.989358073874337e-06,
      "loss": 0.8724,
      "step": 1639
    },
    {
      "epoch": 0.08841923657537201,
      "grad_norm": 0.8909460306167603,
      "learning_rate": 9.989344244633564e-06,
      "loss": 0.7618,
      "step": 1640
    },
    {
      "epoch": 0.08847315074401553,
      "grad_norm": 1.0117470026016235,
      "learning_rate": 9.98933040642262e-06,
      "loss": 0.8191,
      "step": 1641
    },
    {
      "epoch": 0.08852706491265905,
      "grad_norm": 0.8317937850952148,
      "learning_rate": 9.989316559241533e-06,
      "loss": 0.8339,
      "step": 1642
    },
    {
      "epoch": 0.08858097908130257,
      "grad_norm": 0.7955135107040405,
      "learning_rate": 9.98930270309033e-06,
      "loss": 0.7799,
      "step": 1643
    },
    {
      "epoch": 0.08863489324994608,
      "grad_norm": 0.996306300163269,
      "learning_rate": 9.98928883796903e-06,
      "loss": 0.8547,
      "step": 1644
    },
    {
      "epoch": 0.0886888074185896,
      "grad_norm": 0.9679511189460754,
      "learning_rate": 9.989274963877664e-06,
      "loss": 1.0831,
      "step": 1645
    },
    {
      "epoch": 0.08874272158723312,
      "grad_norm": 0.8471615314483643,
      "learning_rate": 9.989261080816253e-06,
      "loss": 0.7765,
      "step": 1646
    },
    {
      "epoch": 0.08879663575587664,
      "grad_norm": 0.8662555813789368,
      "learning_rate": 9.989247188784826e-06,
      "loss": 0.8894,
      "step": 1647
    },
    {
      "epoch": 0.08885054992452017,
      "grad_norm": 0.9549373388290405,
      "learning_rate": 9.989233287783402e-06,
      "loss": 0.8341,
      "step": 1648
    },
    {
      "epoch": 0.08890446409316369,
      "grad_norm": 0.8179014325141907,
      "learning_rate": 9.989219377812014e-06,
      "loss": 0.8653,
      "step": 1649
    },
    {
      "epoch": 0.0889583782618072,
      "grad_norm": 0.9237802624702454,
      "learning_rate": 9.989205458870678e-06,
      "loss": 0.8206,
      "step": 1650
    },
    {
      "epoch": 0.08901229243045072,
      "grad_norm": 0.940217137336731,
      "learning_rate": 9.989191530959426e-06,
      "loss": 0.8695,
      "step": 1651
    },
    {
      "epoch": 0.08906620659909424,
      "grad_norm": 0.9200409054756165,
      "learning_rate": 9.98917759407828e-06,
      "loss": 0.7984,
      "step": 1652
    },
    {
      "epoch": 0.08912012076773776,
      "grad_norm": 0.9270562529563904,
      "learning_rate": 9.989163648227265e-06,
      "loss": 0.8265,
      "step": 1653
    },
    {
      "epoch": 0.08917403493638128,
      "grad_norm": 0.9945223331451416,
      "learning_rate": 9.989149693406408e-06,
      "loss": 0.84,
      "step": 1654
    },
    {
      "epoch": 0.0892279491050248,
      "grad_norm": 0.826195478439331,
      "learning_rate": 9.98913572961573e-06,
      "loss": 0.7862,
      "step": 1655
    },
    {
      "epoch": 0.08928186327366831,
      "grad_norm": 0.9132022857666016,
      "learning_rate": 9.989121756855263e-06,
      "loss": 0.826,
      "step": 1656
    },
    {
      "epoch": 0.08933577744231185,
      "grad_norm": 0.8559401631355286,
      "learning_rate": 9.989107775125023e-06,
      "loss": 0.8007,
      "step": 1657
    },
    {
      "epoch": 0.08938969161095536,
      "grad_norm": 0.8000867366790771,
      "learning_rate": 9.989093784425044e-06,
      "loss": 0.7547,
      "step": 1658
    },
    {
      "epoch": 0.08944360577959888,
      "grad_norm": 0.7761433720588684,
      "learning_rate": 9.989079784755346e-06,
      "loss": 0.8083,
      "step": 1659
    },
    {
      "epoch": 0.0894975199482424,
      "grad_norm": 0.8072230815887451,
      "learning_rate": 9.989065776115956e-06,
      "loss": 0.892,
      "step": 1660
    },
    {
      "epoch": 0.08955143411688592,
      "grad_norm": 0.9021360874176025,
      "learning_rate": 9.989051758506898e-06,
      "loss": 0.8715,
      "step": 1661
    },
    {
      "epoch": 0.08960534828552943,
      "grad_norm": 0.7585147023200989,
      "learning_rate": 9.989037731928197e-06,
      "loss": 0.7115,
      "step": 1662
    },
    {
      "epoch": 0.08965926245417295,
      "grad_norm": 0.9388399124145508,
      "learning_rate": 9.98902369637988e-06,
      "loss": 0.8976,
      "step": 1663
    },
    {
      "epoch": 0.08971317662281647,
      "grad_norm": 0.8454418778419495,
      "learning_rate": 9.989009651861972e-06,
      "loss": 0.8063,
      "step": 1664
    },
    {
      "epoch": 0.08976709079146,
      "grad_norm": 0.82308030128479,
      "learning_rate": 9.988995598374496e-06,
      "loss": 0.8044,
      "step": 1665
    },
    {
      "epoch": 0.08982100496010352,
      "grad_norm": 1.006800651550293,
      "learning_rate": 9.98898153591748e-06,
      "loss": 0.8609,
      "step": 1666
    },
    {
      "epoch": 0.08987491912874704,
      "grad_norm": 0.8325724601745605,
      "learning_rate": 9.988967464490947e-06,
      "loss": 0.8295,
      "step": 1667
    },
    {
      "epoch": 0.08992883329739056,
      "grad_norm": 0.7575547695159912,
      "learning_rate": 9.988953384094923e-06,
      "loss": 0.8252,
      "step": 1668
    },
    {
      "epoch": 0.08998274746603407,
      "grad_norm": 0.869877278804779,
      "learning_rate": 9.988939294729436e-06,
      "loss": 0.8304,
      "step": 1669
    },
    {
      "epoch": 0.09003666163467759,
      "grad_norm": 0.7840037941932678,
      "learning_rate": 9.988925196394508e-06,
      "loss": 0.7742,
      "step": 1670
    },
    {
      "epoch": 0.09009057580332111,
      "grad_norm": 0.8044409155845642,
      "learning_rate": 9.988911089090163e-06,
      "loss": 0.8371,
      "step": 1671
    },
    {
      "epoch": 0.09014448997196463,
      "grad_norm": 0.8635613322257996,
      "learning_rate": 9.988896972816431e-06,
      "loss": 0.7693,
      "step": 1672
    },
    {
      "epoch": 0.09019840414060815,
      "grad_norm": 0.7780656814575195,
      "learning_rate": 9.988882847573335e-06,
      "loss": 0.841,
      "step": 1673
    },
    {
      "epoch": 0.09025231830925168,
      "grad_norm": 0.8938048481941223,
      "learning_rate": 9.9888687133609e-06,
      "loss": 0.8149,
      "step": 1674
    },
    {
      "epoch": 0.0903062324778952,
      "grad_norm": 0.8432002663612366,
      "learning_rate": 9.988854570179152e-06,
      "loss": 0.853,
      "step": 1675
    },
    {
      "epoch": 0.09036014664653871,
      "grad_norm": 0.8222450613975525,
      "learning_rate": 9.988840418028118e-06,
      "loss": 0.897,
      "step": 1676
    },
    {
      "epoch": 0.09041406081518223,
      "grad_norm": 0.8370371460914612,
      "learning_rate": 9.98882625690782e-06,
      "loss": 0.8288,
      "step": 1677
    },
    {
      "epoch": 0.09046797498382575,
      "grad_norm": 0.8510713577270508,
      "learning_rate": 9.988812086818285e-06,
      "loss": 0.7637,
      "step": 1678
    },
    {
      "epoch": 0.09052188915246927,
      "grad_norm": 0.8271141648292542,
      "learning_rate": 9.98879790775954e-06,
      "loss": 0.853,
      "step": 1679
    },
    {
      "epoch": 0.09057580332111279,
      "grad_norm": 1.0627025365829468,
      "learning_rate": 9.988783719731607e-06,
      "loss": 0.7569,
      "step": 1680
    },
    {
      "epoch": 0.0906297174897563,
      "grad_norm": 0.880283534526825,
      "learning_rate": 9.988769522734517e-06,
      "loss": 0.8362,
      "step": 1681
    },
    {
      "epoch": 0.09068363165839982,
      "grad_norm": 0.8721734881401062,
      "learning_rate": 9.988755316768288e-06,
      "loss": 0.8585,
      "step": 1682
    },
    {
      "epoch": 0.09073754582704335,
      "grad_norm": 0.8830682039260864,
      "learning_rate": 9.988741101832952e-06,
      "loss": 0.8853,
      "step": 1683
    },
    {
      "epoch": 0.09079145999568687,
      "grad_norm": 0.7676220536231995,
      "learning_rate": 9.988726877928534e-06,
      "loss": 0.7832,
      "step": 1684
    },
    {
      "epoch": 0.09084537416433039,
      "grad_norm": 0.866149365901947,
      "learning_rate": 9.988712645055055e-06,
      "loss": 0.8534,
      "step": 1685
    },
    {
      "epoch": 0.09089928833297391,
      "grad_norm": 0.8467028141021729,
      "learning_rate": 9.988698403212546e-06,
      "loss": 0.8637,
      "step": 1686
    },
    {
      "epoch": 0.09095320250161743,
      "grad_norm": 0.913436770439148,
      "learning_rate": 9.988684152401028e-06,
      "loss": 0.855,
      "step": 1687
    },
    {
      "epoch": 0.09100711667026094,
      "grad_norm": 0.8307977914810181,
      "learning_rate": 9.98866989262053e-06,
      "loss": 0.8538,
      "step": 1688
    },
    {
      "epoch": 0.09106103083890446,
      "grad_norm": 1.13442862033844,
      "learning_rate": 9.988655623871075e-06,
      "loss": 0.8129,
      "step": 1689
    },
    {
      "epoch": 0.09111494500754798,
      "grad_norm": 0.8950080871582031,
      "learning_rate": 9.988641346152692e-06,
      "loss": 0.8674,
      "step": 1690
    },
    {
      "epoch": 0.0911688591761915,
      "grad_norm": 0.9107043147087097,
      "learning_rate": 9.988627059465403e-06,
      "loss": 0.9507,
      "step": 1691
    },
    {
      "epoch": 0.09122277334483503,
      "grad_norm": 0.8210874795913696,
      "learning_rate": 9.988612763809237e-06,
      "loss": 0.8913,
      "step": 1692
    },
    {
      "epoch": 0.09127668751347855,
      "grad_norm": 1.0306476354599,
      "learning_rate": 9.988598459184217e-06,
      "loss": 0.8589,
      "step": 1693
    },
    {
      "epoch": 0.09133060168212206,
      "grad_norm": 0.7582615613937378,
      "learning_rate": 9.98858414559037e-06,
      "loss": 0.7482,
      "step": 1694
    },
    {
      "epoch": 0.09138451585076558,
      "grad_norm": 0.8572216629981995,
      "learning_rate": 9.98856982302772e-06,
      "loss": 0.822,
      "step": 1695
    },
    {
      "epoch": 0.0914384300194091,
      "grad_norm": 0.9358139038085938,
      "learning_rate": 9.988555491496297e-06,
      "loss": 0.8298,
      "step": 1696
    },
    {
      "epoch": 0.09149234418805262,
      "grad_norm": 0.8705672025680542,
      "learning_rate": 9.988541150996123e-06,
      "loss": 0.8818,
      "step": 1697
    },
    {
      "epoch": 0.09154625835669614,
      "grad_norm": 0.9081273674964905,
      "learning_rate": 9.988526801527224e-06,
      "loss": 0.8994,
      "step": 1698
    },
    {
      "epoch": 0.09160017252533965,
      "grad_norm": 0.7358905076980591,
      "learning_rate": 9.988512443089627e-06,
      "loss": 0.7752,
      "step": 1699
    },
    {
      "epoch": 0.09165408669398317,
      "grad_norm": 0.8570963740348816,
      "learning_rate": 9.988498075683357e-06,
      "loss": 0.908,
      "step": 1700
    },
    {
      "epoch": 0.0917080008626267,
      "grad_norm": 0.8998208045959473,
      "learning_rate": 9.988483699308442e-06,
      "loss": 0.8561,
      "step": 1701
    },
    {
      "epoch": 0.09176191503127022,
      "grad_norm": 0.7481779456138611,
      "learning_rate": 9.988469313964903e-06,
      "loss": 0.7184,
      "step": 1702
    },
    {
      "epoch": 0.09181582919991374,
      "grad_norm": 1.052809238433838,
      "learning_rate": 9.988454919652772e-06,
      "loss": 0.8579,
      "step": 1703
    },
    {
      "epoch": 0.09186974336855726,
      "grad_norm": 0.8492130637168884,
      "learning_rate": 9.988440516372071e-06,
      "loss": 0.8796,
      "step": 1704
    },
    {
      "epoch": 0.09192365753720078,
      "grad_norm": 0.884483277797699,
      "learning_rate": 9.988426104122826e-06,
      "loss": 0.8781,
      "step": 1705
    },
    {
      "epoch": 0.0919775717058443,
      "grad_norm": 0.8844857811927795,
      "learning_rate": 9.988411682905065e-06,
      "loss": 0.8981,
      "step": 1706
    },
    {
      "epoch": 0.09203148587448781,
      "grad_norm": 0.906216025352478,
      "learning_rate": 9.988397252718811e-06,
      "loss": 0.8741,
      "step": 1707
    },
    {
      "epoch": 0.09208540004313133,
      "grad_norm": 0.8565787076950073,
      "learning_rate": 9.988382813564092e-06,
      "loss": 0.7358,
      "step": 1708
    },
    {
      "epoch": 0.09213931421177485,
      "grad_norm": 0.8036391139030457,
      "learning_rate": 9.988368365440935e-06,
      "loss": 0.7966,
      "step": 1709
    },
    {
      "epoch": 0.09219322838041838,
      "grad_norm": 1.1708556413650513,
      "learning_rate": 9.988353908349361e-06,
      "loss": 0.8385,
      "step": 1710
    },
    {
      "epoch": 0.0922471425490619,
      "grad_norm": 0.8536746501922607,
      "learning_rate": 9.988339442289403e-06,
      "loss": 0.7387,
      "step": 1711
    },
    {
      "epoch": 0.09230105671770542,
      "grad_norm": 0.8376518487930298,
      "learning_rate": 9.988324967261083e-06,
      "loss": 0.8537,
      "step": 1712
    },
    {
      "epoch": 0.09235497088634893,
      "grad_norm": 0.8793227672576904,
      "learning_rate": 9.988310483264426e-06,
      "loss": 0.8028,
      "step": 1713
    },
    {
      "epoch": 0.09240888505499245,
      "grad_norm": 0.8186830282211304,
      "learning_rate": 9.98829599029946e-06,
      "loss": 0.8478,
      "step": 1714
    },
    {
      "epoch": 0.09246279922363597,
      "grad_norm": 0.8845428824424744,
      "learning_rate": 9.98828148836621e-06,
      "loss": 0.8524,
      "step": 1715
    },
    {
      "epoch": 0.09251671339227949,
      "grad_norm": 1.0494492053985596,
      "learning_rate": 9.988266977464704e-06,
      "loss": 0.8542,
      "step": 1716
    },
    {
      "epoch": 0.092570627560923,
      "grad_norm": 0.8876493573188782,
      "learning_rate": 9.988252457594966e-06,
      "loss": 0.8989,
      "step": 1717
    },
    {
      "epoch": 0.09262454172956654,
      "grad_norm": 0.8787088394165039,
      "learning_rate": 9.988237928757024e-06,
      "loss": 0.8214,
      "step": 1718
    },
    {
      "epoch": 0.09267845589821005,
      "grad_norm": 1.069684624671936,
      "learning_rate": 9.988223390950901e-06,
      "loss": 0.9714,
      "step": 1719
    },
    {
      "epoch": 0.09273237006685357,
      "grad_norm": 0.7957501411437988,
      "learning_rate": 9.988208844176626e-06,
      "loss": 0.7562,
      "step": 1720
    },
    {
      "epoch": 0.09278628423549709,
      "grad_norm": 0.8354908227920532,
      "learning_rate": 9.988194288434225e-06,
      "loss": 0.7494,
      "step": 1721
    },
    {
      "epoch": 0.09284019840414061,
      "grad_norm": 0.8205936551094055,
      "learning_rate": 9.988179723723722e-06,
      "loss": 0.7727,
      "step": 1722
    },
    {
      "epoch": 0.09289411257278413,
      "grad_norm": 0.8364951014518738,
      "learning_rate": 9.988165150045146e-06,
      "loss": 0.861,
      "step": 1723
    },
    {
      "epoch": 0.09294802674142764,
      "grad_norm": 0.8664119243621826,
      "learning_rate": 9.98815056739852e-06,
      "loss": 0.8512,
      "step": 1724
    },
    {
      "epoch": 0.09300194091007116,
      "grad_norm": 0.9565482139587402,
      "learning_rate": 9.988135975783874e-06,
      "loss": 0.8606,
      "step": 1725
    },
    {
      "epoch": 0.09305585507871468,
      "grad_norm": 0.8696085214614868,
      "learning_rate": 9.988121375201232e-06,
      "loss": 0.8614,
      "step": 1726
    },
    {
      "epoch": 0.09310976924735821,
      "grad_norm": 0.8623467683792114,
      "learning_rate": 9.98810676565062e-06,
      "loss": 0.8547,
      "step": 1727
    },
    {
      "epoch": 0.09316368341600173,
      "grad_norm": 0.8284831047058105,
      "learning_rate": 9.988092147132064e-06,
      "loss": 0.8376,
      "step": 1728
    },
    {
      "epoch": 0.09321759758464525,
      "grad_norm": 0.7768245339393616,
      "learning_rate": 9.988077519645591e-06,
      "loss": 0.7472,
      "step": 1729
    },
    {
      "epoch": 0.09327151175328877,
      "grad_norm": 1.221225619316101,
      "learning_rate": 9.988062883191228e-06,
      "loss": 0.9052,
      "step": 1730
    },
    {
      "epoch": 0.09332542592193228,
      "grad_norm": 1.0027954578399658,
      "learning_rate": 9.988048237769002e-06,
      "loss": 0.9411,
      "step": 1731
    },
    {
      "epoch": 0.0933793400905758,
      "grad_norm": 0.8029824495315552,
      "learning_rate": 9.988033583378937e-06,
      "loss": 0.8141,
      "step": 1732
    },
    {
      "epoch": 0.09343325425921932,
      "grad_norm": 0.8081389665603638,
      "learning_rate": 9.98801892002106e-06,
      "loss": 0.7977,
      "step": 1733
    },
    {
      "epoch": 0.09348716842786284,
      "grad_norm": 0.887438952922821,
      "learning_rate": 9.988004247695398e-06,
      "loss": 0.8574,
      "step": 1734
    },
    {
      "epoch": 0.09354108259650636,
      "grad_norm": 0.887238085269928,
      "learning_rate": 9.987989566401977e-06,
      "loss": 0.9041,
      "step": 1735
    },
    {
      "epoch": 0.09359499676514989,
      "grad_norm": 0.9135997891426086,
      "learning_rate": 9.987974876140822e-06,
      "loss": 0.738,
      "step": 1736
    },
    {
      "epoch": 0.0936489109337934,
      "grad_norm": 0.7749861478805542,
      "learning_rate": 9.987960176911964e-06,
      "loss": 0.773,
      "step": 1737
    },
    {
      "epoch": 0.09370282510243692,
      "grad_norm": 0.7850096225738525,
      "learning_rate": 9.987945468715425e-06,
      "loss": 0.7924,
      "step": 1738
    },
    {
      "epoch": 0.09375673927108044,
      "grad_norm": 0.8044145107269287,
      "learning_rate": 9.987930751551231e-06,
      "loss": 0.8196,
      "step": 1739
    },
    {
      "epoch": 0.09381065343972396,
      "grad_norm": 0.8781464695930481,
      "learning_rate": 9.987916025419413e-06,
      "loss": 0.9337,
      "step": 1740
    },
    {
      "epoch": 0.09386456760836748,
      "grad_norm": 1.0839952230453491,
      "learning_rate": 9.987901290319993e-06,
      "loss": 0.8092,
      "step": 1741
    },
    {
      "epoch": 0.093918481777011,
      "grad_norm": 0.7910736203193665,
      "learning_rate": 9.987886546253e-06,
      "loss": 0.8775,
      "step": 1742
    },
    {
      "epoch": 0.09397239594565451,
      "grad_norm": 0.887287974357605,
      "learning_rate": 9.98787179321846e-06,
      "loss": 0.8271,
      "step": 1743
    },
    {
      "epoch": 0.09402631011429803,
      "grad_norm": 1.1318427324295044,
      "learning_rate": 9.987857031216397e-06,
      "loss": 0.8328,
      "step": 1744
    },
    {
      "epoch": 0.09408022428294156,
      "grad_norm": 0.8660401105880737,
      "learning_rate": 9.987842260246842e-06,
      "loss": 0.8647,
      "step": 1745
    },
    {
      "epoch": 0.09413413845158508,
      "grad_norm": 0.9396790266036987,
      "learning_rate": 9.98782748030982e-06,
      "loss": 0.9373,
      "step": 1746
    },
    {
      "epoch": 0.0941880526202286,
      "grad_norm": 0.8715323209762573,
      "learning_rate": 9.987812691405353e-06,
      "loss": 0.8621,
      "step": 1747
    },
    {
      "epoch": 0.09424196678887212,
      "grad_norm": 0.7882347106933594,
      "learning_rate": 9.987797893533475e-06,
      "loss": 0.7283,
      "step": 1748
    },
    {
      "epoch": 0.09429588095751563,
      "grad_norm": 0.9641733765602112,
      "learning_rate": 9.987783086694208e-06,
      "loss": 0.8038,
      "step": 1749
    },
    {
      "epoch": 0.09434979512615915,
      "grad_norm": 0.8808518648147583,
      "learning_rate": 9.98776827088758e-06,
      "loss": 0.8072,
      "step": 1750
    },
    {
      "epoch": 0.09440370929480267,
      "grad_norm": 0.7720713019371033,
      "learning_rate": 9.987753446113618e-06,
      "loss": 0.7786,
      "step": 1751
    },
    {
      "epoch": 0.09445762346344619,
      "grad_norm": 1.0507936477661133,
      "learning_rate": 9.987738612372346e-06,
      "loss": 0.9302,
      "step": 1752
    },
    {
      "epoch": 0.0945115376320897,
      "grad_norm": 0.7705017328262329,
      "learning_rate": 9.987723769663795e-06,
      "loss": 0.7366,
      "step": 1753
    },
    {
      "epoch": 0.09456545180073324,
      "grad_norm": 0.82464200258255,
      "learning_rate": 9.987708917987989e-06,
      "loss": 0.8063,
      "step": 1754
    },
    {
      "epoch": 0.09461936596937676,
      "grad_norm": 0.9387272000312805,
      "learning_rate": 9.987694057344953e-06,
      "loss": 0.8108,
      "step": 1755
    },
    {
      "epoch": 0.09467328013802027,
      "grad_norm": 0.9161933064460754,
      "learning_rate": 9.987679187734717e-06,
      "loss": 0.8331,
      "step": 1756
    },
    {
      "epoch": 0.09472719430666379,
      "grad_norm": 0.9379769563674927,
      "learning_rate": 9.987664309157306e-06,
      "loss": 0.9064,
      "step": 1757
    },
    {
      "epoch": 0.09478110847530731,
      "grad_norm": 0.9597976803779602,
      "learning_rate": 9.987649421612748e-06,
      "loss": 0.7785,
      "step": 1758
    },
    {
      "epoch": 0.09483502264395083,
      "grad_norm": 0.8689720630645752,
      "learning_rate": 9.98763452510107e-06,
      "loss": 0.7828,
      "step": 1759
    },
    {
      "epoch": 0.09488893681259435,
      "grad_norm": 0.9207726716995239,
      "learning_rate": 9.987619619622296e-06,
      "loss": 0.7853,
      "step": 1760
    },
    {
      "epoch": 0.09494285098123786,
      "grad_norm": 0.8130320310592651,
      "learning_rate": 9.987604705176455e-06,
      "loss": 0.858,
      "step": 1761
    },
    {
      "epoch": 0.09499676514988138,
      "grad_norm": 0.9004638195037842,
      "learning_rate": 9.987589781763574e-06,
      "loss": 0.8148,
      "step": 1762
    },
    {
      "epoch": 0.09505067931852491,
      "grad_norm": 0.8554181456565857,
      "learning_rate": 9.987574849383678e-06,
      "loss": 0.8103,
      "step": 1763
    },
    {
      "epoch": 0.09510459348716843,
      "grad_norm": 0.9148527979850769,
      "learning_rate": 9.987559908036797e-06,
      "loss": 0.9467,
      "step": 1764
    },
    {
      "epoch": 0.09515850765581195,
      "grad_norm": 0.890083909034729,
      "learning_rate": 9.987544957722956e-06,
      "loss": 0.8338,
      "step": 1765
    },
    {
      "epoch": 0.09521242182445547,
      "grad_norm": 0.8118012547492981,
      "learning_rate": 9.98752999844218e-06,
      "loss": 0.8355,
      "step": 1766
    },
    {
      "epoch": 0.09526633599309899,
      "grad_norm": 0.8115151524543762,
      "learning_rate": 9.987515030194498e-06,
      "loss": 0.9172,
      "step": 1767
    },
    {
      "epoch": 0.0953202501617425,
      "grad_norm": 0.8750082850456238,
      "learning_rate": 9.987500052979938e-06,
      "loss": 0.8301,
      "step": 1768
    },
    {
      "epoch": 0.09537416433038602,
      "grad_norm": 0.9008756875991821,
      "learning_rate": 9.987485066798525e-06,
      "loss": 0.8642,
      "step": 1769
    },
    {
      "epoch": 0.09542807849902954,
      "grad_norm": 0.8335922956466675,
      "learning_rate": 9.987470071650287e-06,
      "loss": 0.8466,
      "step": 1770
    },
    {
      "epoch": 0.09548199266767307,
      "grad_norm": 0.8604272603988647,
      "learning_rate": 9.987455067535249e-06,
      "loss": 0.8801,
      "step": 1771
    },
    {
      "epoch": 0.09553590683631659,
      "grad_norm": 0.889854371547699,
      "learning_rate": 9.98744005445344e-06,
      "loss": 0.8804,
      "step": 1772
    },
    {
      "epoch": 0.09558982100496011,
      "grad_norm": 0.8756876587867737,
      "learning_rate": 9.987425032404887e-06,
      "loss": 0.8367,
      "step": 1773
    },
    {
      "epoch": 0.09564373517360363,
      "grad_norm": 0.9071298837661743,
      "learning_rate": 9.987410001389616e-06,
      "loss": 0.8875,
      "step": 1774
    },
    {
      "epoch": 0.09569764934224714,
      "grad_norm": 0.8214284777641296,
      "learning_rate": 9.987394961407654e-06,
      "loss": 0.7859,
      "step": 1775
    },
    {
      "epoch": 0.09575156351089066,
      "grad_norm": 0.940034806728363,
      "learning_rate": 9.98737991245903e-06,
      "loss": 0.8272,
      "step": 1776
    },
    {
      "epoch": 0.09580547767953418,
      "grad_norm": 0.8156501054763794,
      "learning_rate": 9.987364854543768e-06,
      "loss": 0.7831,
      "step": 1777
    },
    {
      "epoch": 0.0958593918481777,
      "grad_norm": 0.8450450301170349,
      "learning_rate": 9.987349787661898e-06,
      "loss": 0.7888,
      "step": 1778
    },
    {
      "epoch": 0.09591330601682121,
      "grad_norm": 0.8143148422241211,
      "learning_rate": 9.987334711813446e-06,
      "loss": 0.7593,
      "step": 1779
    },
    {
      "epoch": 0.09596722018546475,
      "grad_norm": 1.0489457845687866,
      "learning_rate": 9.987319626998437e-06,
      "loss": 0.8248,
      "step": 1780
    },
    {
      "epoch": 0.09602113435410826,
      "grad_norm": 0.9584689140319824,
      "learning_rate": 9.987304533216901e-06,
      "loss": 0.9025,
      "step": 1781
    },
    {
      "epoch": 0.09607504852275178,
      "grad_norm": 0.8366501331329346,
      "learning_rate": 9.987289430468862e-06,
      "loss": 0.7513,
      "step": 1782
    },
    {
      "epoch": 0.0961289626913953,
      "grad_norm": 0.9896461963653564,
      "learning_rate": 9.987274318754352e-06,
      "loss": 0.8598,
      "step": 1783
    },
    {
      "epoch": 0.09618287686003882,
      "grad_norm": 1.1904568672180176,
      "learning_rate": 9.987259198073396e-06,
      "loss": 0.9143,
      "step": 1784
    },
    {
      "epoch": 0.09623679102868234,
      "grad_norm": 0.8100086450576782,
      "learning_rate": 9.987244068426019e-06,
      "loss": 0.7733,
      "step": 1785
    },
    {
      "epoch": 0.09629070519732585,
      "grad_norm": 0.7814387083053589,
      "learning_rate": 9.987228929812249e-06,
      "loss": 0.7735,
      "step": 1786
    },
    {
      "epoch": 0.09634461936596937,
      "grad_norm": 0.8880924582481384,
      "learning_rate": 9.987213782232115e-06,
      "loss": 0.8377,
      "step": 1787
    },
    {
      "epoch": 0.09639853353461289,
      "grad_norm": 0.8739203810691833,
      "learning_rate": 9.987198625685643e-06,
      "loss": 0.8851,
      "step": 1788
    },
    {
      "epoch": 0.09645244770325642,
      "grad_norm": 0.8984062671661377,
      "learning_rate": 9.987183460172861e-06,
      "loss": 0.8773,
      "step": 1789
    },
    {
      "epoch": 0.09650636187189994,
      "grad_norm": 1.2485296726226807,
      "learning_rate": 9.987168285693795e-06,
      "loss": 0.787,
      "step": 1790
    },
    {
      "epoch": 0.09656027604054346,
      "grad_norm": 0.8414161205291748,
      "learning_rate": 9.987153102248474e-06,
      "loss": 0.7895,
      "step": 1791
    },
    {
      "epoch": 0.09661419020918698,
      "grad_norm": 0.7895180583000183,
      "learning_rate": 9.987137909836924e-06,
      "loss": 0.7592,
      "step": 1792
    },
    {
      "epoch": 0.0966681043778305,
      "grad_norm": 1.0752787590026855,
      "learning_rate": 9.987122708459173e-06,
      "loss": 0.8472,
      "step": 1793
    },
    {
      "epoch": 0.09672201854647401,
      "grad_norm": 0.9069424271583557,
      "learning_rate": 9.987107498115247e-06,
      "loss": 0.8746,
      "step": 1794
    },
    {
      "epoch": 0.09677593271511753,
      "grad_norm": 0.8566716909408569,
      "learning_rate": 9.987092278805175e-06,
      "loss": 0.7604,
      "step": 1795
    },
    {
      "epoch": 0.09682984688376105,
      "grad_norm": 0.833852231502533,
      "learning_rate": 9.987077050528983e-06,
      "loss": 0.8645,
      "step": 1796
    },
    {
      "epoch": 0.09688376105240457,
      "grad_norm": 0.8439596891403198,
      "learning_rate": 9.9870618132867e-06,
      "loss": 0.7673,
      "step": 1797
    },
    {
      "epoch": 0.0969376752210481,
      "grad_norm": 0.9743669629096985,
      "learning_rate": 9.987046567078352e-06,
      "loss": 0.7754,
      "step": 1798
    },
    {
      "epoch": 0.09699158938969162,
      "grad_norm": 0.9291634559631348,
      "learning_rate": 9.987031311903968e-06,
      "loss": 0.8431,
      "step": 1799
    },
    {
      "epoch": 0.09704550355833513,
      "grad_norm": 1.169450283050537,
      "learning_rate": 9.987016047763571e-06,
      "loss": 0.9321,
      "step": 1800
    },
    {
      "epoch": 0.09709941772697865,
      "grad_norm": 0.7758163809776306,
      "learning_rate": 9.987000774657195e-06,
      "loss": 0.7832,
      "step": 1801
    },
    {
      "epoch": 0.09715333189562217,
      "grad_norm": 0.9673672914505005,
      "learning_rate": 9.986985492584863e-06,
      "loss": 0.9822,
      "step": 1802
    },
    {
      "epoch": 0.09720724606426569,
      "grad_norm": 1.1516417264938354,
      "learning_rate": 9.986970201546605e-06,
      "loss": 0.9956,
      "step": 1803
    },
    {
      "epoch": 0.0972611602329092,
      "grad_norm": 0.9660587906837463,
      "learning_rate": 9.986954901542445e-06,
      "loss": 0.8248,
      "step": 1804
    },
    {
      "epoch": 0.09731507440155272,
      "grad_norm": 0.9452739953994751,
      "learning_rate": 9.986939592572413e-06,
      "loss": 0.8805,
      "step": 1805
    },
    {
      "epoch": 0.09736898857019624,
      "grad_norm": 0.9339364171028137,
      "learning_rate": 9.986924274636538e-06,
      "loss": 0.8819,
      "step": 1806
    },
    {
      "epoch": 0.09742290273883977,
      "grad_norm": 0.9344542026519775,
      "learning_rate": 9.986908947734844e-06,
      "loss": 0.8531,
      "step": 1807
    },
    {
      "epoch": 0.09747681690748329,
      "grad_norm": 0.8910528421401978,
      "learning_rate": 9.986893611867362e-06,
      "loss": 0.8949,
      "step": 1808
    },
    {
      "epoch": 0.09753073107612681,
      "grad_norm": 0.8484895825386047,
      "learning_rate": 9.986878267034115e-06,
      "loss": 0.8028,
      "step": 1809
    },
    {
      "epoch": 0.09758464524477033,
      "grad_norm": 1.0784810781478882,
      "learning_rate": 9.986862913235135e-06,
      "loss": 0.9564,
      "step": 1810
    },
    {
      "epoch": 0.09763855941341384,
      "grad_norm": 0.8350296020507812,
      "learning_rate": 9.98684755047045e-06,
      "loss": 0.8672,
      "step": 1811
    },
    {
      "epoch": 0.09769247358205736,
      "grad_norm": 0.8558050990104675,
      "learning_rate": 9.986832178740084e-06,
      "loss": 0.8538,
      "step": 1812
    },
    {
      "epoch": 0.09774638775070088,
      "grad_norm": 0.8633396029472351,
      "learning_rate": 9.986816798044066e-06,
      "loss": 0.8356,
      "step": 1813
    },
    {
      "epoch": 0.0978003019193444,
      "grad_norm": 0.8256344199180603,
      "learning_rate": 9.986801408382424e-06,
      "loss": 0.7552,
      "step": 1814
    },
    {
      "epoch": 0.09785421608798792,
      "grad_norm": 0.872844398021698,
      "learning_rate": 9.986786009755186e-06,
      "loss": 0.9153,
      "step": 1815
    },
    {
      "epoch": 0.09790813025663145,
      "grad_norm": 0.842241108417511,
      "learning_rate": 9.986770602162378e-06,
      "loss": 0.7965,
      "step": 1816
    },
    {
      "epoch": 0.09796204442527497,
      "grad_norm": 0.9673634171485901,
      "learning_rate": 9.98675518560403e-06,
      "loss": 0.8317,
      "step": 1817
    },
    {
      "epoch": 0.09801595859391848,
      "grad_norm": 0.8744896650314331,
      "learning_rate": 9.98673976008017e-06,
      "loss": 0.7342,
      "step": 1818
    },
    {
      "epoch": 0.098069872762562,
      "grad_norm": 0.7830422520637512,
      "learning_rate": 9.986724325590825e-06,
      "loss": 0.721,
      "step": 1819
    },
    {
      "epoch": 0.09812378693120552,
      "grad_norm": 1.0335441827774048,
      "learning_rate": 9.986708882136021e-06,
      "loss": 0.8088,
      "step": 1820
    },
    {
      "epoch": 0.09817770109984904,
      "grad_norm": 0.841342568397522,
      "learning_rate": 9.986693429715785e-06,
      "loss": 0.8847,
      "step": 1821
    },
    {
      "epoch": 0.09823161526849256,
      "grad_norm": 0.9405834674835205,
      "learning_rate": 9.98667796833015e-06,
      "loss": 0.8878,
      "step": 1822
    },
    {
      "epoch": 0.09828552943713607,
      "grad_norm": 0.8358225226402283,
      "learning_rate": 9.986662497979138e-06,
      "loss": 0.7377,
      "step": 1823
    },
    {
      "epoch": 0.0983394436057796,
      "grad_norm": 0.8844004273414612,
      "learning_rate": 9.98664701866278e-06,
      "loss": 0.7236,
      "step": 1824
    },
    {
      "epoch": 0.09839335777442312,
      "grad_norm": 0.8165417313575745,
      "learning_rate": 9.986631530381105e-06,
      "loss": 0.819,
      "step": 1825
    },
    {
      "epoch": 0.09844727194306664,
      "grad_norm": 0.9569553732872009,
      "learning_rate": 9.986616033134137e-06,
      "loss": 0.9337,
      "step": 1826
    },
    {
      "epoch": 0.09850118611171016,
      "grad_norm": 0.8311771750450134,
      "learning_rate": 9.986600526921907e-06,
      "loss": 0.8516,
      "step": 1827
    },
    {
      "epoch": 0.09855510028035368,
      "grad_norm": 0.9444357752799988,
      "learning_rate": 9.986585011744441e-06,
      "loss": 0.805,
      "step": 1828
    },
    {
      "epoch": 0.0986090144489972,
      "grad_norm": 1.0128875970840454,
      "learning_rate": 9.986569487601769e-06,
      "loss": 0.8514,
      "step": 1829
    },
    {
      "epoch": 0.09866292861764071,
      "grad_norm": 0.8973994255065918,
      "learning_rate": 9.986553954493917e-06,
      "loss": 0.7938,
      "step": 1830
    },
    {
      "epoch": 0.09871684278628423,
      "grad_norm": 0.8571779131889343,
      "learning_rate": 9.986538412420912e-06,
      "loss": 0.7506,
      "step": 1831
    },
    {
      "epoch": 0.09877075695492775,
      "grad_norm": 0.9053436517715454,
      "learning_rate": 9.986522861382785e-06,
      "loss": 0.8551,
      "step": 1832
    },
    {
      "epoch": 0.09882467112357128,
      "grad_norm": 0.9941746592521667,
      "learning_rate": 9.986507301379562e-06,
      "loss": 0.8828,
      "step": 1833
    },
    {
      "epoch": 0.0988785852922148,
      "grad_norm": 0.9620066285133362,
      "learning_rate": 9.986491732411272e-06,
      "loss": 0.8982,
      "step": 1834
    },
    {
      "epoch": 0.09893249946085832,
      "grad_norm": 0.9470074772834778,
      "learning_rate": 9.986476154477941e-06,
      "loss": 0.8295,
      "step": 1835
    },
    {
      "epoch": 0.09898641362950183,
      "grad_norm": 0.9962137937545776,
      "learning_rate": 9.986460567579599e-06,
      "loss": 0.8714,
      "step": 1836
    },
    {
      "epoch": 0.09904032779814535,
      "grad_norm": 0.8492829203605652,
      "learning_rate": 9.986444971716273e-06,
      "loss": 0.8234,
      "step": 1837
    },
    {
      "epoch": 0.09909424196678887,
      "grad_norm": 0.9463719725608826,
      "learning_rate": 9.986429366887994e-06,
      "loss": 0.7769,
      "step": 1838
    },
    {
      "epoch": 0.09914815613543239,
      "grad_norm": 0.8588153123855591,
      "learning_rate": 9.986413753094786e-06,
      "loss": 0.8883,
      "step": 1839
    },
    {
      "epoch": 0.0992020703040759,
      "grad_norm": 0.7692183256149292,
      "learning_rate": 9.986398130336677e-06,
      "loss": 0.7691,
      "step": 1840
    },
    {
      "epoch": 0.09925598447271942,
      "grad_norm": 0.8377199172973633,
      "learning_rate": 9.986382498613699e-06,
      "loss": 0.789,
      "step": 1841
    },
    {
      "epoch": 0.09930989864136296,
      "grad_norm": 0.9783869385719299,
      "learning_rate": 9.986366857925876e-06,
      "loss": 0.8517,
      "step": 1842
    },
    {
      "epoch": 0.09936381281000647,
      "grad_norm": 0.8233169913291931,
      "learning_rate": 9.986351208273239e-06,
      "loss": 0.8701,
      "step": 1843
    },
    {
      "epoch": 0.09941772697864999,
      "grad_norm": 0.9393780827522278,
      "learning_rate": 9.986335549655814e-06,
      "loss": 0.8837,
      "step": 1844
    },
    {
      "epoch": 0.09947164114729351,
      "grad_norm": 0.8517693877220154,
      "learning_rate": 9.986319882073631e-06,
      "loss": 0.9043,
      "step": 1845
    },
    {
      "epoch": 0.09952555531593703,
      "grad_norm": 0.8296724557876587,
      "learning_rate": 9.986304205526718e-06,
      "loss": 0.7406,
      "step": 1846
    },
    {
      "epoch": 0.09957946948458055,
      "grad_norm": 0.8372161388397217,
      "learning_rate": 9.986288520015102e-06,
      "loss": 0.7763,
      "step": 1847
    },
    {
      "epoch": 0.09963338365322406,
      "grad_norm": 0.8086470365524292,
      "learning_rate": 9.986272825538812e-06,
      "loss": 0.8786,
      "step": 1848
    },
    {
      "epoch": 0.09968729782186758,
      "grad_norm": 0.8562842011451721,
      "learning_rate": 9.986257122097875e-06,
      "loss": 0.8391,
      "step": 1849
    },
    {
      "epoch": 0.0997412119905111,
      "grad_norm": 0.9052720665931702,
      "learning_rate": 9.986241409692321e-06,
      "loss": 0.948,
      "step": 1850
    },
    {
      "epoch": 0.09979512615915463,
      "grad_norm": 0.8220609426498413,
      "learning_rate": 9.986225688322178e-06,
      "loss": 0.8039,
      "step": 1851
    },
    {
      "epoch": 0.09984904032779815,
      "grad_norm": 0.8018030524253845,
      "learning_rate": 9.98620995798747e-06,
      "loss": 0.7748,
      "step": 1852
    },
    {
      "epoch": 0.09990295449644167,
      "grad_norm": 0.8150879144668579,
      "learning_rate": 9.986194218688235e-06,
      "loss": 0.7304,
      "step": 1853
    },
    {
      "epoch": 0.09995686866508519,
      "grad_norm": 0.8677535653114319,
      "learning_rate": 9.98617847042449e-06,
      "loss": 0.8756,
      "step": 1854
    },
    {
      "epoch": 0.1000107828337287,
      "grad_norm": 0.8889294862747192,
      "learning_rate": 9.986162713196272e-06,
      "loss": 0.8926,
      "step": 1855
    }
  ],
  "logging_steps": 1,
  "max_steps": 74192,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 1855,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.474209732975657e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}