{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9998924384209961,
  "eval_steps": 500,
  "global_step": 13944,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 27.85062599182129,
      "learning_rate": 2.3866348448687353e-08,
      "loss": 0.4259,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 39.31293487548828,
      "learning_rate": 4.7732696897374706e-08,
      "loss": 3.0322,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 30.092140197753906,
      "learning_rate": 7.159904534606206e-08,
      "loss": 3.5234,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.957374572753906,
      "learning_rate": 9.546539379474941e-08,
      "loss": 3.3389,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.18878936767578,
      "learning_rate": 1.1933174224343676e-07,
      "loss": 1.6447,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 35.24954605102539,
      "learning_rate": 1.431980906921241e-07,
      "loss": 3.4297,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.25824737548828,
      "learning_rate": 1.6706443914081147e-07,
      "loss": 3.2727,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 28.20185661315918,
      "learning_rate": 1.9093078758949883e-07,
      "loss": 0.3376,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.952945709228516,
      "learning_rate": 2.1479713603818616e-07,
      "loss": 1.5743,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 30.74711799621582,
      "learning_rate": 2.386634844868735e-07,
      "loss": 0.3597,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 30.868654251098633,
      "learning_rate": 2.625298329355609e-07,
      "loss": 3.5078,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 45.23512268066406,
      "learning_rate": 2.863961813842482e-07,
      "loss": 3.375,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.499176025390625,
      "learning_rate": 3.1026252983293556e-07,
      "loss": 1.6408,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 47.16597366333008,
      "learning_rate": 3.3412887828162294e-07,
      "loss": 3.5166,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.835880279541016,
      "learning_rate": 3.579952267303103e-07,
      "loss": 3.0488,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.780803680419922,
      "learning_rate": 3.8186157517899765e-07,
      "loss": 3.4766,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.998685836791992,
      "learning_rate": 4.0572792362768503e-07,
      "loss": 0.3748,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.0872802734375,
      "learning_rate": 4.295942720763723e-07,
      "loss": 3.4121,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.669025421142578,
      "learning_rate": 4.534606205250597e-07,
      "loss": 0.2806,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.377641677856445,
      "learning_rate": 4.77326968973747e-07,
      "loss": 0.2245,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.93362808227539,
      "learning_rate": 5.011933174224344e-07,
      "loss": 3.2227,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 66.31773376464844,
      "learning_rate": 5.250596658711218e-07,
      "loss": 3.5215,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.73788070678711,
      "learning_rate": 5.489260143198091e-07,
      "loss": 1.51,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.923440933227539,
      "learning_rate": 5.727923627684965e-07,
      "loss": 0.3576,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.545602798461914,
      "learning_rate": 5.966587112171838e-07,
      "loss": 0.2775,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.80500030517578,
      "learning_rate": 6.205250596658711e-07,
      "loss": 3.3257,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.005102157592773,
      "learning_rate": 6.443914081145585e-07,
      "loss": 0.3961,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 28.592844009399414,
      "learning_rate": 6.682577565632459e-07,
      "loss": 3.1182,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.781620979309082,
      "learning_rate": 6.921241050119332e-07,
      "loss": 0.3737,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.963949203491211,
      "learning_rate": 7.159904534606206e-07,
      "loss": 1.5498,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.929140090942383,
      "learning_rate": 7.398568019093079e-07,
      "loss": 3.2988,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.522132873535156,
      "learning_rate": 7.637231503579953e-07,
      "loss": 3.0735,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.669593811035156,
      "learning_rate": 7.875894988066826e-07,
      "loss": 1.5161,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.902339935302734,
      "learning_rate": 8.114558472553701e-07,
      "loss": 3.3896,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.835033416748047,
      "learning_rate": 8.353221957040573e-07,
      "loss": 1.4601,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.08356285095215,
      "learning_rate": 8.591885441527446e-07,
      "loss": 1.5425,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.164640426635742,
      "learning_rate": 8.830548926014321e-07,
      "loss": 2.814,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.3609724044799805,
      "learning_rate": 9.069212410501194e-07,
      "loss": 0.185,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4838056564331055,
      "learning_rate": 9.307875894988068e-07,
      "loss": 0.1378,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.611600875854492,
      "learning_rate": 9.54653937947494e-07,
      "loss": 3.2607,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.814241886138916,
      "learning_rate": 9.785202863961816e-07,
      "loss": 0.2118,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.2152681350708,
      "learning_rate": 1.0023866348448687e-06,
      "loss": 0.3055,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.908705711364746,
      "learning_rate": 1.026252983293556e-06,
      "loss": 0.1819,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.999817848205566,
      "learning_rate": 1.0501193317422436e-06,
      "loss": 0.238,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.44059371948242,
      "learning_rate": 1.073985680190931e-06,
      "loss": 3.2627,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.282682418823242,
      "learning_rate": 1.0978520286396182e-06,
      "loss": 3.2314,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.57794952392578,
      "learning_rate": 1.1217183770883056e-06,
      "loss": 3.1787,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.920943260192871,
      "learning_rate": 1.145584725536993e-06,
      "loss": 3.4941,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.535140514373779,
      "learning_rate": 1.1694510739856802e-06,
      "loss": 0.1558,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.935603618621826,
      "learning_rate": 1.1933174224343676e-06,
      "loss": 0.1668,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.099740982055664,
      "learning_rate": 1.217183770883055e-06,
      "loss": 3.3555,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.804582595825195,
      "learning_rate": 1.2410501193317422e-06,
      "loss": 3.2471,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.085600852966309,
      "learning_rate": 1.2649164677804296e-06,
      "loss": 1.2823,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.849931716918945,
      "learning_rate": 1.288782816229117e-06,
      "loss": 3.418,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 33.43053436279297,
      "learning_rate": 1.3126491646778044e-06,
      "loss": 3.2402,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.957284927368164,
      "learning_rate": 1.3365155131264918e-06,
      "loss": 3.2324,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.370464324951172,
      "learning_rate": 1.360381861575179e-06,
      "loss": 3.2002,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.382131576538086,
      "learning_rate": 1.3842482100238664e-06,
      "loss": 3.0625,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.713335990905762,
      "learning_rate": 1.408114558472554e-06,
      "loss": 0.2261,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.54577922821045,
      "learning_rate": 1.4319809069212413e-06,
      "loss": 1.3122,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.637004852294922,
      "learning_rate": 1.4558472553699284e-06,
      "loss": 3.2876,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.773542404174805,
      "learning_rate": 1.4797136038186157e-06,
      "loss": 0.2336,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4341206550598145,
      "learning_rate": 1.5035799522673033e-06,
      "loss": 0.1858,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.744496822357178,
      "learning_rate": 1.5274463007159906e-06,
      "loss": 0.2304,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.05228042602539,
      "learning_rate": 1.551312649164678e-06,
      "loss": 3.2109,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.883865356445312,
      "learning_rate": 1.5751789976133653e-06,
      "loss": 1.1608,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.316129684448242,
      "learning_rate": 1.5990453460620526e-06,
      "loss": 0.1877,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.956028938293457,
      "learning_rate": 1.6229116945107401e-06,
      "loss": 0.2154,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.69055700302124,
      "learning_rate": 1.6467780429594275e-06,
      "loss": 0.1308,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.286288261413574,
      "learning_rate": 1.6706443914081146e-06,
      "loss": 1.2117,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.748434066772461,
      "learning_rate": 1.694510739856802e-06,
      "loss": 0.328,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.583340167999268,
      "learning_rate": 1.7183770883054892e-06,
      "loss": 0.2104,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.373615264892578,
      "learning_rate": 1.7422434367541768e-06,
      "loss": 0.1457,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.73750114440918,
      "learning_rate": 1.7661097852028641e-06,
      "loss": 3.1689,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.935587406158447,
      "learning_rate": 1.7899761336515514e-06,
      "loss": 0.1674,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.761417388916016,
      "learning_rate": 1.8138424821002388e-06,
      "loss": 1.2183,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.207029342651367,
      "learning_rate": 1.8377088305489263e-06,
      "loss": 0.1454,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.245403289794922,
      "learning_rate": 1.8615751789976136e-06,
      "loss": 3.1348,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.609433174133301,
      "learning_rate": 1.885441527446301e-06,
      "loss": 0.2465,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 42.88984680175781,
      "learning_rate": 1.909307875894988e-06,
      "loss": 3.2554,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.97998571395874,
      "learning_rate": 1.9331742243436756e-06,
      "loss": 0.1559,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.083598136901855,
      "learning_rate": 1.957040572792363e-06,
      "loss": 3.2695,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.450130462646484,
      "learning_rate": 1.9809069212410503e-06,
      "loss": 3.0518,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.856487274169922,
      "learning_rate": 2.0047732696897374e-06,
      "loss": 3.2969,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.22026252746582,
      "learning_rate": 2.028639618138425e-06,
      "loss": 0.1003,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.612014770507812,
      "learning_rate": 2.052505966587112e-06,
      "loss": 1.0473,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.097400665283203,
      "learning_rate": 2.0763723150357996e-06,
      "loss": 3.2625,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.052860260009766,
      "learning_rate": 2.100238663484487e-06,
      "loss": 2.9622,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0120439529418945,
      "learning_rate": 2.1241050119331743e-06,
      "loss": 0.131,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.379283905029297,
      "learning_rate": 2.147971360381862e-06,
      "loss": 3.2681,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.300370216369629,
      "learning_rate": 2.1718377088305494e-06,
      "loss": 3.3433,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.266372680664062,
      "learning_rate": 2.1957040572792365e-06,
      "loss": 0.1695,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.751738548278809,
      "learning_rate": 2.219570405727924e-06,
      "loss": 1.0286,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.499706268310547,
      "learning_rate": 2.243436754176611e-06,
      "loss": 0.1968,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.859902381896973,
      "learning_rate": 2.2673031026252983e-06,
      "loss": 0.1306,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.597694396972656,
      "learning_rate": 2.291169451073986e-06,
      "loss": 1.0781,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.340027809143066,
      "learning_rate": 2.3150357995226733e-06,
      "loss": 3.1816,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.156256675720215,
      "learning_rate": 2.3389021479713605e-06,
      "loss": 0.1586,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.322132110595703,
      "learning_rate": 2.362768496420048e-06,
      "loss": 0.1633,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.958500862121582,
      "learning_rate": 2.386634844868735e-06,
      "loss": 3.4863,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.30799102783203,
      "learning_rate": 2.4105011933174227e-06,
      "loss": 3.022,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.461362361907959,
      "learning_rate": 2.43436754176611e-06,
      "loss": 0.0891,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.631566047668457,
      "learning_rate": 2.4582338902147973e-06,
      "loss": 1.1196,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.225022315979004,
      "learning_rate": 2.4821002386634844e-06,
      "loss": 0.2158,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.276027202606201,
      "learning_rate": 2.505966587112172e-06,
      "loss": 0.2619,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.695845603942871,
      "learning_rate": 2.529832935560859e-06,
      "loss": 0.1825,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.335461616516113,
      "learning_rate": 2.5536992840095466e-06,
      "loss": 0.1855,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.440604209899902,
      "learning_rate": 2.577565632458234e-06,
      "loss": 1.041,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.887969970703125,
      "learning_rate": 2.6014319809069217e-06,
      "loss": 3.3052,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.76985740661621,
      "learning_rate": 2.625298329355609e-06,
      "loss": 3.2354,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.00410270690918,
      "learning_rate": 2.6491646778042964e-06,
      "loss": 1.0489,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.999443054199219,
      "learning_rate": 2.6730310262529835e-06,
      "loss": 0.1049,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.593664169311523,
      "learning_rate": 2.6968973747016706e-06,
      "loss": 3.2206,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.557558536529541,
      "learning_rate": 2.720763723150358e-06,
      "loss": 0.2009,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.1054205894470215,
      "learning_rate": 2.7446300715990453e-06,
      "loss": 0.1956,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.702746868133545,
      "learning_rate": 2.768496420047733e-06,
      "loss": 0.1351,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.824159622192383,
      "learning_rate": 2.7923627684964204e-06,
      "loss": 3.1331,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0172648429870605,
      "learning_rate": 2.816229116945108e-06,
      "loss": 0.1259,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.570687294006348,
      "learning_rate": 2.840095465393795e-06,
      "loss": 0.9513,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.516415119171143,
      "learning_rate": 2.8639618138424826e-06,
      "loss": 0.1362,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.770516395568848,
      "learning_rate": 2.8878281622911697e-06,
      "loss": 0.1244,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.12515926361084,
      "learning_rate": 2.911694510739857e-06,
      "loss": 0.2494,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.425390243530273,
      "learning_rate": 2.9355608591885443e-06,
      "loss": 1.0796,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.851609230041504,
      "learning_rate": 2.9594272076372315e-06,
      "loss": 0.1112,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.3731231689453125,
      "learning_rate": 2.983293556085919e-06,
      "loss": 0.088,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.966947555541992,
      "learning_rate": 3.0071599045346066e-06,
      "loss": 0.1321,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.72164249420166,
      "learning_rate": 3.031026252983294e-06,
      "loss": 0.2138,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.300365447998047,
      "learning_rate": 3.054892601431981e-06,
      "loss": 3.2856,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.600420951843262,
      "learning_rate": 3.0787589498806688e-06,
      "loss": 0.9896,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.9525532722473145,
      "learning_rate": 3.102625298329356e-06,
      "loss": 0.2309,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.665853977203369,
      "learning_rate": 3.126491646778043e-06,
      "loss": 1.0104,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.871527671813965,
      "learning_rate": 3.1503579952267305e-06,
      "loss": 0.1761,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.871618270874023,
      "learning_rate": 3.1742243436754176e-06,
      "loss": 3.291,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.242939472198486,
      "learning_rate": 3.198090692124105e-06,
      "loss": 0.1757,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.807085037231445,
      "learning_rate": 3.2219570405727923e-06,
      "loss": 2.9844,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.183164119720459,
      "learning_rate": 3.2458233890214803e-06,
      "loss": 0.1597,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.616246223449707,
      "learning_rate": 3.2696897374701674e-06,
      "loss": 1.0841,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.841903686523438,
      "learning_rate": 3.293556085918855e-06,
      "loss": 2.8345,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.954968452453613,
      "learning_rate": 3.317422434367542e-06,
      "loss": 2.9351,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.193575859069824,
      "learning_rate": 3.341288782816229e-06,
      "loss": 0.9626,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.750083923339844,
      "learning_rate": 3.3651551312649167e-06,
      "loss": 3.1191,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.033980846405029,
      "learning_rate": 3.389021479713604e-06,
      "loss": 0.1129,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.4968318939209,
      "learning_rate": 3.4128878281622914e-06,
      "loss": 2.7393,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.291167259216309,
      "learning_rate": 3.4367541766109785e-06,
      "loss": 0.9285,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.868911266326904,
      "learning_rate": 3.4606205250596665e-06,
      "loss": 0.0919,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.884643077850342,
      "learning_rate": 3.4844868735083536e-06,
      "loss": 0.237,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.985185146331787,
      "learning_rate": 3.508353221957041e-06,
      "loss": 0.9678,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.112469673156738,
      "learning_rate": 3.5322195704057282e-06,
      "loss": 0.8806,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.415334701538086,
      "learning_rate": 3.5560859188544154e-06,
      "loss": 0.1157,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.942835330963135,
      "learning_rate": 3.579952267303103e-06,
      "loss": 0.9596,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.326348781585693,
      "learning_rate": 3.60381861575179e-06,
      "loss": 0.082,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.637968063354492,
      "learning_rate": 3.6276849642004776e-06,
      "loss": 0.2781,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.688216209411621,
      "learning_rate": 3.6515513126491647e-06,
      "loss": 0.1766,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.780233383178711,
      "learning_rate": 3.6754176610978526e-06,
      "loss": 1.1554,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.775963306427002,
      "learning_rate": 3.6992840095465398e-06,
      "loss": 0.1349,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.239860534667969,
      "learning_rate": 3.7231503579952273e-06,
      "loss": 1.1265,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.847940921783447,
      "learning_rate": 3.7470167064439144e-06,
      "loss": 0.2615,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.219351291656494,
      "learning_rate": 3.770883054892602e-06,
      "loss": 0.1427,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 36.71184539794922,
      "learning_rate": 3.794749403341289e-06,
      "loss": 2.8997,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.575949668884277,
      "learning_rate": 3.818615751789976e-06,
      "loss": 3.2417,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.381652355194092,
      "learning_rate": 3.842482100238663e-06,
      "loss": 0.2087,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.872273921966553,
      "learning_rate": 3.866348448687351e-06,
      "loss": 0.1533,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.686379432678223,
      "learning_rate": 3.890214797136038e-06,
      "loss": 0.1951,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.827803611755371,
      "learning_rate": 3.914081145584726e-06,
      "loss": 3.2324,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.905025959014893,
      "learning_rate": 3.9379474940334135e-06,
      "loss": 0.19,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.119263648986816,
      "learning_rate": 3.961813842482101e-06,
      "loss": 3.2871,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.190079689025879,
      "learning_rate": 3.985680190930788e-06,
      "loss": 0.1476,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.885051965713501,
      "learning_rate": 4.009546539379475e-06,
      "loss": 0.1069,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.3723726272583,
      "learning_rate": 4.033412887828163e-06,
      "loss": 3.2341,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.520909309387207,
      "learning_rate": 4.05727923627685e-06,
      "loss": 1.0563,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.60108470916748,
      "learning_rate": 4.081145584725537e-06,
      "loss": 3.2114,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.580583572387695,
      "learning_rate": 4.105011933174224e-06,
      "loss": 3.4688,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.274011611938477,
      "learning_rate": 4.128878281622912e-06,
      "loss": 1.1075,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.746452808380127,
      "learning_rate": 4.152744630071599e-06,
      "loss": 0.1461,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.474963188171387,
      "learning_rate": 4.176610978520287e-06,
      "loss": 3.1855,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.7621660232543945,
      "learning_rate": 4.200477326968974e-06,
      "loss": 0.2212,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.608884811401367,
      "learning_rate": 4.2243436754176614e-06,
      "loss": 2.9385,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.955658435821533,
      "learning_rate": 4.2482100238663486e-06,
      "loss": 0.2167,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.393836498260498,
      "learning_rate": 4.272076372315036e-06,
      "loss": 0.0921,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.874258041381836,
      "learning_rate": 4.295942720763724e-06,
      "loss": 2.8491,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.280537605285645,
      "learning_rate": 4.319809069212411e-06,
      "loss": 2.6738,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.821020126342773,
      "learning_rate": 4.343675417661099e-06,
      "loss": 2.4326,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.665031433105469,
      "learning_rate": 4.367541766109786e-06,
      "loss": 2.8262,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.361561298370361,
      "learning_rate": 4.391408114558473e-06,
      "loss": 0.2523,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.284908294677734,
      "learning_rate": 4.41527446300716e-06,
      "loss": 2.9561,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 23.71626853942871,
      "learning_rate": 4.439140811455848e-06,
      "loss": 2.7056,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 30.785314559936523,
      "learning_rate": 4.463007159904535e-06,
      "loss": 2.9258,
      "step": 187
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.976282119750977,
      "learning_rate": 4.486873508353222e-06,
      "loss": 0.1461,
      "step": 188
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.143136024475098,
      "learning_rate": 4.510739856801909e-06,
      "loss": 0.2509,
      "step": 189
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.355886459350586,
      "learning_rate": 4.5346062052505965e-06,
      "loss": 0.9759,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.734907150268555,
      "learning_rate": 4.5584725536992845e-06,
      "loss": 2.4233,
      "step": 191
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.497225761413574,
      "learning_rate": 4.582338902147972e-06,
      "loss": 1.1165,
      "step": 192
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.406531810760498,
      "learning_rate": 4.6062052505966596e-06,
      "loss": 0.2744,
      "step": 193
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.819290161132812,
      "learning_rate": 4.630071599045347e-06,
      "loss": 2.8643,
      "step": 194
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.781013011932373,
      "learning_rate": 4.653937947494034e-06,
      "loss": 0.2837,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.145565032958984,
      "learning_rate": 4.677804295942721e-06,
      "loss": 2.5845,
      "step": 196
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.693892002105713,
      "learning_rate": 4.701670644391408e-06,
      "loss": 0.2092,
      "step": 197
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.683666706085205,
      "learning_rate": 4.725536992840096e-06,
      "loss": 0.1526,
      "step": 198
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.976194858551025,
      "learning_rate": 4.749403341288783e-06,
      "loss": 0.1793,
      "step": 199
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.213768482208252,
      "learning_rate": 4.77326968973747e-06,
      "loss": 0.14,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 43.59369659423828,
      "learning_rate": 4.797136038186158e-06,
      "loss": 2.3232,
      "step": 201
    },
    {
      "epoch": 0.01,
      "grad_norm": 46.719703674316406,
      "learning_rate": 4.821002386634845e-06,
      "loss": 2.4138,
      "step": 202
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.142409801483154,
      "learning_rate": 4.8448687350835324e-06,
      "loss": 0.1316,
      "step": 203
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.384970664978027,
      "learning_rate": 4.86873508353222e-06,
      "loss": 0.1562,
      "step": 204
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.805733680725098,
      "learning_rate": 4.8926014319809075e-06,
      "loss": 0.312,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.9200897216796875,
      "learning_rate": 4.916467780429595e-06,
      "loss": 0.2255,
      "step": 206
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.8262858390808105,
      "learning_rate": 4.940334128878282e-06,
      "loss": 1.094,
      "step": 207
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.518152236938477,
      "learning_rate": 4.964200477326969e-06,
      "loss": 2.3672,
      "step": 208
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.696699619293213,
      "learning_rate": 4.988066825775657e-06,
      "loss": 0.1207,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.26022481918335,
      "learning_rate": 5.011933174224344e-06,
      "loss": 0.1789,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.955089092254639,
      "learning_rate": 5.035799522673032e-06,
      "loss": 0.9197,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.587421417236328,
      "learning_rate": 5.059665871121718e-06,
      "loss": 0.2683,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.085227012634277,
      "learning_rate": 5.083532219570406e-06,
      "loss": 0.2124,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.62516975402832,
      "learning_rate": 5.107398568019093e-06,
      "loss": 2.3101,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.739049911499023,
      "learning_rate": 5.13126491646778e-06,
      "loss": 1.0492,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.92177152633667,
      "learning_rate": 5.155131264916468e-06,
      "loss": 1.0776,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.897430896759033,
      "learning_rate": 5.1789976133651555e-06,
      "loss": 0.1671,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.501258850097656,
      "learning_rate": 5.2028639618138435e-06,
      "loss": 2.533,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.146834850311279,
      "learning_rate": 5.22673031026253e-06,
      "loss": 0.1594,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.293914794921875,
      "learning_rate": 5.250596658711218e-06,
      "loss": 0.2086,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.837182998657227,
      "learning_rate": 5.274463007159905e-06,
      "loss": 0.2411,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.814523696899414,
      "learning_rate": 5.298329355608593e-06,
      "loss": 2.1064,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.105995178222656,
      "learning_rate": 5.322195704057279e-06,
      "loss": 1.9734,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.168612480163574,
      "learning_rate": 5.346062052505967e-06,
      "loss": 0.2582,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.108205795288086,
      "learning_rate": 5.369928400954655e-06,
      "loss": 1.9709,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.469554901123047,
      "learning_rate": 5.393794749403341e-06,
      "loss": 1.9321,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.614267349243164,
      "learning_rate": 5.417661097852029e-06,
      "loss": 2.016,
      "step": 227
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.4017720222473145,
      "learning_rate": 5.441527446300716e-06,
      "loss": 0.2097,
      "step": 228
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.513286113739014,
      "learning_rate": 5.465393794749404e-06,
      "loss": 0.1284,
      "step": 229
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.413851261138916,
      "learning_rate": 5.4892601431980906e-06,
      "loss": 0.1139,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.998461723327637,
      "learning_rate": 5.5131264916467785e-06,
      "loss": 0.1742,
      "step": 231
    },
    {
      "epoch": 0.02,
      "grad_norm": 25.13848304748535,
      "learning_rate": 5.536992840095466e-06,
      "loss": 2.0214,
      "step": 232
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.982104301452637,
      "learning_rate": 5.560859188544153e-06,
      "loss": 1.0243,
      "step": 233
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.687582015991211,
      "learning_rate": 5.584725536992841e-06,
      "loss": 1.682,
      "step": 234
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.2003173828125,
      "learning_rate": 5.608591885441528e-06,
      "loss": 1.2847,
      "step": 235
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.791386604309082,
      "learning_rate": 5.632458233890216e-06,
      "loss": 2.0698,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.371529579162598,
      "learning_rate": 5.656324582338902e-06,
      "loss": 1.4911,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.529620170593262,
      "learning_rate": 5.68019093078759e-06,
      "loss": 1.5923,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.034557342529297,
      "learning_rate": 5.704057279236277e-06,
      "loss": 0.23,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.476567268371582,
      "learning_rate": 5.727923627684965e-06,
      "loss": 0.0893,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.551618576049805,
      "learning_rate": 5.751789976133651e-06,
      "loss": 0.1445,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.49770450592041,
      "learning_rate": 5.775656324582339e-06,
      "loss": 1.7434,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.335045337677002,
      "learning_rate": 5.799522673031027e-06,
      "loss": 0.1479,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.3712568283081055,
      "learning_rate": 5.823389021479714e-06,
      "loss": 0.2189,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.287660598754883,
      "learning_rate": 5.847255369928402e-06,
      "loss": 2.2551,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.115777969360352,
      "learning_rate": 5.871121718377089e-06,
      "loss": 0.1221,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 51.66255569458008,
      "learning_rate": 5.894988066825777e-06,
      "loss": 2.1548,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.614236831665039,
      "learning_rate": 5.918854415274463e-06,
      "loss": 0.2169,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 26.18594741821289,
      "learning_rate": 5.942720763723151e-06,
      "loss": 2.1689,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.457281589508057,
      "learning_rate": 5.966587112171838e-06,
      "loss": 0.1493,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.34776782989502,
      "learning_rate": 5.990453460620526e-06,
      "loss": 1.6108,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 58.661380767822266,
      "learning_rate": 6.014319809069213e-06,
      "loss": 1.8885,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8144659996032715,
      "learning_rate": 6.0381861575179e-06,
      "loss": 0.2408,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.479042053222656,
      "learning_rate": 6.062052505966588e-06,
      "loss": 0.1555,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.317495346069336,
      "learning_rate": 6.0859188544152745e-06,
      "loss": 0.1449,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 26.920459747314453,
      "learning_rate": 6.109785202863962e-06,
      "loss": 2.0545,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.758576393127441,
      "learning_rate": 6.1336515513126495e-06,
      "loss": 1.3064,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.888866424560547,
      "learning_rate": 6.1575178997613375e-06,
      "loss": 0.212,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.559238433837891,
      "learning_rate": 6.181384248210024e-06,
      "loss": 0.1705,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.847671031951904,
      "learning_rate": 6.205250596658712e-06,
      "loss": 0.1451,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.11386775970459,
      "learning_rate": 6.2291169451074e-06,
      "loss": 0.1977,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.6174235343933105,
      "learning_rate": 6.252983293556086e-06,
      "loss": 1.0975,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.055359363555908,
      "learning_rate": 6.276849642004774e-06,
      "loss": 0.1845,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.271392822265625,
      "learning_rate": 6.300715990453461e-06,
      "loss": 1.8544,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.006891250610352,
      "learning_rate": 6.324582338902149e-06,
      "loss": 1.8776,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.408445835113525,
      "learning_rate": 6.348448687350835e-06,
      "loss": 0.1862,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.617584228515625,
      "learning_rate": 6.372315035799523e-06,
      "loss": 1.4482,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.798867702484131,
      "learning_rate": 6.39618138424821e-06,
      "loss": 0.1909,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.626176357269287,
      "learning_rate": 6.420047732696898e-06,
      "loss": 1.1745,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9295244216918945,
      "learning_rate": 6.443914081145585e-06,
      "loss": 0.2033,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.971166133880615,
      "learning_rate": 6.467780429594273e-06,
      "loss": 0.1929,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.8090291023254395,
      "learning_rate": 6.4916467780429605e-06,
      "loss": 0.2055,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.266201496124268,
      "learning_rate": 6.515513126491647e-06,
      "loss": 0.9017,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.42595100402832,
      "learning_rate": 6.539379474940335e-06,
      "loss": 1.9542,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.932315826416016,
      "learning_rate": 6.563245823389022e-06,
      "loss": 1.7467,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 24.637062072753906,
      "learning_rate": 6.58711217183771e-06,
      "loss": 2.3748,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.142692565917969,
      "learning_rate": 6.610978520286396e-06,
      "loss": 1.5859,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.317706108093262,
      "learning_rate": 6.634844868735084e-06,
      "loss": 1.8748,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.013027667999268,
      "learning_rate": 6.658711217183771e-06,
      "loss": 0.194,
      "step": 279
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.957297325134277,
      "learning_rate": 6.682577565632458e-06,
      "loss": 0.1893,
      "step": 280
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.454446792602539,
      "learning_rate": 6.706443914081146e-06,
      "loss": 0.2432,
      "step": 281
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.193295955657959,
      "learning_rate": 6.7303102625298334e-06,
      "loss": 0.2869,
      "step": 282
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.080166816711426,
      "learning_rate": 6.754176610978521e-06,
      "loss": 1.1206,
      "step": 283
    },
    {
      "epoch": 0.02,
      "grad_norm": 33.960304260253906,
      "learning_rate": 6.778042959427208e-06,
      "loss": 2.1177,
      "step": 284
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.37203025817871,
      "learning_rate": 6.801909307875896e-06,
      "loss": 1.3185,
      "step": 285
    },
    {
      "epoch": 0.02,
      "grad_norm": 60.37544631958008,
      "learning_rate": 6.825775656324583e-06,
      "loss": 1.6853,
      "step": 286
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.666402339935303,
      "learning_rate": 6.849642004773271e-06,
      "loss": 0.1626,
      "step": 287
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.257932662963867,
      "learning_rate": 6.873508353221957e-06,
      "loss": 1.5982,
      "step": 288
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.473372459411621,
      "learning_rate": 6.897374701670645e-06,
      "loss": 1.0391,
      "step": 289
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.810182571411133,
      "learning_rate": 6.921241050119333e-06,
      "loss": 1.6919,
      "step": 290
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.479059219360352,
      "learning_rate": 6.945107398568019e-06,
      "loss": 1.0348,
      "step": 291
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.432766914367676,
      "learning_rate": 6.968973747016707e-06,
      "loss": 1.6929,
      "step": 292
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.268370151519775,
      "learning_rate": 6.992840095465394e-06,
      "loss": 0.2113,
      "step": 293
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.99972152709961,
      "learning_rate": 7.016706443914082e-06,
      "loss": 1.5163,
      "step": 294
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.182229042053223,
      "learning_rate": 7.0405727923627685e-06,
      "loss": 0.2798,
      "step": 295
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.406827926635742,
      "learning_rate": 7.0644391408114565e-06,
      "loss": 0.1711,
      "step": 296
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.753603458404541,
      "learning_rate": 7.088305489260144e-06,
      "loss": 0.2368,
      "step": 297
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.459636688232422,
      "learning_rate": 7.112171837708831e-06,
      "loss": 1.3699,
      "step": 298
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.605556011199951,
      "learning_rate": 7.136038186157519e-06,
      "loss": 0.2902,
      "step": 299
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.58261489868164,
      "learning_rate": 7.159904534606206e-06,
      "loss": 1.5798,
      "step": 300
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.338225841522217,
      "learning_rate": 7.183770883054894e-06,
      "loss": 1.1042,
      "step": 301
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.750434875488281,
      "learning_rate": 7.20763723150358e-06,
      "loss": 1.435,
      "step": 302
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.371431827545166,
      "learning_rate": 7.231503579952268e-06,
      "loss": 0.165,
      "step": 303
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.868019104003906,
      "learning_rate": 7.255369928400955e-06,
      "loss": 0.2186,
      "step": 304
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.6646370887756348,
      "learning_rate": 7.279236276849643e-06,
      "loss": 0.0917,
      "step": 305
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.1093621253967285,
      "learning_rate": 7.303102625298329e-06,
      "loss": 0.2173,
      "step": 306
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.123207092285156,
      "learning_rate": 7.326968973747017e-06,
      "loss": 1.0714,
      "step": 307
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.045269012451172,
      "learning_rate": 7.350835322195705e-06,
      "loss": 1.4079,
      "step": 308
    },
    {
      "epoch": 0.02,
      "grad_norm": 28.59505271911621,
      "learning_rate": 7.3747016706443915e-06,
      "loss": 1.1853,
      "step": 309
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.74015235900879,
      "learning_rate": 7.3985680190930795e-06,
      "loss": 1.4961,
      "step": 310
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.694206237792969,
      "learning_rate": 7.422434367541767e-06,
      "loss": 0.208,
      "step": 311
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.65590763092041,
      "learning_rate": 7.446300715990455e-06,
      "loss": 1.0341,
      "step": 312
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.165308952331543,
      "learning_rate": 7.470167064439141e-06,
      "loss": 0.9286,
      "step": 313
    },
    {
      "epoch": 0.02,
      "grad_norm": 28.373455047607422,
      "learning_rate": 7.494033412887829e-06,
      "loss": 2.1813,
      "step": 314
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.034715175628662,
      "learning_rate": 7.517899761336516e-06,
      "loss": 1.1113,
      "step": 315
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.080165386199951,
      "learning_rate": 7.541766109785204e-06,
      "loss": 1.0006,
      "step": 316
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.59237813949585,
      "learning_rate": 7.565632458233891e-06,
      "loss": 0.1366,
      "step": 317
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.253661155700684,
      "learning_rate": 7.589498806682578e-06,
      "loss": 0.2052,
      "step": 318
    },
    {
      "epoch": 0.02,
      "grad_norm": 42.04719161987305,
      "learning_rate": 7.613365155131266e-06,
      "loss": 1.5516,
      "step": 319
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.367011547088623,
      "learning_rate": 7.637231503579952e-06,
      "loss": 0.2567,
      "step": 320
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.846184730529785,
      "learning_rate": 7.661097852028641e-06,
      "loss": 2.3658,
      "step": 321
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.594919204711914,
      "learning_rate": 7.684964200477327e-06,
      "loss": 1.7204,
      "step": 322
    },
    {
      "epoch": 0.02,
      "grad_norm": 27.61686134338379,
      "learning_rate": 7.708830548926015e-06,
      "loss": 2.0536,
      "step": 323
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.22140884399414,
      "learning_rate": 7.732696897374703e-06,
      "loss": 1.0462,
      "step": 324
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.613796234130859,
      "learning_rate": 7.75656324582339e-06,
      "loss": 0.2287,
      "step": 325
    },
    {
      "epoch": 0.02,
      "grad_norm": 27.421138763427734,
      "learning_rate": 7.780429594272077e-06,
      "loss": 1.7311,
      "step": 326
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.231576919555664,
      "learning_rate": 7.804295942720764e-06,
      "loss": 1.5178,
      "step": 327
    },
    {
      "epoch": 0.02,
      "grad_norm": 32.60564422607422,
      "learning_rate": 7.828162291169453e-06,
      "loss": 1.9094,
      "step": 328
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.749217987060547,
      "learning_rate": 7.852028639618138e-06,
      "loss": 1.6864,
      "step": 329
    },
    {
      "epoch": 0.02,
      "grad_norm": 37.396488189697266,
      "learning_rate": 7.875894988066827e-06,
      "loss": 2.0648,
      "step": 330
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.25080680847168,
      "learning_rate": 7.899761336515514e-06,
      "loss": 0.1811,
      "step": 331
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.450233459472656,
      "learning_rate": 7.923627684964201e-06,
      "loss": 0.2472,
      "step": 332
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.110951900482178,
      "learning_rate": 7.947494033412888e-06,
      "loss": 0.1411,
      "step": 333
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.167850494384766,
      "learning_rate": 7.971360381861575e-06,
      "loss": 1.8119,
      "step": 334
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.082810401916504,
      "learning_rate": 7.995226730310263e-06,
      "loss": 0.2698,
      "step": 335
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.830267429351807,
      "learning_rate": 8.01909307875895e-06,
      "loss": 0.1374,
      "step": 336
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.371277332305908,
      "learning_rate": 8.042959427207638e-06,
      "loss": 0.1937,
      "step": 337
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.64510726928711,
      "learning_rate": 8.066825775656326e-06,
      "loss": 1.3392,
      "step": 338
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9497387409210205,
      "learning_rate": 8.090692124105013e-06,
      "loss": 0.1343,
      "step": 339
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.76235580444336,
      "learning_rate": 8.1145584725537e-06,
      "loss": 1.399,
      "step": 340
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4064435958862305,
      "learning_rate": 8.138424821002387e-06,
      "loss": 0.2381,
      "step": 341
    },
    {
      "epoch": 0.02,
      "grad_norm": 54.01567077636719,
      "learning_rate": 8.162291169451074e-06,
      "loss": 2.5981,
      "step": 342
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.679837703704834,
      "learning_rate": 8.186157517899763e-06,
      "loss": 0.9594,
      "step": 343
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.945934295654297,
      "learning_rate": 8.210023866348448e-06,
      "loss": 1.7007,
      "step": 344
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.748713970184326,
      "learning_rate": 8.233890214797137e-06,
      "loss": 0.252,
      "step": 345
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.329431533813477,
      "learning_rate": 8.257756563245824e-06,
      "loss": 2.7065,
      "step": 346
    },
    {
      "epoch": 0.02,
      "grad_norm": 25.446884155273438,
      "learning_rate": 8.281622911694511e-06,
      "loss": 1.8872,
      "step": 347
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.27498722076416,
      "learning_rate": 8.305489260143198e-06,
      "loss": 0.1707,
      "step": 348
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.6156158447265625,
      "learning_rate": 8.329355608591886e-06,
      "loss": 1.0156,
      "step": 349
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.72133445739746,
      "learning_rate": 8.353221957040574e-06,
      "loss": 1.8402,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.100526809692383,
      "learning_rate": 8.37708830548926e-06,
      "loss": 2.0554,
      "step": 351
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.120012283325195,
      "learning_rate": 8.400954653937949e-06,
      "loss": 1.5596,
      "step": 352
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.02634048461914,
      "learning_rate": 8.424821002386636e-06,
      "loss": 2.0026,
      "step": 353
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.880990028381348,
      "learning_rate": 8.448687350835323e-06,
      "loss": 1.2395,
      "step": 354
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.221147537231445,
      "learning_rate": 8.47255369928401e-06,
      "loss": 1.3752,
      "step": 355
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.004133224487305,
      "learning_rate": 8.496420047732697e-06,
      "loss": 1.6024,
      "step": 356
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.276809692382812,
      "learning_rate": 8.520286396181386e-06,
      "loss": 0.2901,
      "step": 357
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.671300172805786,
      "learning_rate": 8.544152744630071e-06,
      "loss": 0.1031,
      "step": 358
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.122105598449707,
      "learning_rate": 8.56801909307876e-06,
      "loss": 0.2141,
      "step": 359
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.224388122558594,
      "learning_rate": 8.591885441527447e-06,
      "loss": 0.1969,
      "step": 360
    },
    {
      "epoch": 0.03,
      "grad_norm": 20.023622512817383,
      "learning_rate": 8.615751789976134e-06,
      "loss": 2.2733,
      "step": 361
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.224896430969238,
      "learning_rate": 8.639618138424822e-06,
      "loss": 1.3861,
      "step": 362
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.129985809326172,
      "learning_rate": 8.663484486873509e-06,
      "loss": 2.0922,
      "step": 363
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.708957672119141,
      "learning_rate": 8.687350835322197e-06,
      "loss": 0.1474,
      "step": 364
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.358415603637695,
      "learning_rate": 8.711217183770883e-06,
      "loss": 2.135,
      "step": 365
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.277490139007568,
      "learning_rate": 8.735083532219572e-06,
      "loss": 0.2629,
      "step": 366
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.754583358764648,
      "learning_rate": 8.758949880668259e-06,
      "loss": 1.6924,
      "step": 367
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.14402961730957,
      "learning_rate": 8.782816229116946e-06,
      "loss": 0.2232,
      "step": 368
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.83326530456543,
      "learning_rate": 8.806682577565633e-06,
      "loss": 2.1696,
      "step": 369
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.66140365600586,
      "learning_rate": 8.83054892601432e-06,
      "loss": 1.7473,
      "step": 370
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.5750837326049805,
      "learning_rate": 8.854415274463007e-06,
      "loss": 0.2378,
      "step": 371
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.6355562210083,
      "learning_rate": 8.878281622911696e-06,
      "loss": 1.6308,
      "step": 372
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.21183204650879,
      "learning_rate": 8.902147971360383e-06,
      "loss": 1.272,
      "step": 373
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.464688301086426,
      "learning_rate": 8.92601431980907e-06,
      "loss": 1.1521,
      "step": 374
    },
    {
      "epoch": 0.03,
      "grad_norm": 22.02095603942871,
      "learning_rate": 8.949880668257757e-06,
      "loss": 1.056,
      "step": 375
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.479707717895508,
      "learning_rate": 8.973747016706445e-06,
      "loss": 0.2284,
      "step": 376
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.961345672607422,
      "learning_rate": 8.997613365155132e-06,
      "loss": 1.2778,
      "step": 377
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.244306564331055,
      "learning_rate": 9.021479713603819e-06,
      "loss": 1.0705,
      "step": 378
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.863973140716553,
      "learning_rate": 9.045346062052508e-06,
      "loss": 1.0616,
      "step": 379
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.787870407104492,
      "learning_rate": 9.069212410501193e-06,
      "loss": 2.2412,
      "step": 380
    },
    {
      "epoch": 0.03,
      "grad_norm": 33.758758544921875,
      "learning_rate": 9.093078758949882e-06,
      "loss": 2.8302,
      "step": 381
    },
    {
      "epoch": 0.03,
      "grad_norm": 32.99989700317383,
      "learning_rate": 9.116945107398569e-06,
      "loss": 1.7407,
      "step": 382
    },
    {
      "epoch": 0.03,
      "grad_norm": 46.49549865722656,
      "learning_rate": 9.140811455847256e-06,
      "loss": 1.6715,
      "step": 383
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.282810211181641,
      "learning_rate": 9.164677804295943e-06,
      "loss": 0.2239,
      "step": 384
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.448434352874756,
      "learning_rate": 9.18854415274463e-06,
      "loss": 1.1658,
      "step": 385
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.066117286682129,
      "learning_rate": 9.212410501193319e-06,
      "loss": 0.9827,
      "step": 386
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.7246270179748535,
      "learning_rate": 9.236276849642005e-06,
      "loss": 0.2242,
      "step": 387
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.207185745239258,
      "learning_rate": 9.260143198090693e-06,
      "loss": 1.9758,
      "step": 388
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.516511917114258,
      "learning_rate": 9.28400954653938e-06,
      "loss": 2.0776,
      "step": 389
    },
    {
      "epoch": 0.03,
      "grad_norm": 53.98463821411133,
      "learning_rate": 9.307875894988068e-06,
      "loss": 1.0902,
      "step": 390
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.34937286376953,
      "learning_rate": 9.331742243436755e-06,
      "loss": 1.8317,
      "step": 391
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.498335361480713,
      "learning_rate": 9.355608591885442e-06,
      "loss": 0.1584,
      "step": 392
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.423630237579346,
      "learning_rate": 9.37947494033413e-06,
      "loss": 1.0493,
      "step": 393
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.903542995452881,
      "learning_rate": 9.403341288782816e-06,
      "loss": 0.2579,
      "step": 394
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.503958702087402,
      "learning_rate": 9.427207637231505e-06,
      "loss": 0.208,
      "step": 395
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.3379902839660645,
      "learning_rate": 9.451073985680192e-06,
      "loss": 0.1934,
      "step": 396
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.214580535888672,
      "learning_rate": 9.474940334128879e-06,
      "loss": 0.323,
      "step": 397
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.364828109741211,
      "learning_rate": 9.498806682577566e-06,
      "loss": 1.4643,
      "step": 398
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.730569839477539,
      "learning_rate": 9.522673031026253e-06,
      "loss": 0.2489,
      "step": 399
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.879016637802124,
      "learning_rate": 9.54653937947494e-06,
      "loss": 0.0948,
      "step": 400
    },
    {
      "epoch": 0.03,
      "grad_norm": 19.969961166381836,
      "learning_rate": 9.570405727923628e-06,
      "loss": 1.7092,
      "step": 401
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.267436981201172,
      "learning_rate": 9.594272076372316e-06,
      "loss": 1.4283,
      "step": 402
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.974484443664551,
      "learning_rate": 9.618138424821004e-06,
      "loss": 1.328,
      "step": 403
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.097031116485596,
      "learning_rate": 9.64200477326969e-06,
      "loss": 0.2503,
      "step": 404
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.788335800170898,
      "learning_rate": 9.665871121718378e-06,
      "loss": 0.2898,
      "step": 405
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.294631004333496,
      "learning_rate": 9.689737470167065e-06,
      "loss": 0.1741,
      "step": 406
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.27406644821167,
      "learning_rate": 9.713603818615752e-06,
      "loss": 1.1416,
      "step": 407
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.619866371154785,
      "learning_rate": 9.73747016706444e-06,
      "loss": 2.0094,
      "step": 408
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.192477703094482,
      "learning_rate": 9.761336515513126e-06,
      "loss": 0.1668,
      "step": 409
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.574115753173828,
      "learning_rate": 9.785202863961815e-06,
      "loss": 1.8927,
      "step": 410
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.609882354736328,
      "learning_rate": 9.809069212410502e-06,
      "loss": 0.1535,
      "step": 411
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.246999740600586,
      "learning_rate": 9.83293556085919e-06,
      "loss": 0.1526,
      "step": 412
    },
    {
      "epoch": 0.03,
      "grad_norm": 26.218774795532227,
      "learning_rate": 9.856801909307876e-06,
      "loss": 2.6511,
      "step": 413
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.023394584655762,
      "learning_rate": 9.880668257756564e-06,
      "loss": 1.0721,
      "step": 414
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.483069896697998,
      "learning_rate": 9.904534606205252e-06,
      "loss": 0.1873,
      "step": 415
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.1530375480651855,
      "learning_rate": 9.928400954653938e-06,
      "loss": 0.1216,
      "step": 416
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3828437328338623,
      "learning_rate": 9.952267303102627e-06,
      "loss": 0.1525,
      "step": 417
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.351907253265381,
      "learning_rate": 9.976133651551314e-06,
      "loss": 0.253,
      "step": 418
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.58116626739502,
      "learning_rate": 1e-05,
      "loss": 2.133,
      "step": 419
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.1378912925720215,
      "learning_rate": 9.999999865134465e-06,
      "loss": 0.2961,
      "step": 420
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.395329475402832,
      "learning_rate": 9.999999460537861e-06,
      "loss": 0.2389,
      "step": 421
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.361896514892578,
      "learning_rate": 9.999998786210212e-06,
      "loss": 1.6761,
      "step": 422
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.174637794494629,
      "learning_rate": 9.999997842151555e-06,
      "loss": 0.1379,
      "step": 423
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.032309532165527,
      "learning_rate": 9.99999662836194e-06,
      "loss": 0.2852,
      "step": 424
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.119866371154785,
      "learning_rate": 9.999995144841433e-06,
      "loss": 0.217,
      "step": 425
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.838905334472656,
      "learning_rate": 9.999993391590115e-06,
      "loss": 1.0335,
      "step": 426
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.847648620605469,
      "learning_rate": 9.999991368608079e-06,
      "loss": 2.0368,
      "step": 427
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.180704593658447,
      "learning_rate": 9.999989075895434e-06,
      "loss": 0.1938,
      "step": 428
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.486629486083984,
      "learning_rate": 9.999986513452305e-06,
      "loss": 2.0475,
      "step": 429
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.826103448867798,
      "learning_rate": 9.99998368127883e-06,
      "loss": 0.1356,
      "step": 430
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.770874500274658,
      "learning_rate": 9.999980579375158e-06,
      "loss": 1.0927,
      "step": 431
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.761422157287598,
      "learning_rate": 9.999977207741466e-06,
      "loss": 0.8926,
      "step": 432
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.713716506958008,
      "learning_rate": 9.999973566377923e-06,
      "loss": 1.1652,
      "step": 433
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.499587535858154,
      "learning_rate": 9.999969655284737e-06,
      "loss": 0.1154,
      "step": 434
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.783328533172607,
      "learning_rate": 9.99996547446211e-06,
      "loss": 1.277,
      "step": 435
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.972327709197998,
      "learning_rate": 9.999961023910274e-06,
      "loss": 0.1469,
      "step": 436
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.15032434463501,
      "learning_rate": 9.999956303629467e-06,
      "loss": 0.199,
      "step": 437
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.401660919189453,
      "learning_rate": 9.999951313619942e-06,
      "loss": 1.7581,
      "step": 438
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.273219585418701,
      "learning_rate": 9.999946053881971e-06,
      "loss": 1.1651,
      "step": 439
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.7883591651916504,
      "learning_rate": 9.999940524415835e-06,
      "loss": 0.1283,
      "step": 440
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.908879280090332,
      "learning_rate": 9.999934725221834e-06,
      "loss": 0.2115,
      "step": 441
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.620343208312988,
      "learning_rate": 9.99992865630028e-06,
      "loss": 1.8725,
      "step": 442
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.615207195281982,
      "learning_rate": 9.9999223176515e-06,
      "loss": 1.0918,
      "step": 443
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.749943256378174,
      "learning_rate": 9.999915709275839e-06,
      "loss": 0.2243,
      "step": 444
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.836407661437988,
      "learning_rate": 9.999908831173649e-06,
      "loss": 0.9391,
      "step": 445
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.880640506744385,
      "learning_rate": 9.999901683345304e-06,
      "loss": 0.1684,
      "step": 446
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.018899917602539,
      "learning_rate": 9.999894265791188e-06,
      "loss": 1.6973,
      "step": 447
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.404901504516602,
      "learning_rate": 9.999886578511703e-06,
      "loss": 1.6696,
      "step": 448
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.5608744621276855,
      "learning_rate": 9.999878621507263e-06,
      "loss": 0.232,
      "step": 449
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.625431537628174,
      "learning_rate": 9.999870394778294e-06,
      "loss": 0.2487,
      "step": 450
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.016963481903076,
      "learning_rate": 9.999861898325246e-06,
      "loss": 0.1538,
      "step": 451
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.771223068237305,
      "learning_rate": 9.999853132148573e-06,
      "loss": 0.165,
      "step": 452
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.838277339935303,
      "learning_rate": 9.99984409624875e-06,
      "loss": 1.0258,
      "step": 453
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.701510429382324,
      "learning_rate": 9.999834790626262e-06,
      "loss": 0.2308,
      "step": 454
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.39403772354126,
      "learning_rate": 9.999825215281613e-06,
      "loss": 1.127,
      "step": 455
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.033388614654541,
      "learning_rate": 9.99981537021532e-06,
      "loss": 0.243,
      "step": 456
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.917449951171875,
      "learning_rate": 9.999805255427912e-06,
      "loss": 1.7782,
      "step": 457
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.544135570526123,
      "learning_rate": 9.999794870919936e-06,
      "loss": 0.9476,
      "step": 458
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.1927947998046875,
      "learning_rate": 9.999784216691953e-06,
      "loss": 0.21,
      "step": 459
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.265624046325684,
      "learning_rate": 9.999773292744534e-06,
      "loss": 0.1365,
      "step": 460
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.19537353515625,
      "learning_rate": 9.999762099078274e-06,
      "loss": 2.0515,
      "step": 461
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.227529525756836,
      "learning_rate": 9.999750635693772e-06,
      "loss": 0.2136,
      "step": 462
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.684843063354492,
      "learning_rate": 9.999738902591648e-06,
      "loss": 1.9701,
      "step": 463
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.724134922027588,
      "learning_rate": 9.999726899772537e-06,
      "loss": 0.1435,
      "step": 464
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.326423645019531,
      "learning_rate": 9.999714627237084e-06,
      "loss": 0.2232,
      "step": 465
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.188990116119385,
      "learning_rate": 9.999702084985952e-06,
      "loss": 0.2072,
      "step": 466
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.342813491821289,
      "learning_rate": 9.999689273019817e-06,
      "loss": 0.2778,
      "step": 467
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.60174560546875,
      "learning_rate": 9.999676191339369e-06,
      "loss": 1.3344,
      "step": 468
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.890679359436035,
      "learning_rate": 9.999662839945317e-06,
      "loss": 0.2923,
      "step": 469
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.353227138519287,
      "learning_rate": 9.999649218838381e-06,
      "loss": 0.2046,
      "step": 470
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.4079365730285645,
      "learning_rate": 9.999635328019293e-06,
      "loss": 1.2815,
      "step": 471
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.909841060638428,
      "learning_rate": 9.999621167488803e-06,
      "loss": 1.0669,
      "step": 472
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.407723426818848,
      "learning_rate": 9.999606737247675e-06,
      "loss": 1.0736,
      "step": 473
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.538171291351318,
      "learning_rate": 9.99959203729669e-06,
      "loss": 0.3591,
      "step": 474
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.770170211791992,
      "learning_rate": 9.999577067636638e-06,
      "loss": 1.7383,
      "step": 475
    },
    {
      "epoch": 0.03,
      "grad_norm": 19.601173400878906,
      "learning_rate": 9.999561828268328e-06,
      "loss": 1.8125,
      "step": 476
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.8768792152404785,
      "learning_rate": 9.999546319192581e-06,
      "loss": 0.208,
      "step": 477
    },
    {
      "epoch": 0.03,
      "grad_norm": 19.513994216918945,
      "learning_rate": 9.999530540410236e-06,
      "loss": 2.2848,
      "step": 478
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.307207107543945,
      "learning_rate": 9.999514491922141e-06,
      "loss": 0.3313,
      "step": 479
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.025099277496338,
      "learning_rate": 9.999498173729164e-06,
      "loss": 0.1773,
      "step": 480
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.649541854858398,
      "learning_rate": 9.999481585832185e-06,
      "loss": 1.407,
      "step": 481
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.822442531585693,
      "learning_rate": 9.999464728232099e-06,
      "loss": 0.1524,
      "step": 482
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.594359874725342,
      "learning_rate": 9.999447600929813e-06,
      "loss": 0.3134,
      "step": 483
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.411062240600586,
      "learning_rate": 9.999430203926253e-06,
      "loss": 1.1027,
      "step": 484
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.858865261077881,
      "learning_rate": 9.99941253722236e-06,
      "loss": 0.2376,
      "step": 485
    },
    {
      "epoch": 0.03,
      "grad_norm": 24.492767333984375,
      "learning_rate": 9.999394600819082e-06,
      "loss": 1.7162,
      "step": 486
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.971532821655273,
      "learning_rate": 9.99937639471739e-06,
      "loss": 1.7843,
      "step": 487
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.038531303405762,
      "learning_rate": 9.999357918918264e-06,
      "loss": 0.1765,
      "step": 488
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.292446613311768,
      "learning_rate": 9.999339173422702e-06,
      "loss": 0.2842,
      "step": 489
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.758338451385498,
      "learning_rate": 9.999320158231717e-06,
      "loss": 0.2384,
      "step": 490
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8782176971435547,
      "learning_rate": 9.999300873346332e-06,
      "loss": 0.2139,
      "step": 491
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.061819076538086,
      "learning_rate": 9.999281318767588e-06,
      "loss": 0.135,
      "step": 492
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.054653167724609,
      "learning_rate": 9.99926149449654e-06,
      "loss": 1.1357,
      "step": 493
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.028374195098877,
      "learning_rate": 9.999241400534257e-06,
      "loss": 1.2402,
      "step": 494
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.445295810699463,
      "learning_rate": 9.999221036881825e-06,
      "loss": 0.1934,
      "step": 495
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.288728713989258,
      "learning_rate": 9.99920040354034e-06,
      "loss": 0.2262,
      "step": 496
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.836421966552734,
      "learning_rate": 9.999179500510916e-06,
      "loss": 1.7585,
      "step": 497
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.427230834960938,
      "learning_rate": 9.999158327794682e-06,
      "loss": 1.8533,
      "step": 498
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.81179428100586,
      "learning_rate": 9.99913688539278e-06,
      "loss": 2.268,
      "step": 499
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.1238017082214355,
      "learning_rate": 9.999115173306365e-06,
      "loss": 1.0947,
      "step": 500
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.152070999145508,
      "learning_rate": 9.999093191536607e-06,
      "loss": 0.9904,
      "step": 501
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.215607166290283,
      "learning_rate": 9.999070940084697e-06,
      "loss": 1.2034,
      "step": 502
    },
    {
      "epoch": 0.04,
      "grad_norm": 50.35987091064453,
      "learning_rate": 9.99904841895183e-06,
      "loss": 1.5343,
      "step": 503
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.718503952026367,
      "learning_rate": 9.999025628139224e-06,
      "loss": 1.4921,
      "step": 504
    },
    {
      "epoch": 0.04,
      "grad_norm": 22.12445640563965,
      "learning_rate": 9.999002567648109e-06,
      "loss": 1.6534,
      "step": 505
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.374300956726074,
      "learning_rate": 9.998979237479725e-06,
      "loss": 0.2074,
      "step": 506
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.032337665557861,
      "learning_rate": 9.998955637635334e-06,
      "loss": 0.3449,
      "step": 507
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.701272010803223,
      "learning_rate": 9.99893176811621e-06,
      "loss": 0.2011,
      "step": 508
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.746965408325195,
      "learning_rate": 9.998907628923638e-06,
      "loss": 1.0603,
      "step": 509
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.00566577911377,
      "learning_rate": 9.998883220058921e-06,
      "loss": 1.6287,
      "step": 510
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.642836093902588,
      "learning_rate": 9.998858541523378e-06,
      "loss": 0.222,
      "step": 511
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.166630744934082,
      "learning_rate": 9.998833593318336e-06,
      "loss": 0.1602,
      "step": 512
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.741205215454102,
      "learning_rate": 9.998808375445145e-06,
      "loss": 1.3624,
      "step": 513
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.442261219024658,
      "learning_rate": 9.998782887905162e-06,
      "loss": 0.2288,
      "step": 514
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.906387805938721,
      "learning_rate": 9.998757130699766e-06,
      "loss": 0.3257,
      "step": 515
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.171483993530273,
      "learning_rate": 9.998731103830342e-06,
      "loss": 1.3432,
      "step": 516
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.9177446365356445,
      "learning_rate": 9.998704807298297e-06,
      "loss": 0.3075,
      "step": 517
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.700723648071289,
      "learning_rate": 9.998678241105048e-06,
      "loss": 1.4842,
      "step": 518
    },
    {
      "epoch": 0.04,
      "grad_norm": 22.469806671142578,
      "learning_rate": 9.99865140525203e-06,
      "loss": 2.2017,
      "step": 519
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.651871681213379,
      "learning_rate": 9.99862429974069e-06,
      "loss": 1.234,
      "step": 520
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.03162956237793,
      "learning_rate": 9.998596924572488e-06,
      "loss": 0.2336,
      "step": 521
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.496156215667725,
      "learning_rate": 9.998569279748904e-06,
      "loss": 0.2892,
      "step": 522
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.244393348693848,
      "learning_rate": 9.998541365271428e-06,
      "loss": 1.1145,
      "step": 523
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.342768669128418,
      "learning_rate": 9.998513181141565e-06,
      "loss": 0.2251,
      "step": 524
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.642395973205566,
      "learning_rate": 9.998484727360836e-06,
      "loss": 0.9357,
      "step": 525
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.027926445007324,
      "learning_rate": 9.998456003930777e-06,
      "loss": 1.5048,
      "step": 526
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.807500123977661,
      "learning_rate": 9.998427010852938e-06,
      "loss": 0.2233,
      "step": 527
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.211508750915527,
      "learning_rate": 9.99839774812888e-06,
      "loss": 1.5903,
      "step": 528
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.53528118133545,
      "learning_rate": 9.998368215760184e-06,
      "loss": 1.3684,
      "step": 529
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.556333541870117,
      "learning_rate": 9.998338413748443e-06,
      "loss": 1.2561,
      "step": 530
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.781978130340576,
      "learning_rate": 9.998308342095265e-06,
      "loss": 0.213,
      "step": 531
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.280364513397217,
      "learning_rate": 9.99827800080227e-06,
      "loss": 1.1797,
      "step": 532
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.736016273498535,
      "learning_rate": 9.998247389871096e-06,
      "loss": 1.2814,
      "step": 533
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.360488891601562,
      "learning_rate": 9.998216509303395e-06,
      "loss": 1.8661,
      "step": 534
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.738839149475098,
      "learning_rate": 9.998185359100834e-06,
      "loss": 1.2401,
      "step": 535
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.397369384765625,
      "learning_rate": 9.998153939265091e-06,
      "loss": 1.3451,
      "step": 536
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.451443195343018,
      "learning_rate": 9.998122249797863e-06,
      "loss": 1.0511,
      "step": 537
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.214555740356445,
      "learning_rate": 9.998090290700856e-06,
      "loss": 0.1411,
      "step": 538
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6544442176818848,
      "learning_rate": 9.9980580619758e-06,
      "loss": 0.23,
      "step": 539
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.920378684997559,
      "learning_rate": 9.998025563624429e-06,
      "loss": 1.4408,
      "step": 540
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9491682052612305,
      "learning_rate": 9.997992795648496e-06,
      "loss": 0.2706,
      "step": 541
    },
    {
      "epoch": 0.04,
      "grad_norm": 30.137548446655273,
      "learning_rate": 9.99795975804977e-06,
      "loss": 2.045,
      "step": 542
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.237855911254883,
      "learning_rate": 9.997926450830036e-06,
      "loss": 0.227,
      "step": 543
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.536004066467285,
      "learning_rate": 9.997892873991086e-06,
      "loss": 1.3096,
      "step": 544
    },
    {
      "epoch": 0.04,
      "grad_norm": 28.957536697387695,
      "learning_rate": 9.997859027534735e-06,
      "loss": 1.2615,
      "step": 545
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.1401143074035645,
      "learning_rate": 9.997824911462805e-06,
      "loss": 0.3312,
      "step": 546
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.519582748413086,
      "learning_rate": 9.997790525777142e-06,
      "loss": 0.2848,
      "step": 547
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.277632713317871,
      "learning_rate": 9.997755870479595e-06,
      "loss": 2.0582,
      "step": 548
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.987781047821045,
      "learning_rate": 9.997720945572038e-06,
      "loss": 1.2578,
      "step": 549
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.882593154907227,
      "learning_rate": 9.997685751056354e-06,
      "loss": 0.3465,
      "step": 550
    },
    {
      "epoch": 0.04,
      "grad_norm": 51.11644744873047,
      "learning_rate": 9.99765028693444e-06,
      "loss": 2.9628,
      "step": 551
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.241851806640625,
      "learning_rate": 9.99761455320821e-06,
      "loss": 1.1988,
      "step": 552
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.781780242919922,
      "learning_rate": 9.997578549879592e-06,
      "loss": 1.7467,
      "step": 553
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.365095138549805,
      "learning_rate": 9.997542276950528e-06,
      "loss": 1.4466,
      "step": 554
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.083973407745361,
      "learning_rate": 9.997505734422974e-06,
      "loss": 0.176,
      "step": 555
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.662355422973633,
      "learning_rate": 9.997468922298904e-06,
      "loss": 0.1913,
      "step": 556
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.886309862136841,
      "learning_rate": 9.997431840580301e-06,
      "loss": 0.1477,
      "step": 557
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.774263381958008,
      "learning_rate": 9.997394489269165e-06,
      "loss": 0.975,
      "step": 558
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.394223213195801,
      "learning_rate": 9.997356868367517e-06,
      "loss": 1.0703,
      "step": 559
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.590246677398682,
      "learning_rate": 9.997318977877377e-06,
      "loss": 0.1528,
      "step": 560
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4274494647979736,
      "learning_rate": 9.997280817800797e-06,
      "loss": 0.106,
      "step": 561
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.604013442993164,
      "learning_rate": 9.997242388139832e-06,
      "loss": 1.9276,
      "step": 562
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.248636245727539,
      "learning_rate": 9.997203688896554e-06,
      "loss": 1.6055,
      "step": 563
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.882675647735596,
      "learning_rate": 9.997164720073055e-06,
      "loss": 0.9898,
      "step": 564
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.007952690124512,
      "learning_rate": 9.997125481671433e-06,
      "loss": 1.4911,
      "step": 565
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.974711894989014,
      "learning_rate": 9.997085973693807e-06,
      "loss": 1.1609,
      "step": 566
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.896945476531982,
      "learning_rate": 9.997046196142309e-06,
      "loss": 0.3133,
      "step": 567
    },
    {
      "epoch": 0.04,
      "grad_norm": 27.517601013183594,
      "learning_rate": 9.997006149019081e-06,
      "loss": 1.4444,
      "step": 568
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.269046783447266,
      "learning_rate": 9.996965832326287e-06,
      "loss": 1.8182,
      "step": 569
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.698917865753174,
      "learning_rate": 9.996925246066101e-06,
      "loss": 1.0796,
      "step": 570
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.908824920654297,
      "learning_rate": 9.996884390240713e-06,
      "loss": 1.6264,
      "step": 571
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.593176364898682,
      "learning_rate": 9.996843264852327e-06,
      "loss": 0.3198,
      "step": 572
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.95229434967041,
      "learning_rate": 9.996801869903159e-06,
      "loss": 0.2457,
      "step": 573
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.751264572143555,
      "learning_rate": 9.996760205395445e-06,
      "loss": 1.1627,
      "step": 574
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.735759735107422,
      "learning_rate": 9.99671827133143e-06,
      "loss": 1.1319,
      "step": 575
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.215450286865234,
      "learning_rate": 9.99667606771338e-06,
      "loss": 1.8104,
      "step": 576
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.561466217041016,
      "learning_rate": 9.99663359454357e-06,
      "loss": 1.3643,
      "step": 577
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.602389812469482,
      "learning_rate": 9.996590851824289e-06,
      "loss": 0.182,
      "step": 578
    },
    {
      "epoch": 0.04,
      "grad_norm": 49.395877838134766,
      "learning_rate": 9.996547839557846e-06,
      "loss": 1.5564,
      "step": 579
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.994291305541992,
      "learning_rate": 9.99650455774656e-06,
      "loss": 1.2315,
      "step": 580
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.689165115356445,
      "learning_rate": 9.996461006392764e-06,
      "loss": 1.9498,
      "step": 581
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.723715305328369,
      "learning_rate": 9.996417185498812e-06,
      "loss": 0.1986,
      "step": 582
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.4522576332092285,
      "learning_rate": 9.996373095067064e-06,
      "loss": 1.148,
      "step": 583
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.637054443359375,
      "learning_rate": 9.9963287350999e-06,
      "loss": 1.9338,
      "step": 584
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.396203994750977,
      "learning_rate": 9.996284105599712e-06,
      "loss": 1.4999,
      "step": 585
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.64026165008545,
      "learning_rate": 9.996239206568909e-06,
      "loss": 0.2746,
      "step": 586
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7909228801727295,
      "learning_rate": 9.996194038009914e-06,
      "loss": 0.1885,
      "step": 587
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.838239669799805,
      "learning_rate": 9.996148599925161e-06,
      "loss": 1.2452,
      "step": 588
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.005929946899414,
      "learning_rate": 9.996102892317102e-06,
      "loss": 1.2567,
      "step": 589
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5360753536224365,
      "learning_rate": 9.996056915188204e-06,
      "loss": 0.2093,
      "step": 590
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.087030410766602,
      "learning_rate": 9.996010668540946e-06,
      "loss": 1.3953,
      "step": 591
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.535155296325684,
      "learning_rate": 9.995964152377824e-06,
      "loss": 0.2747,
      "step": 592
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.817758798599243,
      "learning_rate": 9.995917366701346e-06,
      "loss": 0.2682,
      "step": 593
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.874335289001465,
      "learning_rate": 9.995870311514037e-06,
      "loss": 1.2432,
      "step": 594
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.6156487464904785,
      "learning_rate": 9.995822986818436e-06,
      "loss": 0.2832,
      "step": 595
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.699280023574829,
      "learning_rate": 9.995775392617095e-06,
      "loss": 0.1578,
      "step": 596
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.946338176727295,
      "learning_rate": 9.99572752891258e-06,
      "loss": 0.3329,
      "step": 597
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.444000244140625,
      "learning_rate": 9.995679395707476e-06,
      "loss": 1.882,
      "step": 598
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.583696365356445,
      "learning_rate": 9.995630993004376e-06,
      "loss": 0.2162,
      "step": 599
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.415683269500732,
      "learning_rate": 9.995582320805896e-06,
      "loss": 0.2598,
      "step": 600
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.632563591003418,
      "learning_rate": 9.995533379114659e-06,
      "loss": 0.1675,
      "step": 601
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.13376522064209,
      "learning_rate": 9.995484167933305e-06,
      "loss": 0.2281,
      "step": 602
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.729438304901123,
      "learning_rate": 9.995434687264489e-06,
      "loss": 1.1139,
      "step": 603
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.511749267578125,
      "learning_rate": 9.99538493711088e-06,
      "loss": 1.732,
      "step": 604
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5875072479248047,
      "learning_rate": 9.995334917475162e-06,
      "loss": 0.1474,
      "step": 605
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.212520122528076,
      "learning_rate": 9.995284628360033e-06,
      "loss": 0.2638,
      "step": 606
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.43911600112915,
      "learning_rate": 9.995234069768208e-06,
      "loss": 0.309,
      "step": 607
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.162487506866455,
      "learning_rate": 9.995183241702412e-06,
      "loss": 0.2281,
      "step": 608
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.635001182556152,
      "learning_rate": 9.99513214416539e-06,
      "loss": 1.3266,
      "step": 609
    },
    {
      "epoch": 0.04,
      "grad_norm": 32.06420135498047,
      "learning_rate": 9.995080777159894e-06,
      "loss": 1.5084,
      "step": 610
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.082242250442505,
      "learning_rate": 9.995029140688697e-06,
      "loss": 0.1461,
      "step": 611
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.221163749694824,
      "learning_rate": 9.994977234754585e-06,
      "loss": 1.2378,
      "step": 612
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.016908645629883,
      "learning_rate": 9.99492505936036e-06,
      "loss": 0.2497,
      "step": 613
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.641006469726562,
      "learning_rate": 9.994872614508835e-06,
      "loss": 1.1127,
      "step": 614
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.475832939147949,
      "learning_rate": 9.994819900202838e-06,
      "loss": 0.3381,
      "step": 615
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.777936935424805,
      "learning_rate": 9.994766916445215e-06,
      "loss": 1.9023,
      "step": 616
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.477105140686035,
      "learning_rate": 9.994713663238822e-06,
      "loss": 0.2643,
      "step": 617
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.250126838684082,
      "learning_rate": 9.994660140586532e-06,
      "loss": 1.0046,
      "step": 618
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.889604091644287,
      "learning_rate": 9.994606348491236e-06,
      "loss": 0.1561,
      "step": 619
    },
    {
      "epoch": 0.04,
      "grad_norm": 26.26430320739746,
      "learning_rate": 9.99455228695583e-06,
      "loss": 1.6493,
      "step": 620
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.333027839660645,
      "learning_rate": 9.994497955983235e-06,
      "loss": 1.1391,
      "step": 621
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.827508926391602,
      "learning_rate": 9.994443355576379e-06,
      "loss": 1.0043,
      "step": 622
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.010416030883789,
      "learning_rate": 9.99438848573821e-06,
      "loss": 0.2735,
      "step": 623
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.9342041015625,
      "learning_rate": 9.99433334647169e-06,
      "loss": 1.7452,
      "step": 624
    },
    {
      "epoch": 0.04,
      "grad_norm": 16.563247680664062,
      "learning_rate": 9.994277937779787e-06,
      "loss": 1.577,
      "step": 625
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9400484561920166,
      "learning_rate": 9.994222259665494e-06,
      "loss": 0.1318,
      "step": 626
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.80966567993164,
      "learning_rate": 9.994166312131814e-06,
      "loss": 1.8585,
      "step": 627
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.199241638183594,
      "learning_rate": 9.994110095181767e-06,
      "loss": 0.2544,
      "step": 628
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.409934997558594,
      "learning_rate": 9.994053608818383e-06,
      "loss": 0.205,
      "step": 629
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.060593605041504,
      "learning_rate": 9.993996853044711e-06,
      "loss": 1.5218,
      "step": 630
    },
    {
      "epoch": 0.05,
      "grad_norm": 39.41031265258789,
      "learning_rate": 9.993939827863812e-06,
      "loss": 2.0559,
      "step": 631
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.5635905265808105,
      "learning_rate": 9.993882533278762e-06,
      "loss": 0.3927,
      "step": 632
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.1053385734558105,
      "learning_rate": 9.993824969292651e-06,
      "loss": 1.285,
      "step": 633
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.351515293121338,
      "learning_rate": 9.993767135908587e-06,
      "loss": 0.3244,
      "step": 634
    },
    {
      "epoch": 0.05,
      "grad_norm": 27.981176376342773,
      "learning_rate": 9.99370903312969e-06,
      "loss": 1.5729,
      "step": 635
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.729844570159912,
      "learning_rate": 9.99365066095909e-06,
      "loss": 0.2217,
      "step": 636
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.988632917404175,
      "learning_rate": 9.99359201939994e-06,
      "loss": 0.1014,
      "step": 637
    },
    {
      "epoch": 0.05,
      "grad_norm": 24.083620071411133,
      "learning_rate": 9.993533108455402e-06,
      "loss": 1.689,
      "step": 638
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.625862121582031,
      "learning_rate": 9.993473928128655e-06,
      "loss": 1.1669,
      "step": 639
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6924240589141846,
      "learning_rate": 9.99341447842289e-06,
      "loss": 0.1657,
      "step": 640
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.342416286468506,
      "learning_rate": 9.993354759341316e-06,
      "loss": 1.0518,
      "step": 641
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.255677700042725,
      "learning_rate": 9.993294770887153e-06,
      "loss": 1.0483,
      "step": 642
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.959400177001953,
      "learning_rate": 9.993234513063639e-06,
      "loss": 1.6289,
      "step": 643
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.620476245880127,
      "learning_rate": 9.993173985874021e-06,
      "loss": 0.1939,
      "step": 644
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.707299709320068,
      "learning_rate": 9.993113189321569e-06,
      "loss": 1.0175,
      "step": 645
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.16036319732666,
      "learning_rate": 9.99305212340956e-06,
      "loss": 0.2007,
      "step": 646
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.453094959259033,
      "learning_rate": 9.992990788141288e-06,
      "loss": 0.2418,
      "step": 647
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.710414409637451,
      "learning_rate": 9.992929183520062e-06,
      "loss": 0.2652,
      "step": 648
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.467668533325195,
      "learning_rate": 9.992867309549206e-06,
      "loss": 1.1594,
      "step": 649
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.567554473876953,
      "learning_rate": 9.992805166232058e-06,
      "loss": 2.1235,
      "step": 650
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.905242919921875,
      "learning_rate": 9.992742753571969e-06,
      "loss": 0.1388,
      "step": 651
    },
    {
      "epoch": 0.05,
      "grad_norm": 24.253725051879883,
      "learning_rate": 9.992680071572309e-06,
      "loss": 2.2881,
      "step": 652
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.505508899688721,
      "learning_rate": 9.992617120236457e-06,
      "loss": 0.1794,
      "step": 653
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.4277777671813965,
      "learning_rate": 9.992553899567809e-06,
      "loss": 0.9965,
      "step": 654
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.272252559661865,
      "learning_rate": 9.992490409569775e-06,
      "loss": 0.2905,
      "step": 655
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.571274757385254,
      "learning_rate": 9.992426650245783e-06,
      "loss": 0.1666,
      "step": 656
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.797722339630127,
      "learning_rate": 9.992362621599269e-06,
      "loss": 0.2091,
      "step": 657
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2274999618530273,
      "learning_rate": 9.992298323633688e-06,
      "loss": 0.154,
      "step": 658
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.818232536315918,
      "learning_rate": 9.992233756352511e-06,
      "loss": 0.9458,
      "step": 659
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.83664894104004,
      "learning_rate": 9.99216891975922e-06,
      "loss": 2.3688,
      "step": 660
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.167415618896484,
      "learning_rate": 9.99210381385731e-06,
      "loss": 1.1702,
      "step": 661
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.656370162963867,
      "learning_rate": 9.992038438650297e-06,
      "loss": 1.6032,
      "step": 662
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.334367752075195,
      "learning_rate": 9.991972794141706e-06,
      "loss": 1.5637,
      "step": 663
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.867382049560547,
      "learning_rate": 9.991906880335079e-06,
      "loss": 1.1465,
      "step": 664
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.2552547454834,
      "learning_rate": 9.99184069723397e-06,
      "loss": 1.6341,
      "step": 665
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.664504528045654,
      "learning_rate": 9.99177424484195e-06,
      "loss": 0.1943,
      "step": 666
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.37862491607666,
      "learning_rate": 9.991707523162606e-06,
      "loss": 0.1258,
      "step": 667
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.024679660797119,
      "learning_rate": 9.991640532199534e-06,
      "loss": 0.9667,
      "step": 668
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.42310333251953,
      "learning_rate": 9.99157327195635e-06,
      "loss": 1.7451,
      "step": 669
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.038665771484375,
      "learning_rate": 9.991505742436683e-06,
      "loss": 1.1731,
      "step": 670
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.313358783721924,
      "learning_rate": 9.991437943644175e-06,
      "loss": 1.1112,
      "step": 671
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.366402626037598,
      "learning_rate": 9.991369875582483e-06,
      "loss": 1.2019,
      "step": 672
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.668091773986816,
      "learning_rate": 9.99130153825528e-06,
      "loss": 1.2501,
      "step": 673
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.22758150100708,
      "learning_rate": 9.991232931666253e-06,
      "loss": 0.4108,
      "step": 674
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.195480346679688,
      "learning_rate": 9.9911640558191e-06,
      "loss": 1.4873,
      "step": 675
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.469003677368164,
      "learning_rate": 9.99109491071754e-06,
      "loss": 0.1381,
      "step": 676
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.901341915130615,
      "learning_rate": 9.991025496365303e-06,
      "loss": 1.2024,
      "step": 677
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.740177631378174,
      "learning_rate": 9.990955812766133e-06,
      "loss": 0.2776,
      "step": 678
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.519429206848145,
      "learning_rate": 9.990885859923786e-06,
      "loss": 1.5248,
      "step": 679
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.951432228088379,
      "learning_rate": 9.990815637842039e-06,
      "loss": 0.1388,
      "step": 680
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.581445693969727,
      "learning_rate": 9.990745146524681e-06,
      "loss": 1.6517,
      "step": 681
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.352225303649902,
      "learning_rate": 9.990674385975512e-06,
      "loss": 1.0401,
      "step": 682
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.231869697570801,
      "learning_rate": 9.990603356198351e-06,
      "loss": 0.1259,
      "step": 683
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.214491844177246,
      "learning_rate": 9.99053205719703e-06,
      "loss": 2.0756,
      "step": 684
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.879469394683838,
      "learning_rate": 9.990460488975392e-06,
      "loss": 0.9557,
      "step": 685
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.876859188079834,
      "learning_rate": 9.990388651537303e-06,
      "loss": 1.1372,
      "step": 686
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.841719627380371,
      "learning_rate": 9.990316544886636e-06,
      "loss": 1.4274,
      "step": 687
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7075657844543457,
      "learning_rate": 9.990244169027281e-06,
      "loss": 0.1543,
      "step": 688
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.210724830627441,
      "learning_rate": 9.990171523963141e-06,
      "loss": 0.2487,
      "step": 689
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.253916263580322,
      "learning_rate": 9.990098609698136e-06,
      "loss": 1.0625,
      "step": 690
    },
    {
      "epoch": 0.05,
      "grad_norm": 20.4244384765625,
      "learning_rate": 9.9900254262362e-06,
      "loss": 1.6104,
      "step": 691
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.546996593475342,
      "learning_rate": 9.989951973581283e-06,
      "loss": 0.2003,
      "step": 692
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.3575029373168945,
      "learning_rate": 9.989878251737342e-06,
      "loss": 0.2004,
      "step": 693
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.54151725769043,
      "learning_rate": 9.989804260708359e-06,
      "loss": 1.0176,
      "step": 694
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.445076942443848,
      "learning_rate": 9.98973000049832e-06,
      "loss": 0.3756,
      "step": 695
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.694364547729492,
      "learning_rate": 9.989655471111237e-06,
      "loss": 1.5444,
      "step": 696
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.857968807220459,
      "learning_rate": 9.98958067255113e-06,
      "loss": 0.1206,
      "step": 697
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.104480743408203,
      "learning_rate": 9.98950560482203e-06,
      "loss": 0.1257,
      "step": 698
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.248236656188965,
      "learning_rate": 9.989430267927988e-06,
      "loss": 1.1178,
      "step": 699
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.854472637176514,
      "learning_rate": 9.989354661873072e-06,
      "loss": 1.0622,
      "step": 700
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.317495822906494,
      "learning_rate": 9.989278786661356e-06,
      "loss": 1.1411,
      "step": 701
    },
    {
      "epoch": 0.05,
      "grad_norm": 27.90595054626465,
      "learning_rate": 9.989202642296935e-06,
      "loss": 1.45,
      "step": 702
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.241791725158691,
      "learning_rate": 9.989126228783916e-06,
      "loss": 1.4871,
      "step": 703
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.471338272094727,
      "learning_rate": 9.989049546126422e-06,
      "loss": 1.189,
      "step": 704
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.80232310295105,
      "learning_rate": 9.988972594328591e-06,
      "loss": 0.15,
      "step": 705
    },
    {
      "epoch": 0.05,
      "grad_norm": 24.64083480834961,
      "learning_rate": 9.988895373394571e-06,
      "loss": 1.6135,
      "step": 706
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.577859401702881,
      "learning_rate": 9.988817883328532e-06,
      "loss": 1.009,
      "step": 707
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.02918243408203,
      "learning_rate": 9.988740124134651e-06,
      "loss": 1.4388,
      "step": 708
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.918377876281738,
      "learning_rate": 9.988662095817122e-06,
      "loss": 1.1604,
      "step": 709
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.649811744689941,
      "learning_rate": 9.988583798380157e-06,
      "loss": 1.3826,
      "step": 710
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.502440452575684,
      "learning_rate": 9.988505231827979e-06,
      "loss": 1.1332,
      "step": 711
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.46412992477417,
      "learning_rate": 9.988426396164826e-06,
      "loss": 1.0501,
      "step": 712
    },
    {
      "epoch": 0.05,
      "grad_norm": 25.15907096862793,
      "learning_rate": 9.988347291394952e-06,
      "loss": 1.4507,
      "step": 713
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.91136360168457,
      "learning_rate": 9.988267917522623e-06,
      "loss": 1.3571,
      "step": 714
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.984882354736328,
      "learning_rate": 9.988188274552122e-06,
      "loss": 0.2464,
      "step": 715
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.094900131225586,
      "learning_rate": 9.988108362487743e-06,
      "loss": 1.354,
      "step": 716
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.999746322631836,
      "learning_rate": 9.9880281813338e-06,
      "loss": 2.1819,
      "step": 717
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.422212600708008,
      "learning_rate": 9.987947731094618e-06,
      "loss": 1.0846,
      "step": 718
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.598791599273682,
      "learning_rate": 9.987867011774537e-06,
      "loss": 0.2611,
      "step": 719
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.020877361297607,
      "learning_rate": 9.987786023377909e-06,
      "loss": 0.1777,
      "step": 720
    },
    {
      "epoch": 0.05,
      "grad_norm": 28.770841598510742,
      "learning_rate": 9.987704765909106e-06,
      "loss": 2.1119,
      "step": 721
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.597056865692139,
      "learning_rate": 9.98762323937251e-06,
      "loss": 0.2219,
      "step": 722
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.239082336425781,
      "learning_rate": 9.987541443772519e-06,
      "loss": 0.3711,
      "step": 723
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.400554656982422,
      "learning_rate": 9.987459379113546e-06,
      "loss": 0.366,
      "step": 724
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.942784309387207,
      "learning_rate": 9.987377045400018e-06,
      "loss": 1.8887,
      "step": 725
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.607375383377075,
      "learning_rate": 9.987294442636378e-06,
      "loss": 0.1909,
      "step": 726
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.947427749633789,
      "learning_rate": 9.98721157082708e-06,
      "loss": 1.1203,
      "step": 727
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.677373170852661,
      "learning_rate": 9.987128429976595e-06,
      "loss": 0.2707,
      "step": 728
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.714303016662598,
      "learning_rate": 9.987045020089408e-06,
      "loss": 1.0495,
      "step": 729
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.384537696838379,
      "learning_rate": 9.98696134117002e-06,
      "loss": 0.1848,
      "step": 730
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.096622467041016,
      "learning_rate": 9.986877393222945e-06,
      "loss": 1.9373,
      "step": 731
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.414405345916748,
      "learning_rate": 9.98679317625271e-06,
      "loss": 0.2365,
      "step": 732
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7510409355163574,
      "learning_rate": 9.986708690263858e-06,
      "loss": 0.1329,
      "step": 733
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.2837982177734375,
      "learning_rate": 9.98662393526095e-06,
      "loss": 0.1892,
      "step": 734
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.110679626464844,
      "learning_rate": 9.986538911248555e-06,
      "loss": 1.7264,
      "step": 735
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.420869827270508,
      "learning_rate": 9.98645361823126e-06,
      "loss": 1.7103,
      "step": 736
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.171655654907227,
      "learning_rate": 9.986368056213669e-06,
      "loss": 1.7027,
      "step": 737
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.958489418029785,
      "learning_rate": 9.986282225200394e-06,
      "loss": 1.5011,
      "step": 738
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.577925682067871,
      "learning_rate": 9.986196125196067e-06,
      "loss": 0.2939,
      "step": 739
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.8596951961517334,
      "learning_rate": 9.986109756205335e-06,
      "loss": 0.1315,
      "step": 740
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.855889320373535,
      "learning_rate": 9.986023118232855e-06,
      "loss": 1.0408,
      "step": 741
    },
    {
      "epoch": 0.05,
      "grad_norm": 25.907487869262695,
      "learning_rate": 9.9859362112833e-06,
      "loss": 2.1812,
      "step": 742
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.124670028686523,
      "learning_rate": 9.98584903536136e-06,
      "loss": 1.7373,
      "step": 743
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.408076763153076,
      "learning_rate": 9.985761590471735e-06,
      "loss": 1.1412,
      "step": 744
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.207207202911377,
      "learning_rate": 9.985673876619143e-06,
      "loss": 0.2029,
      "step": 745
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.756026268005371,
      "learning_rate": 9.985585893808321e-06,
      "loss": 0.2339,
      "step": 746
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.069007396697998,
      "learning_rate": 9.98549764204401e-06,
      "loss": 0.2676,
      "step": 747
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.700265407562256,
      "learning_rate": 9.985409121330972e-06,
      "loss": 0.2306,
      "step": 748
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.058094024658203,
      "learning_rate": 9.985320331673983e-06,
      "loss": 0.194,
      "step": 749
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.757126569747925,
      "learning_rate": 9.985231273077834e-06,
      "loss": 0.1267,
      "step": 750
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.27551555633545,
      "learning_rate": 9.985141945547325e-06,
      "loss": 1.6213,
      "step": 751
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.464486122131348,
      "learning_rate": 9.98505234908728e-06,
      "loss": 0.2386,
      "step": 752
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.606345176696777,
      "learning_rate": 9.98496248370253e-06,
      "loss": 1.5603,
      "step": 753
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.565373420715332,
      "learning_rate": 9.984872349397923e-06,
      "loss": 0.1985,
      "step": 754
    },
    {
      "epoch": 0.05,
      "grad_norm": 43.823081970214844,
      "learning_rate": 9.98478194617832e-06,
      "loss": 1.7207,
      "step": 755
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.527823448181152,
      "learning_rate": 9.9846912740486e-06,
      "loss": 2.0421,
      "step": 756
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.358763694763184,
      "learning_rate": 9.984600333013655e-06,
      "loss": 0.2002,
      "step": 757
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.208784103393555,
      "learning_rate": 9.98450912307839e-06,
      "loss": 1.2622,
      "step": 758
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.0232930183410645,
      "learning_rate": 9.984417644247724e-06,
      "loss": 1.1205,
      "step": 759
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.8283281326293945,
      "learning_rate": 9.984325896526594e-06,
      "loss": 1.1872,
      "step": 760
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.852667808532715,
      "learning_rate": 9.984233879919947e-06,
      "loss": 1.6678,
      "step": 761
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.460629940032959,
      "learning_rate": 9.98414159443275e-06,
      "loss": 0.3057,
      "step": 762
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.06594467163086,
      "learning_rate": 9.984049040069981e-06,
      "loss": 1.6608,
      "step": 763
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.32802677154541,
      "learning_rate": 9.98395621683663e-06,
      "loss": 1.0419,
      "step": 764
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.386711597442627,
      "learning_rate": 9.983863124737707e-06,
      "loss": 0.1371,
      "step": 765
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6262950897216797,
      "learning_rate": 9.983769763778232e-06,
      "loss": 0.1781,
      "step": 766
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.781742095947266,
      "learning_rate": 9.983676133963245e-06,
      "loss": 1.1593,
      "step": 767
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.346411228179932,
      "learning_rate": 9.983582235297794e-06,
      "loss": 0.2022,
      "step": 768
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.691783905029297,
      "learning_rate": 9.983488067786946e-06,
      "loss": 0.8844,
      "step": 769
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.43714141845703,
      "learning_rate": 9.983393631435778e-06,
      "loss": 1.7018,
      "step": 770
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.011190891265869,
      "learning_rate": 9.983298926249389e-06,
      "loss": 1.0916,
      "step": 771
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.045363426208496,
      "learning_rate": 9.983203952232885e-06,
      "loss": 1.2107,
      "step": 772
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4354732036590576,
      "learning_rate": 9.983108709391392e-06,
      "loss": 0.2127,
      "step": 773
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.923701286315918,
      "learning_rate": 9.983013197730043e-06,
      "loss": 0.9397,
      "step": 774
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4263806343078613,
      "learning_rate": 9.982917417253995e-06,
      "loss": 0.1844,
      "step": 775
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.627791881561279,
      "learning_rate": 9.982821367968413e-06,
      "loss": 0.8902,
      "step": 776
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.219080448150635,
      "learning_rate": 9.98272504987848e-06,
      "loss": 0.9941,
      "step": 777
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.792088031768799,
      "learning_rate": 9.982628462989393e-06,
      "loss": 1.368,
      "step": 778
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.849386692047119,
      "learning_rate": 9.982531607306358e-06,
      "loss": 0.2809,
      "step": 779
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.808527946472168,
      "learning_rate": 9.982434482834603e-06,
      "loss": 1.7569,
      "step": 780
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.417469024658203,
      "learning_rate": 9.982337089579369e-06,
      "loss": 1.4137,
      "step": 781
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.316526412963867,
      "learning_rate": 9.982239427545908e-06,
      "loss": 1.1,
      "step": 782
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.76253604888916,
      "learning_rate": 9.982141496739486e-06,
      "loss": 0.3366,
      "step": 783
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.241515159606934,
      "learning_rate": 9.982043297165392e-06,
      "loss": 0.2064,
      "step": 784
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.855693817138672,
      "learning_rate": 9.981944828828917e-06,
      "loss": 0.96,
      "step": 785
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.406031131744385,
      "learning_rate": 9.981846091735378e-06,
      "loss": 0.1746,
      "step": 786
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.297788619995117,
      "learning_rate": 9.981747085890101e-06,
      "loss": 1.9278,
      "step": 787
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.074715614318848,
      "learning_rate": 9.981647811298424e-06,
      "loss": 0.9575,
      "step": 788
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.731443405151367,
      "learning_rate": 9.981548267965704e-06,
      "loss": 1.364,
      "step": 789
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.154314041137695,
      "learning_rate": 9.981448455897312e-06,
      "loss": 1.0439,
      "step": 790
    },
    {
      "epoch": 0.06,
      "grad_norm": 14.696914672851562,
      "learning_rate": 9.981348375098629e-06,
      "loss": 1.5708,
      "step": 791
    },
    {
      "epoch": 0.06,
      "grad_norm": 17.10193634033203,
      "learning_rate": 9.98124802557506e-06,
      "loss": 1.4879,
      "step": 792
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.053723335266113,
      "learning_rate": 9.981147407332014e-06,
      "loss": 1.1571,
      "step": 793
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5716774463653564,
      "learning_rate": 9.981046520374919e-06,
      "loss": 0.189,
      "step": 794
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.806672096252441,
      "learning_rate": 9.980945364709218e-06,
      "loss": 1.2595,
      "step": 795
    },
    {
      "epoch": 0.06,
      "grad_norm": 18.44878578186035,
      "learning_rate": 9.98084394034037e-06,
      "loss": 1.481,
      "step": 796
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3831708431243896,
      "learning_rate": 9.980742247273846e-06,
      "loss": 0.1549,
      "step": 797
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.016735553741455,
      "learning_rate": 9.980640285515128e-06,
      "loss": 0.2475,
      "step": 798
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.275846004486084,
      "learning_rate": 9.980538055069721e-06,
      "loss": 0.2382,
      "step": 799
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.00758171081543,
      "learning_rate": 9.98043555594314e-06,
      "loss": 1.1695,
      "step": 800
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.109618186950684,
      "learning_rate": 9.980332788140909e-06,
      "loss": 0.1659,
      "step": 801
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.092493534088135,
      "learning_rate": 9.980229751668578e-06,
      "loss": 0.1981,
      "step": 802
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6991262435913086,
      "learning_rate": 9.980126446531705e-06,
      "loss": 0.1652,
      "step": 803
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.653724193572998,
      "learning_rate": 9.980022872735857e-06,
      "loss": 1.1229,
      "step": 804
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5375816822052,
      "learning_rate": 9.979919030286626e-06,
      "loss": 0.1246,
      "step": 805
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.498637676239014,
      "learning_rate": 9.979814919189616e-06,
      "loss": 0.2481,
      "step": 806
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.363789081573486,
      "learning_rate": 9.979710539450438e-06,
      "loss": 0.3061,
      "step": 807
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5797064304351807,
      "learning_rate": 9.979605891074727e-06,
      "loss": 0.1487,
      "step": 808
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.120511531829834,
      "learning_rate": 9.979500974068126e-06,
      "loss": 1.1201,
      "step": 809
    },
    {
      "epoch": 0.06,
      "grad_norm": 31.69936752319336,
      "learning_rate": 9.979395788436297e-06,
      "loss": 1.7813,
      "step": 810
    },
    {
      "epoch": 0.06,
      "grad_norm": 23.749187469482422,
      "learning_rate": 9.979290334184913e-06,
      "loss": 1.3849,
      "step": 811
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.231563091278076,
      "learning_rate": 9.979184611319664e-06,
      "loss": 0.1494,
      "step": 812
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.9892377853393555,
      "learning_rate": 9.979078619846251e-06,
      "loss": 0.278,
      "step": 813
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.47850799560547,
      "learning_rate": 9.978972359770394e-06,
      "loss": 1.465,
      "step": 814
    },
    {
      "epoch": 0.06,
      "grad_norm": 26.654258728027344,
      "learning_rate": 9.978865831097824e-06,
      "loss": 1.4059,
      "step": 815
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6847424507141113,
      "learning_rate": 9.978759033834289e-06,
      "loss": 0.223,
      "step": 816
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.871658325195312,
      "learning_rate": 9.978651967985548e-06,
      "loss": 1.2228,
      "step": 817
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.582578659057617,
      "learning_rate": 9.978544633557381e-06,
      "loss": 1.4995,
      "step": 818
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.64096736907959,
      "learning_rate": 9.978437030555576e-06,
      "loss": 0.9517,
      "step": 819
    },
    {
      "epoch": 0.06,
      "grad_norm": 20.502107620239258,
      "learning_rate": 9.978329158985936e-06,
      "loss": 1.2675,
      "step": 820
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.375232696533203,
      "learning_rate": 9.978221018854282e-06,
      "loss": 1.2054,
      "step": 821
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.5976362228393555,
      "learning_rate": 9.978112610166448e-06,
      "loss": 1.3194,
      "step": 822
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.007765769958496,
      "learning_rate": 9.978003932928281e-06,
      "loss": 0.1955,
      "step": 823
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.100627899169922,
      "learning_rate": 9.977894987145646e-06,
      "loss": 1.0818,
      "step": 824
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.484675884246826,
      "learning_rate": 9.97778577282442e-06,
      "loss": 0.2694,
      "step": 825
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.463666915893555,
      "learning_rate": 9.97767628997049e-06,
      "loss": 1.206,
      "step": 826
    },
    {
      "epoch": 0.06,
      "grad_norm": 14.941417694091797,
      "learning_rate": 9.97756653858977e-06,
      "loss": 1.7193,
      "step": 827
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4905283451080322,
      "learning_rate": 9.977456518688173e-06,
      "loss": 0.2105,
      "step": 828
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.712163925170898,
      "learning_rate": 9.977346230271639e-06,
      "loss": 1.837,
      "step": 829
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.630441665649414,
      "learning_rate": 9.977235673346115e-06,
      "loss": 0.2868,
      "step": 830
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.250532150268555,
      "learning_rate": 9.977124847917565e-06,
      "loss": 0.2035,
      "step": 831
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.5686445236206055,
      "learning_rate": 9.97701375399197e-06,
      "loss": 1.0235,
      "step": 832
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6093673706054688,
      "learning_rate": 9.976902391575325e-06,
      "loss": 0.1788,
      "step": 833
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1731173992156982,
      "learning_rate": 9.976790760673632e-06,
      "loss": 0.1027,
      "step": 834
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.198331594467163,
      "learning_rate": 9.976678861292915e-06,
      "loss": 0.1368,
      "step": 835
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.4312005043029785,
      "learning_rate": 9.976566693439212e-06,
      "loss": 0.2197,
      "step": 836
    },
    {
      "epoch": 0.06,
      "grad_norm": 19.71101188659668,
      "learning_rate": 9.976454257118573e-06,
      "loss": 2.15,
      "step": 837
    },
    {
      "epoch": 0.06,
      "grad_norm": 29.138595581054688,
      "learning_rate": 9.976341552337066e-06,
      "loss": 1.7692,
      "step": 838
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.006170749664307,
      "learning_rate": 9.976228579100765e-06,
      "loss": 0.1718,
      "step": 839
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.191370964050293,
      "learning_rate": 9.97611533741577e-06,
      "loss": 2.0762,
      "step": 840
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.228923320770264,
      "learning_rate": 9.97600182728819e-06,
      "loss": 1.0329,
      "step": 841
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.1090669631958,
      "learning_rate": 9.975888048724143e-06,
      "loss": 1.1892,
      "step": 842
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5114848613739014,
      "learning_rate": 9.975774001729773e-06,
      "loss": 0.1676,
      "step": 843
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.809741020202637,
      "learning_rate": 9.975659686311228e-06,
      "loss": 1.6203,
      "step": 844
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.444591760635376,
      "learning_rate": 9.975545102474679e-06,
      "loss": 0.1689,
      "step": 845
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.929079532623291,
      "learning_rate": 9.975430250226304e-06,
      "loss": 0.7156,
      "step": 846
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.320666790008545,
      "learning_rate": 9.9753151295723e-06,
      "loss": 0.9035,
      "step": 847
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.794825553894043,
      "learning_rate": 9.975199740518877e-06,
      "loss": 1.7378,
      "step": 848
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.22800350189209,
      "learning_rate": 9.975084083072262e-06,
      "loss": 1.0317,
      "step": 849
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.588905334472656,
      "learning_rate": 9.974968157238691e-06,
      "loss": 0.2227,
      "step": 850
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.101860046386719,
      "learning_rate": 9.974851963024419e-06,
      "loss": 1.0885,
      "step": 851
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.831782341003418,
      "learning_rate": 9.974735500435715e-06,
      "loss": 2.1576,
      "step": 852
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.016918659210205,
      "learning_rate": 9.974618769478862e-06,
      "loss": 0.1732,
      "step": 853
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.096602439880371,
      "learning_rate": 9.974501770160154e-06,
      "loss": 0.2823,
      "step": 854
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.674158573150635,
      "learning_rate": 9.974384502485906e-06,
      "loss": 0.2606,
      "step": 855
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.849853515625,
      "learning_rate": 9.974266966462444e-06,
      "loss": 1.4273,
      "step": 856
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.934659004211426,
      "learning_rate": 9.974149162096108e-06,
      "loss": 1.0401,
      "step": 857
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.54241681098938,
      "learning_rate": 9.974031089393251e-06,
      "loss": 0.2435,
      "step": 858
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.105855464935303,
      "learning_rate": 9.973912748360245e-06,
      "loss": 1.1416,
      "step": 859
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.785295486450195,
      "learning_rate": 9.973794139003475e-06,
      "loss": 0.2507,
      "step": 860
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.79770040512085,
      "learning_rate": 9.973675261329336e-06,
      "loss": 0.2622,
      "step": 861
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.671207427978516,
      "learning_rate": 9.973556115344246e-06,
      "loss": 0.3067,
      "step": 862
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.681049823760986,
      "learning_rate": 9.973436701054627e-06,
      "loss": 1.1473,
      "step": 863
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.621525764465332,
      "learning_rate": 9.973317018466924e-06,
      "loss": 1.6814,
      "step": 864
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5526764392852783,
      "learning_rate": 9.973197067587593e-06,
      "loss": 0.1562,
      "step": 865
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.381828308105469,
      "learning_rate": 9.973076848423105e-06,
      "loss": 0.2159,
      "step": 866
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.615251064300537,
      "learning_rate": 9.972956360979944e-06,
      "loss": 1.0938,
      "step": 867
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.895399570465088,
      "learning_rate": 9.972835605264612e-06,
      "loss": 0.1442,
      "step": 868
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.944989204406738,
      "learning_rate": 9.972714581283623e-06,
      "loss": 0.9266,
      "step": 869
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.1279296875,
      "learning_rate": 9.972593289043504e-06,
      "loss": 2.0416,
      "step": 870
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.279372692108154,
      "learning_rate": 9.972471728550799e-06,
      "loss": 0.1839,
      "step": 871
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.382205963134766,
      "learning_rate": 9.972349899812067e-06,
      "loss": 1.0624,
      "step": 872
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.778101921081543,
      "learning_rate": 9.97222780283388e-06,
      "loss": 0.8683,
      "step": 873
    },
    {
      "epoch": 0.06,
      "grad_norm": 20.449316024780273,
      "learning_rate": 9.972105437622822e-06,
      "loss": 2.3457,
      "step": 874
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.782371520996094,
      "learning_rate": 9.971982804185497e-06,
      "loss": 2.0218,
      "step": 875
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.31382942199707,
      "learning_rate": 9.97185990252852e-06,
      "loss": 1.366,
      "step": 876
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.8159494400024414,
      "learning_rate": 9.97173673265852e-06,
      "loss": 0.1971,
      "step": 877
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.049508094787598,
      "learning_rate": 9.971613294582143e-06,
      "loss": 0.2121,
      "step": 878
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.24026107788086,
      "learning_rate": 9.971489588306047e-06,
      "loss": 1.4169,
      "step": 879
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.58404541015625,
      "learning_rate": 9.971365613836905e-06,
      "loss": 0.9788,
      "step": 880
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.7991790771484375,
      "learning_rate": 9.971241371181409e-06,
      "loss": 0.1811,
      "step": 881
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.628178119659424,
      "learning_rate": 9.971116860346254e-06,
      "loss": 0.1776,
      "step": 882
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.350831985473633,
      "learning_rate": 9.970992081338163e-06,
      "loss": 1.9078,
      "step": 883
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.059969425201416,
      "learning_rate": 9.970867034163865e-06,
      "loss": 0.2099,
      "step": 884
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.655264377593994,
      "learning_rate": 9.970741718830108e-06,
      "loss": 0.2901,
      "step": 885
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.8947556018829346,
      "learning_rate": 9.970616135343649e-06,
      "loss": 0.2263,
      "step": 886
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.10819149017334,
      "learning_rate": 9.970490283711265e-06,
      "loss": 1.5275,
      "step": 887
    },
    {
      "epoch": 0.06,
      "grad_norm": 14.062370300292969,
      "learning_rate": 9.970364163939742e-06,
      "loss": 1.0995,
      "step": 888
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.297138690948486,
      "learning_rate": 9.970237776035888e-06,
      "loss": 0.175,
      "step": 889
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4596762657165527,
      "learning_rate": 9.97011112000652e-06,
      "loss": 0.1062,
      "step": 890
    },
    {
      "epoch": 0.06,
      "grad_norm": 18.91884422302246,
      "learning_rate": 9.96998419585847e-06,
      "loss": 1.6912,
      "step": 891
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.595026969909668,
      "learning_rate": 9.969857003598586e-06,
      "loss": 0.1703,
      "step": 892
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.212413787841797,
      "learning_rate": 9.969729543233727e-06,
      "loss": 0.924,
      "step": 893
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.0734758377075195,
      "learning_rate": 9.969601814770771e-06,
      "loss": 0.1597,
      "step": 894
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.672718048095703,
      "learning_rate": 9.969473818216608e-06,
      "loss": 0.3536,
      "step": 895
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.52903938293457,
      "learning_rate": 9.969345553578143e-06,
      "loss": 1.4356,
      "step": 896
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.648919105529785,
      "learning_rate": 9.969217020862294e-06,
      "loss": 1.2253,
      "step": 897
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2697675228118896,
      "learning_rate": 9.969088220075998e-06,
      "loss": 0.1579,
      "step": 898
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.257129192352295,
      "learning_rate": 9.968959151226201e-06,
      "loss": 0.4263,
      "step": 899
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.375749588012695,
      "learning_rate": 9.968829814319865e-06,
      "loss": 0.9645,
      "step": 900
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.10562515258789,
      "learning_rate": 9.968700209363971e-06,
      "loss": 1.4868,
      "step": 901
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.292746543884277,
      "learning_rate": 9.968570336365507e-06,
      "loss": 1.3291,
      "step": 902
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.335055351257324,
      "learning_rate": 9.968440195331482e-06,
      "loss": 0.1431,
      "step": 903
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.62394905090332,
      "learning_rate": 9.968309786268913e-06,
      "loss": 0.2018,
      "step": 904
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.652385711669922,
      "learning_rate": 9.968179109184837e-06,
      "loss": 1.4493,
      "step": 905
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.427290916442871,
      "learning_rate": 9.968048164086304e-06,
      "loss": 1.0376,
      "step": 906
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.9282283782958984,
      "learning_rate": 9.967916950980379e-06,
      "loss": 0.2179,
      "step": 907
    },
    {
      "epoch": 0.07,
      "grad_norm": 31.194189071655273,
      "learning_rate": 9.967785469874137e-06,
      "loss": 2.2592,
      "step": 908
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.646177768707275,
      "learning_rate": 9.967653720774674e-06,
      "loss": 1.0751,
      "step": 909
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.626405715942383,
      "learning_rate": 9.967521703689096e-06,
      "loss": 1.1877,
      "step": 910
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.664020538330078,
      "learning_rate": 9.967389418624527e-06,
      "loss": 0.9522,
      "step": 911
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.7849578857421875,
      "learning_rate": 9.9672568655881e-06,
      "loss": 0.1762,
      "step": 912
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.373814105987549,
      "learning_rate": 9.967124044586966e-06,
      "loss": 0.2464,
      "step": 913
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.136683940887451,
      "learning_rate": 9.966990955628293e-06,
      "loss": 1.0821,
      "step": 914
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.290602207183838,
      "learning_rate": 9.966857598719259e-06,
      "loss": 0.3102,
      "step": 915
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.38869571685791,
      "learning_rate": 9.966723973867058e-06,
      "loss": 0.24,
      "step": 916
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.019811630249023,
      "learning_rate": 9.966590081078898e-06,
      "loss": 1.4209,
      "step": 917
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.29668664932251,
      "learning_rate": 9.966455920362003e-06,
      "loss": 0.2101,
      "step": 918
    },
    {
      "epoch": 0.07,
      "grad_norm": 22.90641975402832,
      "learning_rate": 9.966321491723609e-06,
      "loss": 1.9028,
      "step": 919
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.115350246429443,
      "learning_rate": 9.966186795170971e-06,
      "loss": 0.159,
      "step": 920
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.7599899768829346,
      "learning_rate": 9.966051830711353e-06,
      "loss": 0.2432,
      "step": 921
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.11130714416504,
      "learning_rate": 9.965916598352037e-06,
      "loss": 1.8512,
      "step": 922
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.338322639465332,
      "learning_rate": 9.965781098100315e-06,
      "loss": 1.0696,
      "step": 923
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3570187091827393,
      "learning_rate": 9.965645329963502e-06,
      "loss": 0.1366,
      "step": 924
    },
    {
      "epoch": 0.07,
      "grad_norm": 22.134218215942383,
      "learning_rate": 9.965509293948918e-06,
      "loss": 1.3009,
      "step": 925
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.638859272003174,
      "learning_rate": 9.965372990063903e-06,
      "loss": 1.0424,
      "step": 926
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.384793758392334,
      "learning_rate": 9.96523641831581e-06,
      "loss": 1.0119,
      "step": 927
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.778343677520752,
      "learning_rate": 9.965099578712008e-06,
      "loss": 0.2453,
      "step": 928
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.732186317443848,
      "learning_rate": 9.964962471259875e-06,
      "loss": 1.4287,
      "step": 929
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.265811920166016,
      "learning_rate": 9.964825095966814e-06,
      "loss": 1.2469,
      "step": 930
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.799042224884033,
      "learning_rate": 9.96468745284023e-06,
      "loss": 0.1788,
      "step": 931
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.589057922363281,
      "learning_rate": 9.96454954188755e-06,
      "loss": 0.973,
      "step": 932
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.277734279632568,
      "learning_rate": 9.964411363116213e-06,
      "loss": 0.2045,
      "step": 933
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.898396492004395,
      "learning_rate": 9.964272916533677e-06,
      "loss": 1.5181,
      "step": 934
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.022498607635498,
      "learning_rate": 9.964134202147406e-06,
      "loss": 0.177,
      "step": 935
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.612159729003906,
      "learning_rate": 9.963995219964884e-06,
      "loss": 0.9796,
      "step": 936
    },
    {
      "epoch": 0.07,
      "grad_norm": 21.442657470703125,
      "learning_rate": 9.963855969993611e-06,
      "loss": 1.4492,
      "step": 937
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.828654766082764,
      "learning_rate": 9.963716452241099e-06,
      "loss": 0.9511,
      "step": 938
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.385120391845703,
      "learning_rate": 9.963576666714871e-06,
      "loss": 1.1126,
      "step": 939
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.465864896774292,
      "learning_rate": 9.96343661342247e-06,
      "loss": 0.186,
      "step": 940
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.488648414611816,
      "learning_rate": 9.963296292371454e-06,
      "loss": 0.305,
      "step": 941
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.366483688354492,
      "learning_rate": 9.963155703569388e-06,
      "loss": 1.63,
      "step": 942
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.593531608581543,
      "learning_rate": 9.963014847023858e-06,
      "loss": 1.2181,
      "step": 943
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.0353901386260986,
      "learning_rate": 9.962873722742463e-06,
      "loss": 0.1007,
      "step": 944
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.749434471130371,
      "learning_rate": 9.962732330732818e-06,
      "loss": 1.2705,
      "step": 945
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3474791049957275,
      "learning_rate": 9.962590671002546e-06,
      "loss": 0.0822,
      "step": 946
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.354191780090332,
      "learning_rate": 9.962448743559295e-06,
      "loss": 1.5302,
      "step": 947
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.007987022399902,
      "learning_rate": 9.962306548410717e-06,
      "loss": 0.2461,
      "step": 948
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.009433269500732,
      "learning_rate": 9.962164085564482e-06,
      "loss": 0.2333,
      "step": 949
    },
    {
      "epoch": 0.07,
      "grad_norm": 44.69111633300781,
      "learning_rate": 9.96202135502828e-06,
      "loss": 2.5001,
      "step": 950
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.9776480197906494,
      "learning_rate": 9.961878356809807e-06,
      "loss": 0.252,
      "step": 951
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.334018707275391,
      "learning_rate": 9.961735090916778e-06,
      "loss": 0.259,
      "step": 952
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.049874305725098,
      "learning_rate": 9.961591557356922e-06,
      "loss": 0.1695,
      "step": 953
    },
    {
      "epoch": 0.07,
      "grad_norm": 26.235212326049805,
      "learning_rate": 9.961447756137982e-06,
      "loss": 1.695,
      "step": 954
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.822461128234863,
      "learning_rate": 9.961303687267716e-06,
      "loss": 1.4537,
      "step": 955
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.6951794624328613,
      "learning_rate": 9.961159350753895e-06,
      "loss": 0.1312,
      "step": 956
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.61787223815918,
      "learning_rate": 9.961014746604308e-06,
      "loss": 1.2625,
      "step": 957
    },
    {
      "epoch": 0.07,
      "grad_norm": 23.0617618560791,
      "learning_rate": 9.960869874826753e-06,
      "loss": 1.1416,
      "step": 958
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.682098388671875,
      "learning_rate": 9.960724735429047e-06,
      "loss": 0.292,
      "step": 959
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.623281478881836,
      "learning_rate": 9.960579328419017e-06,
      "loss": 1.2112,
      "step": 960
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.514970779418945,
      "learning_rate": 9.960433653804509e-06,
      "loss": 0.1577,
      "step": 961
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.029726028442383,
      "learning_rate": 9.960287711593383e-06,
      "loss": 1.3764,
      "step": 962
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.748559474945068,
      "learning_rate": 9.96014150179351e-06,
      "loss": 1.0677,
      "step": 963
    },
    {
      "epoch": 0.07,
      "grad_norm": 30.224946975708008,
      "learning_rate": 9.959995024412778e-06,
      "loss": 1.9758,
      "step": 964
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.234914779663086,
      "learning_rate": 9.95984827945909e-06,
      "loss": 1.5059,
      "step": 965
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.815462112426758,
      "learning_rate": 9.95970126694036e-06,
      "loss": 1.0147,
      "step": 966
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.477725028991699,
      "learning_rate": 9.959553986864521e-06,
      "loss": 0.233,
      "step": 967
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.2830171585083,
      "learning_rate": 9.959406439239517e-06,
      "loss": 0.9133,
      "step": 968
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.109032392501831,
      "learning_rate": 9.95925862407331e-06,
      "loss": 0.1384,
      "step": 969
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.88412094116211,
      "learning_rate": 9.95911054137387e-06,
      "loss": 0.9393,
      "step": 970
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.190434455871582,
      "learning_rate": 9.958962191149187e-06,
      "loss": 0.9143,
      "step": 971
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.214535713195801,
      "learning_rate": 9.958813573407264e-06,
      "loss": 0.994,
      "step": 972
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.417541027069092,
      "learning_rate": 9.95866468815612e-06,
      "loss": 0.2967,
      "step": 973
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.7727060317993164,
      "learning_rate": 9.958515535403787e-06,
      "loss": 0.149,
      "step": 974
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.21804428100586,
      "learning_rate": 9.958366115158308e-06,
      "loss": 1.3067,
      "step": 975
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.699132204055786,
      "learning_rate": 9.958216427427746e-06,
      "loss": 0.2281,
      "step": 976
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.615583658218384,
      "learning_rate": 9.958066472220176e-06,
      "loss": 0.1974,
      "step": 977
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.500351428985596,
      "learning_rate": 9.957916249543686e-06,
      "loss": 0.2225,
      "step": 978
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.099822521209717,
      "learning_rate": 9.957765759406383e-06,
      "loss": 0.3205,
      "step": 979
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.403736591339111,
      "learning_rate": 9.957615001816381e-06,
      "loss": 0.2757,
      "step": 980
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.485078811645508,
      "learning_rate": 9.957463976781818e-06,
      "loss": 1.5475,
      "step": 981
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2485287189483643,
      "learning_rate": 9.957312684310835e-06,
      "loss": 0.1001,
      "step": 982
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.009557723999023,
      "learning_rate": 9.957161124411601e-06,
      "loss": 2.1508,
      "step": 983
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.96927833557129,
      "learning_rate": 9.957009297092286e-06,
      "loss": 1.6724,
      "step": 984
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.076412200927734,
      "learning_rate": 9.956857202361082e-06,
      "loss": 0.3051,
      "step": 985
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.539407253265381,
      "learning_rate": 9.956704840226197e-06,
      "loss": 1.0639,
      "step": 986
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.157520771026611,
      "learning_rate": 9.956552210695847e-06,
      "loss": 0.3059,
      "step": 987
    },
    {
      "epoch": 0.07,
      "grad_norm": 37.81488037109375,
      "learning_rate": 9.956399313778267e-06,
      "loss": 1.3444,
      "step": 988
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.674591064453125,
      "learning_rate": 9.956246149481703e-06,
      "loss": 1.3561,
      "step": 989
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.421346664428711,
      "learning_rate": 9.956092717814421e-06,
      "loss": 1.4048,
      "step": 990
    },
    {
      "epoch": 0.07,
      "grad_norm": 44.441349029541016,
      "learning_rate": 9.955939018784698e-06,
      "loss": 1.408,
      "step": 991
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.897730588912964,
      "learning_rate": 9.955785052400824e-06,
      "loss": 0.207,
      "step": 992
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.197593688964844,
      "learning_rate": 9.955630818671103e-06,
      "loss": 0.2345,
      "step": 993
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.352502822875977,
      "learning_rate": 9.955476317603858e-06,
      "loss": 1.8841,
      "step": 994
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.358023643493652,
      "learning_rate": 9.955321549207423e-06,
      "loss": 1.3089,
      "step": 995
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.0168914794921875,
      "learning_rate": 9.955166513490149e-06,
      "loss": 1.1268,
      "step": 996
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.76189422607422,
      "learning_rate": 9.955011210460398e-06,
      "loss": 1.6399,
      "step": 997
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.002405166625977,
      "learning_rate": 9.954855640126546e-06,
      "loss": 1.1592,
      "step": 998
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.296775817871094,
      "learning_rate": 9.95469980249699e-06,
      "loss": 1.0906,
      "step": 999
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.943358421325684,
      "learning_rate": 9.954543697580132e-06,
      "loss": 1.4037,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.8427836894989014,
      "learning_rate": 9.954387325384397e-06,
      "loss": 0.1643,
      "step": 1001
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.65923309326172,
      "learning_rate": 9.954230685918216e-06,
      "loss": 1.6262,
      "step": 1002
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.286722421646118,
      "learning_rate": 9.954073779190046e-06,
      "loss": 0.1666,
      "step": 1003
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.818149089813232,
      "learning_rate": 9.953916605208346e-06,
      "loss": 0.1655,
      "step": 1004
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.88496732711792,
      "learning_rate": 9.953759163981597e-06,
      "loss": 1.1992,
      "step": 1005
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.206100463867188,
      "learning_rate": 9.953601455518292e-06,
      "loss": 1.6216,
      "step": 1006
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.8569235801696777,
      "learning_rate": 9.953443479826939e-06,
      "loss": 0.145,
      "step": 1007
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.131827354431152,
      "learning_rate": 9.95328523691606e-06,
      "loss": 0.2102,
      "step": 1008
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.876476287841797,
      "learning_rate": 9.953126726794192e-06,
      "loss": 1.5769,
      "step": 1009
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.187329292297363,
      "learning_rate": 9.952967949469886e-06,
      "loss": 1.0134,
      "step": 1010
    },
    {
      "epoch": 0.07,
      "grad_norm": 76.46361541748047,
      "learning_rate": 9.952808904951706e-06,
      "loss": 1.3705,
      "step": 1011
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.9026527404785156,
      "learning_rate": 9.952649593248233e-06,
      "loss": 0.3198,
      "step": 1012
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.720800399780273,
      "learning_rate": 9.952490014368062e-06,
      "loss": 1.2789,
      "step": 1013
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.360846519470215,
      "learning_rate": 9.9523301683198e-06,
      "loss": 1.267,
      "step": 1014
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.6684627532959,
      "learning_rate": 9.952170055112072e-06,
      "loss": 2.1169,
      "step": 1015
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.277995586395264,
      "learning_rate": 9.952009674753514e-06,
      "loss": 1.1245,
      "step": 1016
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.652012825012207,
      "learning_rate": 9.95184902725278e-06,
      "loss": 1.3211,
      "step": 1017
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.734283685684204,
      "learning_rate": 9.951688112618531e-06,
      "loss": 0.2954,
      "step": 1018
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.845208168029785,
      "learning_rate": 9.951526930859455e-06,
      "loss": 1.5682,
      "step": 1019
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.8226430416107178,
      "learning_rate": 9.951365481984243e-06,
      "loss": 0.2802,
      "step": 1020
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.751898765563965,
      "learning_rate": 9.951203766001605e-06,
      "loss": 0.2449,
      "step": 1021
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.8313238620758057,
      "learning_rate": 9.951041782920263e-06,
      "loss": 0.2761,
      "step": 1022
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.932676315307617,
      "learning_rate": 9.95087953274896e-06,
      "loss": 1.1327,
      "step": 1023
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.243513584136963,
      "learning_rate": 9.950717015496447e-06,
      "loss": 0.1535,
      "step": 1024
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.784003973007202,
      "learning_rate": 9.950554231171488e-06,
      "loss": 0.1982,
      "step": 1025
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.497347116470337,
      "learning_rate": 9.950391179782868e-06,
      "loss": 0.2131,
      "step": 1026
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.7917633056640625,
      "learning_rate": 9.950227861339383e-06,
      "loss": 1.1563,
      "step": 1027
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.217779159545898,
      "learning_rate": 9.950064275849842e-06,
      "loss": 0.1878,
      "step": 1028
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8267123699188232,
      "learning_rate": 9.949900423323068e-06,
      "loss": 0.1176,
      "step": 1029
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.265015602111816,
      "learning_rate": 9.949736303767906e-06,
      "loss": 0.3015,
      "step": 1030
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3566336631774902,
      "learning_rate": 9.949571917193206e-06,
      "loss": 0.173,
      "step": 1031
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.24682903289795,
      "learning_rate": 9.949407263607835e-06,
      "loss": 0.8351,
      "step": 1032
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.152237892150879,
      "learning_rate": 9.949242343020676e-06,
      "loss": 1.0779,
      "step": 1033
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.216475009918213,
      "learning_rate": 9.949077155440627e-06,
      "loss": 0.2974,
      "step": 1034
    },
    {
      "epoch": 0.07,
      "grad_norm": 27.89541244506836,
      "learning_rate": 9.9489117008766e-06,
      "loss": 1.508,
      "step": 1035
    },
    {
      "epoch": 0.07,
      "grad_norm": 22.13656234741211,
      "learning_rate": 9.948745979337518e-06,
      "loss": 1.2397,
      "step": 1036
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.889971733093262,
      "learning_rate": 9.948579990832322e-06,
      "loss": 1.0123,
      "step": 1037
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1296048164367676,
      "learning_rate": 9.948413735369968e-06,
      "loss": 0.0947,
      "step": 1038
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.7040324211120605,
      "learning_rate": 9.948247212959424e-06,
      "loss": 0.1956,
      "step": 1039
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.647492408752441,
      "learning_rate": 9.948080423609672e-06,
      "loss": 1.027,
      "step": 1040
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.188313484191895,
      "learning_rate": 9.947913367329711e-06,
      "loss": 1.1062,
      "step": 1041
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.169330596923828,
      "learning_rate": 9.947746044128552e-06,
      "loss": 0.3053,
      "step": 1042
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.738101959228516,
      "learning_rate": 9.947578454015224e-06,
      "loss": 1.8182,
      "step": 1043
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.18891143798828,
      "learning_rate": 9.947410596998765e-06,
      "loss": 1.4672,
      "step": 1044
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.590545654296875,
      "learning_rate": 9.947242473088232e-06,
      "loss": 0.8577,
      "step": 1045
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0772194862365723,
      "learning_rate": 9.947074082292692e-06,
      "loss": 0.1549,
      "step": 1046
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.587944030761719,
      "learning_rate": 9.946905424621232e-06,
      "loss": 0.9951,
      "step": 1047
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.423672676086426,
      "learning_rate": 9.946736500082949e-06,
      "loss": 1.1173,
      "step": 1048
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.150685787200928,
      "learning_rate": 9.946567308686957e-06,
      "loss": 0.9879,
      "step": 1049
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.428102493286133,
      "learning_rate": 9.946397850442382e-06,
      "loss": 0.227,
      "step": 1050
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1976778507232666,
      "learning_rate": 9.946228125358365e-06,
      "loss": 0.1596,
      "step": 1051
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7297680377960205,
      "learning_rate": 9.946058133444066e-06,
      "loss": 0.2163,
      "step": 1052
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.176034927368164,
      "learning_rate": 9.945887874708651e-06,
      "loss": 0.9556,
      "step": 1053
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.194331169128418,
      "learning_rate": 9.945717349161306e-06,
      "loss": 1.134,
      "step": 1054
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.492671012878418,
      "learning_rate": 9.94554655681123e-06,
      "loss": 1.7867,
      "step": 1055
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.59260082244873,
      "learning_rate": 9.94537549766764e-06,
      "loss": 1.0384,
      "step": 1056
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.539525508880615,
      "learning_rate": 9.945204171739759e-06,
      "loss": 0.2144,
      "step": 1057
    },
    {
      "epoch": 0.08,
      "grad_norm": 25.00383758544922,
      "learning_rate": 9.94503257903683e-06,
      "loss": 1.3821,
      "step": 1058
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.197775840759277,
      "learning_rate": 9.944860719568115e-06,
      "loss": 0.8621,
      "step": 1059
    },
    {
      "epoch": 0.08,
      "grad_norm": 28.44542694091797,
      "learning_rate": 9.94468859334288e-06,
      "loss": 1.5824,
      "step": 1060
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.983018398284912,
      "learning_rate": 9.94451620037041e-06,
      "loss": 1.0485,
      "step": 1061
    },
    {
      "epoch": 0.08,
      "grad_norm": 22.088476181030273,
      "learning_rate": 9.944343540660009e-06,
      "loss": 1.1417,
      "step": 1062
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.107381820678711,
      "learning_rate": 9.94417061422099e-06,
      "loss": 0.2809,
      "step": 1063
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.543365478515625,
      "learning_rate": 9.943997421062679e-06,
      "loss": 0.2603,
      "step": 1064
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.073062419891357,
      "learning_rate": 9.943823961194422e-06,
      "loss": 0.2587,
      "step": 1065
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.708248615264893,
      "learning_rate": 9.943650234625576e-06,
      "loss": 0.2855,
      "step": 1066
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.90576171875,
      "learning_rate": 9.943476241365513e-06,
      "loss": 1.52,
      "step": 1067
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.2038204669952393,
      "learning_rate": 9.943301981423618e-06,
      "loss": 0.1531,
      "step": 1068
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.538870811462402,
      "learning_rate": 9.943127454809293e-06,
      "loss": 1.7578,
      "step": 1069
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.877978324890137,
      "learning_rate": 9.942952661531952e-06,
      "loss": 0.8118,
      "step": 1070
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.207956790924072,
      "learning_rate": 9.942777601601025e-06,
      "loss": 0.8913,
      "step": 1071
    },
    {
      "epoch": 0.08,
      "grad_norm": 27.07394027709961,
      "learning_rate": 9.942602275025956e-06,
      "loss": 1.2434,
      "step": 1072
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0766701698303223,
      "learning_rate": 9.942426681816204e-06,
      "loss": 0.1289,
      "step": 1073
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.237187385559082,
      "learning_rate": 9.94225082198124e-06,
      "loss": 0.2238,
      "step": 1074
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.9947638511657715,
      "learning_rate": 9.94207469553055e-06,
      "loss": 0.2452,
      "step": 1075
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.667441368103027,
      "learning_rate": 9.94189830247364e-06,
      "loss": 1.3398,
      "step": 1076
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.126637935638428,
      "learning_rate": 9.94172164282002e-06,
      "loss": 1.1292,
      "step": 1077
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7417805194854736,
      "learning_rate": 9.941544716579224e-06,
      "loss": 0.1827,
      "step": 1078
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.007020473480225,
      "learning_rate": 9.941367523760796e-06,
      "loss": 0.1546,
      "step": 1079
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.478809356689453,
      "learning_rate": 9.941190064374294e-06,
      "loss": 1.4162,
      "step": 1080
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0784385204315186,
      "learning_rate": 9.94101233842929e-06,
      "loss": 0.11,
      "step": 1081
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.8695149421691895,
      "learning_rate": 9.940834345935374e-06,
      "loss": 0.2828,
      "step": 1082
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.625448226928711,
      "learning_rate": 9.940656086902147e-06,
      "loss": 0.339,
      "step": 1083
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.3943915367126465,
      "learning_rate": 9.940477561339225e-06,
      "loss": 0.296,
      "step": 1084
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.588335037231445,
      "learning_rate": 9.94029876925624e-06,
      "loss": 0.3083,
      "step": 1085
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.595217704772949,
      "learning_rate": 9.940119710662837e-06,
      "loss": 0.1213,
      "step": 1086
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.45607852935791,
      "learning_rate": 9.939940385568673e-06,
      "loss": 1.4633,
      "step": 1087
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.096630573272705,
      "learning_rate": 9.939760793983426e-06,
      "loss": 0.2677,
      "step": 1088
    },
    {
      "epoch": 0.08,
      "grad_norm": 20.4110050201416,
      "learning_rate": 9.939580935916782e-06,
      "loss": 1.5427,
      "step": 1089
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.6019463539123535,
      "learning_rate": 9.93940081137844e-06,
      "loss": 0.2669,
      "step": 1090
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.469991683959961,
      "learning_rate": 9.939220420378126e-06,
      "loss": 1.2214,
      "step": 1091
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.078134536743164,
      "learning_rate": 9.939039762925565e-06,
      "loss": 0.981,
      "step": 1092
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.1450629234313965,
      "learning_rate": 9.938858839030503e-06,
      "loss": 0.3114,
      "step": 1093
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.216594696044922,
      "learning_rate": 9.9386776487027e-06,
      "loss": 2.0569,
      "step": 1094
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7769827842712402,
      "learning_rate": 9.938496191951934e-06,
      "loss": 0.1566,
      "step": 1095
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.741845607757568,
      "learning_rate": 9.938314468787992e-06,
      "loss": 0.969,
      "step": 1096
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.6650407314300537,
      "learning_rate": 9.938132479220675e-06,
      "loss": 0.1988,
      "step": 1097
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.440617561340332,
      "learning_rate": 9.937950223259804e-06,
      "loss": 0.9362,
      "step": 1098
    },
    {
      "epoch": 0.08,
      "grad_norm": 20.416738510131836,
      "learning_rate": 9.93776770091521e-06,
      "loss": 2.0585,
      "step": 1099
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4264869689941406,
      "learning_rate": 9.937584912196738e-06,
      "loss": 0.1549,
      "step": 1100
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.4808030128479,
      "learning_rate": 9.937401857114249e-06,
      "loss": 0.2621,
      "step": 1101
    },
    {
      "epoch": 0.08,
      "grad_norm": 35.75822067260742,
      "learning_rate": 9.93721853567762e-06,
      "loss": 1.989,
      "step": 1102
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.852687120437622,
      "learning_rate": 9.937034947896739e-06,
      "loss": 0.131,
      "step": 1103
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.71227264404297,
      "learning_rate": 9.936851093781512e-06,
      "loss": 1.4509,
      "step": 1104
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.015422821044922,
      "learning_rate": 9.936666973341853e-06,
      "loss": 1.8793,
      "step": 1105
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.6196277141571045,
      "learning_rate": 9.936482586587698e-06,
      "loss": 0.2013,
      "step": 1106
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.044247627258301,
      "learning_rate": 9.936297933528992e-06,
      "loss": 0.1722,
      "step": 1107
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.070391654968262,
      "learning_rate": 9.9361130141757e-06,
      "loss": 1.2459,
      "step": 1108
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.2434818744659424,
      "learning_rate": 9.935927828537793e-06,
      "loss": 0.1243,
      "step": 1109
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.74853777885437,
      "learning_rate": 9.935742376625264e-06,
      "loss": 0.2265,
      "step": 1110
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.3153743743896484,
      "learning_rate": 9.935556658448115e-06,
      "loss": 0.2144,
      "step": 1111
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.200043201446533,
      "learning_rate": 9.935370674016369e-06,
      "loss": 0.2618,
      "step": 1112
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.526752471923828,
      "learning_rate": 9.935184423340054e-06,
      "loss": 0.2519,
      "step": 1113
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.176612854003906,
      "learning_rate": 9.934997906429222e-06,
      "loss": 2.0883,
      "step": 1114
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.192365646362305,
      "learning_rate": 9.934811123293933e-06,
      "loss": 0.9088,
      "step": 1115
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.46162748336792,
      "learning_rate": 9.93462407394426e-06,
      "loss": 0.2118,
      "step": 1116
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.551731586456299,
      "learning_rate": 9.9344367583903e-06,
      "loss": 1.0432,
      "step": 1117
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.446638584136963,
      "learning_rate": 9.934249176642155e-06,
      "loss": 1.0126,
      "step": 1118
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7243454456329346,
      "learning_rate": 9.934061328709943e-06,
      "loss": 0.1034,
      "step": 1119
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.623844623565674,
      "learning_rate": 9.933873214603798e-06,
      "loss": 0.1985,
      "step": 1120
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.269323348999023,
      "learning_rate": 9.93368483433387e-06,
      "loss": 0.971,
      "step": 1121
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.033571243286133,
      "learning_rate": 9.933496187910319e-06,
      "loss": 0.2505,
      "step": 1122
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.006040573120117,
      "learning_rate": 9.933307275343324e-06,
      "loss": 1.0024,
      "step": 1123
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.791678428649902,
      "learning_rate": 9.933118096643075e-06,
      "loss": 0.3865,
      "step": 1124
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.260610580444336,
      "learning_rate": 9.932928651819777e-06,
      "loss": 1.3276,
      "step": 1125
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.026941776275635,
      "learning_rate": 9.93273894088365e-06,
      "loss": 0.1831,
      "step": 1126
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.505718231201172,
      "learning_rate": 9.93254896384493e-06,
      "loss": 1.0762,
      "step": 1127
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.826434135437012,
      "learning_rate": 9.932358720713863e-06,
      "loss": 1.4217,
      "step": 1128
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.8435869216918945,
      "learning_rate": 9.932168211500714e-06,
      "loss": 0.9092,
      "step": 1129
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.884497880935669,
      "learning_rate": 9.931977436215759e-06,
      "loss": 0.1763,
      "step": 1130
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.57691764831543,
      "learning_rate": 9.93178639486929e-06,
      "loss": 0.803,
      "step": 1131
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4224050045013428,
      "learning_rate": 9.93159508747161e-06,
      "loss": 0.1721,
      "step": 1132
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.343262195587158,
      "learning_rate": 9.931403514033046e-06,
      "loss": 0.1969,
      "step": 1133
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.69195032119751,
      "learning_rate": 9.931211674563926e-06,
      "loss": 0.1751,
      "step": 1134
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.390392780303955,
      "learning_rate": 9.931019569074604e-06,
      "loss": 0.2629,
      "step": 1135
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.8672046661376953,
      "learning_rate": 9.93082719757544e-06,
      "loss": 0.1933,
      "step": 1136
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.44472599029541,
      "learning_rate": 9.930634560076813e-06,
      "loss": 0.3303,
      "step": 1137
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.497509479522705,
      "learning_rate": 9.930441656589115e-06,
      "loss": 0.3554,
      "step": 1138
    },
    {
      "epoch": 0.08,
      "grad_norm": 35.84265899658203,
      "learning_rate": 9.930248487122752e-06,
      "loss": 1.3312,
      "step": 1139
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.7674689292907715,
      "learning_rate": 9.930055051688146e-06,
      "loss": 0.2872,
      "step": 1140
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.579472541809082,
      "learning_rate": 9.929861350295732e-06,
      "loss": 1.0463,
      "step": 1141
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.522286891937256,
      "learning_rate": 9.929667382955956e-06,
      "loss": 0.9367,
      "step": 1142
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.03921890258789,
      "learning_rate": 9.929473149679285e-06,
      "loss": 1.4068,
      "step": 1143
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.03004789352417,
      "learning_rate": 9.929278650476199e-06,
      "loss": 0.1417,
      "step": 1144
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.5579378604888916,
      "learning_rate": 9.929083885357186e-06,
      "loss": 0.1712,
      "step": 1145
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.149375915527344,
      "learning_rate": 9.928888854332756e-06,
      "loss": 1.2839,
      "step": 1146
    },
    {
      "epoch": 0.08,
      "grad_norm": 20.395103454589844,
      "learning_rate": 9.928693557413429e-06,
      "loss": 1.3921,
      "step": 1147
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.682611465454102,
      "learning_rate": 9.928497994609741e-06,
      "loss": 1.558,
      "step": 1148
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7247278690338135,
      "learning_rate": 9.92830216593224e-06,
      "loss": 0.2185,
      "step": 1149
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.434350490570068,
      "learning_rate": 9.928106071391493e-06,
      "loss": 1.0111,
      "step": 1150
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.725472450256348,
      "learning_rate": 9.927909710998079e-06,
      "loss": 1.0502,
      "step": 1151
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.374797344207764,
      "learning_rate": 9.927713084762587e-06,
      "loss": 1.0385,
      "step": 1152
    },
    {
      "epoch": 0.08,
      "grad_norm": 19.51506805419922,
      "learning_rate": 9.927516192695625e-06,
      "loss": 1.935,
      "step": 1153
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.148155212402344,
      "learning_rate": 9.927319034807819e-06,
      "loss": 0.1305,
      "step": 1154
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0832293033599854,
      "learning_rate": 9.927121611109802e-06,
      "loss": 0.1001,
      "step": 1155
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.837215900421143,
      "learning_rate": 9.926923921612222e-06,
      "loss": 1.2088,
      "step": 1156
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.577264308929443,
      "learning_rate": 9.926725966325748e-06,
      "loss": 1.1725,
      "step": 1157
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.849409103393555,
      "learning_rate": 9.926527745261055e-06,
      "loss": 0.3606,
      "step": 1158
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.994741916656494,
      "learning_rate": 9.92632925842884e-06,
      "loss": 0.2176,
      "step": 1159
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.01142406463623,
      "learning_rate": 9.926130505839807e-06,
      "loss": 1.3942,
      "step": 1160
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.080160617828369,
      "learning_rate": 9.92593148750468e-06,
      "loss": 0.1632,
      "step": 1161
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.303258895874023,
      "learning_rate": 9.925732203434195e-06,
      "loss": 0.7089,
      "step": 1162
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.737188339233398,
      "learning_rate": 9.925532653639102e-06,
      "loss": 1.0758,
      "step": 1163
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.021963596343994,
      "learning_rate": 9.925332838130169e-06,
      "loss": 0.2207,
      "step": 1164
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.9313695430755615,
      "learning_rate": 9.92513275691817e-06,
      "loss": 0.1911,
      "step": 1165
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8915841579437256,
      "learning_rate": 9.924932410013902e-06,
      "loss": 0.1905,
      "step": 1166
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.826519012451172,
      "learning_rate": 9.924731797428174e-06,
      "loss": 0.2003,
      "step": 1167
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.555332660675049,
      "learning_rate": 9.924530919171806e-06,
      "loss": 1.0061,
      "step": 1168
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.9644157886505127,
      "learning_rate": 9.924329775255635e-06,
      "loss": 0.1604,
      "step": 1169
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.524539947509766,
      "learning_rate": 9.92412836569051e-06,
      "loss": 0.2247,
      "step": 1170
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.373236656188965,
      "learning_rate": 9.9239266904873e-06,
      "loss": 1.508,
      "step": 1171
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.6784231662750244,
      "learning_rate": 9.923724749656884e-06,
      "loss": 0.1928,
      "step": 1172
    },
    {
      "epoch": 0.08,
      "grad_norm": 37.26552200317383,
      "learning_rate": 9.923522543210155e-06,
      "loss": 1.3944,
      "step": 1173
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.886837482452393,
      "learning_rate": 9.923320071158022e-06,
      "loss": 0.9954,
      "step": 1174
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.06859302520752,
      "learning_rate": 9.923117333511405e-06,
      "loss": 0.894,
      "step": 1175
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.125773429870605,
      "learning_rate": 9.922914330281244e-06,
      "loss": 1.3989,
      "step": 1176
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.634920835494995,
      "learning_rate": 9.922711061478489e-06,
      "loss": 0.1983,
      "step": 1177
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.494105339050293,
      "learning_rate": 9.922507527114105e-06,
      "loss": 0.1739,
      "step": 1178
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9848899841308594,
      "learning_rate": 9.922303727199074e-06,
      "loss": 0.1668,
      "step": 1179
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.322721481323242,
      "learning_rate": 9.922099661744388e-06,
      "loss": 0.1672,
      "step": 1180
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.8892931938171387,
      "learning_rate": 9.921895330761055e-06,
      "loss": 0.2781,
      "step": 1181
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.6099178791046143,
      "learning_rate": 9.921690734260102e-06,
      "loss": 0.1479,
      "step": 1182
    },
    {
      "epoch": 0.08,
      "grad_norm": 17.97439956665039,
      "learning_rate": 9.921485872252563e-06,
      "loss": 1.5915,
      "step": 1183
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.646812438964844,
      "learning_rate": 9.921280744749488e-06,
      "loss": 1.4456,
      "step": 1184
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.405263900756836,
      "learning_rate": 9.921075351761947e-06,
      "loss": 0.2105,
      "step": 1185
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.987311601638794,
      "learning_rate": 9.920869693301016e-06,
      "loss": 0.1167,
      "step": 1186
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5945537090301514,
      "learning_rate": 9.920663769377794e-06,
      "loss": 0.2255,
      "step": 1187
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.577862739562988,
      "learning_rate": 9.920457580003385e-06,
      "loss": 0.9841,
      "step": 1188
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.4561824798584,
      "learning_rate": 9.920251125188915e-06,
      "loss": 1.4828,
      "step": 1189
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6343882083892822,
      "learning_rate": 9.920044404945522e-06,
      "loss": 0.141,
      "step": 1190
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.404818534851074,
      "learning_rate": 9.919837419284357e-06,
      "loss": 1.0955,
      "step": 1191
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.993502378463745,
      "learning_rate": 9.919630168216586e-06,
      "loss": 0.2136,
      "step": 1192
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.286456108093262,
      "learning_rate": 9.919422651753388e-06,
      "loss": 1.1253,
      "step": 1193
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.9939780235290527,
      "learning_rate": 9.919214869905959e-06,
      "loss": 0.1529,
      "step": 1194
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.152338027954102,
      "learning_rate": 9.919006822685508e-06,
      "loss": 0.1602,
      "step": 1195
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.7781901359558105,
      "learning_rate": 9.91879851010326e-06,
      "loss": 0.2568,
      "step": 1196
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.670380592346191,
      "learning_rate": 9.918589932170448e-06,
      "loss": 1.1631,
      "step": 1197
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.083042621612549,
      "learning_rate": 9.91838108889833e-06,
      "loss": 0.1948,
      "step": 1198
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.360422611236572,
      "learning_rate": 9.918171980298167e-06,
      "loss": 0.2372,
      "step": 1199
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.424715518951416,
      "learning_rate": 9.917962606381244e-06,
      "loss": 0.3137,
      "step": 1200
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.7883501052856445,
      "learning_rate": 9.917752967158852e-06,
      "loss": 0.3034,
      "step": 1201
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6628031730651855,
      "learning_rate": 9.917543062642303e-06,
      "loss": 0.2093,
      "step": 1202
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.01384162902832,
      "learning_rate": 9.917332892842922e-06,
      "loss": 1.1222,
      "step": 1203
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.9853618144989014,
      "learning_rate": 9.917122457772042e-06,
      "loss": 0.1866,
      "step": 1204
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0315933227539062,
      "learning_rate": 9.916911757441017e-06,
      "loss": 0.1367,
      "step": 1205
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.961125373840332,
      "learning_rate": 9.916700791861215e-06,
      "loss": 0.9358,
      "step": 1206
    },
    {
      "epoch": 0.09,
      "grad_norm": 20.302993774414062,
      "learning_rate": 9.916489561044016e-06,
      "loss": 1.0635,
      "step": 1207
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.953972816467285,
      "learning_rate": 9.916278065000815e-06,
      "loss": 0.9691,
      "step": 1208
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.146875381469727,
      "learning_rate": 9.916066303743023e-06,
      "loss": 1.6524,
      "step": 1209
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.033422470092773,
      "learning_rate": 9.915854277282061e-06,
      "loss": 1.2652,
      "step": 1210
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.007145881652832,
      "learning_rate": 9.915641985629368e-06,
      "loss": 0.1435,
      "step": 1211
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.660061836242676,
      "learning_rate": 9.915429428796397e-06,
      "loss": 0.8304,
      "step": 1212
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.758929252624512,
      "learning_rate": 9.915216606794615e-06,
      "loss": 1.3211,
      "step": 1213
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.1785173416137695,
      "learning_rate": 9.9150035196355e-06,
      "loss": 0.2197,
      "step": 1214
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.202383041381836,
      "learning_rate": 9.914790167330552e-06,
      "loss": 1.2988,
      "step": 1215
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.983184337615967,
      "learning_rate": 9.914576549891276e-06,
      "loss": 0.1499,
      "step": 1216
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.147887229919434,
      "learning_rate": 9.914362667329197e-06,
      "loss": 0.2062,
      "step": 1217
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.5463337898254395,
      "learning_rate": 9.914148519655857e-06,
      "loss": 0.9583,
      "step": 1218
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.621294975280762,
      "learning_rate": 9.913934106882805e-06,
      "loss": 1.2819,
      "step": 1219
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.37663459777832,
      "learning_rate": 9.913719429021607e-06,
      "loss": 1.6005,
      "step": 1220
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.1457953453063965,
      "learning_rate": 9.913504486083845e-06,
      "loss": 1.3099,
      "step": 1221
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.239138126373291,
      "learning_rate": 9.913289278081115e-06,
      "loss": 0.9378,
      "step": 1222
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.715312957763672,
      "learning_rate": 9.913073805025027e-06,
      "loss": 1.1756,
      "step": 1223
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.400039196014404,
      "learning_rate": 9.912858066927202e-06,
      "loss": 0.9488,
      "step": 1224
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.24488639831543,
      "learning_rate": 9.912642063799283e-06,
      "loss": 1.5385,
      "step": 1225
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.081244945526123,
      "learning_rate": 9.912425795652918e-06,
      "loss": 0.1628,
      "step": 1226
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.013315200805664,
      "learning_rate": 9.912209262499777e-06,
      "loss": 0.2801,
      "step": 1227
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9643380641937256,
      "learning_rate": 9.91199246435154e-06,
      "loss": 0.1035,
      "step": 1228
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.260542869567871,
      "learning_rate": 9.911775401219902e-06,
      "loss": 0.1721,
      "step": 1229
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.706669807434082,
      "learning_rate": 9.911558073116574e-06,
      "loss": 1.7866,
      "step": 1230
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1277925968170166,
      "learning_rate": 9.911340480053277e-06,
      "loss": 0.1236,
      "step": 1231
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.35386848449707,
      "learning_rate": 9.911122622041754e-06,
      "loss": 1.1085,
      "step": 1232
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.335397720336914,
      "learning_rate": 9.910904499093754e-06,
      "loss": 0.1351,
      "step": 1233
    },
    {
      "epoch": 0.09,
      "grad_norm": 43.67336654663086,
      "learning_rate": 9.910686111221047e-06,
      "loss": 0.9415,
      "step": 1234
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.552379846572876,
      "learning_rate": 9.91046745843541e-06,
      "loss": 0.2732,
      "step": 1235
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.812910318374634,
      "learning_rate": 9.910248540748642e-06,
      "loss": 0.1626,
      "step": 1236
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.903214454650879,
      "learning_rate": 9.91002935817255e-06,
      "loss": 1.146,
      "step": 1237
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.674181938171387,
      "learning_rate": 9.90980991071896e-06,
      "loss": 1.0851,
      "step": 1238
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6380395889282227,
      "learning_rate": 9.90959019839971e-06,
      "loss": 0.1626,
      "step": 1239
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.353751182556152,
      "learning_rate": 9.909370221226654e-06,
      "loss": 1.134,
      "step": 1240
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.15925407409668,
      "learning_rate": 9.909149979211656e-06,
      "loss": 1.761,
      "step": 1241
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.713423728942871,
      "learning_rate": 9.9089294723666e-06,
      "loss": 0.9968,
      "step": 1242
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0097172260284424,
      "learning_rate": 9.90870870070338e-06,
      "loss": 0.1558,
      "step": 1243
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.187614440917969,
      "learning_rate": 9.908487664233906e-06,
      "loss": 0.9433,
      "step": 1244
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.755218505859375,
      "learning_rate": 9.908266362970101e-06,
      "loss": 0.1198,
      "step": 1245
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.91285514831543,
      "learning_rate": 9.908044796923906e-06,
      "loss": 0.7977,
      "step": 1246
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.7427237033843994,
      "learning_rate": 9.90782296610727e-06,
      "loss": 0.254,
      "step": 1247
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.773791313171387,
      "learning_rate": 9.907600870532165e-06,
      "loss": 0.9663,
      "step": 1248
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.954700469970703,
      "learning_rate": 9.907378510210568e-06,
      "loss": 1.2283,
      "step": 1249
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0311198234558105,
      "learning_rate": 9.907155885154476e-06,
      "loss": 0.107,
      "step": 1250
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.75332260131836,
      "learning_rate": 9.906932995375898e-06,
      "loss": 1.2175,
      "step": 1251
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.8071506023406982,
      "learning_rate": 9.906709840886858e-06,
      "loss": 0.2145,
      "step": 1252
    },
    {
      "epoch": 0.09,
      "grad_norm": 34.0540771484375,
      "learning_rate": 9.906486421699397e-06,
      "loss": 1.1672,
      "step": 1253
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.524105548858643,
      "learning_rate": 9.906262737825565e-06,
      "loss": 0.8656,
      "step": 1254
    },
    {
      "epoch": 0.09,
      "grad_norm": 20.05674171447754,
      "learning_rate": 9.906038789277428e-06,
      "loss": 1.5368,
      "step": 1255
    },
    {
      "epoch": 0.09,
      "grad_norm": 33.3026237487793,
      "learning_rate": 9.90581457606707e-06,
      "loss": 1.0186,
      "step": 1256
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.666166305541992,
      "learning_rate": 9.905590098206585e-06,
      "loss": 1.4524,
      "step": 1257
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6718990802764893,
      "learning_rate": 9.905365355708084e-06,
      "loss": 0.2496,
      "step": 1258
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.809666872024536,
      "learning_rate": 9.905140348583688e-06,
      "loss": 0.1047,
      "step": 1259
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.355252265930176,
      "learning_rate": 9.904915076845538e-06,
      "loss": 0.671,
      "step": 1260
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.396124839782715,
      "learning_rate": 9.904689540505785e-06,
      "loss": 0.2185,
      "step": 1261
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.003965854644775,
      "learning_rate": 9.904463739576598e-06,
      "loss": 0.2548,
      "step": 1262
    },
    {
      "epoch": 0.09,
      "grad_norm": 22.737184524536133,
      "learning_rate": 9.904237674070155e-06,
      "loss": 1.571,
      "step": 1263
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0177433490753174,
      "learning_rate": 9.904011343998654e-06,
      "loss": 0.1502,
      "step": 1264
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.839568614959717,
      "learning_rate": 9.903784749374303e-06,
      "loss": 1.0384,
      "step": 1265
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3911306858062744,
      "learning_rate": 9.903557890209327e-06,
      "loss": 0.2139,
      "step": 1266
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.882798194885254,
      "learning_rate": 9.903330766515965e-06,
      "loss": 1.0251,
      "step": 1267
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5166635513305664,
      "learning_rate": 9.903103378306466e-06,
      "loss": 0.153,
      "step": 1268
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9299817085266113,
      "learning_rate": 9.9028757255931e-06,
      "loss": 0.1218,
      "step": 1269
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.26445198059082,
      "learning_rate": 9.902647808388149e-06,
      "loss": 0.3116,
      "step": 1270
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.1405510902404785,
      "learning_rate": 9.902419626703903e-06,
      "loss": 1.1915,
      "step": 1271
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.8506994247436523,
      "learning_rate": 9.902191180552676e-06,
      "loss": 0.2302,
      "step": 1272
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.785433530807495,
      "learning_rate": 9.90196246994679e-06,
      "loss": 0.1267,
      "step": 1273
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.521512985229492,
      "learning_rate": 9.901733494898585e-06,
      "loss": 0.1658,
      "step": 1274
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.3724775314331055,
      "learning_rate": 9.901504255420411e-06,
      "loss": 0.2893,
      "step": 1275
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.570353507995605,
      "learning_rate": 9.901274751524636e-06,
      "loss": 1.0925,
      "step": 1276
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.547623634338379,
      "learning_rate": 9.901044983223642e-06,
      "loss": 0.9487,
      "step": 1277
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.125594139099121,
      "learning_rate": 9.90081495052982e-06,
      "loss": 0.2743,
      "step": 1278
    },
    {
      "epoch": 0.09,
      "grad_norm": 22.07520866394043,
      "learning_rate": 9.900584653455584e-06,
      "loss": 0.6704,
      "step": 1279
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.291956663131714,
      "learning_rate": 9.900354092013355e-06,
      "loss": 0.1537,
      "step": 1280
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.659837245941162,
      "learning_rate": 9.90012326621557e-06,
      "loss": 0.1565,
      "step": 1281
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.420018672943115,
      "learning_rate": 9.899892176074687e-06,
      "loss": 0.1226,
      "step": 1282
    },
    {
      "epoch": 0.09,
      "grad_norm": 38.9077033996582,
      "learning_rate": 9.899660821603165e-06,
      "loss": 1.1737,
      "step": 1283
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3213651180267334,
      "learning_rate": 9.899429202813487e-06,
      "loss": 0.1032,
      "step": 1284
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.182603359222412,
      "learning_rate": 9.89919731971815e-06,
      "loss": 0.2812,
      "step": 1285
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.327254295349121,
      "learning_rate": 9.898965172329663e-06,
      "loss": 0.9321,
      "step": 1286
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.8287363052368164,
      "learning_rate": 9.898732760660549e-06,
      "loss": 0.2523,
      "step": 1287
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.174894332885742,
      "learning_rate": 9.898500084723344e-06,
      "loss": 0.9508,
      "step": 1288
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.760002136230469,
      "learning_rate": 9.8982671445306e-06,
      "loss": 0.2531,
      "step": 1289
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.237426519393921,
      "learning_rate": 9.898033940094886e-06,
      "loss": 0.1213,
      "step": 1290
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.396337509155273,
      "learning_rate": 9.89780047142878e-06,
      "loss": 1.6606,
      "step": 1291
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.549808502197266,
      "learning_rate": 9.89756673854488e-06,
      "loss": 1.8771,
      "step": 1292
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.444509506225586,
      "learning_rate": 9.897332741455791e-06,
      "loss": 0.232,
      "step": 1293
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.2459611892700195,
      "learning_rate": 9.897098480174138e-06,
      "loss": 0.9717,
      "step": 1294
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.492798805236816,
      "learning_rate": 9.896863954712558e-06,
      "loss": 1.1118,
      "step": 1295
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3044869899749756,
      "learning_rate": 9.896629165083705e-06,
      "loss": 0.1489,
      "step": 1296
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.099847316741943,
      "learning_rate": 9.896394111300242e-06,
      "loss": 0.2154,
      "step": 1297
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.182825088500977,
      "learning_rate": 9.896158793374852e-06,
      "loss": 1.1442,
      "step": 1298
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.527493476867676,
      "learning_rate": 9.895923211320227e-06,
      "loss": 0.1075,
      "step": 1299
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.565707206726074,
      "learning_rate": 9.895687365149077e-06,
      "loss": 0.7748,
      "step": 1300
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.401540756225586,
      "learning_rate": 9.895451254874124e-06,
      "loss": 1.2303,
      "step": 1301
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.685039520263672,
      "learning_rate": 9.895214880508107e-06,
      "loss": 1.4314,
      "step": 1302
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.463653087615967,
      "learning_rate": 9.894978242063777e-06,
      "loss": 1.0936,
      "step": 1303
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.074282169342041,
      "learning_rate": 9.894741339553898e-06,
      "loss": 0.1953,
      "step": 1304
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.88322925567627,
      "learning_rate": 9.894504172991252e-06,
      "loss": 1.3246,
      "step": 1305
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.876561164855957,
      "learning_rate": 9.894266742388633e-06,
      "loss": 0.9569,
      "step": 1306
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.449714183807373,
      "learning_rate": 9.894029047758848e-06,
      "loss": 1.0475,
      "step": 1307
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.23015022277832,
      "learning_rate": 9.893791089114722e-06,
      "loss": 0.3173,
      "step": 1308
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.47113037109375,
      "learning_rate": 9.89355286646909e-06,
      "loss": 0.1779,
      "step": 1309
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.317330360412598,
      "learning_rate": 9.893314379834804e-06,
      "loss": 1.0144,
      "step": 1310
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.81078815460205,
      "learning_rate": 9.89307562922473e-06,
      "loss": 1.3838,
      "step": 1311
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.36014461517334,
      "learning_rate": 9.892836614651747e-06,
      "loss": 1.3497,
      "step": 1312
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.929587364196777,
      "learning_rate": 9.89259733612875e-06,
      "loss": 1.3274,
      "step": 1313
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.8853936195373535,
      "learning_rate": 9.892357793668643e-06,
      "loss": 0.1681,
      "step": 1314
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.320490837097168,
      "learning_rate": 9.892117987284354e-06,
      "loss": 0.1095,
      "step": 1315
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.344123840332031,
      "learning_rate": 9.891877916988817e-06,
      "loss": 0.3117,
      "step": 1316
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.428056240081787,
      "learning_rate": 9.891637582794983e-06,
      "loss": 0.108,
      "step": 1317
    },
    {
      "epoch": 0.09,
      "grad_norm": 36.97405242919922,
      "learning_rate": 9.891396984715817e-06,
      "loss": 1.3685,
      "step": 1318
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.485784530639648,
      "learning_rate": 9.891156122764297e-06,
      "loss": 0.9804,
      "step": 1319
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.957452774047852,
      "learning_rate": 9.890914996953422e-06,
      "loss": 1.1654,
      "step": 1320
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.900535583496094,
      "learning_rate": 9.890673607296194e-06,
      "loss": 0.8844,
      "step": 1321
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.159750461578369,
      "learning_rate": 9.890431953805637e-06,
      "loss": 1.0114,
      "step": 1322
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7274351119995117,
      "learning_rate": 9.890190036494787e-06,
      "loss": 0.1294,
      "step": 1323
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.381221771240234,
      "learning_rate": 9.889947855376696e-06,
      "loss": 1.5988,
      "step": 1324
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.6459176540374756,
      "learning_rate": 9.889705410464428e-06,
      "loss": 0.1715,
      "step": 1325
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.416633605957031,
      "learning_rate": 9.88946270177106e-06,
      "loss": 0.9858,
      "step": 1326
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5822315216064453,
      "learning_rate": 9.889219729309686e-06,
      "loss": 0.2312,
      "step": 1327
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.1130428314209,
      "learning_rate": 9.888976493093417e-06,
      "loss": 1.8745,
      "step": 1328
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.8425168991088867,
      "learning_rate": 9.88873299313537e-06,
      "loss": 0.1305,
      "step": 1329
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.066742420196533,
      "learning_rate": 9.888489229448686e-06,
      "loss": 0.2346,
      "step": 1330
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.221304416656494,
      "learning_rate": 9.888245202046509e-06,
      "loss": 1.0399,
      "step": 1331
    },
    {
      "epoch": 0.1,
      "grad_norm": 41.7017707824707,
      "learning_rate": 9.888000910942006e-06,
      "loss": 1.7073,
      "step": 1332
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.257852554321289,
      "learning_rate": 9.887756356148359e-06,
      "loss": 1.1971,
      "step": 1333
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.866288900375366,
      "learning_rate": 9.887511537678754e-06,
      "loss": 0.1296,
      "step": 1334
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.605417013168335,
      "learning_rate": 9.887266455546404e-06,
      "loss": 0.1992,
      "step": 1335
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.364129066467285,
      "learning_rate": 9.887021109764527e-06,
      "loss": 1.7312,
      "step": 1336
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.61745834350586,
      "learning_rate": 9.886775500346362e-06,
      "loss": 1.4198,
      "step": 1337
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5403945446014404,
      "learning_rate": 9.886529627305152e-06,
      "loss": 0.2669,
      "step": 1338
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.591720581054688,
      "learning_rate": 9.88628349065417e-06,
      "loss": 1.0244,
      "step": 1339
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5660481452941895,
      "learning_rate": 9.886037090406685e-06,
      "loss": 0.1639,
      "step": 1340
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.7232089042663574,
      "learning_rate": 9.885790426575996e-06,
      "loss": 0.2295,
      "step": 1341
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.69907283782959,
      "learning_rate": 9.885543499175406e-06,
      "loss": 0.2272,
      "step": 1342
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.404844284057617,
      "learning_rate": 9.885296308218238e-06,
      "loss": 0.1217,
      "step": 1343
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.897635459899902,
      "learning_rate": 9.885048853717825e-06,
      "loss": 1.3515,
      "step": 1344
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.487855911254883,
      "learning_rate": 9.884801135687519e-06,
      "loss": 1.3098,
      "step": 1345
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.086919784545898,
      "learning_rate": 9.88455315414068e-06,
      "loss": 0.1065,
      "step": 1346
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.434908151626587,
      "learning_rate": 9.884304909090688e-06,
      "loss": 0.2614,
      "step": 1347
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.0309858322143555,
      "learning_rate": 9.884056400550934e-06,
      "loss": 0.2162,
      "step": 1348
    },
    {
      "epoch": 0.1,
      "grad_norm": 28.20795249938965,
      "learning_rate": 9.883807628534825e-06,
      "loss": 2.0709,
      "step": 1349
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.190674781799316,
      "learning_rate": 9.88355859305578e-06,
      "loss": 0.2043,
      "step": 1350
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.600149154663086,
      "learning_rate": 9.883309294127234e-06,
      "loss": 0.2471,
      "step": 1351
    },
    {
      "epoch": 0.1,
      "grad_norm": 35.286739349365234,
      "learning_rate": 9.883059731762637e-06,
      "loss": 0.9834,
      "step": 1352
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.768688201904297,
      "learning_rate": 9.88280990597545e-06,
      "loss": 0.1523,
      "step": 1353
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.60287857055664,
      "learning_rate": 9.882559816779152e-06,
      "loss": 1.4176,
      "step": 1354
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.280874252319336,
      "learning_rate": 9.882309464187231e-06,
      "loss": 0.337,
      "step": 1355
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.039989471435547,
      "learning_rate": 9.882058848213196e-06,
      "loss": 1.4771,
      "step": 1356
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.029749870300293,
      "learning_rate": 9.881807968870568e-06,
      "loss": 1.1698,
      "step": 1357
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.2182884216308594,
      "learning_rate": 9.881556826172875e-06,
      "loss": 0.1162,
      "step": 1358
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.291902542114258,
      "learning_rate": 9.881305420133671e-06,
      "loss": 1.1704,
      "step": 1359
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.6339921951293945,
      "learning_rate": 9.881053750766517e-06,
      "loss": 1.2081,
      "step": 1360
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8829705715179443,
      "learning_rate": 9.880801818084988e-06,
      "loss": 0.0998,
      "step": 1361
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.74262809753418,
      "learning_rate": 9.880549622102676e-06,
      "loss": 1.0738,
      "step": 1362
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.3757195472717285,
      "learning_rate": 9.880297162833186e-06,
      "loss": 0.2451,
      "step": 1363
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.027692794799805,
      "learning_rate": 9.880044440290136e-06,
      "loss": 0.2035,
      "step": 1364
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.499086380004883,
      "learning_rate": 9.879791454487161e-06,
      "loss": 1.3478,
      "step": 1365
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.58035945892334,
      "learning_rate": 9.879538205437909e-06,
      "loss": 1.1065,
      "step": 1366
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.37774658203125,
      "learning_rate": 9.87928469315604e-06,
      "loss": 1.4383,
      "step": 1367
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.119558811187744,
      "learning_rate": 9.87903091765523e-06,
      "loss": 0.1847,
      "step": 1368
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.485809803009033,
      "learning_rate": 9.878776878949171e-06,
      "loss": 0.8096,
      "step": 1369
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.622949600219727,
      "learning_rate": 9.878522577051567e-06,
      "loss": 1.4574,
      "step": 1370
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.176223754882812,
      "learning_rate": 9.878268011976135e-06,
      "loss": 1.1496,
      "step": 1371
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.628866195678711,
      "learning_rate": 9.87801318373661e-06,
      "loss": 0.1861,
      "step": 1372
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.452880859375,
      "learning_rate": 9.87775809234674e-06,
      "loss": 0.9863,
      "step": 1373
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.9767324924468994,
      "learning_rate": 9.877502737820279e-06,
      "loss": 0.236,
      "step": 1374
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.695029258728027,
      "learning_rate": 9.877247120171012e-06,
      "loss": 0.3188,
      "step": 1375
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.9201858043670654,
      "learning_rate": 9.876991239412723e-06,
      "loss": 0.1729,
      "step": 1376
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.472376823425293,
      "learning_rate": 9.876735095559217e-06,
      "loss": 1.1714,
      "step": 1377
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.81708025932312,
      "learning_rate": 9.87647868862431e-06,
      "loss": 0.1993,
      "step": 1378
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.036365985870361,
      "learning_rate": 9.876222018621839e-06,
      "loss": 0.3764,
      "step": 1379
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.364053249359131,
      "learning_rate": 9.875965085565648e-06,
      "loss": 0.8808,
      "step": 1380
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.663402557373047,
      "learning_rate": 9.875707889469595e-06,
      "loss": 0.9374,
      "step": 1381
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.816634178161621,
      "learning_rate": 9.875450430347558e-06,
      "loss": 1.4315,
      "step": 1382
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.054262161254883,
      "learning_rate": 9.875192708213425e-06,
      "loss": 0.2076,
      "step": 1383
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.815258979797363,
      "learning_rate": 9.874934723081099e-06,
      "loss": 1.58,
      "step": 1384
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.446205139160156,
      "learning_rate": 9.874676474964497e-06,
      "loss": 1.1479,
      "step": 1385
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.7314698696136475,
      "learning_rate": 9.87441796387755e-06,
      "loss": 0.2107,
      "step": 1386
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.64333724975586,
      "learning_rate": 9.874159189834206e-06,
      "loss": 1.3394,
      "step": 1387
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.899712562561035,
      "learning_rate": 9.873900152848424e-06,
      "loss": 0.9467,
      "step": 1388
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.655942916870117,
      "learning_rate": 9.873640852934176e-06,
      "loss": 1.6368,
      "step": 1389
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.568121910095215,
      "learning_rate": 9.873381290105451e-06,
      "loss": 1.4223,
      "step": 1390
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.635213851928711,
      "learning_rate": 9.873121464376255e-06,
      "loss": 0.1549,
      "step": 1391
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.03825569152832,
      "learning_rate": 9.872861375760598e-06,
      "loss": 1.5144,
      "step": 1392
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.0389010906219482,
      "learning_rate": 9.872601024272517e-06,
      "loss": 0.125,
      "step": 1393
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.451642990112305,
      "learning_rate": 9.872340409926053e-06,
      "loss": 0.2304,
      "step": 1394
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.245147705078125,
      "learning_rate": 9.872079532735268e-06,
      "loss": 1.2448,
      "step": 1395
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.36198902130127,
      "learning_rate": 9.871818392714234e-06,
      "loss": 1.3998,
      "step": 1396
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.560783863067627,
      "learning_rate": 9.871556989877038e-06,
      "loss": 1.1009,
      "step": 1397
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.011608600616455,
      "learning_rate": 9.871295324237783e-06,
      "loss": 0.9242,
      "step": 1398
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.78962230682373,
      "learning_rate": 9.871033395810583e-06,
      "loss": 1.2007,
      "step": 1399
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.724549293518066,
      "learning_rate": 9.870771204609571e-06,
      "loss": 0.6552,
      "step": 1400
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.148155450820923,
      "learning_rate": 9.870508750648887e-06,
      "loss": 0.1062,
      "step": 1401
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.152446746826172,
      "learning_rate": 9.870246033942694e-06,
      "loss": 0.1111,
      "step": 1402
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.449497222900391,
      "learning_rate": 9.869983054505161e-06,
      "loss": 1.1003,
      "step": 1403
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.959516763687134,
      "learning_rate": 9.869719812350475e-06,
      "loss": 0.1341,
      "step": 1404
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.410678863525391,
      "learning_rate": 9.86945630749284e-06,
      "loss": 0.3653,
      "step": 1405
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5219411849975586,
      "learning_rate": 9.869192539946468e-06,
      "loss": 0.1507,
      "step": 1406
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.864381790161133,
      "learning_rate": 9.868928509725591e-06,
      "loss": 1.7393,
      "step": 1407
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.947678565979004,
      "learning_rate": 9.86866421684445e-06,
      "loss": 1.076,
      "step": 1408
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.166224956512451,
      "learning_rate": 9.868399661317303e-06,
      "loss": 1.07,
      "step": 1409
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.579044342041016,
      "learning_rate": 9.868134843158421e-06,
      "loss": 1.6025,
      "step": 1410
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.378654718399048,
      "learning_rate": 9.867869762382093e-06,
      "loss": 0.1271,
      "step": 1411
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.841790199279785,
      "learning_rate": 9.867604419002616e-06,
      "loss": 0.9485,
      "step": 1412
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.130681037902832,
      "learning_rate": 9.867338813034307e-06,
      "loss": 0.9372,
      "step": 1413
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.3568830490112305,
      "learning_rate": 9.867072944491492e-06,
      "loss": 0.1838,
      "step": 1414
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.220094203948975,
      "learning_rate": 9.866806813388514e-06,
      "loss": 1.0828,
      "step": 1415
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.039098739624023,
      "learning_rate": 9.86654041973973e-06,
      "loss": 1.181,
      "step": 1416
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1317241191864014,
      "learning_rate": 9.866273763559512e-06,
      "loss": 0.1082,
      "step": 1417
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.074370384216309,
      "learning_rate": 9.866006844862246e-06,
      "loss": 0.2157,
      "step": 1418
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.655025005340576,
      "learning_rate": 9.865739663662326e-06,
      "loss": 1.1873,
      "step": 1419
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.771806716918945,
      "learning_rate": 9.86547221997417e-06,
      "loss": 1.6055,
      "step": 1420
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.634392261505127,
      "learning_rate": 9.865204513812208e-06,
      "loss": 1.0511,
      "step": 1421
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1192874908447266,
      "learning_rate": 9.864936545190876e-06,
      "loss": 0.1507,
      "step": 1422
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1123270988464355,
      "learning_rate": 9.86466831412463e-06,
      "loss": 0.1076,
      "step": 1423
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.030678749084473,
      "learning_rate": 9.864399820627944e-06,
      "loss": 1.0301,
      "step": 1424
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.7355809211730957,
      "learning_rate": 9.864131064715302e-06,
      "loss": 0.1822,
      "step": 1425
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.87462043762207,
      "learning_rate": 9.863862046401198e-06,
      "loss": 1.7813,
      "step": 1426
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.595348834991455,
      "learning_rate": 9.863592765700149e-06,
      "loss": 1.0358,
      "step": 1427
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.706003189086914,
      "learning_rate": 9.86332322262668e-06,
      "loss": 0.3289,
      "step": 1428
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.20396614074707,
      "learning_rate": 9.86305341719533e-06,
      "loss": 1.5221,
      "step": 1429
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.460779190063477,
      "learning_rate": 9.862783349420657e-06,
      "loss": 1.6012,
      "step": 1430
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.584102630615234,
      "learning_rate": 9.86251301931723e-06,
      "loss": 1.1275,
      "step": 1431
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.881664276123047,
      "learning_rate": 9.86224242689963e-06,
      "loss": 1.4889,
      "step": 1432
    },
    {
      "epoch": 0.1,
      "grad_norm": 20.019065856933594,
      "learning_rate": 9.861971572182456e-06,
      "loss": 1.91,
      "step": 1433
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.981452465057373,
      "learning_rate": 9.86170045518032e-06,
      "loss": 0.226,
      "step": 1434
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.843503952026367,
      "learning_rate": 9.861429075907845e-06,
      "loss": 1.0804,
      "step": 1435
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9785661697387695,
      "learning_rate": 9.861157434379673e-06,
      "loss": 0.1559,
      "step": 1436
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.265893936157227,
      "learning_rate": 9.86088553061046e-06,
      "loss": 0.3322,
      "step": 1437
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.506650924682617,
      "learning_rate": 9.86061336461487e-06,
      "loss": 0.7968,
      "step": 1438
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.669819831848145,
      "learning_rate": 9.860340936407589e-06,
      "loss": 1.3307,
      "step": 1439
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.489107131958008,
      "learning_rate": 9.860068246003311e-06,
      "loss": 0.9398,
      "step": 1440
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.317109107971191,
      "learning_rate": 9.859795293416748e-06,
      "loss": 0.3136,
      "step": 1441
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.734450340270996,
      "learning_rate": 9.859522078662625e-06,
      "loss": 0.8783,
      "step": 1442
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.798953533172607,
      "learning_rate": 9.859248601755679e-06,
      "loss": 0.3554,
      "step": 1443
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.178080558776855,
      "learning_rate": 9.858974862710664e-06,
      "loss": 1.5269,
      "step": 1444
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.092634677886963,
      "learning_rate": 9.858700861542348e-06,
      "loss": 0.8861,
      "step": 1445
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.676863670349121,
      "learning_rate": 9.858426598265511e-06,
      "loss": 1.0091,
      "step": 1446
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.443914413452148,
      "learning_rate": 9.85815207289495e-06,
      "loss": 1.1376,
      "step": 1447
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.935353994369507,
      "learning_rate": 9.857877285445473e-06,
      "loss": 0.0798,
      "step": 1448
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.221005439758301,
      "learning_rate": 9.857602235931904e-06,
      "loss": 0.2595,
      "step": 1449
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.724729537963867,
      "learning_rate": 9.857326924369082e-06,
      "loss": 0.2495,
      "step": 1450
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.091803550720215,
      "learning_rate": 9.85705135077186e-06,
      "loss": 1.6624,
      "step": 1451
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.61551570892334,
      "learning_rate": 9.856775515155101e-06,
      "loss": 0.1721,
      "step": 1452
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5468387603759766,
      "learning_rate": 9.856499417533687e-06,
      "loss": 0.1657,
      "step": 1453
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.410910606384277,
      "learning_rate": 9.856223057922512e-06,
      "loss": 0.2419,
      "step": 1454
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.183123588562012,
      "learning_rate": 9.855946436336486e-06,
      "loss": 0.9814,
      "step": 1455
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.231118202209473,
      "learning_rate": 9.85566955279053e-06,
      "loss": 0.2459,
      "step": 1456
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.619039535522461,
      "learning_rate": 9.85539240729958e-06,
      "loss": 0.1301,
      "step": 1457
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.20236873626709,
      "learning_rate": 9.85511499987859e-06,
      "loss": 0.8724,
      "step": 1458
    },
    {
      "epoch": 0.1,
      "grad_norm": 50.27764129638672,
      "learning_rate": 9.854837330542521e-06,
      "loss": 1.6613,
      "step": 1459
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.603214740753174,
      "learning_rate": 9.854559399306358e-06,
      "loss": 0.332,
      "step": 1460
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1328582763671875,
      "learning_rate": 9.854281206185088e-06,
      "loss": 0.0973,
      "step": 1461
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.8595962524414062,
      "learning_rate": 9.854002751193724e-06,
      "loss": 0.2502,
      "step": 1462
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.106230735778809,
      "learning_rate": 9.853724034347283e-06,
      "loss": 1.9094,
      "step": 1463
    },
    {
      "epoch": 0.1,
      "grad_norm": 33.106651306152344,
      "learning_rate": 9.853445055660803e-06,
      "loss": 1.5742,
      "step": 1464
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.661048650741577,
      "learning_rate": 9.853165815149333e-06,
      "loss": 0.2233,
      "step": 1465
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.546961307525635,
      "learning_rate": 9.852886312827938e-06,
      "loss": 0.2021,
      "step": 1466
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.508948802947998,
      "learning_rate": 9.852606548711697e-06,
      "loss": 0.1412,
      "step": 1467
    },
    {
      "epoch": 0.11,
      "grad_norm": 22.730987548828125,
      "learning_rate": 9.8523265228157e-06,
      "loss": 1.5609,
      "step": 1468
    },
    {
      "epoch": 0.11,
      "grad_norm": 19.392696380615234,
      "learning_rate": 9.852046235155053e-06,
      "loss": 1.3961,
      "step": 1469
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.977898359298706,
      "learning_rate": 9.85176568574488e-06,
      "loss": 0.3401,
      "step": 1470
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.200067043304443,
      "learning_rate": 9.85148487460031e-06,
      "loss": 1.2746,
      "step": 1471
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.010694980621338,
      "learning_rate": 9.851203801736496e-06,
      "loss": 1.0231,
      "step": 1472
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.7874670028686523,
      "learning_rate": 9.8509224671686e-06,
      "loss": 0.2458,
      "step": 1473
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6770577430725098,
      "learning_rate": 9.8506408709118e-06,
      "loss": 0.2153,
      "step": 1474
    },
    {
      "epoch": 0.11,
      "grad_norm": 17.384296417236328,
      "learning_rate": 9.850359012981284e-06,
      "loss": 1.5493,
      "step": 1475
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.7003204822540283,
      "learning_rate": 9.85007689339226e-06,
      "loss": 0.2635,
      "step": 1476
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.537616729736328,
      "learning_rate": 9.849794512159944e-06,
      "loss": 1.4053,
      "step": 1477
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.581594467163086,
      "learning_rate": 9.849511869299574e-06,
      "loss": 1.3709,
      "step": 1478
    },
    {
      "epoch": 0.11,
      "grad_norm": 20.104942321777344,
      "learning_rate": 9.849228964826394e-06,
      "loss": 1.7841,
      "step": 1479
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.9386098384857178,
      "learning_rate": 9.848945798755667e-06,
      "loss": 0.2319,
      "step": 1480
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.832286357879639,
      "learning_rate": 9.848662371102668e-06,
      "loss": 0.9543,
      "step": 1481
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.51386547088623,
      "learning_rate": 9.848378681882687e-06,
      "loss": 1.0204,
      "step": 1482
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.264516353607178,
      "learning_rate": 9.848094731111028e-06,
      "loss": 0.2757,
      "step": 1483
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.095860481262207,
      "learning_rate": 9.847810518803011e-06,
      "loss": 1.1423,
      "step": 1484
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.618060350418091,
      "learning_rate": 9.847526044973964e-06,
      "loss": 0.0944,
      "step": 1485
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7914655208587646,
      "learning_rate": 9.847241309639238e-06,
      "loss": 0.1975,
      "step": 1486
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.927055358886719,
      "learning_rate": 9.84695631281419e-06,
      "loss": 1.9698,
      "step": 1487
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.768080711364746,
      "learning_rate": 9.846671054514196e-06,
      "loss": 1.3171,
      "step": 1488
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.921008110046387,
      "learning_rate": 9.846385534754646e-06,
      "loss": 1.5087,
      "step": 1489
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.2641983032226562,
      "learning_rate": 9.846099753550938e-06,
      "loss": 0.1326,
      "step": 1490
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.776296615600586,
      "learning_rate": 9.845813710918492e-06,
      "loss": 0.9684,
      "step": 1491
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.5510149002075195,
      "learning_rate": 9.845527406872742e-06,
      "loss": 0.9496,
      "step": 1492
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.426608085632324,
      "learning_rate": 9.845240841429126e-06,
      "loss": 0.8368,
      "step": 1493
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.111198902130127,
      "learning_rate": 9.844954014603108e-06,
      "loss": 0.2712,
      "step": 1494
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.6145734786987305,
      "learning_rate": 9.844666926410161e-06,
      "loss": 1.0348,
      "step": 1495
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.68517017364502,
      "learning_rate": 9.84437957686577e-06,
      "loss": 1.1871,
      "step": 1496
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.7648448944091797,
      "learning_rate": 9.844091965985438e-06,
      "loss": 0.1544,
      "step": 1497
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.11738657951355,
      "learning_rate": 9.843804093784682e-06,
      "loss": 0.1514,
      "step": 1498
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.563495635986328,
      "learning_rate": 9.843515960279029e-06,
      "loss": 0.8733,
      "step": 1499
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.094138145446777,
      "learning_rate": 9.843227565484022e-06,
      "loss": 1.7637,
      "step": 1500
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.245957374572754,
      "learning_rate": 9.842938909415222e-06,
      "loss": 0.9174,
      "step": 1501
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.163412094116211,
      "learning_rate": 9.842649992088199e-06,
      "loss": 1.0415,
      "step": 1502
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.720177173614502,
      "learning_rate": 9.842360813518538e-06,
      "loss": 0.1299,
      "step": 1503
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.419018268585205,
      "learning_rate": 9.842071373721844e-06,
      "loss": 0.1484,
      "step": 1504
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.9379825592041,
      "learning_rate": 9.841781672713725e-06,
      "loss": 1.7216,
      "step": 1505
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.9781179428100586,
      "learning_rate": 9.841491710509813e-06,
      "loss": 0.217,
      "step": 1506
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.0593647956848145,
      "learning_rate": 9.841201487125749e-06,
      "loss": 0.3316,
      "step": 1507
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.06661605834961,
      "learning_rate": 9.840911002577189e-06,
      "loss": 1.0233,
      "step": 1508
    },
    {
      "epoch": 0.11,
      "grad_norm": 17.91058349609375,
      "learning_rate": 9.840620256879806e-06,
      "loss": 1.6078,
      "step": 1509
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.526772499084473,
      "learning_rate": 9.840329250049282e-06,
      "loss": 1.0516,
      "step": 1510
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.358610153198242,
      "learning_rate": 9.840037982101314e-06,
      "loss": 1.0054,
      "step": 1511
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6692354679107666,
      "learning_rate": 9.839746453051619e-06,
      "loss": 0.1754,
      "step": 1512
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.364377498626709,
      "learning_rate": 9.839454662915923e-06,
      "loss": 0.2804,
      "step": 1513
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.336708068847656,
      "learning_rate": 9.839162611709968e-06,
      "loss": 1.4946,
      "step": 1514
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6497724056243896,
      "learning_rate": 9.838870299449504e-06,
      "loss": 0.197,
      "step": 1515
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.260359525680542,
      "learning_rate": 9.838577726150305e-06,
      "loss": 0.1973,
      "step": 1516
    },
    {
      "epoch": 0.11,
      "grad_norm": 27.188873291015625,
      "learning_rate": 9.838284891828154e-06,
      "loss": 1.3266,
      "step": 1517
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.639321327209473,
      "learning_rate": 9.837991796498845e-06,
      "loss": 1.1247,
      "step": 1518
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.035640716552734,
      "learning_rate": 9.837698440178191e-06,
      "loss": 0.8321,
      "step": 1519
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.284334659576416,
      "learning_rate": 9.83740482288202e-06,
      "loss": 1.1486,
      "step": 1520
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.8620026111602783,
      "learning_rate": 9.837110944626168e-06,
      "loss": 0.2696,
      "step": 1521
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.8324520587921143,
      "learning_rate": 9.83681680542649e-06,
      "loss": 0.2162,
      "step": 1522
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.030567169189453,
      "learning_rate": 9.836522405298853e-06,
      "loss": 1.3291,
      "step": 1523
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.4104411602020264,
      "learning_rate": 9.836227744259141e-06,
      "loss": 0.2,
      "step": 1524
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.4746623039245605,
      "learning_rate": 9.835932822323247e-06,
      "loss": 0.2324,
      "step": 1525
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.516877174377441,
      "learning_rate": 9.835637639507082e-06,
      "loss": 0.9289,
      "step": 1526
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.8094494342803955,
      "learning_rate": 9.835342195826572e-06,
      "loss": 0.1877,
      "step": 1527
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.831025123596191,
      "learning_rate": 9.835046491297653e-06,
      "loss": 1.5836,
      "step": 1528
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.3984878063201904,
      "learning_rate": 9.834750525936277e-06,
      "loss": 0.063,
      "step": 1529
    },
    {
      "epoch": 0.11,
      "grad_norm": 17.481840133666992,
      "learning_rate": 9.83445429975841e-06,
      "loss": 1.598,
      "step": 1530
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.532362937927246,
      "learning_rate": 9.834157812780035e-06,
      "loss": 0.967,
      "step": 1531
    },
    {
      "epoch": 0.11,
      "grad_norm": 25.801666259765625,
      "learning_rate": 9.833861065017142e-06,
      "loss": 0.9181,
      "step": 1532
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.048044204711914,
      "learning_rate": 9.833564056485742e-06,
      "loss": 0.8459,
      "step": 1533
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.439344644546509,
      "learning_rate": 9.833266787201858e-06,
      "loss": 0.151,
      "step": 1534
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0660903453826904,
      "learning_rate": 9.832969257181525e-06,
      "loss": 0.1104,
      "step": 1535
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6125807762145996,
      "learning_rate": 9.832671466440795e-06,
      "loss": 0.1529,
      "step": 1536
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.822551727294922,
      "learning_rate": 9.83237341499573e-06,
      "loss": 0.3028,
      "step": 1537
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.469480037689209,
      "learning_rate": 9.832075102862414e-06,
      "loss": 0.2246,
      "step": 1538
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.2479305267334,
      "learning_rate": 9.831776530056936e-06,
      "loss": 2.0151,
      "step": 1539
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.16267728805542,
      "learning_rate": 9.831477696595403e-06,
      "loss": 1.0384,
      "step": 1540
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.934087753295898,
      "learning_rate": 9.831178602493935e-06,
      "loss": 1.4546,
      "step": 1541
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.057218551635742,
      "learning_rate": 9.830879247768669e-06,
      "loss": 0.8656,
      "step": 1542
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.44724464416504,
      "learning_rate": 9.830579632435754e-06,
      "loss": 1.172,
      "step": 1543
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.1214137077331543,
      "learning_rate": 9.830279756511352e-06,
      "loss": 0.2177,
      "step": 1544
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.376391887664795,
      "learning_rate": 9.829979620011641e-06,
      "loss": 0.1634,
      "step": 1545
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.645354270935059,
      "learning_rate": 9.829679222952813e-06,
      "loss": 0.8792,
      "step": 1546
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.340595245361328,
      "learning_rate": 9.829378565351071e-06,
      "loss": 1.0112,
      "step": 1547
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.650851249694824,
      "learning_rate": 9.829077647222636e-06,
      "loss": 0.1466,
      "step": 1548
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.780759811401367,
      "learning_rate": 9.82877646858374e-06,
      "loss": 0.1387,
      "step": 1549
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.4110825061798096,
      "learning_rate": 9.828475029450633e-06,
      "loss": 0.1817,
      "step": 1550
    },
    {
      "epoch": 0.11,
      "grad_norm": 21.192411422729492,
      "learning_rate": 9.828173329839574e-06,
      "loss": 1.2092,
      "step": 1551
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.136307239532471,
      "learning_rate": 9.827871369766841e-06,
      "loss": 0.1751,
      "step": 1552
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.439509391784668,
      "learning_rate": 9.82756914924872e-06,
      "loss": 1.1601,
      "step": 1553
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.9333388805389404,
      "learning_rate": 9.827266668301518e-06,
      "loss": 0.1568,
      "step": 1554
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.575227975845337,
      "learning_rate": 9.82696392694155e-06,
      "loss": 0.1656,
      "step": 1555
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.819108486175537,
      "learning_rate": 9.826660925185151e-06,
      "loss": 0.7991,
      "step": 1556
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.254276275634766,
      "learning_rate": 9.826357663048665e-06,
      "loss": 0.2378,
      "step": 1557
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0556535720825195,
      "learning_rate": 9.826054140548451e-06,
      "loss": 0.1864,
      "step": 1558
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.152022838592529,
      "learning_rate": 9.825750357700884e-06,
      "loss": 0.8445,
      "step": 1559
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.192803382873535,
      "learning_rate": 9.825446314522351e-06,
      "loss": 1.0057,
      "step": 1560
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.653295516967773,
      "learning_rate": 9.825142011029255e-06,
      "loss": 1.3228,
      "step": 1561
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.1938323974609375,
      "learning_rate": 9.824837447238012e-06,
      "loss": 0.9691,
      "step": 1562
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.140986442565918,
      "learning_rate": 9.824532623165051e-06,
      "loss": 0.1538,
      "step": 1563
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.575438022613525,
      "learning_rate": 9.824227538826818e-06,
      "loss": 0.2521,
      "step": 1564
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.122982978820801,
      "learning_rate": 9.823922194239769e-06,
      "loss": 0.262,
      "step": 1565
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.316142559051514,
      "learning_rate": 9.823616589420378e-06,
      "loss": 0.9315,
      "step": 1566
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.229416847229004,
      "learning_rate": 9.823310724385128e-06,
      "loss": 0.0852,
      "step": 1567
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.769855499267578,
      "learning_rate": 9.823004599150525e-06,
      "loss": 1.5289,
      "step": 1568
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.60298752784729,
      "learning_rate": 9.822698213733077e-06,
      "loss": 0.2086,
      "step": 1569
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.2221221923828125,
      "learning_rate": 9.822391568149318e-06,
      "loss": 0.1737,
      "step": 1570
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.4774768352508545,
      "learning_rate": 9.822084662415786e-06,
      "loss": 0.2619,
      "step": 1571
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.725266456604004,
      "learning_rate": 9.821777496549039e-06,
      "loss": 1.556,
      "step": 1572
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.73742413520813,
      "learning_rate": 9.821470070565648e-06,
      "loss": 0.2712,
      "step": 1573
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.556142807006836,
      "learning_rate": 9.821162384482196e-06,
      "loss": 1.572,
      "step": 1574
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.701679229736328,
      "learning_rate": 9.820854438315283e-06,
      "loss": 0.8097,
      "step": 1575
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.539632797241211,
      "learning_rate": 9.820546232081523e-06,
      "loss": 1.2942,
      "step": 1576
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8611741065979004,
      "learning_rate": 9.820237765797538e-06,
      "loss": 0.1485,
      "step": 1577
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.790428638458252,
      "learning_rate": 9.819929039479972e-06,
      "loss": 0.2565,
      "step": 1578
    },
    {
      "epoch": 0.11,
      "grad_norm": 22.979999542236328,
      "learning_rate": 9.819620053145478e-06,
      "loss": 1.9299,
      "step": 1579
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0872511863708496,
      "learning_rate": 9.819310806810727e-06,
      "loss": 0.1367,
      "step": 1580
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.709670066833496,
      "learning_rate": 9.819001300492398e-06,
      "loss": 1.1831,
      "step": 1581
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.250016689300537,
      "learning_rate": 9.818691534207192e-06,
      "loss": 0.1646,
      "step": 1582
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.812612295150757,
      "learning_rate": 9.818381507971815e-06,
      "loss": 0.2723,
      "step": 1583
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.8159570693969727,
      "learning_rate": 9.818071221802996e-06,
      "loss": 0.1781,
      "step": 1584
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.2544100284576416,
      "learning_rate": 9.81776067571747e-06,
      "loss": 0.2354,
      "step": 1585
    },
    {
      "epoch": 0.11,
      "grad_norm": 17.162288665771484,
      "learning_rate": 9.817449869731995e-06,
      "loss": 1.9566,
      "step": 1586
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.907869338989258,
      "learning_rate": 9.817138803863332e-06,
      "loss": 1.1086,
      "step": 1587
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.580927848815918,
      "learning_rate": 9.816827478128265e-06,
      "loss": 0.1919,
      "step": 1588
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.3630146980285645,
      "learning_rate": 9.81651589254359e-06,
      "loss": 0.1396,
      "step": 1589
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.131373882293701,
      "learning_rate": 9.816204047126111e-06,
      "loss": 0.2065,
      "step": 1590
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.4270212650299072,
      "learning_rate": 9.815891941892657e-06,
      "loss": 0.1391,
      "step": 1591
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.003899574279785,
      "learning_rate": 9.81557957686006e-06,
      "loss": 1.0993,
      "step": 1592
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.470678329467773,
      "learning_rate": 9.815266952045173e-06,
      "loss": 1.362,
      "step": 1593
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.69726824760437,
      "learning_rate": 9.81495406746486e-06,
      "loss": 0.2746,
      "step": 1594
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0907957553863525,
      "learning_rate": 9.814640923136003e-06,
      "loss": 0.2313,
      "step": 1595
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.239919185638428,
      "learning_rate": 9.81432751907549e-06,
      "loss": 0.8787,
      "step": 1596
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.841450691223145,
      "learning_rate": 9.814013855300231e-06,
      "loss": 1.4303,
      "step": 1597
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.041932106018066,
      "learning_rate": 9.813699931827147e-06,
      "loss": 0.8985,
      "step": 1598
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.489378452301025,
      "learning_rate": 9.813385748673175e-06,
      "loss": 0.1949,
      "step": 1599
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.2379817962646484,
      "learning_rate": 9.813071305855259e-06,
      "loss": 0.2142,
      "step": 1600
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.876520872116089,
      "learning_rate": 9.812756603390362e-06,
      "loss": 0.1796,
      "step": 1601
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.532240867614746,
      "learning_rate": 9.812441641295468e-06,
      "loss": 1.3312,
      "step": 1602
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.738271713256836,
      "learning_rate": 9.812126419587562e-06,
      "loss": 1.5074,
      "step": 1603
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.748154163360596,
      "learning_rate": 9.81181093828365e-06,
      "loss": 0.1821,
      "step": 1604
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0293095111846924,
      "learning_rate": 9.811495197400753e-06,
      "loss": 0.1352,
      "step": 1605
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7153337001800537,
      "learning_rate": 9.811179196955902e-06,
      "loss": 0.182,
      "step": 1606
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.736285209655762,
      "learning_rate": 9.810862936966144e-06,
      "loss": 0.1823,
      "step": 1607
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.550849914550781,
      "learning_rate": 9.81054641744854e-06,
      "loss": 1.5098,
      "step": 1608
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.259852409362793,
      "learning_rate": 9.810229638420166e-06,
      "loss": 1.1909,
      "step": 1609
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7279882431030273,
      "learning_rate": 9.809912599898112e-06,
      "loss": 0.1319,
      "step": 1610
    },
    {
      "epoch": 0.12,
      "grad_norm": 20.380329132080078,
      "learning_rate": 9.80959530189948e-06,
      "loss": 1.4254,
      "step": 1611
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.02982759475708,
      "learning_rate": 9.809277744441384e-06,
      "loss": 0.1446,
      "step": 1612
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.4762208461761475,
      "learning_rate": 9.808959927540959e-06,
      "loss": 0.1811,
      "step": 1613
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.317538261413574,
      "learning_rate": 9.80864185121535e-06,
      "loss": 0.961,
      "step": 1614
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.845116138458252,
      "learning_rate": 9.808323515481714e-06,
      "loss": 1.1649,
      "step": 1615
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.50478744506836,
      "learning_rate": 9.808004920357226e-06,
      "loss": 1.4429,
      "step": 1616
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1041171550750732,
      "learning_rate": 9.807686065859071e-06,
      "loss": 0.1846,
      "step": 1617
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.129193305969238,
      "learning_rate": 9.80736695200445e-06,
      "loss": 1.2049,
      "step": 1618
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.214450359344482,
      "learning_rate": 9.80704757881058e-06,
      "loss": 1.1211,
      "step": 1619
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.633133888244629,
      "learning_rate": 9.806727946294688e-06,
      "loss": 1.4134,
      "step": 1620
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9328982830047607,
      "learning_rate": 9.806408054474017e-06,
      "loss": 0.1701,
      "step": 1621
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.8803653717041016,
      "learning_rate": 9.806087903365826e-06,
      "loss": 0.1909,
      "step": 1622
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.993401527404785,
      "learning_rate": 9.805767492987385e-06,
      "loss": 1.3794,
      "step": 1623
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0597739219665527,
      "learning_rate": 9.805446823355978e-06,
      "loss": 0.1575,
      "step": 1624
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.839095115661621,
      "learning_rate": 9.805125894488905e-06,
      "loss": 1.2713,
      "step": 1625
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.60671615600586,
      "learning_rate": 9.804804706403476e-06,
      "loss": 1.0739,
      "step": 1626
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.120072364807129,
      "learning_rate": 9.804483259117024e-06,
      "loss": 0.2834,
      "step": 1627
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.117929458618164,
      "learning_rate": 9.804161552646882e-06,
      "loss": 0.8888,
      "step": 1628
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.28278923034668,
      "learning_rate": 9.803839587010413e-06,
      "loss": 0.9362,
      "step": 1629
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.525413513183594,
      "learning_rate": 9.80351736222498e-06,
      "loss": 1.9899,
      "step": 1630
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.738076210021973,
      "learning_rate": 9.803194878307967e-06,
      "loss": 0.8403,
      "step": 1631
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.750765800476074,
      "learning_rate": 9.802872135276771e-06,
      "loss": 0.2206,
      "step": 1632
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.79340124130249,
      "learning_rate": 9.802549133148804e-06,
      "loss": 1.1511,
      "step": 1633
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.198713779449463,
      "learning_rate": 9.80222587194149e-06,
      "loss": 0.2182,
      "step": 1634
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.266284942626953,
      "learning_rate": 9.80190235167227e-06,
      "loss": 1.1368,
      "step": 1635
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.770800828933716,
      "learning_rate": 9.80157857235859e-06,
      "loss": 0.1824,
      "step": 1636
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.648249626159668,
      "learning_rate": 9.801254534017923e-06,
      "loss": 0.9271,
      "step": 1637
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.581627368927002,
      "learning_rate": 9.800930236667748e-06,
      "loss": 0.2037,
      "step": 1638
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.147141456604004,
      "learning_rate": 9.80060568032556e-06,
      "loss": 0.1177,
      "step": 1639
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.817143678665161,
      "learning_rate": 9.800280865008864e-06,
      "loss": 0.0996,
      "step": 1640
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.0345516204834,
      "learning_rate": 9.799955790735187e-06,
      "loss": 1.448,
      "step": 1641
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5098440647125244,
      "learning_rate": 9.799630457522065e-06,
      "loss": 0.2209,
      "step": 1642
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.2241902351379395,
      "learning_rate": 9.799304865387045e-06,
      "loss": 0.3065,
      "step": 1643
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1580708026885986,
      "learning_rate": 9.798979014347696e-06,
      "loss": 0.2035,
      "step": 1644
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.6335439682006836,
      "learning_rate": 9.798652904421593e-06,
      "loss": 0.2089,
      "step": 1645
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.479948043823242,
      "learning_rate": 9.798326535626331e-06,
      "loss": 0.1214,
      "step": 1646
    },
    {
      "epoch": 0.12,
      "grad_norm": 23.307430267333984,
      "learning_rate": 9.797999907979515e-06,
      "loss": 2.1962,
      "step": 1647
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.232641220092773,
      "learning_rate": 9.797673021498764e-06,
      "loss": 1.5178,
      "step": 1648
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.460296154022217,
      "learning_rate": 9.797345876201714e-06,
      "loss": 0.1999,
      "step": 1649
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.37320613861084,
      "learning_rate": 9.797018472106013e-06,
      "loss": 0.2031,
      "step": 1650
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.320541381835938,
      "learning_rate": 9.796690809229323e-06,
      "loss": 1.5454,
      "step": 1651
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.154484987258911,
      "learning_rate": 9.796362887589321e-06,
      "loss": 0.1422,
      "step": 1652
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.747184753417969,
      "learning_rate": 9.796034707203697e-06,
      "loss": 1.2253,
      "step": 1653
    },
    {
      "epoch": 0.12,
      "grad_norm": 12.300724983215332,
      "learning_rate": 9.795706268090153e-06,
      "loss": 1.4428,
      "step": 1654
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.164287567138672,
      "learning_rate": 9.795377570266409e-06,
      "loss": 1.0052,
      "step": 1655
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.360777854919434,
      "learning_rate": 9.795048613750197e-06,
      "loss": 0.2467,
      "step": 1656
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.12449836730957,
      "learning_rate": 9.794719398559263e-06,
      "loss": 1.0433,
      "step": 1657
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.719354629516602,
      "learning_rate": 9.794389924711365e-06,
      "loss": 1.2868,
      "step": 1658
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.122771263122559,
      "learning_rate": 9.79406019222428e-06,
      "loss": 0.2572,
      "step": 1659
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.553377389907837,
      "learning_rate": 9.793730201115792e-06,
      "loss": 0.1403,
      "step": 1660
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.456069946289062,
      "learning_rate": 9.793399951403706e-06,
      "loss": 1.6437,
      "step": 1661
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.756826400756836,
      "learning_rate": 9.793069443105836e-06,
      "loss": 0.9937,
      "step": 1662
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.039206504821777,
      "learning_rate": 9.792738676240013e-06,
      "loss": 1.6254,
      "step": 1663
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.727163791656494,
      "learning_rate": 9.792407650824081e-06,
      "loss": 0.8653,
      "step": 1664
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.514394760131836,
      "learning_rate": 9.792076366875895e-06,
      "loss": 1.5615,
      "step": 1665
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1783065795898438,
      "learning_rate": 9.791744824413329e-06,
      "loss": 0.0892,
      "step": 1666
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.181712627410889,
      "learning_rate": 9.791413023454266e-06,
      "loss": 0.8125,
      "step": 1667
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.425811767578125,
      "learning_rate": 9.791080964016609e-06,
      "loss": 1.8423,
      "step": 1668
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.205951690673828,
      "learning_rate": 9.790748646118266e-06,
      "loss": 1.1819,
      "step": 1669
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.33835506439209,
      "learning_rate": 9.79041606977717e-06,
      "loss": 1.4455,
      "step": 1670
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.8455276489257812,
      "learning_rate": 9.79008323501126e-06,
      "loss": 0.1332,
      "step": 1671
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.6268486976623535,
      "learning_rate": 9.78975014183849e-06,
      "loss": 1.09,
      "step": 1672
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6451590061187744,
      "learning_rate": 9.78941679027683e-06,
      "loss": 0.1885,
      "step": 1673
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.48021125793457,
      "learning_rate": 9.789083180344263e-06,
      "loss": 1.8187,
      "step": 1674
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.242030143737793,
      "learning_rate": 9.788749312058787e-06,
      "loss": 0.9387,
      "step": 1675
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5782618522644043,
      "learning_rate": 9.788415185438411e-06,
      "loss": 0.1401,
      "step": 1676
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.801339149475098,
      "learning_rate": 9.78808080050116e-06,
      "loss": 1.1571,
      "step": 1677
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.361963272094727,
      "learning_rate": 9.787746157265077e-06,
      "loss": 0.9236,
      "step": 1678
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.825510501861572,
      "learning_rate": 9.787411255748207e-06,
      "loss": 1.0327,
      "step": 1679
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.671525001525879,
      "learning_rate": 9.787076095968624e-06,
      "loss": 0.9184,
      "step": 1680
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.373629570007324,
      "learning_rate": 9.786740677944407e-06,
      "loss": 0.9436,
      "step": 1681
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.876495122909546,
      "learning_rate": 9.786405001693646e-06,
      "loss": 0.1683,
      "step": 1682
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.4789788722991943,
      "learning_rate": 9.786069067234454e-06,
      "loss": 0.2073,
      "step": 1683
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.2245612144470215,
      "learning_rate": 9.785732874584951e-06,
      "loss": 0.9645,
      "step": 1684
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.210757255554199,
      "learning_rate": 9.785396423763276e-06,
      "loss": 0.7879,
      "step": 1685
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.243076801300049,
      "learning_rate": 9.785059714787579e-06,
      "loss": 1.2023,
      "step": 1686
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.526307106018066,
      "learning_rate": 9.78472274767602e-06,
      "loss": 0.1817,
      "step": 1687
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.964899063110352,
      "learning_rate": 9.78438552244678e-06,
      "loss": 0.2516,
      "step": 1688
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.1992926597595215,
      "learning_rate": 9.784048039118053e-06,
      "loss": 0.2403,
      "step": 1689
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.029666423797607,
      "learning_rate": 9.78371029770804e-06,
      "loss": 0.1616,
      "step": 1690
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.560153007507324,
      "learning_rate": 9.783372298234966e-06,
      "loss": 0.2474,
      "step": 1691
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.622244358062744,
      "learning_rate": 9.783034040717063e-06,
      "loss": 0.1291,
      "step": 1692
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.770482063293457,
      "learning_rate": 9.782695525172577e-06,
      "loss": 0.1946,
      "step": 1693
    },
    {
      "epoch": 0.12,
      "grad_norm": 25.00404930114746,
      "learning_rate": 9.78235675161977e-06,
      "loss": 1.159,
      "step": 1694
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.287704467773438,
      "learning_rate": 9.782017720076922e-06,
      "loss": 0.1913,
      "step": 1695
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.724944591522217,
      "learning_rate": 9.781678430562315e-06,
      "loss": 0.8709,
      "step": 1696
    },
    {
      "epoch": 0.12,
      "grad_norm": 25.669342041015625,
      "learning_rate": 9.781338883094258e-06,
      "loss": 1.6485,
      "step": 1697
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.199229717254639,
      "learning_rate": 9.780999077691067e-06,
      "loss": 0.2893,
      "step": 1698
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.257030487060547,
      "learning_rate": 9.780659014371072e-06,
      "loss": 1.0603,
      "step": 1699
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.985515594482422,
      "learning_rate": 9.78031869315262e-06,
      "loss": 1.4314,
      "step": 1700
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5077574253082275,
      "learning_rate": 9.779978114054067e-06,
      "loss": 0.1506,
      "step": 1701
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.666869640350342,
      "learning_rate": 9.77963727709379e-06,
      "loss": 0.1403,
      "step": 1702
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.887974262237549,
      "learning_rate": 9.779296182290173e-06,
      "loss": 0.9229,
      "step": 1703
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.3403947353363037,
      "learning_rate": 9.778954829661618e-06,
      "loss": 0.2681,
      "step": 1704
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.91504955291748,
      "learning_rate": 9.778613219226539e-06,
      "loss": 1.1455,
      "step": 1705
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.898677825927734,
      "learning_rate": 9.778271351003365e-06,
      "loss": 2.0563,
      "step": 1706
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.191592216491699,
      "learning_rate": 9.777929225010538e-06,
      "loss": 0.2741,
      "step": 1707
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.527915954589844,
      "learning_rate": 9.777586841266514e-06,
      "loss": 1.1861,
      "step": 1708
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.123993158340454,
      "learning_rate": 9.777244199789767e-06,
      "loss": 0.179,
      "step": 1709
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.237450361251831,
      "learning_rate": 9.776901300598775e-06,
      "loss": 0.1141,
      "step": 1710
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5513405799865723,
      "learning_rate": 9.77655814371204e-06,
      "loss": 0.098,
      "step": 1711
    },
    {
      "epoch": 0.12,
      "grad_norm": 24.710607528686523,
      "learning_rate": 9.776214729148075e-06,
      "loss": 1.8811,
      "step": 1712
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.56718921661377,
      "learning_rate": 9.775871056925403e-06,
      "loss": 1.2316,
      "step": 1713
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.3074774742126465,
      "learning_rate": 9.775527127062566e-06,
      "loss": 1.1871,
      "step": 1714
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.6833882331848145,
      "learning_rate": 9.775182939578115e-06,
      "loss": 0.984,
      "step": 1715
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.804471254348755,
      "learning_rate": 9.77483849449062e-06,
      "loss": 0.2123,
      "step": 1716
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.4174857139587402,
      "learning_rate": 9.774493791818663e-06,
      "loss": 0.2281,
      "step": 1717
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.501598834991455,
      "learning_rate": 9.774148831580838e-06,
      "loss": 0.1162,
      "step": 1718
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.550223350524902,
      "learning_rate": 9.773803613795757e-06,
      "loss": 0.9463,
      "step": 1719
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.963236808776855,
      "learning_rate": 9.773458138482037e-06,
      "loss": 1.6589,
      "step": 1720
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.966086387634277,
      "learning_rate": 9.77311240565832e-06,
      "loss": 0.9537,
      "step": 1721
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.448789119720459,
      "learning_rate": 9.772766415343257e-06,
      "loss": 0.1738,
      "step": 1722
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.79829978942871,
      "learning_rate": 9.772420167555512e-06,
      "loss": 1.5446,
      "step": 1723
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.3246753215789795,
      "learning_rate": 9.772073662313761e-06,
      "loss": 0.21,
      "step": 1724
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.2718505859375,
      "learning_rate": 9.771726899636701e-06,
      "loss": 1.8347,
      "step": 1725
    },
    {
      "epoch": 0.12,
      "grad_norm": 25.509859085083008,
      "learning_rate": 9.771379879543038e-06,
      "loss": 1.9206,
      "step": 1726
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6564621925354004,
      "learning_rate": 9.771032602051489e-06,
      "loss": 0.1561,
      "step": 1727
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.3145692348480225,
      "learning_rate": 9.770685067180792e-06,
      "loss": 0.1595,
      "step": 1728
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.142828464508057,
      "learning_rate": 9.770337274949693e-06,
      "loss": 1.017,
      "step": 1729
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.855834007263184,
      "learning_rate": 9.769989225376954e-06,
      "loss": 0.9796,
      "step": 1730
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.125455856323242,
      "learning_rate": 9.769640918481354e-06,
      "loss": 0.9437,
      "step": 1731
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.75150203704834,
      "learning_rate": 9.769292354281679e-06,
      "loss": 1.4006,
      "step": 1732
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.337838172912598,
      "learning_rate": 9.768943532796732e-06,
      "loss": 1.4794,
      "step": 1733
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5832948684692383,
      "learning_rate": 9.768594454045337e-06,
      "loss": 0.0973,
      "step": 1734
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7536070346832275,
      "learning_rate": 9.768245118046319e-06,
      "loss": 0.1925,
      "step": 1735
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.837528228759766,
      "learning_rate": 9.767895524818526e-06,
      "loss": 0.9396,
      "step": 1736
    },
    {
      "epoch": 0.12,
      "grad_norm": 15.641388893127441,
      "learning_rate": 9.767545674380816e-06,
      "loss": 1.0395,
      "step": 1737
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.677794933319092,
      "learning_rate": 9.767195566752064e-06,
      "loss": 0.9,
      "step": 1738
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.116865158081055,
      "learning_rate": 9.766845201951156e-06,
      "loss": 0.3645,
      "step": 1739
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.5513386726379395,
      "learning_rate": 9.766494579996992e-06,
      "loss": 1.0768,
      "step": 1740
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.40962791442871,
      "learning_rate": 9.766143700908489e-06,
      "loss": 1.1779,
      "step": 1741
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.3977344036102295,
      "learning_rate": 9.765792564704574e-06,
      "loss": 0.1342,
      "step": 1742
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0440356731414795,
      "learning_rate": 9.765441171404188e-06,
      "loss": 0.1843,
      "step": 1743
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.551520824432373,
      "learning_rate": 9.76508952102629e-06,
      "loss": 0.2199,
      "step": 1744
    },
    {
      "epoch": 0.13,
      "grad_norm": 19.4486141204834,
      "learning_rate": 9.76473761358985e-06,
      "loss": 1.2388,
      "step": 1745
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.087019920349121,
      "learning_rate": 9.764385449113852e-06,
      "loss": 0.0994,
      "step": 1746
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.4981956481933594,
      "learning_rate": 9.76403302761729e-06,
      "loss": 0.1957,
      "step": 1747
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.269723892211914,
      "learning_rate": 9.763680349119181e-06,
      "loss": 2.0845,
      "step": 1748
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.071861267089844,
      "learning_rate": 9.763327413638549e-06,
      "loss": 1.0808,
      "step": 1749
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0803637504577637,
      "learning_rate": 9.762974221194433e-06,
      "loss": 0.1598,
      "step": 1750
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.206488609313965,
      "learning_rate": 9.762620771805884e-06,
      "loss": 0.1736,
      "step": 1751
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.297355651855469,
      "learning_rate": 9.762267065491975e-06,
      "loss": 1.291,
      "step": 1752
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.6974258422851562,
      "learning_rate": 9.761913102271782e-06,
      "loss": 0.139,
      "step": 1753
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.3042826652526855,
      "learning_rate": 9.761558882164404e-06,
      "loss": 0.1641,
      "step": 1754
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.79511022567749,
      "learning_rate": 9.761204405188947e-06,
      "loss": 0.9868,
      "step": 1755
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.2280733585357666,
      "learning_rate": 9.760849671364533e-06,
      "loss": 0.2206,
      "step": 1756
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.5174787044525146,
      "learning_rate": 9.760494680710303e-06,
      "loss": 0.1496,
      "step": 1757
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.2350382804870605,
      "learning_rate": 9.760139433245403e-06,
      "loss": 0.9337,
      "step": 1758
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.3165178298950195,
      "learning_rate": 9.759783928988997e-06,
      "loss": 0.8923,
      "step": 1759
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.2796711921691895,
      "learning_rate": 9.759428167960267e-06,
      "loss": 0.8547,
      "step": 1760
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.106170177459717,
      "learning_rate": 9.7590721501784e-06,
      "loss": 0.2949,
      "step": 1761
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.169034481048584,
      "learning_rate": 9.758715875662608e-06,
      "loss": 0.9196,
      "step": 1762
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.087031841278076,
      "learning_rate": 9.758359344432107e-06,
      "loss": 0.1689,
      "step": 1763
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.887112617492676,
      "learning_rate": 9.75800255650613e-06,
      "loss": 0.9804,
      "step": 1764
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.795257568359375,
      "learning_rate": 9.757645511903924e-06,
      "loss": 0.1485,
      "step": 1765
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.719614028930664,
      "learning_rate": 9.75728821064475e-06,
      "loss": 1.0625,
      "step": 1766
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.180347442626953,
      "learning_rate": 9.756930652747887e-06,
      "loss": 1.1267,
      "step": 1767
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.511608600616455,
      "learning_rate": 9.756572838232621e-06,
      "loss": 1.0114,
      "step": 1768
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4796862602233887,
      "learning_rate": 9.756214767118255e-06,
      "loss": 0.0765,
      "step": 1769
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.458967208862305,
      "learning_rate": 9.755856439424105e-06,
      "loss": 0.8901,
      "step": 1770
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0006916522979736,
      "learning_rate": 9.755497855169502e-06,
      "loss": 0.1671,
      "step": 1771
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.8927783966064453,
      "learning_rate": 9.755139014373789e-06,
      "loss": 0.9578,
      "step": 1772
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.102849006652832,
      "learning_rate": 9.754779917056327e-06,
      "loss": 1.267,
      "step": 1773
    },
    {
      "epoch": 0.13,
      "grad_norm": 22.27863311767578,
      "learning_rate": 9.754420563236487e-06,
      "loss": 1.1573,
      "step": 1774
    },
    {
      "epoch": 0.13,
      "grad_norm": 31.563800811767578,
      "learning_rate": 9.754060952933652e-06,
      "loss": 1.367,
      "step": 1775
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.271739959716797,
      "learning_rate": 9.753701086167224e-06,
      "loss": 0.1473,
      "step": 1776
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.9839725494384766,
      "learning_rate": 9.753340962956617e-06,
      "loss": 0.2784,
      "step": 1777
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.655642032623291,
      "learning_rate": 9.752980583321258e-06,
      "loss": 0.9998,
      "step": 1778
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.357290267944336,
      "learning_rate": 9.752619947280587e-06,
      "loss": 1.2334,
      "step": 1779
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.136338710784912,
      "learning_rate": 9.752259054854058e-06,
      "loss": 0.386,
      "step": 1780
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.8029561042785645,
      "learning_rate": 9.751897906061143e-06,
      "loss": 0.812,
      "step": 1781
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.918668746948242,
      "learning_rate": 9.751536500921322e-06,
      "loss": 0.8354,
      "step": 1782
    },
    {
      "epoch": 0.13,
      "grad_norm": 26.73061180114746,
      "learning_rate": 9.751174839454094e-06,
      "loss": 1.3182,
      "step": 1783
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.477599859237671,
      "learning_rate": 9.750812921678965e-06,
      "loss": 0.2221,
      "step": 1784
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.9181809425354004,
      "learning_rate": 9.750450747615463e-06,
      "loss": 0.1828,
      "step": 1785
    },
    {
      "epoch": 0.13,
      "grad_norm": 24.223121643066406,
      "learning_rate": 9.750088317283125e-06,
      "loss": 1.141,
      "step": 1786
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4874024391174316,
      "learning_rate": 9.749725630701503e-06,
      "loss": 0.0811,
      "step": 1787
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.299013614654541,
      "learning_rate": 9.74936268789016e-06,
      "loss": 0.2008,
      "step": 1788
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.6095380783081055,
      "learning_rate": 9.748999488868677e-06,
      "loss": 0.1479,
      "step": 1789
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.8061633110046387,
      "learning_rate": 9.74863603365665e-06,
      "loss": 0.3199,
      "step": 1790
    },
    {
      "epoch": 0.13,
      "grad_norm": 31.802581787109375,
      "learning_rate": 9.748272322273681e-06,
      "loss": 1.5735,
      "step": 1791
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.961166381835938,
      "learning_rate": 9.747908354739394e-06,
      "loss": 1.3716,
      "step": 1792
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.307753562927246,
      "learning_rate": 9.747544131073425e-06,
      "loss": 0.9062,
      "step": 1793
    },
    {
      "epoch": 0.13,
      "grad_norm": 18.72035789489746,
      "learning_rate": 9.74717965129542e-06,
      "loss": 0.7314,
      "step": 1794
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6879374980926514,
      "learning_rate": 9.74681491542504e-06,
      "loss": 0.0872,
      "step": 1795
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7011666297912598,
      "learning_rate": 9.746449923481965e-06,
      "loss": 0.0909,
      "step": 1796
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.388063669204712,
      "learning_rate": 9.746084675485881e-06,
      "loss": 0.2858,
      "step": 1797
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.736567735671997,
      "learning_rate": 9.745719171456495e-06,
      "loss": 0.1894,
      "step": 1798
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.561927795410156,
      "learning_rate": 9.745353411413523e-06,
      "loss": 1.3186,
      "step": 1799
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.340452194213867,
      "learning_rate": 9.744987395376696e-06,
      "loss": 1.3309,
      "step": 1800
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.85903000831604,
      "learning_rate": 9.74462112336576e-06,
      "loss": 0.1943,
      "step": 1801
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.431395530700684,
      "learning_rate": 9.744254595400474e-06,
      "loss": 0.2712,
      "step": 1802
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.29075813293457,
      "learning_rate": 9.74388781150061e-06,
      "loss": 1.5806,
      "step": 1803
    },
    {
      "epoch": 0.13,
      "grad_norm": 20.498266220092773,
      "learning_rate": 9.743520771685957e-06,
      "loss": 1.6741,
      "step": 1804
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.791541576385498,
      "learning_rate": 9.743153475976312e-06,
      "loss": 0.1228,
      "step": 1805
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.438924551010132,
      "learning_rate": 9.74278592439149e-06,
      "loss": 0.1984,
      "step": 1806
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.496175765991211,
      "learning_rate": 9.742418116951322e-06,
      "loss": 1.1347,
      "step": 1807
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.7397966384887695,
      "learning_rate": 9.742050053675645e-06,
      "loss": 1.0798,
      "step": 1808
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.2502055168151855,
      "learning_rate": 9.74168173458432e-06,
      "loss": 0.1144,
      "step": 1809
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.389044761657715,
      "learning_rate": 9.741313159697212e-06,
      "loss": 1.4469,
      "step": 1810
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.745922803878784,
      "learning_rate": 9.740944329034206e-06,
      "loss": 0.1387,
      "step": 1811
    },
    {
      "epoch": 0.13,
      "grad_norm": 20.80751609802246,
      "learning_rate": 9.7405752426152e-06,
      "loss": 1.4698,
      "step": 1812
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0226242542266846,
      "learning_rate": 9.740205900460102e-06,
      "loss": 0.1232,
      "step": 1813
    },
    {
      "epoch": 0.13,
      "grad_norm": 25.5186824798584,
      "learning_rate": 9.739836302588839e-06,
      "loss": 1.3143,
      "step": 1814
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.1866607666015625,
      "learning_rate": 9.739466449021348e-06,
      "loss": 0.9491,
      "step": 1815
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.468506813049316,
      "learning_rate": 9.739096339777584e-06,
      "loss": 1.0425,
      "step": 1816
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.49345064163208,
      "learning_rate": 9.73872597487751e-06,
      "loss": 0.3029,
      "step": 1817
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.652796983718872,
      "learning_rate": 9.738355354341107e-06,
      "loss": 0.306,
      "step": 1818
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.971500396728516,
      "learning_rate": 9.737984478188368e-06,
      "loss": 0.3468,
      "step": 1819
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.1637020111084,
      "learning_rate": 9.7376133464393e-06,
      "loss": 1.2645,
      "step": 1820
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.342608451843262,
      "learning_rate": 9.737241959113925e-06,
      "loss": 1.4585,
      "step": 1821
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.869625091552734,
      "learning_rate": 9.736870316232278e-06,
      "loss": 1.2462,
      "step": 1822
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.8491058349609375,
      "learning_rate": 9.736498417814406e-06,
      "loss": 0.8432,
      "step": 1823
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.5918691158294678,
      "learning_rate": 9.736126263880376e-06,
      "loss": 0.2369,
      "step": 1824
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.936806678771973,
      "learning_rate": 9.735753854450258e-06,
      "loss": 1.1061,
      "step": 1825
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.620856285095215,
      "learning_rate": 9.735381189544149e-06,
      "loss": 1.0215,
      "step": 1826
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.139918088912964,
      "learning_rate": 9.735008269182147e-06,
      "loss": 0.19,
      "step": 1827
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.888744354248047,
      "learning_rate": 9.734635093384372e-06,
      "loss": 1.1774,
      "step": 1828
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.096431255340576,
      "learning_rate": 9.734261662170953e-06,
      "loss": 0.9412,
      "step": 1829
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7108426094055176,
      "learning_rate": 9.733887975562041e-06,
      "loss": 0.1405,
      "step": 1830
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.725452423095703,
      "learning_rate": 9.733514033577791e-06,
      "loss": 1.9014,
      "step": 1831
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.997764587402344,
      "learning_rate": 9.733139836238376e-06,
      "loss": 1.0637,
      "step": 1832
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.410281658172607,
      "learning_rate": 9.732765383563983e-06,
      "loss": 1.0295,
      "step": 1833
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.467117786407471,
      "learning_rate": 9.73239067557481e-06,
      "loss": 0.2537,
      "step": 1834
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.198081016540527,
      "learning_rate": 9.732015712291075e-06,
      "loss": 1.6181,
      "step": 1835
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.512094974517822,
      "learning_rate": 9.731640493733004e-06,
      "loss": 0.886,
      "step": 1836
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.863938331604004,
      "learning_rate": 9.731265019920836e-06,
      "loss": 1.0294,
      "step": 1837
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.403618812561035,
      "learning_rate": 9.730889290874833e-06,
      "loss": 1.3057,
      "step": 1838
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.81894850730896,
      "learning_rate": 9.730513306615258e-06,
      "loss": 0.1908,
      "step": 1839
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.486428737640381,
      "learning_rate": 9.730137067162396e-06,
      "loss": 0.1844,
      "step": 1840
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.490869998931885,
      "learning_rate": 9.729760572536542e-06,
      "loss": 0.8603,
      "step": 1841
    },
    {
      "epoch": 0.13,
      "grad_norm": 18.147123336791992,
      "learning_rate": 9.729383822758012e-06,
      "loss": 1.24,
      "step": 1842
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.596998453140259,
      "learning_rate": 9.729006817847124e-06,
      "loss": 0.1989,
      "step": 1843
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.739780902862549,
      "learning_rate": 9.728629557824218e-06,
      "loss": 0.2762,
      "step": 1844
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5540337562561035,
      "learning_rate": 9.728252042709648e-06,
      "loss": 0.1248,
      "step": 1845
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.927225589752197,
      "learning_rate": 9.727874272523777e-06,
      "loss": 1.3029,
      "step": 1846
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.2804172039031982,
      "learning_rate": 9.727496247286985e-06,
      "loss": 0.19,
      "step": 1847
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.1348934173584,
      "learning_rate": 9.727117967019664e-06,
      "loss": 2.3435,
      "step": 1848
    },
    {
      "epoch": 0.13,
      "grad_norm": 22.070127487182617,
      "learning_rate": 9.726739431742224e-06,
      "loss": 1.5385,
      "step": 1849
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.4765353202819824,
      "learning_rate": 9.726360641475081e-06,
      "loss": 0.147,
      "step": 1850
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.016174793243408,
      "learning_rate": 9.725981596238675e-06,
      "loss": 0.1231,
      "step": 1851
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.61055850982666,
      "learning_rate": 9.725602296053447e-06,
      "loss": 0.7741,
      "step": 1852
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0476877689361572,
      "learning_rate": 9.725222740939863e-06,
      "loss": 0.1301,
      "step": 1853
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.08096694946289,
      "learning_rate": 9.724842930918398e-06,
      "loss": 0.8194,
      "step": 1854
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.946939468383789,
      "learning_rate": 9.724462866009544e-06,
      "loss": 1.0039,
      "step": 1855
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.854684829711914,
      "learning_rate": 9.7240825462338e-06,
      "loss": 1.6536,
      "step": 1856
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.572047233581543,
      "learning_rate": 9.723701971611683e-06,
      "loss": 1.1654,
      "step": 1857
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.31001877784729,
      "learning_rate": 9.723321142163726e-06,
      "loss": 0.1729,
      "step": 1858
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.853214263916016,
      "learning_rate": 9.722940057910473e-06,
      "loss": 0.8716,
      "step": 1859
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.108271360397339,
      "learning_rate": 9.72255871887248e-06,
      "loss": 0.1677,
      "step": 1860
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.793107986450195,
      "learning_rate": 9.72217712507032e-06,
      "loss": 1.4145,
      "step": 1861
    },
    {
      "epoch": 0.13,
      "grad_norm": 19.52044677734375,
      "learning_rate": 9.721795276524579e-06,
      "loss": 1.1464,
      "step": 1862
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.4116082191467285,
      "learning_rate": 9.721413173255857e-06,
      "loss": 1.161,
      "step": 1863
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.654584884643555,
      "learning_rate": 9.721030815284763e-06,
      "loss": 1.7438,
      "step": 1864
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.546472549438477,
      "learning_rate": 9.720648202631927e-06,
      "loss": 0.8844,
      "step": 1865
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.58388090133667,
      "learning_rate": 9.72026533531799e-06,
      "loss": 0.1035,
      "step": 1866
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.719399929046631,
      "learning_rate": 9.719882213363605e-06,
      "loss": 0.9511,
      "step": 1867
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.6123592853546143,
      "learning_rate": 9.719498836789441e-06,
      "loss": 0.2212,
      "step": 1868
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.500956535339355,
      "learning_rate": 9.719115205616178e-06,
      "loss": 1.4,
      "step": 1869
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.3053205013275146,
      "learning_rate": 9.718731319864514e-06,
      "loss": 0.1884,
      "step": 1870
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.713679552078247,
      "learning_rate": 9.718347179555155e-06,
      "loss": 0.2544,
      "step": 1871
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.872351169586182,
      "learning_rate": 9.717962784708827e-06,
      "loss": 1.0839,
      "step": 1872
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.2468109130859375,
      "learning_rate": 9.717578135346264e-06,
      "loss": 0.8615,
      "step": 1873
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.42303729057312,
      "learning_rate": 9.717193231488218e-06,
      "loss": 0.187,
      "step": 1874
    },
    {
      "epoch": 0.13,
      "grad_norm": 32.528995513916016,
      "learning_rate": 9.716808073155454e-06,
      "loss": 1.1894,
      "step": 1875
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.761847496032715,
      "learning_rate": 9.716422660368747e-06,
      "loss": 0.9347,
      "step": 1876
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.244523048400879,
      "learning_rate": 9.71603699314889e-06,
      "loss": 1.2253,
      "step": 1877
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.2064666748046875,
      "learning_rate": 9.71565107151669e-06,
      "loss": 1.0826,
      "step": 1878
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.564750671386719,
      "learning_rate": 9.715264895492962e-06,
      "loss": 1.0165,
      "step": 1879
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.1277806758880615,
      "learning_rate": 9.714878465098543e-06,
      "loss": 0.1622,
      "step": 1880
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.157463788986206,
      "learning_rate": 9.714491780354277e-06,
      "loss": 0.0968,
      "step": 1881
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.939328670501709,
      "learning_rate": 9.714104841281024e-06,
      "loss": 0.9049,
      "step": 1882
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.17530632019043,
      "learning_rate": 9.713717647899662e-06,
      "loss": 1.6674,
      "step": 1883
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.03761100769043,
      "learning_rate": 9.71333020023107e-06,
      "loss": 1.4102,
      "step": 1884
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.598487377166748,
      "learning_rate": 9.712942498296159e-06,
      "loss": 0.2523,
      "step": 1885
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6315362453460693,
      "learning_rate": 9.712554542115838e-06,
      "loss": 0.0881,
      "step": 1886
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.033503532409668,
      "learning_rate": 9.712166331711037e-06,
      "loss": 0.2908,
      "step": 1887
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.52482795715332,
      "learning_rate": 9.7117778671027e-06,
      "loss": 0.2376,
      "step": 1888
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.432248592376709,
      "learning_rate": 9.711389148311781e-06,
      "loss": 0.3881,
      "step": 1889
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.324443817138672,
      "learning_rate": 9.711000175359251e-06,
      "loss": 1.8487,
      "step": 1890
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.89382266998291,
      "learning_rate": 9.710610948266096e-06,
      "loss": 1.1488,
      "step": 1891
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.659177303314209,
      "learning_rate": 9.710221467053308e-06,
      "loss": 0.2409,
      "step": 1892
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.261019706726074,
      "learning_rate": 9.709831731741902e-06,
      "loss": 1.2293,
      "step": 1893
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.6333088874816895,
      "learning_rate": 9.709441742352902e-06,
      "loss": 0.938,
      "step": 1894
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.6357839107513428,
      "learning_rate": 9.709051498907345e-06,
      "loss": 0.2617,
      "step": 1895
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.1840732097625732,
      "learning_rate": 9.708661001426287e-06,
      "loss": 0.1623,
      "step": 1896
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.9032986164093018,
      "learning_rate": 9.708270249930789e-06,
      "loss": 0.185,
      "step": 1897
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.567410469055176,
      "learning_rate": 9.707879244441933e-06,
      "loss": 1.2479,
      "step": 1898
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.59634780883789,
      "learning_rate": 9.707487984980814e-06,
      "loss": 0.9497,
      "step": 1899
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.531733512878418,
      "learning_rate": 9.707096471568536e-06,
      "loss": 0.1896,
      "step": 1900
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.500524520874023,
      "learning_rate": 9.70670470422622e-06,
      "loss": 1.129,
      "step": 1901
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.9372551441192627,
      "learning_rate": 9.706312682975002e-06,
      "loss": 0.13,
      "step": 1902
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.9145219326019287,
      "learning_rate": 9.705920407836026e-06,
      "loss": 0.3917,
      "step": 1903
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.251033306121826,
      "learning_rate": 9.70552787883046e-06,
      "loss": 0.1537,
      "step": 1904
    },
    {
      "epoch": 0.14,
      "grad_norm": 26.275514602661133,
      "learning_rate": 9.705135095979477e-06,
      "loss": 2.3602,
      "step": 1905
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.914057493209839,
      "learning_rate": 9.704742059304263e-06,
      "loss": 0.2622,
      "step": 1906
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.493529796600342,
      "learning_rate": 9.704348768826023e-06,
      "loss": 1.0005,
      "step": 1907
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.727938652038574,
      "learning_rate": 9.703955224565977e-06,
      "loss": 1.0126,
      "step": 1908
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.901433944702148,
      "learning_rate": 9.70356142654535e-06,
      "loss": 0.8173,
      "step": 1909
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.485055446624756,
      "learning_rate": 9.703167374785388e-06,
      "loss": 0.1692,
      "step": 1910
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.2351737022399902,
      "learning_rate": 9.70277306930735e-06,
      "loss": 0.1515,
      "step": 1911
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.7177064418792725,
      "learning_rate": 9.702378510132505e-06,
      "loss": 0.0779,
      "step": 1912
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.996959209442139,
      "learning_rate": 9.701983697282138e-06,
      "loss": 0.9416,
      "step": 1913
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.437577486038208,
      "learning_rate": 9.70158863077755e-06,
      "loss": 0.1918,
      "step": 1914
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.986425399780273,
      "learning_rate": 9.701193310640049e-06,
      "loss": 1.1042,
      "step": 1915
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.756556510925293,
      "learning_rate": 9.700797736890967e-06,
      "loss": 0.8512,
      "step": 1916
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.161729097366333,
      "learning_rate": 9.700401909551638e-06,
      "loss": 0.173,
      "step": 1917
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.344950675964355,
      "learning_rate": 9.70000582864342e-06,
      "loss": 0.97,
      "step": 1918
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.57353138923645,
      "learning_rate": 9.699609494187679e-06,
      "loss": 0.1137,
      "step": 1919
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.686438798904419,
      "learning_rate": 9.699212906205793e-06,
      "loss": 0.1191,
      "step": 1920
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.600643157958984,
      "learning_rate": 9.698816064719158e-06,
      "loss": 1.0114,
      "step": 1921
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.4354355335235596,
      "learning_rate": 9.698418969749181e-06,
      "loss": 0.1686,
      "step": 1922
    },
    {
      "epoch": 0.14,
      "grad_norm": 35.29090881347656,
      "learning_rate": 9.698021621317288e-06,
      "loss": 1.45,
      "step": 1923
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.6147713661193848,
      "learning_rate": 9.697624019444911e-06,
      "loss": 0.1605,
      "step": 1924
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.77957010269165,
      "learning_rate": 9.697226164153498e-06,
      "loss": 0.9345,
      "step": 1925
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.317383289337158,
      "learning_rate": 9.696828055464515e-06,
      "loss": 0.2982,
      "step": 1926
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.8618311882019043,
      "learning_rate": 9.696429693399436e-06,
      "loss": 0.1698,
      "step": 1927
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.221551895141602,
      "learning_rate": 9.696031077979752e-06,
      "loss": 1.5676,
      "step": 1928
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.4418320655822754,
      "learning_rate": 9.695632209226967e-06,
      "loss": 0.2256,
      "step": 1929
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.536098957061768,
      "learning_rate": 9.6952330871626e-06,
      "loss": 0.2536,
      "step": 1930
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.051658630371094,
      "learning_rate": 9.694833711808175e-06,
      "loss": 0.95,
      "step": 1931
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.020015716552734,
      "learning_rate": 9.694434083185247e-06,
      "loss": 1.8456,
      "step": 1932
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.191449165344238,
      "learning_rate": 9.69403420131537e-06,
      "loss": 1.1185,
      "step": 1933
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.836642265319824,
      "learning_rate": 9.693634066220112e-06,
      "loss": 0.5882,
      "step": 1934
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.7691516876220703,
      "learning_rate": 9.693233677921065e-06,
      "loss": 0.145,
      "step": 1935
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.3705732822418213,
      "learning_rate": 9.692833036439827e-06,
      "loss": 0.2351,
      "step": 1936
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.117159366607666,
      "learning_rate": 9.692432141798008e-06,
      "loss": 0.2616,
      "step": 1937
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.5018999576568604,
      "learning_rate": 9.692030994017236e-06,
      "loss": 0.2581,
      "step": 1938
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.1944193840026855,
      "learning_rate": 9.691629593119154e-06,
      "loss": 0.9444,
      "step": 1939
    },
    {
      "epoch": 0.14,
      "grad_norm": 25.02295684814453,
      "learning_rate": 9.691227939125414e-06,
      "loss": 1.711,
      "step": 1940
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.127017974853516,
      "learning_rate": 9.690826032057683e-06,
      "loss": 0.2455,
      "step": 1941
    },
    {
      "epoch": 0.14,
      "grad_norm": 52.57400894165039,
      "learning_rate": 9.690423871937645e-06,
      "loss": 1.9704,
      "step": 1942
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.86298942565918,
      "learning_rate": 9.690021458786991e-06,
      "loss": 1.2995,
      "step": 1943
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.279035568237305,
      "learning_rate": 9.689618792627434e-06,
      "loss": 0.88,
      "step": 1944
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.8075995445251465,
      "learning_rate": 9.689215873480694e-06,
      "loss": 0.1798,
      "step": 1945
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.921020030975342,
      "learning_rate": 9.688812701368506e-06,
      "loss": 1.0173,
      "step": 1946
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.046889305114746,
      "learning_rate": 9.688409276312622e-06,
      "loss": 0.2033,
      "step": 1947
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.2100741863250732,
      "learning_rate": 9.688005598334804e-06,
      "loss": 0.1675,
      "step": 1948
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.318879127502441,
      "learning_rate": 9.687601667456829e-06,
      "loss": 0.2062,
      "step": 1949
    },
    {
      "epoch": 0.14,
      "grad_norm": 19.71500015258789,
      "learning_rate": 9.687197483700488e-06,
      "loss": 0.9621,
      "step": 1950
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.527086019515991,
      "learning_rate": 9.686793047087584e-06,
      "loss": 0.1372,
      "step": 1951
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.269138813018799,
      "learning_rate": 9.686388357639936e-06,
      "loss": 0.1701,
      "step": 1952
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4478259086608887,
      "learning_rate": 9.685983415379374e-06,
      "loss": 0.0891,
      "step": 1953
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.671096801757812,
      "learning_rate": 9.685578220327744e-06,
      "loss": 1.4063,
      "step": 1954
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.65535306930542,
      "learning_rate": 9.685172772506908e-06,
      "loss": 0.2799,
      "step": 1955
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4812629222869873,
      "learning_rate": 9.684767071938732e-06,
      "loss": 0.1568,
      "step": 1956
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.134533882141113,
      "learning_rate": 9.684361118645108e-06,
      "loss": 0.9773,
      "step": 1957
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.8086166381835938,
      "learning_rate": 9.683954912647931e-06,
      "loss": 0.3123,
      "step": 1958
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.632165431976318,
      "learning_rate": 9.683548453969117e-06,
      "loss": 0.8978,
      "step": 1959
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.844097375869751,
      "learning_rate": 9.683141742630591e-06,
      "loss": 0.1313,
      "step": 1960
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.0809097290039062,
      "learning_rate": 9.682734778654294e-06,
      "loss": 0.3009,
      "step": 1961
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.902270793914795,
      "learning_rate": 9.682327562062183e-06,
      "loss": 0.1859,
      "step": 1962
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.335949897766113,
      "learning_rate": 9.681920092876222e-06,
      "loss": 1.1151,
      "step": 1963
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.827122688293457,
      "learning_rate": 9.681512371118395e-06,
      "loss": 0.9368,
      "step": 1964
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.252735137939453,
      "learning_rate": 9.681104396810696e-06,
      "loss": 1.282,
      "step": 1965
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.948546886444092,
      "learning_rate": 9.680696169975133e-06,
      "loss": 0.8984,
      "step": 1966
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.856147766113281,
      "learning_rate": 9.68028769063373e-06,
      "loss": 0.5983,
      "step": 1967
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.279843807220459,
      "learning_rate": 9.679878958808521e-06,
      "loss": 1.0233,
      "step": 1968
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.429913282394409,
      "learning_rate": 9.679469974521558e-06,
      "loss": 0.116,
      "step": 1969
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.893506050109863,
      "learning_rate": 9.6790607377949e-06,
      "loss": 1.0113,
      "step": 1970
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.77771282196045,
      "learning_rate": 9.678651248650628e-06,
      "loss": 0.983,
      "step": 1971
    },
    {
      "epoch": 0.14,
      "grad_norm": 24.499065399169922,
      "learning_rate": 9.678241507110833e-06,
      "loss": 1.3681,
      "step": 1972
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.207149028778076,
      "learning_rate": 9.677831513197614e-06,
      "loss": 0.6111,
      "step": 1973
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.5355567932128906,
      "learning_rate": 9.677421266933091e-06,
      "loss": 0.2315,
      "step": 1974
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.1260132789611816,
      "learning_rate": 9.677010768339397e-06,
      "loss": 0.1723,
      "step": 1975
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.676078796386719,
      "learning_rate": 9.676600017438675e-06,
      "loss": 0.8566,
      "step": 1976
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.154817581176758,
      "learning_rate": 9.676189014253085e-06,
      "loss": 0.457,
      "step": 1977
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.639553546905518,
      "learning_rate": 9.675777758804796e-06,
      "loss": 0.9528,
      "step": 1978
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.608999252319336,
      "learning_rate": 9.675366251115998e-06,
      "loss": 1.0004,
      "step": 1979
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.281124591827393,
      "learning_rate": 9.674954491208886e-06,
      "loss": 0.8433,
      "step": 1980
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.050676345825195,
      "learning_rate": 9.674542479105676e-06,
      "loss": 0.9414,
      "step": 1981
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.1263349056243896,
      "learning_rate": 9.674130214828592e-06,
      "loss": 0.1204,
      "step": 1982
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.352166175842285,
      "learning_rate": 9.673717698399878e-06,
      "loss": 0.9809,
      "step": 1983
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.373276710510254,
      "learning_rate": 9.673304929841782e-06,
      "loss": 0.2083,
      "step": 1984
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.889559030532837,
      "learning_rate": 9.672891909176577e-06,
      "loss": 0.2215,
      "step": 1985
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.220156669616699,
      "learning_rate": 9.67247863642654e-06,
      "loss": 0.316,
      "step": 1986
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.950822830200195,
      "learning_rate": 9.672065111613968e-06,
      "loss": 0.5252,
      "step": 1987
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.493032455444336,
      "learning_rate": 9.671651334761167e-06,
      "loss": 1.3089,
      "step": 1988
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.553999900817871,
      "learning_rate": 9.67123730589046e-06,
      "loss": 0.2848,
      "step": 1989
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.358596801757812,
      "learning_rate": 9.670823025024183e-06,
      "loss": 1.3653,
      "step": 1990
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.7284069061279297,
      "learning_rate": 9.670408492184682e-06,
      "loss": 0.1413,
      "step": 1991
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.760220527648926,
      "learning_rate": 9.66999370739432e-06,
      "loss": 0.18,
      "step": 1992
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.0202207565307617,
      "learning_rate": 9.669578670675476e-06,
      "loss": 0.0787,
      "step": 1993
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.806095123291016,
      "learning_rate": 9.669163382050539e-06,
      "loss": 0.2724,
      "step": 1994
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.558430194854736,
      "learning_rate": 9.668747841541912e-06,
      "loss": 1.125,
      "step": 1995
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.74130916595459,
      "learning_rate": 9.668332049172009e-06,
      "loss": 0.9979,
      "step": 1996
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.471333503723145,
      "learning_rate": 9.667916004963264e-06,
      "loss": 1.1064,
      "step": 1997
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.819490671157837,
      "learning_rate": 9.667499708938119e-06,
      "loss": 0.1503,
      "step": 1998
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.7089767456054688,
      "learning_rate": 9.66708316111903e-06,
      "loss": 0.1772,
      "step": 1999
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.876626491546631,
      "learning_rate": 9.666666361528474e-06,
      "loss": 0.0847,
      "step": 2000
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.518651008605957,
      "learning_rate": 9.666249310188932e-06,
      "loss": 0.1509,
      "step": 2001
    },
    {
      "epoch": 0.14,
      "grad_norm": 21.321840286254883,
      "learning_rate": 9.665832007122899e-06,
      "loss": 1.3613,
      "step": 2002
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.02342700958252,
      "learning_rate": 9.665414452352893e-06,
      "loss": 1.2135,
      "step": 2003
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.385352611541748,
      "learning_rate": 9.664996645901436e-06,
      "loss": 0.8382,
      "step": 2004
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.254938125610352,
      "learning_rate": 9.664578587791067e-06,
      "loss": 1.5994,
      "step": 2005
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.976360321044922,
      "learning_rate": 9.66416027804434e-06,
      "loss": 1.0769,
      "step": 2006
    },
    {
      "epoch": 0.14,
      "grad_norm": 20.51070213317871,
      "learning_rate": 9.663741716683822e-06,
      "loss": 1.8376,
      "step": 2007
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.777775764465332,
      "learning_rate": 9.66332290373209e-06,
      "loss": 1.0365,
      "step": 2008
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.6900811195373535,
      "learning_rate": 9.662903839211742e-06,
      "loss": 0.2221,
      "step": 2009
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.179304599761963,
      "learning_rate": 9.662484523145379e-06,
      "loss": 0.6858,
      "step": 2010
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.879019021987915,
      "learning_rate": 9.662064955555625e-06,
      "loss": 0.1801,
      "step": 2011
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.658252716064453,
      "learning_rate": 9.661645136465114e-06,
      "loss": 1.1143,
      "step": 2012
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.937427043914795,
      "learning_rate": 9.661225065896493e-06,
      "loss": 0.1975,
      "step": 2013
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.103673934936523,
      "learning_rate": 9.660804743872423e-06,
      "loss": 1.0256,
      "step": 2014
    },
    {
      "epoch": 0.14,
      "grad_norm": 45.66421127319336,
      "learning_rate": 9.66038417041558e-06,
      "loss": 1.276,
      "step": 2015
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.902955055236816,
      "learning_rate": 9.65996334554865e-06,
      "loss": 0.9083,
      "step": 2016
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.908502578735352,
      "learning_rate": 9.659542269294338e-06,
      "loss": 0.8058,
      "step": 2017
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.614635705947876,
      "learning_rate": 9.659120941675357e-06,
      "loss": 0.1112,
      "step": 2018
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.652543067932129,
      "learning_rate": 9.658699362714438e-06,
      "loss": 1.9035,
      "step": 2019
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.8334994316101074,
      "learning_rate": 9.658277532434322e-06,
      "loss": 0.1579,
      "step": 2020
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.282428741455078,
      "learning_rate": 9.657855450857765e-06,
      "loss": 1.076,
      "step": 2021
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.28358268737793,
      "learning_rate": 9.657433118007538e-06,
      "loss": 0.8165,
      "step": 2022
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.113452434539795,
      "learning_rate": 9.657010533906424e-06,
      "loss": 0.201,
      "step": 2023
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.40118932723999,
      "learning_rate": 9.656587698577219e-06,
      "loss": 0.2818,
      "step": 2024
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.328226089477539,
      "learning_rate": 9.656164612042734e-06,
      "loss": 0.8029,
      "step": 2025
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.380638122558594,
      "learning_rate": 9.655741274325791e-06,
      "loss": 1.0596,
      "step": 2026
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.87686824798584,
      "learning_rate": 9.655317685449233e-06,
      "loss": 0.1901,
      "step": 2027
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.217700958251953,
      "learning_rate": 9.654893845435904e-06,
      "loss": 0.1315,
      "step": 2028
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.343107223510742,
      "learning_rate": 9.654469754308673e-06,
      "loss": 0.9226,
      "step": 2029
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.222505569458008,
      "learning_rate": 9.654045412090415e-06,
      "loss": 0.8799,
      "step": 2030
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.91943359375,
      "learning_rate": 9.653620818804026e-06,
      "loss": 0.2101,
      "step": 2031
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.2541630268096924,
      "learning_rate": 9.653195974472407e-06,
      "loss": 0.1782,
      "step": 2032
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.494174957275391,
      "learning_rate": 9.65277087911848e-06,
      "loss": 0.8644,
      "step": 2033
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.460762977600098,
      "learning_rate": 9.652345532765174e-06,
      "loss": 1.029,
      "step": 2034
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.381223678588867,
      "learning_rate": 9.651919935435438e-06,
      "loss": 1.4632,
      "step": 2035
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.600857734680176,
      "learning_rate": 9.65149408715223e-06,
      "loss": 1.0416,
      "step": 2036
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.0146331787109375,
      "learning_rate": 9.651067987938522e-06,
      "loss": 0.1224,
      "step": 2037
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.279592514038086,
      "learning_rate": 9.650641637817301e-06,
      "loss": 0.9467,
      "step": 2038
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.7580864429473877,
      "learning_rate": 9.650215036811568e-06,
      "loss": 0.1016,
      "step": 2039
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.249646186828613,
      "learning_rate": 9.649788184944337e-06,
      "loss": 1.3601,
      "step": 2040
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.602764844894409,
      "learning_rate": 9.649361082238633e-06,
      "loss": 0.2008,
      "step": 2041
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.570150375366211,
      "learning_rate": 9.648933728717497e-06,
      "loss": 1.3167,
      "step": 2042
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.349982261657715,
      "learning_rate": 9.648506124403982e-06,
      "loss": 0.1702,
      "step": 2043
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.292166233062744,
      "learning_rate": 9.64807826932116e-06,
      "loss": 0.1379,
      "step": 2044
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.1849822998046875,
      "learning_rate": 9.647650163492108e-06,
      "loss": 0.1787,
      "step": 2045
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.5348587036132812,
      "learning_rate": 9.647221806939922e-06,
      "loss": 0.1801,
      "step": 2046
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.502243995666504,
      "learning_rate": 9.646793199687709e-06,
      "loss": 1.0153,
      "step": 2047
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.7667477130889893,
      "learning_rate": 9.646364341758593e-06,
      "loss": 0.1179,
      "step": 2048
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.522982597351074,
      "learning_rate": 9.64593523317571e-06,
      "loss": 0.9846,
      "step": 2049
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.3093719482421875,
      "learning_rate": 9.645505873962205e-06,
      "loss": 0.1545,
      "step": 2050
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.212429046630859,
      "learning_rate": 9.645076264141242e-06,
      "loss": 0.2977,
      "step": 2051
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.282095909118652,
      "learning_rate": 9.644646403735998e-06,
      "loss": 0.2193,
      "step": 2052
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.907639503479004,
      "learning_rate": 9.644216292769661e-06,
      "loss": 0.2182,
      "step": 2053
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.192800045013428,
      "learning_rate": 9.643785931265435e-06,
      "loss": 0.2816,
      "step": 2054
    },
    {
      "epoch": 0.15,
      "grad_norm": 28.081432342529297,
      "learning_rate": 9.643355319246535e-06,
      "loss": 1.3224,
      "step": 2055
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.602406978607178,
      "learning_rate": 9.64292445673619e-06,
      "loss": 0.834,
      "step": 2056
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.39701509475708,
      "learning_rate": 9.642493343757648e-06,
      "loss": 0.2699,
      "step": 2057
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4628865718841553,
      "learning_rate": 9.642061980334161e-06,
      "loss": 0.0903,
      "step": 2058
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.297352313995361,
      "learning_rate": 9.641630366489003e-06,
      "loss": 0.2381,
      "step": 2059
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.604198932647705,
      "learning_rate": 9.641198502245455e-06,
      "loss": 0.2223,
      "step": 2060
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.003594398498535,
      "learning_rate": 9.640766387626815e-06,
      "loss": 1.3301,
      "step": 2061
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.470142364501953,
      "learning_rate": 9.640334022656396e-06,
      "loss": 0.2118,
      "step": 2062
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.81695556640625,
      "learning_rate": 9.639901407357521e-06,
      "loss": 0.8938,
      "step": 2063
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.173506259918213,
      "learning_rate": 9.639468541753527e-06,
      "loss": 0.1218,
      "step": 2064
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.322266101837158,
      "learning_rate": 9.639035425867767e-06,
      "loss": 0.9092,
      "step": 2065
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.880090236663818,
      "learning_rate": 9.638602059723607e-06,
      "loss": 0.8103,
      "step": 2066
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.397453308105469,
      "learning_rate": 9.638168443344422e-06,
      "loss": 1.7016,
      "step": 2067
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.737143516540527,
      "learning_rate": 9.637734576753606e-06,
      "loss": 0.9254,
      "step": 2068
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.7768073081970215,
      "learning_rate": 9.637300459974566e-06,
      "loss": 0.1114,
      "step": 2069
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.019187927246094,
      "learning_rate": 9.636866093030717e-06,
      "loss": 1.9381,
      "step": 2070
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.863450765609741,
      "learning_rate": 9.636431475945495e-06,
      "loss": 0.3024,
      "step": 2071
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.55871868133545,
      "learning_rate": 9.635996608742345e-06,
      "loss": 0.8699,
      "step": 2072
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.745410442352295,
      "learning_rate": 9.635561491444726e-06,
      "loss": 0.9253,
      "step": 2073
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.3270716667175293,
      "learning_rate": 9.63512612407611e-06,
      "loss": 0.194,
      "step": 2074
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.421716213226318,
      "learning_rate": 9.634690506659986e-06,
      "loss": 0.9087,
      "step": 2075
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.019907474517822,
      "learning_rate": 9.634254639219852e-06,
      "loss": 0.2549,
      "step": 2076
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.129538536071777,
      "learning_rate": 9.633818521779221e-06,
      "loss": 1.2248,
      "step": 2077
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.7162580490112305,
      "learning_rate": 9.633382154361623e-06,
      "loss": 0.9469,
      "step": 2078
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.05525016784668,
      "learning_rate": 9.632945536990595e-06,
      "loss": 1.479,
      "step": 2079
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.558476209640503,
      "learning_rate": 9.632508669689691e-06,
      "loss": 0.2479,
      "step": 2080
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.349125862121582,
      "learning_rate": 9.632071552482479e-06,
      "loss": 1.5142,
      "step": 2081
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.30592942237854,
      "learning_rate": 9.631634185392541e-06,
      "loss": 0.1346,
      "step": 2082
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.732703685760498,
      "learning_rate": 9.631196568443471e-06,
      "loss": 0.7722,
      "step": 2083
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.303284645080566,
      "learning_rate": 9.630758701658874e-06,
      "loss": 1.0473,
      "step": 2084
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.2253623008728027,
      "learning_rate": 9.630320585062375e-06,
      "loss": 0.1928,
      "step": 2085
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.819886207580566,
      "learning_rate": 9.629882218677606e-06,
      "loss": 1.4058,
      "step": 2086
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.7655651569366455,
      "learning_rate": 9.629443602528217e-06,
      "loss": 0.1201,
      "step": 2087
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.0509510040283203,
      "learning_rate": 9.62900473663787e-06,
      "loss": 0.1291,
      "step": 2088
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.486306190490723,
      "learning_rate": 9.62856562103024e-06,
      "loss": 0.7924,
      "step": 2089
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.8189213275909424,
      "learning_rate": 9.628126255729013e-06,
      "loss": 0.2829,
      "step": 2090
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.117879390716553,
      "learning_rate": 9.627686640757891e-06,
      "loss": 1.2818,
      "step": 2091
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.894007682800293,
      "learning_rate": 9.627246776140595e-06,
      "loss": 1.0528,
      "step": 2092
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.79813814163208,
      "learning_rate": 9.62680666190085e-06,
      "loss": 0.2464,
      "step": 2093
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.362245559692383,
      "learning_rate": 9.626366298062397e-06,
      "loss": 1.4606,
      "step": 2094
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.622009754180908,
      "learning_rate": 9.625925684648995e-06,
      "loss": 0.216,
      "step": 2095
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.24604606628418,
      "learning_rate": 9.625484821684413e-06,
      "loss": 1.1491,
      "step": 2096
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.397596836090088,
      "learning_rate": 9.625043709192433e-06,
      "loss": 0.1712,
      "step": 2097
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.1068549156188965,
      "learning_rate": 9.624602347196852e-06,
      "loss": 0.1631,
      "step": 2098
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.088080406188965,
      "learning_rate": 9.624160735721479e-06,
      "loss": 0.9071,
      "step": 2099
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.822866201400757,
      "learning_rate": 9.623718874790137e-06,
      "loss": 0.2313,
      "step": 2100
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.92546272277832,
      "learning_rate": 9.623276764426663e-06,
      "loss": 0.8522,
      "step": 2101
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.987454652786255,
      "learning_rate": 9.62283440465491e-06,
      "loss": 0.8063,
      "step": 2102
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.3463878631591797,
      "learning_rate": 9.622391795498738e-06,
      "loss": 0.2282,
      "step": 2103
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.092488288879395,
      "learning_rate": 9.621948936982024e-06,
      "loss": 1.01,
      "step": 2104
    },
    {
      "epoch": 0.15,
      "grad_norm": 18.804054260253906,
      "learning_rate": 9.621505829128661e-06,
      "loss": 1.2637,
      "step": 2105
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.674654960632324,
      "learning_rate": 9.621062471962553e-06,
      "loss": 0.2101,
      "step": 2106
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.132452011108398,
      "learning_rate": 9.620618865507615e-06,
      "loss": 1.5145,
      "step": 2107
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.90686297416687,
      "learning_rate": 9.62017500978778e-06,
      "loss": 0.2545,
      "step": 2108
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.149988174438477,
      "learning_rate": 9.61973090482699e-06,
      "loss": 0.936,
      "step": 2109
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.207387924194336,
      "learning_rate": 9.619286550649205e-06,
      "loss": 1.4454,
      "step": 2110
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.833172559738159,
      "learning_rate": 9.618841947278396e-06,
      "loss": 0.2377,
      "step": 2111
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.7554965019226074,
      "learning_rate": 9.618397094738547e-06,
      "loss": 0.1376,
      "step": 2112
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.9417688846588135,
      "learning_rate": 9.617951993053658e-06,
      "loss": 0.1228,
      "step": 2113
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.103332042694092,
      "learning_rate": 9.617506642247735e-06,
      "loss": 0.1467,
      "step": 2114
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4369497299194336,
      "learning_rate": 9.617061042344812e-06,
      "loss": 0.2101,
      "step": 2115
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.27540922164917,
      "learning_rate": 9.616615193368917e-06,
      "loss": 0.2139,
      "step": 2116
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.260189056396484,
      "learning_rate": 9.61616909534411e-06,
      "loss": 0.2077,
      "step": 2117
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7176895141601562,
      "learning_rate": 9.615722748294454e-06,
      "loss": 0.1266,
      "step": 2118
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.2063136100769043,
      "learning_rate": 9.615276152244027e-06,
      "loss": 0.1766,
      "step": 2119
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.084796905517578,
      "learning_rate": 9.61482930721692e-06,
      "loss": 1.5701,
      "step": 2120
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.328549861907959,
      "learning_rate": 9.61438221323724e-06,
      "loss": 0.7312,
      "step": 2121
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.753108024597168,
      "learning_rate": 9.613934870329107e-06,
      "loss": 0.155,
      "step": 2122
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.399599075317383,
      "learning_rate": 9.61348727851665e-06,
      "loss": 0.118,
      "step": 2123
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.821772813796997,
      "learning_rate": 9.613039437824022e-06,
      "loss": 0.3218,
      "step": 2124
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.398229122161865,
      "learning_rate": 9.612591348275373e-06,
      "loss": 0.2922,
      "step": 2125
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.905421733856201,
      "learning_rate": 9.612143009894883e-06,
      "loss": 0.1819,
      "step": 2126
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.078972816467285,
      "learning_rate": 9.611694422706734e-06,
      "loss": 1.3185,
      "step": 2127
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.287263870239258,
      "learning_rate": 9.611245586735128e-06,
      "loss": 1.2496,
      "step": 2128
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6583175659179688,
      "learning_rate": 9.610796502004278e-06,
      "loss": 0.1556,
      "step": 2129
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.842867851257324,
      "learning_rate": 9.610347168538408e-06,
      "loss": 1.0724,
      "step": 2130
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.722309112548828,
      "learning_rate": 9.60989758636176e-06,
      "loss": 0.9986,
      "step": 2131
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.834226608276367,
      "learning_rate": 9.609447755498586e-06,
      "loss": 1.1434,
      "step": 2132
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.424683094024658,
      "learning_rate": 9.608997675973155e-06,
      "loss": 0.2662,
      "step": 2133
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.331483364105225,
      "learning_rate": 9.608547347809744e-06,
      "loss": 0.6632,
      "step": 2134
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.242280960083008,
      "learning_rate": 9.608096771032648e-06,
      "loss": 0.2159,
      "step": 2135
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.796021461486816,
      "learning_rate": 9.607645945666174e-06,
      "loss": 1.3303,
      "step": 2136
    },
    {
      "epoch": 0.15,
      "grad_norm": 17.17946434020996,
      "learning_rate": 9.607194871734643e-06,
      "loss": 0.9518,
      "step": 2137
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.571222305297852,
      "learning_rate": 9.606743549262386e-06,
      "loss": 1.6748,
      "step": 2138
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.91213321685791,
      "learning_rate": 9.606291978273752e-06,
      "loss": 1.0211,
      "step": 2139
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.663773536682129,
      "learning_rate": 9.605840158793102e-06,
      "loss": 1.0122,
      "step": 2140
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.582804203033447,
      "learning_rate": 9.60538809084481e-06,
      "loss": 0.8853,
      "step": 2141
    },
    {
      "epoch": 0.15,
      "grad_norm": 16.36347007751465,
      "learning_rate": 9.604935774453263e-06,
      "loss": 1.584,
      "step": 2142
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.17276930809021,
      "learning_rate": 9.604483209642861e-06,
      "loss": 0.1665,
      "step": 2143
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.481816291809082,
      "learning_rate": 9.60403039643802e-06,
      "loss": 0.13,
      "step": 2144
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.310188293457031,
      "learning_rate": 9.603577334863166e-06,
      "loss": 1.4227,
      "step": 2145
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.181827545166016,
      "learning_rate": 9.603124024942739e-06,
      "loss": 1.1901,
      "step": 2146
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.508474111557007,
      "learning_rate": 9.602670466701195e-06,
      "loss": 0.1184,
      "step": 2147
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.3281800746917725,
      "learning_rate": 9.602216660163001e-06,
      "loss": 0.2547,
      "step": 2148
    },
    {
      "epoch": 0.15,
      "grad_norm": 17.959049224853516,
      "learning_rate": 9.601762605352641e-06,
      "loss": 1.2629,
      "step": 2149
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.410063743591309,
      "learning_rate": 9.601308302294605e-06,
      "loss": 1.0391,
      "step": 2150
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.7345194816589355,
      "learning_rate": 9.600853751013402e-06,
      "loss": 0.9239,
      "step": 2151
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.639814615249634,
      "learning_rate": 9.600398951533557e-06,
      "loss": 0.1508,
      "step": 2152
    },
    {
      "epoch": 0.15,
      "grad_norm": 18.718307495117188,
      "learning_rate": 9.599943903879599e-06,
      "loss": 0.9197,
      "step": 2153
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.905706405639648,
      "learning_rate": 9.59948860807608e-06,
      "loss": 0.8791,
      "step": 2154
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.206671714782715,
      "learning_rate": 9.599033064147563e-06,
      "loss": 1.254,
      "step": 2155
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.285445213317871,
      "learning_rate": 9.598577272118618e-06,
      "loss": 1.1124,
      "step": 2156
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.26230812072754,
      "learning_rate": 9.598121232013837e-06,
      "loss": 1.2998,
      "step": 2157
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.205742835998535,
      "learning_rate": 9.59766494385782e-06,
      "loss": 1.5652,
      "step": 2158
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.532905101776123,
      "learning_rate": 9.597208407675181e-06,
      "loss": 1.0449,
      "step": 2159
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.373947143554688,
      "learning_rate": 9.596751623490553e-06,
      "loss": 0.965,
      "step": 2160
    },
    {
      "epoch": 0.15,
      "grad_norm": 21.28253936767578,
      "learning_rate": 9.596294591328571e-06,
      "loss": 1.3904,
      "step": 2161
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.081828594207764,
      "learning_rate": 9.595837311213896e-06,
      "loss": 0.1576,
      "step": 2162
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8179945945739746,
      "learning_rate": 9.595379783171192e-06,
      "loss": 0.2891,
      "step": 2163
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.159666538238525,
      "learning_rate": 9.594922007225144e-06,
      "loss": 0.7871,
      "step": 2164
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.0435791015625,
      "learning_rate": 9.594463983400449e-06,
      "loss": 1.3722,
      "step": 2165
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.511645317077637,
      "learning_rate": 9.594005711721811e-06,
      "loss": 1.2244,
      "step": 2166
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.1120476722717285,
      "learning_rate": 9.593547192213955e-06,
      "loss": 1.0625,
      "step": 2167
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.5556745529174805,
      "learning_rate": 9.593088424901615e-06,
      "loss": 1.0606,
      "step": 2168
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1314496994018555,
      "learning_rate": 9.59262940980954e-06,
      "loss": 0.1726,
      "step": 2169
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1161000728607178,
      "learning_rate": 9.592170146962492e-06,
      "loss": 0.1685,
      "step": 2170
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.58799171447754,
      "learning_rate": 9.591710636385248e-06,
      "loss": 1.0365,
      "step": 2171
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.226937770843506,
      "learning_rate": 9.591250878102594e-06,
      "loss": 0.8211,
      "step": 2172
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7329559326171875,
      "learning_rate": 9.590790872139335e-06,
      "loss": 0.0745,
      "step": 2173
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.249016761779785,
      "learning_rate": 9.590330618520284e-06,
      "loss": 0.2498,
      "step": 2174
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.65255069732666,
      "learning_rate": 9.589870117270274e-06,
      "loss": 0.8772,
      "step": 2175
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.6043548583984375,
      "learning_rate": 9.589409368414144e-06,
      "loss": 0.1525,
      "step": 2176
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.029115676879883,
      "learning_rate": 9.58894837197675e-06,
      "loss": 1.799,
      "step": 2177
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.38209342956543,
      "learning_rate": 9.588487127982962e-06,
      "loss": 1.064,
      "step": 2178
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.891541481018066,
      "learning_rate": 9.588025636457662e-06,
      "loss": 1.552,
      "step": 2179
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.031394958496094,
      "learning_rate": 9.587563897425744e-06,
      "loss": 0.1797,
      "step": 2180
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.518272399902344,
      "learning_rate": 9.587101910912121e-06,
      "loss": 1.5965,
      "step": 2181
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.857667922973633,
      "learning_rate": 9.586639676941712e-06,
      "loss": 0.94,
      "step": 2182
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.654722690582275,
      "learning_rate": 9.586177195539452e-06,
      "loss": 0.9878,
      "step": 2183
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.668612480163574,
      "learning_rate": 9.585714466730294e-06,
      "loss": 1.7508,
      "step": 2184
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5507304668426514,
      "learning_rate": 9.585251490539198e-06,
      "loss": 0.1706,
      "step": 2185
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.6984939575195312,
      "learning_rate": 9.58478826699114e-06,
      "loss": 0.1978,
      "step": 2186
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.462133407592773,
      "learning_rate": 9.58432479611111e-06,
      "loss": 1.281,
      "step": 2187
    },
    {
      "epoch": 0.16,
      "grad_norm": 17.16120147705078,
      "learning_rate": 9.583861077924109e-06,
      "loss": 1.2168,
      "step": 2188
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.661581516265869,
      "learning_rate": 9.583397112455152e-06,
      "loss": 1.0081,
      "step": 2189
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.894930124282837,
      "learning_rate": 9.582932899729274e-06,
      "loss": 0.2345,
      "step": 2190
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.368805408477783,
      "learning_rate": 9.582468439771511e-06,
      "loss": 0.1374,
      "step": 2191
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.9738895893096924,
      "learning_rate": 9.582003732606921e-06,
      "loss": 0.1601,
      "step": 2192
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.030647277832031,
      "learning_rate": 9.581538778260574e-06,
      "loss": 0.1909,
      "step": 2193
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.253812789916992,
      "learning_rate": 9.581073576757551e-06,
      "loss": 1.0063,
      "step": 2194
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5482139587402344,
      "learning_rate": 9.58060812812295e-06,
      "loss": 0.2197,
      "step": 2195
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.065762758255005,
      "learning_rate": 9.580142432381881e-06,
      "loss": 0.235,
      "step": 2196
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.521998405456543,
      "learning_rate": 9.579676489559462e-06,
      "loss": 1.1436,
      "step": 2197
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.948287010192871,
      "learning_rate": 9.57921029968083e-06,
      "loss": 1.0847,
      "step": 2198
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.276700973510742,
      "learning_rate": 9.578743862771137e-06,
      "loss": 0.2241,
      "step": 2199
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.410935640335083,
      "learning_rate": 9.578277178855545e-06,
      "loss": 0.2677,
      "step": 2200
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.521038770675659,
      "learning_rate": 9.57781024795923e-06,
      "loss": 0.1744,
      "step": 2201
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.3100171089172363,
      "learning_rate": 9.577343070107377e-06,
      "loss": 0.19,
      "step": 2202
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.349071979522705,
      "learning_rate": 9.576875645325193e-06,
      "loss": 0.718,
      "step": 2203
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.636296272277832,
      "learning_rate": 9.576407973637892e-06,
      "loss": 1.5029,
      "step": 2204
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.645451068878174,
      "learning_rate": 9.575940055070705e-06,
      "loss": 0.1979,
      "step": 2205
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.388911247253418,
      "learning_rate": 9.575471889648871e-06,
      "loss": 0.9764,
      "step": 2206
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.985505104064941,
      "learning_rate": 9.57500347739765e-06,
      "loss": 1.557,
      "step": 2207
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.417073726654053,
      "learning_rate": 9.574534818342304e-06,
      "loss": 0.8681,
      "step": 2208
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.980625867843628,
      "learning_rate": 9.574065912508124e-06,
      "loss": 0.1762,
      "step": 2209
    },
    {
      "epoch": 0.16,
      "grad_norm": 29.866716384887695,
      "learning_rate": 9.573596759920401e-06,
      "loss": 1.7886,
      "step": 2210
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.233823299407959,
      "learning_rate": 9.573127360604445e-06,
      "loss": 0.2304,
      "step": 2211
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.026279926300049,
      "learning_rate": 9.572657714585578e-06,
      "loss": 0.2547,
      "step": 2212
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.76287317276001,
      "learning_rate": 9.572187821889136e-06,
      "loss": 0.9739,
      "step": 2213
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.947810649871826,
      "learning_rate": 9.571717682540467e-06,
      "loss": 0.1757,
      "step": 2214
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8120646476745605,
      "learning_rate": 9.571247296564933e-06,
      "loss": 0.1962,
      "step": 2215
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.7111146450042725,
      "learning_rate": 9.570776663987914e-06,
      "loss": 0.163,
      "step": 2216
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.0019636154174805,
      "learning_rate": 9.570305784834792e-06,
      "loss": 0.234,
      "step": 2217
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.366501808166504,
      "learning_rate": 9.569834659130974e-06,
      "loss": 1.2528,
      "step": 2218
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.3857038021087646,
      "learning_rate": 9.569363286901873e-06,
      "loss": 0.1601,
      "step": 2219
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.013938903808594,
      "learning_rate": 9.568891668172919e-06,
      "loss": 1.5063,
      "step": 2220
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.655313730239868,
      "learning_rate": 9.568419802969553e-06,
      "loss": 0.1251,
      "step": 2221
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.9371883869171143,
      "learning_rate": 9.567947691317231e-06,
      "loss": 0.1021,
      "step": 2222
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.110422134399414,
      "learning_rate": 9.567475333241423e-06,
      "loss": 0.2096,
      "step": 2223
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.9522552490234375,
      "learning_rate": 9.567002728767609e-06,
      "loss": 0.1394,
      "step": 2224
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.423377275466919,
      "learning_rate": 9.566529877921284e-06,
      "loss": 0.1571,
      "step": 2225
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1707210540771484,
      "learning_rate": 9.56605678072796e-06,
      "loss": 0.1718,
      "step": 2226
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.131199836730957,
      "learning_rate": 9.565583437213154e-06,
      "loss": 1.1226,
      "step": 2227
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.3590433597564697,
      "learning_rate": 9.565109847402403e-06,
      "loss": 0.1927,
      "step": 2228
    },
    {
      "epoch": 0.16,
      "grad_norm": 17.21759033203125,
      "learning_rate": 9.564636011321256e-06,
      "loss": 1.574,
      "step": 2229
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.180258750915527,
      "learning_rate": 9.564161928995275e-06,
      "loss": 1.2397,
      "step": 2230
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.925943374633789,
      "learning_rate": 9.563687600450034e-06,
      "loss": 0.1686,
      "step": 2231
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.7207295894622803,
      "learning_rate": 9.563213025711122e-06,
      "loss": 0.294,
      "step": 2232
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.6645348072052,
      "learning_rate": 9.562738204804138e-06,
      "loss": 0.1674,
      "step": 2233
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.384199142456055,
      "learning_rate": 9.5622631377547e-06,
      "loss": 1.4209,
      "step": 2234
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.60464334487915,
      "learning_rate": 9.561787824588434e-06,
      "loss": 0.9288,
      "step": 2235
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.370473861694336,
      "learning_rate": 9.561312265330983e-06,
      "loss": 0.9851,
      "step": 2236
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.7214393615722656,
      "learning_rate": 9.560836460008e-06,
      "loss": 0.1533,
      "step": 2237
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.24022388458252,
      "learning_rate": 9.560360408645155e-06,
      "loss": 1.5747,
      "step": 2238
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.378293037414551,
      "learning_rate": 9.559884111268128e-06,
      "loss": 0.2744,
      "step": 2239
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.325429916381836,
      "learning_rate": 9.559407567902612e-06,
      "loss": 1.1719,
      "step": 2240
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.881653308868408,
      "learning_rate": 9.558930778574316e-06,
      "loss": 0.2286,
      "step": 2241
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.465372562408447,
      "learning_rate": 9.558453743308962e-06,
      "loss": 0.3009,
      "step": 2242
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.74232292175293,
      "learning_rate": 9.557976462132282e-06,
      "loss": 1.0787,
      "step": 2243
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8498787879943848,
      "learning_rate": 9.557498935070027e-06,
      "loss": 0.1751,
      "step": 2244
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.2045552730560303,
      "learning_rate": 9.557021162147952e-06,
      "loss": 0.1696,
      "step": 2245
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.144731044769287,
      "learning_rate": 9.556543143391836e-06,
      "loss": 0.1906,
      "step": 2246
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.605093955993652,
      "learning_rate": 9.556064878827467e-06,
      "loss": 0.9743,
      "step": 2247
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.3616361618042,
      "learning_rate": 9.555586368480642e-06,
      "loss": 1.3203,
      "step": 2248
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.4425625801086426,
      "learning_rate": 9.555107612377174e-06,
      "loss": 0.1132,
      "step": 2249
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8828818798065186,
      "learning_rate": 9.554628610542895e-06,
      "loss": 0.1496,
      "step": 2250
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.583095073699951,
      "learning_rate": 9.554149363003643e-06,
      "loss": 0.6848,
      "step": 2251
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.702300071716309,
      "learning_rate": 9.55366986978527e-06,
      "loss": 0.8869,
      "step": 2252
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.137383460998535,
      "learning_rate": 9.553190130913646e-06,
      "loss": 0.1338,
      "step": 2253
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.563185930252075,
      "learning_rate": 9.552710146414646e-06,
      "loss": 0.1905,
      "step": 2254
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.516636848449707,
      "learning_rate": 9.552229916314168e-06,
      "loss": 0.9384,
      "step": 2255
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.273566722869873,
      "learning_rate": 9.55174944063812e-06,
      "loss": 0.3113,
      "step": 2256
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.520325660705566,
      "learning_rate": 9.551268719412414e-06,
      "loss": 0.7828,
      "step": 2257
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.62884521484375,
      "learning_rate": 9.550787752662992e-06,
      "loss": 1.3442,
      "step": 2258
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.3708181381225586,
      "learning_rate": 9.550306540415793e-06,
      "loss": 0.1941,
      "step": 2259
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.820876598358154,
      "learning_rate": 9.549825082696782e-06,
      "loss": 0.9076,
      "step": 2260
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.967702627182007,
      "learning_rate": 9.549343379531931e-06,
      "loss": 0.1062,
      "step": 2261
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.6372272968292236,
      "learning_rate": 9.548861430947222e-06,
      "loss": 0.2326,
      "step": 2262
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.235971450805664,
      "learning_rate": 9.548379236968659e-06,
      "loss": 0.9652,
      "step": 2263
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.261711359024048,
      "learning_rate": 9.547896797622252e-06,
      "loss": 0.1976,
      "step": 2264
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.6642067432403564,
      "learning_rate": 9.547414112934028e-06,
      "loss": 0.1938,
      "step": 2265
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.172354698181152,
      "learning_rate": 9.546931182930024e-06,
      "loss": 1.2212,
      "step": 2266
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.9429073333740234,
      "learning_rate": 9.546448007636295e-06,
      "loss": 0.3301,
      "step": 2267
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.9454190731048584,
      "learning_rate": 9.545964587078906e-06,
      "loss": 0.1282,
      "step": 2268
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.3538899421691895,
      "learning_rate": 9.545480921283934e-06,
      "loss": 1.0572,
      "step": 2269
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.811741828918457,
      "learning_rate": 9.544997010277472e-06,
      "loss": 0.1104,
      "step": 2270
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.612871170043945,
      "learning_rate": 9.544512854085626e-06,
      "loss": 0.9328,
      "step": 2271
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.966136455535889,
      "learning_rate": 9.544028452734513e-06,
      "loss": 0.2385,
      "step": 2272
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5169966220855713,
      "learning_rate": 9.543543806250264e-06,
      "loss": 0.2614,
      "step": 2273
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.304843425750732,
      "learning_rate": 9.543058914659026e-06,
      "loss": 1.0658,
      "step": 2274
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.2578482627868652,
      "learning_rate": 9.542573777986955e-06,
      "loss": 0.1696,
      "step": 2275
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5814735889434814,
      "learning_rate": 9.542088396260224e-06,
      "loss": 0.1098,
      "step": 2276
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.313300371170044,
      "learning_rate": 9.541602769505017e-06,
      "loss": 0.1874,
      "step": 2277
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.159225463867188,
      "learning_rate": 9.541116897747532e-06,
      "loss": 1.0739,
      "step": 2278
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.705228328704834,
      "learning_rate": 9.540630781013978e-06,
      "loss": 0.1156,
      "step": 2279
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.3845648765563965,
      "learning_rate": 9.54014441933058e-06,
      "loss": 0.1282,
      "step": 2280
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.824277877807617,
      "learning_rate": 9.539657812723578e-06,
      "loss": 0.202,
      "step": 2281
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.337512969970703,
      "learning_rate": 9.539170961219219e-06,
      "loss": 1.7402,
      "step": 2282
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.765043020248413,
      "learning_rate": 9.538683864843769e-06,
      "loss": 0.1566,
      "step": 2283
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.000441551208496,
      "learning_rate": 9.538196523623505e-06,
      "loss": 0.1925,
      "step": 2284
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.2506823539733887,
      "learning_rate": 9.537708937584714e-06,
      "loss": 0.1691,
      "step": 2285
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.104345798492432,
      "learning_rate": 9.537221106753705e-06,
      "loss": 0.2458,
      "step": 2286
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.416677474975586,
      "learning_rate": 9.536733031156789e-06,
      "loss": 0.1385,
      "step": 2287
    },
    {
      "epoch": 0.16,
      "grad_norm": 42.89938735961914,
      "learning_rate": 9.5362447108203e-06,
      "loss": 0.9546,
      "step": 2288
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.0466461181640625,
      "learning_rate": 9.53575614577058e-06,
      "loss": 0.9606,
      "step": 2289
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.000297784805298,
      "learning_rate": 9.535267336033984e-06,
      "loss": 0.1143,
      "step": 2290
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.896048545837402,
      "learning_rate": 9.534778281636882e-06,
      "loss": 1.73,
      "step": 2291
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.383526563644409,
      "learning_rate": 9.534288982605658e-06,
      "loss": 0.1897,
      "step": 2292
    },
    {
      "epoch": 0.16,
      "grad_norm": 25.00267791748047,
      "learning_rate": 9.533799438966704e-06,
      "loss": 1.5991,
      "step": 2293
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1994333267211914,
      "learning_rate": 9.533309650746433e-06,
      "loss": 0.1767,
      "step": 2294
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.127007484436035,
      "learning_rate": 9.532819617971266e-06,
      "loss": 0.1625,
      "step": 2295
    },
    {
      "epoch": 0.16,
      "grad_norm": 22.28851890563965,
      "learning_rate": 9.532329340667638e-06,
      "loss": 2.208,
      "step": 2296
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8039262294769287,
      "learning_rate": 9.531838818861998e-06,
      "loss": 0.2154,
      "step": 2297
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.759381294250488,
      "learning_rate": 9.53134805258081e-06,
      "loss": 0.894,
      "step": 2298
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.730814933776855,
      "learning_rate": 9.530857041850542e-06,
      "loss": 1.7202,
      "step": 2299
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.309789657592773,
      "learning_rate": 9.530365786697691e-06,
      "loss": 0.9877,
      "step": 2300
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.739618301391602,
      "learning_rate": 9.52987428714875e-06,
      "loss": 1.2683,
      "step": 2301
    },
    {
      "epoch": 0.17,
      "grad_norm": 31.244834899902344,
      "learning_rate": 9.529382543230242e-06,
      "loss": 1.9088,
      "step": 2302
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.4124512672424316,
      "learning_rate": 9.528890554968688e-06,
      "loss": 0.1213,
      "step": 2303
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.932226181030273,
      "learning_rate": 9.528398322390631e-06,
      "loss": 1.1225,
      "step": 2304
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.8334779739379883,
      "learning_rate": 9.527905845522628e-06,
      "loss": 0.2032,
      "step": 2305
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.825733184814453,
      "learning_rate": 9.52741312439124e-06,
      "loss": 0.1553,
      "step": 2306
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.064456939697266,
      "learning_rate": 9.526920159023055e-06,
      "loss": 0.9413,
      "step": 2307
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.421639442443848,
      "learning_rate": 9.52642694944466e-06,
      "loss": 0.3031,
      "step": 2308
    },
    {
      "epoch": 0.17,
      "grad_norm": 19.60984230041504,
      "learning_rate": 9.525933495682666e-06,
      "loss": 0.8884,
      "step": 2309
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.00214672088623,
      "learning_rate": 9.52543979776369e-06,
      "loss": 0.8099,
      "step": 2310
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0586111545562744,
      "learning_rate": 9.524945855714368e-06,
      "loss": 0.1582,
      "step": 2311
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.613866806030273,
      "learning_rate": 9.524451669561345e-06,
      "loss": 0.9002,
      "step": 2312
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.261247634887695,
      "learning_rate": 9.523957239331278e-06,
      "loss": 0.621,
      "step": 2313
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.862764835357666,
      "learning_rate": 9.523462565050845e-06,
      "loss": 0.2248,
      "step": 2314
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3473429679870605,
      "learning_rate": 9.522967646746728e-06,
      "loss": 0.2631,
      "step": 2315
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.701450347900391,
      "learning_rate": 9.522472484445626e-06,
      "loss": 1.0847,
      "step": 2316
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.629209280014038,
      "learning_rate": 9.521977078174253e-06,
      "loss": 0.0885,
      "step": 2317
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.940750122070312,
      "learning_rate": 9.521481427959332e-06,
      "loss": 1.3017,
      "step": 2318
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8117120265960693,
      "learning_rate": 9.520985533827605e-06,
      "loss": 0.155,
      "step": 2319
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.097954750061035,
      "learning_rate": 9.52048939580582e-06,
      "loss": 0.8056,
      "step": 2320
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.9723641872406006,
      "learning_rate": 9.519993013920742e-06,
      "loss": 0.1744,
      "step": 2321
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.32719612121582,
      "learning_rate": 9.51949638819915e-06,
      "loss": 1.6749,
      "step": 2322
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.591658592224121,
      "learning_rate": 9.518999518667837e-06,
      "loss": 0.9478,
      "step": 2323
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.428008079528809,
      "learning_rate": 9.518502405353605e-06,
      "loss": 0.9077,
      "step": 2324
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.785569190979004,
      "learning_rate": 9.51800504828327e-06,
      "loss": 1.0609,
      "step": 2325
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3911209106445312,
      "learning_rate": 9.517507447483665e-06,
      "loss": 0.1422,
      "step": 2326
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.957242012023926,
      "learning_rate": 9.517009602981632e-06,
      "loss": 1.62,
      "step": 2327
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.086160659790039,
      "learning_rate": 9.516511514804031e-06,
      "loss": 0.6363,
      "step": 2328
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.382735729217529,
      "learning_rate": 9.516013182977728e-06,
      "loss": 1.0995,
      "step": 2329
    },
    {
      "epoch": 0.17,
      "grad_norm": 22.444339752197266,
      "learning_rate": 9.51551460752961e-06,
      "loss": 1.1293,
      "step": 2330
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.727903842926025,
      "learning_rate": 9.515015788486568e-06,
      "loss": 0.8543,
      "step": 2331
    },
    {
      "epoch": 0.17,
      "grad_norm": 32.74965286254883,
      "learning_rate": 9.514516725875517e-06,
      "loss": 1.29,
      "step": 2332
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.688870668411255,
      "learning_rate": 9.514017419723376e-06,
      "loss": 0.2977,
      "step": 2333
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.146736145019531,
      "learning_rate": 9.513517870057081e-06,
      "loss": 1.7036,
      "step": 2334
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.0977349281311035,
      "learning_rate": 9.513018076903581e-06,
      "loss": 0.2009,
      "step": 2335
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.2657411098480225,
      "learning_rate": 9.51251804028984e-06,
      "loss": 0.1998,
      "step": 2336
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.607784271240234,
      "learning_rate": 9.51201776024283e-06,
      "loss": 1.388,
      "step": 2337
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0090463161468506,
      "learning_rate": 9.511517236789543e-06,
      "loss": 0.1813,
      "step": 2338
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.109184265136719,
      "learning_rate": 9.511016469956977e-06,
      "loss": 0.5241,
      "step": 2339
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.487557411193848,
      "learning_rate": 9.510515459772148e-06,
      "loss": 1.0679,
      "step": 2340
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3519160747528076,
      "learning_rate": 9.510014206262083e-06,
      "loss": 0.2296,
      "step": 2341
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.196262836456299,
      "learning_rate": 9.509512709453821e-06,
      "loss": 0.94,
      "step": 2342
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.485539436340332,
      "learning_rate": 9.509010969374421e-06,
      "loss": 1.1806,
      "step": 2343
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.532151937484741,
      "learning_rate": 9.508508986050945e-06,
      "loss": 0.2054,
      "step": 2344
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.036693811416626,
      "learning_rate": 9.508006759510474e-06,
      "loss": 0.2496,
      "step": 2345
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3933539390563965,
      "learning_rate": 9.507504289780103e-06,
      "loss": 0.3016,
      "step": 2346
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.75271224975586,
      "learning_rate": 9.507001576886939e-06,
      "loss": 0.8016,
      "step": 2347
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.8102736473083496,
      "learning_rate": 9.506498620858097e-06,
      "loss": 0.1931,
      "step": 2348
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.037345886230469,
      "learning_rate": 9.505995421720713e-06,
      "loss": 1.447,
      "step": 2349
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.661952495574951,
      "learning_rate": 9.505491979501934e-06,
      "loss": 0.917,
      "step": 2350
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3825323581695557,
      "learning_rate": 9.504988294228914e-06,
      "loss": 0.1948,
      "step": 2351
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.3037691116333,
      "learning_rate": 9.50448436592883e-06,
      "loss": 0.9657,
      "step": 2352
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.021044969558716,
      "learning_rate": 9.503980194628865e-06,
      "loss": 0.1426,
      "step": 2353
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6150732040405273,
      "learning_rate": 9.503475780356216e-06,
      "loss": 0.131,
      "step": 2354
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.701838970184326,
      "learning_rate": 9.502971123138097e-06,
      "loss": 1.0912,
      "step": 2355
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.924607276916504,
      "learning_rate": 9.502466223001728e-06,
      "loss": 1.066,
      "step": 2356
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.711330413818359,
      "learning_rate": 9.50196107997435e-06,
      "loss": 0.7627,
      "step": 2357
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0790436267852783,
      "learning_rate": 9.501455694083213e-06,
      "loss": 0.1015,
      "step": 2358
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.085824489593506,
      "learning_rate": 9.500950065355579e-06,
      "loss": 0.123,
      "step": 2359
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.2576959133148193,
      "learning_rate": 9.500444193818728e-06,
      "loss": 0.2446,
      "step": 2360
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.463944435119629,
      "learning_rate": 9.499938079499947e-06,
      "loss": 0.8286,
      "step": 2361
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.295198678970337,
      "learning_rate": 9.49943172242654e-06,
      "loss": 0.1853,
      "step": 2362
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.578311443328857,
      "learning_rate": 9.498925122625822e-06,
      "loss": 0.8577,
      "step": 2363
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.729142427444458,
      "learning_rate": 9.498418280125124e-06,
      "loss": 0.1371,
      "step": 2364
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.663201332092285,
      "learning_rate": 9.497911194951786e-06,
      "loss": 0.1251,
      "step": 2365
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.2838053703308105,
      "learning_rate": 9.497403867133166e-06,
      "loss": 0.7659,
      "step": 2366
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.07968521118164,
      "learning_rate": 9.49689629669663e-06,
      "loss": 1.6697,
      "step": 2367
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.375901937484741,
      "learning_rate": 9.49638848366956e-06,
      "loss": 0.1502,
      "step": 2368
    },
    {
      "epoch": 0.17,
      "grad_norm": 20.940073013305664,
      "learning_rate": 9.495880428079352e-06,
      "loss": 2.0574,
      "step": 2369
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.660184860229492,
      "learning_rate": 9.495372129953412e-06,
      "loss": 1.2051,
      "step": 2370
    },
    {
      "epoch": 0.17,
      "grad_norm": 22.018308639526367,
      "learning_rate": 9.494863589319161e-06,
      "loss": 1.7651,
      "step": 2371
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.174297332763672,
      "learning_rate": 9.494354806204036e-06,
      "loss": 1.5241,
      "step": 2372
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.9292213916778564,
      "learning_rate": 9.493845780635478e-06,
      "loss": 0.1684,
      "step": 2373
    },
    {
      "epoch": 0.17,
      "grad_norm": 19.67542266845703,
      "learning_rate": 9.493336512640955e-06,
      "loss": 1.4878,
      "step": 2374
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.936762571334839,
      "learning_rate": 9.492827002247931e-06,
      "loss": 0.157,
      "step": 2375
    },
    {
      "epoch": 0.17,
      "grad_norm": 25.805665969848633,
      "learning_rate": 9.492317249483897e-06,
      "loss": 1.8528,
      "step": 2376
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.078366756439209,
      "learning_rate": 9.491807254376353e-06,
      "loss": 0.1793,
      "step": 2377
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.113070487976074,
      "learning_rate": 9.491297016952809e-06,
      "loss": 1.171,
      "step": 2378
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.2245330810546875,
      "learning_rate": 9.490786537240792e-06,
      "loss": 0.1472,
      "step": 2379
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.256169319152832,
      "learning_rate": 9.490275815267839e-06,
      "loss": 0.142,
      "step": 2380
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.679542541503906,
      "learning_rate": 9.489764851061504e-06,
      "loss": 0.8622,
      "step": 2381
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.472679138183594,
      "learning_rate": 9.489253644649348e-06,
      "loss": 1.0199,
      "step": 2382
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.033998489379883,
      "learning_rate": 9.488742196058953e-06,
      "loss": 0.2013,
      "step": 2383
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.886030673980713,
      "learning_rate": 9.488230505317905e-06,
      "loss": 0.6595,
      "step": 2384
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.299691200256348,
      "learning_rate": 9.487718572453812e-06,
      "loss": 0.7073,
      "step": 2385
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.752386569976807,
      "learning_rate": 9.48720639749429e-06,
      "loss": 0.3234,
      "step": 2386
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.1212892532348633,
      "learning_rate": 9.486693980466963e-06,
      "loss": 0.2301,
      "step": 2387
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.886155605316162,
      "learning_rate": 9.486181321399483e-06,
      "loss": 1.1896,
      "step": 2388
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6966023445129395,
      "learning_rate": 9.4856684203195e-06,
      "loss": 0.0798,
      "step": 2389
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.8811147212982178,
      "learning_rate": 9.485155277254686e-06,
      "loss": 0.2087,
      "step": 2390
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.851930856704712,
      "learning_rate": 9.48464189223272e-06,
      "loss": 0.131,
      "step": 2391
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.664958000183105,
      "learning_rate": 9.484128265281303e-06,
      "loss": 1.0249,
      "step": 2392
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.416055202484131,
      "learning_rate": 9.483614396428136e-06,
      "loss": 0.1824,
      "step": 2393
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.361411094665527,
      "learning_rate": 9.483100285700946e-06,
      "loss": 0.7713,
      "step": 2394
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.953547954559326,
      "learning_rate": 9.482585933127463e-06,
      "loss": 0.3049,
      "step": 2395
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.157768249511719,
      "learning_rate": 9.482071338735437e-06,
      "loss": 1.1728,
      "step": 2396
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0437023639678955,
      "learning_rate": 9.481556502552629e-06,
      "loss": 0.1446,
      "step": 2397
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8444700241088867,
      "learning_rate": 9.48104142460681e-06,
      "loss": 0.2047,
      "step": 2398
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.824846267700195,
      "learning_rate": 9.480526104925769e-06,
      "loss": 1.5959,
      "step": 2399
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.08230209350586,
      "learning_rate": 9.480010543537303e-06,
      "loss": 1.1816,
      "step": 2400
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.2609024047851562,
      "learning_rate": 9.479494740469226e-06,
      "loss": 0.1565,
      "step": 2401
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.494643211364746,
      "learning_rate": 9.478978695749366e-06,
      "loss": 0.683,
      "step": 2402
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.910301685333252,
      "learning_rate": 9.478462409405555e-06,
      "loss": 1.1819,
      "step": 2403
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.378589391708374,
      "learning_rate": 9.477945881465654e-06,
      "loss": 0.215,
      "step": 2404
    },
    {
      "epoch": 0.17,
      "grad_norm": 22.900379180908203,
      "learning_rate": 9.47742911195752e-06,
      "loss": 1.7918,
      "step": 2405
    },
    {
      "epoch": 0.17,
      "grad_norm": 19.056833267211914,
      "learning_rate": 9.476912100909032e-06,
      "loss": 1.6873,
      "step": 2406
    },
    {
      "epoch": 0.17,
      "grad_norm": 50.83261489868164,
      "learning_rate": 9.476394848348083e-06,
      "loss": 0.8428,
      "step": 2407
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.707582473754883,
      "learning_rate": 9.475877354302578e-06,
      "loss": 0.1865,
      "step": 2408
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.212255477905273,
      "learning_rate": 9.47535961880043e-06,
      "loss": 1.348,
      "step": 2409
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.632453918457031,
      "learning_rate": 9.47484164186957e-06,
      "loss": 1.0712,
      "step": 2410
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.069390296936035,
      "learning_rate": 9.474323423537944e-06,
      "loss": 1.2677,
      "step": 2411
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.522040367126465,
      "learning_rate": 9.473804963833503e-06,
      "loss": 0.6149,
      "step": 2412
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.6014111042022705,
      "learning_rate": 9.473286262784221e-06,
      "loss": 0.2543,
      "step": 2413
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.994207859039307,
      "learning_rate": 9.472767320418074e-06,
      "loss": 0.717,
      "step": 2414
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.17799949645996,
      "learning_rate": 9.472248136763063e-06,
      "loss": 1.3291,
      "step": 2415
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.415285110473633,
      "learning_rate": 9.471728711847195e-06,
      "loss": 1.0089,
      "step": 2416
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.949458599090576,
      "learning_rate": 9.471209045698486e-06,
      "loss": 0.9939,
      "step": 2417
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.6388559341430664,
      "learning_rate": 9.470689138344975e-06,
      "loss": 0.1813,
      "step": 2418
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.555744171142578,
      "learning_rate": 9.470168989814708e-06,
      "loss": 0.1813,
      "step": 2419
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.297693252563477,
      "learning_rate": 9.469648600135743e-06,
      "loss": 1.6509,
      "step": 2420
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.980031967163086,
      "learning_rate": 9.469127969336156e-06,
      "loss": 0.1894,
      "step": 2421
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.465456962585449,
      "learning_rate": 9.468607097444031e-06,
      "loss": 0.1887,
      "step": 2422
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0827152729034424,
      "learning_rate": 9.468085984487467e-06,
      "loss": 0.2452,
      "step": 2423
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.414389610290527,
      "learning_rate": 9.467564630494577e-06,
      "loss": 0.9782,
      "step": 2424
    },
    {
      "epoch": 0.17,
      "grad_norm": 28.150972366333008,
      "learning_rate": 9.467043035493486e-06,
      "loss": 0.6939,
      "step": 2425
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.195561170578003,
      "learning_rate": 9.466521199512333e-06,
      "loss": 0.224,
      "step": 2426
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.5943827629089355,
      "learning_rate": 9.465999122579267e-06,
      "loss": 0.9964,
      "step": 2427
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.124607563018799,
      "learning_rate": 9.465476804722453e-06,
      "loss": 0.0738,
      "step": 2428
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.600163459777832,
      "learning_rate": 9.464954245970068e-06,
      "loss": 1.2726,
      "step": 2429
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.156117916107178,
      "learning_rate": 9.464431446350304e-06,
      "loss": 0.7891,
      "step": 2430
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3091790676116943,
      "learning_rate": 9.46390840589136e-06,
      "loss": 0.1362,
      "step": 2431
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.731165885925293,
      "learning_rate": 9.463385124621454e-06,
      "loss": 0.2248,
      "step": 2432
    },
    {
      "epoch": 0.17,
      "grad_norm": 20.024805068969727,
      "learning_rate": 9.462861602568817e-06,
      "loss": 1.6253,
      "step": 2433
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.5269603729248047,
      "learning_rate": 9.46233783976169e-06,
      "loss": 0.1101,
      "step": 2434
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.704139709472656,
      "learning_rate": 9.461813836228326e-06,
      "loss": 0.6632,
      "step": 2435
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.784064292907715,
      "learning_rate": 9.461289591996994e-06,
      "loss": 0.1248,
      "step": 2436
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.27882194519043,
      "learning_rate": 9.460765107095976e-06,
      "loss": 0.9023,
      "step": 2437
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.407623291015625,
      "learning_rate": 9.460240381553564e-06,
      "loss": 1.5855,
      "step": 2438
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.779458522796631,
      "learning_rate": 9.459715415398069e-06,
      "loss": 0.3075,
      "step": 2439
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.971574783325195,
      "learning_rate": 9.459190208657807e-06,
      "loss": 1.3687,
      "step": 2440
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.259450912475586,
      "learning_rate": 9.458664761361112e-06,
      "loss": 1.4255,
      "step": 2441
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.9060134887695312,
      "learning_rate": 9.458139073536331e-06,
      "loss": 0.1672,
      "step": 2442
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.646353244781494,
      "learning_rate": 9.457613145211821e-06,
      "loss": 0.284,
      "step": 2443
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9340660572052,
      "learning_rate": 9.457086976415956e-06,
      "loss": 0.2226,
      "step": 2444
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.540607929229736,
      "learning_rate": 9.456560567177118e-06,
      "loss": 0.8277,
      "step": 2445
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.562009811401367,
      "learning_rate": 9.456033917523708e-06,
      "loss": 0.9946,
      "step": 2446
    },
    {
      "epoch": 0.18,
      "grad_norm": 16.817909240722656,
      "learning_rate": 9.455507027484134e-06,
      "loss": 1.498,
      "step": 2447
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.249384880065918,
      "learning_rate": 9.45497989708682e-06,
      "loss": 1.0031,
      "step": 2448
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.606060028076172,
      "learning_rate": 9.454452526360205e-06,
      "loss": 1.3982,
      "step": 2449
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6922998428344727,
      "learning_rate": 9.453924915332737e-06,
      "loss": 0.1665,
      "step": 2450
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.517987251281738,
      "learning_rate": 9.453397064032879e-06,
      "loss": 1.1654,
      "step": 2451
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.649144172668457,
      "learning_rate": 9.452868972489106e-06,
      "loss": 0.1126,
      "step": 2452
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.434260368347168,
      "learning_rate": 9.452340640729908e-06,
      "loss": 0.2543,
      "step": 2453
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.58724594116211,
      "learning_rate": 9.451812068783784e-06,
      "loss": 0.9046,
      "step": 2454
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.128225326538086,
      "learning_rate": 9.45128325667925e-06,
      "loss": 1.054,
      "step": 2455
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.49062728881836,
      "learning_rate": 9.450754204444834e-06,
      "loss": 0.5953,
      "step": 2456
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4289395809173584,
      "learning_rate": 9.450224912109076e-06,
      "loss": 0.1496,
      "step": 2457
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.195755481719971,
      "learning_rate": 9.44969537970053e-06,
      "loss": 0.944,
      "step": 2458
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.566673278808594,
      "learning_rate": 9.44916560724776e-06,
      "loss": 1.0796,
      "step": 2459
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.371973991394043,
      "learning_rate": 9.448635594779345e-06,
      "loss": 0.1156,
      "step": 2460
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.932480812072754,
      "learning_rate": 9.44810534232388e-06,
      "loss": 1.1235,
      "step": 2461
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.9431190490722656,
      "learning_rate": 9.447574849909969e-06,
      "loss": 0.1907,
      "step": 2462
    },
    {
      "epoch": 0.18,
      "grad_norm": 16.149799346923828,
      "learning_rate": 9.447044117566228e-06,
      "loss": 1.6646,
      "step": 2463
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.56183624267578,
      "learning_rate": 9.446513145321293e-06,
      "loss": 1.1455,
      "step": 2464
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.048731327056885,
      "learning_rate": 9.445981933203803e-06,
      "loss": 1.0235,
      "step": 2465
    },
    {
      "epoch": 0.18,
      "grad_norm": 14.513916969299316,
      "learning_rate": 9.445450481242417e-06,
      "loss": 0.8402,
      "step": 2466
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9513602256774902,
      "learning_rate": 9.444918789465804e-06,
      "loss": 0.1865,
      "step": 2467
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.841301918029785,
      "learning_rate": 9.444386857902647e-06,
      "loss": 1.0507,
      "step": 2468
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.732715606689453,
      "learning_rate": 9.443854686581644e-06,
      "loss": 0.9931,
      "step": 2469
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.8022050857543945,
      "learning_rate": 9.443322275531498e-06,
      "loss": 0.2517,
      "step": 2470
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.15549373626709,
      "learning_rate": 9.442789624780937e-06,
      "loss": 1.2507,
      "step": 2471
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.608857154846191,
      "learning_rate": 9.442256734358692e-06,
      "loss": 0.9753,
      "step": 2472
    },
    {
      "epoch": 0.18,
      "grad_norm": 14.411566734313965,
      "learning_rate": 9.44172360429351e-06,
      "loss": 1.146,
      "step": 2473
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.13414478302002,
      "learning_rate": 9.441190234614155e-06,
      "loss": 1.1728,
      "step": 2474
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.4349093437194824,
      "learning_rate": 9.440656625349396e-06,
      "loss": 0.1794,
      "step": 2475
    },
    {
      "epoch": 0.18,
      "grad_norm": 28.91055679321289,
      "learning_rate": 9.44012277652802e-06,
      "loss": 1.3973,
      "step": 2476
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6135714054107666,
      "learning_rate": 9.439588688178829e-06,
      "loss": 0.1555,
      "step": 2477
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.091653347015381,
      "learning_rate": 9.439054360330631e-06,
      "loss": 0.2333,
      "step": 2478
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1302130222320557,
      "learning_rate": 9.438519793012254e-06,
      "loss": 0.0753,
      "step": 2479
    },
    {
      "epoch": 0.18,
      "grad_norm": 28.899330139160156,
      "learning_rate": 9.437984986252536e-06,
      "loss": 1.806,
      "step": 2480
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.3078205585479736,
      "learning_rate": 9.437449940080322e-06,
      "loss": 0.2635,
      "step": 2481
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.344773292541504,
      "learning_rate": 9.436914654524486e-06,
      "loss": 0.2498,
      "step": 2482
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.9124691486358643,
      "learning_rate": 9.436379129613896e-06,
      "loss": 0.2353,
      "step": 2483
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.576817512512207,
      "learning_rate": 9.435843365377446e-06,
      "loss": 1.3044,
      "step": 2484
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1966466903686523,
      "learning_rate": 9.435307361844035e-06,
      "loss": 0.1921,
      "step": 2485
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.024789333343506,
      "learning_rate": 9.434771119042584e-06,
      "loss": 0.9585,
      "step": 2486
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.055250883102417,
      "learning_rate": 9.434234637002014e-06,
      "loss": 0.1011,
      "step": 2487
    },
    {
      "epoch": 0.18,
      "grad_norm": 14.49535083770752,
      "learning_rate": 9.433697915751271e-06,
      "loss": 1.4444,
      "step": 2488
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.775069236755371,
      "learning_rate": 9.433160955319308e-06,
      "loss": 0.1658,
      "step": 2489
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.539717674255371,
      "learning_rate": 9.432623755735092e-06,
      "loss": 0.6185,
      "step": 2490
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.698471069335938,
      "learning_rate": 9.432086317027602e-06,
      "loss": 1.716,
      "step": 2491
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.320414543151855,
      "learning_rate": 9.431548639225834e-06,
      "loss": 1.0369,
      "step": 2492
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.91954231262207,
      "learning_rate": 9.431010722358788e-06,
      "loss": 0.8952,
      "step": 2493
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3623838424682617,
      "learning_rate": 9.430472566455488e-06,
      "loss": 0.1728,
      "step": 2494
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.4212939739227295,
      "learning_rate": 9.429934171544964e-06,
      "loss": 0.223,
      "step": 2495
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.472788095474243,
      "learning_rate": 9.429395537656257e-06,
      "loss": 0.1755,
      "step": 2496
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.37802791595459,
      "learning_rate": 9.428856664818427e-06,
      "loss": 1.0893,
      "step": 2497
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.1412272453308105,
      "learning_rate": 9.428317553060546e-06,
      "loss": 1.084,
      "step": 2498
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.546282768249512,
      "learning_rate": 9.427778202411695e-06,
      "loss": 1.2112,
      "step": 2499
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1252803802490234,
      "learning_rate": 9.427238612900968e-06,
      "loss": 0.1575,
      "step": 2500
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1635777950286865,
      "learning_rate": 9.426698784557478e-06,
      "loss": 0.1391,
      "step": 2501
    },
    {
      "epoch": 0.18,
      "grad_norm": 15.827073097229004,
      "learning_rate": 9.426158717410345e-06,
      "loss": 1.168,
      "step": 2502
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.407527923583984,
      "learning_rate": 9.425618411488702e-06,
      "loss": 1.1614,
      "step": 2503
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3669493198394775,
      "learning_rate": 9.425077866821696e-06,
      "loss": 0.091,
      "step": 2504
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.054330825805664,
      "learning_rate": 9.42453708343849e-06,
      "loss": 0.1927,
      "step": 2505
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.92004108428955,
      "learning_rate": 9.423996061368258e-06,
      "loss": 1.1156,
      "step": 2506
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.6224629878997803,
      "learning_rate": 9.423454800640182e-06,
      "loss": 0.2159,
      "step": 2507
    },
    {
      "epoch": 0.18,
      "grad_norm": 14.34748363494873,
      "learning_rate": 9.422913301283465e-06,
      "loss": 1.2316,
      "step": 2508
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9808168411254883,
      "learning_rate": 9.422371563327316e-06,
      "loss": 0.214,
      "step": 2509
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.726420402526855,
      "learning_rate": 9.421829586800959e-06,
      "loss": 0.7308,
      "step": 2510
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.778477668762207,
      "learning_rate": 9.421287371733636e-06,
      "loss": 1.2929,
      "step": 2511
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.576149940490723,
      "learning_rate": 9.420744918154594e-06,
      "loss": 1.2058,
      "step": 2512
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.2913689613342285,
      "learning_rate": 9.420202226093096e-06,
      "loss": 0.9416,
      "step": 2513
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.133821725845337,
      "learning_rate": 9.41965929557842e-06,
      "loss": 0.2217,
      "step": 2514
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.704504013061523,
      "learning_rate": 9.419116126639855e-06,
      "loss": 1.0656,
      "step": 2515
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.357940196990967,
      "learning_rate": 9.418572719306703e-06,
      "loss": 0.4436,
      "step": 2516
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.7181806564331055,
      "learning_rate": 9.418029073608274e-06,
      "loss": 1.0169,
      "step": 2517
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.699920892715454,
      "learning_rate": 9.417485189573903e-06,
      "loss": 0.227,
      "step": 2518
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.405648708343506,
      "learning_rate": 9.416941067232927e-06,
      "loss": 0.1343,
      "step": 2519
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.481426239013672,
      "learning_rate": 9.4163967066147e-06,
      "loss": 1.656,
      "step": 2520
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.729726791381836,
      "learning_rate": 9.415852107748587e-06,
      "loss": 0.901,
      "step": 2521
    },
    {
      "epoch": 0.18,
      "grad_norm": 14.219523429870605,
      "learning_rate": 9.415307270663966e-06,
      "loss": 1.4038,
      "step": 2522
    },
    {
      "epoch": 0.18,
      "grad_norm": 15.638139724731445,
      "learning_rate": 9.414762195390233e-06,
      "loss": 1.6652,
      "step": 2523
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.8778793811798096,
      "learning_rate": 9.414216881956787e-06,
      "loss": 0.1657,
      "step": 2524
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.4260382652282715,
      "learning_rate": 9.413671330393053e-06,
      "loss": 1.0065,
      "step": 2525
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.65059232711792,
      "learning_rate": 9.413125540728456e-06,
      "loss": 0.2446,
      "step": 2526
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.841453552246094,
      "learning_rate": 9.41257951299244e-06,
      "loss": 1.1416,
      "step": 2527
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.7674543857574463,
      "learning_rate": 9.412033247214461e-06,
      "loss": 0.2894,
      "step": 2528
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.7418627738952637,
      "learning_rate": 9.41148674342399e-06,
      "loss": 0.1962,
      "step": 2529
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.693967819213867,
      "learning_rate": 9.410940001650507e-06,
      "loss": 0.1835,
      "step": 2530
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.261454582214355,
      "learning_rate": 9.410393021923508e-06,
      "loss": 0.976,
      "step": 2531
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.218835830688477,
      "learning_rate": 9.409845804272499e-06,
      "loss": 1.3702,
      "step": 2532
    },
    {
      "epoch": 0.18,
      "grad_norm": 25.827617645263672,
      "learning_rate": 9.409298348727001e-06,
      "loss": 1.3597,
      "step": 2533
    },
    {
      "epoch": 0.18,
      "grad_norm": 20.108476638793945,
      "learning_rate": 9.408750655316547e-06,
      "loss": 1.5996,
      "step": 2534
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.711272239685059,
      "learning_rate": 9.408202724070683e-06,
      "loss": 0.8807,
      "step": 2535
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9733641147613525,
      "learning_rate": 9.40765455501897e-06,
      "loss": 0.1854,
      "step": 2536
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3864223957061768,
      "learning_rate": 9.407106148190977e-06,
      "loss": 0.0843,
      "step": 2537
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.67551851272583,
      "learning_rate": 9.406557503616286e-06,
      "loss": 1.0005,
      "step": 2538
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.6221351623535156,
      "learning_rate": 9.406008621324501e-06,
      "loss": 0.1155,
      "step": 2539
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1487112045288086,
      "learning_rate": 9.405459501345228e-06,
      "loss": 0.1396,
      "step": 2540
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.745940685272217,
      "learning_rate": 9.404910143708089e-06,
      "loss": 0.9487,
      "step": 2541
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1279594898223877,
      "learning_rate": 9.404360548442722e-06,
      "loss": 0.1698,
      "step": 2542
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.538606882095337,
      "learning_rate": 9.403810715578776e-06,
      "loss": 0.2759,
      "step": 2543
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.246020317077637,
      "learning_rate": 9.403260645145908e-06,
      "loss": 0.2045,
      "step": 2544
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.6955039501190186,
      "learning_rate": 9.402710337173799e-06,
      "loss": 0.2055,
      "step": 2545
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1295697689056396,
      "learning_rate": 9.402159791692132e-06,
      "loss": 0.2037,
      "step": 2546
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.550723075866699,
      "learning_rate": 9.401609008730605e-06,
      "loss": 0.7108,
      "step": 2547
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.774153709411621,
      "learning_rate": 9.401057988318933e-06,
      "loss": 0.2139,
      "step": 2548
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.936607360839844,
      "learning_rate": 9.400506730486845e-06,
      "loss": 0.7519,
      "step": 2549
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.0778632164001465,
      "learning_rate": 9.399955235264072e-06,
      "loss": 0.756,
      "step": 2550
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.034532070159912,
      "learning_rate": 9.399403502680371e-06,
      "loss": 0.5089,
      "step": 2551
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.2795307636260986,
      "learning_rate": 9.398851532765502e-06,
      "loss": 0.1519,
      "step": 2552
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.2762291431427,
      "learning_rate": 9.398299325549244e-06,
      "loss": 0.1672,
      "step": 2553
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.286746978759766,
      "learning_rate": 9.397746881061387e-06,
      "loss": 0.9782,
      "step": 2554
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.539025783538818,
      "learning_rate": 9.397194199331732e-06,
      "loss": 0.2108,
      "step": 2555
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.467007637023926,
      "learning_rate": 9.396641280390092e-06,
      "loss": 1.1137,
      "step": 2556
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.281261444091797,
      "learning_rate": 9.396088124266298e-06,
      "loss": 0.9287,
      "step": 2557
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.2822887897491455,
      "learning_rate": 9.395534730990192e-06,
      "loss": 0.1772,
      "step": 2558
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.868099212646484,
      "learning_rate": 9.394981100591622e-06,
      "loss": 1.2602,
      "step": 2559
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.12526512145996,
      "learning_rate": 9.39442723310046e-06,
      "loss": 1.2483,
      "step": 2560
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.358901977539062,
      "learning_rate": 9.393873128546582e-06,
      "loss": 1.3701,
      "step": 2561
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.426589965820312,
      "learning_rate": 9.39331878695988e-06,
      "loss": 1.1248,
      "step": 2562
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.548637628555298,
      "learning_rate": 9.39276420837026e-06,
      "loss": 0.1845,
      "step": 2563
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.7503201961517334,
      "learning_rate": 9.392209392807637e-06,
      "loss": 0.2325,
      "step": 2564
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1260337829589844,
      "learning_rate": 9.391654340301944e-06,
      "loss": 0.1842,
      "step": 2565
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.483011245727539,
      "learning_rate": 9.391099050883122e-06,
      "loss": 0.3447,
      "step": 2566
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.631744861602783,
      "learning_rate": 9.390543524581126e-06,
      "loss": 0.2138,
      "step": 2567
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.131403923034668,
      "learning_rate": 9.389987761425929e-06,
      "loss": 0.8836,
      "step": 2568
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.2916364669799805,
      "learning_rate": 9.389431761447507e-06,
      "loss": 0.1774,
      "step": 2569
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.97688102722168,
      "learning_rate": 9.388875524675857e-06,
      "loss": 1.1369,
      "step": 2570
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.895349502563477,
      "learning_rate": 9.388319051140986e-06,
      "loss": 0.8735,
      "step": 2571
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.939342975616455,
      "learning_rate": 9.387762340872912e-06,
      "loss": 0.1052,
      "step": 2572
    },
    {
      "epoch": 0.18,
      "grad_norm": 20.666702270507812,
      "learning_rate": 9.387205393901668e-06,
      "loss": 1.4065,
      "step": 2573
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.284713745117188,
      "learning_rate": 9.3866482102573e-06,
      "loss": 0.8437,
      "step": 2574
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.5373241901397705,
      "learning_rate": 9.386090789969866e-06,
      "loss": 0.088,
      "step": 2575
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.911600112915039,
      "learning_rate": 9.385533133069436e-06,
      "loss": 0.7603,
      "step": 2576
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.843514919281006,
      "learning_rate": 9.384975239586092e-06,
      "loss": 0.8956,
      "step": 2577
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.4460418224334717,
      "learning_rate": 9.384417109549933e-06,
      "loss": 0.2333,
      "step": 2578
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.696913719177246,
      "learning_rate": 9.383858742991067e-06,
      "loss": 0.1576,
      "step": 2579
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.2921111583709717,
      "learning_rate": 9.383300139939615e-06,
      "loss": 0.2778,
      "step": 2580
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.9886529445648193,
      "learning_rate": 9.382741300425713e-06,
      "loss": 0.1895,
      "step": 2581
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.447125196456909,
      "learning_rate": 9.382182224479508e-06,
      "loss": 0.1405,
      "step": 2582
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.292627334594727,
      "learning_rate": 9.381622912131158e-06,
      "loss": 0.9834,
      "step": 2583
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.556983709335327,
      "learning_rate": 9.381063363410837e-06,
      "loss": 0.2003,
      "step": 2584
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.527267456054688,
      "learning_rate": 9.380503578348732e-06,
      "loss": 1.2793,
      "step": 2585
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.827157497406006,
      "learning_rate": 9.37994355697504e-06,
      "loss": 0.6953,
      "step": 2586
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.838904857635498,
      "learning_rate": 9.37938329931997e-06,
      "loss": 0.9303,
      "step": 2587
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.299413681030273,
      "learning_rate": 9.37882280541375e-06,
      "loss": 0.8455,
      "step": 2588
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5242457389831543,
      "learning_rate": 9.378262075286615e-06,
      "loss": 0.0907,
      "step": 2589
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.753171920776367,
      "learning_rate": 9.377701108968811e-06,
      "loss": 0.5954,
      "step": 2590
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.793468475341797,
      "learning_rate": 9.377139906490605e-06,
      "loss": 1.1362,
      "step": 2591
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.036170721054077,
      "learning_rate": 9.37657846788227e-06,
      "loss": 0.1134,
      "step": 2592
    },
    {
      "epoch": 0.19,
      "grad_norm": 23.531280517578125,
      "learning_rate": 9.37601679317409e-06,
      "loss": 0.9193,
      "step": 2593
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.650600910186768,
      "learning_rate": 9.37545488239637e-06,
      "loss": 1.0689,
      "step": 2594
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.416563034057617,
      "learning_rate": 9.374892735579422e-06,
      "loss": 0.9099,
      "step": 2595
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.189233779907227,
      "learning_rate": 9.374330352753569e-06,
      "loss": 0.9892,
      "step": 2596
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.724115371704102,
      "learning_rate": 9.373767733949151e-06,
      "loss": 0.7973,
      "step": 2597
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.6643121242523193,
      "learning_rate": 9.373204879196521e-06,
      "loss": 0.1524,
      "step": 2598
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.784503221511841,
      "learning_rate": 9.37264178852604e-06,
      "loss": 0.3431,
      "step": 2599
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.2618842124938965,
      "learning_rate": 9.372078461968086e-06,
      "loss": 0.4017,
      "step": 2600
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3214995861053467,
      "learning_rate": 9.371514899553047e-06,
      "loss": 0.1357,
      "step": 2601
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.523812294006348,
      "learning_rate": 9.37095110131133e-06,
      "loss": 1.5161,
      "step": 2602
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.2194390296936035,
      "learning_rate": 9.370387067273342e-06,
      "loss": 0.1033,
      "step": 2603
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9585490226745605,
      "learning_rate": 9.369822797469517e-06,
      "loss": 0.1541,
      "step": 2604
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.180684804916382,
      "learning_rate": 9.369258291930292e-06,
      "loss": 0.2078,
      "step": 2605
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.7914624214172363,
      "learning_rate": 9.36869355068612e-06,
      "loss": 0.2915,
      "step": 2606
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.881542921066284,
      "learning_rate": 9.368128573767468e-06,
      "loss": 0.1983,
      "step": 2607
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.080543518066406,
      "learning_rate": 9.367563361204814e-06,
      "loss": 1.1057,
      "step": 2608
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.386143684387207,
      "learning_rate": 9.366997913028649e-06,
      "loss": 1.2286,
      "step": 2609
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.2477335929870605,
      "learning_rate": 9.366432229269476e-06,
      "loss": 0.696,
      "step": 2610
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.668393850326538,
      "learning_rate": 9.365866309957812e-06,
      "loss": 0.1621,
      "step": 2611
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.6385931968688965,
      "learning_rate": 9.365300155124186e-06,
      "loss": 0.2295,
      "step": 2612
    },
    {
      "epoch": 0.19,
      "grad_norm": 33.393218994140625,
      "learning_rate": 9.36473376479914e-06,
      "loss": 1.2417,
      "step": 2613
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.198997497558594,
      "learning_rate": 9.36416713901323e-06,
      "loss": 0.898,
      "step": 2614
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.0237014293670654,
      "learning_rate": 9.363600277797023e-06,
      "loss": 0.129,
      "step": 2615
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.824865818023682,
      "learning_rate": 9.363033181181096e-06,
      "loss": 0.7899,
      "step": 2616
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4286768436431885,
      "learning_rate": 9.362465849196044e-06,
      "loss": 0.341,
      "step": 2617
    },
    {
      "epoch": 0.19,
      "grad_norm": 25.289836883544922,
      "learning_rate": 9.361898281872474e-06,
      "loss": 1.4899,
      "step": 2618
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.637411117553711,
      "learning_rate": 9.361330479241e-06,
      "loss": 1.1903,
      "step": 2619
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.263806343078613,
      "learning_rate": 9.360762441332256e-06,
      "loss": 0.9226,
      "step": 2620
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.085941314697266,
      "learning_rate": 9.360194168176887e-06,
      "loss": 1.2624,
      "step": 2621
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9941446781158447,
      "learning_rate": 9.359625659805543e-06,
      "loss": 0.0878,
      "step": 2622
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4665591716766357,
      "learning_rate": 9.359056916248898e-06,
      "loss": 0.1649,
      "step": 2623
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.0798144340515137,
      "learning_rate": 9.358487937537635e-06,
      "loss": 0.1755,
      "step": 2624
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.0280442237854004,
      "learning_rate": 9.357918723702443e-06,
      "loss": 0.1807,
      "step": 2625
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2885830402374268,
      "learning_rate": 9.357349274774032e-06,
      "loss": 0.1125,
      "step": 2626
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.194838047027588,
      "learning_rate": 9.35677959078312e-06,
      "loss": 0.2389,
      "step": 2627
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.710905075073242,
      "learning_rate": 9.356209671760443e-06,
      "loss": 0.9193,
      "step": 2628
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4832100868225098,
      "learning_rate": 9.355639517736742e-06,
      "loss": 0.255,
      "step": 2629
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.151353359222412,
      "learning_rate": 9.355069128742775e-06,
      "loss": 0.1353,
      "step": 2630
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6722095012664795,
      "learning_rate": 9.354498504809313e-06,
      "loss": 0.1313,
      "step": 2631
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.923968315124512,
      "learning_rate": 9.353927645967142e-06,
      "loss": 1.1767,
      "step": 2632
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.977903366088867,
      "learning_rate": 9.353356552247052e-06,
      "loss": 1.108,
      "step": 2633
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.333343982696533,
      "learning_rate": 9.352785223679857e-06,
      "loss": 0.2263,
      "step": 2634
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.939363718032837,
      "learning_rate": 9.352213660296374e-06,
      "loss": 0.271,
      "step": 2635
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8383054733276367,
      "learning_rate": 9.351641862127438e-06,
      "loss": 0.236,
      "step": 2636
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.614631175994873,
      "learning_rate": 9.351069829203896e-06,
      "loss": 0.2026,
      "step": 2637
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.652910232543945,
      "learning_rate": 9.350497561556606e-06,
      "loss": 1.3206,
      "step": 2638
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.207006931304932,
      "learning_rate": 9.349925059216441e-06,
      "loss": 0.2108,
      "step": 2639
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.209299087524414,
      "learning_rate": 9.349352322214282e-06,
      "loss": 1.3644,
      "step": 2640
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.874880313873291,
      "learning_rate": 9.34877935058103e-06,
      "loss": 0.198,
      "step": 2641
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.4494547843933105,
      "learning_rate": 9.348206144347593e-06,
      "loss": 0.9904,
      "step": 2642
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.565636157989502,
      "learning_rate": 9.347632703544893e-06,
      "loss": 1.2047,
      "step": 2643
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.879610776901245,
      "learning_rate": 9.347059028203866e-06,
      "loss": 0.2438,
      "step": 2644
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.8511505126953125,
      "learning_rate": 9.34648511835546e-06,
      "loss": 0.9934,
      "step": 2645
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.897060871124268,
      "learning_rate": 9.345910974030632e-06,
      "loss": 0.1968,
      "step": 2646
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4824752807617188,
      "learning_rate": 9.345336595260358e-06,
      "loss": 0.1325,
      "step": 2647
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.351438522338867,
      "learning_rate": 9.344761982075622e-06,
      "loss": 1.0178,
      "step": 2648
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.713492393493652,
      "learning_rate": 9.344187134507425e-06,
      "loss": 1.3227,
      "step": 2649
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.5624983310699463,
      "learning_rate": 9.343612052586774e-06,
      "loss": 0.1726,
      "step": 2650
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.772265434265137,
      "learning_rate": 9.343036736344695e-06,
      "loss": 0.9676,
      "step": 2651
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.592134952545166,
      "learning_rate": 9.34246118581222e-06,
      "loss": 0.2807,
      "step": 2652
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.049394607543945,
      "learning_rate": 9.341885401020405e-06,
      "loss": 0.8596,
      "step": 2653
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.1544759273529053,
      "learning_rate": 9.341309382000307e-06,
      "loss": 0.1604,
      "step": 2654
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.091894149780273,
      "learning_rate": 9.340733128783e-06,
      "loss": 1.4755,
      "step": 2655
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.341187953948975,
      "learning_rate": 9.340156641399572e-06,
      "loss": 1.0338,
      "step": 2656
    },
    {
      "epoch": 0.19,
      "grad_norm": 24.726282119750977,
      "learning_rate": 9.339579919881122e-06,
      "loss": 0.9277,
      "step": 2657
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.135340690612793,
      "learning_rate": 9.339002964258758e-06,
      "loss": 0.9509,
      "step": 2658
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.706071615219116,
      "learning_rate": 9.33842577456361e-06,
      "loss": 0.1488,
      "step": 2659
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.541613817214966,
      "learning_rate": 9.337848350826815e-06,
      "loss": 0.0795,
      "step": 2660
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.120638132095337,
      "learning_rate": 9.337270693079521e-06,
      "loss": 0.2014,
      "step": 2661
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.751246929168701,
      "learning_rate": 9.336692801352889e-06,
      "loss": 1.0041,
      "step": 2662
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.56098747253418,
      "learning_rate": 9.336114675678096e-06,
      "loss": 1.0416,
      "step": 2663
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.1431503295898438,
      "learning_rate": 9.33553631608633e-06,
      "loss": 0.1659,
      "step": 2664
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9167320728302,
      "learning_rate": 9.33495772260879e-06,
      "loss": 0.1446,
      "step": 2665
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.62926197052002,
      "learning_rate": 9.33437889527669e-06,
      "loss": 0.9667,
      "step": 2666
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.250364303588867,
      "learning_rate": 9.333799834121256e-06,
      "loss": 0.9713,
      "step": 2667
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.085865497589111,
      "learning_rate": 9.333220539173722e-06,
      "loss": 0.2689,
      "step": 2668
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.689435958862305,
      "learning_rate": 9.332641010465346e-06,
      "loss": 0.9613,
      "step": 2669
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.923173904418945,
      "learning_rate": 9.332061248027385e-06,
      "loss": 1.0185,
      "step": 2670
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.880165100097656,
      "learning_rate": 9.331481251891117e-06,
      "loss": 1.6412,
      "step": 2671
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.200841426849365,
      "learning_rate": 9.330901022087833e-06,
      "loss": 0.6896,
      "step": 2672
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.2310047149658203,
      "learning_rate": 9.330320558648831e-06,
      "loss": 0.1517,
      "step": 2673
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.878687858581543,
      "learning_rate": 9.329739861605427e-06,
      "loss": 1.2548,
      "step": 2674
    },
    {
      "epoch": 0.19,
      "grad_norm": 18.625659942626953,
      "learning_rate": 9.329158930988944e-06,
      "loss": 1.1644,
      "step": 2675
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.506759643554688,
      "learning_rate": 9.328577766830726e-06,
      "loss": 1.0877,
      "step": 2676
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.767807006835938,
      "learning_rate": 9.32799636916212e-06,
      "loss": 1.2066,
      "step": 2677
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.7302162647247314,
      "learning_rate": 9.327414738014496e-06,
      "loss": 0.2027,
      "step": 2678
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.266034126281738,
      "learning_rate": 9.326832873419222e-06,
      "loss": 0.9317,
      "step": 2679
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9835450649261475,
      "learning_rate": 9.326250775407696e-06,
      "loss": 0.1492,
      "step": 2680
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.04509973526001,
      "learning_rate": 9.325668444011315e-06,
      "loss": 0.9713,
      "step": 2681
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.0610294342041016,
      "learning_rate": 9.325085879261494e-06,
      "loss": 0.1402,
      "step": 2682
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.999900817871094,
      "learning_rate": 9.324503081189665e-06,
      "loss": 1.8414,
      "step": 2683
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4945967197418213,
      "learning_rate": 9.323920049827259e-06,
      "loss": 0.1568,
      "step": 2684
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.422374963760376,
      "learning_rate": 9.323336785205737e-06,
      "loss": 0.1369,
      "step": 2685
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.0675389766693115,
      "learning_rate": 9.322753287356559e-06,
      "loss": 0.2076,
      "step": 2686
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.195417404174805,
      "learning_rate": 9.322169556311203e-06,
      "loss": 0.9806,
      "step": 2687
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.577984809875488,
      "learning_rate": 9.32158559210116e-06,
      "loss": 1.0024,
      "step": 2688
    },
    {
      "epoch": 0.19,
      "grad_norm": 23.67171859741211,
      "learning_rate": 9.321001394757932e-06,
      "loss": 1.9379,
      "step": 2689
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.686534881591797,
      "learning_rate": 9.320416964313034e-06,
      "loss": 0.16,
      "step": 2690
    },
    {
      "epoch": 0.19,
      "grad_norm": 33.809017181396484,
      "learning_rate": 9.319832300797997e-06,
      "loss": 0.7796,
      "step": 2691
    },
    {
      "epoch": 0.19,
      "grad_norm": 24.653528213500977,
      "learning_rate": 9.319247404244357e-06,
      "loss": 1.8169,
      "step": 2692
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.636938095092773,
      "learning_rate": 9.31866227468367e-06,
      "loss": 1.0561,
      "step": 2693
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.29718589782715,
      "learning_rate": 9.3180769121475e-06,
      "loss": 0.8014,
      "step": 2694
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4960951805114746,
      "learning_rate": 9.317491316667423e-06,
      "loss": 0.3033,
      "step": 2695
    },
    {
      "epoch": 0.19,
      "grad_norm": 25.192880630493164,
      "learning_rate": 9.316905488275035e-06,
      "loss": 1.117,
      "step": 2696
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.587412357330322,
      "learning_rate": 9.316319427001936e-06,
      "loss": 0.9089,
      "step": 2697
    },
    {
      "epoch": 0.19,
      "grad_norm": 18.87272834777832,
      "learning_rate": 9.31573313287974e-06,
      "loss": 1.2624,
      "step": 2698
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.885823249816895,
      "learning_rate": 9.31514660594008e-06,
      "loss": 1.094,
      "step": 2699
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.677677154541016,
      "learning_rate": 9.314559846214593e-06,
      "loss": 1.0148,
      "step": 2700
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.842366218566895,
      "learning_rate": 9.313972853734932e-06,
      "loss": 1.2357,
      "step": 2701
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.6431033611297607,
      "learning_rate": 9.313385628532767e-06,
      "loss": 0.216,
      "step": 2702
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.3375444412231445,
      "learning_rate": 9.312798170639774e-06,
      "loss": 0.2561,
      "step": 2703
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.749300003051758,
      "learning_rate": 9.312210480087645e-06,
      "loss": 0.1687,
      "step": 2704
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.809762001037598,
      "learning_rate": 9.311622556908082e-06,
      "loss": 1.3376,
      "step": 2705
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.772726535797119,
      "learning_rate": 9.311034401132804e-06,
      "loss": 0.2404,
      "step": 2706
    },
    {
      "epoch": 0.19,
      "grad_norm": 21.24615478515625,
      "learning_rate": 9.310446012793537e-06,
      "loss": 1.0502,
      "step": 2707
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9607086181640625,
      "learning_rate": 9.309857391922024e-06,
      "loss": 0.1895,
      "step": 2708
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.934116840362549,
      "learning_rate": 9.309268538550017e-06,
      "loss": 0.8472,
      "step": 2709
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8827455043792725,
      "learning_rate": 9.308679452709287e-06,
      "loss": 0.2102,
      "step": 2710
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6949193477630615,
      "learning_rate": 9.308090134431607e-06,
      "loss": 0.2088,
      "step": 2711
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.416539192199707,
      "learning_rate": 9.307500583748772e-06,
      "loss": 0.9587,
      "step": 2712
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.136409282684326,
      "learning_rate": 9.306910800692586e-06,
      "loss": 0.9821,
      "step": 2713
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.926033020019531,
      "learning_rate": 9.306320785294865e-06,
      "loss": 1.1753,
      "step": 2714
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.512986183166504,
      "learning_rate": 9.305730537587436e-06,
      "loss": 0.8002,
      "step": 2715
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.2149224281311035,
      "learning_rate": 9.305140057602145e-06,
      "loss": 0.7547,
      "step": 2716
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.84565258026123,
      "learning_rate": 9.304549345370841e-06,
      "loss": 0.7753,
      "step": 2717
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.884396553039551,
      "learning_rate": 9.303958400925396e-06,
      "loss": 0.9705,
      "step": 2718
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.498124122619629,
      "learning_rate": 9.303367224297687e-06,
      "loss": 0.9179,
      "step": 2719
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.508824586868286,
      "learning_rate": 9.302775815519603e-06,
      "loss": 0.1777,
      "step": 2720
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8358943462371826,
      "learning_rate": 9.302184174623052e-06,
      "loss": 0.18,
      "step": 2721
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.493263244628906,
      "learning_rate": 9.301592301639948e-06,
      "loss": 0.9686,
      "step": 2722
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.0986602306365967,
      "learning_rate": 9.301000196602223e-06,
      "loss": 0.2021,
      "step": 2723
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9728784561157227,
      "learning_rate": 9.300407859541817e-06,
      "loss": 0.1496,
      "step": 2724
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.6772959232330322,
      "learning_rate": 9.299815290490686e-06,
      "loss": 0.2888,
      "step": 2725
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.449040651321411,
      "learning_rate": 9.299222489480795e-06,
      "loss": 0.2494,
      "step": 2726
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.3291566371917725,
      "learning_rate": 9.298629456544123e-06,
      "loss": 0.1224,
      "step": 2727
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.187983989715576,
      "learning_rate": 9.298036191712665e-06,
      "loss": 0.1933,
      "step": 2728
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.529912948608398,
      "learning_rate": 9.297442695018421e-06,
      "loss": 1.011,
      "step": 2729
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6697020530700684,
      "learning_rate": 9.296848966493412e-06,
      "loss": 0.0669,
      "step": 2730
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.331397533416748,
      "learning_rate": 9.296255006169667e-06,
      "loss": 0.257,
      "step": 2731
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.348870277404785,
      "learning_rate": 9.295660814079224e-06,
      "loss": 0.1948,
      "step": 2732
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.003155469894409,
      "learning_rate": 9.29506639025414e-06,
      "loss": 0.1521,
      "step": 2733
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.79599928855896,
      "learning_rate": 9.294471734726486e-06,
      "loss": 0.253,
      "step": 2734
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.391707181930542,
      "learning_rate": 9.293876847528333e-06,
      "loss": 0.1729,
      "step": 2735
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.7061123847961426,
      "learning_rate": 9.29328172869178e-06,
      "loss": 0.2598,
      "step": 2736
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.409526824951172,
      "learning_rate": 9.292686378248927e-06,
      "loss": 1.2318,
      "step": 2737
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5231449604034424,
      "learning_rate": 9.292090796231893e-06,
      "loss": 0.1117,
      "step": 2738
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.011224269866943,
      "learning_rate": 9.291494982672808e-06,
      "loss": 0.28,
      "step": 2739
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.4531989097595215,
      "learning_rate": 9.290898937603811e-06,
      "loss": 0.2873,
      "step": 2740
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.391073703765869,
      "learning_rate": 9.29030266105706e-06,
      "loss": 0.7753,
      "step": 2741
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.130692481994629,
      "learning_rate": 9.28970615306472e-06,
      "loss": 1.1432,
      "step": 2742
    },
    {
      "epoch": 0.2,
      "grad_norm": 18.301334381103516,
      "learning_rate": 9.28910941365897e-06,
      "loss": 1.5576,
      "step": 2743
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.02585768699646,
      "learning_rate": 9.288512442872002e-06,
      "loss": 0.1794,
      "step": 2744
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.770378112792969,
      "learning_rate": 9.28791524073602e-06,
      "loss": 1.0378,
      "step": 2745
    },
    {
      "epoch": 0.2,
      "grad_norm": 30.758499145507812,
      "learning_rate": 9.287317807283243e-06,
      "loss": 0.8832,
      "step": 2746
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.286774158477783,
      "learning_rate": 9.286720142545897e-06,
      "loss": 0.1355,
      "step": 2747
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.53621768951416,
      "learning_rate": 9.286122246556227e-06,
      "loss": 1.2084,
      "step": 2748
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.48896598815918,
      "learning_rate": 9.285524119346483e-06,
      "loss": 0.8569,
      "step": 2749
    },
    {
      "epoch": 0.2,
      "grad_norm": 21.987319946289062,
      "learning_rate": 9.284925760948938e-06,
      "loss": 1.7803,
      "step": 2750
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.65814733505249,
      "learning_rate": 9.284327171395866e-06,
      "loss": 0.8195,
      "step": 2751
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.5178914070129395,
      "learning_rate": 9.283728350719559e-06,
      "loss": 0.8166,
      "step": 2752
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.276876926422119,
      "learning_rate": 9.283129298952322e-06,
      "loss": 0.1125,
      "step": 2753
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.907129287719727,
      "learning_rate": 9.282530016126474e-06,
      "loss": 1.2046,
      "step": 2754
    },
    {
      "epoch": 0.2,
      "grad_norm": 19.98345184326172,
      "learning_rate": 9.281930502274341e-06,
      "loss": 1.0831,
      "step": 2755
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.065156936645508,
      "learning_rate": 9.281330757428265e-06,
      "loss": 1.5648,
      "step": 2756
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.147533416748047,
      "learning_rate": 9.280730781620602e-06,
      "loss": 0.8583,
      "step": 2757
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.406851768493652,
      "learning_rate": 9.280130574883714e-06,
      "loss": 0.2404,
      "step": 2758
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.646657943725586,
      "learning_rate": 9.279530137249985e-06,
      "loss": 0.1681,
      "step": 2759
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.222623825073242,
      "learning_rate": 9.278929468751804e-06,
      "loss": 0.9384,
      "step": 2760
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.407957077026367,
      "learning_rate": 9.278328569421574e-06,
      "loss": 1.0059,
      "step": 2761
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.68196177482605,
      "learning_rate": 9.27772743929171e-06,
      "loss": 0.194,
      "step": 2762
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.465167999267578,
      "learning_rate": 9.277126078394645e-06,
      "loss": 0.6242,
      "step": 2763
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.174593687057495,
      "learning_rate": 9.276524486762819e-06,
      "loss": 0.0932,
      "step": 2764
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.5288727283477783,
      "learning_rate": 9.275922664428682e-06,
      "loss": 0.1598,
      "step": 2765
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.919507026672363,
      "learning_rate": 9.275320611424703e-06,
      "loss": 0.8777,
      "step": 2766
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.572737216949463,
      "learning_rate": 9.27471832778336e-06,
      "loss": 0.5041,
      "step": 2767
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.037841796875,
      "learning_rate": 9.274115813537145e-06,
      "loss": 1.2481,
      "step": 2768
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.960662364959717,
      "learning_rate": 9.27351306871856e-06,
      "loss": 0.2056,
      "step": 2769
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.20881175994873,
      "learning_rate": 9.272910093360122e-06,
      "loss": 1.0518,
      "step": 2770
    },
    {
      "epoch": 0.2,
      "grad_norm": 23.55914878845215,
      "learning_rate": 9.272306887494355e-06,
      "loss": 0.9744,
      "step": 2771
    },
    {
      "epoch": 0.2,
      "grad_norm": 19.67010498046875,
      "learning_rate": 9.271703451153806e-06,
      "loss": 1.1973,
      "step": 2772
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.56431245803833,
      "learning_rate": 9.271099784371024e-06,
      "loss": 0.2684,
      "step": 2773
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.922235488891602,
      "learning_rate": 9.270495887178575e-06,
      "loss": 0.9749,
      "step": 2774
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.497174263000488,
      "learning_rate": 9.269891759609038e-06,
      "loss": 0.9433,
      "step": 2775
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.295557498931885,
      "learning_rate": 9.269287401695005e-06,
      "loss": 0.9096,
      "step": 2776
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.40404224395752,
      "learning_rate": 9.268682813469074e-06,
      "loss": 1.064,
      "step": 2777
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.0849404335021973,
      "learning_rate": 9.268077994963864e-06,
      "loss": 0.3105,
      "step": 2778
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.72822904586792,
      "learning_rate": 9.267472946212002e-06,
      "loss": 1.0144,
      "step": 2779
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.425432205200195,
      "learning_rate": 9.266867667246127e-06,
      "loss": 1.0013,
      "step": 2780
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.262612342834473,
      "learning_rate": 9.266262158098893e-06,
      "loss": 0.8762,
      "step": 2781
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.87933874130249,
      "learning_rate": 9.265656418802964e-06,
      "loss": 1.0044,
      "step": 2782
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.214265823364258,
      "learning_rate": 9.265050449391018e-06,
      "loss": 1.3823,
      "step": 2783
    },
    {
      "epoch": 0.2,
      "grad_norm": 35.80508041381836,
      "learning_rate": 9.264444249895745e-06,
      "loss": 1.0508,
      "step": 2784
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.281342506408691,
      "learning_rate": 9.263837820349846e-06,
      "loss": 0.7045,
      "step": 2785
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.8561220169067383,
      "learning_rate": 9.263231160786035e-06,
      "loss": 0.2108,
      "step": 2786
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.6715545654296875,
      "learning_rate": 9.262624271237042e-06,
      "loss": 1.069,
      "step": 2787
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.0235977172851562,
      "learning_rate": 9.262017151735602e-06,
      "loss": 0.091,
      "step": 2788
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.8006272315979,
      "learning_rate": 9.261409802314472e-06,
      "loss": 0.3098,
      "step": 2789
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.205580472946167,
      "learning_rate": 9.26080222300641e-06,
      "loss": 0.1661,
      "step": 2790
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2396273612976074,
      "learning_rate": 9.2601944138442e-06,
      "loss": 0.1126,
      "step": 2791
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9327433109283447,
      "learning_rate": 9.259586374860625e-06,
      "loss": 0.1641,
      "step": 2792
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.466054916381836,
      "learning_rate": 9.258978106088488e-06,
      "loss": 1.5765,
      "step": 2793
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.9273903369903564,
      "learning_rate": 9.258369607560605e-06,
      "loss": 0.2296,
      "step": 2794
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.10890531539917,
      "learning_rate": 9.257760879309799e-06,
      "loss": 0.1688,
      "step": 2795
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.067045211791992,
      "learning_rate": 9.25715192136891e-06,
      "loss": 1.0495,
      "step": 2796
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.862573623657227,
      "learning_rate": 9.25654273377079e-06,
      "loss": 0.2242,
      "step": 2797
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.36393928527832,
      "learning_rate": 9.255933316548302e-06,
      "loss": 0.9664,
      "step": 2798
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.0317769050598145,
      "learning_rate": 9.25532366973432e-06,
      "loss": 0.2629,
      "step": 2799
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.2845876216888428,
      "learning_rate": 9.254713793361733e-06,
      "loss": 0.1896,
      "step": 2800
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.73851203918457,
      "learning_rate": 9.254103687463443e-06,
      "loss": 0.9145,
      "step": 2801
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.66867733001709,
      "learning_rate": 9.25349335207236e-06,
      "loss": 0.801,
      "step": 2802
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.789694786071777,
      "learning_rate": 9.252882787221413e-06,
      "loss": 0.7549,
      "step": 2803
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8469014167785645,
      "learning_rate": 9.252271992943536e-06,
      "loss": 0.3946,
      "step": 2804
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.346832752227783,
      "learning_rate": 9.251660969271682e-06,
      "loss": 1.0512,
      "step": 2805
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.168092727661133,
      "learning_rate": 9.251049716238811e-06,
      "loss": 0.7762,
      "step": 2806
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.482666015625,
      "learning_rate": 9.2504382338779e-06,
      "loss": 0.59,
      "step": 2807
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.550238609313965,
      "learning_rate": 9.249826522221935e-06,
      "loss": 0.8566,
      "step": 2808
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.5511248111724854,
      "learning_rate": 9.249214581303913e-06,
      "loss": 0.2594,
      "step": 2809
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.44143295288086,
      "learning_rate": 9.248602411156852e-06,
      "loss": 1.1587,
      "step": 2810
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.594829797744751,
      "learning_rate": 9.247990011813771e-06,
      "loss": 0.149,
      "step": 2811
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.538214206695557,
      "learning_rate": 9.247377383307709e-06,
      "loss": 1.2128,
      "step": 2812
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.442255020141602,
      "learning_rate": 9.246764525671714e-06,
      "loss": 0.5694,
      "step": 2813
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1564157009124756,
      "learning_rate": 9.246151438938847e-06,
      "loss": 0.0923,
      "step": 2814
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.9953837394714355,
      "learning_rate": 9.245538123142184e-06,
      "loss": 0.1663,
      "step": 2815
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.7865447998046875,
      "learning_rate": 9.244924578314808e-06,
      "loss": 0.107,
      "step": 2816
    },
    {
      "epoch": 0.2,
      "grad_norm": 33.99858093261719,
      "learning_rate": 9.24431080448982e-06,
      "loss": 0.9153,
      "step": 2817
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.364986419677734,
      "learning_rate": 9.24369680170033e-06,
      "loss": 1.2833,
      "step": 2818
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9866530895233154,
      "learning_rate": 9.243082569979458e-06,
      "loss": 0.1731,
      "step": 2819
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.826971054077148,
      "learning_rate": 9.242468109360345e-06,
      "loss": 1.2413,
      "step": 2820
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.171114444732666,
      "learning_rate": 9.241853419876136e-06,
      "loss": 0.8804,
      "step": 2821
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.516740798950195,
      "learning_rate": 9.24123850155999e-06,
      "loss": 0.6634,
      "step": 2822
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.234861850738525,
      "learning_rate": 9.240623354445083e-06,
      "loss": 0.8196,
      "step": 2823
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3925106525421143,
      "learning_rate": 9.240007978564596e-06,
      "loss": 0.0711,
      "step": 2824
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.532957077026367,
      "learning_rate": 9.239392373951727e-06,
      "loss": 0.4086,
      "step": 2825
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.886420488357544,
      "learning_rate": 9.238776540639688e-06,
      "loss": 0.1531,
      "step": 2826
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.917294502258301,
      "learning_rate": 9.2381604786617e-06,
      "loss": 0.1184,
      "step": 2827
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.392552852630615,
      "learning_rate": 9.237544188050994e-06,
      "loss": 0.8247,
      "step": 2828
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9027440547943115,
      "learning_rate": 9.23692766884082e-06,
      "loss": 0.1427,
      "step": 2829
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.621631145477295,
      "learning_rate": 9.236310921064437e-06,
      "loss": 0.3007,
      "step": 2830
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.753511905670166,
      "learning_rate": 9.235693944755114e-06,
      "loss": 0.1444,
      "step": 2831
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.283461093902588,
      "learning_rate": 9.235076739946136e-06,
      "loss": 0.7861,
      "step": 2832
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.138086795806885,
      "learning_rate": 9.234459306670798e-06,
      "loss": 0.8309,
      "step": 2833
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9313297271728516,
      "learning_rate": 9.23384164496241e-06,
      "loss": 0.2054,
      "step": 2834
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.565382719039917,
      "learning_rate": 9.23322375485429e-06,
      "loss": 0.2274,
      "step": 2835
    },
    {
      "epoch": 0.2,
      "grad_norm": 71.68184661865234,
      "learning_rate": 9.232605636379772e-06,
      "loss": 1.5293,
      "step": 2836
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5463342666625977,
      "learning_rate": 9.231987289572201e-06,
      "loss": 0.0913,
      "step": 2837
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8783586025238037,
      "learning_rate": 9.231368714464937e-06,
      "loss": 0.3087,
      "step": 2838
    },
    {
      "epoch": 0.2,
      "grad_norm": 19.813804626464844,
      "learning_rate": 9.230749911091346e-06,
      "loss": 1.5679,
      "step": 2839
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.850502967834473,
      "learning_rate": 9.230130879484811e-06,
      "loss": 1.6826,
      "step": 2840
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.8307132720947266,
      "learning_rate": 9.229511619678727e-06,
      "loss": 0.2607,
      "step": 2841
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.5501959323883057,
      "learning_rate": 9.228892131706502e-06,
      "loss": 0.2379,
      "step": 2842
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.351670503616333,
      "learning_rate": 9.228272415601552e-06,
      "loss": 0.2305,
      "step": 2843
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.883230209350586,
      "learning_rate": 9.22765247139731e-06,
      "loss": 0.2148,
      "step": 2844
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.8316938877105713,
      "learning_rate": 9.227032299127221e-06,
      "loss": 0.1837,
      "step": 2845
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.4191668033599854,
      "learning_rate": 9.226411898824738e-06,
      "loss": 0.1737,
      "step": 2846
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.0577142238616943,
      "learning_rate": 9.225791270523332e-06,
      "loss": 0.1134,
      "step": 2847
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.58112096786499,
      "learning_rate": 9.225170414256483e-06,
      "loss": 0.8402,
      "step": 2848
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.3586316108703613,
      "learning_rate": 9.224549330057684e-06,
      "loss": 0.1929,
      "step": 2849
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.219071865081787,
      "learning_rate": 9.223928017960437e-06,
      "loss": 0.8616,
      "step": 2850
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5745885372161865,
      "learning_rate": 9.223306477998265e-06,
      "loss": 0.1565,
      "step": 2851
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.9246392250061035,
      "learning_rate": 9.22268471020469e-06,
      "loss": 1.1563,
      "step": 2852
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.3752310276031494,
      "learning_rate": 9.222062714613263e-06,
      "loss": 0.2628,
      "step": 2853
    },
    {
      "epoch": 0.2,
      "grad_norm": 19.790983200073242,
      "learning_rate": 9.221440491257531e-06,
      "loss": 1.4232,
      "step": 2854
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.001333236694336,
      "learning_rate": 9.220818040171065e-06,
      "loss": 1.3622,
      "step": 2855
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6472349166870117,
      "learning_rate": 9.220195361387442e-06,
      "loss": 0.1378,
      "step": 2856
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.2196567058563232,
      "learning_rate": 9.219572454940253e-06,
      "loss": 0.2686,
      "step": 2857
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.672839641571045,
      "learning_rate": 9.218949320863104e-06,
      "loss": 0.9807,
      "step": 2858
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.282482147216797,
      "learning_rate": 9.218325959189608e-06,
      "loss": 0.1863,
      "step": 2859
    },
    {
      "epoch": 0.21,
      "grad_norm": 28.604055404663086,
      "learning_rate": 9.217702369953392e-06,
      "loss": 0.9459,
      "step": 2860
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.58562469482422,
      "learning_rate": 9.217078553188099e-06,
      "loss": 1.6457,
      "step": 2861
    },
    {
      "epoch": 0.21,
      "grad_norm": 20.01325798034668,
      "learning_rate": 9.216454508927381e-06,
      "loss": 1.5226,
      "step": 2862
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.780285596847534,
      "learning_rate": 9.215830237204902e-06,
      "loss": 0.1321,
      "step": 2863
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.104820966720581,
      "learning_rate": 9.215205738054339e-06,
      "loss": 0.1114,
      "step": 2864
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.3378472328186035,
      "learning_rate": 9.214581011509383e-06,
      "loss": 0.234,
      "step": 2865
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.4815168380737305,
      "learning_rate": 9.213956057603734e-06,
      "loss": 0.2414,
      "step": 2866
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.9712657928466797,
      "learning_rate": 9.213330876371105e-06,
      "loss": 0.4146,
      "step": 2867
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.38607406616211,
      "learning_rate": 9.212705467845226e-06,
      "loss": 0.9761,
      "step": 2868
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.088327407836914,
      "learning_rate": 9.212079832059831e-06,
      "loss": 0.9823,
      "step": 2869
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.759021282196045,
      "learning_rate": 9.211453969048673e-06,
      "loss": 0.8872,
      "step": 2870
    },
    {
      "epoch": 0.21,
      "grad_norm": 41.99214553833008,
      "learning_rate": 9.210827878845516e-06,
      "loss": 1.865,
      "step": 2871
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8496999740600586,
      "learning_rate": 9.210201561484132e-06,
      "loss": 0.1235,
      "step": 2872
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.146448612213135,
      "learning_rate": 9.20957501699831e-06,
      "loss": 0.9146,
      "step": 2873
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.057476043701172,
      "learning_rate": 9.208948245421853e-06,
      "loss": 1.1599,
      "step": 2874
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8264729976654053,
      "learning_rate": 9.208321246788567e-06,
      "loss": 0.143,
      "step": 2875
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.420790195465088,
      "learning_rate": 9.207694021132281e-06,
      "loss": 0.2058,
      "step": 2876
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.735781669616699,
      "learning_rate": 9.207066568486827e-06,
      "loss": 0.1908,
      "step": 2877
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.694308757781982,
      "learning_rate": 9.206438888886058e-06,
      "loss": 0.7265,
      "step": 2878
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.725190162658691,
      "learning_rate": 9.205810982363835e-06,
      "loss": 1.2507,
      "step": 2879
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.15093469619751,
      "learning_rate": 9.205182848954027e-06,
      "loss": 0.2796,
      "step": 2880
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.513794422149658,
      "learning_rate": 9.20455448869052e-06,
      "loss": 1.0641,
      "step": 2881
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.3770344257354736,
      "learning_rate": 9.203925901607216e-06,
      "loss": 0.2817,
      "step": 2882
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5844714641571045,
      "learning_rate": 9.203297087738023e-06,
      "loss": 0.2101,
      "step": 2883
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.611845970153809,
      "learning_rate": 9.20266804711686e-06,
      "loss": 1.7252,
      "step": 2884
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.075446128845215,
      "learning_rate": 9.202038779777666e-06,
      "loss": 1.2661,
      "step": 2885
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.999151229858398,
      "learning_rate": 9.201409285754384e-06,
      "loss": 1.4336,
      "step": 2886
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.808654546737671,
      "learning_rate": 9.200779565080976e-06,
      "loss": 0.1732,
      "step": 2887
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.293582916259766,
      "learning_rate": 9.20014961779141e-06,
      "loss": 1.1493,
      "step": 2888
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.683685302734375,
      "learning_rate": 9.19951944391967e-06,
      "loss": 0.9637,
      "step": 2889
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.59009838104248,
      "learning_rate": 9.198889043499752e-06,
      "loss": 0.9051,
      "step": 2890
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.444679260253906,
      "learning_rate": 9.198258416565663e-06,
      "loss": 0.7301,
      "step": 2891
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.268258094787598,
      "learning_rate": 9.197627563151427e-06,
      "loss": 1.2757,
      "step": 2892
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5452451705932617,
      "learning_rate": 9.196996483291071e-06,
      "loss": 0.1374,
      "step": 2893
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.717588186264038,
      "learning_rate": 9.19636517701864e-06,
      "loss": 0.134,
      "step": 2894
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8668980598449707,
      "learning_rate": 9.195733644368195e-06,
      "loss": 0.1622,
      "step": 2895
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.586726427078247,
      "learning_rate": 9.195101885373798e-06,
      "loss": 0.3412,
      "step": 2896
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.76254940032959,
      "learning_rate": 9.194469900069536e-06,
      "loss": 0.9494,
      "step": 2897
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.221315860748291,
      "learning_rate": 9.193837688489499e-06,
      "loss": 0.2114,
      "step": 2898
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.908536911010742,
      "learning_rate": 9.193205250667792e-06,
      "loss": 1.5418,
      "step": 2899
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.30324935913086,
      "learning_rate": 9.192572586638537e-06,
      "loss": 0.8954,
      "step": 2900
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.7098262310028076,
      "learning_rate": 9.191939696435857e-06,
      "loss": 0.2506,
      "step": 2901
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5435409545898438,
      "learning_rate": 9.191306580093901e-06,
      "loss": 0.2464,
      "step": 2902
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.1497628688812256,
      "learning_rate": 9.190673237646817e-06,
      "loss": 0.2375,
      "step": 2903
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.263763427734375,
      "learning_rate": 9.190039669128776e-06,
      "loss": 1.4105,
      "step": 2904
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.404533863067627,
      "learning_rate": 9.189405874573955e-06,
      "loss": 0.2485,
      "step": 2905
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8694868087768555,
      "learning_rate": 9.188771854016546e-06,
      "loss": 0.2399,
      "step": 2906
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.665217161178589,
      "learning_rate": 9.18813760749075e-06,
      "loss": 0.1595,
      "step": 2907
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.4573898315429688,
      "learning_rate": 9.187503135030782e-06,
      "loss": 0.2532,
      "step": 2908
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.810412645339966,
      "learning_rate": 9.18686843667087e-06,
      "loss": 0.2015,
      "step": 2909
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.172001838684082,
      "learning_rate": 9.186233512445255e-06,
      "loss": 0.7245,
      "step": 2910
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.327399492263794,
      "learning_rate": 9.185598362388188e-06,
      "loss": 0.2226,
      "step": 2911
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7555155754089355,
      "learning_rate": 9.184962986533933e-06,
      "loss": 0.1421,
      "step": 2912
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.361807346343994,
      "learning_rate": 9.184327384916765e-06,
      "loss": 0.1628,
      "step": 2913
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.936225891113281,
      "learning_rate": 9.183691557570973e-06,
      "loss": 1.0788,
      "step": 2914
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.557157516479492,
      "learning_rate": 9.183055504530857e-06,
      "loss": 1.0348,
      "step": 2915
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.957542419433594,
      "learning_rate": 9.182419225830733e-06,
      "loss": 0.8974,
      "step": 2916
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.70486831665039,
      "learning_rate": 9.18178272150492e-06,
      "loss": 1.1396,
      "step": 2917
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.845495223999023,
      "learning_rate": 9.181145991587762e-06,
      "loss": 1.2354,
      "step": 2918
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.138643503189087,
      "learning_rate": 9.180509036113601e-06,
      "loss": 0.1662,
      "step": 2919
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.510936737060547,
      "learning_rate": 9.179871855116802e-06,
      "loss": 0.8724,
      "step": 2920
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8723623752593994,
      "learning_rate": 9.179234448631737e-06,
      "loss": 0.1748,
      "step": 2921
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.351768493652344,
      "learning_rate": 9.178596816692796e-06,
      "loss": 0.9656,
      "step": 2922
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.974350690841675,
      "learning_rate": 9.17795895933437e-06,
      "loss": 0.1193,
      "step": 2923
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.1766090393066406,
      "learning_rate": 9.177320876590874e-06,
      "loss": 0.2415,
      "step": 2924
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.9202256202697754,
      "learning_rate": 9.17668256849673e-06,
      "loss": 0.1745,
      "step": 2925
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.0973896980285645,
      "learning_rate": 9.176044035086372e-06,
      "loss": 0.2616,
      "step": 2926
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.957859039306641,
      "learning_rate": 9.175405276394243e-06,
      "loss": 0.8459,
      "step": 2927
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.076824426651001,
      "learning_rate": 9.174766292454804e-06,
      "loss": 0.2598,
      "step": 2928
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.2448277473449707,
      "learning_rate": 9.174127083302527e-06,
      "loss": 0.1376,
      "step": 2929
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.233881950378418,
      "learning_rate": 9.173487648971893e-06,
      "loss": 0.7036,
      "step": 2930
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.136392116546631,
      "learning_rate": 9.172847989497397e-06,
      "loss": 0.8711,
      "step": 2931
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.816132068634033,
      "learning_rate": 9.172208104913549e-06,
      "loss": 0.7363,
      "step": 2932
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.3319973945617676,
      "learning_rate": 9.171567995254867e-06,
      "loss": 0.2913,
      "step": 2933
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.987983226776123,
      "learning_rate": 9.17092766055588e-06,
      "loss": 0.148,
      "step": 2934
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.093076705932617,
      "learning_rate": 9.170287100851133e-06,
      "loss": 0.7123,
      "step": 2935
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5349411964416504,
      "learning_rate": 9.169646316175183e-06,
      "loss": 0.1633,
      "step": 2936
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5979342460632324,
      "learning_rate": 9.169005306562598e-06,
      "loss": 0.1556,
      "step": 2937
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.832676887512207,
      "learning_rate": 9.168364072047957e-06,
      "loss": 1.1964,
      "step": 2938
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.7374870777130127,
      "learning_rate": 9.167722612665853e-06,
      "loss": 0.2282,
      "step": 2939
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.761624574661255,
      "learning_rate": 9.167080928450888e-06,
      "loss": 0.1799,
      "step": 2940
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.2324583530426025,
      "learning_rate": 9.16643901943768e-06,
      "loss": 0.2118,
      "step": 2941
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.070383071899414,
      "learning_rate": 9.16579688566086e-06,
      "loss": 0.925,
      "step": 2942
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.741703033447266,
      "learning_rate": 9.165154527155065e-06,
      "loss": 1.3244,
      "step": 2943
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.298766613006592,
      "learning_rate": 9.16451194395495e-06,
      "loss": 0.9176,
      "step": 2944
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.9744319915771484,
      "learning_rate": 9.163869136095178e-06,
      "loss": 0.2173,
      "step": 2945
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.6996684074401855,
      "learning_rate": 9.163226103610429e-06,
      "loss": 0.1242,
      "step": 2946
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.606008291244507,
      "learning_rate": 9.162582846535388e-06,
      "loss": 0.1043,
      "step": 2947
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.878764152526855,
      "learning_rate": 9.161939364904761e-06,
      "loss": 0.9576,
      "step": 2948
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.74566388130188,
      "learning_rate": 9.161295658753257e-06,
      "loss": 0.2658,
      "step": 2949
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.30720329284668,
      "learning_rate": 9.160651728115606e-06,
      "loss": 1.9285,
      "step": 2950
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.070989608764648,
      "learning_rate": 9.160007573026542e-06,
      "loss": 0.8032,
      "step": 2951
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.293943881988525,
      "learning_rate": 9.159363193520814e-06,
      "loss": 1.1175,
      "step": 2952
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.509190082550049,
      "learning_rate": 9.158718589633189e-06,
      "loss": 0.8728,
      "step": 2953
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.24274206161499,
      "learning_rate": 9.158073761398435e-06,
      "loss": 0.3173,
      "step": 2954
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.718416929244995,
      "learning_rate": 9.157428708851341e-06,
      "loss": 0.1627,
      "step": 2955
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.5332441329956055,
      "learning_rate": 9.156783432026707e-06,
      "loss": 0.1151,
      "step": 2956
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.285185813903809,
      "learning_rate": 9.15613793095934e-06,
      "loss": 0.8331,
      "step": 2957
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.95568561553955,
      "learning_rate": 9.155492205684062e-06,
      "loss": 0.9327,
      "step": 2958
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.656949043273926,
      "learning_rate": 9.154846256235712e-06,
      "loss": 0.2201,
      "step": 2959
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.78057861328125,
      "learning_rate": 9.154200082649131e-06,
      "loss": 1.1004,
      "step": 2960
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.0538554191589355,
      "learning_rate": 9.15355368495918e-06,
      "loss": 0.1135,
      "step": 2961
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.6450181007385254,
      "learning_rate": 9.15290706320073e-06,
      "loss": 0.8066,
      "step": 2962
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7076210975646973,
      "learning_rate": 9.152260217408664e-06,
      "loss": 0.1588,
      "step": 2963
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.62447738647461,
      "learning_rate": 9.151613147617878e-06,
      "loss": 1.0838,
      "step": 2964
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.191338300704956,
      "learning_rate": 9.150965853863275e-06,
      "loss": 0.1786,
      "step": 2965
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2807443141937256,
      "learning_rate": 9.150318336179778e-06,
      "loss": 0.0929,
      "step": 2966
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.2355527877807617,
      "learning_rate": 9.149670594602315e-06,
      "loss": 0.1865,
      "step": 2967
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.216065406799316,
      "learning_rate": 9.149022629165831e-06,
      "loss": 1.0748,
      "step": 2968
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.088498592376709,
      "learning_rate": 9.148374439905281e-06,
      "loss": 0.0821,
      "step": 2969
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.4840047359466553,
      "learning_rate": 9.147726026855635e-06,
      "loss": 0.2185,
      "step": 2970
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.688595771789551,
      "learning_rate": 9.147077390051867e-06,
      "loss": 0.8055,
      "step": 2971
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.7732951641082764,
      "learning_rate": 9.146428529528972e-06,
      "loss": 0.9081,
      "step": 2972
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.240026473999023,
      "learning_rate": 9.145779445321953e-06,
      "loss": 1.4299,
      "step": 2973
    },
    {
      "epoch": 0.21,
      "grad_norm": 21.887062072753906,
      "learning_rate": 9.145130137465826e-06,
      "loss": 0.6215,
      "step": 2974
    },
    {
      "epoch": 0.21,
      "grad_norm": 17.89809226989746,
      "learning_rate": 9.144480605995618e-06,
      "loss": 1.2036,
      "step": 2975
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.524801254272461,
      "learning_rate": 9.143830850946369e-06,
      "loss": 1.2636,
      "step": 2976
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.140510559082031,
      "learning_rate": 9.143180872353131e-06,
      "loss": 0.9244,
      "step": 2977
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.133908748626709,
      "learning_rate": 9.142530670250967e-06,
      "loss": 0.2308,
      "step": 2978
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.610553741455078,
      "learning_rate": 9.141880244674954e-06,
      "loss": 0.9473,
      "step": 2979
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.715606689453125,
      "learning_rate": 9.141229595660181e-06,
      "loss": 0.7871,
      "step": 2980
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.891142845153809,
      "learning_rate": 9.140578723241745e-06,
      "loss": 0.932,
      "step": 2981
    },
    {
      "epoch": 0.21,
      "grad_norm": 21.221067428588867,
      "learning_rate": 9.139927627454761e-06,
      "loss": 1.2776,
      "step": 2982
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.274014472961426,
      "learning_rate": 9.139276308334352e-06,
      "loss": 0.2431,
      "step": 2983
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.1506428718566895,
      "learning_rate": 9.138624765915653e-06,
      "loss": 0.1424,
      "step": 2984
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.2600510120391846,
      "learning_rate": 9.137973000233814e-06,
      "loss": 0.213,
      "step": 2985
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.653330326080322,
      "learning_rate": 9.137321011323994e-06,
      "loss": 0.7715,
      "step": 2986
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.5357666015625,
      "learning_rate": 9.136668799221367e-06,
      "loss": 0.4861,
      "step": 2987
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8340630531311035,
      "learning_rate": 9.136016363961117e-06,
      "loss": 0.2416,
      "step": 2988
    },
    {
      "epoch": 0.21,
      "grad_norm": 33.09947967529297,
      "learning_rate": 9.13536370557844e-06,
      "loss": 1.5143,
      "step": 2989
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.0846033096313477,
      "learning_rate": 9.13471082410854e-06,
      "loss": 0.1361,
      "step": 2990
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.782886981964111,
      "learning_rate": 9.134057719586647e-06,
      "loss": 0.8187,
      "step": 2991
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.637687683105469,
      "learning_rate": 9.133404392047986e-06,
      "loss": 0.8779,
      "step": 2992
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.401811361312866,
      "learning_rate": 9.132750841527804e-06,
      "loss": 0.0821,
      "step": 2993
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.62626838684082,
      "learning_rate": 9.13209706806136e-06,
      "loss": 1.1673,
      "step": 2994
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.871782302856445,
      "learning_rate": 9.131443071683917e-06,
      "loss": 1.1846,
      "step": 2995
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.6333489418029785,
      "learning_rate": 9.130788852430759e-06,
      "loss": 0.187,
      "step": 2996
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.768144607543945,
      "learning_rate": 9.130134410337179e-06,
      "loss": 0.8887,
      "step": 2997
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.0710012912750244,
      "learning_rate": 9.12947974543848e-06,
      "loss": 0.1745,
      "step": 2998
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.0495717525482178,
      "learning_rate": 9.12882485776998e-06,
      "loss": 0.2224,
      "step": 2999
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.37161922454834,
      "learning_rate": 9.128169747367007e-06,
      "loss": 0.7878,
      "step": 3000
    },
    {
      "epoch": 0.22,
      "grad_norm": 36.568519592285156,
      "learning_rate": 9.127514414264902e-06,
      "loss": 1.2649,
      "step": 3001
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.5228402614593506,
      "learning_rate": 9.12685885849902e-06,
      "loss": 0.3006,
      "step": 3002
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.033921241760254,
      "learning_rate": 9.126203080104721e-06,
      "loss": 1.3298,
      "step": 3003
    },
    {
      "epoch": 0.22,
      "grad_norm": 16.63180160522461,
      "learning_rate": 9.125547079117386e-06,
      "loss": 1.0709,
      "step": 3004
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.586618900299072,
      "learning_rate": 9.124890855572402e-06,
      "loss": 1.0175,
      "step": 3005
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.263263702392578,
      "learning_rate": 9.12423440950517e-06,
      "loss": 1.2016,
      "step": 3006
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.702919006347656,
      "learning_rate": 9.123577740951102e-06,
      "loss": 0.7563,
      "step": 3007
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.9913668632507324,
      "learning_rate": 9.122920849945623e-06,
      "loss": 0.1943,
      "step": 3008
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.525822639465332,
      "learning_rate": 9.12226373652417e-06,
      "loss": 1.1088,
      "step": 3009
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.51592254638672,
      "learning_rate": 9.121606400722194e-06,
      "loss": 1.4062,
      "step": 3010
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.681846618652344,
      "learning_rate": 9.120948842575152e-06,
      "loss": 1.06,
      "step": 3011
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.310091495513916,
      "learning_rate": 9.12029106211852e-06,
      "loss": 0.3186,
      "step": 3012
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.26157283782959,
      "learning_rate": 9.11963305938778e-06,
      "loss": 0.1129,
      "step": 3013
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.050293922424316,
      "learning_rate": 9.118974834418432e-06,
      "loss": 1.1211,
      "step": 3014
    },
    {
      "epoch": 0.22,
      "grad_norm": 20.758136749267578,
      "learning_rate": 9.11831638724598e-06,
      "loss": 1.4425,
      "step": 3015
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.693673133850098,
      "learning_rate": 9.11765771790595e-06,
      "loss": 1.1376,
      "step": 3016
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.3200416564941406,
      "learning_rate": 9.116998826433872e-06,
      "loss": 0.1809,
      "step": 3017
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.237931489944458,
      "learning_rate": 9.11633971286529e-06,
      "loss": 0.3032,
      "step": 3018
    },
    {
      "epoch": 0.22,
      "grad_norm": 41.69808578491211,
      "learning_rate": 9.115680377235762e-06,
      "loss": 1.4016,
      "step": 3019
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.490622043609619,
      "learning_rate": 9.115020819580858e-06,
      "loss": 0.2913,
      "step": 3020
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.086595058441162,
      "learning_rate": 9.114361039936155e-06,
      "loss": 0.9282,
      "step": 3021
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.307992935180664,
      "learning_rate": 9.113701038337248e-06,
      "loss": 0.2058,
      "step": 3022
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.0142645835876465,
      "learning_rate": 9.11304081481974e-06,
      "loss": 0.1741,
      "step": 3023
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.022430419921875,
      "learning_rate": 9.112380369419252e-06,
      "loss": 0.1982,
      "step": 3024
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.138863563537598,
      "learning_rate": 9.111719702171407e-06,
      "loss": 0.1451,
      "step": 3025
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.4633569717407227,
      "learning_rate": 9.111058813111848e-06,
      "loss": 0.1322,
      "step": 3026
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9839859008789062,
      "learning_rate": 9.110397702276226e-06,
      "loss": 0.139,
      "step": 3027
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.602115631103516,
      "learning_rate": 9.109736369700209e-06,
      "loss": 1.0496,
      "step": 3028
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.247779369354248,
      "learning_rate": 9.10907481541947e-06,
      "loss": 0.522,
      "step": 3029
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8161990642547607,
      "learning_rate": 9.108413039469699e-06,
      "loss": 0.124,
      "step": 3030
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.094574928283691,
      "learning_rate": 9.107751041886594e-06,
      "loss": 1.0002,
      "step": 3031
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.2911808490753174,
      "learning_rate": 9.107088822705871e-06,
      "loss": 0.1779,
      "step": 3032
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.006009101867676,
      "learning_rate": 9.106426381963252e-06,
      "loss": 0.873,
      "step": 3033
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.3451452255249023,
      "learning_rate": 9.105763719694472e-06,
      "loss": 0.1856,
      "step": 3034
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.9296765327453613,
      "learning_rate": 9.105100835935281e-06,
      "loss": 0.2224,
      "step": 3035
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.901412010192871,
      "learning_rate": 9.104437730721439e-06,
      "loss": 0.8968,
      "step": 3036
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.144843339920044,
      "learning_rate": 9.103774404088717e-06,
      "loss": 0.1449,
      "step": 3037
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.969000816345215,
      "learning_rate": 9.1031108560729e-06,
      "loss": 0.9579,
      "step": 3038
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.659397602081299,
      "learning_rate": 9.102447086709783e-06,
      "loss": 1.2686,
      "step": 3039
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.632997512817383,
      "learning_rate": 9.101783096035174e-06,
      "loss": 0.1433,
      "step": 3040
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.5605878829956055,
      "learning_rate": 9.101118884084893e-06,
      "loss": 0.9707,
      "step": 3041
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.44025993347168,
      "learning_rate": 9.100454450894774e-06,
      "loss": 0.6445,
      "step": 3042
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.597338676452637,
      "learning_rate": 9.099789796500656e-06,
      "loss": 0.8875,
      "step": 3043
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.013192176818848,
      "learning_rate": 9.099124920938398e-06,
      "loss": 1.3988,
      "step": 3044
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.47758960723877,
      "learning_rate": 9.098459824243865e-06,
      "loss": 0.8849,
      "step": 3045
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.33447527885437,
      "learning_rate": 9.09779450645294e-06,
      "loss": 0.1234,
      "step": 3046
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.4183311462402344,
      "learning_rate": 9.097128967601512e-06,
      "loss": 0.2073,
      "step": 3047
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.82105541229248,
      "learning_rate": 9.096463207725485e-06,
      "loss": 0.8126,
      "step": 3048
    },
    {
      "epoch": 0.22,
      "grad_norm": 31.355627059936523,
      "learning_rate": 9.095797226860774e-06,
      "loss": 1.1351,
      "step": 3049
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.133194446563721,
      "learning_rate": 9.095131025043305e-06,
      "loss": 0.2722,
      "step": 3050
    },
    {
      "epoch": 0.22,
      "grad_norm": 24.50668716430664,
      "learning_rate": 9.094464602309017e-06,
      "loss": 1.6535,
      "step": 3051
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.006777763366699,
      "learning_rate": 9.093797958693864e-06,
      "loss": 0.8414,
      "step": 3052
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.799127578735352,
      "learning_rate": 9.093131094233807e-06,
      "loss": 0.986,
      "step": 3053
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.969933271408081,
      "learning_rate": 9.09246400896482e-06,
      "loss": 0.3653,
      "step": 3054
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.203428268432617,
      "learning_rate": 9.09179670292289e-06,
      "loss": 0.6702,
      "step": 3055
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.40153980255127,
      "learning_rate": 9.091129176144018e-06,
      "loss": 1.1359,
      "step": 3056
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.3237247467041016,
      "learning_rate": 9.090461428664212e-06,
      "loss": 0.1561,
      "step": 3057
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.823590278625488,
      "learning_rate": 9.089793460519496e-06,
      "loss": 1.0782,
      "step": 3058
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.80734920501709,
      "learning_rate": 9.0891252717459e-06,
      "loss": 0.8897,
      "step": 3059
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.402946949005127,
      "learning_rate": 9.088456862379477e-06,
      "loss": 0.2078,
      "step": 3060
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.3073010444641113,
      "learning_rate": 9.087788232456282e-06,
      "loss": 0.1549,
      "step": 3061
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8572869300842285,
      "learning_rate": 9.087119382012384e-06,
      "loss": 0.1487,
      "step": 3062
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.591032981872559,
      "learning_rate": 9.086450311083865e-06,
      "loss": 1.062,
      "step": 3063
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.7318506240844727,
      "learning_rate": 9.085781019706821e-06,
      "loss": 0.1391,
      "step": 3064
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8321664333343506,
      "learning_rate": 9.085111507917356e-06,
      "loss": 0.1548,
      "step": 3065
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.548561096191406,
      "learning_rate": 9.084441775751589e-06,
      "loss": 1.1827,
      "step": 3066
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.077216148376465,
      "learning_rate": 9.083771823245648e-06,
      "loss": 0.1788,
      "step": 3067
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.88909912109375,
      "learning_rate": 9.083101650435676e-06,
      "loss": 0.111,
      "step": 3068
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.788105964660645,
      "learning_rate": 9.082431257357825e-06,
      "loss": 1.1934,
      "step": 3069
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.1247899532318115,
      "learning_rate": 9.081760644048258e-06,
      "loss": 0.2147,
      "step": 3070
    },
    {
      "epoch": 0.22,
      "grad_norm": 19.075702667236328,
      "learning_rate": 9.081089810543158e-06,
      "loss": 1.2771,
      "step": 3071
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9111804962158203,
      "learning_rate": 9.080418756878709e-06,
      "loss": 0.1655,
      "step": 3072
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.179722309112549,
      "learning_rate": 9.079747483091112e-06,
      "loss": 0.1974,
      "step": 3073
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.0968523025512695,
      "learning_rate": 9.079075989216583e-06,
      "loss": 0.9147,
      "step": 3074
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.9456787109375,
      "learning_rate": 9.078404275291344e-06,
      "loss": 1.2186,
      "step": 3075
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.74497389793396,
      "learning_rate": 9.077732341351632e-06,
      "loss": 0.2849,
      "step": 3076
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.786133766174316,
      "learning_rate": 9.077060187433696e-06,
      "loss": 1.0972,
      "step": 3077
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.051285743713379,
      "learning_rate": 9.076387813573793e-06,
      "loss": 0.9763,
      "step": 3078
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.094763278961182,
      "learning_rate": 9.075715219808199e-06,
      "loss": 0.1241,
      "step": 3079
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.2002675533294678,
      "learning_rate": 9.075042406173196e-06,
      "loss": 0.1193,
      "step": 3080
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.496564865112305,
      "learning_rate": 9.07436937270508e-06,
      "loss": 1.0655,
      "step": 3081
    },
    {
      "epoch": 0.22,
      "grad_norm": 27.09364891052246,
      "learning_rate": 9.07369611944016e-06,
      "loss": 1.3015,
      "step": 3082
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.638641357421875,
      "learning_rate": 9.073022646414753e-06,
      "loss": 0.7214,
      "step": 3083
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.85014533996582,
      "learning_rate": 9.07234895366519e-06,
      "loss": 1.2871,
      "step": 3084
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.8004729747772217,
      "learning_rate": 9.071675041227819e-06,
      "loss": 0.2547,
      "step": 3085
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.054144382476807,
      "learning_rate": 9.071000909138989e-06,
      "loss": 0.7041,
      "step": 3086
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.903865337371826,
      "learning_rate": 9.070326557435069e-06,
      "loss": 0.1315,
      "step": 3087
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.7418975830078125,
      "learning_rate": 9.069651986152442e-06,
      "loss": 0.9793,
      "step": 3088
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.1808319091796875,
      "learning_rate": 9.068977195327491e-06,
      "loss": 0.1416,
      "step": 3089
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.324289321899414,
      "learning_rate": 9.068302184996624e-06,
      "loss": 0.9999,
      "step": 3090
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.64450740814209,
      "learning_rate": 9.067626955196253e-06,
      "loss": 0.9301,
      "step": 3091
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.268097400665283,
      "learning_rate": 9.066951505962804e-06,
      "loss": 0.1644,
      "step": 3092
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.509668350219727,
      "learning_rate": 9.066275837332716e-06,
      "loss": 1.52,
      "step": 3093
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.6470046043395996,
      "learning_rate": 9.065599949342439e-06,
      "loss": 0.1823,
      "step": 3094
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.8318493366241455,
      "learning_rate": 9.064923842028432e-06,
      "loss": 0.2026,
      "step": 3095
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.523395299911499,
      "learning_rate": 9.064247515427172e-06,
      "loss": 0.1002,
      "step": 3096
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.878536224365234,
      "learning_rate": 9.063570969575142e-06,
      "loss": 0.1256,
      "step": 3097
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.839153528213501,
      "learning_rate": 9.06289420450884e-06,
      "loss": 0.1683,
      "step": 3098
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.21083927154541,
      "learning_rate": 9.062217220264774e-06,
      "loss": 0.9335,
      "step": 3099
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.5107014179229736,
      "learning_rate": 9.061540016879467e-06,
      "loss": 0.2012,
      "step": 3100
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.749850034713745,
      "learning_rate": 9.060862594389448e-06,
      "loss": 0.1091,
      "step": 3101
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.9139556884765625,
      "learning_rate": 9.060184952831263e-06,
      "loss": 0.8503,
      "step": 3102
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.257131576538086,
      "learning_rate": 9.059507092241469e-06,
      "loss": 0.8293,
      "step": 3103
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.980654239654541,
      "learning_rate": 9.058829012656634e-06,
      "loss": 0.1028,
      "step": 3104
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.4107253551483154,
      "learning_rate": 9.058150714113336e-06,
      "loss": 0.2243,
      "step": 3105
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.04246711730957,
      "learning_rate": 9.05747219664817e-06,
      "loss": 0.8334,
      "step": 3106
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9005308151245117,
      "learning_rate": 9.056793460297734e-06,
      "loss": 0.1027,
      "step": 3107
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.387017726898193,
      "learning_rate": 9.05611450509865e-06,
      "loss": 0.9415,
      "step": 3108
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.269723415374756,
      "learning_rate": 9.055435331087541e-06,
      "loss": 0.1894,
      "step": 3109
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.407599687576294,
      "learning_rate": 9.054755938301046e-06,
      "loss": 0.1792,
      "step": 3110
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.952792167663574,
      "learning_rate": 9.054076326775815e-06,
      "loss": 0.8515,
      "step": 3111
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8527278900146484,
      "learning_rate": 9.053396496548514e-06,
      "loss": 0.1312,
      "step": 3112
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.615025520324707,
      "learning_rate": 9.052716447655814e-06,
      "loss": 0.9221,
      "step": 3113
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.831181049346924,
      "learning_rate": 9.0520361801344e-06,
      "loss": 0.8509,
      "step": 3114
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.262648582458496,
      "learning_rate": 9.051355694020975e-06,
      "loss": 0.1362,
      "step": 3115
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.014747619628906,
      "learning_rate": 9.050674989352243e-06,
      "loss": 1.1174,
      "step": 3116
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3486480712890625,
      "learning_rate": 9.04999406616493e-06,
      "loss": 0.129,
      "step": 3117
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.234870433807373,
      "learning_rate": 9.049312924495765e-06,
      "loss": 0.176,
      "step": 3118
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.51844596862793,
      "learning_rate": 9.048631564381498e-06,
      "loss": 0.9714,
      "step": 3119
    },
    {
      "epoch": 0.22,
      "grad_norm": 16.632713317871094,
      "learning_rate": 9.047949985858882e-06,
      "loss": 1.478,
      "step": 3120
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.62026596069336,
      "learning_rate": 9.047268188964687e-06,
      "loss": 1.1854,
      "step": 3121
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.19714641571045,
      "learning_rate": 9.046586173735691e-06,
      "loss": 1.3777,
      "step": 3122
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.678762435913086,
      "learning_rate": 9.04590394020869e-06,
      "loss": 0.207,
      "step": 3123
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.648825168609619,
      "learning_rate": 9.045221488420484e-06,
      "loss": 0.3024,
      "step": 3124
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.813904762268066,
      "learning_rate": 9.04453881840789e-06,
      "loss": 0.8489,
      "step": 3125
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.724608421325684,
      "learning_rate": 9.043855930207739e-06,
      "loss": 0.7637,
      "step": 3126
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.415851593017578,
      "learning_rate": 9.043172823856867e-06,
      "loss": 0.1087,
      "step": 3127
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.180417060852051,
      "learning_rate": 9.042489499392125e-06,
      "loss": 0.2276,
      "step": 3128
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.217188835144043,
      "learning_rate": 9.041805956850377e-06,
      "loss": 0.1771,
      "step": 3129
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.306659698486328,
      "learning_rate": 9.041122196268495e-06,
      "loss": 1.2696,
      "step": 3130
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.039468765258789,
      "learning_rate": 9.040438217683367e-06,
      "loss": 0.1473,
      "step": 3131
    },
    {
      "epoch": 0.22,
      "grad_norm": 19.6474552154541,
      "learning_rate": 9.039754021131892e-06,
      "loss": 0.6778,
      "step": 3132
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.7317922115325928,
      "learning_rate": 9.039069606650978e-06,
      "loss": 0.1933,
      "step": 3133
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.54013991355896,
      "learning_rate": 9.038384974277548e-06,
      "loss": 0.1355,
      "step": 3134
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.289541244506836,
      "learning_rate": 9.037700124048534e-06,
      "loss": 0.2639,
      "step": 3135
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.831461906433105,
      "learning_rate": 9.037015056000885e-06,
      "loss": 1.0589,
      "step": 3136
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.1410508155822754,
      "learning_rate": 9.03632977017155e-06,
      "loss": 0.2029,
      "step": 3137
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.846431255340576,
      "learning_rate": 9.035644266597505e-06,
      "loss": 0.1593,
      "step": 3138
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.480672836303711,
      "learning_rate": 9.034958545315728e-06,
      "loss": 0.1483,
      "step": 3139
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.0368220806121826,
      "learning_rate": 9.034272606363209e-06,
      "loss": 0.0986,
      "step": 3140
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.30552339553833,
      "learning_rate": 9.033586449776954e-06,
      "loss": 0.8921,
      "step": 3141
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.437187671661377,
      "learning_rate": 9.032900075593978e-06,
      "loss": 0.3104,
      "step": 3142
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8151943683624268,
      "learning_rate": 9.032213483851308e-06,
      "loss": 0.2458,
      "step": 3143
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.387480735778809,
      "learning_rate": 9.031526674585985e-06,
      "loss": 0.6834,
      "step": 3144
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5064728260040283,
      "learning_rate": 9.030839647835055e-06,
      "loss": 0.1165,
      "step": 3145
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.495934963226318,
      "learning_rate": 9.030152403635586e-06,
      "loss": 0.7645,
      "step": 3146
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.310022354125977,
      "learning_rate": 9.029464942024648e-06,
      "loss": 1.4252,
      "step": 3147
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.218351364135742,
      "learning_rate": 9.02877726303933e-06,
      "loss": 0.8418,
      "step": 3148
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.000790596008301,
      "learning_rate": 9.028089366716729e-06,
      "loss": 0.0741,
      "step": 3149
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.29085922241211,
      "learning_rate": 9.027401253093953e-06,
      "loss": 0.8919,
      "step": 3150
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.1604363918304443,
      "learning_rate": 9.026712922208123e-06,
      "loss": 0.2117,
      "step": 3151
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.5830862522125244,
      "learning_rate": 9.026024374096374e-06,
      "loss": 0.3046,
      "step": 3152
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.2561872005462646,
      "learning_rate": 9.02533560879585e-06,
      "loss": 0.2024,
      "step": 3153
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.650157928466797,
      "learning_rate": 9.024646626343704e-06,
      "loss": 1.319,
      "step": 3154
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.944436073303223,
      "learning_rate": 9.02395742677711e-06,
      "loss": 0.8129,
      "step": 3155
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.3167150020599365,
      "learning_rate": 9.023268010133241e-06,
      "loss": 0.2666,
      "step": 3156
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9496009349823,
      "learning_rate": 9.022578376449294e-06,
      "loss": 0.1725,
      "step": 3157
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.65777015686035,
      "learning_rate": 9.02188852576247e-06,
      "loss": 0.9265,
      "step": 3158
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9874863624572754,
      "learning_rate": 9.021198458109982e-06,
      "loss": 0.1524,
      "step": 3159
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.035560131072998,
      "learning_rate": 9.02050817352906e-06,
      "loss": 0.3101,
      "step": 3160
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.498598575592041,
      "learning_rate": 9.01981767205694e-06,
      "loss": 0.2681,
      "step": 3161
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.356119155883789,
      "learning_rate": 9.019126953730873e-06,
      "loss": 0.95,
      "step": 3162
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.1238667964935303,
      "learning_rate": 9.018436018588118e-06,
      "loss": 0.1529,
      "step": 3163
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.126539707183838,
      "learning_rate": 9.017744866665952e-06,
      "loss": 0.8113,
      "step": 3164
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3900442123413086,
      "learning_rate": 9.017053498001658e-06,
      "loss": 0.0954,
      "step": 3165
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9706249237060547,
      "learning_rate": 9.016361912632533e-06,
      "loss": 0.2215,
      "step": 3166
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8176791667938232,
      "learning_rate": 9.015670110595886e-06,
      "loss": 0.1234,
      "step": 3167
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9638609886169434,
      "learning_rate": 9.014978091929037e-06,
      "loss": 0.1406,
      "step": 3168
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4252030849456787,
      "learning_rate": 9.014285856669316e-06,
      "loss": 0.105,
      "step": 3169
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.437047243118286,
      "learning_rate": 9.013593404854068e-06,
      "loss": 0.1792,
      "step": 3170
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.04850435256958,
      "learning_rate": 9.012900736520648e-06,
      "loss": 0.6687,
      "step": 3171
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.587364196777344,
      "learning_rate": 9.012207851706425e-06,
      "loss": 1.7385,
      "step": 3172
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8735525608062744,
      "learning_rate": 9.011514750448775e-06,
      "loss": 0.1557,
      "step": 3173
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.4314029216766357,
      "learning_rate": 9.010821432785087e-06,
      "loss": 0.2282,
      "step": 3174
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.648726463317871,
      "learning_rate": 9.010127898752765e-06,
      "loss": 1.2994,
      "step": 3175
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.451676368713379,
      "learning_rate": 9.009434148389224e-06,
      "loss": 0.6655,
      "step": 3176
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.049045085906982,
      "learning_rate": 9.008740181731885e-06,
      "loss": 0.4345,
      "step": 3177
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.960183143615723,
      "learning_rate": 9.008045998818187e-06,
      "loss": 0.5361,
      "step": 3178
    },
    {
      "epoch": 0.23,
      "grad_norm": 17.129701614379883,
      "learning_rate": 9.007351599685581e-06,
      "loss": 0.9273,
      "step": 3179
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.285346269607544,
      "learning_rate": 9.006656984371523e-06,
      "loss": 0.2225,
      "step": 3180
    },
    {
      "epoch": 0.23,
      "grad_norm": 25.517019271850586,
      "learning_rate": 9.005962152913485e-06,
      "loss": 0.8881,
      "step": 3181
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6511521339416504,
      "learning_rate": 9.005267105348957e-06,
      "loss": 0.1324,
      "step": 3182
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.293754577636719,
      "learning_rate": 9.004571841715424e-06,
      "loss": 0.9322,
      "step": 3183
    },
    {
      "epoch": 0.23,
      "grad_norm": 34.981544494628906,
      "learning_rate": 9.003876362050401e-06,
      "loss": 0.961,
      "step": 3184
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.236231803894043,
      "learning_rate": 9.003180666391405e-06,
      "loss": 0.1282,
      "step": 3185
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.9091625213623047,
      "learning_rate": 9.002484754775964e-06,
      "loss": 0.1565,
      "step": 3186
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.2846531867980957,
      "learning_rate": 9.001788627241619e-06,
      "loss": 0.1266,
      "step": 3187
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.997334957122803,
      "learning_rate": 9.001092283825927e-06,
      "loss": 1.2494,
      "step": 3188
    },
    {
      "epoch": 0.23,
      "grad_norm": 23.496726989746094,
      "learning_rate": 9.000395724566451e-06,
      "loss": 1.1702,
      "step": 3189
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.601071834564209,
      "learning_rate": 8.999698949500767e-06,
      "loss": 0.2614,
      "step": 3190
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.4521565437316895,
      "learning_rate": 8.999001958666464e-06,
      "loss": 0.8454,
      "step": 3191
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.207900047302246,
      "learning_rate": 8.998304752101143e-06,
      "loss": 0.1081,
      "step": 3192
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.358241081237793,
      "learning_rate": 8.997607329842416e-06,
      "loss": 0.6068,
      "step": 3193
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.06884765625,
      "learning_rate": 8.996909691927906e-06,
      "loss": 0.2986,
      "step": 3194
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.2862119674682617,
      "learning_rate": 8.996211838395246e-06,
      "loss": 0.239,
      "step": 3195
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.186507701873779,
      "learning_rate": 8.995513769282083e-06,
      "loss": 0.1503,
      "step": 3196
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.5737738609313965,
      "learning_rate": 8.994815484626076e-06,
      "loss": 0.2571,
      "step": 3197
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.1225714683532715,
      "learning_rate": 8.994116984464896e-06,
      "loss": 0.9844,
      "step": 3198
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.113941192626953,
      "learning_rate": 8.993418268836223e-06,
      "loss": 0.9514,
      "step": 3199
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.492166519165039,
      "learning_rate": 8.99271933777775e-06,
      "loss": 0.6548,
      "step": 3200
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.3255980014801025,
      "learning_rate": 8.992020191327181e-06,
      "loss": 0.1264,
      "step": 3201
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.0203065872192383,
      "learning_rate": 8.991320829522235e-06,
      "loss": 0.1903,
      "step": 3202
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.615459442138672,
      "learning_rate": 8.990621252400637e-06,
      "loss": 0.8701,
      "step": 3203
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.4319262504577637,
      "learning_rate": 8.989921460000128e-06,
      "loss": 0.2622,
      "step": 3204
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.47119903564453,
      "learning_rate": 8.989221452358462e-06,
      "loss": 1.8502,
      "step": 3205
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.392512798309326,
      "learning_rate": 8.988521229513394e-06,
      "loss": 0.1863,
      "step": 3206
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.126009702682495,
      "learning_rate": 8.987820791502704e-06,
      "loss": 0.141,
      "step": 3207
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.8892240524292,
      "learning_rate": 8.98712013836418e-06,
      "loss": 0.9483,
      "step": 3208
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.137482643127441,
      "learning_rate": 8.986419270135613e-06,
      "loss": 0.9741,
      "step": 3209
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.027373790740967,
      "learning_rate": 8.985718186854818e-06,
      "loss": 0.0697,
      "step": 3210
    },
    {
      "epoch": 0.23,
      "grad_norm": 25.650218963623047,
      "learning_rate": 8.985016888559612e-06,
      "loss": 1.3439,
      "step": 3211
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.872220993041992,
      "learning_rate": 8.984315375287831e-06,
      "loss": 1.2069,
      "step": 3212
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.099249839782715,
      "learning_rate": 8.983613647077315e-06,
      "loss": 0.9488,
      "step": 3213
    },
    {
      "epoch": 0.23,
      "grad_norm": 28.977752685546875,
      "learning_rate": 8.982911703965923e-06,
      "loss": 1.5939,
      "step": 3214
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.97616195678711,
      "learning_rate": 8.98220954599152e-06,
      "loss": 1.1848,
      "step": 3215
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.6676924228668213,
      "learning_rate": 8.981507173191986e-06,
      "loss": 0.2497,
      "step": 3216
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.642974853515625,
      "learning_rate": 8.980804585605211e-06,
      "loss": 0.9746,
      "step": 3217
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.597650051116943,
      "learning_rate": 8.980101783269095e-06,
      "loss": 0.9494,
      "step": 3218
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6707465648651123,
      "learning_rate": 8.979398766221556e-06,
      "loss": 0.1799,
      "step": 3219
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.41928482055664,
      "learning_rate": 8.978695534500515e-06,
      "loss": 0.8561,
      "step": 3220
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.005169868469238,
      "learning_rate": 8.97799208814391e-06,
      "loss": 1.3837,
      "step": 3221
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.714120388031006,
      "learning_rate": 8.977288427189688e-06,
      "loss": 0.1218,
      "step": 3222
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.44294548034668,
      "learning_rate": 8.976584551675812e-06,
      "loss": 1.6955,
      "step": 3223
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8020174503326416,
      "learning_rate": 8.975880461640251e-06,
      "loss": 0.1303,
      "step": 3224
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.866995096206665,
      "learning_rate": 8.975176157120988e-06,
      "loss": 0.1285,
      "step": 3225
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.944664001464844,
      "learning_rate": 8.97447163815602e-06,
      "loss": 1.0642,
      "step": 3226
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.643624305725098,
      "learning_rate": 8.97376690478335e-06,
      "loss": 0.8289,
      "step": 3227
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.1046218872070312,
      "learning_rate": 8.973061957040998e-06,
      "loss": 0.2121,
      "step": 3228
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.377106189727783,
      "learning_rate": 8.972356794966991e-06,
      "loss": 0.88,
      "step": 3229
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.992248773574829,
      "learning_rate": 8.971651418599371e-06,
      "loss": 0.1969,
      "step": 3230
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.513615608215332,
      "learning_rate": 8.970945827976192e-06,
      "loss": 1.0672,
      "step": 3231
    },
    {
      "epoch": 0.23,
      "grad_norm": 23.48374366760254,
      "learning_rate": 8.970240023135514e-06,
      "loss": 1.0462,
      "step": 3232
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8368115425109863,
      "learning_rate": 8.969534004115419e-06,
      "loss": 0.1615,
      "step": 3233
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.838066101074219,
      "learning_rate": 8.968827770953986e-06,
      "loss": 1.0585,
      "step": 3234
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.063634395599365,
      "learning_rate": 8.96812132368932e-06,
      "loss": 0.1909,
      "step": 3235
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8870933055877686,
      "learning_rate": 8.967414662359529e-06,
      "loss": 0.1261,
      "step": 3236
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9642443656921387,
      "learning_rate": 8.966707787002732e-06,
      "loss": 0.1504,
      "step": 3237
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.965714931488037,
      "learning_rate": 8.966000697657064e-06,
      "loss": 1.1191,
      "step": 3238
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.018167495727539,
      "learning_rate": 8.965293394360674e-06,
      "loss": 1.2763,
      "step": 3239
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.99779224395752,
      "learning_rate": 8.964585877151714e-06,
      "loss": 0.9859,
      "step": 3240
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.66013240814209,
      "learning_rate": 8.96387814606835e-06,
      "loss": 0.8974,
      "step": 3241
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.946852207183838,
      "learning_rate": 8.963170201148767e-06,
      "loss": 0.829,
      "step": 3242
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.111255645751953,
      "learning_rate": 8.962462042431152e-06,
      "loss": 0.9397,
      "step": 3243
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.722329139709473,
      "learning_rate": 8.96175366995371e-06,
      "loss": 1.2477,
      "step": 3244
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.577774047851562,
      "learning_rate": 8.961045083754651e-06,
      "loss": 1.4675,
      "step": 3245
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.688838958740234,
      "learning_rate": 8.960336283872204e-06,
      "loss": 1.101,
      "step": 3246
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.741943836212158,
      "learning_rate": 8.959627270344606e-06,
      "loss": 0.7468,
      "step": 3247
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.268838882446289,
      "learning_rate": 8.958918043210106e-06,
      "loss": 0.8161,
      "step": 3248
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.1102817058563232,
      "learning_rate": 8.95820860250696e-06,
      "loss": 0.1561,
      "step": 3249
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.129760265350342,
      "learning_rate": 8.957498948273444e-06,
      "loss": 0.1531,
      "step": 3250
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.057751178741455,
      "learning_rate": 8.956789080547838e-06,
      "loss": 0.7694,
      "step": 3251
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.754127025604248,
      "learning_rate": 8.95607899936844e-06,
      "loss": 0.8845,
      "step": 3252
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.994779109954834,
      "learning_rate": 8.955368704773555e-06,
      "loss": 0.2215,
      "step": 3253
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.299508094787598,
      "learning_rate": 8.9546581968015e-06,
      "loss": 1.6453,
      "step": 3254
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5729284286499023,
      "learning_rate": 8.953947475490603e-06,
      "loss": 0.0998,
      "step": 3255
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.9614648818969727,
      "learning_rate": 8.953236540879208e-06,
      "loss": 0.266,
      "step": 3256
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.74042272567749,
      "learning_rate": 8.952525393005664e-06,
      "loss": 1.0689,
      "step": 3257
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.7581048011779785,
      "learning_rate": 8.951814031908337e-06,
      "loss": 0.8171,
      "step": 3258
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.5013294219970703,
      "learning_rate": 8.951102457625603e-06,
      "loss": 0.1836,
      "step": 3259
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.434038162231445,
      "learning_rate": 8.950390670195843e-06,
      "loss": 1.5973,
      "step": 3260
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8784525394439697,
      "learning_rate": 8.949678669657462e-06,
      "loss": 0.0979,
      "step": 3261
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.40568733215332,
      "learning_rate": 8.948966456048868e-06,
      "loss": 0.8508,
      "step": 3262
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.056772232055664,
      "learning_rate": 8.94825402940848e-06,
      "loss": 0.9136,
      "step": 3263
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.002987861633301,
      "learning_rate": 8.947541389774732e-06,
      "loss": 0.1649,
      "step": 3264
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.436513900756836,
      "learning_rate": 8.946828537186068e-06,
      "loss": 0.2577,
      "step": 3265
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.303218841552734,
      "learning_rate": 8.946115471680946e-06,
      "loss": 1.1484,
      "step": 3266
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.776313304901123,
      "learning_rate": 8.94540219329783e-06,
      "loss": 0.797,
      "step": 3267
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8077645301818848,
      "learning_rate": 8.9446887020752e-06,
      "loss": 0.1331,
      "step": 3268
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.0135140419006348,
      "learning_rate": 8.943974998051544e-06,
      "loss": 0.1464,
      "step": 3269
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9909729957580566,
      "learning_rate": 8.943261081265367e-06,
      "loss": 0.1228,
      "step": 3270
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9400012493133545,
      "learning_rate": 8.942546951755182e-06,
      "loss": 0.1582,
      "step": 3271
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.083400726318359,
      "learning_rate": 8.94183260955951e-06,
      "loss": 0.742,
      "step": 3272
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.252608299255371,
      "learning_rate": 8.941118054716891e-06,
      "loss": 0.2873,
      "step": 3273
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.7276771068573,
      "learning_rate": 8.94040328726587e-06,
      "loss": 0.1373,
      "step": 3274
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.81938362121582,
      "learning_rate": 8.939688307245009e-06,
      "loss": 1.3526,
      "step": 3275
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.3433735370635986,
      "learning_rate": 8.938973114692874e-06,
      "loss": 0.1819,
      "step": 3276
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.669937133789062,
      "learning_rate": 8.93825770964805e-06,
      "loss": 0.5692,
      "step": 3277
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4128756523132324,
      "learning_rate": 8.93754209214913e-06,
      "loss": 0.2816,
      "step": 3278
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.96707820892334,
      "learning_rate": 8.936826262234717e-06,
      "loss": 0.2539,
      "step": 3279
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.524844646453857,
      "learning_rate": 8.936110219943431e-06,
      "loss": 0.8529,
      "step": 3280
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.897435188293457,
      "learning_rate": 8.935393965313897e-06,
      "loss": 1.2539,
      "step": 3281
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.63888931274414,
      "learning_rate": 8.934677498384753e-06,
      "loss": 1.3287,
      "step": 3282
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.521111488342285,
      "learning_rate": 8.933960819194654e-06,
      "loss": 0.9536,
      "step": 3283
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.538966178894043,
      "learning_rate": 8.933243927782259e-06,
      "loss": 0.5835,
      "step": 3284
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.981033325195312,
      "learning_rate": 8.93252682418624e-06,
      "loss": 0.9993,
      "step": 3285
    },
    {
      "epoch": 0.24,
      "grad_norm": 20.473581314086914,
      "learning_rate": 8.931809508445287e-06,
      "loss": 1.6386,
      "step": 3286
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.349581480026245,
      "learning_rate": 8.931091980598093e-06,
      "loss": 0.2061,
      "step": 3287
    },
    {
      "epoch": 0.24,
      "grad_norm": 15.355157852172852,
      "learning_rate": 8.930374240683366e-06,
      "loss": 1.0218,
      "step": 3288
    },
    {
      "epoch": 0.24,
      "grad_norm": 26.441579818725586,
      "learning_rate": 8.929656288739828e-06,
      "loss": 1.1488,
      "step": 3289
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.0757505893707275,
      "learning_rate": 8.928938124806206e-06,
      "loss": 0.1504,
      "step": 3290
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.482525825500488,
      "learning_rate": 8.928219748921244e-06,
      "loss": 1.5564,
      "step": 3291
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.598480224609375,
      "learning_rate": 8.927501161123695e-06,
      "loss": 0.5775,
      "step": 3292
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6503312587738037,
      "learning_rate": 8.926782361452326e-06,
      "loss": 0.1529,
      "step": 3293
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.390744209289551,
      "learning_rate": 8.926063349945912e-06,
      "loss": 0.1569,
      "step": 3294
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.5903825759887695,
      "learning_rate": 8.925344126643241e-06,
      "loss": 1.0006,
      "step": 3295
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.586823463439941,
      "learning_rate": 8.924624691583114e-06,
      "loss": 1.0813,
      "step": 3296
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.209139823913574,
      "learning_rate": 8.923905044804338e-06,
      "loss": 0.7538,
      "step": 3297
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.646315097808838,
      "learning_rate": 8.92318518634574e-06,
      "loss": 0.8221,
      "step": 3298
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.5413527488708496,
      "learning_rate": 8.92246511624615e-06,
      "loss": 0.2173,
      "step": 3299
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.689218044281006,
      "learning_rate": 8.921744834544415e-06,
      "loss": 0.9813,
      "step": 3300
    },
    {
      "epoch": 0.24,
      "grad_norm": 30.541584014892578,
      "learning_rate": 8.92102434127939e-06,
      "loss": 1.3274,
      "step": 3301
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.76471710205078,
      "learning_rate": 8.920303636489943e-06,
      "loss": 1.1592,
      "step": 3302
    },
    {
      "epoch": 0.24,
      "grad_norm": 21.3282470703125,
      "learning_rate": 8.919582720214955e-06,
      "loss": 1.6606,
      "step": 3303
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.962520599365234,
      "learning_rate": 8.918861592493316e-06,
      "loss": 0.846,
      "step": 3304
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.319941520690918,
      "learning_rate": 8.918140253363927e-06,
      "loss": 1.3046,
      "step": 3305
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.5190813541412354,
      "learning_rate": 8.917418702865701e-06,
      "loss": 0.1869,
      "step": 3306
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.973762512207031,
      "learning_rate": 8.916696941037566e-06,
      "loss": 0.771,
      "step": 3307
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.100841999053955,
      "learning_rate": 8.915974967918457e-06,
      "loss": 0.1521,
      "step": 3308
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.902118682861328,
      "learning_rate": 8.915252783547321e-06,
      "loss": 0.8224,
      "step": 3309
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.587026834487915,
      "learning_rate": 8.914530387963116e-06,
      "loss": 0.1258,
      "step": 3310
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.769362449645996,
      "learning_rate": 8.913807781204816e-06,
      "loss": 0.2736,
      "step": 3311
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.9754464626312256,
      "learning_rate": 8.9130849633114e-06,
      "loss": 0.0926,
      "step": 3312
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1176724433898926,
      "learning_rate": 8.91236193432186e-06,
      "loss": 0.0475,
      "step": 3313
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.8640775680542,
      "learning_rate": 8.911638694275207e-06,
      "loss": 1.3025,
      "step": 3314
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.637084007263184,
      "learning_rate": 8.910915243210448e-06,
      "loss": 0.942,
      "step": 3315
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.509592294692993,
      "learning_rate": 8.910191581166619e-06,
      "loss": 0.2794,
      "step": 3316
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.853187084197998,
      "learning_rate": 8.909467708182753e-06,
      "loss": 0.846,
      "step": 3317
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.682267189025879,
      "learning_rate": 8.908743624297901e-06,
      "loss": 1.0442,
      "step": 3318
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.846597909927368,
      "learning_rate": 8.90801932955113e-06,
      "loss": 0.1228,
      "step": 3319
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.833499908447266,
      "learning_rate": 8.907294823981507e-06,
      "loss": 1.2452,
      "step": 3320
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.287138938903809,
      "learning_rate": 8.906570107628116e-06,
      "loss": 0.9947,
      "step": 3321
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.189695835113525,
      "learning_rate": 8.905845180530057e-06,
      "loss": 1.026,
      "step": 3322
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.578473091125488,
      "learning_rate": 8.905120042726434e-06,
      "loss": 1.0583,
      "step": 3323
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.648106575012207,
      "learning_rate": 8.904394694256367e-06,
      "loss": 0.995,
      "step": 3324
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.607075691223145,
      "learning_rate": 8.903669135158985e-06,
      "loss": 1.6181,
      "step": 3325
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.845690727233887,
      "learning_rate": 8.90294336547343e-06,
      "loss": 1.8189,
      "step": 3326
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.949864387512207,
      "learning_rate": 8.902217385238852e-06,
      "loss": 0.9342,
      "step": 3327
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.98075008392334,
      "learning_rate": 8.901491194494418e-06,
      "loss": 0.1462,
      "step": 3328
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.101104736328125,
      "learning_rate": 8.900764793279302e-06,
      "loss": 0.8352,
      "step": 3329
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.924719333648682,
      "learning_rate": 8.90003818163269e-06,
      "loss": 0.7478,
      "step": 3330
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.266112804412842,
      "learning_rate": 8.899311359593784e-06,
      "loss": 0.1274,
      "step": 3331
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.831724643707275,
      "learning_rate": 8.898584327201785e-06,
      "loss": 1.0065,
      "step": 3332
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.781190872192383,
      "learning_rate": 8.897857084495921e-06,
      "loss": 1.008,
      "step": 3333
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.5508129596710205,
      "learning_rate": 8.897129631515422e-06,
      "loss": 0.2643,
      "step": 3334
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1187009811401367,
      "learning_rate": 8.896401968299528e-06,
      "loss": 0.1012,
      "step": 3335
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.911822319030762,
      "learning_rate": 8.895674094887498e-06,
      "loss": 1.0571,
      "step": 3336
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.4114990234375,
      "learning_rate": 8.894946011318597e-06,
      "loss": 0.9277,
      "step": 3337
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.333850383758545,
      "learning_rate": 8.894217717632104e-06,
      "loss": 0.9683,
      "step": 3338
    },
    {
      "epoch": 0.24,
      "grad_norm": 18.792804718017578,
      "learning_rate": 8.893489213867302e-06,
      "loss": 0.542,
      "step": 3339
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.854748249053955,
      "learning_rate": 8.892760500063495e-06,
      "loss": 0.5786,
      "step": 3340
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.082038402557373,
      "learning_rate": 8.892031576259996e-06,
      "loss": 0.5107,
      "step": 3341
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.418750047683716,
      "learning_rate": 8.891302442496123e-06,
      "loss": 0.1407,
      "step": 3342
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.168185710906982,
      "learning_rate": 8.890573098811213e-06,
      "loss": 0.8186,
      "step": 3343
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.438587188720703,
      "learning_rate": 8.889843545244614e-06,
      "loss": 1.1511,
      "step": 3344
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.702693939208984,
      "learning_rate": 8.889113781835676e-06,
      "loss": 1.1758,
      "step": 3345
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.22663402557373,
      "learning_rate": 8.888383808623772e-06,
      "loss": 0.9755,
      "step": 3346
    },
    {
      "epoch": 0.24,
      "grad_norm": 15.897064208984375,
      "learning_rate": 8.88765362564828e-06,
      "loss": 1.3345,
      "step": 3347
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.56986927986145,
      "learning_rate": 8.88692323294859e-06,
      "loss": 0.1616,
      "step": 3348
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.2015931606292725,
      "learning_rate": 8.886192630564106e-06,
      "loss": 0.113,
      "step": 3349
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.399712562561035,
      "learning_rate": 8.885461818534237e-06,
      "loss": 0.8541,
      "step": 3350
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.0588088035583496,
      "learning_rate": 8.884730796898412e-06,
      "loss": 0.2079,
      "step": 3351
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.031238555908203,
      "learning_rate": 8.883999565696064e-06,
      "loss": 0.2381,
      "step": 3352
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.7943155765533447,
      "learning_rate": 8.883268124966643e-06,
      "loss": 0.0988,
      "step": 3353
    },
    {
      "epoch": 0.24,
      "grad_norm": 24.52372932434082,
      "learning_rate": 8.882536474749604e-06,
      "loss": 1.0618,
      "step": 3354
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.3661580085754395,
      "learning_rate": 8.881804615084418e-06,
      "loss": 0.1718,
      "step": 3355
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.3336124420166,
      "learning_rate": 8.881072546010567e-06,
      "loss": 1.0678,
      "step": 3356
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6390764713287354,
      "learning_rate": 8.880340267567543e-06,
      "loss": 0.1287,
      "step": 3357
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4713568687438965,
      "learning_rate": 8.87960777979485e-06,
      "loss": 0.3311,
      "step": 3358
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.790002822875977,
      "learning_rate": 8.878875082732e-06,
      "loss": 0.7044,
      "step": 3359
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.649567127227783,
      "learning_rate": 8.878142176418525e-06,
      "loss": 0.9439,
      "step": 3360
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.416668176651001,
      "learning_rate": 8.877409060893956e-06,
      "loss": 0.2912,
      "step": 3361
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.2324090003967285,
      "learning_rate": 8.876675736197847e-06,
      "loss": 0.1519,
      "step": 3362
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.88204026222229,
      "learning_rate": 8.875942202369757e-06,
      "loss": 0.289,
      "step": 3363
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.527923107147217,
      "learning_rate": 8.875208459449254e-06,
      "loss": 0.6879,
      "step": 3364
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.2963898181915283,
      "learning_rate": 8.874474507475923e-06,
      "loss": 0.211,
      "step": 3365
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.7444300651550293,
      "learning_rate": 8.87374034648936e-06,
      "loss": 0.3054,
      "step": 3366
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.8280885219573975,
      "learning_rate": 8.873005976529167e-06,
      "loss": 0.1698,
      "step": 3367
    },
    {
      "epoch": 0.24,
      "grad_norm": 15.616776466369629,
      "learning_rate": 8.872271397634962e-06,
      "loss": 0.8137,
      "step": 3368
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.214742660522461,
      "learning_rate": 8.871536609846371e-06,
      "loss": 0.1795,
      "step": 3369
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.130558967590332,
      "learning_rate": 8.870801613203037e-06,
      "loss": 0.1805,
      "step": 3370
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.54507827758789,
      "learning_rate": 8.870066407744607e-06,
      "loss": 1.8124,
      "step": 3371
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.6783249378204346,
      "learning_rate": 8.869330993510743e-06,
      "loss": 0.1964,
      "step": 3372
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.905982971191406,
      "learning_rate": 8.868595370541117e-06,
      "loss": 0.6972,
      "step": 3373
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.553712844848633,
      "learning_rate": 8.867859538875417e-06,
      "loss": 0.7376,
      "step": 3374
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.9267563819885254,
      "learning_rate": 8.867123498553333e-06,
      "loss": 0.2582,
      "step": 3375
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.734492063522339,
      "learning_rate": 8.866387249614573e-06,
      "loss": 0.1689,
      "step": 3376
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.6983091831207275,
      "learning_rate": 8.86565079209886e-06,
      "loss": 0.2435,
      "step": 3377
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.225045204162598,
      "learning_rate": 8.864914126045916e-06,
      "loss": 0.6842,
      "step": 3378
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.091399669647217,
      "learning_rate": 8.864177251495484e-06,
      "loss": 0.1596,
      "step": 3379
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.24845814704895,
      "learning_rate": 8.863440168487318e-06,
      "loss": 0.1913,
      "step": 3380
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.524185180664062,
      "learning_rate": 8.862702877061178e-06,
      "loss": 1.2216,
      "step": 3381
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.365227699279785,
      "learning_rate": 8.86196537725684e-06,
      "loss": 0.9653,
      "step": 3382
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.100797653198242,
      "learning_rate": 8.861227669114085e-06,
      "loss": 0.1004,
      "step": 3383
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.004018783569336,
      "learning_rate": 8.860489752672715e-06,
      "loss": 0.1742,
      "step": 3384
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6085355281829834,
      "learning_rate": 8.859751627972536e-06,
      "loss": 0.0664,
      "step": 3385
    },
    {
      "epoch": 0.24,
      "grad_norm": 18.551429748535156,
      "learning_rate": 8.859013295053366e-06,
      "loss": 1.1507,
      "step": 3386
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.454112529754639,
      "learning_rate": 8.858274753955036e-06,
      "loss": 0.7887,
      "step": 3387
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.645933151245117,
      "learning_rate": 8.857536004717389e-06,
      "loss": 0.1556,
      "step": 3388
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.744903326034546,
      "learning_rate": 8.856797047380275e-06,
      "loss": 0.1474,
      "step": 3389
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.465235233306885,
      "learning_rate": 8.856057881983557e-06,
      "loss": 0.8526,
      "step": 3390
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.671526908874512,
      "learning_rate": 8.855318508567115e-06,
      "loss": 1.024,
      "step": 3391
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4120571613311768,
      "learning_rate": 8.854578927170832e-06,
      "loss": 0.1167,
      "step": 3392
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.7141048908233643,
      "learning_rate": 8.853839137834605e-06,
      "loss": 0.1537,
      "step": 3393
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.037478446960449,
      "learning_rate": 8.853099140598344e-06,
      "loss": 0.0945,
      "step": 3394
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.9752357006073,
      "learning_rate": 8.85235893550197e-06,
      "loss": 0.1381,
      "step": 3395
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.986065864562988,
      "learning_rate": 8.851618522585414e-06,
      "loss": 0.6645,
      "step": 3396
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.349930763244629,
      "learning_rate": 8.850877901888618e-06,
      "loss": 0.1082,
      "step": 3397
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4459238052368164,
      "learning_rate": 8.850137073451534e-06,
      "loss": 0.1374,
      "step": 3398
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.375278949737549,
      "learning_rate": 8.849396037314128e-06,
      "loss": 0.1308,
      "step": 3399
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.17384147644043,
      "learning_rate": 8.848654793516379e-06,
      "loss": 1.283,
      "step": 3400
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.782709121704102,
      "learning_rate": 8.84791334209827e-06,
      "loss": 1.4376,
      "step": 3401
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.732570171356201,
      "learning_rate": 8.847171683099803e-06,
      "loss": 0.1539,
      "step": 3402
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.8716299533843994,
      "learning_rate": 8.846429816560984e-06,
      "loss": 0.1521,
      "step": 3403
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.016188144683838,
      "learning_rate": 8.845687742521836e-06,
      "loss": 0.2915,
      "step": 3404
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4364044666290283,
      "learning_rate": 8.844945461022393e-06,
      "loss": 0.1576,
      "step": 3405
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.9790127277374268,
      "learning_rate": 8.844202972102696e-06,
      "loss": 0.1841,
      "step": 3406
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.759633779525757,
      "learning_rate": 8.843460275802798e-06,
      "loss": 0.2154,
      "step": 3407
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.246771812438965,
      "learning_rate": 8.842717372162768e-06,
      "loss": 0.2011,
      "step": 3408
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.2166168689727783,
      "learning_rate": 8.84197426122268e-06,
      "loss": 0.2089,
      "step": 3409
    },
    {
      "epoch": 0.24,
      "grad_norm": 22.5726318359375,
      "learning_rate": 8.841230943022624e-06,
      "loss": 1.5992,
      "step": 3410
    },
    {
      "epoch": 0.24,
      "grad_norm": 31.4986515045166,
      "learning_rate": 8.840487417602699e-06,
      "loss": 1.2675,
      "step": 3411
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.244766712188721,
      "learning_rate": 8.839743685003014e-06,
      "loss": 0.815,
      "step": 3412
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.458810329437256,
      "learning_rate": 8.83899974526369e-06,
      "loss": 0.2289,
      "step": 3413
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.000284194946289,
      "learning_rate": 8.838255598424864e-06,
      "loss": 0.1179,
      "step": 3414
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.910381317138672,
      "learning_rate": 8.837511244526675e-06,
      "loss": 0.8174,
      "step": 3415
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.628876686096191,
      "learning_rate": 8.836766683609282e-06,
      "loss": 0.7028,
      "step": 3416
    },
    {
      "epoch": 0.25,
      "grad_norm": 30.427104949951172,
      "learning_rate": 8.836021915712847e-06,
      "loss": 1.3563,
      "step": 3417
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.075287342071533,
      "learning_rate": 8.835276940877551e-06,
      "loss": 0.1856,
      "step": 3418
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.267923355102539,
      "learning_rate": 8.834531759143583e-06,
      "loss": 1.1022,
      "step": 3419
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.071022987365723,
      "learning_rate": 8.833786370551138e-06,
      "loss": 0.2548,
      "step": 3420
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.142563819885254,
      "learning_rate": 8.833040775140432e-06,
      "loss": 0.1713,
      "step": 3421
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.06659984588623,
      "learning_rate": 8.832294972951685e-06,
      "loss": 0.7411,
      "step": 3422
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.1595048904418945,
      "learning_rate": 8.831548964025129e-06,
      "loss": 0.1825,
      "step": 3423
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.381627321243286,
      "learning_rate": 8.830802748401011e-06,
      "loss": 0.0842,
      "step": 3424
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.6836793422698975,
      "learning_rate": 8.830056326119586e-06,
      "loss": 0.1166,
      "step": 3425
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.274350166320801,
      "learning_rate": 8.829309697221116e-06,
      "loss": 0.2127,
      "step": 3426
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.298768997192383,
      "learning_rate": 8.828562861745886e-06,
      "loss": 0.85,
      "step": 3427
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.1861572265625,
      "learning_rate": 8.82781581973418e-06,
      "loss": 0.9442,
      "step": 3428
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4599921703338623,
      "learning_rate": 8.8270685712263e-06,
      "loss": 0.0789,
      "step": 3429
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.997218608856201,
      "learning_rate": 8.826321116262556e-06,
      "loss": 0.1622,
      "step": 3430
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.7858574390411377,
      "learning_rate": 8.825573454883271e-06,
      "loss": 0.1366,
      "step": 3431
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.0429699420928955,
      "learning_rate": 8.82482558712878e-06,
      "loss": 0.2701,
      "step": 3432
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.725153923034668,
      "learning_rate": 8.824077513039426e-06,
      "loss": 0.8457,
      "step": 3433
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.898299217224121,
      "learning_rate": 8.823329232655566e-06,
      "loss": 0.9678,
      "step": 3434
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.7027838230133057,
      "learning_rate": 8.822580746017564e-06,
      "loss": 0.1628,
      "step": 3435
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9878085851669312,
      "learning_rate": 8.821832053165802e-06,
      "loss": 0.1135,
      "step": 3436
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.9536359310150146,
      "learning_rate": 8.821083154140667e-06,
      "loss": 0.1243,
      "step": 3437
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.237447500228882,
      "learning_rate": 8.820334048982557e-06,
      "loss": 0.2295,
      "step": 3438
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.2315194606781006,
      "learning_rate": 8.819584737731888e-06,
      "loss": 0.1441,
      "step": 3439
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.532492637634277,
      "learning_rate": 8.81883522042908e-06,
      "loss": 0.6885,
      "step": 3440
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.270757675170898,
      "learning_rate": 8.818085497114565e-06,
      "loss": 1.3468,
      "step": 3441
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.168239116668701,
      "learning_rate": 8.817335567828792e-06,
      "loss": 0.7978,
      "step": 3442
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.253885507583618,
      "learning_rate": 8.816585432612215e-06,
      "loss": 0.1668,
      "step": 3443
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.8051164150238037,
      "learning_rate": 8.815835091505298e-06,
      "loss": 0.21,
      "step": 3444
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.363850116729736,
      "learning_rate": 8.815084544548523e-06,
      "loss": 0.2491,
      "step": 3445
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.877016067504883,
      "learning_rate": 8.814333791782377e-06,
      "loss": 0.8988,
      "step": 3446
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.26621150970459,
      "learning_rate": 8.813582833247363e-06,
      "loss": 0.868,
      "step": 3447
    },
    {
      "epoch": 0.25,
      "grad_norm": 16.421113967895508,
      "learning_rate": 8.812831668983987e-06,
      "loss": 1.2005,
      "step": 3448
    },
    {
      "epoch": 0.25,
      "grad_norm": 15.433029174804688,
      "learning_rate": 8.812080299032776e-06,
      "loss": 0.7376,
      "step": 3449
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.9309959411621094,
      "learning_rate": 8.811328723434265e-06,
      "loss": 0.9412,
      "step": 3450
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.7095534801483154,
      "learning_rate": 8.810576942228994e-06,
      "loss": 0.8922,
      "step": 3451
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.214864253997803,
      "learning_rate": 8.80982495545752e-06,
      "loss": 0.9946,
      "step": 3452
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.422937393188477,
      "learning_rate": 8.80907276316041e-06,
      "loss": 1.4972,
      "step": 3453
    },
    {
      "epoch": 0.25,
      "grad_norm": 26.182079315185547,
      "learning_rate": 8.808320365378246e-06,
      "loss": 1.9231,
      "step": 3454
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.738308429718018,
      "learning_rate": 8.807567762151611e-06,
      "loss": 0.8557,
      "step": 3455
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.090645790100098,
      "learning_rate": 8.80681495352111e-06,
      "loss": 0.6336,
      "step": 3456
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.379228115081787,
      "learning_rate": 8.806061939527348e-06,
      "loss": 0.1309,
      "step": 3457
    },
    {
      "epoch": 0.25,
      "grad_norm": 18.95085334777832,
      "learning_rate": 8.805308720210955e-06,
      "loss": 1.1703,
      "step": 3458
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.5601725578308105,
      "learning_rate": 8.804555295612558e-06,
      "loss": 0.1672,
      "step": 3459
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.367326259613037,
      "learning_rate": 8.803801665772804e-06,
      "loss": 0.1793,
      "step": 3460
    },
    {
      "epoch": 0.25,
      "grad_norm": 16.925201416015625,
      "learning_rate": 8.803047830732348e-06,
      "loss": 1.5759,
      "step": 3461
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.491528511047363,
      "learning_rate": 8.802293790531858e-06,
      "loss": 0.7161,
      "step": 3462
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.234106540679932,
      "learning_rate": 8.80153954521201e-06,
      "loss": 0.8119,
      "step": 3463
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.35544776916504,
      "learning_rate": 8.800785094813493e-06,
      "loss": 1.1355,
      "step": 3464
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.9460179805755615,
      "learning_rate": 8.800030439377007e-06,
      "loss": 0.1201,
      "step": 3465
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.506248712539673,
      "learning_rate": 8.799275578943263e-06,
      "loss": 0.1618,
      "step": 3466
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.299746513366699,
      "learning_rate": 8.798520513552981e-06,
      "loss": 0.9744,
      "step": 3467
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.084218978881836,
      "learning_rate": 8.797765243246898e-06,
      "loss": 0.6826,
      "step": 3468
    },
    {
      "epoch": 0.25,
      "grad_norm": 15.488914489746094,
      "learning_rate": 8.797009768065755e-06,
      "loss": 0.9698,
      "step": 3469
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.668675422668457,
      "learning_rate": 8.796254088050305e-06,
      "loss": 0.1348,
      "step": 3470
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.155791759490967,
      "learning_rate": 8.795498203241319e-06,
      "loss": 0.8153,
      "step": 3471
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.2528164386749268,
      "learning_rate": 8.79474211367957e-06,
      "loss": 0.1987,
      "step": 3472
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.813027381896973,
      "learning_rate": 8.793985819405851e-06,
      "loss": 0.6498,
      "step": 3473
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.805790424346924,
      "learning_rate": 8.793229320460955e-06,
      "loss": 0.62,
      "step": 3474
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.7593002319335938,
      "learning_rate": 8.792472616885696e-06,
      "loss": 0.1192,
      "step": 3475
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.487211227416992,
      "learning_rate": 8.791715708720893e-06,
      "loss": 1.6435,
      "step": 3476
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.1125335693359375,
      "learning_rate": 8.790958596007382e-06,
      "loss": 1.0067,
      "step": 3477
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.7385880947113037,
      "learning_rate": 8.790201278786004e-06,
      "loss": 0.2299,
      "step": 3478
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.115399360656738,
      "learning_rate": 8.789443757097615e-06,
      "loss": 0.9531,
      "step": 3479
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.528020143508911,
      "learning_rate": 8.788686030983077e-06,
      "loss": 0.1143,
      "step": 3480
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2720792293548584,
      "learning_rate": 8.787928100483269e-06,
      "loss": 0.1508,
      "step": 3481
    },
    {
      "epoch": 0.25,
      "grad_norm": 21.722225189208984,
      "learning_rate": 8.78716996563908e-06,
      "loss": 1.3256,
      "step": 3482
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.113974571228027,
      "learning_rate": 8.786411626491405e-06,
      "loss": 0.8156,
      "step": 3483
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.2767322063446045,
      "learning_rate": 8.785653083081153e-06,
      "loss": 0.1568,
      "step": 3484
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.297073364257812,
      "learning_rate": 8.78489433544925e-06,
      "loss": 0.5232,
      "step": 3485
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.905820608139038,
      "learning_rate": 8.784135383636623e-06,
      "loss": 0.1507,
      "step": 3486
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.973756790161133,
      "learning_rate": 8.783376227684215e-06,
      "loss": 0.1819,
      "step": 3487
    },
    {
      "epoch": 0.25,
      "grad_norm": 18.196746826171875,
      "learning_rate": 8.782616867632982e-06,
      "loss": 1.156,
      "step": 3488
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.599611282348633,
      "learning_rate": 8.781857303523886e-06,
      "loss": 0.1871,
      "step": 3489
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.65114426612854,
      "learning_rate": 8.781097535397904e-06,
      "loss": 0.2649,
      "step": 3490
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.069324493408203,
      "learning_rate": 8.780337563296022e-06,
      "loss": 0.2,
      "step": 3491
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.857264518737793,
      "learning_rate": 8.779577387259237e-06,
      "loss": 0.8215,
      "step": 3492
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.650775909423828,
      "learning_rate": 8.778817007328559e-06,
      "loss": 1.5299,
      "step": 3493
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.027890682220459,
      "learning_rate": 8.778056423545008e-06,
      "loss": 0.1674,
      "step": 3494
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.830077886581421,
      "learning_rate": 8.777295635949612e-06,
      "loss": 0.1257,
      "step": 3495
    },
    {
      "epoch": 0.25,
      "grad_norm": 15.022753715515137,
      "learning_rate": 8.776534644583416e-06,
      "loss": 1.6172,
      "step": 3496
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.373356819152832,
      "learning_rate": 8.775773449487469e-06,
      "loss": 1.0776,
      "step": 3497
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.062749862670898,
      "learning_rate": 8.775012050702838e-06,
      "loss": 1.8122,
      "step": 3498
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.844263076782227,
      "learning_rate": 8.774250448270596e-06,
      "loss": 0.9899,
      "step": 3499
    },
    {
      "epoch": 0.25,
      "grad_norm": 15.559508323669434,
      "learning_rate": 8.773488642231828e-06,
      "loss": 1.2905,
      "step": 3500
    },
    {
      "epoch": 0.25,
      "grad_norm": 20.67278480529785,
      "learning_rate": 8.77272663262763e-06,
      "loss": 1.1343,
      "step": 3501
    },
    {
      "epoch": 0.25,
      "grad_norm": 27.648813247680664,
      "learning_rate": 8.771964419499113e-06,
      "loss": 0.9865,
      "step": 3502
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.921749114990234,
      "learning_rate": 8.771202002887392e-06,
      "loss": 0.6547,
      "step": 3503
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.994030475616455,
      "learning_rate": 8.770439382833598e-06,
      "loss": 0.6817,
      "step": 3504
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.703680992126465,
      "learning_rate": 8.76967655937887e-06,
      "loss": 0.2598,
      "step": 3505
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.566256046295166,
      "learning_rate": 8.768913532564363e-06,
      "loss": 0.1278,
      "step": 3506
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3275723457336426,
      "learning_rate": 8.768150302431235e-06,
      "loss": 0.1031,
      "step": 3507
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.250544548034668,
      "learning_rate": 8.767386869020662e-06,
      "loss": 0.0939,
      "step": 3508
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.0091395378112793,
      "learning_rate": 8.766623232373829e-06,
      "loss": 0.1631,
      "step": 3509
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.5956146717071533,
      "learning_rate": 8.76585939253193e-06,
      "loss": 0.2722,
      "step": 3510
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.167787551879883,
      "learning_rate": 8.76509534953617e-06,
      "loss": 0.7592,
      "step": 3511
    },
    {
      "epoch": 0.25,
      "grad_norm": 34.943321228027344,
      "learning_rate": 8.764331103427768e-06,
      "loss": 1.3454,
      "step": 3512
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.8008604049682617,
      "learning_rate": 8.763566654247951e-06,
      "loss": 0.2007,
      "step": 3513
    },
    {
      "epoch": 0.25,
      "grad_norm": 15.301290512084961,
      "learning_rate": 8.76280200203796e-06,
      "loss": 1.5792,
      "step": 3514
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.456066370010376,
      "learning_rate": 8.762037146839043e-06,
      "loss": 0.2552,
      "step": 3515
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.0043575763702393,
      "learning_rate": 8.761272088692463e-06,
      "loss": 0.2125,
      "step": 3516
    },
    {
      "epoch": 0.25,
      "grad_norm": 25.08172607421875,
      "learning_rate": 8.76050682763949e-06,
      "loss": 2.0912,
      "step": 3517
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.440638303756714,
      "learning_rate": 8.759741363721408e-06,
      "loss": 0.1047,
      "step": 3518
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.640551567077637,
      "learning_rate": 8.758975696979513e-06,
      "loss": 1.7325,
      "step": 3519
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.440399646759033,
      "learning_rate": 8.758209827455105e-06,
      "loss": 0.2888,
      "step": 3520
    },
    {
      "epoch": 0.25,
      "grad_norm": 26.734533309936523,
      "learning_rate": 8.757443755189503e-06,
      "loss": 0.2735,
      "step": 3521
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.6662049293518066,
      "learning_rate": 8.756677480224034e-06,
      "loss": 0.2425,
      "step": 3522
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.823296308517456,
      "learning_rate": 8.755911002600034e-06,
      "loss": 0.1385,
      "step": 3523
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.8531558513641357,
      "learning_rate": 8.755144322358853e-06,
      "loss": 0.3453,
      "step": 3524
    },
    {
      "epoch": 0.25,
      "grad_norm": 26.26752281188965,
      "learning_rate": 8.754377439541848e-06,
      "loss": 1.3678,
      "step": 3525
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.12773323059082,
      "learning_rate": 8.753610354190394e-06,
      "loss": 0.9856,
      "step": 3526
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.0185675621032715,
      "learning_rate": 8.752843066345868e-06,
      "loss": 0.9176,
      "step": 3527
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.178989887237549,
      "learning_rate": 8.752075576049664e-06,
      "loss": 0.9745,
      "step": 3528
    },
    {
      "epoch": 0.25,
      "grad_norm": 22.654903411865234,
      "learning_rate": 8.751307883343186e-06,
      "loss": 0.7675,
      "step": 3529
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.558788537979126,
      "learning_rate": 8.750539988267845e-06,
      "loss": 0.2054,
      "step": 3530
    },
    {
      "epoch": 0.25,
      "grad_norm": 20.728147506713867,
      "learning_rate": 8.749771890865069e-06,
      "loss": 1.1144,
      "step": 3531
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.65529441833496,
      "learning_rate": 8.749003591176293e-06,
      "loss": 1.3698,
      "step": 3532
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.860008716583252,
      "learning_rate": 8.748235089242964e-06,
      "loss": 0.1933,
      "step": 3533
    },
    {
      "epoch": 0.25,
      "grad_norm": 16.895540237426758,
      "learning_rate": 8.747466385106542e-06,
      "loss": 1.2732,
      "step": 3534
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.51763916015625,
      "learning_rate": 8.746697478808491e-06,
      "loss": 1.6472,
      "step": 3535
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.689727783203125,
      "learning_rate": 8.745928370390295e-06,
      "loss": 0.8664,
      "step": 3536
    },
    {
      "epoch": 0.25,
      "grad_norm": 17.004968643188477,
      "learning_rate": 8.74515905989344e-06,
      "loss": 1.4162,
      "step": 3537
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.0567407608032227,
      "learning_rate": 8.744389547359432e-06,
      "loss": 0.2541,
      "step": 3538
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.870331764221191,
      "learning_rate": 8.74361983282978e-06,
      "loss": 0.9871,
      "step": 3539
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.914487838745117,
      "learning_rate": 8.742849916346008e-06,
      "loss": 0.8964,
      "step": 3540
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.9194984436035156,
      "learning_rate": 8.742079797949653e-06,
      "loss": 0.1521,
      "step": 3541
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.5741100311279297,
      "learning_rate": 8.741309477682255e-06,
      "loss": 0.1514,
      "step": 3542
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.611295700073242,
      "learning_rate": 8.740538955585374e-06,
      "loss": 1.5197,
      "step": 3543
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.054387092590332,
      "learning_rate": 8.739768231700574e-06,
      "loss": 0.2201,
      "step": 3544
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.747147560119629,
      "learning_rate": 8.738997306069432e-06,
      "loss": 0.7918,
      "step": 3545
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.5635502338409424,
      "learning_rate": 8.73822617873354e-06,
      "loss": 0.2002,
      "step": 3546
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.9840729236602783,
      "learning_rate": 8.737454849734498e-06,
      "loss": 0.912,
      "step": 3547
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.380089282989502,
      "learning_rate": 8.73668331911391e-06,
      "loss": 0.9972,
      "step": 3548
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.06112003326416,
      "learning_rate": 8.735911586913402e-06,
      "loss": 1.1902,
      "step": 3549
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.871826171875,
      "learning_rate": 8.735139653174608e-06,
      "loss": 1.1871,
      "step": 3550
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.3437461853027344,
      "learning_rate": 8.734367517939163e-06,
      "loss": 0.2336,
      "step": 3551
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.912376880645752,
      "learning_rate": 8.733595181248729e-06,
      "loss": 0.9661,
      "step": 3552
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.598069190979004,
      "learning_rate": 8.732822643144967e-06,
      "loss": 0.9385,
      "step": 3553
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.7630746364593506,
      "learning_rate": 8.732049903669552e-06,
      "loss": 0.2236,
      "step": 3554
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.334928035736084,
      "learning_rate": 8.731276962864172e-06,
      "loss": 0.2886,
      "step": 3555
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.98043441772461,
      "learning_rate": 8.730503820770523e-06,
      "loss": 1.4992,
      "step": 3556
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.222707748413086,
      "learning_rate": 8.729730477430315e-06,
      "loss": 0.6983,
      "step": 3557
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.930184364318848,
      "learning_rate": 8.728956932885264e-06,
      "loss": 1.3073,
      "step": 3558
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.662258148193359,
      "learning_rate": 8.728183187177101e-06,
      "loss": 1.082,
      "step": 3559
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.977810859680176,
      "learning_rate": 8.727409240347569e-06,
      "loss": 1.2042,
      "step": 3560
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.0091323852539062,
      "learning_rate": 8.726635092438416e-06,
      "loss": 0.1467,
      "step": 3561
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.924165725708008,
      "learning_rate": 8.725860743491407e-06,
      "loss": 1.0258,
      "step": 3562
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.115586280822754,
      "learning_rate": 8.725086193548313e-06,
      "loss": 0.8071,
      "step": 3563
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1485648155212402,
      "learning_rate": 8.72431144265092e-06,
      "loss": 0.0782,
      "step": 3564
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.852924346923828,
      "learning_rate": 8.72353649084102e-06,
      "loss": 0.6071,
      "step": 3565
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.364192485809326,
      "learning_rate": 8.722761338160422e-06,
      "loss": 1.011,
      "step": 3566
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.157099723815918,
      "learning_rate": 8.721985984650942e-06,
      "loss": 0.9228,
      "step": 3567
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.401063442230225,
      "learning_rate": 8.721210430354406e-06,
      "loss": 1.2329,
      "step": 3568
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.086515426635742,
      "learning_rate": 8.720434675312652e-06,
      "loss": 1.1435,
      "step": 3569
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.860114574432373,
      "learning_rate": 8.719658719567532e-06,
      "loss": 0.1366,
      "step": 3570
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.9720797538757324,
      "learning_rate": 8.718882563160902e-06,
      "loss": 0.1021,
      "step": 3571
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.322578430175781,
      "learning_rate": 8.718106206134633e-06,
      "loss": 0.9749,
      "step": 3572
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.660400390625,
      "learning_rate": 8.71732964853061e-06,
      "loss": 0.88,
      "step": 3573
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.663121461868286,
      "learning_rate": 8.716552890390723e-06,
      "loss": 0.1519,
      "step": 3574
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.076938629150391,
      "learning_rate": 8.715775931756876e-06,
      "loss": 0.9642,
      "step": 3575
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.770272970199585,
      "learning_rate": 8.714998772670981e-06,
      "loss": 0.281,
      "step": 3576
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.883331060409546,
      "learning_rate": 8.714221413174967e-06,
      "loss": 0.9261,
      "step": 3577
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.060202121734619,
      "learning_rate": 8.713443853310765e-06,
      "loss": 0.196,
      "step": 3578
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.4082067012786865,
      "learning_rate": 8.712666093120323e-06,
      "loss": 0.3256,
      "step": 3579
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.383091926574707,
      "learning_rate": 8.711888132645598e-06,
      "loss": 1.2625,
      "step": 3580
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.085688591003418,
      "learning_rate": 8.71110997192856e-06,
      "loss": 0.2199,
      "step": 3581
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8095169067382812,
      "learning_rate": 8.710331611011184e-06,
      "loss": 0.168,
      "step": 3582
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.490185737609863,
      "learning_rate": 8.709553049935464e-06,
      "loss": 1.0987,
      "step": 3583
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.546158790588379,
      "learning_rate": 8.708774288743399e-06,
      "loss": 0.1643,
      "step": 3584
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.277813911437988,
      "learning_rate": 8.707995327476998e-06,
      "loss": 0.2412,
      "step": 3585
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.803617238998413,
      "learning_rate": 8.707216166178285e-06,
      "loss": 0.1151,
      "step": 3586
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.450141429901123,
      "learning_rate": 8.706436804889293e-06,
      "loss": 0.8603,
      "step": 3587
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.645053863525391,
      "learning_rate": 8.705657243652064e-06,
      "loss": 0.7997,
      "step": 3588
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.972545623779297,
      "learning_rate": 8.704877482508655e-06,
      "loss": 0.5658,
      "step": 3589
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.993062019348145,
      "learning_rate": 8.704097521501128e-06,
      "loss": 0.8653,
      "step": 3590
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.638288974761963,
      "learning_rate": 8.703317360671562e-06,
      "loss": 0.7237,
      "step": 3591
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.21026349067688,
      "learning_rate": 8.70253700006204e-06,
      "loss": 0.227,
      "step": 3592
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.559710502624512,
      "learning_rate": 8.701756439714664e-06,
      "loss": 1.0775,
      "step": 3593
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5737738609313965,
      "learning_rate": 8.700975679671538e-06,
      "loss": 0.1319,
      "step": 3594
    },
    {
      "epoch": 0.26,
      "grad_norm": 20.902681350708008,
      "learning_rate": 8.700194719974783e-06,
      "loss": 1.5154,
      "step": 3595
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.444269180297852,
      "learning_rate": 8.69941356066653e-06,
      "loss": 0.7228,
      "step": 3596
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8570868968963623,
      "learning_rate": 8.698632201788919e-06,
      "loss": 0.233,
      "step": 3597
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.704746723175049,
      "learning_rate": 8.6978506433841e-06,
      "loss": 0.1055,
      "step": 3598
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.3332481384277344,
      "learning_rate": 8.697068885494236e-06,
      "loss": 0.1441,
      "step": 3599
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.6923017501831055,
      "learning_rate": 8.696286928161499e-06,
      "loss": 0.9897,
      "step": 3600
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.079593658447266,
      "learning_rate": 8.695504771428075e-06,
      "loss": 0.1987,
      "step": 3601
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.541881799697876,
      "learning_rate": 8.694722415336155e-06,
      "loss": 0.2899,
      "step": 3602
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.722195863723755,
      "learning_rate": 8.693939859927946e-06,
      "loss": 0.1094,
      "step": 3603
    },
    {
      "epoch": 0.26,
      "grad_norm": 16.303550720214844,
      "learning_rate": 8.693157105245667e-06,
      "loss": 1.0979,
      "step": 3604
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.2922909259796143,
      "learning_rate": 8.692374151331539e-06,
      "loss": 0.152,
      "step": 3605
    },
    {
      "epoch": 0.26,
      "grad_norm": 16.67504119873047,
      "learning_rate": 8.6915909982278e-06,
      "loss": 0.832,
      "step": 3606
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.230364799499512,
      "learning_rate": 8.690807645976705e-06,
      "loss": 0.7393,
      "step": 3607
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8717451095581055,
      "learning_rate": 8.690024094620504e-06,
      "loss": 0.1699,
      "step": 3608
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.619947671890259,
      "learning_rate": 8.689240344201473e-06,
      "loss": 0.1277,
      "step": 3609
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.280949831008911,
      "learning_rate": 8.688456394761888e-06,
      "loss": 0.1757,
      "step": 3610
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.673839569091797,
      "learning_rate": 8.687672246344044e-06,
      "loss": 1.212,
      "step": 3611
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.253798961639404,
      "learning_rate": 8.686887898990239e-06,
      "loss": 0.8954,
      "step": 3612
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.7265660762786865,
      "learning_rate": 8.686103352742788e-06,
      "loss": 0.1147,
      "step": 3613
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.824063777923584,
      "learning_rate": 8.685318607644015e-06,
      "loss": 0.9759,
      "step": 3614
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.787280559539795,
      "learning_rate": 8.684533663736252e-06,
      "loss": 0.8809,
      "step": 3615
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.7661139965057373,
      "learning_rate": 8.683748521061844e-06,
      "loss": 0.2675,
      "step": 3616
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.17425012588501,
      "learning_rate": 8.682963179663147e-06,
      "loss": 1.0365,
      "step": 3617
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.790897369384766,
      "learning_rate": 8.682177639582529e-06,
      "loss": 1.1992,
      "step": 3618
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.1519172191619873,
      "learning_rate": 8.681391900862364e-06,
      "loss": 0.1427,
      "step": 3619
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.89077091217041,
      "learning_rate": 8.68060596354504e-06,
      "loss": 0.2367,
      "step": 3620
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.9405808448791504,
      "learning_rate": 8.679819827672957e-06,
      "loss": 0.1747,
      "step": 3621
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.627796173095703,
      "learning_rate": 8.679033493288522e-06,
      "loss": 0.8611,
      "step": 3622
    },
    {
      "epoch": 0.26,
      "grad_norm": 25.342159271240234,
      "learning_rate": 8.678246960434157e-06,
      "loss": 0.9803,
      "step": 3623
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.43882942199707,
      "learning_rate": 8.677460229152292e-06,
      "loss": 1.3006,
      "step": 3624
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.069403886795044,
      "learning_rate": 8.676673299485367e-06,
      "loss": 0.2268,
      "step": 3625
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.8007960319519043,
      "learning_rate": 8.675886171475833e-06,
      "loss": 0.8887,
      "step": 3626
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.394465446472168,
      "learning_rate": 8.675098845166157e-06,
      "loss": 1.3295,
      "step": 3627
    },
    {
      "epoch": 0.26,
      "grad_norm": 46.75782012939453,
      "learning_rate": 8.674311320598807e-06,
      "loss": 1.3061,
      "step": 3628
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.3037819862365723,
      "learning_rate": 8.67352359781627e-06,
      "loss": 0.1755,
      "step": 3629
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.195365905761719,
      "learning_rate": 8.67273567686104e-06,
      "loss": 0.8314,
      "step": 3630
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.278402328491211,
      "learning_rate": 8.671947557775622e-06,
      "loss": 0.087,
      "step": 3631
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.41349983215332,
      "learning_rate": 8.671159240602533e-06,
      "loss": 1.3244,
      "step": 3632
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0239813327789307,
      "learning_rate": 8.670370725384298e-06,
      "loss": 0.0837,
      "step": 3633
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.4622063636779785,
      "learning_rate": 8.669582012163457e-06,
      "loss": 0.9416,
      "step": 3634
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.373422384262085,
      "learning_rate": 8.668793100982555e-06,
      "loss": 0.0936,
      "step": 3635
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.03679084777832,
      "learning_rate": 8.668003991884154e-06,
      "loss": 0.9059,
      "step": 3636
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.117292642593384,
      "learning_rate": 8.66721468491082e-06,
      "loss": 0.1899,
      "step": 3637
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.685295343399048,
      "learning_rate": 8.666425180105136e-06,
      "loss": 0.186,
      "step": 3638
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.198758602142334,
      "learning_rate": 8.66563547750969e-06,
      "loss": 0.9034,
      "step": 3639
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5741126537323,
      "learning_rate": 8.664845577167087e-06,
      "loss": 0.1366,
      "step": 3640
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.725096225738525,
      "learning_rate": 8.664055479119937e-06,
      "loss": 1.0068,
      "step": 3641
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.991544723510742,
      "learning_rate": 8.663265183410863e-06,
      "loss": 1.3359,
      "step": 3642
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.84371018409729,
      "learning_rate": 8.662474690082498e-06,
      "loss": 0.2313,
      "step": 3643
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.491502046585083,
      "learning_rate": 8.661683999177489e-06,
      "loss": 0.1277,
      "step": 3644
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.723956823348999,
      "learning_rate": 8.660893110738485e-06,
      "loss": 0.135,
      "step": 3645
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.200973033905029,
      "learning_rate": 8.660102024808158e-06,
      "loss": 0.904,
      "step": 3646
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.18211555480957,
      "learning_rate": 8.659310741429178e-06,
      "loss": 0.7602,
      "step": 3647
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.2752718925476074,
      "learning_rate": 8.658519260644236e-06,
      "loss": 0.2006,
      "step": 3648
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.58223295211792,
      "learning_rate": 8.657727582496029e-06,
      "loss": 0.1283,
      "step": 3649
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.919059753417969,
      "learning_rate": 8.656935707027264e-06,
      "loss": 0.7697,
      "step": 3650
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.722399711608887,
      "learning_rate": 8.656143634280659e-06,
      "loss": 1.5899,
      "step": 3651
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.6403160095214844,
      "learning_rate": 8.655351364298945e-06,
      "loss": 0.1082,
      "step": 3652
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.954197883605957,
      "learning_rate": 8.65455889712486e-06,
      "loss": 0.3059,
      "step": 3653
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.615890026092529,
      "learning_rate": 8.653766232801158e-06,
      "loss": 0.6791,
      "step": 3654
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.546842575073242,
      "learning_rate": 8.652973371370597e-06,
      "loss": 0.2588,
      "step": 3655
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.897366523742676,
      "learning_rate": 8.652180312875948e-06,
      "loss": 0.9464,
      "step": 3656
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.566391944885254,
      "learning_rate": 8.651387057359996e-06,
      "loss": 0.0964,
      "step": 3657
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.79098892211914,
      "learning_rate": 8.650593604865535e-06,
      "loss": 0.9609,
      "step": 3658
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.321464538574219,
      "learning_rate": 8.649799955435367e-06,
      "loss": 0.6245,
      "step": 3659
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.155757188796997,
      "learning_rate": 8.649006109112305e-06,
      "loss": 0.2833,
      "step": 3660
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.27310037612915,
      "learning_rate": 8.64821206593918e-06,
      "loss": 0.475,
      "step": 3661
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.3919055461883545,
      "learning_rate": 8.64741782595882e-06,
      "loss": 0.1785,
      "step": 3662
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.273439407348633,
      "learning_rate": 8.646623389214073e-06,
      "loss": 0.7866,
      "step": 3663
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.876194000244141,
      "learning_rate": 8.6458287557478e-06,
      "loss": 0.9613,
      "step": 3664
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.5663533210754395,
      "learning_rate": 8.645033925602864e-06,
      "loss": 0.2754,
      "step": 3665
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.8662025928497314,
      "learning_rate": 8.644238898822146e-06,
      "loss": 0.2312,
      "step": 3666
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.956388473510742,
      "learning_rate": 8.643443675448533e-06,
      "loss": 0.7782,
      "step": 3667
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3507540225982666,
      "learning_rate": 8.642648255524926e-06,
      "loss": 0.0969,
      "step": 3668
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.754924774169922,
      "learning_rate": 8.641852639094232e-06,
      "loss": 0.115,
      "step": 3669
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.6016721725463867,
      "learning_rate": 8.641056826199375e-06,
      "loss": 0.2741,
      "step": 3670
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.540754795074463,
      "learning_rate": 8.640260816883284e-06,
      "loss": 0.1558,
      "step": 3671
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.2427845001220703,
      "learning_rate": 8.6394646111889e-06,
      "loss": 0.14,
      "step": 3672
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.747777938842773,
      "learning_rate": 8.638668209159177e-06,
      "loss": 1.3659,
      "step": 3673
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.654476165771484,
      "learning_rate": 8.637871610837078e-06,
      "loss": 1.5291,
      "step": 3674
    },
    {
      "epoch": 0.26,
      "grad_norm": 16.425697326660156,
      "learning_rate": 8.637074816265574e-06,
      "loss": 0.5023,
      "step": 3675
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.356470823287964,
      "learning_rate": 8.636277825487654e-06,
      "loss": 0.2982,
      "step": 3676
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.4033620357513428,
      "learning_rate": 8.635480638546305e-06,
      "loss": 0.2043,
      "step": 3677
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.442436456680298,
      "learning_rate": 8.63468325548454e-06,
      "loss": 0.1892,
      "step": 3678
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.925070285797119,
      "learning_rate": 8.63388567634537e-06,
      "loss": 0.3568,
      "step": 3679
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.774106979370117,
      "learning_rate": 8.633087901171822e-06,
      "loss": 1.2923,
      "step": 3680
    },
    {
      "epoch": 0.26,
      "grad_norm": 19.44649887084961,
      "learning_rate": 8.632289930006935e-06,
      "loss": 1.1787,
      "step": 3681
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.647125244140625,
      "learning_rate": 8.631491762893755e-06,
      "loss": 0.9181,
      "step": 3682
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8119494915008545,
      "learning_rate": 8.630693399875341e-06,
      "loss": 0.148,
      "step": 3683
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.181823492050171,
      "learning_rate": 8.62989484099476e-06,
      "loss": 0.2496,
      "step": 3684
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.14570140838623,
      "learning_rate": 8.629096086295092e-06,
      "loss": 0.9188,
      "step": 3685
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.378843307495117,
      "learning_rate": 8.628297135819428e-06,
      "loss": 0.1142,
      "step": 3686
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.882722854614258,
      "learning_rate": 8.627497989610868e-06,
      "loss": 0.6783,
      "step": 3687
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.472635269165039,
      "learning_rate": 8.62669864771252e-06,
      "loss": 1.0601,
      "step": 3688
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.946168899536133,
      "learning_rate": 8.625899110167508e-06,
      "loss": 0.1864,
      "step": 3689
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8321621417999268,
      "learning_rate": 8.625099377018965e-06,
      "loss": 0.1524,
      "step": 3690
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.453338623046875,
      "learning_rate": 8.624299448310031e-06,
      "loss": 0.8994,
      "step": 3691
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.823899030685425,
      "learning_rate": 8.623499324083862e-06,
      "loss": 0.1703,
      "step": 3692
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.206641674041748,
      "learning_rate": 8.62269900438362e-06,
      "loss": 0.9784,
      "step": 3693
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.094876289367676,
      "learning_rate": 8.621898489252479e-06,
      "loss": 1.138,
      "step": 3694
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.89679479598999,
      "learning_rate": 8.621097778733624e-06,
      "loss": 0.8857,
      "step": 3695
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.300184726715088,
      "learning_rate": 8.620296872870251e-06,
      "loss": 0.6629,
      "step": 3696
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.396074295043945,
      "learning_rate": 8.619495771705564e-06,
      "loss": 1.1362,
      "step": 3697
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.6719963550567627,
      "learning_rate": 8.618694475282783e-06,
      "loss": 0.1048,
      "step": 3698
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.002962112426758,
      "learning_rate": 8.61789298364513e-06,
      "loss": 1.4525,
      "step": 3699
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.635820388793945,
      "learning_rate": 8.617091296835847e-06,
      "loss": 0.9838,
      "step": 3700
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.1258609294891357,
      "learning_rate": 8.616289414898178e-06,
      "loss": 0.1449,
      "step": 3701
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.130570888519287,
      "learning_rate": 8.615487337875385e-06,
      "loss": 0.214,
      "step": 3702
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.8822615146636963,
      "learning_rate": 8.614685065810734e-06,
      "loss": 0.7922,
      "step": 3703
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4986448287963867,
      "learning_rate": 8.613882598747509e-06,
      "loss": 0.1263,
      "step": 3704
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.267916202545166,
      "learning_rate": 8.613079936728994e-06,
      "loss": 0.9485,
      "step": 3705
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.07821273803711,
      "learning_rate": 8.612277079798494e-06,
      "loss": 0.6687,
      "step": 3706
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.577009916305542,
      "learning_rate": 8.611474027999319e-06,
      "loss": 0.1322,
      "step": 3707
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.761931896209717,
      "learning_rate": 8.610670781374789e-06,
      "loss": 0.6103,
      "step": 3708
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.481982707977295,
      "learning_rate": 8.609867339968239e-06,
      "loss": 0.2145,
      "step": 3709
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.587241172790527,
      "learning_rate": 8.609063703823009e-06,
      "loss": 1.3512,
      "step": 3710
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.476158857345581,
      "learning_rate": 8.608259872982454e-06,
      "loss": 0.2939,
      "step": 3711
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.312610626220703,
      "learning_rate": 8.607455847489935e-06,
      "loss": 0.2009,
      "step": 3712
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.965741157531738,
      "learning_rate": 8.606651627388831e-06,
      "loss": 0.5081,
      "step": 3713
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.2136361598968506,
      "learning_rate": 8.605847212722523e-06,
      "loss": 0.2194,
      "step": 3714
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.704914093017578,
      "learning_rate": 8.605042603534405e-06,
      "loss": 0.7883,
      "step": 3715
    },
    {
      "epoch": 0.27,
      "grad_norm": 24.221242904663086,
      "learning_rate": 8.604237799867885e-06,
      "loss": 2.0341,
      "step": 3716
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.628326416015625,
      "learning_rate": 8.603432801766379e-06,
      "loss": 1.0412,
      "step": 3717
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.70244312286377,
      "learning_rate": 8.602627609273313e-06,
      "loss": 1.0678,
      "step": 3718
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.885362386703491,
      "learning_rate": 8.601822222432124e-06,
      "loss": 0.091,
      "step": 3719
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.5375547409057617,
      "learning_rate": 8.601016641286259e-06,
      "loss": 0.2544,
      "step": 3720
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.440054416656494,
      "learning_rate": 8.600210865879178e-06,
      "loss": 0.1045,
      "step": 3721
    },
    {
      "epoch": 0.27,
      "grad_norm": 19.17667579650879,
      "learning_rate": 8.599404896254349e-06,
      "loss": 1.2803,
      "step": 3722
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.517989158630371,
      "learning_rate": 8.598598732455248e-06,
      "loss": 0.1194,
      "step": 3723
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.182754516601562,
      "learning_rate": 8.59779237452537e-06,
      "loss": 1.1891,
      "step": 3724
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.827540397644043,
      "learning_rate": 8.596985822508208e-06,
      "loss": 1.2801,
      "step": 3725
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.674622058868408,
      "learning_rate": 8.596179076447278e-06,
      "loss": 0.3017,
      "step": 3726
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.933795928955078,
      "learning_rate": 8.5953721363861e-06,
      "loss": 1.356,
      "step": 3727
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.062785148620605,
      "learning_rate": 8.594565002368202e-06,
      "loss": 0.9959,
      "step": 3728
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.852598190307617,
      "learning_rate": 8.593757674437131e-06,
      "loss": 1.3545,
      "step": 3729
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.354941368103027,
      "learning_rate": 8.592950152636435e-06,
      "loss": 1.1961,
      "step": 3730
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.57625675201416,
      "learning_rate": 8.592142437009678e-06,
      "loss": 0.0721,
      "step": 3731
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.207991600036621,
      "learning_rate": 8.591334527600434e-06,
      "loss": 1.4466,
      "step": 3732
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.632596254348755,
      "learning_rate": 8.590526424452287e-06,
      "loss": 0.1375,
      "step": 3733
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.671817779541016,
      "learning_rate": 8.58971812760883e-06,
      "loss": 1.3099,
      "step": 3734
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.04507064819336,
      "learning_rate": 8.588909637113668e-06,
      "loss": 1.169,
      "step": 3735
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.709009170532227,
      "learning_rate": 8.588100953010415e-06,
      "loss": 0.7868,
      "step": 3736
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.459301233291626,
      "learning_rate": 8.587292075342698e-06,
      "loss": 0.2431,
      "step": 3737
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.039473533630371,
      "learning_rate": 8.586483004154151e-06,
      "loss": 0.1065,
      "step": 3738
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.315274238586426,
      "learning_rate": 8.585673739488422e-06,
      "loss": 0.2455,
      "step": 3739
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.5939364433288574,
      "learning_rate": 8.584864281389168e-06,
      "loss": 0.1603,
      "step": 3740
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3195104598999023,
      "learning_rate": 8.584054629900057e-06,
      "loss": 0.2341,
      "step": 3741
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.263704299926758,
      "learning_rate": 8.583244785064762e-06,
      "loss": 0.3101,
      "step": 3742
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.65164041519165,
      "learning_rate": 8.582434746926976e-06,
      "loss": 1.0358,
      "step": 3743
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.305780410766602,
      "learning_rate": 8.581624515530393e-06,
      "loss": 0.7875,
      "step": 3744
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.6338930130004883,
      "learning_rate": 8.580814090918727e-06,
      "loss": 0.1267,
      "step": 3745
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.559101104736328,
      "learning_rate": 8.580003473135695e-06,
      "loss": 0.1442,
      "step": 3746
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.4874467849731445,
      "learning_rate": 8.579192662225026e-06,
      "loss": 1.0229,
      "step": 3747
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.03961944580078,
      "learning_rate": 8.57838165823046e-06,
      "loss": 1.2621,
      "step": 3748
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.9314815998077393,
      "learning_rate": 8.57757046119575e-06,
      "loss": 0.1734,
      "step": 3749
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.486356735229492,
      "learning_rate": 8.576759071164656e-06,
      "loss": 0.8356,
      "step": 3750
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.8576369285583496,
      "learning_rate": 8.575947488180946e-06,
      "loss": 0.3693,
      "step": 3751
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.553979873657227,
      "learning_rate": 8.575135712288409e-06,
      "loss": 1.1878,
      "step": 3752
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.14134979248047,
      "learning_rate": 8.57432374353083e-06,
      "loss": 1.1208,
      "step": 3753
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.244577884674072,
      "learning_rate": 8.573511581952015e-06,
      "loss": 0.9583,
      "step": 3754
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3637728691101074,
      "learning_rate": 8.572699227595775e-06,
      "loss": 0.1623,
      "step": 3755
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.20672082901001,
      "learning_rate": 8.571886680505938e-06,
      "loss": 1.0537,
      "step": 3756
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.308609008789062,
      "learning_rate": 8.571073940726334e-06,
      "loss": 0.8182,
      "step": 3757
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.0042266845703125,
      "learning_rate": 8.570261008300808e-06,
      "loss": 0.1678,
      "step": 3758
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.337312698364258,
      "learning_rate": 8.569447883273215e-06,
      "loss": 0.7059,
      "step": 3759
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7166972160339355,
      "learning_rate": 8.568634565687418e-06,
      "loss": 0.1257,
      "step": 3760
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.479681015014648,
      "learning_rate": 8.567821055587295e-06,
      "loss": 0.7448,
      "step": 3761
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.372957468032837,
      "learning_rate": 8.567007353016732e-06,
      "loss": 0.1452,
      "step": 3762
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.0404093265533447,
      "learning_rate": 8.566193458019624e-06,
      "loss": 0.1935,
      "step": 3763
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.900275945663452,
      "learning_rate": 8.565379370639876e-06,
      "loss": 0.1813,
      "step": 3764
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.55798864364624,
      "learning_rate": 8.56456509092141e-06,
      "loss": 0.9148,
      "step": 3765
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.0498595237731934,
      "learning_rate": 8.563750618908148e-06,
      "loss": 0.1407,
      "step": 3766
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.3928728103637695,
      "learning_rate": 8.56293595464403e-06,
      "loss": 0.6074,
      "step": 3767
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.7120838165283203,
      "learning_rate": 8.562121098173002e-06,
      "loss": 0.8735,
      "step": 3768
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.478297233581543,
      "learning_rate": 8.561306049539027e-06,
      "loss": 0.9935,
      "step": 3769
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.2666616439819336,
      "learning_rate": 8.560490808786069e-06,
      "loss": 0.1374,
      "step": 3770
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.556129455566406,
      "learning_rate": 8.559675375958108e-06,
      "loss": 1.3826,
      "step": 3771
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.449899435043335,
      "learning_rate": 8.558859751099137e-06,
      "loss": 0.0954,
      "step": 3772
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.659180641174316,
      "learning_rate": 8.558043934253152e-06,
      "loss": 1.3959,
      "step": 3773
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.087007522583008,
      "learning_rate": 8.557227925464165e-06,
      "loss": 1.5432,
      "step": 3774
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.5859060287475586,
      "learning_rate": 8.556411724776196e-06,
      "loss": 0.1222,
      "step": 3775
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.274656295776367,
      "learning_rate": 8.555595332233277e-06,
      "loss": 0.9582,
      "step": 3776
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.3255293369293213,
      "learning_rate": 8.554778747879448e-06,
      "loss": 0.1244,
      "step": 3777
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.299494743347168,
      "learning_rate": 8.553961971758762e-06,
      "loss": 0.7845,
      "step": 3778
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7924790382385254,
      "learning_rate": 8.55314500391528e-06,
      "loss": 0.1595,
      "step": 3779
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.093368530273438,
      "learning_rate": 8.552327844393072e-06,
      "loss": 1.7982,
      "step": 3780
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.938546895980835,
      "learning_rate": 8.551510493236227e-06,
      "loss": 0.1438,
      "step": 3781
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.322392463684082,
      "learning_rate": 8.55069295048883e-06,
      "loss": 1.2008,
      "step": 3782
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.2197744846343994,
      "learning_rate": 8.549875216194991e-06,
      "loss": 0.2191,
      "step": 3783
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.703437328338623,
      "learning_rate": 8.54905729039882e-06,
      "loss": 0.9949,
      "step": 3784
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8207218647003174,
      "learning_rate": 8.548239173144443e-06,
      "loss": 0.1491,
      "step": 3785
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.789146661758423,
      "learning_rate": 8.547420864475992e-06,
      "loss": 0.0974,
      "step": 3786
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.0027313232421875,
      "learning_rate": 8.546602364437615e-06,
      "loss": 0.8847,
      "step": 3787
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.272705078125,
      "learning_rate": 8.545783673073463e-06,
      "loss": 0.7189,
      "step": 3788
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.121644496917725,
      "learning_rate": 8.544964790427704e-06,
      "loss": 0.7829,
      "step": 3789
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.676911354064941,
      "learning_rate": 8.544145716544515e-06,
      "loss": 1.0376,
      "step": 3790
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.897464752197266,
      "learning_rate": 8.543326451468078e-06,
      "loss": 0.8045,
      "step": 3791
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.0906167030334473,
      "learning_rate": 8.542506995242591e-06,
      "loss": 0.2231,
      "step": 3792
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.248319149017334,
      "learning_rate": 8.541687347912262e-06,
      "loss": 0.9702,
      "step": 3793
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.5745153427124023,
      "learning_rate": 8.540867509521305e-06,
      "loss": 0.3349,
      "step": 3794
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3891685009002686,
      "learning_rate": 8.54004748011395e-06,
      "loss": 0.2313,
      "step": 3795
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.376321315765381,
      "learning_rate": 8.539227259734432e-06,
      "loss": 1.0649,
      "step": 3796
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.394845008850098,
      "learning_rate": 8.538406848427002e-06,
      "loss": 1.0232,
      "step": 3797
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.103670120239258,
      "learning_rate": 8.537586246235914e-06,
      "loss": 1.0255,
      "step": 3798
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.569535970687866,
      "learning_rate": 8.53676545320544e-06,
      "loss": 0.7015,
      "step": 3799
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.08653450012207,
      "learning_rate": 8.535944469379855e-06,
      "loss": 0.839,
      "step": 3800
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.6158385276794434,
      "learning_rate": 8.535123294803453e-06,
      "loss": 0.1754,
      "step": 3801
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.586395263671875,
      "learning_rate": 8.534301929520529e-06,
      "loss": 0.1372,
      "step": 3802
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.784417152404785,
      "learning_rate": 8.533480373575393e-06,
      "loss": 0.1656,
      "step": 3803
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.641611576080322,
      "learning_rate": 8.532658627012369e-06,
      "loss": 1.1393,
      "step": 3804
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.006233215332031,
      "learning_rate": 8.531836689875781e-06,
      "loss": 0.8119,
      "step": 3805
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.323812961578369,
      "learning_rate": 8.531014562209974e-06,
      "loss": 0.8922,
      "step": 3806
    },
    {
      "epoch": 0.27,
      "grad_norm": 24.766061782836914,
      "learning_rate": 8.530192244059296e-06,
      "loss": 1.4628,
      "step": 3807
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3837602138519287,
      "learning_rate": 8.529369735468109e-06,
      "loss": 0.2684,
      "step": 3808
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3437116146087646,
      "learning_rate": 8.528547036480782e-06,
      "loss": 0.1995,
      "step": 3809
    },
    {
      "epoch": 0.27,
      "grad_norm": 28.032506942749023,
      "learning_rate": 8.527724147141702e-06,
      "loss": 0.8859,
      "step": 3810
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3768630027770996,
      "learning_rate": 8.526901067495257e-06,
      "loss": 0.1752,
      "step": 3811
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8936851024627686,
      "learning_rate": 8.526077797585849e-06,
      "loss": 0.2099,
      "step": 3812
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.404975891113281,
      "learning_rate": 8.525254337457891e-06,
      "loss": 0.7563,
      "step": 3813
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.3212227821350098,
      "learning_rate": 8.524430687155805e-06,
      "loss": 0.0732,
      "step": 3814
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.666351318359375,
      "learning_rate": 8.523606846724023e-06,
      "loss": 0.1445,
      "step": 3815
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.184680938720703,
      "learning_rate": 8.522782816206992e-06,
      "loss": 1.3531,
      "step": 3816
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.1304922103881836,
      "learning_rate": 8.52195859564916e-06,
      "loss": 0.1399,
      "step": 3817
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.396167755126953,
      "learning_rate": 8.521134185094992e-06,
      "loss": 1.2471,
      "step": 3818
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.203950881958008,
      "learning_rate": 8.520309584588965e-06,
      "loss": 0.1209,
      "step": 3819
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.874737501144409,
      "learning_rate": 8.51948479417556e-06,
      "loss": 0.1674,
      "step": 3820
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.9043450355529785,
      "learning_rate": 8.518659813899275e-06,
      "loss": 0.103,
      "step": 3821
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.18096351623535,
      "learning_rate": 8.51783464380461e-06,
      "loss": 1.4376,
      "step": 3822
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.95280647277832,
      "learning_rate": 8.517009283936082e-06,
      "loss": 0.9071,
      "step": 3823
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.596090316772461,
      "learning_rate": 8.516183734338216e-06,
      "loss": 0.6426,
      "step": 3824
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.502511739730835,
      "learning_rate": 8.515357995055545e-06,
      "loss": 0.2818,
      "step": 3825
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.37551212310791,
      "learning_rate": 8.514532066132618e-06,
      "loss": 0.784,
      "step": 3826
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7349026203155518,
      "learning_rate": 8.513705947613988e-06,
      "loss": 0.1268,
      "step": 3827
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.861551284790039,
      "learning_rate": 8.512879639544224e-06,
      "loss": 0.9294,
      "step": 3828
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.6137185096740723,
      "learning_rate": 8.512053141967902e-06,
      "loss": 0.264,
      "step": 3829
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.863602876663208,
      "learning_rate": 8.511226454929604e-06,
      "loss": 0.1291,
      "step": 3830
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.1773529052734375,
      "learning_rate": 8.510399578473929e-06,
      "loss": 0.7643,
      "step": 3831
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.1915338039398193,
      "learning_rate": 8.509572512645486e-06,
      "loss": 0.161,
      "step": 3832
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.6395857334136963,
      "learning_rate": 8.508745257488888e-06,
      "loss": 0.2308,
      "step": 3833
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0626299381256104,
      "learning_rate": 8.507917813048768e-06,
      "loss": 0.1011,
      "step": 3834
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.579959392547607,
      "learning_rate": 8.50709017936976e-06,
      "loss": 0.896,
      "step": 3835
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.089292526245117,
      "learning_rate": 8.506262356496509e-06,
      "loss": 0.1883,
      "step": 3836
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.59047794342041,
      "learning_rate": 8.505434344473678e-06,
      "loss": 0.1219,
      "step": 3837
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.922087669372559,
      "learning_rate": 8.504606143345931e-06,
      "loss": 1.1849,
      "step": 3838
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.625929832458496,
      "learning_rate": 8.50377775315795e-06,
      "loss": 0.7555,
      "step": 3839
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.879437446594238,
      "learning_rate": 8.502949173954423e-06,
      "loss": 1.0257,
      "step": 3840
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0795347690582275,
      "learning_rate": 8.502120405780045e-06,
      "loss": 0.094,
      "step": 3841
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4893274307250977,
      "learning_rate": 8.501291448679527e-06,
      "loss": 0.1271,
      "step": 3842
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.535813093185425,
      "learning_rate": 8.50046230269759e-06,
      "loss": 0.1429,
      "step": 3843
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.904728412628174,
      "learning_rate": 8.499632967878962e-06,
      "loss": 0.2088,
      "step": 3844
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.944016695022583,
      "learning_rate": 8.49880344426838e-06,
      "loss": 0.1947,
      "step": 3845
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9236249923706055,
      "learning_rate": 8.497973731910599e-06,
      "loss": 0.2295,
      "step": 3846
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.402427673339844,
      "learning_rate": 8.497143830850374e-06,
      "loss": 1.0994,
      "step": 3847
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.882864475250244,
      "learning_rate": 8.496313741132476e-06,
      "loss": 0.9955,
      "step": 3848
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.398696899414062,
      "learning_rate": 8.495483462801686e-06,
      "loss": 0.9291,
      "step": 3849
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.2227783203125,
      "learning_rate": 8.494652995902795e-06,
      "loss": 1.2299,
      "step": 3850
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.285126686096191,
      "learning_rate": 8.493822340480602e-06,
      "loss": 0.6359,
      "step": 3851
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.1899309158325195,
      "learning_rate": 8.492991496579921e-06,
      "loss": 0.2022,
      "step": 3852
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.14113998413086,
      "learning_rate": 8.492160464245569e-06,
      "loss": 1.6074,
      "step": 3853
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.02531099319458,
      "learning_rate": 8.491329243522378e-06,
      "loss": 1.1417,
      "step": 3854
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.7834270000457764,
      "learning_rate": 8.490497834455189e-06,
      "loss": 0.1098,
      "step": 3855
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.528621196746826,
      "learning_rate": 8.489666237088857e-06,
      "loss": 0.9783,
      "step": 3856
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.079597473144531,
      "learning_rate": 8.488834451468238e-06,
      "loss": 0.9354,
      "step": 3857
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.485868215560913,
      "learning_rate": 8.488002477638207e-06,
      "loss": 0.256,
      "step": 3858
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4581522941589355,
      "learning_rate": 8.487170315643645e-06,
      "loss": 0.0945,
      "step": 3859
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.686817169189453,
      "learning_rate": 8.486337965529445e-06,
      "loss": 1.1278,
      "step": 3860
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.0585548877716064,
      "learning_rate": 8.485505427340508e-06,
      "loss": 0.1481,
      "step": 3861
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.123680830001831,
      "learning_rate": 8.484672701121746e-06,
      "loss": 0.259,
      "step": 3862
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.060366153717041,
      "learning_rate": 8.48383978691808e-06,
      "loss": 0.1694,
      "step": 3863
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.839876890182495,
      "learning_rate": 8.483006684774448e-06,
      "loss": 0.1122,
      "step": 3864
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.1093251705169678,
      "learning_rate": 8.48217339473579e-06,
      "loss": 0.1722,
      "step": 3865
    },
    {
      "epoch": 0.28,
      "grad_norm": 18.905099868774414,
      "learning_rate": 8.481339916847054e-06,
      "loss": 1.1335,
      "step": 3866
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.693322658538818,
      "learning_rate": 8.480506251153208e-06,
      "loss": 0.8703,
      "step": 3867
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0812387466430664,
      "learning_rate": 8.479672397699227e-06,
      "loss": 0.1328,
      "step": 3868
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.046366691589355,
      "learning_rate": 8.478838356530089e-06,
      "loss": 0.9781,
      "step": 3869
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.688445568084717,
      "learning_rate": 8.478004127690791e-06,
      "loss": 0.8605,
      "step": 3870
    },
    {
      "epoch": 0.28,
      "grad_norm": 19.78893280029297,
      "learning_rate": 8.477169711226335e-06,
      "loss": 1.4868,
      "step": 3871
    },
    {
      "epoch": 0.28,
      "grad_norm": 25.754711151123047,
      "learning_rate": 8.476335107181734e-06,
      "loss": 1.0635,
      "step": 3872
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.0285000801086426,
      "learning_rate": 8.475500315602013e-06,
      "loss": 0.0945,
      "step": 3873
    },
    {
      "epoch": 0.28,
      "grad_norm": 21.193838119506836,
      "learning_rate": 8.474665336532207e-06,
      "loss": 1.0476,
      "step": 3874
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.551163673400879,
      "learning_rate": 8.473830170017357e-06,
      "loss": 2.0313,
      "step": 3875
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.702085494995117,
      "learning_rate": 8.472994816102518e-06,
      "loss": 0.1391,
      "step": 3876
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.990373134613037,
      "learning_rate": 8.472159274832755e-06,
      "loss": 0.7293,
      "step": 3877
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6151671409606934,
      "learning_rate": 8.471323546253143e-06,
      "loss": 0.1793,
      "step": 3878
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.22208309173584,
      "learning_rate": 8.470487630408763e-06,
      "loss": 0.9874,
      "step": 3879
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.893350601196289,
      "learning_rate": 8.469651527344714e-06,
      "loss": 1.2555,
      "step": 3880
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.4034526348114014,
      "learning_rate": 8.468815237106098e-06,
      "loss": 0.1752,
      "step": 3881
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.297041416168213,
      "learning_rate": 8.46797875973803e-06,
      "loss": 0.1166,
      "step": 3882
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.838545560836792,
      "learning_rate": 8.467142095285635e-06,
      "loss": 0.1874,
      "step": 3883
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.093825340270996,
      "learning_rate": 8.466305243794049e-06,
      "loss": 1.0699,
      "step": 3884
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.126856803894043,
      "learning_rate": 8.465468205308414e-06,
      "loss": 1.0658,
      "step": 3885
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.137354850769043,
      "learning_rate": 8.464630979873887e-06,
      "loss": 0.8972,
      "step": 3886
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.514949798583984,
      "learning_rate": 8.463793567535633e-06,
      "loss": 0.8921,
      "step": 3887
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.498000144958496,
      "learning_rate": 8.462955968338828e-06,
      "loss": 1.4387,
      "step": 3888
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.602092742919922,
      "learning_rate": 8.462118182328655e-06,
      "loss": 0.9448,
      "step": 3889
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.433192253112793,
      "learning_rate": 8.461280209550313e-06,
      "loss": 0.4329,
      "step": 3890
    },
    {
      "epoch": 0.28,
      "grad_norm": 21.993453979492188,
      "learning_rate": 8.460442050049004e-06,
      "loss": 1.1178,
      "step": 3891
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.70986270904541,
      "learning_rate": 8.459603703869945e-06,
      "loss": 1.0791,
      "step": 3892
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.13008975982666,
      "learning_rate": 8.45876517105836e-06,
      "loss": 0.804,
      "step": 3893
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.22420072555542,
      "learning_rate": 8.45792645165949e-06,
      "loss": 0.5072,
      "step": 3894
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.03928279876709,
      "learning_rate": 8.457087545718574e-06,
      "loss": 0.1226,
      "step": 3895
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.83039665222168,
      "learning_rate": 8.456248453280873e-06,
      "loss": 1.115,
      "step": 3896
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.051531791687012,
      "learning_rate": 8.455409174391648e-06,
      "loss": 0.8692,
      "step": 3897
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.558713912963867,
      "learning_rate": 8.454569709096179e-06,
      "loss": 1.0078,
      "step": 3898
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.554507255554199,
      "learning_rate": 8.453730057439749e-06,
      "loss": 0.6093,
      "step": 3899
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.83810043334961,
      "learning_rate": 8.452890219467657e-06,
      "loss": 1.2797,
      "step": 3900
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.1086459159851074,
      "learning_rate": 8.452050195225207e-06,
      "loss": 0.1432,
      "step": 3901
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.91906476020813,
      "learning_rate": 8.451209984757715e-06,
      "loss": 0.154,
      "step": 3902
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.218061923980713,
      "learning_rate": 8.450369588110508e-06,
      "loss": 0.2228,
      "step": 3903
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.726616382598877,
      "learning_rate": 8.449529005328922e-06,
      "loss": 0.172,
      "step": 3904
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.533439636230469,
      "learning_rate": 8.448688236458305e-06,
      "loss": 0.9831,
      "step": 3905
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.8798649311065674,
      "learning_rate": 8.44784728154401e-06,
      "loss": 0.1175,
      "step": 3906
    },
    {
      "epoch": 0.28,
      "grad_norm": 19.129179000854492,
      "learning_rate": 8.447006140631407e-06,
      "loss": 1.1009,
      "step": 3907
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.425772190093994,
      "learning_rate": 8.446164813765868e-06,
      "loss": 0.2042,
      "step": 3908
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.143651008605957,
      "learning_rate": 8.445323300992783e-06,
      "loss": 0.9014,
      "step": 3909
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.423334836959839,
      "learning_rate": 8.444481602357546e-06,
      "loss": 0.1223,
      "step": 3910
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.81655502319336,
      "learning_rate": 8.443639717905565e-06,
      "loss": 0.3803,
      "step": 3911
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.867452621459961,
      "learning_rate": 8.442797647682257e-06,
      "loss": 0.9738,
      "step": 3912
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.506850004196167,
      "learning_rate": 8.441955391733048e-06,
      "loss": 0.6054,
      "step": 3913
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.1352763175964355,
      "learning_rate": 8.441112950103371e-06,
      "loss": 0.1932,
      "step": 3914
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.8777480125427246,
      "learning_rate": 8.440270322838679e-06,
      "loss": 0.1969,
      "step": 3915
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.250516891479492,
      "learning_rate": 8.439427509984424e-06,
      "loss": 0.7391,
      "step": 3916
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.436243057250977,
      "learning_rate": 8.438584511586074e-06,
      "loss": 0.9841,
      "step": 3917
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.7774417400360107,
      "learning_rate": 8.437741327689105e-06,
      "loss": 0.8231,
      "step": 3918
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.789440393447876,
      "learning_rate": 8.436897958339005e-06,
      "loss": 0.2576,
      "step": 3919
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.689435958862305,
      "learning_rate": 8.436054403581268e-06,
      "loss": 0.84,
      "step": 3920
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.273359775543213,
      "learning_rate": 8.435210663461403e-06,
      "loss": 0.9285,
      "step": 3921
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.281172037124634,
      "learning_rate": 8.434366738024925e-06,
      "loss": 0.2064,
      "step": 3922
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.1087050437927246,
      "learning_rate": 8.433522627317361e-06,
      "loss": 0.1988,
      "step": 3923
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.664382934570312,
      "learning_rate": 8.432678331384249e-06,
      "loss": 1.639,
      "step": 3924
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9450974464416504,
      "learning_rate": 8.431833850271134e-06,
      "loss": 0.2186,
      "step": 3925
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.993860244750977,
      "learning_rate": 8.430989184023572e-06,
      "loss": 1.0735,
      "step": 3926
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.561415672302246,
      "learning_rate": 8.430144332687133e-06,
      "loss": 0.2061,
      "step": 3927
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.096847534179688,
      "learning_rate": 8.42929929630739e-06,
      "loss": 1.07,
      "step": 3928
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.068852424621582,
      "learning_rate": 8.428454074929929e-06,
      "loss": 0.2281,
      "step": 3929
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.30285930633545,
      "learning_rate": 8.427608668600348e-06,
      "loss": 1.0301,
      "step": 3930
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9017858505249023,
      "learning_rate": 8.426763077364257e-06,
      "loss": 0.1928,
      "step": 3931
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.2737667560577393,
      "learning_rate": 8.425917301267266e-06,
      "loss": 0.174,
      "step": 3932
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.867708921432495,
      "learning_rate": 8.425071340355005e-06,
      "loss": 0.1354,
      "step": 3933
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.5618934631347656,
      "learning_rate": 8.42422519467311e-06,
      "loss": 0.1127,
      "step": 3934
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.636375427246094,
      "learning_rate": 8.423378864267226e-06,
      "loss": 1.7006,
      "step": 3935
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.970264434814453,
      "learning_rate": 8.422532349183012e-06,
      "loss": 0.9898,
      "step": 3936
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.625408411026001,
      "learning_rate": 8.421685649466133e-06,
      "loss": 0.1106,
      "step": 3937
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.7974400520324707,
      "learning_rate": 8.420838765162264e-06,
      "loss": 0.1902,
      "step": 3938
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.4376819133758545,
      "learning_rate": 8.419991696317091e-06,
      "loss": 0.2489,
      "step": 3939
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.429146766662598,
      "learning_rate": 8.419144442976313e-06,
      "loss": 0.7268,
      "step": 3940
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.214077949523926,
      "learning_rate": 8.418297005185634e-06,
      "loss": 0.545,
      "step": 3941
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.003347396850586,
      "learning_rate": 8.417449382990771e-06,
      "loss": 2.1372,
      "step": 3942
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.971964836120605,
      "learning_rate": 8.41660157643745e-06,
      "loss": 0.997,
      "step": 3943
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.6806914806365967,
      "learning_rate": 8.415753585571406e-06,
      "loss": 0.3721,
      "step": 3944
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.798247337341309,
      "learning_rate": 8.414905410438385e-06,
      "loss": 0.5192,
      "step": 3945
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.3597495555877686,
      "learning_rate": 8.414057051084144e-06,
      "loss": 0.261,
      "step": 3946
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.721397399902344,
      "learning_rate": 8.413208507554447e-06,
      "loss": 0.94,
      "step": 3947
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.839911460876465,
      "learning_rate": 8.41235977989507e-06,
      "loss": 0.9036,
      "step": 3948
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.56887435913086,
      "learning_rate": 8.411510868151801e-06,
      "loss": 1.3346,
      "step": 3949
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7979869842529297,
      "learning_rate": 8.410661772370433e-06,
      "loss": 0.0669,
      "step": 3950
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.4697039127349854,
      "learning_rate": 8.409812492596774e-06,
      "loss": 0.2777,
      "step": 3951
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.537973165512085,
      "learning_rate": 8.408963028876636e-06,
      "loss": 0.194,
      "step": 3952
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.400599479675293,
      "learning_rate": 8.408113381255847e-06,
      "loss": 0.988,
      "step": 3953
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.679001808166504,
      "learning_rate": 8.40726354978024e-06,
      "loss": 1.0622,
      "step": 3954
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.302691459655762,
      "learning_rate": 8.406413534495664e-06,
      "loss": 0.7917,
      "step": 3955
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.39115047454834,
      "learning_rate": 8.40556333544797e-06,
      "loss": 0.7417,
      "step": 3956
    },
    {
      "epoch": 0.28,
      "grad_norm": 19.892393112182617,
      "learning_rate": 8.404712952683027e-06,
      "loss": 0.838,
      "step": 3957
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9891231060028076,
      "learning_rate": 8.403862386246706e-06,
      "loss": 0.2748,
      "step": 3958
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.01379108428955,
      "learning_rate": 8.403011636184894e-06,
      "loss": 1.0859,
      "step": 3959
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.887070417404175,
      "learning_rate": 8.402160702543485e-06,
      "loss": 0.2362,
      "step": 3960
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.4718968868255615,
      "learning_rate": 8.401309585368385e-06,
      "loss": 0.1898,
      "step": 3961
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.820819854736328,
      "learning_rate": 8.400458284705506e-06,
      "loss": 0.8059,
      "step": 3962
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.85147762298584,
      "learning_rate": 8.399606800600774e-06,
      "loss": 0.9162,
      "step": 3963
    },
    {
      "epoch": 0.28,
      "grad_norm": 20.69620704650879,
      "learning_rate": 8.398755133100125e-06,
      "loss": 1.1581,
      "step": 3964
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.83495044708252,
      "learning_rate": 8.3979032822495e-06,
      "loss": 1.4009,
      "step": 3965
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.092970848083496,
      "learning_rate": 8.397051248094856e-06,
      "loss": 0.1123,
      "step": 3966
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.706342697143555,
      "learning_rate": 8.396199030682156e-06,
      "loss": 0.5681,
      "step": 3967
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5055246353149414,
      "learning_rate": 8.395346630057372e-06,
      "loss": 0.1132,
      "step": 3968
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.428218841552734,
      "learning_rate": 8.39449404626649e-06,
      "loss": 0.8455,
      "step": 3969
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.097907304763794,
      "learning_rate": 8.393641279355502e-06,
      "loss": 0.1844,
      "step": 3970
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.6444034576416016,
      "learning_rate": 8.392788329370414e-06,
      "loss": 0.2936,
      "step": 3971
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.474303245544434,
      "learning_rate": 8.391935196357239e-06,
      "loss": 0.5732,
      "step": 3972
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.94560718536377,
      "learning_rate": 8.391081880361998e-06,
      "loss": 1.2565,
      "step": 3973
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.319413185119629,
      "learning_rate": 8.390228381430725e-06,
      "loss": 1.0296,
      "step": 3974
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.8347012996673584,
      "learning_rate": 8.389374699609465e-06,
      "loss": 1.0086,
      "step": 3975
    },
    {
      "epoch": 0.29,
      "grad_norm": 21.051311492919922,
      "learning_rate": 8.388520834944268e-06,
      "loss": 1.3461,
      "step": 3976
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.377649307250977,
      "learning_rate": 8.3876667874812e-06,
      "loss": 1.1015,
      "step": 3977
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.5206615924835205,
      "learning_rate": 8.386812557266329e-06,
      "loss": 0.1413,
      "step": 3978
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.551892280578613,
      "learning_rate": 8.385958144345743e-06,
      "loss": 1.3588,
      "step": 3979
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.52565336227417,
      "learning_rate": 8.38510354876553e-06,
      "loss": 1.0239,
      "step": 3980
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.945376396179199,
      "learning_rate": 8.384248770571794e-06,
      "loss": 0.2125,
      "step": 3981
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.0285801887512207,
      "learning_rate": 8.383393809810648e-06,
      "loss": 0.18,
      "step": 3982
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2634856700897217,
      "learning_rate": 8.382538666528212e-06,
      "loss": 0.1762,
      "step": 3983
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.689896583557129,
      "learning_rate": 8.381683340770618e-06,
      "loss": 0.9217,
      "step": 3984
    },
    {
      "epoch": 0.29,
      "grad_norm": 21.02092742919922,
      "learning_rate": 8.380827832584008e-06,
      "loss": 1.2005,
      "step": 3985
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8201797008514404,
      "learning_rate": 8.379972142014534e-06,
      "loss": 0.2286,
      "step": 3986
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9132626056671143,
      "learning_rate": 8.379116269108359e-06,
      "loss": 0.0893,
      "step": 3987
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.240445375442505,
      "learning_rate": 8.378260213911648e-06,
      "loss": 0.1079,
      "step": 3988
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.318277359008789,
      "learning_rate": 8.37740397647059e-06,
      "loss": 0.9571,
      "step": 3989
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.517770290374756,
      "learning_rate": 8.376547556831369e-06,
      "loss": 0.8096,
      "step": 3990
    },
    {
      "epoch": 0.29,
      "grad_norm": 20.838111877441406,
      "learning_rate": 8.37569095504019e-06,
      "loss": 1.3199,
      "step": 3991
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.355390548706055,
      "learning_rate": 8.37483417114326e-06,
      "loss": 0.9164,
      "step": 3992
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.0847012996673584,
      "learning_rate": 8.373977205186801e-06,
      "loss": 0.153,
      "step": 3993
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.856548309326172,
      "learning_rate": 8.373120057217044e-06,
      "loss": 1.1315,
      "step": 3994
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.038788318634033,
      "learning_rate": 8.372262727280228e-06,
      "loss": 1.0569,
      "step": 3995
    },
    {
      "epoch": 0.29,
      "grad_norm": 18.49404525756836,
      "learning_rate": 8.3714052154226e-06,
      "loss": 2.2471,
      "step": 3996
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.6111953258514404,
      "learning_rate": 8.370547521690424e-06,
      "loss": 0.2366,
      "step": 3997
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.298098087310791,
      "learning_rate": 8.369689646129968e-06,
      "loss": 0.0978,
      "step": 3998
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.013155460357666,
      "learning_rate": 8.36883158878751e-06,
      "loss": 0.1451,
      "step": 3999
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.063262939453125,
      "learning_rate": 8.36797334970934e-06,
      "loss": 0.1964,
      "step": 4000
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9479598999023438,
      "learning_rate": 8.367114928941755e-06,
      "loss": 0.1616,
      "step": 4001
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.05134654045105,
      "learning_rate": 8.366256326531065e-06,
      "loss": 0.1818,
      "step": 4002
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.259897708892822,
      "learning_rate": 8.36539754252359e-06,
      "loss": 0.9125,
      "step": 4003
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.101262092590332,
      "learning_rate": 8.364538576965653e-06,
      "loss": 1.2444,
      "step": 4004
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.476824998855591,
      "learning_rate": 8.363679429903597e-06,
      "loss": 0.1656,
      "step": 4005
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.278332471847534,
      "learning_rate": 8.36282010138377e-06,
      "loss": 0.1521,
      "step": 4006
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.236632823944092,
      "learning_rate": 8.361960591452525e-06,
      "loss": 0.8758,
      "step": 4007
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.349040985107422,
      "learning_rate": 8.361100900156232e-06,
      "loss": 1.1766,
      "step": 4008
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.994928359985352,
      "learning_rate": 8.36024102754127e-06,
      "loss": 0.8875,
      "step": 4009
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.282524108886719,
      "learning_rate": 8.35938097365402e-06,
      "loss": 1.8644,
      "step": 4010
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9525060653686523,
      "learning_rate": 8.358520738540886e-06,
      "loss": 0.1301,
      "step": 4011
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.4095282554626465,
      "learning_rate": 8.357660322248269e-06,
      "loss": 0.3898,
      "step": 4012
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.5336110591888428,
      "learning_rate": 8.356799724822586e-06,
      "loss": 0.2397,
      "step": 4013
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.893233060836792,
      "learning_rate": 8.355938946310265e-06,
      "loss": 0.0892,
      "step": 4014
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.6991820335388184,
      "learning_rate": 8.355077986757743e-06,
      "loss": 0.2366,
      "step": 4015
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.082921504974365,
      "learning_rate": 8.354216846211462e-06,
      "loss": 0.7983,
      "step": 4016
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7657270431518555,
      "learning_rate": 8.353355524717877e-06,
      "loss": 0.2147,
      "step": 4017
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.6077213287353516,
      "learning_rate": 8.352494022323457e-06,
      "loss": 0.1739,
      "step": 4018
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.8382768630981445,
      "learning_rate": 8.35163233907467e-06,
      "loss": 0.9487,
      "step": 4019
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5710031986236572,
      "learning_rate": 8.350770475018009e-06,
      "loss": 0.0932,
      "step": 4020
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.675590515136719,
      "learning_rate": 8.349908430199965e-06,
      "loss": 1.2062,
      "step": 4021
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2375810146331787,
      "learning_rate": 8.349046204667038e-06,
      "loss": 0.223,
      "step": 4022
    },
    {
      "epoch": 0.29,
      "grad_norm": 16.568864822387695,
      "learning_rate": 8.348183798465746e-06,
      "loss": 1.0347,
      "step": 4023
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.910677194595337,
      "learning_rate": 8.347321211642611e-06,
      "loss": 0.2389,
      "step": 4024
    },
    {
      "epoch": 0.29,
      "grad_norm": 33.49213790893555,
      "learning_rate": 8.34645844424417e-06,
      "loss": 1.6053,
      "step": 4025
    },
    {
      "epoch": 0.29,
      "grad_norm": 277.6747741699219,
      "learning_rate": 8.345595496316961e-06,
      "loss": 0.9783,
      "step": 4026
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5686728954315186,
      "learning_rate": 8.344732367907538e-06,
      "loss": 0.1744,
      "step": 4027
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.8601226806640625,
      "learning_rate": 8.343869059062466e-06,
      "loss": 1.1016,
      "step": 4028
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.200630187988281,
      "learning_rate": 8.343005569828313e-06,
      "loss": 0.7698,
      "step": 4029
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.560925483703613,
      "learning_rate": 8.342141900251666e-06,
      "loss": 0.9614,
      "step": 4030
    },
    {
      "epoch": 0.29,
      "grad_norm": 18.570405960083008,
      "learning_rate": 8.341278050379113e-06,
      "loss": 1.1982,
      "step": 4031
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.339859962463379,
      "learning_rate": 8.340414020257258e-06,
      "loss": 1.0961,
      "step": 4032
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7404866218566895,
      "learning_rate": 8.339549809932709e-06,
      "loss": 0.1408,
      "step": 4033
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2093448638916016,
      "learning_rate": 8.338685419452089e-06,
      "loss": 0.1874,
      "step": 4034
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.579092025756836,
      "learning_rate": 8.337820848862028e-06,
      "loss": 0.8895,
      "step": 4035
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.44447660446167,
      "learning_rate": 8.336956098209169e-06,
      "loss": 0.141,
      "step": 4036
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.833732604980469,
      "learning_rate": 8.336091167540154e-06,
      "loss": 0.9466,
      "step": 4037
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.102055549621582,
      "learning_rate": 8.335226056901653e-06,
      "loss": 1.2355,
      "step": 4038
    },
    {
      "epoch": 0.29,
      "grad_norm": 16.828887939453125,
      "learning_rate": 8.33436076634033e-06,
      "loss": 1.0965,
      "step": 4039
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.785415649414062,
      "learning_rate": 8.333495295902865e-06,
      "loss": 1.002,
      "step": 4040
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.985234260559082,
      "learning_rate": 8.332629645635946e-06,
      "loss": 0.2462,
      "step": 4041
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.3783111572265625,
      "learning_rate": 8.331763815586272e-06,
      "loss": 0.2459,
      "step": 4042
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.289749145507812,
      "learning_rate": 8.330897805800553e-06,
      "loss": 0.9785,
      "step": 4043
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.288039445877075,
      "learning_rate": 8.330031616325504e-06,
      "loss": 0.1478,
      "step": 4044
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.658350944519043,
      "learning_rate": 8.329165247207856e-06,
      "loss": 0.8828,
      "step": 4045
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.4242353439331055,
      "learning_rate": 8.328298698494342e-06,
      "loss": 0.8263,
      "step": 4046
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9469213485717773,
      "learning_rate": 8.327431970231712e-06,
      "loss": 0.1834,
      "step": 4047
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2640881538391113,
      "learning_rate": 8.326565062466723e-06,
      "loss": 0.2547,
      "step": 4048
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.998935341835022,
      "learning_rate": 8.325697975246142e-06,
      "loss": 0.1115,
      "step": 4049
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.666274070739746,
      "learning_rate": 8.324830708616741e-06,
      "loss": 1.2394,
      "step": 4050
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.782803773880005,
      "learning_rate": 8.32396326262531e-06,
      "loss": 0.1199,
      "step": 4051
    },
    {
      "epoch": 0.29,
      "grad_norm": 18.115400314331055,
      "learning_rate": 8.323095637318643e-06,
      "loss": 1.2714,
      "step": 4052
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.106067419052124,
      "learning_rate": 8.322227832743544e-06,
      "loss": 0.2376,
      "step": 4053
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.489015579223633,
      "learning_rate": 8.321359848946828e-06,
      "loss": 0.8261,
      "step": 4054
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.112651824951172,
      "learning_rate": 8.320491685975322e-06,
      "loss": 1.0439,
      "step": 4055
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.9539783000946045,
      "learning_rate": 8.319623343875858e-06,
      "loss": 0.9316,
      "step": 4056
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.588397026062012,
      "learning_rate": 8.318754822695282e-06,
      "loss": 1.3834,
      "step": 4057
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8959739208221436,
      "learning_rate": 8.317886122480441e-06,
      "loss": 0.1278,
      "step": 4058
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.990289688110352,
      "learning_rate": 8.317017243278205e-06,
      "loss": 1.0001,
      "step": 4059
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0012145042419434,
      "learning_rate": 8.316148185135444e-06,
      "loss": 0.0764,
      "step": 4060
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.788879871368408,
      "learning_rate": 8.315278948099041e-06,
      "loss": 0.2172,
      "step": 4061
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3333349227905273,
      "learning_rate": 8.314409532215887e-06,
      "loss": 0.1251,
      "step": 4062
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.81517219543457,
      "learning_rate": 8.313539937532886e-06,
      "loss": 0.8209,
      "step": 4063
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.73404598236084,
      "learning_rate": 8.312670164096947e-06,
      "loss": 1.347,
      "step": 4064
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3999836444854736,
      "learning_rate": 8.311800211954993e-06,
      "loss": 0.1762,
      "step": 4065
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.7432243824005127,
      "learning_rate": 8.310930081153951e-06,
      "loss": 0.1498,
      "step": 4066
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.095125198364258,
      "learning_rate": 8.310059771740764e-06,
      "loss": 1.4418,
      "step": 4067
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.566680908203125,
      "learning_rate": 8.309189283762382e-06,
      "loss": 0.2188,
      "step": 4068
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.143998146057129,
      "learning_rate": 8.308318617265764e-06,
      "loss": 1.2849,
      "step": 4069
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.463528633117676,
      "learning_rate": 8.307447772297882e-06,
      "loss": 0.7962,
      "step": 4070
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2273645401000977,
      "learning_rate": 8.306576748905709e-06,
      "loss": 0.1122,
      "step": 4071
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.022811889648438,
      "learning_rate": 8.305705547136237e-06,
      "loss": 0.8205,
      "step": 4072
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.054410934448242,
      "learning_rate": 8.304834167036464e-06,
      "loss": 0.8101,
      "step": 4073
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.556113243103027,
      "learning_rate": 8.303962608653396e-06,
      "loss": 1.1466,
      "step": 4074
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.377622127532959,
      "learning_rate": 8.303090872034053e-06,
      "loss": 0.1544,
      "step": 4075
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.351604700088501,
      "learning_rate": 8.30221895722546e-06,
      "loss": 0.2651,
      "step": 4076
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.542955160140991,
      "learning_rate": 8.301346864274652e-06,
      "loss": 0.1915,
      "step": 4077
    },
    {
      "epoch": 0.29,
      "grad_norm": 17.16863250732422,
      "learning_rate": 8.30047459322868e-06,
      "loss": 0.8366,
      "step": 4078
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7858011722564697,
      "learning_rate": 8.299602144134594e-06,
      "loss": 0.1806,
      "step": 4079
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.175204277038574,
      "learning_rate": 8.298729517039463e-06,
      "loss": 0.0702,
      "step": 4080
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.175907611846924,
      "learning_rate": 8.297856711990361e-06,
      "loss": 1.012,
      "step": 4081
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.117030143737793,
      "learning_rate": 8.296983729034374e-06,
      "loss": 0.1836,
      "step": 4082
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5399889945983887,
      "learning_rate": 8.296110568218592e-06,
      "loss": 0.1346,
      "step": 4083
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.403763771057129,
      "learning_rate": 8.295237229590123e-06,
      "loss": 0.6708,
      "step": 4084
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.703540325164795,
      "learning_rate": 8.294363713196077e-06,
      "loss": 0.8353,
      "step": 4085
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.898330688476562,
      "learning_rate": 8.29349001908358e-06,
      "loss": 1.2534,
      "step": 4086
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8520021438598633,
      "learning_rate": 8.292616147299763e-06,
      "loss": 0.231,
      "step": 4087
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2977001667022705,
      "learning_rate": 8.291742097891768e-06,
      "loss": 0.099,
      "step": 4088
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9007296562194824,
      "learning_rate": 8.290867870906746e-06,
      "loss": 0.1833,
      "step": 4089
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.1599082946777344,
      "learning_rate": 8.28999346639186e-06,
      "loss": 0.2162,
      "step": 4090
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.709149122238159,
      "learning_rate": 8.289118884394279e-06,
      "loss": 0.8817,
      "step": 4091
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2353549003601074,
      "learning_rate": 8.288244124961182e-06,
      "loss": 0.2048,
      "step": 4092
    },
    {
      "epoch": 0.29,
      "grad_norm": 14.628946304321289,
      "learning_rate": 8.287369188139765e-06,
      "loss": 1.222,
      "step": 4093
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8553833961486816,
      "learning_rate": 8.286494073977221e-06,
      "loss": 0.2021,
      "step": 4094
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.527453422546387,
      "learning_rate": 8.285618782520763e-06,
      "loss": 1.0632,
      "step": 4095
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.045003890991211,
      "learning_rate": 8.284743313817609e-06,
      "loss": 1.1686,
      "step": 4096
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.9658122062683105,
      "learning_rate": 8.283867667914986e-06,
      "loss": 0.9184,
      "step": 4097
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9158501625061035,
      "learning_rate": 8.28299184486013e-06,
      "loss": 0.136,
      "step": 4098
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2039828300476074,
      "learning_rate": 8.282115844700296e-06,
      "loss": 0.2172,
      "step": 4099
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.983036041259766,
      "learning_rate": 8.281239667482733e-06,
      "loss": 1.3263,
      "step": 4100
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.161806106567383,
      "learning_rate": 8.280363313254708e-06,
      "loss": 1.4234,
      "step": 4101
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.0074172019958496,
      "learning_rate": 8.279486782063502e-06,
      "loss": 0.1559,
      "step": 4102
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.637517929077148,
      "learning_rate": 8.278610073956396e-06,
      "loss": 0.7826,
      "step": 4103
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.269408226013184,
      "learning_rate": 8.277733188980688e-06,
      "loss": 0.9713,
      "step": 4104
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.7471654415130615,
      "learning_rate": 8.276856127183681e-06,
      "loss": 0.2659,
      "step": 4105
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.77639627456665,
      "learning_rate": 8.275978888612688e-06,
      "loss": 0.8053,
      "step": 4106
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.278784990310669,
      "learning_rate": 8.275101473315037e-06,
      "loss": 0.1381,
      "step": 4107
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.202277421951294,
      "learning_rate": 8.274223881338057e-06,
      "loss": 0.1705,
      "step": 4108
    },
    {
      "epoch": 0.29,
      "grad_norm": 16.72820281982422,
      "learning_rate": 8.27334611272909e-06,
      "loss": 1.1683,
      "step": 4109
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1243457794189453,
      "learning_rate": 8.272468167535494e-06,
      "loss": 0.0917,
      "step": 4110
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.994150161743164,
      "learning_rate": 8.271590045804624e-06,
      "loss": 0.7567,
      "step": 4111
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7547619342803955,
      "learning_rate": 8.270711747583858e-06,
      "loss": 0.1347,
      "step": 4112
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.5286271572113037,
      "learning_rate": 8.269833272920572e-06,
      "loss": 0.178,
      "step": 4113
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8157010078430176,
      "learning_rate": 8.268954621862158e-06,
      "loss": 0.0691,
      "step": 4114
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.1855602264404297,
      "learning_rate": 8.268075794456016e-06,
      "loss": 0.2305,
      "step": 4115
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.119885444641113,
      "learning_rate": 8.267196790749555e-06,
      "loss": 0.4839,
      "step": 4116
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5895559787750244,
      "learning_rate": 8.266317610790194e-06,
      "loss": 0.1133,
      "step": 4117
    },
    {
      "epoch": 0.3,
      "grad_norm": 14.040252685546875,
      "learning_rate": 8.265438254625362e-06,
      "loss": 1.3811,
      "step": 4118
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.236719846725464,
      "learning_rate": 8.264558722302496e-06,
      "loss": 0.2238,
      "step": 4119
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6467379331588745,
      "learning_rate": 8.263679013869044e-06,
      "loss": 0.094,
      "step": 4120
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7660231590270996,
      "learning_rate": 8.262799129372464e-06,
      "loss": 0.0609,
      "step": 4121
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.218109607696533,
      "learning_rate": 8.261919068860221e-06,
      "loss": 0.1518,
      "step": 4122
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.930939197540283,
      "learning_rate": 8.26103883237979e-06,
      "loss": 0.9103,
      "step": 4123
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.78514575958252,
      "learning_rate": 8.260158419978659e-06,
      "loss": 1.1493,
      "step": 4124
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.0354464054107666,
      "learning_rate": 8.25927783170432e-06,
      "loss": 0.1952,
      "step": 4125
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.6705920696258545,
      "learning_rate": 8.25839706760428e-06,
      "loss": 0.2178,
      "step": 4126
    },
    {
      "epoch": 0.3,
      "grad_norm": 16.713869094848633,
      "learning_rate": 8.257516127726054e-06,
      "loss": 1.5798,
      "step": 4127
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.0712480545043945,
      "learning_rate": 8.25663501211716e-06,
      "loss": 0.1646,
      "step": 4128
    },
    {
      "epoch": 0.3,
      "grad_norm": 15.217180252075195,
      "learning_rate": 8.255753720825134e-06,
      "loss": 1.0966,
      "step": 4129
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.271543264389038,
      "learning_rate": 8.25487225389752e-06,
      "loss": 0.1663,
      "step": 4130
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.012160778045654,
      "learning_rate": 8.25399061138187e-06,
      "loss": 0.7412,
      "step": 4131
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.550637245178223,
      "learning_rate": 8.253108793325742e-06,
      "loss": 1.4059,
      "step": 4132
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.01666259765625,
      "learning_rate": 8.252226799776706e-06,
      "loss": 0.2405,
      "step": 4133
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.9662563800811768,
      "learning_rate": 8.251344630782348e-06,
      "loss": 0.1085,
      "step": 4134
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.9007318019866943,
      "learning_rate": 8.250462286390253e-06,
      "loss": 0.1229,
      "step": 4135
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.665828227996826,
      "learning_rate": 8.24957976664802e-06,
      "loss": 1.014,
      "step": 4136
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.960903167724609,
      "learning_rate": 8.24869707160326e-06,
      "loss": 0.9186,
      "step": 4137
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9886281490325928,
      "learning_rate": 8.24781420130359e-06,
      "loss": 0.0713,
      "step": 4138
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.1405606269836426,
      "learning_rate": 8.246931155796637e-06,
      "loss": 0.1023,
      "step": 4139
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.7274200916290283,
      "learning_rate": 8.24604793513004e-06,
      "loss": 0.1226,
      "step": 4140
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.806622266769409,
      "learning_rate": 8.245164539351443e-06,
      "loss": 0.2639,
      "step": 4141
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.912792205810547,
      "learning_rate": 8.2442809685085e-06,
      "loss": 0.1322,
      "step": 4142
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.623866558074951,
      "learning_rate": 8.243397222648883e-06,
      "loss": 0.2208,
      "step": 4143
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.29396915435791,
      "learning_rate": 8.24251330182026e-06,
      "loss": 0.5978,
      "step": 4144
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.634528398513794,
      "learning_rate": 8.24162920607032e-06,
      "loss": 0.1797,
      "step": 4145
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.142045021057129,
      "learning_rate": 8.240744935446752e-06,
      "loss": 1.2616,
      "step": 4146
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6761724948883057,
      "learning_rate": 8.239860489997265e-06,
      "loss": 0.0422,
      "step": 4147
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.251199722290039,
      "learning_rate": 8.238975869769565e-06,
      "loss": 0.0843,
      "step": 4148
    },
    {
      "epoch": 0.3,
      "grad_norm": 17.430103302001953,
      "learning_rate": 8.23809107481138e-06,
      "loss": 1.9147,
      "step": 4149
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.292757034301758,
      "learning_rate": 8.237206105170436e-06,
      "loss": 0.938,
      "step": 4150
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.831773042678833,
      "learning_rate": 8.236320960894478e-06,
      "loss": 0.1651,
      "step": 4151
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.275128364562988,
      "learning_rate": 8.235435642031252e-06,
      "loss": 0.6981,
      "step": 4152
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.72672438621521,
      "learning_rate": 8.234550148628521e-06,
      "loss": 0.0846,
      "step": 4153
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.331655502319336,
      "learning_rate": 8.233664480734054e-06,
      "loss": 0.1991,
      "step": 4154
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.015830993652344,
      "learning_rate": 8.232778638395627e-06,
      "loss": 0.7594,
      "step": 4155
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.9596266746521,
      "learning_rate": 8.231892621661029e-06,
      "loss": 1.1686,
      "step": 4156
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.595189094543457,
      "learning_rate": 8.231006430578058e-06,
      "loss": 0.8195,
      "step": 4157
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8277082443237305,
      "learning_rate": 8.23012006519452e-06,
      "loss": 0.2304,
      "step": 4158
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.109856128692627,
      "learning_rate": 8.22923352555823e-06,
      "loss": 0.1712,
      "step": 4159
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.575376510620117,
      "learning_rate": 8.228346811717018e-06,
      "loss": 0.1275,
      "step": 4160
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.4329915046691895,
      "learning_rate": 8.227459923718713e-06,
      "loss": 0.2204,
      "step": 4161
    },
    {
      "epoch": 0.3,
      "grad_norm": 19.538902282714844,
      "learning_rate": 8.226572861611162e-06,
      "loss": 0.9848,
      "step": 4162
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.6515660285949707,
      "learning_rate": 8.225685625442217e-06,
      "loss": 0.1049,
      "step": 4163
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.995394706726074,
      "learning_rate": 8.224798215259743e-06,
      "loss": 0.8186,
      "step": 4164
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4807026386260986,
      "learning_rate": 8.223910631111613e-06,
      "loss": 0.121,
      "step": 4165
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.852132320404053,
      "learning_rate": 8.223022873045708e-06,
      "loss": 0.9328,
      "step": 4166
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.3420262336730957,
      "learning_rate": 8.222134941109919e-06,
      "loss": 0.125,
      "step": 4167
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.060938835144043,
      "learning_rate": 8.221246835352145e-06,
      "loss": 0.2608,
      "step": 4168
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.995333194732666,
      "learning_rate": 8.220358555820296e-06,
      "loss": 0.9026,
      "step": 4169
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.3864521980285645,
      "learning_rate": 8.219470102562294e-06,
      "loss": 0.7287,
      "step": 4170
    },
    {
      "epoch": 0.3,
      "grad_norm": 16.081377029418945,
      "learning_rate": 8.218581475626067e-06,
      "loss": 1.023,
      "step": 4171
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.442086219787598,
      "learning_rate": 8.217692675059552e-06,
      "loss": 0.9666,
      "step": 4172
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.624505996704102,
      "learning_rate": 8.216803700910698e-06,
      "loss": 1.1399,
      "step": 4173
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.79641056060791,
      "learning_rate": 8.215914553227459e-06,
      "loss": 0.7244,
      "step": 4174
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.557767868041992,
      "learning_rate": 8.215025232057803e-06,
      "loss": 0.6842,
      "step": 4175
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8576366901397705,
      "learning_rate": 8.214135737449704e-06,
      "loss": 0.1613,
      "step": 4176
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.227459907531738,
      "learning_rate": 8.21324606945115e-06,
      "loss": 0.6294,
      "step": 4177
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.46561050415039,
      "learning_rate": 8.212356228110134e-06,
      "loss": 0.8852,
      "step": 4178
    },
    {
      "epoch": 0.3,
      "grad_norm": 18.128637313842773,
      "learning_rate": 8.211466213474657e-06,
      "loss": 1.843,
      "step": 4179
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.16457462310791,
      "learning_rate": 8.210576025592736e-06,
      "loss": 0.8377,
      "step": 4180
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.464289426803589,
      "learning_rate": 8.20968566451239e-06,
      "loss": 0.2028,
      "step": 4181
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.340783596038818,
      "learning_rate": 8.208795130281651e-06,
      "loss": 0.9358,
      "step": 4182
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.276045322418213,
      "learning_rate": 8.20790442294856e-06,
      "loss": 1.1153,
      "step": 4183
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.324042558670044,
      "learning_rate": 8.207013542561171e-06,
      "loss": 0.1382,
      "step": 4184
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.580596923828125,
      "learning_rate": 8.206122489167537e-06,
      "loss": 0.9325,
      "step": 4185
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.516736030578613,
      "learning_rate": 8.205231262815733e-06,
      "loss": 0.8037,
      "step": 4186
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.343850135803223,
      "learning_rate": 8.204339863553834e-06,
      "loss": 1.1277,
      "step": 4187
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.919084072113037,
      "learning_rate": 8.203448291429927e-06,
      "loss": 0.8365,
      "step": 4188
    },
    {
      "epoch": 0.3,
      "grad_norm": 14.718472480773926,
      "learning_rate": 8.202556546492113e-06,
      "loss": 1.2326,
      "step": 4189
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2195441722869873,
      "learning_rate": 8.201664628788494e-06,
      "loss": 0.0652,
      "step": 4190
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.217757701873779,
      "learning_rate": 8.200772538367188e-06,
      "loss": 0.8116,
      "step": 4191
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.30385422706604,
      "learning_rate": 8.199880275276319e-06,
      "loss": 0.1235,
      "step": 4192
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.175147294998169,
      "learning_rate": 8.19898783956402e-06,
      "loss": 0.1839,
      "step": 4193
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.211395502090454,
      "learning_rate": 8.198095231278438e-06,
      "loss": 0.1972,
      "step": 4194
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.6297760009765625,
      "learning_rate": 8.197202450467723e-06,
      "loss": 0.2058,
      "step": 4195
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.3818857669830322,
      "learning_rate": 8.196309497180038e-06,
      "loss": 0.2018,
      "step": 4196
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3591017723083496,
      "learning_rate": 8.195416371463555e-06,
      "loss": 0.135,
      "step": 4197
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.462413787841797,
      "learning_rate": 8.194523073366452e-06,
      "loss": 1.0949,
      "step": 4198
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.7321627140045166,
      "learning_rate": 8.193629602936922e-06,
      "loss": 0.1387,
      "step": 4199
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.500476837158203,
      "learning_rate": 8.192735960223164e-06,
      "loss": 0.6198,
      "step": 4200
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.4077532291412354,
      "learning_rate": 8.191842145273386e-06,
      "loss": 0.1127,
      "step": 4201
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.078341484069824,
      "learning_rate": 8.190948158135806e-06,
      "loss": 0.1242,
      "step": 4202
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.919137001037598,
      "learning_rate": 8.19005399885865e-06,
      "loss": 1.0998,
      "step": 4203
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.330915927886963,
      "learning_rate": 8.189159667490159e-06,
      "loss": 0.1741,
      "step": 4204
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.148736000061035,
      "learning_rate": 8.188265164078573e-06,
      "loss": 0.1026,
      "step": 4205
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.9570183753967285,
      "learning_rate": 8.187370488672149e-06,
      "loss": 0.769,
      "step": 4206
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.557243824005127,
      "learning_rate": 8.186475641319153e-06,
      "loss": 0.5092,
      "step": 4207
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.142472505569458,
      "learning_rate": 8.18558062206786e-06,
      "loss": 0.1604,
      "step": 4208
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.553647994995117,
      "learning_rate": 8.184685430966547e-06,
      "loss": 0.1362,
      "step": 4209
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1496095657348633,
      "learning_rate": 8.18379006806351e-06,
      "loss": 0.0621,
      "step": 4210
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.657137870788574,
      "learning_rate": 8.182894533407051e-06,
      "loss": 0.6991,
      "step": 4211
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.485966682434082,
      "learning_rate": 8.18199882704548e-06,
      "loss": 0.5821,
      "step": 4212
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.174439430236816,
      "learning_rate": 8.181102949027116e-06,
      "loss": 1.2038,
      "step": 4213
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.853670597076416,
      "learning_rate": 8.180206899400291e-06,
      "loss": 0.1413,
      "step": 4214
    },
    {
      "epoch": 0.3,
      "grad_norm": 24.122610092163086,
      "learning_rate": 8.17931067821334e-06,
      "loss": 1.59,
      "step": 4215
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.7012128829956055,
      "learning_rate": 8.178414285514613e-06,
      "loss": 0.858,
      "step": 4216
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8231801986694336,
      "learning_rate": 8.177517721352464e-06,
      "loss": 0.166,
      "step": 4217
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.240896224975586,
      "learning_rate": 8.176620985775264e-06,
      "loss": 0.8876,
      "step": 4218
    },
    {
      "epoch": 0.3,
      "grad_norm": 19.576326370239258,
      "learning_rate": 8.175724078831386e-06,
      "loss": 1.2749,
      "step": 4219
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.423359870910645,
      "learning_rate": 8.174827000569214e-06,
      "loss": 1.2914,
      "step": 4220
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.614795684814453,
      "learning_rate": 8.173929751037142e-06,
      "loss": 0.8957,
      "step": 4221
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.981671333312988,
      "learning_rate": 8.173032330283574e-06,
      "loss": 0.8806,
      "step": 4222
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.553495168685913,
      "learning_rate": 8.172134738356923e-06,
      "loss": 0.1299,
      "step": 4223
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.123672485351562,
      "learning_rate": 8.17123697530561e-06,
      "loss": 0.8726,
      "step": 4224
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.441058397293091,
      "learning_rate": 8.170339041178067e-06,
      "loss": 0.2274,
      "step": 4225
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.5669236183166504,
      "learning_rate": 8.169440936022731e-06,
      "loss": 0.1336,
      "step": 4226
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8861331939697266,
      "learning_rate": 8.168542659888054e-06,
      "loss": 0.1789,
      "step": 4227
    },
    {
      "epoch": 0.3,
      "grad_norm": 18.375667572021484,
      "learning_rate": 8.167644212822495e-06,
      "loss": 0.8645,
      "step": 4228
    },
    {
      "epoch": 0.3,
      "grad_norm": 17.229217529296875,
      "learning_rate": 8.166745594874521e-06,
      "loss": 1.4336,
      "step": 4229
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.5253446102142334,
      "learning_rate": 8.165846806092607e-06,
      "loss": 0.1203,
      "step": 4230
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.263700485229492,
      "learning_rate": 8.164947846525243e-06,
      "loss": 0.2739,
      "step": 4231
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.657198190689087,
      "learning_rate": 8.164048716220922e-06,
      "loss": 0.1661,
      "step": 4232
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.872667074203491,
      "learning_rate": 8.163149415228152e-06,
      "loss": 0.1659,
      "step": 4233
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.03263258934021,
      "learning_rate": 8.162249943595439e-06,
      "loss": 0.2314,
      "step": 4234
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.3557562828063965,
      "learning_rate": 8.161350301371315e-06,
      "loss": 0.2149,
      "step": 4235
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.481296539306641,
      "learning_rate": 8.160450488604307e-06,
      "loss": 0.1828,
      "step": 4236
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.145288944244385,
      "learning_rate": 8.159550505342959e-06,
      "loss": 0.7778,
      "step": 4237
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.9596943855285645,
      "learning_rate": 8.158650351635821e-06,
      "loss": 0.1505,
      "step": 4238
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.890303134918213,
      "learning_rate": 8.157750027531452e-06,
      "loss": 0.1172,
      "step": 4239
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.4465746879577637,
      "learning_rate": 8.156849533078422e-06,
      "loss": 0.2285,
      "step": 4240
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.487563133239746,
      "learning_rate": 8.15594886832531e-06,
      "loss": 0.0935,
      "step": 4241
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8012101650238037,
      "learning_rate": 8.155048033320701e-06,
      "loss": 0.1362,
      "step": 4242
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.811694622039795,
      "learning_rate": 8.154147028113194e-06,
      "loss": 0.8046,
      "step": 4243
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.7591793537139893,
      "learning_rate": 8.153245852751394e-06,
      "loss": 0.1734,
      "step": 4244
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.950852394104004,
      "learning_rate": 8.152344507283914e-06,
      "loss": 0.234,
      "step": 4245
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.415546894073486,
      "learning_rate": 8.151442991759381e-06,
      "loss": 0.1687,
      "step": 4246
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.955677509307861,
      "learning_rate": 8.150541306226429e-06,
      "loss": 1.0059,
      "step": 4247
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.016944885253906,
      "learning_rate": 8.149639450733698e-06,
      "loss": 0.803,
      "step": 4248
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.202603578567505,
      "learning_rate": 8.14873742532984e-06,
      "loss": 0.1434,
      "step": 4249
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.7052676677703857,
      "learning_rate": 8.14783523006352e-06,
      "loss": 0.1553,
      "step": 4250
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.0969510078430176,
      "learning_rate": 8.1469328649834e-06,
      "loss": 0.161,
      "step": 4251
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.17032527923584,
      "learning_rate": 8.146030330138166e-06,
      "loss": 1.1327,
      "step": 4252
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.109394550323486,
      "learning_rate": 8.145127625576502e-06,
      "loss": 1.0106,
      "step": 4253
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7381575107574463,
      "learning_rate": 8.14422475134711e-06,
      "loss": 0.0811,
      "step": 4254
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.506488800048828,
      "learning_rate": 8.143321707498692e-06,
      "loss": 0.172,
      "step": 4255
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7516915798187256,
      "learning_rate": 8.142418494079969e-06,
      "loss": 0.1107,
      "step": 4256
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.78156852722168,
      "learning_rate": 8.14151511113966e-06,
      "loss": 1.236,
      "step": 4257
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0635509490966797,
      "learning_rate": 8.140611558726503e-06,
      "loss": 0.1244,
      "step": 4258
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.6446995735168457,
      "learning_rate": 8.139707836889242e-06,
      "loss": 0.1279,
      "step": 4259
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.2487339973449707,
      "learning_rate": 8.138803945676626e-06,
      "loss": 0.2514,
      "step": 4260
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.798912763595581,
      "learning_rate": 8.137899885137417e-06,
      "loss": 0.2259,
      "step": 4261
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.274639368057251,
      "learning_rate": 8.136995655320389e-06,
      "loss": 0.2051,
      "step": 4262
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.688298225402832,
      "learning_rate": 8.13609125627432e-06,
      "loss": 1.214,
      "step": 4263
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0643792152404785,
      "learning_rate": 8.135186688047995e-06,
      "loss": 0.1219,
      "step": 4264
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.923439979553223,
      "learning_rate": 8.134281950690218e-06,
      "loss": 1.1059,
      "step": 4265
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.706815719604492,
      "learning_rate": 8.133377044249793e-06,
      "loss": 0.5904,
      "step": 4266
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.146839141845703,
      "learning_rate": 8.132471968775537e-06,
      "loss": 0.1137,
      "step": 4267
    },
    {
      "epoch": 0.31,
      "grad_norm": 23.546796798706055,
      "learning_rate": 8.131566724316276e-06,
      "loss": 1.1232,
      "step": 4268
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.763113021850586,
      "learning_rate": 8.130661310920843e-06,
      "loss": 1.2957,
      "step": 4269
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1412272453308105,
      "learning_rate": 8.129755728638082e-06,
      "loss": 0.0543,
      "step": 4270
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.9691531658172607,
      "learning_rate": 8.128849977516848e-06,
      "loss": 0.157,
      "step": 4271
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.782063484191895,
      "learning_rate": 8.127944057606e-06,
      "loss": 0.9441,
      "step": 4272
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.269124507904053,
      "learning_rate": 8.12703796895441e-06,
      "loss": 0.8061,
      "step": 4273
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3913676738739014,
      "learning_rate": 8.126131711610958e-06,
      "loss": 0.1088,
      "step": 4274
    },
    {
      "epoch": 0.31,
      "grad_norm": 11.120064735412598,
      "learning_rate": 8.125225285624534e-06,
      "loss": 1.2231,
      "step": 4275
    },
    {
      "epoch": 0.31,
      "grad_norm": 18.382089614868164,
      "learning_rate": 8.124318691044036e-06,
      "loss": 0.9947,
      "step": 4276
    },
    {
      "epoch": 0.31,
      "grad_norm": 25.22840118408203,
      "learning_rate": 8.123411927918369e-06,
      "loss": 2.3059,
      "step": 4277
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4676287174224854,
      "learning_rate": 8.122504996296453e-06,
      "loss": 0.2656,
      "step": 4278
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.606041669845581,
      "learning_rate": 8.121597896227208e-06,
      "loss": 0.1158,
      "step": 4279
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.329864501953125,
      "learning_rate": 8.120690627759576e-06,
      "loss": 0.157,
      "step": 4280
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.715263366699219,
      "learning_rate": 8.119783190942497e-06,
      "loss": 0.5608,
      "step": 4281
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.296446800231934,
      "learning_rate": 8.118875585824923e-06,
      "loss": 0.713,
      "step": 4282
    },
    {
      "epoch": 0.31,
      "grad_norm": 11.346074104309082,
      "learning_rate": 8.117967812455816e-06,
      "loss": 0.9375,
      "step": 4283
    },
    {
      "epoch": 0.31,
      "grad_norm": 20.365734100341797,
      "learning_rate": 8.117059870884148e-06,
      "loss": 1.6891,
      "step": 4284
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.529841423034668,
      "learning_rate": 8.116151761158898e-06,
      "loss": 1.1596,
      "step": 4285
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7286518812179565,
      "learning_rate": 8.115243483329059e-06,
      "loss": 0.0415,
      "step": 4286
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.234091758728027,
      "learning_rate": 8.114335037443624e-06,
      "loss": 0.7476,
      "step": 4287
    },
    {
      "epoch": 0.31,
      "grad_norm": 11.905817031860352,
      "learning_rate": 8.113426423551602e-06,
      "loss": 0.9354,
      "step": 4288
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0245378017425537,
      "learning_rate": 8.112517641702009e-06,
      "loss": 0.1601,
      "step": 4289
    },
    {
      "epoch": 0.31,
      "grad_norm": 23.094484329223633,
      "learning_rate": 8.111608691943872e-06,
      "loss": 1.0248,
      "step": 4290
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.270027160644531,
      "learning_rate": 8.110699574326222e-06,
      "loss": 1.033,
      "step": 4291
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.523574829101562,
      "learning_rate": 8.109790288898108e-06,
      "loss": 1.3353,
      "step": 4292
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0352048873901367,
      "learning_rate": 8.108880835708574e-06,
      "loss": 0.1728,
      "step": 4293
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.957057476043701,
      "learning_rate": 8.107971214806691e-06,
      "loss": 0.1909,
      "step": 4294
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.8956751823425293,
      "learning_rate": 8.107061426241523e-06,
      "loss": 0.2831,
      "step": 4295
    },
    {
      "epoch": 0.31,
      "grad_norm": 14.094900131225586,
      "learning_rate": 8.106151470062154e-06,
      "loss": 1.4276,
      "step": 4296
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.215487003326416,
      "learning_rate": 8.105241346317668e-06,
      "loss": 0.1159,
      "step": 4297
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9621883630752563,
      "learning_rate": 8.104331055057168e-06,
      "loss": 0.0829,
      "step": 4298
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.47582745552063,
      "learning_rate": 8.103420596329756e-06,
      "loss": 0.1547,
      "step": 4299
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.580503463745117,
      "learning_rate": 8.102509970184552e-06,
      "loss": 0.7166,
      "step": 4300
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.668515205383301,
      "learning_rate": 8.101599176670676e-06,
      "loss": 0.2275,
      "step": 4301
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4782395362854004,
      "learning_rate": 8.100688215837267e-06,
      "loss": 0.1807,
      "step": 4302
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.707925796508789,
      "learning_rate": 8.099777087733464e-06,
      "loss": 0.0918,
      "step": 4303
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.9077377319335938,
      "learning_rate": 8.098865792408422e-06,
      "loss": 0.824,
      "step": 4304
    },
    {
      "epoch": 0.31,
      "grad_norm": 17.555545806884766,
      "learning_rate": 8.0979543299113e-06,
      "loss": 0.7692,
      "step": 4305
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.828606605529785,
      "learning_rate": 8.097042700291266e-06,
      "loss": 1.2508,
      "step": 4306
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4599483013153076,
      "learning_rate": 8.096130903597503e-06,
      "loss": 0.2716,
      "step": 4307
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.107448577880859,
      "learning_rate": 8.095218939879198e-06,
      "loss": 0.9145,
      "step": 4308
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.1714401245117188,
      "learning_rate": 8.094306809185546e-06,
      "loss": 0.192,
      "step": 4309
    },
    {
      "epoch": 0.31,
      "grad_norm": 11.909939765930176,
      "learning_rate": 8.093394511565755e-06,
      "loss": 1.5253,
      "step": 4310
    },
    {
      "epoch": 0.31,
      "grad_norm": 21.333520889282227,
      "learning_rate": 8.09248204706904e-06,
      "loss": 1.379,
      "step": 4311
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.2641735076904297,
      "learning_rate": 8.091569415744622e-06,
      "loss": 0.1694,
      "step": 4312
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.895337104797363,
      "learning_rate": 8.090656617641738e-06,
      "loss": 0.9924,
      "step": 4313
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.869619369506836,
      "learning_rate": 8.089743652809626e-06,
      "loss": 1.3502,
      "step": 4314
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.857149124145508,
      "learning_rate": 8.088830521297542e-06,
      "loss": 0.9126,
      "step": 4315
    },
    {
      "epoch": 0.31,
      "grad_norm": 18.888416290283203,
      "learning_rate": 8.087917223154742e-06,
      "loss": 1.3115,
      "step": 4316
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.257673501968384,
      "learning_rate": 8.087003758430496e-06,
      "loss": 0.209,
      "step": 4317
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.6298234462738037,
      "learning_rate": 8.086090127174082e-06,
      "loss": 0.1431,
      "step": 4318
    },
    {
      "epoch": 0.31,
      "grad_norm": 17.51776123046875,
      "learning_rate": 8.085176329434787e-06,
      "loss": 0.9372,
      "step": 4319
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3203577995300293,
      "learning_rate": 8.084262365261907e-06,
      "loss": 0.0894,
      "step": 4320
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.295899391174316,
      "learning_rate": 8.083348234704748e-06,
      "loss": 0.8484,
      "step": 4321
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.076310634613037,
      "learning_rate": 8.082433937812624e-06,
      "loss": 0.8484,
      "step": 4322
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.393707752227783,
      "learning_rate": 8.081519474634855e-06,
      "loss": 0.1275,
      "step": 4323
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.580345153808594,
      "learning_rate": 8.080604845220774e-06,
      "loss": 1.243,
      "step": 4324
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.195770740509033,
      "learning_rate": 8.079690049619725e-06,
      "loss": 0.8172,
      "step": 4325
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.696931838989258,
      "learning_rate": 8.078775087881052e-06,
      "loss": 0.8396,
      "step": 4326
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.4885101318359375,
      "learning_rate": 8.077859960054118e-06,
      "loss": 0.135,
      "step": 4327
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.848787546157837,
      "learning_rate": 8.076944666188288e-06,
      "loss": 0.1136,
      "step": 4328
    },
    {
      "epoch": 0.31,
      "grad_norm": 20.533260345458984,
      "learning_rate": 8.076029206332943e-06,
      "loss": 1.446,
      "step": 4329
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.176753520965576,
      "learning_rate": 8.075113580537463e-06,
      "loss": 0.2124,
      "step": 4330
    },
    {
      "epoch": 0.31,
      "grad_norm": 14.517207145690918,
      "learning_rate": 8.074197788851247e-06,
      "loss": 0.9126,
      "step": 4331
    },
    {
      "epoch": 0.31,
      "grad_norm": 19.115636825561523,
      "learning_rate": 8.073281831323695e-06,
      "loss": 1.2016,
      "step": 4332
    },
    {
      "epoch": 0.31,
      "grad_norm": 14.34183120727539,
      "learning_rate": 8.072365708004224e-06,
      "loss": 1.1754,
      "step": 4333
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.9335715770721436,
      "learning_rate": 8.071449418942251e-06,
      "loss": 0.2156,
      "step": 4334
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.4142351150512695,
      "learning_rate": 8.070532964187208e-06,
      "loss": 0.9642,
      "step": 4335
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.9206316471099854,
      "learning_rate": 8.069616343788534e-06,
      "loss": 0.1536,
      "step": 4336
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.2012367248535156,
      "learning_rate": 8.068699557795677e-06,
      "loss": 0.1254,
      "step": 4337
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.46402359008789,
      "learning_rate": 8.067782606258094e-06,
      "loss": 0.9264,
      "step": 4338
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.003963947296143,
      "learning_rate": 8.066865489225254e-06,
      "loss": 0.7963,
      "step": 4339
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.520467281341553,
      "learning_rate": 8.065948206746627e-06,
      "loss": 0.9396,
      "step": 4340
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.185277223587036,
      "learning_rate": 8.065030758871702e-06,
      "loss": 0.3231,
      "step": 4341
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.156391143798828,
      "learning_rate": 8.064113145649968e-06,
      "loss": 0.2419,
      "step": 4342
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5832290649414062,
      "learning_rate": 8.063195367130928e-06,
      "loss": 0.1287,
      "step": 4343
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.021773099899292,
      "learning_rate": 8.062277423364094e-06,
      "loss": 0.183,
      "step": 4344
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.5913889408111572,
      "learning_rate": 8.061359314398982e-06,
      "loss": 0.1648,
      "step": 4345
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.2760229110717773,
      "learning_rate": 8.060441040285125e-06,
      "loss": 0.1409,
      "step": 4346
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.141303062438965,
      "learning_rate": 8.059522601072057e-06,
      "loss": 0.1586,
      "step": 4347
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.783326625823975,
      "learning_rate": 8.058603996809325e-06,
      "loss": 0.8989,
      "step": 4348
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.2458372116088867,
      "learning_rate": 8.057685227546487e-06,
      "loss": 0.1507,
      "step": 4349
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.8707780838012695,
      "learning_rate": 8.056766293333103e-06,
      "loss": 0.8993,
      "step": 4350
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.425694465637207,
      "learning_rate": 8.05584719421875e-06,
      "loss": 0.1603,
      "step": 4351
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0589404106140137,
      "learning_rate": 8.054927930253005e-06,
      "loss": 0.1912,
      "step": 4352
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.336133003234863,
      "learning_rate": 8.054008501485465e-06,
      "loss": 0.7278,
      "step": 4353
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.9287500381469727,
      "learning_rate": 8.053088907965722e-06,
      "loss": 0.3114,
      "step": 4354
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.670742034912109,
      "learning_rate": 8.052169149743393e-06,
      "loss": 1.0883,
      "step": 4355
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.901339530944824,
      "learning_rate": 8.051249226868089e-06,
      "loss": 1.44,
      "step": 4356
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.415398597717285,
      "learning_rate": 8.050329139389442e-06,
      "loss": 0.9679,
      "step": 4357
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.717487096786499,
      "learning_rate": 8.04940888735708e-06,
      "loss": 0.0713,
      "step": 4358
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.03978157043457,
      "learning_rate": 8.048488470820655e-06,
      "loss": 1.2243,
      "step": 4359
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.005053997039795,
      "learning_rate": 8.047567889829814e-06,
      "loss": 0.0421,
      "step": 4360
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.619483947753906,
      "learning_rate": 8.04664714443422e-06,
      "loss": 1.0064,
      "step": 4361
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.339330673217773,
      "learning_rate": 8.045726234683547e-06,
      "loss": 0.8846,
      "step": 4362
    },
    {
      "epoch": 0.31,
      "grad_norm": 24.822179794311523,
      "learning_rate": 8.04480516062747e-06,
      "loss": 1.6631,
      "step": 4363
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.803126573562622,
      "learning_rate": 8.043883922315683e-06,
      "loss": 0.2094,
      "step": 4364
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.8320608139038086,
      "learning_rate": 8.042962519797877e-06,
      "loss": 0.1542,
      "step": 4365
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.118852376937866,
      "learning_rate": 8.042040953123764e-06,
      "loss": 0.1712,
      "step": 4366
    },
    {
      "epoch": 0.31,
      "grad_norm": 26.78178596496582,
      "learning_rate": 8.041119222343054e-06,
      "loss": 0.685,
      "step": 4367
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.824671745300293,
      "learning_rate": 8.040197327505475e-06,
      "loss": 0.9375,
      "step": 4368
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.5350911617279053,
      "learning_rate": 8.039275268660756e-06,
      "loss": 0.3135,
      "step": 4369
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.153500080108643,
      "learning_rate": 8.038353045858642e-06,
      "loss": 1.1247,
      "step": 4370
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.495451927185059,
      "learning_rate": 8.037430659148882e-06,
      "loss": 1.0055,
      "step": 4371
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.099343776702881,
      "learning_rate": 8.036508108581235e-06,
      "loss": 0.1928,
      "step": 4372
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.853680372238159,
      "learning_rate": 8.03558539420547e-06,
      "loss": 0.1375,
      "step": 4373
    },
    {
      "epoch": 0.31,
      "grad_norm": 14.725759506225586,
      "learning_rate": 8.034662516071361e-06,
      "loss": 0.8994,
      "step": 4374
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.211424350738525,
      "learning_rate": 8.033739474228696e-06,
      "loss": 0.5328,
      "step": 4375
    },
    {
      "epoch": 0.31,
      "grad_norm": 34.4471549987793,
      "learning_rate": 8.032816268727273e-06,
      "loss": 0.8835,
      "step": 4376
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.572580337524414,
      "learning_rate": 8.031892899616891e-06,
      "loss": 1.2263,
      "step": 4377
    },
    {
      "epoch": 0.31,
      "grad_norm": 21.29865837097168,
      "learning_rate": 8.030969366947361e-06,
      "loss": 1.4209,
      "step": 4378
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.193061113357544,
      "learning_rate": 8.030045670768509e-06,
      "loss": 0.2643,
      "step": 4379
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.018740653991699,
      "learning_rate": 8.029121811130163e-06,
      "loss": 0.2024,
      "step": 4380
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.947896003723145,
      "learning_rate": 8.02819778808216e-06,
      "loss": 1.3718,
      "step": 4381
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4822957515716553,
      "learning_rate": 8.027273601674346e-06,
      "loss": 0.1548,
      "step": 4382
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.81882095336914,
      "learning_rate": 8.026349251956585e-06,
      "loss": 1.0571,
      "step": 4383
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7996885776519775,
      "learning_rate": 8.025424738978733e-06,
      "loss": 0.1358,
      "step": 4384
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5913312435150146,
      "learning_rate": 8.02450006279067e-06,
      "loss": 0.1063,
      "step": 4385
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.174457550048828,
      "learning_rate": 8.023575223442278e-06,
      "loss": 0.8823,
      "step": 4386
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.935226440429688,
      "learning_rate": 8.022650220983446e-06,
      "loss": 0.9312,
      "step": 4387
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.272311210632324,
      "learning_rate": 8.021725055464076e-06,
      "loss": 0.2258,
      "step": 4388
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.992941379547119,
      "learning_rate": 8.020799726934079e-06,
      "loss": 0.1461,
      "step": 4389
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5869865417480469,
      "learning_rate": 8.019874235443368e-06,
      "loss": 0.1057,
      "step": 4390
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.664798736572266,
      "learning_rate": 8.018948581041874e-06,
      "loss": 1.1593,
      "step": 4391
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.8767831325531006,
      "learning_rate": 8.018022763779533e-06,
      "loss": 0.1523,
      "step": 4392
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.7407631874084473,
      "learning_rate": 8.017096783706286e-06,
      "loss": 0.8958,
      "step": 4393
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0171351432800293,
      "learning_rate": 8.01617064087209e-06,
      "loss": 0.1141,
      "step": 4394
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.7700259685516357,
      "learning_rate": 8.015244335326901e-06,
      "loss": 0.1416,
      "step": 4395
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.317467212677002,
      "learning_rate": 8.014317867120696e-06,
      "loss": 0.1461,
      "step": 4396
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.0716352462768555,
      "learning_rate": 8.013391236303451e-06,
      "loss": 0.907,
      "step": 4397
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.415306568145752,
      "learning_rate": 8.012464442925156e-06,
      "loss": 0.1521,
      "step": 4398
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.155092239379883,
      "learning_rate": 8.011537487035807e-06,
      "loss": 0.7748,
      "step": 4399
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.528543710708618,
      "learning_rate": 8.01061036868541e-06,
      "loss": 0.159,
      "step": 4400
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.7467451095581055,
      "learning_rate": 8.009683087923979e-06,
      "loss": 0.837,
      "step": 4401
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.273348331451416,
      "learning_rate": 8.008755644801538e-06,
      "loss": 0.3003,
      "step": 4402
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.249424934387207,
      "learning_rate": 8.007828039368118e-06,
      "loss": 0.2816,
      "step": 4403
    },
    {
      "epoch": 0.32,
      "grad_norm": 21.49958038330078,
      "learning_rate": 8.00690027167376e-06,
      "loss": 1.2383,
      "step": 4404
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.18609356880188,
      "learning_rate": 8.005972341768515e-06,
      "loss": 0.1259,
      "step": 4405
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.976595878601074,
      "learning_rate": 8.00504424970244e-06,
      "loss": 1.0814,
      "step": 4406
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.9551494121551514,
      "learning_rate": 8.004115995525605e-06,
      "loss": 0.2217,
      "step": 4407
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.352458953857422,
      "learning_rate": 8.00318757928808e-06,
      "loss": 1.3266,
      "step": 4408
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.948563575744629,
      "learning_rate": 8.002259001039954e-06,
      "loss": 0.884,
      "step": 4409
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.7534291744232178,
      "learning_rate": 8.001330260831319e-06,
      "loss": 0.1111,
      "step": 4410
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.991850852966309,
      "learning_rate": 8.000401358712277e-06,
      "loss": 1.155,
      "step": 4411
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.125394105911255,
      "learning_rate": 7.999472294732938e-06,
      "loss": 0.2024,
      "step": 4412
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.960232734680176,
      "learning_rate": 7.998543068943422e-06,
      "loss": 0.2265,
      "step": 4413
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.117680072784424,
      "learning_rate": 7.99761368139386e-06,
      "loss": 0.6778,
      "step": 4414
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8312768936157227,
      "learning_rate": 7.996684132134384e-06,
      "loss": 0.1977,
      "step": 4415
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.461465358734131,
      "learning_rate": 7.995754421215144e-06,
      "loss": 0.8584,
      "step": 4416
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.448981761932373,
      "learning_rate": 7.99482454868629e-06,
      "loss": 0.2083,
      "step": 4417
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8576858043670654,
      "learning_rate": 7.993894514597991e-06,
      "loss": 0.2406,
      "step": 4418
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.590956211090088,
      "learning_rate": 7.992964319000414e-06,
      "loss": 0.2262,
      "step": 4419
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.913393974304199,
      "learning_rate": 7.992033961943742e-06,
      "loss": 0.2097,
      "step": 4420
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.5040946006774902,
      "learning_rate": 7.99110344347816e-06,
      "loss": 0.165,
      "step": 4421
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5210461616516113,
      "learning_rate": 7.990172763653872e-06,
      "loss": 0.1151,
      "step": 4422
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.844282627105713,
      "learning_rate": 7.989241922521081e-06,
      "loss": 0.4145,
      "step": 4423
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.734329223632812,
      "learning_rate": 7.988310920130002e-06,
      "loss": 1.1796,
      "step": 4424
    },
    {
      "epoch": 0.32,
      "grad_norm": 42.18605422973633,
      "learning_rate": 7.987379756530862e-06,
      "loss": 1.1462,
      "step": 4425
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.506353378295898,
      "learning_rate": 7.986448431773893e-06,
      "loss": 0.9738,
      "step": 4426
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.9086103439331055,
      "learning_rate": 7.985516945909333e-06,
      "loss": 0.7569,
      "step": 4427
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.071929931640625,
      "learning_rate": 7.984585298987435e-06,
      "loss": 0.1491,
      "step": 4428
    },
    {
      "epoch": 0.32,
      "grad_norm": 14.80405330657959,
      "learning_rate": 7.983653491058457e-06,
      "loss": 0.892,
      "step": 4429
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.910019874572754,
      "learning_rate": 7.982721522172667e-06,
      "loss": 0.9896,
      "step": 4430
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3486053943634033,
      "learning_rate": 7.981789392380344e-06,
      "loss": 0.0909,
      "step": 4431
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.1991918087005615,
      "learning_rate": 7.980857101731766e-06,
      "loss": 0.2159,
      "step": 4432
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8050875663757324,
      "learning_rate": 7.979924650277233e-06,
      "loss": 0.2243,
      "step": 4433
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.23966121673584,
      "learning_rate": 7.978992038067044e-06,
      "loss": 0.27,
      "step": 4434
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.749836921691895,
      "learning_rate": 7.97805926515151e-06,
      "loss": 1.314,
      "step": 4435
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0866382122039795,
      "learning_rate": 7.97712633158095e-06,
      "loss": 0.1916,
      "step": 4436
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.097457885742188,
      "learning_rate": 7.976193237405699e-06,
      "loss": 1.0429,
      "step": 4437
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.579172134399414,
      "learning_rate": 7.975259982676083e-06,
      "loss": 0.3948,
      "step": 4438
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.9040651321411133,
      "learning_rate": 7.974326567442454e-06,
      "loss": 0.1962,
      "step": 4439
    },
    {
      "epoch": 0.32,
      "grad_norm": 68.3354721069336,
      "learning_rate": 7.973392991755167e-06,
      "loss": 1.1191,
      "step": 4440
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.689516067504883,
      "learning_rate": 7.97245925566458e-06,
      "loss": 0.9653,
      "step": 4441
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3113560676574707,
      "learning_rate": 7.97152535922107e-06,
      "loss": 0.2598,
      "step": 4442
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.957202911376953,
      "learning_rate": 7.970591302475015e-06,
      "loss": 1.1439,
      "step": 4443
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.7904775142669678,
      "learning_rate": 7.969657085476802e-06,
      "loss": 0.2881,
      "step": 4444
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.156943321228027,
      "learning_rate": 7.96872270827683e-06,
      "loss": 1.2142,
      "step": 4445
    },
    {
      "epoch": 0.32,
      "grad_norm": 17.591787338256836,
      "learning_rate": 7.967788170925505e-06,
      "loss": 1.0023,
      "step": 4446
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8540241718292236,
      "learning_rate": 7.966853473473242e-06,
      "loss": 0.2195,
      "step": 4447
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5844533443450928,
      "learning_rate": 7.965918615970466e-06,
      "loss": 0.1329,
      "step": 4448
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.020275115966797,
      "learning_rate": 7.964983598467606e-06,
      "loss": 0.9269,
      "step": 4449
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8333451747894287,
      "learning_rate": 7.964048421015102e-06,
      "loss": 0.2454,
      "step": 4450
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.705058097839355,
      "learning_rate": 7.963113083663408e-06,
      "loss": 0.498,
      "step": 4451
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.715638160705566,
      "learning_rate": 7.962177586462978e-06,
      "loss": 1.3413,
      "step": 4452
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.588075876235962,
      "learning_rate": 7.961241929464279e-06,
      "loss": 0.1397,
      "step": 4453
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3669967651367188,
      "learning_rate": 7.960306112717788e-06,
      "loss": 0.2936,
      "step": 4454
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.135318756103516,
      "learning_rate": 7.959370136273988e-06,
      "loss": 0.8708,
      "step": 4455
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5308351516723633,
      "learning_rate": 7.95843400018337e-06,
      "loss": 0.1747,
      "step": 4456
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6883084774017334,
      "learning_rate": 7.957497704496436e-06,
      "loss": 0.1875,
      "step": 4457
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.081707954406738,
      "learning_rate": 7.956561249263696e-06,
      "loss": 0.8455,
      "step": 4458
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.70584774017334,
      "learning_rate": 7.955624634535669e-06,
      "loss": 0.8761,
      "step": 4459
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.008850336074829,
      "learning_rate": 7.954687860362877e-06,
      "loss": 0.2331,
      "step": 4460
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.84323501586914,
      "learning_rate": 7.953750926795863e-06,
      "loss": 1.4518,
      "step": 4461
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.587035894393921,
      "learning_rate": 7.952813833885165e-06,
      "loss": 0.1616,
      "step": 4462
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.9827933311462402,
      "learning_rate": 7.951876581681337e-06,
      "loss": 0.1824,
      "step": 4463
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1645421981811523,
      "learning_rate": 7.95093917023494e-06,
      "loss": 0.1177,
      "step": 4464
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.014586448669434,
      "learning_rate": 7.950001599596549e-06,
      "loss": 0.9303,
      "step": 4465
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.871403932571411,
      "learning_rate": 7.949063869816736e-06,
      "loss": 0.2516,
      "step": 4466
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.120675086975098,
      "learning_rate": 7.94812598094609e-06,
      "loss": 0.882,
      "step": 4467
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.045214891433716,
      "learning_rate": 7.947187933035206e-06,
      "loss": 0.2083,
      "step": 4468
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.201459884643555,
      "learning_rate": 7.946249726134687e-06,
      "loss": 0.9408,
      "step": 4469
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.151686668395996,
      "learning_rate": 7.945311360295151e-06,
      "loss": 0.8577,
      "step": 4470
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.518312454223633,
      "learning_rate": 7.944372835567213e-06,
      "loss": 0.7189,
      "step": 4471
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.27713394165039,
      "learning_rate": 7.943434152001506e-06,
      "loss": 0.9133,
      "step": 4472
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.916751861572266,
      "learning_rate": 7.94249530964867e-06,
      "loss": 0.9067,
      "step": 4473
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.789576530456543,
      "learning_rate": 7.941556308559349e-06,
      "loss": 1.078,
      "step": 4474
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.2174112796783447,
      "learning_rate": 7.940617148784198e-06,
      "loss": 0.1512,
      "step": 4475
    },
    {
      "epoch": 0.32,
      "grad_norm": 25.029722213745117,
      "learning_rate": 7.939677830373884e-06,
      "loss": 0.995,
      "step": 4476
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.283576011657715,
      "learning_rate": 7.938738353379079e-06,
      "loss": 0.6358,
      "step": 4477
    },
    {
      "epoch": 0.32,
      "grad_norm": 15.501226425170898,
      "learning_rate": 7.937798717850462e-06,
      "loss": 1.2801,
      "step": 4478
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.311788558959961,
      "learning_rate": 7.936858923838726e-06,
      "loss": 0.8337,
      "step": 4479
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.519695281982422,
      "learning_rate": 7.935918971394566e-06,
      "loss": 1.076,
      "step": 4480
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.616152763366699,
      "learning_rate": 7.934978860568692e-06,
      "loss": 0.1011,
      "step": 4481
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.1619274616241455,
      "learning_rate": 7.934038591411817e-06,
      "loss": 0.2528,
      "step": 4482
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8030195236206055,
      "learning_rate": 7.933098163974667e-06,
      "loss": 0.1624,
      "step": 4483
    },
    {
      "epoch": 0.32,
      "grad_norm": 17.208406448364258,
      "learning_rate": 7.932157578307971e-06,
      "loss": 1.254,
      "step": 4484
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.4726388454437256,
      "learning_rate": 7.931216834462475e-06,
      "loss": 0.248,
      "step": 4485
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.486880302429199,
      "learning_rate": 7.930275932488925e-06,
      "loss": 1.0666,
      "step": 4486
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.053432464599609,
      "learning_rate": 7.929334872438081e-06,
      "loss": 0.8496,
      "step": 4487
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.056798219680786,
      "learning_rate": 7.928393654360708e-06,
      "loss": 0.2627,
      "step": 4488
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.63766622543335,
      "learning_rate": 7.927452278307581e-06,
      "loss": 0.8485,
      "step": 4489
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.5710642337799072,
      "learning_rate": 7.926510744329486e-06,
      "loss": 0.1706,
      "step": 4490
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.666353702545166,
      "learning_rate": 7.925569052477214e-06,
      "loss": 0.1321,
      "step": 4491
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.420896530151367,
      "learning_rate": 7.924627202801564e-06,
      "loss": 0.3147,
      "step": 4492
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.276540756225586,
      "learning_rate": 7.923685195353347e-06,
      "loss": 1.2324,
      "step": 4493
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.4935872554779053,
      "learning_rate": 7.922743030183383e-06,
      "loss": 0.1071,
      "step": 4494
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.440840482711792,
      "learning_rate": 7.921800707342494e-06,
      "loss": 0.1477,
      "step": 4495
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.436941623687744,
      "learning_rate": 7.920858226881515e-06,
      "loss": 1.0919,
      "step": 4496
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.716090202331543,
      "learning_rate": 7.919915588851292e-06,
      "loss": 1.5737,
      "step": 4497
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.41760540008545,
      "learning_rate": 7.918972793302674e-06,
      "loss": 0.7985,
      "step": 4498
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.2811174392700195,
      "learning_rate": 7.918029840286525e-06,
      "loss": 0.1844,
      "step": 4499
    },
    {
      "epoch": 0.32,
      "grad_norm": 17.622751235961914,
      "learning_rate": 7.91708672985371e-06,
      "loss": 1.2244,
      "step": 4500
    },
    {
      "epoch": 0.32,
      "grad_norm": 21.72211265563965,
      "learning_rate": 7.916143462055109e-06,
      "loss": 1.1603,
      "step": 4501
    },
    {
      "epoch": 0.32,
      "grad_norm": 15.414556503295898,
      "learning_rate": 7.915200036941606e-06,
      "loss": 0.838,
      "step": 4502
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0042195320129395,
      "learning_rate": 7.914256454564095e-06,
      "loss": 0.1531,
      "step": 4503
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.295681953430176,
      "learning_rate": 7.913312714973481e-06,
      "loss": 0.1863,
      "step": 4504
    },
    {
      "epoch": 0.32,
      "grad_norm": 14.87752628326416,
      "learning_rate": 7.91236881822067e-06,
      "loss": 1.1793,
      "step": 4505
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.9903814792633057,
      "learning_rate": 7.911424764356588e-06,
      "loss": 0.1521,
      "step": 4506
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.7143189907073975,
      "learning_rate": 7.91048055343216e-06,
      "loss": 0.1993,
      "step": 4507
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8144567012786865,
      "learning_rate": 7.909536185498323e-06,
      "loss": 0.1644,
      "step": 4508
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3096580505371094,
      "learning_rate": 7.908591660606021e-06,
      "loss": 0.1995,
      "step": 4509
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.134422302246094,
      "learning_rate": 7.90764697880621e-06,
      "loss": 1.0293,
      "step": 4510
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.940500497817993,
      "learning_rate": 7.906702140149848e-06,
      "loss": 0.1759,
      "step": 4511
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.951969861984253,
      "learning_rate": 7.90575714468791e-06,
      "loss": 0.0771,
      "step": 4512
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.232682466506958,
      "learning_rate": 7.904811992471373e-06,
      "loss": 0.1507,
      "step": 4513
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3292319774627686,
      "learning_rate": 7.903866683551224e-06,
      "loss": 0.1058,
      "step": 4514
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.08901309967041,
      "learning_rate": 7.902921217978458e-06,
      "loss": 0.203,
      "step": 4515
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.98809051513672,
      "learning_rate": 7.901975595804083e-06,
      "loss": 0.6582,
      "step": 4516
    },
    {
      "epoch": 0.32,
      "grad_norm": 14.82683277130127,
      "learning_rate": 7.901029817079106e-06,
      "loss": 1.3517,
      "step": 4517
    },
    {
      "epoch": 0.32,
      "grad_norm": 18.31149673461914,
      "learning_rate": 7.900083881854554e-06,
      "loss": 0.8392,
      "step": 4518
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.340227127075195,
      "learning_rate": 7.899137790181453e-06,
      "loss": 0.8407,
      "step": 4519
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.3172712326049805,
      "learning_rate": 7.898191542110841e-06,
      "loss": 0.9968,
      "step": 4520
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.417423248291016,
      "learning_rate": 7.897245137693768e-06,
      "loss": 1.1745,
      "step": 4521
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6596436500549316,
      "learning_rate": 7.896298576981285e-06,
      "loss": 0.1109,
      "step": 4522
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.2761008739471436,
      "learning_rate": 7.895351860024456e-06,
      "loss": 0.1464,
      "step": 4523
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.162915229797363,
      "learning_rate": 7.894404986874356e-06,
      "loss": 0.9224,
      "step": 4524
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1122071743011475,
      "learning_rate": 7.89345795758206e-06,
      "loss": 0.1091,
      "step": 4525
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.855762481689453,
      "learning_rate": 7.89251077219866e-06,
      "loss": 0.2768,
      "step": 4526
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.97677993774414,
      "learning_rate": 7.891563430775256e-06,
      "loss": 0.7302,
      "step": 4527
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.318231582641602,
      "learning_rate": 7.890615933362946e-06,
      "loss": 1.0598,
      "step": 4528
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.001105308532715,
      "learning_rate": 7.88966828001285e-06,
      "loss": 0.1336,
      "step": 4529
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.882739305496216,
      "learning_rate": 7.888720470776088e-06,
      "loss": 0.1704,
      "step": 4530
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.144750595092773,
      "learning_rate": 7.88777250570379e-06,
      "loss": 0.7667,
      "step": 4531
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.728300094604492,
      "learning_rate": 7.886824384847098e-06,
      "loss": 1.1083,
      "step": 4532
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.719538450241089,
      "learning_rate": 7.885876108257158e-06,
      "loss": 0.2156,
      "step": 4533
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.7590746879577637,
      "learning_rate": 7.884927675985123e-06,
      "loss": 0.1504,
      "step": 4534
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.265931129455566,
      "learning_rate": 7.88397908808216e-06,
      "loss": 1.0557,
      "step": 4535
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.094980239868164,
      "learning_rate": 7.883030344599443e-06,
      "loss": 0.0823,
      "step": 4536
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.372920036315918,
      "learning_rate": 7.882081445588152e-06,
      "loss": 0.8271,
      "step": 4537
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.063030242919922,
      "learning_rate": 7.881132391099476e-06,
      "loss": 0.2687,
      "step": 4538
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.550873279571533,
      "learning_rate": 7.880183181184614e-06,
      "loss": 0.9139,
      "step": 4539
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.8204405307769775,
      "learning_rate": 7.879233815894772e-06,
      "loss": 0.1026,
      "step": 4540
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.45037841796875,
      "learning_rate": 7.87828429528116e-06,
      "loss": 1.1501,
      "step": 4541
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.50939655303955,
      "learning_rate": 7.877334619395009e-06,
      "loss": 0.945,
      "step": 4542
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.288302421569824,
      "learning_rate": 7.876384788287546e-06,
      "loss": 0.1409,
      "step": 4543
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.497162342071533,
      "learning_rate": 7.87543480201001e-06,
      "loss": 0.1412,
      "step": 4544
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.212499141693115,
      "learning_rate": 7.874484660613652e-06,
      "loss": 0.96,
      "step": 4545
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.522650718688965,
      "learning_rate": 7.873534364149726e-06,
      "loss": 0.8432,
      "step": 4546
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.738285064697266,
      "learning_rate": 7.872583912669498e-06,
      "loss": 1.2417,
      "step": 4547
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.1013054847717285,
      "learning_rate": 7.871633306224241e-06,
      "loss": 0.2103,
      "step": 4548
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.428556442260742,
      "learning_rate": 7.870682544865237e-06,
      "loss": 0.8772,
      "step": 4549
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.800793170928955,
      "learning_rate": 7.869731628643777e-06,
      "loss": 1.0388,
      "step": 4550
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.479705810546875,
      "learning_rate": 7.86878055761116e-06,
      "loss": 0.8705,
      "step": 4551
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.8454790115356445,
      "learning_rate": 7.867829331818686e-06,
      "loss": 0.1051,
      "step": 4552
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.142500877380371,
      "learning_rate": 7.866877951317678e-06,
      "loss": 0.8873,
      "step": 4553
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.1300249099731445,
      "learning_rate": 7.865926416159455e-06,
      "loss": 0.8468,
      "step": 4554
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.4956705570220947,
      "learning_rate": 7.864974726395351e-06,
      "loss": 0.1721,
      "step": 4555
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.039095163345337,
      "learning_rate": 7.864022882076704e-06,
      "loss": 0.1372,
      "step": 4556
    },
    {
      "epoch": 0.33,
      "grad_norm": 14.642386436462402,
      "learning_rate": 7.863070883254863e-06,
      "loss": 1.2398,
      "step": 4557
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0044405460357666,
      "learning_rate": 7.862118729981188e-06,
      "loss": 0.1352,
      "step": 4558
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.26646089553833,
      "learning_rate": 7.861166422307038e-06,
      "loss": 0.8174,
      "step": 4559
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.638916015625,
      "learning_rate": 7.860213960283792e-06,
      "loss": 0.1447,
      "step": 4560
    },
    {
      "epoch": 0.33,
      "grad_norm": 26.30522346496582,
      "learning_rate": 7.859261343962827e-06,
      "loss": 1.3488,
      "step": 4561
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.423732042312622,
      "learning_rate": 7.858308573395536e-06,
      "loss": 0.1787,
      "step": 4562
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.4329380989074707,
      "learning_rate": 7.857355648633318e-06,
      "loss": 0.1832,
      "step": 4563
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2621238231658936,
      "learning_rate": 7.856402569727578e-06,
      "loss": 0.1014,
      "step": 4564
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.516537666320801,
      "learning_rate": 7.855449336729732e-06,
      "loss": 0.1453,
      "step": 4565
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.475648880004883,
      "learning_rate": 7.854495949691201e-06,
      "loss": 0.1892,
      "step": 4566
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.754389762878418,
      "learning_rate": 7.853542408663419e-06,
      "loss": 0.761,
      "step": 4567
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5816580057144165,
      "learning_rate": 7.852588713697826e-06,
      "loss": 0.0657,
      "step": 4568
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.71541428565979,
      "learning_rate": 7.851634864845868e-06,
      "loss": 0.8121,
      "step": 4569
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.7123148441314697,
      "learning_rate": 7.850680862159005e-06,
      "loss": 0.2403,
      "step": 4570
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2920193672180176,
      "learning_rate": 7.8497267056887e-06,
      "loss": 0.2218,
      "step": 4571
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.3121254444122314,
      "learning_rate": 7.848772395486424e-06,
      "loss": 0.1951,
      "step": 4572
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.769006729125977,
      "learning_rate": 7.847817931603662e-06,
      "loss": 1.3069,
      "step": 4573
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.109542369842529,
      "learning_rate": 7.846863314091902e-06,
      "loss": 0.1667,
      "step": 4574
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.0066580772399902,
      "learning_rate": 7.845908543002642e-06,
      "loss": 0.1964,
      "step": 4575
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.23271656036377,
      "learning_rate": 7.844953618387388e-06,
      "loss": 0.9538,
      "step": 4576
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1166832447052,
      "learning_rate": 7.843998540297654e-06,
      "loss": 0.1693,
      "step": 4577
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.842903137207031,
      "learning_rate": 7.843043308784967e-06,
      "loss": 1.3988,
      "step": 4578
    },
    {
      "epoch": 0.33,
      "grad_norm": 22.612674713134766,
      "learning_rate": 7.842087923900854e-06,
      "loss": 1.0453,
      "step": 4579
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2497334480285645,
      "learning_rate": 7.841132385696853e-06,
      "loss": 0.0725,
      "step": 4580
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.850243091583252,
      "learning_rate": 7.840176694224517e-06,
      "loss": 0.1141,
      "step": 4581
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.766274452209473,
      "learning_rate": 7.839220849535398e-06,
      "loss": 0.7468,
      "step": 4582
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.5670223236083984,
      "learning_rate": 7.83826485168106e-06,
      "loss": 0.0962,
      "step": 4583
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.86941909790039,
      "learning_rate": 7.837308700713078e-06,
      "loss": 0.8453,
      "step": 4584
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2190093994140625,
      "learning_rate": 7.836352396683031e-06,
      "loss": 0.1773,
      "step": 4585
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0743396282196045,
      "learning_rate": 7.835395939642508e-06,
      "loss": 0.1182,
      "step": 4586
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.026221752166748,
      "learning_rate": 7.834439329643106e-06,
      "loss": 0.8141,
      "step": 4587
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.8007924556732178,
      "learning_rate": 7.833482566736433e-06,
      "loss": 0.0738,
      "step": 4588
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.708272933959961,
      "learning_rate": 7.8325256509741e-06,
      "loss": 0.1511,
      "step": 4589
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.9510416984558105,
      "learning_rate": 7.831568582407729e-06,
      "loss": 0.2207,
      "step": 4590
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.3191640377044678,
      "learning_rate": 7.830611361088952e-06,
      "loss": 0.2073,
      "step": 4591
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.506589889526367,
      "learning_rate": 7.829653987069407e-06,
      "loss": 0.1141,
      "step": 4592
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2494523525238037,
      "learning_rate": 7.82869646040074e-06,
      "loss": 0.2129,
      "step": 4593
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.413084983825684,
      "learning_rate": 7.827738781134606e-06,
      "loss": 0.1582,
      "step": 4594
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.18751049041748,
      "learning_rate": 7.826780949322668e-06,
      "loss": 0.8346,
      "step": 4595
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.271283149719238,
      "learning_rate": 7.825822965016596e-06,
      "loss": 0.7514,
      "step": 4596
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.001437187194824,
      "learning_rate": 7.824864828268073e-06,
      "loss": 0.7591,
      "step": 4597
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.09749698638916,
      "learning_rate": 7.823906539128786e-06,
      "loss": 0.6679,
      "step": 4598
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.315703392028809,
      "learning_rate": 7.82294809765043e-06,
      "loss": 0.9195,
      "step": 4599
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.919105529785156,
      "learning_rate": 7.821989503884709e-06,
      "loss": 0.8035,
      "step": 4600
    },
    {
      "epoch": 0.33,
      "grad_norm": 22.373979568481445,
      "learning_rate": 7.821030757883337e-06,
      "loss": 1.0572,
      "step": 4601
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.7104692459106445,
      "learning_rate": 7.820071859698033e-06,
      "loss": 0.8,
      "step": 4602
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.562370777130127,
      "learning_rate": 7.819112809380526e-06,
      "loss": 0.7823,
      "step": 4603
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.78889274597168,
      "learning_rate": 7.818153606982555e-06,
      "loss": 1.2792,
      "step": 4604
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.47573471069336,
      "learning_rate": 7.817194252555864e-06,
      "loss": 0.6289,
      "step": 4605
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.003035068511963,
      "learning_rate": 7.816234746152207e-06,
      "loss": 0.9745,
      "step": 4606
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.076926231384277,
      "learning_rate": 7.815275087823343e-06,
      "loss": 0.5352,
      "step": 4607
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.99980354309082,
      "learning_rate": 7.814315277621047e-06,
      "loss": 0.6802,
      "step": 4608
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.120671510696411,
      "learning_rate": 7.813355315597094e-06,
      "loss": 0.1397,
      "step": 4609
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.534919023513794,
      "learning_rate": 7.812395201803272e-06,
      "loss": 0.275,
      "step": 4610
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.4934041500091553,
      "learning_rate": 7.81143493629137e-06,
      "loss": 0.2309,
      "step": 4611
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.654522895812988,
      "learning_rate": 7.8104745191132e-06,
      "loss": 0.6997,
      "step": 4612
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.644519805908203,
      "learning_rate": 7.809513950320566e-06,
      "loss": 1.4734,
      "step": 4613
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.821456909179688,
      "learning_rate": 7.80855322996529e-06,
      "loss": 1.4594,
      "step": 4614
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.251857042312622,
      "learning_rate": 7.807592358099197e-06,
      "loss": 0.1596,
      "step": 4615
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.4059648513793945,
      "learning_rate": 7.806631334774125e-06,
      "loss": 0.2152,
      "step": 4616
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.276810646057129,
      "learning_rate": 7.805670160041914e-06,
      "loss": 0.9125,
      "step": 4617
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.109184265136719,
      "learning_rate": 7.80470883395442e-06,
      "loss": 1.1495,
      "step": 4618
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.440401077270508,
      "learning_rate": 7.803747356563498e-06,
      "loss": 0.0953,
      "step": 4619
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.0624473094940186,
      "learning_rate": 7.802785727921021e-06,
      "loss": 0.1911,
      "step": 4620
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.293135643005371,
      "learning_rate": 7.801823948078863e-06,
      "loss": 0.7692,
      "step": 4621
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.357582092285156,
      "learning_rate": 7.800862017088908e-06,
      "loss": 0.5445,
      "step": 4622
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.992502212524414,
      "learning_rate": 7.799899935003048e-06,
      "loss": 0.6604,
      "step": 4623
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7449100017547607,
      "learning_rate": 7.798937701873188e-06,
      "loss": 0.0743,
      "step": 4624
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.635924339294434,
      "learning_rate": 7.79797531775123e-06,
      "loss": 0.6864,
      "step": 4625
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.589493751525879,
      "learning_rate": 7.797012782689097e-06,
      "loss": 1.0862,
      "step": 4626
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.259016513824463,
      "learning_rate": 7.79605009673871e-06,
      "loss": 0.8875,
      "step": 4627
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.490600109100342,
      "learning_rate": 7.795087259952005e-06,
      "loss": 0.1666,
      "step": 4628
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.87271785736084,
      "learning_rate": 7.794124272380921e-06,
      "loss": 0.9877,
      "step": 4629
    },
    {
      "epoch": 0.33,
      "grad_norm": 25.555849075317383,
      "learning_rate": 7.79316113407741e-06,
      "loss": 1.2439,
      "step": 4630
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.113100051879883,
      "learning_rate": 7.792197845093428e-06,
      "loss": 0.9178,
      "step": 4631
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.598352432250977,
      "learning_rate": 7.791234405480942e-06,
      "loss": 1.1154,
      "step": 4632
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.3468804359436035,
      "learning_rate": 7.790270815291924e-06,
      "loss": 0.3039,
      "step": 4633
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.011892795562744,
      "learning_rate": 7.789307074578358e-06,
      "loss": 0.162,
      "step": 4634
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.6872992515563965,
      "learning_rate": 7.788343183392232e-06,
      "loss": 1.0213,
      "step": 4635
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.096609354019165,
      "learning_rate": 7.78737914178555e-06,
      "loss": 0.1975,
      "step": 4636
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0423331260681152,
      "learning_rate": 7.786414949810311e-06,
      "loss": 0.0992,
      "step": 4637
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.1603546142578125,
      "learning_rate": 7.785450607518532e-06,
      "loss": 0.9232,
      "step": 4638
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.687654972076416,
      "learning_rate": 7.784486114962237e-06,
      "loss": 0.2126,
      "step": 4639
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.938001871109009,
      "learning_rate": 7.783521472193456e-06,
      "loss": 0.1043,
      "step": 4640
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.188401699066162,
      "learning_rate": 7.782556679264228e-06,
      "loss": 0.1753,
      "step": 4641
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.7577567100524902,
      "learning_rate": 7.7815917362266e-06,
      "loss": 0.1987,
      "step": 4642
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.3624773025512695,
      "learning_rate": 7.780626643132626e-06,
      "loss": 0.8819,
      "step": 4643
    },
    {
      "epoch": 0.33,
      "grad_norm": 20.554811477661133,
      "learning_rate": 7.779661400034368e-06,
      "loss": 1.407,
      "step": 4644
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.7138636112213135,
      "learning_rate": 7.778696006983903e-06,
      "loss": 0.1368,
      "step": 4645
    },
    {
      "epoch": 0.33,
      "grad_norm": 15.190347671508789,
      "learning_rate": 7.777730464033304e-06,
      "loss": 1.2843,
      "step": 4646
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2650086879730225,
      "learning_rate": 7.776764771234662e-06,
      "loss": 0.2271,
      "step": 4647
    },
    {
      "epoch": 0.33,
      "grad_norm": 25.272619247436523,
      "learning_rate": 7.775798928640071e-06,
      "loss": 0.8188,
      "step": 4648
    },
    {
      "epoch": 0.33,
      "grad_norm": 22.884794235229492,
      "learning_rate": 7.774832936301632e-06,
      "loss": 0.6727,
      "step": 4649
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.74905776977539,
      "learning_rate": 7.773866794271464e-06,
      "loss": 1.5672,
      "step": 4650
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.1549296379089355,
      "learning_rate": 7.772900502601678e-06,
      "loss": 0.2704,
      "step": 4651
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.080925464630127,
      "learning_rate": 7.771934061344407e-06,
      "loss": 0.6874,
      "step": 4652
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.5788872241973877,
      "learning_rate": 7.770967470551787e-06,
      "loss": 0.1132,
      "step": 4653
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.7743804454803467,
      "learning_rate": 7.770000730275959e-06,
      "loss": 0.2594,
      "step": 4654
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.5457072257995605,
      "learning_rate": 7.769033840569078e-06,
      "loss": 0.2538,
      "step": 4655
    },
    {
      "epoch": 0.33,
      "grad_norm": 22.445955276489258,
      "learning_rate": 7.768066801483303e-06,
      "loss": 1.7462,
      "step": 4656
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1191017627716064,
      "learning_rate": 7.767099613070802e-06,
      "loss": 0.1225,
      "step": 4657
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.434774875640869,
      "learning_rate": 7.76613227538375e-06,
      "loss": 0.6828,
      "step": 4658
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.878033638000488,
      "learning_rate": 7.76516478847433e-06,
      "loss": 0.4317,
      "step": 4659
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.289509296417236,
      "learning_rate": 7.764197152394738e-06,
      "loss": 0.9129,
      "step": 4660
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.848413944244385,
      "learning_rate": 7.763229367197173e-06,
      "loss": 0.8692,
      "step": 4661
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.988802194595337,
      "learning_rate": 7.762261432933844e-06,
      "loss": 0.1565,
      "step": 4662
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3217270374298096,
      "learning_rate": 7.761293349656966e-06,
      "loss": 0.1373,
      "step": 4663
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.385951995849609,
      "learning_rate": 7.760325117418763e-06,
      "loss": 0.9569,
      "step": 4664
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.356113910675049,
      "learning_rate": 7.759356736271469e-06,
      "loss": 0.5865,
      "step": 4665
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.237249374389648,
      "learning_rate": 7.758388206267324e-06,
      "loss": 0.9347,
      "step": 4666
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.571282148361206,
      "learning_rate": 7.757419527458577e-06,
      "loss": 0.1181,
      "step": 4667
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3577322959899902,
      "learning_rate": 7.75645069989748e-06,
      "loss": 0.1054,
      "step": 4668
    },
    {
      "epoch": 0.33,
      "grad_norm": 14.087679862976074,
      "learning_rate": 7.755481723636305e-06,
      "loss": 1.5596,
      "step": 4669
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.881197929382324,
      "learning_rate": 7.75451259872732e-06,
      "loss": 0.1149,
      "step": 4670
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.6862902641296387,
      "learning_rate": 7.753543325222807e-06,
      "loss": 0.1424,
      "step": 4671
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.017249822616577,
      "learning_rate": 7.752573903175055e-06,
      "loss": 0.1704,
      "step": 4672
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3011670112609863,
      "learning_rate": 7.75160433263636e-06,
      "loss": 0.1109,
      "step": 4673
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1367876529693604,
      "learning_rate": 7.750634613659026e-06,
      "loss": 0.1845,
      "step": 4674
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.577188014984131,
      "learning_rate": 7.749664746295364e-06,
      "loss": 0.8825,
      "step": 4675
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5515096187591553,
      "learning_rate": 7.7486947305977e-06,
      "loss": 0.1251,
      "step": 4676
    },
    {
      "epoch": 0.34,
      "grad_norm": 18.06385612487793,
      "learning_rate": 7.74772456661836e-06,
      "loss": 0.8297,
      "step": 4677
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.161862373352051,
      "learning_rate": 7.746754254409679e-06,
      "loss": 0.3006,
      "step": 4678
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.724593162536621,
      "learning_rate": 7.745783794024004e-06,
      "loss": 0.2196,
      "step": 4679
    },
    {
      "epoch": 0.34,
      "grad_norm": 25.649211883544922,
      "learning_rate": 7.744813185513685e-06,
      "loss": 1.4271,
      "step": 4680
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.515346050262451,
      "learning_rate": 7.743842428931086e-06,
      "loss": 1.0211,
      "step": 4681
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9725630283355713,
      "learning_rate": 7.742871524328572e-06,
      "loss": 0.2163,
      "step": 4682
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5168168544769287,
      "learning_rate": 7.741900471758523e-06,
      "loss": 0.1283,
      "step": 4683
    },
    {
      "epoch": 0.34,
      "grad_norm": 29.907196044921875,
      "learning_rate": 7.740929271273323e-06,
      "loss": 0.6606,
      "step": 4684
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1595873832702637,
      "learning_rate": 7.739957922925362e-06,
      "loss": 0.1355,
      "step": 4685
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.17049241065979,
      "learning_rate": 7.738986426767044e-06,
      "loss": 0.2099,
      "step": 4686
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.117385387420654,
      "learning_rate": 7.738014782850773e-06,
      "loss": 0.979,
      "step": 4687
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.8049824237823486,
      "learning_rate": 7.737042991228972e-06,
      "loss": 0.3691,
      "step": 4688
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.496891021728516,
      "learning_rate": 7.73607105195406e-06,
      "loss": 0.9857,
      "step": 4689
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.174270629882812,
      "learning_rate": 7.735098965078472e-06,
      "loss": 1.1808,
      "step": 4690
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.6228394508361816,
      "learning_rate": 7.734126730654648e-06,
      "loss": 0.1243,
      "step": 4691
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.543551206588745,
      "learning_rate": 7.733154348735035e-06,
      "loss": 0.1789,
      "step": 4692
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9396255016326904,
      "learning_rate": 7.73218181937209e-06,
      "loss": 0.2499,
      "step": 4693
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.141023874282837,
      "learning_rate": 7.731209142618279e-06,
      "loss": 0.0771,
      "step": 4694
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.634486198425293,
      "learning_rate": 7.730236318526072e-06,
      "loss": 0.823,
      "step": 4695
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.406465530395508,
      "learning_rate": 7.729263347147952e-06,
      "loss": 0.9715,
      "step": 4696
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.18309211730957,
      "learning_rate": 7.728290228536404e-06,
      "loss": 1.0514,
      "step": 4697
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.6777772903442383,
      "learning_rate": 7.727316962743924e-06,
      "loss": 0.1935,
      "step": 4698
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.5188164710998535,
      "learning_rate": 7.72634354982302e-06,
      "loss": 0.8878,
      "step": 4699
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.252198696136475,
      "learning_rate": 7.7253699898262e-06,
      "loss": 0.3372,
      "step": 4700
    },
    {
      "epoch": 0.34,
      "grad_norm": 18.8830509185791,
      "learning_rate": 7.724396282805986e-06,
      "loss": 1.573,
      "step": 4701
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1919355392456055,
      "learning_rate": 7.723422428814903e-06,
      "loss": 0.1278,
      "step": 4702
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.8312578201293945,
      "learning_rate": 7.722448427905489e-06,
      "loss": 0.1374,
      "step": 4703
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.123594284057617,
      "learning_rate": 7.721474280130288e-06,
      "loss": 1.1751,
      "step": 4704
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3879544734954834,
      "learning_rate": 7.720499985541852e-06,
      "loss": 0.1424,
      "step": 4705
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.300567150115967,
      "learning_rate": 7.719525544192737e-06,
      "loss": 0.1391,
      "step": 4706
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.540412902832031,
      "learning_rate": 7.718550956135515e-06,
      "loss": 0.9006,
      "step": 4707
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.2206778526306152,
      "learning_rate": 7.717576221422757e-06,
      "loss": 0.1376,
      "step": 4708
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5867462158203125,
      "learning_rate": 7.716601340107052e-06,
      "loss": 0.0628,
      "step": 4709
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.719175338745117,
      "learning_rate": 7.715626312240985e-06,
      "loss": 1.0086,
      "step": 4710
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.83556866645813,
      "learning_rate": 7.714651137877159e-06,
      "loss": 0.2014,
      "step": 4711
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.610742092132568,
      "learning_rate": 7.713675817068176e-06,
      "loss": 0.7785,
      "step": 4712
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6949408054351807,
      "learning_rate": 7.712700349866657e-06,
      "loss": 1.0513,
      "step": 4713
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.543334484100342,
      "learning_rate": 7.711724736325223e-06,
      "loss": 0.8745,
      "step": 4714
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.0892465114593506,
      "learning_rate": 7.710748976496506e-06,
      "loss": 0.1601,
      "step": 4715
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.245230674743652,
      "learning_rate": 7.709773070433139e-06,
      "loss": 0.7796,
      "step": 4716
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6507067680358887,
      "learning_rate": 7.708797018187773e-06,
      "loss": 0.8712,
      "step": 4717
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.8278884887695312,
      "learning_rate": 7.70782081981306e-06,
      "loss": 0.1496,
      "step": 4718
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.809030055999756,
      "learning_rate": 7.706844475361664e-06,
      "loss": 0.7322,
      "step": 4719
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.619668960571289,
      "learning_rate": 7.705867984886255e-06,
      "loss": 1.0182,
      "step": 4720
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.98078989982605,
      "learning_rate": 7.70489134843951e-06,
      "loss": 0.8471,
      "step": 4721
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.643211364746094,
      "learning_rate": 7.703914566074117e-06,
      "loss": 1.0902,
      "step": 4722
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.4677369594573975,
      "learning_rate": 7.702937637842767e-06,
      "loss": 0.2501,
      "step": 4723
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.393730163574219,
      "learning_rate": 7.701960563798161e-06,
      "loss": 1.2813,
      "step": 4724
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1803736686706543,
      "learning_rate": 7.700983343993012e-06,
      "loss": 0.1251,
      "step": 4725
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.614607334136963,
      "learning_rate": 7.700005978480036e-06,
      "loss": 0.17,
      "step": 4726
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.332775115966797,
      "learning_rate": 7.699028467311955e-06,
      "loss": 0.1282,
      "step": 4727
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.472050666809082,
      "learning_rate": 7.698050810541506e-06,
      "loss": 1.0392,
      "step": 4728
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.917169570922852,
      "learning_rate": 7.697073008221428e-06,
      "loss": 1.1359,
      "step": 4729
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.807342052459717,
      "learning_rate": 7.696095060404471e-06,
      "loss": 0.9455,
      "step": 4730
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9855294227600098,
      "learning_rate": 7.69511696714339e-06,
      "loss": 0.1528,
      "step": 4731
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.066118240356445,
      "learning_rate": 7.694138728490951e-06,
      "loss": 0.597,
      "step": 4732
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.640809059143066,
      "learning_rate": 7.693160344499924e-06,
      "loss": 0.1217,
      "step": 4733
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.40427827835083,
      "learning_rate": 7.692181815223093e-06,
      "loss": 0.8559,
      "step": 4734
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.483299255371094,
      "learning_rate": 7.691203140713242e-06,
      "loss": 0.9361,
      "step": 4735
    },
    {
      "epoch": 0.34,
      "grad_norm": 14.950200080871582,
      "learning_rate": 7.690224321023168e-06,
      "loss": 0.5384,
      "step": 4736
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.0915110111236572,
      "learning_rate": 7.689245356205675e-06,
      "loss": 0.2237,
      "step": 4737
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.884266376495361,
      "learning_rate": 7.688266246313574e-06,
      "loss": 0.8807,
      "step": 4738
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.237699508666992,
      "learning_rate": 7.687286991399686e-06,
      "loss": 0.0825,
      "step": 4739
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.988581657409668,
      "learning_rate": 7.686307591516836e-06,
      "loss": 0.2765,
      "step": 4740
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.280163764953613,
      "learning_rate": 7.685328046717859e-06,
      "loss": 0.1807,
      "step": 4741
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.988571166992188,
      "learning_rate": 7.6843483570556e-06,
      "loss": 0.7953,
      "step": 4742
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3972361087799072,
      "learning_rate": 7.683368522582905e-06,
      "loss": 0.1219,
      "step": 4743
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.163151741027832,
      "learning_rate": 7.682388543352638e-06,
      "loss": 0.1381,
      "step": 4744
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6757307052612305,
      "learning_rate": 7.681408419417663e-06,
      "loss": 0.1796,
      "step": 4745
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.311796188354492,
      "learning_rate": 7.680428150830852e-06,
      "loss": 0.9129,
      "step": 4746
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.289051055908203,
      "learning_rate": 7.679447737645088e-06,
      "loss": 0.7196,
      "step": 4747
    },
    {
      "epoch": 0.34,
      "grad_norm": 16.466367721557617,
      "learning_rate": 7.678467179913263e-06,
      "loss": 1.1366,
      "step": 4748
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.13201904296875,
      "learning_rate": 7.67748647768827e-06,
      "loss": 1.0419,
      "step": 4749
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.312890529632568,
      "learning_rate": 7.676505631023017e-06,
      "loss": 1.0067,
      "step": 4750
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.516700744628906,
      "learning_rate": 7.675524639970416e-06,
      "loss": 1.306,
      "step": 4751
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.54512357711792,
      "learning_rate": 7.674543504583388e-06,
      "loss": 0.8274,
      "step": 4752
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1989176273345947,
      "learning_rate": 7.673562224914864e-06,
      "loss": 0.1938,
      "step": 4753
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.510815382003784,
      "learning_rate": 7.672580801017774e-06,
      "loss": 0.2393,
      "step": 4754
    },
    {
      "epoch": 0.34,
      "grad_norm": 14.675544738769531,
      "learning_rate": 7.671599232945069e-06,
      "loss": 1.0903,
      "step": 4755
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.606475353240967,
      "learning_rate": 7.670617520749697e-06,
      "loss": 0.0914,
      "step": 4756
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.710186958312988,
      "learning_rate": 7.669635664484619e-06,
      "loss": 0.908,
      "step": 4757
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.78538990020752,
      "learning_rate": 7.6686536642028e-06,
      "loss": 0.8804,
      "step": 4758
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.192602157592773,
      "learning_rate": 7.66767151995722e-06,
      "loss": 1.1559,
      "step": 4759
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.4998438358306885,
      "learning_rate": 7.666689231800856e-06,
      "loss": 0.2036,
      "step": 4760
    },
    {
      "epoch": 0.34,
      "grad_norm": 14.333269119262695,
      "learning_rate": 7.665706799786705e-06,
      "loss": 0.8461,
      "step": 4761
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.22127628326416,
      "learning_rate": 7.664724223967761e-06,
      "loss": 0.8886,
      "step": 4762
    },
    {
      "epoch": 0.34,
      "grad_norm": 16.20050048828125,
      "learning_rate": 7.663741504397032e-06,
      "loss": 1.1171,
      "step": 4763
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.395224094390869,
      "learning_rate": 7.662758641127531e-06,
      "loss": 0.9238,
      "step": 4764
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.82241153717041,
      "learning_rate": 7.66177563421228e-06,
      "loss": 0.1568,
      "step": 4765
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.432409286499023,
      "learning_rate": 7.660792483704312e-06,
      "loss": 0.9009,
      "step": 4766
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.986292600631714,
      "learning_rate": 7.659809189656659e-06,
      "loss": 0.1662,
      "step": 4767
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.095210313796997,
      "learning_rate": 7.658825752122367e-06,
      "loss": 0.2014,
      "step": 4768
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.908305168151855,
      "learning_rate": 7.65784217115449e-06,
      "loss": 0.7746,
      "step": 4769
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.0662598609924316,
      "learning_rate": 7.65685844680609e-06,
      "loss": 0.1646,
      "step": 4770
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5821213722229004,
      "learning_rate": 7.655874579130234e-06,
      "loss": 0.1001,
      "step": 4771
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.3307929039001465,
      "learning_rate": 7.654890568179995e-06,
      "loss": 0.8509,
      "step": 4772
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.965593338012695,
      "learning_rate": 7.65390641400846e-06,
      "loss": 1.4404,
      "step": 4773
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.9768080711364746,
      "learning_rate": 7.652922116668722e-06,
      "loss": 0.9743,
      "step": 4774
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.862563371658325,
      "learning_rate": 7.651937676213876e-06,
      "loss": 0.8154,
      "step": 4775
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0302932262420654,
      "learning_rate": 7.650953092697029e-06,
      "loss": 0.083,
      "step": 4776
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.075276851654053,
      "learning_rate": 7.649968366171298e-06,
      "loss": 0.9109,
      "step": 4777
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.4317190647125244,
      "learning_rate": 7.648983496689804e-06,
      "loss": 0.1398,
      "step": 4778
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.36586332321167,
      "learning_rate": 7.647998484305678e-06,
      "loss": 0.2667,
      "step": 4779
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.694833755493164,
      "learning_rate": 7.647013329072058e-06,
      "loss": 0.1284,
      "step": 4780
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6626574993133545,
      "learning_rate": 7.646028031042088e-06,
      "loss": 0.145,
      "step": 4781
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2276577949523926,
      "learning_rate": 7.645042590268921e-06,
      "loss": 0.1423,
      "step": 4782
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.396082639694214,
      "learning_rate": 7.644057006805716e-06,
      "loss": 0.2062,
      "step": 4783
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.120304822921753,
      "learning_rate": 7.643071280705647e-06,
      "loss": 0.1878,
      "step": 4784
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.122096538543701,
      "learning_rate": 7.642085412021886e-06,
      "loss": 0.8907,
      "step": 4785
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.455334663391113,
      "learning_rate": 7.641099400807616e-06,
      "loss": 1.2286,
      "step": 4786
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.809576988220215,
      "learning_rate": 7.640113247116034e-06,
      "loss": 0.8298,
      "step": 4787
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.153135299682617,
      "learning_rate": 7.639126951000333e-06,
      "loss": 1.5342,
      "step": 4788
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.810080051422119,
      "learning_rate": 7.638140512513722e-06,
      "loss": 0.9447,
      "step": 4789
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.487138271331787,
      "learning_rate": 7.63715393170942e-06,
      "loss": 0.1002,
      "step": 4790
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.953609466552734,
      "learning_rate": 7.636167208640642e-06,
      "loss": 0.836,
      "step": 4791
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.013246536254883,
      "learning_rate": 7.635180343360622e-06,
      "loss": 0.8715,
      "step": 4792
    },
    {
      "epoch": 0.34,
      "grad_norm": 15.583481788635254,
      "learning_rate": 7.634193335922599e-06,
      "loss": 0.8514,
      "step": 4793
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.728911399841309,
      "learning_rate": 7.633206186379815e-06,
      "loss": 0.8136,
      "step": 4794
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.681389808654785,
      "learning_rate": 7.632218894785526e-06,
      "loss": 0.9727,
      "step": 4795
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.2751104831695557,
      "learning_rate": 7.63123146119299e-06,
      "loss": 0.1728,
      "step": 4796
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.2104341983795166,
      "learning_rate": 7.630243885655476e-06,
      "loss": 0.2314,
      "step": 4797
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.056594848632812,
      "learning_rate": 7.62925616822626e-06,
      "loss": 1.0013,
      "step": 4798
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.7129480838775635,
      "learning_rate": 7.628268308958629e-06,
      "loss": 0.1366,
      "step": 4799
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.16242790222168,
      "learning_rate": 7.6272803079058675e-06,
      "loss": 0.6914,
      "step": 4800
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.270237922668457,
      "learning_rate": 7.6262921651212794e-06,
      "loss": 0.8478,
      "step": 4801
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.1465864181518555,
      "learning_rate": 7.62530388065817e-06,
      "loss": 0.983,
      "step": 4802
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.855730056762695,
      "learning_rate": 7.624315454569855e-06,
      "loss": 0.9804,
      "step": 4803
    },
    {
      "epoch": 0.34,
      "grad_norm": 16.83928108215332,
      "learning_rate": 7.623326886909655e-06,
      "loss": 0.9873,
      "step": 4804
    },
    {
      "epoch": 0.34,
      "grad_norm": 16.577816009521484,
      "learning_rate": 7.622338177730897e-06,
      "loss": 1.0511,
      "step": 4805
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.326534271240234,
      "learning_rate": 7.621349327086921e-06,
      "loss": 0.924,
      "step": 4806
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.316156387329102,
      "learning_rate": 7.620360335031071e-06,
      "loss": 0.8028,
      "step": 4807
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9868736267089844,
      "learning_rate": 7.619371201616701e-06,
      "loss": 0.173,
      "step": 4808
    },
    {
      "epoch": 0.34,
      "grad_norm": 14.6183500289917,
      "learning_rate": 7.6183819268971685e-06,
      "loss": 1.065,
      "step": 4809
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.7693209648132324,
      "learning_rate": 7.617392510925843e-06,
      "loss": 0.2706,
      "step": 4810
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.653261423110962,
      "learning_rate": 7.616402953756098e-06,
      "loss": 0.0958,
      "step": 4811
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6665399074554443,
      "learning_rate": 7.615413255441317e-06,
      "loss": 0.2206,
      "step": 4812
    },
    {
      "epoch": 0.35,
      "grad_norm": 18.208873748779297,
      "learning_rate": 7.6144234160348915e-06,
      "loss": 1.057,
      "step": 4813
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4288225173950195,
      "learning_rate": 7.6134334355902186e-06,
      "loss": 0.1067,
      "step": 4814
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.620673656463623,
      "learning_rate": 7.612443314160704e-06,
      "loss": 0.1577,
      "step": 4815
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.354556083679199,
      "learning_rate": 7.611453051799762e-06,
      "loss": 0.12,
      "step": 4816
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.142473220825195,
      "learning_rate": 7.610462648560812e-06,
      "loss": 0.5303,
      "step": 4817
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.298485040664673,
      "learning_rate": 7.609472104497283e-06,
      "loss": 0.1625,
      "step": 4818
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.891895294189453,
      "learning_rate": 7.608481419662612e-06,
      "loss": 0.153,
      "step": 4819
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.1106350421905518,
      "learning_rate": 7.6074905941102405e-06,
      "loss": 0.2138,
      "step": 4820
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.7533652782440186,
      "learning_rate": 7.606499627893624e-06,
      "loss": 0.21,
      "step": 4821
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.869091749191284,
      "learning_rate": 7.605508521066216e-06,
      "loss": 0.351,
      "step": 4822
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.555386543273926,
      "learning_rate": 7.604517273681489e-06,
      "loss": 1.1074,
      "step": 4823
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.058560848236084,
      "learning_rate": 7.603525885792911e-06,
      "loss": 0.8106,
      "step": 4824
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.447098731994629,
      "learning_rate": 7.602534357453968e-06,
      "loss": 0.8296,
      "step": 4825
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.110638380050659,
      "learning_rate": 7.601542688718146e-06,
      "loss": 0.09,
      "step": 4826
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.550702095031738,
      "learning_rate": 7.600550879638944e-06,
      "loss": 1.3026,
      "step": 4827
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.4199867248535156,
      "learning_rate": 7.599558930269865e-06,
      "loss": 0.2665,
      "step": 4828
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.916510105133057,
      "learning_rate": 7.598566840664423e-06,
      "loss": 0.7943,
      "step": 4829
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6737630367279053,
      "learning_rate": 7.597574610876135e-06,
      "loss": 0.2281,
      "step": 4830
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.530842781066895,
      "learning_rate": 7.596582240958529e-06,
      "loss": 0.7213,
      "step": 4831
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.780617713928223,
      "learning_rate": 7.595589730965139e-06,
      "loss": 0.7455,
      "step": 4832
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.360484600067139,
      "learning_rate": 7.594597080949511e-06,
      "loss": 0.8111,
      "step": 4833
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.508974075317383,
      "learning_rate": 7.593604290965189e-06,
      "loss": 1.1262,
      "step": 4834
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.078738212585449,
      "learning_rate": 7.592611361065732e-06,
      "loss": 0.2317,
      "step": 4835
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.318672180175781,
      "learning_rate": 7.591618291304707e-06,
      "loss": 1.138,
      "step": 4836
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.523918151855469,
      "learning_rate": 7.590625081735685e-06,
      "loss": 0.9098,
      "step": 4837
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.352978706359863,
      "learning_rate": 7.589631732412246e-06,
      "loss": 0.4638,
      "step": 4838
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.8978347778320312,
      "learning_rate": 7.588638243387978e-06,
      "loss": 0.8222,
      "step": 4839
    },
    {
      "epoch": 0.35,
      "grad_norm": 24.660049438476562,
      "learning_rate": 7.587644614716474e-06,
      "loss": 1.0589,
      "step": 4840
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.154013156890869,
      "learning_rate": 7.586650846451338e-06,
      "loss": 0.1682,
      "step": 4841
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.210247278213501,
      "learning_rate": 7.58565693864618e-06,
      "loss": 0.3041,
      "step": 4842
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3807873725891113,
      "learning_rate": 7.584662891354617e-06,
      "loss": 0.1213,
      "step": 4843
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.6406989097595215,
      "learning_rate": 7.583668704630276e-06,
      "loss": 0.7448,
      "step": 4844
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.4387736320495605,
      "learning_rate": 7.582674378526786e-06,
      "loss": 0.7524,
      "step": 4845
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.170940399169922,
      "learning_rate": 7.58167991309779e-06,
      "loss": 0.6598,
      "step": 4846
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.912238121032715,
      "learning_rate": 7.580685308396938e-06,
      "loss": 0.944,
      "step": 4847
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6714320182800293,
      "learning_rate": 7.579690564477877e-06,
      "loss": 0.1132,
      "step": 4848
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.720771789550781,
      "learning_rate": 7.578695681394277e-06,
      "loss": 1.1627,
      "step": 4849
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8230347633361816,
      "learning_rate": 7.577700659199807e-06,
      "loss": 0.0969,
      "step": 4850
    },
    {
      "epoch": 0.35,
      "grad_norm": 14.752702713012695,
      "learning_rate": 7.576705497948141e-06,
      "loss": 1.2489,
      "step": 4851
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6371102333068848,
      "learning_rate": 7.5757101976929684e-06,
      "loss": 0.0969,
      "step": 4852
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.828598976135254,
      "learning_rate": 7.574714758487981e-06,
      "loss": 0.7573,
      "step": 4853
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.275949478149414,
      "learning_rate": 7.573719180386876e-06,
      "loss": 0.0663,
      "step": 4854
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2542591094970703,
      "learning_rate": 7.572723463443365e-06,
      "loss": 0.0957,
      "step": 4855
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.283947467803955,
      "learning_rate": 7.571727607711161e-06,
      "loss": 0.7123,
      "step": 4856
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.218151092529297,
      "learning_rate": 7.5707316132439866e-06,
      "loss": 0.49,
      "step": 4857
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.765862226486206,
      "learning_rate": 7.569735480095572e-06,
      "loss": 0.3574,
      "step": 4858
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.7088117599487305,
      "learning_rate": 7.5687392083196575e-06,
      "loss": 0.1387,
      "step": 4859
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.834127187728882,
      "learning_rate": 7.567742797969984e-06,
      "loss": 0.1729,
      "step": 4860
    },
    {
      "epoch": 0.35,
      "grad_norm": 18.8779296875,
      "learning_rate": 7.566746249100306e-06,
      "loss": 1.4384,
      "step": 4861
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.754568576812744,
      "learning_rate": 7.565749561764384e-06,
      "loss": 0.2953,
      "step": 4862
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.55444073677063,
      "learning_rate": 7.564752736015985e-06,
      "loss": 0.1336,
      "step": 4863
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.70637035369873,
      "learning_rate": 7.563755771908884e-06,
      "loss": 1.1361,
      "step": 4864
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.0625720024108887,
      "learning_rate": 7.562758669496863e-06,
      "loss": 0.1309,
      "step": 4865
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8519763946533203,
      "learning_rate": 7.561761428833713e-06,
      "loss": 0.1176,
      "step": 4866
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8693385124206543,
      "learning_rate": 7.560764049973231e-06,
      "loss": 0.215,
      "step": 4867
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4403414726257324,
      "learning_rate": 7.55976653296922e-06,
      "loss": 0.1083,
      "step": 4868
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.413917541503906,
      "learning_rate": 7.558768877875495e-06,
      "loss": 1.0871,
      "step": 4869
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.2089381217956543,
      "learning_rate": 7.557771084745874e-06,
      "loss": 0.1231,
      "step": 4870
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8269591331481934,
      "learning_rate": 7.556773153634185e-06,
      "loss": 0.1561,
      "step": 4871
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4141666889190674,
      "learning_rate": 7.555775084594263e-06,
      "loss": 0.108,
      "step": 4872
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.5405893325805664,
      "learning_rate": 7.554776877679948e-06,
      "loss": 0.3058,
      "step": 4873
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.08262825012207,
      "learning_rate": 7.553778532945091e-06,
      "loss": 0.9124,
      "step": 4874
    },
    {
      "epoch": 0.35,
      "grad_norm": 18.750215530395508,
      "learning_rate": 7.552780050443549e-06,
      "loss": 1.0642,
      "step": 4875
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.789843559265137,
      "learning_rate": 7.551781430229186e-06,
      "loss": 0.9687,
      "step": 4876
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.86733055114746,
      "learning_rate": 7.5507826723558746e-06,
      "loss": 0.9113,
      "step": 4877
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.1003384590148926,
      "learning_rate": 7.549783776877493e-06,
      "loss": 0.1707,
      "step": 4878
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.6111421585083,
      "learning_rate": 7.548784743847928e-06,
      "loss": 1.4681,
      "step": 4879
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.403285980224609,
      "learning_rate": 7.547785573321072e-06,
      "loss": 0.8155,
      "step": 4880
    },
    {
      "epoch": 0.35,
      "grad_norm": 15.513958930969238,
      "learning_rate": 7.546786265350831e-06,
      "loss": 1.0154,
      "step": 4881
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.8565008640289307,
      "learning_rate": 7.5457868199911086e-06,
      "loss": 0.37,
      "step": 4882
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.1866824626922607,
      "learning_rate": 7.544787237295825e-06,
      "loss": 0.1932,
      "step": 4883
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9531991481781006,
      "learning_rate": 7.543787517318904e-06,
      "loss": 1.0029,
      "step": 4884
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.583179235458374,
      "learning_rate": 7.542787660114272e-06,
      "loss": 0.2391,
      "step": 4885
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.075812816619873,
      "learning_rate": 7.541787665735871e-06,
      "loss": 0.3122,
      "step": 4886
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.1654272079467773,
      "learning_rate": 7.540787534237649e-06,
      "loss": 0.1423,
      "step": 4887
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.108147144317627,
      "learning_rate": 7.539787265673555e-06,
      "loss": 0.6235,
      "step": 4888
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3284687995910645,
      "learning_rate": 7.5387868600975535e-06,
      "loss": 0.1396,
      "step": 4889
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.377547264099121,
      "learning_rate": 7.53778631756361e-06,
      "loss": 0.9726,
      "step": 4890
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.038299083709717,
      "learning_rate": 7.5367856381257e-06,
      "loss": 0.1737,
      "step": 4891
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.819032669067383,
      "learning_rate": 7.535784821837806e-06,
      "loss": 0.1209,
      "step": 4892
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.120330333709717,
      "learning_rate": 7.534783868753921e-06,
      "loss": 1.0892,
      "step": 4893
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.592216730117798,
      "learning_rate": 7.53378277892804e-06,
      "loss": 0.2725,
      "step": 4894
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1246798038482666,
      "learning_rate": 7.532781552414171e-06,
      "loss": 0.1517,
      "step": 4895
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.862966537475586,
      "learning_rate": 7.531780189266324e-06,
      "loss": 0.8135,
      "step": 4896
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8953871726989746,
      "learning_rate": 7.530778689538517e-06,
      "loss": 0.2013,
      "step": 4897
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.984404563903809,
      "learning_rate": 7.529777053284782e-06,
      "loss": 0.8866,
      "step": 4898
    },
    {
      "epoch": 0.35,
      "grad_norm": 22.2523193359375,
      "learning_rate": 7.5287752805591476e-06,
      "loss": 1.297,
      "step": 4899
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.143317461013794,
      "learning_rate": 7.527773371415659e-06,
      "loss": 0.065,
      "step": 4900
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.3834941387176514,
      "learning_rate": 7.526771325908369e-06,
      "loss": 0.2167,
      "step": 4901
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.438712120056152,
      "learning_rate": 7.525769144091326e-06,
      "loss": 0.9377,
      "step": 4902
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.7915427684783936,
      "learning_rate": 7.524766826018599e-06,
      "loss": 0.1231,
      "step": 4903
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.6657650470733643,
      "learning_rate": 7.52376437174426e-06,
      "loss": 0.7977,
      "step": 4904
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.0416715145111084,
      "learning_rate": 7.522761781322385e-06,
      "loss": 0.1093,
      "step": 4905
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.393514633178711,
      "learning_rate": 7.521759054807061e-06,
      "loss": 1.0564,
      "step": 4906
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8915858268737793,
      "learning_rate": 7.520756192252383e-06,
      "loss": 0.1543,
      "step": 4907
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.399266242980957,
      "learning_rate": 7.519753193712449e-06,
      "loss": 1.0857,
      "step": 4908
    },
    {
      "epoch": 0.35,
      "grad_norm": 17.599817276000977,
      "learning_rate": 7.518750059241367e-06,
      "loss": 1.1613,
      "step": 4909
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.943532943725586,
      "learning_rate": 7.517746788893255e-06,
      "loss": 0.9065,
      "step": 4910
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.198966026306152,
      "learning_rate": 7.516743382722232e-06,
      "loss": 0.8188,
      "step": 4911
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.73505973815918,
      "learning_rate": 7.515739840782432e-06,
      "loss": 0.664,
      "step": 4912
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.541267156600952,
      "learning_rate": 7.5147361631279895e-06,
      "loss": 0.1554,
      "step": 4913
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4824957847595215,
      "learning_rate": 7.51373234981305e-06,
      "loss": 0.1654,
      "step": 4914
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.453948736190796,
      "learning_rate": 7.5127284008917645e-06,
      "loss": 0.1483,
      "step": 4915
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.675987958908081,
      "learning_rate": 7.511724316418292e-06,
      "loss": 0.1488,
      "step": 4916
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.294069766998291,
      "learning_rate": 7.510720096446802e-06,
      "loss": 0.7196,
      "step": 4917
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.077743530273438,
      "learning_rate": 7.509715741031464e-06,
      "loss": 1.0965,
      "step": 4918
    },
    {
      "epoch": 0.35,
      "grad_norm": 33.52392578125,
      "learning_rate": 7.508711250226465e-06,
      "loss": 0.9956,
      "step": 4919
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.4165971279144287,
      "learning_rate": 7.507706624085988e-06,
      "loss": 0.1366,
      "step": 4920
    },
    {
      "epoch": 0.35,
      "grad_norm": 42.00212478637695,
      "learning_rate": 7.50670186266423e-06,
      "loss": 0.9197,
      "step": 4921
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.218473434448242,
      "learning_rate": 7.505696966015397e-06,
      "loss": 1.0446,
      "step": 4922
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.280776500701904,
      "learning_rate": 7.504691934193695e-06,
      "loss": 0.9085,
      "step": 4923
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.232227325439453,
      "learning_rate": 7.503686767253346e-06,
      "loss": 0.9197,
      "step": 4924
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.3520820140838623,
      "learning_rate": 7.502681465248571e-06,
      "loss": 0.2867,
      "step": 4925
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.0906829833984375,
      "learning_rate": 7.501676028233605e-06,
      "loss": 0.3433,
      "step": 4926
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4758925437927246,
      "learning_rate": 7.500670456262687e-06,
      "loss": 0.1045,
      "step": 4927
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.700645446777344,
      "learning_rate": 7.499664749390063e-06,
      "loss": 0.9238,
      "step": 4928
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.050441265106201,
      "learning_rate": 7.498658907669988e-06,
      "loss": 0.0639,
      "step": 4929
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.383262634277344,
      "learning_rate": 7.497652931156723e-06,
      "loss": 1.0037,
      "step": 4930
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.6649293899536133,
      "learning_rate": 7.496646819904536e-06,
      "loss": 0.9534,
      "step": 4931
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.655611991882324,
      "learning_rate": 7.495640573967703e-06,
      "loss": 1.1094,
      "step": 4932
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.6748764514923096,
      "learning_rate": 7.494634193400508e-06,
      "loss": 0.7713,
      "step": 4933
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.99657940864563,
      "learning_rate": 7.49362767825724e-06,
      "loss": 0.2388,
      "step": 4934
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.8764312267303467,
      "learning_rate": 7.492621028592199e-06,
      "loss": 0.2432,
      "step": 4935
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.9697771072387695,
      "learning_rate": 7.491614244459688e-06,
      "loss": 0.9561,
      "step": 4936
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.944233894348145,
      "learning_rate": 7.490607325914021e-06,
      "loss": 0.6477,
      "step": 4937
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.4735324382781982,
      "learning_rate": 7.4896002730095154e-06,
      "loss": 0.8238,
      "step": 4938
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.875801086425781,
      "learning_rate": 7.488593085800498e-06,
      "loss": 0.6532,
      "step": 4939
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.308149814605713,
      "learning_rate": 7.487585764341303e-06,
      "loss": 0.0982,
      "step": 4940
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.111783027648926,
      "learning_rate": 7.486578308686273e-06,
      "loss": 0.0735,
      "step": 4941
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.7451748847961426,
      "learning_rate": 7.485570718889755e-06,
      "loss": 0.3335,
      "step": 4942
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.380502700805664,
      "learning_rate": 7.484562995006106e-06,
      "loss": 0.9879,
      "step": 4943
    },
    {
      "epoch": 0.35,
      "grad_norm": 19.972105026245117,
      "learning_rate": 7.483555137089689e-06,
      "loss": 0.667,
      "step": 4944
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.0497021675109863,
      "learning_rate": 7.48254714519487e-06,
      "loss": 0.2633,
      "step": 4945
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.7516753673553467,
      "learning_rate": 7.4815390193760314e-06,
      "loss": 0.1403,
      "step": 4946
    },
    {
      "epoch": 0.35,
      "grad_norm": 24.930261611938477,
      "learning_rate": 7.480530759687555e-06,
      "loss": 1.2609,
      "step": 4947
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.733661651611328,
      "learning_rate": 7.479522366183834e-06,
      "loss": 1.0939,
      "step": 4948
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.7359962463378906,
      "learning_rate": 7.478513838919268e-06,
      "loss": 0.0766,
      "step": 4949
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.2425994873046875,
      "learning_rate": 7.4775051779482615e-06,
      "loss": 0.2239,
      "step": 4950
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.39075756072998,
      "learning_rate": 7.476496383325227e-06,
      "loss": 0.9378,
      "step": 4951
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8166096210479736,
      "learning_rate": 7.475487455104589e-06,
      "loss": 0.1359,
      "step": 4952
    },
    {
      "epoch": 0.36,
      "grad_norm": 13.730646133422852,
      "learning_rate": 7.474478393340769e-06,
      "loss": 0.8697,
      "step": 4953
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.270735025405884,
      "learning_rate": 7.473469198088209e-06,
      "loss": 0.1287,
      "step": 4954
    },
    {
      "epoch": 0.36,
      "grad_norm": 15.050780296325684,
      "learning_rate": 7.47245986940135e-06,
      "loss": 1.2218,
      "step": 4955
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8149335384368896,
      "learning_rate": 7.471450407334637e-06,
      "loss": 0.1186,
      "step": 4956
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.290149211883545,
      "learning_rate": 7.4704408119425296e-06,
      "loss": 0.7311,
      "step": 4957
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.166351556777954,
      "learning_rate": 7.469431083279492e-06,
      "loss": 0.1832,
      "step": 4958
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6317262649536133,
      "learning_rate": 7.468421221399995e-06,
      "loss": 0.2963,
      "step": 4959
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.522149085998535,
      "learning_rate": 7.4674112263585166e-06,
      "loss": 0.1226,
      "step": 4960
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.280121803283691,
      "learning_rate": 7.466401098209544e-06,
      "loss": 0.9397,
      "step": 4961
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.472233772277832,
      "learning_rate": 7.465390837007566e-06,
      "loss": 0.8851,
      "step": 4962
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.588717460632324,
      "learning_rate": 7.464380442807084e-06,
      "loss": 1.2784,
      "step": 4963
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.2614564895629883,
      "learning_rate": 7.463369915662606e-06,
      "loss": 0.2305,
      "step": 4964
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.923713207244873,
      "learning_rate": 7.462359255628645e-06,
      "loss": 0.6766,
      "step": 4965
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.174336433410645,
      "learning_rate": 7.461348462759723e-06,
      "loss": 0.9636,
      "step": 4966
    },
    {
      "epoch": 0.36,
      "grad_norm": 17.047861099243164,
      "learning_rate": 7.460337537110369e-06,
      "loss": 1.0156,
      "step": 4967
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.6578562259674072,
      "learning_rate": 7.459326478735116e-06,
      "loss": 0.165,
      "step": 4968
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.514452934265137,
      "learning_rate": 7.45831528768851e-06,
      "loss": 0.8179,
      "step": 4969
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.163287401199341,
      "learning_rate": 7.457303964025098e-06,
      "loss": 0.2141,
      "step": 4970
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.672333240509033,
      "learning_rate": 7.456292507799438e-06,
      "loss": 0.6931,
      "step": 4971
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.004265308380127,
      "learning_rate": 7.455280919066098e-06,
      "loss": 0.0674,
      "step": 4972
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.977195382118225,
      "learning_rate": 7.4542691978796445e-06,
      "loss": 0.1021,
      "step": 4973
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.7775535583496094,
      "learning_rate": 7.453257344294658e-06,
      "loss": 0.1503,
      "step": 4974
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.05694580078125,
      "learning_rate": 7.4522453583657226e-06,
      "loss": 0.9574,
      "step": 4975
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.05307388305664,
      "learning_rate": 7.451233240147433e-06,
      "loss": 1.2959,
      "step": 4976
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.172600269317627,
      "learning_rate": 7.4502209896943875e-06,
      "loss": 1.0175,
      "step": 4977
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.418454885482788,
      "learning_rate": 7.449208607061197e-06,
      "loss": 0.1249,
      "step": 4978
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2700793743133545,
      "learning_rate": 7.448196092302471e-06,
      "loss": 0.1091,
      "step": 4979
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.150790214538574,
      "learning_rate": 7.447183445472833e-06,
      "loss": 0.931,
      "step": 4980
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6664254665374756,
      "learning_rate": 7.446170666626912e-06,
      "loss": 0.121,
      "step": 4981
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.367384195327759,
      "learning_rate": 7.445157755819341e-06,
      "loss": 0.1762,
      "step": 4982
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.1001524925231934,
      "learning_rate": 7.4441447131047664e-06,
      "loss": 0.2238,
      "step": 4983
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.632292747497559,
      "learning_rate": 7.443131538537834e-06,
      "loss": 0.6124,
      "step": 4984
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.5402004718780518,
      "learning_rate": 7.442118232173206e-06,
      "loss": 0.1316,
      "step": 4985
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.629934549331665,
      "learning_rate": 7.44110479406554e-06,
      "loss": 0.1355,
      "step": 4986
    },
    {
      "epoch": 0.36,
      "grad_norm": 18.58985710144043,
      "learning_rate": 7.440091224269512e-06,
      "loss": 0.8531,
      "step": 4987
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.860465049743652,
      "learning_rate": 7.4390775228397985e-06,
      "loss": 1.0957,
      "step": 4988
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6657989025115967,
      "learning_rate": 7.438063689831086e-06,
      "loss": 0.1117,
      "step": 4989
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8727636337280273,
      "learning_rate": 7.437049725298063e-06,
      "loss": 0.1458,
      "step": 4990
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2230565547943115,
      "learning_rate": 7.436035629295435e-06,
      "loss": 0.1047,
      "step": 4991
    },
    {
      "epoch": 0.36,
      "grad_norm": 16.96685218811035,
      "learning_rate": 7.435021401877906e-06,
      "loss": 1.3973,
      "step": 4992
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.406099319458008,
      "learning_rate": 7.434007043100188e-06,
      "loss": 0.8404,
      "step": 4993
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.109750747680664,
      "learning_rate": 7.432992553017002e-06,
      "loss": 1.2422,
      "step": 4994
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.239673376083374,
      "learning_rate": 7.4319779316830785e-06,
      "loss": 0.1356,
      "step": 4995
    },
    {
      "epoch": 0.36,
      "grad_norm": 22.470232009887695,
      "learning_rate": 7.430963179153151e-06,
      "loss": 1.086,
      "step": 4996
    },
    {
      "epoch": 0.36,
      "grad_norm": 13.88762378692627,
      "learning_rate": 7.429948295481964e-06,
      "loss": 0.9334,
      "step": 4997
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.9347217082977295,
      "learning_rate": 7.428933280724262e-06,
      "loss": 0.2662,
      "step": 4998
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.257467269897461,
      "learning_rate": 7.427918134934803e-06,
      "loss": 0.1986,
      "step": 4999
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.521291494369507,
      "learning_rate": 7.426902858168351e-06,
      "loss": 0.1873,
      "step": 5000
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.750354290008545,
      "learning_rate": 7.425887450479678e-06,
      "loss": 0.1186,
      "step": 5001
    },
    {
      "epoch": 0.36,
      "grad_norm": 13.407685279846191,
      "learning_rate": 7.4248719119235566e-06,
      "loss": 0.8976,
      "step": 5002
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.780242681503296,
      "learning_rate": 7.4238562425547766e-06,
      "loss": 0.0922,
      "step": 5003
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9738965034484863,
      "learning_rate": 7.422840442428126e-06,
      "loss": 0.8939,
      "step": 5004
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6349329948425293,
      "learning_rate": 7.421824511598405e-06,
      "loss": 0.226,
      "step": 5005
    },
    {
      "epoch": 0.36,
      "grad_norm": 16.778921127319336,
      "learning_rate": 7.420808450120419e-06,
      "loss": 0.6846,
      "step": 5006
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.390873908996582,
      "learning_rate": 7.419792258048979e-06,
      "loss": 0.5689,
      "step": 5007
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.284186840057373,
      "learning_rate": 7.418775935438907e-06,
      "loss": 0.1841,
      "step": 5008
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.443887710571289,
      "learning_rate": 7.417759482345028e-06,
      "loss": 0.914,
      "step": 5009
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.630990982055664,
      "learning_rate": 7.416742898822179e-06,
      "loss": 0.161,
      "step": 5010
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.737186908721924,
      "learning_rate": 7.415726184925196e-06,
      "loss": 0.3247,
      "step": 5011
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.171115875244141,
      "learning_rate": 7.414709340708931e-06,
      "loss": 0.1881,
      "step": 5012
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.6099135875701904,
      "learning_rate": 7.413692366228236e-06,
      "loss": 0.1531,
      "step": 5013
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.012449741363525,
      "learning_rate": 7.412675261537974e-06,
      "loss": 0.8578,
      "step": 5014
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.670966386795044,
      "learning_rate": 7.411658026693016e-06,
      "loss": 0.1454,
      "step": 5015
    },
    {
      "epoch": 0.36,
      "grad_norm": 19.855297088623047,
      "learning_rate": 7.410640661748234e-06,
      "loss": 1.0993,
      "step": 5016
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3816778659820557,
      "learning_rate": 7.4096231667585135e-06,
      "loss": 0.1495,
      "step": 5017
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.527682304382324,
      "learning_rate": 7.408605541778746e-06,
      "loss": 0.9336,
      "step": 5018
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.282783508300781,
      "learning_rate": 7.407587786863823e-06,
      "loss": 0.6735,
      "step": 5019
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.577988386154175,
      "learning_rate": 7.406569902068657e-06,
      "loss": 0.313,
      "step": 5020
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8868223428726196,
      "learning_rate": 7.40555188744815e-06,
      "loss": 0.0766,
      "step": 5021
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.506958484649658,
      "learning_rate": 7.404533743057225e-06,
      "loss": 0.6253,
      "step": 5022
    },
    {
      "epoch": 0.36,
      "grad_norm": 15.149784088134766,
      "learning_rate": 7.403515468950807e-06,
      "loss": 0.9066,
      "step": 5023
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.326829671859741,
      "learning_rate": 7.402497065183827e-06,
      "loss": 0.2264,
      "step": 5024
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.297430038452148,
      "learning_rate": 7.401478531811223e-06,
      "loss": 0.6948,
      "step": 5025
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.767858505249023,
      "learning_rate": 7.400459868887946e-06,
      "loss": 0.9342,
      "step": 5026
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.6456098556518555,
      "learning_rate": 7.399441076468941e-06,
      "loss": 0.8932,
      "step": 5027
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.350851058959961,
      "learning_rate": 7.398422154609175e-06,
      "loss": 0.6263,
      "step": 5028
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.573836803436279,
      "learning_rate": 7.397403103363611e-06,
      "loss": 0.7814,
      "step": 5029
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.493917942047119,
      "learning_rate": 7.396383922787225e-06,
      "loss": 1.0227,
      "step": 5030
    },
    {
      "epoch": 0.36,
      "grad_norm": 16.837034225463867,
      "learning_rate": 7.395364612934997e-06,
      "loss": 1.1799,
      "step": 5031
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.125504493713379,
      "learning_rate": 7.394345173861915e-06,
      "loss": 0.1946,
      "step": 5032
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4086995124816895,
      "learning_rate": 7.393325605622974e-06,
      "loss": 0.0998,
      "step": 5033
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.956493616104126,
      "learning_rate": 7.392305908273176e-06,
      "loss": 0.1613,
      "step": 5034
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.915841102600098,
      "learning_rate": 7.391286081867529e-06,
      "loss": 1.2566,
      "step": 5035
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.40201735496521,
      "learning_rate": 7.39026612646105e-06,
      "loss": 0.2126,
      "step": 5036
    },
    {
      "epoch": 0.36,
      "grad_norm": 20.504499435424805,
      "learning_rate": 7.38924604210876e-06,
      "loss": 1.4949,
      "step": 5037
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.0262768268585205,
      "learning_rate": 7.388225828865691e-06,
      "loss": 0.1242,
      "step": 5038
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.684021949768066,
      "learning_rate": 7.387205486786878e-06,
      "loss": 0.9676,
      "step": 5039
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.440448760986328,
      "learning_rate": 7.386185015927365e-06,
      "loss": 0.4308,
      "step": 5040
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8821346759796143,
      "learning_rate": 7.3851644163422036e-06,
      "loss": 0.1026,
      "step": 5041
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.0286383628845215,
      "learning_rate": 7.384143688086449e-06,
      "loss": 0.908,
      "step": 5042
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3495278358459473,
      "learning_rate": 7.383122831215168e-06,
      "loss": 0.1473,
      "step": 5043
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8378331661224365,
      "learning_rate": 7.382101845783429e-06,
      "loss": 0.1453,
      "step": 5044
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.917500495910645,
      "learning_rate": 7.381080731846313e-06,
      "loss": 0.7481,
      "step": 5045
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.3508317470550537,
      "learning_rate": 7.380059489458906e-06,
      "loss": 0.1931,
      "step": 5046
    },
    {
      "epoch": 0.36,
      "grad_norm": 24.682518005371094,
      "learning_rate": 7.379038118676297e-06,
      "loss": 0.5534,
      "step": 5047
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.442351341247559,
      "learning_rate": 7.3780166195535874e-06,
      "loss": 0.5287,
      "step": 5048
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.68889331817627,
      "learning_rate": 7.376994992145881e-06,
      "loss": 0.7898,
      "step": 5049
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.04150104522705,
      "learning_rate": 7.3759732365082936e-06,
      "loss": 0.9068,
      "step": 5050
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.677830457687378,
      "learning_rate": 7.374951352695943e-06,
      "loss": 0.3074,
      "step": 5051
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.41654634475708,
      "learning_rate": 7.373929340763958e-06,
      "loss": 0.229,
      "step": 5052
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4164183139801025,
      "learning_rate": 7.3729072007674695e-06,
      "loss": 0.1037,
      "step": 5053
    },
    {
      "epoch": 0.36,
      "grad_norm": 15.576800346374512,
      "learning_rate": 7.3718849327616195e-06,
      "loss": 1.0372,
      "step": 5054
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.372862815856934,
      "learning_rate": 7.370862536801555e-06,
      "loss": 0.7617,
      "step": 5055
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.130066156387329,
      "learning_rate": 7.369840012942432e-06,
      "loss": 0.1128,
      "step": 5056
    },
    {
      "epoch": 0.36,
      "grad_norm": 18.94695472717285,
      "learning_rate": 7.368817361239411e-06,
      "loss": 1.3788,
      "step": 5057
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.17134428024292,
      "learning_rate": 7.367794581747658e-06,
      "loss": 0.9741,
      "step": 5058
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.444718360900879,
      "learning_rate": 7.366771674522351e-06,
      "loss": 0.9032,
      "step": 5059
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8433687686920166,
      "learning_rate": 7.365748639618671e-06,
      "loss": 0.2184,
      "step": 5060
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.013381958007812,
      "learning_rate": 7.3647254770918065e-06,
      "loss": 1.1584,
      "step": 5061
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.959115982055664,
      "learning_rate": 7.363702186996953e-06,
      "loss": 0.8588,
      "step": 5062
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.643325805664062,
      "learning_rate": 7.362678769389314e-06,
      "loss": 0.6613,
      "step": 5063
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.034698009490967,
      "learning_rate": 7.361655224324099e-06,
      "loss": 0.1493,
      "step": 5064
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.9229328632354736,
      "learning_rate": 7.360631551856523e-06,
      "loss": 0.2481,
      "step": 5065
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.956249713897705,
      "learning_rate": 7.359607752041812e-06,
      "loss": 0.3379,
      "step": 5066
    },
    {
      "epoch": 0.36,
      "grad_norm": 17.76042938232422,
      "learning_rate": 7.358583824935193e-06,
      "loss": 1.3544,
      "step": 5067
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.413004159927368,
      "learning_rate": 7.357559770591904e-06,
      "loss": 0.1306,
      "step": 5068
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.606311082839966,
      "learning_rate": 7.356535589067191e-06,
      "loss": 0.1573,
      "step": 5069
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.789534568786621,
      "learning_rate": 7.355511280416301e-06,
      "loss": 1.292,
      "step": 5070
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.655467987060547,
      "learning_rate": 7.354486844694494e-06,
      "loss": 0.1624,
      "step": 5071
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.685960531234741,
      "learning_rate": 7.353462281957035e-06,
      "loss": 0.1916,
      "step": 5072
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.18438458442688,
      "learning_rate": 7.352437592259193e-06,
      "loss": 0.1992,
      "step": 5073
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.037373065948486,
      "learning_rate": 7.351412775656248e-06,
      "loss": 0.9311,
      "step": 5074
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.815849781036377,
      "learning_rate": 7.3503878322034834e-06,
      "loss": 0.0674,
      "step": 5075
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.5815513134002686,
      "learning_rate": 7.349362761956194e-06,
      "loss": 0.2696,
      "step": 5076
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.377358436584473,
      "learning_rate": 7.348337564969674e-06,
      "loss": 0.94,
      "step": 5077
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.1601362228393555,
      "learning_rate": 7.347312241299234e-06,
      "loss": 1.0158,
      "step": 5078
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.0204520225524902,
      "learning_rate": 7.346286791000181e-06,
      "loss": 0.1117,
      "step": 5079
    },
    {
      "epoch": 0.36,
      "grad_norm": 15.922263145446777,
      "learning_rate": 7.345261214127839e-06,
      "loss": 0.7421,
      "step": 5080
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.783235549926758,
      "learning_rate": 7.3442355107375296e-06,
      "loss": 0.1428,
      "step": 5081
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.8396453857421875,
      "learning_rate": 7.34320968088459e-06,
      "loss": 0.9146,
      "step": 5082
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.098783016204834,
      "learning_rate": 7.342183724624357e-06,
      "loss": 0.6951,
      "step": 5083
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.7557122707366943,
      "learning_rate": 7.3411576420121765e-06,
      "loss": 0.1157,
      "step": 5084
    },
    {
      "epoch": 0.36,
      "grad_norm": 13.88252067565918,
      "learning_rate": 7.3401314331034035e-06,
      "loss": 0.834,
      "step": 5085
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.903164863586426,
      "learning_rate": 7.339105097953398e-06,
      "loss": 0.9203,
      "step": 5086
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8664346933364868,
      "learning_rate": 7.338078636617526e-06,
      "loss": 0.0808,
      "step": 5087
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.541516065597534,
      "learning_rate": 7.337052049151162e-06,
      "loss": 0.1129,
      "step": 5088
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.50734281539917,
      "learning_rate": 7.3360253356096865e-06,
      "loss": 0.9936,
      "step": 5089
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.7207961082458496,
      "learning_rate": 7.3349984960484856e-06,
      "loss": 0.2484,
      "step": 5090
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.513929843902588,
      "learning_rate": 7.333971530522954e-06,
      "loss": 0.2313,
      "step": 5091
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.32576847076416,
      "learning_rate": 7.332944439088494e-06,
      "loss": 0.3001,
      "step": 5092
    },
    {
      "epoch": 0.37,
      "grad_norm": 21.382129669189453,
      "learning_rate": 7.331917221800512e-06,
      "loss": 1.7127,
      "step": 5093
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.513020038604736,
      "learning_rate": 7.3308898787144235e-06,
      "loss": 0.885,
      "step": 5094
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.584336757659912,
      "learning_rate": 7.3298624098856486e-06,
      "loss": 0.8992,
      "step": 5095
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.498171329498291,
      "learning_rate": 7.3288348153696145e-06,
      "loss": 0.9854,
      "step": 5096
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.9565160274505615,
      "learning_rate": 7.327807095221757e-06,
      "loss": 0.8583,
      "step": 5097
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.971063613891602,
      "learning_rate": 7.326779249497519e-06,
      "loss": 0.981,
      "step": 5098
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1740171909332275,
      "learning_rate": 7.325751278252346e-06,
      "loss": 0.0774,
      "step": 5099
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.51494836807251,
      "learning_rate": 7.324723181541698e-06,
      "loss": 0.3949,
      "step": 5100
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0721945762634277,
      "learning_rate": 7.323694959421033e-06,
      "loss": 0.2157,
      "step": 5101
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0880937576293945,
      "learning_rate": 7.322666611945819e-06,
      "loss": 0.228,
      "step": 5102
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.941342830657959,
      "learning_rate": 7.321638139171535e-06,
      "loss": 0.5944,
      "step": 5103
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.342085123062134,
      "learning_rate": 7.320609541153661e-06,
      "loss": 0.2019,
      "step": 5104
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.93371319770813,
      "learning_rate": 7.3195808179476846e-06,
      "loss": 0.1736,
      "step": 5105
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3187103271484375,
      "learning_rate": 7.318551969609104e-06,
      "loss": 0.1154,
      "step": 5106
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.735330820083618,
      "learning_rate": 7.3175229961934215e-06,
      "loss": 0.1167,
      "step": 5107
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.524962425231934,
      "learning_rate": 7.316493897756145e-06,
      "loss": 0.4527,
      "step": 5108
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.6994400024414062,
      "learning_rate": 7.315464674352791e-06,
      "loss": 0.1765,
      "step": 5109
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8290891647338867,
      "learning_rate": 7.3144353260388825e-06,
      "loss": 0.1336,
      "step": 5110
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.3707313537597656,
      "learning_rate": 7.313405852869949e-06,
      "loss": 0.24,
      "step": 5111
    },
    {
      "epoch": 0.37,
      "grad_norm": 17.91295051574707,
      "learning_rate": 7.312376254901528e-06,
      "loss": 0.9847,
      "step": 5112
    },
    {
      "epoch": 0.37,
      "grad_norm": 16.813446044921875,
      "learning_rate": 7.311346532189159e-06,
      "loss": 1.4722,
      "step": 5113
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2677528858184814,
      "learning_rate": 7.310316684788393e-06,
      "loss": 0.0839,
      "step": 5114
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8876354694366455,
      "learning_rate": 7.309286712754787e-06,
      "loss": 0.1287,
      "step": 5115
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.291229248046875,
      "learning_rate": 7.308256616143905e-06,
      "loss": 0.7426,
      "step": 5116
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.4855122566223145,
      "learning_rate": 7.307226395011316e-06,
      "loss": 0.6494,
      "step": 5117
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.703749179840088,
      "learning_rate": 7.306196049412595e-06,
      "loss": 0.6478,
      "step": 5118
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3750689029693604,
      "learning_rate": 7.305165579403327e-06,
      "loss": 0.0923,
      "step": 5119
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.811850547790527,
      "learning_rate": 7.304134985039102e-06,
      "loss": 1.1234,
      "step": 5120
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.81275749206543,
      "learning_rate": 7.3031042663755156e-06,
      "loss": 1.6421,
      "step": 5121
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.912071228027344,
      "learning_rate": 7.302073423468173e-06,
      "loss": 0.5562,
      "step": 5122
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.162311553955078,
      "learning_rate": 7.301042456372681e-06,
      "loss": 0.1992,
      "step": 5123
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.2166457176208496,
      "learning_rate": 7.3000113651446605e-06,
      "loss": 0.2131,
      "step": 5124
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9931223392486572,
      "learning_rate": 7.298980149839731e-06,
      "loss": 0.1486,
      "step": 5125
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.33343505859375,
      "learning_rate": 7.297948810513526e-06,
      "loss": 0.9446,
      "step": 5126
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.709509372711182,
      "learning_rate": 7.29691734722168e-06,
      "loss": 0.328,
      "step": 5127
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.709930896759033,
      "learning_rate": 7.295885760019838e-06,
      "loss": 0.1437,
      "step": 5128
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3715786933898926,
      "learning_rate": 7.294854048963649e-06,
      "loss": 0.0824,
      "step": 5129
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.357173919677734,
      "learning_rate": 7.293822214108771e-06,
      "loss": 1.0878,
      "step": 5130
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.4427475929260254,
      "learning_rate": 7.2927902555108665e-06,
      "loss": 0.143,
      "step": 5131
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.592174053192139,
      "learning_rate": 7.291758173225608e-06,
      "loss": 0.8175,
      "step": 5132
    },
    {
      "epoch": 0.37,
      "grad_norm": 39.36619567871094,
      "learning_rate": 7.290725967308669e-06,
      "loss": 1.7387,
      "step": 5133
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.315417289733887,
      "learning_rate": 7.2896936378157355e-06,
      "loss": 1.137,
      "step": 5134
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.22851824760437,
      "learning_rate": 7.288661184802498e-06,
      "loss": 0.1798,
      "step": 5135
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.198528289794922,
      "learning_rate": 7.287628608324652e-06,
      "loss": 0.2227,
      "step": 5136
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9180476665496826,
      "learning_rate": 7.286595908437902e-06,
      "loss": 0.1649,
      "step": 5137
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.633812665939331,
      "learning_rate": 7.2855630851979575e-06,
      "loss": 0.1291,
      "step": 5138
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.3498036861419678,
      "learning_rate": 7.284530138660536e-06,
      "loss": 0.2427,
      "step": 5139
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.3028106689453125,
      "learning_rate": 7.283497068881361e-06,
      "loss": 0.9446,
      "step": 5140
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.968301296234131,
      "learning_rate": 7.282463875916162e-06,
      "loss": 0.2874,
      "step": 5141
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0129177570343018,
      "learning_rate": 7.281430559820678e-06,
      "loss": 0.1428,
      "step": 5142
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.062557697296143,
      "learning_rate": 7.28039712065065e-06,
      "loss": 0.7647,
      "step": 5143
    },
    {
      "epoch": 0.37,
      "grad_norm": 20.41753578186035,
      "learning_rate": 7.27936355846183e-06,
      "loss": 1.1423,
      "step": 5144
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.898650050163269,
      "learning_rate": 7.278329873309973e-06,
      "loss": 0.106,
      "step": 5145
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.5067951679229736,
      "learning_rate": 7.277296065250843e-06,
      "loss": 0.2648,
      "step": 5146
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.321563720703125,
      "learning_rate": 7.276262134340211e-06,
      "loss": 0.1248,
      "step": 5147
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.7945773601531982,
      "learning_rate": 7.275228080633853e-06,
      "loss": 0.9498,
      "step": 5148
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.284545660018921,
      "learning_rate": 7.274193904187552e-06,
      "loss": 0.1999,
      "step": 5149
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.210003852844238,
      "learning_rate": 7.273159605057098e-06,
      "loss": 1.1854,
      "step": 5150
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.029271125793457,
      "learning_rate": 7.2721251832982875e-06,
      "loss": 0.8759,
      "step": 5151
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.608834743499756,
      "learning_rate": 7.271090638966924e-06,
      "loss": 0.3163,
      "step": 5152
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.530630350112915,
      "learning_rate": 7.270055972118817e-06,
      "loss": 0.111,
      "step": 5153
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2706029415130615,
      "learning_rate": 7.269021182809785e-06,
      "loss": 0.0965,
      "step": 5154
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.241036891937256,
      "learning_rate": 7.2679862710956465e-06,
      "loss": 0.2938,
      "step": 5155
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.8791680335998535,
      "learning_rate": 7.266951237032232e-06,
      "loss": 0.8596,
      "step": 5156
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.496885299682617,
      "learning_rate": 7.265916080675383e-06,
      "loss": 0.3994,
      "step": 5157
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.150147438049316,
      "learning_rate": 7.264880802080936e-06,
      "loss": 0.7634,
      "step": 5158
    },
    {
      "epoch": 0.37,
      "grad_norm": 12.041993141174316,
      "learning_rate": 7.263845401304742e-06,
      "loss": 0.9558,
      "step": 5159
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.141373634338379,
      "learning_rate": 7.2628098784026615e-06,
      "loss": 0.1665,
      "step": 5160
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.104431629180908,
      "learning_rate": 7.26177423343055e-06,
      "loss": 0.208,
      "step": 5161
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.296441078186035,
      "learning_rate": 7.26073846644428e-06,
      "loss": 1.1253,
      "step": 5162
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.3326194286346436,
      "learning_rate": 7.259702577499728e-06,
      "loss": 0.1433,
      "step": 5163
    },
    {
      "epoch": 0.37,
      "grad_norm": 13.410021781921387,
      "learning_rate": 7.258666566652774e-06,
      "loss": 0.8806,
      "step": 5164
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.058321714401245,
      "learning_rate": 7.257630433959311e-06,
      "loss": 0.2043,
      "step": 5165
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.498013019561768,
      "learning_rate": 7.256594179475231e-06,
      "loss": 0.7876,
      "step": 5166
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.5492749214172363,
      "learning_rate": 7.255557803256434e-06,
      "loss": 0.1459,
      "step": 5167
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.882083415985107,
      "learning_rate": 7.254521305358835e-06,
      "loss": 0.5728,
      "step": 5168
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.728192090988159,
      "learning_rate": 7.253484685838343e-06,
      "loss": 0.2047,
      "step": 5169
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.668449640274048,
      "learning_rate": 7.252447944750881e-06,
      "loss": 0.1385,
      "step": 5170
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.568369746208191,
      "learning_rate": 7.251411082152381e-06,
      "loss": 0.0747,
      "step": 5171
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.367377281188965,
      "learning_rate": 7.250374098098773e-06,
      "loss": 1.0499,
      "step": 5172
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.42080020904541,
      "learning_rate": 7.249336992646002e-06,
      "loss": 0.1211,
      "step": 5173
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.5278944969177246,
      "learning_rate": 7.248299765850016e-06,
      "loss": 0.1206,
      "step": 5174
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.921939849853516,
      "learning_rate": 7.247262417766765e-06,
      "loss": 1.3019,
      "step": 5175
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.7040159702301025,
      "learning_rate": 7.246224948452213e-06,
      "loss": 0.1544,
      "step": 5176
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.033246040344238,
      "learning_rate": 7.245187357962329e-06,
      "loss": 1.092,
      "step": 5177
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.984907388687134,
      "learning_rate": 7.2441496463530845e-06,
      "loss": 0.117,
      "step": 5178
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0687003135681152,
      "learning_rate": 7.243111813680461e-06,
      "loss": 0.2119,
      "step": 5179
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.992896556854248,
      "learning_rate": 7.242073860000447e-06,
      "loss": 0.8729,
      "step": 5180
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.021839141845703,
      "learning_rate": 7.241035785369034e-06,
      "loss": 0.3738,
      "step": 5181
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.7585911750793457,
      "learning_rate": 7.239997589842223e-06,
      "loss": 0.9363,
      "step": 5182
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3877575397491455,
      "learning_rate": 7.238959273476022e-06,
      "loss": 0.1437,
      "step": 5183
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.456559181213379,
      "learning_rate": 7.237920836326443e-06,
      "loss": 0.8635,
      "step": 5184
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.999457359313965,
      "learning_rate": 7.236882278449504e-06,
      "loss": 0.7676,
      "step": 5185
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.864264488220215,
      "learning_rate": 7.235843599901235e-06,
      "loss": 0.1484,
      "step": 5186
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.4840457439422607,
      "learning_rate": 7.234804800737667e-06,
      "loss": 0.8797,
      "step": 5187
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.162583351135254,
      "learning_rate": 7.233765881014838e-06,
      "loss": 0.781,
      "step": 5188
    },
    {
      "epoch": 0.37,
      "grad_norm": 14.274267196655273,
      "learning_rate": 7.232726840788795e-06,
      "loss": 1.4559,
      "step": 5189
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.516021728515625,
      "learning_rate": 7.231687680115591e-06,
      "loss": 0.3385,
      "step": 5190
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.617876052856445,
      "learning_rate": 7.230648399051285e-06,
      "loss": 0.9351,
      "step": 5191
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.761577606201172,
      "learning_rate": 7.22960899765194e-06,
      "loss": 0.1487,
      "step": 5192
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.304812431335449,
      "learning_rate": 7.228569475973628e-06,
      "loss": 0.0951,
      "step": 5193
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.4805727005004883,
      "learning_rate": 7.227529834072431e-06,
      "loss": 0.2247,
      "step": 5194
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.6071794033050537,
      "learning_rate": 7.226490072004428e-06,
      "loss": 0.0632,
      "step": 5195
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.463109493255615,
      "learning_rate": 7.225450189825717e-06,
      "loss": 0.9606,
      "step": 5196
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8943428993225098,
      "learning_rate": 7.22441018759239e-06,
      "loss": 0.2585,
      "step": 5197
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.650480270385742,
      "learning_rate": 7.2233700653605534e-06,
      "loss": 0.7963,
      "step": 5198
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.025178909301758,
      "learning_rate": 7.222329823186317e-06,
      "loss": 0.1205,
      "step": 5199
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.070739269256592,
      "learning_rate": 7.2212894611258e-06,
      "loss": 0.6661,
      "step": 5200
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.448754787445068,
      "learning_rate": 7.220248979235123e-06,
      "loss": 0.5653,
      "step": 5201
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.189923286437988,
      "learning_rate": 7.219208377570419e-06,
      "loss": 0.9421,
      "step": 5202
    },
    {
      "epoch": 0.37,
      "grad_norm": 44.72954177856445,
      "learning_rate": 7.218167656187823e-06,
      "loss": 1.1782,
      "step": 5203
    },
    {
      "epoch": 0.37,
      "grad_norm": 13.769665718078613,
      "learning_rate": 7.217126815143478e-06,
      "loss": 0.9744,
      "step": 5204
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.3325276374816895,
      "learning_rate": 7.216085854493534e-06,
      "loss": 0.1959,
      "step": 5205
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.177857398986816,
      "learning_rate": 7.215044774294145e-06,
      "loss": 0.9039,
      "step": 5206
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.198053359985352,
      "learning_rate": 7.214003574601476e-06,
      "loss": 0.8415,
      "step": 5207
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2963361740112305,
      "learning_rate": 7.212962255471696e-06,
      "loss": 0.1245,
      "step": 5208
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9866740703582764,
      "learning_rate": 7.211920816960976e-06,
      "loss": 0.1665,
      "step": 5209
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.2674148082733154,
      "learning_rate": 7.210879259125502e-06,
      "loss": 0.195,
      "step": 5210
    },
    {
      "epoch": 0.37,
      "grad_norm": 20.943246841430664,
      "learning_rate": 7.209837582021461e-06,
      "loss": 0.7228,
      "step": 5211
    },
    {
      "epoch": 0.37,
      "grad_norm": 12.179662704467773,
      "learning_rate": 7.2087957857050454e-06,
      "loss": 1.3468,
      "step": 5212
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.2096586227417,
      "learning_rate": 7.207753870232459e-06,
      "loss": 0.6108,
      "step": 5213
    },
    {
      "epoch": 0.37,
      "grad_norm": 12.203096389770508,
      "learning_rate": 7.20671183565991e-06,
      "loss": 1.1949,
      "step": 5214
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.647597312927246,
      "learning_rate": 7.205669682043608e-06,
      "loss": 0.8774,
      "step": 5215
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.667799949645996,
      "learning_rate": 7.204627409439775e-06,
      "loss": 0.8183,
      "step": 5216
    },
    {
      "epoch": 0.37,
      "grad_norm": 14.162248611450195,
      "learning_rate": 7.20358501790464e-06,
      "loss": 0.7318,
      "step": 5217
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.164416551589966,
      "learning_rate": 7.202542507494432e-06,
      "loss": 0.2023,
      "step": 5218
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9529452323913574,
      "learning_rate": 7.201499878265396e-06,
      "loss": 0.228,
      "step": 5219
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.656495213508606,
      "learning_rate": 7.200457130273772e-06,
      "loss": 0.0483,
      "step": 5220
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.966113805770874,
      "learning_rate": 7.199414263575817e-06,
      "loss": 0.1781,
      "step": 5221
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.48490571975708,
      "learning_rate": 7.198371278227787e-06,
      "loss": 0.132,
      "step": 5222
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.817744731903076,
      "learning_rate": 7.197328174285947e-06,
      "loss": 0.2162,
      "step": 5223
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.168372869491577,
      "learning_rate": 7.1962849518065695e-06,
      "loss": 0.1512,
      "step": 5224
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8686134815216064,
      "learning_rate": 7.195241610845934e-06,
      "loss": 0.157,
      "step": 5225
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.35203742980957,
      "learning_rate": 7.1941981514603195e-06,
      "loss": 1.0636,
      "step": 5226
    },
    {
      "epoch": 0.37,
      "grad_norm": 13.324507713317871,
      "learning_rate": 7.1931545737060234e-06,
      "loss": 0.7046,
      "step": 5227
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.612225532531738,
      "learning_rate": 7.19211087763934e-06,
      "loss": 1.358,
      "step": 5228
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.847691535949707,
      "learning_rate": 7.19106706331657e-06,
      "loss": 0.8467,
      "step": 5229
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.3299167156219482,
      "learning_rate": 7.190023130794026e-06,
      "loss": 0.1452,
      "step": 5230
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6831676959991455,
      "learning_rate": 7.188979080128024e-06,
      "loss": 0.0702,
      "step": 5231
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.384594917297363,
      "learning_rate": 7.187934911374886e-06,
      "loss": 0.986,
      "step": 5232
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.891510009765625,
      "learning_rate": 7.186890624590943e-06,
      "loss": 1.3485,
      "step": 5233
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.469616889953613,
      "learning_rate": 7.185846219832528e-06,
      "loss": 0.7762,
      "step": 5234
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.32863712310791,
      "learning_rate": 7.184801697155983e-06,
      "loss": 1.0657,
      "step": 5235
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.9979336261749268,
      "learning_rate": 7.183757056617655e-06,
      "loss": 0.2336,
      "step": 5236
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.798501491546631,
      "learning_rate": 7.182712298273902e-06,
      "loss": 0.0985,
      "step": 5237
    },
    {
      "epoch": 0.38,
      "grad_norm": 26.43146514892578,
      "learning_rate": 7.181667422181081e-06,
      "loss": 1.3772,
      "step": 5238
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.053902626037598,
      "learning_rate": 7.180622428395562e-06,
      "loss": 0.8878,
      "step": 5239
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1446030139923096,
      "learning_rate": 7.179577316973717e-06,
      "loss": 0.1459,
      "step": 5240
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.76420783996582,
      "learning_rate": 7.1785320879719246e-06,
      "loss": 0.8891,
      "step": 5241
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.477489948272705,
      "learning_rate": 7.177486741446573e-06,
      "loss": 0.8941,
      "step": 5242
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.088459968566895,
      "learning_rate": 7.176441277454054e-06,
      "loss": 0.8956,
      "step": 5243
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.761130332946777,
      "learning_rate": 7.175395696050766e-06,
      "loss": 0.8512,
      "step": 5244
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.380803108215332,
      "learning_rate": 7.174349997293115e-06,
      "loss": 0.8052,
      "step": 5245
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.364364147186279,
      "learning_rate": 7.173304181237512e-06,
      "loss": 0.8063,
      "step": 5246
    },
    {
      "epoch": 0.38,
      "grad_norm": 15.673524856567383,
      "learning_rate": 7.172258247940373e-06,
      "loss": 0.9546,
      "step": 5247
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.060466289520264,
      "learning_rate": 7.171212197458126e-06,
      "loss": 0.8833,
      "step": 5248
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.8930907249450684,
      "learning_rate": 7.170166029847199e-06,
      "loss": 0.8125,
      "step": 5249
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.179396390914917,
      "learning_rate": 7.169119745164028e-06,
      "loss": 0.2212,
      "step": 5250
    },
    {
      "epoch": 0.38,
      "grad_norm": 15.052206039428711,
      "learning_rate": 7.1680733434650605e-06,
      "loss": 1.0063,
      "step": 5251
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.835391998291016,
      "learning_rate": 7.16702682480674e-06,
      "loss": 1.2178,
      "step": 5252
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.097670555114746,
      "learning_rate": 7.165980189245525e-06,
      "loss": 0.7539,
      "step": 5253
    },
    {
      "epoch": 0.38,
      "grad_norm": 22.25922203063965,
      "learning_rate": 7.164933436837879e-06,
      "loss": 0.7046,
      "step": 5254
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.9494075775146484,
      "learning_rate": 7.1638865676402665e-06,
      "loss": 0.1681,
      "step": 5255
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.50362491607666,
      "learning_rate": 7.162839581709167e-06,
      "loss": 0.1415,
      "step": 5256
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.3143563270568848,
      "learning_rate": 7.161792479101059e-06,
      "loss": 0.2345,
      "step": 5257
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.483457326889038,
      "learning_rate": 7.160745259872427e-06,
      "loss": 0.2635,
      "step": 5258
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.834622383117676,
      "learning_rate": 7.159697924079769e-06,
      "loss": 0.8153,
      "step": 5259
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.0623605251312256,
      "learning_rate": 7.158650471779581e-06,
      "loss": 0.1705,
      "step": 5260
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.497754096984863,
      "learning_rate": 7.157602903028372e-06,
      "loss": 0.9656,
      "step": 5261
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.82512378692627,
      "learning_rate": 7.156555217882655e-06,
      "loss": 0.984,
      "step": 5262
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.204044342041016,
      "learning_rate": 7.155507416398946e-06,
      "loss": 1.231,
      "step": 5263
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.9847490787506104,
      "learning_rate": 7.154459498633769e-06,
      "loss": 0.1825,
      "step": 5264
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.0183446407318115,
      "learning_rate": 7.1534114646436596e-06,
      "loss": 0.1201,
      "step": 5265
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.856626510620117,
      "learning_rate": 7.152363314485151e-06,
      "loss": 0.2381,
      "step": 5266
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.965510368347168,
      "learning_rate": 7.151315048214789e-06,
      "loss": 0.9748,
      "step": 5267
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.079838275909424,
      "learning_rate": 7.150266665889125e-06,
      "loss": 0.9204,
      "step": 5268
    },
    {
      "epoch": 0.38,
      "grad_norm": 16.96560287475586,
      "learning_rate": 7.149218167564712e-06,
      "loss": 1.3032,
      "step": 5269
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.540570259094238,
      "learning_rate": 7.1481695532981145e-06,
      "loss": 0.9312,
      "step": 5270
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6300514936447144,
      "learning_rate": 7.147120823145901e-06,
      "loss": 0.0646,
      "step": 5271
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5569639205932617,
      "learning_rate": 7.146071977164646e-06,
      "loss": 0.1175,
      "step": 5272
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9553309679031372,
      "learning_rate": 7.1450230154109325e-06,
      "loss": 0.0945,
      "step": 5273
    },
    {
      "epoch": 0.38,
      "grad_norm": 12.970060348510742,
      "learning_rate": 7.1439739379413445e-06,
      "loss": 1.362,
      "step": 5274
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.187320709228516,
      "learning_rate": 7.14292474481248e-06,
      "loss": 1.1822,
      "step": 5275
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1744585037231445,
      "learning_rate": 7.141875436080935e-06,
      "loss": 0.1478,
      "step": 5276
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.454512119293213,
      "learning_rate": 7.140826011803319e-06,
      "loss": 0.1392,
      "step": 5277
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.268495082855225,
      "learning_rate": 7.139776472036242e-06,
      "loss": 0.564,
      "step": 5278
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3336262702941895,
      "learning_rate": 7.138726816836325e-06,
      "loss": 0.0774,
      "step": 5279
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8170876502990723,
      "learning_rate": 7.137677046260191e-06,
      "loss": 0.0988,
      "step": 5280
    },
    {
      "epoch": 0.38,
      "grad_norm": 14.891763687133789,
      "learning_rate": 7.136627160364473e-06,
      "loss": 1.4614,
      "step": 5281
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1373724937438965,
      "learning_rate": 7.135577159205807e-06,
      "loss": 0.2138,
      "step": 5282
    },
    {
      "epoch": 0.38,
      "grad_norm": 15.346732139587402,
      "learning_rate": 7.134527042840835e-06,
      "loss": 1.1168,
      "step": 5283
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.051906108856201,
      "learning_rate": 7.1334768113262096e-06,
      "loss": 0.2256,
      "step": 5284
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1814682483673096,
      "learning_rate": 7.132426464718587e-06,
      "loss": 0.1118,
      "step": 5285
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.457536220550537,
      "learning_rate": 7.131376003074627e-06,
      "loss": 0.1278,
      "step": 5286
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6373815536499023,
      "learning_rate": 7.130325426451002e-06,
      "loss": 0.1088,
      "step": 5287
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.401608467102051,
      "learning_rate": 7.129274734904381e-06,
      "loss": 0.1943,
      "step": 5288
    },
    {
      "epoch": 0.38,
      "grad_norm": 16.16236114501953,
      "learning_rate": 7.128223928491449e-06,
      "loss": 1.3875,
      "step": 5289
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.909681797027588,
      "learning_rate": 7.127173007268891e-06,
      "loss": 0.9036,
      "step": 5290
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5743002891540527,
      "learning_rate": 7.126121971293402e-06,
      "loss": 0.1699,
      "step": 5291
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.823534965515137,
      "learning_rate": 7.125070820621679e-06,
      "loss": 1.1504,
      "step": 5292
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.336620807647705,
      "learning_rate": 7.124019555310431e-06,
      "loss": 0.8969,
      "step": 5293
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.052638053894043,
      "learning_rate": 7.122968175416367e-06,
      "loss": 0.9664,
      "step": 5294
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.084172010421753,
      "learning_rate": 7.121916680996205e-06,
      "loss": 0.1825,
      "step": 5295
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.74445104598999,
      "learning_rate": 7.1208650721066705e-06,
      "loss": 0.6855,
      "step": 5296
    },
    {
      "epoch": 0.38,
      "grad_norm": 79.6749038696289,
      "learning_rate": 7.119813348804494e-06,
      "loss": 1.4007,
      "step": 5297
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.927577972412109,
      "learning_rate": 7.118761511146411e-06,
      "loss": 0.9027,
      "step": 5298
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.3378872871398926,
      "learning_rate": 7.117709559189165e-06,
      "loss": 0.2934,
      "step": 5299
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6371448040008545,
      "learning_rate": 7.1166574929895026e-06,
      "loss": 0.0426,
      "step": 5300
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.73244571685791,
      "learning_rate": 7.11560531260418e-06,
      "loss": 0.6309,
      "step": 5301
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.9972617626190186,
      "learning_rate": 7.11455301808996e-06,
      "loss": 0.1393,
      "step": 5302
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.601076602935791,
      "learning_rate": 7.11350060950361e-06,
      "loss": 0.8974,
      "step": 5303
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.156611204147339,
      "learning_rate": 7.1124480869018995e-06,
      "loss": 0.232,
      "step": 5304
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.804024696350098,
      "learning_rate": 7.111395450341614e-06,
      "loss": 1.0391,
      "step": 5305
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.932177543640137,
      "learning_rate": 7.110342699879533e-06,
      "loss": 0.8413,
      "step": 5306
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.328266143798828,
      "learning_rate": 7.109289835572452e-06,
      "loss": 0.5475,
      "step": 5307
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.464007616043091,
      "learning_rate": 7.10823685747717e-06,
      "loss": 0.1704,
      "step": 5308
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.02845287322998,
      "learning_rate": 7.107183765650487e-06,
      "loss": 0.5243,
      "step": 5309
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.088887691497803,
      "learning_rate": 7.106130560149218e-06,
      "loss": 1.0308,
      "step": 5310
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.519313812255859,
      "learning_rate": 7.105077241030178e-06,
      "loss": 1.0676,
      "step": 5311
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.7179479598999023,
      "learning_rate": 7.104023808350188e-06,
      "loss": 0.3286,
      "step": 5312
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.255537509918213,
      "learning_rate": 7.102970262166078e-06,
      "loss": 0.9491,
      "step": 5313
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.413334369659424,
      "learning_rate": 7.101916602534682e-06,
      "loss": 0.1563,
      "step": 5314
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.13401460647583,
      "learning_rate": 7.100862829512841e-06,
      "loss": 0.9558,
      "step": 5315
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6689720153808594,
      "learning_rate": 7.099808943157403e-06,
      "loss": 0.305,
      "step": 5316
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.856374979019165,
      "learning_rate": 7.0987549435252225e-06,
      "loss": 0.2194,
      "step": 5317
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.697176933288574,
      "learning_rate": 7.097700830673155e-06,
      "loss": 1.222,
      "step": 5318
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.244558334350586,
      "learning_rate": 7.096646604658068e-06,
      "loss": 0.9455,
      "step": 5319
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.298550605773926,
      "learning_rate": 7.095592265536834e-06,
      "loss": 0.7141,
      "step": 5320
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.9850239753723145,
      "learning_rate": 7.094537813366328e-06,
      "loss": 0.1329,
      "step": 5321
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.195126533508301,
      "learning_rate": 7.093483248203436e-06,
      "loss": 0.141,
      "step": 5322
    },
    {
      "epoch": 0.38,
      "grad_norm": 12.614649772644043,
      "learning_rate": 7.0924285701050475e-06,
      "loss": 0.7773,
      "step": 5323
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.9678492546081543,
      "learning_rate": 7.091373779128058e-06,
      "loss": 1.0015,
      "step": 5324
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.211848735809326,
      "learning_rate": 7.090318875329369e-06,
      "loss": 0.13,
      "step": 5325
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.627896308898926,
      "learning_rate": 7.089263858765889e-06,
      "loss": 1.0863,
      "step": 5326
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6446239948272705,
      "learning_rate": 7.0882087294945325e-06,
      "loss": 0.1938,
      "step": 5327
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.2335734367370605,
      "learning_rate": 7.087153487572218e-06,
      "loss": 0.9951,
      "step": 5328
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.585373878479004,
      "learning_rate": 7.086098133055874e-06,
      "loss": 0.9804,
      "step": 5329
    },
    {
      "epoch": 0.38,
      "grad_norm": 14.212227821350098,
      "learning_rate": 7.085042666002432e-06,
      "loss": 1.2882,
      "step": 5330
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.758255958557129,
      "learning_rate": 7.083987086468831e-06,
      "loss": 0.9233,
      "step": 5331
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.961108207702637,
      "learning_rate": 7.082931394512013e-06,
      "loss": 0.759,
      "step": 5332
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5488905906677246,
      "learning_rate": 7.081875590188934e-06,
      "loss": 0.1121,
      "step": 5333
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.573791027069092,
      "learning_rate": 7.080819673556545e-06,
      "loss": 0.7277,
      "step": 5334
    },
    {
      "epoch": 0.38,
      "grad_norm": 16.377363204956055,
      "learning_rate": 7.079763644671811e-06,
      "loss": 0.9063,
      "step": 5335
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6543242931365967,
      "learning_rate": 7.078707503591702e-06,
      "loss": 0.2075,
      "step": 5336
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.032752990722656,
      "learning_rate": 7.07765125037319e-06,
      "loss": 0.851,
      "step": 5337
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.9426169395446777,
      "learning_rate": 7.076594885073257e-06,
      "loss": 0.9197,
      "step": 5338
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6035380363464355,
      "learning_rate": 7.075538407748892e-06,
      "loss": 0.1559,
      "step": 5339
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.103360176086426,
      "learning_rate": 7.074481818457084e-06,
      "loss": 0.2089,
      "step": 5340
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.55731201171875,
      "learning_rate": 7.073425117254838e-06,
      "loss": 1.0199,
      "step": 5341
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.7478976249694824,
      "learning_rate": 7.072368304199152e-06,
      "loss": 0.0891,
      "step": 5342
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.057966232299805,
      "learning_rate": 7.0713113793470415e-06,
      "loss": 0.914,
      "step": 5343
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.264172554016113,
      "learning_rate": 7.07025434275552e-06,
      "loss": 0.8897,
      "step": 5344
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.576478958129883,
      "learning_rate": 7.069197194481616e-06,
      "loss": 0.1362,
      "step": 5345
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.965652346611023,
      "learning_rate": 7.068139934582355e-06,
      "loss": 0.0845,
      "step": 5346
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.244164228439331,
      "learning_rate": 7.067082563114773e-06,
      "loss": 0.1527,
      "step": 5347
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1165573596954346,
      "learning_rate": 7.066025080135912e-06,
      "loss": 0.1417,
      "step": 5348
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.367778778076172,
      "learning_rate": 7.064967485702816e-06,
      "loss": 0.0934,
      "step": 5349
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.246878623962402,
      "learning_rate": 7.063909779872542e-06,
      "loss": 0.747,
      "step": 5350
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.890291690826416,
      "learning_rate": 7.062851962702148e-06,
      "loss": 0.1486,
      "step": 5351
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.498520851135254,
      "learning_rate": 7.0617940342486965e-06,
      "loss": 1.138,
      "step": 5352
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.456752061843872,
      "learning_rate": 7.0607359945692645e-06,
      "loss": 0.1512,
      "step": 5353
    },
    {
      "epoch": 0.38,
      "grad_norm": 12.278377532958984,
      "learning_rate": 7.059677843720924e-06,
      "loss": 0.9564,
      "step": 5354
    },
    {
      "epoch": 0.38,
      "grad_norm": 12.5022611618042,
      "learning_rate": 7.0586195817607595e-06,
      "loss": 1.0591,
      "step": 5355
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.25756549835205,
      "learning_rate": 7.057561208745862e-06,
      "loss": 0.6539,
      "step": 5356
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.950228214263916,
      "learning_rate": 7.056502724733325e-06,
      "loss": 0.2468,
      "step": 5357
    },
    {
      "epoch": 0.38,
      "grad_norm": 14.764368057250977,
      "learning_rate": 7.05544412978025e-06,
      "loss": 1.1478,
      "step": 5358
    },
    {
      "epoch": 0.38,
      "grad_norm": 21.965625762939453,
      "learning_rate": 7.054385423943747e-06,
      "loss": 1.1578,
      "step": 5359
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4669270515441895,
      "learning_rate": 7.053326607280924e-06,
      "loss": 0.1087,
      "step": 5360
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.974698066711426,
      "learning_rate": 7.052267679848903e-06,
      "loss": 0.1501,
      "step": 5361
    },
    {
      "epoch": 0.38,
      "grad_norm": 15.666451454162598,
      "learning_rate": 7.05120864170481e-06,
      "loss": 0.9226,
      "step": 5362
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.250504016876221,
      "learning_rate": 7.050149492905775e-06,
      "loss": 0.9035,
      "step": 5363
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.260448932647705,
      "learning_rate": 7.049090233508934e-06,
      "loss": 0.4736,
      "step": 5364
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.690932273864746,
      "learning_rate": 7.048030863571432e-06,
      "loss": 0.1519,
      "step": 5365
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.0404748916625977,
      "learning_rate": 7.0469713831504175e-06,
      "loss": 0.2165,
      "step": 5366
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.073938369750977,
      "learning_rate": 7.045911792303044e-06,
      "loss": 1.1757,
      "step": 5367
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.626930236816406,
      "learning_rate": 7.044852091086474e-06,
      "loss": 0.859,
      "step": 5368
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.595343589782715,
      "learning_rate": 7.043792279557873e-06,
      "loss": 0.1953,
      "step": 5369
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.210716962814331,
      "learning_rate": 7.042732357774416e-06,
      "loss": 0.1123,
      "step": 5370
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.915306091308594,
      "learning_rate": 7.04167232579328e-06,
      "loss": 0.8486,
      "step": 5371
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6535770893096924,
      "learning_rate": 7.04061218367165e-06,
      "loss": 0.1307,
      "step": 5372
    },
    {
      "epoch": 0.39,
      "grad_norm": 29.04788589477539,
      "learning_rate": 7.039551931466717e-06,
      "loss": 0.6607,
      "step": 5373
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.472811222076416,
      "learning_rate": 7.038491569235677e-06,
      "loss": 0.0949,
      "step": 5374
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3216090202331543,
      "learning_rate": 7.037431097035733e-06,
      "loss": 0.1582,
      "step": 5375
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.06667423248291,
      "learning_rate": 7.036370514924095e-06,
      "loss": 0.159,
      "step": 5376
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.793184995651245,
      "learning_rate": 7.035309822957974e-06,
      "loss": 0.2331,
      "step": 5377
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.7592597007751465,
      "learning_rate": 7.0342490211945925e-06,
      "loss": 0.7719,
      "step": 5378
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.507868766784668,
      "learning_rate": 7.033188109691177e-06,
      "loss": 0.1051,
      "step": 5379
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.048153877258301,
      "learning_rate": 7.032127088504959e-06,
      "loss": 0.9012,
      "step": 5380
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.629340410232544,
      "learning_rate": 7.031065957693176e-06,
      "loss": 0.3192,
      "step": 5381
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.272808074951172,
      "learning_rate": 7.030004717313072e-06,
      "loss": 0.7944,
      "step": 5382
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.490428447723389,
      "learning_rate": 7.028943367421899e-06,
      "loss": 1.0911,
      "step": 5383
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.792206287384033,
      "learning_rate": 7.02788190807691e-06,
      "loss": 0.1252,
      "step": 5384
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0869557857513428,
      "learning_rate": 7.026820339335369e-06,
      "loss": 0.0874,
      "step": 5385
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2307114601135254,
      "learning_rate": 7.0257586612545414e-06,
      "loss": 0.0767,
      "step": 5386
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.056832313537598,
      "learning_rate": 7.024696873891704e-06,
      "loss": 0.81,
      "step": 5387
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.920626401901245,
      "learning_rate": 7.0236349773041315e-06,
      "loss": 0.216,
      "step": 5388
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6639654636383057,
      "learning_rate": 7.0225729715491135e-06,
      "loss": 0.1301,
      "step": 5389
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.887688398361206,
      "learning_rate": 7.021510856683939e-06,
      "loss": 0.1689,
      "step": 5390
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.566728591918945,
      "learning_rate": 7.020448632765906e-06,
      "loss": 0.6254,
      "step": 5391
    },
    {
      "epoch": 0.39,
      "grad_norm": 14.98349380493164,
      "learning_rate": 7.019386299852316e-06,
      "loss": 2.3547,
      "step": 5392
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.131125450134277,
      "learning_rate": 7.018323858000481e-06,
      "loss": 0.9596,
      "step": 5393
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7119956016540527,
      "learning_rate": 7.0172613072677125e-06,
      "loss": 0.206,
      "step": 5394
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.038280487060547,
      "learning_rate": 7.016198647711332e-06,
      "loss": 0.1396,
      "step": 5395
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.580249309539795,
      "learning_rate": 7.015135879388667e-06,
      "loss": 0.8693,
      "step": 5396
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.8837850093841553,
      "learning_rate": 7.014073002357047e-06,
      "loss": 0.2129,
      "step": 5397
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.337266445159912,
      "learning_rate": 7.013010016673813e-06,
      "loss": 0.1151,
      "step": 5398
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.669329643249512,
      "learning_rate": 7.01194692239631e-06,
      "loss": 0.7805,
      "step": 5399
    },
    {
      "epoch": 0.39,
      "grad_norm": 16.443531036376953,
      "learning_rate": 7.0108837195818845e-06,
      "loss": 1.2519,
      "step": 5400
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.41287899017334,
      "learning_rate": 7.009820408287895e-06,
      "loss": 0.9217,
      "step": 5401
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.728351593017578,
      "learning_rate": 7.0087569885717015e-06,
      "loss": 0.118,
      "step": 5402
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.734492778778076,
      "learning_rate": 7.007693460490672e-06,
      "loss": 0.2157,
      "step": 5403
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.600372791290283,
      "learning_rate": 7.00662982410218e-06,
      "loss": 0.1856,
      "step": 5404
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.559983968734741,
      "learning_rate": 7.005566079463604e-06,
      "loss": 0.1476,
      "step": 5405
    },
    {
      "epoch": 0.39,
      "grad_norm": 19.755691528320312,
      "learning_rate": 7.004502226632329e-06,
      "loss": 1.0556,
      "step": 5406
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.4858479499816895,
      "learning_rate": 7.003438265665748e-06,
      "loss": 0.1687,
      "step": 5407
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.98272180557251,
      "learning_rate": 7.0023741966212555e-06,
      "loss": 0.8231,
      "step": 5408
    },
    {
      "epoch": 0.39,
      "grad_norm": 15.854585647583008,
      "learning_rate": 7.0013100195562535e-06,
      "loss": 1.1672,
      "step": 5409
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.2490220069885254,
      "learning_rate": 7.0002457345281526e-06,
      "loss": 0.1155,
      "step": 5410
    },
    {
      "epoch": 0.39,
      "grad_norm": 12.551800727844238,
      "learning_rate": 6.9991813415943645e-06,
      "loss": 0.6431,
      "step": 5411
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.7663166522979736,
      "learning_rate": 6.99811684081231e-06,
      "loss": 0.186,
      "step": 5412
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.580138206481934,
      "learning_rate": 6.997052232239419e-06,
      "loss": 0.923,
      "step": 5413
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.424075126647949,
      "learning_rate": 6.9959875159331145e-06,
      "loss": 0.7429,
      "step": 5414
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.144588470458984,
      "learning_rate": 6.99492269195084e-06,
      "loss": 0.8658,
      "step": 5415
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.571002960205078,
      "learning_rate": 6.993857760350039e-06,
      "loss": 0.8143,
      "step": 5416
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.857241630554199,
      "learning_rate": 6.992792721188158e-06,
      "loss": 0.2414,
      "step": 5417
    },
    {
      "epoch": 0.39,
      "grad_norm": 14.71638011932373,
      "learning_rate": 6.991727574522653e-06,
      "loss": 0.9634,
      "step": 5418
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.9907331466674805,
      "learning_rate": 6.990662320410985e-06,
      "loss": 0.2085,
      "step": 5419
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.042077541351318,
      "learning_rate": 6.989596958910618e-06,
      "loss": 0.6349,
      "step": 5420
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.432367324829102,
      "learning_rate": 6.988531490079028e-06,
      "loss": 0.562,
      "step": 5421
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.987910270690918,
      "learning_rate": 6.98746591397369e-06,
      "loss": 0.9032,
      "step": 5422
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.115936040878296,
      "learning_rate": 6.986400230652089e-06,
      "loss": 0.1567,
      "step": 5423
    },
    {
      "epoch": 0.39,
      "grad_norm": 13.424097061157227,
      "learning_rate": 6.985334440171716e-06,
      "loss": 1.0768,
      "step": 5424
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.049901247024536,
      "learning_rate": 6.984268542590063e-06,
      "loss": 0.1591,
      "step": 5425
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.268720865249634,
      "learning_rate": 6.983202537964634e-06,
      "loss": 0.2452,
      "step": 5426
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.558272361755371,
      "learning_rate": 6.982136426352934e-06,
      "loss": 0.7829,
      "step": 5427
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2703585624694824,
      "learning_rate": 6.981070207812477e-06,
      "loss": 0.067,
      "step": 5428
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.9522340297698975,
      "learning_rate": 6.980003882400781e-06,
      "loss": 0.1332,
      "step": 5429
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.240145683288574,
      "learning_rate": 6.978937450175372e-06,
      "loss": 0.8648,
      "step": 5430
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.53151535987854,
      "learning_rate": 6.977870911193776e-06,
      "loss": 0.1977,
      "step": 5431
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3137731552124023,
      "learning_rate": 6.976804265513532e-06,
      "loss": 0.1319,
      "step": 5432
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.8123788833618164,
      "learning_rate": 6.97573751319218e-06,
      "loss": 0.1833,
      "step": 5433
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3178751468658447,
      "learning_rate": 6.974670654287268e-06,
      "loss": 0.1348,
      "step": 5434
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.334117412567139,
      "learning_rate": 6.973603688856349e-06,
      "loss": 0.6211,
      "step": 5435
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7806451320648193,
      "learning_rate": 6.972536616956981e-06,
      "loss": 0.1126,
      "step": 5436
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.845785140991211,
      "learning_rate": 6.971469438646729e-06,
      "loss": 0.0643,
      "step": 5437
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.481468677520752,
      "learning_rate": 6.970402153983164e-06,
      "loss": 0.8117,
      "step": 5438
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.3676581382751465,
      "learning_rate": 6.96933476302386e-06,
      "loss": 0.287,
      "step": 5439
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.914379119873047,
      "learning_rate": 6.968267265826401e-06,
      "loss": 0.9451,
      "step": 5440
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.138936996459961,
      "learning_rate": 6.967199662448372e-06,
      "loss": 1.1758,
      "step": 5441
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.31937313079834,
      "learning_rate": 6.966131952947368e-06,
      "loss": 0.5361,
      "step": 5442
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.1377553939819336,
      "learning_rate": 6.965064137380987e-06,
      "loss": 0.1628,
      "step": 5443
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.14564037322998,
      "learning_rate": 6.963996215806836e-06,
      "loss": 0.8094,
      "step": 5444
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.225896835327148,
      "learning_rate": 6.962928188282522e-06,
      "loss": 0.2906,
      "step": 5445
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.3853025436401367,
      "learning_rate": 6.961860054865661e-06,
      "loss": 0.2152,
      "step": 5446
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.890792846679688,
      "learning_rate": 6.960791815613878e-06,
      "loss": 1.0113,
      "step": 5447
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6908884048461914,
      "learning_rate": 6.959723470584797e-06,
      "loss": 0.0957,
      "step": 5448
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.20700740814209,
      "learning_rate": 6.958655019836052e-06,
      "loss": 0.2124,
      "step": 5449
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.846900224685669,
      "learning_rate": 6.9575864634252866e-06,
      "loss": 0.1087,
      "step": 5450
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6427087783813477,
      "learning_rate": 6.9565178014101375e-06,
      "loss": 0.1545,
      "step": 5451
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.134119987487793,
      "learning_rate": 6.95544903384826e-06,
      "loss": 0.3128,
      "step": 5452
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6461946964263916,
      "learning_rate": 6.9543801607973095e-06,
      "loss": 0.1472,
      "step": 5453
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.100834846496582,
      "learning_rate": 6.953311182314946e-06,
      "loss": 0.4514,
      "step": 5454
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3199925422668457,
      "learning_rate": 6.952242098458839e-06,
      "loss": 0.0973,
      "step": 5455
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.207007646560669,
      "learning_rate": 6.951172909286661e-06,
      "loss": 0.2146,
      "step": 5456
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.844284772872925,
      "learning_rate": 6.950103614856087e-06,
      "loss": 0.1868,
      "step": 5457
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.142391681671143,
      "learning_rate": 6.949034215224808e-06,
      "loss": 0.8374,
      "step": 5458
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.154209852218628,
      "learning_rate": 6.947964710450509e-06,
      "loss": 0.1066,
      "step": 5459
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.19703483581543,
      "learning_rate": 6.946895100590887e-06,
      "loss": 0.9256,
      "step": 5460
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.79628324508667,
      "learning_rate": 6.9458253857036435e-06,
      "loss": 0.086,
      "step": 5461
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.550283908843994,
      "learning_rate": 6.944755565846488e-06,
      "loss": 0.8006,
      "step": 5462
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7630841732025146,
      "learning_rate": 6.943685641077129e-06,
      "loss": 0.1874,
      "step": 5463
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.3186192512512207,
      "learning_rate": 6.942615611453287e-06,
      "loss": 0.1463,
      "step": 5464
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.645384311676025,
      "learning_rate": 6.941545477032687e-06,
      "loss": 0.706,
      "step": 5465
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.8518779277801514,
      "learning_rate": 6.940475237873057e-06,
      "loss": 0.3516,
      "step": 5466
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.8367490768432617,
      "learning_rate": 6.939404894032132e-06,
      "loss": 0.2215,
      "step": 5467
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.188509941101074,
      "learning_rate": 6.938334445567656e-06,
      "loss": 0.5679,
      "step": 5468
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.421807289123535,
      "learning_rate": 6.9372638925373725e-06,
      "loss": 0.8167,
      "step": 5469
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.0412750244140625,
      "learning_rate": 6.936193234999036e-06,
      "loss": 0.121,
      "step": 5470
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.609835386276245,
      "learning_rate": 6.935122473010402e-06,
      "loss": 0.145,
      "step": 5471
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.59435749053955,
      "learning_rate": 6.9340516066292364e-06,
      "loss": 0.8474,
      "step": 5472
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4690768718719482,
      "learning_rate": 6.932980635913307e-06,
      "loss": 0.1754,
      "step": 5473
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.5724892616271973,
      "learning_rate": 6.931909560920389e-06,
      "loss": 0.1514,
      "step": 5474
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.723560333251953,
      "learning_rate": 6.930838381708263e-06,
      "loss": 1.0546,
      "step": 5475
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.617505073547363,
      "learning_rate": 6.929767098334714e-06,
      "loss": 0.8653,
      "step": 5476
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7979371547698975,
      "learning_rate": 6.928695710857535e-06,
      "loss": 0.1034,
      "step": 5477
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.5833823680877686,
      "learning_rate": 6.927624219334523e-06,
      "loss": 0.9039,
      "step": 5478
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.8620357513427734,
      "learning_rate": 6.926552623823482e-06,
      "loss": 0.2238,
      "step": 5479
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.822953224182129,
      "learning_rate": 6.925480924382218e-06,
      "loss": 0.1203,
      "step": 5480
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.966365337371826,
      "learning_rate": 6.9244091210685475e-06,
      "loss": 1.0112,
      "step": 5481
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.239340782165527,
      "learning_rate": 6.923337213940289e-06,
      "loss": 0.5057,
      "step": 5482
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6271235942840576,
      "learning_rate": 6.922265203055266e-06,
      "loss": 0.0886,
      "step": 5483
    },
    {
      "epoch": 0.39,
      "grad_norm": 18.494747161865234,
      "learning_rate": 6.921193088471314e-06,
      "loss": 1.2266,
      "step": 5484
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.85886287689209,
      "learning_rate": 6.920120870246265e-06,
      "loss": 0.8548,
      "step": 5485
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.971796035766602,
      "learning_rate": 6.919048548437966e-06,
      "loss": 0.8527,
      "step": 5486
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.305066108703613,
      "learning_rate": 6.91797612310426e-06,
      "loss": 0.8397,
      "step": 5487
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.190934181213379,
      "learning_rate": 6.916903594303001e-06,
      "loss": 0.2472,
      "step": 5488
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.391557216644287,
      "learning_rate": 6.91583096209205e-06,
      "loss": 0.7292,
      "step": 5489
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.9055533409118652,
      "learning_rate": 6.914758226529271e-06,
      "loss": 0.1802,
      "step": 5490
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.875375747680664,
      "learning_rate": 6.913685387672533e-06,
      "loss": 0.848,
      "step": 5491
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.431441783905029,
      "learning_rate": 6.912612445579712e-06,
      "loss": 0.8899,
      "step": 5492
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.871878147125244,
      "learning_rate": 6.911539400308689e-06,
      "loss": 0.3802,
      "step": 5493
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.8968346118927,
      "learning_rate": 6.910466251917352e-06,
      "loss": 0.8456,
      "step": 5494
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.7851433753967285,
      "learning_rate": 6.909393000463592e-06,
      "loss": 0.7689,
      "step": 5495
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.918240547180176,
      "learning_rate": 6.908319646005305e-06,
      "loss": 0.6116,
      "step": 5496
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.669656753540039,
      "learning_rate": 6.907246188600397e-06,
      "loss": 1.1633,
      "step": 5497
    },
    {
      "epoch": 0.39,
      "grad_norm": 14.252010345458984,
      "learning_rate": 6.906172628306779e-06,
      "loss": 0.4104,
      "step": 5498
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.096311092376709,
      "learning_rate": 6.90509896518236e-06,
      "loss": 0.7708,
      "step": 5499
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.0676357746124268,
      "learning_rate": 6.904025199285064e-06,
      "loss": 0.2013,
      "step": 5500
    },
    {
      "epoch": 0.39,
      "grad_norm": 32.60609436035156,
      "learning_rate": 6.902951330672816e-06,
      "loss": 0.9287,
      "step": 5501
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7919373512268066,
      "learning_rate": 6.901877359403546e-06,
      "loss": 0.1955,
      "step": 5502
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.101912498474121,
      "learning_rate": 6.900803285535191e-06,
      "loss": 0.8194,
      "step": 5503
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.855050086975098,
      "learning_rate": 6.899729109125697e-06,
      "loss": 0.8623,
      "step": 5504
    },
    {
      "epoch": 0.39,
      "grad_norm": 13.77261734008789,
      "learning_rate": 6.898654830233006e-06,
      "loss": 0.6902,
      "step": 5505
    },
    {
      "epoch": 0.39,
      "grad_norm": 14.555625915527344,
      "learning_rate": 6.8975804489150735e-06,
      "loss": 0.9158,
      "step": 5506
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4314489364624023,
      "learning_rate": 6.89650596522986e-06,
      "loss": 0.1002,
      "step": 5507
    },
    {
      "epoch": 0.39,
      "grad_norm": 18.731006622314453,
      "learning_rate": 6.895431379235327e-06,
      "loss": 0.8531,
      "step": 5508
    },
    {
      "epoch": 0.4,
      "grad_norm": 14.339290618896484,
      "learning_rate": 6.894356690989445e-06,
      "loss": 1.4079,
      "step": 5509
    },
    {
      "epoch": 0.4,
      "grad_norm": 13.954771041870117,
      "learning_rate": 6.893281900550192e-06,
      "loss": 0.7371,
      "step": 5510
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8031539916992188,
      "learning_rate": 6.892207007975545e-06,
      "loss": 0.1876,
      "step": 5511
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7659687995910645,
      "learning_rate": 6.891132013323494e-06,
      "loss": 0.1511,
      "step": 5512
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.723506450653076,
      "learning_rate": 6.8900569166520285e-06,
      "loss": 0.8086,
      "step": 5513
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.9778478145599365,
      "learning_rate": 6.888981718019145e-06,
      "loss": 0.212,
      "step": 5514
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.53315806388855,
      "learning_rate": 6.88790641748285e-06,
      "loss": 0.1507,
      "step": 5515
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.369292259216309,
      "learning_rate": 6.886831015101151e-06,
      "loss": 1.1366,
      "step": 5516
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.818778991699219,
      "learning_rate": 6.885755510932058e-06,
      "loss": 1.2266,
      "step": 5517
    },
    {
      "epoch": 0.4,
      "grad_norm": 14.464984893798828,
      "learning_rate": 6.884679905033596e-06,
      "loss": 1.2339,
      "step": 5518
    },
    {
      "epoch": 0.4,
      "grad_norm": 19.21283721923828,
      "learning_rate": 6.883604197463784e-06,
      "loss": 1.0616,
      "step": 5519
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.4857873916625977,
      "learning_rate": 6.882528388280656e-06,
      "loss": 0.2723,
      "step": 5520
    },
    {
      "epoch": 0.4,
      "grad_norm": 14.310903549194336,
      "learning_rate": 6.881452477542249e-06,
      "loss": 1.0361,
      "step": 5521
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2440059185028076,
      "learning_rate": 6.880376465306602e-06,
      "loss": 0.1268,
      "step": 5522
    },
    {
      "epoch": 0.4,
      "grad_norm": 17.025453567504883,
      "learning_rate": 6.879300351631762e-06,
      "loss": 1.1159,
      "step": 5523
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.07649040222168,
      "learning_rate": 6.878224136575781e-06,
      "loss": 0.7686,
      "step": 5524
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.981122016906738,
      "learning_rate": 6.877147820196717e-06,
      "loss": 1.5925,
      "step": 5525
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.410175323486328,
      "learning_rate": 6.876071402552634e-06,
      "loss": 0.2911,
      "step": 5526
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.649217367172241,
      "learning_rate": 6.8749948837016e-06,
      "loss": 0.1304,
      "step": 5527
    },
    {
      "epoch": 0.4,
      "grad_norm": 16.858001708984375,
      "learning_rate": 6.873918263701691e-06,
      "loss": 0.744,
      "step": 5528
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2856030464172363,
      "learning_rate": 6.872841542610982e-06,
      "loss": 0.0703,
      "step": 5529
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.747156143188477,
      "learning_rate": 6.871764720487564e-06,
      "loss": 0.6707,
      "step": 5530
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.497338056564331,
      "learning_rate": 6.8706877973895215e-06,
      "loss": 0.1205,
      "step": 5531
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.090934753417969,
      "learning_rate": 6.8696107733749555e-06,
      "loss": 0.9164,
      "step": 5532
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.371113300323486,
      "learning_rate": 6.868533648501962e-06,
      "loss": 1.1856,
      "step": 5533
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5088791847229004,
      "learning_rate": 6.867456422828654e-06,
      "loss": 0.2028,
      "step": 5534
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.3121659755706787,
      "learning_rate": 6.866379096413141e-06,
      "loss": 0.1424,
      "step": 5535
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.451214551925659,
      "learning_rate": 6.865301669313538e-06,
      "loss": 0.1336,
      "step": 5536
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.3520448207855225,
      "learning_rate": 6.864224141587971e-06,
      "loss": 0.6198,
      "step": 5537
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.974488258361816,
      "learning_rate": 6.8631465132945685e-06,
      "loss": 0.1832,
      "step": 5538
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.0935187339782715,
      "learning_rate": 6.862068784491464e-06,
      "loss": 0.1535,
      "step": 5539
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.786938667297363,
      "learning_rate": 6.860990955236797e-06,
      "loss": 1.3354,
      "step": 5540
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7047953605651855,
      "learning_rate": 6.859913025588713e-06,
      "loss": 0.0791,
      "step": 5541
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.304962158203125,
      "learning_rate": 6.85883499560536e-06,
      "loss": 1.0117,
      "step": 5542
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7012007236480713,
      "learning_rate": 6.857756865344895e-06,
      "loss": 0.1123,
      "step": 5543
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.568686485290527,
      "learning_rate": 6.85667863486548e-06,
      "loss": 1.1739,
      "step": 5544
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.154447555541992,
      "learning_rate": 6.855600304225281e-06,
      "loss": 0.8495,
      "step": 5545
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.477684020996094,
      "learning_rate": 6.8545218734824685e-06,
      "loss": 0.8615,
      "step": 5546
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.896976470947266,
      "learning_rate": 6.853443342695221e-06,
      "loss": 0.8045,
      "step": 5547
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8306257724761963,
      "learning_rate": 6.85236471192172e-06,
      "loss": 0.1579,
      "step": 5548
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.4681665897369385,
      "learning_rate": 6.851285981220156e-06,
      "loss": 0.1302,
      "step": 5549
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.310873985290527,
      "learning_rate": 6.85020715064872e-06,
      "loss": 1.4201,
      "step": 5550
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.956937789916992,
      "learning_rate": 6.8491282202656106e-06,
      "loss": 0.7343,
      "step": 5551
    },
    {
      "epoch": 0.4,
      "grad_norm": 30.336578369140625,
      "learning_rate": 6.8480491901290356e-06,
      "loss": 0.7388,
      "step": 5552
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.033907651901245,
      "learning_rate": 6.8469700602972e-06,
      "loss": 0.1905,
      "step": 5553
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.876399517059326,
      "learning_rate": 6.84589083082832e-06,
      "loss": 0.9449,
      "step": 5554
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.3023059368133545,
      "learning_rate": 6.844811501780619e-06,
      "loss": 0.0805,
      "step": 5555
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.888305425643921,
      "learning_rate": 6.843732073212319e-06,
      "loss": 0.1341,
      "step": 5556
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.355558395385742,
      "learning_rate": 6.842652545181652e-06,
      "loss": 0.244,
      "step": 5557
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2579293251037598,
      "learning_rate": 6.841572917746857e-06,
      "loss": 0.1347,
      "step": 5558
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.9985077381134033,
      "learning_rate": 6.840493190966173e-06,
      "loss": 0.8667,
      "step": 5559
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.824347972869873,
      "learning_rate": 6.839413364897846e-06,
      "loss": 0.1457,
      "step": 5560
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.636181831359863,
      "learning_rate": 6.838333439600135e-06,
      "loss": 0.949,
      "step": 5561
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7640230655670166,
      "learning_rate": 6.83725341513129e-06,
      "loss": 0.1544,
      "step": 5562
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8001480102539062,
      "learning_rate": 6.836173291549578e-06,
      "loss": 0.1941,
      "step": 5563
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8495585918426514,
      "learning_rate": 6.835093068913267e-06,
      "loss": 0.2117,
      "step": 5564
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5288331508636475,
      "learning_rate": 6.834012747280633e-06,
      "loss": 0.1041,
      "step": 5565
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.681227445602417,
      "learning_rate": 6.8329323267099525e-06,
      "loss": 0.2394,
      "step": 5566
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.617342948913574,
      "learning_rate": 6.83185180725951e-06,
      "loss": 1.2808,
      "step": 5567
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.361928939819336,
      "learning_rate": 6.830771188987596e-06,
      "loss": 0.604,
      "step": 5568
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.039451599121094,
      "learning_rate": 6.829690471952508e-06,
      "loss": 0.4144,
      "step": 5569
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.562185287475586,
      "learning_rate": 6.828609656212545e-06,
      "loss": 0.1326,
      "step": 5570
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.886679172515869,
      "learning_rate": 6.827528741826011e-06,
      "loss": 0.1864,
      "step": 5571
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.588736534118652,
      "learning_rate": 6.82644772885122e-06,
      "loss": 0.9927,
      "step": 5572
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.111225605010986,
      "learning_rate": 6.8253666173464876e-06,
      "loss": 0.953,
      "step": 5573
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.835580348968506,
      "learning_rate": 6.824285407370134e-06,
      "loss": 0.117,
      "step": 5574
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.578390121459961,
      "learning_rate": 6.823204098980489e-06,
      "loss": 0.8052,
      "step": 5575
    },
    {
      "epoch": 0.4,
      "grad_norm": 25.140840530395508,
      "learning_rate": 6.822122692235885e-06,
      "loss": 0.7874,
      "step": 5576
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.080495834350586,
      "learning_rate": 6.821041187194658e-06,
      "loss": 0.1926,
      "step": 5577
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.169532060623169,
      "learning_rate": 6.8199595839151526e-06,
      "loss": 0.1742,
      "step": 5578
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.082882881164551,
      "learning_rate": 6.818877882455716e-06,
      "loss": 0.9065,
      "step": 5579
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.520149230957031,
      "learning_rate": 6.817796082874702e-06,
      "loss": 1.588,
      "step": 5580
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.188344955444336,
      "learning_rate": 6.8167141852304705e-06,
      "loss": 0.9735,
      "step": 5581
    },
    {
      "epoch": 0.4,
      "grad_norm": 13.515519142150879,
      "learning_rate": 6.815632189581386e-06,
      "loss": 1.7396,
      "step": 5582
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.666753768920898,
      "learning_rate": 6.814550095985817e-06,
      "loss": 1.3142,
      "step": 5583
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.005579948425293,
      "learning_rate": 6.813467904502138e-06,
      "loss": 0.7438,
      "step": 5584
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7345714569091797,
      "learning_rate": 6.81238561518873e-06,
      "loss": 0.1529,
      "step": 5585
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.515465021133423,
      "learning_rate": 6.811303228103979e-06,
      "loss": 0.1608,
      "step": 5586
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.9596376419067383,
      "learning_rate": 6.810220743306275e-06,
      "loss": 0.1453,
      "step": 5587
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.017350673675537,
      "learning_rate": 6.809138160854014e-06,
      "loss": 0.1676,
      "step": 5588
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.1740012168884277,
      "learning_rate": 6.808055480805597e-06,
      "loss": 0.2183,
      "step": 5589
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.415849685668945,
      "learning_rate": 6.80697270321943e-06,
      "loss": 0.8998,
      "step": 5590
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.64565372467041,
      "learning_rate": 6.8058898281539265e-06,
      "loss": 0.1483,
      "step": 5591
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.649191856384277,
      "learning_rate": 6.804806855667501e-06,
      "loss": 1.2057,
      "step": 5592
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.774937868118286,
      "learning_rate": 6.803723785818579e-06,
      "loss": 0.1486,
      "step": 5593
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6313161849975586,
      "learning_rate": 6.802640618665585e-06,
      "loss": 0.191,
      "step": 5594
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.2466580867767334,
      "learning_rate": 6.801557354266955e-06,
      "loss": 0.1355,
      "step": 5595
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.704415798187256,
      "learning_rate": 6.800473992681122e-06,
      "loss": 0.1382,
      "step": 5596
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.159027099609375,
      "learning_rate": 6.799390533966533e-06,
      "loss": 0.8104,
      "step": 5597
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.755506753921509,
      "learning_rate": 6.798306978181637e-06,
      "loss": 0.1306,
      "step": 5598
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.5717949867248535,
      "learning_rate": 6.7972233253848855e-06,
      "loss": 0.7238,
      "step": 5599
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.011937618255615,
      "learning_rate": 6.7961395756347395e-06,
      "loss": 0.9305,
      "step": 5600
    },
    {
      "epoch": 0.4,
      "grad_norm": 14.121397018432617,
      "learning_rate": 6.795055728989662e-06,
      "loss": 0.627,
      "step": 5601
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.3688080310821533,
      "learning_rate": 6.7939717855081225e-06,
      "loss": 0.1009,
      "step": 5602
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.431504249572754,
      "learning_rate": 6.792887745248595e-06,
      "loss": 0.8338,
      "step": 5603
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.757314682006836,
      "learning_rate": 6.79180360826956e-06,
      "loss": 0.7387,
      "step": 5604
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.185975074768066,
      "learning_rate": 6.7907193746295016e-06,
      "loss": 0.9653,
      "step": 5605
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.166177749633789,
      "learning_rate": 6.789635044386915e-06,
      "loss": 0.9251,
      "step": 5606
    },
    {
      "epoch": 0.4,
      "grad_norm": 19.04930305480957,
      "learning_rate": 6.788550617600289e-06,
      "loss": 1.1661,
      "step": 5607
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.422466278076172,
      "learning_rate": 6.787466094328128e-06,
      "loss": 1.084,
      "step": 5608
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.741326332092285,
      "learning_rate": 6.786381474628937e-06,
      "loss": 1.0555,
      "step": 5609
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.00621223449707,
      "learning_rate": 6.785296758561226e-06,
      "loss": 0.8871,
      "step": 5610
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.754476547241211,
      "learning_rate": 6.784211946183513e-06,
      "loss": 1.1606,
      "step": 5611
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.988335132598877,
      "learning_rate": 6.783127037554322e-06,
      "loss": 0.7454,
      "step": 5612
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.537798881530762,
      "learning_rate": 6.782042032732175e-06,
      "loss": 0.8567,
      "step": 5613
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6975255012512207,
      "learning_rate": 6.780956931775604e-06,
      "loss": 0.1301,
      "step": 5614
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.361785650253296,
      "learning_rate": 6.77987173474315e-06,
      "loss": 0.1697,
      "step": 5615
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.4770772457122803,
      "learning_rate": 6.778786441693351e-06,
      "loss": 0.0947,
      "step": 5616
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.364981174468994,
      "learning_rate": 6.777701052684759e-06,
      "loss": 1.0302,
      "step": 5617
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.80391263961792,
      "learning_rate": 6.7766155677759236e-06,
      "loss": 0.2275,
      "step": 5618
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.127688884735107,
      "learning_rate": 6.775529987025402e-06,
      "loss": 0.6375,
      "step": 5619
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.2577543258667,
      "learning_rate": 6.774444310491759e-06,
      "loss": 0.6915,
      "step": 5620
    },
    {
      "epoch": 0.4,
      "grad_norm": 17.581037521362305,
      "learning_rate": 6.773358538233562e-06,
      "loss": 1.0428,
      "step": 5621
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.2466509342193604,
      "learning_rate": 6.7722726703093836e-06,
      "loss": 0.1511,
      "step": 5622
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.020066261291504,
      "learning_rate": 6.7711867067778044e-06,
      "loss": 0.5507,
      "step": 5623
    },
    {
      "epoch": 0.4,
      "grad_norm": 11.647672653198242,
      "learning_rate": 6.770100647697406e-06,
      "loss": 0.9781,
      "step": 5624
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.249826669692993,
      "learning_rate": 6.769014493126778e-06,
      "loss": 0.1305,
      "step": 5625
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.937422752380371,
      "learning_rate": 6.767928243124516e-06,
      "loss": 0.8476,
      "step": 5626
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.309476137161255,
      "learning_rate": 6.766841897749213e-06,
      "loss": 0.1176,
      "step": 5627
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.245361328125,
      "learning_rate": 6.76575545705948e-06,
      "loss": 0.6312,
      "step": 5628
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2414751052856445,
      "learning_rate": 6.764668921113923e-06,
      "loss": 0.1683,
      "step": 5629
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.1721460819244385,
      "learning_rate": 6.76358228997116e-06,
      "loss": 0.1446,
      "step": 5630
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.0970516204833984,
      "learning_rate": 6.762495563689806e-06,
      "loss": 0.215,
      "step": 5631
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.5042366981506348,
      "learning_rate": 6.761408742328488e-06,
      "loss": 0.7959,
      "step": 5632
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.572076797485352,
      "learning_rate": 6.7603218259458344e-06,
      "loss": 0.5979,
      "step": 5633
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.986553192138672,
      "learning_rate": 6.7592348146004836e-06,
      "loss": 0.7129,
      "step": 5634
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.503801107406616,
      "learning_rate": 6.758147708351072e-06,
      "loss": 0.3241,
      "step": 5635
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.993279933929443,
      "learning_rate": 6.757060507256248e-06,
      "loss": 0.7913,
      "step": 5636
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.749504566192627,
      "learning_rate": 6.755973211374658e-06,
      "loss": 0.1406,
      "step": 5637
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.715325355529785,
      "learning_rate": 6.754885820764962e-06,
      "loss": 0.3425,
      "step": 5638
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.3427302837371826,
      "learning_rate": 6.753798335485818e-06,
      "loss": 0.127,
      "step": 5639
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.123676300048828,
      "learning_rate": 6.752710755595893e-06,
      "loss": 0.878,
      "step": 5640
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.293579578399658,
      "learning_rate": 6.751623081153855e-06,
      "loss": 0.8356,
      "step": 5641
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9738280773162842,
      "learning_rate": 6.750535312218385e-06,
      "loss": 0.1157,
      "step": 5642
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.250044107437134,
      "learning_rate": 6.74944744884816e-06,
      "loss": 0.1961,
      "step": 5643
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7878124713897705,
      "learning_rate": 6.748359491101866e-06,
      "loss": 0.1443,
      "step": 5644
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.608217239379883,
      "learning_rate": 6.747271439038195e-06,
      "loss": 0.5109,
      "step": 5645
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.293252468109131,
      "learning_rate": 6.746183292715845e-06,
      "loss": 0.1118,
      "step": 5646
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.545401096343994,
      "learning_rate": 6.745095052193515e-06,
      "loss": 0.7532,
      "step": 5647
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9844584465026855,
      "learning_rate": 6.744006717529914e-06,
      "loss": 0.0811,
      "step": 5648
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.144975185394287,
      "learning_rate": 6.742918288783752e-06,
      "loss": 0.743,
      "step": 5649
    },
    {
      "epoch": 0.41,
      "grad_norm": 16.339279174804688,
      "learning_rate": 6.741829766013744e-06,
      "loss": 0.9501,
      "step": 5650
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8521029949188232,
      "learning_rate": 6.7407411492786135e-06,
      "loss": 0.122,
      "step": 5651
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.7307400703430176,
      "learning_rate": 6.739652438637089e-06,
      "loss": 0.1308,
      "step": 5652
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5999724864959717,
      "learning_rate": 6.738563634147898e-06,
      "loss": 0.1608,
      "step": 5653
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.548293590545654,
      "learning_rate": 6.7374747358697815e-06,
      "loss": 0.1382,
      "step": 5654
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.534099578857422,
      "learning_rate": 6.73638574386148e-06,
      "loss": 0.2068,
      "step": 5655
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.7675318717956543,
      "learning_rate": 6.735296658181738e-06,
      "loss": 0.1528,
      "step": 5656
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.652014970779419,
      "learning_rate": 6.734207478889312e-06,
      "loss": 0.1218,
      "step": 5657
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.925245761871338,
      "learning_rate": 6.733118206042955e-06,
      "loss": 0.2433,
      "step": 5658
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.947298526763916,
      "learning_rate": 6.732028839701431e-06,
      "loss": 0.1774,
      "step": 5659
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.29573392868042,
      "learning_rate": 6.730939379923509e-06,
      "loss": 0.9324,
      "step": 5660
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.26467227935791,
      "learning_rate": 6.729849826767958e-06,
      "loss": 0.2139,
      "step": 5661
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.7988719940185547,
      "learning_rate": 6.728760180293557e-06,
      "loss": 0.348,
      "step": 5662
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.894384860992432,
      "learning_rate": 6.727670440559089e-06,
      "loss": 0.6195,
      "step": 5663
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.4033455848693848,
      "learning_rate": 6.726580607623339e-06,
      "loss": 0.1494,
      "step": 5664
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.142617702484131,
      "learning_rate": 6.725490681545099e-06,
      "loss": 1.0217,
      "step": 5665
    },
    {
      "epoch": 0.41,
      "grad_norm": 18.985570907592773,
      "learning_rate": 6.724400662383172e-06,
      "loss": 1.2383,
      "step": 5666
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.960386276245117,
      "learning_rate": 6.723310550196355e-06,
      "loss": 0.1974,
      "step": 5667
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.955226898193359,
      "learning_rate": 6.7222203450434555e-06,
      "loss": 0.7886,
      "step": 5668
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.8865764141082764,
      "learning_rate": 6.721130046983288e-06,
      "loss": 0.84,
      "step": 5669
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.00065279006958,
      "learning_rate": 6.720039656074669e-06,
      "loss": 0.7017,
      "step": 5670
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.535150527954102,
      "learning_rate": 6.718949172376423e-06,
      "loss": 0.1879,
      "step": 5671
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5840439796447754,
      "learning_rate": 6.7178585959473745e-06,
      "loss": 0.1347,
      "step": 5672
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9010710716247559,
      "learning_rate": 6.716767926846357e-06,
      "loss": 0.064,
      "step": 5673
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.7410504817962646,
      "learning_rate": 6.715677165132208e-06,
      "loss": 0.1337,
      "step": 5674
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.6112258434295654,
      "learning_rate": 6.714586310863771e-06,
      "loss": 0.186,
      "step": 5675
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5698487758636475,
      "learning_rate": 6.713495364099892e-06,
      "loss": 0.2011,
      "step": 5676
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2322566509246826,
      "learning_rate": 6.712404324899426e-06,
      "loss": 0.0739,
      "step": 5677
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.940441131591797,
      "learning_rate": 6.711313193321225e-06,
      "loss": 0.2173,
      "step": 5678
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3617377281188965,
      "learning_rate": 6.710221969424158e-06,
      "loss": 0.1001,
      "step": 5679
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.0651767253875732,
      "learning_rate": 6.709130653267087e-06,
      "loss": 0.1487,
      "step": 5680
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.33707857131958,
      "learning_rate": 6.708039244908888e-06,
      "loss": 0.4215,
      "step": 5681
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.9301774501800537,
      "learning_rate": 6.706947744408436e-06,
      "loss": 0.1612,
      "step": 5682
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1553962230682373,
      "learning_rate": 6.705856151824615e-06,
      "loss": 0.0858,
      "step": 5683
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.600163459777832,
      "learning_rate": 6.704764467216312e-06,
      "loss": 0.1377,
      "step": 5684
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.777895212173462,
      "learning_rate": 6.7036726906424174e-06,
      "loss": 0.1676,
      "step": 5685
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.224245071411133,
      "learning_rate": 6.7025808221618325e-06,
      "loss": 0.1169,
      "step": 5686
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2004196643829346,
      "learning_rate": 6.701488861833454e-06,
      "loss": 0.0934,
      "step": 5687
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.804073333740234,
      "learning_rate": 6.7003968097161935e-06,
      "loss": 0.7973,
      "step": 5688
    },
    {
      "epoch": 0.41,
      "grad_norm": 12.22189998626709,
      "learning_rate": 6.699304665868961e-06,
      "loss": 0.7569,
      "step": 5689
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4872748851776123,
      "learning_rate": 6.698212430350673e-06,
      "loss": 0.2027,
      "step": 5690
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2705979347229004,
      "learning_rate": 6.697120103220255e-06,
      "loss": 0.1314,
      "step": 5691
    },
    {
      "epoch": 0.41,
      "grad_norm": 24.449068069458008,
      "learning_rate": 6.69602768453663e-06,
      "loss": 1.6606,
      "step": 5692
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.0398571491241455,
      "learning_rate": 6.694935174358731e-06,
      "loss": 0.1877,
      "step": 5693
    },
    {
      "epoch": 0.41,
      "grad_norm": 15.661907196044922,
      "learning_rate": 6.693842572745496e-06,
      "loss": 1.0155,
      "step": 5694
    },
    {
      "epoch": 0.41,
      "grad_norm": 14.141727447509766,
      "learning_rate": 6.692749879755864e-06,
      "loss": 0.9813,
      "step": 5695
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.783662796020508,
      "learning_rate": 6.691657095448784e-06,
      "loss": 0.8931,
      "step": 5696
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.246397018432617,
      "learning_rate": 6.690564219883209e-06,
      "loss": 0.5115,
      "step": 5697
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.9204373359680176,
      "learning_rate": 6.6894712531180915e-06,
      "loss": 0.1311,
      "step": 5698
    },
    {
      "epoch": 0.41,
      "grad_norm": 14.059284210205078,
      "learning_rate": 6.688378195212395e-06,
      "loss": 0.9423,
      "step": 5699
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0981221199035645,
      "learning_rate": 6.687285046225086e-06,
      "loss": 0.1285,
      "step": 5700
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.0013160705566406,
      "learning_rate": 6.686191806215134e-06,
      "loss": 0.1846,
      "step": 5701
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.819549083709717,
      "learning_rate": 6.685098475241518e-06,
      "loss": 0.5522,
      "step": 5702
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.758452892303467,
      "learning_rate": 6.68400505336322e-06,
      "loss": 0.0914,
      "step": 5703
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.255590438842773,
      "learning_rate": 6.682911540639221e-06,
      "loss": 0.8179,
      "step": 5704
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.837156295776367,
      "learning_rate": 6.681817937128514e-06,
      "loss": 0.5714,
      "step": 5705
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.23836898803711,
      "learning_rate": 6.680724242890097e-06,
      "loss": 0.8993,
      "step": 5706
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.551638126373291,
      "learning_rate": 6.6796304579829676e-06,
      "loss": 0.3913,
      "step": 5707
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.362487316131592,
      "learning_rate": 6.678536582466133e-06,
      "loss": 0.1221,
      "step": 5708
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2287561893463135,
      "learning_rate": 6.677442616398605e-06,
      "loss": 0.1216,
      "step": 5709
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6257004737854004,
      "learning_rate": 6.676348559839395e-06,
      "loss": 0.0549,
      "step": 5710
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.699882984161377,
      "learning_rate": 6.675254412847527e-06,
      "loss": 0.7227,
      "step": 5711
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.928101062774658,
      "learning_rate": 6.674160175482024e-06,
      "loss": 0.0929,
      "step": 5712
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.70823860168457,
      "learning_rate": 6.673065847801916e-06,
      "loss": 1.0531,
      "step": 5713
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.47744083404541,
      "learning_rate": 6.67197142986624e-06,
      "loss": 0.1822,
      "step": 5714
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.828028440475464,
      "learning_rate": 6.670876921734034e-06,
      "loss": 0.1369,
      "step": 5715
    },
    {
      "epoch": 0.41,
      "grad_norm": 13.891085624694824,
      "learning_rate": 6.669782323464341e-06,
      "loss": 1.1617,
      "step": 5716
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.788393974304199,
      "learning_rate": 6.668687635116214e-06,
      "loss": 0.1719,
      "step": 5717
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.738246202468872,
      "learning_rate": 6.667592856748703e-06,
      "loss": 0.7228,
      "step": 5718
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.374539375305176,
      "learning_rate": 6.6664979884208716e-06,
      "loss": 0.0969,
      "step": 5719
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.479330539703369,
      "learning_rate": 6.665403030191782e-06,
      "loss": 0.6826,
      "step": 5720
    },
    {
      "epoch": 0.41,
      "grad_norm": 16.424312591552734,
      "learning_rate": 6.6643079821205025e-06,
      "loss": 1.5175,
      "step": 5721
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5859408378601074,
      "learning_rate": 6.663212844266106e-06,
      "loss": 0.0802,
      "step": 5722
    },
    {
      "epoch": 0.41,
      "grad_norm": 11.274689674377441,
      "learning_rate": 6.6621176166876735e-06,
      "loss": 0.7631,
      "step": 5723
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.917872428894043,
      "learning_rate": 6.661022299444286e-06,
      "loss": 0.6898,
      "step": 5724
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.9174258708953857,
      "learning_rate": 6.659926892595033e-06,
      "loss": 0.2276,
      "step": 5725
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.186534404754639,
      "learning_rate": 6.658831396199007e-06,
      "loss": 0.9599,
      "step": 5726
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.3363373279571533,
      "learning_rate": 6.657735810315307e-06,
      "loss": 0.1365,
      "step": 5727
    },
    {
      "epoch": 0.41,
      "grad_norm": 16.53277015686035,
      "learning_rate": 6.6566401350030344e-06,
      "loss": 0.9617,
      "step": 5728
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5783419609069824,
      "learning_rate": 6.655544370321297e-06,
      "loss": 0.1441,
      "step": 5729
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.318081855773926,
      "learning_rate": 6.654448516329208e-06,
      "loss": 0.2224,
      "step": 5730
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.987942695617676,
      "learning_rate": 6.6533525730858835e-06,
      "loss": 0.8087,
      "step": 5731
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.848395586013794,
      "learning_rate": 6.652256540650446e-06,
      "loss": 0.1085,
      "step": 5732
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.375058174133301,
      "learning_rate": 6.651160419082024e-06,
      "loss": 0.8746,
      "step": 5733
    },
    {
      "epoch": 0.41,
      "grad_norm": 15.563003540039062,
      "learning_rate": 6.650064208439746e-06,
      "loss": 0.664,
      "step": 5734
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.876977920532227,
      "learning_rate": 6.6489679087827485e-06,
      "loss": 0.9724,
      "step": 5735
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.7959177494049072,
      "learning_rate": 6.647871520170175e-06,
      "loss": 0.1458,
      "step": 5736
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.749453544616699,
      "learning_rate": 6.646775042661171e-06,
      "loss": 0.1792,
      "step": 5737
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5814318656921387,
      "learning_rate": 6.645678476314886e-06,
      "loss": 0.0818,
      "step": 5738
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.298441410064697,
      "learning_rate": 6.644581821190478e-06,
      "loss": 0.9425,
      "step": 5739
    },
    {
      "epoch": 0.41,
      "grad_norm": 12.037821769714355,
      "learning_rate": 6.643485077347105e-06,
      "loss": 0.8286,
      "step": 5740
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.304684638977051,
      "learning_rate": 6.642388244843931e-06,
      "loss": 1.0567,
      "step": 5741
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.097854614257812,
      "learning_rate": 6.64129132374013e-06,
      "loss": 1.5002,
      "step": 5742
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.307095527648926,
      "learning_rate": 6.640194314094874e-06,
      "loss": 0.6283,
      "step": 5743
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3558337688446045,
      "learning_rate": 6.639097215967344e-06,
      "loss": 0.0822,
      "step": 5744
    },
    {
      "epoch": 0.41,
      "grad_norm": 14.84233283996582,
      "learning_rate": 6.638000029416724e-06,
      "loss": 1.2575,
      "step": 5745
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8973779678344727,
      "learning_rate": 6.636902754502203e-06,
      "loss": 0.1678,
      "step": 5746
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.657292127609253,
      "learning_rate": 6.635805391282972e-06,
      "loss": 0.8543,
      "step": 5747
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.4432058334350586,
      "learning_rate": 6.634707939818233e-06,
      "loss": 0.126,
      "step": 5748
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8460640907287598,
      "learning_rate": 6.633610400167189e-06,
      "loss": 0.133,
      "step": 5749
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.723191738128662,
      "learning_rate": 6.632512772389046e-06,
      "loss": 0.3854,
      "step": 5750
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.9954609870910645,
      "learning_rate": 6.63141505654302e-06,
      "loss": 0.0999,
      "step": 5751
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.0982718467712402,
      "learning_rate": 6.630317252688326e-06,
      "loss": 0.2142,
      "step": 5752
    },
    {
      "epoch": 0.41,
      "grad_norm": 13.300171852111816,
      "learning_rate": 6.629219360884187e-06,
      "loss": 1.4167,
      "step": 5753
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.1267447471618652,
      "learning_rate": 6.628121381189832e-06,
      "loss": 0.1736,
      "step": 5754
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.319631576538086,
      "learning_rate": 6.627023313664489e-06,
      "loss": 0.123,
      "step": 5755
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.086050033569336,
      "learning_rate": 6.6259251583673975e-06,
      "loss": 0.217,
      "step": 5756
    },
    {
      "epoch": 0.41,
      "grad_norm": 12.681012153625488,
      "learning_rate": 6.6248269153578e-06,
      "loss": 1.044,
      "step": 5757
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.5393195152282715,
      "learning_rate": 6.62372858469494e-06,
      "loss": 0.8972,
      "step": 5758
    },
    {
      "epoch": 0.41,
      "grad_norm": 28.522647857666016,
      "learning_rate": 6.622630166438068e-06,
      "loss": 1.195,
      "step": 5759
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.536642551422119,
      "learning_rate": 6.62153166064644e-06,
      "loss": 0.9117,
      "step": 5760
    },
    {
      "epoch": 0.41,
      "grad_norm": 14.58309268951416,
      "learning_rate": 6.620433067379318e-06,
      "loss": 1.0819,
      "step": 5761
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.539823293685913,
      "learning_rate": 6.619334386695966e-06,
      "loss": 0.123,
      "step": 5762
    },
    {
      "epoch": 0.41,
      "grad_norm": 21.402273178100586,
      "learning_rate": 6.618235618655653e-06,
      "loss": 0.8136,
      "step": 5763
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.9852137565612793,
      "learning_rate": 6.617136763317653e-06,
      "loss": 0.8662,
      "step": 5764
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.648226022720337,
      "learning_rate": 6.616037820741246e-06,
      "loss": 0.13,
      "step": 5765
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.919534206390381,
      "learning_rate": 6.614938790985716e-06,
      "loss": 0.0889,
      "step": 5766
    },
    {
      "epoch": 0.41,
      "grad_norm": 14.750734329223633,
      "learning_rate": 6.61383967411035e-06,
      "loss": 1.3874,
      "step": 5767
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.9740705490112305,
      "learning_rate": 6.612740470174445e-06,
      "loss": 0.1882,
      "step": 5768
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2633495330810547,
      "learning_rate": 6.6116411792372945e-06,
      "loss": 0.1338,
      "step": 5769
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.632391929626465,
      "learning_rate": 6.610541801358203e-06,
      "loss": 0.6883,
      "step": 5770
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.421175003051758,
      "learning_rate": 6.609442336596476e-06,
      "loss": 0.112,
      "step": 5771
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2811262607574463,
      "learning_rate": 6.6083427850114294e-06,
      "loss": 0.1337,
      "step": 5772
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.14937686920166,
      "learning_rate": 6.607243146662375e-06,
      "loss": 0.4847,
      "step": 5773
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.317826509475708,
      "learning_rate": 6.606143421608638e-06,
      "loss": 0.2023,
      "step": 5774
    },
    {
      "epoch": 0.41,
      "grad_norm": 12.768814086914062,
      "learning_rate": 6.6050436099095435e-06,
      "loss": 1.0377,
      "step": 5775
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.6348142623901367,
      "learning_rate": 6.603943711624421e-06,
      "loss": 0.2321,
      "step": 5776
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.62860369682312,
      "learning_rate": 6.602843726812606e-06,
      "loss": 0.1615,
      "step": 5777
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.016993761062622,
      "learning_rate": 6.6017436555334405e-06,
      "loss": 0.1042,
      "step": 5778
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.91325044631958,
      "learning_rate": 6.600643497846268e-06,
      "loss": 0.6598,
      "step": 5779
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4114573001861572,
      "learning_rate": 6.599543253810435e-06,
      "loss": 0.0937,
      "step": 5780
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.999547243118286,
      "learning_rate": 6.5984429234853e-06,
      "loss": 0.2258,
      "step": 5781
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2086808681488037,
      "learning_rate": 6.5973425069302185e-06,
      "loss": 0.1048,
      "step": 5782
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4413530826568604,
      "learning_rate": 6.596242004204556e-06,
      "loss": 0.12,
      "step": 5783
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.3243377208709717,
      "learning_rate": 6.595141415367678e-06,
      "loss": 0.2387,
      "step": 5784
    },
    {
      "epoch": 0.41,
      "grad_norm": 14.571833610534668,
      "learning_rate": 6.594040740478962e-06,
      "loss": 1.0167,
      "step": 5785
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.6254308223724365,
      "learning_rate": 6.5929399795977786e-06,
      "loss": 0.1482,
      "step": 5786
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.9102768898010254,
      "learning_rate": 6.591839132783514e-06,
      "loss": 0.935,
      "step": 5787
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.842111587524414,
      "learning_rate": 6.5907382000955546e-06,
      "loss": 0.6658,
      "step": 5788
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4030494689941406,
      "learning_rate": 6.5896371815932895e-06,
      "loss": 0.2909,
      "step": 5789
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.440455436706543,
      "learning_rate": 6.588536077336114e-06,
      "loss": 1.0261,
      "step": 5790
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.9490132331848145,
      "learning_rate": 6.587434887383433e-06,
      "loss": 0.1513,
      "step": 5791
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.543192863464355,
      "learning_rate": 6.5863336117946465e-06,
      "loss": 1.5359,
      "step": 5792
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8475204706192017,
      "learning_rate": 6.585232250629167e-06,
      "loss": 0.049,
      "step": 5793
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.015726089477539,
      "learning_rate": 6.584130803946408e-06,
      "loss": 0.4118,
      "step": 5794
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.397608280181885,
      "learning_rate": 6.583029271805788e-06,
      "loss": 0.6395,
      "step": 5795
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.51696252822876,
      "learning_rate": 6.581927654266731e-06,
      "loss": 0.716,
      "step": 5796
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.5919032096862793,
      "learning_rate": 6.580825951388665e-06,
      "loss": 0.1726,
      "step": 5797
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.0965423583984375,
      "learning_rate": 6.579724163231023e-06,
      "loss": 0.5421,
      "step": 5798
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.970991611480713,
      "learning_rate": 6.578622289853242e-06,
      "loss": 1.004,
      "step": 5799
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.181187868118286,
      "learning_rate": 6.577520331314764e-06,
      "loss": 0.0947,
      "step": 5800
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.355743408203125,
      "learning_rate": 6.576418287675035e-06,
      "loss": 0.2341,
      "step": 5801
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9535043239593506,
      "learning_rate": 6.575316158993507e-06,
      "loss": 0.1119,
      "step": 5802
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.939568042755127,
      "learning_rate": 6.574213945329634e-06,
      "loss": 0.1985,
      "step": 5803
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.100254535675049,
      "learning_rate": 6.573111646742878e-06,
      "loss": 0.8896,
      "step": 5804
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.092838764190674,
      "learning_rate": 6.572009263292704e-06,
      "loss": 0.7679,
      "step": 5805
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.6787984371185303,
      "learning_rate": 6.57090679503858e-06,
      "loss": 0.1428,
      "step": 5806
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.344935894012451,
      "learning_rate": 6.569804242039978e-06,
      "loss": 0.0896,
      "step": 5807
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.454908847808838,
      "learning_rate": 6.568701604356383e-06,
      "loss": 0.1202,
      "step": 5808
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.835449457168579,
      "learning_rate": 6.5675988820472725e-06,
      "loss": 0.2607,
      "step": 5809
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.8407421112060547,
      "learning_rate": 6.5664960751721366e-06,
      "loss": 0.1514,
      "step": 5810
    },
    {
      "epoch": 0.42,
      "grad_norm": 23.163362503051758,
      "learning_rate": 6.565393183790468e-06,
      "loss": 1.2851,
      "step": 5811
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.953825950622559,
      "learning_rate": 6.564290207961761e-06,
      "loss": 0.5849,
      "step": 5812
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.918600082397461,
      "learning_rate": 6.563187147745518e-06,
      "loss": 0.925,
      "step": 5813
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.647313117980957,
      "learning_rate": 6.5620840032012465e-06,
      "loss": 0.6955,
      "step": 5814
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.971953392028809,
      "learning_rate": 6.560980774388455e-06,
      "loss": 0.8329,
      "step": 5815
    },
    {
      "epoch": 0.42,
      "grad_norm": 15.8052339553833,
      "learning_rate": 6.559877461366659e-06,
      "loss": 1.3449,
      "step": 5816
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0959625244140625,
      "learning_rate": 6.558774064195381e-06,
      "loss": 0.1393,
      "step": 5817
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.6626944541931152,
      "learning_rate": 6.557670582934139e-06,
      "loss": 0.2055,
      "step": 5818
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.7761988639831543,
      "learning_rate": 6.556567017642468e-06,
      "loss": 0.8472,
      "step": 5819
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.4215738773345947,
      "learning_rate": 6.555463368379897e-06,
      "loss": 0.1021,
      "step": 5820
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.3489410877227783,
      "learning_rate": 6.554359635205965e-06,
      "loss": 0.1548,
      "step": 5821
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.6755735874176025,
      "learning_rate": 6.553255818180214e-06,
      "loss": 0.8242,
      "step": 5822
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.667055368423462,
      "learning_rate": 6.552151917362193e-06,
      "loss": 0.1484,
      "step": 5823
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.237597942352295,
      "learning_rate": 6.5510479328114485e-06,
      "loss": 0.0637,
      "step": 5824
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.362079620361328,
      "learning_rate": 6.54994386458754e-06,
      "loss": 0.5291,
      "step": 5825
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.8220269680023193,
      "learning_rate": 6.548839712750027e-06,
      "loss": 0.1421,
      "step": 5826
    },
    {
      "epoch": 0.42,
      "grad_norm": 29.46271514892578,
      "learning_rate": 6.547735477358474e-06,
      "loss": 1.4723,
      "step": 5827
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7073464393615723,
      "learning_rate": 6.546631158472452e-06,
      "loss": 0.2356,
      "step": 5828
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.774195432662964,
      "learning_rate": 6.545526756151532e-06,
      "loss": 0.1129,
      "step": 5829
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.607325553894043,
      "learning_rate": 6.544422270455293e-06,
      "loss": 0.7997,
      "step": 5830
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.066598415374756,
      "learning_rate": 6.54331770144332e-06,
      "loss": 0.1875,
      "step": 5831
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.091453552246094,
      "learning_rate": 6.542213049175197e-06,
      "loss": 0.8604,
      "step": 5832
    },
    {
      "epoch": 0.42,
      "grad_norm": 16.53289222717285,
      "learning_rate": 6.541108313710519e-06,
      "loss": 0.8777,
      "step": 5833
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.6450445652008057,
      "learning_rate": 6.54000349510888e-06,
      "loss": 0.1006,
      "step": 5834
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.140445709228516,
      "learning_rate": 6.538898593429883e-06,
      "loss": 0.777,
      "step": 5835
    },
    {
      "epoch": 0.42,
      "grad_norm": 13.14768123626709,
      "learning_rate": 6.537793608733131e-06,
      "loss": 0.5812,
      "step": 5836
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9689111709594727,
      "learning_rate": 6.536688541078236e-06,
      "loss": 0.1565,
      "step": 5837
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.706376075744629,
      "learning_rate": 6.535583390524808e-06,
      "loss": 0.1822,
      "step": 5838
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.416284561157227,
      "learning_rate": 6.534478157132471e-06,
      "loss": 0.8001,
      "step": 5839
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.656754493713379,
      "learning_rate": 6.533372840960844e-06,
      "loss": 1.0672,
      "step": 5840
    },
    {
      "epoch": 0.42,
      "grad_norm": 13.431079864501953,
      "learning_rate": 6.532267442069558e-06,
      "loss": 1.1226,
      "step": 5841
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.2711892127990723,
      "learning_rate": 6.531161960518243e-06,
      "loss": 0.1557,
      "step": 5842
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7393290996551514,
      "learning_rate": 6.530056396366535e-06,
      "loss": 0.1295,
      "step": 5843
    },
    {
      "epoch": 0.42,
      "grad_norm": 16.034454345703125,
      "learning_rate": 6.528950749674077e-06,
      "loss": 0.9647,
      "step": 5844
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.521756172180176,
      "learning_rate": 6.527845020500514e-06,
      "loss": 1.0754,
      "step": 5845
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.223973274230957,
      "learning_rate": 6.526739208905494e-06,
      "loss": 0.7159,
      "step": 5846
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1868648529052734,
      "learning_rate": 6.525633314948673e-06,
      "loss": 0.0924,
      "step": 5847
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.2103047370910645,
      "learning_rate": 6.52452733868971e-06,
      "loss": 0.9083,
      "step": 5848
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.055383205413818,
      "learning_rate": 6.523421280188267e-06,
      "loss": 0.2486,
      "step": 5849
    },
    {
      "epoch": 0.42,
      "grad_norm": 15.621162414550781,
      "learning_rate": 6.522315139504013e-06,
      "loss": 1.1896,
      "step": 5850
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.382949113845825,
      "learning_rate": 6.5212089166966195e-06,
      "loss": 0.2291,
      "step": 5851
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.481952667236328,
      "learning_rate": 6.520102611825763e-06,
      "loss": 0.6398,
      "step": 5852
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.3239612579345703,
      "learning_rate": 6.5189962249511244e-06,
      "loss": 0.1263,
      "step": 5853
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.641979694366455,
      "learning_rate": 6.517889756132389e-06,
      "loss": 0.1269,
      "step": 5854
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.035675048828125,
      "learning_rate": 6.516783205429247e-06,
      "loss": 0.6738,
      "step": 5855
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.899109840393066,
      "learning_rate": 6.515676572901392e-06,
      "loss": 0.5195,
      "step": 5856
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9228084087371826,
      "learning_rate": 6.514569858608524e-06,
      "loss": 0.164,
      "step": 5857
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.316786289215088,
      "learning_rate": 6.513463062610344e-06,
      "loss": 0.572,
      "step": 5858
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.575024604797363,
      "learning_rate": 6.512356184966562e-06,
      "loss": 0.6891,
      "step": 5859
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.710633277893066,
      "learning_rate": 6.511249225736888e-06,
      "loss": 0.8158,
      "step": 5860
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.617913246154785,
      "learning_rate": 6.510142184981036e-06,
      "loss": 0.9382,
      "step": 5861
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.1114659309387207,
      "learning_rate": 6.5090350627587325e-06,
      "loss": 0.2094,
      "step": 5862
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.0997276306152344,
      "learning_rate": 6.507927859129698e-06,
      "loss": 0.1567,
      "step": 5863
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8421639204025269,
      "learning_rate": 6.506820574153662e-06,
      "loss": 0.0885,
      "step": 5864
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.6040151119232178,
      "learning_rate": 6.505713207890364e-06,
      "loss": 0.1371,
      "step": 5865
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.343789100646973,
      "learning_rate": 6.504605760399534e-06,
      "loss": 0.8274,
      "step": 5866
    },
    {
      "epoch": 0.42,
      "grad_norm": 12.5361909866333,
      "learning_rate": 6.503498231740919e-06,
      "loss": 0.6284,
      "step": 5867
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.6065163612365723,
      "learning_rate": 6.502390621974267e-06,
      "loss": 0.1715,
      "step": 5868
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.441183090209961,
      "learning_rate": 6.501282931159326e-06,
      "loss": 0.1672,
      "step": 5869
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.482560157775879,
      "learning_rate": 6.500175159355855e-06,
      "loss": 0.8542,
      "step": 5870
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9998855590820312,
      "learning_rate": 6.499067306623612e-06,
      "loss": 0.2444,
      "step": 5871
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.621125221252441,
      "learning_rate": 6.497959373022362e-06,
      "loss": 0.5111,
      "step": 5872
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.131278991699219,
      "learning_rate": 6.496851358611874e-06,
      "loss": 0.6849,
      "step": 5873
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0700831413269043,
      "learning_rate": 6.495743263451921e-06,
      "loss": 0.1424,
      "step": 5874
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.95937728881836,
      "learning_rate": 6.49463508760228e-06,
      "loss": 0.7606,
      "step": 5875
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.8617920875549316,
      "learning_rate": 6.493526831122735e-06,
      "loss": 0.1668,
      "step": 5876
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.684346675872803,
      "learning_rate": 6.4924184940730696e-06,
      "loss": 0.7563,
      "step": 5877
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.189572334289551,
      "learning_rate": 6.491310076513076e-06,
      "loss": 0.8743,
      "step": 5878
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.098965167999268,
      "learning_rate": 6.4902015785025494e-06,
      "loss": 0.374,
      "step": 5879
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.032729148864746,
      "learning_rate": 6.489093000101288e-06,
      "loss": 0.7972,
      "step": 5880
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4897358417510986,
      "learning_rate": 6.487984341369093e-06,
      "loss": 0.2654,
      "step": 5881
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.092984676361084,
      "learning_rate": 6.486875602365778e-06,
      "loss": 0.8599,
      "step": 5882
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.294508457183838,
      "learning_rate": 6.485766783151153e-06,
      "loss": 0.1093,
      "step": 5883
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.292454957962036,
      "learning_rate": 6.484657883785032e-06,
      "loss": 0.2732,
      "step": 5884
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.547504425048828,
      "learning_rate": 6.4835489043272395e-06,
      "loss": 1.3641,
      "step": 5885
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.0690224170684814,
      "learning_rate": 6.482439844837598e-06,
      "loss": 0.1556,
      "step": 5886
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.171391487121582,
      "learning_rate": 6.481330705375939e-06,
      "loss": 0.585,
      "step": 5887
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.783137321472168,
      "learning_rate": 6.480221486002095e-06,
      "loss": 0.822,
      "step": 5888
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9880788326263428,
      "learning_rate": 6.4791121867759045e-06,
      "loss": 0.1912,
      "step": 5889
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.764956474304199,
      "learning_rate": 6.478002807757211e-06,
      "loss": 0.7914,
      "step": 5890
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6415835618972778,
      "learning_rate": 6.476893349005861e-06,
      "loss": 0.0621,
      "step": 5891
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.388064861297607,
      "learning_rate": 6.4757838105817035e-06,
      "loss": 0.9354,
      "step": 5892
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.7878963947296143,
      "learning_rate": 6.474674192544598e-06,
      "loss": 0.9079,
      "step": 5893
    },
    {
      "epoch": 0.42,
      "grad_norm": 18.288042068481445,
      "learning_rate": 6.473564494954399e-06,
      "loss": 1.3005,
      "step": 5894
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.918119430541992,
      "learning_rate": 6.472454717870974e-06,
      "loss": 1.0156,
      "step": 5895
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.0576090812683105,
      "learning_rate": 6.471344861354193e-06,
      "loss": 0.1928,
      "step": 5896
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.830085754394531,
      "learning_rate": 6.470234925463923e-06,
      "loss": 0.9366,
      "step": 5897
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.430872917175293,
      "learning_rate": 6.4691249102600454e-06,
      "loss": 0.9741,
      "step": 5898
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.743119239807129,
      "learning_rate": 6.468014815802439e-06,
      "loss": 1.334,
      "step": 5899
    },
    {
      "epoch": 0.42,
      "grad_norm": 20.94127655029297,
      "learning_rate": 6.466904642150991e-06,
      "loss": 0.7528,
      "step": 5900
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.542970657348633,
      "learning_rate": 6.465794389365589e-06,
      "loss": 1.0365,
      "step": 5901
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.7281720638275146,
      "learning_rate": 6.4646840575061295e-06,
      "loss": 0.9611,
      "step": 5902
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.482203960418701,
      "learning_rate": 6.463573646632508e-06,
      "loss": 0.1529,
      "step": 5903
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.5760269165039062,
      "learning_rate": 6.462463156804628e-06,
      "loss": 0.2068,
      "step": 5904
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.230319976806641,
      "learning_rate": 6.461352588082397e-06,
      "loss": 0.6919,
      "step": 5905
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9711601734161377,
      "learning_rate": 6.460241940525724e-06,
      "loss": 0.1388,
      "step": 5906
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.649962902069092,
      "learning_rate": 6.459131214194528e-06,
      "loss": 0.1837,
      "step": 5907
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9267971515655518,
      "learning_rate": 6.458020409148727e-06,
      "loss": 0.132,
      "step": 5908
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.777182579040527,
      "learning_rate": 6.45690952544824e-06,
      "loss": 0.6351,
      "step": 5909
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.6635334491729736,
      "learning_rate": 6.455798563153002e-06,
      "loss": 0.0759,
      "step": 5910
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.021687507629395,
      "learning_rate": 6.45468752232294e-06,
      "loss": 0.8702,
      "step": 5911
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.0812199115753174,
      "learning_rate": 6.453576403017992e-06,
      "loss": 0.1987,
      "step": 5912
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.234639644622803,
      "learning_rate": 6.452465205298101e-06,
      "loss": 0.972,
      "step": 5913
    },
    {
      "epoch": 0.42,
      "grad_norm": 21.370689392089844,
      "learning_rate": 6.45135392922321e-06,
      "loss": 1.4432,
      "step": 5914
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.097420692443848,
      "learning_rate": 6.450242574853267e-06,
      "loss": 1.2275,
      "step": 5915
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.196736335754395,
      "learning_rate": 6.4491311422482276e-06,
      "loss": 0.9262,
      "step": 5916
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.5480799674987793,
      "learning_rate": 6.448019631468048e-06,
      "loss": 0.161,
      "step": 5917
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.141777992248535,
      "learning_rate": 6.4469080425726885e-06,
      "loss": 0.8785,
      "step": 5918
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.720037460327148,
      "learning_rate": 6.44579637562212e-06,
      "loss": 0.9577,
      "step": 5919
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.7518367767333984,
      "learning_rate": 6.444684630676308e-06,
      "loss": 0.8705,
      "step": 5920
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.5355172157287598,
      "learning_rate": 6.443572807795229e-06,
      "loss": 0.1015,
      "step": 5921
    },
    {
      "epoch": 0.42,
      "grad_norm": 25.49697494506836,
      "learning_rate": 6.442460907038861e-06,
      "loss": 1.0591,
      "step": 5922
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.844198226928711,
      "learning_rate": 6.441348928467187e-06,
      "loss": 0.143,
      "step": 5923
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.506565093994141,
      "learning_rate": 6.4402368721401955e-06,
      "loss": 0.8375,
      "step": 5924
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.019144058227539,
      "learning_rate": 6.4391247381178744e-06,
      "loss": 0.784,
      "step": 5925
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.275196552276611,
      "learning_rate": 6.438012526460223e-06,
      "loss": 1.0238,
      "step": 5926
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2383055686950684,
      "learning_rate": 6.436900237227236e-06,
      "loss": 0.061,
      "step": 5927
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.7297253608703613,
      "learning_rate": 6.435787870478922e-06,
      "loss": 0.2322,
      "step": 5928
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.106345176696777,
      "learning_rate": 6.434675426275287e-06,
      "loss": 0.8157,
      "step": 5929
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.3115999698638916,
      "learning_rate": 6.433562904676345e-06,
      "loss": 0.2147,
      "step": 5930
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.491360664367676,
      "learning_rate": 6.432450305742107e-06,
      "loss": 1.1336,
      "step": 5931
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1857190132141113,
      "learning_rate": 6.4313376295326005e-06,
      "loss": 0.1329,
      "step": 5932
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.421431541442871,
      "learning_rate": 6.430224876107846e-06,
      "loss": 1.2238,
      "step": 5933
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.477218627929688,
      "learning_rate": 6.429112045527872e-06,
      "loss": 0.6452,
      "step": 5934
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.8543062210083,
      "learning_rate": 6.427999137852712e-06,
      "loss": 0.9807,
      "step": 5935
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.640300989151001,
      "learning_rate": 6.426886153142407e-06,
      "loss": 0.189,
      "step": 5936
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.925610542297363,
      "learning_rate": 6.4257730914569915e-06,
      "loss": 0.7635,
      "step": 5937
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.787459373474121,
      "learning_rate": 6.424659952856516e-06,
      "loss": 0.7728,
      "step": 5938
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8641855716705322,
      "learning_rate": 6.4235467374010295e-06,
      "loss": 0.1259,
      "step": 5939
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.3239169120788574,
      "learning_rate": 6.4224334451505845e-06,
      "loss": 0.1573,
      "step": 5940
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.038015842437744,
      "learning_rate": 6.4213200761652386e-06,
      "loss": 0.1179,
      "step": 5941
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3492982387542725,
      "learning_rate": 6.420206630505055e-06,
      "loss": 0.1543,
      "step": 5942
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.326326847076416,
      "learning_rate": 6.419093108230099e-06,
      "loss": 0.1606,
      "step": 5943
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1515095233917236,
      "learning_rate": 6.417979509400443e-06,
      "loss": 0.1565,
      "step": 5944
    },
    {
      "epoch": 0.43,
      "grad_norm": 14.028761863708496,
      "learning_rate": 6.416865834076158e-06,
      "loss": 1.2356,
      "step": 5945
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.7319767475128174,
      "learning_rate": 6.415752082317326e-06,
      "loss": 0.1981,
      "step": 5946
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8468966484069824,
      "learning_rate": 6.414638254184027e-06,
      "loss": 0.1653,
      "step": 5947
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.9845831394195557,
      "learning_rate": 6.413524349736348e-06,
      "loss": 0.8088,
      "step": 5948
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.044163227081299,
      "learning_rate": 6.412410369034382e-06,
      "loss": 0.0796,
      "step": 5949
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3029658794403076,
      "learning_rate": 6.411296312138224e-06,
      "loss": 0.111,
      "step": 5950
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.355725288391113,
      "learning_rate": 6.41018217910797e-06,
      "loss": 0.9605,
      "step": 5951
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.284919738769531,
      "learning_rate": 6.409067970003725e-06,
      "loss": 0.7535,
      "step": 5952
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5658721923828125,
      "learning_rate": 6.407953684885598e-06,
      "loss": 0.116,
      "step": 5953
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.7761831283569336,
      "learning_rate": 6.406839323813699e-06,
      "loss": 0.8638,
      "step": 5954
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.9416050910949707,
      "learning_rate": 6.405724886848143e-06,
      "loss": 0.9838,
      "step": 5955
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.874084234237671,
      "learning_rate": 6.404610374049052e-06,
      "loss": 0.1241,
      "step": 5956
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.4684410095214844,
      "learning_rate": 6.403495785476546e-06,
      "loss": 0.7241,
      "step": 5957
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.646839141845703,
      "learning_rate": 6.402381121190755e-06,
      "loss": 0.1153,
      "step": 5958
    },
    {
      "epoch": 0.43,
      "grad_norm": 16.035839080810547,
      "learning_rate": 6.401266381251812e-06,
      "loss": 0.7251,
      "step": 5959
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.962916374206543,
      "learning_rate": 6.400151565719851e-06,
      "loss": 0.7225,
      "step": 5960
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8849434852600098,
      "learning_rate": 6.399036674655014e-06,
      "loss": 0.1072,
      "step": 5961
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.026030540466309,
      "learning_rate": 6.3979217081174425e-06,
      "loss": 0.7982,
      "step": 5962
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8546369075775146,
      "learning_rate": 6.396806666167287e-06,
      "loss": 0.2002,
      "step": 5963
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5936954021453857,
      "learning_rate": 6.395691548864699e-06,
      "loss": 0.0986,
      "step": 5964
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.57080078125,
      "learning_rate": 6.394576356269835e-06,
      "loss": 0.9037,
      "step": 5965
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.0577242374420166,
      "learning_rate": 6.3934610884428535e-06,
      "loss": 0.1427,
      "step": 5966
    },
    {
      "epoch": 0.43,
      "grad_norm": 16.836061477661133,
      "learning_rate": 6.392345745443924e-06,
      "loss": 0.7216,
      "step": 5967
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.418591499328613,
      "learning_rate": 6.391230327333211e-06,
      "loss": 0.4402,
      "step": 5968
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.862484455108643,
      "learning_rate": 6.390114834170887e-06,
      "loss": 0.7177,
      "step": 5969
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.152195453643799,
      "learning_rate": 6.388999266017131e-06,
      "loss": 0.1851,
      "step": 5970
    },
    {
      "epoch": 0.43,
      "grad_norm": 14.958252906799316,
      "learning_rate": 6.3878836229321195e-06,
      "loss": 1.1796,
      "step": 5971
    },
    {
      "epoch": 0.43,
      "grad_norm": 17.428686141967773,
      "learning_rate": 6.386767904976041e-06,
      "loss": 1.4804,
      "step": 5972
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.874126434326172,
      "learning_rate": 6.385652112209085e-06,
      "loss": 0.8631,
      "step": 5973
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.966129779815674,
      "learning_rate": 6.384536244691442e-06,
      "loss": 0.1697,
      "step": 5974
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1065547466278076,
      "learning_rate": 6.383420302483308e-06,
      "loss": 0.1432,
      "step": 5975
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.820228099822998,
      "learning_rate": 6.382304285644887e-06,
      "loss": 0.2235,
      "step": 5976
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.205411434173584,
      "learning_rate": 6.38118819423638e-06,
      "loss": 0.1136,
      "step": 5977
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.676462650299072,
      "learning_rate": 6.380072028318e-06,
      "loss": 1.0367,
      "step": 5978
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.728010654449463,
      "learning_rate": 6.378955787949956e-06,
      "loss": 0.164,
      "step": 5979
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.02685546875,
      "learning_rate": 6.3778394731924684e-06,
      "loss": 0.8785,
      "step": 5980
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5514636039733887,
      "learning_rate": 6.376723084105756e-06,
      "loss": 0.0796,
      "step": 5981
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.230253219604492,
      "learning_rate": 6.375606620750045e-06,
      "loss": 1.0118,
      "step": 5982
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.426972389221191,
      "learning_rate": 6.374490083185563e-06,
      "loss": 0.1281,
      "step": 5983
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.788980722427368,
      "learning_rate": 6.3733734714725434e-06,
      "loss": 0.1878,
      "step": 5984
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.9151644706726074,
      "learning_rate": 6.372256785671224e-06,
      "loss": 0.1467,
      "step": 5985
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.927000999450684,
      "learning_rate": 6.371140025841845e-06,
      "loss": 1.1284,
      "step": 5986
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.81056547164917,
      "learning_rate": 6.370023192044653e-06,
      "loss": 0.0596,
      "step": 5987
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.796532154083252,
      "learning_rate": 6.3689062843398945e-06,
      "loss": 0.1838,
      "step": 5988
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1946167945861816,
      "learning_rate": 6.3677893027878234e-06,
      "loss": 0.1688,
      "step": 5989
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.954973220825195,
      "learning_rate": 6.366672247448697e-06,
      "loss": 0.6872,
      "step": 5990
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.921651840209961,
      "learning_rate": 6.365555118382776e-06,
      "loss": 0.8478,
      "step": 5991
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.611519813537598,
      "learning_rate": 6.364437915650326e-06,
      "loss": 0.4561,
      "step": 5992
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.7626371383667,
      "learning_rate": 6.363320639311615e-06,
      "loss": 0.8528,
      "step": 5993
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.3515844345092773,
      "learning_rate": 6.362203289426913e-06,
      "loss": 0.2098,
      "step": 5994
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.5199613571167,
      "learning_rate": 6.361085866056502e-06,
      "loss": 0.6856,
      "step": 5995
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1681697368621826,
      "learning_rate": 6.359968369260662e-06,
      "loss": 0.2239,
      "step": 5996
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.513619422912598,
      "learning_rate": 6.358850799099674e-06,
      "loss": 1.15,
      "step": 5997
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.2059478759765625,
      "learning_rate": 6.357733155633831e-06,
      "loss": 0.8477,
      "step": 5998
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.093019962310791,
      "learning_rate": 6.356615438923422e-06,
      "loss": 0.0608,
      "step": 5999
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.53430461883545,
      "learning_rate": 6.3554976490287456e-06,
      "loss": 0.7056,
      "step": 6000
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5882668495178223,
      "learning_rate": 6.354379786010102e-06,
      "loss": 0.1199,
      "step": 6001
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.9666476249694824,
      "learning_rate": 6.353261849927796e-06,
      "loss": 0.1603,
      "step": 6002
    },
    {
      "epoch": 0.43,
      "grad_norm": 15.293228149414062,
      "learning_rate": 6.352143840842133e-06,
      "loss": 0.9067,
      "step": 6003
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.254301071166992,
      "learning_rate": 6.3510257588134315e-06,
      "loss": 0.1579,
      "step": 6004
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.364196300506592,
      "learning_rate": 6.3499076039020025e-06,
      "loss": 0.3983,
      "step": 6005
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.011345863342285,
      "learning_rate": 6.348789376168167e-06,
      "loss": 0.829,
      "step": 6006
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.565458297729492,
      "learning_rate": 6.347671075672252e-06,
      "loss": 0.1507,
      "step": 6007
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.7788329124450684,
      "learning_rate": 6.3465527024745825e-06,
      "loss": 0.1509,
      "step": 6008
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.835688591003418,
      "learning_rate": 6.345434256635491e-06,
      "loss": 0.8256,
      "step": 6009
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.9295058250427246,
      "learning_rate": 6.344315738215316e-06,
      "loss": 0.1584,
      "step": 6010
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.008113861083984,
      "learning_rate": 6.343197147274395e-06,
      "loss": 0.167,
      "step": 6011
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8386311531066895,
      "learning_rate": 6.3420784838730705e-06,
      "loss": 0.1684,
      "step": 6012
    },
    {
      "epoch": 0.43,
      "grad_norm": 19.52285385131836,
      "learning_rate": 6.340959748071695e-06,
      "loss": 0.967,
      "step": 6013
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5027570724487305,
      "learning_rate": 6.339840939930614e-06,
      "loss": 0.1325,
      "step": 6014
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.566340446472168,
      "learning_rate": 6.338722059510187e-06,
      "loss": 0.9765,
      "step": 6015
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.668217658996582,
      "learning_rate": 6.337603106870772e-06,
      "loss": 0.6133,
      "step": 6016
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.332464218139648,
      "learning_rate": 6.336484082072732e-06,
      "loss": 0.9923,
      "step": 6017
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.010980606079102,
      "learning_rate": 6.335364985176435e-06,
      "loss": 1.2923,
      "step": 6018
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.629063606262207,
      "learning_rate": 6.334245816242253e-06,
      "loss": 0.5439,
      "step": 6019
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.418156385421753,
      "learning_rate": 6.333126575330557e-06,
      "loss": 0.1416,
      "step": 6020
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.038692951202393,
      "learning_rate": 6.332007262501731e-06,
      "loss": 0.7707,
      "step": 6021
    },
    {
      "epoch": 0.43,
      "grad_norm": 15.11922836303711,
      "learning_rate": 6.330887877816154e-06,
      "loss": 1.6812,
      "step": 6022
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.756718635559082,
      "learning_rate": 6.3297684213342126e-06,
      "loss": 0.0946,
      "step": 6023
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.130000114440918,
      "learning_rate": 6.3286488931163e-06,
      "loss": 0.8181,
      "step": 6024
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.776658296585083,
      "learning_rate": 6.327529293222808e-06,
      "loss": 0.8178,
      "step": 6025
    },
    {
      "epoch": 0.43,
      "grad_norm": 14.675040245056152,
      "learning_rate": 6.326409621714135e-06,
      "loss": 1.2221,
      "step": 6026
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.297731399536133,
      "learning_rate": 6.325289878650686e-06,
      "loss": 0.0937,
      "step": 6027
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.170948028564453,
      "learning_rate": 6.324170064092863e-06,
      "loss": 0.1336,
      "step": 6028
    },
    {
      "epoch": 0.43,
      "grad_norm": 18.08177947998047,
      "learning_rate": 6.323050178101076e-06,
      "loss": 0.8033,
      "step": 6029
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8172128200531006,
      "learning_rate": 6.3219302207357425e-06,
      "loss": 0.2049,
      "step": 6030
    },
    {
      "epoch": 0.43,
      "grad_norm": 18.859071731567383,
      "learning_rate": 6.320810192057275e-06,
      "loss": 1.2919,
      "step": 6031
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.9483091831207275,
      "learning_rate": 6.3196900921260985e-06,
      "loss": 0.1657,
      "step": 6032
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.016507625579834,
      "learning_rate": 6.318569921002636e-06,
      "loss": 0.6431,
      "step": 6033
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.495723009109497,
      "learning_rate": 6.317449678747318e-06,
      "loss": 0.1465,
      "step": 6034
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.8642072677612305,
      "learning_rate": 6.316329365420577e-06,
      "loss": 0.9693,
      "step": 6035
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.628289222717285,
      "learning_rate": 6.315208981082849e-06,
      "loss": 0.8745,
      "step": 6036
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.675324440002441,
      "learning_rate": 6.314088525794574e-06,
      "loss": 0.854,
      "step": 6037
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.501890659332275,
      "learning_rate": 6.312967999616199e-06,
      "loss": 0.7062,
      "step": 6038
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.9443376064300537,
      "learning_rate": 6.311847402608168e-06,
      "loss": 0.1562,
      "step": 6039
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.786128044128418,
      "learning_rate": 6.310726734830936e-06,
      "loss": 0.1612,
      "step": 6040
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.531996250152588,
      "learning_rate": 6.30960599634496e-06,
      "loss": 0.9847,
      "step": 6041
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.192736625671387,
      "learning_rate": 6.308485187210696e-06,
      "loss": 0.8182,
      "step": 6042
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.4159462451934814,
      "learning_rate": 6.307364307488607e-06,
      "loss": 0.1927,
      "step": 6043
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6380727291107178,
      "learning_rate": 6.306243357239166e-06,
      "loss": 0.1369,
      "step": 6044
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.959585428237915,
      "learning_rate": 6.3051223365228376e-06,
      "loss": 0.9672,
      "step": 6045
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.9515702724456787,
      "learning_rate": 6.3040012454001e-06,
      "loss": 0.832,
      "step": 6046
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.865669250488281,
      "learning_rate": 6.302880083931432e-06,
      "loss": 0.9109,
      "step": 6047
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.756895065307617,
      "learning_rate": 6.301758852177313e-06,
      "loss": 0.756,
      "step": 6048
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.8515944480896,
      "learning_rate": 6.300637550198233e-06,
      "loss": 0.7982,
      "step": 6049
    },
    {
      "epoch": 0.43,
      "grad_norm": 16.871835708618164,
      "learning_rate": 6.299516178054681e-06,
      "loss": 1.482,
      "step": 6050
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1646406650543213,
      "learning_rate": 6.2983947358071485e-06,
      "loss": 0.2055,
      "step": 6051
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.840834617614746,
      "learning_rate": 6.297273223516136e-06,
      "loss": 1.2217,
      "step": 6052
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.41524600982666,
      "learning_rate": 6.296151641242143e-06,
      "loss": 0.8946,
      "step": 6053
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.393472194671631,
      "learning_rate": 6.2950299890456755e-06,
      "loss": 0.251,
      "step": 6054
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8074917793273926,
      "learning_rate": 6.293908266987242e-06,
      "loss": 0.8631,
      "step": 6055
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.71891188621521,
      "learning_rate": 6.292786475127355e-06,
      "loss": 0.1857,
      "step": 6056
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.0151591300964355,
      "learning_rate": 6.291664613526531e-06,
      "loss": 0.9868,
      "step": 6057
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.461549758911133,
      "learning_rate": 6.290542682245291e-06,
      "loss": 1.2296,
      "step": 6058
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3217928409576416,
      "learning_rate": 6.289420681344157e-06,
      "loss": 0.1005,
      "step": 6059
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.866372585296631,
      "learning_rate": 6.288298610883659e-06,
      "loss": 0.1706,
      "step": 6060
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.398428201675415,
      "learning_rate": 6.2871764709243265e-06,
      "loss": 0.1661,
      "step": 6061
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.15523624420166,
      "learning_rate": 6.286054261526695e-06,
      "loss": 0.7246,
      "step": 6062
    },
    {
      "epoch": 0.43,
      "grad_norm": 19.0081729888916,
      "learning_rate": 6.2849319827513054e-06,
      "loss": 1.5252,
      "step": 6063
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2854158878326416,
      "learning_rate": 6.283809634658699e-06,
      "loss": 0.1028,
      "step": 6064
    },
    {
      "epoch": 0.43,
      "grad_norm": 26.377338409423828,
      "learning_rate": 6.282687217309423e-06,
      "loss": 1.2162,
      "step": 6065
    },
    {
      "epoch": 0.43,
      "grad_norm": 15.391912460327148,
      "learning_rate": 6.2815647307640244e-06,
      "loss": 1.3177,
      "step": 6066
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.607959270477295,
      "learning_rate": 6.280442175083063e-06,
      "loss": 0.825,
      "step": 6067
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.367272138595581,
      "learning_rate": 6.2793195503270895e-06,
      "loss": 0.0361,
      "step": 6068
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.8250763416290283,
      "learning_rate": 6.278196856556672e-06,
      "loss": 0.1521,
      "step": 6069
    },
    {
      "epoch": 0.44,
      "grad_norm": 13.485076904296875,
      "learning_rate": 6.277074093832371e-06,
      "loss": 0.7952,
      "step": 6070
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.0942296981811523,
      "learning_rate": 6.275951262214755e-06,
      "loss": 0.1879,
      "step": 6071
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8362478017807007,
      "learning_rate": 6.274828361764401e-06,
      "loss": 0.0563,
      "step": 6072
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.19979190826416,
      "learning_rate": 6.27370539254188e-06,
      "loss": 0.7513,
      "step": 6073
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.0888772010803223,
      "learning_rate": 6.272582354607775e-06,
      "loss": 0.221,
      "step": 6074
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.673801422119141,
      "learning_rate": 6.2714592480226696e-06,
      "loss": 0.5036,
      "step": 6075
    },
    {
      "epoch": 0.44,
      "grad_norm": 13.962189674377441,
      "learning_rate": 6.270336072847151e-06,
      "loss": 0.8245,
      "step": 6076
    },
    {
      "epoch": 0.44,
      "grad_norm": 13.441689491271973,
      "learning_rate": 6.269212829141808e-06,
      "loss": 0.6676,
      "step": 6077
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8053127527236938,
      "learning_rate": 6.268089516967239e-06,
      "loss": 0.0526,
      "step": 6078
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8128130435943604,
      "learning_rate": 6.266966136384038e-06,
      "loss": 0.0727,
      "step": 6079
    },
    {
      "epoch": 0.44,
      "grad_norm": 15.780427932739258,
      "learning_rate": 6.26584268745281e-06,
      "loss": 0.8207,
      "step": 6080
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.226046323776245,
      "learning_rate": 6.2647191702341625e-06,
      "loss": 0.0989,
      "step": 6081
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.810931444168091,
      "learning_rate": 6.263595584788703e-06,
      "loss": 0.1707,
      "step": 6082
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.59595251083374,
      "learning_rate": 6.262471931177043e-06,
      "loss": 0.3588,
      "step": 6083
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.142445087432861,
      "learning_rate": 6.261348209459802e-06,
      "loss": 0.7315,
      "step": 6084
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.427177906036377,
      "learning_rate": 6.260224419697599e-06,
      "loss": 0.0993,
      "step": 6085
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.577509641647339,
      "learning_rate": 6.259100561951059e-06,
      "loss": 0.2359,
      "step": 6086
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.405735969543457,
      "learning_rate": 6.2579766362808105e-06,
      "loss": 0.9326,
      "step": 6087
    },
    {
      "epoch": 0.44,
      "grad_norm": 22.84750747680664,
      "learning_rate": 6.256852642747484e-06,
      "loss": 0.8042,
      "step": 6088
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.783820152282715,
      "learning_rate": 6.255728581411713e-06,
      "loss": 0.9272,
      "step": 6089
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.042826652526855,
      "learning_rate": 6.254604452334141e-06,
      "loss": 0.7945,
      "step": 6090
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.698440551757812,
      "learning_rate": 6.253480255575405e-06,
      "loss": 1.1389,
      "step": 6091
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.2997593879699707,
      "learning_rate": 6.252355991196155e-06,
      "loss": 0.2472,
      "step": 6092
    },
    {
      "epoch": 0.44,
      "grad_norm": 18.315818786621094,
      "learning_rate": 6.25123165925704e-06,
      "loss": 0.2881,
      "step": 6093
    },
    {
      "epoch": 0.44,
      "grad_norm": 20.48809814453125,
      "learning_rate": 6.250107259818711e-06,
      "loss": 1.4437,
      "step": 6094
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.459249496459961,
      "learning_rate": 6.248982792941832e-06,
      "loss": 0.1066,
      "step": 6095
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.261301279067993,
      "learning_rate": 6.247858258687056e-06,
      "loss": 0.1696,
      "step": 6096
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.000524044036865,
      "learning_rate": 6.246733657115049e-06,
      "loss": 0.4557,
      "step": 6097
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.9718165397644043,
      "learning_rate": 6.2456089882864834e-06,
      "loss": 0.1269,
      "step": 6098
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.698227405548096,
      "learning_rate": 6.2444842522620256e-06,
      "loss": 0.806,
      "step": 6099
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.8464763164520264,
      "learning_rate": 6.243359449102353e-06,
      "loss": 0.1792,
      "step": 6100
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4122610092163086,
      "learning_rate": 6.242234578868147e-06,
      "loss": 0.1245,
      "step": 6101
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7899532318115234,
      "learning_rate": 6.241109641620086e-06,
      "loss": 0.0455,
      "step": 6102
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.193105459213257,
      "learning_rate": 6.2399846374188566e-06,
      "loss": 0.1617,
      "step": 6103
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.349413871765137,
      "learning_rate": 6.23885956632515e-06,
      "loss": 0.8181,
      "step": 6104
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.764707088470459,
      "learning_rate": 6.23773442839966e-06,
      "loss": 0.153,
      "step": 6105
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.0061869621276855,
      "learning_rate": 6.236609223703083e-06,
      "loss": 0.3062,
      "step": 6106
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.7125725746154785,
      "learning_rate": 6.2354839522961186e-06,
      "loss": 0.8745,
      "step": 6107
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.117087364196777,
      "learning_rate": 6.234358614239471e-06,
      "loss": 0.7955,
      "step": 6108
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.885385513305664,
      "learning_rate": 6.233233209593848e-06,
      "loss": 0.5966,
      "step": 6109
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.31489372253418,
      "learning_rate": 6.232107738419962e-06,
      "loss": 0.1068,
      "step": 6110
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.1134068965911865,
      "learning_rate": 6.230982200778528e-06,
      "loss": 0.2031,
      "step": 6111
    },
    {
      "epoch": 0.44,
      "grad_norm": 18.722360610961914,
      "learning_rate": 6.2298565967302635e-06,
      "loss": 0.92,
      "step": 6112
    },
    {
      "epoch": 0.44,
      "grad_norm": 15.364537239074707,
      "learning_rate": 6.228730926335891e-06,
      "loss": 1.0075,
      "step": 6113
    },
    {
      "epoch": 0.44,
      "grad_norm": 21.1574764251709,
      "learning_rate": 6.227605189656135e-06,
      "loss": 0.8682,
      "step": 6114
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.209662437438965,
      "learning_rate": 6.226479386751727e-06,
      "loss": 1.0501,
      "step": 6115
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.5337166786193848,
      "learning_rate": 6.225353517683398e-06,
      "loss": 0.1169,
      "step": 6116
    },
    {
      "epoch": 0.44,
      "grad_norm": 18.519317626953125,
      "learning_rate": 6.224227582511884e-06,
      "loss": 0.8905,
      "step": 6117
    },
    {
      "epoch": 0.44,
      "grad_norm": 15.069014549255371,
      "learning_rate": 6.223101581297928e-06,
      "loss": 1.1169,
      "step": 6118
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.8999435901641846,
      "learning_rate": 6.22197551410227e-06,
      "loss": 0.2349,
      "step": 6119
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.052816390991211,
      "learning_rate": 6.220849380985658e-06,
      "loss": 0.937,
      "step": 6120
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.526472330093384,
      "learning_rate": 6.219723182008844e-06,
      "loss": 0.2212,
      "step": 6121
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.8489813804626465,
      "learning_rate": 6.2185969172325815e-06,
      "loss": 0.2544,
      "step": 6122
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.71851921081543,
      "learning_rate": 6.2174705867176275e-06,
      "loss": 0.4798,
      "step": 6123
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.09753704071045,
      "learning_rate": 6.216344190524743e-06,
      "loss": 0.5497,
      "step": 6124
    },
    {
      "epoch": 0.44,
      "grad_norm": 12.951369285583496,
      "learning_rate": 6.215217728714695e-06,
      "loss": 0.8671,
      "step": 6125
    },
    {
      "epoch": 0.44,
      "grad_norm": 11.819787979125977,
      "learning_rate": 6.214091201348249e-06,
      "loss": 0.9016,
      "step": 6126
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.48237681388855,
      "learning_rate": 6.212964608486179e-06,
      "loss": 0.1333,
      "step": 6127
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.783626556396484,
      "learning_rate": 6.211837950189259e-06,
      "loss": 0.8034,
      "step": 6128
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.9150612354278564,
      "learning_rate": 6.21071122651827e-06,
      "loss": 0.8771,
      "step": 6129
    },
    {
      "epoch": 0.44,
      "grad_norm": 12.355501174926758,
      "learning_rate": 6.209584437533992e-06,
      "loss": 0.7257,
      "step": 6130
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.203105449676514,
      "learning_rate": 6.208457583297213e-06,
      "loss": 0.878,
      "step": 6131
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.7593870162963867,
      "learning_rate": 6.207330663868721e-06,
      "loss": 0.1586,
      "step": 6132
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.8168768882751465,
      "learning_rate": 6.20620367930931e-06,
      "loss": 0.7506,
      "step": 6133
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.130147695541382,
      "learning_rate": 6.205076629679775e-06,
      "loss": 0.2184,
      "step": 6134
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.743783950805664,
      "learning_rate": 6.203949515040919e-06,
      "loss": 0.1592,
      "step": 6135
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.0527143478393555,
      "learning_rate": 6.2028223354535425e-06,
      "loss": 0.9265,
      "step": 6136
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.0787274837493896,
      "learning_rate": 6.201695090978454e-06,
      "loss": 0.2042,
      "step": 6137
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.7486162185668945,
      "learning_rate": 6.200567781676465e-06,
      "loss": 0.2022,
      "step": 6138
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.378349304199219,
      "learning_rate": 6.199440407608388e-06,
      "loss": 0.7989,
      "step": 6139
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.673142910003662,
      "learning_rate": 6.198312968835041e-06,
      "loss": 0.6779,
      "step": 6140
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0608839988708496,
      "learning_rate": 6.197185465417246e-06,
      "loss": 0.0832,
      "step": 6141
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3678574562072754,
      "learning_rate": 6.196057897415826e-06,
      "loss": 0.0762,
      "step": 6142
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.070466041564941,
      "learning_rate": 6.19493026489161e-06,
      "loss": 0.1688,
      "step": 6143
    },
    {
      "epoch": 0.44,
      "grad_norm": 13.627692222595215,
      "learning_rate": 6.1938025679054295e-06,
      "loss": 1.2301,
      "step": 6144
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.6019506454467773,
      "learning_rate": 6.1926748065181185e-06,
      "loss": 0.1627,
      "step": 6145
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.692070484161377,
      "learning_rate": 6.191546980790517e-06,
      "loss": 0.1858,
      "step": 6146
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.566851615905762,
      "learning_rate": 6.190419090783466e-06,
      "loss": 0.6499,
      "step": 6147
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.6867048740386963,
      "learning_rate": 6.18929113655781e-06,
      "loss": 0.1088,
      "step": 6148
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0003151893615723,
      "learning_rate": 6.188163118174402e-06,
      "loss": 0.0678,
      "step": 6149
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.3176093101501465,
      "learning_rate": 6.187035035694088e-06,
      "loss": 1.0242,
      "step": 6150
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.931370735168457,
      "learning_rate": 6.185906889177727e-06,
      "loss": 0.7273,
      "step": 6151
    },
    {
      "epoch": 0.44,
      "grad_norm": 12.021435737609863,
      "learning_rate": 6.184778678686179e-06,
      "loss": 0.835,
      "step": 6152
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.987791538238525,
      "learning_rate": 6.183650404280306e-06,
      "loss": 0.6386,
      "step": 6153
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.020184516906738,
      "learning_rate": 6.182522066020974e-06,
      "loss": 0.9041,
      "step": 6154
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.093376636505127,
      "learning_rate": 6.181393663969054e-06,
      "loss": 0.6041,
      "step": 6155
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.266923427581787,
      "learning_rate": 6.180265198185416e-06,
      "loss": 0.8863,
      "step": 6156
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.726644277572632,
      "learning_rate": 6.179136668730938e-06,
      "loss": 0.8027,
      "step": 6157
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.0721817016601562,
      "learning_rate": 6.178008075666502e-06,
      "loss": 0.1113,
      "step": 6158
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4808497428894043,
      "learning_rate": 6.176879419052987e-06,
      "loss": 0.2716,
      "step": 6159
    },
    {
      "epoch": 0.44,
      "grad_norm": 11.980377197265625,
      "learning_rate": 6.175750698951283e-06,
      "loss": 0.7166,
      "step": 6160
    },
    {
      "epoch": 0.44,
      "grad_norm": 23.924631118774414,
      "learning_rate": 6.17462191542228e-06,
      "loss": 0.8648,
      "step": 6161
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.96017074584961,
      "learning_rate": 6.173493068526869e-06,
      "loss": 1.2548,
      "step": 6162
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.7697927951812744,
      "learning_rate": 6.172364158325951e-06,
      "loss": 0.1474,
      "step": 6163
    },
    {
      "epoch": 0.44,
      "grad_norm": 17.759910583496094,
      "learning_rate": 6.171235184880422e-06,
      "loss": 0.5327,
      "step": 6164
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9968525171279907,
      "learning_rate": 6.17010614825119e-06,
      "loss": 0.0908,
      "step": 6165
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4417312145233154,
      "learning_rate": 6.1689770484991585e-06,
      "loss": 0.1078,
      "step": 6166
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4893479347229004,
      "learning_rate": 6.167847885685242e-06,
      "loss": 0.088,
      "step": 6167
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.438100576400757,
      "learning_rate": 6.166718659870352e-06,
      "loss": 0.2256,
      "step": 6168
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.379227638244629,
      "learning_rate": 6.165589371115407e-06,
      "loss": 0.6534,
      "step": 6169
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.7059459686279297,
      "learning_rate": 6.164460019481326e-06,
      "loss": 0.1078,
      "step": 6170
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.177149772644043,
      "learning_rate": 6.1633306050290345e-06,
      "loss": 0.9743,
      "step": 6171
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4294769763946533,
      "learning_rate": 6.1622011278194625e-06,
      "loss": 0.1245,
      "step": 6172
    },
    {
      "epoch": 0.44,
      "grad_norm": 29.4661808013916,
      "learning_rate": 6.1610715879135375e-06,
      "loss": 1.418,
      "step": 6173
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.5694117546081543,
      "learning_rate": 6.159941985372195e-06,
      "loss": 0.1611,
      "step": 6174
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.6444692611694336,
      "learning_rate": 6.158812320256374e-06,
      "loss": 0.2227,
      "step": 6175
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.5240538120269775,
      "learning_rate": 6.1576825926270135e-06,
      "loss": 0.7159,
      "step": 6176
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.309534072875977,
      "learning_rate": 6.156552802545061e-06,
      "loss": 0.0855,
      "step": 6177
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.8138346672058105,
      "learning_rate": 6.15542295007146e-06,
      "loss": 0.1681,
      "step": 6178
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.9888393878936768,
      "learning_rate": 6.154293035267167e-06,
      "loss": 0.1468,
      "step": 6179
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.341657638549805,
      "learning_rate": 6.153163058193131e-06,
      "loss": 0.7268,
      "step": 6180
    },
    {
      "epoch": 0.44,
      "grad_norm": 12.564131736755371,
      "learning_rate": 6.152033018910316e-06,
      "loss": 0.7592,
      "step": 6181
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.30715012550354,
      "learning_rate": 6.150902917479678e-06,
      "loss": 0.1457,
      "step": 6182
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.1222095489501953,
      "learning_rate": 6.149772753962185e-06,
      "loss": 0.1896,
      "step": 6183
    },
    {
      "epoch": 0.44,
      "grad_norm": 15.161857604980469,
      "learning_rate": 6.148642528418804e-06,
      "loss": 1.1783,
      "step": 6184
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.9686121940612793,
      "learning_rate": 6.147512240910507e-06,
      "loss": 0.919,
      "step": 6185
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.9282588958740234,
      "learning_rate": 6.146381891498269e-06,
      "loss": 0.1474,
      "step": 6186
    },
    {
      "epoch": 0.44,
      "grad_norm": 34.92990493774414,
      "learning_rate": 6.145251480243064e-06,
      "loss": 1.2859,
      "step": 6187
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3658337593078613,
      "learning_rate": 6.144121007205879e-06,
      "loss": 0.1224,
      "step": 6188
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.7679693698883057,
      "learning_rate": 6.142990472447696e-06,
      "loss": 0.1403,
      "step": 6189
    },
    {
      "epoch": 0.44,
      "grad_norm": 22.15004539489746,
      "learning_rate": 6.141859876029503e-06,
      "loss": 0.9979,
      "step": 6190
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.9882187843322754,
      "learning_rate": 6.140729218012292e-06,
      "loss": 0.1177,
      "step": 6191
    },
    {
      "epoch": 0.44,
      "grad_norm": 18.402944564819336,
      "learning_rate": 6.139598498457059e-06,
      "loss": 1.0114,
      "step": 6192
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.452960014343262,
      "learning_rate": 6.138467717424798e-06,
      "loss": 0.6827,
      "step": 6193
    },
    {
      "epoch": 0.44,
      "grad_norm": 14.297884941101074,
      "learning_rate": 6.137336874976515e-06,
      "loss": 1.3124,
      "step": 6194
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.05166244506836,
      "learning_rate": 6.136205971173213e-06,
      "loss": 0.7422,
      "step": 6195
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.1954827308654785,
      "learning_rate": 6.135075006075898e-06,
      "loss": 0.8583,
      "step": 6196
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.8039872646331787,
      "learning_rate": 6.133943979745583e-06,
      "loss": 0.2096,
      "step": 6197
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.328752517700195,
      "learning_rate": 6.132812892243285e-06,
      "loss": 0.9298,
      "step": 6198
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.0085883140563965,
      "learning_rate": 6.131681743630017e-06,
      "loss": 0.8339,
      "step": 6199
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.233104705810547,
      "learning_rate": 6.130550533966804e-06,
      "loss": 0.9124,
      "step": 6200
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2395761013031006,
      "learning_rate": 6.129419263314666e-06,
      "loss": 0.122,
      "step": 6201
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.5968198776245117,
      "learning_rate": 6.128287931734636e-06,
      "loss": 0.7805,
      "step": 6202
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.490518569946289,
      "learning_rate": 6.127156539287744e-06,
      "loss": 0.2003,
      "step": 6203
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.6661622524261475,
      "learning_rate": 6.126025086035022e-06,
      "loss": 0.1234,
      "step": 6204
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4967174530029297,
      "learning_rate": 6.124893572037509e-06,
      "loss": 0.1775,
      "step": 6205
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.599807262420654,
      "learning_rate": 6.123761997356246e-06,
      "loss": 0.1887,
      "step": 6206
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.056016206741333,
      "learning_rate": 6.122630362052276e-06,
      "loss": 0.206,
      "step": 6207
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1626081466674805,
      "learning_rate": 6.121498666186647e-06,
      "loss": 0.0686,
      "step": 6208
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.920395612716675,
      "learning_rate": 6.120366909820413e-06,
      "loss": 0.13,
      "step": 6209
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2606699466705322,
      "learning_rate": 6.119235093014623e-06,
      "loss": 0.0784,
      "step": 6210
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.361499786376953,
      "learning_rate": 6.118103215830336e-06,
      "loss": 0.9115,
      "step": 6211
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.20522689819336,
      "learning_rate": 6.116971278328613e-06,
      "loss": 1.1446,
      "step": 6212
    },
    {
      "epoch": 0.45,
      "grad_norm": 16.61461639404297,
      "learning_rate": 6.115839280570517e-06,
      "loss": 1.0882,
      "step": 6213
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2946953773498535,
      "learning_rate": 6.114707222617117e-06,
      "loss": 0.0804,
      "step": 6214
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.395822525024414,
      "learning_rate": 6.113575104529481e-06,
      "loss": 0.3089,
      "step": 6215
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.936414957046509,
      "learning_rate": 6.112442926368682e-06,
      "loss": 0.1628,
      "step": 6216
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.1960601806640625,
      "learning_rate": 6.111310688195799e-06,
      "loss": 0.2243,
      "step": 6217
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.333909511566162,
      "learning_rate": 6.1101783900719115e-06,
      "loss": 0.108,
      "step": 6218
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.420109272003174,
      "learning_rate": 6.1090460320581e-06,
      "loss": 0.9062,
      "step": 6219
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6060678958892822,
      "learning_rate": 6.107913614215455e-06,
      "loss": 0.0521,
      "step": 6220
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.635342597961426,
      "learning_rate": 6.106781136605064e-06,
      "loss": 0.1878,
      "step": 6221
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.899888277053833,
      "learning_rate": 6.10564859928802e-06,
      "loss": 0.2035,
      "step": 6222
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.7658252716064453,
      "learning_rate": 6.104516002325419e-06,
      "loss": 0.8569,
      "step": 6223
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.633495569229126,
      "learning_rate": 6.1033833457783605e-06,
      "loss": 0.1087,
      "step": 6224
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3070647716522217,
      "learning_rate": 6.102250629707946e-06,
      "loss": 0.0987,
      "step": 6225
    },
    {
      "epoch": 0.45,
      "grad_norm": 15.592764854431152,
      "learning_rate": 6.101117854175285e-06,
      "loss": 1.4543,
      "step": 6226
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.030240535736084,
      "learning_rate": 6.0999850192414815e-06,
      "loss": 0.1506,
      "step": 6227
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.038942337036133,
      "learning_rate": 6.09885212496765e-06,
      "loss": 1.4878,
      "step": 6228
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.38846492767334,
      "learning_rate": 6.097719171414907e-06,
      "loss": 0.1221,
      "step": 6229
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2736098766326904,
      "learning_rate": 6.096586158644368e-06,
      "loss": 0.1333,
      "step": 6230
    },
    {
      "epoch": 0.45,
      "grad_norm": 12.54434585571289,
      "learning_rate": 6.095453086717158e-06,
      "loss": 1.1073,
      "step": 6231
    },
    {
      "epoch": 0.45,
      "grad_norm": 17.252901077270508,
      "learning_rate": 6.094319955694399e-06,
      "loss": 0.5733,
      "step": 6232
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.808116912841797,
      "learning_rate": 6.093186765637223e-06,
      "loss": 0.6834,
      "step": 6233
    },
    {
      "epoch": 0.45,
      "grad_norm": 15.38717269897461,
      "learning_rate": 6.092053516606756e-06,
      "loss": 1.1665,
      "step": 6234
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.691502094268799,
      "learning_rate": 6.090920208664137e-06,
      "loss": 0.2106,
      "step": 6235
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.947966575622559,
      "learning_rate": 6.089786841870502e-06,
      "loss": 0.5118,
      "step": 6236
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.511131286621094,
      "learning_rate": 6.088653416286993e-06,
      "loss": 0.9185,
      "step": 6237
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.392859697341919,
      "learning_rate": 6.087519931974751e-06,
      "loss": 0.1073,
      "step": 6238
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.430557250976562,
      "learning_rate": 6.0863863889949275e-06,
      "loss": 1.1219,
      "step": 6239
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.468937873840332,
      "learning_rate": 6.085252787408669e-06,
      "loss": 0.9363,
      "step": 6240
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.590117454528809,
      "learning_rate": 6.084119127277131e-06,
      "loss": 0.7053,
      "step": 6241
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.397647380828857,
      "learning_rate": 6.082985408661468e-06,
      "loss": 0.5881,
      "step": 6242
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.969064712524414,
      "learning_rate": 6.0818516316228425e-06,
      "loss": 0.436,
      "step": 6243
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.8693575859069824,
      "learning_rate": 6.080717796222417e-06,
      "loss": 0.1295,
      "step": 6244
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.1250345706939697,
      "learning_rate": 6.079583902521358e-06,
      "loss": 0.2013,
      "step": 6245
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.344456195831299,
      "learning_rate": 6.078449950580833e-06,
      "loss": 0.7937,
      "step": 6246
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.9990921020507812,
      "learning_rate": 6.077315940462014e-06,
      "loss": 0.1706,
      "step": 6247
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.9431564807891846,
      "learning_rate": 6.076181872226079e-06,
      "loss": 0.1756,
      "step": 6248
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.259123802185059,
      "learning_rate": 6.075047745934206e-06,
      "loss": 1.2081,
      "step": 6249
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3233330249786377,
      "learning_rate": 6.073913561647575e-06,
      "loss": 0.0921,
      "step": 6250
    },
    {
      "epoch": 0.45,
      "grad_norm": 17.749692916870117,
      "learning_rate": 6.0727793194273744e-06,
      "loss": 0.8839,
      "step": 6251
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.883035182952881,
      "learning_rate": 6.07164501933479e-06,
      "loss": 0.7943,
      "step": 6252
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.371185064315796,
      "learning_rate": 6.070510661431012e-06,
      "loss": 0.1057,
      "step": 6253
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.927057266235352,
      "learning_rate": 6.0693762457772366e-06,
      "loss": 0.9753,
      "step": 6254
    },
    {
      "epoch": 0.45,
      "grad_norm": 15.894341468811035,
      "learning_rate": 6.06824177243466e-06,
      "loss": 1.2343,
      "step": 6255
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.738042116165161,
      "learning_rate": 6.067107241464483e-06,
      "loss": 0.0831,
      "step": 6256
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.3387956619262695,
      "learning_rate": 6.065972652927911e-06,
      "loss": 0.4959,
      "step": 6257
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.520453453063965,
      "learning_rate": 6.064838006886148e-06,
      "loss": 0.9748,
      "step": 6258
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.415308952331543,
      "learning_rate": 6.063703303400406e-06,
      "loss": 0.9105,
      "step": 6259
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.16485333442688,
      "learning_rate": 6.062568542531896e-06,
      "loss": 0.179,
      "step": 6260
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.038450241088867,
      "learning_rate": 6.0614337243418354e-06,
      "loss": 0.7821,
      "step": 6261
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.9609007835388184,
      "learning_rate": 6.060298848891442e-06,
      "loss": 0.1721,
      "step": 6262
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0106632709503174,
      "learning_rate": 6.0591639162419434e-06,
      "loss": 0.123,
      "step": 6263
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.129069328308105,
      "learning_rate": 6.058028926454558e-06,
      "loss": 0.3894,
      "step": 6264
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.298105478286743,
      "learning_rate": 6.056893879590515e-06,
      "loss": 0.1215,
      "step": 6265
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.150641918182373,
      "learning_rate": 6.055758775711052e-06,
      "loss": 0.6995,
      "step": 6266
    },
    {
      "epoch": 0.45,
      "grad_norm": 12.331208229064941,
      "learning_rate": 6.054623614877395e-06,
      "loss": 0.8963,
      "step": 6267
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.619361400604248,
      "learning_rate": 6.053488397150789e-06,
      "loss": 0.1694,
      "step": 6268
    },
    {
      "epoch": 0.45,
      "grad_norm": 15.864991188049316,
      "learning_rate": 6.052353122592471e-06,
      "loss": 0.8263,
      "step": 6269
    },
    {
      "epoch": 0.45,
      "grad_norm": 14.417516708374023,
      "learning_rate": 6.0512177912636835e-06,
      "loss": 0.9273,
      "step": 6270
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.032243251800537,
      "learning_rate": 6.050082403225678e-06,
      "loss": 0.4725,
      "step": 6271
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.5625195503234863,
      "learning_rate": 6.048946958539699e-06,
      "loss": 0.1948,
      "step": 6272
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.1545162200927734,
      "learning_rate": 6.047811457267003e-06,
      "loss": 0.1648,
      "step": 6273
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.894605875015259,
      "learning_rate": 6.0466758994688454e-06,
      "loss": 0.2229,
      "step": 6274
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.415135383605957,
      "learning_rate": 6.045540285206486e-06,
      "loss": 0.8382,
      "step": 6275
    },
    {
      "epoch": 0.45,
      "grad_norm": 16.3867130279541,
      "learning_rate": 6.044404614541184e-06,
      "loss": 0.7076,
      "step": 6276
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.4199509620666504,
      "learning_rate": 6.043268887534208e-06,
      "loss": 0.7308,
      "step": 6277
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.669649600982666,
      "learning_rate": 6.042133104246822e-06,
      "loss": 1.0238,
      "step": 6278
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4973387718200684,
      "learning_rate": 6.040997264740301e-06,
      "loss": 0.1616,
      "step": 6279
    },
    {
      "epoch": 0.45,
      "grad_norm": 17.0791015625,
      "learning_rate": 6.039861369075919e-06,
      "loss": 0.4242,
      "step": 6280
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.603030204772949,
      "learning_rate": 6.038725417314952e-06,
      "loss": 0.8214,
      "step": 6281
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.677031993865967,
      "learning_rate": 6.037589409518679e-06,
      "loss": 0.2714,
      "step": 6282
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.227740287780762,
      "learning_rate": 6.036453345748387e-06,
      "loss": 0.7383,
      "step": 6283
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.82598876953125,
      "learning_rate": 6.035317226065358e-06,
      "loss": 0.832,
      "step": 6284
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.1259284019470215,
      "learning_rate": 6.034181050530885e-06,
      "loss": 0.1667,
      "step": 6285
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3400142192840576,
      "learning_rate": 6.033044819206258e-06,
      "loss": 0.1187,
      "step": 6286
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.515731811523438,
      "learning_rate": 6.031908532152775e-06,
      "loss": 0.7841,
      "step": 6287
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.31264066696167,
      "learning_rate": 6.030772189431731e-06,
      "loss": 0.8513,
      "step": 6288
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5979056358337402,
      "learning_rate": 6.029635791104429e-06,
      "loss": 0.1496,
      "step": 6289
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.645097732543945,
      "learning_rate": 6.028499337232175e-06,
      "loss": 0.4805,
      "step": 6290
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4157731533050537,
      "learning_rate": 6.027362827876273e-06,
      "loss": 0.1215,
      "step": 6291
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.991357803344727,
      "learning_rate": 6.026226263098035e-06,
      "loss": 1.3468,
      "step": 6292
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0326905250549316,
      "learning_rate": 6.0250896429587765e-06,
      "loss": 0.1576,
      "step": 6293
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.5231754779815674,
      "learning_rate": 6.023952967519812e-06,
      "loss": 0.2153,
      "step": 6294
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.8833060264587402,
      "learning_rate": 6.022816236842459e-06,
      "loss": 0.6847,
      "step": 6295
    },
    {
      "epoch": 0.45,
      "grad_norm": 22.52227210998535,
      "learning_rate": 6.021679450988042e-06,
      "loss": 0.9885,
      "step": 6296
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3276374340057373,
      "learning_rate": 6.020542610017886e-06,
      "loss": 0.0717,
      "step": 6297
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.987105369567871,
      "learning_rate": 6.019405713993319e-06,
      "loss": 0.4632,
      "step": 6298
    },
    {
      "epoch": 0.45,
      "grad_norm": 17.72905921936035,
      "learning_rate": 6.018268762975674e-06,
      "loss": 0.3612,
      "step": 6299
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2614972591400146,
      "learning_rate": 6.017131757026282e-06,
      "loss": 0.1009,
      "step": 6300
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8683818578720093,
      "learning_rate": 6.015994696206482e-06,
      "loss": 0.1007,
      "step": 6301
    },
    {
      "epoch": 0.45,
      "grad_norm": 14.053206443786621,
      "learning_rate": 6.0148575805776134e-06,
      "loss": 1.2879,
      "step": 6302
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.9057681560516357,
      "learning_rate": 6.013720410201022e-06,
      "loss": 0.8966,
      "step": 6303
    },
    {
      "epoch": 0.45,
      "grad_norm": 14.340136528015137,
      "learning_rate": 6.012583185138049e-06,
      "loss": 1.2171,
      "step": 6304
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.3494009971618652,
      "learning_rate": 6.011445905450047e-06,
      "loss": 0.163,
      "step": 6305
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0326218605041504,
      "learning_rate": 6.010308571198368e-06,
      "loss": 0.1355,
      "step": 6306
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.9855921268463135,
      "learning_rate": 6.009171182444364e-06,
      "loss": 0.9296,
      "step": 6307
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.362120628356934,
      "learning_rate": 6.008033739249396e-06,
      "loss": 0.682,
      "step": 6308
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.22637939453125,
      "learning_rate": 6.006896241674822e-06,
      "loss": 0.773,
      "step": 6309
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.6175122261047363,
      "learning_rate": 6.005758689782008e-06,
      "loss": 0.1333,
      "step": 6310
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.293973922729492,
      "learning_rate": 6.004621083632319e-06,
      "loss": 0.0935,
      "step": 6311
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.080155849456787,
      "learning_rate": 6.003483423287126e-06,
      "loss": 0.3119,
      "step": 6312
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.786722183227539,
      "learning_rate": 6.0023457088078e-06,
      "loss": 1.0569,
      "step": 6313
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.8214948177337646,
      "learning_rate": 6.001207940255719e-06,
      "loss": 0.1151,
      "step": 6314
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.8490943908691406,
      "learning_rate": 6.0000701176922574e-06,
      "loss": 0.09,
      "step": 6315
    },
    {
      "epoch": 0.45,
      "grad_norm": 23.898771286010742,
      "learning_rate": 5.998932241178798e-06,
      "loss": 0.6069,
      "step": 6316
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.327894687652588,
      "learning_rate": 5.997794310776727e-06,
      "loss": 0.1103,
      "step": 6317
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.1036088466644287,
      "learning_rate": 5.99665632654743e-06,
      "loss": 0.2561,
      "step": 6318
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.704172134399414,
      "learning_rate": 5.995518288552294e-06,
      "loss": 0.5129,
      "step": 6319
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.133648872375488,
      "learning_rate": 5.9943801968527185e-06,
      "loss": 1.1249,
      "step": 6320
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.06058406829834,
      "learning_rate": 5.993242051510092e-06,
      "loss": 0.7918,
      "step": 6321
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.066490888595581,
      "learning_rate": 5.992103852585818e-06,
      "loss": 0.1667,
      "step": 6322
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.307752847671509,
      "learning_rate": 5.990965600141297e-06,
      "loss": 0.1062,
      "step": 6323
    },
    {
      "epoch": 0.45,
      "grad_norm": 14.258599281311035,
      "learning_rate": 5.989827294237931e-06,
      "loss": 1.6367,
      "step": 6324
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0599191188812256,
      "learning_rate": 5.98868893493713e-06,
      "loss": 0.0903,
      "step": 6325
    },
    {
      "epoch": 0.45,
      "grad_norm": 12.96230411529541,
      "learning_rate": 5.987550522300304e-06,
      "loss": 1.5366,
      "step": 6326
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.794061660766602,
      "learning_rate": 5.986412056388865e-06,
      "loss": 0.4344,
      "step": 6327
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9393820762634277,
      "learning_rate": 5.985273537264229e-06,
      "loss": 0.1444,
      "step": 6328
    },
    {
      "epoch": 0.45,
      "grad_norm": 31.71111297607422,
      "learning_rate": 5.984134964987815e-06,
      "loss": 1.3816,
      "step": 6329
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.087750434875488,
      "learning_rate": 5.982996339621047e-06,
      "loss": 0.7625,
      "step": 6330
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5448217391967773,
      "learning_rate": 5.981857661225345e-06,
      "loss": 0.1274,
      "step": 6331
    },
    {
      "epoch": 0.45,
      "grad_norm": 17.16866683959961,
      "learning_rate": 5.9807189298621395e-06,
      "loss": 1.2009,
      "step": 6332
    },
    {
      "epoch": 0.45,
      "grad_norm": 14.27739429473877,
      "learning_rate": 5.9795801455928584e-06,
      "loss": 0.5389,
      "step": 6333
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.476618766784668,
      "learning_rate": 5.978441308478938e-06,
      "loss": 0.078,
      "step": 6334
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.08779764175415,
      "learning_rate": 5.977302418581812e-06,
      "loss": 0.9283,
      "step": 6335
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.222419500350952,
      "learning_rate": 5.976163475962921e-06,
      "loss": 0.089,
      "step": 6336
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.3428239822387695,
      "learning_rate": 5.975024480683704e-06,
      "loss": 0.1004,
      "step": 6337
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.12327766418457,
      "learning_rate": 5.973885432805607e-06,
      "loss": 0.2979,
      "step": 6338
    },
    {
      "epoch": 0.45,
      "grad_norm": 25.20401954650879,
      "learning_rate": 5.972746332390078e-06,
      "loss": 0.4728,
      "step": 6339
    },
    {
      "epoch": 0.45,
      "grad_norm": 17.69309425354004,
      "learning_rate": 5.971607179498567e-06,
      "loss": 0.9395,
      "step": 6340
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.354823112487793,
      "learning_rate": 5.970467974192525e-06,
      "loss": 0.7859,
      "step": 6341
    },
    {
      "epoch": 0.45,
      "grad_norm": 22.4034423828125,
      "learning_rate": 5.96932871653341e-06,
      "loss": 0.991,
      "step": 6342
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0511434078216553,
      "learning_rate": 5.96818940658268e-06,
      "loss": 0.0807,
      "step": 6343
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.504488706588745,
      "learning_rate": 5.967050044401796e-06,
      "loss": 0.7042,
      "step": 6344
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1713147163391113,
      "learning_rate": 5.965910630052223e-06,
      "loss": 0.1031,
      "step": 6345
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.897001028060913,
      "learning_rate": 5.964771163595427e-06,
      "loss": 0.1632,
      "step": 6346
    },
    {
      "epoch": 0.46,
      "grad_norm": 16.15854835510254,
      "learning_rate": 5.9636316450928775e-06,
      "loss": 1.6309,
      "step": 6347
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.918750762939453,
      "learning_rate": 5.962492074606051e-06,
      "loss": 0.9353,
      "step": 6348
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.890237808227539,
      "learning_rate": 5.961352452196419e-06,
      "loss": 0.6381,
      "step": 6349
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.443538665771484,
      "learning_rate": 5.960212777925459e-06,
      "loss": 0.8379,
      "step": 6350
    },
    {
      "epoch": 0.46,
      "grad_norm": 23.491256713867188,
      "learning_rate": 5.959073051854658e-06,
      "loss": 1.445,
      "step": 6351
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5652098655700684,
      "learning_rate": 5.957933274045493e-06,
      "loss": 0.1323,
      "step": 6352
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.042832374572754,
      "learning_rate": 5.956793444559454e-06,
      "loss": 0.9469,
      "step": 6353
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.037773132324219,
      "learning_rate": 5.955653563458034e-06,
      "loss": 0.9597,
      "step": 6354
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1273865699768066,
      "learning_rate": 5.954513630802719e-06,
      "loss": 0.1845,
      "step": 6355
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.0734102725982666,
      "learning_rate": 5.9533736466550054e-06,
      "loss": 0.2131,
      "step": 6356
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.91789722442627,
      "learning_rate": 5.952233611076394e-06,
      "loss": 0.7362,
      "step": 6357
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.758790969848633,
      "learning_rate": 5.9510935241283825e-06,
      "loss": 0.7349,
      "step": 6358
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.62969970703125,
      "learning_rate": 5.949953385872478e-06,
      "loss": 0.5771,
      "step": 6359
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.370732307434082,
      "learning_rate": 5.948813196370182e-06,
      "loss": 0.0897,
      "step": 6360
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.730862617492676,
      "learning_rate": 5.947672955683006e-06,
      "loss": 0.7977,
      "step": 6361
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.5675313472747803,
      "learning_rate": 5.946532663872462e-06,
      "loss": 0.1857,
      "step": 6362
    },
    {
      "epoch": 0.46,
      "grad_norm": 14.53116226196289,
      "learning_rate": 5.945392321000062e-06,
      "loss": 1.5862,
      "step": 6363
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.734601020812988,
      "learning_rate": 5.944251927127325e-06,
      "loss": 0.8124,
      "step": 6364
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.7456047534942627,
      "learning_rate": 5.943111482315771e-06,
      "loss": 0.0962,
      "step": 6365
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.105919122695923,
      "learning_rate": 5.941970986626923e-06,
      "loss": 0.1631,
      "step": 6366
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.745338439941406,
      "learning_rate": 5.940830440122305e-06,
      "loss": 0.5619,
      "step": 6367
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.145332336425781,
      "learning_rate": 5.939689842863448e-06,
      "loss": 1.1532,
      "step": 6368
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.085855484008789,
      "learning_rate": 5.938549194911878e-06,
      "loss": 1.1356,
      "step": 6369
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.729783535003662,
      "learning_rate": 5.937408496329132e-06,
      "loss": 0.0896,
      "step": 6370
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.8979499340057373,
      "learning_rate": 5.936267747176747e-06,
      "loss": 0.1722,
      "step": 6371
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.751089096069336,
      "learning_rate": 5.935126947516262e-06,
      "loss": 0.6122,
      "step": 6372
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7987468242645264,
      "learning_rate": 5.933986097409215e-06,
      "loss": 0.0533,
      "step": 6373
    },
    {
      "epoch": 0.46,
      "grad_norm": 24.532743453979492,
      "learning_rate": 5.9328451969171565e-06,
      "loss": 0.801,
      "step": 6374
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.188304424285889,
      "learning_rate": 5.9317042461016285e-06,
      "loss": 0.8592,
      "step": 6375
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.484183311462402,
      "learning_rate": 5.930563245024184e-06,
      "loss": 1.1302,
      "step": 6376
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.984833717346191,
      "learning_rate": 5.929422193746374e-06,
      "loss": 0.7273,
      "step": 6377
    },
    {
      "epoch": 0.46,
      "grad_norm": 18.200077056884766,
      "learning_rate": 5.928281092329755e-06,
      "loss": 1.0205,
      "step": 6378
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3505401611328125,
      "learning_rate": 5.927139940835885e-06,
      "loss": 0.0705,
      "step": 6379
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.041687488555908,
      "learning_rate": 5.925998739326325e-06,
      "loss": 0.1176,
      "step": 6380
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7467050552368164,
      "learning_rate": 5.924857487862637e-06,
      "loss": 0.0576,
      "step": 6381
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1830320358276367,
      "learning_rate": 5.92371618650639e-06,
      "loss": 0.1564,
      "step": 6382
    },
    {
      "epoch": 0.46,
      "grad_norm": 11.388071060180664,
      "learning_rate": 5.9225748353191495e-06,
      "loss": 1.0063,
      "step": 6383
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.7427384853363037,
      "learning_rate": 5.92143343436249e-06,
      "loss": 0.1582,
      "step": 6384
    },
    {
      "epoch": 0.46,
      "grad_norm": 29.537837982177734,
      "learning_rate": 5.920291983697984e-06,
      "loss": 1.2704,
      "step": 6385
    },
    {
      "epoch": 0.46,
      "grad_norm": 19.81719398498535,
      "learning_rate": 5.919150483387209e-06,
      "loss": 1.3697,
      "step": 6386
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.6557044982910156,
      "learning_rate": 5.918008933491744e-06,
      "loss": 0.1603,
      "step": 6387
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.772865295410156,
      "learning_rate": 5.916867334073173e-06,
      "loss": 0.8491,
      "step": 6388
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.174001932144165,
      "learning_rate": 5.9157256851930785e-06,
      "loss": 0.1345,
      "step": 6389
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.7477548122406006,
      "learning_rate": 5.914583986913052e-06,
      "loss": 0.822,
      "step": 6390
    },
    {
      "epoch": 0.46,
      "grad_norm": 18.639699935913086,
      "learning_rate": 5.91344223929468e-06,
      "loss": 1.0778,
      "step": 6391
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.703538179397583,
      "learning_rate": 5.9123004423995554e-06,
      "loss": 0.166,
      "step": 6392
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.759013652801514,
      "learning_rate": 5.911158596289276e-06,
      "loss": 0.9185,
      "step": 6393
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.075716972351074,
      "learning_rate": 5.9100167010254394e-06,
      "loss": 0.5123,
      "step": 6394
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.900076866149902,
      "learning_rate": 5.908874756669646e-06,
      "loss": 0.7058,
      "step": 6395
    },
    {
      "epoch": 0.46,
      "grad_norm": 15.788897514343262,
      "learning_rate": 5.9077327632835e-06,
      "loss": 0.8348,
      "step": 6396
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.099185943603516,
      "learning_rate": 5.906590720928609e-06,
      "loss": 1.0072,
      "step": 6397
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.9711382389068604,
      "learning_rate": 5.905448629666578e-06,
      "loss": 0.2378,
      "step": 6398
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.08604097366333,
      "learning_rate": 5.904306489559022e-06,
      "loss": 0.2683,
      "step": 6399
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1399741172790527,
      "learning_rate": 5.903164300667554e-06,
      "loss": 0.2471,
      "step": 6400
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.012587070465088,
      "learning_rate": 5.902022063053788e-06,
      "loss": 0.8905,
      "step": 6401
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.8444647789001465,
      "learning_rate": 5.90087977677935e-06,
      "loss": 0.1196,
      "step": 6402
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.721473455429077,
      "learning_rate": 5.899737441905855e-06,
      "loss": 0.1702,
      "step": 6403
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.671096801757812,
      "learning_rate": 5.898595058494932e-06,
      "loss": 1.0551,
      "step": 6404
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.82408618927002,
      "learning_rate": 5.897452626608207e-06,
      "loss": 1.1804,
      "step": 6405
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.451076030731201,
      "learning_rate": 5.8963101463073094e-06,
      "loss": 0.1473,
      "step": 6406
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.00328254699707,
      "learning_rate": 5.895167617653872e-06,
      "loss": 0.9668,
      "step": 6407
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.056882858276367,
      "learning_rate": 5.8940250407095316e-06,
      "loss": 0.7165,
      "step": 6408
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.180215835571289,
      "learning_rate": 5.892882415535922e-06,
      "loss": 0.8955,
      "step": 6409
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.341139793395996,
      "learning_rate": 5.8917397421946865e-06,
      "loss": 0.8058,
      "step": 6410
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5541412830352783,
      "learning_rate": 5.890597020747468e-06,
      "loss": 0.1242,
      "step": 6411
    },
    {
      "epoch": 0.46,
      "grad_norm": 27.066143035888672,
      "learning_rate": 5.88945425125591e-06,
      "loss": 0.5352,
      "step": 6412
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.83959698677063,
      "learning_rate": 5.888311433781665e-06,
      "loss": 0.2264,
      "step": 6413
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4238905906677246,
      "learning_rate": 5.887168568386379e-06,
      "loss": 0.0796,
      "step": 6414
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.9650461673736572,
      "learning_rate": 5.886025655131706e-06,
      "loss": 0.1982,
      "step": 6415
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.1525468826293945,
      "learning_rate": 5.884882694079304e-06,
      "loss": 0.7387,
      "step": 6416
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.125318765640259,
      "learning_rate": 5.883739685290828e-06,
      "loss": 0.0934,
      "step": 6417
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.395465612411499,
      "learning_rate": 5.882596628827941e-06,
      "loss": 0.1281,
      "step": 6418
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.56177043914795,
      "learning_rate": 5.88145352475231e-06,
      "loss": 0.9947,
      "step": 6419
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.736344575881958,
      "learning_rate": 5.880310373125595e-06,
      "loss": 0.193,
      "step": 6420
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4620087146759033,
      "learning_rate": 5.879167174009468e-06,
      "loss": 0.0887,
      "step": 6421
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.76422119140625,
      "learning_rate": 5.878023927465602e-06,
      "loss": 0.8259,
      "step": 6422
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.021551132202148,
      "learning_rate": 5.876880633555665e-06,
      "loss": 0.8506,
      "step": 6423
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.6412768363952637,
      "learning_rate": 5.875737292341338e-06,
      "loss": 0.1649,
      "step": 6424
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9264158010482788,
      "learning_rate": 5.8745939038843e-06,
      "loss": 0.0613,
      "step": 6425
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.63773250579834,
      "learning_rate": 5.8734504682462325e-06,
      "loss": 0.1324,
      "step": 6426
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.04486083984375,
      "learning_rate": 5.872306985488816e-06,
      "loss": 0.7233,
      "step": 6427
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.256588935852051,
      "learning_rate": 5.871163455673742e-06,
      "loss": 0.8234,
      "step": 6428
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.057374000549316,
      "learning_rate": 5.870019878862694e-06,
      "loss": 0.8639,
      "step": 6429
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.046253204345703,
      "learning_rate": 5.86887625511737e-06,
      "loss": 0.8098,
      "step": 6430
    },
    {
      "epoch": 0.46,
      "grad_norm": 11.868216514587402,
      "learning_rate": 5.867732584499458e-06,
      "loss": 0.7447,
      "step": 6431
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.435662746429443,
      "learning_rate": 5.86658886707066e-06,
      "loss": 0.4267,
      "step": 6432
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.459271430969238,
      "learning_rate": 5.86544510289267e-06,
      "loss": 0.8379,
      "step": 6433
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.156914710998535,
      "learning_rate": 5.864301292027196e-06,
      "loss": 0.3814,
      "step": 6434
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.264980316162109,
      "learning_rate": 5.863157434535935e-06,
      "loss": 0.969,
      "step": 6435
    },
    {
      "epoch": 0.46,
      "grad_norm": 11.168984413146973,
      "learning_rate": 5.862013530480601e-06,
      "loss": 0.5816,
      "step": 6436
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.645484685897827,
      "learning_rate": 5.860869579922899e-06,
      "loss": 0.1861,
      "step": 6437
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.279578685760498,
      "learning_rate": 5.859725582924541e-06,
      "loss": 0.084,
      "step": 6438
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7651498317718506,
      "learning_rate": 5.858581539547244e-06,
      "loss": 0.0747,
      "step": 6439
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.382430076599121,
      "learning_rate": 5.857437449852721e-06,
      "loss": 1.0898,
      "step": 6440
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.0154166221618652,
      "learning_rate": 5.856293313902692e-06,
      "loss": 0.1563,
      "step": 6441
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.7726736068725586,
      "learning_rate": 5.855149131758883e-06,
      "loss": 0.1482,
      "step": 6442
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.791109561920166,
      "learning_rate": 5.854004903483013e-06,
      "loss": 0.1788,
      "step": 6443
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.701561450958252,
      "learning_rate": 5.852860629136812e-06,
      "loss": 0.236,
      "step": 6444
    },
    {
      "epoch": 0.46,
      "grad_norm": 14.12303638458252,
      "learning_rate": 5.8517163087820084e-06,
      "loss": 1.0367,
      "step": 6445
    },
    {
      "epoch": 0.46,
      "grad_norm": 20.583744049072266,
      "learning_rate": 5.850571942480333e-06,
      "loss": 0.7127,
      "step": 6446
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.302165985107422,
      "learning_rate": 5.8494275302935215e-06,
      "loss": 0.6913,
      "step": 6447
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4172046184539795,
      "learning_rate": 5.84828307228331e-06,
      "loss": 0.2215,
      "step": 6448
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.913707733154297,
      "learning_rate": 5.847138568511436e-06,
      "loss": 0.6822,
      "step": 6449
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.93477201461792,
      "learning_rate": 5.845994019039645e-06,
      "loss": 0.1972,
      "step": 6450
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5048816204071045,
      "learning_rate": 5.844849423929679e-06,
      "loss": 0.1329,
      "step": 6451
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.870338439941406,
      "learning_rate": 5.843704783243284e-06,
      "loss": 0.8878,
      "step": 6452
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9818388223648071,
      "learning_rate": 5.84256009704221e-06,
      "loss": 0.0602,
      "step": 6453
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.79846715927124,
      "learning_rate": 5.8414153653882065e-06,
      "loss": 0.8552,
      "step": 6454
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.148501873016357,
      "learning_rate": 5.84027058834303e-06,
      "loss": 0.6743,
      "step": 6455
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1938655376434326,
      "learning_rate": 5.839125765968436e-06,
      "loss": 0.1257,
      "step": 6456
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.890353202819824,
      "learning_rate": 5.837980898326183e-06,
      "loss": 1.0172,
      "step": 6457
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.414273738861084,
      "learning_rate": 5.836835985478032e-06,
      "loss": 0.489,
      "step": 6458
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.293478965759277,
      "learning_rate": 5.8356910274857485e-06,
      "loss": 1.0909,
      "step": 6459
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.286898374557495,
      "learning_rate": 5.8345460244110955e-06,
      "loss": 0.1175,
      "step": 6460
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.163439750671387,
      "learning_rate": 5.833400976315844e-06,
      "loss": 0.7596,
      "step": 6461
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.033167600631714,
      "learning_rate": 5.8322558832617656e-06,
      "loss": 0.123,
      "step": 6462
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9110537767410278,
      "learning_rate": 5.831110745310632e-06,
      "loss": 0.0494,
      "step": 6463
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.9595019817352295,
      "learning_rate": 5.829965562524218e-06,
      "loss": 0.1562,
      "step": 6464
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.383620500564575,
      "learning_rate": 5.828820334964306e-06,
      "loss": 0.303,
      "step": 6465
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.33000659942627,
      "learning_rate": 5.827675062692673e-06,
      "loss": 0.7471,
      "step": 6466
    },
    {
      "epoch": 0.46,
      "grad_norm": 29.66350746154785,
      "learning_rate": 5.826529745771103e-06,
      "loss": 0.8279,
      "step": 6467
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.523818016052246,
      "learning_rate": 5.825384384261384e-06,
      "loss": 0.5097,
      "step": 6468
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.497194290161133,
      "learning_rate": 5.824238978225299e-06,
      "loss": 1.0113,
      "step": 6469
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.275249481201172,
      "learning_rate": 5.823093527724641e-06,
      "loss": 0.5435,
      "step": 6470
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.764428973197937,
      "learning_rate": 5.821948032821203e-06,
      "loss": 0.0591,
      "step": 6471
    },
    {
      "epoch": 0.46,
      "grad_norm": 16.2867374420166,
      "learning_rate": 5.82080249357678e-06,
      "loss": 0.6397,
      "step": 6472
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.370691299438477,
      "learning_rate": 5.819656910053171e-06,
      "loss": 0.8097,
      "step": 6473
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.520925045013428,
      "learning_rate": 5.8185112823121735e-06,
      "loss": 0.9733,
      "step": 6474
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.167111873626709,
      "learning_rate": 5.81736561041559e-06,
      "loss": 0.1406,
      "step": 6475
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.087580680847168,
      "learning_rate": 5.816219894425226e-06,
      "loss": 0.107,
      "step": 6476
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.975886344909668,
      "learning_rate": 5.815074134402887e-06,
      "loss": 0.845,
      "step": 6477
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.986618757247925,
      "learning_rate": 5.8139283304103845e-06,
      "loss": 0.9031,
      "step": 6478
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4239933490753174,
      "learning_rate": 5.81278248250953e-06,
      "loss": 0.114,
      "step": 6479
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.078961372375488,
      "learning_rate": 5.811636590762138e-06,
      "loss": 0.4712,
      "step": 6480
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.169988632202148,
      "learning_rate": 5.810490655230022e-06,
      "loss": 1.0629,
      "step": 6481
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.562542200088501,
      "learning_rate": 5.809344675975005e-06,
      "loss": 0.1581,
      "step": 6482
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3583409786224365,
      "learning_rate": 5.808198653058905e-06,
      "loss": 0.1484,
      "step": 6483
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.745686531066895,
      "learning_rate": 5.807052586543548e-06,
      "loss": 1.2497,
      "step": 6484
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.744007110595703,
      "learning_rate": 5.805906476490757e-06,
      "loss": 0.141,
      "step": 6485
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.6269330978393555,
      "learning_rate": 5.804760322962364e-06,
      "loss": 0.2077,
      "step": 6486
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.6734037399292,
      "learning_rate": 5.803614126020196e-06,
      "loss": 1.2598,
      "step": 6487
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.488502502441406,
      "learning_rate": 5.8024678857260896e-06,
      "loss": 0.7973,
      "step": 6488
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.9618914127349854,
      "learning_rate": 5.801321602141876e-06,
      "loss": 0.1028,
      "step": 6489
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.926107406616211,
      "learning_rate": 5.800175275329397e-06,
      "loss": 0.7928,
      "step": 6490
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.41539478302002,
      "learning_rate": 5.79902890535049e-06,
      "loss": 0.8458,
      "step": 6491
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.20759916305542,
      "learning_rate": 5.797882492266999e-06,
      "loss": 0.8995,
      "step": 6492
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.867905616760254,
      "learning_rate": 5.7967360361407665e-06,
      "loss": 1.1821,
      "step": 6493
    },
    {
      "epoch": 0.47,
      "grad_norm": 19.9490966796875,
      "learning_rate": 5.795589537033641e-06,
      "loss": 1.7599,
      "step": 6494
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.659337043762207,
      "learning_rate": 5.794442995007471e-06,
      "loss": 0.7758,
      "step": 6495
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.702011585235596,
      "learning_rate": 5.793296410124109e-06,
      "loss": 0.6199,
      "step": 6496
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6051981449127197,
      "learning_rate": 5.792149782445408e-06,
      "loss": 0.1487,
      "step": 6497
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.048666000366211,
      "learning_rate": 5.791003112033227e-06,
      "loss": 0.9585,
      "step": 6498
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.073250770568848,
      "learning_rate": 5.7898563989494204e-06,
      "loss": 0.6957,
      "step": 6499
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.261427402496338,
      "learning_rate": 5.78870964325585e-06,
      "loss": 0.7097,
      "step": 6500
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.7823486328125,
      "learning_rate": 5.787562845014381e-06,
      "loss": 0.9513,
      "step": 6501
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.06409740447998,
      "learning_rate": 5.786416004286877e-06,
      "loss": 0.5794,
      "step": 6502
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.761638641357422,
      "learning_rate": 5.785269121135206e-06,
      "loss": 0.1735,
      "step": 6503
    },
    {
      "epoch": 0.47,
      "grad_norm": 18.771636962890625,
      "learning_rate": 5.784122195621239e-06,
      "loss": 1.2759,
      "step": 6504
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.970693826675415,
      "learning_rate": 5.782975227806848e-06,
      "loss": 0.8063,
      "step": 6505
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9057081937789917,
      "learning_rate": 5.781828217753906e-06,
      "loss": 0.0698,
      "step": 6506
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.992654323577881,
      "learning_rate": 5.780681165524294e-06,
      "loss": 0.1868,
      "step": 6507
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.1089978218078613,
      "learning_rate": 5.779534071179884e-06,
      "loss": 0.1287,
      "step": 6508
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.7795538902282715,
      "learning_rate": 5.7783869347825626e-06,
      "loss": 0.8518,
      "step": 6509
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.806735515594482,
      "learning_rate": 5.777239756394215e-06,
      "loss": 0.2613,
      "step": 6510
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1898066997528076,
      "learning_rate": 5.776092536076723e-06,
      "loss": 0.073,
      "step": 6511
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.654337882995605,
      "learning_rate": 5.7749452738919754e-06,
      "loss": 0.6715,
      "step": 6512
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.7304697036743164,
      "learning_rate": 5.773797969901864e-06,
      "loss": 0.1089,
      "step": 6513
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3206543922424316,
      "learning_rate": 5.772650624168281e-06,
      "loss": 0.0647,
      "step": 6514
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.973095178604126,
      "learning_rate": 5.771503236753122e-06,
      "loss": 0.9233,
      "step": 6515
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.7406721115112305,
      "learning_rate": 5.770355807718285e-06,
      "loss": 0.3233,
      "step": 6516
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.2053098678588867,
      "learning_rate": 5.769208337125666e-06,
      "loss": 0.2561,
      "step": 6517
    },
    {
      "epoch": 0.47,
      "grad_norm": 13.788408279418945,
      "learning_rate": 5.768060825037169e-06,
      "loss": 0.9622,
      "step": 6518
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.338592052459717,
      "learning_rate": 5.766913271514699e-06,
      "loss": 1.0824,
      "step": 6519
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.83380126953125,
      "learning_rate": 5.76576567662016e-06,
      "loss": 1.4321,
      "step": 6520
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.646730422973633,
      "learning_rate": 5.764618040415462e-06,
      "loss": 0.115,
      "step": 6521
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.7122812271118164,
      "learning_rate": 5.763470362962515e-06,
      "loss": 0.1634,
      "step": 6522
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.7212047576904297,
      "learning_rate": 5.762322644323231e-06,
      "loss": 0.1742,
      "step": 6523
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.5196595191955566,
      "learning_rate": 5.761174884559526e-06,
      "loss": 0.1635,
      "step": 6524
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.9921953678131104,
      "learning_rate": 5.760027083733317e-06,
      "loss": 0.9287,
      "step": 6525
    },
    {
      "epoch": 0.47,
      "grad_norm": 20.973997116088867,
      "learning_rate": 5.758879241906523e-06,
      "loss": 1.0188,
      "step": 6526
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.85032320022583,
      "learning_rate": 5.757731359141069e-06,
      "loss": 0.1852,
      "step": 6527
    },
    {
      "epoch": 0.47,
      "grad_norm": 14.614124298095703,
      "learning_rate": 5.756583435498874e-06,
      "loss": 0.666,
      "step": 6528
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.398686408996582,
      "learning_rate": 5.7554354710418684e-06,
      "loss": 0.1335,
      "step": 6529
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.715936660766602,
      "learning_rate": 5.754287465831977e-06,
      "loss": 1.1884,
      "step": 6530
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.309936761856079,
      "learning_rate": 5.753139419931132e-06,
      "loss": 0.1209,
      "step": 6531
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.3416857719421387,
      "learning_rate": 5.751991333401266e-06,
      "loss": 0.2315,
      "step": 6532
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.190261125564575,
      "learning_rate": 5.7508432063043155e-06,
      "loss": 0.1499,
      "step": 6533
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.223145961761475,
      "learning_rate": 5.749695038702214e-06,
      "loss": 0.701,
      "step": 6534
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.423814296722412,
      "learning_rate": 5.7485468306569045e-06,
      "loss": 0.1401,
      "step": 6535
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.75657320022583,
      "learning_rate": 5.747398582230328e-06,
      "loss": 0.1818,
      "step": 6536
    },
    {
      "epoch": 0.47,
      "grad_norm": 17.612018585205078,
      "learning_rate": 5.7462502934844244e-06,
      "loss": 1.1928,
      "step": 6537
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.92918872833252,
      "learning_rate": 5.745101964481144e-06,
      "loss": 0.8217,
      "step": 6538
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6907402276992798,
      "learning_rate": 5.743953595282432e-06,
      "loss": 0.0705,
      "step": 6539
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.125211477279663,
      "learning_rate": 5.742805185950239e-06,
      "loss": 0.1362,
      "step": 6540
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.27030611038208,
      "learning_rate": 5.741656736546519e-06,
      "loss": 0.9466,
      "step": 6541
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.9018709659576416,
      "learning_rate": 5.740508247133226e-06,
      "loss": 0.1699,
      "step": 6542
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.542557716369629,
      "learning_rate": 5.739359717772314e-06,
      "loss": 0.7442,
      "step": 6543
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.375901937484741,
      "learning_rate": 5.738211148525746e-06,
      "loss": 0.3166,
      "step": 6544
    },
    {
      "epoch": 0.47,
      "grad_norm": 57.049415588378906,
      "learning_rate": 5.737062539455479e-06,
      "loss": 0.6863,
      "step": 6545
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.266988277435303,
      "learning_rate": 5.735913890623479e-06,
      "loss": 0.6158,
      "step": 6546
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.058180809020996,
      "learning_rate": 5.734765202091711e-06,
      "loss": 0.7843,
      "step": 6547
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.914375066757202,
      "learning_rate": 5.73361647392214e-06,
      "loss": 0.1588,
      "step": 6548
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.063718795776367,
      "learning_rate": 5.732467706176736e-06,
      "loss": 0.2049,
      "step": 6549
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.830815315246582,
      "learning_rate": 5.731318898917475e-06,
      "loss": 0.1507,
      "step": 6550
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.724264144897461,
      "learning_rate": 5.730170052206325e-06,
      "loss": 0.9428,
      "step": 6551
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.074391841888428,
      "learning_rate": 5.7290211661052655e-06,
      "loss": 0.8133,
      "step": 6552
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.510215997695923,
      "learning_rate": 5.727872240676275e-06,
      "loss": 0.1585,
      "step": 6553
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.715008735656738,
      "learning_rate": 5.7267232759813304e-06,
      "loss": 0.6166,
      "step": 6554
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.060256004333496,
      "learning_rate": 5.725574272082416e-06,
      "loss": 1.1055,
      "step": 6555
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.9537172317504883,
      "learning_rate": 5.724425229041517e-06,
      "loss": 0.2309,
      "step": 6556
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.798923015594482,
      "learning_rate": 5.723276146920618e-06,
      "loss": 0.6826,
      "step": 6557
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.297244071960449,
      "learning_rate": 5.72212702578171e-06,
      "loss": 0.1395,
      "step": 6558
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.409150123596191,
      "learning_rate": 5.720977865686781e-06,
      "loss": 0.5563,
      "step": 6559
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.4431920051574707,
      "learning_rate": 5.719828666697825e-06,
      "loss": 0.7753,
      "step": 6560
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.896019697189331,
      "learning_rate": 5.718679428876839e-06,
      "loss": 0.1329,
      "step": 6561
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.7162647247314453,
      "learning_rate": 5.717530152285816e-06,
      "loss": 0.1194,
      "step": 6562
    },
    {
      "epoch": 0.47,
      "grad_norm": 14.476567268371582,
      "learning_rate": 5.716380836986758e-06,
      "loss": 1.3293,
      "step": 6563
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.626204013824463,
      "learning_rate": 5.7152314830416654e-06,
      "loss": 0.7023,
      "step": 6564
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.5151729583740234,
      "learning_rate": 5.714082090512542e-06,
      "loss": 0.23,
      "step": 6565
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.34816312789917,
      "learning_rate": 5.712932659461393e-06,
      "loss": 0.0843,
      "step": 6566
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.9511752128601074,
      "learning_rate": 5.711783189950226e-06,
      "loss": 0.1598,
      "step": 6567
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.507232189178467,
      "learning_rate": 5.710633682041049e-06,
      "loss": 0.2066,
      "step": 6568
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.7776715755462646,
      "learning_rate": 5.709484135795874e-06,
      "loss": 0.1262,
      "step": 6569
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6918978691101074,
      "learning_rate": 5.708334551276718e-06,
      "loss": 0.1353,
      "step": 6570
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2673728466033936,
      "learning_rate": 5.707184928545593e-06,
      "loss": 0.1624,
      "step": 6571
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.135922431945801,
      "learning_rate": 5.706035267664517e-06,
      "loss": 0.0723,
      "step": 6572
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.662061929702759,
      "learning_rate": 5.704885568695512e-06,
      "loss": 0.1032,
      "step": 6573
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.8595829010009766,
      "learning_rate": 5.703735831700597e-06,
      "loss": 0.1797,
      "step": 6574
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.3812150955200195,
      "learning_rate": 5.702586056741799e-06,
      "loss": 0.9847,
      "step": 6575
    },
    {
      "epoch": 0.47,
      "grad_norm": 41.4074592590332,
      "learning_rate": 5.701436243881142e-06,
      "loss": 0.6963,
      "step": 6576
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.01289176940918,
      "learning_rate": 5.700286393180656e-06,
      "loss": 0.6648,
      "step": 6577
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.941676616668701,
      "learning_rate": 5.699136504702369e-06,
      "loss": 0.1683,
      "step": 6578
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.447129011154175,
      "learning_rate": 5.697986578508314e-06,
      "loss": 0.1758,
      "step": 6579
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.640588760375977,
      "learning_rate": 5.696836614660525e-06,
      "loss": 0.4323,
      "step": 6580
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.588183879852295,
      "learning_rate": 5.695686613221038e-06,
      "loss": 0.12,
      "step": 6581
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.898735046386719,
      "learning_rate": 5.694536574251892e-06,
      "loss": 1.015,
      "step": 6582
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.252561569213867,
      "learning_rate": 5.693386497815128e-06,
      "loss": 1.076,
      "step": 6583
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.9842658042907715,
      "learning_rate": 5.692236383972786e-06,
      "loss": 0.1455,
      "step": 6584
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.813484191894531,
      "learning_rate": 5.69108623278691e-06,
      "loss": 0.8824,
      "step": 6585
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.4424877166748047,
      "learning_rate": 5.6899360443195485e-06,
      "loss": 0.092,
      "step": 6586
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.8225789070129395,
      "learning_rate": 5.688785818632749e-06,
      "loss": 0.1152,
      "step": 6587
    },
    {
      "epoch": 0.47,
      "grad_norm": 16.031652450561523,
      "learning_rate": 5.687635555788562e-06,
      "loss": 0.8925,
      "step": 6588
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.497551679611206,
      "learning_rate": 5.6864852558490405e-06,
      "loss": 0.2954,
      "step": 6589
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.754286050796509,
      "learning_rate": 5.685334918876237e-06,
      "loss": 0.1457,
      "step": 6590
    },
    {
      "epoch": 0.47,
      "grad_norm": 20.000102996826172,
      "learning_rate": 5.6841845449322084e-06,
      "loss": 0.6344,
      "step": 6591
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.39406418800354,
      "learning_rate": 5.683034134079013e-06,
      "loss": 0.2156,
      "step": 6592
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.166159152984619,
      "learning_rate": 5.681883686378713e-06,
      "loss": 0.9162,
      "step": 6593
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.553549289703369,
      "learning_rate": 5.680733201893368e-06,
      "loss": 0.072,
      "step": 6594
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6366100311279297,
      "learning_rate": 5.679582680685044e-06,
      "loss": 0.142,
      "step": 6595
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.632987022399902,
      "learning_rate": 5.678432122815808e-06,
      "loss": 0.8684,
      "step": 6596
    },
    {
      "epoch": 0.47,
      "grad_norm": 11.738269805908203,
      "learning_rate": 5.6772815283477246e-06,
      "loss": 1.2895,
      "step": 6597
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.080442428588867,
      "learning_rate": 5.676130897342868e-06,
      "loss": 0.7265,
      "step": 6598
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.360846996307373,
      "learning_rate": 5.674980229863307e-06,
      "loss": 0.2801,
      "step": 6599
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.0704851150512695,
      "learning_rate": 5.673829525971118e-06,
      "loss": 0.4368,
      "step": 6600
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.02376651763916,
      "learning_rate": 5.67267878572838e-06,
      "loss": 1.1304,
      "step": 6601
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.583115577697754,
      "learning_rate": 5.671528009197165e-06,
      "loss": 0.8094,
      "step": 6602
    },
    {
      "epoch": 0.47,
      "grad_norm": 13.935036659240723,
      "learning_rate": 5.670377196439555e-06,
      "loss": 0.6301,
      "step": 6603
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1199374198913574,
      "learning_rate": 5.669226347517634e-06,
      "loss": 0.114,
      "step": 6604
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.290247917175293,
      "learning_rate": 5.668075462493484e-06,
      "loss": 0.4305,
      "step": 6605
    },
    {
      "epoch": 0.47,
      "grad_norm": 14.068161964416504,
      "learning_rate": 5.6669245414291915e-06,
      "loss": 0.8934,
      "step": 6606
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.863792419433594,
      "learning_rate": 5.665773584386847e-06,
      "loss": 0.6546,
      "step": 6607
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.34421968460083,
      "learning_rate": 5.6646225914285346e-06,
      "loss": 0.9508,
      "step": 6608
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.039468765258789,
      "learning_rate": 5.6634715626163495e-06,
      "loss": 0.8609,
      "step": 6609
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.337550401687622,
      "learning_rate": 5.662320498012386e-06,
      "loss": 0.1662,
      "step": 6610
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.79694128036499,
      "learning_rate": 5.6611693976787385e-06,
      "loss": 0.3601,
      "step": 6611
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.3577094078063965,
      "learning_rate": 5.6600182616775045e-06,
      "loss": 0.634,
      "step": 6612
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.7482287883758545,
      "learning_rate": 5.658867090070784e-06,
      "loss": 0.1544,
      "step": 6613
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.919831275939941,
      "learning_rate": 5.657715882920677e-06,
      "loss": 0.4874,
      "step": 6614
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.2632474899292,
      "learning_rate": 5.6565646402892895e-06,
      "loss": 1.016,
      "step": 6615
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.784897804260254,
      "learning_rate": 5.655413362238722e-06,
      "loss": 0.1016,
      "step": 6616
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6545755863189697,
      "learning_rate": 5.6542620488310865e-06,
      "loss": 0.1088,
      "step": 6617
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.5611255168914795,
      "learning_rate": 5.653110700128489e-06,
      "loss": 0.1666,
      "step": 6618
    },
    {
      "epoch": 0.47,
      "grad_norm": 11.237664222717285,
      "learning_rate": 5.651959316193043e-06,
      "loss": 0.5095,
      "step": 6619
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.578974723815918,
      "learning_rate": 5.6508078970868585e-06,
      "loss": 0.7747,
      "step": 6620
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.352383613586426,
      "learning_rate": 5.649656442872051e-06,
      "loss": 1.1883,
      "step": 6621
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.688370704650879,
      "learning_rate": 5.6485049536107386e-06,
      "loss": 0.7617,
      "step": 6622
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.4867708683013916,
      "learning_rate": 5.647353429365036e-06,
      "loss": 0.1529,
      "step": 6623
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.658182144165039,
      "learning_rate": 5.64620187019707e-06,
      "loss": 0.1043,
      "step": 6624
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.302725791931152,
      "learning_rate": 5.645050276168958e-06,
      "loss": 0.2858,
      "step": 6625
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.9364049434661865,
      "learning_rate": 5.643898647342824e-06,
      "loss": 0.1743,
      "step": 6626
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.099762916564941,
      "learning_rate": 5.642746983780797e-06,
      "loss": 0.8934,
      "step": 6627
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.2501614093780518,
      "learning_rate": 5.6415952855450005e-06,
      "loss": 0.226,
      "step": 6628
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0529215335845947,
      "learning_rate": 5.64044355269757e-06,
      "loss": 0.0792,
      "step": 6629
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.026857852935791,
      "learning_rate": 5.639291785300631e-06,
      "loss": 0.0829,
      "step": 6630
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.0948963165283203,
      "learning_rate": 5.638139983416323e-06,
      "loss": 0.192,
      "step": 6631
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.138137340545654,
      "learning_rate": 5.636988147106777e-06,
      "loss": 0.8053,
      "step": 6632
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.42484188079834,
      "learning_rate": 5.635836276434131e-06,
      "loss": 0.1344,
      "step": 6633
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.680274248123169,
      "learning_rate": 5.634684371460525e-06,
      "loss": 0.7438,
      "step": 6634
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.985203504562378,
      "learning_rate": 5.6335324322481e-06,
      "loss": 0.185,
      "step": 6635
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.092060089111328,
      "learning_rate": 5.632380458858998e-06,
      "loss": 0.1972,
      "step": 6636
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5933172702789307,
      "learning_rate": 5.631228451355364e-06,
      "loss": 0.1489,
      "step": 6637
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.90539288520813,
      "learning_rate": 5.630076409799346e-06,
      "loss": 0.1797,
      "step": 6638
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.930652618408203,
      "learning_rate": 5.6289243342530875e-06,
      "loss": 0.5817,
      "step": 6639
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.251723766326904,
      "learning_rate": 5.627772224778743e-06,
      "loss": 0.8192,
      "step": 6640
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.987414360046387,
      "learning_rate": 5.626620081438464e-06,
      "loss": 0.9732,
      "step": 6641
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.778871774673462,
      "learning_rate": 5.625467904294403e-06,
      "loss": 0.1507,
      "step": 6642
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.8441591262817383,
      "learning_rate": 5.624315693408718e-06,
      "loss": 0.2935,
      "step": 6643
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.987091064453125,
      "learning_rate": 5.623163448843562e-06,
      "loss": 0.5564,
      "step": 6644
    },
    {
      "epoch": 0.48,
      "grad_norm": 14.50638484954834,
      "learning_rate": 5.622011170661097e-06,
      "loss": 1.1093,
      "step": 6645
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.280339241027832,
      "learning_rate": 5.620858858923484e-06,
      "loss": 0.1296,
      "step": 6646
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.699462652206421,
      "learning_rate": 5.6197065136928865e-06,
      "loss": 0.1394,
      "step": 6647
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.462065696716309,
      "learning_rate": 5.6185541350314665e-06,
      "loss": 0.9777,
      "step": 6648
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.431391716003418,
      "learning_rate": 5.6174017230013944e-06,
      "loss": 0.1399,
      "step": 6649
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.577770709991455,
      "learning_rate": 5.616249277664836e-06,
      "loss": 0.0877,
      "step": 6650
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.640404462814331,
      "learning_rate": 5.61509679908396e-06,
      "loss": 0.1377,
      "step": 6651
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.6244282722473145,
      "learning_rate": 5.613944287320942e-06,
      "loss": 0.8899,
      "step": 6652
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4151315689086914,
      "learning_rate": 5.612791742437952e-06,
      "loss": 0.1188,
      "step": 6653
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.68204927444458,
      "learning_rate": 5.611639164497167e-06,
      "loss": 0.7047,
      "step": 6654
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.3950741291046143,
      "learning_rate": 5.610486553560767e-06,
      "loss": 0.3881,
      "step": 6655
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5321004390716553,
      "learning_rate": 5.609333909690925e-06,
      "loss": 0.1828,
      "step": 6656
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.6984918117523193,
      "learning_rate": 5.608181232949825e-06,
      "loss": 0.284,
      "step": 6657
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3789710998535156,
      "learning_rate": 5.607028523399652e-06,
      "loss": 0.1199,
      "step": 6658
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.869051456451416,
      "learning_rate": 5.605875781102585e-06,
      "loss": 0.7353,
      "step": 6659
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.13683819770813,
      "learning_rate": 5.6047230061208145e-06,
      "loss": 0.1057,
      "step": 6660
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.196549654006958,
      "learning_rate": 5.603570198516529e-06,
      "loss": 0.1279,
      "step": 6661
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.432111978530884,
      "learning_rate": 5.602417358351914e-06,
      "loss": 0.1272,
      "step": 6662
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.33868408203125,
      "learning_rate": 5.601264485689162e-06,
      "loss": 0.1402,
      "step": 6663
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.889310121536255,
      "learning_rate": 5.600111580590468e-06,
      "loss": 1.0157,
      "step": 6664
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.61294412612915,
      "learning_rate": 5.598958643118024e-06,
      "loss": 0.1926,
      "step": 6665
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.259461402893066,
      "learning_rate": 5.597805673334031e-06,
      "loss": 0.9614,
      "step": 6666
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7270859479904175,
      "learning_rate": 5.596652671300684e-06,
      "loss": 0.0483,
      "step": 6667
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.043170928955078,
      "learning_rate": 5.595499637080183e-06,
      "loss": 0.5269,
      "step": 6668
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.047359466552734,
      "learning_rate": 5.594346570734731e-06,
      "loss": 0.932,
      "step": 6669
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.426027536392212,
      "learning_rate": 5.593193472326531e-06,
      "loss": 0.6229,
      "step": 6670
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.9908032417297363,
      "learning_rate": 5.592040341917788e-06,
      "loss": 0.2114,
      "step": 6671
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.814613342285156,
      "learning_rate": 5.590887179570711e-06,
      "loss": 1.1812,
      "step": 6672
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6850833892822266,
      "learning_rate": 5.589733985347507e-06,
      "loss": 0.1484,
      "step": 6673
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.071320056915283,
      "learning_rate": 5.588580759310385e-06,
      "loss": 0.6635,
      "step": 6674
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7531473636627197,
      "learning_rate": 5.587427501521562e-06,
      "loss": 0.1376,
      "step": 6675
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.165477991104126,
      "learning_rate": 5.5862742120432455e-06,
      "loss": 0.1339,
      "step": 6676
    },
    {
      "epoch": 0.48,
      "grad_norm": 13.571264266967773,
      "learning_rate": 5.5851208909376564e-06,
      "loss": 0.6236,
      "step": 6677
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.044196128845215,
      "learning_rate": 5.58396753826701e-06,
      "loss": 0.0728,
      "step": 6678
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.1040172576904297,
      "learning_rate": 5.582814154093526e-06,
      "loss": 0.133,
      "step": 6679
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.717834949493408,
      "learning_rate": 5.581660738479422e-06,
      "loss": 0.8118,
      "step": 6680
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4920623302459717,
      "learning_rate": 5.580507291486925e-06,
      "loss": 0.1491,
      "step": 6681
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.064718723297119,
      "learning_rate": 5.579353813178255e-06,
      "loss": 0.4076,
      "step": 6682
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.706030368804932,
      "learning_rate": 5.578200303615643e-06,
      "loss": 0.7906,
      "step": 6683
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3276174068450928,
      "learning_rate": 5.57704676286131e-06,
      "loss": 0.1296,
      "step": 6684
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.206961631774902,
      "learning_rate": 5.575893190977491e-06,
      "loss": 0.9763,
      "step": 6685
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.8878016471862793,
      "learning_rate": 5.574739588026412e-06,
      "loss": 0.8078,
      "step": 6686
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4941020011901855,
      "learning_rate": 5.573585954070309e-06,
      "loss": 0.0672,
      "step": 6687
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.202989101409912,
      "learning_rate": 5.572432289171415e-06,
      "loss": 0.8723,
      "step": 6688
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4019405841827393,
      "learning_rate": 5.571278593391965e-06,
      "loss": 0.0661,
      "step": 6689
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.8886566162109375,
      "learning_rate": 5.570124866794196e-06,
      "loss": 0.3662,
      "step": 6690
    },
    {
      "epoch": 0.48,
      "grad_norm": 13.725378036499023,
      "learning_rate": 5.568971109440351e-06,
      "loss": 0.3763,
      "step": 6691
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.866788387298584,
      "learning_rate": 5.5678173213926674e-06,
      "loss": 0.732,
      "step": 6692
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.8174779415130615,
      "learning_rate": 5.5666635027133885e-06,
      "loss": 0.4276,
      "step": 6693
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.1158103942871094,
      "learning_rate": 5.565509653464759e-06,
      "loss": 0.1397,
      "step": 6694
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.512201309204102,
      "learning_rate": 5.564355773709023e-06,
      "loss": 0.9457,
      "step": 6695
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.488330125808716,
      "learning_rate": 5.563201863508431e-06,
      "loss": 0.1279,
      "step": 6696
    },
    {
      "epoch": 0.48,
      "grad_norm": 32.92778015136719,
      "learning_rate": 5.56204792292523e-06,
      "loss": 0.5885,
      "step": 6697
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.550128936767578,
      "learning_rate": 5.560893952021671e-06,
      "loss": 0.1103,
      "step": 6698
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4395101070404053,
      "learning_rate": 5.559739950860005e-06,
      "loss": 0.0873,
      "step": 6699
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.0686769485473633,
      "learning_rate": 5.558585919502489e-06,
      "loss": 0.158,
      "step": 6700
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.530503273010254,
      "learning_rate": 5.557431858011375e-06,
      "loss": 0.961,
      "step": 6701
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9804631471633911,
      "learning_rate": 5.556277766448923e-06,
      "loss": 0.1252,
      "step": 6702
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.7271311283111572,
      "learning_rate": 5.5551236448773925e-06,
      "loss": 0.7864,
      "step": 6703
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.848785400390625,
      "learning_rate": 5.553969493359043e-06,
      "loss": 1.3031,
      "step": 6704
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8967690467834473,
      "learning_rate": 5.552815311956133e-06,
      "loss": 0.0944,
      "step": 6705
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.268951416015625,
      "learning_rate": 5.551661100730933e-06,
      "loss": 0.7447,
      "step": 6706
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.583533525466919,
      "learning_rate": 5.5505068597457024e-06,
      "loss": 0.128,
      "step": 6707
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.0001633167266846,
      "learning_rate": 5.5493525890627105e-06,
      "loss": 0.1616,
      "step": 6708
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.691324710845947,
      "learning_rate": 5.548198288744229e-06,
      "loss": 1.1621,
      "step": 6709
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.509457111358643,
      "learning_rate": 5.547043958852522e-06,
      "loss": 0.9481,
      "step": 6710
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.749436378479004,
      "learning_rate": 5.545889599449863e-06,
      "loss": 0.7953,
      "step": 6711
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.6717209815979,
      "learning_rate": 5.544735210598529e-06,
      "loss": 0.1565,
      "step": 6712
    },
    {
      "epoch": 0.48,
      "grad_norm": 14.97298812866211,
      "learning_rate": 5.543580792360791e-06,
      "loss": 0.944,
      "step": 6713
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.407238006591797,
      "learning_rate": 5.542426344798926e-06,
      "loss": 1.0775,
      "step": 6714
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.553358316421509,
      "learning_rate": 5.541271867975216e-06,
      "loss": 0.0887,
      "step": 6715
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.735990524291992,
      "learning_rate": 5.540117361951934e-06,
      "loss": 0.1148,
      "step": 6716
    },
    {
      "epoch": 0.48,
      "grad_norm": 21.381397247314453,
      "learning_rate": 5.538962826791366e-06,
      "loss": 0.8796,
      "step": 6717
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.520969390869141,
      "learning_rate": 5.537808262555795e-06,
      "loss": 0.7973,
      "step": 6718
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.684489250183105,
      "learning_rate": 5.536653669307502e-06,
      "loss": 1.4004,
      "step": 6719
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.550161838531494,
      "learning_rate": 5.535499047108776e-06,
      "loss": 0.1326,
      "step": 6720
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.842339038848877,
      "learning_rate": 5.5343443960219045e-06,
      "loss": 0.792,
      "step": 6721
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.294413566589355,
      "learning_rate": 5.533189716109173e-06,
      "loss": 0.6854,
      "step": 6722
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4179739952087402,
      "learning_rate": 5.532035007432876e-06,
      "loss": 0.0898,
      "step": 6723
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.334853172302246,
      "learning_rate": 5.530880270055304e-06,
      "loss": 0.7168,
      "step": 6724
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2374768257141113,
      "learning_rate": 5.52972550403875e-06,
      "loss": 0.0879,
      "step": 6725
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.204300880432129,
      "learning_rate": 5.528570709445511e-06,
      "loss": 0.6719,
      "step": 6726
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.7084951400756836,
      "learning_rate": 5.527415886337883e-06,
      "loss": 0.8092,
      "step": 6727
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.187677383422852,
      "learning_rate": 5.526261034778164e-06,
      "loss": 0.978,
      "step": 6728
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.8880984783172607,
      "learning_rate": 5.525106154828656e-06,
      "loss": 0.7431,
      "step": 6729
    },
    {
      "epoch": 0.48,
      "grad_norm": 13.397013664245605,
      "learning_rate": 5.523951246551657e-06,
      "loss": 0.6042,
      "step": 6730
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.11585807800293,
      "learning_rate": 5.522796310009469e-06,
      "loss": 0.8968,
      "step": 6731
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7260963916778564,
      "learning_rate": 5.521641345264402e-06,
      "loss": 0.1571,
      "step": 6732
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7544126510620117,
      "learning_rate": 5.52048635237876e-06,
      "loss": 0.1128,
      "step": 6733
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.1905059814453125,
      "learning_rate": 5.519331331414846e-06,
      "loss": 0.2488,
      "step": 6734
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.89155101776123,
      "learning_rate": 5.518176282434976e-06,
      "loss": 0.8575,
      "step": 6735
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.328343391418457,
      "learning_rate": 5.517021205501455e-06,
      "loss": 0.095,
      "step": 6736
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.252011775970459,
      "learning_rate": 5.5158661006765975e-06,
      "loss": 0.9643,
      "step": 6737
    },
    {
      "epoch": 0.48,
      "grad_norm": 18.014331817626953,
      "learning_rate": 5.5147109680227165e-06,
      "loss": 0.7122,
      "step": 6738
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.156545877456665,
      "learning_rate": 5.513555807602126e-06,
      "loss": 0.1203,
      "step": 6739
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.9891114234924316,
      "learning_rate": 5.512400619477144e-06,
      "loss": 0.1989,
      "step": 6740
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.383279800415039,
      "learning_rate": 5.511245403710089e-06,
      "loss": 0.7812,
      "step": 6741
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.255810022354126,
      "learning_rate": 5.51009016036328e-06,
      "loss": 0.1268,
      "step": 6742
    },
    {
      "epoch": 0.48,
      "grad_norm": 14.57345962524414,
      "learning_rate": 5.508934889499037e-06,
      "loss": 0.7173,
      "step": 6743
    },
    {
      "epoch": 0.48,
      "grad_norm": 15.04857349395752,
      "learning_rate": 5.507779591179681e-06,
      "loss": 0.6645,
      "step": 6744
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.6664116382598877,
      "learning_rate": 5.506624265467542e-06,
      "loss": 0.2551,
      "step": 6745
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.947211503982544,
      "learning_rate": 5.5054689124249394e-06,
      "loss": 0.1964,
      "step": 6746
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.167558193206787,
      "learning_rate": 5.504313532114202e-06,
      "loss": 0.5684,
      "step": 6747
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5197720527648926,
      "learning_rate": 5.503158124597658e-06,
      "loss": 0.1113,
      "step": 6748
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3154070377349854,
      "learning_rate": 5.50200268993764e-06,
      "loss": 0.1474,
      "step": 6749
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.058757781982422,
      "learning_rate": 5.500847228196475e-06,
      "loss": 0.2092,
      "step": 6750
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.739823818206787,
      "learning_rate": 5.499691739436498e-06,
      "loss": 0.219,
      "step": 6751
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.4950788021087646,
      "learning_rate": 5.4985362237200455e-06,
      "loss": 0.1923,
      "step": 6752
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.288471221923828,
      "learning_rate": 5.497380681109449e-06,
      "loss": 0.3896,
      "step": 6753
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.883372783660889,
      "learning_rate": 5.496225111667046e-06,
      "loss": 0.8607,
      "step": 6754
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.884669303894043,
      "learning_rate": 5.495069515455179e-06,
      "loss": 0.1094,
      "step": 6755
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.245828151702881,
      "learning_rate": 5.493913892536184e-06,
      "loss": 0.9457,
      "step": 6756
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.0202791690826416,
      "learning_rate": 5.492758242972405e-06,
      "loss": 0.1684,
      "step": 6757
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.071927547454834,
      "learning_rate": 5.491602566826184e-06,
      "loss": 0.4039,
      "step": 6758
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.685799598693848,
      "learning_rate": 5.490446864159865e-06,
      "loss": 1.1093,
      "step": 6759
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2206757068634033,
      "learning_rate": 5.489291135035795e-06,
      "loss": 0.1658,
      "step": 6760
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6651852130889893,
      "learning_rate": 5.488135379516318e-06,
      "loss": 0.1146,
      "step": 6761
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.870401382446289,
      "learning_rate": 5.486979597663786e-06,
      "loss": 1.1417,
      "step": 6762
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.421466588973999,
      "learning_rate": 5.485823789540548e-06,
      "loss": 0.1064,
      "step": 6763
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1782474517822266,
      "learning_rate": 5.484667955208958e-06,
      "loss": 0.0944,
      "step": 6764
    },
    {
      "epoch": 0.49,
      "grad_norm": 12.153481483459473,
      "learning_rate": 5.483512094731362e-06,
      "loss": 0.6663,
      "step": 6765
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.4685709476470947,
      "learning_rate": 5.482356208170122e-06,
      "loss": 0.1084,
      "step": 6766
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.296482801437378,
      "learning_rate": 5.48120029558759e-06,
      "loss": 0.273,
      "step": 6767
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.552541255950928,
      "learning_rate": 5.480044357046122e-06,
      "loss": 0.8928,
      "step": 6768
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.161220073699951,
      "learning_rate": 5.4788883926080794e-06,
      "loss": 0.0731,
      "step": 6769
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.841573476791382,
      "learning_rate": 5.47773240233582e-06,
      "loss": 0.9035,
      "step": 6770
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.7623116970062256,
      "learning_rate": 5.476576386291705e-06,
      "loss": 0.1778,
      "step": 6771
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.894285202026367,
      "learning_rate": 5.475420344538101e-06,
      "loss": 0.7793,
      "step": 6772
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.116223335266113,
      "learning_rate": 5.474264277137365e-06,
      "loss": 0.5571,
      "step": 6773
    },
    {
      "epoch": 0.49,
      "grad_norm": 17.485992431640625,
      "learning_rate": 5.47310818415187e-06,
      "loss": 1.3297,
      "step": 6774
    },
    {
      "epoch": 0.49,
      "grad_norm": 23.157716751098633,
      "learning_rate": 5.471952065643976e-06,
      "loss": 0.9601,
      "step": 6775
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.264354705810547,
      "learning_rate": 5.470795921676056e-06,
      "loss": 0.1092,
      "step": 6776
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.180832624435425,
      "learning_rate": 5.4696397523104786e-06,
      "loss": 0.0877,
      "step": 6777
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.7470381259918213,
      "learning_rate": 5.468483557609614e-06,
      "loss": 0.1957,
      "step": 6778
    },
    {
      "epoch": 0.49,
      "grad_norm": 15.322029113769531,
      "learning_rate": 5.4673273376358334e-06,
      "loss": 1.2329,
      "step": 6779
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.7701117992401123,
      "learning_rate": 5.466171092451514e-06,
      "loss": 0.1732,
      "step": 6780
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.042272567749023,
      "learning_rate": 5.465014822119028e-06,
      "loss": 0.3962,
      "step": 6781
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.1552734375,
      "learning_rate": 5.463858526700753e-06,
      "loss": 0.7887,
      "step": 6782
    },
    {
      "epoch": 0.49,
      "grad_norm": 16.797353744506836,
      "learning_rate": 5.462702206259065e-06,
      "loss": 1.3068,
      "step": 6783
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.6219828128814697,
      "learning_rate": 5.4615458608563444e-06,
      "loss": 0.1454,
      "step": 6784
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.3396897315979004,
      "learning_rate": 5.460389490554972e-06,
      "loss": 0.2004,
      "step": 6785
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.8142824172973633,
      "learning_rate": 5.45923309541733e-06,
      "loss": 0.4033,
      "step": 6786
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.499669075012207,
      "learning_rate": 5.4580766755058e-06,
      "loss": 0.1431,
      "step": 6787
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.9370503425598145,
      "learning_rate": 5.456920230882768e-06,
      "loss": 0.5728,
      "step": 6788
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.597405910491943,
      "learning_rate": 5.455763761610618e-06,
      "loss": 0.8233,
      "step": 6789
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.1887125968933105,
      "learning_rate": 5.454607267751739e-06,
      "loss": 1.0663,
      "step": 6790
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3271608352661133,
      "learning_rate": 5.453450749368518e-06,
      "loss": 0.1233,
      "step": 6791
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7862151861190796,
      "learning_rate": 5.452294206523347e-06,
      "loss": 0.0767,
      "step": 6792
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.045109987258911,
      "learning_rate": 5.451137639278613e-06,
      "loss": 0.0788,
      "step": 6793
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.500083923339844,
      "learning_rate": 5.449981047696713e-06,
      "loss": 0.9507,
      "step": 6794
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.244239807128906,
      "learning_rate": 5.4488244318400395e-06,
      "loss": 0.8996,
      "step": 6795
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.916879892349243,
      "learning_rate": 5.447667791770986e-06,
      "loss": 0.2905,
      "step": 6796
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.948573589324951,
      "learning_rate": 5.446511127551949e-06,
      "loss": 0.664,
      "step": 6797
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.957610845565796,
      "learning_rate": 5.445354439245326e-06,
      "loss": 0.151,
      "step": 6798
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.8723764419555664,
      "learning_rate": 5.444197726913517e-06,
      "loss": 0.2168,
      "step": 6799
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.113210201263428,
      "learning_rate": 5.443040990618925e-06,
      "loss": 0.9256,
      "step": 6800
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.156022548675537,
      "learning_rate": 5.441884230423946e-06,
      "loss": 0.1048,
      "step": 6801
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.657897472381592,
      "learning_rate": 5.440727446390985e-06,
      "loss": 0.8167,
      "step": 6802
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.614415168762207,
      "learning_rate": 5.439570638582448e-06,
      "loss": 1.0557,
      "step": 6803
    },
    {
      "epoch": 0.49,
      "grad_norm": 15.051764488220215,
      "learning_rate": 5.4384138070607374e-06,
      "loss": 0.882,
      "step": 6804
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.897552490234375,
      "learning_rate": 5.437256951888262e-06,
      "loss": 0.196,
      "step": 6805
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.513503551483154,
      "learning_rate": 5.43610007312743e-06,
      "loss": 0.5593,
      "step": 6806
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.765864133834839,
      "learning_rate": 5.43494317084065e-06,
      "loss": 0.145,
      "step": 6807
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.606735229492188,
      "learning_rate": 5.43378624509033e-06,
      "loss": 1.0808,
      "step": 6808
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.754261016845703,
      "learning_rate": 5.432629295938886e-06,
      "loss": 0.8418,
      "step": 6809
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.476998805999756,
      "learning_rate": 5.4314723234487285e-06,
      "loss": 0.7226,
      "step": 6810
    },
    {
      "epoch": 0.49,
      "grad_norm": 16.52519989013672,
      "learning_rate": 5.430315327682273e-06,
      "loss": 0.8347,
      "step": 6811
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.277793884277344,
      "learning_rate": 5.429158308701934e-06,
      "loss": 0.9901,
      "step": 6812
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.842139720916748,
      "learning_rate": 5.428001266570127e-06,
      "loss": 0.7803,
      "step": 6813
    },
    {
      "epoch": 0.49,
      "grad_norm": 12.573620796203613,
      "learning_rate": 5.426844201349275e-06,
      "loss": 0.9757,
      "step": 6814
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.991071939468384,
      "learning_rate": 5.425687113101792e-06,
      "loss": 0.2244,
      "step": 6815
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3555421829223633,
      "learning_rate": 5.4245300018901005e-06,
      "loss": 0.1523,
      "step": 6816
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.170689344406128,
      "learning_rate": 5.423372867776623e-06,
      "loss": 0.1578,
      "step": 6817
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7038381099700928,
      "learning_rate": 5.422215710823782e-06,
      "loss": 0.7479,
      "step": 6818
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.471742153167725,
      "learning_rate": 5.4210585310940004e-06,
      "loss": 0.8125,
      "step": 6819
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.2894961833953857,
      "learning_rate": 5.419901328649706e-06,
      "loss": 0.2345,
      "step": 6820
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.706267356872559,
      "learning_rate": 5.4187441035533225e-06,
      "loss": 0.8185,
      "step": 6821
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0798120498657227,
      "learning_rate": 5.41758685586728e-06,
      "loss": 0.0654,
      "step": 6822
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.3106300830841064,
      "learning_rate": 5.41642958565401e-06,
      "loss": 0.2193,
      "step": 6823
    },
    {
      "epoch": 0.49,
      "grad_norm": 16.589853286743164,
      "learning_rate": 5.415272292975938e-06,
      "loss": 0.489,
      "step": 6824
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.158233642578125,
      "learning_rate": 5.4141149778954985e-06,
      "loss": 0.7438,
      "step": 6825
    },
    {
      "epoch": 0.49,
      "grad_norm": 17.71464729309082,
      "learning_rate": 5.412957640475124e-06,
      "loss": 1.5361,
      "step": 6826
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7794277667999268,
      "learning_rate": 5.4118002807772465e-06,
      "loss": 0.7128,
      "step": 6827
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.4899396896362305,
      "learning_rate": 5.410642898864303e-06,
      "loss": 0.2622,
      "step": 6828
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.50521469116211,
      "learning_rate": 5.409485494798729e-06,
      "loss": 0.5651,
      "step": 6829
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.632735252380371,
      "learning_rate": 5.408328068642964e-06,
      "loss": 0.8834,
      "step": 6830
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.815382957458496,
      "learning_rate": 5.407170620459443e-06,
      "loss": 0.1298,
      "step": 6831
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.4230785369873047,
      "learning_rate": 5.40601315031061e-06,
      "loss": 0.1547,
      "step": 6832
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.2271900177001953,
      "learning_rate": 5.404855658258904e-06,
      "loss": 0.1981,
      "step": 6833
    },
    {
      "epoch": 0.49,
      "grad_norm": 26.261852264404297,
      "learning_rate": 5.403698144366768e-06,
      "loss": 0.973,
      "step": 6834
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.630120038986206,
      "learning_rate": 5.402540608696645e-06,
      "loss": 0.1331,
      "step": 6835
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.873988151550293,
      "learning_rate": 5.40138305131098e-06,
      "loss": 1.1587,
      "step": 6836
    },
    {
      "epoch": 0.49,
      "grad_norm": 21.074859619140625,
      "learning_rate": 5.400225472272219e-06,
      "loss": 1.1039,
      "step": 6837
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.899020195007324,
      "learning_rate": 5.399067871642808e-06,
      "loss": 0.7953,
      "step": 6838
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.295458793640137,
      "learning_rate": 5.3979102494851955e-06,
      "loss": 1.0551,
      "step": 6839
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.039547443389893,
      "learning_rate": 5.396752605861831e-06,
      "loss": 0.839,
      "step": 6840
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.1676435470581055,
      "learning_rate": 5.395594940835166e-06,
      "loss": 0.8879,
      "step": 6841
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.4168052673339844,
      "learning_rate": 5.394437254467653e-06,
      "loss": 0.1164,
      "step": 6842
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.421131134033203,
      "learning_rate": 5.393279546821741e-06,
      "loss": 1.0421,
      "step": 6843
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.3361873626708984,
      "learning_rate": 5.392121817959887e-06,
      "loss": 0.1029,
      "step": 6844
    },
    {
      "epoch": 0.49,
      "grad_norm": 18.050325393676758,
      "learning_rate": 5.390964067944543e-06,
      "loss": 1.2265,
      "step": 6845
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.901432752609253,
      "learning_rate": 5.38980629683817e-06,
      "loss": 0.1521,
      "step": 6846
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.0219390392303467,
      "learning_rate": 5.388648504703221e-06,
      "loss": 0.1935,
      "step": 6847
    },
    {
      "epoch": 0.49,
      "grad_norm": 17.24956703186035,
      "learning_rate": 5.387490691602158e-06,
      "loss": 1.2862,
      "step": 6848
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.782229423522949,
      "learning_rate": 5.386332857597439e-06,
      "loss": 0.1069,
      "step": 6849
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.482044219970703,
      "learning_rate": 5.3851750027515235e-06,
      "loss": 0.8625,
      "step": 6850
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.7345056533813477,
      "learning_rate": 5.384017127126875e-06,
      "loss": 0.1206,
      "step": 6851
    },
    {
      "epoch": 0.49,
      "grad_norm": 16.57011604309082,
      "learning_rate": 5.382859230785955e-06,
      "loss": 0.8818,
      "step": 6852
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.9564967155456543,
      "learning_rate": 5.38170131379123e-06,
      "loss": 0.7357,
      "step": 6853
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.6594479084014893,
      "learning_rate": 5.380543376205165e-06,
      "loss": 0.1743,
      "step": 6854
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.725799083709717,
      "learning_rate": 5.3793854180902225e-06,
      "loss": 0.1802,
      "step": 6855
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.14591121673584,
      "learning_rate": 5.378227439508874e-06,
      "loss": 0.4795,
      "step": 6856
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.685238838195801,
      "learning_rate": 5.377069440523588e-06,
      "loss": 0.4165,
      "step": 6857
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.4578897953033447,
      "learning_rate": 5.375911421196831e-06,
      "loss": 0.2182,
      "step": 6858
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.080073356628418,
      "learning_rate": 5.374753381591076e-06,
      "loss": 0.2,
      "step": 6859
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.314394950866699,
      "learning_rate": 5.3735953217687976e-06,
      "loss": 0.2175,
      "step": 6860
    },
    {
      "epoch": 0.49,
      "grad_norm": 19.032167434692383,
      "learning_rate": 5.372437241792462e-06,
      "loss": 1.6708,
      "step": 6861
    },
    {
      "epoch": 0.49,
      "grad_norm": 12.551080703735352,
      "learning_rate": 5.371279141724548e-06,
      "loss": 0.9471,
      "step": 6862
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.596463203430176,
      "learning_rate": 5.37012102162753e-06,
      "loss": 0.1334,
      "step": 6863
    },
    {
      "epoch": 0.49,
      "grad_norm": 18.99374008178711,
      "learning_rate": 5.3689628815638835e-06,
      "loss": 1.4987,
      "step": 6864
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.945690393447876,
      "learning_rate": 5.367804721596087e-06,
      "loss": 0.1817,
      "step": 6865
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.274316310882568,
      "learning_rate": 5.366646541786619e-06,
      "loss": 0.9026,
      "step": 6866
    },
    {
      "epoch": 0.49,
      "grad_norm": 18.275081634521484,
      "learning_rate": 5.365488342197955e-06,
      "loss": 0.9598,
      "step": 6867
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.628401756286621,
      "learning_rate": 5.364330122892581e-06,
      "loss": 0.9297,
      "step": 6868
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.797816276550293,
      "learning_rate": 5.363171883932974e-06,
      "loss": 0.1634,
      "step": 6869
    },
    {
      "epoch": 0.49,
      "grad_norm": 14.434748649597168,
      "learning_rate": 5.362013625381618e-06,
      "loss": 0.9566,
      "step": 6870
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.25811767578125,
      "learning_rate": 5.360855347301e-06,
      "loss": 0.9853,
      "step": 6871
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.324587106704712,
      "learning_rate": 5.3596970497536004e-06,
      "loss": 0.0973,
      "step": 6872
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.791787624359131,
      "learning_rate": 5.3585387328019055e-06,
      "loss": 0.7039,
      "step": 6873
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.2572014331817627,
      "learning_rate": 5.357380396508405e-06,
      "loss": 0.1236,
      "step": 6874
    },
    {
      "epoch": 0.49,
      "grad_norm": 19.10042953491211,
      "learning_rate": 5.3562220409355834e-06,
      "loss": 0.7242,
      "step": 6875
    },
    {
      "epoch": 0.49,
      "grad_norm": 15.367011070251465,
      "learning_rate": 5.355063666145931e-06,
      "loss": 1.2186,
      "step": 6876
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.830657005310059,
      "learning_rate": 5.353905272201939e-06,
      "loss": 0.9001,
      "step": 6877
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1489274501800537,
      "learning_rate": 5.352746859166097e-06,
      "loss": 0.1413,
      "step": 6878
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.475449562072754,
      "learning_rate": 5.351588427100897e-06,
      "loss": 1.1031,
      "step": 6879
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.174131870269775,
      "learning_rate": 5.3504299760688315e-06,
      "loss": 0.9739,
      "step": 6880
    },
    {
      "epoch": 0.49,
      "grad_norm": 12.850561141967773,
      "learning_rate": 5.349271506132395e-06,
      "loss": 1.1424,
      "step": 6881
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.965825080871582,
      "learning_rate": 5.348113017354084e-06,
      "loss": 0.1903,
      "step": 6882
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1388399600982666,
      "learning_rate": 5.346954509796393e-06,
      "loss": 0.141,
      "step": 6883
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.474905490875244,
      "learning_rate": 5.34579598352182e-06,
      "loss": 0.3081,
      "step": 6884
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.5899739265441895,
      "learning_rate": 5.344637438592861e-06,
      "loss": 0.8589,
      "step": 6885
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.493683338165283,
      "learning_rate": 5.343478875072018e-06,
      "loss": 0.1121,
      "step": 6886
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.9211385250091553,
      "learning_rate": 5.34232029302179e-06,
      "loss": 0.4418,
      "step": 6887
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.152930974960327,
      "learning_rate": 5.341161692504677e-06,
      "loss": 0.2936,
      "step": 6888
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.577589750289917,
      "learning_rate": 5.340003073583182e-06,
      "loss": 0.146,
      "step": 6889
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.780285358428955,
      "learning_rate": 5.33884443631981e-06,
      "loss": 0.2642,
      "step": 6890
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.482837438583374,
      "learning_rate": 5.337685780777062e-06,
      "loss": 0.1468,
      "step": 6891
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.64186429977417,
      "learning_rate": 5.336527107017444e-06,
      "loss": 0.8094,
      "step": 6892
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.52242374420166,
      "learning_rate": 5.335368415103461e-06,
      "loss": 0.6459,
      "step": 6893
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.997211933135986,
      "learning_rate": 5.334209705097625e-06,
      "loss": 0.7084,
      "step": 6894
    },
    {
      "epoch": 0.49,
      "grad_norm": 25.23040771484375,
      "learning_rate": 5.3330509770624375e-06,
      "loss": 1.1892,
      "step": 6895
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.70635724067688,
      "learning_rate": 5.331892231060411e-06,
      "loss": 0.1449,
      "step": 6896
    },
    {
      "epoch": 0.49,
      "grad_norm": 14.08711051940918,
      "learning_rate": 5.330733467154057e-06,
      "loss": 1.3492,
      "step": 6897
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7275538444519043,
      "learning_rate": 5.329574685405883e-06,
      "loss": 0.6244,
      "step": 6898
    },
    {
      "epoch": 0.49,
      "grad_norm": 20.663658142089844,
      "learning_rate": 5.328415885878401e-06,
      "loss": 1.1785,
      "step": 6899
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.464698314666748,
      "learning_rate": 5.3272570686341265e-06,
      "loss": 0.7336,
      "step": 6900
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.271448135375977,
      "learning_rate": 5.326098233735571e-06,
      "loss": 0.7666,
      "step": 6901
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.398315906524658,
      "learning_rate": 5.324939381245251e-06,
      "loss": 0.138,
      "step": 6902
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.7762858867645264,
      "learning_rate": 5.323780511225682e-06,
      "loss": 0.1767,
      "step": 6903
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.184584617614746,
      "learning_rate": 5.322621623739378e-06,
      "loss": 0.129,
      "step": 6904
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7676100730895996,
      "learning_rate": 5.32146271884886e-06,
      "loss": 0.191,
      "step": 6905
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.104360580444336,
      "learning_rate": 5.320303796616644e-06,
      "loss": 0.864,
      "step": 6906
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.784419059753418,
      "learning_rate": 5.319144857105251e-06,
      "loss": 0.7123,
      "step": 6907
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.883378982543945,
      "learning_rate": 5.317985900377203e-06,
      "loss": 1.0566,
      "step": 6908
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.782102584838867,
      "learning_rate": 5.316826926495016e-06,
      "loss": 0.2215,
      "step": 6909
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.852545738220215,
      "learning_rate": 5.315667935521217e-06,
      "loss": 0.8877,
      "step": 6910
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.392398834228516,
      "learning_rate": 5.314508927518328e-06,
      "loss": 0.8245,
      "step": 6911
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.2979111671447754,
      "learning_rate": 5.313349902548872e-06,
      "loss": 0.2866,
      "step": 6912
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.080273151397705,
      "learning_rate": 5.312190860675376e-06,
      "loss": 0.0886,
      "step": 6913
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.733335494995117,
      "learning_rate": 5.311031801960366e-06,
      "loss": 0.8064,
      "step": 6914
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2990028858184814,
      "learning_rate": 5.3098727264663655e-06,
      "loss": 0.1322,
      "step": 6915
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.235487461090088,
      "learning_rate": 5.3087136342559045e-06,
      "loss": 0.5805,
      "step": 6916
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.4912140369415283,
      "learning_rate": 5.307554525391512e-06,
      "loss": 0.1671,
      "step": 6917
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9054553508758545,
      "learning_rate": 5.306395399935715e-06,
      "loss": 0.0945,
      "step": 6918
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1312642097473145,
      "learning_rate": 5.305236257951049e-06,
      "loss": 0.0715,
      "step": 6919
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.85225248336792,
      "learning_rate": 5.3040770995000415e-06,
      "loss": 0.7169,
      "step": 6920
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.2987961769104004,
      "learning_rate": 5.302917924645225e-06,
      "loss": 0.1961,
      "step": 6921
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.539132118225098,
      "learning_rate": 5.301758733449132e-06,
      "loss": 1.2088,
      "step": 6922
    },
    {
      "epoch": 0.5,
      "grad_norm": 14.497913360595703,
      "learning_rate": 5.300599525974299e-06,
      "loss": 1.0921,
      "step": 6923
    },
    {
      "epoch": 0.5,
      "grad_norm": 15.558096885681152,
      "learning_rate": 5.299440302283258e-06,
      "loss": 1.2882,
      "step": 6924
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.9588027000427246,
      "learning_rate": 5.298281062438549e-06,
      "loss": 0.9034,
      "step": 6925
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.378283500671387,
      "learning_rate": 5.297121806502703e-06,
      "loss": 0.9487,
      "step": 6926
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.22400164604187,
      "learning_rate": 5.295962534538261e-06,
      "loss": 0.182,
      "step": 6927
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7005536556243896,
      "learning_rate": 5.294803246607761e-06,
      "loss": 0.1167,
      "step": 6928
    },
    {
      "epoch": 0.5,
      "grad_norm": 14.695283889770508,
      "learning_rate": 5.293643942773741e-06,
      "loss": 0.9963,
      "step": 6929
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2350008487701416,
      "learning_rate": 5.2924846230987416e-06,
      "loss": 0.0918,
      "step": 6930
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.225976943969727,
      "learning_rate": 5.291325287645305e-06,
      "loss": 0.9643,
      "step": 6931
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9834115505218506,
      "learning_rate": 5.290165936475971e-06,
      "loss": 0.0773,
      "step": 6932
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.1273274421691895,
      "learning_rate": 5.289006569653284e-06,
      "loss": 0.6644,
      "step": 6933
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.185359001159668,
      "learning_rate": 5.287847187239786e-06,
      "loss": 0.8266,
      "step": 6934
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2484445571899414,
      "learning_rate": 5.28668778929802e-06,
      "loss": 0.1747,
      "step": 6935
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.0257716178894043,
      "learning_rate": 5.285528375890536e-06,
      "loss": 0.2185,
      "step": 6936
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.820408821105957,
      "learning_rate": 5.284368947079875e-06,
      "loss": 0.9488,
      "step": 6937
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.908677577972412,
      "learning_rate": 5.283209502928587e-06,
      "loss": 0.1061,
      "step": 6938
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7030107975006104,
      "learning_rate": 5.2820500434992174e-06,
      "loss": 0.0635,
      "step": 6939
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.140401840209961,
      "learning_rate": 5.280890568854315e-06,
      "loss": 0.1555,
      "step": 6940
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.54437255859375,
      "learning_rate": 5.27973107905643e-06,
      "loss": 0.4421,
      "step": 6941
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.395134925842285,
      "learning_rate": 5.278571574168113e-06,
      "loss": 0.6735,
      "step": 6942
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.453813314437866,
      "learning_rate": 5.277412054251914e-06,
      "loss": 0.1399,
      "step": 6943
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.793320178985596,
      "learning_rate": 5.276252519370384e-06,
      "loss": 0.716,
      "step": 6944
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.8350372314453125,
      "learning_rate": 5.275092969586077e-06,
      "loss": 0.1061,
      "step": 6945
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.24946403503418,
      "learning_rate": 5.273933404961544e-06,
      "loss": 1.0666,
      "step": 6946
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.148733615875244,
      "learning_rate": 5.2727738255593416e-06,
      "loss": 0.8764,
      "step": 6947
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.07430362701416,
      "learning_rate": 5.271614231442025e-06,
      "loss": 0.5787,
      "step": 6948
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.424025058746338,
      "learning_rate": 5.270454622672147e-06,
      "loss": 0.6575,
      "step": 6949
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.192106246948242,
      "learning_rate": 5.269294999312267e-06,
      "loss": 0.9202,
      "step": 6950
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5553174018859863,
      "learning_rate": 5.268135361424941e-06,
      "loss": 0.113,
      "step": 6951
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.491791248321533,
      "learning_rate": 5.266975709072727e-06,
      "loss": 0.7078,
      "step": 6952
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.0790679454803467,
      "learning_rate": 5.265816042318183e-06,
      "loss": 0.1353,
      "step": 6953
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.194458484649658,
      "learning_rate": 5.264656361223871e-06,
      "loss": 0.1478,
      "step": 6954
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.8847970962524414,
      "learning_rate": 5.263496665852349e-06,
      "loss": 0.1431,
      "step": 6955
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.964948654174805,
      "learning_rate": 5.262336956266181e-06,
      "loss": 0.752,
      "step": 6956
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.114041328430176,
      "learning_rate": 5.261177232527926e-06,
      "loss": 0.1577,
      "step": 6957
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.657561779022217,
      "learning_rate": 5.260017494700148e-06,
      "loss": 0.1116,
      "step": 6958
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.991654396057129,
      "learning_rate": 5.258857742845412e-06,
      "loss": 0.6438,
      "step": 6959
    },
    {
      "epoch": 0.5,
      "grad_norm": 17.57351303100586,
      "learning_rate": 5.257697977026277e-06,
      "loss": 2.1065,
      "step": 6960
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.506199836730957,
      "learning_rate": 5.256538197305313e-06,
      "loss": 0.9917,
      "step": 6961
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.655411720275879,
      "learning_rate": 5.255378403745087e-06,
      "loss": 0.8436,
      "step": 6962
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.260075092315674,
      "learning_rate": 5.25421859640816e-06,
      "loss": 0.1911,
      "step": 6963
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.759223222732544,
      "learning_rate": 5.253058775357103e-06,
      "loss": 0.1849,
      "step": 6964
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.289855480194092,
      "learning_rate": 5.251898940654483e-06,
      "loss": 0.8059,
      "step": 6965
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.0818376541137695,
      "learning_rate": 5.2507390923628675e-06,
      "loss": 0.8275,
      "step": 6966
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.881756544113159,
      "learning_rate": 5.249579230544827e-06,
      "loss": 0.1663,
      "step": 6967
    },
    {
      "epoch": 0.5,
      "grad_norm": 25.074296951293945,
      "learning_rate": 5.248419355262934e-06,
      "loss": 1.0437,
      "step": 6968
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.422341823577881,
      "learning_rate": 5.247259466579755e-06,
      "loss": 0.8754,
      "step": 6969
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.886590003967285,
      "learning_rate": 5.246099564557864e-06,
      "loss": 1.0725,
      "step": 6970
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.128389596939087,
      "learning_rate": 5.244939649259835e-06,
      "loss": 0.2073,
      "step": 6971
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.857957363128662,
      "learning_rate": 5.2437797207482365e-06,
      "loss": 0.8025,
      "step": 6972
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.122498989105225,
      "learning_rate": 5.242619779085646e-06,
      "loss": 0.73,
      "step": 6973
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.979227066040039,
      "learning_rate": 5.241459824334636e-06,
      "loss": 0.9182,
      "step": 6974
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.6856954097747803,
      "learning_rate": 5.240299856557783e-06,
      "loss": 0.1479,
      "step": 6975
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.04371166229248,
      "learning_rate": 5.239139875817662e-06,
      "loss": 0.9727,
      "step": 6976
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6226794719696045,
      "learning_rate": 5.237979882176851e-06,
      "loss": 0.2105,
      "step": 6977
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.983934164047241,
      "learning_rate": 5.236819875697925e-06,
      "loss": 0.1922,
      "step": 6978
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.023285865783691,
      "learning_rate": 5.235659856443465e-06,
      "loss": 1.2079,
      "step": 6979
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.100804328918457,
      "learning_rate": 5.234499824476045e-06,
      "loss": 1.0764,
      "step": 6980
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.4655892848968506,
      "learning_rate": 5.2333397798582484e-06,
      "loss": 0.1847,
      "step": 6981
    },
    {
      "epoch": 0.5,
      "grad_norm": 14.773565292358398,
      "learning_rate": 5.2321797226526545e-06,
      "loss": 1.0293,
      "step": 6982
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.520915985107422,
      "learning_rate": 5.231019652921843e-06,
      "loss": 1.3335,
      "step": 6983
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7494068145751953,
      "learning_rate": 5.229859570728395e-06,
      "loss": 0.2002,
      "step": 6984
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.970657825469971,
      "learning_rate": 5.228699476134894e-06,
      "loss": 0.673,
      "step": 6985
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.342178821563721,
      "learning_rate": 5.227539369203921e-06,
      "loss": 0.7969,
      "step": 6986
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.704598426818848,
      "learning_rate": 5.2263792499980615e-06,
      "loss": 0.6526,
      "step": 6987
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.2829458713531494,
      "learning_rate": 5.225219118579899e-06,
      "loss": 0.1405,
      "step": 6988
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5794429779052734,
      "learning_rate": 5.224058975012016e-06,
      "loss": 0.1495,
      "step": 6989
    },
    {
      "epoch": 0.5,
      "grad_norm": 17.84427261352539,
      "learning_rate": 5.222898819356999e-06,
      "loss": 1.1061,
      "step": 6990
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.296159505844116,
      "learning_rate": 5.221738651677436e-06,
      "loss": 0.0978,
      "step": 6991
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7138638496398926,
      "learning_rate": 5.22057847203591e-06,
      "loss": 0.1371,
      "step": 6992
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.4131603240966797,
      "learning_rate": 5.219418280495013e-06,
      "loss": 0.1272,
      "step": 6993
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.960792303085327,
      "learning_rate": 5.218258077117328e-06,
      "loss": 0.784,
      "step": 6994
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.007632732391357,
      "learning_rate": 5.2170978619654475e-06,
      "loss": 0.2132,
      "step": 6995
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.46112585067749,
      "learning_rate": 5.215937635101959e-06,
      "loss": 0.3342,
      "step": 6996
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.505363464355469,
      "learning_rate": 5.214777396589451e-06,
      "loss": 1.0799,
      "step": 6997
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.805098056793213,
      "learning_rate": 5.2136171464905164e-06,
      "loss": 0.2133,
      "step": 6998
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.615818977355957,
      "learning_rate": 5.212456884867747e-06,
      "loss": 0.113,
      "step": 6999
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.0646133422851562,
      "learning_rate": 5.211296611783731e-06,
      "loss": 0.2787,
      "step": 7000
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.0047316551208496,
      "learning_rate": 5.2101363273010606e-06,
      "loss": 0.2001,
      "step": 7001
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.494405746459961,
      "learning_rate": 5.208976031482335e-06,
      "loss": 1.2936,
      "step": 7002
    },
    {
      "epoch": 0.5,
      "grad_norm": 21.02206039428711,
      "learning_rate": 5.207815724390139e-06,
      "loss": 1.2035,
      "step": 7003
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.436125755310059,
      "learning_rate": 5.206655406087074e-06,
      "loss": 0.7972,
      "step": 7004
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9260292053222656,
      "learning_rate": 5.205495076635734e-06,
      "loss": 0.0599,
      "step": 7005
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6225709915161133,
      "learning_rate": 5.20433473609871e-06,
      "loss": 0.1527,
      "step": 7006
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.299656867980957,
      "learning_rate": 5.203174384538599e-06,
      "loss": 0.1307,
      "step": 7007
    },
    {
      "epoch": 0.5,
      "grad_norm": 15.170188903808594,
      "learning_rate": 5.202014022018002e-06,
      "loss": 0.9388,
      "step": 7008
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.8776216506958,
      "learning_rate": 5.200853648599511e-06,
      "loss": 0.9939,
      "step": 7009
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.637459754943848,
      "learning_rate": 5.199693264345728e-06,
      "loss": 0.5366,
      "step": 7010
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7638871669769287,
      "learning_rate": 5.198532869319248e-06,
      "loss": 0.0888,
      "step": 7011
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.79677677154541,
      "learning_rate": 5.197372463582671e-06,
      "loss": 0.1264,
      "step": 7012
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.374124765396118,
      "learning_rate": 5.196212047198597e-06,
      "loss": 0.1645,
      "step": 7013
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.55448842048645,
      "learning_rate": 5.195051620229625e-06,
      "loss": 0.1842,
      "step": 7014
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.798769950866699,
      "learning_rate": 5.193891182738357e-06,
      "loss": 0.1545,
      "step": 7015
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.966343879699707,
      "learning_rate": 5.1927307347873935e-06,
      "loss": 0.9648,
      "step": 7016
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.124767303466797,
      "learning_rate": 5.191570276439337e-06,
      "loss": 0.2263,
      "step": 7017
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.069148063659668,
      "learning_rate": 5.190409807756789e-06,
      "loss": 0.5701,
      "step": 7018
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.896665096282959,
      "learning_rate": 5.189249328802352e-06,
      "loss": 0.6861,
      "step": 7019
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.193716049194336,
      "learning_rate": 5.188088839638631e-06,
      "loss": 0.1403,
      "step": 7020
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.658909797668457,
      "learning_rate": 5.186928340328228e-06,
      "loss": 0.775,
      "step": 7021
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.429229259490967,
      "learning_rate": 5.185767830933749e-06,
      "loss": 0.8268,
      "step": 7022
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.346606254577637,
      "learning_rate": 5.1846073115178e-06,
      "loss": 0.8709,
      "step": 7023
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.868361473083496,
      "learning_rate": 5.183446782142983e-06,
      "loss": 0.6732,
      "step": 7024
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0266449451446533,
      "learning_rate": 5.182286242871908e-06,
      "loss": 0.0997,
      "step": 7025
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.173090696334839,
      "learning_rate": 5.181125693767179e-06,
      "loss": 0.0674,
      "step": 7026
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.210979461669922,
      "learning_rate": 5.179965134891406e-06,
      "loss": 0.8491,
      "step": 7027
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.2983784675598145,
      "learning_rate": 5.1788045663071936e-06,
      "loss": 0.3024,
      "step": 7028
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.210638999938965,
      "learning_rate": 5.177643988077154e-06,
      "loss": 0.1182,
      "step": 7029
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.0819754600524902,
      "learning_rate": 5.176483400263891e-06,
      "loss": 0.1233,
      "step": 7030
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2745392322540283,
      "learning_rate": 5.175322802930018e-06,
      "loss": 0.1078,
      "step": 7031
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.303621292114258,
      "learning_rate": 5.174162196138143e-06,
      "loss": 1.5843,
      "step": 7032
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.015338897705078,
      "learning_rate": 5.173001579950877e-06,
      "loss": 1.4132,
      "step": 7033
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.7118608951568604,
      "learning_rate": 5.17184095443083e-06,
      "loss": 0.7224,
      "step": 7034
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.785813808441162,
      "learning_rate": 5.170680319640615e-06,
      "loss": 0.7364,
      "step": 7035
    },
    {
      "epoch": 0.5,
      "grad_norm": 20.751632690429688,
      "learning_rate": 5.1695196756428435e-06,
      "loss": 0.7405,
      "step": 7036
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.2220683097839355,
      "learning_rate": 5.168359022500126e-06,
      "loss": 0.9581,
      "step": 7037
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.379527568817139,
      "learning_rate": 5.167198360275075e-06,
      "loss": 0.1527,
      "step": 7038
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.321193218231201,
      "learning_rate": 5.166037689030308e-06,
      "loss": 0.782,
      "step": 7039
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.85841703414917,
      "learning_rate": 5.164877008828434e-06,
      "loss": 0.1865,
      "step": 7040
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.039969444274902,
      "learning_rate": 5.163716319732071e-06,
      "loss": 0.6038,
      "step": 7041
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.210753440856934,
      "learning_rate": 5.162555621803832e-06,
      "loss": 1.0915,
      "step": 7042
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.8548948764801025,
      "learning_rate": 5.161394915106331e-06,
      "loss": 0.1848,
      "step": 7043
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.484311103820801,
      "learning_rate": 5.160234199702186e-06,
      "loss": 0.1597,
      "step": 7044
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.93287992477417,
      "learning_rate": 5.159073475654014e-06,
      "loss": 0.4001,
      "step": 7045
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1812052726745605,
      "learning_rate": 5.157912743024428e-06,
      "loss": 0.1889,
      "step": 7046
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.812036514282227,
      "learning_rate": 5.156752001876048e-06,
      "loss": 0.7354,
      "step": 7047
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.564760446548462,
      "learning_rate": 5.155591252271492e-06,
      "loss": 0.1074,
      "step": 7048
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.0669777393341064,
      "learning_rate": 5.154430494273376e-06,
      "loss": 0.1272,
      "step": 7049
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.301396369934082,
      "learning_rate": 5.15326972794432e-06,
      "loss": 0.5394,
      "step": 7050
    },
    {
      "epoch": 0.51,
      "grad_norm": 20.616165161132812,
      "learning_rate": 5.152108953346941e-06,
      "loss": 1.0182,
      "step": 7051
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.141336679458618,
      "learning_rate": 5.1509481705438605e-06,
      "loss": 0.094,
      "step": 7052
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.699932336807251,
      "learning_rate": 5.149787379597698e-06,
      "loss": 0.125,
      "step": 7053
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.805615425109863,
      "learning_rate": 5.148626580571073e-06,
      "loss": 0.8712,
      "step": 7054
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6879074573516846,
      "learning_rate": 5.147465773526605e-06,
      "loss": 0.1332,
      "step": 7055
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1591436862945557,
      "learning_rate": 5.1463049585269195e-06,
      "loss": 0.1539,
      "step": 7056
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.69414234161377,
      "learning_rate": 5.145144135634633e-06,
      "loss": 1.0027,
      "step": 7057
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.4354166984558105,
      "learning_rate": 5.14398330491237e-06,
      "loss": 0.5604,
      "step": 7058
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.0036721229553223,
      "learning_rate": 5.142822466422756e-06,
      "loss": 0.1311,
      "step": 7059
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.0616214275360107,
      "learning_rate": 5.141661620228406e-06,
      "loss": 0.19,
      "step": 7060
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.390727043151855,
      "learning_rate": 5.140500766391949e-06,
      "loss": 0.4589,
      "step": 7061
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.223433971405029,
      "learning_rate": 5.13933990497601e-06,
      "loss": 0.7686,
      "step": 7062
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.174476623535156,
      "learning_rate": 5.138179036043207e-06,
      "loss": 0.7454,
      "step": 7063
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.691484451293945,
      "learning_rate": 5.137018159656171e-06,
      "loss": 0.9142,
      "step": 7064
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6569864749908447,
      "learning_rate": 5.135857275877523e-06,
      "loss": 0.113,
      "step": 7065
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.6718201637268066,
      "learning_rate": 5.134696384769888e-06,
      "loss": 0.2436,
      "step": 7066
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.084631443023682,
      "learning_rate": 5.133535486395895e-06,
      "loss": 0.9404,
      "step": 7067
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.795123338699341,
      "learning_rate": 5.132374580818166e-06,
      "loss": 0.1632,
      "step": 7068
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1473987102508545,
      "learning_rate": 5.13121366809933e-06,
      "loss": 0.09,
      "step": 7069
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6983747482299805,
      "learning_rate": 5.130052748302013e-06,
      "loss": 0.188,
      "step": 7070
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.209561824798584,
      "learning_rate": 5.128891821488844e-06,
      "loss": 0.9615,
      "step": 7071
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.462048053741455,
      "learning_rate": 5.127730887722449e-06,
      "loss": 0.7724,
      "step": 7072
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.840229511260986,
      "learning_rate": 5.126569947065455e-06,
      "loss": 0.4426,
      "step": 7073
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.8941195011138916,
      "learning_rate": 5.125408999580491e-06,
      "loss": 0.2244,
      "step": 7074
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.382996082305908,
      "learning_rate": 5.124248045330186e-06,
      "loss": 0.7609,
      "step": 7075
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5855915546417236,
      "learning_rate": 5.1230870843771705e-06,
      "loss": 0.1136,
      "step": 7076
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.058497428894043,
      "learning_rate": 5.1219261167840725e-06,
      "loss": 0.7124,
      "step": 7077
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.118941307067871,
      "learning_rate": 5.120765142613522e-06,
      "loss": 0.8384,
      "step": 7078
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.138000011444092,
      "learning_rate": 5.119604161928149e-06,
      "loss": 0.15,
      "step": 7079
    },
    {
      "epoch": 0.51,
      "grad_norm": 13.574618339538574,
      "learning_rate": 5.118443174790584e-06,
      "loss": 1.1046,
      "step": 7080
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8453959226608276,
      "learning_rate": 5.117282181263459e-06,
      "loss": 0.0689,
      "step": 7081
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7934582233428955,
      "learning_rate": 5.116121181409403e-06,
      "loss": 0.0528,
      "step": 7082
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.571986198425293,
      "learning_rate": 5.114960175291049e-06,
      "loss": 1.211,
      "step": 7083
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.98233699798584,
      "learning_rate": 5.1137991629710295e-06,
      "loss": 0.9565,
      "step": 7084
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.088736057281494,
      "learning_rate": 5.112638144511976e-06,
      "loss": 0.8509,
      "step": 7085
    },
    {
      "epoch": 0.51,
      "grad_norm": 13.45777702331543,
      "learning_rate": 5.111477119976519e-06,
      "loss": 0.7714,
      "step": 7086
    },
    {
      "epoch": 0.51,
      "grad_norm": 16.23153305053711,
      "learning_rate": 5.110316089427295e-06,
      "loss": 0.7735,
      "step": 7087
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.945376873016357,
      "learning_rate": 5.109155052926935e-06,
      "loss": 0.55,
      "step": 7088
    },
    {
      "epoch": 0.51,
      "grad_norm": 18.601198196411133,
      "learning_rate": 5.1079940105380744e-06,
      "loss": 1.2123,
      "step": 7089
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.638619899749756,
      "learning_rate": 5.1068329623233446e-06,
      "loss": 0.3671,
      "step": 7090
    },
    {
      "epoch": 0.51,
      "grad_norm": 12.391399383544922,
      "learning_rate": 5.105671908345381e-06,
      "loss": 0.6858,
      "step": 7091
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.2601215839385986,
      "learning_rate": 5.104510848666817e-06,
      "loss": 0.2227,
      "step": 7092
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4201815128326416,
      "learning_rate": 5.10334978335029e-06,
      "loss": 0.1221,
      "step": 7093
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5545599460601807,
      "learning_rate": 5.102188712458431e-06,
      "loss": 0.148,
      "step": 7094
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.424168586730957,
      "learning_rate": 5.1010276360538804e-06,
      "loss": 0.1284,
      "step": 7095
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.770672798156738,
      "learning_rate": 5.099866554199271e-06,
      "loss": 0.5024,
      "step": 7096
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.014656066894531,
      "learning_rate": 5.098705466957236e-06,
      "loss": 0.7245,
      "step": 7097
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.2669026851654053,
      "learning_rate": 5.097544374390417e-06,
      "loss": 0.2321,
      "step": 7098
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.392310619354248,
      "learning_rate": 5.09638327656145e-06,
      "loss": 0.0878,
      "step": 7099
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.1289520263671875,
      "learning_rate": 5.0952221735329665e-06,
      "loss": 0.2578,
      "step": 7100
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.170140504837036,
      "learning_rate": 5.0940610653676095e-06,
      "loss": 0.097,
      "step": 7101
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.225003242492676,
      "learning_rate": 5.0928999521280144e-06,
      "loss": 0.2689,
      "step": 7102
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.998929500579834,
      "learning_rate": 5.091738833876816e-06,
      "loss": 0.603,
      "step": 7103
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.913592576980591,
      "learning_rate": 5.090577710676657e-06,
      "loss": 0.1381,
      "step": 7104
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.551321983337402,
      "learning_rate": 5.089416582590172e-06,
      "loss": 0.7947,
      "step": 7105
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.107466220855713,
      "learning_rate": 5.088255449680001e-06,
      "loss": 0.0651,
      "step": 7106
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.022684097290039,
      "learning_rate": 5.087094312008784e-06,
      "loss": 0.1315,
      "step": 7107
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5970535278320312,
      "learning_rate": 5.085933169639158e-06,
      "loss": 0.1619,
      "step": 7108
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.4504082202911377,
      "learning_rate": 5.0847720226337614e-06,
      "loss": 0.2548,
      "step": 7109
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.7121808528900146,
      "learning_rate": 5.083610871055237e-06,
      "loss": 0.6926,
      "step": 7110
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6251115798950195,
      "learning_rate": 5.082449714966221e-06,
      "loss": 0.1897,
      "step": 7111
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.867579460144043,
      "learning_rate": 5.081288554429355e-06,
      "loss": 0.1356,
      "step": 7112
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6174042224884033,
      "learning_rate": 5.080127389507281e-06,
      "loss": 0.1175,
      "step": 7113
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.155557155609131,
      "learning_rate": 5.078966220262635e-06,
      "loss": 0.724,
      "step": 7114
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.754330635070801,
      "learning_rate": 5.077805046758062e-06,
      "loss": 0.1282,
      "step": 7115
    },
    {
      "epoch": 0.51,
      "grad_norm": 14.057090759277344,
      "learning_rate": 5.076643869056201e-06,
      "loss": 1.1073,
      "step": 7116
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.569941997528076,
      "learning_rate": 5.075482687219693e-06,
      "loss": 0.0894,
      "step": 7117
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.467376232147217,
      "learning_rate": 5.07432150131118e-06,
      "loss": 0.7555,
      "step": 7118
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.925597667694092,
      "learning_rate": 5.073160311393305e-06,
      "loss": 0.9107,
      "step": 7119
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.670640468597412,
      "learning_rate": 5.071999117528705e-06,
      "loss": 0.9629,
      "step": 7120
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.2937686443328857,
      "learning_rate": 5.070837919780027e-06,
      "loss": 0.2211,
      "step": 7121
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.6421144008636475,
      "learning_rate": 5.06967671820991e-06,
      "loss": 0.9138,
      "step": 7122
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.6272711753845215,
      "learning_rate": 5.068515512880999e-06,
      "loss": 0.2006,
      "step": 7123
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9222476482391357,
      "learning_rate": 5.067354303855936e-06,
      "loss": 0.1513,
      "step": 7124
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.148873329162598,
      "learning_rate": 5.066193091197363e-06,
      "loss": 0.739,
      "step": 7125
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5380654335021973,
      "learning_rate": 5.0650318749679215e-06,
      "loss": 0.1227,
      "step": 7126
    },
    {
      "epoch": 0.51,
      "grad_norm": 15.057205200195312,
      "learning_rate": 5.063870655230258e-06,
      "loss": 0.6317,
      "step": 7127
    },
    {
      "epoch": 0.51,
      "grad_norm": 14.831502914428711,
      "learning_rate": 5.062709432047015e-06,
      "loss": 0.9763,
      "step": 7128
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1549108028411865,
      "learning_rate": 5.061548205480834e-06,
      "loss": 0.0479,
      "step": 7129
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.577592372894287,
      "learning_rate": 5.060386975594361e-06,
      "loss": 0.1689,
      "step": 7130
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6094982624053955,
      "learning_rate": 5.05922574245024e-06,
      "loss": 0.1597,
      "step": 7131
    },
    {
      "epoch": 0.51,
      "grad_norm": 13.121167182922363,
      "learning_rate": 5.0580645061111135e-06,
      "loss": 0.6838,
      "step": 7132
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9035892486572266,
      "learning_rate": 5.056903266639626e-06,
      "loss": 0.1435,
      "step": 7133
    },
    {
      "epoch": 0.51,
      "grad_norm": 15.497703552246094,
      "learning_rate": 5.055742024098423e-06,
      "loss": 0.8202,
      "step": 7134
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.614983081817627,
      "learning_rate": 5.05458077855015e-06,
      "loss": 0.1397,
      "step": 7135
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.780487537384033,
      "learning_rate": 5.053419530057449e-06,
      "loss": 0.1717,
      "step": 7136
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.98077392578125,
      "learning_rate": 5.052258278682969e-06,
      "loss": 0.7381,
      "step": 7137
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.686889410018921,
      "learning_rate": 5.05109702448935e-06,
      "loss": 0.2019,
      "step": 7138
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.2058749198913574,
      "learning_rate": 5.049935767539241e-06,
      "loss": 0.2274,
      "step": 7139
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.1977081298828125,
      "learning_rate": 5.0487745078952865e-06,
      "loss": 0.5487,
      "step": 7140
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.3541486263275146,
      "learning_rate": 5.047613245620132e-06,
      "loss": 0.2158,
      "step": 7141
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5411343574523926,
      "learning_rate": 5.046451980776422e-06,
      "loss": 0.1545,
      "step": 7142
    },
    {
      "epoch": 0.51,
      "grad_norm": 16.668630599975586,
      "learning_rate": 5.045290713426805e-06,
      "loss": 1.1515,
      "step": 7143
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.131688117980957,
      "learning_rate": 5.044129443633925e-06,
      "loss": 0.9444,
      "step": 7144
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.03812837600708,
      "learning_rate": 5.042968171460428e-06,
      "loss": 0.0871,
      "step": 7145
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.286914825439453,
      "learning_rate": 5.04180689696896e-06,
      "loss": 0.1113,
      "step": 7146
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9497907161712646,
      "learning_rate": 5.040645620222171e-06,
      "loss": 0.2029,
      "step": 7147
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.6310110092163086,
      "learning_rate": 5.039484341282702e-06,
      "loss": 0.0939,
      "step": 7148
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4150936603546143,
      "learning_rate": 5.038323060213204e-06,
      "loss": 0.1149,
      "step": 7149
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.758092880249023,
      "learning_rate": 5.0371617770763214e-06,
      "loss": 0.5385,
      "step": 7150
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.002135992050171,
      "learning_rate": 5.036000491934701e-06,
      "loss": 0.0935,
      "step": 7151
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.6844024658203125,
      "learning_rate": 5.034839204850989e-06,
      "loss": 0.8765,
      "step": 7152
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.568599224090576,
      "learning_rate": 5.0336779158878364e-06,
      "loss": 0.1139,
      "step": 7153
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.4327967166900635,
      "learning_rate": 5.032516625107888e-06,
      "loss": 0.2221,
      "step": 7154
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.482724666595459,
      "learning_rate": 5.0313553325737886e-06,
      "loss": 0.0952,
      "step": 7155
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.3376662731170654,
      "learning_rate": 5.030194038348189e-06,
      "loss": 0.1965,
      "step": 7156
    },
    {
      "epoch": 0.51,
      "grad_norm": 16.972002029418945,
      "learning_rate": 5.029032742493735e-06,
      "loss": 0.7252,
      "step": 7157
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.746386170387268,
      "learning_rate": 5.027871445073075e-06,
      "loss": 0.0633,
      "step": 7158
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.686285972595215,
      "learning_rate": 5.026710146148854e-06,
      "loss": 0.9291,
      "step": 7159
    },
    {
      "epoch": 0.51,
      "grad_norm": 14.47597599029541,
      "learning_rate": 5.025548845783723e-06,
      "loss": 1.0346,
      "step": 7160
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5348074436187744,
      "learning_rate": 5.02438754404033e-06,
      "loss": 0.188,
      "step": 7161
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.4698495864868164,
      "learning_rate": 5.023226240981319e-06,
      "loss": 0.2391,
      "step": 7162
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.241485357284546,
      "learning_rate": 5.022064936669341e-06,
      "loss": 0.1037,
      "step": 7163
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.228954315185547,
      "learning_rate": 5.020903631167043e-06,
      "loss": 0.0917,
      "step": 7164
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.407486915588379,
      "learning_rate": 5.019742324537073e-06,
      "loss": 1.3931,
      "step": 7165
    },
    {
      "epoch": 0.51,
      "grad_norm": 15.176389694213867,
      "learning_rate": 5.01858101684208e-06,
      "loss": 0.8212,
      "step": 7166
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3869786262512207,
      "learning_rate": 5.0174197081447126e-06,
      "loss": 0.1776,
      "step": 7167
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.048456907272339,
      "learning_rate": 5.0162583985076165e-06,
      "loss": 0.1052,
      "step": 7168
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1094236373901367,
      "learning_rate": 5.0150970879934415e-06,
      "loss": 0.0932,
      "step": 7169
    },
    {
      "epoch": 0.51,
      "grad_norm": 12.503087043762207,
      "learning_rate": 5.013935776664836e-06,
      "loss": 0.8388,
      "step": 7170
    },
    {
      "epoch": 0.51,
      "grad_norm": 14.351838111877441,
      "learning_rate": 5.012774464584447e-06,
      "loss": 0.7763,
      "step": 7171
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.2583208084106445,
      "learning_rate": 5.011613151814926e-06,
      "loss": 0.7747,
      "step": 7172
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6761717796325684,
      "learning_rate": 5.0104518384189185e-06,
      "loss": 0.109,
      "step": 7173
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.1112685203552246,
      "learning_rate": 5.009290524459073e-06,
      "loss": 0.1545,
      "step": 7174
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.190705299377441,
      "learning_rate": 5.008129209998039e-06,
      "loss": 0.6739,
      "step": 7175
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.575478553771973,
      "learning_rate": 5.006967895098466e-06,
      "loss": 1.0122,
      "step": 7176
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.806344985961914,
      "learning_rate": 5.005806579822999e-06,
      "loss": 0.0967,
      "step": 7177
    },
    {
      "epoch": 0.51,
      "grad_norm": 16.7582950592041,
      "learning_rate": 5.004645264234292e-06,
      "loss": 1.0995,
      "step": 7178
    },
    {
      "epoch": 0.51,
      "grad_norm": 14.887175559997559,
      "learning_rate": 5.0034839483949895e-06,
      "loss": 0.917,
      "step": 7179
    },
    {
      "epoch": 0.51,
      "grad_norm": 16.361906051635742,
      "learning_rate": 5.002322632367741e-06,
      "loss": 1.438,
      "step": 7180
    },
    {
      "epoch": 0.51,
      "grad_norm": 24.35605812072754,
      "learning_rate": 5.001161316215195e-06,
      "loss": 0.7311,
      "step": 7181
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.074355125427246,
      "learning_rate": 5e-06,
      "loss": 0.403,
      "step": 7182
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.932365417480469,
      "learning_rate": 4.998838683784806e-06,
      "loss": 0.5829,
      "step": 7183
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.3441214561462402,
      "learning_rate": 4.99767736763226e-06,
      "loss": 0.2282,
      "step": 7184
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.5329530239105225,
      "learning_rate": 4.996516051605011e-06,
      "loss": 0.1972,
      "step": 7185
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.861734390258789,
      "learning_rate": 4.99535473576571e-06,
      "loss": 0.7024,
      "step": 7186
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.3375847339630127,
      "learning_rate": 4.994193420177001e-06,
      "loss": 0.152,
      "step": 7187
    },
    {
      "epoch": 0.52,
      "grad_norm": 23.26999282836914,
      "learning_rate": 4.993032104901535e-06,
      "loss": 1.9902,
      "step": 7188
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.7687456607818604,
      "learning_rate": 4.991870790001963e-06,
      "loss": 0.1066,
      "step": 7189
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.505341053009033,
      "learning_rate": 4.990709475540931e-06,
      "loss": 0.1845,
      "step": 7190
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.45141863822937,
      "learning_rate": 4.989548161581084e-06,
      "loss": 0.0953,
      "step": 7191
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.35209584236145,
      "learning_rate": 4.988386848185076e-06,
      "loss": 0.0687,
      "step": 7192
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.17456579208374,
      "learning_rate": 4.987225535415555e-06,
      "loss": 0.7713,
      "step": 7193
    },
    {
      "epoch": 0.52,
      "grad_norm": 15.431589126586914,
      "learning_rate": 4.986064223335166e-06,
      "loss": 0.8196,
      "step": 7194
    },
    {
      "epoch": 0.52,
      "grad_norm": 16.024879455566406,
      "learning_rate": 4.984902912006561e-06,
      "loss": 0.7797,
      "step": 7195
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.716632843017578,
      "learning_rate": 4.983741601492385e-06,
      "loss": 0.1443,
      "step": 7196
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.634355545043945,
      "learning_rate": 4.982580291855288e-06,
      "loss": 0.5549,
      "step": 7197
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.077442169189453,
      "learning_rate": 4.98141898315792e-06,
      "loss": 0.6818,
      "step": 7198
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.673628330230713,
      "learning_rate": 4.980257675462927e-06,
      "loss": 0.5745,
      "step": 7199
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.937359094619751,
      "learning_rate": 4.979096368832958e-06,
      "loss": 0.7738,
      "step": 7200
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.460305690765381,
      "learning_rate": 4.977935063330661e-06,
      "loss": 0.1778,
      "step": 7201
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.3182573318481445,
      "learning_rate": 4.9767737590186836e-06,
      "loss": 0.2509,
      "step": 7202
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.1495256423950195,
      "learning_rate": 4.9756124559596725e-06,
      "loss": 0.8782,
      "step": 7203
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.1005663871765137,
      "learning_rate": 4.974451154216279e-06,
      "loss": 0.1056,
      "step": 7204
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.055776834487915,
      "learning_rate": 4.973289853851147e-06,
      "loss": 0.1786,
      "step": 7205
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.569076061248779,
      "learning_rate": 4.972128554926927e-06,
      "loss": 1.0797,
      "step": 7206
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.024354934692383,
      "learning_rate": 4.970967257506267e-06,
      "loss": 0.6927,
      "step": 7207
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5091023445129395,
      "learning_rate": 4.969805961651812e-06,
      "loss": 0.0877,
      "step": 7208
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.3305134773254395,
      "learning_rate": 4.968644667426213e-06,
      "loss": 0.1279,
      "step": 7209
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.6751976013183594,
      "learning_rate": 4.967483374892113e-06,
      "loss": 0.1544,
      "step": 7210
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.289955139160156,
      "learning_rate": 4.9663220841121635e-06,
      "loss": 0.5784,
      "step": 7211
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.461219072341919,
      "learning_rate": 4.9651607951490106e-06,
      "loss": 0.1983,
      "step": 7212
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.260972738265991,
      "learning_rate": 4.963999508065302e-06,
      "loss": 0.0945,
      "step": 7213
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.50076675415039,
      "learning_rate": 4.962838222923681e-06,
      "loss": 0.9454,
      "step": 7214
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.311245918273926,
      "learning_rate": 4.961676939786798e-06,
      "loss": 1.095,
      "step": 7215
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.408490180969238,
      "learning_rate": 4.960515658717299e-06,
      "loss": 0.7236,
      "step": 7216
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8414041996002197,
      "learning_rate": 4.959354379777831e-06,
      "loss": 0.2175,
      "step": 7217
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.86000919342041,
      "learning_rate": 4.958193103031041e-06,
      "loss": 0.8358,
      "step": 7218
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8828041553497314,
      "learning_rate": 4.957031828539573e-06,
      "loss": 0.1381,
      "step": 7219
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.09548020362854,
      "learning_rate": 4.9558705563660755e-06,
      "loss": 0.1847,
      "step": 7220
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7466368675231934,
      "learning_rate": 4.954709286573196e-06,
      "loss": 0.6211,
      "step": 7221
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8716487884521484,
      "learning_rate": 4.953548019223578e-06,
      "loss": 0.133,
      "step": 7222
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.6303577423095703,
      "learning_rate": 4.952386754379869e-06,
      "loss": 0.1796,
      "step": 7223
    },
    {
      "epoch": 0.52,
      "grad_norm": 14.707518577575684,
      "learning_rate": 4.951225492104714e-06,
      "loss": 0.6946,
      "step": 7224
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.626389741897583,
      "learning_rate": 4.9500642324607615e-06,
      "loss": 0.7446,
      "step": 7225
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.320011138916016,
      "learning_rate": 4.948902975510651e-06,
      "loss": 0.7097,
      "step": 7226
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7833218574523926,
      "learning_rate": 4.947741721317034e-06,
      "loss": 0.9326,
      "step": 7227
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.6971986293792725,
      "learning_rate": 4.9465804699425516e-06,
      "loss": 0.2628,
      "step": 7228
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.9050049781799316,
      "learning_rate": 4.945419221449851e-06,
      "loss": 0.1656,
      "step": 7229
    },
    {
      "epoch": 0.52,
      "grad_norm": 14.141057968139648,
      "learning_rate": 4.944257975901578e-06,
      "loss": 0.9172,
      "step": 7230
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.801012992858887,
      "learning_rate": 4.9430967333603745e-06,
      "loss": 0.939,
      "step": 7231
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.691707611083984,
      "learning_rate": 4.941935493888888e-06,
      "loss": 0.5433,
      "step": 7232
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.621412754058838,
      "learning_rate": 4.940774257549761e-06,
      "loss": 0.7693,
      "step": 7233
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7679877281188965,
      "learning_rate": 4.93961302440564e-06,
      "loss": 0.0647,
      "step": 7234
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.5287599563598633,
      "learning_rate": 4.9384517945191665e-06,
      "loss": 0.2323,
      "step": 7235
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8159968852996826,
      "learning_rate": 4.937290567952986e-06,
      "loss": 0.0429,
      "step": 7236
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.255682468414307,
      "learning_rate": 4.9361293447697435e-06,
      "loss": 0.5053,
      "step": 7237
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.5061681270599365,
      "learning_rate": 4.93496812503208e-06,
      "loss": 0.8225,
      "step": 7238
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.561548233032227,
      "learning_rate": 4.933806908802639e-06,
      "loss": 1.1647,
      "step": 7239
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5155677795410156,
      "learning_rate": 4.932645696144066e-06,
      "loss": 0.0726,
      "step": 7240
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.517390251159668,
      "learning_rate": 4.9314844871190025e-06,
      "loss": 0.9937,
      "step": 7241
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.450723648071289,
      "learning_rate": 4.93032328179009e-06,
      "loss": 0.8634,
      "step": 7242
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1594862937927246,
      "learning_rate": 4.929162080219974e-06,
      "loss": 0.1014,
      "step": 7243
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8023319244384766,
      "learning_rate": 4.9280008824712964e-06,
      "loss": 0.2079,
      "step": 7244
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.6233253479003906,
      "learning_rate": 4.9268396886066975e-06,
      "loss": 0.1072,
      "step": 7245
    },
    {
      "epoch": 0.52,
      "grad_norm": 12.623659133911133,
      "learning_rate": 4.925678498688821e-06,
      "loss": 0.7732,
      "step": 7246
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.260096549987793,
      "learning_rate": 4.924517312780307e-06,
      "loss": 0.0789,
      "step": 7247
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.6956446170806885,
      "learning_rate": 4.923356130943799e-06,
      "loss": 0.1747,
      "step": 7248
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.6989004611968994,
      "learning_rate": 4.92219495324194e-06,
      "loss": 0.1491,
      "step": 7249
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.2921142578125,
      "learning_rate": 4.9210337797373665e-06,
      "loss": 0.13,
      "step": 7250
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2694802284240723,
      "learning_rate": 4.919872610492721e-06,
      "loss": 0.1739,
      "step": 7251
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.90762710571289,
      "learning_rate": 4.918711445570647e-06,
      "loss": 1.0057,
      "step": 7252
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.04297399520874,
      "learning_rate": 4.917550285033781e-06,
      "loss": 0.4996,
      "step": 7253
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.1259889602661133,
      "learning_rate": 4.916389128944765e-06,
      "loss": 0.2065,
      "step": 7254
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.5182523727417,
      "learning_rate": 4.915227977366239e-06,
      "loss": 0.405,
      "step": 7255
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.826406478881836,
      "learning_rate": 4.914066830360844e-06,
      "loss": 0.8012,
      "step": 7256
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.3471221923828125,
      "learning_rate": 4.912905687991217e-06,
      "loss": 0.145,
      "step": 7257
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.648652076721191,
      "learning_rate": 4.911744550319999e-06,
      "loss": 1.0348,
      "step": 7258
    },
    {
      "epoch": 0.52,
      "grad_norm": 17.26963996887207,
      "learning_rate": 4.910583417409828e-06,
      "loss": 0.9973,
      "step": 7259
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.62273120880127,
      "learning_rate": 4.909422289323344e-06,
      "loss": 0.604,
      "step": 7260
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.930511474609375,
      "learning_rate": 4.908261166123186e-06,
      "loss": 0.8246,
      "step": 7261
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.802840709686279,
      "learning_rate": 4.907100047871989e-06,
      "loss": 0.9032,
      "step": 7262
    },
    {
      "epoch": 0.52,
      "grad_norm": 21.51910972595215,
      "learning_rate": 4.905938934632392e-06,
      "loss": 1.275,
      "step": 7263
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.018213272094727,
      "learning_rate": 4.904777826467035e-06,
      "loss": 0.3749,
      "step": 7264
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.499316692352295,
      "learning_rate": 4.903616723438553e-06,
      "loss": 0.1999,
      "step": 7265
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.991150379180908,
      "learning_rate": 4.902455625609584e-06,
      "loss": 0.7611,
      "step": 7266
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.058401107788086,
      "learning_rate": 4.9012945330427645e-06,
      "loss": 0.6091,
      "step": 7267
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.613283634185791,
      "learning_rate": 4.900133445800731e-06,
      "loss": 0.1778,
      "step": 7268
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.278618812561035,
      "learning_rate": 4.898972363946121e-06,
      "loss": 0.8355,
      "step": 7269
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.4764556884765625,
      "learning_rate": 4.897811287541569e-06,
      "loss": 0.1232,
      "step": 7270
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.414459705352783,
      "learning_rate": 4.896650216649711e-06,
      "loss": 0.0939,
      "step": 7271
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.9257726669311523,
      "learning_rate": 4.895489151333184e-06,
      "loss": 0.273,
      "step": 7272
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.1639418601989746,
      "learning_rate": 4.894328091654622e-06,
      "loss": 0.1584,
      "step": 7273
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.820111274719238,
      "learning_rate": 4.893167037676658e-06,
      "loss": 0.9133,
      "step": 7274
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.553392171859741,
      "learning_rate": 4.892005989461929e-06,
      "loss": 0.1044,
      "step": 7275
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5639238357543945,
      "learning_rate": 4.890844947073066e-06,
      "loss": 0.1612,
      "step": 7276
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.7988333702087402,
      "learning_rate": 4.8896839105727055e-06,
      "loss": 0.1153,
      "step": 7277
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.797440528869629,
      "learning_rate": 4.888522880023482e-06,
      "loss": 0.2357,
      "step": 7278
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9196057319641113,
      "learning_rate": 4.887361855488026e-06,
      "loss": 0.7932,
      "step": 7279
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.8914480209350586,
      "learning_rate": 4.886200837028972e-06,
      "loss": 0.7385,
      "step": 7280
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.489880084991455,
      "learning_rate": 4.885039824708952e-06,
      "loss": 0.7717,
      "step": 7281
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.7296056747436523,
      "learning_rate": 4.883878818590598e-06,
      "loss": 0.1369,
      "step": 7282
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.3481521606445312,
      "learning_rate": 4.8827178187365425e-06,
      "loss": 0.1897,
      "step": 7283
    },
    {
      "epoch": 0.52,
      "grad_norm": 14.13259506225586,
      "learning_rate": 4.881556825209417e-06,
      "loss": 0.9121,
      "step": 7284
    },
    {
      "epoch": 0.52,
      "grad_norm": 19.25196647644043,
      "learning_rate": 4.880395838071853e-06,
      "loss": 0.9928,
      "step": 7285
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.940207481384277,
      "learning_rate": 4.8792348573864795e-06,
      "loss": 0.8797,
      "step": 7286
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.617687225341797,
      "learning_rate": 4.87807388321593e-06,
      "loss": 0.1629,
      "step": 7287
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0624430179595947,
      "learning_rate": 4.87691291562283e-06,
      "loss": 0.1295,
      "step": 7288
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.305229187011719,
      "learning_rate": 4.875751954669815e-06,
      "loss": 0.498,
      "step": 7289
    },
    {
      "epoch": 0.52,
      "grad_norm": 16.25248908996582,
      "learning_rate": 4.874591000419511e-06,
      "loss": 0.8617,
      "step": 7290
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.182856559753418,
      "learning_rate": 4.873430052934547e-06,
      "loss": 0.6855,
      "step": 7291
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.81890869140625,
      "learning_rate": 4.872269112277554e-06,
      "loss": 0.9289,
      "step": 7292
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.158269882202148,
      "learning_rate": 4.871108178511157e-06,
      "loss": 0.8062,
      "step": 7293
    },
    {
      "epoch": 0.52,
      "grad_norm": 17.821142196655273,
      "learning_rate": 4.8699472516979875e-06,
      "loss": 1.2682,
      "step": 7294
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.423684120178223,
      "learning_rate": 4.868786331900671e-06,
      "loss": 0.9939,
      "step": 7295
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.9522719383239746,
      "learning_rate": 4.867625419181834e-06,
      "loss": 0.2208,
      "step": 7296
    },
    {
      "epoch": 0.52,
      "grad_norm": 23.724491119384766,
      "learning_rate": 4.8664645136041075e-06,
      "loss": 1.3821,
      "step": 7297
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.819512128829956,
      "learning_rate": 4.865303615230114e-06,
      "loss": 0.2111,
      "step": 7298
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.560331344604492,
      "learning_rate": 4.864142724122479e-06,
      "loss": 0.9115,
      "step": 7299
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5789225101470947,
      "learning_rate": 4.86298184034383e-06,
      "loss": 0.1933,
      "step": 7300
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.863422870635986,
      "learning_rate": 4.861820963956794e-06,
      "loss": 0.8192,
      "step": 7301
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.539844274520874,
      "learning_rate": 4.860660095023992e-06,
      "loss": 0.1175,
      "step": 7302
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.711435556411743,
      "learning_rate": 4.859499233608052e-06,
      "loss": 0.1638,
      "step": 7303
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6296508312225342,
      "learning_rate": 4.858338379771595e-06,
      "loss": 0.0597,
      "step": 7304
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.584335803985596,
      "learning_rate": 4.857177533577247e-06,
      "loss": 0.8064,
      "step": 7305
    },
    {
      "epoch": 0.52,
      "grad_norm": 15.074847221374512,
      "learning_rate": 4.8560166950876305e-06,
      "loss": 0.67,
      "step": 7306
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.0926246643066406,
      "learning_rate": 4.854855864365367e-06,
      "loss": 0.1385,
      "step": 7307
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.9732861518859863,
      "learning_rate": 4.853695041473082e-06,
      "loss": 0.2271,
      "step": 7308
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.614563465118408,
      "learning_rate": 4.852534226473396e-06,
      "loss": 0.7372,
      "step": 7309
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1350250244140625,
      "learning_rate": 4.85137341942893e-06,
      "loss": 0.1032,
      "step": 7310
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.410783052444458,
      "learning_rate": 4.8502126204023044e-06,
      "loss": 0.1378,
      "step": 7311
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8183958530426025,
      "learning_rate": 4.849051829456142e-06,
      "loss": 0.1633,
      "step": 7312
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.719205856323242,
      "learning_rate": 4.84789104665306e-06,
      "loss": 0.2001,
      "step": 7313
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.063973903656006,
      "learning_rate": 4.846730272055682e-06,
      "loss": 0.8218,
      "step": 7314
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.422814130783081,
      "learning_rate": 4.845569505726625e-06,
      "loss": 0.2278,
      "step": 7315
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.764432907104492,
      "learning_rate": 4.844408747728509e-06,
      "loss": 0.9361,
      "step": 7316
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.806976556777954,
      "learning_rate": 4.843247998123953e-06,
      "loss": 0.8948,
      "step": 7317
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.978283643722534,
      "learning_rate": 4.842087256975572e-06,
      "loss": 0.1379,
      "step": 7318
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.566009044647217,
      "learning_rate": 4.8409265243459865e-06,
      "loss": 0.1354,
      "step": 7319
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.766648769378662,
      "learning_rate": 4.839765800297814e-06,
      "loss": 0.1378,
      "step": 7320
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.49294114112854,
      "learning_rate": 4.838605084893671e-06,
      "loss": 0.1283,
      "step": 7321
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.524990081787109,
      "learning_rate": 4.837444378196171e-06,
      "loss": 0.9146,
      "step": 7322
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.710254669189453,
      "learning_rate": 4.836283680267931e-06,
      "loss": 0.8498,
      "step": 7323
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.282626152038574,
      "learning_rate": 4.835122991171567e-06,
      "loss": 0.1231,
      "step": 7324
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.7361412048339844,
      "learning_rate": 4.833962310969694e-06,
      "loss": 0.5813,
      "step": 7325
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.355899810791016,
      "learning_rate": 4.832801639724927e-06,
      "loss": 0.5418,
      "step": 7326
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2246673107147217,
      "learning_rate": 4.831640977499876e-06,
      "loss": 0.0873,
      "step": 7327
    },
    {
      "epoch": 0.53,
      "grad_norm": 13.125011444091797,
      "learning_rate": 4.830480324357158e-06,
      "loss": 0.7457,
      "step": 7328
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.316063642501831,
      "learning_rate": 4.829319680359386e-06,
      "loss": 0.1154,
      "step": 7329
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.740203380584717,
      "learning_rate": 4.82815904556917e-06,
      "loss": 0.1484,
      "step": 7330
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.534353256225586,
      "learning_rate": 4.826998420049124e-06,
      "loss": 0.7212,
      "step": 7331
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.9119699001312256,
      "learning_rate": 4.825837803861857e-06,
      "loss": 0.9326,
      "step": 7332
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.570917129516602,
      "learning_rate": 4.824677197069984e-06,
      "loss": 0.8652,
      "step": 7333
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4064173698425293,
      "learning_rate": 4.82351659973611e-06,
      "loss": 0.0988,
      "step": 7334
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.988138198852539,
      "learning_rate": 4.822356011922849e-06,
      "loss": 0.6845,
      "step": 7335
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.6255099773406982,
      "learning_rate": 4.821195433692807e-06,
      "loss": 0.6694,
      "step": 7336
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.525524377822876,
      "learning_rate": 4.820034865108595e-06,
      "loss": 0.213,
      "step": 7337
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.680835008621216,
      "learning_rate": 4.818874306232822e-06,
      "loss": 0.1225,
      "step": 7338
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.9334981441497803,
      "learning_rate": 4.817713757128093e-06,
      "loss": 0.7304,
      "step": 7339
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.0671041011810303,
      "learning_rate": 4.816553217857018e-06,
      "loss": 0.1877,
      "step": 7340
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.195441246032715,
      "learning_rate": 4.815392688482202e-06,
      "loss": 0.1069,
      "step": 7341
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.203578472137451,
      "learning_rate": 4.81423216906625e-06,
      "loss": 0.1524,
      "step": 7342
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.027698993682861,
      "learning_rate": 4.813071659671773e-06,
      "loss": 0.8343,
      "step": 7343
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.744745969772339,
      "learning_rate": 4.81191116036137e-06,
      "loss": 0.1319,
      "step": 7344
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.402005195617676,
      "learning_rate": 4.81075067119765e-06,
      "loss": 0.1043,
      "step": 7345
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.19947624206543,
      "learning_rate": 4.809590192243213e-06,
      "loss": 0.8111,
      "step": 7346
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3385558128356934,
      "learning_rate": 4.808429723560665e-06,
      "loss": 0.1126,
      "step": 7347
    },
    {
      "epoch": 0.53,
      "grad_norm": 17.13032341003418,
      "learning_rate": 4.807269265212607e-06,
      "loss": 1.3099,
      "step": 7348
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.060155153274536,
      "learning_rate": 4.806108817261645e-06,
      "loss": 0.1723,
      "step": 7349
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9677830934524536,
      "learning_rate": 4.804948379770376e-06,
      "loss": 0.0843,
      "step": 7350
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.9208433628082275,
      "learning_rate": 4.803787952801404e-06,
      "loss": 0.2114,
      "step": 7351
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2701289653778076,
      "learning_rate": 4.802627536417331e-06,
      "loss": 0.1508,
      "step": 7352
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3768725395202637,
      "learning_rate": 4.801467130680753e-06,
      "loss": 0.0584,
      "step": 7353
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.257904052734375,
      "learning_rate": 4.8003067356542745e-06,
      "loss": 0.9159,
      "step": 7354
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.935381531715393,
      "learning_rate": 4.7991463514004895e-06,
      "loss": 0.0748,
      "step": 7355
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6555960178375244,
      "learning_rate": 4.797985977981999e-06,
      "loss": 0.1319,
      "step": 7356
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.927188873291016,
      "learning_rate": 4.7968256154614015e-06,
      "loss": 0.6096,
      "step": 7357
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.8532443046569824,
      "learning_rate": 4.7956652639012934e-06,
      "loss": 0.6854,
      "step": 7358
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.9813857078552246,
      "learning_rate": 4.794504923364269e-06,
      "loss": 0.1671,
      "step": 7359
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8731030225753784,
      "learning_rate": 4.793344593912928e-06,
      "loss": 0.0722,
      "step": 7360
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.430633544921875,
      "learning_rate": 4.7921842756098615e-06,
      "loss": 0.3441,
      "step": 7361
    },
    {
      "epoch": 0.53,
      "grad_norm": 20.640878677368164,
      "learning_rate": 4.791023968517668e-06,
      "loss": 1.0895,
      "step": 7362
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.76997971534729,
      "learning_rate": 4.78986367269894e-06,
      "loss": 0.1486,
      "step": 7363
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.59625244140625,
      "learning_rate": 4.788703388216272e-06,
      "loss": 1.013,
      "step": 7364
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9734137058258057,
      "learning_rate": 4.787543115132255e-06,
      "loss": 0.1101,
      "step": 7365
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.549994468688965,
      "learning_rate": 4.786382853509484e-06,
      "loss": 0.8606,
      "step": 7366
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.341590166091919,
      "learning_rate": 4.7852226034105495e-06,
      "loss": 0.2679,
      "step": 7367
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.0863447189331055,
      "learning_rate": 4.784062364898043e-06,
      "loss": 0.218,
      "step": 7368
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.4498021602630615,
      "learning_rate": 4.782902138034555e-06,
      "loss": 0.2876,
      "step": 7369
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.230294227600098,
      "learning_rate": 4.781741922882673e-06,
      "loss": 1.023,
      "step": 7370
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.065986633300781,
      "learning_rate": 4.780581719504988e-06,
      "loss": 0.2364,
      "step": 7371
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.893051624298096,
      "learning_rate": 4.7794215279640906e-06,
      "loss": 0.3845,
      "step": 7372
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.71608304977417,
      "learning_rate": 4.778261348322566e-06,
      "loss": 0.1102,
      "step": 7373
    },
    {
      "epoch": 0.53,
      "grad_norm": 13.933046340942383,
      "learning_rate": 4.777101180643003e-06,
      "loss": 0.6752,
      "step": 7374
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.3766043186187744,
      "learning_rate": 4.775941024987985e-06,
      "loss": 0.1086,
      "step": 7375
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2888684272766113,
      "learning_rate": 4.774780881420102e-06,
      "loss": 0.1544,
      "step": 7376
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.692039966583252,
      "learning_rate": 4.773620750001939e-06,
      "loss": 0.1085,
      "step": 7377
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.263065338134766,
      "learning_rate": 4.772460630796079e-06,
      "loss": 0.9104,
      "step": 7378
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0483486652374268,
      "learning_rate": 4.771300523865106e-06,
      "loss": 0.0839,
      "step": 7379
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.838721990585327,
      "learning_rate": 4.770140429271606e-06,
      "loss": 0.801,
      "step": 7380
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.965667724609375,
      "learning_rate": 4.7689803470781595e-06,
      "loss": 0.6872,
      "step": 7381
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7876205444335938,
      "learning_rate": 4.767820277347347e-06,
      "loss": 0.0503,
      "step": 7382
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.570329189300537,
      "learning_rate": 4.766660220141753e-06,
      "loss": 0.1664,
      "step": 7383
    },
    {
      "epoch": 0.53,
      "grad_norm": 33.741912841796875,
      "learning_rate": 4.765500175523957e-06,
      "loss": 1.0118,
      "step": 7384
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.128089904785156,
      "learning_rate": 4.764340143556537e-06,
      "loss": 0.9841,
      "step": 7385
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5838935375213623,
      "learning_rate": 4.763180124302076e-06,
      "loss": 0.0828,
      "step": 7386
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.583067893981934,
      "learning_rate": 4.76202011782315e-06,
      "loss": 1.0345,
      "step": 7387
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6024177074432373,
      "learning_rate": 4.7608601241823385e-06,
      "loss": 0.1295,
      "step": 7388
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.032801628112793,
      "learning_rate": 4.759700143442218e-06,
      "loss": 0.3914,
      "step": 7389
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.528916835784912,
      "learning_rate": 4.758540175665365e-06,
      "loss": 0.1352,
      "step": 7390
    },
    {
      "epoch": 0.53,
      "grad_norm": 15.251715660095215,
      "learning_rate": 4.757380220914355e-06,
      "loss": 0.9361,
      "step": 7391
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.2191386222839355,
      "learning_rate": 4.756220279251764e-06,
      "loss": 1.1059,
      "step": 7392
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8746534585952759,
      "learning_rate": 4.755060350740168e-06,
      "loss": 0.076,
      "step": 7393
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.626680374145508,
      "learning_rate": 4.753900435442137e-06,
      "loss": 0.6029,
      "step": 7394
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.1843032836914062,
      "learning_rate": 4.752740533420247e-06,
      "loss": 0.1933,
      "step": 7395
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.064257621765137,
      "learning_rate": 4.751580644737069e-06,
      "loss": 0.8531,
      "step": 7396
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.371893882751465,
      "learning_rate": 4.750420769455174e-06,
      "loss": 1.0453,
      "step": 7397
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.947539329528809,
      "learning_rate": 4.749260907637134e-06,
      "loss": 0.7014,
      "step": 7398
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.420722007751465,
      "learning_rate": 4.748101059345519e-06,
      "loss": 0.5252,
      "step": 7399
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.60419225692749,
      "learning_rate": 4.7469412246428996e-06,
      "loss": 0.5248,
      "step": 7400
    },
    {
      "epoch": 0.53,
      "grad_norm": 17.89799690246582,
      "learning_rate": 4.745781403591841e-06,
      "loss": 0.6799,
      "step": 7401
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.2767715454101562,
      "learning_rate": 4.744621596254914e-06,
      "loss": 0.2073,
      "step": 7402
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0336763858795166,
      "learning_rate": 4.743461802694687e-06,
      "loss": 0.1165,
      "step": 7403
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.5433268547058105,
      "learning_rate": 4.742302022973723e-06,
      "loss": 0.1278,
      "step": 7404
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.2052135467529297,
      "learning_rate": 4.741142257154591e-06,
      "loss": 0.1737,
      "step": 7405
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.32406997680664,
      "learning_rate": 4.739982505299855e-06,
      "loss": 0.6857,
      "step": 7406
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.5746989250183105,
      "learning_rate": 4.738822767472076e-06,
      "loss": 0.9767,
      "step": 7407
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.8016464710235596,
      "learning_rate": 4.7376630437338205e-06,
      "loss": 0.8374,
      "step": 7408
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.681565523147583,
      "learning_rate": 4.736503334147652e-06,
      "loss": 0.1448,
      "step": 7409
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3101823329925537,
      "learning_rate": 4.735343638776131e-06,
      "loss": 0.167,
      "step": 7410
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.504725933074951,
      "learning_rate": 4.734183957681818e-06,
      "loss": 0.5331,
      "step": 7411
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.442905902862549,
      "learning_rate": 4.733024290927275e-06,
      "loss": 0.1474,
      "step": 7412
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.566535949707031,
      "learning_rate": 4.73186463857506e-06,
      "loss": 0.9558,
      "step": 7413
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.250915765762329,
      "learning_rate": 4.730705000687734e-06,
      "loss": 0.0894,
      "step": 7414
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.149133682250977,
      "learning_rate": 4.729545377327853e-06,
      "loss": 0.5822,
      "step": 7415
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.8025083541870117,
      "learning_rate": 4.728385768557976e-06,
      "loss": 0.1237,
      "step": 7416
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7286906242370605,
      "learning_rate": 4.727226174440658e-06,
      "loss": 0.1129,
      "step": 7417
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.784600734710693,
      "learning_rate": 4.726066595038457e-06,
      "loss": 0.3095,
      "step": 7418
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.022047281265259,
      "learning_rate": 4.724907030413925e-06,
      "loss": 0.1396,
      "step": 7419
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.16513729095459,
      "learning_rate": 4.723747480629619e-06,
      "loss": 0.7443,
      "step": 7420
    },
    {
      "epoch": 0.53,
      "grad_norm": 13.272684097290039,
      "learning_rate": 4.722587945748088e-06,
      "loss": 0.8707,
      "step": 7421
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.497267484664917,
      "learning_rate": 4.721428425831888e-06,
      "loss": 0.0763,
      "step": 7422
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7743823528289795,
      "learning_rate": 4.720268920943571e-06,
      "loss": 0.1205,
      "step": 7423
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.879540920257568,
      "learning_rate": 4.719109431145686e-06,
      "loss": 0.954,
      "step": 7424
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.000189781188965,
      "learning_rate": 4.717949956500785e-06,
      "loss": 0.5047,
      "step": 7425
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.4475018978118896,
      "learning_rate": 4.716790497071414e-06,
      "loss": 0.1288,
      "step": 7426
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.183046340942383,
      "learning_rate": 4.715631052920126e-06,
      "loss": 0.9401,
      "step": 7427
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.9754831790924072,
      "learning_rate": 4.714471624109465e-06,
      "loss": 0.1774,
      "step": 7428
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.528312683105469,
      "learning_rate": 4.713312210701979e-06,
      "loss": 0.6258,
      "step": 7429
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5239861011505127,
      "learning_rate": 4.712152812760217e-06,
      "loss": 0.0791,
      "step": 7430
    },
    {
      "epoch": 0.53,
      "grad_norm": 15.385984420776367,
      "learning_rate": 4.710993430346718e-06,
      "loss": 1.0023,
      "step": 7431
    },
    {
      "epoch": 0.53,
      "grad_norm": 13.59044361114502,
      "learning_rate": 4.709834063524031e-06,
      "loss": 0.6298,
      "step": 7432
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.313578605651855,
      "learning_rate": 4.708674712354697e-06,
      "loss": 0.4858,
      "step": 7433
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0939836502075195,
      "learning_rate": 4.70751537690126e-06,
      "loss": 0.0861,
      "step": 7434
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.6236319541931152,
      "learning_rate": 4.706356057226261e-06,
      "loss": 0.2378,
      "step": 7435
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.654639959335327,
      "learning_rate": 4.7051967533922405e-06,
      "loss": 0.2509,
      "step": 7436
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.264331817626953,
      "learning_rate": 4.7040374654617406e-06,
      "loss": 0.2018,
      "step": 7437
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.983928918838501,
      "learning_rate": 4.702878193497298e-06,
      "loss": 0.3033,
      "step": 7438
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.464517593383789,
      "learning_rate": 4.701718937561454e-06,
      "loss": 0.4759,
      "step": 7439
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.248017311096191,
      "learning_rate": 4.700559697716742e-06,
      "loss": 0.7115,
      "step": 7440
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.6202592849731445,
      "learning_rate": 4.699400474025702e-06,
      "loss": 0.8169,
      "step": 7441
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3997952938079834,
      "learning_rate": 4.69824126655087e-06,
      "loss": 0.1175,
      "step": 7442
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.344099044799805,
      "learning_rate": 4.697082075354778e-06,
      "loss": 0.4762,
      "step": 7443
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.217241287231445,
      "learning_rate": 4.695922900499961e-06,
      "loss": 0.679,
      "step": 7444
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.092809677124023,
      "learning_rate": 4.694763742048952e-06,
      "loss": 1.0219,
      "step": 7445
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.471895217895508,
      "learning_rate": 4.6936046000642856e-06,
      "loss": 1.1788,
      "step": 7446
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.640007019042969,
      "learning_rate": 4.69244547460849e-06,
      "loss": 1.42,
      "step": 7447
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.9149539470672607,
      "learning_rate": 4.691286365744098e-06,
      "loss": 0.2142,
      "step": 7448
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.334506034851074,
      "learning_rate": 4.690127273533636e-06,
      "loss": 0.2741,
      "step": 7449
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.9606757164001465,
      "learning_rate": 4.688968198039636e-06,
      "loss": 0.1786,
      "step": 7450
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4089367389678955,
      "learning_rate": 4.687809139324624e-06,
      "loss": 0.0968,
      "step": 7451
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.0501396656036377,
      "learning_rate": 4.686650097451127e-06,
      "loss": 0.1318,
      "step": 7452
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.509982109069824,
      "learning_rate": 4.685491072481673e-06,
      "loss": 0.1304,
      "step": 7453
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.243470191955566,
      "learning_rate": 4.684332064478784e-06,
      "loss": 1.0566,
      "step": 7454
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.8492116928100586,
      "learning_rate": 4.683173073504986e-06,
      "loss": 0.1872,
      "step": 7455
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.298110008239746,
      "learning_rate": 4.6820140996228e-06,
      "loss": 0.0887,
      "step": 7456
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.225672721862793,
      "learning_rate": 4.680855142894751e-06,
      "loss": 0.7069,
      "step": 7457
    },
    {
      "epoch": 0.53,
      "grad_norm": 23.040746688842773,
      "learning_rate": 4.679696203383358e-06,
      "loss": 0.6467,
      "step": 7458
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.4824137687683105,
      "learning_rate": 4.678537281151141e-06,
      "loss": 0.6352,
      "step": 7459
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.4193196296691895,
      "learning_rate": 4.677378376260623e-06,
      "loss": 0.8427,
      "step": 7460
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.708413600921631,
      "learning_rate": 4.67621948877432e-06,
      "loss": 0.8581,
      "step": 7461
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5114994049072266,
      "learning_rate": 4.67506061875475e-06,
      "loss": 0.13,
      "step": 7462
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.954113721847534,
      "learning_rate": 4.67390176626443e-06,
      "loss": 0.0758,
      "step": 7463
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.26424241065979,
      "learning_rate": 4.672742931365874e-06,
      "loss": 0.2131,
      "step": 7464
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.906162738800049,
      "learning_rate": 4.671584114121599e-06,
      "loss": 0.1466,
      "step": 7465
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.4115161895751953,
      "learning_rate": 4.67042531459412e-06,
      "loss": 0.1012,
      "step": 7466
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.772655963897705,
      "learning_rate": 4.669266532845945e-06,
      "loss": 1.0499,
      "step": 7467
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.587648868560791,
      "learning_rate": 4.6681077689395896e-06,
      "loss": 0.1039,
      "step": 7468
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.47299575805664,
      "learning_rate": 4.666949022937564e-06,
      "loss": 0.6996,
      "step": 7469
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.396627426147461,
      "learning_rate": 4.665790294902377e-06,
      "loss": 0.6661,
      "step": 7470
    },
    {
      "epoch": 0.54,
      "grad_norm": 15.208300590515137,
      "learning_rate": 4.66463158489654e-06,
      "loss": 1.4474,
      "step": 7471
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.454220294952393,
      "learning_rate": 4.663472892982558e-06,
      "loss": 0.7516,
      "step": 7472
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.126234531402588,
      "learning_rate": 4.662314219222939e-06,
      "loss": 0.8977,
      "step": 7473
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.630902051925659,
      "learning_rate": 4.661155563680192e-06,
      "loss": 0.8857,
      "step": 7474
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.374330520629883,
      "learning_rate": 4.6599969264168185e-06,
      "loss": 0.0909,
      "step": 7475
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.857545852661133,
      "learning_rate": 4.658838307495324e-06,
      "loss": 0.4801,
      "step": 7476
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.03532886505127,
      "learning_rate": 4.657679706978211e-06,
      "loss": 0.5988,
      "step": 7477
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.872957706451416,
      "learning_rate": 4.656521124927984e-06,
      "loss": 0.1433,
      "step": 7478
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.517192840576172,
      "learning_rate": 4.65536256140714e-06,
      "loss": 0.1155,
      "step": 7479
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.054446220397949,
      "learning_rate": 4.654204016478183e-06,
      "loss": 0.2175,
      "step": 7480
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.094963312149048,
      "learning_rate": 4.653045490203609e-06,
      "loss": 0.2335,
      "step": 7481
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.5890274047851562,
      "learning_rate": 4.651886982645917e-06,
      "loss": 0.0955,
      "step": 7482
    },
    {
      "epoch": 0.54,
      "grad_norm": 15.52548599243164,
      "learning_rate": 4.650728493867606e-06,
      "loss": 1.0817,
      "step": 7483
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.431488037109375,
      "learning_rate": 4.64957002393117e-06,
      "loss": 0.8227,
      "step": 7484
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7445380687713623,
      "learning_rate": 4.6484115728991054e-06,
      "loss": 0.1262,
      "step": 7485
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.122747421264648,
      "learning_rate": 4.647253140833904e-06,
      "loss": 0.4354,
      "step": 7486
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5824480056762695,
      "learning_rate": 4.646094727798061e-06,
      "loss": 0.1203,
      "step": 7487
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.952082872390747,
      "learning_rate": 4.644936333854069e-06,
      "loss": 0.136,
      "step": 7488
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.0147223472595215,
      "learning_rate": 4.643777959064417e-06,
      "loss": 0.1966,
      "step": 7489
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.391681671142578,
      "learning_rate": 4.642619603491598e-06,
      "loss": 1.2276,
      "step": 7490
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.975040435791016,
      "learning_rate": 4.641461267198097e-06,
      "loss": 0.509,
      "step": 7491
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.584404945373535,
      "learning_rate": 4.640302950246402e-06,
      "loss": 0.8125,
      "step": 7492
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.46754264831543,
      "learning_rate": 4.639144652699002e-06,
      "loss": 0.6575,
      "step": 7493
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.803123950958252,
      "learning_rate": 4.6379863746183826e-06,
      "loss": 0.9607,
      "step": 7494
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.041860580444336,
      "learning_rate": 4.6368281160670276e-06,
      "loss": 0.9286,
      "step": 7495
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2823007106781006,
      "learning_rate": 4.635669877107421e-06,
      "loss": 0.072,
      "step": 7496
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5477467775344849,
      "learning_rate": 4.634511657802046e-06,
      "loss": 0.048,
      "step": 7497
    },
    {
      "epoch": 0.54,
      "grad_norm": 19.102340698242188,
      "learning_rate": 4.633353458213384e-06,
      "loss": 1.1065,
      "step": 7498
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8958102464675903,
      "learning_rate": 4.632195278403914e-06,
      "loss": 0.0428,
      "step": 7499
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.152251243591309,
      "learning_rate": 4.6310371184361165e-06,
      "loss": 0.9609,
      "step": 7500
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.763772964477539,
      "learning_rate": 4.629878978372469e-06,
      "loss": 0.4642,
      "step": 7501
    },
    {
      "epoch": 0.54,
      "grad_norm": 13.203325271606445,
      "learning_rate": 4.628720858275454e-06,
      "loss": 1.2896,
      "step": 7502
    },
    {
      "epoch": 0.54,
      "grad_norm": 13.420275688171387,
      "learning_rate": 4.62756275820754e-06,
      "loss": 0.5954,
      "step": 7503
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.72457504272461,
      "learning_rate": 4.626404678231206e-06,
      "loss": 0.8321,
      "step": 7504
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.04948616027832,
      "learning_rate": 4.625246618408925e-06,
      "loss": 0.9483,
      "step": 7505
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.346686363220215,
      "learning_rate": 4.62408857880317e-06,
      "loss": 0.5541,
      "step": 7506
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.9291677474975586,
      "learning_rate": 4.6229305594764135e-06,
      "loss": 0.6167,
      "step": 7507
    },
    {
      "epoch": 0.54,
      "grad_norm": 14.333006858825684,
      "learning_rate": 4.6217725604911274e-06,
      "loss": 1.0705,
      "step": 7508
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.700196266174316,
      "learning_rate": 4.620614581909778e-06,
      "loss": 0.8549,
      "step": 7509
    },
    {
      "epoch": 0.54,
      "grad_norm": 17.526697158813477,
      "learning_rate": 4.6194566237948365e-06,
      "loss": 1.0294,
      "step": 7510
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.2947416305542,
      "learning_rate": 4.618298686208771e-06,
      "loss": 0.4905,
      "step": 7511
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.273394584655762,
      "learning_rate": 4.617140769214045e-06,
      "loss": 0.9807,
      "step": 7512
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.5849969387054443,
      "learning_rate": 4.615982872873127e-06,
      "loss": 0.2311,
      "step": 7513
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.85565185546875,
      "learning_rate": 4.614824997248479e-06,
      "loss": 1.5132,
      "step": 7514
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.692910194396973,
      "learning_rate": 4.613667142402563e-06,
      "loss": 0.4646,
      "step": 7515
    },
    {
      "epoch": 0.54,
      "grad_norm": 13.866721153259277,
      "learning_rate": 4.6125093083978435e-06,
      "loss": 0.9168,
      "step": 7516
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.438631057739258,
      "learning_rate": 4.6113514952967795e-06,
      "loss": 0.067,
      "step": 7517
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.362966537475586,
      "learning_rate": 4.610193703161832e-06,
      "loss": 0.6159,
      "step": 7518
    },
    {
      "epoch": 0.54,
      "grad_norm": 19.01886558532715,
      "learning_rate": 4.6090359320554575e-06,
      "loss": 1.0445,
      "step": 7519
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.9225504398345947,
      "learning_rate": 4.607878182040115e-06,
      "loss": 0.1786,
      "step": 7520
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.671767234802246,
      "learning_rate": 4.6067204531782605e-06,
      "loss": 0.7266,
      "step": 7521
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.552061080932617,
      "learning_rate": 4.6055627455323495e-06,
      "loss": 0.5662,
      "step": 7522
    },
    {
      "epoch": 0.54,
      "grad_norm": 14.102363586425781,
      "learning_rate": 4.6044050591648345e-06,
      "loss": 0.8845,
      "step": 7523
    },
    {
      "epoch": 0.54,
      "grad_norm": 16.490337371826172,
      "learning_rate": 4.603247394138168e-06,
      "loss": 1.0919,
      "step": 7524
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7809057235717773,
      "learning_rate": 4.602089750514805e-06,
      "loss": 0.1151,
      "step": 7525
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.047035217285156,
      "learning_rate": 4.6009321283571954e-06,
      "loss": 0.6511,
      "step": 7526
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.540555000305176,
      "learning_rate": 4.599774527727784e-06,
      "loss": 1.0319,
      "step": 7527
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.0563836097717285,
      "learning_rate": 4.598616948689023e-06,
      "loss": 0.6297,
      "step": 7528
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.6848273277282715,
      "learning_rate": 4.597459391303357e-06,
      "loss": 0.6199,
      "step": 7529
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.9525110721588135,
      "learning_rate": 4.596301855633233e-06,
      "loss": 0.1753,
      "step": 7530
    },
    {
      "epoch": 0.54,
      "grad_norm": 31.949308395385742,
      "learning_rate": 4.595144341741097e-06,
      "loss": 0.7605,
      "step": 7531
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.5879604816436768,
      "learning_rate": 4.59398684968939e-06,
      "loss": 0.1443,
      "step": 7532
    },
    {
      "epoch": 0.54,
      "grad_norm": 14.465895652770996,
      "learning_rate": 4.592829379540558e-06,
      "loss": 0.7288,
      "step": 7533
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.393710613250732,
      "learning_rate": 4.591671931357038e-06,
      "loss": 0.8219,
      "step": 7534
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.587890386581421,
      "learning_rate": 4.590514505201271e-06,
      "loss": 0.0789,
      "step": 7535
    },
    {
      "epoch": 0.54,
      "grad_norm": 22.213878631591797,
      "learning_rate": 4.589357101135699e-06,
      "loss": 1.5919,
      "step": 7536
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.436047554016113,
      "learning_rate": 4.588199719222754e-06,
      "loss": 0.4871,
      "step": 7537
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.582922458648682,
      "learning_rate": 4.5870423595248795e-06,
      "loss": 0.7667,
      "step": 7538
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.534949541091919,
      "learning_rate": 4.585885022104503e-06,
      "loss": 0.1846,
      "step": 7539
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.98353099822998,
      "learning_rate": 4.5847277070240635e-06,
      "loss": 0.9007,
      "step": 7540
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.764923095703125,
      "learning_rate": 4.583570414345991e-06,
      "loss": 0.7814,
      "step": 7541
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.128210544586182,
      "learning_rate": 4.5824131441327205e-06,
      "loss": 0.4246,
      "step": 7542
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.337153911590576,
      "learning_rate": 4.581255896446678e-06,
      "loss": 0.737,
      "step": 7543
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.431185007095337,
      "learning_rate": 4.580098671350296e-06,
      "loss": 0.1526,
      "step": 7544
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.6840248107910156,
      "learning_rate": 4.578941468906001e-06,
      "loss": 0.0483,
      "step": 7545
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.940401792526245,
      "learning_rate": 4.57778428917622e-06,
      "loss": 0.1718,
      "step": 7546
    },
    {
      "epoch": 0.54,
      "grad_norm": 13.464526176452637,
      "learning_rate": 4.576627132223379e-06,
      "loss": 0.9419,
      "step": 7547
    },
    {
      "epoch": 0.54,
      "grad_norm": 18.58356475830078,
      "learning_rate": 4.5754699981099e-06,
      "loss": 0.7376,
      "step": 7548
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.090775012969971,
      "learning_rate": 4.574312886898209e-06,
      "loss": 0.1876,
      "step": 7549
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.580827713012695,
      "learning_rate": 4.573155798650728e-06,
      "loss": 0.6137,
      "step": 7550
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2921431064605713,
      "learning_rate": 4.571998733429874e-06,
      "loss": 0.1409,
      "step": 7551
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.333041191101074,
      "learning_rate": 4.5708416912980684e-06,
      "loss": 0.166,
      "step": 7552
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7163400650024414,
      "learning_rate": 4.569684672317729e-06,
      "loss": 0.1349,
      "step": 7553
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.0383951663970947,
      "learning_rate": 4.568527676551273e-06,
      "loss": 0.0784,
      "step": 7554
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.095108985900879,
      "learning_rate": 4.567370704061115e-06,
      "loss": 0.0741,
      "step": 7555
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.63175630569458,
      "learning_rate": 4.566213754909671e-06,
      "loss": 0.2273,
      "step": 7556
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.859145164489746,
      "learning_rate": 4.565056829159352e-06,
      "loss": 0.7506,
      "step": 7557
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0947487354278564,
      "learning_rate": 4.5638999268725706e-06,
      "loss": 0.0738,
      "step": 7558
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.9660468101501465,
      "learning_rate": 4.562743048111739e-06,
      "loss": 0.4453,
      "step": 7559
    },
    {
      "epoch": 0.54,
      "grad_norm": 13.526263236999512,
      "learning_rate": 4.561586192939263e-06,
      "loss": 1.0759,
      "step": 7560
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.71898353099823,
      "learning_rate": 4.560429361417553e-06,
      "loss": 0.0524,
      "step": 7561
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.4427578449249268,
      "learning_rate": 4.559272553609016e-06,
      "loss": 0.1034,
      "step": 7562
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.457408905029297,
      "learning_rate": 4.558115769576057e-06,
      "loss": 0.8096,
      "step": 7563
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.405695915222168,
      "learning_rate": 4.556959009381077e-06,
      "loss": 1.0299,
      "step": 7564
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.590616226196289,
      "learning_rate": 4.5558022730864835e-06,
      "loss": 0.1371,
      "step": 7565
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5990893840789795,
      "learning_rate": 4.5546455607546755e-06,
      "loss": 0.1548,
      "step": 7566
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.823020935058594,
      "learning_rate": 4.5534888724480525e-06,
      "loss": 0.9033,
      "step": 7567
    },
    {
      "epoch": 0.54,
      "grad_norm": 19.814821243286133,
      "learning_rate": 4.552332208229017e-06,
      "loss": 0.8831,
      "step": 7568
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.555666923522949,
      "learning_rate": 4.551175568159961e-06,
      "loss": 0.1281,
      "step": 7569
    },
    {
      "epoch": 0.54,
      "grad_norm": 16.264833450317383,
      "learning_rate": 4.5500189523032875e-06,
      "loss": 0.944,
      "step": 7570
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.653963804244995,
      "learning_rate": 4.548862360721387e-06,
      "loss": 0.2002,
      "step": 7571
    },
    {
      "epoch": 0.54,
      "grad_norm": 23.21731948852539,
      "learning_rate": 4.547705793476654e-06,
      "loss": 0.9824,
      "step": 7572
    },
    {
      "epoch": 0.54,
      "grad_norm": 15.197431564331055,
      "learning_rate": 4.546549250631483e-06,
      "loss": 0.8781,
      "step": 7573
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.7453699111938477,
      "learning_rate": 4.545392732248264e-06,
      "loss": 0.6767,
      "step": 7574
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.434550285339355,
      "learning_rate": 4.544236238389384e-06,
      "loss": 0.9683,
      "step": 7575
    },
    {
      "epoch": 0.54,
      "grad_norm": 14.878315925598145,
      "learning_rate": 4.543079769117233e-06,
      "loss": 0.8726,
      "step": 7576
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.629725456237793,
      "learning_rate": 4.541923324494202e-06,
      "loss": 0.8653,
      "step": 7577
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.934957981109619,
      "learning_rate": 4.5407669045826714e-06,
      "loss": 0.2329,
      "step": 7578
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.2596435546875,
      "learning_rate": 4.539610509445029e-06,
      "loss": 0.7403,
      "step": 7579
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.00768780708313,
      "learning_rate": 4.538454139143656e-06,
      "loss": 0.1029,
      "step": 7580
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.915311813354492,
      "learning_rate": 4.537297793740937e-06,
      "loss": 0.7598,
      "step": 7581
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.8686370849609375,
      "learning_rate": 4.53614147329925e-06,
      "loss": 0.1174,
      "step": 7582
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5127689838409424,
      "learning_rate": 4.534985177880972e-06,
      "loss": 0.1135,
      "step": 7583
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.468791961669922,
      "learning_rate": 4.533828907548487e-06,
      "loss": 0.1157,
      "step": 7584
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7561607360839844,
      "learning_rate": 4.532672662364166e-06,
      "loss": 0.1531,
      "step": 7585
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.728726863861084,
      "learning_rate": 4.531516442390389e-06,
      "loss": 0.0954,
      "step": 7586
    },
    {
      "epoch": 0.54,
      "grad_norm": 18.166940689086914,
      "learning_rate": 4.530360247689523e-06,
      "loss": 0.5485,
      "step": 7587
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.094403266906738,
      "learning_rate": 4.529204078323946e-06,
      "loss": 1.0376,
      "step": 7588
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.243706226348877,
      "learning_rate": 4.528047934356025e-06,
      "loss": 0.1916,
      "step": 7589
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7181339263916016,
      "learning_rate": 4.526891815848133e-06,
      "loss": 0.2523,
      "step": 7590
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.68562912940979,
      "learning_rate": 4.525735722862636e-06,
      "loss": 0.1278,
      "step": 7591
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.238979816436768,
      "learning_rate": 4.524579655461902e-06,
      "loss": 0.3337,
      "step": 7592
    },
    {
      "epoch": 0.54,
      "grad_norm": 13.413365364074707,
      "learning_rate": 4.5234236137082955e-06,
      "loss": 0.7416,
      "step": 7593
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.226884365081787,
      "learning_rate": 4.522267597664181e-06,
      "loss": 0.1438,
      "step": 7594
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.755855083465576,
      "learning_rate": 4.521111607391921e-06,
      "loss": 0.1564,
      "step": 7595
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.763912677764893,
      "learning_rate": 4.519955642953879e-06,
      "loss": 0.8478,
      "step": 7596
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.735029220581055,
      "learning_rate": 4.518799704412411e-06,
      "loss": 0.9472,
      "step": 7597
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.356038093566895,
      "learning_rate": 4.5176437918298796e-06,
      "loss": 0.7006,
      "step": 7598
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.8070836067199707,
      "learning_rate": 4.516487905268639e-06,
      "loss": 0.8011,
      "step": 7599
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.0871338844299316,
      "learning_rate": 4.515332044791045e-06,
      "loss": 0.1573,
      "step": 7600
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.147902488708496,
      "learning_rate": 4.5141762104594525e-06,
      "loss": 0.15,
      "step": 7601
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.020303726196289,
      "learning_rate": 4.513020402336215e-06,
      "loss": 0.1845,
      "step": 7602
    },
    {
      "epoch": 0.55,
      "grad_norm": 20.503833770751953,
      "learning_rate": 4.511864620483684e-06,
      "loss": 0.8055,
      "step": 7603
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.759169816970825,
      "learning_rate": 4.510708864964207e-06,
      "loss": 0.2129,
      "step": 7604
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.884073495864868,
      "learning_rate": 4.509553135840137e-06,
      "loss": 0.2904,
      "step": 7605
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.071014642715454,
      "learning_rate": 4.508397433173817e-06,
      "loss": 0.113,
      "step": 7606
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.1351382732391357,
      "learning_rate": 4.507241757027596e-06,
      "loss": 0.2256,
      "step": 7607
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.663825750350952,
      "learning_rate": 4.506086107463816e-06,
      "loss": 0.1477,
      "step": 7608
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.680959701538086,
      "learning_rate": 4.504930484544822e-06,
      "loss": 0.8671,
      "step": 7609
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3763577938079834,
      "learning_rate": 4.5037748883329555e-06,
      "loss": 0.1449,
      "step": 7610
    },
    {
      "epoch": 0.55,
      "grad_norm": 14.814225196838379,
      "learning_rate": 4.502619318890554e-06,
      "loss": 0.9051,
      "step": 7611
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.561017036437988,
      "learning_rate": 4.501463776279957e-06,
      "loss": 0.6907,
      "step": 7612
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.7059760093688965,
      "learning_rate": 4.500308260563504e-06,
      "loss": 0.1386,
      "step": 7613
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.926135301589966,
      "learning_rate": 4.499152771803527e-06,
      "loss": 0.7775,
      "step": 7614
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8211909532546997,
      "learning_rate": 4.497997310062362e-06,
      "loss": 0.0777,
      "step": 7615
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.119755506515503,
      "learning_rate": 4.496841875402343e-06,
      "loss": 0.0667,
      "step": 7616
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.054446697235107,
      "learning_rate": 4.495686467885799e-06,
      "loss": 0.6781,
      "step": 7617
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.275440216064453,
      "learning_rate": 4.494531087575062e-06,
      "loss": 0.4036,
      "step": 7618
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.228224754333496,
      "learning_rate": 4.49337573453246e-06,
      "loss": 0.1457,
      "step": 7619
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.8122425079345703,
      "learning_rate": 4.492220408820319e-06,
      "loss": 0.1912,
      "step": 7620
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.041400909423828,
      "learning_rate": 4.491065110500965e-06,
      "loss": 0.1607,
      "step": 7621
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.022902011871338,
      "learning_rate": 4.489909839636721e-06,
      "loss": 0.6428,
      "step": 7622
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3229849338531494,
      "learning_rate": 4.488754596289913e-06,
      "loss": 0.0839,
      "step": 7623
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.255924701690674,
      "learning_rate": 4.487599380522856e-06,
      "loss": 0.1353,
      "step": 7624
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.391866207122803,
      "learning_rate": 4.486444192397876e-06,
      "loss": 0.858,
      "step": 7625
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.539097785949707,
      "learning_rate": 4.485289031977286e-06,
      "loss": 0.8522,
      "step": 7626
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.479194164276123,
      "learning_rate": 4.484133899323404e-06,
      "loss": 0.1615,
      "step": 7627
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.068037033081055,
      "learning_rate": 4.482978794498547e-06,
      "loss": 0.8423,
      "step": 7628
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.43210506439209,
      "learning_rate": 4.481823717565026e-06,
      "loss": 0.1107,
      "step": 7629
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.88655948638916,
      "learning_rate": 4.4806686685851545e-06,
      "loss": 0.9122,
      "step": 7630
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5815515518188477,
      "learning_rate": 4.479513647621241e-06,
      "loss": 0.1247,
      "step": 7631
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3756744861602783,
      "learning_rate": 4.478358654735598e-06,
      "loss": 0.213,
      "step": 7632
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.412248611450195,
      "learning_rate": 4.47720368999053e-06,
      "loss": 0.5837,
      "step": 7633
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.288266658782959,
      "learning_rate": 4.476048753448344e-06,
      "loss": 0.7059,
      "step": 7634
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.5417842864990234,
      "learning_rate": 4.474893845171347e-06,
      "loss": 0.1714,
      "step": 7635
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.968932628631592,
      "learning_rate": 4.473738965221838e-06,
      "loss": 0.6054,
      "step": 7636
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.046666145324707,
      "learning_rate": 4.4725841136621185e-06,
      "loss": 0.7362,
      "step": 7637
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1390602588653564,
      "learning_rate": 4.47142929055449e-06,
      "loss": 0.1338,
      "step": 7638
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.720559597015381,
      "learning_rate": 4.470274495961252e-06,
      "loss": 0.1332,
      "step": 7639
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.575155019760132,
      "learning_rate": 4.4691197299446975e-06,
      "loss": 0.0948,
      "step": 7640
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.2478463649749756,
      "learning_rate": 4.4679649925671244e-06,
      "loss": 0.1111,
      "step": 7641
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4242758750915527,
      "learning_rate": 4.466810283890828e-06,
      "loss": 0.0916,
      "step": 7642
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.799530982971191,
      "learning_rate": 4.465655603978097e-06,
      "loss": 0.1866,
      "step": 7643
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.529566764831543,
      "learning_rate": 4.464500952891225e-06,
      "loss": 0.7182,
      "step": 7644
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.7965309619903564,
      "learning_rate": 4.463346330692498e-06,
      "loss": 0.1621,
      "step": 7645
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.7678778171539307,
      "learning_rate": 4.4621917374442055e-06,
      "loss": 0.1382,
      "step": 7646
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.856968879699707,
      "learning_rate": 4.461037173208635e-06,
      "loss": 0.1565,
      "step": 7647
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.159763813018799,
      "learning_rate": 4.459882638048068e-06,
      "loss": 0.8536,
      "step": 7648
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.2227625846862793,
      "learning_rate": 4.458728132024787e-06,
      "loss": 0.0845,
      "step": 7649
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.092367172241211,
      "learning_rate": 4.457573655201075e-06,
      "loss": 0.4317,
      "step": 7650
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.6285042762756348,
      "learning_rate": 4.456419207639211e-06,
      "loss": 0.1921,
      "step": 7651
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.8024728298187256,
      "learning_rate": 4.455264789401472e-06,
      "loss": 0.0897,
      "step": 7652
    },
    {
      "epoch": 0.55,
      "grad_norm": 12.399250030517578,
      "learning_rate": 4.454110400550138e-06,
      "loss": 0.9103,
      "step": 7653
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.660727024078369,
      "learning_rate": 4.45295604114748e-06,
      "loss": 0.8683,
      "step": 7654
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.638282299041748,
      "learning_rate": 4.451801711255773e-06,
      "loss": 1.0258,
      "step": 7655
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.362870693206787,
      "learning_rate": 4.4506474109372894e-06,
      "loss": 0.1691,
      "step": 7656
    },
    {
      "epoch": 0.55,
      "grad_norm": 18.842891693115234,
      "learning_rate": 4.4494931402542975e-06,
      "loss": 1.716,
      "step": 7657
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.538661479949951,
      "learning_rate": 4.4483388992690685e-06,
      "loss": 0.1127,
      "step": 7658
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.24210786819458,
      "learning_rate": 4.447184688043869e-06,
      "loss": 0.7683,
      "step": 7659
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.7398483753204346,
      "learning_rate": 4.446030506640961e-06,
      "loss": 0.1536,
      "step": 7660
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5074920654296875,
      "learning_rate": 4.444876355122608e-06,
      "loss": 0.1179,
      "step": 7661
    },
    {
      "epoch": 0.55,
      "grad_norm": 15.293807983398438,
      "learning_rate": 4.443722233551078e-06,
      "loss": 0.9554,
      "step": 7662
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4068424701690674,
      "learning_rate": 4.4425681419886255e-06,
      "loss": 0.1114,
      "step": 7663
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.890346527099609,
      "learning_rate": 4.441414080497514e-06,
      "loss": 0.4995,
      "step": 7664
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.86893892288208,
      "learning_rate": 4.440260049139996e-06,
      "loss": 0.8104,
      "step": 7665
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.08218002319336,
      "learning_rate": 4.439106047978331e-06,
      "loss": 0.6794,
      "step": 7666
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.629760980606079,
      "learning_rate": 4.437952077074772e-06,
      "loss": 0.1624,
      "step": 7667
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.869786262512207,
      "learning_rate": 4.436798136491569e-06,
      "loss": 0.8745,
      "step": 7668
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.115379810333252,
      "learning_rate": 4.435644226290976e-06,
      "loss": 0.8965,
      "step": 7669
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.50241756439209,
      "learning_rate": 4.434490346535242e-06,
      "loss": 0.84,
      "step": 7670
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.608758449554443,
      "learning_rate": 4.433336497286613e-06,
      "loss": 0.879,
      "step": 7671
    },
    {
      "epoch": 0.55,
      "grad_norm": 12.806743621826172,
      "learning_rate": 4.432182678607334e-06,
      "loss": 0.9056,
      "step": 7672
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.153020858764648,
      "learning_rate": 4.431028890559651e-06,
      "loss": 0.8531,
      "step": 7673
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3505382537841797,
      "learning_rate": 4.4298751332058045e-06,
      "loss": 0.222,
      "step": 7674
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.021491527557373,
      "learning_rate": 4.428721406608036e-06,
      "loss": 0.112,
      "step": 7675
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.848978281021118,
      "learning_rate": 4.427567710828588e-06,
      "loss": 0.1984,
      "step": 7676
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.002556800842285,
      "learning_rate": 4.426414045929692e-06,
      "loss": 0.2969,
      "step": 7677
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.156209468841553,
      "learning_rate": 4.42526041197359e-06,
      "loss": 0.637,
      "step": 7678
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.362534999847412,
      "learning_rate": 4.424106809022511e-06,
      "loss": 0.8752,
      "step": 7679
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.605008125305176,
      "learning_rate": 4.42295323713869e-06,
      "loss": 0.5073,
      "step": 7680
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0703330039978027,
      "learning_rate": 4.421799696384359e-06,
      "loss": 0.1118,
      "step": 7681
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.57065486907959,
      "learning_rate": 4.420646186821744e-06,
      "loss": 0.1568,
      "step": 7682
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.417104721069336,
      "learning_rate": 4.4194927085130775e-06,
      "loss": 0.1348,
      "step": 7683
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.108586549758911,
      "learning_rate": 4.4183392615205785e-06,
      "loss": 0.12,
      "step": 7684
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.459664344787598,
      "learning_rate": 4.417185845906477e-06,
      "loss": 0.4628,
      "step": 7685
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.896251678466797,
      "learning_rate": 4.4160324617329915e-06,
      "loss": 0.8513,
      "step": 7686
    },
    {
      "epoch": 0.55,
      "grad_norm": 12.227497100830078,
      "learning_rate": 4.414879109062345e-06,
      "loss": 0.8567,
      "step": 7687
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.129716396331787,
      "learning_rate": 4.413725787956755e-06,
      "loss": 0.2144,
      "step": 7688
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.564568519592285,
      "learning_rate": 4.4125724984784405e-06,
      "loss": 0.6401,
      "step": 7689
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.964686632156372,
      "learning_rate": 4.411419240689615e-06,
      "loss": 0.2096,
      "step": 7690
    },
    {
      "epoch": 0.55,
      "grad_norm": 17.230873107910156,
      "learning_rate": 4.410266014652495e-06,
      "loss": 0.8654,
      "step": 7691
    },
    {
      "epoch": 0.55,
      "grad_norm": 18.05014419555664,
      "learning_rate": 4.40911282042929e-06,
      "loss": 1.1287,
      "step": 7692
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.6848793029785156,
      "learning_rate": 4.407959658082213e-06,
      "loss": 0.1123,
      "step": 7693
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1337172985076904,
      "learning_rate": 4.4068065276734695e-06,
      "loss": 0.1105,
      "step": 7694
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.246764659881592,
      "learning_rate": 4.405653429265272e-06,
      "loss": 0.1704,
      "step": 7695
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.512399673461914,
      "learning_rate": 4.40450036291982e-06,
      "loss": 0.1477,
      "step": 7696
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.146968364715576,
      "learning_rate": 4.403347328699319e-06,
      "loss": 0.8107,
      "step": 7697
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.512953758239746,
      "learning_rate": 4.40219432666597e-06,
      "loss": 0.1709,
      "step": 7698
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.662543535232544,
      "learning_rate": 4.401041356881977e-06,
      "loss": 0.1635,
      "step": 7699
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.6947641372680664,
      "learning_rate": 4.399888419409534e-06,
      "loss": 0.1014,
      "step": 7700
    },
    {
      "epoch": 0.55,
      "grad_norm": 16.434473037719727,
      "learning_rate": 4.39873551431084e-06,
      "loss": 1.1404,
      "step": 7701
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.970792770385742,
      "learning_rate": 4.397582641648088e-06,
      "loss": 0.5119,
      "step": 7702
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.272457122802734,
      "learning_rate": 4.396429801483472e-06,
      "loss": 0.6967,
      "step": 7703
    },
    {
      "epoch": 0.55,
      "grad_norm": 17.943838119506836,
      "learning_rate": 4.395276993879185e-06,
      "loss": 1.2648,
      "step": 7704
    },
    {
      "epoch": 0.55,
      "grad_norm": 16.261463165283203,
      "learning_rate": 4.394124218897414e-06,
      "loss": 0.8134,
      "step": 7705
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.288058280944824,
      "learning_rate": 4.392971476600349e-06,
      "loss": 0.5483,
      "step": 7706
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.249682426452637,
      "learning_rate": 4.391818767050176e-06,
      "loss": 0.7204,
      "step": 7707
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3777525424957275,
      "learning_rate": 4.390666090309078e-06,
      "loss": 0.0966,
      "step": 7708
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.2656190395355225,
      "learning_rate": 4.3895134464392365e-06,
      "loss": 0.1891,
      "step": 7709
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1684670448303223,
      "learning_rate": 4.388360835502835e-06,
      "loss": 0.1347,
      "step": 7710
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.450617790222168,
      "learning_rate": 4.3872082575620494e-06,
      "loss": 0.7632,
      "step": 7711
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.993352174758911,
      "learning_rate": 4.38605571267906e-06,
      "loss": 0.214,
      "step": 7712
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.7777292728424072,
      "learning_rate": 4.3849032009160415e-06,
      "loss": 0.2545,
      "step": 7713
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.02799129486084,
      "learning_rate": 4.383750722335166e-06,
      "loss": 0.0984,
      "step": 7714
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.096570014953613,
      "learning_rate": 4.382598276998607e-06,
      "loss": 0.9744,
      "step": 7715
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.44250202178955,
      "learning_rate": 4.3814458649685335e-06,
      "loss": 0.8124,
      "step": 7716
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.50929856300354,
      "learning_rate": 4.380293486307114e-06,
      "loss": 0.2194,
      "step": 7717
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.825628280639648,
      "learning_rate": 4.379141141076517e-06,
      "loss": 0.566,
      "step": 7718
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.316494941711426,
      "learning_rate": 4.377988829338906e-06,
      "loss": 0.7939,
      "step": 7719
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3584723472595215,
      "learning_rate": 4.3768365511564405e-06,
      "loss": 0.0966,
      "step": 7720
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.127513408660889,
      "learning_rate": 4.375684306591285e-06,
      "loss": 0.8951,
      "step": 7721
    },
    {
      "epoch": 0.55,
      "grad_norm": 16.97591781616211,
      "learning_rate": 4.374532095705598e-06,
      "loss": 1.1116,
      "step": 7722
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.10920786857605,
      "learning_rate": 4.373379918561537e-06,
      "loss": 0.087,
      "step": 7723
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1624057292938232,
      "learning_rate": 4.3722277752212585e-06,
      "loss": 0.1176,
      "step": 7724
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.256774425506592,
      "learning_rate": 4.371075665746913e-06,
      "loss": 0.6191,
      "step": 7725
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1102418899536133,
      "learning_rate": 4.369923590200656e-06,
      "loss": 0.0963,
      "step": 7726
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5489635467529297,
      "learning_rate": 4.368771548644637e-06,
      "loss": 0.1355,
      "step": 7727
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.0135765075683594,
      "learning_rate": 4.367619541141003e-06,
      "loss": 0.1728,
      "step": 7728
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.8273425102233887,
      "learning_rate": 4.366467567751901e-06,
      "loss": 0.1001,
      "step": 7729
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4566566944122314,
      "learning_rate": 4.365315628539475e-06,
      "loss": 0.0955,
      "step": 7730
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4903106689453125,
      "learning_rate": 4.364163723565872e-06,
      "loss": 0.1412,
      "step": 7731
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.258092164993286,
      "learning_rate": 4.363011852893226e-06,
      "loss": 0.1847,
      "step": 7732
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.285221099853516,
      "learning_rate": 4.36186001658368e-06,
      "loss": 0.6035,
      "step": 7733
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.32274866104126,
      "learning_rate": 4.36070821469937e-06,
      "loss": 0.848,
      "step": 7734
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.438028812408447,
      "learning_rate": 4.359556447302432e-06,
      "loss": 1.0069,
      "step": 7735
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.174945831298828,
      "learning_rate": 4.358404714455e-06,
      "loss": 0.084,
      "step": 7736
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.859663963317871,
      "learning_rate": 4.357253016219205e-06,
      "loss": 1.5523,
      "step": 7737
    },
    {
      "epoch": 0.55,
      "grad_norm": 13.367768287658691,
      "learning_rate": 4.356101352657178e-06,
      "loss": 0.7452,
      "step": 7738
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.017251014709473,
      "learning_rate": 4.354949723831043e-06,
      "loss": 0.7282,
      "step": 7739
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5900726318359375,
      "learning_rate": 4.3537981298029305e-06,
      "loss": 0.1539,
      "step": 7740
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.6384785175323486,
      "learning_rate": 4.352646570634964e-06,
      "loss": 0.1576,
      "step": 7741
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.9934558868408203,
      "learning_rate": 4.351495046389262e-06,
      "loss": 0.2264,
      "step": 7742
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.271928310394287,
      "learning_rate": 4.35034355712795e-06,
      "loss": 0.1523,
      "step": 7743
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.8346526622772217,
      "learning_rate": 4.349192102913145e-06,
      "loss": 0.1338,
      "step": 7744
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.990457057952881,
      "learning_rate": 4.34804068380696e-06,
      "loss": 0.672,
      "step": 7745
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.6205930709838867,
      "learning_rate": 4.3468892998715114e-06,
      "loss": 0.1613,
      "step": 7746
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.812689781188965,
      "learning_rate": 4.345737951168916e-06,
      "loss": 0.8373,
      "step": 7747
    },
    {
      "epoch": 0.56,
      "grad_norm": 17.407629013061523,
      "learning_rate": 4.344586637761279e-06,
      "loss": 0.9886,
      "step": 7748
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4725329875946045,
      "learning_rate": 4.343435359710713e-06,
      "loss": 0.1353,
      "step": 7749
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.2664155960083,
      "learning_rate": 4.3422841170793244e-06,
      "loss": 0.8503,
      "step": 7750
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.162057399749756,
      "learning_rate": 4.341132909929217e-06,
      "loss": 0.0775,
      "step": 7751
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5261731147766113,
      "learning_rate": 4.339981738322497e-06,
      "loss": 0.2105,
      "step": 7752
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.920635223388672,
      "learning_rate": 4.338830602321262e-06,
      "loss": 0.6956,
      "step": 7753
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3243982791900635,
      "learning_rate": 4.337679501987614e-06,
      "loss": 0.2276,
      "step": 7754
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.094985961914062,
      "learning_rate": 4.336528437383652e-06,
      "loss": 0.6406,
      "step": 7755
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1609177589416504,
      "learning_rate": 4.335377408571468e-06,
      "loss": 0.0765,
      "step": 7756
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7599236965179443,
      "learning_rate": 4.334226415613156e-06,
      "loss": 0.0634,
      "step": 7757
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.0376946926116943,
      "learning_rate": 4.33307545857081e-06,
      "loss": 0.1093,
      "step": 7758
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.72933292388916,
      "learning_rate": 4.331924537506518e-06,
      "loss": 0.1454,
      "step": 7759
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.9004931449890137,
      "learning_rate": 4.330773652482367e-06,
      "loss": 0.8752,
      "step": 7760
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.040538787841797,
      "learning_rate": 4.329622803560447e-06,
      "loss": 0.1117,
      "step": 7761
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.974839925765991,
      "learning_rate": 4.328471990802837e-06,
      "loss": 0.1456,
      "step": 7762
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7729320526123047,
      "learning_rate": 4.327321214271622e-06,
      "loss": 0.9024,
      "step": 7763
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.0937066078186035,
      "learning_rate": 4.326170474028881e-06,
      "loss": 0.1297,
      "step": 7764
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.006904602050781,
      "learning_rate": 4.325019770136692e-06,
      "loss": 0.6431,
      "step": 7765
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.374835252761841,
      "learning_rate": 4.323869102657133e-06,
      "loss": 0.1209,
      "step": 7766
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7017154693603516,
      "learning_rate": 4.322718471652278e-06,
      "loss": 0.1849,
      "step": 7767
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.779311418533325,
      "learning_rate": 4.3215678771841955e-06,
      "loss": 0.1481,
      "step": 7768
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.330153465270996,
      "learning_rate": 4.3204173193149565e-06,
      "loss": 0.9667,
      "step": 7769
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.1320667266845703,
      "learning_rate": 4.319266798106634e-06,
      "loss": 0.1595,
      "step": 7770
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.341664791107178,
      "learning_rate": 4.318116313621288e-06,
      "loss": 0.57,
      "step": 7771
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.820849657058716,
      "learning_rate": 4.316965865920989e-06,
      "loss": 0.1033,
      "step": 7772
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.056827068328857,
      "learning_rate": 4.315815455067793e-06,
      "loss": 0.5403,
      "step": 7773
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7455861568450928,
      "learning_rate": 4.3146650811237645e-06,
      "loss": 0.057,
      "step": 7774
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.51459264755249,
      "learning_rate": 4.313514744150961e-06,
      "loss": 0.8126,
      "step": 7775
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5964133739471436,
      "learning_rate": 4.312364444211438e-06,
      "loss": 0.0928,
      "step": 7776
    },
    {
      "epoch": 0.56,
      "grad_norm": 14.484294891357422,
      "learning_rate": 4.311214181367251e-06,
      "loss": 1.1558,
      "step": 7777
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.280766487121582,
      "learning_rate": 4.3100639556804515e-06,
      "loss": 0.8409,
      "step": 7778
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.33875036239624,
      "learning_rate": 4.308913767213092e-06,
      "loss": 1.0396,
      "step": 7779
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.129956245422363,
      "learning_rate": 4.307763616027217e-06,
      "loss": 0.9581,
      "step": 7780
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.256321668624878,
      "learning_rate": 4.3066135021848756e-06,
      "loss": 0.1521,
      "step": 7781
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3695294857025146,
      "learning_rate": 4.305463425748109e-06,
      "loss": 0.1039,
      "step": 7782
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.9409713745117188,
      "learning_rate": 4.3043133867789625e-06,
      "loss": 0.9643,
      "step": 7783
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.906975746154785,
      "learning_rate": 4.303163385339477e-06,
      "loss": 0.85,
      "step": 7784
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.46875,
      "learning_rate": 4.302013421491687e-06,
      "loss": 0.1422,
      "step": 7785
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.105839252471924,
      "learning_rate": 4.300863495297632e-06,
      "loss": 0.5345,
      "step": 7786
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.968452453613281,
      "learning_rate": 4.299713606819345e-06,
      "loss": 0.6266,
      "step": 7787
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.280069589614868,
      "learning_rate": 4.298563756118858e-06,
      "loss": 0.1432,
      "step": 7788
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.800048828125,
      "learning_rate": 4.297413943258202e-06,
      "loss": 0.2755,
      "step": 7789
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.487404823303223,
      "learning_rate": 4.296264168299404e-06,
      "loss": 0.5403,
      "step": 7790
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.372788906097412,
      "learning_rate": 4.2951144313044915e-06,
      "loss": 0.1128,
      "step": 7791
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.853900909423828,
      "learning_rate": 4.293964732335485e-06,
      "loss": 0.734,
      "step": 7792
    },
    {
      "epoch": 0.56,
      "grad_norm": 31.775146484375,
      "learning_rate": 4.29281507145441e-06,
      "loss": 1.0097,
      "step": 7793
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.766537666320801,
      "learning_rate": 4.291665448723285e-06,
      "loss": 0.4988,
      "step": 7794
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.91982650756836,
      "learning_rate": 4.290515864204127e-06,
      "loss": 0.5233,
      "step": 7795
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.8807268142700195,
      "learning_rate": 4.289366317958953e-06,
      "loss": 0.9924,
      "step": 7796
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.6468963623046875,
      "learning_rate": 4.2882168100497755e-06,
      "loss": 0.8951,
      "step": 7797
    },
    {
      "epoch": 0.56,
      "grad_norm": 13.598251342773438,
      "learning_rate": 4.2870673405386085e-06,
      "loss": 0.8399,
      "step": 7798
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.454594373703003,
      "learning_rate": 4.2859179094874585e-06,
      "loss": 0.1433,
      "step": 7799
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.671605110168457,
      "learning_rate": 4.284768516958335e-06,
      "loss": 0.1572,
      "step": 7800
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.9238455295562744,
      "learning_rate": 4.283619163013243e-06,
      "loss": 0.1254,
      "step": 7801
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.87562096118927,
      "learning_rate": 4.282469847714185e-06,
      "loss": 0.0494,
      "step": 7802
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4348621368408203,
      "learning_rate": 4.281320571123164e-06,
      "loss": 0.1778,
      "step": 7803
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.6005609035491943,
      "learning_rate": 4.2801713333021775e-06,
      "loss": 0.1208,
      "step": 7804
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.917028427124023,
      "learning_rate": 4.279022134313221e-06,
      "loss": 0.4609,
      "step": 7805
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.8321175575256348,
      "learning_rate": 4.277872974218292e-06,
      "loss": 0.1324,
      "step": 7806
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4109907150268555,
      "learning_rate": 4.276723853079384e-06,
      "loss": 0.1001,
      "step": 7807
    },
    {
      "epoch": 0.56,
      "grad_norm": 22.438732147216797,
      "learning_rate": 4.275574770958485e-06,
      "loss": 1.4089,
      "step": 7808
    },
    {
      "epoch": 0.56,
      "grad_norm": 19.84918975830078,
      "learning_rate": 4.274425727917586e-06,
      "loss": 1.0223,
      "step": 7809
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.8781135082244873,
      "learning_rate": 4.27327672401867e-06,
      "loss": 0.0693,
      "step": 7810
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.260943412780762,
      "learning_rate": 4.272127759323726e-06,
      "loss": 0.3758,
      "step": 7811
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7079386711120605,
      "learning_rate": 4.270978833894735e-06,
      "loss": 0.6098,
      "step": 7812
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7074878215789795,
      "learning_rate": 4.269829947793675e-06,
      "loss": 0.9639,
      "step": 7813
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.810103178024292,
      "learning_rate": 4.268681101082525e-06,
      "loss": 0.116,
      "step": 7814
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3800721168518066,
      "learning_rate": 4.267532293823265e-06,
      "loss": 0.1147,
      "step": 7815
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.907096862792969,
      "learning_rate": 4.266383526077863e-06,
      "loss": 0.7079,
      "step": 7816
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.7028121948242188,
      "learning_rate": 4.265234797908291e-06,
      "loss": 0.1738,
      "step": 7817
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.218101501464844,
      "learning_rate": 4.264086109376523e-06,
      "loss": 0.5047,
      "step": 7818
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.8063626289367676,
      "learning_rate": 4.262937460544522e-06,
      "loss": 0.1326,
      "step": 7819
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.0891928672790527,
      "learning_rate": 4.261788851474255e-06,
      "loss": 0.2041,
      "step": 7820
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.5493011474609375,
      "learning_rate": 4.260640282227687e-06,
      "loss": 0.9396,
      "step": 7821
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.37479305267334,
      "learning_rate": 4.259491752866776e-06,
      "loss": 0.6744,
      "step": 7822
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.466304063796997,
      "learning_rate": 4.258343263453482e-06,
      "loss": 0.0537,
      "step": 7823
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.477778196334839,
      "learning_rate": 4.257194814049761e-06,
      "loss": 0.1627,
      "step": 7824
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.873382091522217,
      "learning_rate": 4.256046404717569e-06,
      "loss": 0.1479,
      "step": 7825
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1371893882751465,
      "learning_rate": 4.254898035518857e-06,
      "loss": 0.0986,
      "step": 7826
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0909948348999023,
      "learning_rate": 4.253749706515578e-06,
      "loss": 0.0723,
      "step": 7827
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.197319030761719,
      "learning_rate": 4.252601417769675e-06,
      "loss": 0.859,
      "step": 7828
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2828147411346436,
      "learning_rate": 4.251453169343096e-06,
      "loss": 0.0966,
      "step": 7829
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.0752882957458496,
      "learning_rate": 4.2503049612977874e-06,
      "loss": 0.1689,
      "step": 7830
    },
    {
      "epoch": 0.56,
      "grad_norm": 74.95733642578125,
      "learning_rate": 4.249156793695686e-06,
      "loss": 1.1084,
      "step": 7831
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.8378684520721436,
      "learning_rate": 4.248008666598735e-06,
      "loss": 0.2094,
      "step": 7832
    },
    {
      "epoch": 0.56,
      "grad_norm": 16.50323486328125,
      "learning_rate": 4.2468605800688685e-06,
      "loss": 0.7013,
      "step": 7833
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.429800033569336,
      "learning_rate": 4.245712534168024e-06,
      "loss": 0.8796,
      "step": 7834
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2965874671936035,
      "learning_rate": 4.244564528958134e-06,
      "loss": 0.1137,
      "step": 7835
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.273332595825195,
      "learning_rate": 4.243416564501127e-06,
      "loss": 0.9022,
      "step": 7836
    },
    {
      "epoch": 0.56,
      "grad_norm": 14.114879608154297,
      "learning_rate": 4.242268640858933e-06,
      "loss": 0.6813,
      "step": 7837
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.8125269412994385,
      "learning_rate": 4.241120758093477e-06,
      "loss": 0.1088,
      "step": 7838
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8606324195861816,
      "learning_rate": 4.239972916266684e-06,
      "loss": 0.0835,
      "step": 7839
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.84488844871521,
      "learning_rate": 4.238825115440476e-06,
      "loss": 0.0722,
      "step": 7840
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.702641010284424,
      "learning_rate": 4.237677355676772e-06,
      "loss": 0.1619,
      "step": 7841
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.381866455078125,
      "learning_rate": 4.236529637037488e-06,
      "loss": 0.4977,
      "step": 7842
    },
    {
      "epoch": 0.56,
      "grad_norm": 18.191556930541992,
      "learning_rate": 4.23538195958454e-06,
      "loss": 1.0468,
      "step": 7843
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.068589210510254,
      "learning_rate": 4.234234323379842e-06,
      "loss": 0.8106,
      "step": 7844
    },
    {
      "epoch": 0.56,
      "grad_norm": 13.556662559509277,
      "learning_rate": 4.233086728485303e-06,
      "loss": 1.1708,
      "step": 7845
    },
    {
      "epoch": 0.56,
      "grad_norm": 23.875059127807617,
      "learning_rate": 4.231939174962833e-06,
      "loss": 0.4861,
      "step": 7846
    },
    {
      "epoch": 0.56,
      "grad_norm": 13.549098014831543,
      "learning_rate": 4.230791662874335e-06,
      "loss": 1.123,
      "step": 7847
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.158412456512451,
      "learning_rate": 4.229644192281717e-06,
      "loss": 0.1529,
      "step": 7848
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.318668365478516,
      "learning_rate": 4.228496763246878e-06,
      "loss": 0.8217,
      "step": 7849
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.752684593200684,
      "learning_rate": 4.227349375831719e-06,
      "loss": 0.6385,
      "step": 7850
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.84157133102417,
      "learning_rate": 4.2262020300981375e-06,
      "loss": 0.2315,
      "step": 7851
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.335612297058105,
      "learning_rate": 4.225054726108028e-06,
      "loss": 0.5224,
      "step": 7852
    },
    {
      "epoch": 0.56,
      "grad_norm": 17.927978515625,
      "learning_rate": 4.22390746392328e-06,
      "loss": 1.1601,
      "step": 7853
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.474325180053711,
      "learning_rate": 4.222760243605787e-06,
      "loss": 0.1686,
      "step": 7854
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.005712985992432,
      "learning_rate": 4.221613065217438e-06,
      "loss": 0.8358,
      "step": 7855
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.252182960510254,
      "learning_rate": 4.220465928820117e-06,
      "loss": 0.0632,
      "step": 7856
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.49990177154541,
      "learning_rate": 4.219318834475709e-06,
      "loss": 1.4509,
      "step": 7857
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3347504138946533,
      "learning_rate": 4.218171782246095e-06,
      "loss": 0.0635,
      "step": 7858
    },
    {
      "epoch": 0.56,
      "grad_norm": 15.044137954711914,
      "learning_rate": 4.217024772193153e-06,
      "loss": 0.4997,
      "step": 7859
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.131077766418457,
      "learning_rate": 4.215877804378762e-06,
      "loss": 0.1004,
      "step": 7860
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4072766304016113,
      "learning_rate": 4.2147308788647944e-06,
      "loss": 0.1175,
      "step": 7861
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.082468032836914,
      "learning_rate": 4.213583995713123e-06,
      "loss": 0.1673,
      "step": 7862
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.7890193462371826,
      "learning_rate": 4.21243715498562e-06,
      "loss": 0.1505,
      "step": 7863
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6735647916793823,
      "learning_rate": 4.211290356744152e-06,
      "loss": 0.0729,
      "step": 7864
    },
    {
      "epoch": 0.56,
      "grad_norm": 27.776887893676758,
      "learning_rate": 4.210143601050583e-06,
      "loss": 0.5671,
      "step": 7865
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.510951042175293,
      "learning_rate": 4.2089968879667756e-06,
      "loss": 0.6707,
      "step": 7866
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.123624563217163,
      "learning_rate": 4.207850217554592e-06,
      "loss": 0.1139,
      "step": 7867
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.194598436355591,
      "learning_rate": 4.2067035898758915e-06,
      "loss": 0.2068,
      "step": 7868
    },
    {
      "epoch": 0.56,
      "grad_norm": 15.250765800476074,
      "learning_rate": 4.205557004992531e-06,
      "loss": 1.1643,
      "step": 7869
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.181861877441406,
      "learning_rate": 4.20441046296636e-06,
      "loss": 0.7157,
      "step": 7870
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.329320907592773,
      "learning_rate": 4.203263963859234e-06,
      "loss": 0.6886,
      "step": 7871
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.299953937530518,
      "learning_rate": 4.202117507733003e-06,
      "loss": 0.2518,
      "step": 7872
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.48126220703125,
      "learning_rate": 4.2009710946495104e-06,
      "loss": 1.2653,
      "step": 7873
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5363049507141113,
      "learning_rate": 4.199824724670604e-06,
      "loss": 0.1182,
      "step": 7874
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.307990074157715,
      "learning_rate": 4.198678397858124e-06,
      "loss": 0.5555,
      "step": 7875
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4851274490356445,
      "learning_rate": 4.197532114273914e-06,
      "loss": 0.1198,
      "step": 7876
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.745147705078125,
      "learning_rate": 4.1963858739798054e-06,
      "loss": 0.9309,
      "step": 7877
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.627378463745117,
      "learning_rate": 4.195239677037639e-06,
      "loss": 0.1468,
      "step": 7878
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.264151573181152,
      "learning_rate": 4.194093523509244e-06,
      "loss": 0.4134,
      "step": 7879
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.72192096710205,
      "learning_rate": 4.192947413456454e-06,
      "loss": 0.9843,
      "step": 7880
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4155277013778687,
      "learning_rate": 4.191801346941097e-06,
      "loss": 0.0465,
      "step": 7881
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.091733455657959,
      "learning_rate": 4.1906553240249965e-06,
      "loss": 0.1873,
      "step": 7882
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.737980365753174,
      "learning_rate": 4.189509344769979e-06,
      "loss": 0.1152,
      "step": 7883
    },
    {
      "epoch": 0.57,
      "grad_norm": 19.431198120117188,
      "learning_rate": 4.188363409237863e-06,
      "loss": 0.7499,
      "step": 7884
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.908496856689453,
      "learning_rate": 4.18721751749047e-06,
      "loss": 0.9156,
      "step": 7885
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.579860210418701,
      "learning_rate": 4.1860716695896155e-06,
      "loss": 0.8359,
      "step": 7886
    },
    {
      "epoch": 0.57,
      "grad_norm": 26.237197875976562,
      "learning_rate": 4.184925865597113e-06,
      "loss": 2.0601,
      "step": 7887
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.130682468414307,
      "learning_rate": 4.183780105574776e-06,
      "loss": 0.3638,
      "step": 7888
    },
    {
      "epoch": 0.57,
      "grad_norm": 17.060951232910156,
      "learning_rate": 4.182634389584413e-06,
      "loss": 1.5759,
      "step": 7889
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.748401641845703,
      "learning_rate": 4.181488717687829e-06,
      "loss": 0.1181,
      "step": 7890
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.6333682537078857,
      "learning_rate": 4.180343089946831e-06,
      "loss": 0.8347,
      "step": 7891
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.654787540435791,
      "learning_rate": 4.179197506423221e-06,
      "loss": 0.063,
      "step": 7892
    },
    {
      "epoch": 0.57,
      "grad_norm": 19.206443786621094,
      "learning_rate": 4.1780519671787974e-06,
      "loss": 0.9105,
      "step": 7893
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.857715606689453,
      "learning_rate": 4.17690647227536e-06,
      "loss": 1.4113,
      "step": 7894
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0601930618286133,
      "learning_rate": 4.175761021774704e-06,
      "loss": 0.08,
      "step": 7895
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.029310464859009,
      "learning_rate": 4.174615615738619e-06,
      "loss": 0.2553,
      "step": 7896
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.798788547515869,
      "learning_rate": 4.173470254228899e-06,
      "loss": 0.7085,
      "step": 7897
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.102541446685791,
      "learning_rate": 4.172324937307328e-06,
      "loss": 0.6499,
      "step": 7898
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0029211044311523,
      "learning_rate": 4.171179665035694e-06,
      "loss": 0.0649,
      "step": 7899
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.755117893218994,
      "learning_rate": 4.170034437475783e-06,
      "loss": 0.7257,
      "step": 7900
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.623302936553955,
      "learning_rate": 4.168889254689371e-06,
      "loss": 0.1059,
      "step": 7901
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.875570297241211,
      "learning_rate": 4.167744116738236e-06,
      "loss": 0.8117,
      "step": 7902
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.386143684387207,
      "learning_rate": 4.166599023684157e-06,
      "loss": 0.9936,
      "step": 7903
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.216588973999023,
      "learning_rate": 4.165453975588905e-06,
      "loss": 0.7482,
      "step": 7904
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.9283857345581055,
      "learning_rate": 4.164308972514253e-06,
      "loss": 0.8227,
      "step": 7905
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.039945602416992,
      "learning_rate": 4.163164014521969e-06,
      "loss": 0.6929,
      "step": 7906
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.849813461303711,
      "learning_rate": 4.162019101673818e-06,
      "loss": 0.7199,
      "step": 7907
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.0759637355804443,
      "learning_rate": 4.1608742340315644e-06,
      "loss": 0.2326,
      "step": 7908
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.384789943695068,
      "learning_rate": 4.159729411656971e-06,
      "loss": 1.0925,
      "step": 7909
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.248650550842285,
      "learning_rate": 4.1585846346117935e-06,
      "loss": 0.5819,
      "step": 7910
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.201882839202881,
      "learning_rate": 4.157439902957791e-06,
      "loss": 0.956,
      "step": 7911
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.811840772628784,
      "learning_rate": 4.156295216756718e-06,
      "loss": 0.1254,
      "step": 7912
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.336000442504883,
      "learning_rate": 4.155150576070323e-06,
      "loss": 1.3019,
      "step": 7913
    },
    {
      "epoch": 0.57,
      "grad_norm": 15.341705322265625,
      "learning_rate": 4.154005980960355e-06,
      "loss": 1.2162,
      "step": 7914
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.106147289276123,
      "learning_rate": 4.152861431488564e-06,
      "loss": 0.6327,
      "step": 7915
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.346275806427002,
      "learning_rate": 4.151716927716692e-06,
      "loss": 0.703,
      "step": 7916
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.8986153602600098,
      "learning_rate": 4.150572469706481e-06,
      "loss": 0.104,
      "step": 7917
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.7804641723632812,
      "learning_rate": 4.149428057519669e-06,
      "loss": 0.2149,
      "step": 7918
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.816437721252441,
      "learning_rate": 4.148283691217993e-06,
      "loss": 0.5673,
      "step": 7919
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.169962882995605,
      "learning_rate": 4.14713937086319e-06,
      "loss": 0.5452,
      "step": 7920
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.966690063476562,
      "learning_rate": 4.145995096516988e-06,
      "loss": 0.9655,
      "step": 7921
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.206979751586914,
      "learning_rate": 4.144850868241118e-06,
      "loss": 0.2345,
      "step": 7922
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.132394552230835,
      "learning_rate": 4.143706686097308e-06,
      "loss": 0.15,
      "step": 7923
    },
    {
      "epoch": 0.57,
      "grad_norm": 18.952909469604492,
      "learning_rate": 4.142562550147282e-06,
      "loss": 0.8484,
      "step": 7924
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.31811237335205,
      "learning_rate": 4.141418460452759e-06,
      "loss": 0.3991,
      "step": 7925
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.342024564743042,
      "learning_rate": 4.140274417075461e-06,
      "loss": 0.0786,
      "step": 7926
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.3820295333862305,
      "learning_rate": 4.139130420077103e-06,
      "loss": 0.1212,
      "step": 7927
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.600358486175537,
      "learning_rate": 4.1379864695194e-06,
      "loss": 0.9071,
      "step": 7928
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2684943675994873,
      "learning_rate": 4.136842565464066e-06,
      "loss": 0.1224,
      "step": 7929
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.449640274047852,
      "learning_rate": 4.135698707972806e-06,
      "loss": 0.616,
      "step": 7930
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.7243752479553223,
      "learning_rate": 4.1345548971073305e-06,
      "loss": 0.0903,
      "step": 7931
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.3073177337646484,
      "learning_rate": 4.133411132929342e-06,
      "loss": 0.0921,
      "step": 7932
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.9673497676849365,
      "learning_rate": 4.1322674155005426e-06,
      "loss": 0.6048,
      "step": 7933
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.87572979927063,
      "learning_rate": 4.131123744882633e-06,
      "loss": 0.9247,
      "step": 7934
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.854128360748291,
      "learning_rate": 4.129980121137306e-06,
      "loss": 0.7529,
      "step": 7935
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.729146003723145,
      "learning_rate": 4.128836544326262e-06,
      "loss": 0.5296,
      "step": 7936
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.776942253112793,
      "learning_rate": 4.127693014511185e-06,
      "loss": 0.7241,
      "step": 7937
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.1212356090545654,
      "learning_rate": 4.12654953175377e-06,
      "loss": 0.135,
      "step": 7938
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.362723350524902,
      "learning_rate": 4.125406096115701e-06,
      "loss": 1.0445,
      "step": 7939
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.403809547424316,
      "learning_rate": 4.124262707658663e-06,
      "loss": 0.813,
      "step": 7940
    },
    {
      "epoch": 0.57,
      "grad_norm": 18.925500869750977,
      "learning_rate": 4.123119366444336e-06,
      "loss": 1.3132,
      "step": 7941
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.277434825897217,
      "learning_rate": 4.121976072534401e-06,
      "loss": 0.6242,
      "step": 7942
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.404869318008423,
      "learning_rate": 4.1208328259905326e-06,
      "loss": 0.1992,
      "step": 7943
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.428698539733887,
      "learning_rate": 4.119689626874406e-06,
      "loss": 0.9703,
      "step": 7944
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.901726722717285,
      "learning_rate": 4.1185464752476925e-06,
      "loss": 1.0509,
      "step": 7945
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.492518424987793,
      "learning_rate": 4.117403371172059e-06,
      "loss": 1.0663,
      "step": 7946
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.446135997772217,
      "learning_rate": 4.116260314709172e-06,
      "loss": 0.153,
      "step": 7947
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.7144534587860107,
      "learning_rate": 4.1151173059207e-06,
      "loss": 0.1465,
      "step": 7948
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.549751281738281,
      "learning_rate": 4.113974344868297e-06,
      "loss": 0.1663,
      "step": 7949
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.184967279434204,
      "learning_rate": 4.112831431613624e-06,
      "loss": 0.1465,
      "step": 7950
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8991434574127197,
      "learning_rate": 4.111688566218337e-06,
      "loss": 0.0503,
      "step": 7951
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5737767219543457,
      "learning_rate": 4.110545748744091e-06,
      "loss": 0.2,
      "step": 7952
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.369920253753662,
      "learning_rate": 4.109402979252533e-06,
      "loss": 0.1052,
      "step": 7953
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.5928587913513184,
      "learning_rate": 4.108260257805314e-06,
      "loss": 0.1232,
      "step": 7954
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7444195747375488,
      "learning_rate": 4.107117584464079e-06,
      "loss": 0.0633,
      "step": 7955
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.368906021118164,
      "learning_rate": 4.105974959290469e-06,
      "loss": 0.1039,
      "step": 7956
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.9065206050872803,
      "learning_rate": 4.104832382346129e-06,
      "loss": 0.9687,
      "step": 7957
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2199618816375732,
      "learning_rate": 4.103689853692691e-06,
      "loss": 0.1,
      "step": 7958
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.183082580566406,
      "learning_rate": 4.102547373391794e-06,
      "loss": 0.8616,
      "step": 7959
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.181315898895264,
      "learning_rate": 4.101404941505069e-06,
      "loss": 0.5414,
      "step": 7960
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.012564659118652,
      "learning_rate": 4.100262558094147e-06,
      "loss": 0.727,
      "step": 7961
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4869842529296875,
      "learning_rate": 4.0991202232206526e-06,
      "loss": 0.0973,
      "step": 7962
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7044028043746948,
      "learning_rate": 4.097977936946213e-06,
      "loss": 0.0619,
      "step": 7963
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.9344217777252197,
      "learning_rate": 4.096835699332449e-06,
      "loss": 0.1336,
      "step": 7964
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.148602485656738,
      "learning_rate": 4.09569351044098e-06,
      "loss": 0.4723,
      "step": 7965
    },
    {
      "epoch": 0.57,
      "grad_norm": 18.813602447509766,
      "learning_rate": 4.094551370333424e-06,
      "loss": 0.9378,
      "step": 7966
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.594055652618408,
      "learning_rate": 4.093409279071392e-06,
      "loss": 0.7834,
      "step": 7967
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.939495801925659,
      "learning_rate": 4.092267236716501e-06,
      "loss": 0.8124,
      "step": 7968
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.4113242626190186,
      "learning_rate": 4.091125243330355e-06,
      "loss": 0.2636,
      "step": 7969
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.725740671157837,
      "learning_rate": 4.0899832989745605e-06,
      "loss": 0.1717,
      "step": 7970
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.092796325683594,
      "learning_rate": 4.088841403710725e-06,
      "loss": 0.6699,
      "step": 7971
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.028398036956787,
      "learning_rate": 4.087699557600447e-06,
      "loss": 0.1325,
      "step": 7972
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.948461651802063,
      "learning_rate": 4.086557760705323e-06,
      "loss": 0.069,
      "step": 7973
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.488203048706055,
      "learning_rate": 4.08541601308695e-06,
      "loss": 0.1682,
      "step": 7974
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.262808322906494,
      "learning_rate": 4.084274314806922e-06,
      "loss": 0.7571,
      "step": 7975
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.004541397094727,
      "learning_rate": 4.083132665926829e-06,
      "loss": 0.8487,
      "step": 7976
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.874647378921509,
      "learning_rate": 4.0819910665082575e-06,
      "loss": 0.1643,
      "step": 7977
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.249675750732422,
      "learning_rate": 4.080849516612792e-06,
      "loss": 0.8798,
      "step": 7978
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.8389289379119873,
      "learning_rate": 4.079708016302017e-06,
      "loss": 0.104,
      "step": 7979
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.8274247646331787,
      "learning_rate": 4.078566565637511e-06,
      "loss": 0.146,
      "step": 7980
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.7823076248168945,
      "learning_rate": 4.0774251646808505e-06,
      "loss": 0.7948,
      "step": 7981
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.901609420776367,
      "learning_rate": 4.076283813493611e-06,
      "loss": 0.8591,
      "step": 7982
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2541873455047607,
      "learning_rate": 4.075142512137363e-06,
      "loss": 0.1634,
      "step": 7983
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.383793830871582,
      "learning_rate": 4.074001260673678e-06,
      "loss": 0.1087,
      "step": 7984
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.4445672035217285,
      "learning_rate": 4.0728600591641165e-06,
      "loss": 0.6354,
      "step": 7985
    },
    {
      "epoch": 0.57,
      "grad_norm": 15.066162109375,
      "learning_rate": 4.0717189076702466e-06,
      "loss": 0.7101,
      "step": 7986
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.6403770446777344,
      "learning_rate": 4.070577806253627e-06,
      "loss": 0.2129,
      "step": 7987
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.426819324493408,
      "learning_rate": 4.069436754975817e-06,
      "loss": 0.1515,
      "step": 7988
    },
    {
      "epoch": 0.57,
      "grad_norm": 17.509929656982422,
      "learning_rate": 4.068295753898373e-06,
      "loss": 1.2478,
      "step": 7989
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.775161743164062,
      "learning_rate": 4.067154803082845e-06,
      "loss": 0.9801,
      "step": 7990
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.913259029388428,
      "learning_rate": 4.0660139025907855e-06,
      "loss": 0.4111,
      "step": 7991
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.551392078399658,
      "learning_rate": 4.0648730524837396e-06,
      "loss": 0.1276,
      "step": 7992
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.712406158447266,
      "learning_rate": 4.063732252823253e-06,
      "loss": 0.7934,
      "step": 7993
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.618640661239624,
      "learning_rate": 4.062591503670869e-06,
      "loss": 0.1186,
      "step": 7994
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.542325019836426,
      "learning_rate": 4.061450805088122e-06,
      "loss": 0.156,
      "step": 7995
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.610164165496826,
      "learning_rate": 4.0603101571365555e-06,
      "loss": 0.2038,
      "step": 7996
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.3497042655944824,
      "learning_rate": 4.059169559877697e-06,
      "loss": 0.1141,
      "step": 7997
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1467204093933105,
      "learning_rate": 4.058029013373079e-06,
      "loss": 0.1116,
      "step": 7998
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.346988201141357,
      "learning_rate": 4.05688851768423e-06,
      "loss": 0.8075,
      "step": 7999
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.1665358543396,
      "learning_rate": 4.055748072872678e-06,
      "loss": 0.7932,
      "step": 8000
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.1048712730407715,
      "learning_rate": 4.05460767899994e-06,
      "loss": 0.9832,
      "step": 8001
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.529154777526855,
      "learning_rate": 4.05346733612754e-06,
      "loss": 1.2113,
      "step": 8002
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.7437140941619873,
      "learning_rate": 4.052327044316996e-06,
      "loss": 0.6221,
      "step": 8003
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.5738210678100586,
      "learning_rate": 4.051186803629819e-06,
      "loss": 0.0615,
      "step": 8004
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.398521661758423,
      "learning_rate": 4.050046614127524e-06,
      "loss": 0.1045,
      "step": 8005
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.047262668609619,
      "learning_rate": 4.048906475871617e-06,
      "loss": 0.0484,
      "step": 8006
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.4860401153564453,
      "learning_rate": 4.047766388923606e-06,
      "loss": 0.5783,
      "step": 8007
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.861310005187988,
      "learning_rate": 4.046626353344995e-06,
      "loss": 0.9949,
      "step": 8008
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2655811309814453,
      "learning_rate": 4.045486369197284e-06,
      "loss": 0.3144,
      "step": 8009
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.509476661682129,
      "learning_rate": 4.044346436541968e-06,
      "loss": 0.837,
      "step": 8010
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.771165609359741,
      "learning_rate": 4.043206555440547e-06,
      "loss": 0.1125,
      "step": 8011
    },
    {
      "epoch": 0.57,
      "grad_norm": 17.876529693603516,
      "learning_rate": 4.042066725954508e-06,
      "loss": 0.6356,
      "step": 8012
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.99342679977417,
      "learning_rate": 4.040926948145344e-06,
      "loss": 0.2133,
      "step": 8013
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.6160452365875244,
      "learning_rate": 4.039787222074542e-06,
      "loss": 0.1429,
      "step": 8014
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.040727376937866,
      "learning_rate": 4.038647547803583e-06,
      "loss": 0.0795,
      "step": 8015
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.656522512435913,
      "learning_rate": 4.0375079253939505e-06,
      "loss": 0.0988,
      "step": 8016
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.29813003540039,
      "learning_rate": 4.036368354907123e-06,
      "loss": 0.9334,
      "step": 8017
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.937209606170654,
      "learning_rate": 4.035228836404574e-06,
      "loss": 0.4037,
      "step": 8018
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.957895278930664,
      "learning_rate": 4.034089369947779e-06,
      "loss": 0.8355,
      "step": 8019
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.067728042602539,
      "learning_rate": 4.032949955598207e-06,
      "loss": 0.714,
      "step": 8020
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.31679105758667,
      "learning_rate": 4.031810593417322e-06,
      "loss": 0.3876,
      "step": 8021
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.8042337894439697,
      "learning_rate": 4.030671283466592e-06,
      "loss": 0.1747,
      "step": 8022
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.512502431869507,
      "learning_rate": 4.029532025807476e-06,
      "loss": 0.1499,
      "step": 8023
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.242558002471924,
      "learning_rate": 4.028392820501435e-06,
      "loss": 0.2586,
      "step": 8024
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.9857170581817627,
      "learning_rate": 4.027253667609923e-06,
      "loss": 0.8712,
      "step": 8025
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.54345703125,
      "learning_rate": 4.0261145671943935e-06,
      "loss": 0.0852,
      "step": 8026
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.8651788234710693,
      "learning_rate": 4.024975519316297e-06,
      "loss": 0.1306,
      "step": 8027
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.855976104736328,
      "learning_rate": 4.023836524037081e-06,
      "loss": 0.7539,
      "step": 8028
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.139892578125,
      "learning_rate": 4.022697581418188e-06,
      "loss": 0.5746,
      "step": 8029
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.66400408744812,
      "learning_rate": 4.0215586915210616e-06,
      "loss": 0.0797,
      "step": 8030
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0754199028015137,
      "learning_rate": 4.0204198544071415e-06,
      "loss": 0.0739,
      "step": 8031
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.7664546966552734,
      "learning_rate": 4.019281070137864e-06,
      "loss": 0.1621,
      "step": 8032
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3581392765045166,
      "learning_rate": 4.018142338774657e-06,
      "loss": 0.1002,
      "step": 8033
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.22780704498291,
      "learning_rate": 4.017003660378957e-06,
      "loss": 0.5256,
      "step": 8034
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.483342170715332,
      "learning_rate": 4.015865035012186e-06,
      "loss": 0.0928,
      "step": 8035
    },
    {
      "epoch": 0.58,
      "grad_norm": 13.970250129699707,
      "learning_rate": 4.014726462735772e-06,
      "loss": 0.7896,
      "step": 8036
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.91591739654541,
      "learning_rate": 4.013587943611137e-06,
      "loss": 1.2497,
      "step": 8037
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.7986817359924316,
      "learning_rate": 4.012449477699697e-06,
      "loss": 0.273,
      "step": 8038
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6591665744781494,
      "learning_rate": 4.011311065062871e-06,
      "loss": 0.1173,
      "step": 8039
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.543933153152466,
      "learning_rate": 4.01017270576207e-06,
      "loss": 0.1731,
      "step": 8040
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9787836074829102,
      "learning_rate": 4.009034399858704e-06,
      "loss": 0.1185,
      "step": 8041
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.495797634124756,
      "learning_rate": 4.007896147414183e-06,
      "loss": 0.8059,
      "step": 8042
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.073795318603516,
      "learning_rate": 4.0067579484899084e-06,
      "loss": 0.7382,
      "step": 8043
    },
    {
      "epoch": 0.58,
      "grad_norm": 20.863204956054688,
      "learning_rate": 4.005619803147285e-06,
      "loss": 0.6377,
      "step": 8044
    },
    {
      "epoch": 0.58,
      "grad_norm": 20.644166946411133,
      "learning_rate": 4.004481711447707e-06,
      "loss": 0.7993,
      "step": 8045
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.993662357330322,
      "learning_rate": 4.003343673452573e-06,
      "loss": 0.3769,
      "step": 8046
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.94357681274414,
      "learning_rate": 4.002205689223274e-06,
      "loss": 0.6373,
      "step": 8047
    },
    {
      "epoch": 0.58,
      "grad_norm": 17.13591957092285,
      "learning_rate": 4.001067758821203e-06,
      "loss": 1.1032,
      "step": 8048
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.1051154136657715,
      "learning_rate": 3.999929882307744e-06,
      "loss": 0.2801,
      "step": 8049
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.787211418151855,
      "learning_rate": 3.9987920597442834e-06,
      "loss": 0.6078,
      "step": 8050
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.171907663345337,
      "learning_rate": 3.997654291192201e-06,
      "loss": 0.1174,
      "step": 8051
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8108196258544922,
      "learning_rate": 3.9965165767128745e-06,
      "loss": 0.1038,
      "step": 8052
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.340242624282837,
      "learning_rate": 3.995378916367681e-06,
      "loss": 0.1301,
      "step": 8053
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.33601450920105,
      "learning_rate": 3.994241310217993e-06,
      "loss": 0.1375,
      "step": 8054
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5250837802886963,
      "learning_rate": 3.993103758325179e-06,
      "loss": 0.2229,
      "step": 8055
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.0366010665893555,
      "learning_rate": 3.991966260750605e-06,
      "loss": 0.7082,
      "step": 8056
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.197478771209717,
      "learning_rate": 3.990828817555639e-06,
      "loss": 0.5551,
      "step": 8057
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.0357298851013184,
      "learning_rate": 3.9896914288016344e-06,
      "loss": 0.1353,
      "step": 8058
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.636610269546509,
      "learning_rate": 3.988554094549955e-06,
      "loss": 0.1218,
      "step": 8059
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.97157096862793,
      "learning_rate": 3.987416814861953e-06,
      "loss": 0.6188,
      "step": 8060
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.209071636199951,
      "learning_rate": 3.986279589798981e-06,
      "loss": 0.1328,
      "step": 8061
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.049313068389893,
      "learning_rate": 3.985142419422387e-06,
      "loss": 0.6997,
      "step": 8062
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.411134719848633,
      "learning_rate": 3.98400530379352e-06,
      "loss": 0.5423,
      "step": 8063
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.135297775268555,
      "learning_rate": 3.982868242973719e-06,
      "loss": 0.4861,
      "step": 8064
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.6415276527404785,
      "learning_rate": 3.981731237024328e-06,
      "loss": 0.8021,
      "step": 8065
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.476770401000977,
      "learning_rate": 3.980594286006681e-06,
      "loss": 0.5332,
      "step": 8066
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.561014175415039,
      "learning_rate": 3.979457389982114e-06,
      "loss": 0.9596,
      "step": 8067
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.372974157333374,
      "learning_rate": 3.978320549011959e-06,
      "loss": 0.2383,
      "step": 8068
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.0519938468933105,
      "learning_rate": 3.977183763157544e-06,
      "loss": 0.1943,
      "step": 8069
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.4932780265808105,
      "learning_rate": 3.976047032480191e-06,
      "loss": 0.1176,
      "step": 8070
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.16965389251709,
      "learning_rate": 3.974910357041226e-06,
      "loss": 0.071,
      "step": 8071
    },
    {
      "epoch": 0.58,
      "grad_norm": 11.41592788696289,
      "learning_rate": 3.973773736901966e-06,
      "loss": 0.9052,
      "step": 8072
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.010721683502197,
      "learning_rate": 3.972637172123728e-06,
      "loss": 0.7932,
      "step": 8073
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.827730655670166,
      "learning_rate": 3.971500662767828e-06,
      "loss": 0.7692,
      "step": 8074
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.2592387199401855,
      "learning_rate": 3.970364208895572e-06,
      "loss": 0.9504,
      "step": 8075
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.09920597076416,
      "learning_rate": 3.969227810568271e-06,
      "loss": 0.7992,
      "step": 8076
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5144450664520264,
      "learning_rate": 3.968091467847227e-06,
      "loss": 0.1368,
      "step": 8077
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6549291610717773,
      "learning_rate": 3.966955180793743e-06,
      "loss": 0.106,
      "step": 8078
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.04086446762085,
      "learning_rate": 3.965818949469116e-06,
      "loss": 0.8889,
      "step": 8079
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.626739501953125,
      "learning_rate": 3.964682773934642e-06,
      "loss": 0.1888,
      "step": 8080
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.3786020278930664,
      "learning_rate": 3.963546654251616e-06,
      "loss": 0.1488,
      "step": 8081
    },
    {
      "epoch": 0.58,
      "grad_norm": 13.55936336517334,
      "learning_rate": 3.962410590481322e-06,
      "loss": 0.6299,
      "step": 8082
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3829710483551025,
      "learning_rate": 3.961274582685051e-06,
      "loss": 0.2122,
      "step": 8083
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.781795501708984,
      "learning_rate": 3.960138630924083e-06,
      "loss": 0.7326,
      "step": 8084
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.4469010829925537,
      "learning_rate": 3.9590027352597e-06,
      "loss": 0.0993,
      "step": 8085
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.070562362670898,
      "learning_rate": 3.957866895753179e-06,
      "loss": 0.7372,
      "step": 8086
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.102075099945068,
      "learning_rate": 3.9567311124657935e-06,
      "loss": 0.8488,
      "step": 8087
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.9373526573181152,
      "learning_rate": 3.955595385458817e-06,
      "loss": 0.2116,
      "step": 8088
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.7693328857421875,
      "learning_rate": 3.954459714793515e-06,
      "loss": 0.2765,
      "step": 8089
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.491023063659668,
      "learning_rate": 3.953324100531155e-06,
      "loss": 0.8717,
      "step": 8090
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.921842098236084,
      "learning_rate": 3.952188542732998e-06,
      "loss": 0.1868,
      "step": 8091
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.571900367736816,
      "learning_rate": 3.951053041460301e-06,
      "loss": 0.7582,
      "step": 8092
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.015946388244629,
      "learning_rate": 3.9499175967743244e-06,
      "loss": 0.9125,
      "step": 8093
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.462875366210938,
      "learning_rate": 3.948782208736319e-06,
      "loss": 1.5081,
      "step": 8094
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3802051544189453,
      "learning_rate": 3.947646877407532e-06,
      "loss": 0.1013,
      "step": 8095
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.791674852371216,
      "learning_rate": 3.946511602849213e-06,
      "loss": 0.1547,
      "step": 8096
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5423946380615234,
      "learning_rate": 3.945376385122606e-06,
      "loss": 0.1117,
      "step": 8097
    },
    {
      "epoch": 0.58,
      "grad_norm": 18.4996337890625,
      "learning_rate": 3.944241224288951e-06,
      "loss": 1.0231,
      "step": 8098
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.387173652648926,
      "learning_rate": 3.943106120409486e-06,
      "loss": 0.9116,
      "step": 8099
    },
    {
      "epoch": 0.58,
      "grad_norm": 26.88948631286621,
      "learning_rate": 3.9419710735454444e-06,
      "loss": 1.5148,
      "step": 8100
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.417145252227783,
      "learning_rate": 3.940836083758058e-06,
      "loss": 0.772,
      "step": 8101
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5837888717651367,
      "learning_rate": 3.939701151108557e-06,
      "loss": 0.1187,
      "step": 8102
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.466192245483398,
      "learning_rate": 3.938566275658165e-06,
      "loss": 0.8972,
      "step": 8103
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.2876524925231934,
      "learning_rate": 3.937431457468105e-06,
      "loss": 0.0714,
      "step": 8104
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6287665367126465,
      "learning_rate": 3.936296696599597e-06,
      "loss": 0.1172,
      "step": 8105
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.326751232147217,
      "learning_rate": 3.9351619931138545e-06,
      "loss": 0.0964,
      "step": 8106
    },
    {
      "epoch": 0.58,
      "grad_norm": 24.3182373046875,
      "learning_rate": 3.934027347072091e-06,
      "loss": 1.048,
      "step": 8107
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.873851776123047,
      "learning_rate": 3.932892758535519e-06,
      "loss": 0.7183,
      "step": 8108
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.158812522888184,
      "learning_rate": 3.931758227565342e-06,
      "loss": 0.519,
      "step": 8109
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6533210277557373,
      "learning_rate": 3.930623754222765e-06,
      "loss": 0.2246,
      "step": 8110
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.060220956802368,
      "learning_rate": 3.92948933856899e-06,
      "loss": 0.2347,
      "step": 8111
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.766739845275879,
      "learning_rate": 3.9283549806652115e-06,
      "loss": 0.5663,
      "step": 8112
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3286826610565186,
      "learning_rate": 3.927220680572626e-06,
      "loss": 0.0781,
      "step": 8113
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.938175201416016,
      "learning_rate": 3.926086438352425e-06,
      "loss": 0.5833,
      "step": 8114
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.715244293212891,
      "learning_rate": 3.9249522540657945e-06,
      "loss": 0.5402,
      "step": 8115
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.3877997398376465,
      "learning_rate": 3.923818127773921e-06,
      "loss": 0.8152,
      "step": 8116
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.950321197509766,
      "learning_rate": 3.9226840595379875e-06,
      "loss": 0.8156,
      "step": 8117
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.347885847091675,
      "learning_rate": 3.92155004941917e-06,
      "loss": 0.1399,
      "step": 8118
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.658651828765869,
      "learning_rate": 3.920416097478644e-06,
      "loss": 0.1761,
      "step": 8119
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.884446620941162,
      "learning_rate": 3.9192822037775845e-06,
      "loss": 0.1982,
      "step": 8120
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.1944146156311035,
      "learning_rate": 3.918148368377158e-06,
      "loss": 0.1673,
      "step": 8121
    },
    {
      "epoch": 0.58,
      "grad_norm": 13.503876686096191,
      "learning_rate": 3.917014591338534e-06,
      "loss": 0.5982,
      "step": 8122
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.609069585800171,
      "learning_rate": 3.915880872722872e-06,
      "loss": 0.1538,
      "step": 8123
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.269571304321289,
      "learning_rate": 3.914747212591332e-06,
      "loss": 0.908,
      "step": 8124
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3793540000915527,
      "learning_rate": 3.913613611005075e-06,
      "loss": 0.0954,
      "step": 8125
    },
    {
      "epoch": 0.58,
      "grad_norm": 14.436097145080566,
      "learning_rate": 3.9124800680252496e-06,
      "loss": 0.8771,
      "step": 8126
    },
    {
      "epoch": 0.58,
      "grad_norm": 18.521060943603516,
      "learning_rate": 3.911346583713009e-06,
      "loss": 1.0758,
      "step": 8127
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.395164489746094,
      "learning_rate": 3.910213158129498e-06,
      "loss": 0.4948,
      "step": 8128
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.031708717346191,
      "learning_rate": 3.909079791335864e-06,
      "loss": 0.8533,
      "step": 8129
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2490720748901367,
      "learning_rate": 3.907946483393245e-06,
      "loss": 0.1984,
      "step": 8130
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3295106887817383,
      "learning_rate": 3.90681323436278e-06,
      "loss": 0.0953,
      "step": 8131
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.548451900482178,
      "learning_rate": 3.905680044305602e-06,
      "loss": 0.3,
      "step": 8132
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.753633975982666,
      "learning_rate": 3.904546913282844e-06,
      "loss": 0.1121,
      "step": 8133
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.878664493560791,
      "learning_rate": 3.903413841355633e-06,
      "loss": 0.13,
      "step": 8134
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.160658359527588,
      "learning_rate": 3.902280828585095e-06,
      "loss": 0.6753,
      "step": 8135
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.396796226501465,
      "learning_rate": 3.9011478750323515e-06,
      "loss": 0.9248,
      "step": 8136
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.893576145172119,
      "learning_rate": 3.900014980758519e-06,
      "loss": 0.1925,
      "step": 8137
    },
    {
      "epoch": 0.58,
      "grad_norm": 11.94554615020752,
      "learning_rate": 3.898882145824716e-06,
      "loss": 1.096,
      "step": 8138
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.2477385997772217,
      "learning_rate": 3.897749370292054e-06,
      "loss": 0.172,
      "step": 8139
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.225586891174316,
      "learning_rate": 3.89661665422164e-06,
      "loss": 0.924,
      "step": 8140
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.541251182556152,
      "learning_rate": 3.895483997674583e-06,
      "loss": 0.5356,
      "step": 8141
    },
    {
      "epoch": 0.58,
      "grad_norm": 13.518221855163574,
      "learning_rate": 3.894351400711982e-06,
      "loss": 1.3793,
      "step": 8142
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.374350070953369,
      "learning_rate": 3.893218863394937e-06,
      "loss": 0.5015,
      "step": 8143
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.7093284130096436,
      "learning_rate": 3.892086385784546e-06,
      "loss": 0.6986,
      "step": 8144
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.816980361938477,
      "learning_rate": 3.8909539679419016e-06,
      "loss": 0.9845,
      "step": 8145
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.464593887329102,
      "learning_rate": 3.88982160992809e-06,
      "loss": 0.5888,
      "step": 8146
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.933154821395874,
      "learning_rate": 3.8886893118042015e-06,
      "loss": 0.8388,
      "step": 8147
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.720694541931152,
      "learning_rate": 3.887557073631319e-06,
      "loss": 0.7998,
      "step": 8148
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.512732982635498,
      "learning_rate": 3.886424895470521e-06,
      "loss": 0.1866,
      "step": 8149
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.087643146514893,
      "learning_rate": 3.885292777382885e-06,
      "loss": 0.7625,
      "step": 8150
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6127192974090576,
      "learning_rate": 3.884160719429484e-06,
      "loss": 0.1231,
      "step": 8151
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.90086555480957,
      "learning_rate": 3.883028721671387e-06,
      "loss": 1.1645,
      "step": 8152
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.679823875427246,
      "learning_rate": 3.881896784169666e-06,
      "loss": 0.3614,
      "step": 8153
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.551179885864258,
      "learning_rate": 3.88076490698538e-06,
      "loss": 0.4422,
      "step": 8154
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.035283088684082,
      "learning_rate": 3.87963309017959e-06,
      "loss": 0.4974,
      "step": 8155
    },
    {
      "epoch": 0.58,
      "grad_norm": 13.115300178527832,
      "learning_rate": 3.878501333813354e-06,
      "loss": 0.5719,
      "step": 8156
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.873077630996704,
      "learning_rate": 3.877369637947725e-06,
      "loss": 0.9036,
      "step": 8157
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3748443126678467,
      "learning_rate": 3.876238002643756e-06,
      "loss": 0.0895,
      "step": 8158
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.5688881874084473,
      "learning_rate": 3.875106427962493e-06,
      "loss": 0.2981,
      "step": 8159
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2172369956970215,
      "learning_rate": 3.873974913964979e-06,
      "loss": 0.11,
      "step": 8160
    },
    {
      "epoch": 0.59,
      "grad_norm": 18.27913475036621,
      "learning_rate": 3.8728434607122566e-06,
      "loss": 1.1284,
      "step": 8161
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.074932813644409,
      "learning_rate": 3.871712068265364e-06,
      "loss": 0.2705,
      "step": 8162
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.294479370117188,
      "learning_rate": 3.870580736685334e-06,
      "loss": 1.1304,
      "step": 8163
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.27378511428833,
      "learning_rate": 3.869449466033198e-06,
      "loss": 0.2951,
      "step": 8164
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9250589609146118,
      "learning_rate": 3.8683182563699866e-06,
      "loss": 0.0691,
      "step": 8165
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.904792547225952,
      "learning_rate": 3.867187107756718e-06,
      "loss": 0.7312,
      "step": 8166
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.205113887786865,
      "learning_rate": 3.866056020254418e-06,
      "loss": 0.9044,
      "step": 8167
    },
    {
      "epoch": 0.59,
      "grad_norm": 18.107717514038086,
      "learning_rate": 3.864924993924104e-06,
      "loss": 1.7245,
      "step": 8168
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2050530910491943,
      "learning_rate": 3.863794028826789e-06,
      "loss": 0.107,
      "step": 8169
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.566988945007324,
      "learning_rate": 3.862663125023487e-06,
      "loss": 0.9796,
      "step": 8170
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.52100944519043,
      "learning_rate": 3.8615322825752025e-06,
      "loss": 1.3209,
      "step": 8171
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.7150816917419434,
      "learning_rate": 3.860401501542943e-06,
      "loss": 0.6191,
      "step": 8172
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9923932552337646,
      "learning_rate": 3.8592707819877084e-06,
      "loss": 0.1978,
      "step": 8173
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.20219087600708,
      "learning_rate": 3.858140123970498e-06,
      "loss": 0.9509,
      "step": 8174
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.1917569637298584,
      "learning_rate": 3.8570095275523045e-06,
      "loss": 0.1466,
      "step": 8175
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.78157901763916,
      "learning_rate": 3.855878992794122e-06,
      "loss": 0.544,
      "step": 8176
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.906134605407715,
      "learning_rate": 3.8547485197569375e-06,
      "loss": 0.8408,
      "step": 8177
    },
    {
      "epoch": 0.59,
      "grad_norm": 13.520547866821289,
      "learning_rate": 3.853618108501735e-06,
      "loss": 0.3996,
      "step": 8178
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.949639320373535,
      "learning_rate": 3.852487759089495e-06,
      "loss": 1.2298,
      "step": 8179
    },
    {
      "epoch": 0.59,
      "grad_norm": 20.607107162475586,
      "learning_rate": 3.851357471581197e-06,
      "loss": 1.4234,
      "step": 8180
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.8253252506256104,
      "learning_rate": 3.850227246037816e-06,
      "loss": 0.2419,
      "step": 8181
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.258626461029053,
      "learning_rate": 3.849097082520323e-06,
      "loss": 0.5533,
      "step": 8182
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.160612106323242,
      "learning_rate": 3.847966981089686e-06,
      "loss": 0.6393,
      "step": 8183
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.266855716705322,
      "learning_rate": 3.846836941806869e-06,
      "loss": 0.8025,
      "step": 8184
    },
    {
      "epoch": 0.59,
      "grad_norm": 13.615227699279785,
      "learning_rate": 3.845706964732835e-06,
      "loss": 0.8935,
      "step": 8185
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6073840856552124,
      "learning_rate": 3.84457704992854e-06,
      "loss": 0.0732,
      "step": 8186
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.255960941314697,
      "learning_rate": 3.843447197454941e-06,
      "loss": 0.2209,
      "step": 8187
    },
    {
      "epoch": 0.59,
      "grad_norm": 13.06768798828125,
      "learning_rate": 3.8423174073729865e-06,
      "loss": 1.033,
      "step": 8188
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.639479637145996,
      "learning_rate": 3.8411876797436285e-06,
      "loss": 0.182,
      "step": 8189
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.4116530418395996,
      "learning_rate": 3.840058014627806e-06,
      "loss": 0.1405,
      "step": 8190
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.71278190612793,
      "learning_rate": 3.838928412086464e-06,
      "loss": 0.9711,
      "step": 8191
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9780936241149902,
      "learning_rate": 3.837798872180539e-06,
      "loss": 0.1543,
      "step": 8192
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.8003106117248535,
      "learning_rate": 3.836669394970966e-06,
      "loss": 0.2585,
      "step": 8193
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.9192800521850586,
      "learning_rate": 3.835539980518675e-06,
      "loss": 0.8368,
      "step": 8194
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9203546047210693,
      "learning_rate": 3.834410628884595e-06,
      "loss": 0.1815,
      "step": 8195
    },
    {
      "epoch": 0.59,
      "grad_norm": 18.081741333007812,
      "learning_rate": 3.83328134012965e-06,
      "loss": 1.5829,
      "step": 8196
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.6348531246185303,
      "learning_rate": 3.832152114314759e-06,
      "loss": 0.1376,
      "step": 8197
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.223515272140503,
      "learning_rate": 3.831022951500842e-06,
      "loss": 0.087,
      "step": 8198
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.009913444519043,
      "learning_rate": 3.829893851748811e-06,
      "loss": 0.8752,
      "step": 8199
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.987825870513916,
      "learning_rate": 3.828764815119579e-06,
      "loss": 0.7693,
      "step": 8200
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.448893070220947,
      "learning_rate": 3.827635841674052e-06,
      "loss": 0.3952,
      "step": 8201
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.466829538345337,
      "learning_rate": 3.8265069314731335e-06,
      "loss": 0.1487,
      "step": 8202
    },
    {
      "epoch": 0.59,
      "grad_norm": 12.951789855957031,
      "learning_rate": 3.825378084577723e-06,
      "loss": 0.946,
      "step": 8203
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.2782764434814453,
      "learning_rate": 3.8242493010487185e-06,
      "loss": 0.2549,
      "step": 8204
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.514158248901367,
      "learning_rate": 3.8231205809470154e-06,
      "loss": 1.0054,
      "step": 8205
    },
    {
      "epoch": 0.59,
      "grad_norm": 13.002402305603027,
      "learning_rate": 3.8219919243335005e-06,
      "loss": 1.3535,
      "step": 8206
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.928370714187622,
      "learning_rate": 3.820863331269064e-06,
      "loss": 0.1603,
      "step": 8207
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.878714561462402,
      "learning_rate": 3.819734801814585e-06,
      "loss": 0.7541,
      "step": 8208
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.84238338470459,
      "learning_rate": 3.8186063360309475e-06,
      "loss": 0.7437,
      "step": 8209
    },
    {
      "epoch": 0.59,
      "grad_norm": 15.536859512329102,
      "learning_rate": 3.817477933979027e-06,
      "loss": 0.986,
      "step": 8210
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3661766052246094,
      "learning_rate": 3.816349595719694e-06,
      "loss": 0.0723,
      "step": 8211
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.12774658203125,
      "learning_rate": 3.8152213213138215e-06,
      "loss": 0.9389,
      "step": 8212
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.154096603393555,
      "learning_rate": 3.8140931108222744e-06,
      "loss": 1.048,
      "step": 8213
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.467707395553589,
      "learning_rate": 3.8129649643059146e-06,
      "loss": 0.1139,
      "step": 8214
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3086421489715576,
      "learning_rate": 3.8118368818256012e-06,
      "loss": 0.0975,
      "step": 8215
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.4679553508758545,
      "learning_rate": 3.8107088634421913e-06,
      "loss": 0.0835,
      "step": 8216
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.767235517501831,
      "learning_rate": 3.809580909216536e-06,
      "loss": 0.2051,
      "step": 8217
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.13277530670166,
      "learning_rate": 3.808453019209484e-06,
      "loss": 0.1706,
      "step": 8218
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.42680549621582,
      "learning_rate": 3.8073251934818827e-06,
      "loss": 0.7526,
      "step": 8219
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.030432939529419,
      "learning_rate": 3.8061974320945717e-06,
      "loss": 0.1024,
      "step": 8220
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.5914149284362793,
      "learning_rate": 3.8050697351083916e-06,
      "loss": 0.088,
      "step": 8221
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.757754802703857,
      "learning_rate": 3.803942102584175e-06,
      "loss": 0.7039,
      "step": 8222
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.13441801071167,
      "learning_rate": 3.8028145345827543e-06,
      "loss": 0.0876,
      "step": 8223
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.4501614570617676,
      "learning_rate": 3.8016870311649597e-06,
      "loss": 0.0977,
      "step": 8224
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.211184978485107,
      "learning_rate": 3.800559592391614e-06,
      "loss": 0.7806,
      "step": 8225
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2543296813964844,
      "learning_rate": 3.7994322183235367e-06,
      "loss": 0.1402,
      "step": 8226
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.5407211780548096,
      "learning_rate": 3.798304909021547e-06,
      "loss": 0.1366,
      "step": 8227
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2256219387054443,
      "learning_rate": 3.797177664546459e-06,
      "loss": 0.0972,
      "step": 8228
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.201181411743164,
      "learning_rate": 3.796050484959083e-06,
      "loss": 0.0969,
      "step": 8229
    },
    {
      "epoch": 0.59,
      "grad_norm": 12.251989364624023,
      "learning_rate": 3.7949233703202265e-06,
      "loss": 0.7701,
      "step": 8230
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.572178363800049,
      "learning_rate": 3.793796320690692e-06,
      "loss": 0.8907,
      "step": 8231
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0651743412017822,
      "learning_rate": 3.7926693361312806e-06,
      "loss": 0.1641,
      "step": 8232
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.77278995513916,
      "learning_rate": 3.7915424167027885e-06,
      "loss": 0.5144,
      "step": 8233
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.931497573852539,
      "learning_rate": 3.790415562466009e-06,
      "loss": 0.2508,
      "step": 8234
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3970072269439697,
      "learning_rate": 3.7892887734817315e-06,
      "loss": 0.1172,
      "step": 8235
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3365907669067383,
      "learning_rate": 3.788162049810741e-06,
      "loss": 0.076,
      "step": 8236
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.76926040649414,
      "learning_rate": 3.787035391513823e-06,
      "loss": 0.5787,
      "step": 8237
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.283300399780273,
      "learning_rate": 3.7859087986517524e-06,
      "loss": 0.8806,
      "step": 8238
    },
    {
      "epoch": 0.59,
      "grad_norm": 14.059928894042969,
      "learning_rate": 3.7847822712853077e-06,
      "loss": 0.9034,
      "step": 8239
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.020460605621338,
      "learning_rate": 3.7836558094752583e-06,
      "loss": 0.8923,
      "step": 8240
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.6864302158355713,
      "learning_rate": 3.7825294132823738e-06,
      "loss": 0.1257,
      "step": 8241
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1031248569488525,
      "learning_rate": 3.78140308276742e-06,
      "loss": 0.0448,
      "step": 8242
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.6756086349487305,
      "learning_rate": 3.7802768179911566e-06,
      "loss": 0.3113,
      "step": 8243
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.990445852279663,
      "learning_rate": 3.7791506190143427e-06,
      "loss": 0.8736,
      "step": 8244
    },
    {
      "epoch": 0.59,
      "grad_norm": 17.603485107421875,
      "learning_rate": 3.778024485897731e-06,
      "loss": 1.291,
      "step": 8245
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.063016414642334,
      "learning_rate": 3.7768984187020728e-06,
      "loss": 0.7566,
      "step": 8246
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.221529483795166,
      "learning_rate": 3.775772417488116e-06,
      "loss": 0.9242,
      "step": 8247
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.79159688949585,
      "learning_rate": 3.7746464823166024e-06,
      "loss": 0.4569,
      "step": 8248
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.2190451622009277,
      "learning_rate": 3.7735206132482748e-06,
      "loss": 0.199,
      "step": 8249
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.195439577102661,
      "learning_rate": 3.7723948103438675e-06,
      "loss": 0.1797,
      "step": 8250
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.996753692626953,
      "learning_rate": 3.7712690736641113e-06,
      "loss": 0.6187,
      "step": 8251
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.39180850982666,
      "learning_rate": 3.770143403269738e-06,
      "loss": 0.93,
      "step": 8252
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.8469398021698,
      "learning_rate": 3.769017799221474e-06,
      "loss": 0.1129,
      "step": 8253
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.784363269805908,
      "learning_rate": 3.767892261580039e-06,
      "loss": 0.5609,
      "step": 8254
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.178789138793945,
      "learning_rate": 3.7667667904061524e-06,
      "loss": 0.5483,
      "step": 8255
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.6583871841430664,
      "learning_rate": 3.765641385760531e-06,
      "loss": 0.186,
      "step": 8256
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1913623809814453,
      "learning_rate": 3.7645160477038835e-06,
      "loss": 0.1624,
      "step": 8257
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.790318012237549,
      "learning_rate": 3.763390776296919e-06,
      "loss": 0.1657,
      "step": 8258
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.7225277423858643,
      "learning_rate": 3.762265571600341e-06,
      "loss": 0.4296,
      "step": 8259
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.1325860023498535,
      "learning_rate": 3.7611404336748497e-06,
      "loss": 0.7418,
      "step": 8260
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.520136833190918,
      "learning_rate": 3.760015362581145e-06,
      "loss": 0.6317,
      "step": 8261
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.7103118896484375,
      "learning_rate": 3.7588903583799172e-06,
      "loss": 0.7994,
      "step": 8262
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7254897356033325,
      "learning_rate": 3.7577654211318556e-06,
      "loss": 0.0753,
      "step": 8263
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.854840278625488,
      "learning_rate": 3.7566405508976483e-06,
      "loss": 0.4491,
      "step": 8264
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.713527679443359,
      "learning_rate": 3.7555157477379757e-06,
      "loss": 0.6895,
      "step": 8265
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.665709972381592,
      "learning_rate": 3.754391011713518e-06,
      "loss": 0.5638,
      "step": 8266
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.9456968307495117,
      "learning_rate": 3.7532663428849514e-06,
      "loss": 0.8598,
      "step": 8267
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.372281074523926,
      "learning_rate": 3.752141741312946e-06,
      "loss": 0.6998,
      "step": 8268
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.016573905944824,
      "learning_rate": 3.751017207058169e-06,
      "loss": 0.7962,
      "step": 8269
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.669947862625122,
      "learning_rate": 3.7498927401812877e-06,
      "loss": 0.1341,
      "step": 8270
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.807266235351562,
      "learning_rate": 3.7487683407429605e-06,
      "loss": 0.4819,
      "step": 8271
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1829471588134766,
      "learning_rate": 3.7476440088038456e-06,
      "loss": 0.0531,
      "step": 8272
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.460148811340332,
      "learning_rate": 3.7465197444245953e-06,
      "loss": 0.1602,
      "step": 8273
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.532694101333618,
      "learning_rate": 3.745395547665862e-06,
      "loss": 0.0789,
      "step": 8274
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.748413324356079,
      "learning_rate": 3.744271418588288e-06,
      "loss": 0.1547,
      "step": 8275
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.284336566925049,
      "learning_rate": 3.743147357252519e-06,
      "loss": 0.1424,
      "step": 8276
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.400176525115967,
      "learning_rate": 3.7420233637191916e-06,
      "loss": 0.0869,
      "step": 8277
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5356621742248535,
      "learning_rate": 3.7408994380489415e-06,
      "loss": 0.0373,
      "step": 8278
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9609408378601074,
      "learning_rate": 3.7397755803024023e-06,
      "loss": 0.1896,
      "step": 8279
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.8408212661743164,
      "learning_rate": 3.7386517905401988e-06,
      "loss": 0.1794,
      "step": 8280
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.419656753540039,
      "learning_rate": 3.7375280688229585e-06,
      "loss": 0.1735,
      "step": 8281
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.617229700088501,
      "learning_rate": 3.7364044152112987e-06,
      "loss": 0.2996,
      "step": 8282
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.2685546875,
      "learning_rate": 3.735280829765837e-06,
      "loss": 0.8641,
      "step": 8283
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.146506309509277,
      "learning_rate": 3.7341573125471895e-06,
      "loss": 0.2385,
      "step": 8284
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.9277920722961426,
      "learning_rate": 3.7330338636159626e-06,
      "loss": 0.9183,
      "step": 8285
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.276070594787598,
      "learning_rate": 3.7319104830327647e-06,
      "loss": 0.7446,
      "step": 8286
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0922398567199707,
      "learning_rate": 3.7307871708581945e-06,
      "loss": 0.1074,
      "step": 8287
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.309631109237671,
      "learning_rate": 3.7296639271528523e-06,
      "loss": 0.0951,
      "step": 8288
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.768418788909912,
      "learning_rate": 3.728540751977332e-06,
      "loss": 0.9051,
      "step": 8289
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.789355278015137,
      "learning_rate": 3.7274176453922263e-06,
      "loss": 0.6077,
      "step": 8290
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.4736831188201904,
      "learning_rate": 3.726294607458121e-06,
      "loss": 0.1575,
      "step": 8291
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.034017562866211,
      "learning_rate": 3.7251716382356007e-06,
      "loss": 0.7727,
      "step": 8292
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.9086437225341797,
      "learning_rate": 3.724048737785245e-06,
      "loss": 0.7721,
      "step": 8293
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3261196613311768,
      "learning_rate": 3.7229259061676305e-06,
      "loss": 0.11,
      "step": 8294
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.413597583770752,
      "learning_rate": 3.72180314344333e-06,
      "loss": 0.142,
      "step": 8295
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0434699058532715,
      "learning_rate": 3.72068044967291e-06,
      "loss": 0.1876,
      "step": 8296
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.031951665878296,
      "learning_rate": 3.719557824916937e-06,
      "loss": 0.076,
      "step": 8297
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.518568754196167,
      "learning_rate": 3.718435269235976e-06,
      "loss": 0.084,
      "step": 8298
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.241260290145874,
      "learning_rate": 3.7173127826905796e-06,
      "loss": 0.1925,
      "step": 8299
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.6316394805908203,
      "learning_rate": 3.7161903653413024e-06,
      "loss": 0.1085,
      "step": 8300
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.794508695602417,
      "learning_rate": 3.7150680172486962e-06,
      "loss": 1.0384,
      "step": 8301
    },
    {
      "epoch": 0.6,
      "grad_norm": 26.242704391479492,
      "learning_rate": 3.7139457384733057e-06,
      "loss": 1.632,
      "step": 8302
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.461554765701294,
      "learning_rate": 3.7128235290756747e-06,
      "loss": 0.2632,
      "step": 8303
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.579237222671509,
      "learning_rate": 3.711701389116343e-06,
      "loss": 0.1473,
      "step": 8304
    },
    {
      "epoch": 0.6,
      "grad_norm": 21.47517204284668,
      "learning_rate": 3.7105793186558447e-06,
      "loss": 0.8159,
      "step": 8305
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.6711065769195557,
      "learning_rate": 3.7094573177547105e-06,
      "loss": 0.2446,
      "step": 8306
    },
    {
      "epoch": 0.6,
      "grad_norm": 12.446288108825684,
      "learning_rate": 3.7083353864734706e-06,
      "loss": 0.7328,
      "step": 8307
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.8162760734558105,
      "learning_rate": 3.707213524872646e-06,
      "loss": 0.121,
      "step": 8308
    },
    {
      "epoch": 0.6,
      "grad_norm": 13.826817512512207,
      "learning_rate": 3.7060917330127592e-06,
      "loss": 0.669,
      "step": 8309
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.657043933868408,
      "learning_rate": 3.7049700109543274e-06,
      "loss": 0.8511,
      "step": 8310
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.9255127906799316,
      "learning_rate": 3.703848358757859e-06,
      "loss": 0.1239,
      "step": 8311
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3699822425842285,
      "learning_rate": 3.7027267764838653e-06,
      "loss": 0.1232,
      "step": 8312
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.475923538208008,
      "learning_rate": 3.7016052641928528e-06,
      "loss": 0.953,
      "step": 8313
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.340325117111206,
      "learning_rate": 3.700483821945321e-06,
      "loss": 0.1641,
      "step": 8314
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.79156231880188,
      "learning_rate": 3.699362449801768e-06,
      "loss": 0.1421,
      "step": 8315
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.851573944091797,
      "learning_rate": 3.6982411478226874e-06,
      "loss": 0.7229,
      "step": 8316
    },
    {
      "epoch": 0.6,
      "grad_norm": 18.290369033813477,
      "learning_rate": 3.6971199160685694e-06,
      "loss": 0.882,
      "step": 8317
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.802679061889648,
      "learning_rate": 3.695998754599901e-06,
      "loss": 0.9161,
      "step": 8318
    },
    {
      "epoch": 0.6,
      "grad_norm": 13.267298698425293,
      "learning_rate": 3.6948776634771633e-06,
      "loss": 0.9075,
      "step": 8319
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.134734630584717,
      "learning_rate": 3.6937566427608352e-06,
      "loss": 0.5841,
      "step": 8320
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.677935838699341,
      "learning_rate": 3.6926356925113926e-06,
      "loss": 0.744,
      "step": 8321
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.626187324523926,
      "learning_rate": 3.6915148127893073e-06,
      "loss": 0.4886,
      "step": 8322
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.059722900390625,
      "learning_rate": 3.690394003655043e-06,
      "loss": 0.7364,
      "step": 8323
    },
    {
      "epoch": 0.6,
      "grad_norm": 19.78825569152832,
      "learning_rate": 3.6892732651690655e-06,
      "loss": 0.8284,
      "step": 8324
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.367741346359253,
      "learning_rate": 3.6881525973918336e-06,
      "loss": 0.1494,
      "step": 8325
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.424652099609375,
      "learning_rate": 3.6870320003838034e-06,
      "loss": 0.713,
      "step": 8326
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.271402359008789,
      "learning_rate": 3.685911474205427e-06,
      "loss": 0.7558,
      "step": 8327
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.52449631690979,
      "learning_rate": 3.684791018917152e-06,
      "loss": 0.1007,
      "step": 8328
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.5700840950012207,
      "learning_rate": 3.6836706345794247e-06,
      "loss": 0.1153,
      "step": 8329
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.1000683307647705,
      "learning_rate": 3.6825503212526824e-06,
      "loss": 0.1719,
      "step": 8330
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7005717754364014,
      "learning_rate": 3.681430078997364e-06,
      "loss": 0.1408,
      "step": 8331
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.930370807647705,
      "learning_rate": 3.6803099078739023e-06,
      "loss": 0.6671,
      "step": 8332
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.877058982849121,
      "learning_rate": 3.679189807942725e-06,
      "loss": 0.1244,
      "step": 8333
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.539001226425171,
      "learning_rate": 3.6780697792642605e-06,
      "loss": 0.106,
      "step": 8334
    },
    {
      "epoch": 0.6,
      "grad_norm": 18.31267547607422,
      "learning_rate": 3.6769498218989252e-06,
      "loss": 0.5805,
      "step": 8335
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2483465671539307,
      "learning_rate": 3.67582993590714e-06,
      "loss": 0.1087,
      "step": 8336
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0655739307403564,
      "learning_rate": 3.6747101213493163e-06,
      "loss": 0.0895,
      "step": 8337
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.537232398986816,
      "learning_rate": 3.6735903782858663e-06,
      "loss": 0.832,
      "step": 8338
    },
    {
      "epoch": 0.6,
      "grad_norm": 22.0687198638916,
      "learning_rate": 3.6724707067771936e-06,
      "loss": 0.7657,
      "step": 8339
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.6501810550689697,
      "learning_rate": 3.671351106883701e-06,
      "loss": 0.116,
      "step": 8340
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.365688323974609,
      "learning_rate": 3.6702315786657882e-06,
      "loss": 0.864,
      "step": 8341
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.141562461853027,
      "learning_rate": 3.6691121221838475e-06,
      "loss": 0.6077,
      "step": 8342
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.072253227233887,
      "learning_rate": 3.667992737498271e-06,
      "loss": 0.7356,
      "step": 8343
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.588998317718506,
      "learning_rate": 3.6668734246694433e-06,
      "loss": 0.8399,
      "step": 8344
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7060816287994385,
      "learning_rate": 3.665754183757748e-06,
      "loss": 0.0842,
      "step": 8345
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.266441822052002,
      "learning_rate": 3.6646350148235665e-06,
      "loss": 0.2457,
      "step": 8346
    },
    {
      "epoch": 0.6,
      "grad_norm": 16.961435317993164,
      "learning_rate": 3.6635159179272705e-06,
      "loss": 1.1901,
      "step": 8347
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4178476333618164,
      "learning_rate": 3.6623968931292296e-06,
      "loss": 0.1046,
      "step": 8348
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.312714576721191,
      "learning_rate": 3.6612779404898146e-06,
      "loss": 0.5103,
      "step": 8349
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.683139443397522,
      "learning_rate": 3.660159060069388e-06,
      "loss": 0.0507,
      "step": 8350
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.449129343032837,
      "learning_rate": 3.6590402519283075e-06,
      "loss": 0.2705,
      "step": 8351
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4044153690338135,
      "learning_rate": 3.6579215161269303e-06,
      "loss": 0.1157,
      "step": 8352
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.194995403289795,
      "learning_rate": 3.656802852725606e-06,
      "loss": 0.8437,
      "step": 8353
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.689835548400879,
      "learning_rate": 3.6556842617846845e-06,
      "loss": 0.3762,
      "step": 8354
    },
    {
      "epoch": 0.6,
      "grad_norm": 21.894006729125977,
      "learning_rate": 3.654565743364509e-06,
      "loss": 0.7514,
      "step": 8355
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.157180309295654,
      "learning_rate": 3.6534472975254175e-06,
      "loss": 0.8669,
      "step": 8356
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.257915496826172,
      "learning_rate": 3.6523289243277493e-06,
      "loss": 0.2184,
      "step": 8357
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.278724193572998,
      "learning_rate": 3.651210623831834e-06,
      "loss": 0.1314,
      "step": 8358
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.6513752937316895,
      "learning_rate": 3.650092396098e-06,
      "loss": 0.1481,
      "step": 8359
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.461151123046875,
      "learning_rate": 3.64897424118657e-06,
      "loss": 0.9189,
      "step": 8360
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.265127420425415,
      "learning_rate": 3.6478561591578676e-06,
      "loss": 0.1298,
      "step": 8361
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7737247943878174,
      "learning_rate": 3.6467381500722065e-06,
      "loss": 0.1132,
      "step": 8362
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.825132369995117,
      "learning_rate": 3.645620213989899e-06,
      "loss": 0.8608,
      "step": 8363
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.807041883468628,
      "learning_rate": 3.6445023509712557e-06,
      "loss": 0.1473,
      "step": 8364
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.16636848449707,
      "learning_rate": 3.6433845610765785e-06,
      "loss": 0.8485,
      "step": 8365
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.549330711364746,
      "learning_rate": 3.642266844366171e-06,
      "loss": 0.3769,
      "step": 8366
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2401320934295654,
      "learning_rate": 3.6411492009003256e-06,
      "loss": 0.0742,
      "step": 8367
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.659311771392822,
      "learning_rate": 3.640031630739338e-06,
      "loss": 0.9651,
      "step": 8368
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.94056510925293,
      "learning_rate": 3.6389141339434976e-06,
      "loss": 0.2971,
      "step": 8369
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.119875192642212,
      "learning_rate": 3.6377967105730884e-06,
      "loss": 0.1603,
      "step": 8370
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.2960383892059326,
      "learning_rate": 3.6366793606883883e-06,
      "loss": 0.4947,
      "step": 8371
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.8975372314453125,
      "learning_rate": 3.6355620843496764e-06,
      "loss": 0.1329,
      "step": 8372
    },
    {
      "epoch": 0.6,
      "grad_norm": 20.134668350219727,
      "learning_rate": 3.634444881617226e-06,
      "loss": 0.8952,
      "step": 8373
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.9712038040161133,
      "learning_rate": 3.6333277525513044e-06,
      "loss": 0.6527,
      "step": 8374
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.726295471191406,
      "learning_rate": 3.6322106972121782e-06,
      "loss": 0.6599,
      "step": 8375
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.838898181915283,
      "learning_rate": 3.631093715660107e-06,
      "loss": 0.1678,
      "step": 8376
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7299838066101074,
      "learning_rate": 3.6299768079553484e-06,
      "loss": 0.111,
      "step": 8377
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.397338628768921,
      "learning_rate": 3.6288599741581554e-06,
      "loss": 0.1536,
      "step": 8378
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.6665754318237305,
      "learning_rate": 3.6277432143287765e-06,
      "loss": 0.3497,
      "step": 8379
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.350260019302368,
      "learning_rate": 3.6266265285274582e-06,
      "loss": 0.1134,
      "step": 8380
    },
    {
      "epoch": 0.6,
      "grad_norm": 13.487424850463867,
      "learning_rate": 3.625509916814438e-06,
      "loss": 0.932,
      "step": 8381
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.5272629261016846,
      "learning_rate": 3.624393379249958e-06,
      "loss": 0.1402,
      "step": 8382
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.188153266906738,
      "learning_rate": 3.623276915894246e-06,
      "loss": 0.5596,
      "step": 8383
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.324127197265625,
      "learning_rate": 3.6221605268075345e-06,
      "loss": 0.1047,
      "step": 8384
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.68947172164917,
      "learning_rate": 3.6210442120500457e-06,
      "loss": 0.1583,
      "step": 8385
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3044934272766113,
      "learning_rate": 3.619927971682002e-06,
      "loss": 0.0918,
      "step": 8386
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.619602203369141,
      "learning_rate": 3.6188118057636213e-06,
      "loss": 0.8184,
      "step": 8387
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7363228797912598,
      "learning_rate": 3.617695714355115e-06,
      "loss": 0.1172,
      "step": 8388
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.7672224044799805,
      "learning_rate": 3.6165796975166932e-06,
      "loss": 0.536,
      "step": 8389
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.698730945587158,
      "learning_rate": 3.6154637553085596e-06,
      "loss": 0.1379,
      "step": 8390
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.425056457519531,
      "learning_rate": 3.6143478877909153e-06,
      "loss": 0.4401,
      "step": 8391
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.8148488998413086,
      "learning_rate": 3.6132320950239584e-06,
      "loss": 0.1699,
      "step": 8392
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.5844388008117676,
      "learning_rate": 3.61211637706788e-06,
      "loss": 0.6308,
      "step": 8393
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.281283140182495,
      "learning_rate": 3.6110007339828723e-06,
      "loss": 0.105,
      "step": 8394
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.994379758834839,
      "learning_rate": 3.6098851658291155e-06,
      "loss": 0.1093,
      "step": 8395
    },
    {
      "epoch": 0.6,
      "grad_norm": 11.258732795715332,
      "learning_rate": 3.6087696726667913e-06,
      "loss": 0.5354,
      "step": 8396
    },
    {
      "epoch": 0.6,
      "grad_norm": 11.328458786010742,
      "learning_rate": 3.607654254556077e-06,
      "loss": 0.9944,
      "step": 8397
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.6880123615264893,
      "learning_rate": 3.606538911557147e-06,
      "loss": 0.1114,
      "step": 8398
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.40005874633789,
      "learning_rate": 3.6054236437301667e-06,
      "loss": 1.073,
      "step": 8399
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0214195251464844,
      "learning_rate": 3.6043084511353023e-06,
      "loss": 0.064,
      "step": 8400
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.6658809185028076,
      "learning_rate": 3.6031933338327146e-06,
      "loss": 0.181,
      "step": 8401
    },
    {
      "epoch": 0.6,
      "grad_norm": 11.959614753723145,
      "learning_rate": 3.6020782918825583e-06,
      "loss": 0.8367,
      "step": 8402
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.393272399902344,
      "learning_rate": 3.6009633253449882e-06,
      "loss": 0.5064,
      "step": 8403
    },
    {
      "epoch": 0.6,
      "grad_norm": 22.595613479614258,
      "learning_rate": 3.5998484342801498e-06,
      "loss": 0.6442,
      "step": 8404
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.182591676712036,
      "learning_rate": 3.598733618748188e-06,
      "loss": 0.1544,
      "step": 8405
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.526700496673584,
      "learning_rate": 3.5976188788092465e-06,
      "loss": 0.2528,
      "step": 8406
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.737537384033203,
      "learning_rate": 3.5965042145234562e-06,
      "loss": 0.92,
      "step": 8407
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.1531918048858643,
      "learning_rate": 3.5953896259509506e-06,
      "loss": 0.1018,
      "step": 8408
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.9492862224578857,
      "learning_rate": 3.594275113151859e-06,
      "loss": 0.9199,
      "step": 8409
    },
    {
      "epoch": 0.6,
      "grad_norm": 15.65947437286377,
      "learning_rate": 3.5931606761863026e-06,
      "loss": 0.819,
      "step": 8410
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.229892730712891,
      "learning_rate": 3.5920463151144024e-06,
      "loss": 0.8302,
      "step": 8411
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.185507297515869,
      "learning_rate": 3.5909320299962757e-06,
      "loss": 0.0979,
      "step": 8412
    },
    {
      "epoch": 0.6,
      "grad_norm": 16.308191299438477,
      "learning_rate": 3.5898178208920313e-06,
      "loss": 0.707,
      "step": 8413
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.927860975265503,
      "learning_rate": 3.5887036878617776e-06,
      "loss": 0.0783,
      "step": 8414
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.957908868789673,
      "learning_rate": 3.587589630965619e-06,
      "loss": 0.1308,
      "step": 8415
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.972162246704102,
      "learning_rate": 3.586475650263652e-06,
      "loss": 0.7287,
      "step": 8416
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7439627647399902,
      "learning_rate": 3.5853617458159744e-06,
      "loss": 0.175,
      "step": 8417
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.49817180633545,
      "learning_rate": 3.584247917682677e-06,
      "loss": 0.8598,
      "step": 8418
    },
    {
      "epoch": 0.6,
      "grad_norm": 22.19693946838379,
      "learning_rate": 3.583134165923844e-06,
      "loss": 1.4043,
      "step": 8419
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3427774906158447,
      "learning_rate": 3.5820204905995593e-06,
      "loss": 0.11,
      "step": 8420
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.090978622436523,
      "learning_rate": 3.5809068917699026e-06,
      "loss": 0.7074,
      "step": 8421
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.610686302185059,
      "learning_rate": 3.579793369494946e-06,
      "loss": 0.6297,
      "step": 8422
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.9651026725769043,
      "learning_rate": 3.5786799238347635e-06,
      "loss": 0.1605,
      "step": 8423
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2302134037017822,
      "learning_rate": 3.5775665548494176e-06,
      "loss": 0.0861,
      "step": 8424
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4959466457366943,
      "learning_rate": 3.5764532625989714e-06,
      "loss": 0.1579,
      "step": 8425
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.1244027614593506,
      "learning_rate": 3.5753400471434853e-06,
      "loss": 0.2257,
      "step": 8426
    },
    {
      "epoch": 0.6,
      "grad_norm": 20.89913558959961,
      "learning_rate": 3.5742269085430093e-06,
      "loss": 0.9653,
      "step": 8427
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.529313087463379,
      "learning_rate": 3.5731138468575944e-06,
      "loss": 1.053,
      "step": 8428
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.480727195739746,
      "learning_rate": 3.572000862147288e-06,
      "loss": 0.0492,
      "step": 8429
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6365997791290283,
      "learning_rate": 3.5708879544721308e-06,
      "loss": 0.0642,
      "step": 8430
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.411763668060303,
      "learning_rate": 3.5697751238921567e-06,
      "loss": 0.7184,
      "step": 8431
    },
    {
      "epoch": 0.6,
      "grad_norm": 12.37303638458252,
      "learning_rate": 3.5686623704674024e-06,
      "loss": 0.8044,
      "step": 8432
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7993741035461426,
      "learning_rate": 3.5675496942578934e-06,
      "loss": 0.7883,
      "step": 8433
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.095607757568359,
      "learning_rate": 3.566437095323657e-06,
      "loss": 0.6597,
      "step": 8434
    },
    {
      "epoch": 0.6,
      "grad_norm": 11.067560195922852,
      "learning_rate": 3.5653245737247143e-06,
      "loss": 0.501,
      "step": 8435
    },
    {
      "epoch": 0.6,
      "grad_norm": 18.714588165283203,
      "learning_rate": 3.5642121295210785e-06,
      "loss": 1.1904,
      "step": 8436
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7570998668670654,
      "learning_rate": 3.5630997627727647e-06,
      "loss": 0.7154,
      "step": 8437
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.639607429504395,
      "learning_rate": 3.5619874735397795e-06,
      "loss": 0.7802,
      "step": 8438
    },
    {
      "epoch": 0.61,
      "grad_norm": 20.585607528686523,
      "learning_rate": 3.5608752618821264e-06,
      "loss": 0.7653,
      "step": 8439
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.630435943603516,
      "learning_rate": 3.5597631278598066e-06,
      "loss": 1.0175,
      "step": 8440
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.447219133377075,
      "learning_rate": 3.5586510715328134e-06,
      "loss": 0.1473,
      "step": 8441
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.338665008544922,
      "learning_rate": 3.5575390929611407e-06,
      "loss": 0.7798,
      "step": 8442
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.73137903213501,
      "learning_rate": 3.5564271922047727e-06,
      "loss": 0.896,
      "step": 8443
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.007318019866943,
      "learning_rate": 3.5553153693236943e-06,
      "loss": 0.8253,
      "step": 8444
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0367047786712646,
      "learning_rate": 3.554203624377882e-06,
      "loss": 0.2193,
      "step": 8445
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.6973183155059814,
      "learning_rate": 3.5530919574273127e-06,
      "loss": 0.172,
      "step": 8446
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3953490257263184,
      "learning_rate": 3.5519803685319545e-06,
      "loss": 0.1009,
      "step": 8447
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.5336074829101562,
      "learning_rate": 3.5508688577517737e-06,
      "loss": 0.1997,
      "step": 8448
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.134289503097534,
      "learning_rate": 3.549757425146735e-06,
      "loss": 0.1016,
      "step": 8449
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.5449419021606445,
      "learning_rate": 3.5486460707767913e-06,
      "loss": 0.2452,
      "step": 8450
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.5370707511901855,
      "learning_rate": 3.5475347947019e-06,
      "loss": 0.0966,
      "step": 8451
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6057602167129517,
      "learning_rate": 3.546423596982008e-06,
      "loss": 0.0521,
      "step": 8452
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.278783798217773,
      "learning_rate": 3.545312477677061e-06,
      "loss": 0.7484,
      "step": 8453
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.177746295928955,
      "learning_rate": 3.544201436847001e-06,
      "loss": 0.7219,
      "step": 8454
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.8643593788146973,
      "learning_rate": 3.5430904745517623e-06,
      "loss": 0.1646,
      "step": 8455
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2013065814971924,
      "learning_rate": 3.5419795908512766e-06,
      "loss": 0.1191,
      "step": 8456
    },
    {
      "epoch": 0.61,
      "grad_norm": 28.216745376586914,
      "learning_rate": 3.540868785805473e-06,
      "loss": 1.0527,
      "step": 8457
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.744898319244385,
      "learning_rate": 3.5397580594742766e-06,
      "loss": 0.8588,
      "step": 8458
    },
    {
      "epoch": 0.61,
      "grad_norm": 23.361669540405273,
      "learning_rate": 3.538647411917604e-06,
      "loss": 0.9148,
      "step": 8459
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0983083248138428,
      "learning_rate": 3.537536843195374e-06,
      "loss": 0.1796,
      "step": 8460
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9894635677337646,
      "learning_rate": 3.5364263533674936e-06,
      "loss": 0.1782,
      "step": 8461
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.589665412902832,
      "learning_rate": 3.535315942493872e-06,
      "loss": 0.5759,
      "step": 8462
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9393982887268066,
      "learning_rate": 3.5342056106344126e-06,
      "loss": 0.1885,
      "step": 8463
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0920910835266113,
      "learning_rate": 3.53309535784901e-06,
      "loss": 0.2441,
      "step": 8464
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0153088569641113,
      "learning_rate": 3.5319851841975622e-06,
      "loss": 0.2029,
      "step": 8465
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0138843059539795,
      "learning_rate": 3.5308750897399566e-06,
      "loss": 0.1031,
      "step": 8466
    },
    {
      "epoch": 0.61,
      "grad_norm": 15.735730171203613,
      "learning_rate": 3.529765074536079e-06,
      "loss": 0.9336,
      "step": 8467
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.079490900039673,
      "learning_rate": 3.52865513864581e-06,
      "loss": 0.1339,
      "step": 8468
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.197683334350586,
      "learning_rate": 3.5275452821290275e-06,
      "loss": 0.1034,
      "step": 8469
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.359118461608887,
      "learning_rate": 3.5264355050456023e-06,
      "loss": 0.5629,
      "step": 8470
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.892500400543213,
      "learning_rate": 3.5253258074554044e-06,
      "loss": 0.4509,
      "step": 8471
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.353614807128906,
      "learning_rate": 3.524216189418298e-06,
      "loss": 0.928,
      "step": 8472
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.068730115890503,
      "learning_rate": 3.52310665099414e-06,
      "loss": 0.0746,
      "step": 8473
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.67393159866333,
      "learning_rate": 3.5219971922427904e-06,
      "loss": 0.1147,
      "step": 8474
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.6607582569122314,
      "learning_rate": 3.520887813224096e-06,
      "loss": 0.0868,
      "step": 8475
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9476587772369385,
      "learning_rate": 3.5197785139979057e-06,
      "loss": 0.0675,
      "step": 8476
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3942651748657227,
      "learning_rate": 3.5186692946240618e-06,
      "loss": 0.1353,
      "step": 8477
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.751506328582764,
      "learning_rate": 3.5175601551624037e-06,
      "loss": 0.5978,
      "step": 8478
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.034026145935059,
      "learning_rate": 3.5164510956727626e-06,
      "loss": 0.2005,
      "step": 8479
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0615968704223633,
      "learning_rate": 3.5153421162149693e-06,
      "loss": 0.08,
      "step": 8480
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.367604732513428,
      "learning_rate": 3.514233216848849e-06,
      "loss": 0.4935,
      "step": 8481
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4681954383850098,
      "learning_rate": 3.513124397634222e-06,
      "loss": 0.1061,
      "step": 8482
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.703758478164673,
      "learning_rate": 3.5120156586309072e-06,
      "loss": 0.1121,
      "step": 8483
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.211390733718872,
      "learning_rate": 3.5109069998987145e-06,
      "loss": 0.1173,
      "step": 8484
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.123045921325684,
      "learning_rate": 3.5097984214974522e-06,
      "loss": 0.962,
      "step": 8485
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.690523147583008,
      "learning_rate": 3.5086899234869253e-06,
      "loss": 1.1759,
      "step": 8486
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.031759738922119,
      "learning_rate": 3.507581505926931e-06,
      "loss": 0.1536,
      "step": 8487
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.348242998123169,
      "learning_rate": 3.506473168877267e-06,
      "loss": 0.1901,
      "step": 8488
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9647363424301147,
      "learning_rate": 3.5053649123977202e-06,
      "loss": 0.0545,
      "step": 8489
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.233911037445068,
      "learning_rate": 3.504256736548082e-06,
      "loss": 0.8145,
      "step": 8490
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.973299264907837,
      "learning_rate": 3.5031486413881276e-06,
      "loss": 0.1651,
      "step": 8491
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.033206939697266,
      "learning_rate": 3.5020406269776406e-06,
      "loss": 0.7376,
      "step": 8492
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.099745273590088,
      "learning_rate": 3.50093269337639e-06,
      "loss": 0.2065,
      "step": 8493
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.829835057258606,
      "learning_rate": 3.499824840644146e-06,
      "loss": 0.043,
      "step": 8494
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.5237648487091064,
      "learning_rate": 3.4987170688406753e-06,
      "loss": 0.0733,
      "step": 8495
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3994133472442627,
      "learning_rate": 3.4976093780257347e-06,
      "loss": 0.1166,
      "step": 8496
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.116166114807129,
      "learning_rate": 3.496501768259082e-06,
      "loss": 1.5249,
      "step": 8497
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.171480178833008,
      "learning_rate": 3.495394239600467e-06,
      "loss": 0.1313,
      "step": 8498
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.630160093307495,
      "learning_rate": 3.494286792109638e-06,
      "loss": 0.1499,
      "step": 8499
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.305543899536133,
      "learning_rate": 3.4931794258463376e-06,
      "loss": 0.0492,
      "step": 8500
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.105932235717773,
      "learning_rate": 3.4920721408703028e-06,
      "loss": 0.8786,
      "step": 8501
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3569371700286865,
      "learning_rate": 3.490964937241269e-06,
      "loss": 0.1161,
      "step": 8502
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.478713274002075,
      "learning_rate": 3.4898578150189653e-06,
      "loss": 0.0814,
      "step": 8503
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.1119914054870605,
      "learning_rate": 3.4887507742631154e-06,
      "loss": 0.5422,
      "step": 8504
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.726477861404419,
      "learning_rate": 3.48764381503344e-06,
      "loss": 0.7394,
      "step": 8505
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.5827488899230957,
      "learning_rate": 3.4865369373896572e-06,
      "loss": 0.1425,
      "step": 8506
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.933139443397522,
      "learning_rate": 3.4854301413914775e-06,
      "loss": 0.0766,
      "step": 8507
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.617111206054688,
      "learning_rate": 3.484323427098608e-06,
      "loss": 0.7129,
      "step": 8508
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.673331260681152,
      "learning_rate": 3.483216794570754e-06,
      "loss": 0.8021,
      "step": 8509
    },
    {
      "epoch": 0.61,
      "grad_norm": 15.971867561340332,
      "learning_rate": 3.4821102438676117e-06,
      "loss": 0.8272,
      "step": 8510
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.2302961349487305,
      "learning_rate": 3.481003775048877e-06,
      "loss": 0.8043,
      "step": 8511
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.10818099975586,
      "learning_rate": 3.4798973881742383e-06,
      "loss": 0.8562,
      "step": 8512
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.52815055847168,
      "learning_rate": 3.478791083303381e-06,
      "loss": 0.6094,
      "step": 8513
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.796889781951904,
      "learning_rate": 3.4776848604959877e-06,
      "loss": 0.5609,
      "step": 8514
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.110553741455078,
      "learning_rate": 3.4765787198117353e-06,
      "loss": 0.5425,
      "step": 8515
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.761198043823242,
      "learning_rate": 3.4754726613102917e-06,
      "loss": 0.8488,
      "step": 8516
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.6124184131622314,
      "learning_rate": 3.474366685051328e-06,
      "loss": 0.1719,
      "step": 8517
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.038959980010986,
      "learning_rate": 3.473260791094508e-06,
      "loss": 0.9041,
      "step": 8518
    },
    {
      "epoch": 0.61,
      "grad_norm": 21.945146560668945,
      "learning_rate": 3.472154979499488e-06,
      "loss": 0.5466,
      "step": 8519
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.010817050933838,
      "learning_rate": 3.4710492503259248e-06,
      "loss": 0.0801,
      "step": 8520
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.783041477203369,
      "learning_rate": 3.469943603633466e-06,
      "loss": 0.1688,
      "step": 8521
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.095219850540161,
      "learning_rate": 3.4688380394817578e-06,
      "loss": 0.2731,
      "step": 8522
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.875889539718628,
      "learning_rate": 3.4677325579304434e-06,
      "loss": 0.1286,
      "step": 8523
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.049461364746094,
      "learning_rate": 3.4666271590391566e-06,
      "loss": 0.7646,
      "step": 8524
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.467650413513184,
      "learning_rate": 3.4655218428675307e-06,
      "loss": 0.996,
      "step": 8525
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.692035675048828,
      "learning_rate": 3.4644166094751925e-06,
      "loss": 0.1735,
      "step": 8526
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.007994651794434,
      "learning_rate": 3.4633114589217677e-06,
      "loss": 0.6162,
      "step": 8527
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1259615421295166,
      "learning_rate": 3.462206391266871e-06,
      "loss": 0.1058,
      "step": 8528
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6259284019470215,
      "learning_rate": 3.4611014065701192e-06,
      "loss": 0.0579,
      "step": 8529
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.670091390609741,
      "learning_rate": 3.459996504891121e-06,
      "loss": 0.1344,
      "step": 8530
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.575514793395996,
      "learning_rate": 3.4588916862894817e-06,
      "loss": 0.1037,
      "step": 8531
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.872550964355469,
      "learning_rate": 3.457786950824804e-06,
      "loss": 0.9067,
      "step": 8532
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4036219120025635,
      "learning_rate": 3.4566822985566817e-06,
      "loss": 0.138,
      "step": 8533
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9575624465942383,
      "learning_rate": 3.455577729544708e-06,
      "loss": 0.1788,
      "step": 8534
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.9375219345092773,
      "learning_rate": 3.4544732438484697e-06,
      "loss": 0.6937,
      "step": 8535
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6875343322753906,
      "learning_rate": 3.453368841527549e-06,
      "loss": 0.0475,
      "step": 8536
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.668314933776855,
      "learning_rate": 3.4522645226415265e-06,
      "loss": 0.9833,
      "step": 8537
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.003255367279053,
      "learning_rate": 3.4511602872499727e-06,
      "loss": 0.8393,
      "step": 8538
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9526753425598145,
      "learning_rate": 3.450056135412462e-06,
      "loss": 0.1523,
      "step": 8539
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.203455924987793,
      "learning_rate": 3.448952067188554e-06,
      "loss": 0.6946,
      "step": 8540
    },
    {
      "epoch": 0.61,
      "grad_norm": 17.04839324951172,
      "learning_rate": 3.4478480826378103e-06,
      "loss": 1.4025,
      "step": 8541
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.943750858306885,
      "learning_rate": 3.4467441818197873e-06,
      "loss": 0.7716,
      "step": 8542
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.82821786403656,
      "learning_rate": 3.4456403647940375e-06,
      "loss": 0.0763,
      "step": 8543
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.7599456310272217,
      "learning_rate": 3.4445366316201046e-06,
      "loss": 0.145,
      "step": 8544
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.684875965118408,
      "learning_rate": 3.443432982357533e-06,
      "loss": 0.0803,
      "step": 8545
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.085123062133789,
      "learning_rate": 3.442329417065862e-06,
      "loss": 0.4156,
      "step": 8546
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.7485151290893555,
      "learning_rate": 3.4412259358046207e-06,
      "loss": 0.128,
      "step": 8547
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9095277786254883,
      "learning_rate": 3.4401225386333415e-06,
      "loss": 0.212,
      "step": 8548
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.769622564315796,
      "learning_rate": 3.4390192256115457e-06,
      "loss": 0.815,
      "step": 8549
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.541135311126709,
      "learning_rate": 3.437915996798754e-06,
      "loss": 0.5142,
      "step": 8550
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.030914545059204,
      "learning_rate": 3.4368128522544837e-06,
      "loss": 0.2262,
      "step": 8551
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.260980129241943,
      "learning_rate": 3.435709792038242e-06,
      "loss": 0.7587,
      "step": 8552
    },
    {
      "epoch": 0.61,
      "grad_norm": 17.06851577758789,
      "learning_rate": 3.4346068162095346e-06,
      "loss": 0.8086,
      "step": 8553
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.487114667892456,
      "learning_rate": 3.433503924827865e-06,
      "loss": 0.0776,
      "step": 8554
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.217911720275879,
      "learning_rate": 3.4324011179527283e-06,
      "loss": 1.1525,
      "step": 8555
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.1458516120910645,
      "learning_rate": 3.431298395643618e-06,
      "loss": 0.571,
      "step": 8556
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.4970526695251465,
      "learning_rate": 3.430195757960022e-06,
      "loss": 0.7788,
      "step": 8557
    },
    {
      "epoch": 0.61,
      "grad_norm": 13.950824737548828,
      "learning_rate": 3.4290932049614225e-06,
      "loss": 1.1509,
      "step": 8558
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3618412017822266,
      "learning_rate": 3.4279907367072974e-06,
      "loss": 0.0628,
      "step": 8559
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.733780860900879,
      "learning_rate": 3.426888353257123e-06,
      "loss": 0.639,
      "step": 8560
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.162938117980957,
      "learning_rate": 3.4257860546703663e-06,
      "loss": 0.5492,
      "step": 8561
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.9099831581115723,
      "learning_rate": 3.4246838410064948e-06,
      "loss": 0.2719,
      "step": 8562
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.315002918243408,
      "learning_rate": 3.4235817123249674e-06,
      "loss": 0.2653,
      "step": 8563
    },
    {
      "epoch": 0.61,
      "grad_norm": 16.30780029296875,
      "learning_rate": 3.422479668685238e-06,
      "loss": 0.7621,
      "step": 8564
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.916410446166992,
      "learning_rate": 3.4213777101467593e-06,
      "loss": 0.5751,
      "step": 8565
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.538408279418945,
      "learning_rate": 3.4202758367689786e-06,
      "loss": 0.4521,
      "step": 8566
    },
    {
      "epoch": 0.61,
      "grad_norm": 16.128881454467773,
      "learning_rate": 3.419174048611336e-06,
      "loss": 1.1771,
      "step": 8567
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.727583885192871,
      "learning_rate": 3.418072345733271e-06,
      "loss": 0.7551,
      "step": 8568
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.990448474884033,
      "learning_rate": 3.4169707281942134e-06,
      "loss": 0.6212,
      "step": 8569
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2945878505706787,
      "learning_rate": 3.4158691960535933e-06,
      "loss": 0.1296,
      "step": 8570
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.349287986755371,
      "learning_rate": 3.414767749370835e-06,
      "loss": 0.7632,
      "step": 8571
    },
    {
      "epoch": 0.61,
      "grad_norm": 25.92430877685547,
      "learning_rate": 3.4136663882053543e-06,
      "loss": 0.7134,
      "step": 8572
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.6432466506958,
      "learning_rate": 3.4125651126165682e-06,
      "loss": 0.4735,
      "step": 8573
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.748596429824829,
      "learning_rate": 3.4114639226638863e-06,
      "loss": 0.1772,
      "step": 8574
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.7630794048309326,
      "learning_rate": 3.410362818406714e-06,
      "loss": 0.1264,
      "step": 8575
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.151216983795166,
      "learning_rate": 3.409261799904449e-06,
      "loss": 0.0514,
      "step": 8576
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4353671073913574,
      "learning_rate": 3.4081608672164883e-06,
      "loss": 0.2103,
      "step": 8577
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.856950759887695,
      "learning_rate": 3.407060020402223e-06,
      "loss": 0.6898,
      "step": 8578
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.8678603172302246,
      "learning_rate": 3.40595925952104e-06,
      "loss": 0.7354,
      "step": 8579
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4712488651275635,
      "learning_rate": 3.4048585846323223e-06,
      "loss": 0.1054,
      "step": 8580
    },
    {
      "epoch": 0.62,
      "grad_norm": 13.145135879516602,
      "learning_rate": 3.403757995795445e-06,
      "loss": 1.2752,
      "step": 8581
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.6054115295410156,
      "learning_rate": 3.4026574930697827e-06,
      "loss": 0.7588,
      "step": 8582
    },
    {
      "epoch": 0.62,
      "grad_norm": 14.65932846069336,
      "learning_rate": 3.401557076514701e-06,
      "loss": 1.2195,
      "step": 8583
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9281036853790283,
      "learning_rate": 3.4004567461895654e-06,
      "loss": 0.5382,
      "step": 8584
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.6362617015838623,
      "learning_rate": 3.3993565021537343e-06,
      "loss": 0.2441,
      "step": 8585
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.29530143737793,
      "learning_rate": 3.39825634446656e-06,
      "loss": 0.514,
      "step": 8586
    },
    {
      "epoch": 0.62,
      "grad_norm": 16.156801223754883,
      "learning_rate": 3.3971562731873954e-06,
      "loss": 0.8068,
      "step": 8587
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.164365291595459,
      "learning_rate": 3.3960562883755797e-06,
      "loss": 0.2442,
      "step": 8588
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.680952072143555,
      "learning_rate": 3.394956390090458e-06,
      "loss": 1.0722,
      "step": 8589
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.556821823120117,
      "learning_rate": 3.3938565783913623e-06,
      "loss": 0.363,
      "step": 8590
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2577621936798096,
      "learning_rate": 3.3927568533376265e-06,
      "loss": 0.0963,
      "step": 8591
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.991325855255127,
      "learning_rate": 3.391657214988573e-06,
      "loss": 0.5599,
      "step": 8592
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.355386734008789,
      "learning_rate": 3.3905576634035244e-06,
      "loss": 0.9779,
      "step": 8593
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8542323112487793,
      "learning_rate": 3.389458198641799e-06,
      "loss": 0.0607,
      "step": 8594
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8522820472717285,
      "learning_rate": 3.388358820762707e-06,
      "loss": 0.1582,
      "step": 8595
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.7664546966552734,
      "learning_rate": 3.3872595298255566e-06,
      "loss": 0.1247,
      "step": 8596
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.475996017456055,
      "learning_rate": 3.386160325889649e-06,
      "loss": 0.9546,
      "step": 8597
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6857728958129883,
      "learning_rate": 3.385061209014284e-06,
      "loss": 0.0899,
      "step": 8598
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5553672313690186,
      "learning_rate": 3.3839621792587553e-06,
      "loss": 0.1585,
      "step": 8599
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4321212768554688,
      "learning_rate": 3.382863236682349e-06,
      "loss": 0.1294,
      "step": 8600
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.29183030128479,
      "learning_rate": 3.381764381344349e-06,
      "loss": 0.0908,
      "step": 8601
    },
    {
      "epoch": 0.62,
      "grad_norm": 14.881169319152832,
      "learning_rate": 3.380665613304035e-06,
      "loss": 0.6003,
      "step": 8602
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.608576536178589,
      "learning_rate": 3.379566932620683e-06,
      "loss": 0.1381,
      "step": 8603
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5755603313446045,
      "learning_rate": 3.37846833935356e-06,
      "loss": 0.0612,
      "step": 8604
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8657784461975098,
      "learning_rate": 3.3773698335619337e-06,
      "loss": 0.0799,
      "step": 8605
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.337759017944336,
      "learning_rate": 3.376271415305062e-06,
      "loss": 0.8063,
      "step": 8606
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.2838964462280273,
      "learning_rate": 3.3751730846422006e-06,
      "loss": 0.7159,
      "step": 8607
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.9272236824035645,
      "learning_rate": 3.3740748416326024e-06,
      "loss": 0.6749,
      "step": 8608
    },
    {
      "epoch": 0.62,
      "grad_norm": 15.833277702331543,
      "learning_rate": 3.3729766863355107e-06,
      "loss": 1.0912,
      "step": 8609
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.9232795238494873,
      "learning_rate": 3.3718786188101692e-06,
      "loss": 0.1921,
      "step": 8610
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.660646677017212,
      "learning_rate": 3.3707806391158137e-06,
      "loss": 0.1162,
      "step": 8611
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.133432626724243,
      "learning_rate": 3.3696827473116762e-06,
      "loss": 0.1524,
      "step": 8612
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.685375928878784,
      "learning_rate": 3.368584943456982e-06,
      "loss": 0.1919,
      "step": 8613
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5493180751800537,
      "learning_rate": 3.3674872276109554e-06,
      "loss": 0.1114,
      "step": 8614
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.04287052154541,
      "learning_rate": 3.3663895998328135e-06,
      "loss": 0.7022,
      "step": 8615
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.7229256629943848,
      "learning_rate": 3.365292060181768e-06,
      "loss": 0.7797,
      "step": 8616
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.2893218994140625,
      "learning_rate": 3.3641946087170297e-06,
      "loss": 0.8271,
      "step": 8617
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.7011027336120605,
      "learning_rate": 3.3630972454977997e-06,
      "loss": 0.1709,
      "step": 8618
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.267722129821777,
      "learning_rate": 3.3619999705832773e-06,
      "loss": 0.7825,
      "step": 8619
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.276806831359863,
      "learning_rate": 3.360902784032656e-06,
      "loss": 0.9694,
      "step": 8620
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.162046432495117,
      "learning_rate": 3.359805685905125e-06,
      "loss": 1.0502,
      "step": 8621
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.715178489685059,
      "learning_rate": 3.3587086762598704e-06,
      "loss": 0.5211,
      "step": 8622
    },
    {
      "epoch": 0.62,
      "grad_norm": 23.03791618347168,
      "learning_rate": 3.3576117551560707e-06,
      "loss": 1.3212,
      "step": 8623
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5247702598571777,
      "learning_rate": 3.3565149226528978e-06,
      "loss": 0.1462,
      "step": 8624
    },
    {
      "epoch": 0.62,
      "grad_norm": 23.368078231811523,
      "learning_rate": 3.355418178809524e-06,
      "loss": 0.6239,
      "step": 8625
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.877664566040039,
      "learning_rate": 3.3543215236851155e-06,
      "loss": 0.6429,
      "step": 8626
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.9596898555755615,
      "learning_rate": 3.3532249573388303e-06,
      "loss": 0.2017,
      "step": 8627
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9695321321487427,
      "learning_rate": 3.3521284798298265e-06,
      "loss": 0.0585,
      "step": 8628
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5661633014678955,
      "learning_rate": 3.3510320912172523e-06,
      "loss": 0.0934,
      "step": 8629
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.587829113006592,
      "learning_rate": 3.3499357915602558e-06,
      "loss": 0.8187,
      "step": 8630
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.3741495609283447,
      "learning_rate": 3.348839580917978e-06,
      "loss": 0.1343,
      "step": 8631
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.279031753540039,
      "learning_rate": 3.347743459349554e-06,
      "loss": 0.4185,
      "step": 8632
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.7314674854278564,
      "learning_rate": 3.346647426914117e-06,
      "loss": 0.1394,
      "step": 8633
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8489346504211426,
      "learning_rate": 3.3455514836707926e-06,
      "loss": 0.0489,
      "step": 8634
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.306698322296143,
      "learning_rate": 3.3444556296787044e-06,
      "loss": 0.8713,
      "step": 8635
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.242392539978027,
      "learning_rate": 3.343359864996967e-06,
      "loss": 0.8844,
      "step": 8636
    },
    {
      "epoch": 0.62,
      "grad_norm": 16.13128089904785,
      "learning_rate": 3.342264189684695e-06,
      "loss": 0.6624,
      "step": 8637
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.74293851852417,
      "learning_rate": 3.3411686038009938e-06,
      "loss": 0.8636,
      "step": 8638
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.7729127407073975,
      "learning_rate": 3.340073107404968e-06,
      "loss": 0.9051,
      "step": 8639
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.303340911865234,
      "learning_rate": 3.3389777005557156e-06,
      "loss": 0.6241,
      "step": 8640
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9090142250061035,
      "learning_rate": 3.337882383312328e-06,
      "loss": 0.7678,
      "step": 8641
    },
    {
      "epoch": 0.62,
      "grad_norm": 20.996780395507812,
      "learning_rate": 3.3367871557338952e-06,
      "loss": 0.9707,
      "step": 8642
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.070570468902588,
      "learning_rate": 3.3356920178794996e-06,
      "loss": 0.0882,
      "step": 8643
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.126801013946533,
      "learning_rate": 3.3345969698082186e-06,
      "loss": 0.9221,
      "step": 8644
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1279687881469727,
      "learning_rate": 3.333502011579129e-06,
      "loss": 0.0782,
      "step": 8645
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8425133228302,
      "learning_rate": 3.3324071432512963e-06,
      "loss": 0.1655,
      "step": 8646
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.57922625541687,
      "learning_rate": 3.331312364883789e-06,
      "loss": 0.8272,
      "step": 8647
    },
    {
      "epoch": 0.62,
      "grad_norm": 13.394514083862305,
      "learning_rate": 3.3302176765356607e-06,
      "loss": 1.0177,
      "step": 8648
    },
    {
      "epoch": 0.62,
      "grad_norm": 13.058717727661133,
      "learning_rate": 3.3291230782659684e-06,
      "loss": 0.6609,
      "step": 8649
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.112779140472412,
      "learning_rate": 3.328028570133761e-06,
      "loss": 0.1402,
      "step": 8650
    },
    {
      "epoch": 0.62,
      "grad_norm": 44.763641357421875,
      "learning_rate": 3.3269341521980847e-06,
      "loss": 0.9762,
      "step": 8651
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.41733169555664,
      "learning_rate": 3.325839824517977e-06,
      "loss": 1.0422,
      "step": 8652
    },
    {
      "epoch": 0.62,
      "grad_norm": 13.29305362701416,
      "learning_rate": 3.3247455871524735e-06,
      "loss": 0.6956,
      "step": 8653
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.588852882385254,
      "learning_rate": 3.3236514401606056e-06,
      "loss": 0.3858,
      "step": 8654
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.169259548187256,
      "learning_rate": 3.322557383601397e-06,
      "loss": 0.084,
      "step": 8655
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.7819859981536865,
      "learning_rate": 3.321463417533868e-06,
      "loss": 0.7548,
      "step": 8656
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.3032658100128174,
      "learning_rate": 3.3203695420170333e-06,
      "loss": 0.1232,
      "step": 8657
    },
    {
      "epoch": 0.62,
      "grad_norm": 15.43002986907959,
      "learning_rate": 3.319275757109904e-06,
      "loss": 0.9345,
      "step": 8658
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.601995468139648,
      "learning_rate": 3.318182062871488e-06,
      "loss": 0.5606,
      "step": 8659
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.0331599712371826,
      "learning_rate": 3.317088459360782e-06,
      "loss": 0.2012,
      "step": 8660
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7804638147354126,
      "learning_rate": 3.315994946636783e-06,
      "loss": 0.0892,
      "step": 8661
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.612774848937988,
      "learning_rate": 3.3149015247584835e-06,
      "loss": 0.8719,
      "step": 8662
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.7531726360321045,
      "learning_rate": 3.3138081937848666e-06,
      "loss": 0.1615,
      "step": 8663
    },
    {
      "epoch": 0.62,
      "grad_norm": 16.431058883666992,
      "learning_rate": 3.312714953774916e-06,
      "loss": 0.6033,
      "step": 8664
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.058462142944336,
      "learning_rate": 3.311621804787607e-06,
      "loss": 0.6399,
      "step": 8665
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.609660148620605,
      "learning_rate": 3.31052874688191e-06,
      "loss": 0.7838,
      "step": 8666
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.0619781017303467,
      "learning_rate": 3.3094357801167925e-06,
      "loss": 0.139,
      "step": 8667
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6670968532562256,
      "learning_rate": 3.308342904551216e-06,
      "loss": 0.1254,
      "step": 8668
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.570792198181152,
      "learning_rate": 3.3072501202441365e-06,
      "loss": 0.8656,
      "step": 8669
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.842773199081421,
      "learning_rate": 3.306157427254506e-06,
      "loss": 0.1155,
      "step": 8670
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.7882349491119385,
      "learning_rate": 3.3050648256412714e-06,
      "loss": 0.7124,
      "step": 8671
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.927609443664551,
      "learning_rate": 3.303972315463372e-06,
      "loss": 0.9303,
      "step": 8672
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.185185670852661,
      "learning_rate": 3.3028798967797464e-06,
      "loss": 0.1345,
      "step": 8673
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.3681466579437256,
      "learning_rate": 3.3017875696493274e-06,
      "loss": 0.0601,
      "step": 8674
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.1538963317871094,
      "learning_rate": 3.30069533413104e-06,
      "loss": 0.1567,
      "step": 8675
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.0873634815216064,
      "learning_rate": 3.299603190283808e-06,
      "loss": 0.1332,
      "step": 8676
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8552563190460205,
      "learning_rate": 3.298511138166547e-06,
      "loss": 0.1333,
      "step": 8677
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.2748823165893555,
      "learning_rate": 3.297419177838169e-06,
      "loss": 0.2322,
      "step": 8678
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.310749530792236,
      "learning_rate": 3.296327309357583e-06,
      "loss": 0.6914,
      "step": 8679
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.988632202148438,
      "learning_rate": 3.2952355327836892e-06,
      "loss": 0.3623,
      "step": 8680
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6854867935180664,
      "learning_rate": 3.294143848175385e-06,
      "loss": 0.1725,
      "step": 8681
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.934131622314453,
      "learning_rate": 3.2930522555915644e-06,
      "loss": 0.5585,
      "step": 8682
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6942145824432373,
      "learning_rate": 3.291960755091115e-06,
      "loss": 0.1592,
      "step": 8683
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.3513102531433105,
      "learning_rate": 3.2908693467329144e-06,
      "loss": 0.8914,
      "step": 8684
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.03782844543457,
      "learning_rate": 3.289778030575845e-06,
      "loss": 0.6277,
      "step": 8685
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.433647155761719,
      "learning_rate": 3.288686806678776e-06,
      "loss": 0.7984,
      "step": 8686
    },
    {
      "epoch": 0.62,
      "grad_norm": 15.90373706817627,
      "learning_rate": 3.2875956751005764e-06,
      "loss": 1.0169,
      "step": 8687
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4189260005950928,
      "learning_rate": 3.286504635900109e-06,
      "loss": 0.1338,
      "step": 8688
    },
    {
      "epoch": 0.62,
      "grad_norm": 18.429101943969727,
      "learning_rate": 3.28541368913623e-06,
      "loss": 0.9604,
      "step": 8689
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.0818099975585938,
      "learning_rate": 3.284322834867793e-06,
      "loss": 0.16,
      "step": 8690
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.918241500854492,
      "learning_rate": 3.2832320731536437e-06,
      "loss": 0.8859,
      "step": 8691
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.804490804672241,
      "learning_rate": 3.2821414040526263e-06,
      "loss": 0.125,
      "step": 8692
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4246973991394043,
      "learning_rate": 3.2810508276235787e-06,
      "loss": 0.1092,
      "step": 8693
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.65558910369873,
      "learning_rate": 3.2799603439253303e-06,
      "loss": 0.6412,
      "step": 8694
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.964241027832031,
      "learning_rate": 3.2788699530167135e-06,
      "loss": 0.8051,
      "step": 8695
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.194253921508789,
      "learning_rate": 3.277779654956546e-06,
      "loss": 0.9923,
      "step": 8696
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.7022581100463867,
      "learning_rate": 3.2766894498036484e-06,
      "loss": 0.096,
      "step": 8697
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.364922523498535,
      "learning_rate": 3.2755993376168306e-06,
      "loss": 0.1065,
      "step": 8698
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.084733963012695,
      "learning_rate": 3.2745093184549015e-06,
      "loss": 0.7936,
      "step": 8699
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.5535244941711426,
      "learning_rate": 3.273419392376663e-06,
      "loss": 0.2071,
      "step": 8700
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.574382781982422,
      "learning_rate": 3.2723295594409134e-06,
      "loss": 0.5265,
      "step": 8701
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.705853462219238,
      "learning_rate": 3.2712398197064444e-06,
      "loss": 0.959,
      "step": 8702
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.660196542739868,
      "learning_rate": 3.270150173232043e-06,
      "loss": 0.077,
      "step": 8703
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.0220272541046143,
      "learning_rate": 3.2690606200764918e-06,
      "loss": 0.1438,
      "step": 8704
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.590179443359375,
      "learning_rate": 3.2679711602985697e-06,
      "loss": 0.7113,
      "step": 8705
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4160733222961426,
      "learning_rate": 3.266881793957046e-06,
      "loss": 0.1113,
      "step": 8706
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.275333881378174,
      "learning_rate": 3.265792521110691e-06,
      "loss": 0.7501,
      "step": 8707
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.432378768920898,
      "learning_rate": 3.2647033418182643e-06,
      "loss": 0.683,
      "step": 8708
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.298126220703125,
      "learning_rate": 3.263614256138523e-06,
      "loss": 0.1132,
      "step": 8709
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.77895975112915,
      "learning_rate": 3.26252526413022e-06,
      "loss": 0.9106,
      "step": 8710
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.363861083984375,
      "learning_rate": 3.261436365852103e-06,
      "loss": 0.2447,
      "step": 8711
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.425225019454956,
      "learning_rate": 3.2603475613629134e-06,
      "loss": 0.0722,
      "step": 8712
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.0717949867248535,
      "learning_rate": 3.2592588507213873e-06,
      "loss": 0.4179,
      "step": 8713
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2474610805511475,
      "learning_rate": 3.2581702339862575e-06,
      "loss": 0.1053,
      "step": 8714
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.6822476387023926,
      "learning_rate": 3.2570817112162496e-06,
      "loss": 0.7546,
      "step": 8715
    },
    {
      "epoch": 0.63,
      "grad_norm": 15.077101707458496,
      "learning_rate": 3.255993282470087e-06,
      "loss": 1.0338,
      "step": 8716
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.503369092941284,
      "learning_rate": 3.254904947806485e-06,
      "loss": 0.1199,
      "step": 8717
    },
    {
      "epoch": 0.63,
      "grad_norm": 12.124336242675781,
      "learning_rate": 3.253816707284155e-06,
      "loss": 1.3739,
      "step": 8718
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.051259994506836,
      "learning_rate": 3.2527285609618055e-06,
      "loss": 0.0983,
      "step": 8719
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.973229169845581,
      "learning_rate": 3.251640508898137e-06,
      "loss": 0.1456,
      "step": 8720
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.1093173027038574,
      "learning_rate": 3.2505525511518427e-06,
      "loss": 0.2713,
      "step": 8721
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5021355152130127,
      "learning_rate": 3.2494646877816177e-06,
      "loss": 0.0943,
      "step": 8722
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.20798397064209,
      "learning_rate": 3.2483769188461457e-06,
      "loss": 0.1091,
      "step": 8723
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.8661177158355713,
      "learning_rate": 3.2472892444041086e-06,
      "loss": 0.137,
      "step": 8724
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.42947006225586,
      "learning_rate": 3.2462016645141838e-06,
      "loss": 0.6408,
      "step": 8725
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.0674564838409424,
      "learning_rate": 3.2451141792350393e-06,
      "loss": 0.1528,
      "step": 8726
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.131728172302246,
      "learning_rate": 3.2440267886253428e-06,
      "loss": 0.7548,
      "step": 8727
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.72122859954834,
      "learning_rate": 3.2429394927437542e-06,
      "loss": 0.1448,
      "step": 8728
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.9187347888946533,
      "learning_rate": 3.241852291648928e-06,
      "loss": 0.1646,
      "step": 8729
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.144584655761719,
      "learning_rate": 3.240765185399518e-06,
      "loss": 0.8669,
      "step": 8730
    },
    {
      "epoch": 0.63,
      "grad_norm": 23.517866134643555,
      "learning_rate": 3.2396781740541655e-06,
      "loss": 0.6012,
      "step": 8731
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.763716697692871,
      "learning_rate": 3.2385912576715143e-06,
      "loss": 0.7299,
      "step": 8732
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.079650402069092,
      "learning_rate": 3.237504436310196e-06,
      "loss": 0.1545,
      "step": 8733
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.515554666519165,
      "learning_rate": 3.2364177100288424e-06,
      "loss": 0.1596,
      "step": 8734
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.8940341472625732,
      "learning_rate": 3.235331078886077e-06,
      "loss": 0.8619,
      "step": 8735
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.094059705734253,
      "learning_rate": 3.234244542940521e-06,
      "loss": 0.0775,
      "step": 8736
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.524106740951538,
      "learning_rate": 3.2331581022507875e-06,
      "loss": 0.1032,
      "step": 8737
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.923025369644165,
      "learning_rate": 3.232071756875487e-06,
      "loss": 0.2086,
      "step": 8738
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.292816162109375,
      "learning_rate": 3.230985506873223e-06,
      "loss": 0.9836,
      "step": 8739
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7250896692276,
      "learning_rate": 3.2298993523025944e-06,
      "loss": 0.0418,
      "step": 8740
    },
    {
      "epoch": 0.63,
      "grad_norm": 19.186418533325195,
      "learning_rate": 3.2288132932221972e-06,
      "loss": 0.9404,
      "step": 8741
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.691434621810913,
      "learning_rate": 3.2277273296906164e-06,
      "loss": 0.1428,
      "step": 8742
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.434950113296509,
      "learning_rate": 3.2266414617664387e-06,
      "loss": 0.1,
      "step": 8743
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1528677940368652,
      "learning_rate": 3.225555689508243e-06,
      "loss": 0.1379,
      "step": 8744
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.100724220275879,
      "learning_rate": 3.224470012974601e-06,
      "loss": 0.2271,
      "step": 8745
    },
    {
      "epoch": 0.63,
      "grad_norm": 18.287208557128906,
      "learning_rate": 3.2233844322240794e-06,
      "loss": 1.0394,
      "step": 8746
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.73346209526062,
      "learning_rate": 3.222298947315243e-06,
      "loss": 0.1513,
      "step": 8747
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.5807318687438965,
      "learning_rate": 3.22121355830665e-06,
      "loss": 0.5822,
      "step": 8748
    },
    {
      "epoch": 0.63,
      "grad_norm": 19.406200408935547,
      "learning_rate": 3.220128265256852e-06,
      "loss": 0.6948,
      "step": 8749
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.18993854522705,
      "learning_rate": 3.2190430682243976e-06,
      "loss": 0.4396,
      "step": 8750
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.644095420837402,
      "learning_rate": 3.217957967267827e-06,
      "loss": 1.0308,
      "step": 8751
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.63941764831543,
      "learning_rate": 3.2168729624456795e-06,
      "loss": 0.6427,
      "step": 8752
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.77917218208313,
      "learning_rate": 3.215788053816487e-06,
      "loss": 0.1544,
      "step": 8753
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.8090951442718506,
      "learning_rate": 3.214703241438774e-06,
      "loss": 0.2296,
      "step": 8754
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.239985942840576,
      "learning_rate": 3.213618525371065e-06,
      "loss": 0.1057,
      "step": 8755
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.370846271514893,
      "learning_rate": 3.212533905671874e-06,
      "loss": 0.3649,
      "step": 8756
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.54052448272705,
      "learning_rate": 3.2114493823997132e-06,
      "loss": 0.5221,
      "step": 8757
    },
    {
      "epoch": 0.63,
      "grad_norm": 12.076011657714844,
      "learning_rate": 3.2103649556130878e-06,
      "loss": 1.0237,
      "step": 8758
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.086844444274902,
      "learning_rate": 3.209280625370499e-06,
      "loss": 0.3267,
      "step": 8759
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5546025037765503,
      "learning_rate": 3.2081963917304416e-06,
      "loss": 0.0484,
      "step": 8760
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6267101764678955,
      "learning_rate": 3.2071122547514067e-06,
      "loss": 0.1495,
      "step": 8761
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6878952980041504,
      "learning_rate": 3.2060282144918796e-06,
      "loss": 0.1294,
      "step": 8762
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.04079532623291,
      "learning_rate": 3.2049442710103395e-06,
      "loss": 0.7853,
      "step": 8763
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.498748779296875,
      "learning_rate": 3.2038604243652617e-06,
      "loss": 0.7467,
      "step": 8764
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.8925557136535645,
      "learning_rate": 3.2027766746151144e-06,
      "loss": 0.1302,
      "step": 8765
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.328379154205322,
      "learning_rate": 3.2016930218183628e-06,
      "loss": 1.0005,
      "step": 8766
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.284675598144531,
      "learning_rate": 3.200609466033467e-06,
      "loss": 0.5946,
      "step": 8767
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.693887710571289,
      "learning_rate": 3.19952600731888e-06,
      "loss": 0.7578,
      "step": 8768
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.699800729751587,
      "learning_rate": 3.1984426457330475e-06,
      "loss": 0.1266,
      "step": 8769
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6608829498291016,
      "learning_rate": 3.1973593813344162e-06,
      "loss": 0.153,
      "step": 8770
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.8296127319335938,
      "learning_rate": 3.196276214181423e-06,
      "loss": 0.2226,
      "step": 8771
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.648313283920288,
      "learning_rate": 3.195193144332499e-06,
      "loss": 0.1576,
      "step": 8772
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5058982372283936,
      "learning_rate": 3.1941101718460756e-06,
      "loss": 0.0759,
      "step": 8773
    },
    {
      "epoch": 0.63,
      "grad_norm": 16.97702407836914,
      "learning_rate": 3.193027296780571e-06,
      "loss": 0.8802,
      "step": 8774
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.0202555656433105,
      "learning_rate": 3.191944519194404e-06,
      "loss": 0.4547,
      "step": 8775
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.562135219573975,
      "learning_rate": 3.1908618391459874e-06,
      "loss": 0.8892,
      "step": 8776
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.157760620117188,
      "learning_rate": 3.189779256693726e-06,
      "loss": 0.6533,
      "step": 8777
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.232048988342285,
      "learning_rate": 3.1886967718960214e-06,
      "loss": 0.0962,
      "step": 8778
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5726535320281982,
      "learning_rate": 3.1876143848112702e-06,
      "loss": 0.1602,
      "step": 8779
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.492239952087402,
      "learning_rate": 3.186532095497864e-06,
      "loss": 0.7301,
      "step": 8780
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.743964910507202,
      "learning_rate": 3.185449904014185e-06,
      "loss": 0.1959,
      "step": 8781
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6545581817626953,
      "learning_rate": 3.1843678104186167e-06,
      "loss": 0.116,
      "step": 8782
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5176312923431396,
      "learning_rate": 3.183285814769531e-06,
      "loss": 0.1425,
      "step": 8783
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.821882247924805,
      "learning_rate": 3.182203917125299e-06,
      "loss": 0.471,
      "step": 8784
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.149299144744873,
      "learning_rate": 3.1811221175442864e-06,
      "loss": 1.0218,
      "step": 8785
    },
    {
      "epoch": 0.63,
      "grad_norm": 12.653944969177246,
      "learning_rate": 3.1800404160848487e-06,
      "loss": 0.9586,
      "step": 8786
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.913485288619995,
      "learning_rate": 3.1789588128053433e-06,
      "loss": 0.1738,
      "step": 8787
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.817871570587158,
      "learning_rate": 3.1778773077641157e-06,
      "loss": 0.1774,
      "step": 8788
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.2817671298980713,
      "learning_rate": 3.17679590101951e-06,
      "loss": 0.1269,
      "step": 8789
    },
    {
      "epoch": 0.63,
      "grad_norm": 16.928464889526367,
      "learning_rate": 3.175714592629866e-06,
      "loss": 0.8643,
      "step": 8790
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.0603740215301514,
      "learning_rate": 3.1746333826535137e-06,
      "loss": 0.0732,
      "step": 8791
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0409276485443115,
      "learning_rate": 3.173552271148782e-06,
      "loss": 0.1356,
      "step": 8792
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.892158031463623,
      "learning_rate": 3.1724712581739914e-06,
      "loss": 0.784,
      "step": 8793
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.594255447387695,
      "learning_rate": 3.1713903437874578e-06,
      "loss": 0.6865,
      "step": 8794
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2981584072113037,
      "learning_rate": 3.1703095280474933e-06,
      "loss": 0.0987,
      "step": 8795
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.096254348754883,
      "learning_rate": 3.169228811012405e-06,
      "loss": 0.8324,
      "step": 8796
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.99742317199707,
      "learning_rate": 3.1681481927404916e-06,
      "loss": 0.5582,
      "step": 8797
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.089985370635986,
      "learning_rate": 3.1670676732900496e-06,
      "loss": 0.9135,
      "step": 8798
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.96384334564209,
      "learning_rate": 3.165987252719369e-06,
      "loss": 0.472,
      "step": 8799
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5109362602233887,
      "learning_rate": 3.164906931086733e-06,
      "loss": 0.0992,
      "step": 8800
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.65230393409729,
      "learning_rate": 3.1638267084504233e-06,
      "loss": 0.1602,
      "step": 8801
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8747061491012573,
      "learning_rate": 3.1627465848687113e-06,
      "loss": 0.0488,
      "step": 8802
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1390089988708496,
      "learning_rate": 3.1616665603998664e-06,
      "loss": 0.1031,
      "step": 8803
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.293097496032715,
      "learning_rate": 3.160586635102154e-06,
      "loss": 0.5329,
      "step": 8804
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.415973424911499,
      "learning_rate": 3.1595068090338292e-06,
      "loss": 0.1431,
      "step": 8805
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.165891647338867,
      "learning_rate": 3.158427082253145e-06,
      "loss": 0.2662,
      "step": 8806
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.4872772693634033,
      "learning_rate": 3.1573474548183493e-06,
      "loss": 0.647,
      "step": 8807
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3497354984283447,
      "learning_rate": 3.1562679267876827e-06,
      "loss": 0.1449,
      "step": 8808
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1999404430389404,
      "learning_rate": 3.1551884982193827e-06,
      "loss": 0.1225,
      "step": 8809
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.877031326293945,
      "learning_rate": 3.1541091691716806e-06,
      "loss": 0.5837,
      "step": 8810
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.163743495941162,
      "learning_rate": 3.153029939702802e-06,
      "loss": 0.6774,
      "step": 8811
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.665919303894043,
      "learning_rate": 3.151950809870966e-06,
      "loss": 1.2218,
      "step": 8812
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.559042930603027,
      "learning_rate": 3.1508717797343903e-06,
      "loss": 0.9117,
      "step": 8813
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.337372064590454,
      "learning_rate": 3.1497928493512808e-06,
      "loss": 0.1141,
      "step": 8814
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.119571685791016,
      "learning_rate": 3.1487140187798453e-06,
      "loss": 0.7624,
      "step": 8815
    },
    {
      "epoch": 0.63,
      "grad_norm": 16.81016731262207,
      "learning_rate": 3.147635288078282e-06,
      "loss": 1.0438,
      "step": 8816
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.469158411026001,
      "learning_rate": 3.1465566573047814e-06,
      "loss": 0.1909,
      "step": 8817
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.3768088817596436,
      "learning_rate": 3.1454781265175327e-06,
      "loss": 0.7116,
      "step": 8818
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.0979456901550293,
      "learning_rate": 3.1443996957747215e-06,
      "loss": 0.1592,
      "step": 8819
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.992790222167969,
      "learning_rate": 3.1433213651345206e-06,
      "loss": 0.8528,
      "step": 8820
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.786016464233398,
      "learning_rate": 3.1422431346551064e-06,
      "loss": 0.5698,
      "step": 8821
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.754289627075195,
      "learning_rate": 3.1411650043946417e-06,
      "loss": 1.1624,
      "step": 8822
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.682586669921875,
      "learning_rate": 3.1400869744112885e-06,
      "loss": 0.5587,
      "step": 8823
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.345458507537842,
      "learning_rate": 3.139009044763204e-06,
      "loss": 0.9204,
      "step": 8824
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.29963493347168,
      "learning_rate": 3.137931215508536e-06,
      "loss": 0.9913,
      "step": 8825
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.229486465454102,
      "learning_rate": 3.136853486705431e-06,
      "loss": 0.8647,
      "step": 8826
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.235265254974365,
      "learning_rate": 3.1357758584120294e-06,
      "loss": 0.8192,
      "step": 8827
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.480478048324585,
      "learning_rate": 3.1346983306864646e-06,
      "loss": 0.0462,
      "step": 8828
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.285215139389038,
      "learning_rate": 3.1336209035868626e-06,
      "loss": 0.0947,
      "step": 8829
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.688563823699951,
      "learning_rate": 3.1325435771713485e-06,
      "loss": 0.1668,
      "step": 8830
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.895747184753418,
      "learning_rate": 3.1314663514980393e-06,
      "loss": 0.1309,
      "step": 8831
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.690118789672852,
      "learning_rate": 3.130389226625047e-06,
      "loss": 0.8356,
      "step": 8832
    },
    {
      "epoch": 0.63,
      "grad_norm": 21.873231887817383,
      "learning_rate": 3.1293122026104806e-06,
      "loss": 0.9547,
      "step": 8833
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.441153526306152,
      "learning_rate": 3.128235279512438e-06,
      "loss": 0.7473,
      "step": 8834
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.481649398803711,
      "learning_rate": 3.1271584573890192e-06,
      "loss": 0.93,
      "step": 8835
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.0577609539031982,
      "learning_rate": 3.126081736298311e-06,
      "loss": 0.1442,
      "step": 8836
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.7327287197113037,
      "learning_rate": 3.1250051162983996e-06,
      "loss": 0.2796,
      "step": 8837
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.359199047088623,
      "learning_rate": 3.1239285974473665e-06,
      "loss": 0.7334,
      "step": 8838
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.336320161819458,
      "learning_rate": 3.1228521798032833e-06,
      "loss": 0.264,
      "step": 8839
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1308765411376953,
      "learning_rate": 3.1217758634242218e-06,
      "loss": 0.0477,
      "step": 8840
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4012577533721924,
      "learning_rate": 3.1206996483682404e-06,
      "loss": 0.0837,
      "step": 8841
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.363861560821533,
      "learning_rate": 3.1196235346934013e-06,
      "loss": 0.2585,
      "step": 8842
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.984398365020752,
      "learning_rate": 3.118547522457753e-06,
      "loss": 0.1734,
      "step": 8843
    },
    {
      "epoch": 0.63,
      "grad_norm": 15.304707527160645,
      "learning_rate": 3.1174716117193447e-06,
      "loss": 0.5963,
      "step": 8844
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.583066940307617,
      "learning_rate": 3.1163958025362173e-06,
      "loss": 1.3673,
      "step": 8845
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6675453186035156,
      "learning_rate": 3.1153200949664065e-06,
      "loss": 0.1709,
      "step": 8846
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9660924673080444,
      "learning_rate": 3.1142444890679436e-06,
      "loss": 0.1155,
      "step": 8847
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.479954242706299,
      "learning_rate": 3.1131689848988513e-06,
      "loss": 0.7949,
      "step": 8848
    },
    {
      "epoch": 0.63,
      "grad_norm": 16.06068992614746,
      "learning_rate": 3.1120935825171505e-06,
      "loss": 0.8079,
      "step": 8849
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.774599075317383,
      "learning_rate": 3.111018281980855e-06,
      "loss": 0.1408,
      "step": 8850
    },
    {
      "epoch": 0.63,
      "grad_norm": 15.590961456298828,
      "learning_rate": 3.1099430833479728e-06,
      "loss": 0.7666,
      "step": 8851
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.314035415649414,
      "learning_rate": 3.108867986676508e-06,
      "loss": 0.9605,
      "step": 8852
    },
    {
      "epoch": 0.63,
      "grad_norm": 19.584367752075195,
      "learning_rate": 3.107792992024457e-06,
      "loss": 1.0353,
      "step": 8853
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.666403770446777,
      "learning_rate": 3.1067180994498104e-06,
      "loss": 0.3275,
      "step": 8854
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.549615859985352,
      "learning_rate": 3.105643309010556e-06,
      "loss": 0.4194,
      "step": 8855
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.0263712406158447,
      "learning_rate": 3.104568620764675e-06,
      "loss": 0.1302,
      "step": 8856
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.879495143890381,
      "learning_rate": 3.103494034770142e-06,
      "loss": 0.9306,
      "step": 8857
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.5412797927856445,
      "learning_rate": 3.102419551084928e-06,
      "loss": 0.9611,
      "step": 8858
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.519531488418579,
      "learning_rate": 3.101345169766995e-06,
      "loss": 0.7358,
      "step": 8859
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.93397045135498,
      "learning_rate": 3.100270890874304e-06,
      "loss": 1.0257,
      "step": 8860
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.069218635559082,
      "learning_rate": 3.099196714464808e-06,
      "loss": 0.8074,
      "step": 8861
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.032261848449707,
      "learning_rate": 3.098122640596454e-06,
      "loss": 0.6173,
      "step": 8862
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.108741283416748,
      "learning_rate": 3.097048669327185e-06,
      "loss": 0.1546,
      "step": 8863
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9616525173187256,
      "learning_rate": 3.095974800714937e-06,
      "loss": 0.101,
      "step": 8864
    },
    {
      "epoch": 0.64,
      "grad_norm": 18.940229415893555,
      "learning_rate": 3.094901034817642e-06,
      "loss": 0.8904,
      "step": 8865
    },
    {
      "epoch": 0.64,
      "grad_norm": 29.0656795501709,
      "learning_rate": 3.093827371693224e-06,
      "loss": 0.6824,
      "step": 8866
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.910533905029297,
      "learning_rate": 3.092753811399605e-06,
      "loss": 0.7288,
      "step": 8867
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.9951019287109375,
      "learning_rate": 3.0916803539946972e-06,
      "loss": 0.6803,
      "step": 8868
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.358907699584961,
      "learning_rate": 3.090606999536411e-06,
      "loss": 0.0739,
      "step": 8869
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.8480849266052246,
      "learning_rate": 3.0895337480826503e-06,
      "loss": 0.6705,
      "step": 8870
    },
    {
      "epoch": 0.64,
      "grad_norm": 15.618127822875977,
      "learning_rate": 3.088460599691312e-06,
      "loss": 0.5377,
      "step": 8871
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.00113296508789,
      "learning_rate": 3.0873875544202895e-06,
      "loss": 0.8281,
      "step": 8872
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.643512487411499,
      "learning_rate": 3.086314612327468e-06,
      "loss": 0.1149,
      "step": 8873
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.0850586891174316,
      "learning_rate": 3.0852417734707296e-06,
      "loss": 0.2119,
      "step": 8874
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.333196640014648,
      "learning_rate": 3.08416903790795e-06,
      "loss": 0.8023,
      "step": 8875
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.7105581760406494,
      "learning_rate": 3.0830964056970007e-06,
      "loss": 0.1672,
      "step": 8876
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.772700786590576,
      "learning_rate": 3.082023876895742e-06,
      "loss": 0.0904,
      "step": 8877
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.8782763481140137,
      "learning_rate": 3.080951451562036e-06,
      "loss": 0.0782,
      "step": 8878
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.663632392883301,
      "learning_rate": 3.0798791297537356e-06,
      "loss": 0.6601,
      "step": 8879
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.288033485412598,
      "learning_rate": 3.0788069115286873e-06,
      "loss": 0.768,
      "step": 8880
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.882341980934143,
      "learning_rate": 3.0777347969447346e-06,
      "loss": 0.0457,
      "step": 8881
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.737321376800537,
      "learning_rate": 3.0766627860597133e-06,
      "loss": 0.5952,
      "step": 8882
    },
    {
      "epoch": 0.64,
      "grad_norm": 16.795469284057617,
      "learning_rate": 3.0755908789314537e-06,
      "loss": 0.6753,
      "step": 8883
    },
    {
      "epoch": 0.64,
      "grad_norm": 15.88371753692627,
      "learning_rate": 3.0745190756177833e-06,
      "loss": 0.7247,
      "step": 8884
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.9140195846557617,
      "learning_rate": 3.0734473761765195e-06,
      "loss": 0.7707,
      "step": 8885
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.608051300048828,
      "learning_rate": 3.0723757806654776e-06,
      "loss": 0.6631,
      "step": 8886
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8831614255905151,
      "learning_rate": 3.071304289142465e-06,
      "loss": 0.0581,
      "step": 8887
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.387319326400757,
      "learning_rate": 3.070232901665288e-06,
      "loss": 0.1244,
      "step": 8888
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.55281925201416,
      "learning_rate": 3.0691616182917394e-06,
      "loss": 0.7888,
      "step": 8889
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.342281341552734,
      "learning_rate": 3.068090439079614e-06,
      "loss": 1.1361,
      "step": 8890
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3988282680511475,
      "learning_rate": 3.067019364086695e-06,
      "loss": 0.1202,
      "step": 8891
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.1177573204040527,
      "learning_rate": 3.0659483933707656e-06,
      "loss": 0.1751,
      "step": 8892
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3085694313049316,
      "learning_rate": 3.0648775269895997e-06,
      "loss": 0.1262,
      "step": 8893
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.608860731124878,
      "learning_rate": 3.0638067650009657e-06,
      "loss": 0.1439,
      "step": 8894
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.749284505844116,
      "learning_rate": 3.0627361074626288e-06,
      "loss": 0.3502,
      "step": 8895
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.99385404586792,
      "learning_rate": 3.0616655544323447e-06,
      "loss": 0.6771,
      "step": 8896
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2358360290527344,
      "learning_rate": 3.060595105967867e-06,
      "loss": 0.1649,
      "step": 8897
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.89311408996582,
      "learning_rate": 3.0595247621269443e-06,
      "loss": 0.7406,
      "step": 8898
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.9290690422058105,
      "learning_rate": 3.0584545229673133e-06,
      "loss": 0.7076,
      "step": 8899
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.249547004699707,
      "learning_rate": 3.0573843885467147e-06,
      "loss": 1.0014,
      "step": 8900
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.964445114135742,
      "learning_rate": 3.056314358922874e-06,
      "loss": 0.1756,
      "step": 8901
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.612100124359131,
      "learning_rate": 3.055244434153515e-06,
      "loss": 0.7382,
      "step": 8902
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0260937213897705,
      "learning_rate": 3.054174614296357e-06,
      "loss": 0.0788,
      "step": 8903
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.108039379119873,
      "learning_rate": 3.053104899409115e-06,
      "loss": 0.1061,
      "step": 8904
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.786550521850586,
      "learning_rate": 3.0520352895494925e-06,
      "loss": 0.8976,
      "step": 8905
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.280351638793945,
      "learning_rate": 3.050965784775193e-06,
      "loss": 0.8599,
      "step": 8906
    },
    {
      "epoch": 0.64,
      "grad_norm": 15.66466236114502,
      "learning_rate": 3.0498963851439136e-06,
      "loss": 0.8741,
      "step": 8907
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.35497522354126,
      "learning_rate": 3.0488270907133412e-06,
      "loss": 0.0549,
      "step": 8908
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.732430934906006,
      "learning_rate": 3.0477579015411625e-06,
      "loss": 0.7581,
      "step": 8909
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8843566179275513,
      "learning_rate": 3.046688817685054e-06,
      "loss": 0.0869,
      "step": 8910
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4906673431396484,
      "learning_rate": 3.045619839202691e-06,
      "loss": 0.1149,
      "step": 8911
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.352766275405884,
      "learning_rate": 3.0445509661517416e-06,
      "loss": 0.0978,
      "step": 8912
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.19185209274292,
      "learning_rate": 3.043482198589864e-06,
      "loss": 0.1057,
      "step": 8913
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2563602924346924,
      "learning_rate": 3.0424135365747164e-06,
      "loss": 0.0995,
      "step": 8914
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.671752452850342,
      "learning_rate": 3.0413449801639484e-06,
      "loss": 0.2021,
      "step": 8915
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.184818744659424,
      "learning_rate": 3.0402765294152048e-06,
      "loss": 0.8482,
      "step": 8916
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.308720827102661,
      "learning_rate": 3.039208184386124e-06,
      "loss": 0.153,
      "step": 8917
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.173837661743164,
      "learning_rate": 3.038139945134341e-06,
      "loss": 0.7798,
      "step": 8918
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.796863317489624,
      "learning_rate": 3.03707181171748e-06,
      "loss": 0.1853,
      "step": 8919
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.075629234313965,
      "learning_rate": 3.0360037841931654e-06,
      "loss": 0.065,
      "step": 8920
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.665930986404419,
      "learning_rate": 3.0349358626190136e-06,
      "loss": 0.6559,
      "step": 8921
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.906554222106934,
      "learning_rate": 3.033868047052632e-06,
      "loss": 0.8252,
      "step": 8922
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.402189254760742,
      "learning_rate": 3.032800337551629e-06,
      "loss": 1.5516,
      "step": 8923
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.1661479473114014,
      "learning_rate": 3.0317327341736024e-06,
      "loss": 0.1511,
      "step": 8924
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9707555770874023,
      "learning_rate": 3.0306652369761425e-06,
      "loss": 0.1003,
      "step": 8925
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.8504390716552734,
      "learning_rate": 3.029597846016838e-06,
      "loss": 0.1397,
      "step": 8926
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.8758575916290283,
      "learning_rate": 3.028530561353273e-06,
      "loss": 0.1581,
      "step": 8927
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1755502223968506,
      "learning_rate": 3.0274633830430205e-06,
      "loss": 0.0668,
      "step": 8928
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.108320951461792,
      "learning_rate": 3.026396311143653e-06,
      "loss": 0.1655,
      "step": 8929
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.0499043464660645,
      "learning_rate": 3.025329345712733e-06,
      "loss": 0.1323,
      "step": 8930
    },
    {
      "epoch": 0.64,
      "grad_norm": 39.73313903808594,
      "learning_rate": 3.024262486807821e-06,
      "loss": 0.8055,
      "step": 8931
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.677069187164307,
      "learning_rate": 3.0231957344864693e-06,
      "loss": 0.7768,
      "step": 8932
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3221664428710938,
      "learning_rate": 3.022129088806225e-06,
      "loss": 0.0787,
      "step": 8933
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.769790649414062,
      "learning_rate": 3.0210625498246295e-06,
      "loss": 1.1194,
      "step": 8934
    },
    {
      "epoch": 0.64,
      "grad_norm": 19.02573013305664,
      "learning_rate": 3.0199961175992187e-06,
      "loss": 1.2324,
      "step": 8935
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.00317645072937,
      "learning_rate": 3.0189297921875226e-06,
      "loss": 0.0922,
      "step": 8936
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.609257221221924,
      "learning_rate": 3.0178635736470674e-06,
      "loss": 0.9268,
      "step": 8937
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.801812171936035,
      "learning_rate": 3.0167974620353687e-06,
      "loss": 0.9687,
      "step": 8938
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.850036382675171,
      "learning_rate": 3.0157314574099384e-06,
      "loss": 0.1007,
      "step": 8939
    },
    {
      "epoch": 0.64,
      "grad_norm": 28.73263168334961,
      "learning_rate": 3.014665559828286e-06,
      "loss": 0.5311,
      "step": 8940
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.400712728500366,
      "learning_rate": 3.0135997693479126e-06,
      "loss": 0.2103,
      "step": 8941
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.495706081390381,
      "learning_rate": 3.0125340860263107e-06,
      "loss": 0.319,
      "step": 8942
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.296550989151001,
      "learning_rate": 3.011468509920973e-06,
      "loss": 0.0545,
      "step": 8943
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.727335929870605,
      "learning_rate": 3.0104030410893827e-06,
      "loss": 0.6132,
      "step": 8944
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.160783767700195,
      "learning_rate": 3.0093376795890167e-06,
      "loss": 0.3923,
      "step": 8945
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.552675724029541,
      "learning_rate": 3.0082724254773487e-06,
      "loss": 0.8124,
      "step": 8946
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.687602996826172,
      "learning_rate": 3.0072072788118427e-06,
      "loss": 0.555,
      "step": 8947
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0957303047180176,
      "learning_rate": 3.0061422396499618e-06,
      "loss": 0.093,
      "step": 8948
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.574179172515869,
      "learning_rate": 3.0050773080491615e-06,
      "loss": 0.7704,
      "step": 8949
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.927292346954346,
      "learning_rate": 3.0040124840668876e-06,
      "loss": 0.7975,
      "step": 8950
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7515835762023926,
      "learning_rate": 3.0029477677605846e-06,
      "loss": 0.0954,
      "step": 8951
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.5968451499938965,
      "learning_rate": 3.0018831591876907e-06,
      "loss": 0.6809,
      "step": 8952
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.879228591918945,
      "learning_rate": 3.000818658405637e-06,
      "loss": 0.8315,
      "step": 8953
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.515209197998047,
      "learning_rate": 2.999754265471849e-06,
      "loss": 1.0445,
      "step": 8954
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5584349632263184,
      "learning_rate": 2.9986899804437473e-06,
      "loss": 0.1319,
      "step": 8955
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1994783878326416,
      "learning_rate": 2.9976258033787457e-06,
      "loss": 0.0564,
      "step": 8956
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.357410430908203,
      "learning_rate": 2.9965617343342525e-06,
      "loss": 0.9495,
      "step": 8957
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2190957069396973,
      "learning_rate": 2.9954977733676714e-06,
      "loss": 0.0808,
      "step": 8958
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.7332639694213867,
      "learning_rate": 2.9944339205363966e-06,
      "loss": 0.2789,
      "step": 8959
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.697751522064209,
      "learning_rate": 2.9933701758978214e-06,
      "loss": 0.7982,
      "step": 8960
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.892444610595703,
      "learning_rate": 2.992306539509331e-06,
      "loss": 1.0086,
      "step": 8961
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.84707260131836,
      "learning_rate": 2.991243011428301e-06,
      "loss": 0.8559,
      "step": 8962
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.8675031661987305,
      "learning_rate": 2.990179591712107e-06,
      "loss": 0.6765,
      "step": 8963
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5566537380218506,
      "learning_rate": 2.989116280418117e-06,
      "loss": 0.1587,
      "step": 8964
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5830471515655518,
      "learning_rate": 2.9880530776036916e-06,
      "loss": 0.1369,
      "step": 8965
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.603846549987793,
      "learning_rate": 2.986989983326188e-06,
      "loss": 0.5641,
      "step": 8966
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.909956932067871,
      "learning_rate": 2.985926997642954e-06,
      "loss": 0.7138,
      "step": 8967
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.934793949127197,
      "learning_rate": 2.984864120611335e-06,
      "loss": 0.5168,
      "step": 8968
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.536638259887695,
      "learning_rate": 2.98380135228867e-06,
      "loss": 0.6353,
      "step": 8969
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.4098238945007324,
      "learning_rate": 2.9827386927322888e-06,
      "loss": 0.2147,
      "step": 8970
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.7843947410583496,
      "learning_rate": 2.98167614199952e-06,
      "loss": 0.7189,
      "step": 8971
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.925921678543091,
      "learning_rate": 2.9806137001476845e-06,
      "loss": 0.176,
      "step": 8972
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.424664497375488,
      "learning_rate": 2.9795513672340963e-06,
      "loss": 0.8871,
      "step": 8973
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.269773483276367,
      "learning_rate": 2.9784891433160623e-06,
      "loss": 0.0941,
      "step": 8974
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.6683220863342285,
      "learning_rate": 2.9774270284508886e-06,
      "loss": 0.2009,
      "step": 8975
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2865073680877686,
      "learning_rate": 2.9763650226958697e-06,
      "loss": 0.0885,
      "step": 8976
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.31566047668457,
      "learning_rate": 2.975303126108298e-06,
      "loss": 0.6254,
      "step": 8977
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4224491119384766,
      "learning_rate": 2.97424133874546e-06,
      "loss": 0.145,
      "step": 8978
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.090012550354004,
      "learning_rate": 2.9731796606646325e-06,
      "loss": 1.0016,
      "step": 8979
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.1319222450256348,
      "learning_rate": 2.972118091923092e-06,
      "loss": 0.2597,
      "step": 8980
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.055716037750244,
      "learning_rate": 2.9710566325781026e-06,
      "loss": 0.3852,
      "step": 8981
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.5552866458892822,
      "learning_rate": 2.9699952826869285e-06,
      "loss": 0.2196,
      "step": 8982
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.253187894821167,
      "learning_rate": 2.968934042306826e-06,
      "loss": 0.1491,
      "step": 8983
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9591031074523926,
      "learning_rate": 2.967872911495042e-06,
      "loss": 0.1572,
      "step": 8984
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.538766622543335,
      "learning_rate": 2.966811890308825e-06,
      "loss": 0.1849,
      "step": 8985
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.078592300415039,
      "learning_rate": 2.9657509788054084e-06,
      "loss": 0.0638,
      "step": 8986
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.1168859004974365,
      "learning_rate": 2.9646901770420277e-06,
      "loss": 0.1395,
      "step": 8987
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.203802585601807,
      "learning_rate": 2.963629485075907e-06,
      "loss": 0.3129,
      "step": 8988
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.2244873046875,
      "learning_rate": 2.962568902964268e-06,
      "loss": 0.3348,
      "step": 8989
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.622404098510742,
      "learning_rate": 2.961508430764324e-06,
      "loss": 0.4044,
      "step": 8990
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.422618389129639,
      "learning_rate": 2.960448068533284e-06,
      "loss": 0.939,
      "step": 8991
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.6262524127960205,
      "learning_rate": 2.9593878163283513e-06,
      "loss": 0.0925,
      "step": 8992
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.719655513763428,
      "learning_rate": 2.9583276742067213e-06,
      "loss": 0.5939,
      "step": 8993
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.376334190368652,
      "learning_rate": 2.9572676422255854e-06,
      "loss": 0.5817,
      "step": 8994
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.30108118057251,
      "learning_rate": 2.9562077204421273e-06,
      "loss": 0.9141,
      "step": 8995
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.4307618141174316,
      "learning_rate": 2.9551479089135268e-06,
      "loss": 0.2321,
      "step": 8996
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1248717308044434,
      "learning_rate": 2.954088207696958e-06,
      "loss": 0.2144,
      "step": 8997
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.399395227432251,
      "learning_rate": 2.953028616849586e-06,
      "loss": 0.1118,
      "step": 8998
    },
    {
      "epoch": 0.65,
      "grad_norm": 17.222761154174805,
      "learning_rate": 2.9519691364285696e-06,
      "loss": 0.7992,
      "step": 8999
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.037490367889404,
      "learning_rate": 2.950909766491067e-06,
      "loss": 0.8387,
      "step": 9000
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.768609046936035,
      "learning_rate": 2.9498505070942274e-06,
      "loss": 1.2415,
      "step": 9001
    },
    {
      "epoch": 0.65,
      "grad_norm": 12.145021438598633,
      "learning_rate": 2.9487913582951912e-06,
      "loss": 0.7011,
      "step": 9002
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.108929395675659,
      "learning_rate": 2.947732320151098e-06,
      "loss": 0.1848,
      "step": 9003
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.0857203006744385,
      "learning_rate": 2.9466733927190773e-06,
      "loss": 0.1068,
      "step": 9004
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.426593065261841,
      "learning_rate": 2.945614576056254e-06,
      "loss": 0.0994,
      "step": 9005
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.6031970977783203,
      "learning_rate": 2.9445558702197495e-06,
      "loss": 0.0909,
      "step": 9006
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4071526527404785,
      "learning_rate": 2.943497275266675e-06,
      "loss": 0.1281,
      "step": 9007
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.403523921966553,
      "learning_rate": 2.9424387912541385e-06,
      "loss": 1.0063,
      "step": 9008
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.341465950012207,
      "learning_rate": 2.9413804182392413e-06,
      "loss": 0.9242,
      "step": 9009
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7891669273376465,
      "learning_rate": 2.9403221562790787e-06,
      "loss": 0.8092,
      "step": 9010
    },
    {
      "epoch": 0.65,
      "grad_norm": 19.615116119384766,
      "learning_rate": 2.9392640054307376e-06,
      "loss": 0.4816,
      "step": 9011
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.242965221405029,
      "learning_rate": 2.9382059657513047e-06,
      "loss": 0.9689,
      "step": 9012
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.348386287689209,
      "learning_rate": 2.9371480372978544e-06,
      "loss": 0.7418,
      "step": 9013
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.112396240234375,
      "learning_rate": 2.936090220127459e-06,
      "loss": 1.0461,
      "step": 9014
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.004039287567139,
      "learning_rate": 2.9350325142971847e-06,
      "loss": 0.8918,
      "step": 9015
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.577829360961914,
      "learning_rate": 2.93397491986409e-06,
      "loss": 0.5926,
      "step": 9016
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.775197982788086,
      "learning_rate": 2.9329174368852283e-06,
      "loss": 0.5158,
      "step": 9017
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.845917701721191,
      "learning_rate": 2.9318600654176454e-06,
      "loss": 0.667,
      "step": 9018
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1590447425842285,
      "learning_rate": 2.9308028055183836e-06,
      "loss": 0.141,
      "step": 9019
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.9024243354797363,
      "learning_rate": 2.9297456572444796e-06,
      "loss": 0.1232,
      "step": 9020
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.165023803710938,
      "learning_rate": 2.928688620652962e-06,
      "loss": 0.9589,
      "step": 9021
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.945909261703491,
      "learning_rate": 2.9276316958008498e-06,
      "loss": 0.9963,
      "step": 9022
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.726581573486328,
      "learning_rate": 2.9265748827451647e-06,
      "loss": 0.5774,
      "step": 9023
    },
    {
      "epoch": 0.65,
      "grad_norm": 15.237038612365723,
      "learning_rate": 2.9255181815429168e-06,
      "loss": 1.3453,
      "step": 9024
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.095846176147461,
      "learning_rate": 2.92446159225111e-06,
      "loss": 0.7712,
      "step": 9025
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.866296768188477,
      "learning_rate": 2.9234051149267446e-06,
      "loss": 0.9427,
      "step": 9026
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.433634757995605,
      "learning_rate": 2.922348749626812e-06,
      "loss": 0.7442,
      "step": 9027
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.976709246635437,
      "learning_rate": 2.9212924964083e-06,
      "loss": 0.0689,
      "step": 9028
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.241039991378784,
      "learning_rate": 2.9202363553281904e-06,
      "loss": 0.0864,
      "step": 9029
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.343095302581787,
      "learning_rate": 2.919180326443456e-06,
      "loss": 0.0819,
      "step": 9030
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.325268745422363,
      "learning_rate": 2.918124409811068e-06,
      "loss": 0.7805,
      "step": 9031
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.975519895553589,
      "learning_rate": 2.9170686054879853e-06,
      "loss": 0.1919,
      "step": 9032
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.0135035514831543,
      "learning_rate": 2.9160129135311717e-06,
      "loss": 0.2238,
      "step": 9033
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.8308794498443604,
      "learning_rate": 2.9149573339975708e-06,
      "loss": 0.1513,
      "step": 9034
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.707616329193115,
      "learning_rate": 2.913901866944127e-06,
      "loss": 0.7374,
      "step": 9035
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.519246816635132,
      "learning_rate": 2.912846512427784e-06,
      "loss": 0.8269,
      "step": 9036
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.9998276233673096,
      "learning_rate": 2.911791270505471e-06,
      "loss": 0.1387,
      "step": 9037
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.7005043029785156,
      "learning_rate": 2.910736141234112e-06,
      "loss": 0.1214,
      "step": 9038
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.39480209350586,
      "learning_rate": 2.9096811246706324e-06,
      "loss": 1.1396,
      "step": 9039
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.500946044921875,
      "learning_rate": 2.908626220871944e-06,
      "loss": 0.1066,
      "step": 9040
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.004416465759277,
      "learning_rate": 2.907571429894954e-06,
      "loss": 0.6641,
      "step": 9041
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7244263887405396,
      "learning_rate": 2.906516751796564e-06,
      "loss": 0.086,
      "step": 9042
    },
    {
      "epoch": 0.65,
      "grad_norm": 18.493595123291016,
      "learning_rate": 2.9054621866336723e-06,
      "loss": 0.7506,
      "step": 9043
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.169363021850586,
      "learning_rate": 2.9044077344631678e-06,
      "loss": 0.2141,
      "step": 9044
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.021859645843506,
      "learning_rate": 2.9033533953419337e-06,
      "loss": 0.1072,
      "step": 9045
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.258803844451904,
      "learning_rate": 2.9022991693268476e-06,
      "loss": 0.6242,
      "step": 9046
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.031019687652588,
      "learning_rate": 2.901245056474779e-06,
      "loss": 0.692,
      "step": 9047
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.632823944091797,
      "learning_rate": 2.9001910568425974e-06,
      "loss": 0.4791,
      "step": 9048
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.3911824226379395,
      "learning_rate": 2.8991371704871607e-06,
      "loss": 0.3583,
      "step": 9049
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.537868499755859,
      "learning_rate": 2.8980833974653206e-06,
      "loss": 0.4517,
      "step": 9050
    },
    {
      "epoch": 0.65,
      "grad_norm": 19.687332153320312,
      "learning_rate": 2.8970297378339236e-06,
      "loss": 0.6502,
      "step": 9051
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.193453311920166,
      "learning_rate": 2.8959761916498136e-06,
      "loss": 0.2136,
      "step": 9052
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.244081974029541,
      "learning_rate": 2.894922758969824e-06,
      "loss": 0.064,
      "step": 9053
    },
    {
      "epoch": 0.65,
      "grad_norm": 16.205013275146484,
      "learning_rate": 2.8938694398507816e-06,
      "loss": 0.9187,
      "step": 9054
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.6168053150177,
      "learning_rate": 2.8928162343495124e-06,
      "loss": 0.1302,
      "step": 9055
    },
    {
      "epoch": 0.65,
      "grad_norm": 15.234685897827148,
      "learning_rate": 2.8917631425228323e-06,
      "loss": 0.472,
      "step": 9056
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.861298084259033,
      "learning_rate": 2.8907101644275493e-06,
      "loss": 0.8172,
      "step": 9057
    },
    {
      "epoch": 0.65,
      "grad_norm": 12.26543140411377,
      "learning_rate": 2.88965730012047e-06,
      "loss": 0.7889,
      "step": 9058
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7260241508483887,
      "learning_rate": 2.8886045496583903e-06,
      "loss": 0.1838,
      "step": 9059
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.804598808288574,
      "learning_rate": 2.8875519130981013e-06,
      "loss": 0.0889,
      "step": 9060
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.051024436950684,
      "learning_rate": 2.886499390496392e-06,
      "loss": 0.9287,
      "step": 9061
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.6628317832946777,
      "learning_rate": 2.8854469819100416e-06,
      "loss": 0.1043,
      "step": 9062
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.77206039428711,
      "learning_rate": 2.88439468739582e-06,
      "loss": 0.9101,
      "step": 9063
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.391231536865234,
      "learning_rate": 2.883342507010499e-06,
      "loss": 0.387,
      "step": 9064
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.914911985397339,
      "learning_rate": 2.8822904408108375e-06,
      "loss": 0.8417,
      "step": 9065
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.224936008453369,
      "learning_rate": 2.8812384888535893e-06,
      "loss": 0.3901,
      "step": 9066
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1107397079467773,
      "learning_rate": 2.8801866511955066e-06,
      "loss": 0.1956,
      "step": 9067
    },
    {
      "epoch": 0.65,
      "grad_norm": 20.735881805419922,
      "learning_rate": 2.87913492789333e-06,
      "loss": 0.8035,
      "step": 9068
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.623055934906006,
      "learning_rate": 2.878083319003796e-06,
      "loss": 0.791,
      "step": 9069
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.113718032836914,
      "learning_rate": 2.8770318245836353e-06,
      "loss": 0.3477,
      "step": 9070
    },
    {
      "epoch": 0.65,
      "grad_norm": 21.948165893554688,
      "learning_rate": 2.8759804446895724e-06,
      "loss": 1.2539,
      "step": 9071
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.926543951034546,
      "learning_rate": 2.874929179378322e-06,
      "loss": 0.682,
      "step": 9072
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.99179220199585,
      "learning_rate": 2.8738780287066e-06,
      "loss": 0.467,
      "step": 9073
    },
    {
      "epoch": 0.65,
      "grad_norm": 16.30853271484375,
      "learning_rate": 2.872826992731111e-06,
      "loss": 0.8015,
      "step": 9074
    },
    {
      "epoch": 0.65,
      "grad_norm": 27.637374877929688,
      "learning_rate": 2.8717760715085517e-06,
      "loss": 1.3228,
      "step": 9075
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.584469795227051,
      "learning_rate": 2.8707252650956202e-06,
      "loss": 0.4874,
      "step": 9076
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.4839253425598145,
      "learning_rate": 2.8696745735490004e-06,
      "loss": 0.1894,
      "step": 9077
    },
    {
      "epoch": 0.65,
      "grad_norm": 15.535033226013184,
      "learning_rate": 2.868623996925374e-06,
      "loss": 0.437,
      "step": 9078
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5834919214248657,
      "learning_rate": 2.867573535281413e-06,
      "loss": 0.0436,
      "step": 9079
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.014207601547241,
      "learning_rate": 2.86652318867379e-06,
      "loss": 0.1338,
      "step": 9080
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.2758853435516357,
      "learning_rate": 2.8654729571591655e-06,
      "loss": 0.1184,
      "step": 9081
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.5148274898529053,
      "learning_rate": 2.8644228407941954e-06,
      "loss": 0.245,
      "step": 9082
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.649662494659424,
      "learning_rate": 2.86337283963553e-06,
      "loss": 0.7614,
      "step": 9083
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.6068365573883057,
      "learning_rate": 2.862322953739809e-06,
      "loss": 0.105,
      "step": 9084
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.807038307189941,
      "learning_rate": 2.8612731831636763e-06,
      "loss": 0.518,
      "step": 9085
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.131817817687988,
      "learning_rate": 2.8602235279637592e-06,
      "loss": 0.7899,
      "step": 9086
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.5004987716674805,
      "learning_rate": 2.859173988196683e-06,
      "loss": 0.0939,
      "step": 9087
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.438072443008423,
      "learning_rate": 2.858124563919066e-06,
      "loss": 0.1045,
      "step": 9088
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4380810260772705,
      "learning_rate": 2.857075255187522e-06,
      "loss": 0.1381,
      "step": 9089
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.839902877807617,
      "learning_rate": 2.8560260620586568e-06,
      "loss": 0.2257,
      "step": 9090
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.685985565185547,
      "learning_rate": 2.8549769845890684e-06,
      "loss": 0.119,
      "step": 9091
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.41091251373291,
      "learning_rate": 2.8539280228353546e-06,
      "loss": 0.0944,
      "step": 9092
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.464810848236084,
      "learning_rate": 2.852879176854102e-06,
      "loss": 0.0523,
      "step": 9093
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.10571026802063,
      "learning_rate": 2.851830446701887e-06,
      "loss": 0.0604,
      "step": 9094
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.395749807357788,
      "learning_rate": 2.85078183243529e-06,
      "loss": 0.091,
      "step": 9095
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.7968857288360596,
      "learning_rate": 2.849733334110878e-06,
      "loss": 0.1313,
      "step": 9096
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.636000871658325,
      "learning_rate": 2.848684951785211e-06,
      "loss": 0.1717,
      "step": 9097
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.701885461807251,
      "learning_rate": 2.84763668551485e-06,
      "loss": 0.1798,
      "step": 9098
    },
    {
      "epoch": 0.65,
      "grad_norm": 18.742900848388672,
      "learning_rate": 2.8465885353563425e-06,
      "loss": 0.8744,
      "step": 9099
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.706136465072632,
      "learning_rate": 2.8455405013662308e-06,
      "loss": 0.0989,
      "step": 9100
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.8158059120178223,
      "learning_rate": 2.844492583601056e-06,
      "loss": 0.1462,
      "step": 9101
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.285682678222656,
      "learning_rate": 2.8434447821173473e-06,
      "loss": 0.6166,
      "step": 9102
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.7236132621765137,
      "learning_rate": 2.8423970969716274e-06,
      "loss": 0.0923,
      "step": 9103
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.408785343170166,
      "learning_rate": 2.841349528220419e-06,
      "loss": 0.4243,
      "step": 9104
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2974283695220947,
      "learning_rate": 2.840302075920235e-06,
      "loss": 0.1436,
      "step": 9105
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.837522506713867,
      "learning_rate": 2.8392547401275744e-06,
      "loss": 0.146,
      "step": 9106
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.041188716888428,
      "learning_rate": 2.838207520898944e-06,
      "loss": 0.6371,
      "step": 9107
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.182731866836548,
      "learning_rate": 2.837160418290835e-06,
      "loss": 0.0693,
      "step": 9108
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.179853439331055,
      "learning_rate": 2.8361134323597326e-06,
      "loss": 0.6604,
      "step": 9109
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.674433708190918,
      "learning_rate": 2.8350665631621222e-06,
      "loss": 0.1293,
      "step": 9110
    },
    {
      "epoch": 0.65,
      "grad_norm": 20.6251163482666,
      "learning_rate": 2.834019810754476e-06,
      "loss": 0.9529,
      "step": 9111
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.528199195861816,
      "learning_rate": 2.83297317519326e-06,
      "loss": 0.5785,
      "step": 9112
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1485798358917236,
      "learning_rate": 2.8319266565349408e-06,
      "loss": 0.2422,
      "step": 9113
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.479328155517578,
      "learning_rate": 2.8308802548359716e-06,
      "loss": 0.6686,
      "step": 9114
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.521477222442627,
      "learning_rate": 2.829833970152802e-06,
      "loss": 0.8629,
      "step": 9115
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.5907950401306152,
      "learning_rate": 2.8287878025418736e-06,
      "loss": 0.1046,
      "step": 9116
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.029919624328613,
      "learning_rate": 2.8277417520596285e-06,
      "loss": 0.6864,
      "step": 9117
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4531707763671875,
      "learning_rate": 2.826695818762489e-06,
      "loss": 0.1035,
      "step": 9118
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7665669918060303,
      "learning_rate": 2.8256500027068867e-06,
      "loss": 0.7892,
      "step": 9119
    },
    {
      "epoch": 0.65,
      "grad_norm": 22.18044662475586,
      "learning_rate": 2.824604303949236e-06,
      "loss": 0.8493,
      "step": 9120
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.1377592086792,
      "learning_rate": 2.8235587225459487e-06,
      "loss": 0.5414,
      "step": 9121
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.9537253379821777,
      "learning_rate": 2.8225132585534277e-06,
      "loss": 0.1718,
      "step": 9122
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.852187156677246,
      "learning_rate": 2.821467912028077e-06,
      "loss": 0.0666,
      "step": 9123
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1521553993225098,
      "learning_rate": 2.8204226830262855e-06,
      "loss": 0.0757,
      "step": 9124
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.935603141784668,
      "learning_rate": 2.819377571604438e-06,
      "loss": 0.1812,
      "step": 9125
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.414778709411621,
      "learning_rate": 2.818332577818919e-06,
      "loss": 0.5395,
      "step": 9126
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.439516544342041,
      "learning_rate": 2.817287701726099e-06,
      "loss": 0.4293,
      "step": 9127
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.120477676391602,
      "learning_rate": 2.8162429433823433e-06,
      "loss": 0.894,
      "step": 9128
    },
    {
      "epoch": 0.65,
      "grad_norm": 15.915898323059082,
      "learning_rate": 2.8151983028440194e-06,
      "loss": 0.9635,
      "step": 9129
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.302842378616333,
      "learning_rate": 2.8141537801674747e-06,
      "loss": 0.099,
      "step": 9130
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.952369689941406,
      "learning_rate": 2.8131093754090577e-06,
      "loss": 0.5559,
      "step": 9131
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1732568740844727,
      "learning_rate": 2.8120650886251143e-06,
      "loss": 0.2018,
      "step": 9132
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.5694408416748047,
      "learning_rate": 2.8110209198719773e-06,
      "loss": 0.1182,
      "step": 9133
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.660039186477661,
      "learning_rate": 2.8099768692059747e-06,
      "loss": 0.6605,
      "step": 9134
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.60512638092041,
      "learning_rate": 2.808932936683432e-06,
      "loss": 0.1226,
      "step": 9135
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8465303182601929,
      "learning_rate": 2.807889122360663e-06,
      "loss": 0.0676,
      "step": 9136
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.631687641143799,
      "learning_rate": 2.806845426293977e-06,
      "loss": 0.1045,
      "step": 9137
    },
    {
      "epoch": 0.66,
      "grad_norm": 13.392822265625,
      "learning_rate": 2.8058018485396797e-06,
      "loss": 0.9502,
      "step": 9138
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.064052104949951,
      "learning_rate": 2.8047583891540685e-06,
      "loss": 0.767,
      "step": 9139
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.528100967407227,
      "learning_rate": 2.80371504819343e-06,
      "loss": 0.671,
      "step": 9140
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.371236562728882,
      "learning_rate": 2.8026718257140556e-06,
      "loss": 0.1253,
      "step": 9141
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.094754457473755,
      "learning_rate": 2.8016287217722165e-06,
      "loss": 0.0797,
      "step": 9142
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2329483032226562,
      "learning_rate": 2.800585736424185e-06,
      "loss": 0.0376,
      "step": 9143
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.2192535400390625,
      "learning_rate": 2.7995428697262295e-06,
      "loss": 0.1244,
      "step": 9144
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.544379472732544,
      "learning_rate": 2.798500121734607e-06,
      "loss": 0.1343,
      "step": 9145
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.11552095413208,
      "learning_rate": 2.7974574925055676e-06,
      "loss": 0.2062,
      "step": 9146
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.791800498962402,
      "learning_rate": 2.796414982095362e-06,
      "loss": 0.919,
      "step": 9147
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.511719822883606,
      "learning_rate": 2.795372590560227e-06,
      "loss": 0.0439,
      "step": 9148
    },
    {
      "epoch": 0.66,
      "grad_norm": 19.77895736694336,
      "learning_rate": 2.7943303179563928e-06,
      "loss": 0.9068,
      "step": 9149
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.104428291320801,
      "learning_rate": 2.793288164340092e-06,
      "loss": 0.0901,
      "step": 9150
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.7930145263671875,
      "learning_rate": 2.792246129767542e-06,
      "loss": 0.2141,
      "step": 9151
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.368261337280273,
      "learning_rate": 2.791204214294956e-06,
      "loss": 1.0315,
      "step": 9152
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.054614067077637,
      "learning_rate": 2.7901624179785402e-06,
      "loss": 0.4104,
      "step": 9153
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.665977478027344,
      "learning_rate": 2.789120740874501e-06,
      "loss": 0.9814,
      "step": 9154
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.869601011276245,
      "learning_rate": 2.7880791830390254e-06,
      "loss": 0.1933,
      "step": 9155
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.45268440246582,
      "learning_rate": 2.787037744528307e-06,
      "loss": 0.2604,
      "step": 9156
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.261605262756348,
      "learning_rate": 2.785996425398526e-06,
      "loss": 0.7882,
      "step": 9157
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.9602203369140625,
      "learning_rate": 2.7849552257058572e-06,
      "loss": 0.4619,
      "step": 9158
    },
    {
      "epoch": 0.66,
      "grad_norm": 12.505982398986816,
      "learning_rate": 2.7839141455064678e-06,
      "loss": 0.788,
      "step": 9159
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.2957210540771484,
      "learning_rate": 2.7828731848565237e-06,
      "loss": 0.1303,
      "step": 9160
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.7072691917419434,
      "learning_rate": 2.781832343812179e-06,
      "loss": 0.1139,
      "step": 9161
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.065800189971924,
      "learning_rate": 2.7807916224295817e-06,
      "loss": 0.73,
      "step": 9162
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.689976930618286,
      "learning_rate": 2.779751020764877e-06,
      "loss": 0.1083,
      "step": 9163
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.1285369396209717,
      "learning_rate": 2.778710538874202e-06,
      "loss": 0.1162,
      "step": 9164
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.654637336730957,
      "learning_rate": 2.7776701768136826e-06,
      "loss": 0.4869,
      "step": 9165
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4170167446136475,
      "learning_rate": 2.7766299346394495e-06,
      "loss": 0.1584,
      "step": 9166
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.297273874282837,
      "learning_rate": 2.775589812407613e-06,
      "loss": 0.1101,
      "step": 9167
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.84026837348938,
      "learning_rate": 2.774549810174285e-06,
      "loss": 0.1021,
      "step": 9168
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.028936862945557,
      "learning_rate": 2.7735099279955724e-06,
      "loss": 0.5974,
      "step": 9169
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.7262601852417,
      "learning_rate": 2.7724701659275716e-06,
      "loss": 0.8398,
      "step": 9170
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.706470489501953,
      "learning_rate": 2.7714305240263716e-06,
      "loss": 0.2798,
      "step": 9171
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.163212776184082,
      "learning_rate": 2.770391002348062e-06,
      "loss": 0.7912,
      "step": 9172
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.6952316761016846,
      "learning_rate": 2.7693516009487174e-06,
      "loss": 0.0923,
      "step": 9173
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.105241537094116,
      "learning_rate": 2.7683123198844084e-06,
      "loss": 0.1705,
      "step": 9174
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.309802293777466,
      "learning_rate": 2.767273159211205e-06,
      "loss": 0.1008,
      "step": 9175
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.911057472229004,
      "learning_rate": 2.7662341189851626e-06,
      "loss": 0.7846,
      "step": 9176
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.684891700744629,
      "learning_rate": 2.7651951992623327e-06,
      "loss": 0.132,
      "step": 9177
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.920194625854492,
      "learning_rate": 2.764156400098767e-06,
      "loss": 0.1642,
      "step": 9178
    },
    {
      "epoch": 0.66,
      "grad_norm": 15.619263648986816,
      "learning_rate": 2.763117721550498e-06,
      "loss": 0.4639,
      "step": 9179
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.19760799407959,
      "learning_rate": 2.762079163673559e-06,
      "loss": 0.1909,
      "step": 9180
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.263963460922241,
      "learning_rate": 2.7610407265239793e-06,
      "loss": 0.2588,
      "step": 9181
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.199031352996826,
      "learning_rate": 2.7600024101577784e-06,
      "loss": 0.9943,
      "step": 9182
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.2044177055358887,
      "learning_rate": 2.7589642146309665e-06,
      "loss": 0.1878,
      "step": 9183
    },
    {
      "epoch": 0.66,
      "grad_norm": 12.845037460327148,
      "learning_rate": 2.757926139999554e-06,
      "loss": 1.2509,
      "step": 9184
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.61952018737793,
      "learning_rate": 2.7568881863195397e-06,
      "loss": 0.6257,
      "step": 9185
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.8715596199035645,
      "learning_rate": 2.755850353646917e-06,
      "loss": 0.4318,
      "step": 9186
    },
    {
      "epoch": 0.66,
      "grad_norm": 17.808940887451172,
      "learning_rate": 2.754812642037672e-06,
      "loss": 0.704,
      "step": 9187
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.215420722961426,
      "learning_rate": 2.753775051547788e-06,
      "loss": 0.4988,
      "step": 9188
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.8398427963256836,
      "learning_rate": 2.752737582233237e-06,
      "loss": 0.1244,
      "step": 9189
    },
    {
      "epoch": 0.66,
      "grad_norm": 26.5367431640625,
      "learning_rate": 2.7517002341499876e-06,
      "loss": 0.9556,
      "step": 9190
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.159140586853027,
      "learning_rate": 2.7506630073539996e-06,
      "loss": 0.7271,
      "step": 9191
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.782032012939453,
      "learning_rate": 2.749625901901227e-06,
      "loss": 0.8066,
      "step": 9192
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.71157455444336,
      "learning_rate": 2.748588917847621e-06,
      "loss": 0.6105,
      "step": 9193
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.251971244812012,
      "learning_rate": 2.7475520552491196e-06,
      "loss": 0.6438,
      "step": 9194
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7816359996795654,
      "learning_rate": 2.74651531416166e-06,
      "loss": 0.0658,
      "step": 9195
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.569707870483398,
      "learning_rate": 2.7454786946411673e-06,
      "loss": 0.782,
      "step": 9196
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.653056621551514,
      "learning_rate": 2.7444421967435663e-06,
      "loss": 0.6005,
      "step": 9197
    },
    {
      "epoch": 0.66,
      "grad_norm": 16.87023162841797,
      "learning_rate": 2.7434058205247714e-06,
      "loss": 1.0415,
      "step": 9198
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.098155975341797,
      "learning_rate": 2.7423695660406895e-06,
      "loss": 0.7178,
      "step": 9199
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.417151689529419,
      "learning_rate": 2.7413334333472254e-06,
      "loss": 0.1077,
      "step": 9200
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.918888568878174,
      "learning_rate": 2.7402974225002733e-06,
      "loss": 0.2023,
      "step": 9201
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.961217403411865,
      "learning_rate": 2.7392615335557217e-06,
      "loss": 0.8051,
      "step": 9202
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.406717300415039,
      "learning_rate": 2.738225766569452e-06,
      "loss": 0.7101,
      "step": 9203
    },
    {
      "epoch": 0.66,
      "grad_norm": 14.46037483215332,
      "learning_rate": 2.7371901215973423e-06,
      "loss": 0.65,
      "step": 9204
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.389087438583374,
      "learning_rate": 2.7361545986952578e-06,
      "loss": 0.0955,
      "step": 9205
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.3327131271362305,
      "learning_rate": 2.7351191979190663e-06,
      "loss": 0.5753,
      "step": 9206
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.8565938472747803,
      "learning_rate": 2.7340839193246195e-06,
      "loss": 0.1765,
      "step": 9207
    },
    {
      "epoch": 0.66,
      "grad_norm": 12.08330249786377,
      "learning_rate": 2.7330487629677673e-06,
      "loss": 0.692,
      "step": 9208
    },
    {
      "epoch": 0.66,
      "grad_norm": 18.07373809814453,
      "learning_rate": 2.732013728904356e-06,
      "loss": 0.5251,
      "step": 9209
    },
    {
      "epoch": 0.66,
      "grad_norm": 12.148280143737793,
      "learning_rate": 2.730978817190218e-06,
      "loss": 0.6576,
      "step": 9210
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0867557525634766,
      "learning_rate": 2.729944027881183e-06,
      "loss": 0.0777,
      "step": 9211
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4797821044921875,
      "learning_rate": 2.7289093610330764e-06,
      "loss": 0.1531,
      "step": 9212
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.843238830566406,
      "learning_rate": 2.7278748167017133e-06,
      "loss": 0.9118,
      "step": 9213
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.502583980560303,
      "learning_rate": 2.7268403949429036e-06,
      "loss": 0.8048,
      "step": 9214
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.418937683105469,
      "learning_rate": 2.72580609581245e-06,
      "loss": 1.0321,
      "step": 9215
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.2437620162963867,
      "learning_rate": 2.72477191936615e-06,
      "loss": 0.1158,
      "step": 9216
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.590580940246582,
      "learning_rate": 2.7237378656597903e-06,
      "loss": 0.077,
      "step": 9217
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.011277914047241,
      "learning_rate": 2.722703934749158e-06,
      "loss": 0.081,
      "step": 9218
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.667556762695312,
      "learning_rate": 2.721670126690029e-06,
      "loss": 0.6591,
      "step": 9219
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.854654312133789,
      "learning_rate": 2.720636441538171e-06,
      "loss": 0.7231,
      "step": 9220
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.284522294998169,
      "learning_rate": 2.719602879349351e-06,
      "loss": 0.0967,
      "step": 9221
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1949965953826904,
      "learning_rate": 2.7185694401793234e-06,
      "loss": 0.1251,
      "step": 9222
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.651202917098999,
      "learning_rate": 2.717536124083839e-06,
      "loss": 0.1267,
      "step": 9223
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.206369400024414,
      "learning_rate": 2.716502931118639e-06,
      "loss": 0.792,
      "step": 9224
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.1778831481933594,
      "learning_rate": 2.715469861339465e-06,
      "loss": 0.0909,
      "step": 9225
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.0331904888153076,
      "learning_rate": 2.7144369148020437e-06,
      "loss": 0.1577,
      "step": 9226
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.997926712036133,
      "learning_rate": 2.7134040915621e-06,
      "loss": 0.8452,
      "step": 9227
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.303497791290283,
      "learning_rate": 2.71237139167535e-06,
      "loss": 0.0933,
      "step": 9228
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.074503898620605,
      "learning_rate": 2.7113388151975027e-06,
      "loss": 0.8292,
      "step": 9229
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5819547176361084,
      "learning_rate": 2.710306362184265e-06,
      "loss": 0.1687,
      "step": 9230
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.438642501831055,
      "learning_rate": 2.7092740326913325e-06,
      "loss": 0.4068,
      "step": 9231
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0508337020874023,
      "learning_rate": 2.708241826774395e-06,
      "loss": 0.0872,
      "step": 9232
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.9524102210998535,
      "learning_rate": 2.707209744489133e-06,
      "loss": 0.6772,
      "step": 9233
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.367501258850098,
      "learning_rate": 2.7061777858912297e-06,
      "loss": 0.4973,
      "step": 9234
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.8350164890289307,
      "learning_rate": 2.705145951036352e-06,
      "loss": 0.146,
      "step": 9235
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.916705369949341,
      "learning_rate": 2.704114239980162e-06,
      "loss": 0.1095,
      "step": 9236
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.592519760131836,
      "learning_rate": 2.7030826527783205e-06,
      "loss": 0.5985,
      "step": 9237
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0125527381896973,
      "learning_rate": 2.7020511894864775e-06,
      "loss": 0.0965,
      "step": 9238
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.8762571811676025,
      "learning_rate": 2.7010198501602703e-06,
      "loss": 0.1935,
      "step": 9239
    },
    {
      "epoch": 0.66,
      "grad_norm": 14.991008758544922,
      "learning_rate": 2.6999886348553415e-06,
      "loss": 0.836,
      "step": 9240
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.09496784210205,
      "learning_rate": 2.698957543627321e-06,
      "loss": 0.8604,
      "step": 9241
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.828381299972534,
      "learning_rate": 2.6979265765318286e-06,
      "loss": 0.2132,
      "step": 9242
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4020743370056152,
      "learning_rate": 2.6968957336244857e-06,
      "loss": 0.1296,
      "step": 9243
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.8774373531341553,
      "learning_rate": 2.6958650149609e-06,
      "loss": 0.1451,
      "step": 9244
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.9186313152313232,
      "learning_rate": 2.6948344205966732e-06,
      "loss": 0.2732,
      "step": 9245
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.532194137573242,
      "learning_rate": 2.693803950587406e-06,
      "loss": 0.661,
      "step": 9246
    },
    {
      "epoch": 0.66,
      "grad_norm": 29.195716857910156,
      "learning_rate": 2.692773604988686e-06,
      "loss": 0.4825,
      "step": 9247
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2448952198028564,
      "learning_rate": 2.691743383856095e-06,
      "loss": 0.06,
      "step": 9248
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3297812938690186,
      "learning_rate": 2.690713287245213e-06,
      "loss": 0.1349,
      "step": 9249
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7009695768356323,
      "learning_rate": 2.6896833152116096e-06,
      "loss": 0.0492,
      "step": 9250
    },
    {
      "epoch": 0.66,
      "grad_norm": 14.79141902923584,
      "learning_rate": 2.6886534678108435e-06,
      "loss": 1.0904,
      "step": 9251
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.0786218643188477,
      "learning_rate": 2.6876237450984747e-06,
      "loss": 0.103,
      "step": 9252
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5130698680877686,
      "learning_rate": 2.686594147130053e-06,
      "loss": 0.1228,
      "step": 9253
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.563156843185425,
      "learning_rate": 2.6855646739611175e-06,
      "loss": 0.5584,
      "step": 9254
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.6394567489624023,
      "learning_rate": 2.6845353256472094e-06,
      "loss": 0.1856,
      "step": 9255
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.706076145172119,
      "learning_rate": 2.6835061022438567e-06,
      "loss": 0.5953,
      "step": 9256
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.294682502746582,
      "learning_rate": 2.6824770038065785e-06,
      "loss": 0.0923,
      "step": 9257
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5476737022399902,
      "learning_rate": 2.681448030390896e-06,
      "loss": 0.1134,
      "step": 9258
    },
    {
      "epoch": 0.66,
      "grad_norm": 14.878023147583008,
      "learning_rate": 2.6804191820523163e-06,
      "loss": 0.6324,
      "step": 9259
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.3954567909240723,
      "learning_rate": 2.6793904588463414e-06,
      "loss": 0.1432,
      "step": 9260
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.436196327209473,
      "learning_rate": 2.6783618608284658e-06,
      "loss": 0.6582,
      "step": 9261
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.5636332035064697,
      "learning_rate": 2.677333388054183e-06,
      "loss": 0.1974,
      "step": 9262
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.4351654052734375,
      "learning_rate": 2.676305040578968e-06,
      "loss": 0.6271,
      "step": 9263
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.7265515327453613,
      "learning_rate": 2.6752768184583033e-06,
      "loss": 0.1537,
      "step": 9264
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.873420715332031,
      "learning_rate": 2.674248721747654e-06,
      "loss": 1.0158,
      "step": 9265
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.460684776306152,
      "learning_rate": 2.673220750502483e-06,
      "loss": 0.8491,
      "step": 9266
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.9795007705688477,
      "learning_rate": 2.6721929047782434e-06,
      "loss": 0.2608,
      "step": 9267
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5848617553710938,
      "learning_rate": 2.6711651846303876e-06,
      "loss": 0.1408,
      "step": 9268
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.889824867248535,
      "learning_rate": 2.6701375901143544e-06,
      "loss": 0.6557,
      "step": 9269
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0479564666748047,
      "learning_rate": 2.6691101212855773e-06,
      "loss": 0.1498,
      "step": 9270
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.914388179779053,
      "learning_rate": 2.6680827781994885e-06,
      "loss": 0.4799,
      "step": 9271
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.9368152618408203,
      "learning_rate": 2.6670555609115065e-06,
      "loss": 0.1501,
      "step": 9272
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.518962860107422,
      "learning_rate": 2.666028469477045e-06,
      "loss": 0.7324,
      "step": 9273
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.294579029083252,
      "learning_rate": 2.6650015039515165e-06,
      "loss": 0.1067,
      "step": 9274
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.326509475708008,
      "learning_rate": 2.6639746643903164e-06,
      "loss": 0.0718,
      "step": 9275
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.512164115905762,
      "learning_rate": 2.6629479508488393e-06,
      "loss": 0.945,
      "step": 9276
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.867459297180176,
      "learning_rate": 2.661921363382476e-06,
      "loss": 0.1129,
      "step": 9277
    },
    {
      "epoch": 0.67,
      "grad_norm": 25.612789154052734,
      "learning_rate": 2.6608949020466048e-06,
      "loss": 1.031,
      "step": 9278
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3215816020965576,
      "learning_rate": 2.6598685668965973e-06,
      "loss": 0.1304,
      "step": 9279
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.3976426124572754,
      "learning_rate": 2.658842357987825e-06,
      "loss": 0.2399,
      "step": 9280
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.119419813156128,
      "learning_rate": 2.657816275375645e-06,
      "loss": 0.1885,
      "step": 9281
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0222787857055664,
      "learning_rate": 2.656790319115411e-06,
      "loss": 0.0842,
      "step": 9282
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.7370407581329346,
      "learning_rate": 2.655764489262471e-06,
      "loss": 0.7701,
      "step": 9283
    },
    {
      "epoch": 0.67,
      "grad_norm": 14.486448287963867,
      "learning_rate": 2.654738785872163e-06,
      "loss": 0.8314,
      "step": 9284
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.3531057834625244,
      "learning_rate": 2.6537132089998184e-06,
      "loss": 0.059,
      "step": 9285
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.112443923950195,
      "learning_rate": 2.652687758700769e-06,
      "loss": 0.8667,
      "step": 9286
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.756590843200684,
      "learning_rate": 2.6516624350303278e-06,
      "loss": 0.8178,
      "step": 9287
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.949342727661133,
      "learning_rate": 2.650637238043807e-06,
      "loss": 0.744,
      "step": 9288
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.163630962371826,
      "learning_rate": 2.649612167796517e-06,
      "loss": 0.2234,
      "step": 9289
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.265108585357666,
      "learning_rate": 2.6485872243437537e-06,
      "loss": 0.1572,
      "step": 9290
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.136354923248291,
      "learning_rate": 2.6475624077408074e-06,
      "loss": 0.1435,
      "step": 9291
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.515719175338745,
      "learning_rate": 2.6465377180429664e-06,
      "loss": 0.0773,
      "step": 9292
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.507659912109375,
      "learning_rate": 2.645513155305507e-06,
      "loss": 0.3313,
      "step": 9293
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.689244031906128,
      "learning_rate": 2.644488719583701e-06,
      "loss": 0.129,
      "step": 9294
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.067826509475708,
      "learning_rate": 2.64346441093281e-06,
      "loss": 0.0392,
      "step": 9295
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.1200361251831055,
      "learning_rate": 2.6424402294080963e-06,
      "loss": 0.1667,
      "step": 9296
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.46932053565979,
      "learning_rate": 2.6414161750648088e-06,
      "loss": 0.6575,
      "step": 9297
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9050260782241821,
      "learning_rate": 2.6403922479581905e-06,
      "loss": 0.0551,
      "step": 9298
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.134798526763916,
      "learning_rate": 2.639368448143479e-06,
      "loss": 0.9718,
      "step": 9299
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.974851608276367,
      "learning_rate": 2.638344775675902e-06,
      "loss": 0.1383,
      "step": 9300
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.6212620735168457,
      "learning_rate": 2.6373212306106873e-06,
      "loss": 0.0967,
      "step": 9301
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.995859980583191,
      "learning_rate": 2.6362978130030485e-06,
      "loss": 0.0797,
      "step": 9302
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.481893539428711,
      "learning_rate": 2.635274522908196e-06,
      "loss": 1.0358,
      "step": 9303
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.7511532306671143,
      "learning_rate": 2.63425136038133e-06,
      "loss": 0.1047,
      "step": 9304
    },
    {
      "epoch": 0.67,
      "grad_norm": 19.1171817779541,
      "learning_rate": 2.63322832547765e-06,
      "loss": 1.5729,
      "step": 9305
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.770202398300171,
      "learning_rate": 2.6322054182523438e-06,
      "loss": 0.1361,
      "step": 9306
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.136317253112793,
      "learning_rate": 2.63118263876059e-06,
      "loss": 0.355,
      "step": 9307
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.350421667098999,
      "learning_rate": 2.630159987057569e-06,
      "loss": 0.2029,
      "step": 9308
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9658836126327515,
      "learning_rate": 2.629137463198446e-06,
      "loss": 0.0464,
      "step": 9309
    },
    {
      "epoch": 0.67,
      "grad_norm": 16.112791061401367,
      "learning_rate": 2.6281150672383826e-06,
      "loss": 1.1845,
      "step": 9310
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.9274802207946777,
      "learning_rate": 2.627092799232533e-06,
      "loss": 0.1618,
      "step": 9311
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.956606149673462,
      "learning_rate": 2.6260706592360453e-06,
      "loss": 0.1305,
      "step": 9312
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.747638702392578,
      "learning_rate": 2.6250486473040575e-06,
      "loss": 0.1814,
      "step": 9313
    },
    {
      "epoch": 0.67,
      "grad_norm": 17.867048263549805,
      "learning_rate": 2.6240267634917077e-06,
      "loss": 0.8182,
      "step": 9314
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.761255264282227,
      "learning_rate": 2.6230050078541203e-06,
      "loss": 1.0097,
      "step": 9315
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.12214183807373,
      "learning_rate": 2.6219833804464134e-06,
      "loss": 0.8181,
      "step": 9316
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.948415517807007,
      "learning_rate": 2.6209618813237035e-06,
      "loss": 0.1683,
      "step": 9317
    },
    {
      "epoch": 0.67,
      "grad_norm": 21.56132698059082,
      "learning_rate": 2.619940510541096e-06,
      "loss": 0.9063,
      "step": 9318
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.018516540527344,
      "learning_rate": 2.6189192681536857e-06,
      "loss": 0.8407,
      "step": 9319
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.626800060272217,
      "learning_rate": 2.6178981542165705e-06,
      "loss": 0.5368,
      "step": 9320
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.314023971557617,
      "learning_rate": 2.6168771687848337e-06,
      "loss": 0.0778,
      "step": 9321
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.9455366134643555,
      "learning_rate": 2.615856311913553e-06,
      "loss": 0.697,
      "step": 9322
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.0398261547088623,
      "learning_rate": 2.614835583657799e-06,
      "loss": 0.1376,
      "step": 9323
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.340080261230469,
      "learning_rate": 2.613814984072637e-06,
      "loss": 0.5438,
      "step": 9324
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.081210136413574,
      "learning_rate": 2.612794513213123e-06,
      "loss": 0.4091,
      "step": 9325
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.5063133239746094,
      "learning_rate": 2.6117741711343103e-06,
      "loss": 0.1357,
      "step": 9326
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.819568157196045,
      "learning_rate": 2.610753957891241e-06,
      "loss": 0.1594,
      "step": 9327
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.8218026161193848,
      "learning_rate": 2.609733873538951e-06,
      "loss": 0.1855,
      "step": 9328
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.696104526519775,
      "learning_rate": 2.6087139181324717e-06,
      "loss": 0.5739,
      "step": 9329
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.67897891998291,
      "learning_rate": 2.6076940917268257e-06,
      "loss": 0.6374,
      "step": 9330
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8329944610595703,
      "learning_rate": 2.606674394377028e-06,
      "loss": 0.1503,
      "step": 9331
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.229259967803955,
      "learning_rate": 2.6056548261380856e-06,
      "loss": 0.231,
      "step": 9332
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8771729469299316,
      "learning_rate": 2.6046353870650036e-06,
      "loss": 0.7907,
      "step": 9333
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.5102298259735107,
      "learning_rate": 2.603616077212776e-06,
      "loss": 0.1393,
      "step": 9334
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2447423934936523,
      "learning_rate": 2.6025968966363903e-06,
      "loss": 0.0591,
      "step": 9335
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9529515504837036,
      "learning_rate": 2.6015778453908267e-06,
      "loss": 0.0703,
      "step": 9336
    },
    {
      "epoch": 0.67,
      "grad_norm": 34.01072692871094,
      "learning_rate": 2.600558923531059e-06,
      "loss": 0.7665,
      "step": 9337
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.5154547691345215,
      "learning_rate": 2.599540131112056e-06,
      "loss": 0.9496,
      "step": 9338
    },
    {
      "epoch": 0.67,
      "grad_norm": 15.05038833618164,
      "learning_rate": 2.598521468188777e-06,
      "loss": 1.2661,
      "step": 9339
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.733760118484497,
      "learning_rate": 2.597502934816175e-06,
      "loss": 0.1545,
      "step": 9340
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.153167486190796,
      "learning_rate": 2.596484531049194e-06,
      "loss": 0.2208,
      "step": 9341
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.334203720092773,
      "learning_rate": 2.5954662569427756e-06,
      "loss": 0.2724,
      "step": 9342
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8931807279586792,
      "learning_rate": 2.594448112551852e-06,
      "loss": 0.0632,
      "step": 9343
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3536694049835205,
      "learning_rate": 2.5934300979313443e-06,
      "loss": 0.0862,
      "step": 9344
    },
    {
      "epoch": 0.67,
      "grad_norm": 14.916830062866211,
      "learning_rate": 2.5924122131361763e-06,
      "loss": 0.724,
      "step": 9345
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2776453495025635,
      "learning_rate": 2.591394458221258e-06,
      "loss": 0.0666,
      "step": 9346
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.4945638179779053,
      "learning_rate": 2.590376833241487e-06,
      "loss": 0.1539,
      "step": 9347
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.107781887054443,
      "learning_rate": 2.589359338251768e-06,
      "loss": 0.14,
      "step": 9348
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.636155128479004,
      "learning_rate": 2.588341973306987e-06,
      "loss": 0.15,
      "step": 9349
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.022550582885742,
      "learning_rate": 2.587324738462026e-06,
      "loss": 0.6624,
      "step": 9350
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.7743313312530518,
      "learning_rate": 2.5863076337717654e-06,
      "loss": 0.1832,
      "step": 9351
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.767611503601074,
      "learning_rate": 2.5852906592910707e-06,
      "loss": 0.8224,
      "step": 9352
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.5718283653259277,
      "learning_rate": 2.5842738150748044e-06,
      "loss": 0.1152,
      "step": 9353
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.307806491851807,
      "learning_rate": 2.5832571011778227e-06,
      "loss": 0.6322,
      "step": 9354
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.059084415435791,
      "learning_rate": 2.5822405176549724e-06,
      "loss": 0.0917,
      "step": 9355
    },
    {
      "epoch": 0.67,
      "grad_norm": 47.27279281616211,
      "learning_rate": 2.5812240645610933e-06,
      "loss": 0.7196,
      "step": 9356
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.879554033279419,
      "learning_rate": 2.5802077419510205e-06,
      "loss": 0.7551,
      "step": 9357
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.4345667362213135,
      "learning_rate": 2.5791915498795844e-06,
      "loss": 0.1869,
      "step": 9358
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.6694490909576416,
      "learning_rate": 2.5781754884015963e-06,
      "loss": 0.1165,
      "step": 9359
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.107860565185547,
      "learning_rate": 2.5771595575718754e-06,
      "loss": 0.186,
      "step": 9360
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.860440492630005,
      "learning_rate": 2.576143757445226e-06,
      "loss": 0.183,
      "step": 9361
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.8915202617645264,
      "learning_rate": 2.575128088076444e-06,
      "loss": 0.3306,
      "step": 9362
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.657096862792969,
      "learning_rate": 2.5741125495203244e-06,
      "loss": 0.7068,
      "step": 9363
    },
    {
      "epoch": 0.67,
      "grad_norm": 19.943445205688477,
      "learning_rate": 2.57309714183165e-06,
      "loss": 0.5418,
      "step": 9364
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.212137222290039,
      "learning_rate": 2.5720818650651975e-06,
      "loss": 0.1877,
      "step": 9365
    },
    {
      "epoch": 0.67,
      "grad_norm": 14.298262596130371,
      "learning_rate": 2.57106671927574e-06,
      "loss": 0.8586,
      "step": 9366
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.751298904418945,
      "learning_rate": 2.5700517045180385e-06,
      "loss": 0.6363,
      "step": 9367
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.897047996520996,
      "learning_rate": 2.5690368208468496e-06,
      "loss": 0.1849,
      "step": 9368
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.1196179389953613,
      "learning_rate": 2.568022068316921e-06,
      "loss": 0.1468,
      "step": 9369
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.277327537536621,
      "learning_rate": 2.567007446982998e-06,
      "loss": 0.2568,
      "step": 9370
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.226667642593384,
      "learning_rate": 2.5659929568998145e-06,
      "loss": 0.0633,
      "step": 9371
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.206178665161133,
      "learning_rate": 2.5649785981220965e-06,
      "loss": 0.1505,
      "step": 9372
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.448838710784912,
      "learning_rate": 2.5639643707045667e-06,
      "loss": 0.6157,
      "step": 9373
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.534341335296631,
      "learning_rate": 2.5629502747019367e-06,
      "loss": 0.3544,
      "step": 9374
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.113828659057617,
      "learning_rate": 2.5619363101689164e-06,
      "loss": 0.828,
      "step": 9375
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.034732341766357,
      "learning_rate": 2.5609224771602036e-06,
      "loss": 0.2516,
      "step": 9376
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.172152042388916,
      "learning_rate": 2.55990877573049e-06,
      "loss": 0.9374,
      "step": 9377
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.2158448696136475,
      "learning_rate": 2.5588952059344608e-06,
      "loss": 0.14,
      "step": 9378
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.880686640739441,
      "learning_rate": 2.5578817678267966e-06,
      "loss": 0.0807,
      "step": 9379
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.826202392578125,
      "learning_rate": 2.556868461462167e-06,
      "loss": 0.9328,
      "step": 9380
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4870405197143555,
      "learning_rate": 2.555855286895235e-06,
      "loss": 0.1571,
      "step": 9381
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.5417144298553467,
      "learning_rate": 2.55484224418066e-06,
      "loss": 0.105,
      "step": 9382
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.815323829650879,
      "learning_rate": 2.5538293333730918e-06,
      "loss": 0.8094,
      "step": 9383
    },
    {
      "epoch": 0.67,
      "grad_norm": 16.918262481689453,
      "learning_rate": 2.552816554527169e-06,
      "loss": 1.169,
      "step": 9384
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.7472455501556396,
      "learning_rate": 2.551803907697531e-06,
      "loss": 0.1534,
      "step": 9385
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8775248527526855,
      "learning_rate": 2.5507913929388063e-06,
      "loss": 0.058,
      "step": 9386
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.1626603603363037,
      "learning_rate": 2.549779010305612e-06,
      "loss": 0.2006,
      "step": 9387
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3806090354919434,
      "learning_rate": 2.5487667598525686e-06,
      "loss": 0.0886,
      "step": 9388
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.340277194976807,
      "learning_rate": 2.547754641634279e-06,
      "loss": 0.9205,
      "step": 9389
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.162973165512085,
      "learning_rate": 2.5467426557053433e-06,
      "loss": 0.0365,
      "step": 9390
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2257323265075684,
      "learning_rate": 2.5457308021203564e-06,
      "loss": 0.0565,
      "step": 9391
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.9444782733917236,
      "learning_rate": 2.5447190809339035e-06,
      "loss": 0.8612,
      "step": 9392
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.789010047912598,
      "learning_rate": 2.54370749220056e-06,
      "loss": 0.7069,
      "step": 9393
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7197383642196655,
      "learning_rate": 2.542696035974902e-06,
      "loss": 0.0783,
      "step": 9394
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.261052131652832,
      "learning_rate": 2.5416847123114927e-06,
      "loss": 0.352,
      "step": 9395
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.196695804595947,
      "learning_rate": 2.5406735212648855e-06,
      "loss": 0.3291,
      "step": 9396
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.239174842834473,
      "learning_rate": 2.5396624628896336e-06,
      "loss": 0.8217,
      "step": 9397
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7701033353805542,
      "learning_rate": 2.5386515372402793e-06,
      "loss": 0.0453,
      "step": 9398
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.7760324478149414,
      "learning_rate": 2.537640744371356e-06,
      "loss": 0.1288,
      "step": 9399
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.341015338897705,
      "learning_rate": 2.5366300843373955e-06,
      "loss": 0.5522,
      "step": 9400
    },
    {
      "epoch": 0.67,
      "grad_norm": 17.846336364746094,
      "learning_rate": 2.5356195571929177e-06,
      "loss": 1.0282,
      "step": 9401
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4682776927948,
      "learning_rate": 2.5346091629924345e-06,
      "loss": 0.1408,
      "step": 9402
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.9099814891815186,
      "learning_rate": 2.5335989017904574e-06,
      "loss": 0.7382,
      "step": 9403
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.871941328048706,
      "learning_rate": 2.5325887736414834e-06,
      "loss": 0.1311,
      "step": 9404
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3723485469818115,
      "learning_rate": 2.531578778600006e-06,
      "loss": 0.0989,
      "step": 9405
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.2483551502227783,
      "learning_rate": 2.5305689167205073e-06,
      "loss": 0.218,
      "step": 9406
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8083720207214355,
      "learning_rate": 2.529559188057472e-06,
      "loss": 0.731,
      "step": 9407
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.2694642543792725,
      "learning_rate": 2.528549592665364e-06,
      "loss": 0.1872,
      "step": 9408
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.6404783725738525,
      "learning_rate": 2.5275401305986524e-06,
      "loss": 0.1235,
      "step": 9409
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.035982847213745,
      "learning_rate": 2.5265308019117917e-06,
      "loss": 0.0512,
      "step": 9410
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.56931209564209,
      "learning_rate": 2.5255216066592326e-06,
      "loss": 0.0931,
      "step": 9411
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0378360748291016,
      "learning_rate": 2.5245125448954133e-06,
      "loss": 0.0757,
      "step": 9412
    },
    {
      "epoch": 0.67,
      "grad_norm": 18.455059051513672,
      "learning_rate": 2.5235036166747745e-06,
      "loss": 0.8265,
      "step": 9413
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.105161666870117,
      "learning_rate": 2.5224948220517415e-06,
      "loss": 0.8044,
      "step": 9414
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.6612930297851562,
      "learning_rate": 2.521486161080733e-06,
      "loss": 0.2834,
      "step": 9415
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.894042491912842,
      "learning_rate": 2.520477633816166e-06,
      "loss": 0.1267,
      "step": 9416
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.09136700630188,
      "learning_rate": 2.5194692403124455e-06,
      "loss": 0.0553,
      "step": 9417
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.689977645874023,
      "learning_rate": 2.518460980623968e-06,
      "loss": 0.5374,
      "step": 9418
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.373429775238037,
      "learning_rate": 2.517452854805132e-06,
      "loss": 0.1066,
      "step": 9419
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.22845458984375,
      "learning_rate": 2.516444862910315e-06,
      "loss": 0.1085,
      "step": 9420
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.14715576171875,
      "learning_rate": 2.5154370049938946e-06,
      "loss": 0.0766,
      "step": 9421
    },
    {
      "epoch": 0.68,
      "grad_norm": 16.778486251831055,
      "learning_rate": 2.5144292811102462e-06,
      "loss": 0.6333,
      "step": 9422
    },
    {
      "epoch": 0.68,
      "grad_norm": 13.794326782226562,
      "learning_rate": 2.5134216913137287e-06,
      "loss": 1.0412,
      "step": 9423
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6555341482162476,
      "learning_rate": 2.5124142356586972e-06,
      "loss": 0.0606,
      "step": 9424
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.97712779045105,
      "learning_rate": 2.5114069141995035e-06,
      "loss": 0.229,
      "step": 9425
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.03629469871521,
      "learning_rate": 2.510399726990487e-06,
      "loss": 0.0765,
      "step": 9426
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3656656742095947,
      "learning_rate": 2.5093926740859797e-06,
      "loss": 0.1262,
      "step": 9427
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.7737512588500977,
      "learning_rate": 2.5083857555403122e-06,
      "loss": 0.0908,
      "step": 9428
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.002603530883789,
      "learning_rate": 2.5073789714078023e-06,
      "loss": 0.7791,
      "step": 9429
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.424408435821533,
      "learning_rate": 2.5063723217427595e-06,
      "loss": 0.8812,
      "step": 9430
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9442082643508911,
      "learning_rate": 2.5053658065994944e-06,
      "loss": 0.1,
      "step": 9431
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.346484899520874,
      "learning_rate": 2.5043594260323e-06,
      "loss": 0.1274,
      "step": 9432
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0331621170043945,
      "learning_rate": 2.5033531800954657e-06,
      "loss": 0.0587,
      "step": 9433
    },
    {
      "epoch": 0.68,
      "grad_norm": 32.695709228515625,
      "learning_rate": 2.5023470688432795e-06,
      "loss": 0.8489,
      "step": 9434
    },
    {
      "epoch": 0.68,
      "grad_norm": 13.530463218688965,
      "learning_rate": 2.5013410923300143e-06,
      "loss": 0.7965,
      "step": 9435
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.644500970840454,
      "learning_rate": 2.5003352506099376e-06,
      "loss": 0.0785,
      "step": 9436
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.128264427185059,
      "learning_rate": 2.499329543737314e-06,
      "loss": 0.6902,
      "step": 9437
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.016129970550537,
      "learning_rate": 2.498323971766396e-06,
      "loss": 0.3216,
      "step": 9438
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5440785884857178,
      "learning_rate": 2.4973185347514308e-06,
      "loss": 0.16,
      "step": 9439
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.137157440185547,
      "learning_rate": 2.4963132327466547e-06,
      "loss": 0.0993,
      "step": 9440
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.8394365310668945,
      "learning_rate": 2.4953080658063055e-06,
      "loss": 0.2301,
      "step": 9441
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.654844760894775,
      "learning_rate": 2.4943030339846054e-06,
      "loss": 0.8605,
      "step": 9442
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.524203062057495,
      "learning_rate": 2.4932981373357713e-06,
      "loss": 0.1636,
      "step": 9443
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.7631101608276367,
      "learning_rate": 2.492293375914015e-06,
      "loss": 0.2096,
      "step": 9444
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.037102222442627,
      "learning_rate": 2.4912887497735367e-06,
      "loss": 0.7374,
      "step": 9445
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.240595817565918,
      "learning_rate": 2.490284258968536e-06,
      "loss": 0.4813,
      "step": 9446
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.209182739257812,
      "learning_rate": 2.4892799035532e-06,
      "loss": 1.1199,
      "step": 9447
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4178245067596436,
      "learning_rate": 2.4882756835817096e-06,
      "loss": 0.1096,
      "step": 9448
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.8016324043273926,
      "learning_rate": 2.4872715991082367e-06,
      "loss": 0.1479,
      "step": 9449
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.665040969848633,
      "learning_rate": 2.486267650186952e-06,
      "loss": 0.1269,
      "step": 9450
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.218906402587891,
      "learning_rate": 2.4852638368720126e-06,
      "loss": 0.7199,
      "step": 9451
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.6204254627227783,
      "learning_rate": 2.4842601592175685e-06,
      "loss": 0.824,
      "step": 9452
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.561025619506836,
      "learning_rate": 2.4832566172777676e-06,
      "loss": 0.0996,
      "step": 9453
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3005197048187256,
      "learning_rate": 2.4822532111067466e-06,
      "loss": 0.0513,
      "step": 9454
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.972907066345215,
      "learning_rate": 2.4812499407586336e-06,
      "loss": 0.633,
      "step": 9455
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.779794692993164,
      "learning_rate": 2.4802468062875534e-06,
      "loss": 0.1659,
      "step": 9456
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.13484525680542,
      "learning_rate": 2.4792438077476193e-06,
      "loss": 0.686,
      "step": 9457
    },
    {
      "epoch": 0.68,
      "grad_norm": 16.94529914855957,
      "learning_rate": 2.478240945192939e-06,
      "loss": 1.0758,
      "step": 9458
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.985517740249634,
      "learning_rate": 2.4772382186776163e-06,
      "loss": 0.164,
      "step": 9459
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.7395472526550293,
      "learning_rate": 2.476235628255742e-06,
      "loss": 0.1265,
      "step": 9460
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.9371044635772705,
      "learning_rate": 2.475233173981401e-06,
      "loss": 0.1479,
      "step": 9461
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.279881477355957,
      "learning_rate": 2.474230855908675e-06,
      "loss": 0.0585,
      "step": 9462
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.453551769256592,
      "learning_rate": 2.4732286740916345e-06,
      "loss": 0.0923,
      "step": 9463
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.521754741668701,
      "learning_rate": 2.47222662858434e-06,
      "loss": 0.123,
      "step": 9464
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.050601959228516,
      "learning_rate": 2.4712247194408533e-06,
      "loss": 0.9061,
      "step": 9465
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.3767778873443604,
      "learning_rate": 2.4702229467152207e-06,
      "loss": 0.2028,
      "step": 9466
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.368367195129395,
      "learning_rate": 2.4692213104614843e-06,
      "loss": 1.303,
      "step": 9467
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.3049864768981934,
      "learning_rate": 2.468219810733679e-06,
      "loss": 0.1901,
      "step": 9468
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3858728408813477,
      "learning_rate": 2.4672184475858312e-06,
      "loss": 0.075,
      "step": 9469
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.7857866287231445,
      "learning_rate": 2.466217221071959e-06,
      "loss": 0.603,
      "step": 9470
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.270264625549316,
      "learning_rate": 2.46521613124608e-06,
      "loss": 0.1729,
      "step": 9471
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.187772274017334,
      "learning_rate": 2.464215178162195e-06,
      "loss": 0.1164,
      "step": 9472
    },
    {
      "epoch": 0.68,
      "grad_norm": 17.325660705566406,
      "learning_rate": 2.463214361874301e-06,
      "loss": 0.8267,
      "step": 9473
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0546276569366455,
      "learning_rate": 2.462213682436391e-06,
      "loss": 0.0943,
      "step": 9474
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8627831935882568,
      "learning_rate": 2.461213139902448e-06,
      "loss": 0.07,
      "step": 9475
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0971219539642334,
      "learning_rate": 2.460212734326446e-06,
      "loss": 0.1349,
      "step": 9476
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.414422988891602,
      "learning_rate": 2.459212465762351e-06,
      "loss": 0.7417,
      "step": 9477
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.109419345855713,
      "learning_rate": 2.4582123342641283e-06,
      "loss": 0.9104,
      "step": 9478
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.70099925994873,
      "learning_rate": 2.4572123398857287e-06,
      "loss": 0.8753,
      "step": 9479
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.846911430358887,
      "learning_rate": 2.456212482681099e-06,
      "loss": 0.4954,
      "step": 9480
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.7519705295562744,
      "learning_rate": 2.4552127627041768e-06,
      "loss": 0.206,
      "step": 9481
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.011909484863281,
      "learning_rate": 2.4542131800088914e-06,
      "loss": 0.7775,
      "step": 9482
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9830772876739502,
      "learning_rate": 2.453213734649171e-06,
      "loss": 0.0897,
      "step": 9483
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.059816598892212,
      "learning_rate": 2.4522144266789287e-06,
      "loss": 0.2107,
      "step": 9484
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.543814659118652,
      "learning_rate": 2.451215256152075e-06,
      "loss": 0.668,
      "step": 9485
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.5020904541015625,
      "learning_rate": 2.450216223122508e-06,
      "loss": 0.6984,
      "step": 9486
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.260650873184204,
      "learning_rate": 2.4492173276441267e-06,
      "loss": 0.1284,
      "step": 9487
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.652542114257812,
      "learning_rate": 2.448218569770815e-06,
      "loss": 0.7655,
      "step": 9488
    },
    {
      "epoch": 0.68,
      "grad_norm": 19.290843963623047,
      "learning_rate": 2.447219949556451e-06,
      "loss": 1.3468,
      "step": 9489
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.132090091705322,
      "learning_rate": 2.446221467054909e-06,
      "loss": 0.9373,
      "step": 9490
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.279829263687134,
      "learning_rate": 2.445223122320055e-06,
      "loss": 0.0975,
      "step": 9491
    },
    {
      "epoch": 0.68,
      "grad_norm": 22.30452537536621,
      "learning_rate": 2.444224915405739e-06,
      "loss": 0.7857,
      "step": 9492
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.994123935699463,
      "learning_rate": 2.4432268463658165e-06,
      "loss": 0.0755,
      "step": 9493
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.210505247116089,
      "learning_rate": 2.4422289152541277e-06,
      "loss": 0.0997,
      "step": 9494
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1282529830932617,
      "learning_rate": 2.4412311221245063e-06,
      "loss": 0.0995,
      "step": 9495
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2201075553894043,
      "learning_rate": 2.440233467030781e-06,
      "loss": 0.0842,
      "step": 9496
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.233883857727051,
      "learning_rate": 2.4392359500267714e-06,
      "loss": 1.0002,
      "step": 9497
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.260421752929688,
      "learning_rate": 2.4382385711662874e-06,
      "loss": 1.0484,
      "step": 9498
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5931954383850098,
      "learning_rate": 2.437241330503138e-06,
      "loss": 0.1237,
      "step": 9499
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.1313533782958984,
      "learning_rate": 2.4362442280911174e-06,
      "loss": 0.2231,
      "step": 9500
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.129202365875244,
      "learning_rate": 2.435247263984015e-06,
      "loss": 0.1229,
      "step": 9501
    },
    {
      "epoch": 0.68,
      "grad_norm": 13.045684814453125,
      "learning_rate": 2.4342504382356163e-06,
      "loss": 0.7649,
      "step": 9502
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.8492226600646973,
      "learning_rate": 2.4332537508996966e-06,
      "loss": 0.7084,
      "step": 9503
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.393817901611328,
      "learning_rate": 2.432257202030018e-06,
      "loss": 0.8441,
      "step": 9504
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1908600330352783,
      "learning_rate": 2.431260791680345e-06,
      "loss": 0.1073,
      "step": 9505
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.125667572021484,
      "learning_rate": 2.4302645199044296e-06,
      "loss": 0.6459,
      "step": 9506
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0747671127319336,
      "learning_rate": 2.4292683867560142e-06,
      "loss": 0.071,
      "step": 9507
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.557525634765625,
      "learning_rate": 2.4282723922888407e-06,
      "loss": 0.4188,
      "step": 9508
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3658957481384277,
      "learning_rate": 2.427276536556637e-06,
      "loss": 0.099,
      "step": 9509
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.050227165222168,
      "learning_rate": 2.426280819613124e-06,
      "loss": 0.8558,
      "step": 9510
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.568176746368408,
      "learning_rate": 2.4252852415120204e-06,
      "loss": 0.7425,
      "step": 9511
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4873409271240234,
      "learning_rate": 2.424289802307032e-06,
      "loss": 0.1272,
      "step": 9512
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.190971851348877,
      "learning_rate": 2.42329450205186e-06,
      "loss": 0.8557,
      "step": 9513
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1105287075042725,
      "learning_rate": 2.422299340800194e-06,
      "loss": 0.0809,
      "step": 9514
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.9874954223632812,
      "learning_rate": 2.421304318605725e-06,
      "loss": 0.1673,
      "step": 9515
    },
    {
      "epoch": 0.68,
      "grad_norm": 18.665708541870117,
      "learning_rate": 2.4203094355221234e-06,
      "loss": 0.6099,
      "step": 9516
    },
    {
      "epoch": 0.68,
      "grad_norm": 16.04480743408203,
      "learning_rate": 2.4193146916030653e-06,
      "loss": 0.7837,
      "step": 9517
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0964701175689697,
      "learning_rate": 2.4183200869022112e-06,
      "loss": 0.1107,
      "step": 9518
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.6614394187927246,
      "learning_rate": 2.4173256214732155e-06,
      "loss": 0.1197,
      "step": 9519
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.7919678688049316,
      "learning_rate": 2.4163312953697254e-06,
      "loss": 0.0952,
      "step": 9520
    },
    {
      "epoch": 0.68,
      "grad_norm": 20.445629119873047,
      "learning_rate": 2.4153371086453835e-06,
      "loss": 0.7848,
      "step": 9521
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.319083213806152,
      "learning_rate": 2.4143430613538215e-06,
      "loss": 0.6816,
      "step": 9522
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.6419248580932617,
      "learning_rate": 2.4133491535486624e-06,
      "loss": 0.8701,
      "step": 9523
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.6383588314056396,
      "learning_rate": 2.4123553852835267e-06,
      "loss": 0.1824,
      "step": 9524
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.375734806060791,
      "learning_rate": 2.411361756612024e-06,
      "loss": 0.1659,
      "step": 9525
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.128394603729248,
      "learning_rate": 2.4103682675877536e-06,
      "loss": 0.7119,
      "step": 9526
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.093585729598999,
      "learning_rate": 2.4093749182643167e-06,
      "loss": 0.1047,
      "step": 9527
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.9653844833374023,
      "learning_rate": 2.4083817086952953e-06,
      "loss": 0.1563,
      "step": 9528
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.944272041320801,
      "learning_rate": 2.407388638934269e-06,
      "loss": 0.1152,
      "step": 9529
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.278933048248291,
      "learning_rate": 2.4063957090348134e-06,
      "loss": 0.7217,
      "step": 9530
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.291522026062012,
      "learning_rate": 2.4054029190504925e-06,
      "loss": 0.6741,
      "step": 9531
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3377022743225098,
      "learning_rate": 2.404410269034861e-06,
      "loss": 0.0866,
      "step": 9532
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.074716567993164,
      "learning_rate": 2.4034177590414727e-06,
      "loss": 0.16,
      "step": 9533
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.848755836486816,
      "learning_rate": 2.402425389123867e-06,
      "loss": 0.3653,
      "step": 9534
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.162899971008301,
      "learning_rate": 2.4014331593355776e-06,
      "loss": 0.1078,
      "step": 9535
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.6968674659729004,
      "learning_rate": 2.4004410697301358e-06,
      "loss": 0.0843,
      "step": 9536
    },
    {
      "epoch": 0.68,
      "grad_norm": 13.258395195007324,
      "learning_rate": 2.3994491203610577e-06,
      "loss": 1.1861,
      "step": 9537
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7876322269439697,
      "learning_rate": 2.3984573112818543e-06,
      "loss": 0.0592,
      "step": 9538
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.9700772762298584,
      "learning_rate": 2.397465642546035e-06,
      "loss": 0.1299,
      "step": 9539
    },
    {
      "epoch": 0.68,
      "grad_norm": 21.400636672973633,
      "learning_rate": 2.3964741142070917e-06,
      "loss": 0.8276,
      "step": 9540
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.2970149517059326,
      "learning_rate": 2.395482726318513e-06,
      "loss": 0.129,
      "step": 9541
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.714239120483398,
      "learning_rate": 2.3944914789337847e-06,
      "loss": 0.7072,
      "step": 9542
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.654386520385742,
      "learning_rate": 2.393500372106378e-06,
      "loss": 0.1395,
      "step": 9543
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.8002054691314697,
      "learning_rate": 2.392509405889759e-06,
      "loss": 0.1252,
      "step": 9544
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.2286055088043213,
      "learning_rate": 2.3915185803373893e-06,
      "loss": 0.2777,
      "step": 9545
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.254727363586426,
      "learning_rate": 2.390527895502718e-06,
      "loss": 0.1638,
      "step": 9546
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.130724906921387,
      "learning_rate": 2.3895373514391906e-06,
      "loss": 0.7751,
      "step": 9547
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4742352962493896,
      "learning_rate": 2.388546948200239e-06,
      "loss": 0.0956,
      "step": 9548
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8613336086273193,
      "learning_rate": 2.3875566858392975e-06,
      "loss": 0.1348,
      "step": 9549
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.845985412597656,
      "learning_rate": 2.386566564409783e-06,
      "loss": 0.6842,
      "step": 9550
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4847137928009033,
      "learning_rate": 2.3855765839651106e-06,
      "loss": 0.8185,
      "step": 9551
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.636767387390137,
      "learning_rate": 2.384586744558685e-06,
      "loss": 0.618,
      "step": 9552
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.149126052856445,
      "learning_rate": 2.3835970462439036e-06,
      "loss": 0.6839,
      "step": 9553
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.240776062011719,
      "learning_rate": 2.3826074890741592e-06,
      "loss": 0.9595,
      "step": 9554
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.387967586517334,
      "learning_rate": 2.381618073102833e-06,
      "loss": 0.0906,
      "step": 9555
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1024982929229736,
      "learning_rate": 2.3806287983833017e-06,
      "loss": 0.1629,
      "step": 9556
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5769824981689453,
      "learning_rate": 2.379639664968929e-06,
      "loss": 0.1841,
      "step": 9557
    },
    {
      "epoch": 0.69,
      "grad_norm": 13.7152099609375,
      "learning_rate": 2.3786506729130803e-06,
      "loss": 0.7636,
      "step": 9558
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.605350971221924,
      "learning_rate": 2.377661822269105e-06,
      "loss": 0.677,
      "step": 9559
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1972546577453613,
      "learning_rate": 2.3766731130903465e-06,
      "loss": 0.1152,
      "step": 9560
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.014686584472656,
      "learning_rate": 2.375684545430146e-06,
      "loss": 0.555,
      "step": 9561
    },
    {
      "epoch": 0.69,
      "grad_norm": 17.615795135498047,
      "learning_rate": 2.3746961193418306e-06,
      "loss": 1.4312,
      "step": 9562
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.379612922668457,
      "learning_rate": 2.373707834878722e-06,
      "loss": 0.7544,
      "step": 9563
    },
    {
      "epoch": 0.69,
      "grad_norm": 16.18536949157715,
      "learning_rate": 2.372719692094134e-06,
      "loss": 0.997,
      "step": 9564
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.443786144256592,
      "learning_rate": 2.371731691041375e-06,
      "loss": 0.8255,
      "step": 9565
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.385246276855469,
      "learning_rate": 2.3707438317737403e-06,
      "loss": 0.1589,
      "step": 9566
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1803207397460938,
      "learning_rate": 2.369756114344526e-06,
      "loss": 0.1582,
      "step": 9567
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3551197052001953,
      "learning_rate": 2.3687685388070124e-06,
      "loss": 0.0944,
      "step": 9568
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.324213027954102,
      "learning_rate": 2.367781105214475e-06,
      "loss": 0.9386,
      "step": 9569
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.968013763427734,
      "learning_rate": 2.3667938136201855e-06,
      "loss": 0.2635,
      "step": 9570
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.501105308532715,
      "learning_rate": 2.3658066640774026e-06,
      "loss": 0.634,
      "step": 9571
    },
    {
      "epoch": 0.69,
      "grad_norm": 16.404632568359375,
      "learning_rate": 2.364819656639377e-06,
      "loss": 0.9921,
      "step": 9572
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0025222301483154,
      "learning_rate": 2.3638327913593577e-06,
      "loss": 0.0417,
      "step": 9573
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.855954885482788,
      "learning_rate": 2.3628460682905817e-06,
      "loss": 0.0618,
      "step": 9574
    },
    {
      "epoch": 0.69,
      "grad_norm": 28.924436569213867,
      "learning_rate": 2.3618594874862777e-06,
      "loss": 0.9109,
      "step": 9575
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9246666431427002,
      "learning_rate": 2.360873048999669e-06,
      "loss": 0.0653,
      "step": 9576
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.918282508850098,
      "learning_rate": 2.3598867528839694e-06,
      "loss": 0.697,
      "step": 9577
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.383556365966797,
      "learning_rate": 2.3589005991923838e-06,
      "loss": 0.663,
      "step": 9578
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5189409255981445,
      "learning_rate": 2.3579145879781155e-06,
      "loss": 0.1115,
      "step": 9579
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.863009452819824,
      "learning_rate": 2.3569287192943553e-06,
      "loss": 0.5967,
      "step": 9580
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.715080499649048,
      "learning_rate": 2.3559429931942833e-06,
      "loss": 0.6982,
      "step": 9581
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.13571310043335,
      "learning_rate": 2.354957409731081e-06,
      "loss": 0.6585,
      "step": 9582
    },
    {
      "epoch": 0.69,
      "grad_norm": 13.600754737854004,
      "learning_rate": 2.3539719689579144e-06,
      "loss": 1.1082,
      "step": 9583
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.10705280303955,
      "learning_rate": 2.3529866709279435e-06,
      "loss": 0.6924,
      "step": 9584
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6913771629333496,
      "learning_rate": 2.3520015156943216e-06,
      "loss": 0.1303,
      "step": 9585
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.710235357284546,
      "learning_rate": 2.3510165033101956e-06,
      "loss": 0.1651,
      "step": 9586
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.9914278984069824,
      "learning_rate": 2.3500316338287025e-06,
      "loss": 0.1958,
      "step": 9587
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.52204704284668,
      "learning_rate": 2.3490469073029726e-06,
      "loss": 0.4987,
      "step": 9588
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5784857273101807,
      "learning_rate": 2.3480623237861273e-06,
      "loss": 0.14,
      "step": 9589
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.043036460876465,
      "learning_rate": 2.3470778833312795e-06,
      "loss": 0.0662,
      "step": 9590
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.284151554107666,
      "learning_rate": 2.34609358599154e-06,
      "loss": 0.1399,
      "step": 9591
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.146300792694092,
      "learning_rate": 2.3451094318200056e-06,
      "loss": 0.8419,
      "step": 9592
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.562878370285034,
      "learning_rate": 2.344125420869769e-06,
      "loss": 0.1438,
      "step": 9593
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.539255142211914,
      "learning_rate": 2.34314155319391e-06,
      "loss": 1.0477,
      "step": 9594
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.994927406311035,
      "learning_rate": 2.34215782884551e-06,
      "loss": 0.6983,
      "step": 9595
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1070780754089355,
      "learning_rate": 2.3411742478776345e-06,
      "loss": 0.1308,
      "step": 9596
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.099777936935425,
      "learning_rate": 2.3401908103433422e-06,
      "loss": 0.1879,
      "step": 9597
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.700995445251465,
      "learning_rate": 2.3392075162956896e-06,
      "loss": 0.9066,
      "step": 9598
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.8478403091430664,
      "learning_rate": 2.33822436578772e-06,
      "loss": 0.7295,
      "step": 9599
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.8050944805145264,
      "learning_rate": 2.3372413588724703e-06,
      "loss": 0.871,
      "step": 9600
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6457998752593994,
      "learning_rate": 2.3362584956029704e-06,
      "loss": 0.1674,
      "step": 9601
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.5430002212524414,
      "learning_rate": 2.3352757760322416e-06,
      "loss": 0.1693,
      "step": 9602
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.329588413238525,
      "learning_rate": 2.334293200213296e-06,
      "loss": 0.2982,
      "step": 9603
    },
    {
      "epoch": 0.69,
      "grad_norm": 18.820037841796875,
      "learning_rate": 2.3333107681991445e-06,
      "loss": 0.6165,
      "step": 9604
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1381008625030518,
      "learning_rate": 2.3323284800427833e-06,
      "loss": 0.0996,
      "step": 9605
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5938034057617188,
      "learning_rate": 2.3313463357972004e-06,
      "loss": 0.176,
      "step": 9606
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.780510663986206,
      "learning_rate": 2.330364335515383e-06,
      "loss": 0.0934,
      "step": 9607
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.9143741130828857,
      "learning_rate": 2.3293824792503056e-06,
      "loss": 0.1997,
      "step": 9608
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.691349983215332,
      "learning_rate": 2.3284007670549317e-06,
      "loss": 0.1062,
      "step": 9609
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.637445449829102,
      "learning_rate": 2.3274191989822265e-06,
      "loss": 0.7758,
      "step": 9610
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.206343173980713,
      "learning_rate": 2.3264377750851385e-06,
      "loss": 0.8018,
      "step": 9611
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0819880962371826,
      "learning_rate": 2.325456495416613e-06,
      "loss": 0.2125,
      "step": 9612
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0557281970977783,
      "learning_rate": 2.324475360029586e-06,
      "loss": 0.1815,
      "step": 9613
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.1001482009887695,
      "learning_rate": 2.3234943689769858e-06,
      "loss": 0.7869,
      "step": 9614
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.917834281921387,
      "learning_rate": 2.3225135223117313e-06,
      "loss": 1.1629,
      "step": 9615
    },
    {
      "epoch": 0.69,
      "grad_norm": 13.358146667480469,
      "learning_rate": 2.3215328200867393e-06,
      "loss": 0.9777,
      "step": 9616
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9088619947433472,
      "learning_rate": 2.3205522623549133e-06,
      "loss": 0.1254,
      "step": 9617
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.057939052581787,
      "learning_rate": 2.3195718491691483e-06,
      "loss": 0.4879,
      "step": 9618
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.794651985168457,
      "learning_rate": 2.3185915805823378e-06,
      "loss": 0.6797,
      "step": 9619
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.95000171661377,
      "learning_rate": 2.3176114566473624e-06,
      "loss": 1.0203,
      "step": 9620
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.0421724319458,
      "learning_rate": 2.3166314774170952e-06,
      "loss": 0.7217,
      "step": 9621
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.948636054992676,
      "learning_rate": 2.315651642944401e-06,
      "loss": 0.8348,
      "step": 9622
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.6594130992889404,
      "learning_rate": 2.314671953282141e-06,
      "loss": 0.3091,
      "step": 9623
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.338694095611572,
      "learning_rate": 2.3136924084831653e-06,
      "loss": 0.8104,
      "step": 9624
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.533833503723145,
      "learning_rate": 2.3127130086003156e-06,
      "loss": 0.6164,
      "step": 9625
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1779682636260986,
      "learning_rate": 2.3117337536864276e-06,
      "loss": 0.1151,
      "step": 9626
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.352980375289917,
      "learning_rate": 2.310754643794326e-06,
      "loss": 0.1461,
      "step": 9627
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.5272908210754395,
      "learning_rate": 2.309775678976834e-06,
      "loss": 0.2804,
      "step": 9628
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0143322944641113,
      "learning_rate": 2.3087968592867604e-06,
      "loss": 0.239,
      "step": 9629
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.147721290588379,
      "learning_rate": 2.3078181847769097e-06,
      "loss": 0.6296,
      "step": 9630
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.322359800338745,
      "learning_rate": 2.306839655500076e-06,
      "loss": 0.1527,
      "step": 9631
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.324720621109009,
      "learning_rate": 2.3058612715090507e-06,
      "loss": 0.0864,
      "step": 9632
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.7755813598632812,
      "learning_rate": 2.3048830328566112e-06,
      "loss": 0.1771,
      "step": 9633
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.898867607116699,
      "learning_rate": 2.3039049395955293e-06,
      "loss": 0.4066,
      "step": 9634
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.804160118103027,
      "learning_rate": 2.3029269917785725e-06,
      "loss": 1.3501,
      "step": 9635
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.514103889465332,
      "learning_rate": 2.301949189458497e-06,
      "loss": 0.7602,
      "step": 9636
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.248680591583252,
      "learning_rate": 2.3009715326880467e-06,
      "loss": 0.0895,
      "step": 9637
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.804482936859131,
      "learning_rate": 2.2999940215199675e-06,
      "loss": 0.1918,
      "step": 9638
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.914468765258789,
      "learning_rate": 2.2990166560069903e-06,
      "loss": 0.8853,
      "step": 9639
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.724377393722534,
      "learning_rate": 2.2980394362018392e-06,
      "loss": 0.1329,
      "step": 9640
    },
    {
      "epoch": 0.69,
      "grad_norm": 15.596677780151367,
      "learning_rate": 2.297062362157235e-06,
      "loss": 0.7908,
      "step": 9641
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.684445858001709,
      "learning_rate": 2.2960854339258853e-06,
      "loss": 0.0817,
      "step": 9642
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.184500217437744,
      "learning_rate": 2.2951086515604897e-06,
      "loss": 0.951,
      "step": 9643
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.377453088760376,
      "learning_rate": 2.2941320151137455e-06,
      "loss": 0.1153,
      "step": 9644
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9705994129180908,
      "learning_rate": 2.293155524638337e-06,
      "loss": 0.1063,
      "step": 9645
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.036614418029785,
      "learning_rate": 2.29217918018694e-06,
      "loss": 0.1118,
      "step": 9646
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9287492036819458,
      "learning_rate": 2.2912029818122282e-06,
      "loss": 0.0538,
      "step": 9647
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.258143424987793,
      "learning_rate": 2.2902269295668644e-06,
      "loss": 0.6479,
      "step": 9648
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.643092632293701,
      "learning_rate": 2.2892510235034964e-06,
      "loss": 0.4185,
      "step": 9649
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.327453374862671,
      "learning_rate": 2.2882752636747773e-06,
      "loss": 0.1253,
      "step": 9650
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.459120750427246,
      "learning_rate": 2.287299650133343e-06,
      "loss": 0.7614,
      "step": 9651
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.790405750274658,
      "learning_rate": 2.2863241829318234e-06,
      "loss": 0.7678,
      "step": 9652
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.202311515808105,
      "learning_rate": 2.285348862122844e-06,
      "loss": 0.5877,
      "step": 9653
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.144561290740967,
      "learning_rate": 2.2843736877590174e-06,
      "loss": 0.8581,
      "step": 9654
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.738809585571289,
      "learning_rate": 2.2833986598929493e-06,
      "loss": 0.777,
      "step": 9655
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.14136028289795,
      "learning_rate": 2.2824237785772424e-06,
      "loss": 0.5137,
      "step": 9656
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7826743125915527,
      "learning_rate": 2.281449043864486e-06,
      "loss": 0.092,
      "step": 9657
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.8890135288238525,
      "learning_rate": 2.2804744558072637e-06,
      "loss": 0.7601,
      "step": 9658
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.018444299697876,
      "learning_rate": 2.2795000144581485e-06,
      "loss": 0.0941,
      "step": 9659
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.7823708057403564,
      "learning_rate": 2.278525719869714e-06,
      "loss": 0.1173,
      "step": 9660
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0265450477600098,
      "learning_rate": 2.277551572094512e-06,
      "loss": 0.2381,
      "step": 9661
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.137291431427002,
      "learning_rate": 2.2765775711850983e-06,
      "loss": 0.765,
      "step": 9662
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.611402988433838,
      "learning_rate": 2.2756037171940172e-06,
      "loss": 0.8754,
      "step": 9663
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.999025583267212,
      "learning_rate": 2.2746300101738027e-06,
      "loss": 0.7195,
      "step": 9664
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.95571517944336,
      "learning_rate": 2.273656450176981e-06,
      "loss": 0.798,
      "step": 9665
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.564729690551758,
      "learning_rate": 2.272683037256076e-06,
      "loss": 1.0473,
      "step": 9666
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.665846824645996,
      "learning_rate": 2.2717097714635983e-06,
      "loss": 0.7496,
      "step": 9667
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6536099910736084,
      "learning_rate": 2.2707366528520488e-06,
      "loss": 0.1013,
      "step": 9668
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.015840768814087,
      "learning_rate": 2.2697636814739278e-06,
      "loss": 0.1473,
      "step": 9669
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.501604080200195,
      "learning_rate": 2.2687908573817217e-06,
      "loss": 0.4674,
      "step": 9670
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.182279586791992,
      "learning_rate": 2.2678181806279093e-06,
      "loss": 0.1935,
      "step": 9671
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.4844155311584473,
      "learning_rate": 2.2668456512649677e-06,
      "loss": 0.2526,
      "step": 9672
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.663736343383789,
      "learning_rate": 2.2658732693453558e-06,
      "loss": 0.4622,
      "step": 9673
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.3543310165405273,
      "learning_rate": 2.2649010349215293e-06,
      "loss": 0.1722,
      "step": 9674
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.228262901306152,
      "learning_rate": 2.263928948045942e-06,
      "loss": 0.5789,
      "step": 9675
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.610218048095703,
      "learning_rate": 2.2629570087710305e-06,
      "loss": 0.143,
      "step": 9676
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.0950775146484375,
      "learning_rate": 2.2619852171492266e-06,
      "loss": 0.3382,
      "step": 9677
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2865312099456787,
      "learning_rate": 2.261013573232958e-06,
      "loss": 0.112,
      "step": 9678
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.326506614685059,
      "learning_rate": 2.2600420770746397e-06,
      "loss": 1.0076,
      "step": 9679
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.390698194503784,
      "learning_rate": 2.2590707287266783e-06,
      "loss": 0.1067,
      "step": 9680
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0076675415039062,
      "learning_rate": 2.2580995282414777e-06,
      "loss": 0.0933,
      "step": 9681
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.51638650894165,
      "learning_rate": 2.257128475671429e-06,
      "loss": 0.3782,
      "step": 9682
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.031432151794434,
      "learning_rate": 2.256157571068915e-06,
      "loss": 0.78,
      "step": 9683
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.052021026611328,
      "learning_rate": 2.255186814486317e-06,
      "loss": 0.314,
      "step": 9684
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.0814690589904785,
      "learning_rate": 2.254216205976e-06,
      "loss": 0.806,
      "step": 9685
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.986072063446045,
      "learning_rate": 2.2532457455903224e-06,
      "loss": 0.8497,
      "step": 9686
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.014735698699951,
      "learning_rate": 2.2522754333816423e-06,
      "loss": 0.1314,
      "step": 9687
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.681457042694092,
      "learning_rate": 2.251305269402301e-06,
      "loss": 0.6326,
      "step": 9688
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.129300594329834,
      "learning_rate": 2.2503352537046354e-06,
      "loss": 0.5326,
      "step": 9689
    },
    {
      "epoch": 0.69,
      "grad_norm": 24.87514305114746,
      "learning_rate": 2.249365386340976e-06,
      "loss": 1.1543,
      "step": 9690
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.305783271789551,
      "learning_rate": 2.2483956673636427e-06,
      "loss": 0.0854,
      "step": 9691
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1735782623291016,
      "learning_rate": 2.247426096824947e-06,
      "loss": 0.1061,
      "step": 9692
    },
    {
      "epoch": 0.7,
      "grad_norm": 20.402311325073242,
      "learning_rate": 2.246456674777193e-06,
      "loss": 1.3385,
      "step": 9693
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8928728103637695,
      "learning_rate": 2.24548740127268e-06,
      "loss": 0.1407,
      "step": 9694
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6519298553466797,
      "learning_rate": 2.2445182763636965e-06,
      "loss": 0.1465,
      "step": 9695
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5453219413757324,
      "learning_rate": 2.243549300102521e-06,
      "loss": 0.2535,
      "step": 9696
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.7875254154205322,
      "learning_rate": 2.2425804725414267e-06,
      "loss": 0.951,
      "step": 9697
    },
    {
      "epoch": 0.7,
      "grad_norm": 18.009130477905273,
      "learning_rate": 2.2416117937326777e-06,
      "loss": 1.118,
      "step": 9698
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.4343695640563965,
      "learning_rate": 2.2406432637285323e-06,
      "loss": 0.7078,
      "step": 9699
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.358996868133545,
      "learning_rate": 2.2396748825812387e-06,
      "loss": 0.2538,
      "step": 9700
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.813507080078125,
      "learning_rate": 2.2387066503430365e-06,
      "loss": 0.0486,
      "step": 9701
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.157484769821167,
      "learning_rate": 2.2377385670661567e-06,
      "loss": 0.1525,
      "step": 9702
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.773334264755249,
      "learning_rate": 2.2367706328028273e-06,
      "loss": 0.1568,
      "step": 9703
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.6100149154663086,
      "learning_rate": 2.2358028476052627e-06,
      "loss": 0.8557,
      "step": 9704
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6849570274353027,
      "learning_rate": 2.23483521152567e-06,
      "loss": 0.1521,
      "step": 9705
    },
    {
      "epoch": 0.7,
      "grad_norm": 26.3109130859375,
      "learning_rate": 2.233867724616252e-06,
      "loss": 0.409,
      "step": 9706
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1563546657562256,
      "learning_rate": 2.2329003869292004e-06,
      "loss": 0.0718,
      "step": 9707
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9377151727676392,
      "learning_rate": 2.231933198516699e-06,
      "loss": 0.1059,
      "step": 9708
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.2713041305542,
      "learning_rate": 2.230966159430924e-06,
      "loss": 0.5543,
      "step": 9709
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9360787868499756,
      "learning_rate": 2.229999269724043e-06,
      "loss": 0.0893,
      "step": 9710
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.7310452461242676,
      "learning_rate": 2.229032529448214e-06,
      "loss": 0.1257,
      "step": 9711
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8296587467193604,
      "learning_rate": 2.228065938655594e-06,
      "loss": 0.0965,
      "step": 9712
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.624895453453064,
      "learning_rate": 2.227099497398324e-06,
      "loss": 0.0432,
      "step": 9713
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.011509895324707,
      "learning_rate": 2.2261332057285384e-06,
      "loss": 0.448,
      "step": 9714
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.7968111038208,
      "learning_rate": 2.2251670636983684e-06,
      "loss": 1.0677,
      "step": 9715
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.0438318252563477,
      "learning_rate": 2.2242010713599322e-06,
      "loss": 0.2032,
      "step": 9716
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.220921039581299,
      "learning_rate": 2.223235228765339e-06,
      "loss": 0.0964,
      "step": 9717
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.16649055480957,
      "learning_rate": 2.2222695359666967e-06,
      "loss": 0.6838,
      "step": 9718
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.600552558898926,
      "learning_rate": 2.2213039930160985e-06,
      "loss": 0.8236,
      "step": 9719
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.202962875366211,
      "learning_rate": 2.2203385999656325e-06,
      "loss": 0.2234,
      "step": 9720
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.073688507080078,
      "learning_rate": 2.2193733568673763e-06,
      "loss": 0.4167,
      "step": 9721
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8355743885040283,
      "learning_rate": 2.218408263773403e-06,
      "loss": 0.208,
      "step": 9722
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.175326347351074,
      "learning_rate": 2.2174433207357732e-06,
      "loss": 0.7523,
      "step": 9723
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8167232275009155,
      "learning_rate": 2.216478527806545e-06,
      "loss": 0.0945,
      "step": 9724
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.418367862701416,
      "learning_rate": 2.215513885037765e-06,
      "loss": 0.3473,
      "step": 9725
    },
    {
      "epoch": 0.7,
      "grad_norm": 11.501703262329102,
      "learning_rate": 2.2145493924814685e-06,
      "loss": 0.5426,
      "step": 9726
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.050699710845947,
      "learning_rate": 2.213585050189691e-06,
      "loss": 0.7134,
      "step": 9727
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6074671745300293,
      "learning_rate": 2.2126208582144525e-06,
      "loss": 0.0869,
      "step": 9728
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.5023651123046875,
      "learning_rate": 2.2116568166077677e-06,
      "loss": 0.6414,
      "step": 9729
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.241889476776123,
      "learning_rate": 2.210692925421642e-06,
      "loss": 0.2963,
      "step": 9730
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8485519886016846,
      "learning_rate": 2.2097291847080765e-06,
      "loss": 0.1934,
      "step": 9731
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.883490562438965,
      "learning_rate": 2.2087655945190595e-06,
      "loss": 0.8249,
      "step": 9732
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.584134817123413,
      "learning_rate": 2.2078021549065737e-06,
      "loss": 0.1806,
      "step": 9733
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.484528541564941,
      "learning_rate": 2.206838865922592e-06,
      "loss": 0.6334,
      "step": 9734
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5028202533721924,
      "learning_rate": 2.2058757276190795e-06,
      "loss": 0.1075,
      "step": 9735
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.0486183166503906,
      "learning_rate": 2.204912740047996e-06,
      "loss": 0.1369,
      "step": 9736
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.1614556312561035,
      "learning_rate": 2.2039499032612915e-06,
      "loss": 0.1753,
      "step": 9737
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.346362590789795,
      "learning_rate": 2.2029872173109056e-06,
      "loss": 0.7008,
      "step": 9738
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.850756645202637,
      "learning_rate": 2.20202468224877e-06,
      "loss": 1.4708,
      "step": 9739
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.755070686340332,
      "learning_rate": 2.201062298126814e-06,
      "loss": 0.088,
      "step": 9740
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.15938663482666,
      "learning_rate": 2.2001000649969524e-06,
      "loss": 0.6577,
      "step": 9741
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1770031452178955,
      "learning_rate": 2.199137982911092e-06,
      "loss": 0.1088,
      "step": 9742
    },
    {
      "epoch": 0.7,
      "grad_norm": 11.062973022460938,
      "learning_rate": 2.198176051921138e-06,
      "loss": 0.644,
      "step": 9743
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8973469734191895,
      "learning_rate": 2.197214272078982e-06,
      "loss": 0.0774,
      "step": 9744
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3672356605529785,
      "learning_rate": 2.1962526434365035e-06,
      "loss": 0.1149,
      "step": 9745
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.6758246421813965,
      "learning_rate": 2.1952911660455834e-06,
      "loss": 0.3693,
      "step": 9746
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.556196928024292,
      "learning_rate": 2.1943298399580885e-06,
      "loss": 0.3113,
      "step": 9747
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3640921115875244,
      "learning_rate": 2.193368665225877e-06,
      "loss": 0.1249,
      "step": 9748
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.979544162750244,
      "learning_rate": 2.1924076419008043e-06,
      "loss": 0.7724,
      "step": 9749
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8884044885635376,
      "learning_rate": 2.1914467700347126e-06,
      "loss": 0.0642,
      "step": 9750
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.3530023097991943,
      "learning_rate": 2.190486049679434e-06,
      "loss": 0.6571,
      "step": 9751
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3550331592559814,
      "learning_rate": 2.1895254808868004e-06,
      "loss": 0.0859,
      "step": 9752
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.7956790924072266,
      "learning_rate": 2.188565063708629e-06,
      "loss": 0.791,
      "step": 9753
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5936295986175537,
      "learning_rate": 2.187604798196729e-06,
      "loss": 0.1489,
      "step": 9754
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.6128511428833,
      "learning_rate": 2.1866446844029067e-06,
      "loss": 0.5573,
      "step": 9755
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2448995113372803,
      "learning_rate": 2.1856847223789557e-06,
      "loss": 0.0738,
      "step": 9756
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8647797107696533,
      "learning_rate": 2.1847249121766574e-06,
      "loss": 0.128,
      "step": 9757
    },
    {
      "epoch": 0.7,
      "grad_norm": 11.461533546447754,
      "learning_rate": 2.183765253847796e-06,
      "loss": 0.879,
      "step": 9758
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6782238483428955,
      "learning_rate": 2.182805747444139e-06,
      "loss": 0.1362,
      "step": 9759
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.283342123031616,
      "learning_rate": 2.1818463930174464e-06,
      "loss": 0.1762,
      "step": 9760
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3625283241271973,
      "learning_rate": 2.1808871906194755e-06,
      "loss": 0.0945,
      "step": 9761
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.242849826812744,
      "learning_rate": 2.1799281403019695e-06,
      "loss": 0.0865,
      "step": 9762
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.846073627471924,
      "learning_rate": 2.1789692421166642e-06,
      "loss": 0.1466,
      "step": 9763
    },
    {
      "epoch": 0.7,
      "grad_norm": 16.087814331054688,
      "learning_rate": 2.178010496115292e-06,
      "loss": 0.6347,
      "step": 9764
    },
    {
      "epoch": 0.7,
      "grad_norm": 15.332185745239258,
      "learning_rate": 2.177051902349571e-06,
      "loss": 1.0685,
      "step": 9765
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.626577615737915,
      "learning_rate": 2.176093460871215e-06,
      "loss": 0.1495,
      "step": 9766
    },
    {
      "epoch": 0.7,
      "grad_norm": 16.4563045501709,
      "learning_rate": 2.175135171731926e-06,
      "loss": 0.879,
      "step": 9767
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9865851402282715,
      "learning_rate": 2.174177034983406e-06,
      "loss": 0.2125,
      "step": 9768
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.022860527038574,
      "learning_rate": 2.173219050677334e-06,
      "loss": 0.0939,
      "step": 9769
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.142637014389038,
      "learning_rate": 2.1722612188653963e-06,
      "loss": 0.1906,
      "step": 9770
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.336750984191895,
      "learning_rate": 2.1713035395992628e-06,
      "loss": 0.4005,
      "step": 9771
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1629481315612793,
      "learning_rate": 2.170346012930596e-06,
      "loss": 0.0948,
      "step": 9772
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4635422229766846,
      "learning_rate": 2.1693886389110487e-06,
      "loss": 0.102,
      "step": 9773
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.88806414604187,
      "learning_rate": 2.1684314175922715e-06,
      "loss": 0.1495,
      "step": 9774
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.2940213680267334,
      "learning_rate": 2.167474349025902e-06,
      "loss": 0.1785,
      "step": 9775
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9780209064483643,
      "learning_rate": 2.1665174332635674e-06,
      "loss": 0.108,
      "step": 9776
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.358593940734863,
      "learning_rate": 2.165560670356894e-06,
      "loss": 0.9958,
      "step": 9777
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.0677297115325928,
      "learning_rate": 2.1646040603574935e-06,
      "loss": 0.1774,
      "step": 9778
    },
    {
      "epoch": 0.7,
      "grad_norm": 15.489622116088867,
      "learning_rate": 2.1636476033169694e-06,
      "loss": 0.6067,
      "step": 9779
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.623748779296875,
      "learning_rate": 2.1626912992869247e-06,
      "loss": 0.183,
      "step": 9780
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.7118988037109375,
      "learning_rate": 2.1617351483189426e-06,
      "loss": 0.6205,
      "step": 9781
    },
    {
      "epoch": 0.7,
      "grad_norm": 11.906020164489746,
      "learning_rate": 2.160779150464604e-06,
      "loss": 0.7022,
      "step": 9782
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8196260929107666,
      "learning_rate": 2.159823305775485e-06,
      "loss": 0.1381,
      "step": 9783
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.252772808074951,
      "learning_rate": 2.1588676143031483e-06,
      "loss": 0.1049,
      "step": 9784
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.193235397338867,
      "learning_rate": 2.157912076099148e-06,
      "loss": 0.7515,
      "step": 9785
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.5137646198272705,
      "learning_rate": 2.1569566912150345e-06,
      "loss": 0.5667,
      "step": 9786
    },
    {
      "epoch": 0.7,
      "grad_norm": 15.995034217834473,
      "learning_rate": 2.156001459702346e-06,
      "loss": 0.7419,
      "step": 9787
    },
    {
      "epoch": 0.7,
      "grad_norm": 18.599462509155273,
      "learning_rate": 2.155046381612612e-06,
      "loss": 1.1874,
      "step": 9788
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4208717346191406,
      "learning_rate": 2.154091456997359e-06,
      "loss": 0.1273,
      "step": 9789
    },
    {
      "epoch": 0.7,
      "grad_norm": 23.273212432861328,
      "learning_rate": 2.153136685908099e-06,
      "loss": 0.7223,
      "step": 9790
    },
    {
      "epoch": 0.7,
      "grad_norm": 15.60736083984375,
      "learning_rate": 2.152182068396338e-06,
      "loss": 0.5123,
      "step": 9791
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.070708274841309,
      "learning_rate": 2.1512276045135777e-06,
      "loss": 0.3434,
      "step": 9792
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.928802490234375,
      "learning_rate": 2.150273294311303e-06,
      "loss": 0.1213,
      "step": 9793
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.7369794845581055,
      "learning_rate": 2.149319137840996e-06,
      "loss": 0.7553,
      "step": 9794
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3518176078796387,
      "learning_rate": 2.1483651351541328e-06,
      "loss": 0.1424,
      "step": 9795
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.3121862411499023,
      "learning_rate": 2.147411286302176e-06,
      "loss": 0.2177,
      "step": 9796
    },
    {
      "epoch": 0.7,
      "grad_norm": 16.7918643951416,
      "learning_rate": 2.1464575913365814e-06,
      "loss": 0.9576,
      "step": 9797
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.929622650146484,
      "learning_rate": 2.1455040503088004e-06,
      "loss": 0.5713,
      "step": 9798
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.663630485534668,
      "learning_rate": 2.1445506632702705e-06,
      "loss": 0.0877,
      "step": 9799
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.329530239105225,
      "learning_rate": 2.143597430272422e-06,
      "loss": 0.8138,
      "step": 9800
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.040204048156738,
      "learning_rate": 2.1426443513666828e-06,
      "loss": 0.6524,
      "step": 9801
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.078885555267334,
      "learning_rate": 2.141691426604464e-06,
      "loss": 0.6849,
      "step": 9802
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.2250752449035645,
      "learning_rate": 2.1407386560371747e-06,
      "loss": 0.7941,
      "step": 9803
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.835178375244141,
      "learning_rate": 2.139786039716209e-06,
      "loss": 0.9774,
      "step": 9804
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.143414258956909,
      "learning_rate": 2.1388335776929643e-06,
      "loss": 0.1361,
      "step": 9805
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.8330039978027344,
      "learning_rate": 2.1378812700188146e-06,
      "loss": 0.1634,
      "step": 9806
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.7789974212646484,
      "learning_rate": 2.136929116745138e-06,
      "loss": 0.1989,
      "step": 9807
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9301772117614746,
      "learning_rate": 2.135977117923298e-06,
      "loss": 0.1105,
      "step": 9808
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.2307238578796387,
      "learning_rate": 2.135025273604652e-06,
      "loss": 0.1844,
      "step": 9809
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.144582271575928,
      "learning_rate": 2.1340735838405455e-06,
      "loss": 0.459,
      "step": 9810
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5595152378082275,
      "learning_rate": 2.133122048682324e-06,
      "loss": 0.1148,
      "step": 9811
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.421846628189087,
      "learning_rate": 2.1321706681813155e-06,
      "loss": 0.1151,
      "step": 9812
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.853806495666504,
      "learning_rate": 2.131219442388842e-06,
      "loss": 0.6117,
      "step": 9813
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0502374172210693,
      "learning_rate": 2.1302683713562234e-06,
      "loss": 0.1077,
      "step": 9814
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.910639762878418,
      "learning_rate": 2.129317455134763e-06,
      "loss": 0.8546,
      "step": 9815
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.94804048538208,
      "learning_rate": 2.128366693775758e-06,
      "loss": 0.112,
      "step": 9816
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.8069043159484863,
      "learning_rate": 2.127416087330504e-06,
      "loss": 0.6778,
      "step": 9817
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9281303882598877,
      "learning_rate": 2.1264656358502765e-06,
      "loss": 0.1429,
      "step": 9818
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6748416423797607,
      "learning_rate": 2.1255153393863496e-06,
      "loss": 0.0483,
      "step": 9819
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.609281539916992,
      "learning_rate": 2.1245651979899916e-06,
      "loss": 0.5071,
      "step": 9820
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.935322284698486,
      "learning_rate": 2.123615211712457e-06,
      "loss": 0.7522,
      "step": 9821
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.6767454147338867,
      "learning_rate": 2.122665380604992e-06,
      "loss": 0.6961,
      "step": 9822
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.920179605484009,
      "learning_rate": 2.12171570471884e-06,
      "loss": 0.1707,
      "step": 9823
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.913544178009033,
      "learning_rate": 2.1207661841052314e-06,
      "loss": 0.7558,
      "step": 9824
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4338560104370117,
      "learning_rate": 2.1198168188153868e-06,
      "loss": 0.0918,
      "step": 9825
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.286251068115234,
      "learning_rate": 2.118867608900524e-06,
      "loss": 0.797,
      "step": 9826
    },
    {
      "epoch": 0.7,
      "grad_norm": 14.096146583557129,
      "learning_rate": 2.1179185544118484e-06,
      "loss": 0.5213,
      "step": 9827
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.8967063426971436,
      "learning_rate": 2.116969655400556e-06,
      "loss": 0.7821,
      "step": 9828
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.1200337409973145,
      "learning_rate": 2.116020911917841e-06,
      "loss": 0.8161,
      "step": 9829
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.6546003818511963,
      "learning_rate": 2.1150723240148797e-06,
      "loss": 0.711,
      "step": 9830
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8217414617538452,
      "learning_rate": 2.1141238917428443e-06,
      "loss": 0.0607,
      "step": 9831
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.243110179901123,
      "learning_rate": 2.1131756151529035e-06,
      "loss": 0.7316,
      "step": 9832
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.011942863464355,
      "learning_rate": 2.1122274942962107e-06,
      "loss": 0.4462,
      "step": 9833
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.746769905090332,
      "learning_rate": 2.1112795292239125e-06,
      "loss": 0.8937,
      "step": 9834
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.827153205871582,
      "learning_rate": 2.1103317199871507e-06,
      "loss": 1.0798,
      "step": 9835
    },
    {
      "epoch": 0.71,
      "grad_norm": 18.791109085083008,
      "learning_rate": 2.109384066637055e-06,
      "loss": 0.8719,
      "step": 9836
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.9000909328460693,
      "learning_rate": 2.108436569224747e-06,
      "loss": 0.7236,
      "step": 9837
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.566760778427124,
      "learning_rate": 2.107489227801339e-06,
      "loss": 0.1314,
      "step": 9838
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.6771397590637207,
      "learning_rate": 2.1065420424179408e-06,
      "loss": 0.1343,
      "step": 9839
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.470914363861084,
      "learning_rate": 2.1055950131256466e-06,
      "loss": 0.0706,
      "step": 9840
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.70298957824707,
      "learning_rate": 2.104648139975545e-06,
      "loss": 1.0103,
      "step": 9841
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.3835651874542236,
      "learning_rate": 2.103701423018718e-06,
      "loss": 0.1321,
      "step": 9842
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.220467567443848,
      "learning_rate": 2.1027548623062337e-06,
      "loss": 0.7139,
      "step": 9843
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.603226900100708,
      "learning_rate": 2.1018084578891597e-06,
      "loss": 0.0826,
      "step": 9844
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0374984741210938,
      "learning_rate": 2.100862209818549e-06,
      "loss": 0.0604,
      "step": 9845
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.188058376312256,
      "learning_rate": 2.0999161181454485e-06,
      "loss": 0.749,
      "step": 9846
    },
    {
      "epoch": 0.71,
      "grad_norm": 15.977128028869629,
      "learning_rate": 2.098970182920894e-06,
      "loss": 0.7847,
      "step": 9847
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2223479747772217,
      "learning_rate": 2.0980244041959192e-06,
      "loss": 0.2023,
      "step": 9848
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.470030784606934,
      "learning_rate": 2.097078782021543e-06,
      "loss": 0.7988,
      "step": 9849
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.5948007106781006,
      "learning_rate": 2.096133316448777e-06,
      "loss": 0.2192,
      "step": 9850
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.337099075317383,
      "learning_rate": 2.095188007528628e-06,
      "loss": 0.6667,
      "step": 9851
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.386388301849365,
      "learning_rate": 2.094242855312091e-06,
      "loss": 0.7796,
      "step": 9852
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.3355159759521484,
      "learning_rate": 2.093297859850153e-06,
      "loss": 0.1367,
      "step": 9853
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.732687473297119,
      "learning_rate": 2.0923530211937933e-06,
      "loss": 0.7904,
      "step": 9854
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.310640811920166,
      "learning_rate": 2.0914083393939816e-06,
      "loss": 0.5865,
      "step": 9855
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.38280200958252,
      "learning_rate": 2.0904638145016785e-06,
      "loss": 0.69,
      "step": 9856
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.3573999404907227,
      "learning_rate": 2.0895194465678412e-06,
      "loss": 0.0731,
      "step": 9857
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5460022687911987,
      "learning_rate": 2.0885752356434137e-06,
      "loss": 0.0522,
      "step": 9858
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.677006721496582,
      "learning_rate": 2.08763118177933e-06,
      "loss": 0.4677,
      "step": 9859
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.3751907348632812,
      "learning_rate": 2.0866872850265212e-06,
      "loss": 0.2076,
      "step": 9860
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.090085983276367,
      "learning_rate": 2.0857435454359065e-06,
      "loss": 0.8742,
      "step": 9861
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.5688281059265137,
      "learning_rate": 2.0847999630583944e-06,
      "loss": 0.2028,
      "step": 9862
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2899587154388428,
      "learning_rate": 2.0838565379448918e-06,
      "loss": 0.1708,
      "step": 9863
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.8762500286102295,
      "learning_rate": 2.0829132701462903e-06,
      "loss": 0.8962,
      "step": 9864
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.711379051208496,
      "learning_rate": 2.081970159713476e-06,
      "loss": 0.334,
      "step": 9865
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.4390034675598145,
      "learning_rate": 2.081027206697327e-06,
      "loss": 0.7546,
      "step": 9866
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.619525909423828,
      "learning_rate": 2.0800844111487107e-06,
      "loss": 0.6882,
      "step": 9867
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.190035820007324,
      "learning_rate": 2.079141773118486e-06,
      "loss": 0.0898,
      "step": 9868
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.74591064453125,
      "learning_rate": 2.078199292657509e-06,
      "loss": 0.9134,
      "step": 9869
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.8903355598449707,
      "learning_rate": 2.07725696981662e-06,
      "loss": 0.8984,
      "step": 9870
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2906298637390137,
      "learning_rate": 2.0763148046466523e-06,
      "loss": 0.0859,
      "step": 9871
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.337170600891113,
      "learning_rate": 2.0753727971984365e-06,
      "loss": 1.0112,
      "step": 9872
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.1888253688812256,
      "learning_rate": 2.074430947522788e-06,
      "loss": 0.1757,
      "step": 9873
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.775999665260315,
      "learning_rate": 2.073489255670516e-06,
      "loss": 0.0498,
      "step": 9874
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.266019821166992,
      "learning_rate": 2.0725477216924188e-06,
      "loss": 0.1695,
      "step": 9875
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.155212879180908,
      "learning_rate": 2.0716063456392933e-06,
      "loss": 0.672,
      "step": 9876
    },
    {
      "epoch": 0.71,
      "grad_norm": 17.578927993774414,
      "learning_rate": 2.070665127561921e-06,
      "loss": 0.7457,
      "step": 9877
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.3142776489257812,
      "learning_rate": 2.069724067511077e-06,
      "loss": 0.2577,
      "step": 9878
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.250366449356079,
      "learning_rate": 2.0687831655375273e-06,
      "loss": 0.099,
      "step": 9879
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.848388671875,
      "learning_rate": 2.067842421692029e-06,
      "loss": 0.0533,
      "step": 9880
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.416240215301514,
      "learning_rate": 2.066901836025335e-06,
      "loss": 0.7952,
      "step": 9881
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.0914978981018066,
      "learning_rate": 2.0659614085881847e-06,
      "loss": 0.2245,
      "step": 9882
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.045044422149658,
      "learning_rate": 2.0650211394313107e-06,
      "loss": 0.0915,
      "step": 9883
    },
    {
      "epoch": 0.71,
      "grad_norm": 13.195756912231445,
      "learning_rate": 2.0640810286054347e-06,
      "loss": 0.6442,
      "step": 9884
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.005734920501709,
      "learning_rate": 2.0631410761612756e-06,
      "loss": 0.111,
      "step": 9885
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.6306703090667725,
      "learning_rate": 2.062201282149539e-06,
      "loss": 0.1741,
      "step": 9886
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.066015243530273,
      "learning_rate": 2.061261646620922e-06,
      "loss": 0.7111,
      "step": 9887
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2048821449279785,
      "learning_rate": 2.060322169626116e-06,
      "loss": 0.1485,
      "step": 9888
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.004467964172363,
      "learning_rate": 2.0593828512158044e-06,
      "loss": 0.3497,
      "step": 9889
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.389127731323242,
      "learning_rate": 2.058443691440653e-06,
      "loss": 0.5447,
      "step": 9890
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.022836208343506,
      "learning_rate": 2.0575046903513324e-06,
      "loss": 0.6945,
      "step": 9891
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7183489799499512,
      "learning_rate": 2.056565847998495e-06,
      "loss": 0.0513,
      "step": 9892
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.7028934955596924,
      "learning_rate": 2.0556271644327874e-06,
      "loss": 0.8602,
      "step": 9893
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.3196935653686523,
      "learning_rate": 2.054688639704851e-06,
      "loss": 0.0826,
      "step": 9894
    },
    {
      "epoch": 0.71,
      "grad_norm": 20.808788299560547,
      "learning_rate": 2.0537502738653137e-06,
      "loss": 0.8634,
      "step": 9895
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.7521352767944336,
      "learning_rate": 2.0528120669647955e-06,
      "loss": 0.1093,
      "step": 9896
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.749913215637207,
      "learning_rate": 2.051874019053912e-06,
      "loss": 0.4802,
      "step": 9897
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.234095573425293,
      "learning_rate": 2.050936130183266e-06,
      "loss": 0.8431,
      "step": 9898
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.076907157897949,
      "learning_rate": 2.049998400403451e-06,
      "loss": 0.0726,
      "step": 9899
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.0759105682373047,
      "learning_rate": 2.0490608297650585e-06,
      "loss": 0.1881,
      "step": 9900
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.7676990032196045,
      "learning_rate": 2.0481234183186656e-06,
      "loss": 0.0757,
      "step": 9901
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.4293346405029297,
      "learning_rate": 2.047186166114837e-06,
      "loss": 0.0409,
      "step": 9902
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.9076223373413086,
      "learning_rate": 2.0462490732041396e-06,
      "loss": 0.176,
      "step": 9903
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.809581279754639,
      "learning_rate": 2.045312139637124e-06,
      "loss": 0.4804,
      "step": 9904
    },
    {
      "epoch": 0.71,
      "grad_norm": 14.367630004882812,
      "learning_rate": 2.0443753654643326e-06,
      "loss": 0.7909,
      "step": 9905
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.129429817199707,
      "learning_rate": 2.0434387507363054e-06,
      "loss": 0.605,
      "step": 9906
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.143836498260498,
      "learning_rate": 2.042502295503565e-06,
      "loss": 0.7522,
      "step": 9907
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.4820847511291504,
      "learning_rate": 2.04156599981663e-06,
      "loss": 0.1075,
      "step": 9908
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.6272130012512207,
      "learning_rate": 2.040629863726013e-06,
      "loss": 0.7467,
      "step": 9909
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.608337163925171,
      "learning_rate": 2.0396938872822124e-06,
      "loss": 0.1473,
      "step": 9910
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.0038251876831055,
      "learning_rate": 2.0387580705357212e-06,
      "loss": 1.0082,
      "step": 9911
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.764991760253906,
      "learning_rate": 2.0378224135370224e-06,
      "loss": 0.5869,
      "step": 9912
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.9701714515686035,
      "learning_rate": 2.0368869163365947e-06,
      "loss": 0.1859,
      "step": 9913
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.855421781539917,
      "learning_rate": 2.035951578984898e-06,
      "loss": 0.8229,
      "step": 9914
    },
    {
      "epoch": 0.71,
      "grad_norm": 14.003132820129395,
      "learning_rate": 2.0350164015323963e-06,
      "loss": 0.8148,
      "step": 9915
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.709825038909912,
      "learning_rate": 2.034081384029537e-06,
      "loss": 0.4146,
      "step": 9916
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.495694160461426,
      "learning_rate": 2.0331465265267597e-06,
      "loss": 0.2013,
      "step": 9917
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.374276638031006,
      "learning_rate": 2.032211829074496e-06,
      "loss": 0.1918,
      "step": 9918
    },
    {
      "epoch": 0.71,
      "grad_norm": 14.617422103881836,
      "learning_rate": 2.0312772917231716e-06,
      "loss": 0.9497,
      "step": 9919
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.935542583465576,
      "learning_rate": 2.0303429145232007e-06,
      "loss": 0.0685,
      "step": 9920
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.8894145488739014,
      "learning_rate": 2.0294086975249865e-06,
      "loss": 0.1162,
      "step": 9921
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.504500150680542,
      "learning_rate": 2.0284746407789307e-06,
      "loss": 0.0772,
      "step": 9922
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.326383590698242,
      "learning_rate": 2.0275407443354204e-06,
      "loss": 1.0962,
      "step": 9923
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.490534543991089,
      "learning_rate": 2.0266070082448334e-06,
      "loss": 0.159,
      "step": 9924
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.377859592437744,
      "learning_rate": 2.0256734325575477e-06,
      "loss": 0.0819,
      "step": 9925
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.469688415527344,
      "learning_rate": 2.0247400173239194e-06,
      "loss": 0.8059,
      "step": 9926
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.4808404445648193,
      "learning_rate": 2.0238067625943035e-06,
      "loss": 0.0793,
      "step": 9927
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.3426027297973633,
      "learning_rate": 2.0228736684190496e-06,
      "loss": 0.1301,
      "step": 9928
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.691043853759766,
      "learning_rate": 2.021940734848492e-06,
      "loss": 0.4865,
      "step": 9929
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.362807750701904,
      "learning_rate": 2.021007961932957e-06,
      "loss": 0.7459,
      "step": 9930
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.025270938873291,
      "learning_rate": 2.0200753497227683e-06,
      "loss": 0.0466,
      "step": 9931
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.98924446105957,
      "learning_rate": 2.019142898268235e-06,
      "loss": 0.5725,
      "step": 9932
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.0769689083099365,
      "learning_rate": 2.018210607619657e-06,
      "loss": 0.1559,
      "step": 9933
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.456174850463867,
      "learning_rate": 2.0172784778273326e-06,
      "loss": 0.7329,
      "step": 9934
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.7011115550994873,
      "learning_rate": 2.0163465089415434e-06,
      "loss": 0.7544,
      "step": 9935
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.764899969100952,
      "learning_rate": 2.0154147010125654e-06,
      "loss": 0.1837,
      "step": 9936
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.4879932403564453,
      "learning_rate": 2.0144830540906695e-06,
      "loss": 0.5892,
      "step": 9937
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.130298137664795,
      "learning_rate": 2.013551568226111e-06,
      "loss": 0.0592,
      "step": 9938
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.577540874481201,
      "learning_rate": 2.012620243469139e-06,
      "loss": 0.2041,
      "step": 9939
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.36626672744751,
      "learning_rate": 2.0116890798699986e-06,
      "loss": 0.8729,
      "step": 9940
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2766306400299072,
      "learning_rate": 2.0107580774789215e-06,
      "loss": 0.1737,
      "step": 9941
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.849405288696289,
      "learning_rate": 2.009827236346129e-06,
      "loss": 0.1534,
      "step": 9942
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.841369390487671,
      "learning_rate": 2.0088965565218406e-06,
      "loss": 0.0835,
      "step": 9943
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.9590702056884766,
      "learning_rate": 2.007966038056261e-06,
      "loss": 0.116,
      "step": 9944
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.474616765975952,
      "learning_rate": 2.0070356809995877e-06,
      "loss": 0.132,
      "step": 9945
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.370838165283203,
      "learning_rate": 2.0061054854020095e-06,
      "loss": 0.789,
      "step": 9946
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.018601179122925,
      "learning_rate": 2.0051754513137095e-06,
      "loss": 0.1917,
      "step": 9947
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.086749792098999,
      "learning_rate": 2.0042455787848575e-06,
      "loss": 0.1032,
      "step": 9948
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.859102964401245,
      "learning_rate": 2.0033158678656173e-06,
      "loss": 0.0753,
      "step": 9949
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.3138298988342285,
      "learning_rate": 2.0023863186061425e-06,
      "loss": 0.6593,
      "step": 9950
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.897357940673828,
      "learning_rate": 2.001456931056578e-06,
      "loss": 0.9654,
      "step": 9951
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.67750883102417,
      "learning_rate": 2.0005277052670633e-06,
      "loss": 0.1232,
      "step": 9952
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.8326637744903564,
      "learning_rate": 1.9995986412877256e-06,
      "loss": 0.7599,
      "step": 9953
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.5928256511688232,
      "learning_rate": 1.9986697391686837e-06,
      "loss": 0.1147,
      "step": 9954
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.19262433052063,
      "learning_rate": 1.9977409989600473e-06,
      "loss": 0.1263,
      "step": 9955
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.879759788513184,
      "learning_rate": 1.9968124207119217e-06,
      "loss": 0.8389,
      "step": 9956
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.665553092956543,
      "learning_rate": 1.9958840044743976e-06,
      "loss": 0.7138,
      "step": 9957
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2711362838745117,
      "learning_rate": 1.994955750297559e-06,
      "loss": 0.2256,
      "step": 9958
    },
    {
      "epoch": 0.71,
      "grad_norm": 14.215291976928711,
      "learning_rate": 1.994027658231485e-06,
      "loss": 0.9563,
      "step": 9959
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2670042514801025,
      "learning_rate": 1.9930997283262406e-06,
      "loss": 0.121,
      "step": 9960
    },
    {
      "epoch": 0.71,
      "grad_norm": 15.999619483947754,
      "learning_rate": 1.992171960631884e-06,
      "loss": 0.4933,
      "step": 9961
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.802182912826538,
      "learning_rate": 1.9912443551984646e-06,
      "loss": 0.2156,
      "step": 9962
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.203839302062988,
      "learning_rate": 1.9903169120760235e-06,
      "loss": 0.6995,
      "step": 9963
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.644659996032715,
      "learning_rate": 1.989389631314591e-06,
      "loss": 0.1316,
      "step": 9964
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.166182994842529,
      "learning_rate": 1.988462512964194e-06,
      "loss": 0.7552,
      "step": 9965
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.5990495681762695,
      "learning_rate": 1.987535557074845e-06,
      "loss": 0.1135,
      "step": 9966
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7322596311569214,
      "learning_rate": 1.9866087636965485e-06,
      "loss": 0.0443,
      "step": 9967
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.109344482421875,
      "learning_rate": 1.985682132879304e-06,
      "loss": 0.7168,
      "step": 9968
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.8469951152801514,
      "learning_rate": 1.984755664673099e-06,
      "loss": 0.1419,
      "step": 9969
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1540021896362305,
      "learning_rate": 1.983829359127911e-06,
      "loss": 0.0883,
      "step": 9970
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.608898639678955,
      "learning_rate": 1.982903216293714e-06,
      "loss": 0.8922,
      "step": 9971
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.216156482696533,
      "learning_rate": 1.981977236220468e-06,
      "loss": 0.0587,
      "step": 9972
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.412724494934082,
      "learning_rate": 1.9810514189581265e-06,
      "loss": 0.8901,
      "step": 9973
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5231640338897705,
      "learning_rate": 1.980125764556633e-06,
      "loss": 0.1215,
      "step": 9974
    },
    {
      "epoch": 0.72,
      "grad_norm": 12.181808471679688,
      "learning_rate": 1.9792002730659242e-06,
      "loss": 0.894,
      "step": 9975
    },
    {
      "epoch": 0.72,
      "grad_norm": 31.751972198486328,
      "learning_rate": 1.9782749445359244e-06,
      "loss": 0.7894,
      "step": 9976
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.08626127243042,
      "learning_rate": 1.977349779016555e-06,
      "loss": 0.0679,
      "step": 9977
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.714125156402588,
      "learning_rate": 1.9764247765577237e-06,
      "loss": 0.6845,
      "step": 9978
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.391080141067505,
      "learning_rate": 1.97549993720933e-06,
      "loss": 0.09,
      "step": 9979
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.5079779624938965,
      "learning_rate": 1.974575261021267e-06,
      "loss": 0.7512,
      "step": 9980
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.082646369934082,
      "learning_rate": 1.9736507480434175e-06,
      "loss": 0.7444,
      "step": 9981
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7849234342575073,
      "learning_rate": 1.9727263983256545e-06,
      "loss": 0.1,
      "step": 9982
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.0221807956695557,
      "learning_rate": 1.9718022119178415e-06,
      "loss": 0.1887,
      "step": 9983
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.232024669647217,
      "learning_rate": 1.9708781888698386e-06,
      "loss": 0.0999,
      "step": 9984
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.624634265899658,
      "learning_rate": 1.9699543292314916e-06,
      "loss": 0.105,
      "step": 9985
    },
    {
      "epoch": 0.72,
      "grad_norm": 15.923519134521484,
      "learning_rate": 1.9690306330526395e-06,
      "loss": 1.1985,
      "step": 9986
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.160707712173462,
      "learning_rate": 1.9681071003831115e-06,
      "loss": 0.0879,
      "step": 9987
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.0698561668396,
      "learning_rate": 1.9671837312727276e-06,
      "loss": 0.7541,
      "step": 9988
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.892002820968628,
      "learning_rate": 1.9662605257713035e-06,
      "loss": 0.0463,
      "step": 9989
    },
    {
      "epoch": 0.72,
      "grad_norm": 12.581464767456055,
      "learning_rate": 1.9653374839286405e-06,
      "loss": 0.7122,
      "step": 9990
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2560346126556396,
      "learning_rate": 1.9644146057945336e-06,
      "loss": 0.0662,
      "step": 9991
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.867810249328613,
      "learning_rate": 1.963491891418766e-06,
      "loss": 0.9841,
      "step": 9992
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.939608097076416,
      "learning_rate": 1.9625693408511193e-06,
      "loss": 0.2376,
      "step": 9993
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.140176773071289,
      "learning_rate": 1.9616469541413595e-06,
      "loss": 0.078,
      "step": 9994
    },
    {
      "epoch": 0.72,
      "grad_norm": 16.30414390563965,
      "learning_rate": 1.9607247313392436e-06,
      "loss": 0.8897,
      "step": 9995
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9518535137176514,
      "learning_rate": 1.9598026724945264e-06,
      "loss": 0.1363,
      "step": 9996
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1598217487335205,
      "learning_rate": 1.9588807776569486e-06,
      "loss": 0.1163,
      "step": 9997
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.8591837882995605,
      "learning_rate": 1.9579590468762382e-06,
      "loss": 0.4451,
      "step": 9998
    },
    {
      "epoch": 0.72,
      "grad_norm": 19.1069393157959,
      "learning_rate": 1.957037480202124e-06,
      "loss": 1.0848,
      "step": 9999
    },
    {
      "epoch": 0.72,
      "grad_norm": 13.27191162109375,
      "learning_rate": 1.95611607768432e-06,
      "loss": 1.5096,
      "step": 10000
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.839588165283203,
      "learning_rate": 1.95519483937253e-06,
      "loss": 0.1803,
      "step": 10001
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.087431907653809,
      "learning_rate": 1.9542737653164546e-06,
      "loss": 0.8655,
      "step": 10002
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5173442363739014,
      "learning_rate": 1.953352855565781e-06,
      "loss": 0.1316,
      "step": 10003
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.169168949127197,
      "learning_rate": 1.952432110170187e-06,
      "loss": 0.8999,
      "step": 10004
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.888527870178223,
      "learning_rate": 1.951511529179347e-06,
      "loss": 0.8431,
      "step": 10005
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.197522163391113,
      "learning_rate": 1.9505911126429204e-06,
      "loss": 0.3875,
      "step": 10006
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.8107383251190186,
      "learning_rate": 1.949670860610559e-06,
      "loss": 0.9146,
      "step": 10007
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.245482444763184,
      "learning_rate": 1.9487507731319105e-06,
      "loss": 0.6366,
      "step": 10008
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.50460433959961,
      "learning_rate": 1.9478308502566096e-06,
      "loss": 0.6123,
      "step": 10009
    },
    {
      "epoch": 0.72,
      "grad_norm": 12.843910217285156,
      "learning_rate": 1.946911092034278e-06,
      "loss": 0.598,
      "step": 10010
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.871845245361328,
      "learning_rate": 1.9459914985145377e-06,
      "loss": 0.1908,
      "step": 10011
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.491919994354248,
      "learning_rate": 1.9450720697469967e-06,
      "loss": 0.13,
      "step": 10012
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.638735294342041,
      "learning_rate": 1.944152805781252e-06,
      "loss": 0.1151,
      "step": 10013
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.4121830463409424,
      "learning_rate": 1.9432337066668982e-06,
      "loss": 0.188,
      "step": 10014
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.7081897258758545,
      "learning_rate": 1.9423147724535152e-06,
      "loss": 0.2417,
      "step": 10015
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5538283586502075,
      "learning_rate": 1.941396003190675e-06,
      "loss": 0.0783,
      "step": 10016
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.976466178894043,
      "learning_rate": 1.940477398927944e-06,
      "loss": 0.7954,
      "step": 10017
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.4910879135131836,
      "learning_rate": 1.9395589597148767e-06,
      "loss": 0.1708,
      "step": 10018
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.659892559051514,
      "learning_rate": 1.938640685601019e-06,
      "loss": 0.7339,
      "step": 10019
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.924799919128418,
      "learning_rate": 1.9377225766359077e-06,
      "loss": 1.2146,
      "step": 10020
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.807741403579712,
      "learning_rate": 1.9368046328690724e-06,
      "loss": 0.7864,
      "step": 10021
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.8431408405303955,
      "learning_rate": 1.935886854350033e-06,
      "loss": 0.0837,
      "step": 10022
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.5283706188201904,
      "learning_rate": 1.9349692411282995e-06,
      "loss": 0.1817,
      "step": 10023
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3262240886688232,
      "learning_rate": 1.9340517932533736e-06,
      "loss": 0.1094,
      "step": 10024
    },
    {
      "epoch": 0.72,
      "grad_norm": 21.83769416809082,
      "learning_rate": 1.9331345107747486e-06,
      "loss": 0.7133,
      "step": 10025
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.653141975402832,
      "learning_rate": 1.9322173937419058e-06,
      "loss": 0.3084,
      "step": 10026
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.269108295440674,
      "learning_rate": 1.931300442204324e-06,
      "loss": 0.5803,
      "step": 10027
    },
    {
      "epoch": 0.72,
      "grad_norm": 18.53939437866211,
      "learning_rate": 1.930383656211468e-06,
      "loss": 0.5158,
      "step": 10028
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.361032962799072,
      "learning_rate": 1.929467035812793e-06,
      "loss": 0.4166,
      "step": 10029
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.9988019466400146,
      "learning_rate": 1.9285505810577505e-06,
      "loss": 0.9362,
      "step": 10030
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.925326347351074,
      "learning_rate": 1.927634291995778e-06,
      "loss": 0.4475,
      "step": 10031
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.6589255332946777,
      "learning_rate": 1.9267181686763037e-06,
      "loss": 0.7477,
      "step": 10032
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2398650646209717,
      "learning_rate": 1.9258022111487533e-06,
      "loss": 0.0952,
      "step": 10033
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.1909568309783936,
      "learning_rate": 1.9248864194625392e-06,
      "loss": 0.2245,
      "step": 10034
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5236799716949463,
      "learning_rate": 1.923970793667059e-06,
      "loss": 0.0828,
      "step": 10035
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.587069034576416,
      "learning_rate": 1.923055333811713e-06,
      "loss": 0.1739,
      "step": 10036
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.5383734703063965,
      "learning_rate": 1.922140039945885e-06,
      "loss": 0.4009,
      "step": 10037
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.8048064708709717,
      "learning_rate": 1.9212249121189498e-06,
      "loss": 0.1122,
      "step": 10038
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.287056922912598,
      "learning_rate": 1.920309950380278e-06,
      "loss": 0.4932,
      "step": 10039
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.580366611480713,
      "learning_rate": 1.9193951547792273e-06,
      "loss": 0.7173,
      "step": 10040
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.483264207839966,
      "learning_rate": 1.9184805253651456e-06,
      "loss": 0.1096,
      "step": 10041
    },
    {
      "epoch": 0.72,
      "grad_norm": 14.269447326660156,
      "learning_rate": 1.9175660621873777e-06,
      "loss": 0.594,
      "step": 10042
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.399045467376709,
      "learning_rate": 1.916651765295252e-06,
      "loss": 0.4334,
      "step": 10043
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.813559532165527,
      "learning_rate": 1.915737634738092e-06,
      "loss": 0.4282,
      "step": 10044
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.0386528968811035,
      "learning_rate": 1.9148236705652128e-06,
      "loss": 0.8522,
      "step": 10045
    },
    {
      "epoch": 0.72,
      "grad_norm": 12.651347160339355,
      "learning_rate": 1.9139098728259205e-06,
      "loss": 0.8653,
      "step": 10046
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.96305775642395,
      "learning_rate": 1.912996241569506e-06,
      "loss": 0.7218,
      "step": 10047
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.345427513122559,
      "learning_rate": 1.9120827768452606e-06,
      "loss": 1.0519,
      "step": 10048
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.134202241897583,
      "learning_rate": 1.911169478702461e-06,
      "loss": 0.1012,
      "step": 10049
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.171622276306152,
      "learning_rate": 1.9102563471903744e-06,
      "loss": 0.8914,
      "step": 10050
    },
    {
      "epoch": 0.72,
      "grad_norm": 16.914962768554688,
      "learning_rate": 1.9093433823582646e-06,
      "loss": 0.9643,
      "step": 10051
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5494890213012695,
      "learning_rate": 1.9084305842553802e-06,
      "loss": 0.1034,
      "step": 10052
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.7789933681488037,
      "learning_rate": 1.907517952930962e-06,
      "loss": 0.087,
      "step": 10053
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.456725597381592,
      "learning_rate": 1.9066054884342461e-06,
      "loss": 0.6203,
      "step": 10054
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.53592586517334,
      "learning_rate": 1.9056931908144555e-06,
      "loss": 0.7137,
      "step": 10055
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.412135362625122,
      "learning_rate": 1.904781060120804e-06,
      "loss": 0.0657,
      "step": 10056
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.411853790283203,
      "learning_rate": 1.903869096402497e-06,
      "loss": 0.8296,
      "step": 10057
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.6566836833953857,
      "learning_rate": 1.9029572997087358e-06,
      "loss": 0.6196,
      "step": 10058
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.6808509826660156,
      "learning_rate": 1.9020456700887024e-06,
      "loss": 0.1568,
      "step": 10059
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.7166316509246826,
      "learning_rate": 1.90113420759158e-06,
      "loss": 0.1721,
      "step": 10060
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.9222354888916016,
      "learning_rate": 1.9002229122665377e-06,
      "loss": 0.5846,
      "step": 10061
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.139251708984375,
      "learning_rate": 1.8993117841627357e-06,
      "loss": 0.6384,
      "step": 10062
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.910238265991211,
      "learning_rate": 1.8984008233293243e-06,
      "loss": 0.747,
      "step": 10063
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.544682502746582,
      "learning_rate": 1.8974900298154502e-06,
      "loss": 0.6506,
      "step": 10064
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.044267654418945,
      "learning_rate": 1.8965794036702456e-06,
      "loss": 0.8636,
      "step": 10065
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.095861911773682,
      "learning_rate": 1.8956689449428333e-06,
      "loss": 0.4915,
      "step": 10066
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.832477569580078,
      "learning_rate": 1.8947586536823326e-06,
      "loss": 0.4199,
      "step": 10067
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.7088968753814697,
      "learning_rate": 1.8938485299378484e-06,
      "loss": 0.1262,
      "step": 10068
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.764700889587402,
      "learning_rate": 1.8929385737584765e-06,
      "loss": 0.6463,
      "step": 10069
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.629500389099121,
      "learning_rate": 1.8920287851933117e-06,
      "loss": 0.9194,
      "step": 10070
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.982477188110352,
      "learning_rate": 1.8911191642914279e-06,
      "loss": 0.7852,
      "step": 10071
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.006781101226807,
      "learning_rate": 1.8902097111018952e-06,
      "loss": 0.7132,
      "step": 10072
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.4995205402374268,
      "learning_rate": 1.8893004256737796e-06,
      "loss": 0.0957,
      "step": 10073
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.427213430404663,
      "learning_rate": 1.8883913080561312e-06,
      "loss": 0.1377,
      "step": 10074
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.613620281219482,
      "learning_rate": 1.8874823582979918e-06,
      "loss": 0.3654,
      "step": 10075
    },
    {
      "epoch": 0.72,
      "grad_norm": 12.058236122131348,
      "learning_rate": 1.8865735764483995e-06,
      "loss": 0.9262,
      "step": 10076
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.97006893157959,
      "learning_rate": 1.8856649625563778e-06,
      "loss": 0.9323,
      "step": 10077
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.663029432296753,
      "learning_rate": 1.8847565166709413e-06,
      "loss": 0.1352,
      "step": 10078
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.369342565536499,
      "learning_rate": 1.8838482388411005e-06,
      "loss": 0.1245,
      "step": 10079
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.801408290863037,
      "learning_rate": 1.8829401291158523e-06,
      "loss": 0.8248,
      "step": 10080
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.6066834926605225,
      "learning_rate": 1.8820321875441832e-06,
      "loss": 0.1649,
      "step": 10081
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9533002376556396,
      "learning_rate": 1.8811244141750795e-06,
      "loss": 0.1437,
      "step": 10082
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.306830644607544,
      "learning_rate": 1.8802168090575056e-06,
      "loss": 0.112,
      "step": 10083
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9208574295043945,
      "learning_rate": 1.8793093722404249e-06,
      "loss": 0.1559,
      "step": 10084
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.526397705078125,
      "learning_rate": 1.8784021037727924e-06,
      "loss": 0.6646,
      "step": 10085
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.557220935821533,
      "learning_rate": 1.87749500370355e-06,
      "loss": 0.647,
      "step": 10086
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.160894870758057,
      "learning_rate": 1.876588072081632e-06,
      "loss": 0.434,
      "step": 10087
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6697129011154175,
      "learning_rate": 1.875681308955966e-06,
      "loss": 0.0466,
      "step": 10088
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.037266731262207,
      "learning_rate": 1.8747747143754675e-06,
      "loss": 0.7474,
      "step": 10089
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.164272785186768,
      "learning_rate": 1.873868288389043e-06,
      "loss": 0.6362,
      "step": 10090
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.00616455078125,
      "learning_rate": 1.8729620310455898e-06,
      "loss": 0.1036,
      "step": 10091
    },
    {
      "epoch": 0.72,
      "grad_norm": 18.37679100036621,
      "learning_rate": 1.8720559423940004e-06,
      "loss": 0.4162,
      "step": 10092
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.061868667602539,
      "learning_rate": 1.871150022483153e-06,
      "loss": 0.31,
      "step": 10093
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.00706672668457,
      "learning_rate": 1.8702442713619184e-06,
      "loss": 0.7251,
      "step": 10094
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.510123252868652,
      "learning_rate": 1.8693386890791587e-06,
      "loss": 0.8158,
      "step": 10095
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.438143014907837,
      "learning_rate": 1.8684332756837249e-06,
      "loss": 0.0674,
      "step": 10096
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5593533515930176,
      "learning_rate": 1.8675280312244638e-06,
      "loss": 0.1681,
      "step": 10097
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.116060733795166,
      "learning_rate": 1.8666229557502086e-06,
      "loss": 0.1448,
      "step": 10098
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.056497573852539,
      "learning_rate": 1.8657180493097842e-06,
      "loss": 0.4261,
      "step": 10099
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.7133543491363525,
      "learning_rate": 1.8648133119520056e-06,
      "loss": 0.6414,
      "step": 10100
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.280787467956543,
      "learning_rate": 1.863908743725683e-06,
      "loss": 0.131,
      "step": 10101
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.3034584522247314,
      "learning_rate": 1.8630043446796132e-06,
      "loss": 0.1091,
      "step": 10102
    },
    {
      "epoch": 0.72,
      "grad_norm": 15.134702682495117,
      "learning_rate": 1.8621001148625827e-06,
      "loss": 1.0923,
      "step": 10103
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.358170747756958,
      "learning_rate": 1.8611960543233753e-06,
      "loss": 0.0829,
      "step": 10104
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.490667819976807,
      "learning_rate": 1.8602921631107602e-06,
      "loss": 0.932,
      "step": 10105
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.333097457885742,
      "learning_rate": 1.859388441273498e-06,
      "loss": 0.3543,
      "step": 10106
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.522997856140137,
      "learning_rate": 1.8584848888603418e-06,
      "loss": 0.6124,
      "step": 10107
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.556959390640259,
      "learning_rate": 1.8575815059200342e-06,
      "loss": 0.0959,
      "step": 10108
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.429735541343689,
      "learning_rate": 1.8566782925013082e-06,
      "loss": 0.0281,
      "step": 10109
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.123344898223877,
      "learning_rate": 1.8557752486528918e-06,
      "loss": 0.0589,
      "step": 10110
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.29442024230957,
      "learning_rate": 1.8548723744234993e-06,
      "loss": 0.9285,
      "step": 10111
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.574397563934326,
      "learning_rate": 1.8539696698618354e-06,
      "loss": 0.2588,
      "step": 10112
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.685868740081787,
      "learning_rate": 1.8530671350166012e-06,
      "loss": 0.1253,
      "step": 10113
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.838428497314453,
      "learning_rate": 1.8521647699364837e-06,
      "loss": 0.1309,
      "step": 10114
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.166166305541992,
      "learning_rate": 1.8512625746701595e-06,
      "loss": 0.8131,
      "step": 10115
    },
    {
      "epoch": 0.73,
      "grad_norm": 11.677679061889648,
      "learning_rate": 1.8503605492663023e-06,
      "loss": 0.7809,
      "step": 10116
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.938523769378662,
      "learning_rate": 1.849458693773572e-06,
      "loss": 0.1761,
      "step": 10117
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.6876220703125,
      "learning_rate": 1.848557008240619e-06,
      "loss": 0.6702,
      "step": 10118
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.8689873218536377,
      "learning_rate": 1.8476554927160872e-06,
      "loss": 0.149,
      "step": 10119
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0789694786071777,
      "learning_rate": 1.8467541472486089e-06,
      "loss": 0.092,
      "step": 10120
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.960468292236328,
      "learning_rate": 1.8458529718868072e-06,
      "loss": 0.0966,
      "step": 10121
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.173149585723877,
      "learning_rate": 1.8449519666793003e-06,
      "loss": 0.4514,
      "step": 10122
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.5181782245635986,
      "learning_rate": 1.8440511316746924e-06,
      "loss": 0.2281,
      "step": 10123
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.080394744873047,
      "learning_rate": 1.8431504669215777e-06,
      "loss": 0.5214,
      "step": 10124
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.72221565246582,
      "learning_rate": 1.8422499724685483e-06,
      "loss": 0.936,
      "step": 10125
    },
    {
      "epoch": 0.73,
      "grad_norm": 25.63665008544922,
      "learning_rate": 1.84134964836418e-06,
      "loss": 0.5953,
      "step": 10126
    },
    {
      "epoch": 0.73,
      "grad_norm": 11.728827476501465,
      "learning_rate": 1.8404494946570424e-06,
      "loss": 0.4757,
      "step": 10127
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8668053150177002,
      "learning_rate": 1.8395495113956923e-06,
      "loss": 0.0619,
      "step": 10128
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.549419403076172,
      "learning_rate": 1.8386496986286856e-06,
      "loss": 0.3706,
      "step": 10129
    },
    {
      "epoch": 0.73,
      "grad_norm": 15.174421310424805,
      "learning_rate": 1.8377500564045615e-06,
      "loss": 0.497,
      "step": 10130
    },
    {
      "epoch": 0.73,
      "grad_norm": 19.118024826049805,
      "learning_rate": 1.8368505847718516e-06,
      "loss": 0.9382,
      "step": 10131
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.178032398223877,
      "learning_rate": 1.8359512837790794e-06,
      "loss": 0.654,
      "step": 10132
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0884764194488525,
      "learning_rate": 1.8350521534747573e-06,
      "loss": 0.0974,
      "step": 10133
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.562002658843994,
      "learning_rate": 1.834153193907393e-06,
      "loss": 0.7636,
      "step": 10134
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.21342396736145,
      "learning_rate": 1.8332544051254813e-06,
      "loss": 0.1126,
      "step": 10135
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.868183135986328,
      "learning_rate": 1.8323557871775072e-06,
      "loss": 0.8441,
      "step": 10136
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.2140278816223145,
      "learning_rate": 1.8314573401119462e-06,
      "loss": 0.6952,
      "step": 10137
    },
    {
      "epoch": 0.73,
      "grad_norm": 16.529090881347656,
      "learning_rate": 1.8305590639772701e-06,
      "loss": 0.9253,
      "step": 10138
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.487229347229004,
      "learning_rate": 1.8296609588219354e-06,
      "loss": 0.609,
      "step": 10139
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.431450128555298,
      "learning_rate": 1.8287630246943899e-06,
      "loss": 0.1362,
      "step": 10140
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.5565004348754883,
      "learning_rate": 1.8278652616430776e-06,
      "loss": 0.097,
      "step": 10141
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.606727123260498,
      "learning_rate": 1.8269676697164285e-06,
      "loss": 0.1619,
      "step": 10142
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.22963285446167,
      "learning_rate": 1.8260702489628595e-06,
      "loss": 0.1028,
      "step": 10143
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.0408236980438232,
      "learning_rate": 1.8251729994307888e-06,
      "loss": 0.2078,
      "step": 10144
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.091111660003662,
      "learning_rate": 1.8242759211686168e-06,
      "loss": 0.1395,
      "step": 10145
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.223016738891602,
      "learning_rate": 1.8233790142247366e-06,
      "loss": 0.539,
      "step": 10146
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.73771858215332,
      "learning_rate": 1.822482278647536e-06,
      "loss": 0.3974,
      "step": 10147
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.0486414432525635,
      "learning_rate": 1.8215857144853894e-06,
      "loss": 0.1149,
      "step": 10148
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.587425947189331,
      "learning_rate": 1.8206893217866606e-06,
      "loss": 0.1842,
      "step": 10149
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.025674343109131,
      "learning_rate": 1.81979310059971e-06,
      "loss": 0.7903,
      "step": 10150
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.8803887367248535,
      "learning_rate": 1.8188970509728842e-06,
      "loss": 0.6084,
      "step": 10151
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.8983380794525146,
      "learning_rate": 1.8180011729545193e-06,
      "loss": 0.1328,
      "step": 10152
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.542008638381958,
      "learning_rate": 1.8171054665929488e-06,
      "loss": 0.1075,
      "step": 10153
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.121389389038086,
      "learning_rate": 1.8162099319364923e-06,
      "loss": 1.0015,
      "step": 10154
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.9743905067443848,
      "learning_rate": 1.8153145690334544e-06,
      "loss": 0.2107,
      "step": 10155
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.522247791290283,
      "learning_rate": 1.8144193779321435e-06,
      "loss": 0.8112,
      "step": 10156
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.334260940551758,
      "learning_rate": 1.8135243586808482e-06,
      "loss": 0.73,
      "step": 10157
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.156765460968018,
      "learning_rate": 1.8126295113278513e-06,
      "loss": 0.451,
      "step": 10158
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.533977031707764,
      "learning_rate": 1.8117348359214287e-06,
      "loss": 0.1805,
      "step": 10159
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.197171211242676,
      "learning_rate": 1.8108403325098434e-06,
      "loss": 0.1814,
      "step": 10160
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.8471174240112305,
      "learning_rate": 1.809946001141349e-06,
      "loss": 0.7858,
      "step": 10161
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.0896875858306885,
      "learning_rate": 1.8090518418641945e-06,
      "loss": 0.0918,
      "step": 10162
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.5980284214019775,
      "learning_rate": 1.8081578547266148e-06,
      "loss": 0.1392,
      "step": 10163
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.468015670776367,
      "learning_rate": 1.8072640397768371e-06,
      "loss": 0.1311,
      "step": 10164
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.5240609645843506,
      "learning_rate": 1.8063703970630775e-06,
      "loss": 0.1063,
      "step": 10165
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.5576000213623047,
      "learning_rate": 1.8054769266335498e-06,
      "loss": 0.1264,
      "step": 10166
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.4205193519592285,
      "learning_rate": 1.8045836285364465e-06,
      "loss": 1.0291,
      "step": 10167
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.596843957901001,
      "learning_rate": 1.803690502819963e-06,
      "loss": 0.1428,
      "step": 10168
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.392602920532227,
      "learning_rate": 1.8027975495322785e-06,
      "loss": 0.5747,
      "step": 10169
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9002764225006104,
      "learning_rate": 1.8019047687215634e-06,
      "loss": 0.0928,
      "step": 10170
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6906317472457886,
      "learning_rate": 1.8010121604359794e-06,
      "loss": 0.0416,
      "step": 10171
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.337726593017578,
      "learning_rate": 1.8001197247236824e-06,
      "loss": 0.6159,
      "step": 10172
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.058701992034912,
      "learning_rate": 1.7992274616328137e-06,
      "loss": 0.8439,
      "step": 10173
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.589903354644775,
      "learning_rate": 1.798335371211506e-06,
      "loss": 0.8993,
      "step": 10174
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.781294107437134,
      "learning_rate": 1.7974434535078876e-06,
      "loss": 0.1335,
      "step": 10175
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.555706024169922,
      "learning_rate": 1.7965517085700735e-06,
      "loss": 0.8478,
      "step": 10176
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.7075982093811035,
      "learning_rate": 1.7956601364461662e-06,
      "loss": 0.429,
      "step": 10177
    },
    {
      "epoch": 0.73,
      "grad_norm": 15.598296165466309,
      "learning_rate": 1.7947687371842698e-06,
      "loss": 0.9102,
      "step": 10178
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.8650989532470703,
      "learning_rate": 1.7938775108324653e-06,
      "loss": 0.1572,
      "step": 10179
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.228673934936523,
      "learning_rate": 1.7929864574388318e-06,
      "loss": 0.7769,
      "step": 10180
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.46658992767334,
      "learning_rate": 1.7920955770514408e-06,
      "loss": 0.5812,
      "step": 10181
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.2581868171691895,
      "learning_rate": 1.7912048697183516e-06,
      "loss": 0.1599,
      "step": 10182
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.911513328552246,
      "learning_rate": 1.7903143354876117e-06,
      "loss": 1.1635,
      "step": 10183
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.310950756072998,
      "learning_rate": 1.789423974407266e-06,
      "loss": 0.0657,
      "step": 10184
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.2139253616333,
      "learning_rate": 1.7885337865253439e-06,
      "loss": 0.5164,
      "step": 10185
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.8961052894592285,
      "learning_rate": 1.7876437718898665e-06,
      "loss": 0.44,
      "step": 10186
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.220541954040527,
      "learning_rate": 1.7867539305488502e-06,
      "loss": 0.93,
      "step": 10187
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.8570685386657715,
      "learning_rate": 1.7858642625502958e-06,
      "loss": 0.1232,
      "step": 10188
    },
    {
      "epoch": 0.73,
      "grad_norm": 15.377299308776855,
      "learning_rate": 1.784974767942197e-06,
      "loss": 0.8216,
      "step": 10189
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2344017028808594,
      "learning_rate": 1.7840854467725433e-06,
      "loss": 0.135,
      "step": 10190
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8865690231323242,
      "learning_rate": 1.7831962990893047e-06,
      "loss": 0.0685,
      "step": 10191
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.977889895439148,
      "learning_rate": 1.7823073249404481e-06,
      "loss": 0.0831,
      "step": 10192
    },
    {
      "epoch": 0.73,
      "grad_norm": 12.724410057067871,
      "learning_rate": 1.781418524373934e-06,
      "loss": 1.098,
      "step": 10193
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.443281412124634,
      "learning_rate": 1.7805298974377067e-06,
      "loss": 0.0994,
      "step": 10194
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.5337777137756348,
      "learning_rate": 1.7796414441797038e-06,
      "loss": 0.1498,
      "step": 10195
    },
    {
      "epoch": 0.73,
      "grad_norm": 16.867788314819336,
      "learning_rate": 1.7787531646478567e-06,
      "loss": 0.767,
      "step": 10196
    },
    {
      "epoch": 0.73,
      "grad_norm": 11.753350257873535,
      "learning_rate": 1.7778650588900837e-06,
      "loss": 0.5889,
      "step": 10197
    },
    {
      "epoch": 0.73,
      "grad_norm": 13.995455741882324,
      "learning_rate": 1.7769771269542941e-06,
      "loss": 1.2208,
      "step": 10198
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.059389591217041,
      "learning_rate": 1.7760893688883868e-06,
      "loss": 0.8238,
      "step": 10199
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.065647125244141,
      "learning_rate": 1.775201784740257e-06,
      "loss": 0.7627,
      "step": 10200
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1265203952789307,
      "learning_rate": 1.774314374557784e-06,
      "loss": 0.0858,
      "step": 10201
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0398027896881104,
      "learning_rate": 1.7734271383888407e-06,
      "loss": 0.0911,
      "step": 10202
    },
    {
      "epoch": 0.73,
      "grad_norm": 13.757495880126953,
      "learning_rate": 1.77254007628129e-06,
      "loss": 0.9465,
      "step": 10203
    },
    {
      "epoch": 0.73,
      "grad_norm": 14.146418571472168,
      "learning_rate": 1.7716531882829841e-06,
      "loss": 0.8223,
      "step": 10204
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.529725074768066,
      "learning_rate": 1.7707664744417702e-06,
      "loss": 0.9065,
      "step": 10205
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1131465435028076,
      "learning_rate": 1.7698799348054818e-06,
      "loss": 0.0888,
      "step": 10206
    },
    {
      "epoch": 0.73,
      "grad_norm": 24.368715286254883,
      "learning_rate": 1.7689935694219446e-06,
      "loss": 0.6602,
      "step": 10207
    },
    {
      "epoch": 0.73,
      "grad_norm": 13.221158981323242,
      "learning_rate": 1.768107378338972e-06,
      "loss": 1.2153,
      "step": 10208
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.719562530517578,
      "learning_rate": 1.7672213616043748e-06,
      "loss": 0.4633,
      "step": 10209
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.9792680740356445,
      "learning_rate": 1.7663355192659488e-06,
      "loss": 0.2374,
      "step": 10210
    },
    {
      "epoch": 0.73,
      "grad_norm": 12.089959144592285,
      "learning_rate": 1.765449851371479e-06,
      "loss": 0.7395,
      "step": 10211
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.409838914871216,
      "learning_rate": 1.7645643579687488e-06,
      "loss": 0.1118,
      "step": 10212
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.899906158447266,
      "learning_rate": 1.7636790391055242e-06,
      "loss": 0.8494,
      "step": 10213
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.279073715209961,
      "learning_rate": 1.7627938948295654e-06,
      "loss": 0.5798,
      "step": 10214
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.915584087371826,
      "learning_rate": 1.761908925188623e-06,
      "loss": 0.4182,
      "step": 10215
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.86332631111145,
      "learning_rate": 1.7610241302304364e-06,
      "loss": 0.9104,
      "step": 10216
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.338193893432617,
      "learning_rate": 1.7601395100027368e-06,
      "loss": 0.9907,
      "step": 10217
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.528467893600464,
      "learning_rate": 1.759255064553248e-06,
      "loss": 0.1072,
      "step": 10218
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.6266372203826904,
      "learning_rate": 1.7583707939296824e-06,
      "loss": 0.7542,
      "step": 10219
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0911312103271484,
      "learning_rate": 1.7574866981797401e-06,
      "loss": 0.0992,
      "step": 10220
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.21907901763916,
      "learning_rate": 1.7566027773511185e-06,
      "loss": 0.2197,
      "step": 10221
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.137852191925049,
      "learning_rate": 1.7557190314915007e-06,
      "loss": 0.1312,
      "step": 10222
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.249369144439697,
      "learning_rate": 1.7548354606485585e-06,
      "loss": 0.3486,
      "step": 10223
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.1059272289276123,
      "learning_rate": 1.7539520648699615e-06,
      "loss": 0.1192,
      "step": 10224
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.934391498565674,
      "learning_rate": 1.7530688442033638e-06,
      "loss": 0.1462,
      "step": 10225
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.2341959476470947,
      "learning_rate": 1.7521857986964113e-06,
      "loss": 0.1133,
      "step": 10226
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.8781239986419678,
      "learning_rate": 1.7513029283967414e-06,
      "loss": 0.1397,
      "step": 10227
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.382372856140137,
      "learning_rate": 1.750420233351982e-06,
      "loss": 0.5569,
      "step": 10228
    },
    {
      "epoch": 0.73,
      "grad_norm": 13.422158241271973,
      "learning_rate": 1.7495377136097486e-06,
      "loss": 0.8312,
      "step": 10229
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.0164520740509033,
      "learning_rate": 1.7486553692176539e-06,
      "loss": 0.1277,
      "step": 10230
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9572184085845947,
      "learning_rate": 1.7477732002232944e-06,
      "loss": 0.0751,
      "step": 10231
    },
    {
      "epoch": 0.73,
      "grad_norm": 12.475065231323242,
      "learning_rate": 1.7468912066742594e-06,
      "loss": 0.6046,
      "step": 10232
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0660221576690674,
      "learning_rate": 1.7460093886181312e-06,
      "loss": 0.0926,
      "step": 10233
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.973825216293335,
      "learning_rate": 1.7451277461024796e-06,
      "loss": 0.0594,
      "step": 10234
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0297727584838867,
      "learning_rate": 1.744246279174866e-06,
      "loss": 0.1323,
      "step": 10235
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.097625732421875,
      "learning_rate": 1.74336498788284e-06,
      "loss": 0.7804,
      "step": 10236
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.072628974914551,
      "learning_rate": 1.7424838722739478e-06,
      "loss": 0.8216,
      "step": 10237
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.7815709114074707,
      "learning_rate": 1.7416029323957207e-06,
      "loss": 0.7476,
      "step": 10238
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.266840934753418,
      "learning_rate": 1.740722168295681e-06,
      "loss": 0.4932,
      "step": 10239
    },
    {
      "epoch": 0.73,
      "grad_norm": 15.702693939208984,
      "learning_rate": 1.7398415800213436e-06,
      "loss": 0.7402,
      "step": 10240
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.6206624507904053,
      "learning_rate": 1.7389611676202107e-06,
      "loss": 0.1132,
      "step": 10241
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.175665855407715,
      "learning_rate": 1.7380809311397807e-06,
      "loss": 0.5544,
      "step": 10242
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.435992956161499,
      "learning_rate": 1.7372008706275378e-06,
      "loss": 0.2265,
      "step": 10243
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3653571605682373,
      "learning_rate": 1.7363209861309571e-06,
      "loss": 0.1511,
      "step": 10244
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.2517850399017334,
      "learning_rate": 1.7354412776975044e-06,
      "loss": 0.188,
      "step": 10245
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.9663844108581543,
      "learning_rate": 1.7345617453746388e-06,
      "loss": 0.149,
      "step": 10246
    },
    {
      "epoch": 0.73,
      "grad_norm": 14.459416389465332,
      "learning_rate": 1.733682389209807e-06,
      "loss": 0.7555,
      "step": 10247
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.030280113220215,
      "learning_rate": 1.7328032092504455e-06,
      "loss": 0.8081,
      "step": 10248
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.044022560119629,
      "learning_rate": 1.7319242055439844e-06,
      "loss": 0.7955,
      "step": 10249
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.356456756591797,
      "learning_rate": 1.7310453781378429e-06,
      "loss": 0.9303,
      "step": 10250
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.346478462219238,
      "learning_rate": 1.7301667270794293e-06,
      "loss": 0.4639,
      "step": 10251
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.406295776367188,
      "learning_rate": 1.7292882524161435e-06,
      "loss": 0.7174,
      "step": 10252
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.136275291442871,
      "learning_rate": 1.728409954195377e-06,
      "loss": 0.5845,
      "step": 10253
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.7951021194458,
      "learning_rate": 1.7275318324645074e-06,
      "loss": 0.633,
      "step": 10254
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.832559108734131,
      "learning_rate": 1.7266538872709104e-06,
      "loss": 0.0932,
      "step": 10255
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.712590217590332,
      "learning_rate": 1.725776118661946e-06,
      "loss": 0.1181,
      "step": 10256
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.279582023620605,
      "learning_rate": 1.7248985266849644e-06,
      "loss": 0.5026,
      "step": 10257
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.098310947418213,
      "learning_rate": 1.7240211113873123e-06,
      "loss": 1.0061,
      "step": 10258
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.8595330715179443,
      "learning_rate": 1.7231438728163208e-06,
      "loss": 0.1122,
      "step": 10259
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.0387797355651855,
      "learning_rate": 1.7222668110193126e-06,
      "loss": 0.8829,
      "step": 10260
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.6693785190582275,
      "learning_rate": 1.721389926043604e-06,
      "loss": 0.1946,
      "step": 10261
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.2784271240234375,
      "learning_rate": 1.7205132179364992e-06,
      "loss": 0.1085,
      "step": 10262
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.2885854244232178,
      "learning_rate": 1.7196366867452923e-06,
      "loss": 0.149,
      "step": 10263
    },
    {
      "epoch": 0.74,
      "grad_norm": 29.968965530395508,
      "learning_rate": 1.7187603325172697e-06,
      "loss": 1.095,
      "step": 10264
    },
    {
      "epoch": 0.74,
      "grad_norm": 15.407709121704102,
      "learning_rate": 1.7178841552997073e-06,
      "loss": 1.0555,
      "step": 10265
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.980401039123535,
      "learning_rate": 1.7170081551398687e-06,
      "loss": 0.7658,
      "step": 10266
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.8171138763427734,
      "learning_rate": 1.7161323320850154e-06,
      "loss": 0.1479,
      "step": 10267
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.999155282974243,
      "learning_rate": 1.7152566861823928e-06,
      "loss": 0.2,
      "step": 10268
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.0478625297546387,
      "learning_rate": 1.7143812174792368e-06,
      "loss": 0.2408,
      "step": 10269
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.579718828201294,
      "learning_rate": 1.7135059260227794e-06,
      "loss": 0.1204,
      "step": 10270
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.032876491546631,
      "learning_rate": 1.7126308118602365e-06,
      "loss": 0.2065,
      "step": 10271
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.092932939529419,
      "learning_rate": 1.7117558750388187e-06,
      "loss": 0.1085,
      "step": 10272
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.736837863922119,
      "learning_rate": 1.7108811156057226e-06,
      "loss": 0.1173,
      "step": 10273
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.6602234840393066,
      "learning_rate": 1.7100065336081417e-06,
      "loss": 0.092,
      "step": 10274
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.584295272827148,
      "learning_rate": 1.7091321290932556e-06,
      "loss": 0.9714,
      "step": 10275
    },
    {
      "epoch": 0.74,
      "grad_norm": 16.13158416748047,
      "learning_rate": 1.7082579021082347e-06,
      "loss": 0.7632,
      "step": 10276
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.181488037109375,
      "learning_rate": 1.7073838527002396e-06,
      "loss": 0.1126,
      "step": 10277
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.3357412815093994,
      "learning_rate": 1.7065099809164205e-06,
      "loss": 0.1573,
      "step": 10278
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7419521808624268,
      "learning_rate": 1.7056362868039234e-06,
      "loss": 0.0892,
      "step": 10279
    },
    {
      "epoch": 0.74,
      "grad_norm": 11.13042163848877,
      "learning_rate": 1.704762770409879e-06,
      "loss": 1.3418,
      "step": 10280
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.7642340660095215,
      "learning_rate": 1.7038894317814097e-06,
      "loss": 0.093,
      "step": 10281
    },
    {
      "epoch": 0.74,
      "grad_norm": 11.6802396774292,
      "learning_rate": 1.7030162709656273e-06,
      "loss": 0.682,
      "step": 10282
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.7507009506225586,
      "learning_rate": 1.7021432880096395e-06,
      "loss": 0.122,
      "step": 10283
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.76204776763916,
      "learning_rate": 1.7012704829605375e-06,
      "loss": 0.8214,
      "step": 10284
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.6870710849761963,
      "learning_rate": 1.7003978558654055e-06,
      "loss": 0.6651,
      "step": 10285
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.3141560554504395,
      "learning_rate": 1.6995254067713214e-06,
      "loss": 0.5299,
      "step": 10286
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.68880558013916,
      "learning_rate": 1.6986531357253493e-06,
      "loss": 0.5051,
      "step": 10287
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.713095664978027,
      "learning_rate": 1.6977810427745417e-06,
      "loss": 0.4745,
      "step": 10288
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.595766544342041,
      "learning_rate": 1.696909127965949e-06,
      "loss": 0.6418,
      "step": 10289
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.391833305358887,
      "learning_rate": 1.6960373913466054e-06,
      "loss": 0.6555,
      "step": 10290
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.66160774230957,
      "learning_rate": 1.6951658329635373e-06,
      "loss": 0.5142,
      "step": 10291
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.414315938949585,
      "learning_rate": 1.6942944528637644e-06,
      "loss": 0.7041,
      "step": 10292
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.79494571685791,
      "learning_rate": 1.6934232510942933e-06,
      "loss": 0.8673,
      "step": 10293
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.434310436248779,
      "learning_rate": 1.6925522277021195e-06,
      "loss": 0.5749,
      "step": 10294
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.287235736846924,
      "learning_rate": 1.6916813827342355e-06,
      "loss": 0.0534,
      "step": 10295
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.5732622146606445,
      "learning_rate": 1.6908107162376187e-06,
      "loss": 0.3852,
      "step": 10296
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.703486442565918,
      "learning_rate": 1.6899402282592352e-06,
      "loss": 0.6763,
      "step": 10297
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.080732822418213,
      "learning_rate": 1.6890699188460497e-06,
      "loss": 0.1282,
      "step": 10298
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.5232203006744385,
      "learning_rate": 1.6881997880450108e-06,
      "loss": 0.0808,
      "step": 10299
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.524095296859741,
      "learning_rate": 1.6873298359030543e-06,
      "loss": 0.1159,
      "step": 10300
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.761782169342041,
      "learning_rate": 1.6864600624671156e-06,
      "loss": 0.3667,
      "step": 10301
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.7554869651794434,
      "learning_rate": 1.685590467784114e-06,
      "loss": 0.7117,
      "step": 10302
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.6262381076812744,
      "learning_rate": 1.6847210519009593e-06,
      "loss": 0.0452,
      "step": 10303
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.609853744506836,
      "learning_rate": 1.6838518148645566e-06,
      "loss": 0.7779,
      "step": 10304
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.8373193740844727,
      "learning_rate": 1.682982756721796e-06,
      "loss": 0.231,
      "step": 10305
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.2334535121917725,
      "learning_rate": 1.682113877519559e-06,
      "loss": 0.6023,
      "step": 10306
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.022035598754883,
      "learning_rate": 1.6812451773047205e-06,
      "loss": 0.8563,
      "step": 10307
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.524369478225708,
      "learning_rate": 1.6803766561241425e-06,
      "loss": 0.1499,
      "step": 10308
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.028614044189453,
      "learning_rate": 1.6795083140246788e-06,
      "loss": 0.5396,
      "step": 10309
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.6096720695495605,
      "learning_rate": 1.6786401510531708e-06,
      "loss": 0.7757,
      "step": 10310
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.9131886959075928,
      "learning_rate": 1.6777721672564585e-06,
      "loss": 0.6453,
      "step": 10311
    },
    {
      "epoch": 0.74,
      "grad_norm": 19.867273330688477,
      "learning_rate": 1.6769043626813585e-06,
      "loss": 1.0776,
      "step": 10312
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3455770015716553,
      "learning_rate": 1.6760367373746917e-06,
      "loss": 0.0966,
      "step": 10313
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.77061939239502,
      "learning_rate": 1.6751692913832608e-06,
      "loss": 0.7005,
      "step": 10314
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.306014060974121,
      "learning_rate": 1.674302024753861e-06,
      "loss": 0.4557,
      "step": 10315
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.5397162437438965,
      "learning_rate": 1.6734349375332775e-06,
      "loss": 0.0451,
      "step": 10316
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.0085501670837402,
      "learning_rate": 1.672568029768289e-06,
      "loss": 0.1117,
      "step": 10317
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.410940170288086,
      "learning_rate": 1.6717013015056598e-06,
      "loss": 0.2951,
      "step": 10318
    },
    {
      "epoch": 0.74,
      "grad_norm": 11.34564208984375,
      "learning_rate": 1.6708347527921459e-06,
      "loss": 0.7003,
      "step": 10319
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.8014748096466064,
      "learning_rate": 1.6699683836744973e-06,
      "loss": 0.1566,
      "step": 10320
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.263473033905029,
      "learning_rate": 1.6691021941994489e-06,
      "loss": 0.7346,
      "step": 10321
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3812763690948486,
      "learning_rate": 1.6682361844137279e-06,
      "loss": 0.0925,
      "step": 10322
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.126013278961182,
      "learning_rate": 1.6673703543640562e-06,
      "loss": 0.8511,
      "step": 10323
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.177131652832031,
      "learning_rate": 1.6665047040971383e-06,
      "loss": 0.6182,
      "step": 10324
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.731011152267456,
      "learning_rate": 1.665639233659671e-06,
      "loss": 0.7236,
      "step": 10325
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.953733205795288,
      "learning_rate": 1.6647739430983483e-06,
      "loss": 0.1116,
      "step": 10326
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.619370937347412,
      "learning_rate": 1.6639088324598463e-06,
      "loss": 0.114,
      "step": 10327
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3455824851989746,
      "learning_rate": 1.663043901790834e-06,
      "loss": 0.0719,
      "step": 10328
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.1923928260803223,
      "learning_rate": 1.6621791511379732e-06,
      "loss": 0.1826,
      "step": 10329
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.5485823154449463,
      "learning_rate": 1.661314580547913e-06,
      "loss": 0.0916,
      "step": 10330
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.8398051261901855,
      "learning_rate": 1.660450190067292e-06,
      "loss": 0.3765,
      "step": 10331
    },
    {
      "epoch": 0.74,
      "grad_norm": 15.025010108947754,
      "learning_rate": 1.6595859797427437e-06,
      "loss": 1.1315,
      "step": 10332
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.579587459564209,
      "learning_rate": 1.6587219496208879e-06,
      "loss": 0.2185,
      "step": 10333
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3955187797546387,
      "learning_rate": 1.657858099748334e-06,
      "loss": 0.078,
      "step": 10334
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7894654273986816,
      "learning_rate": 1.6569944301716877e-06,
      "loss": 0.0485,
      "step": 10335
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.370172500610352,
      "learning_rate": 1.656130940937537e-06,
      "loss": 0.7484,
      "step": 10336
    },
    {
      "epoch": 0.74,
      "grad_norm": 15.166187286376953,
      "learning_rate": 1.6552676320924627e-06,
      "loss": 0.6592,
      "step": 10337
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8410053253173828,
      "learning_rate": 1.654404503683041e-06,
      "loss": 0.0808,
      "step": 10338
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.539388179779053,
      "learning_rate": 1.653541555755832e-06,
      "loss": 0.857,
      "step": 10339
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.9696462154388428,
      "learning_rate": 1.652678788357388e-06,
      "loss": 0.7359,
      "step": 10340
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.1923837661743164,
      "learning_rate": 1.6518162015342542e-06,
      "loss": 0.1566,
      "step": 10341
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.9469504356384277,
      "learning_rate": 1.6509537953329635e-06,
      "loss": 0.0998,
      "step": 10342
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.7636053562164307,
      "learning_rate": 1.650091569800038e-06,
      "loss": 0.8634,
      "step": 10343
    },
    {
      "epoch": 0.74,
      "grad_norm": 18.702247619628906,
      "learning_rate": 1.649229524981991e-06,
      "loss": 0.937,
      "step": 10344
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.1988372802734375,
      "learning_rate": 1.6483676609253297e-06,
      "loss": 0.8258,
      "step": 10345
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.3487277030944824,
      "learning_rate": 1.6475059776765456e-06,
      "loss": 0.1245,
      "step": 10346
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.259588241577148,
      "learning_rate": 1.6466444752821248e-06,
      "loss": 0.4627,
      "step": 10347
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.639710426330566,
      "learning_rate": 1.645783153788541e-06,
      "loss": 0.5843,
      "step": 10348
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.124274969100952,
      "learning_rate": 1.6449220132422582e-06,
      "loss": 0.0886,
      "step": 10349
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.152754783630371,
      "learning_rate": 1.644061053689735e-06,
      "loss": 0.2625,
      "step": 10350
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.845047950744629,
      "learning_rate": 1.6432002751774151e-06,
      "loss": 0.1005,
      "step": 10351
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0497941970825195,
      "learning_rate": 1.6423396777517337e-06,
      "loss": 0.0623,
      "step": 10352
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.980561256408691,
      "learning_rate": 1.6414792614591157e-06,
      "loss": 0.5183,
      "step": 10353
    },
    {
      "epoch": 0.74,
      "grad_norm": 13.41122817993164,
      "learning_rate": 1.640619026345981e-06,
      "loss": 0.5947,
      "step": 10354
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.371414184570312,
      "learning_rate": 1.6397589724587331e-06,
      "loss": 0.6083,
      "step": 10355
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.734468460083008,
      "learning_rate": 1.6388990998437682e-06,
      "loss": 0.8929,
      "step": 10356
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.750908851623535,
      "learning_rate": 1.638039408547476e-06,
      "loss": 0.354,
      "step": 10357
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.269925355911255,
      "learning_rate": 1.6371798986162325e-06,
      "loss": 0.1161,
      "step": 10358
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.14490032196045,
      "learning_rate": 1.6363205700964036e-06,
      "loss": 0.7834,
      "step": 10359
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.7490835189819336,
      "learning_rate": 1.6354614230343484e-06,
      "loss": 0.1218,
      "step": 10360
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.493361473083496,
      "learning_rate": 1.6346024574764136e-06,
      "loss": 0.367,
      "step": 10361
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.8238537311553955,
      "learning_rate": 1.6337436734689361e-06,
      "loss": 0.7629,
      "step": 10362
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.34179973602295,
      "learning_rate": 1.6328850710582467e-06,
      "loss": 0.6126,
      "step": 10363
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.818769931793213,
      "learning_rate": 1.6320266502906629e-06,
      "loss": 0.2864,
      "step": 10364
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.147421836853027,
      "learning_rate": 1.631168411212491e-06,
      "loss": 0.5106,
      "step": 10365
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.921613693237305,
      "learning_rate": 1.6303103538700331e-06,
      "loss": 0.8346,
      "step": 10366
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.303135871887207,
      "learning_rate": 1.6294524783095772e-06,
      "loss": 0.7053,
      "step": 10367
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.871468544006348,
      "learning_rate": 1.6285947845773998e-06,
      "loss": 0.4071,
      "step": 10368
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.406195640563965,
      "learning_rate": 1.6277372727197739e-06,
      "loss": 0.6607,
      "step": 10369
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8038218021392822,
      "learning_rate": 1.626879942782958e-06,
      "loss": 0.0437,
      "step": 10370
    },
    {
      "epoch": 0.74,
      "grad_norm": 11.90097427368164,
      "learning_rate": 1.6260227948132006e-06,
      "loss": 0.5239,
      "step": 10371
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.460487127304077,
      "learning_rate": 1.6251658288567423e-06,
      "loss": 0.1109,
      "step": 10372
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.550098419189453,
      "learning_rate": 1.624309044959813e-06,
      "loss": 0.7772,
      "step": 10373
    },
    {
      "epoch": 0.74,
      "grad_norm": 19.353214263916016,
      "learning_rate": 1.6234524431686315e-06,
      "loss": 0.61,
      "step": 10374
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.956148862838745,
      "learning_rate": 1.622596023529412e-06,
      "loss": 0.1804,
      "step": 10375
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.935354232788086,
      "learning_rate": 1.621739786088352e-06,
      "loss": 0.1703,
      "step": 10376
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.540036916732788,
      "learning_rate": 1.620883730891642e-06,
      "loss": 0.172,
      "step": 10377
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.7725541591644287,
      "learning_rate": 1.6200278579854655e-06,
      "loss": 0.1873,
      "step": 10378
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.146663188934326,
      "learning_rate": 1.6191721674159921e-06,
      "loss": 0.9409,
      "step": 10379
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.877920150756836,
      "learning_rate": 1.6183166592293837e-06,
      "loss": 0.8669,
      "step": 10380
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.350201606750488,
      "learning_rate": 1.6174613334717893e-06,
      "loss": 0.4499,
      "step": 10381
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.926713943481445,
      "learning_rate": 1.6166061901893537e-06,
      "loss": 0.7233,
      "step": 10382
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.822056770324707,
      "learning_rate": 1.6157512294282074e-06,
      "loss": 0.4198,
      "step": 10383
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.725121974945068,
      "learning_rate": 1.6148964512344722e-06,
      "loss": 0.8468,
      "step": 10384
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.362938165664673,
      "learning_rate": 1.6140418556542597e-06,
      "loss": 0.0858,
      "step": 10385
    },
    {
      "epoch": 0.74,
      "grad_norm": 11.074581146240234,
      "learning_rate": 1.6131874427336714e-06,
      "loss": 0.7322,
      "step": 10386
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.2786495685577393,
      "learning_rate": 1.6123332125188024e-06,
      "loss": 0.1402,
      "step": 10387
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.893261432647705,
      "learning_rate": 1.6114791650557338e-06,
      "loss": 0.8003,
      "step": 10388
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6034538745880127,
      "learning_rate": 1.6106253003905376e-06,
      "loss": 0.0441,
      "step": 10389
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.593542575836182,
      "learning_rate": 1.6097716185692758e-06,
      "loss": 0.8573,
      "step": 10390
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.6612792015075684,
      "learning_rate": 1.6089181196380034e-06,
      "loss": 0.176,
      "step": 10391
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.137513637542725,
      "learning_rate": 1.6080648036427631e-06,
      "loss": 0.6465,
      "step": 10392
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.84858512878418,
      "learning_rate": 1.6072116706295854e-06,
      "loss": 0.5121,
      "step": 10393
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.5107555389404297,
      "learning_rate": 1.6063587206444976e-06,
      "loss": 0.0979,
      "step": 10394
    },
    {
      "epoch": 0.75,
      "grad_norm": 14.808280944824219,
      "learning_rate": 1.6055059537335128e-06,
      "loss": 1.0568,
      "step": 10395
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.15757155418396,
      "learning_rate": 1.6046533699426297e-06,
      "loss": 0.0695,
      "step": 10396
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.110457420349121,
      "learning_rate": 1.6038009693178464e-06,
      "loss": 0.1304,
      "step": 10397
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.5782790184021,
      "learning_rate": 1.6029487519051462e-06,
      "loss": 0.5836,
      "step": 10398
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.906484603881836,
      "learning_rate": 1.6020967177505003e-06,
      "loss": 0.3822,
      "step": 10399
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.323448181152344,
      "learning_rate": 1.6012448668998759e-06,
      "loss": 0.0631,
      "step": 10400
    },
    {
      "epoch": 0.75,
      "grad_norm": 15.57811164855957,
      "learning_rate": 1.6003931993992267e-06,
      "loss": 0.6706,
      "step": 10401
    },
    {
      "epoch": 0.75,
      "grad_norm": 15.078125,
      "learning_rate": 1.5995417152944942e-06,
      "loss": 0.4015,
      "step": 10402
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.760897159576416,
      "learning_rate": 1.5986904146316162e-06,
      "loss": 0.8469,
      "step": 10403
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.380554437637329,
      "learning_rate": 1.597839297456516e-06,
      "loss": 0.0953,
      "step": 10404
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.808780670166016,
      "learning_rate": 1.5969883638151063e-06,
      "loss": 0.8925,
      "step": 10405
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.3304872512817383,
      "learning_rate": 1.5961376137532946e-06,
      "loss": 0.1808,
      "step": 10406
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.9422590732574463,
      "learning_rate": 1.5952870473169757e-06,
      "loss": 0.6633,
      "step": 10407
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.2678258419036865,
      "learning_rate": 1.5944366645520304e-06,
      "loss": 0.2197,
      "step": 10408
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.37827467918396,
      "learning_rate": 1.5935864655043376e-06,
      "loss": 0.1008,
      "step": 10409
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.7125229835510254,
      "learning_rate": 1.592736450219761e-06,
      "loss": 0.1757,
      "step": 10410
    },
    {
      "epoch": 0.75,
      "grad_norm": 43.96757888793945,
      "learning_rate": 1.5918866187441546e-06,
      "loss": 0.7795,
      "step": 10411
    },
    {
      "epoch": 0.75,
      "grad_norm": 23.72187042236328,
      "learning_rate": 1.5910369711233653e-06,
      "loss": 0.8698,
      "step": 10412
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.345871448516846,
      "learning_rate": 1.5901875074032285e-06,
      "loss": 0.2625,
      "step": 10413
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.754213809967041,
      "learning_rate": 1.5893382276295672e-06,
      "loss": 0.191,
      "step": 10414
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.672075033187866,
      "learning_rate": 1.5884891318482e-06,
      "loss": 0.7255,
      "step": 10415
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.205221176147461,
      "learning_rate": 1.5876402201049307e-06,
      "loss": 0.796,
      "step": 10416
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8136839866638184,
      "learning_rate": 1.5867914924455553e-06,
      "loss": 0.041,
      "step": 10417
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.1015748977661133,
      "learning_rate": 1.5859429489158574e-06,
      "loss": 0.1113,
      "step": 10418
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.443256139755249,
      "learning_rate": 1.5850945895616176e-06,
      "loss": 0.1846,
      "step": 10419
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.254698753356934,
      "learning_rate": 1.5842464144285957e-06,
      "loss": 0.4928,
      "step": 10420
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.864875078201294,
      "learning_rate": 1.5833984235625515e-06,
      "loss": 0.7105,
      "step": 10421
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.600416660308838,
      "learning_rate": 1.5825506170092303e-06,
      "loss": 0.1966,
      "step": 10422
    },
    {
      "epoch": 0.75,
      "grad_norm": 16.89422035217285,
      "learning_rate": 1.5817029948143674e-06,
      "loss": 0.8301,
      "step": 10423
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3729567527770996,
      "learning_rate": 1.5808555570236877e-06,
      "loss": 0.1223,
      "step": 10424
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.7031021118164062,
      "learning_rate": 1.5800083036829096e-06,
      "loss": 0.0913,
      "step": 10425
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.090494632720947,
      "learning_rate": 1.5791612348377383e-06,
      "loss": 0.8571,
      "step": 10426
    },
    {
      "epoch": 0.75,
      "grad_norm": 14.291454315185547,
      "learning_rate": 1.5783143505338683e-06,
      "loss": 0.5531,
      "step": 10427
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.8937156200408936,
      "learning_rate": 1.577467650816989e-06,
      "loss": 0.1293,
      "step": 10428
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.102302551269531,
      "learning_rate": 1.576621135732775e-06,
      "loss": 0.3442,
      "step": 10429
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.304465293884277,
      "learning_rate": 1.5757748053268905e-06,
      "loss": 0.7356,
      "step": 10430
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.805392742156982,
      "learning_rate": 1.5749286596449976e-06,
      "loss": 0.9102,
      "step": 10431
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.607593536376953,
      "learning_rate": 1.5740826987327369e-06,
      "loss": 0.3364,
      "step": 10432
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.642560005187988,
      "learning_rate": 1.5732369226357452e-06,
      "loss": 0.5761,
      "step": 10433
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0551748275756836,
      "learning_rate": 1.5723913313996526e-06,
      "loss": 0.0758,
      "step": 10434
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.311295509338379,
      "learning_rate": 1.5715459250700727e-06,
      "loss": 0.6968,
      "step": 10435
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8671674728393555,
      "learning_rate": 1.5707007036926114e-06,
      "loss": 0.0924,
      "step": 10436
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3214774131774902,
      "learning_rate": 1.5698556673128684e-06,
      "loss": 0.0631,
      "step": 10437
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.0204105377197266,
      "learning_rate": 1.5690108159764283e-06,
      "loss": 0.1209,
      "step": 10438
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.713726997375488,
      "learning_rate": 1.568166149728866e-06,
      "loss": 0.5936,
      "step": 10439
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.035340309143066,
      "learning_rate": 1.5673216686157517e-06,
      "loss": 0.5751,
      "step": 10440
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.7091829776763916,
      "learning_rate": 1.5664773726826393e-06,
      "loss": 0.1678,
      "step": 10441
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.0723185539245605,
      "learning_rate": 1.5656332619750746e-06,
      "loss": 0.7616,
      "step": 10442
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.6965456008911133,
      "learning_rate": 1.5647893365385991e-06,
      "loss": 0.1858,
      "step": 10443
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.943791151046753,
      "learning_rate": 1.5639455964187345e-06,
      "loss": 0.7924,
      "step": 10444
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.5029001235961914,
      "learning_rate": 1.5631020416609966e-06,
      "loss": 0.0902,
      "step": 10445
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.1886425018310547,
      "learning_rate": 1.5622586723108963e-06,
      "loss": 0.1637,
      "step": 10446
    },
    {
      "epoch": 0.75,
      "grad_norm": 21.795032501220703,
      "learning_rate": 1.5614154884139281e-06,
      "loss": 0.8079,
      "step": 10447
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.5082221031188965,
      "learning_rate": 1.5605724900155766e-06,
      "loss": 0.6458,
      "step": 10448
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.558080673217773,
      "learning_rate": 1.559729677161322e-06,
      "loss": 0.7201,
      "step": 10449
    },
    {
      "epoch": 0.75,
      "grad_norm": 17.18353271484375,
      "learning_rate": 1.558887049896629e-06,
      "loss": 0.986,
      "step": 10450
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.444448232650757,
      "learning_rate": 1.5580446082669532e-06,
      "loss": 0.1769,
      "step": 10451
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.264302253723145,
      "learning_rate": 1.5572023523177438e-06,
      "loss": 0.4164,
      "step": 10452
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.4295952320098877,
      "learning_rate": 1.5563602820944357e-06,
      "loss": 0.1197,
      "step": 10453
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9227335453033447,
      "learning_rate": 1.5555183976424553e-06,
      "loss": 0.1155,
      "step": 10454
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.092609882354736,
      "learning_rate": 1.5546766990072176e-06,
      "loss": 0.8751,
      "step": 10455
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.404550552368164,
      "learning_rate": 1.5538351862341345e-06,
      "loss": 0.6558,
      "step": 10456
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.057841300964355,
      "learning_rate": 1.5529938593685951e-06,
      "loss": 0.3433,
      "step": 10457
    },
    {
      "epoch": 0.75,
      "grad_norm": 20.607675552368164,
      "learning_rate": 1.5521527184559908e-06,
      "loss": 0.8898,
      "step": 10458
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.32245922088623,
      "learning_rate": 1.551311763541697e-06,
      "loss": 0.8438,
      "step": 10459
    },
    {
      "epoch": 0.75,
      "grad_norm": 14.493500709533691,
      "learning_rate": 1.5504709946710795e-06,
      "loss": 0.8619,
      "step": 10460
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.1868607997894287,
      "learning_rate": 1.5496304118894922e-06,
      "loss": 0.1016,
      "step": 10461
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.371746778488159,
      "learning_rate": 1.5487900152422864e-06,
      "loss": 0.1063,
      "step": 10462
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.6002261638641357,
      "learning_rate": 1.5479498047747953e-06,
      "loss": 0.2086,
      "step": 10463
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.542469024658203,
      "learning_rate": 1.5471097805323442e-06,
      "loss": 0.067,
      "step": 10464
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.251152992248535,
      "learning_rate": 1.5462699425602517e-06,
      "loss": 0.219,
      "step": 10465
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.6970505714416504,
      "learning_rate": 1.5454302909038232e-06,
      "loss": 0.1129,
      "step": 10466
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.725602626800537,
      "learning_rate": 1.5445908256083525e-06,
      "loss": 0.1105,
      "step": 10467
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.599733829498291,
      "learning_rate": 1.5437515467191305e-06,
      "loss": 0.5054,
      "step": 10468
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8690513372421265,
      "learning_rate": 1.5429124542814283e-06,
      "loss": 0.0775,
      "step": 10469
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.4353504180908203,
      "learning_rate": 1.5420735483405115e-06,
      "loss": 0.2072,
      "step": 10470
    },
    {
      "epoch": 0.75,
      "grad_norm": 29.016681671142578,
      "learning_rate": 1.5412348289416396e-06,
      "loss": 0.8328,
      "step": 10471
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.4817919731140137,
      "learning_rate": 1.5403962961300566e-06,
      "loss": 0.228,
      "step": 10472
    },
    {
      "epoch": 0.75,
      "grad_norm": 14.672749519348145,
      "learning_rate": 1.5395579499509966e-06,
      "loss": 0.4756,
      "step": 10473
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.5629770755767822,
      "learning_rate": 1.5387197904496887e-06,
      "loss": 0.7664,
      "step": 10474
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7137987613677979,
      "learning_rate": 1.5378818176713455e-06,
      "loss": 0.0448,
      "step": 10475
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.73962116241455,
      "learning_rate": 1.5370440316611723e-06,
      "loss": 0.6479,
      "step": 10476
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.308610916137695,
      "learning_rate": 1.5362064324643676e-06,
      "loss": 0.7272,
      "step": 10477
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.4914588928222656,
      "learning_rate": 1.535369020126114e-06,
      "loss": 0.1894,
      "step": 10478
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5754581689834595,
      "learning_rate": 1.5345317946915867e-06,
      "loss": 0.0594,
      "step": 10479
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.568896293640137,
      "learning_rate": 1.533694756205954e-06,
      "loss": 0.6206,
      "step": 10480
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2288174629211426,
      "learning_rate": 1.5328579047143672e-06,
      "loss": 0.081,
      "step": 10481
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.7861738204956055,
      "learning_rate": 1.5320212402619704e-06,
      "loss": 0.1403,
      "step": 10482
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.163247585296631,
      "learning_rate": 1.5311847628939031e-06,
      "loss": 0.8384,
      "step": 10483
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.252385139465332,
      "learning_rate": 1.5303484726552876e-06,
      "loss": 0.6615,
      "step": 10484
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.794902324676514,
      "learning_rate": 1.5295123695912366e-06,
      "loss": 0.43,
      "step": 10485
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.165308952331543,
      "learning_rate": 1.5286764537468585e-06,
      "loss": 0.543,
      "step": 10486
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.916406631469727,
      "learning_rate": 1.527840725167246e-06,
      "loss": 0.6943,
      "step": 10487
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9102437496185303,
      "learning_rate": 1.5270051838974837e-06,
      "loss": 0.0813,
      "step": 10488
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.798925876617432,
      "learning_rate": 1.5261698299826444e-06,
      "loss": 0.411,
      "step": 10489
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7822964191436768,
      "learning_rate": 1.5253346634677946e-06,
      "loss": 0.0567,
      "step": 10490
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.2499494552612305,
      "learning_rate": 1.524499684397988e-06,
      "loss": 0.1778,
      "step": 10491
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.642024278640747,
      "learning_rate": 1.5236648928182673e-06,
      "loss": 0.1044,
      "step": 10492
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.510080337524414,
      "learning_rate": 1.5228302887736674e-06,
      "loss": 0.6995,
      "step": 10493
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4673831462860107,
      "learning_rate": 1.5219958723092099e-06,
      "loss": 0.0742,
      "step": 10494
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.460578441619873,
      "learning_rate": 1.5211616434699117e-06,
      "loss": 0.4519,
      "step": 10495
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.535463333129883,
      "learning_rate": 1.520327602300775e-06,
      "loss": 0.1279,
      "step": 10496
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.229909896850586,
      "learning_rate": 1.5194937488467926e-06,
      "loss": 0.9054,
      "step": 10497
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.8482043743133545,
      "learning_rate": 1.5186600831529463e-06,
      "loss": 0.6289,
      "step": 10498
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.859844207763672,
      "learning_rate": 1.5178266052642132e-06,
      "loss": 0.1881,
      "step": 10499
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.916865110397339,
      "learning_rate": 1.5169933152255535e-06,
      "loss": 0.7921,
      "step": 10500
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7228341102600098,
      "learning_rate": 1.5161602130819186e-06,
      "loss": 0.042,
      "step": 10501
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.401401996612549,
      "learning_rate": 1.5153272988782553e-06,
      "loss": 0.8618,
      "step": 10502
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.0361995697021484,
      "learning_rate": 1.514494572659494e-06,
      "loss": 0.1603,
      "step": 10503
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.989665150642395,
      "learning_rate": 1.5136620344705566e-06,
      "loss": 0.0875,
      "step": 10504
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9947891235351562,
      "learning_rate": 1.5128296843563568e-06,
      "loss": 0.181,
      "step": 10505
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.8373987674713135,
      "learning_rate": 1.5119975223617949e-06,
      "loss": 0.0988,
      "step": 10506
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3435122966766357,
      "learning_rate": 1.5111655485317628e-06,
      "loss": 0.1303,
      "step": 10507
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.718926429748535,
      "learning_rate": 1.5103337629111453e-06,
      "loss": 0.6612,
      "step": 10508
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.217465877532959,
      "learning_rate": 1.5095021655448116e-06,
      "loss": 0.1098,
      "step": 10509
    },
    {
      "epoch": 0.75,
      "grad_norm": 19.856306076049805,
      "learning_rate": 1.5086707564776227e-06,
      "loss": 0.7038,
      "step": 10510
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.135054588317871,
      "learning_rate": 1.5078395357544323e-06,
      "loss": 0.2791,
      "step": 10511
    },
    {
      "epoch": 0.75,
      "grad_norm": 18.632076263427734,
      "learning_rate": 1.5070085034200804e-06,
      "loss": 1.0545,
      "step": 10512
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.405737400054932,
      "learning_rate": 1.5061776595193966e-06,
      "loss": 0.9678,
      "step": 10513
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.595071315765381,
      "learning_rate": 1.5053470040972051e-06,
      "loss": 0.1209,
      "step": 10514
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3603036403656006,
      "learning_rate": 1.5045165371983145e-06,
      "loss": 0.0729,
      "step": 10515
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.8730406761169434,
      "learning_rate": 1.5036862588675254e-06,
      "loss": 0.1526,
      "step": 10516
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8891485929489136,
      "learning_rate": 1.5028561691496286e-06,
      "loss": 0.0498,
      "step": 10517
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.594343900680542,
      "learning_rate": 1.5020262680894043e-06,
      "loss": 0.1356,
      "step": 10518
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.975560188293457,
      "learning_rate": 1.5011965557316199e-06,
      "loss": 0.8329,
      "step": 10519
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.493906021118164,
      "learning_rate": 1.50036703212104e-06,
      "loss": 0.1166,
      "step": 10520
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.538829803466797,
      "learning_rate": 1.4995376973024118e-06,
      "loss": 0.6409,
      "step": 10521
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.430436849594116,
      "learning_rate": 1.498708551320473e-06,
      "loss": 0.1725,
      "step": 10522
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.170722484588623,
      "learning_rate": 1.4978795942199564e-06,
      "loss": 0.0982,
      "step": 10523
    },
    {
      "epoch": 0.75,
      "grad_norm": 15.778495788574219,
      "learning_rate": 1.4970508260455796e-06,
      "loss": 0.6725,
      "step": 10524
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9291021823883057,
      "learning_rate": 1.496222246842051e-06,
      "loss": 0.168,
      "step": 10525
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.115370512008667,
      "learning_rate": 1.4953938566540683e-06,
      "loss": 0.1069,
      "step": 10526
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.0493364334106445,
      "learning_rate": 1.4945656555263233e-06,
      "loss": 0.7767,
      "step": 10527
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9118008613586426,
      "learning_rate": 1.4937376435034917e-06,
      "loss": 0.1587,
      "step": 10528
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.829601526260376,
      "learning_rate": 1.4929098206302429e-06,
      "loss": 0.145,
      "step": 10529
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2190908193588257,
      "learning_rate": 1.4920821869512337e-06,
      "loss": 0.0354,
      "step": 10530
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.993612289428711,
      "learning_rate": 1.4912547425111113e-06,
      "loss": 0.6886,
      "step": 10531
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.780229330062866,
      "learning_rate": 1.4904274873545155e-06,
      "loss": 0.1769,
      "step": 10532
    },
    {
      "epoch": 0.76,
      "grad_norm": 16.062870025634766,
      "learning_rate": 1.4896004215260723e-06,
      "loss": 1.2005,
      "step": 10533
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.4178905487060547,
      "learning_rate": 1.488773545070399e-06,
      "loss": 0.1126,
      "step": 10534
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.2676761150360107,
      "learning_rate": 1.4879468580321006e-06,
      "loss": 0.1826,
      "step": 10535
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.091688632965088,
      "learning_rate": 1.4871203604557767e-06,
      "loss": 0.1546,
      "step": 10536
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.972826957702637,
      "learning_rate": 1.486294052386012e-06,
      "loss": 0.6032,
      "step": 10537
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.195996284484863,
      "learning_rate": 1.4854679338673821e-06,
      "loss": 0.6151,
      "step": 10538
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.23429274559021,
      "learning_rate": 1.4846420049444553e-06,
      "loss": 0.0939,
      "step": 10539
    },
    {
      "epoch": 0.76,
      "grad_norm": 26.069398880004883,
      "learning_rate": 1.4838162656617878e-06,
      "loss": 0.7833,
      "step": 10540
    },
    {
      "epoch": 0.76,
      "grad_norm": 18.68767738342285,
      "learning_rate": 1.48299071606392e-06,
      "loss": 0.6919,
      "step": 10541
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.3576254844665527,
      "learning_rate": 1.482165356195392e-06,
      "loss": 0.2081,
      "step": 10542
    },
    {
      "epoch": 0.76,
      "grad_norm": 21.76236343383789,
      "learning_rate": 1.4813401861007277e-06,
      "loss": 0.5932,
      "step": 10543
    },
    {
      "epoch": 0.76,
      "grad_norm": 13.198902130126953,
      "learning_rate": 1.4805152058244392e-06,
      "loss": 1.0238,
      "step": 10544
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.326185941696167,
      "learning_rate": 1.4796904154110353e-06,
      "loss": 0.1446,
      "step": 10545
    },
    {
      "epoch": 0.76,
      "grad_norm": 12.640639305114746,
      "learning_rate": 1.4788658149050083e-06,
      "loss": 1.0403,
      "step": 10546
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.944557189941406,
      "learning_rate": 1.4780414043508412e-06,
      "loss": 0.2905,
      "step": 10547
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.187800407409668,
      "learning_rate": 1.4772171837930104e-06,
      "loss": 0.0402,
      "step": 10548
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.300849914550781,
      "learning_rate": 1.476393153275978e-06,
      "loss": 0.5826,
      "step": 10549
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2403345108032227,
      "learning_rate": 1.4755693128441956e-06,
      "loss": 0.1734,
      "step": 10550
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8932762145996094,
      "learning_rate": 1.4747456625421096e-06,
      "loss": 0.1551,
      "step": 10551
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.0981035232543945,
      "learning_rate": 1.4739222024141536e-06,
      "loss": 0.8829,
      "step": 10552
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1464993953704834,
      "learning_rate": 1.473098932504744e-06,
      "loss": 0.0917,
      "step": 10553
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.811891555786133,
      "learning_rate": 1.4722758528582986e-06,
      "loss": 0.1229,
      "step": 10554
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.655824661254883,
      "learning_rate": 1.4714529635192182e-06,
      "loss": 0.6053,
      "step": 10555
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.457088470458984,
      "learning_rate": 1.470630264531892e-06,
      "loss": 1.286,
      "step": 10556
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.157529592514038,
      "learning_rate": 1.4698077559407059e-06,
      "loss": 0.1277,
      "step": 10557
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.654557228088379,
      "learning_rate": 1.468985437790028e-06,
      "loss": 0.7431,
      "step": 10558
    },
    {
      "epoch": 0.76,
      "grad_norm": 24.352811813354492,
      "learning_rate": 1.4681633101242194e-06,
      "loss": 1.3864,
      "step": 10559
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5141048431396484,
      "learning_rate": 1.4673413729876324e-06,
      "loss": 0.0372,
      "step": 10560
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.121403455734253,
      "learning_rate": 1.4665196264246067e-06,
      "loss": 0.0922,
      "step": 10561
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.574026107788086,
      "learning_rate": 1.4656980704794727e-06,
      "loss": 0.7137,
      "step": 10562
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.901690483093262,
      "learning_rate": 1.4648767051965473e-06,
      "loss": 0.9702,
      "step": 10563
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0412909984588623,
      "learning_rate": 1.4640555306201464e-06,
      "loss": 0.0944,
      "step": 10564
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3544843196868896,
      "learning_rate": 1.463234546794562e-06,
      "loss": 0.1137,
      "step": 10565
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.044328212738037,
      "learning_rate": 1.4624137537640877e-06,
      "loss": 0.872,
      "step": 10566
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.548328161239624,
      "learning_rate": 1.4615931515730008e-06,
      "loss": 0.8838,
      "step": 10567
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.282874345779419,
      "learning_rate": 1.4607727402655703e-06,
      "loss": 0.1479,
      "step": 10568
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.367234945297241,
      "learning_rate": 1.4599525198860515e-06,
      "loss": 0.0718,
      "step": 10569
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3639330863952637,
      "learning_rate": 1.459132490478697e-06,
      "loss": 0.1044,
      "step": 10570
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.051175594329834,
      "learning_rate": 1.4583126520877406e-06,
      "loss": 0.1173,
      "step": 10571
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.102031946182251,
      "learning_rate": 1.45749300475741e-06,
      "loss": 0.055,
      "step": 10572
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.165626049041748,
      "learning_rate": 1.4566735485319233e-06,
      "loss": 0.7512,
      "step": 10573
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5225269794464111,
      "learning_rate": 1.4558542834554873e-06,
      "loss": 0.0387,
      "step": 10574
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.6855432987213135,
      "learning_rate": 1.4550352095722952e-06,
      "loss": 0.1516,
      "step": 10575
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3499239683151245,
      "learning_rate": 1.4542163269265386e-06,
      "loss": 0.0412,
      "step": 10576
    },
    {
      "epoch": 0.76,
      "grad_norm": 11.082423210144043,
      "learning_rate": 1.453397635562388e-06,
      "loss": 0.356,
      "step": 10577
    },
    {
      "epoch": 0.76,
      "grad_norm": 22.383960723876953,
      "learning_rate": 1.4525791355240088e-06,
      "loss": 0.7065,
      "step": 10578
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.8228940963745117,
      "learning_rate": 1.4517608268555594e-06,
      "loss": 0.7536,
      "step": 10579
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.5173838138580322,
      "learning_rate": 1.4509427096011818e-06,
      "loss": 0.5602,
      "step": 10580
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0071282386779785,
      "learning_rate": 1.4501247838050103e-06,
      "loss": 0.094,
      "step": 10581
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.0123515129089355,
      "learning_rate": 1.4493070495111711e-06,
      "loss": 0.7996,
      "step": 10582
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5868980884552002,
      "learning_rate": 1.448489506763776e-06,
      "loss": 0.0661,
      "step": 10583
    },
    {
      "epoch": 0.76,
      "grad_norm": 18.8773136138916,
      "learning_rate": 1.447672155606928e-06,
      "loss": 1.0817,
      "step": 10584
    },
    {
      "epoch": 0.76,
      "grad_norm": 16.850971221923828,
      "learning_rate": 1.4468549960847216e-06,
      "loss": 0.6971,
      "step": 10585
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.162308931350708,
      "learning_rate": 1.4460380282412395e-06,
      "loss": 0.0723,
      "step": 10586
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.971414566040039,
      "learning_rate": 1.4452212521205516e-06,
      "loss": 0.179,
      "step": 10587
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.551844358444214,
      "learning_rate": 1.4444046677667245e-06,
      "loss": 0.1391,
      "step": 10588
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0289053916931152,
      "learning_rate": 1.4435882752238055e-06,
      "loss": 0.1042,
      "step": 10589
    },
    {
      "epoch": 0.76,
      "grad_norm": 13.558889389038086,
      "learning_rate": 1.442772074535836e-06,
      "loss": 0.8407,
      "step": 10590
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.7790870666503906,
      "learning_rate": 1.441956065746849e-06,
      "loss": 0.2694,
      "step": 10591
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.8079960346221924,
      "learning_rate": 1.4411402489008652e-06,
      "loss": 0.7225,
      "step": 10592
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.920596599578857,
      "learning_rate": 1.4403246240418917e-06,
      "loss": 0.3276,
      "step": 10593
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.1761536598205566,
      "learning_rate": 1.4395091912139325e-06,
      "loss": 0.1937,
      "step": 10594
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.6466851234436035,
      "learning_rate": 1.4386939504609754e-06,
      "loss": 0.7573,
      "step": 10595
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.41238784790039,
      "learning_rate": 1.437878901826999e-06,
      "loss": 0.4642,
      "step": 10596
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1687636375427246,
      "learning_rate": 1.4370640453559714e-06,
      "loss": 0.1064,
      "step": 10597
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3724634647369385,
      "learning_rate": 1.436249381091853e-06,
      "loss": 0.0734,
      "step": 10598
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.5087103843688965,
      "learning_rate": 1.435434909078592e-06,
      "loss": 0.1097,
      "step": 10599
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.223472833633423,
      "learning_rate": 1.4346206293601245e-06,
      "loss": 0.0897,
      "step": 10600
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.3314833641052246,
      "learning_rate": 1.4338065419803782e-06,
      "loss": 0.1691,
      "step": 10601
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7623326778411865,
      "learning_rate": 1.4329926469832683e-06,
      "loss": 0.1298,
      "step": 10602
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.9268312454223633,
      "learning_rate": 1.4321789444127055e-06,
      "loss": 0.0914,
      "step": 10603
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.078869581222534,
      "learning_rate": 1.4313654343125833e-06,
      "loss": 0.0695,
      "step": 10604
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.025076866149902,
      "learning_rate": 1.4305521167267882e-06,
      "loss": 0.454,
      "step": 10605
    },
    {
      "epoch": 0.76,
      "grad_norm": 12.568467140197754,
      "learning_rate": 1.4297389916991939e-06,
      "loss": 0.66,
      "step": 10606
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.2473316192626953,
      "learning_rate": 1.4289260592736676e-06,
      "loss": 0.1209,
      "step": 10607
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.940423011779785,
      "learning_rate": 1.4281133194940638e-06,
      "loss": 0.3692,
      "step": 10608
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.357036590576172,
      "learning_rate": 1.4273007724042244e-06,
      "loss": 0.1191,
      "step": 10609
    },
    {
      "epoch": 0.76,
      "grad_norm": 11.548421859741211,
      "learning_rate": 1.4264884180479865e-06,
      "loss": 0.4909,
      "step": 10610
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.866036891937256,
      "learning_rate": 1.4256762564691718e-06,
      "loss": 0.1394,
      "step": 10611
    },
    {
      "epoch": 0.76,
      "grad_norm": 15.168493270874023,
      "learning_rate": 1.4248642877115937e-06,
      "loss": 1.1544,
      "step": 10612
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7667770385742188,
      "learning_rate": 1.4240525118190547e-06,
      "loss": 0.1293,
      "step": 10613
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.435826778411865,
      "learning_rate": 1.4232409288353467e-06,
      "loss": 0.9501,
      "step": 10614
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.465365171432495,
      "learning_rate": 1.4224295388042503e-06,
      "loss": 0.1154,
      "step": 10615
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1748135089874268,
      "learning_rate": 1.42161834176954e-06,
      "loss": 0.0889,
      "step": 10616
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.152177572250366,
      "learning_rate": 1.4208073377749753e-06,
      "loss": 0.0433,
      "step": 10617
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3710429668426514,
      "learning_rate": 1.4199965268643057e-06,
      "loss": 0.07,
      "step": 10618
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.5849609375,
      "learning_rate": 1.4191859090812732e-06,
      "loss": 0.9419,
      "step": 10619
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.930402994155884,
      "learning_rate": 1.4183754844696074e-06,
      "loss": 0.1418,
      "step": 10620
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6768155097961426,
      "learning_rate": 1.4175652530730249e-06,
      "loss": 0.0621,
      "step": 10621
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.22857403755188,
      "learning_rate": 1.4167552149352386e-06,
      "loss": 0.1253,
      "step": 10622
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.105541229248047,
      "learning_rate": 1.4159453700999453e-06,
      "loss": 0.8997,
      "step": 10623
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.830488204956055,
      "learning_rate": 1.4151357186108327e-06,
      "loss": 0.4529,
      "step": 10624
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.6825919151306152,
      "learning_rate": 1.4143262605115788e-06,
      "loss": 0.2798,
      "step": 10625
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3861944675445557,
      "learning_rate": 1.4135169958458505e-06,
      "loss": 0.13,
      "step": 10626
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.5801877975463867,
      "learning_rate": 1.4127079246573033e-06,
      "loss": 0.0878,
      "step": 10627
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.2129716873168945,
      "learning_rate": 1.4118990469895866e-06,
      "loss": 0.3763,
      "step": 10628
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.548863410949707,
      "learning_rate": 1.4110903628863343e-06,
      "loss": 0.872,
      "step": 10629
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.0946993827819824,
      "learning_rate": 1.4102818723911704e-06,
      "loss": 0.1418,
      "step": 10630
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.0107858180999756,
      "learning_rate": 1.4094735755477135e-06,
      "loss": 0.0867,
      "step": 10631
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8857297897338867,
      "learning_rate": 1.4086654723995663e-06,
      "loss": 0.1524,
      "step": 10632
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.582395553588867,
      "learning_rate": 1.4078575629903229e-06,
      "loss": 0.3922,
      "step": 10633
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.6851165294647217,
      "learning_rate": 1.4070498473635657e-06,
      "loss": 0.1417,
      "step": 10634
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.235339164733887,
      "learning_rate": 1.4062423255628704e-06,
      "loss": 0.419,
      "step": 10635
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.728867292404175,
      "learning_rate": 1.4054349976317988e-06,
      "loss": 0.6771,
      "step": 10636
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.203547954559326,
      "learning_rate": 1.4046278636139026e-06,
      "loss": 0.2153,
      "step": 10637
    },
    {
      "epoch": 0.76,
      "grad_norm": 43.61389923095703,
      "learning_rate": 1.403820923552724e-06,
      "loss": 1.6431,
      "step": 10638
    },
    {
      "epoch": 0.76,
      "grad_norm": 13.748260498046875,
      "learning_rate": 1.4030141774917926e-06,
      "loss": 1.2843,
      "step": 10639
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.851632833480835,
      "learning_rate": 1.4022076254746331e-06,
      "loss": 0.1036,
      "step": 10640
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8828933238983154,
      "learning_rate": 1.401401267544753e-06,
      "loss": 0.1388,
      "step": 10641
    },
    {
      "epoch": 0.76,
      "grad_norm": 18.93425178527832,
      "learning_rate": 1.4005951037456538e-06,
      "loss": 0.7048,
      "step": 10642
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.261715888977051,
      "learning_rate": 1.3997891341208226e-06,
      "loss": 0.7951,
      "step": 10643
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.0460751056671143,
      "learning_rate": 1.3989833587137414e-06,
      "loss": 0.1342,
      "step": 10644
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.1792070865631104,
      "learning_rate": 1.3981777775678773e-06,
      "loss": 0.1561,
      "step": 10645
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0068790912628174,
      "learning_rate": 1.3973723907266873e-06,
      "loss": 0.0969,
      "step": 10646
    },
    {
      "epoch": 0.76,
      "grad_norm": 13.93374252319336,
      "learning_rate": 1.3965671982336216e-06,
      "loss": 0.6568,
      "step": 10647
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8725521564483643,
      "learning_rate": 1.395762200132117e-06,
      "loss": 0.1346,
      "step": 10648
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.544854164123535,
      "learning_rate": 1.394957396465596e-06,
      "loss": 0.0914,
      "step": 10649
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9024477005004883,
      "learning_rate": 1.3941527872774796e-06,
      "loss": 0.0407,
      "step": 10650
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1555163860321045,
      "learning_rate": 1.3933483726111713e-06,
      "loss": 0.0963,
      "step": 10651
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3145546913146973,
      "learning_rate": 1.3925441525100647e-06,
      "loss": 0.1248,
      "step": 10652
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3427906036376953,
      "learning_rate": 1.3917401270175473e-06,
      "loss": 0.0913,
      "step": 10653
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.551735877990723,
      "learning_rate": 1.3909362961769922e-06,
      "loss": 1.0176,
      "step": 10654
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.324016094207764,
      "learning_rate": 1.3901326600317615e-06,
      "loss": 0.7808,
      "step": 10655
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.319209337234497,
      "learning_rate": 1.3893292186252116e-06,
      "loss": 0.0572,
      "step": 10656
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.78000259399414,
      "learning_rate": 1.3885259720006828e-06,
      "loss": 0.7763,
      "step": 10657
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2051990032196045,
      "learning_rate": 1.3877229202015064e-06,
      "loss": 0.0773,
      "step": 10658
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.166728973388672,
      "learning_rate": 1.3869200632710061e-06,
      "loss": 0.7784,
      "step": 10659
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4641034603118896,
      "learning_rate": 1.3861174012524942e-06,
      "loss": 0.0357,
      "step": 10660
    },
    {
      "epoch": 0.76,
      "grad_norm": 13.77861499786377,
      "learning_rate": 1.3853149341892664e-06,
      "loss": 0.6807,
      "step": 10661
    },
    {
      "epoch": 0.76,
      "grad_norm": 12.78648567199707,
      "learning_rate": 1.3845126621246164e-06,
      "loss": 0.8439,
      "step": 10662
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.381983757019043,
      "learning_rate": 1.3837105851018235e-06,
      "loss": 0.6748,
      "step": 10663
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.15616512298584,
      "learning_rate": 1.3829087031641542e-06,
      "loss": 0.0551,
      "step": 10664
    },
    {
      "epoch": 0.76,
      "grad_norm": 12.53077507019043,
      "learning_rate": 1.3821070163548705e-06,
      "loss": 0.7868,
      "step": 10665
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8077532052993774,
      "learning_rate": 1.3813055247172192e-06,
      "loss": 0.0814,
      "step": 10666
    },
    {
      "epoch": 0.76,
      "grad_norm": 12.991155624389648,
      "learning_rate": 1.3805042282944353e-06,
      "loss": 0.7697,
      "step": 10667
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9414087533950806,
      "learning_rate": 1.3797031271297495e-06,
      "loss": 0.1119,
      "step": 10668
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.0817999839782715,
      "learning_rate": 1.3789022212663766e-06,
      "loss": 0.1758,
      "step": 10669
    },
    {
      "epoch": 0.77,
      "grad_norm": 23.87038230895996,
      "learning_rate": 1.378101510747522e-06,
      "loss": 0.5755,
      "step": 10670
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.009562492370605,
      "learning_rate": 1.3773009956163802e-06,
      "loss": 0.5284,
      "step": 10671
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.255174398422241,
      "learning_rate": 1.3765006759161397e-06,
      "loss": 0.0741,
      "step": 10672
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.3613996505737305,
      "learning_rate": 1.3757005516899691e-06,
      "loss": 0.8947,
      "step": 10673
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.979256629943848,
      "learning_rate": 1.3749006229810364e-06,
      "loss": 0.3386,
      "step": 10674
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.41100025177002,
      "learning_rate": 1.374100889832493e-06,
      "loss": 0.7822,
      "step": 10675
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.973663330078125,
      "learning_rate": 1.3733013522874826e-06,
      "loss": 0.7735,
      "step": 10676
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2819530963897705,
      "learning_rate": 1.3725020103891346e-06,
      "loss": 0.0856,
      "step": 10677
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.195635795593262,
      "learning_rate": 1.3717028641805736e-06,
      "loss": 0.6712,
      "step": 10678
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.8860621452331543,
      "learning_rate": 1.3709039137049097e-06,
      "loss": 0.7866,
      "step": 10679
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.214920997619629,
      "learning_rate": 1.3701051590052406e-06,
      "loss": 0.6129,
      "step": 10680
    },
    {
      "epoch": 0.77,
      "grad_norm": 14.548415184020996,
      "learning_rate": 1.3693066001246601e-06,
      "loss": 0.662,
      "step": 10681
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.103820323944092,
      "learning_rate": 1.368508237106246e-06,
      "loss": 0.5804,
      "step": 10682
    },
    {
      "epoch": 0.77,
      "grad_norm": 19.482454299926758,
      "learning_rate": 1.3677100699930647e-06,
      "loss": 0.654,
      "step": 10683
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.160840034484863,
      "learning_rate": 1.3669120988281776e-06,
      "loss": 0.4373,
      "step": 10684
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.875514268875122,
      "learning_rate": 1.3661143236546325e-06,
      "loss": 0.1864,
      "step": 10685
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.0602920055389404,
      "learning_rate": 1.3653167445154613e-06,
      "loss": 0.1503,
      "step": 10686
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.398220062255859,
      "learning_rate": 1.3645193614536955e-06,
      "loss": 0.3992,
      "step": 10687
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.519765615463257,
      "learning_rate": 1.3637221745123492e-06,
      "loss": 0.1155,
      "step": 10688
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1192874908447266,
      "learning_rate": 1.362925183734426e-06,
      "loss": 0.078,
      "step": 10689
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8387336730957031,
      "learning_rate": 1.3621283891629234e-06,
      "loss": 0.0344,
      "step": 10690
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6682344675064087,
      "learning_rate": 1.361331790840824e-06,
      "loss": 0.0671,
      "step": 10691
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.858841896057129,
      "learning_rate": 1.3605353888111e-06,
      "loss": 0.1696,
      "step": 10692
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.174778938293457,
      "learning_rate": 1.3597391831167174e-06,
      "loss": 0.6342,
      "step": 10693
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.2301220893859863,
      "learning_rate": 1.3589431738006264e-06,
      "loss": 0.1463,
      "step": 10694
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.138634443283081,
      "learning_rate": 1.3581473609057677e-06,
      "loss": 0.1166,
      "step": 10695
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8713949918746948,
      "learning_rate": 1.357351744475075e-06,
      "loss": 0.0635,
      "step": 10696
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.797134399414062,
      "learning_rate": 1.356556324551469e-06,
      "loss": 0.8313,
      "step": 10697
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.042632579803467,
      "learning_rate": 1.3557611011778554e-06,
      "loss": 0.1219,
      "step": 10698
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.575772762298584,
      "learning_rate": 1.3549660743971377e-06,
      "loss": 0.6996,
      "step": 10699
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6309666633605957,
      "learning_rate": 1.3541712442522025e-06,
      "loss": 0.1358,
      "step": 10700
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.3665006160736084,
      "learning_rate": 1.3533766107859275e-06,
      "loss": 0.1348,
      "step": 10701
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.23856782913208,
      "learning_rate": 1.3525821740411826e-06,
      "loss": 0.26,
      "step": 10702
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.216801643371582,
      "learning_rate": 1.351787934060823e-06,
      "loss": 0.3133,
      "step": 10703
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.0153143405914307,
      "learning_rate": 1.350993890887694e-06,
      "loss": 0.1536,
      "step": 10704
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.1105878353118896,
      "learning_rate": 1.3502000445646336e-06,
      "loss": 0.1419,
      "step": 10705
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.731254816055298,
      "learning_rate": 1.3494063951344654e-06,
      "loss": 0.1046,
      "step": 10706
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.851750135421753,
      "learning_rate": 1.3486129426400042e-06,
      "loss": 0.7939,
      "step": 10707
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.1857872009277344,
      "learning_rate": 1.3478196871240518e-06,
      "loss": 0.1462,
      "step": 10708
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.966640591621399,
      "learning_rate": 1.3470266286294065e-06,
      "loss": 0.0711,
      "step": 10709
    },
    {
      "epoch": 0.77,
      "grad_norm": 16.982101440429688,
      "learning_rate": 1.3462337671988434e-06,
      "loss": 1.1045,
      "step": 10710
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.481341361999512,
      "learning_rate": 1.3454411028751408e-06,
      "loss": 0.546,
      "step": 10711
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.0211405754089355,
      "learning_rate": 1.344648635701057e-06,
      "loss": 0.8562,
      "step": 10712
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.654128074645996,
      "learning_rate": 1.3438563657193427e-06,
      "loss": 0.1243,
      "step": 10713
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.025403022766113,
      "learning_rate": 1.3430642929727373e-06,
      "loss": 0.5988,
      "step": 10714
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3877575397491455,
      "learning_rate": 1.342272417503972e-06,
      "loss": 0.1346,
      "step": 10715
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.448889255523682,
      "learning_rate": 1.341480739355765e-06,
      "loss": 0.7594,
      "step": 10716
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.39216423034668,
      "learning_rate": 1.340689258570822e-06,
      "loss": 0.2515,
      "step": 10717
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.143284797668457,
      "learning_rate": 1.339897975191844e-06,
      "loss": 0.1141,
      "step": 10718
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.125366687774658,
      "learning_rate": 1.339106889261516e-06,
      "loss": 0.495,
      "step": 10719
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.228095054626465,
      "learning_rate": 1.3383160008225126e-06,
      "loss": 0.9419,
      "step": 10720
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.504684448242188,
      "learning_rate": 1.3375253099175035e-06,
      "loss": 0.3609,
      "step": 10721
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.46696138381958,
      "learning_rate": 1.3367348165891397e-06,
      "loss": 0.2988,
      "step": 10722
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.600180625915527,
      "learning_rate": 1.3359445208800642e-06,
      "loss": 0.9113,
      "step": 10723
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.818305969238281,
      "learning_rate": 1.3351544228329144e-06,
      "loss": 0.7982,
      "step": 10724
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.473362922668457,
      "learning_rate": 1.3343645224903113e-06,
      "loss": 0.6945,
      "step": 10725
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.763391494750977,
      "learning_rate": 1.3335748198948656e-06,
      "loss": 0.8157,
      "step": 10726
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.343323707580566,
      "learning_rate": 1.3327853150891817e-06,
      "loss": 0.7662,
      "step": 10727
    },
    {
      "epoch": 0.77,
      "grad_norm": 18.67729949951172,
      "learning_rate": 1.331996008115849e-06,
      "loss": 1.0941,
      "step": 10728
    },
    {
      "epoch": 0.77,
      "grad_norm": 15.597786903381348,
      "learning_rate": 1.3312068990174453e-06,
      "loss": 0.8804,
      "step": 10729
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.0135116577148438,
      "learning_rate": 1.3304179878365447e-06,
      "loss": 0.1643,
      "step": 10730
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.8305821418762207,
      "learning_rate": 1.329629274615703e-06,
      "loss": 0.1656,
      "step": 10731
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.334935665130615,
      "learning_rate": 1.3288407593974673e-06,
      "loss": 0.8457,
      "step": 10732
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.392434597015381,
      "learning_rate": 1.3280524422243796e-06,
      "loss": 0.1266,
      "step": 10733
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.9655184745788574,
      "learning_rate": 1.327264323138962e-06,
      "loss": 0.1256,
      "step": 10734
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9880969524383545,
      "learning_rate": 1.3264764021837313e-06,
      "loss": 0.0548,
      "step": 10735
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.5086448192596436,
      "learning_rate": 1.3256886794011942e-06,
      "loss": 0.153,
      "step": 10736
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.943124532699585,
      "learning_rate": 1.3249011548338452e-06,
      "loss": 0.0364,
      "step": 10737
    },
    {
      "epoch": 0.77,
      "grad_norm": 23.221914291381836,
      "learning_rate": 1.3241138285241668e-06,
      "loss": 0.7384,
      "step": 10738
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.251448154449463,
      "learning_rate": 1.3233267005146343e-06,
      "loss": 0.6121,
      "step": 10739
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.706875801086426,
      "learning_rate": 1.3225397708477094e-06,
      "loss": 0.1134,
      "step": 10740
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.5008959770202637,
      "learning_rate": 1.321753039565844e-06,
      "loss": 0.3351,
      "step": 10741
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.876697540283203,
      "learning_rate": 1.320966506711478e-06,
      "loss": 0.7842,
      "step": 10742
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.082700729370117,
      "learning_rate": 1.320180172327044e-06,
      "loss": 0.1045,
      "step": 10743
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.241673469543457,
      "learning_rate": 1.3193940364549613e-06,
      "loss": 0.4374,
      "step": 10744
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6061190366744995,
      "learning_rate": 1.3186080991376382e-06,
      "loss": 0.052,
      "step": 10745
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.919332265853882,
      "learning_rate": 1.3178223604174734e-06,
      "loss": 0.6808,
      "step": 10746
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3304953575134277,
      "learning_rate": 1.3170368203368534e-06,
      "loss": 0.138,
      "step": 10747
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.201565980911255,
      "learning_rate": 1.316251478938157e-06,
      "loss": 0.1462,
      "step": 10748
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.3051180839538574,
      "learning_rate": 1.3154663362637498e-06,
      "loss": 0.0748,
      "step": 10749
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.859461545944214,
      "learning_rate": 1.314681392355987e-06,
      "loss": 0.7271,
      "step": 10750
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.73963737487793,
      "learning_rate": 1.3138966472572124e-06,
      "loss": 0.7648,
      "step": 10751
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.809962749481201,
      "learning_rate": 1.313112101009762e-06,
      "loss": 0.1515,
      "step": 10752
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.186493873596191,
      "learning_rate": 1.312327753655958e-06,
      "loss": 0.8603,
      "step": 10753
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.8012497425079346,
      "learning_rate": 1.311543605238112e-06,
      "loss": 0.2405,
      "step": 10754
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.530888080596924,
      "learning_rate": 1.3107596557985281e-06,
      "loss": 0.9924,
      "step": 10755
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.364877700805664,
      "learning_rate": 1.3099759053794963e-06,
      "loss": 0.0995,
      "step": 10756
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.114675045013428,
      "learning_rate": 1.3091923540232971e-06,
      "loss": 0.6354,
      "step": 10757
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.092911720275879,
      "learning_rate": 1.3084090017722e-06,
      "loss": 0.649,
      "step": 10758
    },
    {
      "epoch": 0.77,
      "grad_norm": 15.048443794250488,
      "learning_rate": 1.3076258486684635e-06,
      "loss": 1.1251,
      "step": 10759
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.295116901397705,
      "learning_rate": 1.3068428947543348e-06,
      "loss": 0.6078,
      "step": 10760
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.453377723693848,
      "learning_rate": 1.306060140072054e-06,
      "loss": 0.4205,
      "step": 10761
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.882472991943359,
      "learning_rate": 1.305277584663846e-06,
      "loss": 0.4773,
      "step": 10762
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1503841876983643,
      "learning_rate": 1.304495228571926e-06,
      "loss": 0.0844,
      "step": 10763
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.644445419311523,
      "learning_rate": 1.3037130718385016e-06,
      "loss": 0.4888,
      "step": 10764
    },
    {
      "epoch": 0.77,
      "grad_norm": 38.162086486816406,
      "learning_rate": 1.3029311145057655e-06,
      "loss": 0.7445,
      "step": 10765
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.002022743225098,
      "learning_rate": 1.3021493566159004e-06,
      "loss": 0.8012,
      "step": 10766
    },
    {
      "epoch": 0.77,
      "grad_norm": 20.279239654541016,
      "learning_rate": 1.3013677982110817e-06,
      "loss": 0.9294,
      "step": 10767
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.2516770362854,
      "learning_rate": 1.3005864393334704e-06,
      "loss": 0.7599,
      "step": 10768
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8993593454360962,
      "learning_rate": 1.2998052800252176e-06,
      "loss": 0.0525,
      "step": 10769
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.184781074523926,
      "learning_rate": 1.2990243203284636e-06,
      "loss": 0.1443,
      "step": 10770
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.3205904960632324,
      "learning_rate": 1.2982435602853389e-06,
      "loss": 0.0578,
      "step": 10771
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.5736355781555176,
      "learning_rate": 1.2974629999379607e-06,
      "loss": 0.2563,
      "step": 10772
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6887171268463135,
      "learning_rate": 1.2966826393284404e-06,
      "loss": 0.122,
      "step": 10773
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.18742561340332,
      "learning_rate": 1.2959024784988739e-06,
      "loss": 0.6912,
      "step": 10774
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6061856746673584,
      "learning_rate": 1.2951225174913462e-06,
      "loss": 0.1301,
      "step": 10775
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.4965624809265137,
      "learning_rate": 1.294342756347936e-06,
      "loss": 0.086,
      "step": 10776
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.5570707321167,
      "learning_rate": 1.293563195110708e-06,
      "loss": 0.8523,
      "step": 10777
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.88054370880127,
      "learning_rate": 1.2927838338217164e-06,
      "loss": 0.5158,
      "step": 10778
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3292616605758667,
      "learning_rate": 1.2920046725230023e-06,
      "loss": 0.049,
      "step": 10779
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0375165939331055,
      "learning_rate": 1.2912257112566019e-06,
      "loss": 0.1016,
      "step": 10780
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.047571182250977,
      "learning_rate": 1.290446950064536e-06,
      "loss": 0.5952,
      "step": 10781
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.574493169784546,
      "learning_rate": 1.2896683889888161e-06,
      "loss": 0.0938,
      "step": 10782
    },
    {
      "epoch": 0.77,
      "grad_norm": 20.344619750976562,
      "learning_rate": 1.2888900280714422e-06,
      "loss": 0.5197,
      "step": 10783
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.1976470947265625,
      "learning_rate": 1.2881118673544024e-06,
      "loss": 0.6873,
      "step": 10784
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.335198879241943,
      "learning_rate": 1.2873339068796786e-06,
      "loss": 0.3878,
      "step": 10785
    },
    {
      "epoch": 0.77,
      "grad_norm": 15.931249618530273,
      "learning_rate": 1.2865561466892373e-06,
      "loss": 0.9096,
      "step": 10786
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.317846298217773,
      "learning_rate": 1.2857785868250356e-06,
      "loss": 0.8999,
      "step": 10787
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.388251304626465,
      "learning_rate": 1.2850012273290186e-06,
      "loss": 0.1927,
      "step": 10788
    },
    {
      "epoch": 0.77,
      "grad_norm": 21.693235397338867,
      "learning_rate": 1.2842240682431251e-06,
      "loss": 0.4096,
      "step": 10789
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.1062822341918945,
      "learning_rate": 1.2834471096092778e-06,
      "loss": 0.187,
      "step": 10790
    },
    {
      "epoch": 0.77,
      "grad_norm": 14.871563911437988,
      "learning_rate": 1.2826703514693894e-06,
      "loss": 1.3623,
      "step": 10791
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.297211647033691,
      "learning_rate": 1.281893793865367e-06,
      "loss": 0.6839,
      "step": 10792
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6924996376037598,
      "learning_rate": 1.2811174368391012e-06,
      "loss": 0.1875,
      "step": 10793
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7407715320587158,
      "learning_rate": 1.2803412804324705e-06,
      "loss": 0.0566,
      "step": 10794
    },
    {
      "epoch": 0.77,
      "grad_norm": 20.270910263061523,
      "learning_rate": 1.279565324687349e-06,
      "loss": 0.7834,
      "step": 10795
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.1502761840820312,
      "learning_rate": 1.2787895696455966e-06,
      "loss": 0.1435,
      "step": 10796
    },
    {
      "epoch": 0.77,
      "grad_norm": 14.145915985107422,
      "learning_rate": 1.2780140153490589e-06,
      "loss": 1.2537,
      "step": 10797
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.732665061950684,
      "learning_rate": 1.2772386618395788e-06,
      "loss": 0.7846,
      "step": 10798
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.983238458633423,
      "learning_rate": 1.2764635091589812e-06,
      "loss": 0.1986,
      "step": 10799
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.592682361602783,
      "learning_rate": 1.2756885573490814e-06,
      "loss": 0.4896,
      "step": 10800
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.9284749031066895,
      "learning_rate": 1.2749138064516881e-06,
      "loss": 0.6374,
      "step": 10801
    },
    {
      "epoch": 0.77,
      "grad_norm": 18.170368194580078,
      "learning_rate": 1.2741392565085943e-06,
      "loss": 0.6926,
      "step": 10802
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.175992727279663,
      "learning_rate": 1.2733649075615833e-06,
      "loss": 0.0832,
      "step": 10803
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.863356351852417,
      "learning_rate": 1.272590759652431e-06,
      "loss": 0.7579,
      "step": 10804
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9410178661346436,
      "learning_rate": 1.2718168128229002e-06,
      "loss": 0.081,
      "step": 10805
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.830118179321289,
      "learning_rate": 1.2710430671147366e-06,
      "loss": 0.3192,
      "step": 10806
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.4198901653289795,
      "learning_rate": 1.2702695225696865e-06,
      "loss": 0.1048,
      "step": 10807
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.123100757598877,
      "learning_rate": 1.269496179229478e-06,
      "loss": 0.8948,
      "step": 10808
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.96744155883789,
      "learning_rate": 1.2687230371358284e-06,
      "loss": 0.6574,
      "step": 10809
    },
    {
      "epoch": 0.78,
      "grad_norm": 16.005264282226562,
      "learning_rate": 1.2679500963304487e-06,
      "loss": 1.1385,
      "step": 10810
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3778040409088135,
      "learning_rate": 1.267177356855035e-06,
      "loss": 0.112,
      "step": 10811
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.172426223754883,
      "learning_rate": 1.2664048187512712e-06,
      "loss": 0.9632,
      "step": 10812
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.330357551574707,
      "learning_rate": 1.2656324820608374e-06,
      "loss": 0.103,
      "step": 10813
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.844223022460938,
      "learning_rate": 1.264860346825395e-06,
      "loss": 0.5231,
      "step": 10814
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7580997943878174,
      "learning_rate": 1.2640884130865988e-06,
      "loss": 0.0893,
      "step": 10815
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7719509601593018,
      "learning_rate": 1.2633166808860903e-06,
      "loss": 0.1004,
      "step": 10816
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.544403076171875,
      "learning_rate": 1.2625451502655055e-06,
      "loss": 0.1326,
      "step": 10817
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7707014083862305,
      "learning_rate": 1.2617738212664604e-06,
      "loss": 0.0449,
      "step": 10818
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6040387153625488,
      "learning_rate": 1.2610026939305686e-06,
      "loss": 0.0603,
      "step": 10819
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.071803092956543,
      "learning_rate": 1.2602317682994285e-06,
      "loss": 0.7465,
      "step": 10820
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5652098655700684,
      "learning_rate": 1.2594610444146293e-06,
      "loss": 0.0801,
      "step": 10821
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5403854846954346,
      "learning_rate": 1.2586905223177458e-06,
      "loss": 0.1267,
      "step": 10822
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.996845245361328,
      "learning_rate": 1.2579202020503494e-06,
      "loss": 0.5231,
      "step": 10823
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.446500778198242,
      "learning_rate": 1.257150083653993e-06,
      "loss": 0.1466,
      "step": 10824
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1242082118988037,
      "learning_rate": 1.256380167170221e-06,
      "loss": 0.085,
      "step": 10825
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.914031028747559,
      "learning_rate": 1.2556104526405694e-06,
      "loss": 0.5572,
      "step": 10826
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.516055107116699,
      "learning_rate": 1.2548409401065609e-06,
      "loss": 0.6388,
      "step": 10827
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.912996530532837,
      "learning_rate": 1.254071629609706e-06,
      "loss": 0.1862,
      "step": 10828
    },
    {
      "epoch": 0.78,
      "grad_norm": 17.515750885009766,
      "learning_rate": 1.2533025211915106e-06,
      "loss": 0.9522,
      "step": 10829
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.171154499053955,
      "learning_rate": 1.2525336148934608e-06,
      "loss": 0.3125,
      "step": 10830
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.54386043548584,
      "learning_rate": 1.2517649107570361e-06,
      "loss": 0.9079,
      "step": 10831
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.9273478984832764,
      "learning_rate": 1.250996408823708e-06,
      "loss": 0.6515,
      "step": 10832
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.877068281173706,
      "learning_rate": 1.2502281091349327e-06,
      "loss": 0.076,
      "step": 10833
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.0220792293548584,
      "learning_rate": 1.249460011732156e-06,
      "loss": 0.1828,
      "step": 10834
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.2983901500701904,
      "learning_rate": 1.2486921166568167e-06,
      "loss": 0.1415,
      "step": 10835
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.531578540802002,
      "learning_rate": 1.247924423950338e-06,
      "loss": 0.415,
      "step": 10836
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5830729007720947,
      "learning_rate": 1.2471569336541329e-06,
      "loss": 0.1232,
      "step": 10837
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.481658935546875,
      "learning_rate": 1.246389645809607e-06,
      "loss": 0.722,
      "step": 10838
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.432740211486816,
      "learning_rate": 1.2456225604581524e-06,
      "loss": 0.5835,
      "step": 10839
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.437760353088379,
      "learning_rate": 1.2448556776411474e-06,
      "loss": 0.8154,
      "step": 10840
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.43644905090332,
      "learning_rate": 1.2440889973999676e-06,
      "loss": 0.5755,
      "step": 10841
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.444899320602417,
      "learning_rate": 1.243322519775968e-06,
      "loss": 0.6219,
      "step": 10842
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5358619689941406,
      "learning_rate": 1.2425562448104977e-06,
      "loss": 0.0897,
      "step": 10843
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.2929842472076416,
      "learning_rate": 1.2417901725448961e-06,
      "loss": 0.1258,
      "step": 10844
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.681234121322632,
      "learning_rate": 1.2410243030204893e-06,
      "loss": 0.2439,
      "step": 10845
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.192142963409424,
      "learning_rate": 1.2402586362785918e-06,
      "loss": 0.1692,
      "step": 10846
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.146345138549805,
      "learning_rate": 1.2394931723605103e-06,
      "loss": 0.9444,
      "step": 10847
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.337310791015625,
      "learning_rate": 1.2387279113075384e-06,
      "loss": 0.1738,
      "step": 10848
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.840409517288208,
      "learning_rate": 1.2379628531609584e-06,
      "loss": 0.1982,
      "step": 10849
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.05578899383545,
      "learning_rate": 1.2371979979620408e-06,
      "loss": 0.8369,
      "step": 10850
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.6446852684021,
      "learning_rate": 1.2364333457520495e-06,
      "loss": 0.5954,
      "step": 10851
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.117397308349609,
      "learning_rate": 1.2356688965722336e-06,
      "loss": 0.4958,
      "step": 10852
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.691552639007568,
      "learning_rate": 1.2349046504638318e-06,
      "loss": 0.7348,
      "step": 10853
    },
    {
      "epoch": 0.78,
      "grad_norm": 15.31158447265625,
      "learning_rate": 1.234140607468073e-06,
      "loss": 1.4312,
      "step": 10854
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.729677677154541,
      "learning_rate": 1.2333767676261715e-06,
      "loss": 0.5967,
      "step": 10855
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.0965514183044434,
      "learning_rate": 1.2326131309793383e-06,
      "loss": 0.1184,
      "step": 10856
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.624691963195801,
      "learning_rate": 1.2318496975687654e-06,
      "loss": 0.0721,
      "step": 10857
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.647574424743652,
      "learning_rate": 1.2310864674356387e-06,
      "loss": 0.5895,
      "step": 10858
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.716402530670166,
      "learning_rate": 1.2303234406211295e-06,
      "loss": 0.5158,
      "step": 10859
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.612952947616577,
      "learning_rate": 1.2295606171664032e-06,
      "loss": 0.1431,
      "step": 10860
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.277323246002197,
      "learning_rate": 1.2287979971126097e-06,
      "loss": 0.7814,
      "step": 10861
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.633134603500366,
      "learning_rate": 1.2280355805008875e-06,
      "loss": 0.1432,
      "step": 10862
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.293708801269531,
      "learning_rate": 1.2272733673723702e-06,
      "loss": 0.8218,
      "step": 10863
    },
    {
      "epoch": 0.78,
      "grad_norm": 15.414522171020508,
      "learning_rate": 1.2265113577681736e-06,
      "loss": 0.5097,
      "step": 10864
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.011793851852417,
      "learning_rate": 1.225749551729406e-06,
      "loss": 0.0845,
      "step": 10865
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.103663206100464,
      "learning_rate": 1.2249879492971639e-06,
      "loss": 0.1813,
      "step": 10866
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.333047866821289,
      "learning_rate": 1.2242265505125329e-06,
      "loss": 0.1624,
      "step": 10867
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.870614528656006,
      "learning_rate": 1.2234653554165859e-06,
      "loss": 0.7392,
      "step": 10868
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1516659259796143,
      "learning_rate": 1.2227043640503895e-06,
      "loss": 0.0969,
      "step": 10869
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.409600257873535,
      "learning_rate": 1.2219435764549948e-06,
      "loss": 0.602,
      "step": 10870
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.662719488143921,
      "learning_rate": 1.2211829926714413e-06,
      "loss": 0.1014,
      "step": 10871
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.960590124130249,
      "learning_rate": 1.2204226127407643e-06,
      "loss": 0.2,
      "step": 10872
    },
    {
      "epoch": 0.78,
      "grad_norm": 15.271533966064453,
      "learning_rate": 1.21966243670398e-06,
      "loss": 0.5445,
      "step": 10873
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.2224552631378174,
      "learning_rate": 1.218902464602097e-06,
      "loss": 0.1531,
      "step": 10874
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.1174750328063965,
      "learning_rate": 1.218142696476115e-06,
      "loss": 0.7911,
      "step": 10875
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7998063564300537,
      "learning_rate": 1.2173831323670194e-06,
      "loss": 0.1328,
      "step": 10876
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6576870679855347,
      "learning_rate": 1.216623772315786e-06,
      "loss": 0.0317,
      "step": 10877
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.194353103637695,
      "learning_rate": 1.2158646163633787e-06,
      "loss": 0.5699,
      "step": 10878
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3319625854492188,
      "learning_rate": 1.2151056645507519e-06,
      "loss": 0.1093,
      "step": 10879
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7222295999526978,
      "learning_rate": 1.214346916918847e-06,
      "loss": 0.0643,
      "step": 10880
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.0670318603515625,
      "learning_rate": 1.2135883735085973e-06,
      "loss": 0.7888,
      "step": 10881
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.386230945587158,
      "learning_rate": 1.2128300343609228e-06,
      "loss": 0.6393,
      "step": 10882
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.6965789794921875,
      "learning_rate": 1.2120718995167308e-06,
      "loss": 0.2145,
      "step": 10883
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.9401936531066895,
      "learning_rate": 1.2113139690169234e-06,
      "loss": 0.1254,
      "step": 10884
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.191470623016357,
      "learning_rate": 1.210556242902387e-06,
      "loss": 0.8744,
      "step": 10885
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.0667481422424316,
      "learning_rate": 1.2097987212139967e-06,
      "loss": 0.121,
      "step": 10886
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3794398307800293,
      "learning_rate": 1.209041403992618e-06,
      "loss": 0.1114,
      "step": 10887
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.41986608505249,
      "learning_rate": 1.2082842912791065e-06,
      "loss": 0.7493,
      "step": 10888
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.500912666320801,
      "learning_rate": 1.2075273831143058e-06,
      "loss": 0.7877,
      "step": 10889
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.216461181640625,
      "learning_rate": 1.2067706795390472e-06,
      "loss": 0.9181,
      "step": 10890
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7817461490631104,
      "learning_rate": 1.2060141805941523e-06,
      "loss": 0.1549,
      "step": 10891
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.202226161956787,
      "learning_rate": 1.20525788632043e-06,
      "loss": 0.1434,
      "step": 10892
    },
    {
      "epoch": 0.78,
      "grad_norm": 13.111917495727539,
      "learning_rate": 1.2045017967586826e-06,
      "loss": 0.43,
      "step": 10893
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.6644723415374756,
      "learning_rate": 1.2037459119496959e-06,
      "loss": 0.5943,
      "step": 10894
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2386651039123535,
      "learning_rate": 1.202990231934248e-06,
      "loss": 0.0498,
      "step": 10895
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.016136169433594,
      "learning_rate": 1.202234756753103e-06,
      "loss": 0.3435,
      "step": 10896
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3452858924865723,
      "learning_rate": 1.201479486447019e-06,
      "loss": 0.071,
      "step": 10897
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.9979100227355957,
      "learning_rate": 1.2007244210567387e-06,
      "loss": 0.8991,
      "step": 10898
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.2386953830718994,
      "learning_rate": 1.1999695606229932e-06,
      "loss": 0.1088,
      "step": 10899
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.4157521724700928,
      "learning_rate": 1.1992149051865076e-06,
      "loss": 0.1614,
      "step": 10900
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.6895949840545654,
      "learning_rate": 1.1984604547879913e-06,
      "loss": 0.6869,
      "step": 10901
    },
    {
      "epoch": 0.78,
      "grad_norm": 26.735185623168945,
      "learning_rate": 1.1977062094681436e-06,
      "loss": 1.1168,
      "step": 10902
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5766842365264893,
      "learning_rate": 1.1969521692676533e-06,
      "loss": 0.124,
      "step": 10903
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.493742942810059,
      "learning_rate": 1.1961983342271981e-06,
      "loss": 0.8623,
      "step": 10904
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.6117117404937744,
      "learning_rate": 1.1954447043874434e-06,
      "loss": 0.1122,
      "step": 10905
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.312425136566162,
      "learning_rate": 1.1946912797890476e-06,
      "loss": 0.497,
      "step": 10906
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.691451072692871,
      "learning_rate": 1.193938060472653e-06,
      "loss": 0.8932,
      "step": 10907
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.3222546577453613,
      "learning_rate": 1.193185046478892e-06,
      "loss": 0.2507,
      "step": 10908
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.783681869506836,
      "learning_rate": 1.1924322378483894e-06,
      "loss": 0.6362,
      "step": 10909
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.799529552459717,
      "learning_rate": 1.1916796346217552e-06,
      "loss": 0.1531,
      "step": 10910
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.298585891723633,
      "learning_rate": 1.1909272368395886e-06,
      "loss": 0.1293,
      "step": 10911
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.867468357086182,
      "learning_rate": 1.1901750445424803e-06,
      "loss": 0.902,
      "step": 10912
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1123204231262207,
      "learning_rate": 1.1894230577710076e-06,
      "loss": 0.1918,
      "step": 10913
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.603332996368408,
      "learning_rate": 1.1886712765657366e-06,
      "loss": 0.5081,
      "step": 10914
    },
    {
      "epoch": 0.78,
      "grad_norm": 17.323867797851562,
      "learning_rate": 1.187919700967224e-06,
      "loss": 0.7934,
      "step": 10915
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.4933247566223145,
      "learning_rate": 1.1871683310160136e-06,
      "loss": 0.3186,
      "step": 10916
    },
    {
      "epoch": 0.78,
      "grad_norm": 17.167673110961914,
      "learning_rate": 1.1864171667526385e-06,
      "loss": 0.7707,
      "step": 10917
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.163378715515137,
      "learning_rate": 1.1856662082176235e-06,
      "loss": 0.6278,
      "step": 10918
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.027549743652344,
      "learning_rate": 1.1849154554514786e-06,
      "loss": 0.5857,
      "step": 10919
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.446416854858398,
      "learning_rate": 1.1841649084947022e-06,
      "loss": 0.5884,
      "step": 10920
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.039677858352661,
      "learning_rate": 1.1834145673877867e-06,
      "loss": 0.1143,
      "step": 10921
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.8933849334716797,
      "learning_rate": 1.182664432171209e-06,
      "loss": 0.6382,
      "step": 10922
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.859790325164795,
      "learning_rate": 1.1819145028854357e-06,
      "loss": 0.8596,
      "step": 10923
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.008378982543945,
      "learning_rate": 1.1811647795709208e-06,
      "loss": 0.7904,
      "step": 10924
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.956449270248413,
      "learning_rate": 1.180415262268113e-06,
      "loss": 0.1196,
      "step": 10925
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.2086310386657715,
      "learning_rate": 1.1796659510174435e-06,
      "loss": 0.1932,
      "step": 10926
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.4050540924072266,
      "learning_rate": 1.1789168458593353e-06,
      "loss": 0.1635,
      "step": 10927
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.7220287322998047,
      "learning_rate": 1.1781679468342e-06,
      "loss": 0.1902,
      "step": 10928
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.856035232543945,
      "learning_rate": 1.1774192539824358e-06,
      "loss": 0.6116,
      "step": 10929
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.001732349395752,
      "learning_rate": 1.176670767344435e-06,
      "loss": 0.0453,
      "step": 10930
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.38619065284729,
      "learning_rate": 1.1759224869605746e-06,
      "loss": 0.0843,
      "step": 10931
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.8674349784851074,
      "learning_rate": 1.1751744128712205e-06,
      "loss": 0.6796,
      "step": 10932
    },
    {
      "epoch": 0.78,
      "grad_norm": 16.33014678955078,
      "learning_rate": 1.174426545116728e-06,
      "loss": 0.7115,
      "step": 10933
    },
    {
      "epoch": 0.78,
      "grad_norm": 19.26421356201172,
      "learning_rate": 1.1736788837374447e-06,
      "loss": 1.5651,
      "step": 10934
    },
    {
      "epoch": 0.78,
      "grad_norm": 26.421707153320312,
      "learning_rate": 1.1729314287737014e-06,
      "loss": 0.8716,
      "step": 10935
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.684492349624634,
      "learning_rate": 1.17218418026582e-06,
      "loss": 0.0965,
      "step": 10936
    },
    {
      "epoch": 0.78,
      "grad_norm": 13.470633506774902,
      "learning_rate": 1.1714371382541151e-06,
      "loss": 0.5547,
      "step": 10937
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3289451599121094,
      "learning_rate": 1.1706903027788858e-06,
      "loss": 0.0423,
      "step": 10938
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5308327674865723,
      "learning_rate": 1.1699436738804166e-06,
      "loss": 0.0802,
      "step": 10939
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.06126070022583,
      "learning_rate": 1.1691972515989902e-06,
      "loss": 0.0796,
      "step": 10940
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.268621444702148,
      "learning_rate": 1.1684510359748725e-06,
      "loss": 0.2263,
      "step": 10941
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.8264596462249756,
      "learning_rate": 1.1677050270483158e-06,
      "loss": 0.1463,
      "step": 10942
    },
    {
      "epoch": 0.78,
      "grad_norm": 15.58566665649414,
      "learning_rate": 1.1669592248595685e-06,
      "loss": 0.6814,
      "step": 10943
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9269441366195679,
      "learning_rate": 1.1662136294488624e-06,
      "loss": 0.0586,
      "step": 10944
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.3264570236206055,
      "learning_rate": 1.1654682408564177e-06,
      "loss": 0.9616,
      "step": 10945
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.585744619369507,
      "learning_rate": 1.164723059122449e-06,
      "loss": 0.1474,
      "step": 10946
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.993261337280273,
      "learning_rate": 1.1639780842871534e-06,
      "loss": 0.5296,
      "step": 10947
    },
    {
      "epoch": 0.79,
      "grad_norm": 14.803810119628906,
      "learning_rate": 1.1632333163907184e-06,
      "loss": 0.6945,
      "step": 10948
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.15927267074585,
      "learning_rate": 1.162488755473325e-06,
      "loss": 0.3018,
      "step": 10949
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.459338188171387,
      "learning_rate": 1.1617444015751384e-06,
      "loss": 0.5154,
      "step": 10950
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.321649551391602,
      "learning_rate": 1.1610002547363103e-06,
      "loss": 0.8123,
      "step": 10951
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.713730812072754,
      "learning_rate": 1.1602563149969881e-06,
      "loss": 0.8691,
      "step": 10952
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.633111953735352,
      "learning_rate": 1.1595125823973035e-06,
      "loss": 0.8527,
      "step": 10953
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.95432710647583,
      "learning_rate": 1.158769056977377e-06,
      "loss": 0.6204,
      "step": 10954
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.036306142807007,
      "learning_rate": 1.1580257387773208e-06,
      "loss": 0.1221,
      "step": 10955
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9027385711669922,
      "learning_rate": 1.157282627837234e-06,
      "loss": 0.0651,
      "step": 10956
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.708177328109741,
      "learning_rate": 1.1565397241972021e-06,
      "loss": 0.745,
      "step": 10957
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7420873641967773,
      "learning_rate": 1.1557970278973058e-06,
      "loss": 0.0382,
      "step": 10958
    },
    {
      "epoch": 0.79,
      "grad_norm": 11.9496431350708,
      "learning_rate": 1.155054538977608e-06,
      "loss": 0.4562,
      "step": 10959
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.706929087638855,
      "learning_rate": 1.1543122574781645e-06,
      "loss": 0.0595,
      "step": 10960
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.4658851623535156,
      "learning_rate": 1.1535701834390161e-06,
      "loss": 0.0883,
      "step": 10961
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.45320725440979,
      "learning_rate": 1.1528283169001997e-06,
      "loss": 0.1121,
      "step": 10962
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.8835461139678955,
      "learning_rate": 1.1520866579017309e-06,
      "loss": 0.6795,
      "step": 10963
    },
    {
      "epoch": 0.79,
      "grad_norm": 22.479440689086914,
      "learning_rate": 1.1513452064836227e-06,
      "loss": 0.9588,
      "step": 10964
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.040479898452759,
      "learning_rate": 1.1506039626858729e-06,
      "loss": 0.1968,
      "step": 10965
    },
    {
      "epoch": 0.79,
      "grad_norm": 14.040021896362305,
      "learning_rate": 1.1498629265484684e-06,
      "loss": 0.6683,
      "step": 10966
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.8576338291168213,
      "learning_rate": 1.1491220981113842e-06,
      "loss": 0.7169,
      "step": 10967
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.8634796142578125,
      "learning_rate": 1.148381477414588e-06,
      "loss": 0.4435,
      "step": 10968
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.696811199188232,
      "learning_rate": 1.1476410644980314e-06,
      "loss": 0.2965,
      "step": 10969
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9739038944244385,
      "learning_rate": 1.1469008594016563e-06,
      "loss": 0.162,
      "step": 10970
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.349857807159424,
      "learning_rate": 1.1461608621653964e-06,
      "loss": 0.6861,
      "step": 10971
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.573392391204834,
      "learning_rate": 1.1454210728291704e-06,
      "loss": 0.1309,
      "step": 10972
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2263760566711426,
      "learning_rate": 1.1446814914328858e-06,
      "loss": 0.0604,
      "step": 10973
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.477048397064209,
      "learning_rate": 1.1439421180164446e-06,
      "loss": 0.7028,
      "step": 10974
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.461026430130005,
      "learning_rate": 1.1432029526197285e-06,
      "loss": 0.1287,
      "step": 10975
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.551186561584473,
      "learning_rate": 1.1424639952826128e-06,
      "loss": 0.519,
      "step": 10976
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.6104438304901123,
      "learning_rate": 1.1417252460449646e-06,
      "loss": 0.144,
      "step": 10977
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9634478092193604,
      "learning_rate": 1.140986704946635e-06,
      "loss": 0.0596,
      "step": 10978
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.328834056854248,
      "learning_rate": 1.1402483720274642e-06,
      "loss": 0.1459,
      "step": 10979
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.5549957752227783,
      "learning_rate": 1.1395102473272851e-06,
      "loss": 0.1473,
      "step": 10980
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.3526313304901123,
      "learning_rate": 1.138772330885915e-06,
      "loss": 0.2019,
      "step": 10981
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.803504467010498,
      "learning_rate": 1.1380346227431615e-06,
      "loss": 0.1059,
      "step": 10982
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8860511779785156,
      "learning_rate": 1.1372971229388223e-06,
      "loss": 0.1002,
      "step": 10983
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.757188558578491,
      "learning_rate": 1.136559831512683e-06,
      "loss": 0.188,
      "step": 10984
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9384312629699707,
      "learning_rate": 1.1358227485045148e-06,
      "loss": 0.1581,
      "step": 10985
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2754788398742676,
      "learning_rate": 1.1350858739540859e-06,
      "loss": 0.1274,
      "step": 10986
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.39242935180664,
      "learning_rate": 1.134349207901143e-06,
      "loss": 0.5803,
      "step": 10987
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.7547218799591064,
      "learning_rate": 1.1336127503854266e-06,
      "loss": 0.1188,
      "step": 10988
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.296174049377441,
      "learning_rate": 1.1328765014466692e-06,
      "loss": 0.7999,
      "step": 10989
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.629899024963379,
      "learning_rate": 1.1321404611245857e-06,
      "loss": 0.1327,
      "step": 10990
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3905062675476074,
      "learning_rate": 1.1314046294588831e-06,
      "loss": 0.0765,
      "step": 10991
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4451099634170532,
      "learning_rate": 1.1306690064892583e-06,
      "loss": 0.0457,
      "step": 10992
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.450538635253906,
      "learning_rate": 1.1299335922553945e-06,
      "loss": 0.6683,
      "step": 10993
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.188040733337402,
      "learning_rate": 1.1291983867969646e-06,
      "loss": 0.7284,
      "step": 10994
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.547804832458496,
      "learning_rate": 1.128463390153628e-06,
      "loss": 0.1591,
      "step": 10995
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.299139022827148,
      "learning_rate": 1.127728602365039e-06,
      "loss": 0.6153,
      "step": 10996
    },
    {
      "epoch": 0.79,
      "grad_norm": 14.173325538635254,
      "learning_rate": 1.1269940234708343e-06,
      "loss": 1.1112,
      "step": 10997
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.487460136413574,
      "learning_rate": 1.1262596535106418e-06,
      "loss": 0.6715,
      "step": 10998
    },
    {
      "epoch": 0.79,
      "grad_norm": 21.77909278869629,
      "learning_rate": 1.1255254925240783e-06,
      "loss": 0.5492,
      "step": 10999
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9907004833221436,
      "learning_rate": 1.124791540550747e-06,
      "loss": 0.2059,
      "step": 11000
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.650869369506836,
      "learning_rate": 1.1240577976302453e-06,
      "loss": 0.5083,
      "step": 11001
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.318728923797607,
      "learning_rate": 1.1233242638021541e-06,
      "loss": 0.4997,
      "step": 11002
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.574997663497925,
      "learning_rate": 1.1225909391060453e-06,
      "loss": 0.1227,
      "step": 11003
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.679654359817505,
      "learning_rate": 1.1218578235814765e-06,
      "loss": 0.1422,
      "step": 11004
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.012326717376709,
      "learning_rate": 1.121124917268e-06,
      "loss": 0.0412,
      "step": 11005
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.4930992126464844,
      "learning_rate": 1.120392220205152e-06,
      "loss": 0.1021,
      "step": 11006
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0876705646514893,
      "learning_rate": 1.1196597324324576e-06,
      "loss": 0.0562,
      "step": 11007
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.4824132919311523,
      "learning_rate": 1.1189274539894335e-06,
      "loss": 0.1199,
      "step": 11008
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.811072826385498,
      "learning_rate": 1.118195384915583e-06,
      "loss": 0.5988,
      "step": 11009
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.7794556617736816,
      "learning_rate": 1.117463525250398e-06,
      "loss": 0.6954,
      "step": 11010
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.140062093734741,
      "learning_rate": 1.11673187503336e-06,
      "loss": 0.1111,
      "step": 11011
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.614241123199463,
      "learning_rate": 1.1160004343039377e-06,
      "loss": 0.1599,
      "step": 11012
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.219350814819336,
      "learning_rate": 1.1152692031015895e-06,
      "loss": 0.1103,
      "step": 11013
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.263679504394531,
      "learning_rate": 1.1145381814657641e-06,
      "loss": 1.1171,
      "step": 11014
    },
    {
      "epoch": 0.79,
      "grad_norm": 12.34735107421875,
      "learning_rate": 1.1138073694358965e-06,
      "loss": 0.4399,
      "step": 11015
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.895711898803711,
      "learning_rate": 1.1130767670514103e-06,
      "loss": 0.8375,
      "step": 11016
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.770888090133667,
      "learning_rate": 1.112346374351721e-06,
      "loss": 0.7866,
      "step": 11017
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.953316688537598,
      "learning_rate": 1.1116161913762292e-06,
      "loss": 0.5883,
      "step": 11018
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.758045196533203,
      "learning_rate": 1.1108862181643238e-06,
      "loss": 0.2081,
      "step": 11019
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1923131942749023,
      "learning_rate": 1.1101564547553878e-06,
      "loss": 0.0837,
      "step": 11020
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.8574717044830322,
      "learning_rate": 1.1094269011887871e-06,
      "loss": 0.086,
      "step": 11021
    },
    {
      "epoch": 0.79,
      "grad_norm": 21.254894256591797,
      "learning_rate": 1.1086975575038783e-06,
      "loss": 0.9644,
      "step": 11022
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.6228950023651123,
      "learning_rate": 1.1079684237400069e-06,
      "loss": 0.1023,
      "step": 11023
    },
    {
      "epoch": 0.79,
      "grad_norm": 14.026168823242188,
      "learning_rate": 1.1072394999365065e-06,
      "loss": 0.7218,
      "step": 11024
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.420836448669434,
      "learning_rate": 1.1065107861326996e-06,
      "loss": 0.3305,
      "step": 11025
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.892462253570557,
      "learning_rate": 1.1057822823678988e-06,
      "loss": 0.6743,
      "step": 11026
    },
    {
      "epoch": 0.79,
      "grad_norm": 12.312664985656738,
      "learning_rate": 1.1050539886814039e-06,
      "loss": 0.7329,
      "step": 11027
    },
    {
      "epoch": 0.79,
      "grad_norm": 19.503446578979492,
      "learning_rate": 1.1043259051125017e-06,
      "loss": 0.6476,
      "step": 11028
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0475587844848633,
      "learning_rate": 1.1035980317004724e-06,
      "loss": 0.2005,
      "step": 11029
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.338094711303711,
      "learning_rate": 1.1028703684845805e-06,
      "loss": 0.4399,
      "step": 11030
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.5560102462768555,
      "learning_rate": 1.1021429155040809e-06,
      "loss": 0.1388,
      "step": 11031
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0890421867370605,
      "learning_rate": 1.101415672798215e-06,
      "loss": 0.1951,
      "step": 11032
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8267742395401,
      "learning_rate": 1.1006886404062183e-06,
      "loss": 0.0586,
      "step": 11033
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2665839195251465,
      "learning_rate": 1.09996181836731e-06,
      "loss": 0.0846,
      "step": 11034
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.888427734375,
      "learning_rate": 1.099235206720699e-06,
      "loss": 0.9284,
      "step": 11035
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.464247703552246,
      "learning_rate": 1.0985088055055832e-06,
      "loss": 0.377,
      "step": 11036
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.383105754852295,
      "learning_rate": 1.0977826147611482e-06,
      "loss": 0.6995,
      "step": 11037
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.94079852104187,
      "learning_rate": 1.0970566345265714e-06,
      "loss": 0.8302,
      "step": 11038
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.015230655670166,
      "learning_rate": 1.0963308648410164e-06,
      "loss": 0.4213,
      "step": 11039
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.1474192142486572,
      "learning_rate": 1.0956053057436344e-06,
      "loss": 0.1237,
      "step": 11040
    },
    {
      "epoch": 0.79,
      "grad_norm": 23.366365432739258,
      "learning_rate": 1.0948799572735664e-06,
      "loss": 0.6556,
      "step": 11041
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.107675552368164,
      "learning_rate": 1.0941548194699442e-06,
      "loss": 0.729,
      "step": 11042
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.7733304500579834,
      "learning_rate": 1.093429892371885e-06,
      "loss": 0.0837,
      "step": 11043
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.713597536087036,
      "learning_rate": 1.0927051760184948e-06,
      "loss": 0.6618,
      "step": 11044
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.929187774658203,
      "learning_rate": 1.0919806704488716e-06,
      "loss": 0.1721,
      "step": 11045
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2921864986419678,
      "learning_rate": 1.0912563757021e-06,
      "loss": 0.0542,
      "step": 11046
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.631777286529541,
      "learning_rate": 1.090532291817249e-06,
      "loss": 0.8979,
      "step": 11047
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7961535453796387,
      "learning_rate": 1.0898084188333834e-06,
      "loss": 0.054,
      "step": 11048
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.444488525390625,
      "learning_rate": 1.0890847567895536e-06,
      "loss": 0.7919,
      "step": 11049
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.9381983280181885,
      "learning_rate": 1.0883613057247955e-06,
      "loss": 0.7139,
      "step": 11050
    },
    {
      "epoch": 0.79,
      "grad_norm": 18.235580444335938,
      "learning_rate": 1.0876380656781404e-06,
      "loss": 1.1233,
      "step": 11051
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.6146469116210938,
      "learning_rate": 1.0869150366886023e-06,
      "loss": 0.135,
      "step": 11052
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.3009655475616455,
      "learning_rate": 1.0861922187951846e-06,
      "loss": 0.1524,
      "step": 11053
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.035840034484863,
      "learning_rate": 1.085469612036884e-06,
      "loss": 0.7104,
      "step": 11054
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.551682472229004,
      "learning_rate": 1.0847472164526802e-06,
      "loss": 0.8262,
      "step": 11055
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.737147331237793,
      "learning_rate": 1.0840250320815427e-06,
      "loss": 0.5647,
      "step": 11056
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.893681526184082,
      "learning_rate": 1.0833030589624333e-06,
      "loss": 0.5588,
      "step": 11057
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.303018569946289,
      "learning_rate": 1.0825812971342997e-06,
      "loss": 0.772,
      "step": 11058
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.226698160171509,
      "learning_rate": 1.0818597466360742e-06,
      "loss": 0.1066,
      "step": 11059
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.1958184242248535,
      "learning_rate": 1.081138407506686e-06,
      "loss": 0.2138,
      "step": 11060
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.57476544380188,
      "learning_rate": 1.080417279785047e-06,
      "loss": 0.5974,
      "step": 11061
    },
    {
      "epoch": 0.79,
      "grad_norm": 14.24272346496582,
      "learning_rate": 1.0796963635100577e-06,
      "loss": 0.6067,
      "step": 11062
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.654860496520996,
      "learning_rate": 1.0789756587206118e-06,
      "loss": 0.7424,
      "step": 11063
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.482378005981445,
      "learning_rate": 1.0782551654555873e-06,
      "loss": 0.3944,
      "step": 11064
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.50056791305542,
      "learning_rate": 1.0775348837538507e-06,
      "loss": 0.4563,
      "step": 11065
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.767627000808716,
      "learning_rate": 1.076814813654261e-06,
      "loss": 0.134,
      "step": 11066
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1888058185577393,
      "learning_rate": 1.0760949551956623e-06,
      "loss": 0.0902,
      "step": 11067
    },
    {
      "epoch": 0.79,
      "grad_norm": 28.312557220458984,
      "learning_rate": 1.0753753084168877e-06,
      "loss": 0.3262,
      "step": 11068
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.87861967086792,
      "learning_rate": 1.0746558733567586e-06,
      "loss": 0.1427,
      "step": 11069
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2703940868377686,
      "learning_rate": 1.0739366500540898e-06,
      "loss": 0.0614,
      "step": 11070
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.908111810684204,
      "learning_rate": 1.0732176385476745e-06,
      "loss": 0.6891,
      "step": 11071
    },
    {
      "epoch": 0.79,
      "grad_norm": 16.684675216674805,
      "learning_rate": 1.0724988388763057e-06,
      "loss": 0.5874,
      "step": 11072
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.727231025695801,
      "learning_rate": 1.0717802510787579e-06,
      "loss": 0.3502,
      "step": 11073
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2845311164855957,
      "learning_rate": 1.0710618751937968e-06,
      "loss": 0.0924,
      "step": 11074
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.874916911125183,
      "learning_rate": 1.0703437112601738e-06,
      "loss": 0.0401,
      "step": 11075
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.035201072692871,
      "learning_rate": 1.0696257593166347e-06,
      "loss": 0.4268,
      "step": 11076
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.417816638946533,
      "learning_rate": 1.0689080194019087e-06,
      "loss": 0.1966,
      "step": 11077
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.212827205657959,
      "learning_rate": 1.0681904915547137e-06,
      "loss": 0.3882,
      "step": 11078
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1809189319610596,
      "learning_rate": 1.0674731758137597e-06,
      "loss": 0.0837,
      "step": 11079
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.603978157043457,
      "learning_rate": 1.0667560722177434e-06,
      "loss": 0.0984,
      "step": 11080
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7443543672561646,
      "learning_rate": 1.0660391808053467e-06,
      "loss": 0.0451,
      "step": 11081
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.8243608474731445,
      "learning_rate": 1.0653225016152485e-06,
      "loss": 0.133,
      "step": 11082
    },
    {
      "epoch": 0.79,
      "grad_norm": 20.726531982421875,
      "learning_rate": 1.0646060346861064e-06,
      "loss": 1.137,
      "step": 11083
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.016840934753418,
      "learning_rate": 1.0638897800565707e-06,
      "loss": 0.169,
      "step": 11084
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.239169120788574,
      "learning_rate": 1.063173737765284e-06,
      "loss": 0.6647,
      "step": 11085
    },
    {
      "epoch": 0.79,
      "grad_norm": 17.496116638183594,
      "learning_rate": 1.0624579078508728e-06,
      "loss": 1.2816,
      "step": 11086
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.076195240020752,
      "learning_rate": 1.0617422903519513e-06,
      "loss": 0.5916,
      "step": 11087
    },
    {
      "epoch": 0.8,
      "grad_norm": 13.8013916015625,
      "learning_rate": 1.0610268853071277e-06,
      "loss": 0.7892,
      "step": 11088
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.4196228981018066,
      "learning_rate": 1.0603116927549934e-06,
      "loss": 0.0626,
      "step": 11089
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.89706039428711,
      "learning_rate": 1.05959671273413e-06,
      "loss": 0.6649,
      "step": 11090
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.843693733215332,
      "learning_rate": 1.0588819452831095e-06,
      "loss": 0.8543,
      "step": 11091
    },
    {
      "epoch": 0.8,
      "grad_norm": 16.39290428161621,
      "learning_rate": 1.0581673904404903e-06,
      "loss": 0.8685,
      "step": 11092
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.6610939502716064,
      "learning_rate": 1.0574530482448185e-06,
      "loss": 0.1196,
      "step": 11093
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.7699735164642334,
      "learning_rate": 1.0567389187346343e-06,
      "loss": 0.1404,
      "step": 11094
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.956509828567505,
      "learning_rate": 1.0560250019484574e-06,
      "loss": 0.1009,
      "step": 11095
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.707488059997559,
      "learning_rate": 1.055311297924802e-06,
      "loss": 0.6531,
      "step": 11096
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.530259132385254,
      "learning_rate": 1.0545978067021722e-06,
      "loss": 0.2407,
      "step": 11097
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.2718963623046875,
      "learning_rate": 1.0538845283190563e-06,
      "loss": 0.186,
      "step": 11098
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.553406238555908,
      "learning_rate": 1.0531714628139317e-06,
      "loss": 0.0992,
      "step": 11099
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.962055206298828,
      "learning_rate": 1.0524586102252687e-06,
      "loss": 0.8127,
      "step": 11100
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.3814826011657715,
      "learning_rate": 1.0517459705915211e-06,
      "loss": 0.742,
      "step": 11101
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.01518440246582,
      "learning_rate": 1.0510335439511327e-06,
      "loss": 1.0262,
      "step": 11102
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.79434585571289,
      "learning_rate": 1.050321330342538e-06,
      "loss": 0.6881,
      "step": 11103
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.745253086090088,
      "learning_rate": 1.0496093298041571e-06,
      "loss": 0.1352,
      "step": 11104
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.054819345474243,
      "learning_rate": 1.0488975423743996e-06,
      "loss": 0.1359,
      "step": 11105
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.134617328643799,
      "learning_rate": 1.048185968091664e-06,
      "loss": 0.0792,
      "step": 11106
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.8958380222320557,
      "learning_rate": 1.0474746069943376e-06,
      "loss": 0.8171,
      "step": 11107
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.061883449554443,
      "learning_rate": 1.0467634591207932e-06,
      "loss": 0.8851,
      "step": 11108
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.670286178588867,
      "learning_rate": 1.0460525245093978e-06,
      "loss": 0.2781,
      "step": 11109
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.400195360183716,
      "learning_rate": 1.0453418031985018e-06,
      "loss": 0.1723,
      "step": 11110
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.106278896331787,
      "learning_rate": 1.044631295226447e-06,
      "loss": 0.7462,
      "step": 11111
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.360147714614868,
      "learning_rate": 1.0439210006315603e-06,
      "loss": 0.1088,
      "step": 11112
    },
    {
      "epoch": 0.8,
      "grad_norm": 11.34377670288086,
      "learning_rate": 1.0432109194521622e-06,
      "loss": 0.821,
      "step": 11113
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.960606813430786,
      "learning_rate": 1.0425010517265583e-06,
      "loss": 0.8902,
      "step": 11114
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.95506763458252,
      "learning_rate": 1.0417913974930405e-06,
      "loss": 0.8021,
      "step": 11115
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.8171255588531494,
      "learning_rate": 1.0410819567898962e-06,
      "loss": 0.1188,
      "step": 11116
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.746394157409668,
      "learning_rate": 1.040372729655395e-06,
      "loss": 0.1039,
      "step": 11117
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.934215784072876,
      "learning_rate": 1.0396637161277955e-06,
      "loss": 0.1368,
      "step": 11118
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.307796955108643,
      "learning_rate": 1.0389549162453506e-06,
      "loss": 0.9828,
      "step": 11119
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.966608762741089,
      "learning_rate": 1.038246330046293e-06,
      "loss": 0.1016,
      "step": 11120
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.620086193084717,
      "learning_rate": 1.0375379575688489e-06,
      "loss": 0.1079,
      "step": 11121
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.1220593452453613,
      "learning_rate": 1.0368297988512343e-06,
      "loss": 0.143,
      "step": 11122
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8315703868865967,
      "learning_rate": 1.0361218539316504e-06,
      "loss": 0.0572,
      "step": 11123
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.642443895339966,
      "learning_rate": 1.0354141228482872e-06,
      "loss": 0.1113,
      "step": 11124
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.921102523803711,
      "learning_rate": 1.0347066056393269e-06,
      "loss": 0.6356,
      "step": 11125
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.784803867340088,
      "learning_rate": 1.0339993023429357e-06,
      "loss": 0.6961,
      "step": 11126
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8773301839828491,
      "learning_rate": 1.0332922129972689e-06,
      "loss": 0.0762,
      "step": 11127
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3639020919799805,
      "learning_rate": 1.032585337640473e-06,
      "loss": 0.0849,
      "step": 11128
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1183836460113525,
      "learning_rate": 1.0318786763106815e-06,
      "loss": 0.0665,
      "step": 11129
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.048596382141113,
      "learning_rate": 1.0311722290460135e-06,
      "loss": 0.5946,
      "step": 11130
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.928408145904541,
      "learning_rate": 1.0304659958845835e-06,
      "loss": 0.7954,
      "step": 11131
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.89133071899414,
      "learning_rate": 1.0297599768644866e-06,
      "loss": 0.5227,
      "step": 11132
    },
    {
      "epoch": 0.8,
      "grad_norm": 14.535785675048828,
      "learning_rate": 1.029054172023809e-06,
      "loss": 0.979,
      "step": 11133
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.6007442474365234,
      "learning_rate": 1.0283485814006293e-06,
      "loss": 0.1377,
      "step": 11134
    },
    {
      "epoch": 0.8,
      "grad_norm": 11.384876251220703,
      "learning_rate": 1.0276432050330104e-06,
      "loss": 0.7325,
      "step": 11135
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.041118144989014,
      "learning_rate": 1.026938042959003e-06,
      "loss": 0.8115,
      "step": 11136
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.484492063522339,
      "learning_rate": 1.026233095216651e-06,
      "loss": 0.0907,
      "step": 11137
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1387932300567627,
      "learning_rate": 1.0255283618439815e-06,
      "loss": 0.0776,
      "step": 11138
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.434202194213867,
      "learning_rate": 1.0248238428790125e-06,
      "loss": 0.5566,
      "step": 11139
    },
    {
      "epoch": 0.8,
      "grad_norm": 19.99468231201172,
      "learning_rate": 1.0241195383597497e-06,
      "loss": 0.7302,
      "step": 11140
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.8412137031555176,
      "learning_rate": 1.0234154483241893e-06,
      "loss": 0.1506,
      "step": 11141
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.0879504680633545,
      "learning_rate": 1.0227115728103126e-06,
      "loss": 0.1398,
      "step": 11142
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.176326751708984,
      "learning_rate": 1.0220079118560927e-06,
      "loss": 0.7285,
      "step": 11143
    },
    {
      "epoch": 0.8,
      "grad_norm": 20.518766403198242,
      "learning_rate": 1.021304465499488e-06,
      "loss": 0.725,
      "step": 11144
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.974569082260132,
      "learning_rate": 1.020601233778446e-06,
      "loss": 0.7901,
      "step": 11145
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.160971164703369,
      "learning_rate": 1.0198982167309056e-06,
      "loss": 0.3336,
      "step": 11146
    },
    {
      "epoch": 0.8,
      "grad_norm": 24.933109283447266,
      "learning_rate": 1.0191954143947913e-06,
      "loss": 0.8456,
      "step": 11147
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4341778755187988,
      "learning_rate": 1.0184928268080157e-06,
      "loss": 0.0269,
      "step": 11148
    },
    {
      "epoch": 0.8,
      "grad_norm": 11.794394493103027,
      "learning_rate": 1.0177904540084804e-06,
      "loss": 0.6626,
      "step": 11149
    },
    {
      "epoch": 0.8,
      "grad_norm": 13.854583740234375,
      "learning_rate": 1.0170882960340777e-06,
      "loss": 0.8585,
      "step": 11150
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.181241989135742,
      "learning_rate": 1.0163863529226852e-06,
      "loss": 0.1193,
      "step": 11151
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.465866565704346,
      "learning_rate": 1.0156846247121694e-06,
      "loss": 0.9486,
      "step": 11152
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.106378555297852,
      "learning_rate": 1.0149831114403873e-06,
      "loss": 0.8527,
      "step": 11153
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.49226975440979,
      "learning_rate": 1.0142818131451827e-06,
      "loss": 0.1411,
      "step": 11154
    },
    {
      "epoch": 0.8,
      "grad_norm": 14.609804153442383,
      "learning_rate": 1.0135807298643874e-06,
      "loss": 0.7214,
      "step": 11155
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.640493392944336,
      "learning_rate": 1.0128798616358227e-06,
      "loss": 0.3925,
      "step": 11156
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.3088631629943848,
      "learning_rate": 1.0121792084972964e-06,
      "loss": 0.1191,
      "step": 11157
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.229217529296875,
      "learning_rate": 1.0114787704866063e-06,
      "loss": 0.7958,
      "step": 11158
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.1602563858032227,
      "learning_rate": 1.010778547641541e-06,
      "loss": 0.2035,
      "step": 11159
    },
    {
      "epoch": 0.8,
      "grad_norm": 26.16951560974121,
      "learning_rate": 1.0100785399998725e-06,
      "loss": 0.4983,
      "step": 11160
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.41843581199646,
      "learning_rate": 1.0093787475993627e-06,
      "loss": 0.0668,
      "step": 11161
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0991108417510986,
      "learning_rate": 1.0086791704777653e-06,
      "loss": 0.0607,
      "step": 11162
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9247915744781494,
      "learning_rate": 1.0079798086728194e-06,
      "loss": 0.1089,
      "step": 11163
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.905745029449463,
      "learning_rate": 1.0072806622222502e-06,
      "loss": 0.7228,
      "step": 11164
    },
    {
      "epoch": 0.8,
      "grad_norm": 13.499223709106445,
      "learning_rate": 1.0065817311637776e-06,
      "loss": 0.6522,
      "step": 11165
    },
    {
      "epoch": 0.8,
      "grad_norm": 35.97216033935547,
      "learning_rate": 1.0058830155351047e-06,
      "loss": 1.1028,
      "step": 11166
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.305910110473633,
      "learning_rate": 1.0051845153739248e-06,
      "loss": 0.613,
      "step": 11167
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.365870475769043,
      "learning_rate": 1.0044862307179188e-06,
      "loss": 0.8478,
      "step": 11168
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3251445293426514,
      "learning_rate": 1.0037881616047567e-06,
      "loss": 0.0805,
      "step": 11169
    },
    {
      "epoch": 0.8,
      "grad_norm": 14.092222213745117,
      "learning_rate": 1.0030903080720956e-06,
      "loss": 0.7891,
      "step": 11170
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9834094047546387,
      "learning_rate": 1.0023926701575848e-06,
      "loss": 0.6794,
      "step": 11171
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.991790294647217,
      "learning_rate": 1.0016952478988573e-06,
      "loss": 0.1325,
      "step": 11172
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.3504199981689453,
      "learning_rate": 1.0009980413335358e-06,
      "loss": 0.1157,
      "step": 11173
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.133423328399658,
      "learning_rate": 1.0003010504992338e-06,
      "loss": 0.0724,
      "step": 11174
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.171802520751953,
      "learning_rate": 9.996042754335511e-07,
      "loss": 0.7823,
      "step": 11175
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.0253987312316895,
      "learning_rate": 9.989077161740751e-07,
      "loss": 0.8009,
      "step": 11176
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.017122268676758,
      "learning_rate": 9.982113727583814e-07,
      "loss": 0.1466,
      "step": 11177
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.113951683044434,
      "learning_rate": 9.975152452240378e-07,
      "loss": 0.3493,
      "step": 11178
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2316408157348633,
      "learning_rate": 9.968193336085968e-07,
      "loss": 0.0665,
      "step": 11179
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.68123197555542,
      "learning_rate": 9.961236379495997e-07,
      "loss": 0.1082,
      "step": 11180
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.9557952880859375,
      "learning_rate": 9.954281582845765e-07,
      "loss": 0.4576,
      "step": 11181
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.309190273284912,
      "learning_rate": 9.947328946510455e-07,
      "loss": 0.9681,
      "step": 11182
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.3810389041900635,
      "learning_rate": 9.940378470865148e-07,
      "loss": 0.1941,
      "step": 11183
    },
    {
      "epoch": 0.8,
      "grad_norm": 11.046977996826172,
      "learning_rate": 9.93343015628479e-07,
      "loss": 0.5068,
      "step": 11184
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.685778617858887,
      "learning_rate": 9.926484003144217e-07,
      "loss": 0.4516,
      "step": 11185
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.588871002197266,
      "learning_rate": 9.91954001181813e-07,
      "loss": 0.7864,
      "step": 11186
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.7354366779327393,
      "learning_rate": 9.91259818268116e-07,
      "loss": 0.5861,
      "step": 11187
    },
    {
      "epoch": 0.8,
      "grad_norm": 13.280733108520508,
      "learning_rate": 9.90565851610778e-07,
      "loss": 0.6523,
      "step": 11188
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9165971279144287,
      "learning_rate": 9.898721012472345e-07,
      "loss": 0.1386,
      "step": 11189
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.024927854537964,
      "learning_rate": 9.891785672149129e-07,
      "loss": 0.0667,
      "step": 11190
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.3726158142089844,
      "learning_rate": 9.884852495512276e-07,
      "loss": 0.1762,
      "step": 11191
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.577820301055908,
      "learning_rate": 9.877921482935759e-07,
      "loss": 0.6586,
      "step": 11192
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.862494945526123,
      "learning_rate": 9.870992634793518e-07,
      "loss": 0.3695,
      "step": 11193
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.994048833847046,
      "learning_rate": 9.864065951459335e-07,
      "loss": 0.6605,
      "step": 11194
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.021021842956543,
      "learning_rate": 9.85714143330685e-07,
      "loss": 0.7019,
      "step": 11195
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.6950278282165527,
      "learning_rate": 9.85021908070965e-07,
      "loss": 0.1516,
      "step": 11196
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.8292958736419678,
      "learning_rate": 9.84329889404116e-07,
      "loss": 0.8292,
      "step": 11197
    },
    {
      "epoch": 0.8,
      "grad_norm": 16.653079986572266,
      "learning_rate": 9.836380873674679e-07,
      "loss": 1.348,
      "step": 11198
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.512829780578613,
      "learning_rate": 9.82946501998343e-07,
      "loss": 0.7545,
      "step": 11199
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5266129970550537,
      "learning_rate": 9.822551333340496e-07,
      "loss": 0.0827,
      "step": 11200
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.824743270874023,
      "learning_rate": 9.815639814118822e-07,
      "loss": 0.5421,
      "step": 11201
    },
    {
      "epoch": 0.8,
      "grad_norm": 19.561508178710938,
      "learning_rate": 9.808730462691285e-07,
      "loss": 0.9916,
      "step": 11202
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9009101390838623,
      "learning_rate": 9.801823279430622e-07,
      "loss": 0.1352,
      "step": 11203
    },
    {
      "epoch": 0.8,
      "grad_norm": 11.701443672180176,
      "learning_rate": 9.794918264709413e-07,
      "loss": 0.8368,
      "step": 11204
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.329153060913086,
      "learning_rate": 9.788015418900188e-07,
      "loss": 0.3533,
      "step": 11205
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.055112361907959,
      "learning_rate": 9.781114742375319e-07,
      "loss": 0.701,
      "step": 11206
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.27244234085083,
      "learning_rate": 9.774216235507062e-07,
      "loss": 0.1787,
      "step": 11207
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.161837100982666,
      "learning_rate": 9.767319898667589e-07,
      "loss": 0.815,
      "step": 11208
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0901215076446533,
      "learning_rate": 9.760425732228917e-07,
      "loss": 0.052,
      "step": 11209
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9488933086395264,
      "learning_rate": 9.753533736562953e-07,
      "loss": 0.1201,
      "step": 11210
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5644443035125732,
      "learning_rate": 9.746643912041515e-07,
      "loss": 0.1251,
      "step": 11211
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.648017406463623,
      "learning_rate": 9.739756259036265e-07,
      "loss": 0.7872,
      "step": 11212
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.6593377590179443,
      "learning_rate": 9.732870777918779e-07,
      "loss": 0.1663,
      "step": 11213
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.912665367126465,
      "learning_rate": 9.725987469060478e-07,
      "loss": 0.6521,
      "step": 11214
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.391407012939453,
      "learning_rate": 9.719106332832733e-07,
      "loss": 0.7478,
      "step": 11215
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.731439590454102,
      "learning_rate": 9.712227369606703e-07,
      "loss": 0.6808,
      "step": 11216
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.5734963417053223,
      "learning_rate": 9.705350579753526e-07,
      "loss": 0.224,
      "step": 11217
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.5381417274475098,
      "learning_rate": 9.698475963644161e-07,
      "loss": 0.5826,
      "step": 11218
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.585068464279175,
      "learning_rate": 9.691603521649463e-07,
      "loss": 0.1234,
      "step": 11219
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.79090690612793,
      "learning_rate": 9.684733254140172e-07,
      "loss": 0.5277,
      "step": 11220
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.506380081176758,
      "learning_rate": 9.677865161486933e-07,
      "loss": 0.8287,
      "step": 11221
    },
    {
      "epoch": 0.8,
      "grad_norm": 26.0333194732666,
      "learning_rate": 9.670999244060237e-07,
      "loss": 0.7779,
      "step": 11222
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.263904571533203,
      "learning_rate": 9.66413550223047e-07,
      "loss": 0.4169,
      "step": 11223
    },
    {
      "epoch": 0.8,
      "grad_norm": 14.100922584533691,
      "learning_rate": 9.657273936367922e-07,
      "loss": 1.1609,
      "step": 11224
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.978121757507324,
      "learning_rate": 9.65041454684274e-07,
      "loss": 0.1679,
      "step": 11225
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5217556953430176,
      "learning_rate": 9.64355733402495e-07,
      "loss": 0.092,
      "step": 11226
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.4208767414093018,
      "learning_rate": 9.636702298284513e-07,
      "loss": 0.1831,
      "step": 11227
    },
    {
      "epoch": 0.81,
      "grad_norm": 19.283777236938477,
      "learning_rate": 9.62984943999119e-07,
      "loss": 0.6206,
      "step": 11228
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.287765026092529,
      "learning_rate": 9.622998759514661e-07,
      "loss": 0.2896,
      "step": 11229
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.370995044708252,
      "learning_rate": 9.616150257224533e-07,
      "loss": 0.1055,
      "step": 11230
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.2304444313049316,
      "learning_rate": 9.609303933490237e-07,
      "loss": 0.1903,
      "step": 11231
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.336270570755005,
      "learning_rate": 9.60245978868109e-07,
      "loss": 0.0722,
      "step": 11232
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.210955619812012,
      "learning_rate": 9.595617823166342e-07,
      "loss": 0.4901,
      "step": 11233
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.9259324073791504,
      "learning_rate": 9.588778037315073e-07,
      "loss": 0.1093,
      "step": 11234
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.212434768676758,
      "learning_rate": 9.58194043149625e-07,
      "loss": 0.8495,
      "step": 11235
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.466017723083496,
      "learning_rate": 9.57510500607876e-07,
      "loss": 0.1161,
      "step": 11236
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.458937406539917,
      "learning_rate": 9.568271761431342e-07,
      "loss": 0.1733,
      "step": 11237
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.876642227172852,
      "learning_rate": 9.561440697922608e-07,
      "loss": 0.866,
      "step": 11238
    },
    {
      "epoch": 0.81,
      "grad_norm": 14.845358848571777,
      "learning_rate": 9.5546118159211e-07,
      "loss": 0.6769,
      "step": 11239
    },
    {
      "epoch": 0.81,
      "grad_norm": 13.032732009887695,
      "learning_rate": 9.547785115795184e-07,
      "loss": 0.8082,
      "step": 11240
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.229306697845459,
      "learning_rate": 9.540960597913125e-07,
      "loss": 0.4523,
      "step": 11241
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1211233139038086,
      "learning_rate": 9.534138262643111e-07,
      "loss": 0.0697,
      "step": 11242
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.598965644836426,
      "learning_rate": 9.527318110353162e-07,
      "loss": 0.3594,
      "step": 11243
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.6858725547790527,
      "learning_rate": 9.520500141411193e-07,
      "loss": 0.6815,
      "step": 11244
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.342916965484619,
      "learning_rate": 9.513684356185032e-07,
      "loss": 0.6193,
      "step": 11245
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.262979507446289,
      "learning_rate": 9.506870755042352e-07,
      "loss": 0.2261,
      "step": 11246
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.4596235752105713,
      "learning_rate": 9.50005933835072e-07,
      "loss": 0.094,
      "step": 11247
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.417628765106201,
      "learning_rate": 9.49325010647757e-07,
      "loss": 0.7698,
      "step": 11248
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.139009952545166,
      "learning_rate": 9.486443059790262e-07,
      "loss": 0.0603,
      "step": 11249
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.480339288711548,
      "learning_rate": 9.479638198656004e-07,
      "loss": 0.0667,
      "step": 11250
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.768474578857422,
      "learning_rate": 9.472835523441886e-07,
      "loss": 0.7313,
      "step": 11251
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.1503021717071533,
      "learning_rate": 9.466035034514886e-07,
      "loss": 0.1231,
      "step": 11252
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.243666887283325,
      "learning_rate": 9.459236732241855e-07,
      "loss": 0.2386,
      "step": 11253
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.8777568340301514,
      "learning_rate": 9.452440616989555e-07,
      "loss": 0.0742,
      "step": 11254
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.012251853942871,
      "learning_rate": 9.44564668912461e-07,
      "loss": 0.9018,
      "step": 11255
    },
    {
      "epoch": 0.81,
      "grad_norm": 13.138367652893066,
      "learning_rate": 9.438854949013516e-07,
      "loss": 0.6512,
      "step": 11256
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.638211727142334,
      "learning_rate": 9.432065397022655e-07,
      "loss": 0.1017,
      "step": 11257
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2698755264282227,
      "learning_rate": 9.425278033518315e-07,
      "loss": 0.056,
      "step": 11258
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5670864582061768,
      "learning_rate": 9.418492858866645e-07,
      "loss": 0.1104,
      "step": 11259
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.275937557220459,
      "learning_rate": 9.411709873433666e-07,
      "loss": 0.0646,
      "step": 11260
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.4474756717681885,
      "learning_rate": 9.404929077585312e-07,
      "loss": 0.0537,
      "step": 11261
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.457709312438965,
      "learning_rate": 9.398150471687378e-07,
      "loss": 0.5341,
      "step": 11262
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.9848005771636963,
      "learning_rate": 9.39137405610554e-07,
      "loss": 0.1016,
      "step": 11263
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1874260902404785,
      "learning_rate": 9.384599831205354e-07,
      "loss": 0.0889,
      "step": 11264
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.950358510017395,
      "learning_rate": 9.377827797352274e-07,
      "loss": 0.0527,
      "step": 11265
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.815641164779663,
      "learning_rate": 9.371057954911606e-07,
      "loss": 0.8723,
      "step": 11266
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.625279664993286,
      "learning_rate": 9.364290304248591e-07,
      "loss": 0.1404,
      "step": 11267
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.801729202270508,
      "learning_rate": 9.357524845728294e-07,
      "loss": 0.644,
      "step": 11268
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.212698936462402,
      "learning_rate": 9.350761579715678e-07,
      "loss": 0.7299,
      "step": 11269
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.156956672668457,
      "learning_rate": 9.344000506575623e-07,
      "loss": 0.4361,
      "step": 11270
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.9003779888153076,
      "learning_rate": 9.33724162667285e-07,
      "loss": 0.1298,
      "step": 11271
    },
    {
      "epoch": 0.81,
      "grad_norm": 15.771173477172852,
      "learning_rate": 9.330484940371964e-07,
      "loss": 0.8987,
      "step": 11272
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.30423355102539,
      "learning_rate": 9.323730448037482e-07,
      "loss": 0.4949,
      "step": 11273
    },
    {
      "epoch": 0.81,
      "grad_norm": 25.247055053710938,
      "learning_rate": 9.316978150033773e-07,
      "loss": 0.7663,
      "step": 11274
    },
    {
      "epoch": 0.81,
      "grad_norm": 14.13038158416748,
      "learning_rate": 9.310228046725101e-07,
      "loss": 0.8166,
      "step": 11275
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9714311361312866,
      "learning_rate": 9.303480138475612e-07,
      "loss": 0.0682,
      "step": 11276
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.234320640563965,
      "learning_rate": 9.296734425649318e-07,
      "loss": 0.8221,
      "step": 11277
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8902767896652222,
      "learning_rate": 9.28999090861012e-07,
      "loss": 0.037,
      "step": 11278
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.1121063232421875,
      "learning_rate": 9.283249587721832e-07,
      "loss": 0.7957,
      "step": 11279
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.4956459999084473,
      "learning_rate": 9.276510463348104e-07,
      "loss": 0.1478,
      "step": 11280
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.6546506881713867,
      "learning_rate": 9.269773535852478e-07,
      "loss": 0.1345,
      "step": 11281
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.888213634490967,
      "learning_rate": 9.263038805598413e-07,
      "loss": 0.2561,
      "step": 11282
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7564351558685303,
      "learning_rate": 9.256306272949206e-07,
      "loss": 0.7534,
      "step": 11283
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.492884635925293,
      "learning_rate": 9.249575938268052e-07,
      "loss": 0.608,
      "step": 11284
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.808413505554199,
      "learning_rate": 9.242847801918014e-07,
      "loss": 0.7251,
      "step": 11285
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.243522644042969,
      "learning_rate": 9.236121864262071e-07,
      "loss": 1.43,
      "step": 11286
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.663687229156494,
      "learning_rate": 9.229398125663058e-07,
      "loss": 0.1086,
      "step": 11287
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0481441020965576,
      "learning_rate": 9.222676586483692e-07,
      "loss": 0.0887,
      "step": 11288
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.658083915710449,
      "learning_rate": 9.215957247086577e-07,
      "loss": 0.1408,
      "step": 11289
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2274887561798096,
      "learning_rate": 9.209240107834172e-07,
      "loss": 0.0863,
      "step": 11290
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.3083863258361816,
      "learning_rate": 9.202525169088878e-07,
      "loss": 0.0875,
      "step": 11291
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.949789524078369,
      "learning_rate": 9.195812431212925e-07,
      "loss": 0.1284,
      "step": 11292
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.611757755279541,
      "learning_rate": 9.189101894568442e-07,
      "loss": 0.4108,
      "step": 11293
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.752146005630493,
      "learning_rate": 9.182393559517417e-07,
      "loss": 0.2112,
      "step": 11294
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.85698890686035,
      "learning_rate": 9.17568742642177e-07,
      "loss": 0.7007,
      "step": 11295
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.001232147216797,
      "learning_rate": 9.168983495643258e-07,
      "loss": 0.825,
      "step": 11296
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5824244022369385,
      "learning_rate": 9.162281767543518e-07,
      "loss": 0.1041,
      "step": 11297
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1600069999694824,
      "learning_rate": 9.155582242484112e-07,
      "loss": 0.1118,
      "step": 11298
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.5936930179595947,
      "learning_rate": 9.148884920826451e-07,
      "loss": 0.2497,
      "step": 11299
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.9236884117126465,
      "learning_rate": 9.142189802931795e-07,
      "loss": 0.2711,
      "step": 11300
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7771029472351074,
      "learning_rate": 9.135496889161361e-07,
      "loss": 0.9523,
      "step": 11301
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.529683828353882,
      "learning_rate": 9.128806179876181e-07,
      "loss": 0.7289,
      "step": 11302
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.165337085723877,
      "learning_rate": 9.122117675437192e-07,
      "loss": 0.5001,
      "step": 11303
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2149810791015625,
      "learning_rate": 9.115431376205236e-07,
      "loss": 0.0676,
      "step": 11304
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5824923515319824,
      "learning_rate": 9.108747282541003e-07,
      "loss": 0.0849,
      "step": 11305
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.54547119140625,
      "learning_rate": 9.10206539480506e-07,
      "loss": 0.7285,
      "step": 11306
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.537005424499512,
      "learning_rate": 9.09538571335789e-07,
      "loss": 0.3594,
      "step": 11307
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.148515462875366,
      "learning_rate": 9.088708238559829e-07,
      "loss": 0.1307,
      "step": 11308
    },
    {
      "epoch": 0.81,
      "grad_norm": 24.124652862548828,
      "learning_rate": 9.082032970771093e-07,
      "loss": 1.071,
      "step": 11309
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.3147904872894287,
      "learning_rate": 9.075359910351806e-07,
      "loss": 0.1127,
      "step": 11310
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.4254767894744873,
      "learning_rate": 9.068689057661955e-07,
      "loss": 0.1012,
      "step": 11311
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.250940322875977,
      "learning_rate": 9.062020413061368e-07,
      "loss": 0.7476,
      "step": 11312
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.263753890991211,
      "learning_rate": 9.055353976909837e-07,
      "loss": 0.1163,
      "step": 11313
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.875678062438965,
      "learning_rate": 9.048689749566975e-07,
      "loss": 0.8487,
      "step": 11314
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.1529343128204346,
      "learning_rate": 9.042027731392278e-07,
      "loss": 0.1811,
      "step": 11315
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.279800415039062,
      "learning_rate": 9.035367922745165e-07,
      "loss": 0.3939,
      "step": 11316
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.454690456390381,
      "learning_rate": 9.028710323984891e-07,
      "loss": 0.0829,
      "step": 11317
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5862650871276855,
      "learning_rate": 9.022054935470598e-07,
      "loss": 0.0347,
      "step": 11318
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.944868087768555,
      "learning_rate": 9.015401757561349e-07,
      "loss": 0.976,
      "step": 11319
    },
    {
      "epoch": 0.81,
      "grad_norm": 13.381741523742676,
      "learning_rate": 9.008750790616033e-07,
      "loss": 0.6462,
      "step": 11320
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.937118053436279,
      "learning_rate": 9.002102034993454e-07,
      "loss": 0.5343,
      "step": 11321
    },
    {
      "epoch": 0.81,
      "grad_norm": 18.90721321105957,
      "learning_rate": 8.995455491052274e-07,
      "loss": 0.9197,
      "step": 11322
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.480748653411865,
      "learning_rate": 8.988811159151084e-07,
      "loss": 0.7004,
      "step": 11323
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.966636657714844,
      "learning_rate": 8.98216903964827e-07,
      "loss": 0.5598,
      "step": 11324
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.522674322128296,
      "learning_rate": 8.975529132902189e-07,
      "loss": 0.1261,
      "step": 11325
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.449610710144043,
      "learning_rate": 8.968891439271022e-07,
      "loss": 0.0665,
      "step": 11326
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.817513465881348,
      "learning_rate": 8.962255959112854e-07,
      "loss": 0.653,
      "step": 11327
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.851066589355469,
      "learning_rate": 8.955622692785626e-07,
      "loss": 0.6479,
      "step": 11328
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.025505304336548,
      "learning_rate": 8.948991640647209e-07,
      "loss": 0.1288,
      "step": 11329
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.948573589324951,
      "learning_rate": 8.942362803055299e-07,
      "loss": 0.7557,
      "step": 11330
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.472902774810791,
      "learning_rate": 8.935736180367499e-07,
      "loss": 0.389,
      "step": 11331
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.33896541595459,
      "learning_rate": 8.929111772941307e-07,
      "loss": 0.3497,
      "step": 11332
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.553695678710938,
      "learning_rate": 8.92248958113407e-07,
      "loss": 0.6082,
      "step": 11333
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8694729804992676,
      "learning_rate": 8.915869605303018e-07,
      "loss": 0.0544,
      "step": 11334
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.423313856124878,
      "learning_rate": 8.909251845805305e-07,
      "loss": 0.0875,
      "step": 11335
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.347243547439575,
      "learning_rate": 8.902636302997936e-07,
      "loss": 0.1796,
      "step": 11336
    },
    {
      "epoch": 0.81,
      "grad_norm": 24.45623207092285,
      "learning_rate": 8.896022977237745e-07,
      "loss": 0.6465,
      "step": 11337
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.1493587493896484,
      "learning_rate": 8.889411868881543e-07,
      "loss": 0.2212,
      "step": 11338
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.61183500289917,
      "learning_rate": 8.882802978285954e-07,
      "loss": 0.4025,
      "step": 11339
    },
    {
      "epoch": 0.81,
      "grad_norm": 29.494434356689453,
      "learning_rate": 8.8761963058075e-07,
      "loss": 1.0999,
      "step": 11340
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.3569493293762207,
      "learning_rate": 8.8695918518026e-07,
      "loss": 0.1389,
      "step": 11341
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.2008092403411865,
      "learning_rate": 8.862989616627538e-07,
      "loss": 0.1466,
      "step": 11342
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.232211589813232,
      "learning_rate": 8.85638960063846e-07,
      "loss": 0.153,
      "step": 11343
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.53924560546875,
      "learning_rate": 8.849791804191437e-07,
      "loss": 1.0073,
      "step": 11344
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.059443950653076,
      "learning_rate": 8.843196227642386e-07,
      "loss": 0.0789,
      "step": 11345
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.081754207611084,
      "learning_rate": 8.836602871347099e-07,
      "loss": 0.743,
      "step": 11346
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.787498950958252,
      "learning_rate": 8.830011735661287e-07,
      "loss": 0.0893,
      "step": 11347
    },
    {
      "epoch": 0.81,
      "grad_norm": 14.879264831542969,
      "learning_rate": 8.82342282094052e-07,
      "loss": 0.5908,
      "step": 11348
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.465128898620605,
      "learning_rate": 8.816836127540202e-07,
      "loss": 0.3081,
      "step": 11349
    },
    {
      "epoch": 0.81,
      "grad_norm": 11.388891220092773,
      "learning_rate": 8.810251655815699e-07,
      "loss": 0.7339,
      "step": 11350
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.871088743209839,
      "learning_rate": 8.803669406122211e-07,
      "loss": 0.1237,
      "step": 11351
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.450923919677734,
      "learning_rate": 8.797089378814804e-07,
      "loss": 0.61,
      "step": 11352
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.32822036743164,
      "learning_rate": 8.790511574248484e-07,
      "loss": 0.3134,
      "step": 11353
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.812162399291992,
      "learning_rate": 8.783935992778075e-07,
      "loss": 0.1965,
      "step": 11354
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.841001510620117,
      "learning_rate": 8.77736263475829e-07,
      "loss": 0.5408,
      "step": 11355
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.847916841506958,
      "learning_rate": 8.770791500543774e-07,
      "loss": 0.1594,
      "step": 11356
    },
    {
      "epoch": 0.81,
      "grad_norm": 11.979007720947266,
      "learning_rate": 8.764222590488997e-07,
      "loss": 1.0154,
      "step": 11357
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.784707069396973,
      "learning_rate": 8.75765590494832e-07,
      "loss": 0.4762,
      "step": 11358
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.412135124206543,
      "learning_rate": 8.751091444275983e-07,
      "loss": 0.7188,
      "step": 11359
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7206645011901855,
      "learning_rate": 8.744529208826158e-07,
      "loss": 0.589,
      "step": 11360
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.0447258949279785,
      "learning_rate": 8.737969198952795e-07,
      "loss": 0.2236,
      "step": 11361
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.287289142608643,
      "learning_rate": 8.731411415009817e-07,
      "loss": 0.4379,
      "step": 11362
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.684153079986572,
      "learning_rate": 8.724855857350983e-07,
      "loss": 0.7669,
      "step": 11363
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.336383104324341,
      "learning_rate": 8.718302526329947e-07,
      "loss": 0.1182,
      "step": 11364
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.528149127960205,
      "learning_rate": 8.711751422300213e-07,
      "loss": 0.6276,
      "step": 11365
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.492598056793213,
      "learning_rate": 8.705202545615216e-07,
      "loss": 0.1563,
      "step": 11366
    },
    {
      "epoch": 0.82,
      "grad_norm": 12.476045608520508,
      "learning_rate": 8.698655896628233e-07,
      "loss": 0.2318,
      "step": 11367
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.3449950218200684,
      "learning_rate": 8.692111475692422e-07,
      "loss": 0.1376,
      "step": 11368
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2237939834594727,
      "learning_rate": 8.685569283160849e-07,
      "loss": 0.0665,
      "step": 11369
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.965752124786377,
      "learning_rate": 8.679029319386429e-07,
      "loss": 0.2186,
      "step": 11370
    },
    {
      "epoch": 0.82,
      "grad_norm": 19.904611587524414,
      "learning_rate": 8.672491584721954e-07,
      "loss": 1.3268,
      "step": 11371
    },
    {
      "epoch": 0.82,
      "grad_norm": 12.009252548217773,
      "learning_rate": 8.665956079520149e-07,
      "loss": 0.8677,
      "step": 11372
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.049323558807373,
      "learning_rate": 8.659422804133549e-07,
      "loss": 0.7344,
      "step": 11373
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.464168548583984,
      "learning_rate": 8.652891758914589e-07,
      "loss": 0.7448,
      "step": 11374
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.1372432708740234,
      "learning_rate": 8.646362944215625e-07,
      "loss": 0.2471,
      "step": 11375
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2719435691833496,
      "learning_rate": 8.639836360388848e-07,
      "loss": 0.0546,
      "step": 11376
    },
    {
      "epoch": 0.82,
      "grad_norm": 15.585246086120605,
      "learning_rate": 8.633312007786332e-07,
      "loss": 0.4753,
      "step": 11377
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.36997127532959,
      "learning_rate": 8.626789886760062e-07,
      "loss": 0.4411,
      "step": 11378
    },
    {
      "epoch": 0.82,
      "grad_norm": 12.521855354309082,
      "learning_rate": 8.620269997661873e-07,
      "loss": 0.6307,
      "step": 11379
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.35770320892334,
      "learning_rate": 8.613752340843473e-07,
      "loss": 0.1819,
      "step": 11380
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.1295690536499023,
      "learning_rate": 8.607236916656491e-07,
      "loss": 0.1795,
      "step": 11381
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.001377820968628,
      "learning_rate": 8.600723725452397e-07,
      "loss": 0.1222,
      "step": 11382
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.870345115661621,
      "learning_rate": 8.594212767582543e-07,
      "loss": 0.4004,
      "step": 11383
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.515848636627197,
      "learning_rate": 8.58770404339821e-07,
      "loss": 0.302,
      "step": 11384
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.559234619140625,
      "learning_rate": 8.58119755325047e-07,
      "loss": 0.625,
      "step": 11385
    },
    {
      "epoch": 0.82,
      "grad_norm": 14.668309211730957,
      "learning_rate": 8.574693297490333e-07,
      "loss": 0.6458,
      "step": 11386
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.140887260437012,
      "learning_rate": 8.568191276468701e-07,
      "loss": 0.8472,
      "step": 11387
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.5397982597351074,
      "learning_rate": 8.561691490536323e-07,
      "loss": 0.2292,
      "step": 11388
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.086792945861816,
      "learning_rate": 8.555193940043821e-07,
      "loss": 0.7465,
      "step": 11389
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.9961845874786377,
      "learning_rate": 8.548698625341745e-07,
      "loss": 0.1547,
      "step": 11390
    },
    {
      "epoch": 0.82,
      "grad_norm": 14.199014663696289,
      "learning_rate": 8.542205546780475e-07,
      "loss": 0.7347,
      "step": 11391
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.027763843536377,
      "learning_rate": 8.53571470471029e-07,
      "loss": 0.7762,
      "step": 11392
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.428782939910889,
      "learning_rate": 8.52922609948133e-07,
      "loss": 0.6314,
      "step": 11393
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.054147243499756,
      "learning_rate": 8.522739731443663e-07,
      "loss": 0.7085,
      "step": 11394
    },
    {
      "epoch": 0.82,
      "grad_norm": 15.6380615234375,
      "learning_rate": 8.516255600947187e-07,
      "loss": 0.5995,
      "step": 11395
    },
    {
      "epoch": 0.82,
      "grad_norm": 21.100831985473633,
      "learning_rate": 8.509773708341695e-07,
      "loss": 0.9999,
      "step": 11396
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.91717267036438,
      "learning_rate": 8.503294053976868e-07,
      "loss": 0.1402,
      "step": 11397
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.063014030456543,
      "learning_rate": 8.496816638202238e-07,
      "loss": 0.1529,
      "step": 11398
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.3568685054779053,
      "learning_rate": 8.490341461367263e-07,
      "loss": 0.137,
      "step": 11399
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.646222114562988,
      "learning_rate": 8.483868523821243e-07,
      "loss": 0.6071,
      "step": 11400
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.583495140075684,
      "learning_rate": 8.477397825913369e-07,
      "loss": 0.5736,
      "step": 11401
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6380631923675537,
      "learning_rate": 8.470929367992697e-07,
      "loss": 0.1392,
      "step": 11402
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.391224384307861,
      "learning_rate": 8.464463150408209e-07,
      "loss": 0.1416,
      "step": 11403
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.9085471630096436,
      "learning_rate": 8.457999173508707e-07,
      "loss": 0.2156,
      "step": 11404
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.917659759521484,
      "learning_rate": 8.451537437642893e-07,
      "loss": 0.7685,
      "step": 11405
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.678267002105713,
      "learning_rate": 8.44507794315938e-07,
      "loss": 0.068,
      "step": 11406
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.926499366760254,
      "learning_rate": 8.438620690406612e-07,
      "loss": 0.3497,
      "step": 11407
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.246667861938477,
      "learning_rate": 8.432165679732945e-07,
      "loss": 0.5599,
      "step": 11408
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.34403133392334,
      "learning_rate": 8.425712911486594e-07,
      "loss": 0.3245,
      "step": 11409
    },
    {
      "epoch": 0.82,
      "grad_norm": 15.995855331420898,
      "learning_rate": 8.41926238601567e-07,
      "loss": 1.0891,
      "step": 11410
    },
    {
      "epoch": 0.82,
      "grad_norm": 13.601264953613281,
      "learning_rate": 8.412814103668127e-07,
      "loss": 0.6027,
      "step": 11411
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.0739047527313232,
      "learning_rate": 8.406368064791865e-07,
      "loss": 0.1319,
      "step": 11412
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8962150812149048,
      "learning_rate": 8.399924269734605e-07,
      "loss": 0.0444,
      "step": 11413
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.3433098793029785,
      "learning_rate": 8.393482718843954e-07,
      "loss": 0.0401,
      "step": 11414
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.321227073669434,
      "learning_rate": 8.387043412467432e-07,
      "loss": 0.3539,
      "step": 11415
    },
    {
      "epoch": 0.82,
      "grad_norm": 37.656063079833984,
      "learning_rate": 8.380606350952403e-07,
      "loss": 1.1175,
      "step": 11416
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.418178558349609,
      "learning_rate": 8.374171534646114e-07,
      "loss": 0.2308,
      "step": 11417
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6685703992843628,
      "learning_rate": 8.367738963895722e-07,
      "loss": 0.0593,
      "step": 11418
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.794142484664917,
      "learning_rate": 8.361308639048222e-07,
      "loss": 0.1859,
      "step": 11419
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.761789321899414,
      "learning_rate": 8.354880560450512e-07,
      "loss": 0.4105,
      "step": 11420
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.278275012969971,
      "learning_rate": 8.348454728449362e-07,
      "loss": 0.6721,
      "step": 11421
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.2547109127044678,
      "learning_rate": 8.342031143391416e-07,
      "loss": 0.102,
      "step": 11422
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.5638620853424072,
      "learning_rate": 8.335609805623196e-07,
      "loss": 0.0957,
      "step": 11423
    },
    {
      "epoch": 0.82,
      "grad_norm": 16.54890251159668,
      "learning_rate": 8.329190715491131e-07,
      "loss": 0.7478,
      "step": 11424
    },
    {
      "epoch": 0.82,
      "grad_norm": 16.614952087402344,
      "learning_rate": 8.322773873341494e-07,
      "loss": 0.681,
      "step": 11425
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.369945526123047,
      "learning_rate": 8.316359279520436e-07,
      "loss": 0.5824,
      "step": 11426
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.058284759521484,
      "learning_rate": 8.30994693437403e-07,
      "loss": 0.7751,
      "step": 11427
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.286269426345825,
      "learning_rate": 8.303536838248177e-07,
      "loss": 0.1031,
      "step": 11428
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.352143287658691,
      "learning_rate": 8.297128991488684e-07,
      "loss": 0.8289,
      "step": 11429
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.8847336769104,
      "learning_rate": 8.290723394441214e-07,
      "loss": 0.5091,
      "step": 11430
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0890891551971436,
      "learning_rate": 8.284320047451349e-07,
      "loss": 0.0628,
      "step": 11431
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6389104127883911,
      "learning_rate": 8.277918950864522e-07,
      "loss": 0.0524,
      "step": 11432
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6955020427703857,
      "learning_rate": 8.271520105026037e-07,
      "loss": 0.1385,
      "step": 11433
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.8480679988861084,
      "learning_rate": 8.265123510281093e-07,
      "loss": 0.7646,
      "step": 11434
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.9671359062194824,
      "learning_rate": 8.258729166974744e-07,
      "loss": 0.0926,
      "step": 11435
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.34519362449646,
      "learning_rate": 8.252337075451971e-07,
      "loss": 0.1055,
      "step": 11436
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1291120052337646,
      "learning_rate": 8.245947236057595e-07,
      "loss": 0.069,
      "step": 11437
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.182643890380859,
      "learning_rate": 8.239559649136314e-07,
      "loss": 0.759,
      "step": 11438
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2002310752868652,
      "learning_rate": 8.233174315032705e-07,
      "loss": 0.0573,
      "step": 11439
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.855058670043945,
      "learning_rate": 8.226791234091259e-07,
      "loss": 0.7977,
      "step": 11440
    },
    {
      "epoch": 0.82,
      "grad_norm": 12.859302520751953,
      "learning_rate": 8.220410406656304e-07,
      "loss": 0.6941,
      "step": 11441
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.604799747467041,
      "learning_rate": 8.214031833072051e-07,
      "loss": 0.1174,
      "step": 11442
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.477473258972168,
      "learning_rate": 8.207655513682627e-07,
      "loss": 0.7461,
      "step": 11443
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.727290391921997,
      "learning_rate": 8.201281448832004e-07,
      "loss": 0.1878,
      "step": 11444
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.261514663696289,
      "learning_rate": 8.194909638864007e-07,
      "loss": 0.4891,
      "step": 11445
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.9091098308563232,
      "learning_rate": 8.188540084122409e-07,
      "loss": 0.7728,
      "step": 11446
    },
    {
      "epoch": 0.82,
      "grad_norm": 13.583664894104004,
      "learning_rate": 8.182172784950803e-07,
      "loss": 0.5715,
      "step": 11447
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.801762342453003,
      "learning_rate": 8.175807741692681e-07,
      "loss": 0.7913,
      "step": 11448
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6069695949554443,
      "learning_rate": 8.169444954691424e-07,
      "loss": 0.1421,
      "step": 11449
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.154655694961548,
      "learning_rate": 8.16308442429028e-07,
      "loss": 0.1383,
      "step": 11450
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.2380406856536865,
      "learning_rate": 8.156726150832356e-07,
      "loss": 0.1253,
      "step": 11451
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.761369228363037,
      "learning_rate": 8.150370134660679e-07,
      "loss": 0.9175,
      "step": 11452
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.324460744857788,
      "learning_rate": 8.144016376118131e-07,
      "loss": 0.2116,
      "step": 11453
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.261811256408691,
      "learning_rate": 8.137664875547447e-07,
      "loss": 0.9394,
      "step": 11454
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.7573533058166504,
      "learning_rate": 8.131315633291298e-07,
      "loss": 0.6646,
      "step": 11455
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2504477500915527,
      "learning_rate": 8.124968649692205e-07,
      "loss": 0.0765,
      "step": 11456
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.951112985610962,
      "learning_rate": 8.118623925092523e-07,
      "loss": 0.0807,
      "step": 11457
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.2017781734466553,
      "learning_rate": 8.112281459834558e-07,
      "loss": 0.1349,
      "step": 11458
    },
    {
      "epoch": 0.82,
      "grad_norm": 25.372482299804688,
      "learning_rate": 8.105941254260463e-07,
      "loss": 0.7431,
      "step": 11459
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.67051887512207,
      "learning_rate": 8.09960330871224e-07,
      "loss": 0.7246,
      "step": 11460
    },
    {
      "epoch": 0.82,
      "grad_norm": 13.100137710571289,
      "learning_rate": 8.093267623531831e-07,
      "loss": 0.6778,
      "step": 11461
    },
    {
      "epoch": 0.82,
      "grad_norm": 32.553199768066406,
      "learning_rate": 8.086934199061008e-07,
      "loss": 0.6269,
      "step": 11462
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6028945446014404,
      "learning_rate": 8.080603035641426e-07,
      "loss": 0.081,
      "step": 11463
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4660964012145996,
      "learning_rate": 8.074274133614645e-07,
      "loss": 0.0874,
      "step": 11464
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4111545085906982,
      "learning_rate": 8.067947493322082e-07,
      "loss": 0.0401,
      "step": 11465
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.121582984924316,
      "learning_rate": 8.061623115105027e-07,
      "loss": 0.4399,
      "step": 11466
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.6603100299835205,
      "learning_rate": 8.055300999304649e-07,
      "loss": 0.1647,
      "step": 11467
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.927333116531372,
      "learning_rate": 8.048981146262036e-07,
      "loss": 0.1781,
      "step": 11468
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.581596612930298,
      "learning_rate": 8.04266355631807e-07,
      "loss": 0.0648,
      "step": 11469
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.209589958190918,
      "learning_rate": 8.036348229813606e-07,
      "loss": 0.1858,
      "step": 11470
    },
    {
      "epoch": 0.82,
      "grad_norm": 23.24662208557129,
      "learning_rate": 8.030035167089311e-07,
      "loss": 0.5526,
      "step": 11471
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.964784622192383,
      "learning_rate": 8.023724368485753e-07,
      "loss": 0.457,
      "step": 11472
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.9461169242858887,
      "learning_rate": 8.017415834343362e-07,
      "loss": 0.2148,
      "step": 11473
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.703943252563477,
      "learning_rate": 8.01110956500249e-07,
      "loss": 0.5593,
      "step": 11474
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.589725494384766,
      "learning_rate": 8.00480556080332e-07,
      "loss": 0.9635,
      "step": 11475
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.004612684249878,
      "learning_rate": 7.998503822085918e-07,
      "loss": 0.1346,
      "step": 11476
    },
    {
      "epoch": 0.82,
      "grad_norm": 19.338115692138672,
      "learning_rate": 7.992204349190258e-07,
      "loss": 0.8609,
      "step": 11477
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.715831756591797,
      "learning_rate": 7.985907142456173e-07,
      "loss": 0.1608,
      "step": 11478
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8697686195373535,
      "learning_rate": 7.979612202223347e-07,
      "loss": 0.0845,
      "step": 11479
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.9013583660125732,
      "learning_rate": 7.973319528831414e-07,
      "loss": 0.856,
      "step": 11480
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.777058601379395,
      "learning_rate": 7.967029122619802e-07,
      "loss": 0.4745,
      "step": 11481
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.09861421585083,
      "learning_rate": 7.96074098392785e-07,
      "loss": 0.7186,
      "step": 11482
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.916390419006348,
      "learning_rate": 7.954455113094806e-07,
      "loss": 0.7457,
      "step": 11483
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.899768114089966,
      "learning_rate": 7.948171510459763e-07,
      "loss": 0.0656,
      "step": 11484
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.5140509605407715,
      "learning_rate": 7.941890176361672e-07,
      "loss": 0.0718,
      "step": 11485
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.792596340179443,
      "learning_rate": 7.935611111139424e-07,
      "loss": 0.4295,
      "step": 11486
    },
    {
      "epoch": 0.82,
      "grad_norm": 15.011239051818848,
      "learning_rate": 7.929334315131737e-07,
      "loss": 0.8438,
      "step": 11487
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.478008508682251,
      "learning_rate": 7.923059788677201e-07,
      "loss": 0.1109,
      "step": 11488
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.582573413848877,
      "learning_rate": 7.916787532114334e-07,
      "loss": 0.4325,
      "step": 11489
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9728037118911743,
      "learning_rate": 7.910517545781487e-07,
      "loss": 0.0587,
      "step": 11490
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.6593596935272217,
      "learning_rate": 7.904249830016885e-07,
      "loss": 0.0835,
      "step": 11491
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.840316772460938,
      "learning_rate": 7.897984385158691e-07,
      "loss": 0.4202,
      "step": 11492
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.7968366146087646,
      "learning_rate": 7.891721211544862e-07,
      "loss": 0.1444,
      "step": 11493
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.630964279174805,
      "learning_rate": 7.885460309513276e-07,
      "loss": 0.4292,
      "step": 11494
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.341361045837402,
      "learning_rate": 7.879201679401699e-07,
      "loss": 0.757,
      "step": 11495
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4010016918182373,
      "learning_rate": 7.872945321547765e-07,
      "loss": 0.0993,
      "step": 11496
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.295228958129883,
      "learning_rate": 7.866691236288953e-07,
      "loss": 0.2275,
      "step": 11497
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.286850452423096,
      "learning_rate": 7.860439423962679e-07,
      "loss": 0.7969,
      "step": 11498
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.968289852142334,
      "learning_rate": 7.854189884906188e-07,
      "loss": 0.1316,
      "step": 11499
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.378765106201172,
      "learning_rate": 7.847942619456628e-07,
      "loss": 0.4091,
      "step": 11500
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.380688428878784,
      "learning_rate": 7.84169762795099e-07,
      "loss": 0.0884,
      "step": 11501
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.8779642581939697,
      "learning_rate": 7.835454910726203e-07,
      "loss": 0.7647,
      "step": 11502
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.690460205078125,
      "learning_rate": 7.829214468119023e-07,
      "loss": 0.2896,
      "step": 11503
    },
    {
      "epoch": 0.82,
      "grad_norm": 19.64668846130371,
      "learning_rate": 7.822976300466101e-07,
      "loss": 0.9174,
      "step": 11504
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.518831491470337,
      "learning_rate": 7.816740408103951e-07,
      "loss": 0.0682,
      "step": 11505
    },
    {
      "epoch": 0.83,
      "grad_norm": 18.386384963989258,
      "learning_rate": 7.810506791368977e-07,
      "loss": 0.7815,
      "step": 11506
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.477964401245117,
      "learning_rate": 7.804275450597476e-07,
      "loss": 0.6971,
      "step": 11507
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.721936225891113,
      "learning_rate": 7.798046386125596e-07,
      "loss": 0.6948,
      "step": 11508
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.766232490539551,
      "learning_rate": 7.791819598289368e-07,
      "loss": 0.7989,
      "step": 11509
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.389512777328491,
      "learning_rate": 7.785595087424696e-07,
      "loss": 0.1149,
      "step": 11510
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.004360198974609,
      "learning_rate": 7.77937285386739e-07,
      "loss": 0.8001,
      "step": 11511
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1677329540252686,
      "learning_rate": 7.773152897953107e-07,
      "loss": 0.0998,
      "step": 11512
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.484896659851074,
      "learning_rate": 7.766935220017374e-07,
      "loss": 0.5471,
      "step": 11513
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.752903461456299,
      "learning_rate": 7.760719820395635e-07,
      "loss": 0.0892,
      "step": 11514
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.141624927520752,
      "learning_rate": 7.754506699423181e-07,
      "loss": 0.401,
      "step": 11515
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.389246702194214,
      "learning_rate": 7.748295857435184e-07,
      "loss": 0.1224,
      "step": 11516
    },
    {
      "epoch": 0.83,
      "grad_norm": 11.41774845123291,
      "learning_rate": 7.742087294766687e-07,
      "loss": 0.7394,
      "step": 11517
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.0472829341888428,
      "learning_rate": 7.735881011752633e-07,
      "loss": 0.1255,
      "step": 11518
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.349043607711792,
      "learning_rate": 7.729677008727804e-07,
      "loss": 0.091,
      "step": 11519
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.3468401432037354,
      "learning_rate": 7.72347528602691e-07,
      "loss": 0.1473,
      "step": 11520
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.230710029602051,
      "learning_rate": 7.717275843984501e-07,
      "loss": 0.0745,
      "step": 11521
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.978446006774902,
      "learning_rate": 7.711078682934997e-07,
      "loss": 0.4106,
      "step": 11522
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.2432942390441895,
      "learning_rate": 7.704883803212743e-07,
      "loss": 0.8693,
      "step": 11523
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.036597728729248,
      "learning_rate": 7.698691205151909e-07,
      "loss": 0.4936,
      "step": 11524
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4414353370666504,
      "learning_rate": 7.692500889086551e-07,
      "loss": 0.1312,
      "step": 11525
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.736135482788086,
      "learning_rate": 7.686312855350642e-07,
      "loss": 1.0328,
      "step": 11526
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.223754644393921,
      "learning_rate": 7.68012710427799e-07,
      "loss": 0.1039,
      "step": 11527
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1624138355255127,
      "learning_rate": 7.673943636202286e-07,
      "loss": 0.0954,
      "step": 11528
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2635915279388428,
      "learning_rate": 7.667762451457117e-07,
      "loss": 0.0665,
      "step": 11529
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.192988872528076,
      "learning_rate": 7.661583550375923e-07,
      "loss": 0.1384,
      "step": 11530
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.7626047134399414,
      "learning_rate": 7.655406933292026e-07,
      "loss": 0.1303,
      "step": 11531
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.119543075561523,
      "learning_rate": 7.649232600538653e-07,
      "loss": 0.3643,
      "step": 11532
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0846877098083496,
      "learning_rate": 7.643060552448873e-07,
      "loss": 0.0939,
      "step": 11533
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.8410582542419434,
      "learning_rate": 7.636890789355638e-07,
      "loss": 0.1005,
      "step": 11534
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.485264778137207,
      "learning_rate": 7.630723311591803e-07,
      "loss": 0.638,
      "step": 11535
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.773608446121216,
      "learning_rate": 7.624558119490066e-07,
      "loss": 0.146,
      "step": 11536
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.620872497558594,
      "learning_rate": 7.618395213383023e-07,
      "loss": 0.7585,
      "step": 11537
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.072190761566162,
      "learning_rate": 7.612234593603123e-07,
      "loss": 0.0867,
      "step": 11538
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.306581497192383,
      "learning_rate": 7.606076260482731e-07,
      "loss": 0.7898,
      "step": 11539
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.934562683105469,
      "learning_rate": 7.599920214354056e-07,
      "loss": 0.3861,
      "step": 11540
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.300188064575195,
      "learning_rate": 7.59376645554919e-07,
      "loss": 0.4934,
      "step": 11541
    },
    {
      "epoch": 0.83,
      "grad_norm": 31.14249610900879,
      "learning_rate": 7.58761498440011e-07,
      "loss": 0.8637,
      "step": 11542
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.9227848052978516,
      "learning_rate": 7.581465801238647e-07,
      "loss": 0.6564,
      "step": 11543
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.574453353881836,
      "learning_rate": 7.575318906396556e-07,
      "loss": 0.1422,
      "step": 11544
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.040550947189331,
      "learning_rate": 7.569174300205423e-07,
      "loss": 0.0482,
      "step": 11545
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.7077739238739014,
      "learning_rate": 7.563031982996727e-07,
      "loss": 0.1603,
      "step": 11546
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2683281898498535,
      "learning_rate": 7.556891955101809e-07,
      "loss": 0.0838,
      "step": 11547
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.69411301612854,
      "learning_rate": 7.55075421685193e-07,
      "loss": 0.0713,
      "step": 11548
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.250420570373535,
      "learning_rate": 7.544618768578177e-07,
      "loss": 0.8679,
      "step": 11549
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.5213305950164795,
      "learning_rate": 7.538485610611529e-07,
      "loss": 0.0568,
      "step": 11550
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.9299230575561523,
      "learning_rate": 7.532354743282872e-07,
      "loss": 0.1411,
      "step": 11551
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.533424377441406,
      "learning_rate": 7.526226166922923e-07,
      "loss": 1.081,
      "step": 11552
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.7555992603302,
      "learning_rate": 7.520099881862302e-07,
      "loss": 0.0419,
      "step": 11553
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.876680612564087,
      "learning_rate": 7.513975888431502e-07,
      "loss": 0.0546,
      "step": 11554
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9849166870117188,
      "learning_rate": 7.507854186960878e-07,
      "loss": 0.0562,
      "step": 11555
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3918890953063965,
      "learning_rate": 7.501734777780667e-07,
      "loss": 0.0758,
      "step": 11556
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5018994808197021,
      "learning_rate": 7.495617661221016e-07,
      "loss": 0.0188,
      "step": 11557
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.6332154273986816,
      "learning_rate": 7.489502837611906e-07,
      "loss": 0.1684,
      "step": 11558
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.320455551147461,
      "learning_rate": 7.48339030728319e-07,
      "loss": 0.5388,
      "step": 11559
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.999938726425171,
      "learning_rate": 7.477280070564646e-07,
      "loss": 0.1496,
      "step": 11560
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9189521074295044,
      "learning_rate": 7.471172127785886e-07,
      "loss": 0.0877,
      "step": 11561
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.233994483947754,
      "learning_rate": 7.465066479276395e-07,
      "loss": 0.5195,
      "step": 11562
    },
    {
      "epoch": 0.83,
      "grad_norm": 32.398658752441406,
      "learning_rate": 7.458963125365581e-07,
      "loss": 0.6902,
      "step": 11563
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.943149089813232,
      "learning_rate": 7.452862066382676e-07,
      "loss": 0.3359,
      "step": 11564
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.3712799549102783,
      "learning_rate": 7.446763302656813e-07,
      "loss": 0.1757,
      "step": 11565
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2960195541381836,
      "learning_rate": 7.440666834516996e-07,
      "loss": 0.0874,
      "step": 11566
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.667149305343628,
      "learning_rate": 7.434572662292111e-07,
      "loss": 0.0462,
      "step": 11567
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.484853744506836,
      "learning_rate": 7.428480786310899e-07,
      "loss": 0.5598,
      "step": 11568
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.277315378189087,
      "learning_rate": 7.422391206902018e-07,
      "loss": 0.0851,
      "step": 11569
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.590971946716309,
      "learning_rate": 7.416303924393964e-07,
      "loss": 0.5583,
      "step": 11570
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.349393606185913,
      "learning_rate": 7.410218939115116e-07,
      "loss": 0.0498,
      "step": 11571
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.7396788597106934,
      "learning_rate": 7.404136251393757e-07,
      "loss": 0.1039,
      "step": 11572
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.0760886669158936,
      "learning_rate": 7.398055861558013e-07,
      "loss": 0.2262,
      "step": 11573
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.760393142700195,
      "learning_rate": 7.391977769935899e-07,
      "loss": 0.3563,
      "step": 11574
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.071382522583008,
      "learning_rate": 7.385901976855292e-07,
      "loss": 0.2333,
      "step": 11575
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.614356756210327,
      "learning_rate": 7.379828482643981e-07,
      "loss": 0.1366,
      "step": 11576
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.421564102172852,
      "learning_rate": 7.373757287629596e-07,
      "loss": 0.6218,
      "step": 11577
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.624862194061279,
      "learning_rate": 7.367688392139655e-07,
      "loss": 0.2889,
      "step": 11578
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.9513816833496094,
      "learning_rate": 7.361621796501556e-07,
      "loss": 0.1385,
      "step": 11579
    },
    {
      "epoch": 0.83,
      "grad_norm": 14.694462776184082,
      "learning_rate": 7.355557501042553e-07,
      "loss": 0.6529,
      "step": 11580
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.054971694946289,
      "learning_rate": 7.349495506089821e-07,
      "loss": 0.196,
      "step": 11581
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.266101360321045,
      "learning_rate": 7.343435811970362e-07,
      "loss": 0.1861,
      "step": 11582
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.220405578613281,
      "learning_rate": 7.337378419011076e-07,
      "loss": 0.8345,
      "step": 11583
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.844144344329834,
      "learning_rate": 7.331323327538731e-07,
      "loss": 0.0444,
      "step": 11584
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.309401512145996,
      "learning_rate": 7.325270537879986e-07,
      "loss": 0.8564,
      "step": 11585
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.087759017944336,
      "learning_rate": 7.31922005036137e-07,
      "loss": 0.8101,
      "step": 11586
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.276998281478882,
      "learning_rate": 7.313171865309266e-07,
      "loss": 0.2322,
      "step": 11587
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.703781843185425,
      "learning_rate": 7.307125983049968e-07,
      "loss": 0.0901,
      "step": 11588
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2544214725494385,
      "learning_rate": 7.301082403909631e-07,
      "loss": 0.0827,
      "step": 11589
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1404733657836914,
      "learning_rate": 7.295041128214258e-07,
      "loss": 0.1019,
      "step": 11590
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4587604999542236,
      "learning_rate": 7.289002156289776e-07,
      "loss": 0.0801,
      "step": 11591
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.324852228164673,
      "learning_rate": 7.282965488461962e-07,
      "loss": 0.1055,
      "step": 11592
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.236116886138916,
      "learning_rate": 7.276931125056453e-07,
      "loss": 0.1694,
      "step": 11593
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.0901541709899902,
      "learning_rate": 7.270899066398807e-07,
      "loss": 0.15,
      "step": 11594
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.0218210220336914,
      "learning_rate": 7.264869312814415e-07,
      "loss": 0.1858,
      "step": 11595
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3382797241210938,
      "learning_rate": 7.258841864628551e-07,
      "loss": 0.1089,
      "step": 11596
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2668325901031494,
      "learning_rate": 7.252816722166401e-07,
      "loss": 0.1517,
      "step": 11597
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3395588397979736,
      "learning_rate": 7.246793885752979e-07,
      "loss": 0.0812,
      "step": 11598
    },
    {
      "epoch": 0.83,
      "grad_norm": 56.1466064453125,
      "learning_rate": 7.240773355713182e-07,
      "loss": 0.8486,
      "step": 11599
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.412511825561523,
      "learning_rate": 7.234755132371824e-07,
      "loss": 0.7562,
      "step": 11600
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.0675313472747803,
      "learning_rate": 7.228739216053565e-07,
      "loss": 0.1413,
      "step": 11601
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2024283409118652,
      "learning_rate": 7.222725607082904e-07,
      "loss": 0.239,
      "step": 11602
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.8034555912017822,
      "learning_rate": 7.216714305784283e-07,
      "loss": 0.1461,
      "step": 11603
    },
    {
      "epoch": 0.83,
      "grad_norm": 20.20967674255371,
      "learning_rate": 7.210705312481986e-07,
      "loss": 0.6274,
      "step": 11604
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2311272621154785,
      "learning_rate": 7.204698627500162e-07,
      "loss": 0.1139,
      "step": 11605
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.9268910884857178,
      "learning_rate": 7.198694251162863e-07,
      "loss": 0.1833,
      "step": 11606
    },
    {
      "epoch": 0.83,
      "grad_norm": 13.71609115600586,
      "learning_rate": 7.192692183793998e-07,
      "loss": 0.8626,
      "step": 11607
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.847180366516113,
      "learning_rate": 7.186692425717351e-07,
      "loss": 0.4603,
      "step": 11608
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.385530471801758,
      "learning_rate": 7.180694977256597e-07,
      "loss": 0.1156,
      "step": 11609
    },
    {
      "epoch": 0.83,
      "grad_norm": 17.237199783325195,
      "learning_rate": 7.174699838735266e-07,
      "loss": 0.9472,
      "step": 11610
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6904391050338745,
      "learning_rate": 7.168707010476778e-07,
      "loss": 0.048,
      "step": 11611
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.665290117263794,
      "learning_rate": 7.162716492804412e-07,
      "loss": 0.0579,
      "step": 11612
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2317070960998535,
      "learning_rate": 7.156728286041365e-07,
      "loss": 0.0761,
      "step": 11613
    },
    {
      "epoch": 0.83,
      "grad_norm": 13.337936401367188,
      "learning_rate": 7.150742390510635e-07,
      "loss": 0.6228,
      "step": 11614
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.5899577140808105,
      "learning_rate": 7.144758806535168e-07,
      "loss": 0.1526,
      "step": 11615
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.044369220733643,
      "learning_rate": 7.138777534437752e-07,
      "loss": 0.8039,
      "step": 11616
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4801576137542725,
      "learning_rate": 7.13279857454105e-07,
      "loss": 0.111,
      "step": 11617
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.20840835571289,
      "learning_rate": 7.126821927167587e-07,
      "loss": 0.3929,
      "step": 11618
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.6357569694519043,
      "learning_rate": 7.120847592639812e-07,
      "loss": 0.0836,
      "step": 11619
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.261497497558594,
      "learning_rate": 7.11487557128e-07,
      "loss": 0.7862,
      "step": 11620
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9787813425064087,
      "learning_rate": 7.108905863410314e-07,
      "loss": 0.059,
      "step": 11621
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.7325146198272705,
      "learning_rate": 7.102938469352816e-07,
      "loss": 0.0965,
      "step": 11622
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.070809364318848,
      "learning_rate": 7.096973389429413e-07,
      "loss": 0.477,
      "step": 11623
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.1692962646484375,
      "learning_rate": 7.091010623961886e-07,
      "loss": 0.1712,
      "step": 11624
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.877915382385254,
      "learning_rate": 7.08505017327194e-07,
      "loss": 0.7515,
      "step": 11625
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.737391710281372,
      "learning_rate": 7.079092037681085e-07,
      "loss": 0.1031,
      "step": 11626
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.765594005584717,
      "learning_rate": 7.073136217510734e-07,
      "loss": 0.0658,
      "step": 11627
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.009925842285156,
      "learning_rate": 7.067182713082216e-07,
      "loss": 1.0166,
      "step": 11628
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.862257480621338,
      "learning_rate": 7.061231524716678e-07,
      "loss": 0.1425,
      "step": 11629
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.774470806121826,
      "learning_rate": 7.055282652735152e-07,
      "loss": 0.1881,
      "step": 11630
    },
    {
      "epoch": 0.83,
      "grad_norm": 11.835736274719238,
      "learning_rate": 7.049336097458586e-07,
      "loss": 0.472,
      "step": 11631
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.26932430267334,
      "learning_rate": 7.043391859207765e-07,
      "loss": 0.6377,
      "step": 11632
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.697098731994629,
      "learning_rate": 7.037449938303337e-07,
      "loss": 0.123,
      "step": 11633
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.439654350280762,
      "learning_rate": 7.031510335065878e-07,
      "loss": 0.9744,
      "step": 11634
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.446723461151123,
      "learning_rate": 7.025573049815787e-07,
      "loss": 0.0856,
      "step": 11635
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.75478982925415,
      "learning_rate": 7.019638082873353e-07,
      "loss": 0.6244,
      "step": 11636
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.19318962097168,
      "learning_rate": 7.013705434558782e-07,
      "loss": 0.6095,
      "step": 11637
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.581648588180542,
      "learning_rate": 7.007775105192071e-07,
      "loss": 0.1112,
      "step": 11638
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.808680295944214,
      "learning_rate": 7.001847095093156e-07,
      "loss": 0.1517,
      "step": 11639
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.928267002105713,
      "learning_rate": 6.995921404581841e-07,
      "loss": 0.8111,
      "step": 11640
    },
    {
      "epoch": 0.83,
      "grad_norm": 13.640008926391602,
      "learning_rate": 6.989998033977786e-07,
      "loss": 0.2361,
      "step": 11641
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.696337938308716,
      "learning_rate": 6.984076983600524e-07,
      "loss": 0.0746,
      "step": 11642
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.815094232559204,
      "learning_rate": 6.978158253769501e-07,
      "loss": 0.1153,
      "step": 11643
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.584681272506714,
      "learning_rate": 6.97224184480399e-07,
      "loss": 0.1583,
      "step": 11644
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.8966119289398193,
      "learning_rate": 6.966327757023161e-07,
      "loss": 0.0999,
      "step": 11645
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.011537551879883,
      "learning_rate": 6.96041599074605e-07,
      "loss": 0.5477,
      "step": 11646
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.9131455421447754,
      "learning_rate": 6.954506546291589e-07,
      "loss": 0.6362,
      "step": 11647
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.221598148345947,
      "learning_rate": 6.94859942397857e-07,
      "loss": 0.8671,
      "step": 11648
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.4296441078186035,
      "learning_rate": 6.942694624125646e-07,
      "loss": 0.7856,
      "step": 11649
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.137515544891357,
      "learning_rate": 6.936792147051375e-07,
      "loss": 0.6263,
      "step": 11650
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.88070547580719,
      "learning_rate": 6.930891993074146e-07,
      "loss": 0.0527,
      "step": 11651
    },
    {
      "epoch": 0.84,
      "grad_norm": 12.085066795349121,
      "learning_rate": 6.924994162512283e-07,
      "loss": 0.9885,
      "step": 11652
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.412341117858887,
      "learning_rate": 6.919098655683937e-07,
      "loss": 0.4694,
      "step": 11653
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.7058744430542,
      "learning_rate": 6.913205472907153e-07,
      "loss": 0.5221,
      "step": 11654
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.657582759857178,
      "learning_rate": 6.907314614499821e-07,
      "loss": 0.4867,
      "step": 11655
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.7648367881774902,
      "learning_rate": 6.901426080779772e-07,
      "loss": 0.1673,
      "step": 11656
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4693198204040527,
      "learning_rate": 6.895539872064644e-07,
      "loss": 0.0671,
      "step": 11657
    },
    {
      "epoch": 0.84,
      "grad_norm": 14.016326904296875,
      "learning_rate": 6.889655988671967e-07,
      "loss": 0.7242,
      "step": 11658
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.301168203353882,
      "learning_rate": 6.883774430919182e-07,
      "loss": 0.0857,
      "step": 11659
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.133640766143799,
      "learning_rate": 6.877895199123563e-07,
      "loss": 0.1114,
      "step": 11660
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.9518797397613525,
      "learning_rate": 6.872018293602267e-07,
      "loss": 0.1715,
      "step": 11661
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.689110279083252,
      "learning_rate": 6.866143714672341e-07,
      "loss": 0.2351,
      "step": 11662
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.600111722946167,
      "learning_rate": 6.860271462650692e-07,
      "loss": 0.0937,
      "step": 11663
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.5124359130859375,
      "learning_rate": 6.854401537854094e-07,
      "loss": 0.11,
      "step": 11664
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.7756896018981934,
      "learning_rate": 6.848533940599222e-07,
      "loss": 0.8338,
      "step": 11665
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.981829047203064,
      "learning_rate": 6.842668671202613e-07,
      "loss": 0.1721,
      "step": 11666
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.473362922668457,
      "learning_rate": 6.836805729980655e-07,
      "loss": 0.5383,
      "step": 11667
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.221134662628174,
      "learning_rate": 6.830945117249666e-07,
      "loss": 0.8339,
      "step": 11668
    },
    {
      "epoch": 0.84,
      "grad_norm": 16.34674072265625,
      "learning_rate": 6.825086833325773e-07,
      "loss": 0.2763,
      "step": 11669
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.561289072036743,
      "learning_rate": 6.819230878525013e-07,
      "loss": 0.5728,
      "step": 11670
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.762253999710083,
      "learning_rate": 6.813377253163311e-07,
      "loss": 0.1149,
      "step": 11671
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.784751534461975,
      "learning_rate": 6.807525957556438e-07,
      "loss": 0.0574,
      "step": 11672
    },
    {
      "epoch": 0.84,
      "grad_norm": 14.286134719848633,
      "learning_rate": 6.801676992020045e-07,
      "loss": 0.4578,
      "step": 11673
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.338893413543701,
      "learning_rate": 6.795830356869665e-07,
      "loss": 0.4889,
      "step": 11674
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.6585471630096436,
      "learning_rate": 6.789986052420694e-07,
      "loss": 0.6308,
      "step": 11675
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.475497245788574,
      "learning_rate": 6.784144078988414e-07,
      "loss": 0.6801,
      "step": 11676
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.365537166595459,
      "learning_rate": 6.778304436887984e-07,
      "loss": 0.9005,
      "step": 11677
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.3638405799865723,
      "learning_rate": 6.772467126434435e-07,
      "loss": 0.1129,
      "step": 11678
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.231437921524048,
      "learning_rate": 6.766632147942643e-07,
      "loss": 0.1018,
      "step": 11679
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.48470401763916,
      "learning_rate": 6.760799501727411e-07,
      "loss": 0.0784,
      "step": 11680
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.792654037475586,
      "learning_rate": 6.754969188103378e-07,
      "loss": 0.6263,
      "step": 11681
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.2471466064453125,
      "learning_rate": 6.749141207385063e-07,
      "loss": 0.8508,
      "step": 11682
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9463831186294556,
      "learning_rate": 6.743315559886854e-07,
      "loss": 0.0817,
      "step": 11683
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.352585792541504,
      "learning_rate": 6.73749224592305e-07,
      "loss": 0.7279,
      "step": 11684
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.959621906280518,
      "learning_rate": 6.731671265807782e-07,
      "loss": 0.6383,
      "step": 11685
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.819660186767578,
      "learning_rate": 6.725852619855067e-07,
      "loss": 0.4929,
      "step": 11686
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.717729091644287,
      "learning_rate": 6.720036308378808e-07,
      "loss": 0.6747,
      "step": 11687
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.184959650039673,
      "learning_rate": 6.714222331692749e-07,
      "loss": 0.1852,
      "step": 11688
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.051139831542969,
      "learning_rate": 6.708410690110562e-07,
      "loss": 0.6481,
      "step": 11689
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.4616260528564453,
      "learning_rate": 6.70260138394575e-07,
      "loss": 0.1807,
      "step": 11690
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.712672472000122,
      "learning_rate": 6.69679441351171e-07,
      "loss": 0.0982,
      "step": 11691
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.3655855655670166,
      "learning_rate": 6.690989779121682e-07,
      "loss": 0.2414,
      "step": 11692
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.555008888244629,
      "learning_rate": 6.685187481088834e-07,
      "loss": 0.7852,
      "step": 11693
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.278008937835693,
      "learning_rate": 6.679387519726165e-07,
      "loss": 0.7234,
      "step": 11694
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.1058669090271,
      "learning_rate": 6.673589895346555e-07,
      "loss": 0.7211,
      "step": 11695
    },
    {
      "epoch": 0.84,
      "grad_norm": 16.83368492126465,
      "learning_rate": 6.667794608262778e-07,
      "loss": 1.0375,
      "step": 11696
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.9742462635040283,
      "learning_rate": 6.662001658787476e-07,
      "loss": 0.6628,
      "step": 11697
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.436214447021484,
      "learning_rate": 6.656211047233113e-07,
      "loss": 1.1213,
      "step": 11698
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2635154724121094,
      "learning_rate": 6.650422773912113e-07,
      "loss": 0.1178,
      "step": 11699
    },
    {
      "epoch": 0.84,
      "grad_norm": 13.751453399658203,
      "learning_rate": 6.64463683913672e-07,
      "loss": 0.5289,
      "step": 11700
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.6656980514526367,
      "learning_rate": 6.638853243219045e-07,
      "loss": 0.6332,
      "step": 11701
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.664420247077942,
      "learning_rate": 6.633071986471123e-07,
      "loss": 0.0397,
      "step": 11702
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.6737189292907715,
      "learning_rate": 6.62729306920481e-07,
      "loss": 0.536,
      "step": 11703
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.3150014877319336,
      "learning_rate": 6.621516491731849e-07,
      "loss": 0.0684,
      "step": 11704
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.637765884399414,
      "learning_rate": 6.615742254363894e-07,
      "loss": 0.8277,
      "step": 11705
    },
    {
      "epoch": 0.84,
      "grad_norm": 17.993989944458008,
      "learning_rate": 6.609970357412421e-07,
      "loss": 0.9427,
      "step": 11706
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.968405246734619,
      "learning_rate": 6.604200801188799e-07,
      "loss": 0.8035,
      "step": 11707
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.804314136505127,
      "learning_rate": 6.598433586004288e-07,
      "loss": 0.2067,
      "step": 11708
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.737633228302002,
      "learning_rate": 6.59266871217002e-07,
      "loss": 0.1684,
      "step": 11709
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.911193370819092,
      "learning_rate": 6.586906179996938e-07,
      "loss": 0.0745,
      "step": 11710
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.209676265716553,
      "learning_rate": 6.581145989795956e-07,
      "loss": 0.5728,
      "step": 11711
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.110065460205078,
      "learning_rate": 6.575388141877804e-07,
      "loss": 0.1674,
      "step": 11712
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.413179874420166,
      "learning_rate": 6.569632636553069e-07,
      "loss": 0.1194,
      "step": 11713
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.9859371185302734,
      "learning_rate": 6.563879474132279e-07,
      "loss": 0.6517,
      "step": 11714
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.0268683433532715,
      "learning_rate": 6.558128654925772e-07,
      "loss": 0.5979,
      "step": 11715
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4732658863067627,
      "learning_rate": 6.552380179243778e-07,
      "loss": 0.1428,
      "step": 11716
    },
    {
      "epoch": 0.84,
      "grad_norm": 15.58327579498291,
      "learning_rate": 6.546634047396427e-07,
      "loss": 0.7445,
      "step": 11717
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.057950973510742,
      "learning_rate": 6.540890259693694e-07,
      "loss": 0.2707,
      "step": 11718
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6711199283599854,
      "learning_rate": 6.535148816445425e-07,
      "loss": 0.1166,
      "step": 11719
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5178375244140625,
      "learning_rate": 6.529409717961343e-07,
      "loss": 0.0504,
      "step": 11720
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6813950538635254,
      "learning_rate": 6.523672964551081e-07,
      "loss": 0.0524,
      "step": 11721
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.29744029045105,
      "learning_rate": 6.517938556524079e-07,
      "loss": 0.063,
      "step": 11722
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.550400733947754,
      "learning_rate": 6.51220649418971e-07,
      "loss": 0.0998,
      "step": 11723
    },
    {
      "epoch": 0.84,
      "grad_norm": 20.595914840698242,
      "learning_rate": 6.50647677785719e-07,
      "loss": 0.864,
      "step": 11724
    },
    {
      "epoch": 0.84,
      "grad_norm": 13.942109107971191,
      "learning_rate": 6.500749407835622e-07,
      "loss": 1.1043,
      "step": 11725
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.634852409362793,
      "learning_rate": 6.495024384433951e-07,
      "loss": 0.1067,
      "step": 11726
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.493250608444214,
      "learning_rate": 6.489301707961054e-07,
      "loss": 0.1402,
      "step": 11727
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6920359134674072,
      "learning_rate": 6.483581378725635e-07,
      "loss": 0.1202,
      "step": 11728
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.585473537445068,
      "learning_rate": 6.477863397036272e-07,
      "loss": 0.8378,
      "step": 11729
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.882676839828491,
      "learning_rate": 6.472147763201442e-07,
      "loss": 0.7205,
      "step": 11730
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4004099369049072,
      "learning_rate": 6.466434477529482e-07,
      "loss": 0.108,
      "step": 11731
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.078544616699219,
      "learning_rate": 6.46072354032859e-07,
      "loss": 1.0567,
      "step": 11732
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.143772125244141,
      "learning_rate": 6.455014951906879e-07,
      "loss": 0.27,
      "step": 11733
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.337251663208008,
      "learning_rate": 6.449308712572272e-07,
      "loss": 0.1163,
      "step": 11734
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.8662776947021484,
      "learning_rate": 6.443604822632604e-07,
      "loss": 0.1517,
      "step": 11735
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.884718418121338,
      "learning_rate": 6.437903282395591e-07,
      "loss": 0.4109,
      "step": 11736
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.016151428222656,
      "learning_rate": 6.432204092168808e-07,
      "loss": 0.7698,
      "step": 11737
    },
    {
      "epoch": 0.84,
      "grad_norm": 15.383170127868652,
      "learning_rate": 6.426507252259689e-07,
      "loss": 0.6626,
      "step": 11738
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.688654899597168,
      "learning_rate": 6.420812762975581e-07,
      "loss": 0.1135,
      "step": 11739
    },
    {
      "epoch": 0.84,
      "grad_norm": 12.17729663848877,
      "learning_rate": 6.415120624623672e-07,
      "loss": 0.9997,
      "step": 11740
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2162301540374756,
      "learning_rate": 6.40943083751101e-07,
      "loss": 0.0755,
      "step": 11741
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4049041271209717,
      "learning_rate": 6.403743401944573e-07,
      "loss": 0.1041,
      "step": 11742
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.850707769393921,
      "learning_rate": 6.398058318231154e-07,
      "loss": 0.0909,
      "step": 11743
    },
    {
      "epoch": 0.84,
      "grad_norm": 14.827214241027832,
      "learning_rate": 6.392375586677435e-07,
      "loss": 0.7841,
      "step": 11744
    },
    {
      "epoch": 0.84,
      "grad_norm": 15.083698272705078,
      "learning_rate": 6.38669520759001e-07,
      "loss": 0.6408,
      "step": 11745
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.203053951263428,
      "learning_rate": 6.381017181275284e-07,
      "loss": 0.8907,
      "step": 11746
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9207171201705933,
      "learning_rate": 6.375341508039567e-07,
      "loss": 0.0766,
      "step": 11747
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.77687668800354,
      "learning_rate": 6.369668188189054e-07,
      "loss": 0.8229,
      "step": 11748
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.306426525115967,
      "learning_rate": 6.363997222029794e-07,
      "loss": 0.8315,
      "step": 11749
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.198077917098999,
      "learning_rate": 6.358328609867703e-07,
      "loss": 0.0743,
      "step": 11750
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.354341506958008,
      "learning_rate": 6.352662352008599e-07,
      "loss": 0.2236,
      "step": 11751
    },
    {
      "epoch": 0.84,
      "grad_norm": 20.750080108642578,
      "learning_rate": 6.346998448758146e-07,
      "loss": 0.8985,
      "step": 11752
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4915897846221924,
      "learning_rate": 6.341336900421896e-07,
      "loss": 0.0899,
      "step": 11753
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.081066131591797,
      "learning_rate": 6.335677707305249e-07,
      "loss": 0.5869,
      "step": 11754
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.438773155212402,
      "learning_rate": 6.330020869713521e-07,
      "loss": 0.8042,
      "step": 11755
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.222707748413086,
      "learning_rate": 6.324366387951869e-07,
      "loss": 0.1479,
      "step": 11756
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.650249481201172,
      "learning_rate": 6.318714262325331e-07,
      "loss": 0.0676,
      "step": 11757
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.124591112136841,
      "learning_rate": 6.31306449313881e-07,
      "loss": 0.0979,
      "step": 11758
    },
    {
      "epoch": 0.84,
      "grad_norm": 13.742652893066406,
      "learning_rate": 6.307417080697093e-07,
      "loss": 0.6922,
      "step": 11759
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.5866880416870117,
      "learning_rate": 6.301772025304843e-07,
      "loss": 0.0914,
      "step": 11760
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9290659427642822,
      "learning_rate": 6.296129327266586e-07,
      "loss": 0.0668,
      "step": 11761
    },
    {
      "epoch": 0.84,
      "grad_norm": 14.515636444091797,
      "learning_rate": 6.290488986886728e-07,
      "loss": 0.9372,
      "step": 11762
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.491945505142212,
      "learning_rate": 6.284851004469528e-07,
      "loss": 0.2621,
      "step": 11763
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.7612180709838867,
      "learning_rate": 6.279215380319154e-07,
      "loss": 0.7126,
      "step": 11764
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.843390464782715,
      "learning_rate": 6.273582114739618e-07,
      "loss": 0.7389,
      "step": 11765
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.382371425628662,
      "learning_rate": 6.267951208034801e-07,
      "loss": 0.2049,
      "step": 11766
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.8762736320495605,
      "learning_rate": 6.262322660508496e-07,
      "loss": 0.668,
      "step": 11767
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.541731357574463,
      "learning_rate": 6.256696472464324e-07,
      "loss": 0.4315,
      "step": 11768
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.9329123497009277,
      "learning_rate": 6.251072644205796e-07,
      "loss": 0.1933,
      "step": 11769
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.995863914489746,
      "learning_rate": 6.245451176036316e-07,
      "loss": 0.0436,
      "step": 11770
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.8188364505767822,
      "learning_rate": 6.239832068259116e-07,
      "loss": 0.1117,
      "step": 11771
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.404555320739746,
      "learning_rate": 6.234215321177323e-07,
      "loss": 0.7421,
      "step": 11772
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.650136470794678,
      "learning_rate": 6.22860093509396e-07,
      "loss": 0.7538,
      "step": 11773
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.6844894886016846,
      "learning_rate": 6.222988910311895e-07,
      "loss": 0.7426,
      "step": 11774
    },
    {
      "epoch": 0.84,
      "grad_norm": 12.567973136901855,
      "learning_rate": 6.217379247133864e-07,
      "loss": 0.7095,
      "step": 11775
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.7623250484466553,
      "learning_rate": 6.211771945862505e-07,
      "loss": 0.6221,
      "step": 11776
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.7825403213500977,
      "learning_rate": 6.206167006800301e-07,
      "loss": 0.0981,
      "step": 11777
    },
    {
      "epoch": 0.84,
      "grad_norm": 21.442916870117188,
      "learning_rate": 6.200564430249612e-07,
      "loss": 0.6813,
      "step": 11778
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2845829725265503,
      "learning_rate": 6.194964216512689e-07,
      "loss": 0.0344,
      "step": 11779
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.929659605026245,
      "learning_rate": 6.189366365891635e-07,
      "loss": 0.6717,
      "step": 11780
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.44171667098999,
      "learning_rate": 6.183770878688422e-07,
      "loss": 0.8626,
      "step": 11781
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.199311256408691,
      "learning_rate": 6.178177755204945e-07,
      "loss": 0.5733,
      "step": 11782
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9581190347671509,
      "learning_rate": 6.172586995742886e-07,
      "loss": 0.0309,
      "step": 11783
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.086888313293457,
      "learning_rate": 6.166998600603852e-07,
      "loss": 0.6399,
      "step": 11784
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.754030466079712,
      "learning_rate": 6.161412570089343e-07,
      "loss": 0.0544,
      "step": 11785
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.774052858352661,
      "learning_rate": 6.155828904500683e-07,
      "loss": 0.1149,
      "step": 11786
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.1266961097717285,
      "learning_rate": 6.150247604139081e-07,
      "loss": 0.1087,
      "step": 11787
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.240755558013916,
      "learning_rate": 6.144668669305659e-07,
      "loss": 0.8627,
      "step": 11788
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4099698066711426,
      "learning_rate": 6.139092100301352e-07,
      "loss": 0.1287,
      "step": 11789
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.941706657409668,
      "learning_rate": 6.133517897427011e-07,
      "loss": 0.4343,
      "step": 11790
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.446834087371826,
      "learning_rate": 6.127946060983325e-07,
      "loss": 0.1144,
      "step": 11791
    },
    {
      "epoch": 0.85,
      "grad_norm": 17.75851821899414,
      "learning_rate": 6.12237659127089e-07,
      "loss": 0.6335,
      "step": 11792
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.458265781402588,
      "learning_rate": 6.116809488590153e-07,
      "loss": 0.8975,
      "step": 11793
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.5409727096557617,
      "learning_rate": 6.111244753241441e-07,
      "loss": 0.1461,
      "step": 11794
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.9601340293884277,
      "learning_rate": 6.105682385524942e-07,
      "loss": 0.1708,
      "step": 11795
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4791414737701416,
      "learning_rate": 6.100122385740726e-07,
      "loss": 0.0624,
      "step": 11796
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.141117095947266,
      "learning_rate": 6.094564754188742e-07,
      "loss": 0.3225,
      "step": 11797
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.997449278831482,
      "learning_rate": 6.089009491168796e-07,
      "loss": 0.0883,
      "step": 11798
    },
    {
      "epoch": 0.85,
      "grad_norm": 17.790170669555664,
      "learning_rate": 6.083456596980581e-07,
      "loss": 0.8648,
      "step": 11799
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.60181188583374,
      "learning_rate": 6.077906071923633e-07,
      "loss": 0.9221,
      "step": 11800
    },
    {
      "epoch": 0.85,
      "grad_norm": 17.051076889038086,
      "learning_rate": 6.072357916297417e-07,
      "loss": 0.3211,
      "step": 11801
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.664153099060059,
      "learning_rate": 6.066812130401212e-07,
      "loss": 0.5836,
      "step": 11802
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.74513053894043,
      "learning_rate": 6.061268714534185e-07,
      "loss": 0.2508,
      "step": 11803
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.868582248687744,
      "learning_rate": 6.055727668995404e-07,
      "loss": 0.124,
      "step": 11804
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.03564715385437,
      "learning_rate": 6.050188994083777e-07,
      "loss": 0.1241,
      "step": 11805
    },
    {
      "epoch": 0.85,
      "grad_norm": 21.09227180480957,
      "learning_rate": 6.044652690098096e-07,
      "loss": 0.6686,
      "step": 11806
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.01159954071045,
      "learning_rate": 6.039118757337026e-07,
      "loss": 0.6403,
      "step": 11807
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.266003608703613,
      "learning_rate": 6.033587196099094e-07,
      "loss": 0.3884,
      "step": 11808
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.367232322692871,
      "learning_rate": 6.028058006682702e-07,
      "loss": 0.9707,
      "step": 11809
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.3318605422973633,
      "learning_rate": 6.022531189386144e-07,
      "loss": 0.0985,
      "step": 11810
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0395267009735107,
      "learning_rate": 6.01700674450757e-07,
      "loss": 0.0933,
      "step": 11811
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.7396111488342285,
      "learning_rate": 6.01148467234498e-07,
      "loss": 0.5248,
      "step": 11812
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.9789607524871826,
      "learning_rate": 6.005964973196304e-07,
      "loss": 0.1634,
      "step": 11813
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.561078071594238,
      "learning_rate": 6.000447647359286e-07,
      "loss": 0.8619,
      "step": 11814
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.372844696044922,
      "learning_rate": 5.994932695131567e-07,
      "loss": 0.7121,
      "step": 11815
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2013721466064453,
      "learning_rate": 5.989420116810663e-07,
      "loss": 0.0992,
      "step": 11816
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.851369619369507,
      "learning_rate": 5.983909912693959e-07,
      "loss": 0.0833,
      "step": 11817
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.35181713104248,
      "learning_rate": 5.978402083078705e-07,
      "loss": 0.7518,
      "step": 11818
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.646220207214355,
      "learning_rate": 5.972896628262026e-07,
      "loss": 0.6394,
      "step": 11819
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2350900173187256,
      "learning_rate": 5.967393548540923e-07,
      "loss": 0.0944,
      "step": 11820
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.590348243713379,
      "learning_rate": 5.961892844212258e-07,
      "loss": 0.5837,
      "step": 11821
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.200002193450928,
      "learning_rate": 5.956394515572789e-07,
      "loss": 0.5161,
      "step": 11822
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.330368995666504,
      "learning_rate": 5.950898562919122e-07,
      "loss": 0.1534,
      "step": 11823
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.7611896991729736,
      "learning_rate": 5.945404986547731e-07,
      "loss": 0.0882,
      "step": 11824
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.8509907722473145,
      "learning_rate": 5.939913786754997e-07,
      "loss": 0.6052,
      "step": 11825
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.306453704833984,
      "learning_rate": 5.934424963837138e-07,
      "loss": 0.836,
      "step": 11826
    },
    {
      "epoch": 0.85,
      "grad_norm": 12.911471366882324,
      "learning_rate": 5.928938518090255e-07,
      "loss": 0.6354,
      "step": 11827
    },
    {
      "epoch": 0.85,
      "grad_norm": 22.72386932373047,
      "learning_rate": 5.923454449810306e-07,
      "loss": 0.6654,
      "step": 11828
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8350154161453247,
      "learning_rate": 5.917972759293166e-07,
      "loss": 0.0389,
      "step": 11829
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.9873738288879395,
      "learning_rate": 5.912493446834533e-07,
      "loss": 0.11,
      "step": 11830
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4301838874816895,
      "learning_rate": 5.907016512730002e-07,
      "loss": 0.059,
      "step": 11831
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.3545660972595215,
      "learning_rate": 5.901541957275026e-07,
      "loss": 0.7469,
      "step": 11832
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8106881380081177,
      "learning_rate": 5.896069780764935e-07,
      "loss": 0.0444,
      "step": 11833
    },
    {
      "epoch": 0.85,
      "grad_norm": 14.852202415466309,
      "learning_rate": 5.890599983494938e-07,
      "loss": 0.7551,
      "step": 11834
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.8959779739379883,
      "learning_rate": 5.885132565760116e-07,
      "loss": 0.872,
      "step": 11835
    },
    {
      "epoch": 0.85,
      "grad_norm": 56.34724807739258,
      "learning_rate": 5.879667527855404e-07,
      "loss": 0.8844,
      "step": 11836
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9108057022094727,
      "learning_rate": 5.874204870075617e-07,
      "loss": 0.0535,
      "step": 11837
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.59138011932373,
      "learning_rate": 5.868744592715464e-07,
      "loss": 0.5615,
      "step": 11838
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.5466718673706055,
      "learning_rate": 5.863286696069492e-07,
      "loss": 0.0578,
      "step": 11839
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.303710460662842,
      "learning_rate": 5.857831180432122e-07,
      "loss": 0.8629,
      "step": 11840
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.386667728424072,
      "learning_rate": 5.85237804609769e-07,
      "loss": 0.2871,
      "step": 11841
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.063279151916504,
      "learning_rate": 5.84692729336036e-07,
      "loss": 0.1345,
      "step": 11842
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4286293983459473,
      "learning_rate": 5.841478922514154e-07,
      "loss": 0.0491,
      "step": 11843
    },
    {
      "epoch": 0.85,
      "grad_norm": 23.813461303710938,
      "learning_rate": 5.836032933853019e-07,
      "loss": 1.4734,
      "step": 11844
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6277196407318115,
      "learning_rate": 5.830589327670744e-07,
      "loss": 0.1604,
      "step": 11845
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6018869876861572,
      "learning_rate": 5.825148104260969e-07,
      "loss": 0.1474,
      "step": 11846
    },
    {
      "epoch": 0.85,
      "grad_norm": 16.483078002929688,
      "learning_rate": 5.819709263917256e-07,
      "loss": 1.0316,
      "step": 11847
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.7994823455810547,
      "learning_rate": 5.814272806932997e-07,
      "loss": 0.1576,
      "step": 11848
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.556534767150879,
      "learning_rate": 5.808838733601458e-07,
      "loss": 0.8107,
      "step": 11849
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1419849395751953,
      "learning_rate": 5.803407044215803e-07,
      "loss": 0.0833,
      "step": 11850
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9740151166915894,
      "learning_rate": 5.79797773906905e-07,
      "loss": 0.0421,
      "step": 11851
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.2879068851470947,
      "learning_rate": 5.792550818454068e-07,
      "loss": 0.1375,
      "step": 11852
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.884176731109619,
      "learning_rate": 5.787126282663647e-07,
      "loss": 0.0934,
      "step": 11853
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.099604606628418,
      "learning_rate": 5.781704131990423e-07,
      "loss": 0.5916,
      "step": 11854
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.078505277633667,
      "learning_rate": 5.776284366726858e-07,
      "loss": 0.0562,
      "step": 11855
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.719733238220215,
      "learning_rate": 5.770866987165374e-07,
      "loss": 0.1215,
      "step": 11856
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3686585426330566,
      "learning_rate": 5.765451993598198e-07,
      "loss": 0.0265,
      "step": 11857
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.253666400909424,
      "learning_rate": 5.760039386317434e-07,
      "loss": 0.059,
      "step": 11858
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.687609672546387,
      "learning_rate": 5.7546291656151e-07,
      "loss": 0.7777,
      "step": 11859
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.5111191272735596,
      "learning_rate": 5.74922133178305e-07,
      "loss": 0.5608,
      "step": 11860
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.3669040203094482,
      "learning_rate": 5.743815885112997e-07,
      "loss": 0.1014,
      "step": 11861
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0061988830566406,
      "learning_rate": 5.738412825896566e-07,
      "loss": 0.1968,
      "step": 11862
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.505903959274292,
      "learning_rate": 5.733012154425227e-07,
      "loss": 0.1943,
      "step": 11863
    },
    {
      "epoch": 0.85,
      "grad_norm": 20.638595581054688,
      "learning_rate": 5.727613870990323e-07,
      "loss": 0.6086,
      "step": 11864
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.8975188732147217,
      "learning_rate": 5.72221797588306e-07,
      "loss": 0.1438,
      "step": 11865
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.4744036197662354,
      "learning_rate": 5.716824469394555e-07,
      "loss": 0.2094,
      "step": 11866
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.521826982498169,
      "learning_rate": 5.711433351815732e-07,
      "loss": 0.1169,
      "step": 11867
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.647866249084473,
      "learning_rate": 5.706044623437446e-07,
      "loss": 0.7903,
      "step": 11868
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4555572271347046,
      "learning_rate": 5.700658284550387e-07,
      "loss": 0.0307,
      "step": 11869
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.076788902282715,
      "learning_rate": 5.695274335445139e-07,
      "loss": 0.7643,
      "step": 11870
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.426104545593262,
      "learning_rate": 5.689892776412126e-07,
      "loss": 0.6723,
      "step": 11871
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.265733242034912,
      "learning_rate": 5.684513607741682e-07,
      "loss": 0.1009,
      "step": 11872
    },
    {
      "epoch": 0.85,
      "grad_norm": 19.475833892822266,
      "learning_rate": 5.679136829723986e-07,
      "loss": 0.3632,
      "step": 11873
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.610173225402832,
      "learning_rate": 5.673762442649084e-07,
      "loss": 0.1013,
      "step": 11874
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.813989162445068,
      "learning_rate": 5.668390446806926e-07,
      "loss": 0.7682,
      "step": 11875
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.8201515674591064,
      "learning_rate": 5.663020842487299e-07,
      "loss": 0.6562,
      "step": 11876
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.825835227966309,
      "learning_rate": 5.657653629979865e-07,
      "loss": 0.2305,
      "step": 11877
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.8365395069122314,
      "learning_rate": 5.652288809574191e-07,
      "loss": 0.8365,
      "step": 11878
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.272524118423462,
      "learning_rate": 5.646926381559658e-07,
      "loss": 0.0754,
      "step": 11879
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.3769328594207764,
      "learning_rate": 5.641566346225557e-07,
      "loss": 0.1191,
      "step": 11880
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.818043231964111,
      "learning_rate": 5.636208703861051e-07,
      "loss": 0.7999,
      "step": 11881
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6715779304504395,
      "learning_rate": 5.630853454755164e-07,
      "loss": 0.1411,
      "step": 11882
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.287515640258789,
      "learning_rate": 5.625500599196771e-07,
      "loss": 0.9392,
      "step": 11883
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.5374178886413574,
      "learning_rate": 5.620150137474667e-07,
      "loss": 0.6809,
      "step": 11884
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.34351634979248,
      "learning_rate": 5.614802069877478e-07,
      "loss": 0.754,
      "step": 11885
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.810308456420898,
      "learning_rate": 5.609456396693697e-07,
      "loss": 0.4981,
      "step": 11886
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4241762161254883,
      "learning_rate": 5.604113118211724e-07,
      "loss": 0.1314,
      "step": 11887
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.924628257751465,
      "learning_rate": 5.598772234719807e-07,
      "loss": 0.2129,
      "step": 11888
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.613506317138672,
      "learning_rate": 5.593433746506044e-07,
      "loss": 0.4955,
      "step": 11889
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.009501457214355,
      "learning_rate": 5.588097653858465e-07,
      "loss": 1.0093,
      "step": 11890
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.434105396270752,
      "learning_rate": 5.582763957064902e-07,
      "loss": 0.1855,
      "step": 11891
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.811406135559082,
      "learning_rate": 5.577432656413084e-07,
      "loss": 0.4875,
      "step": 11892
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.155349731445312,
      "learning_rate": 5.572103752190632e-07,
      "loss": 0.6752,
      "step": 11893
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.720653057098389,
      "learning_rate": 5.566777244685017e-07,
      "loss": 0.1767,
      "step": 11894
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.3461527824401855,
      "learning_rate": 5.56145313418357e-07,
      "loss": 0.3653,
      "step": 11895
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.5712175369262695,
      "learning_rate": 5.556131420973531e-07,
      "loss": 0.7937,
      "step": 11896
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4384350776672363,
      "learning_rate": 5.550812105341968e-07,
      "loss": 0.1657,
      "step": 11897
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.6728339195251465,
      "learning_rate": 5.545495187575845e-07,
      "loss": 0.6883,
      "step": 11898
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.2184011936187744,
      "learning_rate": 5.540180667961975e-07,
      "loss": 0.0869,
      "step": 11899
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.475471019744873,
      "learning_rate": 5.534868546787081e-07,
      "loss": 0.8574,
      "step": 11900
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.201474666595459,
      "learning_rate": 5.529558824337716e-07,
      "loss": 0.1757,
      "step": 11901
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.625681161880493,
      "learning_rate": 5.524251500900324e-07,
      "loss": 0.0763,
      "step": 11902
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.191973924636841,
      "learning_rate": 5.518946576761213e-07,
      "loss": 0.2059,
      "step": 11903
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.853616952896118,
      "learning_rate": 5.513644052206557e-07,
      "loss": 0.0991,
      "step": 11904
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.277063846588135,
      "learning_rate": 5.508343927522425e-07,
      "loss": 1.0558,
      "step": 11905
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6434721946716309,
      "learning_rate": 5.503046202994722e-07,
      "loss": 0.0488,
      "step": 11906
    },
    {
      "epoch": 0.85,
      "grad_norm": 22.971982955932617,
      "learning_rate": 5.497750878909253e-07,
      "loss": 0.8104,
      "step": 11907
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.8798489570617676,
      "learning_rate": 5.492457955551661e-07,
      "loss": 0.1441,
      "step": 11908
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.006202220916748,
      "learning_rate": 5.487167433207502e-07,
      "loss": 0.721,
      "step": 11909
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0536715984344482,
      "learning_rate": 5.48187931216217e-07,
      "loss": 0.054,
      "step": 11910
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.344959259033203,
      "learning_rate": 5.47659359270093e-07,
      "loss": 0.5087,
      "step": 11911
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.80643892288208,
      "learning_rate": 5.471310275108943e-07,
      "loss": 0.6015,
      "step": 11912
    },
    {
      "epoch": 0.85,
      "grad_norm": 48.08750915527344,
      "learning_rate": 5.466029359671215e-07,
      "loss": 1.0201,
      "step": 11913
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.873693585395813,
      "learning_rate": 5.460750846672635e-07,
      "loss": 0.0493,
      "step": 11914
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4931893348693848,
      "learning_rate": 5.455474736397959e-07,
      "loss": 0.1483,
      "step": 11915
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.3683063983917236,
      "learning_rate": 5.450201029131808e-07,
      "loss": 0.0949,
      "step": 11916
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.5438241958618164,
      "learning_rate": 5.444929725158671e-07,
      "loss": 0.638,
      "step": 11917
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.7415735721588135,
      "learning_rate": 5.439660824762938e-07,
      "loss": 0.1148,
      "step": 11918
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.86407732963562,
      "learning_rate": 5.434394328228831e-07,
      "loss": 0.1302,
      "step": 11919
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2371392250061035,
      "learning_rate": 5.429130235840451e-07,
      "loss": 0.0785,
      "step": 11920
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.7563207149505615,
      "learning_rate": 5.423868547881795e-07,
      "loss": 0.0971,
      "step": 11921
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.077217102050781,
      "learning_rate": 5.418609264636698e-07,
      "loss": 0.6701,
      "step": 11922
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9856786727905273,
      "learning_rate": 5.413352386388876e-07,
      "loss": 0.0614,
      "step": 11923
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.025938034057617,
      "learning_rate": 5.408097913421934e-07,
      "loss": 0.7676,
      "step": 11924
    },
    {
      "epoch": 0.86,
      "grad_norm": 16.15333366394043,
      "learning_rate": 5.402845846019317e-07,
      "loss": 0.5198,
      "step": 11925
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3281147480010986,
      "learning_rate": 5.397596184464354e-07,
      "loss": 0.0889,
      "step": 11926
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.739344596862793,
      "learning_rate": 5.392348929040254e-07,
      "loss": 0.3119,
      "step": 11927
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.69482946395874,
      "learning_rate": 5.387104080030076e-07,
      "loss": 0.4476,
      "step": 11928
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.999353408813477,
      "learning_rate": 5.381861637716757e-07,
      "loss": 0.6471,
      "step": 11929
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.118889808654785,
      "learning_rate": 5.37662160238312e-07,
      "loss": 0.2972,
      "step": 11930
    },
    {
      "epoch": 0.86,
      "grad_norm": 18.56865119934082,
      "learning_rate": 5.371383974311839e-07,
      "loss": 0.4774,
      "step": 11931
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.120372772216797,
      "learning_rate": 5.36614875378546e-07,
      "loss": 0.7923,
      "step": 11932
    },
    {
      "epoch": 0.86,
      "grad_norm": 15.262406349182129,
      "learning_rate": 5.360915941086414e-07,
      "loss": 0.677,
      "step": 11933
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.2723827362060547,
      "learning_rate": 5.35568553649698e-07,
      "loss": 0.0787,
      "step": 11934
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.309427738189697,
      "learning_rate": 5.350457540299331e-07,
      "loss": 0.9092,
      "step": 11935
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.724877119064331,
      "learning_rate": 5.345231952775476e-07,
      "loss": 0.6781,
      "step": 11936
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0376830101013184,
      "learning_rate": 5.34000877420734e-07,
      "loss": 0.0832,
      "step": 11937
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3004150390625,
      "learning_rate": 5.334788004876684e-07,
      "loss": 0.0993,
      "step": 11938
    },
    {
      "epoch": 0.86,
      "grad_norm": 17.43992042541504,
      "learning_rate": 5.329569645065147e-07,
      "loss": 0.5481,
      "step": 11939
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.071606159210205,
      "learning_rate": 5.324353695054241e-07,
      "loss": 0.6548,
      "step": 11940
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.548562049865723,
      "learning_rate": 5.319140155125341e-07,
      "loss": 0.7472,
      "step": 11941
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.147027969360352,
      "learning_rate": 5.313929025559705e-07,
      "loss": 0.3595,
      "step": 11942
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.37604284286499,
      "learning_rate": 5.308720306638459e-07,
      "loss": 0.9049,
      "step": 11943
    },
    {
      "epoch": 0.86,
      "grad_norm": 12.08402156829834,
      "learning_rate": 5.303513998642584e-07,
      "loss": 0.5797,
      "step": 11944
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.5126953125,
      "learning_rate": 5.298310101852938e-07,
      "loss": 0.4024,
      "step": 11945
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.5514659881591797,
      "learning_rate": 5.293108616550263e-07,
      "loss": 0.0992,
      "step": 11946
    },
    {
      "epoch": 0.86,
      "grad_norm": 14.727596282958984,
      "learning_rate": 5.287909543015152e-07,
      "loss": 0.9366,
      "step": 11947
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.683134078979492,
      "learning_rate": 5.282712881528068e-07,
      "loss": 1.0091,
      "step": 11948
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.679312705993652,
      "learning_rate": 5.277518632369372e-07,
      "loss": 0.8624,
      "step": 11949
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9659935235977173,
      "learning_rate": 5.272326795819266e-07,
      "loss": 0.0723,
      "step": 11950
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.22739315032959,
      "learning_rate": 5.26713737215781e-07,
      "loss": 0.7274,
      "step": 11951
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.819345951080322,
      "learning_rate": 5.26195036166498e-07,
      "loss": 0.4829,
      "step": 11952
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.6328511238098145,
      "learning_rate": 5.256765764620586e-07,
      "loss": 0.1447,
      "step": 11953
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.967865943908691,
      "learning_rate": 5.251583581304304e-07,
      "loss": 0.4545,
      "step": 11954
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.865483283996582,
      "learning_rate": 5.24640381199572e-07,
      "loss": 0.6946,
      "step": 11955
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9410171508789062,
      "learning_rate": 5.241226456974247e-07,
      "loss": 0.1878,
      "step": 11956
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.233593940734863,
      "learning_rate": 5.236051516519175e-07,
      "loss": 0.7644,
      "step": 11957
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9101550579071045,
      "learning_rate": 5.230878990909688e-07,
      "loss": 0.1406,
      "step": 11958
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.755330801010132,
      "learning_rate": 5.225708880424823e-07,
      "loss": 0.1596,
      "step": 11959
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.322198867797852,
      "learning_rate": 5.220541185343475e-07,
      "loss": 0.691,
      "step": 11960
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.512996673583984,
      "learning_rate": 5.215375905944442e-07,
      "loss": 0.8064,
      "step": 11961
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.855961561203003,
      "learning_rate": 5.210213042506368e-07,
      "loss": 0.1806,
      "step": 11962
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.796696424484253,
      "learning_rate": 5.205052595307741e-07,
      "loss": 0.7849,
      "step": 11963
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.4402341842651367,
      "learning_rate": 5.199894564626984e-07,
      "loss": 0.7246,
      "step": 11964
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.2994120121002197,
      "learning_rate": 5.194738950742328e-07,
      "loss": 0.1288,
      "step": 11965
    },
    {
      "epoch": 0.86,
      "grad_norm": 23.215797424316406,
      "learning_rate": 5.189585753931908e-07,
      "loss": 0.7933,
      "step": 11966
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.8436439037323,
      "learning_rate": 5.184434974473723e-07,
      "loss": 0.1918,
      "step": 11967
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.174843192100525,
      "learning_rate": 5.17928661264564e-07,
      "loss": 0.0385,
      "step": 11968
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6501293182373047,
      "learning_rate": 5.174140668725374e-07,
      "loss": 0.1092,
      "step": 11969
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.355869770050049,
      "learning_rate": 5.168997142990556e-07,
      "loss": 0.4096,
      "step": 11970
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.861992835998535,
      "learning_rate": 5.163856035718645e-07,
      "loss": 0.1758,
      "step": 11971
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2274811267852783,
      "learning_rate": 5.158717347186993e-07,
      "loss": 0.1002,
      "step": 11972
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9826247692108154,
      "learning_rate": 5.153581077672792e-07,
      "loss": 0.2177,
      "step": 11973
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.923084020614624,
      "learning_rate": 5.148447227453157e-07,
      "loss": 0.6305,
      "step": 11974
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.590062141418457,
      "learning_rate": 5.14331579680501e-07,
      "loss": 0.3102,
      "step": 11975
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2513678073883057,
      "learning_rate": 5.138186786005184e-07,
      "loss": 0.1247,
      "step": 11976
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.6030771732330322,
      "learning_rate": 5.133060195330375e-07,
      "loss": 0.1526,
      "step": 11977
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.732184648513794,
      "learning_rate": 5.12793602505714e-07,
      "loss": 0.1642,
      "step": 11978
    },
    {
      "epoch": 0.86,
      "grad_norm": 14.756502151489258,
      "learning_rate": 5.12281427546189e-07,
      "loss": 0.542,
      "step": 11979
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9596920013427734,
      "learning_rate": 5.117694946820956e-07,
      "loss": 0.1894,
      "step": 11980
    },
    {
      "epoch": 0.86,
      "grad_norm": 22.392000198364258,
      "learning_rate": 5.112578039410487e-07,
      "loss": 1.4088,
      "step": 11981
    },
    {
      "epoch": 0.86,
      "grad_norm": 27.85207176208496,
      "learning_rate": 5.10746355350652e-07,
      "loss": 0.5049,
      "step": 11982
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.508944511413574,
      "learning_rate": 5.102351489384972e-07,
      "loss": 0.2498,
      "step": 11983
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.764993906021118,
      "learning_rate": 5.097241847321615e-07,
      "loss": 0.1479,
      "step": 11984
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1214184761047363,
      "learning_rate": 5.092134627592088e-07,
      "loss": 0.0991,
      "step": 11985
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.2110435962677,
      "learning_rate": 5.087029830471918e-07,
      "loss": 0.227,
      "step": 11986
    },
    {
      "epoch": 0.86,
      "grad_norm": 20.211822509765625,
      "learning_rate": 5.081927456236496e-07,
      "loss": 0.3496,
      "step": 11987
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.604954957962036,
      "learning_rate": 5.076827505161042e-07,
      "loss": 0.1177,
      "step": 11988
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9761204719543457,
      "learning_rate": 5.071729977520712e-07,
      "loss": 0.8805,
      "step": 11989
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.376537322998047,
      "learning_rate": 5.066634873590486e-07,
      "loss": 0.115,
      "step": 11990
    },
    {
      "epoch": 0.86,
      "grad_norm": 32.155006408691406,
      "learning_rate": 5.061542193645219e-07,
      "loss": 1.0819,
      "step": 11991
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.504487037658691,
      "learning_rate": 5.056451937959656e-07,
      "loss": 0.909,
      "step": 11992
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.04129695892334,
      "learning_rate": 5.051364106808392e-07,
      "loss": 0.4932,
      "step": 11993
    },
    {
      "epoch": 0.86,
      "grad_norm": 24.15570640563965,
      "learning_rate": 5.046278700465884e-07,
      "loss": 0.9337,
      "step": 11994
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.227585792541504,
      "learning_rate": 5.041195719206493e-07,
      "loss": 0.5795,
      "step": 11995
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.096043586730957,
      "learning_rate": 5.036115163304411e-07,
      "loss": 0.5715,
      "step": 11996
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.358750581741333,
      "learning_rate": 5.031037033033709e-07,
      "loss": 0.0516,
      "step": 11997
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.4733476638793945,
      "learning_rate": 5.025961328668355e-07,
      "loss": 0.6228,
      "step": 11998
    },
    {
      "epoch": 0.86,
      "grad_norm": 18.214805603027344,
      "learning_rate": 5.020888050482158e-07,
      "loss": 1.3197,
      "step": 11999
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.762453317642212,
      "learning_rate": 5.01581719874878e-07,
      "loss": 0.1169,
      "step": 12000
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.540266990661621,
      "learning_rate": 5.010748773741798e-07,
      "loss": 0.4524,
      "step": 12001
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.032468557357788,
      "learning_rate": 5.005682775734627e-07,
      "loss": 0.1411,
      "step": 12002
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.636413335800171,
      "learning_rate": 5.000619205000545e-07,
      "loss": 0.0543,
      "step": 12003
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.941506862640381,
      "learning_rate": 4.995558061812734e-07,
      "loss": 0.1518,
      "step": 12004
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.1531405448913574,
      "learning_rate": 4.990499346444216e-07,
      "loss": 0.3057,
      "step": 12005
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.009311556816101,
      "learning_rate": 4.985443059167877e-07,
      "loss": 0.0233,
      "step": 12006
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.457991600036621,
      "learning_rate": 4.980389200256508e-07,
      "loss": 0.6167,
      "step": 12007
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6159005165100098,
      "learning_rate": 4.975337769982736e-07,
      "loss": 0.1189,
      "step": 12008
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.4909827709198,
      "learning_rate": 4.970288768619058e-07,
      "loss": 0.0805,
      "step": 12009
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.724943161010742,
      "learning_rate": 4.965242196437848e-07,
      "loss": 0.1358,
      "step": 12010
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.896045446395874,
      "learning_rate": 4.960198053711374e-07,
      "loss": 0.7624,
      "step": 12011
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2298614978790283,
      "learning_rate": 4.955156340711709e-07,
      "loss": 0.0462,
      "step": 12012
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.1268386840820312,
      "learning_rate": 4.950117057710863e-07,
      "loss": 0.2106,
      "step": 12013
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.402795791625977,
      "learning_rate": 4.945080204980685e-07,
      "loss": 0.2831,
      "step": 12014
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.406520366668701,
      "learning_rate": 4.940045782792885e-07,
      "loss": 0.1206,
      "step": 12015
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.806143283843994,
      "learning_rate": 4.935013791419041e-07,
      "loss": 0.8516,
      "step": 12016
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.7907347679138184,
      "learning_rate": 4.929984231130635e-07,
      "loss": 0.6622,
      "step": 12017
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.996631622314453,
      "learning_rate": 4.92495710219898e-07,
      "loss": 0.6374,
      "step": 12018
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.284885883331299,
      "learning_rate": 4.919932404895267e-07,
      "loss": 0.2928,
      "step": 12019
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.47796893119812,
      "learning_rate": 4.914910139490564e-07,
      "loss": 0.1258,
      "step": 12020
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.8846404552459717,
      "learning_rate": 4.90989030625581e-07,
      "loss": 0.7563,
      "step": 12021
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.886512279510498,
      "learning_rate": 4.904872905461782e-07,
      "loss": 0.0777,
      "step": 12022
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9733314514160156,
      "learning_rate": 4.899857937379193e-07,
      "loss": 0.0853,
      "step": 12023
    },
    {
      "epoch": 0.86,
      "grad_norm": 11.897187232971191,
      "learning_rate": 4.894845402278542e-07,
      "loss": 0.4902,
      "step": 12024
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5680103302001953,
      "learning_rate": 4.889835300430241e-07,
      "loss": 0.0434,
      "step": 12025
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.56642484664917,
      "learning_rate": 4.884827632104577e-07,
      "loss": 0.8422,
      "step": 12026
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.797703981399536,
      "learning_rate": 4.879822397571698e-07,
      "loss": 0.1604,
      "step": 12027
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.571220636367798,
      "learning_rate": 4.874819597101605e-07,
      "loss": 0.0839,
      "step": 12028
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.82088851928711,
      "learning_rate": 4.869819230964191e-07,
      "loss": 0.6287,
      "step": 12029
    },
    {
      "epoch": 0.86,
      "grad_norm": 13.055742263793945,
      "learning_rate": 4.864821299429207e-07,
      "loss": 0.5852,
      "step": 12030
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.003962993621826,
      "learning_rate": 4.859825802766255e-07,
      "loss": 0.7351,
      "step": 12031
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.437793016433716,
      "learning_rate": 4.854832741244841e-07,
      "loss": 0.0926,
      "step": 12032
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6052486896514893,
      "learning_rate": 4.849842115134323e-07,
      "loss": 0.0861,
      "step": 12033
    },
    {
      "epoch": 0.86,
      "grad_norm": 13.113801956176758,
      "learning_rate": 4.844853924703913e-07,
      "loss": 1.0392,
      "step": 12034
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.811009168624878,
      "learning_rate": 4.839868170222728e-07,
      "loss": 0.1124,
      "step": 12035
    },
    {
      "epoch": 0.86,
      "grad_norm": 15.379765510559082,
      "learning_rate": 4.834884851959703e-07,
      "loss": 0.6881,
      "step": 12036
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.361388206481934,
      "learning_rate": 4.829903970183675e-07,
      "loss": 0.6918,
      "step": 12037
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.358542442321777,
      "learning_rate": 4.82492552516336e-07,
      "loss": 0.6473,
      "step": 12038
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.946211814880371,
      "learning_rate": 4.81994951716731e-07,
      "loss": 0.5565,
      "step": 12039
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.802157163619995,
      "learning_rate": 4.814975946463968e-07,
      "loss": 0.0473,
      "step": 12040
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.592254638671875,
      "learning_rate": 4.81000481332164e-07,
      "loss": 0.3373,
      "step": 12041
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.962998867034912,
      "learning_rate": 4.805036118008499e-07,
      "loss": 0.132,
      "step": 12042
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.801036834716797,
      "learning_rate": 4.800069860792594e-07,
      "loss": 0.5795,
      "step": 12043
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.905803918838501,
      "learning_rate": 4.795106041941816e-07,
      "loss": 0.1738,
      "step": 12044
    },
    {
      "epoch": 0.86,
      "grad_norm": 59.003204345703125,
      "learning_rate": 4.790144661723967e-07,
      "loss": 0.7022,
      "step": 12045
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.943629264831543,
      "learning_rate": 4.785185720406683e-07,
      "loss": 0.4675,
      "step": 12046
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.531569004058838,
      "learning_rate": 4.780229218257481e-07,
      "loss": 0.0883,
      "step": 12047
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.8278825283050537,
      "learning_rate": 4.775275155543752e-07,
      "loss": 0.1442,
      "step": 12048
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1528897285461426,
      "learning_rate": 4.770323532532728e-07,
      "loss": 0.0956,
      "step": 12049
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.5375068187713623,
      "learning_rate": 4.765374349491558e-07,
      "loss": 0.1792,
      "step": 12050
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.72684383392334,
      "learning_rate": 4.760427606687218e-07,
      "loss": 0.9058,
      "step": 12051
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.2337491512298584,
      "learning_rate": 4.755483304386571e-07,
      "loss": 0.0739,
      "step": 12052
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.441525459289551,
      "learning_rate": 4.750541442856327e-07,
      "loss": 0.1124,
      "step": 12053
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.922187566757202,
      "learning_rate": 4.745602022363105e-07,
      "loss": 0.1397,
      "step": 12054
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3739004135131836,
      "learning_rate": 4.740665043173359e-07,
      "loss": 0.0758,
      "step": 12055
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.313695192337036,
      "learning_rate": 4.735730505553404e-07,
      "loss": 0.1296,
      "step": 12056
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.88252854347229,
      "learning_rate": 4.7307984097694657e-07,
      "loss": 0.1022,
      "step": 12057
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.975872993469238,
      "learning_rate": 4.7258687560875984e-07,
      "loss": 0.4247,
      "step": 12058
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.0317044258117676,
      "learning_rate": 4.720941544773744e-07,
      "loss": 0.1481,
      "step": 12059
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.844895601272583,
      "learning_rate": 4.716016776093696e-07,
      "loss": 0.714,
      "step": 12060
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.603681802749634,
      "learning_rate": 4.7110944503131363e-07,
      "loss": 0.6767,
      "step": 12061
    },
    {
      "epoch": 0.86,
      "grad_norm": 11.796320915222168,
      "learning_rate": 4.7061745676975966e-07,
      "loss": 0.7794,
      "step": 12062
    },
    {
      "epoch": 0.87,
      "grad_norm": 24.733884811401367,
      "learning_rate": 4.7012571285124985e-07,
      "loss": 1.147,
      "step": 12063
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.162052154541016,
      "learning_rate": 4.6963421330231187e-07,
      "loss": 0.596,
      "step": 12064
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.61405611038208,
      "learning_rate": 4.6914295814945834e-07,
      "loss": 0.3261,
      "step": 12065
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4608964920043945,
      "learning_rate": 4.686519474191925e-07,
      "loss": 0.1191,
      "step": 12066
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.210803031921387,
      "learning_rate": 4.681611811380027e-07,
      "loss": 0.6941,
      "step": 12067
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4824612140655518,
      "learning_rate": 4.676706593323621e-07,
      "loss": 0.0813,
      "step": 12068
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.45697021484375,
      "learning_rate": 4.6718038202873463e-07,
      "loss": 0.3833,
      "step": 12069
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.8944525718688965,
      "learning_rate": 4.6669034925356794e-07,
      "loss": 0.1682,
      "step": 12070
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2310879230499268,
      "learning_rate": 4.66200561033297e-07,
      "loss": 0.038,
      "step": 12071
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.016892671585083,
      "learning_rate": 4.6571101739434465e-07,
      "loss": 0.0596,
      "step": 12072
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.084235668182373,
      "learning_rate": 4.652217183631197e-07,
      "loss": 0.1431,
      "step": 12073
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.540313959121704,
      "learning_rate": 4.647326639660171e-07,
      "loss": 0.1335,
      "step": 12074
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.346052169799805,
      "learning_rate": 4.642438542294214e-07,
      "loss": 0.5816,
      "step": 12075
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.792459011077881,
      "learning_rate": 4.6375528917970093e-07,
      "loss": 0.7718,
      "step": 12076
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9565927982330322,
      "learning_rate": 4.632669688432112e-07,
      "loss": 0.0944,
      "step": 12077
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.2007393836975098,
      "learning_rate": 4.6277889324629677e-07,
      "loss": 0.1879,
      "step": 12078
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.316704750061035,
      "learning_rate": 4.622910624152865e-07,
      "loss": 0.7152,
      "step": 12079
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.317644119262695,
      "learning_rate": 4.618034763764978e-07,
      "loss": 0.7744,
      "step": 12080
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.241267204284668,
      "learning_rate": 4.6131613515623233e-07,
      "loss": 0.9397,
      "step": 12081
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.5971271991729736,
      "learning_rate": 4.6082903878078235e-07,
      "loss": 0.1122,
      "step": 12082
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.107327699661255,
      "learning_rate": 4.603421872764241e-07,
      "loss": 0.0625,
      "step": 12083
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.280817031860352,
      "learning_rate": 4.5985558066942104e-07,
      "loss": 0.6199,
      "step": 12084
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.186267614364624,
      "learning_rate": 4.5936921898602437e-07,
      "loss": 0.0776,
      "step": 12085
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.5044491291046143,
      "learning_rate": 4.5888310225247037e-07,
      "loss": 0.1034,
      "step": 12086
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.477710723876953,
      "learning_rate": 4.5839723049498465e-07,
      "loss": 0.541,
      "step": 12087
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.9592785835266113,
      "learning_rate": 4.579116037397774e-07,
      "loss": 0.1763,
      "step": 12088
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.69472885131836,
      "learning_rate": 4.57426222013046e-07,
      "loss": 0.6227,
      "step": 12089
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.7236921787261963,
      "learning_rate": 4.5694108534097505e-07,
      "loss": 0.0746,
      "step": 12090
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.499610424041748,
      "learning_rate": 4.564561937497369e-07,
      "loss": 0.4314,
      "step": 12091
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.272929191589355,
      "learning_rate": 4.5597154726548853e-07,
      "loss": 0.7682,
      "step": 12092
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.023736000061035,
      "learning_rate": 4.554871459143745e-07,
      "loss": 0.3974,
      "step": 12093
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.487123012542725,
      "learning_rate": 4.5500298972252776e-07,
      "loss": 0.852,
      "step": 12094
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.9956271648406982,
      "learning_rate": 4.545190787160675e-07,
      "loss": 0.1414,
      "step": 12095
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1894609928131104,
      "learning_rate": 4.54035412921095e-07,
      "loss": 0.0946,
      "step": 12096
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.499197483062744,
      "learning_rate": 4.535519923637055e-07,
      "loss": 0.2571,
      "step": 12097
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.385162353515625,
      "learning_rate": 4.5306881706997653e-07,
      "loss": 0.1485,
      "step": 12098
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9486072063446045,
      "learning_rate": 4.525858870659733e-07,
      "loss": 0.1001,
      "step": 12099
    },
    {
      "epoch": 0.87,
      "grad_norm": 15.813056945800781,
      "learning_rate": 4.5210320237774943e-07,
      "loss": 0.935,
      "step": 12100
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.393566131591797,
      "learning_rate": 4.5162076303134295e-07,
      "loss": 0.3651,
      "step": 12101
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.116243362426758,
      "learning_rate": 4.511385690527786e-07,
      "loss": 0.8657,
      "step": 12102
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.132368087768555,
      "learning_rate": 4.5065662046807126e-07,
      "loss": 0.7797,
      "step": 12103
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4917428493499756,
      "learning_rate": 4.501749173032194e-07,
      "loss": 0.1276,
      "step": 12104
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.441228866577148,
      "learning_rate": 4.496934595842073e-07,
      "loss": 0.6203,
      "step": 12105
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.741786479949951,
      "learning_rate": 4.492122473370103e-07,
      "loss": 0.063,
      "step": 12106
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1684207916259766,
      "learning_rate": 4.4873128058758766e-07,
      "loss": 0.0574,
      "step": 12107
    },
    {
      "epoch": 0.87,
      "grad_norm": 21.114768981933594,
      "learning_rate": 4.482505593618836e-07,
      "loss": 0.3963,
      "step": 12108
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.047403573989868,
      "learning_rate": 4.4777008368583286e-07,
      "loss": 0.1987,
      "step": 12109
    },
    {
      "epoch": 0.87,
      "grad_norm": 16.49457359313965,
      "learning_rate": 4.472898535853554e-07,
      "loss": 0.7226,
      "step": 12110
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2646021842956543,
      "learning_rate": 4.468098690863565e-07,
      "loss": 0.079,
      "step": 12111
    },
    {
      "epoch": 0.87,
      "grad_norm": 24.32021141052246,
      "learning_rate": 4.4633013021473105e-07,
      "loss": 0.9551,
      "step": 12112
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.919441223144531,
      "learning_rate": 4.458506369963583e-07,
      "loss": 0.7558,
      "step": 12113
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3791182041168213,
      "learning_rate": 4.453713894571049e-07,
      "loss": 0.1333,
      "step": 12114
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.115739345550537,
      "learning_rate": 4.448923876228256e-07,
      "loss": 0.1876,
      "step": 12115
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.958569526672363,
      "learning_rate": 4.4441363151935977e-07,
      "loss": 0.7732,
      "step": 12116
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.507726669311523,
      "learning_rate": 4.4393512117253455e-07,
      "loss": 0.4072,
      "step": 12117
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.34117317199707,
      "learning_rate": 4.434568566081632e-07,
      "loss": 0.5013,
      "step": 12118
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.222026824951172,
      "learning_rate": 4.429788378520489e-07,
      "loss": 0.0517,
      "step": 12119
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.8300399780273438,
      "learning_rate": 4.4250106492997547e-07,
      "loss": 0.0995,
      "step": 12120
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9152421951293945,
      "learning_rate": 4.4202353786771903e-07,
      "loss": 0.0593,
      "step": 12121
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.938103675842285,
      "learning_rate": 4.4154625669104003e-07,
      "loss": 0.7069,
      "step": 12122
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.708470344543457,
      "learning_rate": 4.4106922142568567e-07,
      "loss": 0.7011,
      "step": 12123
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.090006828308105,
      "learning_rate": 4.4059243209738924e-07,
      "loss": 0.5748,
      "step": 12124
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.319552421569824,
      "learning_rate": 4.4011588873187407e-07,
      "loss": 0.1631,
      "step": 12125
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.4183285236358643,
      "learning_rate": 4.396395913548462e-07,
      "loss": 0.1223,
      "step": 12126
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.00996732711792,
      "learning_rate": 4.3916353999199967e-07,
      "loss": 0.1807,
      "step": 12127
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.608020544052124,
      "learning_rate": 4.386877346690177e-07,
      "loss": 0.743,
      "step": 12128
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.890880823135376,
      "learning_rate": 4.38212175411567e-07,
      "loss": 0.1787,
      "step": 12129
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.811629295349121,
      "learning_rate": 4.37736862245301e-07,
      "loss": 0.0732,
      "step": 12130
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0972771644592285,
      "learning_rate": 4.3726179519586354e-07,
      "loss": 0.0721,
      "step": 12131
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.238476753234863,
      "learning_rate": 4.367869742888808e-07,
      "loss": 0.5514,
      "step": 12132
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.269979953765869,
      "learning_rate": 4.363123995499674e-07,
      "loss": 0.8498,
      "step": 12133
    },
    {
      "epoch": 0.87,
      "grad_norm": 19.380931854248047,
      "learning_rate": 4.358380710047261e-07,
      "loss": 0.5337,
      "step": 12134
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.920732021331787,
      "learning_rate": 4.353639886787447e-07,
      "loss": 0.1061,
      "step": 12135
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.10291862487793,
      "learning_rate": 4.348901525975974e-07,
      "loss": 0.681,
      "step": 12136
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.5667033195495605,
      "learning_rate": 4.344165627868474e-07,
      "loss": 0.2756,
      "step": 12137
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.161208152770996,
      "learning_rate": 4.3394321927204163e-07,
      "loss": 0.1943,
      "step": 12138
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.354727745056152,
      "learning_rate": 4.334701220787152e-07,
      "loss": 0.3856,
      "step": 12139
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.885614037513733,
      "learning_rate": 4.3299727123239156e-07,
      "loss": 0.0703,
      "step": 12140
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.6262259483337402,
      "learning_rate": 4.325246667585775e-07,
      "loss": 0.0796,
      "step": 12141
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.102738857269287,
      "learning_rate": 4.320523086827683e-07,
      "loss": 0.0683,
      "step": 12142
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8207929134368896,
      "learning_rate": 4.315801970304484e-07,
      "loss": 0.051,
      "step": 12143
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7134101390838623,
      "learning_rate": 4.3110833182708313e-07,
      "loss": 0.0509,
      "step": 12144
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.4306640625,
      "learning_rate": 4.306367130981287e-07,
      "loss": 0.5326,
      "step": 12145
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.3295440673828125,
      "learning_rate": 4.3016534086902815e-07,
      "loss": 0.8053,
      "step": 12146
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1756224632263184,
      "learning_rate": 4.2969421516520946e-07,
      "loss": 0.0921,
      "step": 12147
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.7059388160705566,
      "learning_rate": 4.2922333601208785e-07,
      "loss": 0.0995,
      "step": 12148
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.318211555480957,
      "learning_rate": 4.287527034350669e-07,
      "loss": 0.0839,
      "step": 12149
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.175230503082275,
      "learning_rate": 4.2828231745953407e-07,
      "loss": 0.9398,
      "step": 12150
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.641941547393799,
      "learning_rate": 4.2781217811086574e-07,
      "loss": 0.2131,
      "step": 12151
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.287385940551758,
      "learning_rate": 4.2734228541442267e-07,
      "loss": 0.1336,
      "step": 12152
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.652069091796875,
      "learning_rate": 4.268726393955558e-07,
      "loss": 0.379,
      "step": 12153
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.391751289367676,
      "learning_rate": 4.2640324007959977e-07,
      "loss": 0.5202,
      "step": 12154
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.782191276550293,
      "learning_rate": 4.259340874918771e-07,
      "loss": 0.3039,
      "step": 12155
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.525700569152832,
      "learning_rate": 4.2546518165769657e-07,
      "loss": 0.744,
      "step": 12156
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.22055435180664,
      "learning_rate": 4.249965226023528e-07,
      "loss": 0.5195,
      "step": 12157
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.358057022094727,
      "learning_rate": 4.245281103511306e-07,
      "loss": 0.5312,
      "step": 12158
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.4047138690948486,
      "learning_rate": 4.24059944929297e-07,
      "loss": 0.1137,
      "step": 12159
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.5126445293426514,
      "learning_rate": 4.2359202636210906e-07,
      "loss": 0.1434,
      "step": 12160
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.872873544692993,
      "learning_rate": 4.231243546748076e-07,
      "loss": 0.1306,
      "step": 12161
    },
    {
      "epoch": 0.87,
      "grad_norm": 16.57489776611328,
      "learning_rate": 4.226569298926242e-07,
      "loss": 1.2277,
      "step": 12162
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4678187370300293,
      "learning_rate": 4.221897520407725e-07,
      "loss": 0.1266,
      "step": 12163
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.738039255142212,
      "learning_rate": 4.217228211444552e-07,
      "loss": 0.1458,
      "step": 12164
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1580190658569336,
      "learning_rate": 4.212561372288632e-07,
      "loss": 0.1064,
      "step": 12165
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.826400756835938,
      "learning_rate": 4.207897003191702e-07,
      "loss": 0.6629,
      "step": 12166
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4059369564056396,
      "learning_rate": 4.2032351044054064e-07,
      "loss": 0.1064,
      "step": 12167
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.7401022911071777,
      "learning_rate": 4.1985756761812215e-07,
      "loss": 0.1223,
      "step": 12168
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.8154640197753906,
      "learning_rate": 4.193918718770512e-07,
      "loss": 0.1661,
      "step": 12169
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.715529680252075,
      "learning_rate": 4.1892642324244894e-07,
      "loss": 0.8035,
      "step": 12170
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.3007733821868896,
      "learning_rate": 4.184612217394274e-07,
      "loss": 0.2011,
      "step": 12171
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.118323802947998,
      "learning_rate": 4.1799626739308043e-07,
      "loss": 0.3492,
      "step": 12172
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.730217933654785,
      "learning_rate": 4.1753156022849016e-07,
      "loss": 0.4971,
      "step": 12173
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.283634185791016,
      "learning_rate": 4.170671002707277e-07,
      "loss": 0.7258,
      "step": 12174
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.264282703399658,
      "learning_rate": 4.1660288754484743e-07,
      "loss": 0.1113,
      "step": 12175
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.452052116394043,
      "learning_rate": 4.161389220758916e-07,
      "loss": 0.0672,
      "step": 12176
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.081170558929443,
      "learning_rate": 4.1567520388889114e-07,
      "loss": 0.8119,
      "step": 12177
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4993741512298584,
      "learning_rate": 4.1521173300886066e-07,
      "loss": 0.0506,
      "step": 12178
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.961817741394043,
      "learning_rate": 4.147485094608028e-07,
      "loss": 0.2419,
      "step": 12179
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.515952110290527,
      "learning_rate": 4.1428553326970713e-07,
      "loss": 0.7811,
      "step": 12180
    },
    {
      "epoch": 0.87,
      "grad_norm": 23.180801391601562,
      "learning_rate": 4.138228044605486e-07,
      "loss": 0.8425,
      "step": 12181
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3656294345855713,
      "learning_rate": 4.133603230582894e-07,
      "loss": 0.1058,
      "step": 12182
    },
    {
      "epoch": 0.87,
      "grad_norm": 26.733848571777344,
      "learning_rate": 4.128980890878803e-07,
      "loss": 0.6604,
      "step": 12183
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4443187713623047,
      "learning_rate": 4.1243610257425626e-07,
      "loss": 0.1024,
      "step": 12184
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1510889530181885,
      "learning_rate": 4.119743635423384e-07,
      "loss": 0.1465,
      "step": 12185
    },
    {
      "epoch": 0.87,
      "grad_norm": 20.66836929321289,
      "learning_rate": 4.115128720170386e-07,
      "loss": 1.0424,
      "step": 12186
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.824677467346191,
      "learning_rate": 4.1105162802325015e-07,
      "loss": 0.5005,
      "step": 12187
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.254432678222656,
      "learning_rate": 4.1059063158585657e-07,
      "loss": 1.0247,
      "step": 12188
    },
    {
      "epoch": 0.87,
      "grad_norm": 15.262482643127441,
      "learning_rate": 4.1012988272972564e-07,
      "loss": 0.5303,
      "step": 12189
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.757134437561035,
      "learning_rate": 4.096693814797154e-07,
      "loss": 0.0617,
      "step": 12190
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.147623538970947,
      "learning_rate": 4.092091278606658e-07,
      "loss": 0.8278,
      "step": 12191
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.4299910068511963,
      "learning_rate": 4.087491218974071e-07,
      "loss": 0.1854,
      "step": 12192
    },
    {
      "epoch": 0.87,
      "grad_norm": 15.03416919708252,
      "learning_rate": 4.0828936361475447e-07,
      "loss": 0.5646,
      "step": 12193
    },
    {
      "epoch": 0.87,
      "grad_norm": 18.30417251586914,
      "learning_rate": 4.078298530375091e-07,
      "loss": 0.7048,
      "step": 12194
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.334014415740967,
      "learning_rate": 4.0737059019046176e-07,
      "loss": 0.8419,
      "step": 12195
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.514425277709961,
      "learning_rate": 4.069115750983871e-07,
      "loss": 0.7775,
      "step": 12196
    },
    {
      "epoch": 0.87,
      "grad_norm": 18.819372177124023,
      "learning_rate": 4.064528077860469e-07,
      "loss": 0.7551,
      "step": 12197
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.735645294189453,
      "learning_rate": 4.0599428827818975e-07,
      "loss": 0.7067,
      "step": 12198
    },
    {
      "epoch": 0.87,
      "grad_norm": 17.27323341369629,
      "learning_rate": 4.05536016599552e-07,
      "loss": 0.3808,
      "step": 12199
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.229499816894531,
      "learning_rate": 4.0507799277485546e-07,
      "loss": 0.5949,
      "step": 12200
    },
    {
      "epoch": 0.87,
      "grad_norm": 20.590208053588867,
      "learning_rate": 4.046202168288077e-07,
      "loss": 1.0784,
      "step": 12201
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5941286087036133,
      "learning_rate": 4.0416268878610555e-07,
      "loss": 0.0498,
      "step": 12202
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.138561248779297,
      "learning_rate": 4.037054086714298e-07,
      "loss": 0.1549,
      "step": 12203
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.124779224395752,
      "learning_rate": 4.032483765094497e-07,
      "loss": 0.8574,
      "step": 12204
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.0401611328125,
      "learning_rate": 4.027915923248199e-07,
      "loss": 0.5251,
      "step": 12205
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.044924259185791,
      "learning_rate": 4.0233505614218237e-07,
      "loss": 0.9592,
      "step": 12206
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.8275198936462402,
      "learning_rate": 4.0187876798616464e-07,
      "loss": 0.8494,
      "step": 12207
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.548675537109375,
      "learning_rate": 4.0142272788138315e-07,
      "loss": 0.9557,
      "step": 12208
    },
    {
      "epoch": 0.88,
      "grad_norm": 31.60715103149414,
      "learning_rate": 4.0096693585243926e-07,
      "loss": 0.8029,
      "step": 12209
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.520754337310791,
      "learning_rate": 4.005113919239195e-07,
      "loss": 0.1589,
      "step": 12210
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.886476039886475,
      "learning_rate": 4.000560961204014e-07,
      "loss": 0.4107,
      "step": 12211
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.636382102966309,
      "learning_rate": 3.996010484664453e-07,
      "loss": 0.454,
      "step": 12212
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.737226724624634,
      "learning_rate": 3.9914624898659827e-07,
      "loss": 0.0367,
      "step": 12213
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.778334617614746,
      "learning_rate": 3.9869169770539675e-07,
      "loss": 0.7584,
      "step": 12214
    },
    {
      "epoch": 0.88,
      "grad_norm": 19.082651138305664,
      "learning_rate": 3.982373946473611e-07,
      "loss": 0.6415,
      "step": 12215
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5686681270599365,
      "learning_rate": 3.9778333983699957e-07,
      "loss": 0.0662,
      "step": 12216
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.491811513900757,
      "learning_rate": 3.973295332988064e-07,
      "loss": 0.0491,
      "step": 12217
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.8442087173461914,
      "learning_rate": 3.968759750572626e-07,
      "loss": 0.6735,
      "step": 12218
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.7187910079956055,
      "learning_rate": 3.964226651368358e-07,
      "loss": 0.1321,
      "step": 12219
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.276286125183105,
      "learning_rate": 3.9596960356198143e-07,
      "loss": 0.4293,
      "step": 12220
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.508196830749512,
      "learning_rate": 3.955167903571394e-07,
      "loss": 0.3426,
      "step": 12221
    },
    {
      "epoch": 0.88,
      "grad_norm": 16.162979125976562,
      "learning_rate": 3.950642255467374e-07,
      "loss": 0.4876,
      "step": 12222
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2136223316192627,
      "learning_rate": 3.946119091551903e-07,
      "loss": 0.0491,
      "step": 12223
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5392067432403564,
      "learning_rate": 3.941598412068981e-07,
      "loss": 0.0551,
      "step": 12224
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.04135799407959,
      "learning_rate": 3.93708021726249e-07,
      "loss": 0.2175,
      "step": 12225
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.852971076965332,
      "learning_rate": 3.932564507376152e-07,
      "loss": 0.6465,
      "step": 12226
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.6575980186462402,
      "learning_rate": 3.928051282653594e-07,
      "loss": 0.1533,
      "step": 12227
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.16900634765625,
      "learning_rate": 3.923540543338272e-07,
      "loss": 0.5241,
      "step": 12228
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.536498069763184,
      "learning_rate": 3.9190322896735357e-07,
      "loss": 0.6574,
      "step": 12229
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4492886066436768,
      "learning_rate": 3.9145265219025785e-07,
      "loss": 0.0906,
      "step": 12230
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.809630274772644,
      "learning_rate": 3.9100232402684624e-07,
      "loss": 0.0462,
      "step": 12231
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.3601579666137695,
      "learning_rate": 3.905522445014143e-07,
      "loss": 0.628,
      "step": 12232
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.55091381072998,
      "learning_rate": 3.9010241363824144e-07,
      "loss": 0.9679,
      "step": 12233
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.51524543762207,
      "learning_rate": 3.8965283146159325e-07,
      "loss": 0.8472,
      "step": 12234
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.3546013832092285,
      "learning_rate": 3.892034979957232e-07,
      "loss": 0.3101,
      "step": 12235
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.521685600280762,
      "learning_rate": 3.887544132648724e-07,
      "loss": 0.3276,
      "step": 12236
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0759987831115723,
      "learning_rate": 3.883055772932665e-07,
      "loss": 0.1001,
      "step": 12237
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.7773637771606445,
      "learning_rate": 3.8785699010511766e-07,
      "loss": 0.1249,
      "step": 12238
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.264195203781128,
      "learning_rate": 3.874086517246273e-07,
      "loss": 0.0803,
      "step": 12239
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1112289428710938,
      "learning_rate": 3.869605621759809e-07,
      "loss": 0.0198,
      "step": 12240
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.401827812194824,
      "learning_rate": 3.865127214833497e-07,
      "loss": 0.1569,
      "step": 12241
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.114953517913818,
      "learning_rate": 3.8606512967089507e-07,
      "loss": 0.871,
      "step": 12242
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.762549877166748,
      "learning_rate": 3.856177867627614e-07,
      "loss": 0.2428,
      "step": 12243
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.2050771713256836,
      "learning_rate": 3.851706927830817e-07,
      "loss": 0.2046,
      "step": 12244
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.4102911949157715,
      "learning_rate": 3.8472384775597505e-07,
      "loss": 0.4323,
      "step": 12245
    },
    {
      "epoch": 0.88,
      "grad_norm": 14.502120971679688,
      "learning_rate": 3.842772517055476e-07,
      "loss": 1.0403,
      "step": 12246
    },
    {
      "epoch": 0.88,
      "grad_norm": 26.399484634399414,
      "learning_rate": 3.8383090465589024e-07,
      "loss": 1.4458,
      "step": 12247
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.17261028289795,
      "learning_rate": 3.83384806631083e-07,
      "loss": 0.8376,
      "step": 12248
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.9291884899139404,
      "learning_rate": 3.829389576551906e-07,
      "loss": 0.1091,
      "step": 12249
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.301917791366577,
      "learning_rate": 3.824933577522638e-07,
      "loss": 0.0468,
      "step": 12250
    },
    {
      "epoch": 0.88,
      "grad_norm": 18.068801879882812,
      "learning_rate": 3.8204800694634335e-07,
      "loss": 0.5795,
      "step": 12251
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.244985580444336,
      "learning_rate": 3.8160290526145396e-07,
      "loss": 0.828,
      "step": 12252
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.138575553894043,
      "learning_rate": 3.8115805272160413e-07,
      "loss": 0.4654,
      "step": 12253
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.836451768875122,
      "learning_rate": 3.8071344935079524e-07,
      "loss": 0.2072,
      "step": 12254
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.582329273223877,
      "learning_rate": 3.8026909517301093e-07,
      "loss": 0.6252,
      "step": 12255
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9406368732452393,
      "learning_rate": 3.798249902122214e-07,
      "loss": 0.7936,
      "step": 12256
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.314476490020752,
      "learning_rate": 3.793811344923859e-07,
      "loss": 0.2032,
      "step": 12257
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0528316497802734,
      "learning_rate": 3.7893752803744855e-07,
      "loss": 0.048,
      "step": 12258
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.510519504547119,
      "learning_rate": 3.784941708713391e-07,
      "loss": 0.6311,
      "step": 12259
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.008877754211426,
      "learning_rate": 3.7805106301797623e-07,
      "loss": 0.6262,
      "step": 12260
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.370194435119629,
      "learning_rate": 3.7760820450126356e-07,
      "loss": 0.0945,
      "step": 12261
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9261513948440552,
      "learning_rate": 3.7716559534509143e-07,
      "loss": 0.0669,
      "step": 12262
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.7740707397460938,
      "learning_rate": 3.767232355733369e-07,
      "loss": 0.5868,
      "step": 12263
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.554487943649292,
      "learning_rate": 3.7628112520986473e-07,
      "loss": 0.1026,
      "step": 12264
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.0017175674438477,
      "learning_rate": 3.7583926427852247e-07,
      "loss": 0.1248,
      "step": 12265
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.76223087310791,
      "learning_rate": 3.7539765280315e-07,
      "loss": 0.7845,
      "step": 12266
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.071172714233398,
      "learning_rate": 3.7495629080756826e-07,
      "loss": 0.5861,
      "step": 12267
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.550901412963867,
      "learning_rate": 3.745151783155887e-07,
      "loss": 0.7537,
      "step": 12268
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.909489154815674,
      "learning_rate": 3.740743153510057e-07,
      "loss": 0.4728,
      "step": 12269
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.523005485534668,
      "learning_rate": 3.736337019376035e-07,
      "loss": 0.7137,
      "step": 12270
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.002614736557007,
      "learning_rate": 3.7319333809915204e-07,
      "loss": 0.1409,
      "step": 12271
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5144765377044678,
      "learning_rate": 3.727532238594056e-07,
      "loss": 0.1662,
      "step": 12272
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.318570137023926,
      "learning_rate": 3.723133592421085e-07,
      "loss": 0.0926,
      "step": 12273
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.4521729946136475,
      "learning_rate": 3.718737442709891e-07,
      "loss": 0.0814,
      "step": 12274
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.34203052520752,
      "learning_rate": 3.7143437896976165e-07,
      "loss": 0.678,
      "step": 12275
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.786332607269287,
      "learning_rate": 3.709952633621311e-07,
      "loss": 0.6498,
      "step": 12276
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.308454513549805,
      "learning_rate": 3.705563974717835e-07,
      "loss": 0.535,
      "step": 12277
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9146382808685303,
      "learning_rate": 3.701177813223944e-07,
      "loss": 0.261,
      "step": 12278
    },
    {
      "epoch": 0.88,
      "grad_norm": 14.965248107910156,
      "learning_rate": 3.696794149376265e-07,
      "loss": 0.8986,
      "step": 12279
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.038938522338867,
      "learning_rate": 3.69241298341127e-07,
      "loss": 0.5798,
      "step": 12280
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.37550687789917,
      "learning_rate": 3.6880343155653085e-07,
      "loss": 0.1023,
      "step": 12281
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.215455055236816,
      "learning_rate": 3.6836581460746034e-07,
      "loss": 1.0265,
      "step": 12282
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9906630516052246,
      "learning_rate": 3.679284475175221e-07,
      "loss": 0.0804,
      "step": 12283
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.295919895172119,
      "learning_rate": 3.6749133031031e-07,
      "loss": 0.2004,
      "step": 12284
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.808910369873047,
      "learning_rate": 3.67054463009407e-07,
      "loss": 0.5039,
      "step": 12285
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.178732872009277,
      "learning_rate": 3.6661784563837845e-07,
      "loss": 0.7862,
      "step": 12286
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.752908229827881,
      "learning_rate": 3.661814782207784e-07,
      "loss": 0.207,
      "step": 12287
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.208844184875488,
      "learning_rate": 3.657453607801492e-07,
      "loss": 0.5668,
      "step": 12288
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.408959865570068,
      "learning_rate": 3.653094933400153e-07,
      "loss": 0.7337,
      "step": 12289
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.119586229324341,
      "learning_rate": 3.648738759238901e-07,
      "loss": 0.1523,
      "step": 12290
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.611410617828369,
      "learning_rate": 3.6443850855527487e-07,
      "loss": 0.0515,
      "step": 12291
    },
    {
      "epoch": 0.88,
      "grad_norm": 17.497751235961914,
      "learning_rate": 3.6400339125765625e-07,
      "loss": 0.6681,
      "step": 12292
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.8243439197540283,
      "learning_rate": 3.6356852405450505e-07,
      "loss": 0.7094,
      "step": 12293
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.759552955627441,
      "learning_rate": 3.6313390696928355e-07,
      "loss": 0.5075,
      "step": 12294
    },
    {
      "epoch": 0.88,
      "grad_norm": 15.518206596374512,
      "learning_rate": 3.626995400254357e-07,
      "loss": 0.78,
      "step": 12295
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.9712941646575928,
      "learning_rate": 3.6226542324639457e-07,
      "loss": 0.1611,
      "step": 12296
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5749552249908447,
      "learning_rate": 3.6183155665557855e-07,
      "loss": 0.1221,
      "step": 12297
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4670863151550293,
      "learning_rate": 3.613979402763945e-07,
      "loss": 0.1987,
      "step": 12298
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.543017864227295,
      "learning_rate": 3.6096457413223317e-07,
      "loss": 0.2796,
      "step": 12299
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.247835397720337,
      "learning_rate": 3.605314582464736e-07,
      "loss": 0.0915,
      "step": 12300
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.486591339111328,
      "learning_rate": 3.600985926424805e-07,
      "loss": 0.1168,
      "step": 12301
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.838444948196411,
      "learning_rate": 3.5966597734360454e-07,
      "loss": 0.0966,
      "step": 12302
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.462865114212036,
      "learning_rate": 3.5923361237318545e-07,
      "loss": 0.2582,
      "step": 12303
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.7501957416534424,
      "learning_rate": 3.5880149775454673e-07,
      "loss": 0.0842,
      "step": 12304
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.7267823219299316,
      "learning_rate": 3.583696335109987e-07,
      "loss": 0.0819,
      "step": 12305
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.159533977508545,
      "learning_rate": 3.579380196658394e-07,
      "loss": 0.1487,
      "step": 12306
    },
    {
      "epoch": 0.88,
      "grad_norm": 13.190673828125,
      "learning_rate": 3.5750665624235293e-07,
      "loss": 0.8445,
      "step": 12307
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8651808500289917,
      "learning_rate": 3.570755432638101e-07,
      "loss": 0.0632,
      "step": 12308
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.009599685668945,
      "learning_rate": 3.566446807534663e-07,
      "loss": 0.6768,
      "step": 12309
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.737450122833252,
      "learning_rate": 3.5621406873456664e-07,
      "loss": 0.4838,
      "step": 12310
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.434817314147949,
      "learning_rate": 3.5578370723033994e-07,
      "loss": 0.1909,
      "step": 12311
    },
    {
      "epoch": 0.88,
      "grad_norm": 21.93039321899414,
      "learning_rate": 3.553535962640037e-07,
      "loss": 0.3728,
      "step": 12312
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.790159225463867,
      "learning_rate": 3.5492373585875927e-07,
      "loss": 0.6856,
      "step": 12313
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.362520694732666,
      "learning_rate": 3.544941260377971e-07,
      "loss": 0.151,
      "step": 12314
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.866222858428955,
      "learning_rate": 3.540647668242919e-07,
      "loss": 0.097,
      "step": 12315
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.4115190505981445,
      "learning_rate": 3.5363565824140747e-07,
      "loss": 0.825,
      "step": 12316
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6090612411499023,
      "learning_rate": 3.5320680031229185e-07,
      "loss": 0.7083,
      "step": 12317
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.1873650550842285,
      "learning_rate": 3.5277819306007933e-07,
      "loss": 0.7553,
      "step": 12318
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.768379211425781,
      "learning_rate": 3.523498365078931e-07,
      "loss": 0.6433,
      "step": 12319
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.238230228424072,
      "learning_rate": 3.519217306788414e-07,
      "loss": 0.4875,
      "step": 12320
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2795464992523193,
      "learning_rate": 3.5149387559601733e-07,
      "loss": 0.0915,
      "step": 12321
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.441476821899414,
      "learning_rate": 3.5106627128250413e-07,
      "loss": 0.3264,
      "step": 12322
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.3699724674224854,
      "learning_rate": 3.506389177613684e-07,
      "loss": 0.1259,
      "step": 12323
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.529632091522217,
      "learning_rate": 3.502118150556644e-07,
      "loss": 0.1385,
      "step": 12324
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.9083423614501953,
      "learning_rate": 3.49784963188432e-07,
      "loss": 0.1842,
      "step": 12325
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.204978942871094,
      "learning_rate": 3.493583621826996e-07,
      "loss": 0.8203,
      "step": 12326
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.14370059967041,
      "learning_rate": 3.4893201206147864e-07,
      "loss": 0.3654,
      "step": 12327
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.98558521270752,
      "learning_rate": 3.4850591284777133e-07,
      "loss": 0.5559,
      "step": 12328
    },
    {
      "epoch": 0.88,
      "grad_norm": 14.341739654541016,
      "learning_rate": 3.4808006456456265e-07,
      "loss": 0.6803,
      "step": 12329
    },
    {
      "epoch": 0.88,
      "grad_norm": 17.625782012939453,
      "learning_rate": 3.476544672348259e-07,
      "loss": 0.5616,
      "step": 12330
    },
    {
      "epoch": 0.88,
      "grad_norm": 18.385404586791992,
      "learning_rate": 3.4722912088152096e-07,
      "loss": 0.9435,
      "step": 12331
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.040998935699463,
      "learning_rate": 3.468040255275934e-07,
      "loss": 0.694,
      "step": 12332
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.612416982650757,
      "learning_rate": 3.463791811959755e-07,
      "loss": 0.6633,
      "step": 12333
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.2831919193267822,
      "learning_rate": 3.459545879095849e-07,
      "loss": 0.311,
      "step": 12334
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.17158317565918,
      "learning_rate": 3.455302456913284e-07,
      "loss": 0.7366,
      "step": 12335
    },
    {
      "epoch": 0.88,
      "grad_norm": 20.217195510864258,
      "learning_rate": 3.451061545640971e-07,
      "loss": 0.6928,
      "step": 12336
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.305336952209473,
      "learning_rate": 3.4468231455076937e-07,
      "loss": 0.8195,
      "step": 12337
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.903383255004883,
      "learning_rate": 3.4425872567420913e-07,
      "loss": 0.6731,
      "step": 12338
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.210685729980469,
      "learning_rate": 3.43835387957267e-07,
      "loss": 0.6635,
      "step": 12339
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.675074815750122,
      "learning_rate": 3.4341230142278195e-07,
      "loss": 0.0657,
      "step": 12340
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.39373779296875,
      "learning_rate": 3.429894660935773e-07,
      "loss": 0.5746,
      "step": 12341
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.353218078613281,
      "learning_rate": 3.4256688199246267e-07,
      "loss": 0.7826,
      "step": 12342
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.940670132637024,
      "learning_rate": 3.4214454914223483e-07,
      "loss": 0.0686,
      "step": 12343
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0456905364990234,
      "learning_rate": 3.4172246756567885e-07,
      "loss": 0.0778,
      "step": 12344
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.049069404602051,
      "learning_rate": 3.413006372855626e-07,
      "loss": 0.1467,
      "step": 12345
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.3734307289123535,
      "learning_rate": 3.408790583246424e-07,
      "loss": 0.4103,
      "step": 12346
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1026079654693604,
      "learning_rate": 3.4045773070566224e-07,
      "loss": 0.223,
      "step": 12347
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4499833583831787,
      "learning_rate": 3.4003665445135056e-07,
      "loss": 0.1229,
      "step": 12348
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.536497116088867,
      "learning_rate": 3.3961582958442095e-07,
      "loss": 0.4295,
      "step": 12349
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.037208080291748,
      "learning_rate": 3.39195256127578e-07,
      "loss": 0.1205,
      "step": 12350
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.808578372001648,
      "learning_rate": 3.387749341035085e-07,
      "loss": 0.0683,
      "step": 12351
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.070355415344238,
      "learning_rate": 3.3835486353488656e-07,
      "loss": 0.8066,
      "step": 12352
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.3769869804382324,
      "learning_rate": 3.3793504444437576e-07,
      "loss": 0.1208,
      "step": 12353
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.057548999786377,
      "learning_rate": 3.375154768546224e-07,
      "loss": 0.8844,
      "step": 12354
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4057376384735107,
      "learning_rate": 3.3709616078825946e-07,
      "loss": 0.0926,
      "step": 12355
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.774138927459717,
      "learning_rate": 3.3667709626790945e-07,
      "loss": 0.8077,
      "step": 12356
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.449415683746338,
      "learning_rate": 3.362582833161787e-07,
      "loss": 0.0854,
      "step": 12357
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.927220344543457,
      "learning_rate": 3.3583972195565975e-07,
      "loss": 0.6715,
      "step": 12358
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.099574565887451,
      "learning_rate": 3.3542141220893334e-07,
      "loss": 0.1587,
      "step": 12359
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.606327056884766,
      "learning_rate": 3.3500335409856643e-07,
      "loss": 0.4331,
      "step": 12360
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1290981769561768,
      "learning_rate": 3.3458554764710885e-07,
      "loss": 0.1269,
      "step": 12361
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.7083985805511475,
      "learning_rate": 3.341679928771019e-07,
      "loss": 0.1634,
      "step": 12362
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.897007942199707,
      "learning_rate": 3.3375068981107094e-07,
      "loss": 0.5381,
      "step": 12363
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.723485946655273,
      "learning_rate": 3.333336384715269e-07,
      "loss": 0.4373,
      "step": 12364
    },
    {
      "epoch": 0.89,
      "grad_norm": 15.98608684539795,
      "learning_rate": 3.329168388809695e-07,
      "loss": 0.5994,
      "step": 12365
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.444554567337036,
      "learning_rate": 3.325002910618824e-07,
      "loss": 0.1787,
      "step": 12366
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.572263717651367,
      "learning_rate": 3.320839950367372e-07,
      "loss": 0.5108,
      "step": 12367
    },
    {
      "epoch": 0.89,
      "grad_norm": 58.18393325805664,
      "learning_rate": 3.316679508279913e-07,
      "loss": 0.7263,
      "step": 12368
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.051109313964844,
      "learning_rate": 3.3125215845808913e-07,
      "loss": 0.5171,
      "step": 12369
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.59794282913208,
      "learning_rate": 3.3083661794946096e-07,
      "loss": 0.6819,
      "step": 12370
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.553943634033203,
      "learning_rate": 3.3042132932452285e-07,
      "loss": 0.0713,
      "step": 12371
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4144928455352783,
      "learning_rate": 3.300062926056802e-07,
      "loss": 0.051,
      "step": 12372
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.363195419311523,
      "learning_rate": 3.2959150781531944e-07,
      "loss": 0.2625,
      "step": 12373
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.358006954193115,
      "learning_rate": 3.2917697497581946e-07,
      "loss": 0.8356,
      "step": 12374
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.991464853286743,
      "learning_rate": 3.2876269410954176e-07,
      "loss": 0.0899,
      "step": 12375
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4529871940612793,
      "learning_rate": 3.2834866523883457e-07,
      "loss": 0.1444,
      "step": 12376
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9640891551971436,
      "learning_rate": 3.2793488838603336e-07,
      "loss": 0.1253,
      "step": 12377
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.088871955871582,
      "learning_rate": 3.2752136357346087e-07,
      "loss": 0.6948,
      "step": 12378
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9993808269500732,
      "learning_rate": 3.271080908234242e-07,
      "loss": 0.2039,
      "step": 12379
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.512551307678223,
      "learning_rate": 3.2669507015821775e-07,
      "loss": 0.4373,
      "step": 12380
    },
    {
      "epoch": 0.89,
      "grad_norm": 16.253570556640625,
      "learning_rate": 3.2628230160012376e-07,
      "loss": 0.8412,
      "step": 12381
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.4842891693115234,
      "learning_rate": 3.2586978517140824e-07,
      "loss": 0.2258,
      "step": 12382
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0605921745300293,
      "learning_rate": 3.254575208943245e-07,
      "loss": 0.0613,
      "step": 12383
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.468078374862671,
      "learning_rate": 3.250455087911153e-07,
      "loss": 0.0824,
      "step": 12384
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.937717914581299,
      "learning_rate": 3.24633748884004e-07,
      "loss": 0.195,
      "step": 12385
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8104625940322876,
      "learning_rate": 3.242222411952045e-07,
      "loss": 0.076,
      "step": 12386
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.918400764465332,
      "learning_rate": 3.2381098574691684e-07,
      "loss": 0.3029,
      "step": 12387
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9935649633407593,
      "learning_rate": 3.2339998256132597e-07,
      "loss": 0.0767,
      "step": 12388
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.7999603748321533,
      "learning_rate": 3.229892316606037e-07,
      "loss": 0.1161,
      "step": 12389
    },
    {
      "epoch": 0.89,
      "grad_norm": 16.37562370300293,
      "learning_rate": 3.2257873306690944e-07,
      "loss": 0.7246,
      "step": 12390
    },
    {
      "epoch": 0.89,
      "grad_norm": 16.168033599853516,
      "learning_rate": 3.221684868023878e-07,
      "loss": 0.8422,
      "step": 12391
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.969299793243408,
      "learning_rate": 3.2175849288916873e-07,
      "loss": 0.3814,
      "step": 12392
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.26676607131958,
      "learning_rate": 3.2134875134937184e-07,
      "loss": 0.8153,
      "step": 12393
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.915968418121338,
      "learning_rate": 3.2093926220509996e-07,
      "loss": 0.6549,
      "step": 12394
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.881873607635498,
      "learning_rate": 3.205300254784427e-07,
      "loss": 0.1665,
      "step": 12395
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2045252323150635,
      "learning_rate": 3.201210411914796e-07,
      "loss": 0.1634,
      "step": 12396
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.987919807434082,
      "learning_rate": 3.197123093662713e-07,
      "loss": 0.7592,
      "step": 12397
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.452052593231201,
      "learning_rate": 3.1930383002486744e-07,
      "loss": 0.0819,
      "step": 12398
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0325121879577637,
      "learning_rate": 3.1889560318930537e-07,
      "loss": 0.1916,
      "step": 12399
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0484492778778076,
      "learning_rate": 3.1848762888160636e-07,
      "loss": 0.1161,
      "step": 12400
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0704309940338135,
      "learning_rate": 3.1807990712377833e-07,
      "loss": 0.0695,
      "step": 12401
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.062999725341797,
      "learning_rate": 3.176724379378182e-07,
      "loss": 0.1267,
      "step": 12402
    },
    {
      "epoch": 0.89,
      "grad_norm": 13.63711166381836,
      "learning_rate": 3.172652213457067e-07,
      "loss": 0.6353,
      "step": 12403
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.2593278884887695,
      "learning_rate": 3.168582573694112e-07,
      "loss": 0.6674,
      "step": 12404
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.452777862548828,
      "learning_rate": 3.164515460308848e-07,
      "loss": 0.1314,
      "step": 12405
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.6690800189971924,
      "learning_rate": 3.1604508735207093e-07,
      "loss": 0.1345,
      "step": 12406
    },
    {
      "epoch": 0.89,
      "grad_norm": 12.921305656433105,
      "learning_rate": 3.156388813548944e-07,
      "loss": 0.568,
      "step": 12407
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.5596632957458496,
      "learning_rate": 3.1523292806126926e-07,
      "loss": 0.0886,
      "step": 12408
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.825237274169922,
      "learning_rate": 3.148272274930941e-07,
      "loss": 0.4711,
      "step": 12409
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1583549976348877,
      "learning_rate": 3.144217796722554e-07,
      "loss": 0.076,
      "step": 12410
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.6584272384643555,
      "learning_rate": 3.1401658462062665e-07,
      "loss": 0.8155,
      "step": 12411
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.347036361694336,
      "learning_rate": 3.13611642360066e-07,
      "loss": 0.1331,
      "step": 12412
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3009514808654785,
      "learning_rate": 3.132069529124182e-07,
      "loss": 0.0221,
      "step": 12413
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.967491626739502,
      "learning_rate": 3.128025162995135e-07,
      "loss": 0.1685,
      "step": 12414
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8019107580184937,
      "learning_rate": 3.123983325431723e-07,
      "loss": 0.076,
      "step": 12415
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.001774787902832,
      "learning_rate": 3.119944016651971e-07,
      "loss": 0.7777,
      "step": 12416
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.068225622177124,
      "learning_rate": 3.1159072368737886e-07,
      "loss": 0.1169,
      "step": 12417
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.270143508911133,
      "learning_rate": 3.111872986314946e-07,
      "loss": 0.6189,
      "step": 12418
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.774566650390625,
      "learning_rate": 3.1078412651930734e-07,
      "loss": 0.0723,
      "step": 12419
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.185169219970703,
      "learning_rate": 3.103812073725665e-07,
      "loss": 0.3409,
      "step": 12420
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4396048784255981,
      "learning_rate": 3.0997854121300965e-07,
      "loss": 0.0217,
      "step": 12421
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.534301280975342,
      "learning_rate": 3.0957612806235717e-07,
      "loss": 0.1746,
      "step": 12422
    },
    {
      "epoch": 0.89,
      "grad_norm": 22.114912033081055,
      "learning_rate": 3.091739679423178e-07,
      "loss": 0.7818,
      "step": 12423
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.9830098152160645,
      "learning_rate": 3.0877206087458755e-07,
      "loss": 0.7274,
      "step": 12424
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0480525493621826,
      "learning_rate": 3.083704068808474e-07,
      "loss": 0.1029,
      "step": 12425
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.000018119812012,
      "learning_rate": 3.079690059827639e-07,
      "loss": 0.4767,
      "step": 12426
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.984572887420654,
      "learning_rate": 3.0756785820199365e-07,
      "loss": 0.5207,
      "step": 12427
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.340801239013672,
      "learning_rate": 3.071669635601754e-07,
      "loss": 0.7334,
      "step": 12428
    },
    {
      "epoch": 0.89,
      "grad_norm": 17.204246520996094,
      "learning_rate": 3.067663220789352e-07,
      "loss": 0.3571,
      "step": 12429
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.441127300262451,
      "learning_rate": 3.06365933779888e-07,
      "loss": 0.1002,
      "step": 12430
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.296363592147827,
      "learning_rate": 3.0596579868463204e-07,
      "loss": 0.0876,
      "step": 12431
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.74086856842041,
      "learning_rate": 3.055659168147529e-07,
      "loss": 0.7948,
      "step": 12432
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.786202430725098,
      "learning_rate": 3.051662881918249e-07,
      "loss": 0.5665,
      "step": 12433
    },
    {
      "epoch": 0.89,
      "grad_norm": 12.187708854675293,
      "learning_rate": 3.047669128374037e-07,
      "loss": 0.6301,
      "step": 12434
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.18277907371521,
      "learning_rate": 3.0436779077303413e-07,
      "loss": 0.0472,
      "step": 12435
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0181453227996826,
      "learning_rate": 3.0396892202024906e-07,
      "loss": 0.0503,
      "step": 12436
    },
    {
      "epoch": 0.89,
      "grad_norm": 19.251171112060547,
      "learning_rate": 3.0357030660056574e-07,
      "loss": 0.8264,
      "step": 12437
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2066047191619873,
      "learning_rate": 3.031719445354864e-07,
      "loss": 0.0727,
      "step": 12438
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9318113327026367,
      "learning_rate": 3.0277383584650264e-07,
      "loss": 0.1288,
      "step": 12439
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.143439769744873,
      "learning_rate": 3.023759805550908e-07,
      "loss": 0.0563,
      "step": 12440
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.336916923522949,
      "learning_rate": 3.0197837868271295e-07,
      "loss": 0.846,
      "step": 12441
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.707400798797607,
      "learning_rate": 3.015810302508182e-07,
      "loss": 0.3158,
      "step": 12442
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.3384957313537598,
      "learning_rate": 3.0118393528084323e-07,
      "loss": 0.1407,
      "step": 12443
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.8055243492126465,
      "learning_rate": 3.0078709379420866e-07,
      "loss": 0.1534,
      "step": 12444
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0954298973083496,
      "learning_rate": 3.0039050581232353e-07,
      "loss": 0.0786,
      "step": 12445
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.725569486618042,
      "learning_rate": 2.999941713565807e-07,
      "loss": 0.1221,
      "step": 12446
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.42655611038208,
      "learning_rate": 2.99598090448362e-07,
      "loss": 0.1282,
      "step": 12447
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.657528400421143,
      "learning_rate": 2.992022631090341e-07,
      "loss": 0.7423,
      "step": 12448
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9045665264129639,
      "learning_rate": 2.9880668935995117e-07,
      "loss": 0.0664,
      "step": 12449
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.560785293579102,
      "learning_rate": 2.9841136922245274e-07,
      "loss": 0.8287,
      "step": 12450
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9312251806259155,
      "learning_rate": 2.9801630271786344e-07,
      "loss": 0.0364,
      "step": 12451
    },
    {
      "epoch": 0.89,
      "grad_norm": 22.57219696044922,
      "learning_rate": 2.976214898674967e-07,
      "loss": 0.719,
      "step": 12452
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.4393956661224365,
      "learning_rate": 2.972269306926517e-07,
      "loss": 0.1518,
      "step": 12453
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.825383424758911,
      "learning_rate": 2.968326252146125e-07,
      "loss": 0.7399,
      "step": 12454
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2887651920318604,
      "learning_rate": 2.96438573454651e-07,
      "loss": 0.0455,
      "step": 12455
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.703884243965149,
      "learning_rate": 2.9604477543402456e-07,
      "loss": 0.0416,
      "step": 12456
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.5850205421447754,
      "learning_rate": 2.9565123117397675e-07,
      "loss": 0.0954,
      "step": 12457
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.112720489501953,
      "learning_rate": 2.952579406957384e-07,
      "loss": 0.7489,
      "step": 12458
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.030317783355713,
      "learning_rate": 2.948649040205259e-07,
      "loss": 0.3448,
      "step": 12459
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.923825263977051,
      "learning_rate": 2.944721211695406e-07,
      "loss": 0.7992,
      "step": 12460
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.8390493392944336,
      "learning_rate": 2.940795921639744e-07,
      "loss": 0.1047,
      "step": 12461
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.8679916858673096,
      "learning_rate": 2.9368731702500096e-07,
      "loss": 0.7113,
      "step": 12462
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.190890312194824,
      "learning_rate": 2.932952957737817e-07,
      "loss": 0.0706,
      "step": 12463
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4224488735198975,
      "learning_rate": 2.9290352843146574e-07,
      "loss": 0.096,
      "step": 12464
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4552314281463623,
      "learning_rate": 2.925120150191879e-07,
      "loss": 0.0699,
      "step": 12465
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.407994747161865,
      "learning_rate": 2.9212075555806685e-07,
      "loss": 0.886,
      "step": 12466
    },
    {
      "epoch": 0.89,
      "grad_norm": 16.45368766784668,
      "learning_rate": 2.9172975006921125e-07,
      "loss": 1.0921,
      "step": 12467
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.409639358520508,
      "learning_rate": 2.913389985737142e-07,
      "loss": 0.4645,
      "step": 12468
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.647392988204956,
      "learning_rate": 2.9094850109265493e-07,
      "loss": 0.1713,
      "step": 12469
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.240992546081543,
      "learning_rate": 2.9055825764709946e-07,
      "loss": 0.1106,
      "step": 12470
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.618550300598145,
      "learning_rate": 2.9016826825809917e-07,
      "loss": 0.4883,
      "step": 12471
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.946819305419922,
      "learning_rate": 2.897785329466929e-07,
      "loss": 0.4779,
      "step": 12472
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4878711700439453,
      "learning_rate": 2.8938905173390653e-07,
      "loss": 0.1023,
      "step": 12473
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.31403112411499,
      "learning_rate": 2.8899982464074994e-07,
      "loss": 0.8127,
      "step": 12474
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.349571704864502,
      "learning_rate": 2.886108516882197e-07,
      "loss": 0.4051,
      "step": 12475
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0764148235321045,
      "learning_rate": 2.8822213289730115e-07,
      "loss": 0.0664,
      "step": 12476
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.274862289428711,
      "learning_rate": 2.878336682889638e-07,
      "loss": 0.0739,
      "step": 12477
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.303785800933838,
      "learning_rate": 2.8744545788416346e-07,
      "loss": 0.7473,
      "step": 12478
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.616731882095337,
      "learning_rate": 2.8705750170384236e-07,
      "loss": 0.1417,
      "step": 12479
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5506476163864136,
      "learning_rate": 2.8666979976892997e-07,
      "loss": 0.0395,
      "step": 12480
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.131645202636719,
      "learning_rate": 2.862823521003405e-07,
      "loss": 0.7407,
      "step": 12481
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.126465797424316,
      "learning_rate": 2.8589515871897623e-07,
      "loss": 0.5347,
      "step": 12482
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.768901348114014,
      "learning_rate": 2.8550821964572426e-07,
      "loss": 0.645,
      "step": 12483
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.356797933578491,
      "learning_rate": 2.85121534901458e-07,
      "loss": 0.1987,
      "step": 12484
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.031864881515503,
      "learning_rate": 2.8473510450703844e-07,
      "loss": 0.0474,
      "step": 12485
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.32443618774414,
      "learning_rate": 2.8434892848331176e-07,
      "loss": 0.8493,
      "step": 12486
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.795175552368164,
      "learning_rate": 2.839630068511107e-07,
      "loss": 0.4217,
      "step": 12487
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.766528844833374,
      "learning_rate": 2.835773396312541e-07,
      "loss": 0.1246,
      "step": 12488
    },
    {
      "epoch": 0.9,
      "grad_norm": 11.377577781677246,
      "learning_rate": 2.831919268445477e-07,
      "loss": 0.6525,
      "step": 12489
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.350142478942871,
      "learning_rate": 2.8280676851178245e-07,
      "loss": 0.3256,
      "step": 12490
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3329722881317139,
      "learning_rate": 2.824218646537363e-07,
      "loss": 0.0298,
      "step": 12491
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.053171634674072,
      "learning_rate": 2.820372152911738e-07,
      "loss": 0.7714,
      "step": 12492
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.36069107055664,
      "learning_rate": 2.81652820444846e-07,
      "loss": 0.5321,
      "step": 12493
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.120163917541504,
      "learning_rate": 2.812686801354875e-07,
      "loss": 0.8908,
      "step": 12494
    },
    {
      "epoch": 0.9,
      "grad_norm": 17.67220687866211,
      "learning_rate": 2.8088479438382275e-07,
      "loss": 0.7172,
      "step": 12495
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.502696514129639,
      "learning_rate": 2.805011632105603e-07,
      "loss": 0.2566,
      "step": 12496
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.758184432983398,
      "learning_rate": 2.8011778663639524e-07,
      "loss": 0.3987,
      "step": 12497
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4282066822052,
      "learning_rate": 2.7973466468201093e-07,
      "loss": 0.2274,
      "step": 12498
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3040645122528076,
      "learning_rate": 2.793517973680737e-07,
      "loss": 0.0711,
      "step": 12499
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.1020078659057617,
      "learning_rate": 2.7896918471523814e-07,
      "loss": 0.0665,
      "step": 12500
    },
    {
      "epoch": 0.9,
      "grad_norm": 15.740233421325684,
      "learning_rate": 2.785868267441455e-07,
      "loss": 0.6233,
      "step": 12501
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7680108547210693,
      "learning_rate": 2.782047234754226e-07,
      "loss": 0.2152,
      "step": 12502
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.187744617462158,
      "learning_rate": 2.778228749296802e-07,
      "loss": 0.1421,
      "step": 12503
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.834120750427246,
      "learning_rate": 2.7744128112752067e-07,
      "loss": 0.0967,
      "step": 12504
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.184361457824707,
      "learning_rate": 2.770599420895287e-07,
      "loss": 1.1787,
      "step": 12505
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3468708992004395,
      "learning_rate": 2.7667885783627446e-07,
      "loss": 0.0712,
      "step": 12506
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.895510673522949,
      "learning_rate": 2.762980283883171e-07,
      "loss": 0.3676,
      "step": 12507
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.546288967132568,
      "learning_rate": 2.7591745376620125e-07,
      "loss": 0.741,
      "step": 12508
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.55837082862854,
      "learning_rate": 2.7553713399045666e-07,
      "loss": 0.1163,
      "step": 12509
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.6826541423797607,
      "learning_rate": 2.7515706908160135e-07,
      "loss": 0.0677,
      "step": 12510
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.6458680629730225,
      "learning_rate": 2.747772590601372e-07,
      "loss": 0.1199,
      "step": 12511
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.306379556655884,
      "learning_rate": 2.7439770394655397e-07,
      "loss": 0.1524,
      "step": 12512
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.799072504043579,
      "learning_rate": 2.740184037613275e-07,
      "loss": 0.8602,
      "step": 12513
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.445436477661133,
      "learning_rate": 2.736393585249192e-07,
      "loss": 0.184,
      "step": 12514
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.608370542526245,
      "learning_rate": 2.732605682577777e-07,
      "loss": 0.1146,
      "step": 12515
    },
    {
      "epoch": 0.9,
      "grad_norm": 15.712837219238281,
      "learning_rate": 2.7288203298033556e-07,
      "loss": 0.9692,
      "step": 12516
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.721858501434326,
      "learning_rate": 2.7250375271301644e-07,
      "loss": 0.156,
      "step": 12517
    },
    {
      "epoch": 0.9,
      "grad_norm": 18.598608016967773,
      "learning_rate": 2.72125727476224e-07,
      "loss": 0.6265,
      "step": 12518
    },
    {
      "epoch": 0.9,
      "grad_norm": 15.653858184814453,
      "learning_rate": 2.717479572903531e-07,
      "loss": 0.3943,
      "step": 12519
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.368098258972168,
      "learning_rate": 2.7137044217578235e-07,
      "loss": 0.8615,
      "step": 12520
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.4954352378845215,
      "learning_rate": 2.709931821528777e-07,
      "loss": 0.4519,
      "step": 12521
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.135989665985107,
      "learning_rate": 2.706161772419896e-07,
      "loss": 0.695,
      "step": 12522
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.232766628265381,
      "learning_rate": 2.702394274634584e-07,
      "loss": 0.8273,
      "step": 12523
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.0316648483276367,
      "learning_rate": 2.698629328376062e-07,
      "loss": 0.1874,
      "step": 12524
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0536930561065674,
      "learning_rate": 2.69486693384744e-07,
      "loss": 0.0596,
      "step": 12525
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.351071834564209,
      "learning_rate": 2.691107091251693e-07,
      "loss": 0.0947,
      "step": 12526
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.012739181518555,
      "learning_rate": 2.6873498007916445e-07,
      "loss": 0.2652,
      "step": 12527
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.367259979248047,
      "learning_rate": 2.6835950626699803e-07,
      "loss": 0.4986,
      "step": 12528
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.638638496398926,
      "learning_rate": 2.679842877089273e-07,
      "loss": 0.0768,
      "step": 12529
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.801723480224609,
      "learning_rate": 2.676093244251915e-07,
      "loss": 1.0774,
      "step": 12530
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.158856153488159,
      "learning_rate": 2.6723461643601953e-07,
      "loss": 0.0993,
      "step": 12531
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.11668062210083,
      "learning_rate": 2.668601637616258e-07,
      "loss": 0.0931,
      "step": 12532
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0548343658447266,
      "learning_rate": 2.664859664222108e-07,
      "loss": 0.0588,
      "step": 12533
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.1210689544677734,
      "learning_rate": 2.6611202443795947e-07,
      "loss": 0.1779,
      "step": 12534
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.7129788398742676,
      "learning_rate": 2.6573833782904625e-07,
      "loss": 0.1183,
      "step": 12535
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.694741249084473,
      "learning_rate": 2.6536490661562995e-07,
      "loss": 0.6157,
      "step": 12536
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.601005554199219,
      "learning_rate": 2.649917308178546e-07,
      "loss": 0.779,
      "step": 12537
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.260586261749268,
      "learning_rate": 2.646188104558528e-07,
      "loss": 0.8131,
      "step": 12538
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.553663730621338,
      "learning_rate": 2.6424614554974195e-07,
      "loss": 0.1324,
      "step": 12539
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.139373302459717,
      "learning_rate": 2.6387373611962476e-07,
      "loss": 0.1329,
      "step": 12540
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.6614468097686768,
      "learning_rate": 2.635015821855941e-07,
      "loss": 0.1006,
      "step": 12541
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.15763521194458,
      "learning_rate": 2.6312968376772384e-07,
      "loss": 0.0672,
      "step": 12542
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.364349603652954,
      "learning_rate": 2.627580408860764e-07,
      "loss": 0.1126,
      "step": 12543
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.207867383956909,
      "learning_rate": 2.623866535607017e-07,
      "loss": 0.0946,
      "step": 12544
    },
    {
      "epoch": 0.9,
      "grad_norm": 18.198694229125977,
      "learning_rate": 2.6201552181163435e-07,
      "loss": 0.6805,
      "step": 12545
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.8729612827301025,
      "learning_rate": 2.6164464565889446e-07,
      "loss": 0.7154,
      "step": 12546
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.6361184120178223,
      "learning_rate": 2.61274025122491e-07,
      "loss": 0.0733,
      "step": 12547
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4689905643463135,
      "learning_rate": 2.609036602224169e-07,
      "loss": 0.1471,
      "step": 12548
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.030699729919434,
      "learning_rate": 2.6053355097865216e-07,
      "loss": 0.3747,
      "step": 12549
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.236319065093994,
      "learning_rate": 2.6016369741116153e-07,
      "loss": 0.7471,
      "step": 12550
    },
    {
      "epoch": 0.9,
      "grad_norm": 17.189834594726562,
      "learning_rate": 2.5979409953989845e-07,
      "loss": 0.4624,
      "step": 12551
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.721862554550171,
      "learning_rate": 2.594247573848019e-07,
      "loss": 0.0481,
      "step": 12552
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.96198570728302,
      "learning_rate": 2.5905567096579486e-07,
      "loss": 0.0894,
      "step": 12553
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.665874481201172,
      "learning_rate": 2.5868684030278926e-07,
      "loss": 0.3033,
      "step": 12554
    },
    {
      "epoch": 0.9,
      "grad_norm": 14.391423225402832,
      "learning_rate": 2.5831826541568127e-07,
      "loss": 0.8968,
      "step": 12555
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.285662651062012,
      "learning_rate": 2.579499463243551e-07,
      "loss": 0.8543,
      "step": 12556
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.990342378616333,
      "learning_rate": 2.5758188304867984e-07,
      "loss": 0.6776,
      "step": 12557
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3828487396240234,
      "learning_rate": 2.5721407560851066e-07,
      "loss": 0.1117,
      "step": 12558
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.105754375457764,
      "learning_rate": 2.5684652402368893e-07,
      "loss": 0.7797,
      "step": 12559
    },
    {
      "epoch": 0.9,
      "grad_norm": 24.231979370117188,
      "learning_rate": 2.564792283140444e-07,
      "loss": 0.8609,
      "step": 12560
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.327877998352051,
      "learning_rate": 2.561121884993906e-07,
      "loss": 0.2236,
      "step": 12561
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.156100273132324,
      "learning_rate": 2.557454045995267e-07,
      "loss": 0.5827,
      "step": 12562
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.611416816711426,
      "learning_rate": 2.5537887663424075e-07,
      "loss": 0.2478,
      "step": 12563
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.6722609996795654,
      "learning_rate": 2.550126046233048e-07,
      "loss": 0.11,
      "step": 12564
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.764381408691406,
      "learning_rate": 2.546465885864785e-07,
      "loss": 0.5539,
      "step": 12565
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.915409088134766,
      "learning_rate": 2.5428082854350664e-07,
      "loss": 0.7368,
      "step": 12566
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.335064172744751,
      "learning_rate": 2.539153245141207e-07,
      "loss": 0.0487,
      "step": 12567
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.4565815925598145,
      "learning_rate": 2.535500765180371e-07,
      "loss": 0.6711,
      "step": 12568
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.084343671798706,
      "learning_rate": 2.5318508457496116e-07,
      "loss": 0.0444,
      "step": 12569
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.54994535446167,
      "learning_rate": 2.528203487045822e-07,
      "loss": 0.1039,
      "step": 12570
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0274038314819336,
      "learning_rate": 2.5245586892657603e-07,
      "loss": 0.0674,
      "step": 12571
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.362783670425415,
      "learning_rate": 2.520916452606059e-07,
      "loss": 0.1164,
      "step": 12572
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.6870648860931396,
      "learning_rate": 2.5172767772631946e-07,
      "loss": 0.0925,
      "step": 12573
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.117609977722168,
      "learning_rate": 2.5136396634335093e-07,
      "loss": 0.6191,
      "step": 12574
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.077474594116211,
      "learning_rate": 2.5100051113132294e-07,
      "loss": 0.7892,
      "step": 12575
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.8468449115753174,
      "learning_rate": 2.5063731210984097e-07,
      "loss": 0.9268,
      "step": 12576
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.916049957275391,
      "learning_rate": 2.5027436929849933e-07,
      "loss": 0.4697,
      "step": 12577
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.279228925704956,
      "learning_rate": 2.499116827168763e-07,
      "loss": 0.2397,
      "step": 12578
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9577971696853638,
      "learning_rate": 2.4954925238453776e-07,
      "loss": 0.0649,
      "step": 12579
    },
    {
      "epoch": 0.9,
      "grad_norm": 24.78845977783203,
      "learning_rate": 2.491870783210354e-07,
      "loss": 0.7662,
      "step": 12580
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.025628089904785,
      "learning_rate": 2.48825160545908e-07,
      "loss": 0.7321,
      "step": 12581
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.2962698936462402,
      "learning_rate": 2.4846349907867886e-07,
      "loss": 0.1621,
      "step": 12582
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4676597118377686,
      "learning_rate": 2.4810209393885786e-07,
      "loss": 0.0528,
      "step": 12583
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.584398031234741,
      "learning_rate": 2.4774094514594225e-07,
      "loss": 0.1351,
      "step": 12584
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.460158109664917,
      "learning_rate": 2.473800527194148e-07,
      "loss": 0.1529,
      "step": 12585
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.45011043548584,
      "learning_rate": 2.4701941667874375e-07,
      "loss": 0.5544,
      "step": 12586
    },
    {
      "epoch": 0.9,
      "grad_norm": 21.675159454345703,
      "learning_rate": 2.4665903704338303e-07,
      "loss": 0.5304,
      "step": 12587
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9012147188186646,
      "learning_rate": 2.462989138327765e-07,
      "loss": 0.0399,
      "step": 12588
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.1320812702178955,
      "learning_rate": 2.459390470663492e-07,
      "loss": 0.186,
      "step": 12589
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8081296682357788,
      "learning_rate": 2.4557943676351505e-07,
      "loss": 0.0377,
      "step": 12590
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.335807800292969,
      "learning_rate": 2.4522008294367406e-07,
      "loss": 0.2661,
      "step": 12591
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.9369099140167236,
      "learning_rate": 2.448609856262113e-07,
      "loss": 0.1287,
      "step": 12592
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.794247150421143,
      "learning_rate": 2.445021448304996e-07,
      "loss": 0.4812,
      "step": 12593
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.49517560005188,
      "learning_rate": 2.441435605758968e-07,
      "loss": 0.091,
      "step": 12594
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.247708320617676,
      "learning_rate": 2.4378523288174695e-07,
      "loss": 0.4827,
      "step": 12595
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.620159149169922,
      "learning_rate": 2.4342716176738e-07,
      "loss": 0.099,
      "step": 12596
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.595175266265869,
      "learning_rate": 2.430693472521134e-07,
      "loss": 0.1375,
      "step": 12597
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9784654378890991,
      "learning_rate": 2.4271178935524997e-07,
      "loss": 0.0802,
      "step": 12598
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.709358215332031,
      "learning_rate": 2.4235448809607707e-07,
      "loss": 0.6449,
      "step": 12599
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.749527931213379,
      "learning_rate": 2.419974434938721e-07,
      "loss": 0.3831,
      "step": 12600
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.931557655334473,
      "learning_rate": 2.416406555678952e-07,
      "loss": 0.3135,
      "step": 12601
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.7509658336639404,
      "learning_rate": 2.412841243373926e-07,
      "loss": 0.0903,
      "step": 12602
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.064497470855713,
      "learning_rate": 2.4092784982159947e-07,
      "loss": 0.2628,
      "step": 12603
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.9959142208099365,
      "learning_rate": 2.4057183203973433e-07,
      "loss": 0.6475,
      "step": 12604
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.451934814453125,
      "learning_rate": 2.4021607101100296e-07,
      "loss": 0.8278,
      "step": 12605
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.728521347045898,
      "learning_rate": 2.398605667545983e-07,
      "loss": 0.3513,
      "step": 12606
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.86571979522705,
      "learning_rate": 2.3950531928969847e-07,
      "loss": 0.7399,
      "step": 12607
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.361314058303833,
      "learning_rate": 2.391503286354663e-07,
      "loss": 0.0565,
      "step": 12608
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.7466611862182617,
      "learning_rate": 2.3879559481105374e-07,
      "loss": 0.1653,
      "step": 12609
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.8590049743652344,
      "learning_rate": 2.3844111783559666e-07,
      "loss": 0.1393,
      "step": 12610
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.387742042541504,
      "learning_rate": 2.380868977282169e-07,
      "loss": 0.6436,
      "step": 12611
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.301571846008301,
      "learning_rate": 2.3773293450802538e-07,
      "loss": 0.9332,
      "step": 12612
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.47843599319458,
      "learning_rate": 2.3737922819411619e-07,
      "loss": 0.1538,
      "step": 12613
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.457209587097168,
      "learning_rate": 2.370257788055691e-07,
      "loss": 0.7179,
      "step": 12614
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.443408489227295,
      "learning_rate": 2.3667258636145273e-07,
      "loss": 0.6751,
      "step": 12615
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.74122142791748,
      "learning_rate": 2.363196508808202e-07,
      "loss": 0.3584,
      "step": 12616
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.5602569580078125,
      "learning_rate": 2.3596697238271015e-07,
      "loss": 0.116,
      "step": 12617
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.9105958938598633,
      "learning_rate": 2.3561455088615014e-07,
      "loss": 0.147,
      "step": 12618
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3675174713134766,
      "learning_rate": 2.352623864101511e-07,
      "loss": 0.0592,
      "step": 12619
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.5541369915008545,
      "learning_rate": 2.3491047897371e-07,
      "loss": 0.0687,
      "step": 12620
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.643072128295898,
      "learning_rate": 2.3455882859581225e-07,
      "loss": 0.4435,
      "step": 12621
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.3431026935577393,
      "learning_rate": 2.342074352954271e-07,
      "loss": 0.1474,
      "step": 12622
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.924839735031128,
      "learning_rate": 2.3385629909151164e-07,
      "loss": 0.745,
      "step": 12623
    },
    {
      "epoch": 0.91,
      "grad_norm": 60.882789611816406,
      "learning_rate": 2.335054200030079e-07,
      "loss": 0.8156,
      "step": 12624
    },
    {
      "epoch": 0.91,
      "grad_norm": 22.903846740722656,
      "learning_rate": 2.3315479804884522e-07,
      "loss": 0.4961,
      "step": 12625
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.3183465003967285,
      "learning_rate": 2.3280443324793677e-07,
      "loss": 0.6275,
      "step": 12626
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.8940508365631104,
      "learning_rate": 2.324543256191847e-07,
      "loss": 0.7419,
      "step": 12627
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.5236518383026123,
      "learning_rate": 2.3210447518147549e-07,
      "loss": 0.1552,
      "step": 12628
    },
    {
      "epoch": 0.91,
      "grad_norm": 12.936868667602539,
      "learning_rate": 2.3175488195368302e-07,
      "loss": 0.468,
      "step": 12629
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.467458724975586,
      "learning_rate": 2.314055459546649e-07,
      "loss": 0.7889,
      "step": 12630
    },
    {
      "epoch": 0.91,
      "grad_norm": 14.885427474975586,
      "learning_rate": 2.310564672032678e-07,
      "loss": 0.5393,
      "step": 12631
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.001253604888916,
      "learning_rate": 2.3070764571832326e-07,
      "loss": 0.5939,
      "step": 12632
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.276041030883789,
      "learning_rate": 2.303590815186474e-07,
      "loss": 0.4852,
      "step": 12633
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.185901641845703,
      "learning_rate": 2.300107746230462e-07,
      "loss": 0.2038,
      "step": 12634
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.92681884765625,
      "learning_rate": 2.2966272505030806e-07,
      "loss": 0.4968,
      "step": 12635
    },
    {
      "epoch": 0.91,
      "grad_norm": 20.27933692932129,
      "learning_rate": 2.2931493281920847e-07,
      "loss": 0.8963,
      "step": 12636
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.1665759086608887,
      "learning_rate": 2.289673979485113e-07,
      "loss": 0.2039,
      "step": 12637
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.232137680053711,
      "learning_rate": 2.286201204569638e-07,
      "loss": 0.3721,
      "step": 12638
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.1294314861297607,
      "learning_rate": 2.282731003632993e-07,
      "loss": 0.1124,
      "step": 12639
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.0753676891326904,
      "learning_rate": 2.279263376862395e-07,
      "loss": 0.2229,
      "step": 12640
    },
    {
      "epoch": 0.91,
      "grad_norm": 18.471092224121094,
      "learning_rate": 2.2757983244449e-07,
      "loss": 0.6857,
      "step": 12641
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.516937017440796,
      "learning_rate": 2.2723358465674417e-07,
      "loss": 0.1196,
      "step": 12642
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.359994888305664,
      "learning_rate": 2.268875943416804e-07,
      "loss": 0.7326,
      "step": 12643
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2398452758789062,
      "learning_rate": 2.2654186151796432e-07,
      "loss": 0.1458,
      "step": 12644
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.2792329788208,
      "learning_rate": 2.2619638620424544e-07,
      "loss": 0.6361,
      "step": 12645
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.941854476928711,
      "learning_rate": 2.2585116841916222e-07,
      "loss": 0.8528,
      "step": 12646
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.5902557373046875,
      "learning_rate": 2.255062081813375e-07,
      "loss": 0.8634,
      "step": 12647
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.588583946228027,
      "learning_rate": 2.2516150550937977e-07,
      "loss": 0.4967,
      "step": 12648
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.595818281173706,
      "learning_rate": 2.2481706042188578e-07,
      "loss": 0.0645,
      "step": 12649
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.215610027313232,
      "learning_rate": 2.2447287293743681e-07,
      "loss": 0.6983,
      "step": 12650
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.308649063110352,
      "learning_rate": 2.2412894307459855e-07,
      "loss": 0.7522,
      "step": 12651
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.818141460418701,
      "learning_rate": 2.2378527085192725e-07,
      "loss": 0.5057,
      "step": 12652
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.794283151626587,
      "learning_rate": 2.2344185628796145e-07,
      "loss": 0.1599,
      "step": 12653
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.192471504211426,
      "learning_rate": 2.230986994012263e-07,
      "loss": 0.7237,
      "step": 12654
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4034085273742676,
      "learning_rate": 2.2275580021023535e-07,
      "loss": 0.0993,
      "step": 12655
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.6724488735198975,
      "learning_rate": 2.2241315873348658e-07,
      "loss": 0.0714,
      "step": 12656
    },
    {
      "epoch": 0.91,
      "grad_norm": 16.549928665161133,
      "learning_rate": 2.2207077498946294e-07,
      "loss": 1.1228,
      "step": 12657
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.541108131408691,
      "learning_rate": 2.2172864899663636e-07,
      "loss": 0.9048,
      "step": 12658
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.7525434494018555,
      "learning_rate": 2.213867807734621e-07,
      "loss": 0.1541,
      "step": 12659
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.335752964019775,
      "learning_rate": 2.2104517033838314e-07,
      "loss": 0.5095,
      "step": 12660
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.270851135253906,
      "learning_rate": 2.2070381770982752e-07,
      "loss": 0.1257,
      "step": 12661
    },
    {
      "epoch": 0.91,
      "grad_norm": 17.854875564575195,
      "learning_rate": 2.2036272290621109e-07,
      "loss": 0.9465,
      "step": 12662
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.066805362701416,
      "learning_rate": 2.2002188594593298e-07,
      "loss": 0.6686,
      "step": 12663
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.874240398406982,
      "learning_rate": 2.196813068473813e-07,
      "loss": 0.3954,
      "step": 12664
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.717852592468262,
      "learning_rate": 2.1934098562892858e-07,
      "loss": 0.3102,
      "step": 12665
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.272226333618164,
      "learning_rate": 2.1900092230893454e-07,
      "loss": 0.8252,
      "step": 12666
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.927056789398193,
      "learning_rate": 2.186611169057423e-07,
      "loss": 0.8082,
      "step": 12667
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2557454109191895,
      "learning_rate": 2.1832156943768557e-07,
      "loss": 0.0257,
      "step": 12668
    },
    {
      "epoch": 0.91,
      "grad_norm": 21.285261154174805,
      "learning_rate": 2.1798227992308018e-07,
      "loss": 1.2095,
      "step": 12669
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.385859727859497,
      "learning_rate": 2.1764324838022932e-07,
      "loss": 0.1012,
      "step": 12670
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.0735063552856445,
      "learning_rate": 2.173044748274239e-07,
      "loss": 0.1832,
      "step": 12671
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.23132586479187,
      "learning_rate": 2.1696595928293817e-07,
      "loss": 0.1955,
      "step": 12672
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2899413108825684,
      "learning_rate": 2.1662770176503367e-07,
      "loss": 0.1818,
      "step": 12673
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.007453441619873,
      "learning_rate": 2.1628970229196023e-07,
      "loss": 0.0974,
      "step": 12674
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.646735191345215,
      "learning_rate": 2.159519608819488e-07,
      "loss": 0.597,
      "step": 12675
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8372331857681274,
      "learning_rate": 2.156144775532204e-07,
      "loss": 0.0725,
      "step": 12676
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.8906595706939697,
      "learning_rate": 2.1527725232398154e-07,
      "loss": 0.1594,
      "step": 12677
    },
    {
      "epoch": 0.91,
      "grad_norm": 13.411100387573242,
      "learning_rate": 2.1494028521242328e-07,
      "loss": 0.6177,
      "step": 12678
    },
    {
      "epoch": 0.91,
      "grad_norm": 14.189372062683105,
      "learning_rate": 2.146035762367238e-07,
      "loss": 0.8944,
      "step": 12679
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4031033515930176,
      "learning_rate": 2.1426712541504858e-07,
      "loss": 0.0976,
      "step": 12680
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1692233085632324,
      "learning_rate": 2.13930932765547e-07,
      "loss": 0.0635,
      "step": 12681
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.347118377685547,
      "learning_rate": 2.1359499830635455e-07,
      "loss": 0.8844,
      "step": 12682
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.056440830230713,
      "learning_rate": 2.1325932205559508e-07,
      "loss": 0.7648,
      "step": 12683
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.5273993015289307,
      "learning_rate": 2.129239040313763e-07,
      "loss": 0.111,
      "step": 12684
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.564906597137451,
      "learning_rate": 2.125887442517921e-07,
      "loss": 0.0859,
      "step": 12685
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.392367362976074,
      "learning_rate": 2.1225384273492522e-07,
      "loss": 0.4481,
      "step": 12686
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.9344186782836914,
      "learning_rate": 2.1191919949884067e-07,
      "loss": 0.7879,
      "step": 12687
    },
    {
      "epoch": 0.91,
      "grad_norm": 19.613985061645508,
      "learning_rate": 2.1158481456159008e-07,
      "loss": 0.5589,
      "step": 12688
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.171245574951172,
      "learning_rate": 2.112506879412146e-07,
      "loss": 0.7715,
      "step": 12689
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4520246982574463,
      "learning_rate": 2.109168196557382e-07,
      "loss": 0.0938,
      "step": 12690
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.446494102478027,
      "learning_rate": 2.105832097231708e-07,
      "loss": 0.6345,
      "step": 12691
    },
    {
      "epoch": 0.91,
      "grad_norm": 20.09239959716797,
      "learning_rate": 2.1024985816151146e-07,
      "loss": 0.8798,
      "step": 12692
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1884047985076904,
      "learning_rate": 2.0991676498874125e-07,
      "loss": 0.0671,
      "step": 12693
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4217371940612793,
      "learning_rate": 2.0958393022283085e-07,
      "loss": 0.0922,
      "step": 12694
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.20625638961792,
      "learning_rate": 2.0925135388173367e-07,
      "loss": 0.7099,
      "step": 12695
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.448038578033447,
      "learning_rate": 2.0891903598339314e-07,
      "loss": 0.888,
      "step": 12696
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.165228843688965,
      "learning_rate": 2.085869765457349e-07,
      "loss": 0.635,
      "step": 12697
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.6345338821411133,
      "learning_rate": 2.0825517558667297e-07,
      "loss": 0.1651,
      "step": 12698
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.562887191772461,
      "learning_rate": 2.0792363312410636e-07,
      "loss": 0.1508,
      "step": 12699
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.969873428344727,
      "learning_rate": 2.0759234917592019e-07,
      "loss": 0.2639,
      "step": 12700
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.6615076065063477,
      "learning_rate": 2.0726132375998742e-07,
      "loss": 0.7161,
      "step": 12701
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.798562049865723,
      "learning_rate": 2.0693055689416431e-07,
      "loss": 0.7242,
      "step": 12702
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.426642417907715,
      "learning_rate": 2.0660004859629544e-07,
      "loss": 0.6465,
      "step": 12703
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.1214349269866943,
      "learning_rate": 2.0626979888420884e-07,
      "loss": 0.198,
      "step": 12704
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.9276039600372314,
      "learning_rate": 2.0593980777572243e-07,
      "loss": 0.1535,
      "step": 12705
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.3316149711608887,
      "learning_rate": 2.056100752886364e-07,
      "loss": 0.1695,
      "step": 12706
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.8538241386413574,
      "learning_rate": 2.052806014407388e-07,
      "loss": 0.0988,
      "step": 12707
    },
    {
      "epoch": 0.91,
      "grad_norm": 13.618178367614746,
      "learning_rate": 2.0495138624980427e-07,
      "loss": 0.7448,
      "step": 12708
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2117698192596436,
      "learning_rate": 2.0462242973359193e-07,
      "loss": 0.0685,
      "step": 12709
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.889410972595215,
      "learning_rate": 2.0429373190984814e-07,
      "loss": 0.512,
      "step": 12710
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4688639640808105,
      "learning_rate": 2.039652927963054e-07,
      "loss": 0.073,
      "step": 12711
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4710488319396973,
      "learning_rate": 2.0363711241068064e-07,
      "loss": 0.0744,
      "step": 12712
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.888217926025391,
      "learning_rate": 2.0330919077067747e-07,
      "loss": 0.6261,
      "step": 12713
    },
    {
      "epoch": 0.91,
      "grad_norm": 14.554941177368164,
      "learning_rate": 2.0298152789398784e-07,
      "loss": 0.8888,
      "step": 12714
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.7978436946868896,
      "learning_rate": 2.0265412379828764e-07,
      "loss": 0.1834,
      "step": 12715
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.582709312438965,
      "learning_rate": 2.0232697850123718e-07,
      "loss": 0.5294,
      "step": 12716
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.635746479034424,
      "learning_rate": 2.0200009202048676e-07,
      "loss": 0.2535,
      "step": 12717
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.486419200897217,
      "learning_rate": 2.0167346437367065e-07,
      "loss": 0.1433,
      "step": 12718
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.377751111984253,
      "learning_rate": 2.0134709557840691e-07,
      "loss": 0.099,
      "step": 12719
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.8738389015197754,
      "learning_rate": 2.0102098565230487e-07,
      "loss": 0.1538,
      "step": 12720
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.168304443359375,
      "learning_rate": 2.0069513461295486e-07,
      "loss": 0.299,
      "step": 12721
    },
    {
      "epoch": 0.91,
      "grad_norm": 13.094108581542969,
      "learning_rate": 2.0036954247793672e-07,
      "loss": 0.7354,
      "step": 12722
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8225551843643188,
      "learning_rate": 2.000442092648136e-07,
      "loss": 0.0395,
      "step": 12723
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.561399459838867,
      "learning_rate": 1.9971913499113705e-07,
      "loss": 0.7309,
      "step": 12724
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.402515888214111,
      "learning_rate": 1.993943196744419e-07,
      "loss": 0.286,
      "step": 12725
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.533314228057861,
      "learning_rate": 1.9906976333225302e-07,
      "loss": 0.5212,
      "step": 12726
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0011723041534424,
      "learning_rate": 1.9874546598207754e-07,
      "loss": 0.0688,
      "step": 12727
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.03205680847168,
      "learning_rate": 1.9842142764141037e-07,
      "loss": 0.3807,
      "step": 12728
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.937721252441406,
      "learning_rate": 1.9809764832773248e-07,
      "loss": 0.6891,
      "step": 12729
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9905359745025635,
      "learning_rate": 1.9777412805850993e-07,
      "loss": 0.0455,
      "step": 12730
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2710049152374268,
      "learning_rate": 1.9745086685119596e-07,
      "loss": 0.115,
      "step": 12731
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.154037952423096,
      "learning_rate": 1.9712786472322886e-07,
      "loss": 0.8522,
      "step": 12732
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.076123237609863,
      "learning_rate": 1.9680512169203413e-07,
      "loss": 0.8211,
      "step": 12733
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.4735212326049805,
      "learning_rate": 1.9648263777502173e-07,
      "loss": 0.7267,
      "step": 12734
    },
    {
      "epoch": 0.91,
      "grad_norm": 15.566096305847168,
      "learning_rate": 1.9616041298958887e-07,
      "loss": 0.5057,
      "step": 12735
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2292678356170654,
      "learning_rate": 1.958384473531183e-07,
      "loss": 0.179,
      "step": 12736
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.230456352233887,
      "learning_rate": 1.955167408829778e-07,
      "loss": 0.7514,
      "step": 12737
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9896095991134644,
      "learning_rate": 1.95195293596524e-07,
      "loss": 0.0637,
      "step": 12738
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.7805614471435547,
      "learning_rate": 1.9487410551109697e-07,
      "loss": 0.1004,
      "step": 12739
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4259841442108154,
      "learning_rate": 1.945531766440234e-07,
      "loss": 0.0686,
      "step": 12740
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3441686630249023,
      "learning_rate": 1.9423250701261608e-07,
      "loss": 0.0547,
      "step": 12741
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.11452865600586,
      "learning_rate": 1.9391209663417455e-07,
      "loss": 0.6559,
      "step": 12742
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.16102123260498,
      "learning_rate": 1.9359194552598326e-07,
      "loss": 0.5174,
      "step": 12743
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.5076773166656494,
      "learning_rate": 1.9327205370531287e-07,
      "loss": 0.2268,
      "step": 12744
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.004333972930908,
      "learning_rate": 1.9295242118942125e-07,
      "loss": 0.0509,
      "step": 12745
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.931553840637207,
      "learning_rate": 1.9263304799555127e-07,
      "loss": 0.1166,
      "step": 12746
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2914345264434814,
      "learning_rate": 1.923139341409308e-07,
      "loss": 0.1678,
      "step": 12747
    },
    {
      "epoch": 0.91,
      "grad_norm": 17.622304916381836,
      "learning_rate": 1.9199507964277552e-07,
      "loss": 1.099,
      "step": 12748
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.268385410308838,
      "learning_rate": 1.9167648451828669e-07,
      "loss": 0.0578,
      "step": 12749
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.584646701812744,
      "learning_rate": 1.9135814878465053e-07,
      "loss": 0.1302,
      "step": 12750
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.1001687049865723,
      "learning_rate": 1.910400724590411e-07,
      "loss": 0.1902,
      "step": 12751
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.079113245010376,
      "learning_rate": 1.907222555586169e-07,
      "loss": 0.0712,
      "step": 12752
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4782941341400146,
      "learning_rate": 1.9040469810052199e-07,
      "loss": 0.0584,
      "step": 12753
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.9776103496551514,
      "learning_rate": 1.900874001018893e-07,
      "loss": 0.1496,
      "step": 12754
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.691898345947266,
      "learning_rate": 1.8977036157983463e-07,
      "loss": 0.3595,
      "step": 12755
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.0925793647766113,
      "learning_rate": 1.894535825514604e-07,
      "loss": 0.156,
      "step": 12756
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.999355316162109,
      "learning_rate": 1.8913706303385736e-07,
      "loss": 0.4924,
      "step": 12757
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.025839328765869,
      "learning_rate": 1.888208030441002e-07,
      "loss": 0.0389,
      "step": 12758
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2132883071899414,
      "learning_rate": 1.8850480259924863e-07,
      "loss": 0.225,
      "step": 12759
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.2663373947143555,
      "learning_rate": 1.8818906171635064e-07,
      "loss": 0.8434,
      "step": 12760
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.178778648376465,
      "learning_rate": 1.878735804124393e-07,
      "loss": 0.5188,
      "step": 12761
    },
    {
      "epoch": 0.92,
      "grad_norm": 16.711715698242188,
      "learning_rate": 1.8755835870453264e-07,
      "loss": 0.9594,
      "step": 12762
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6566715240478516,
      "learning_rate": 1.8724339660963707e-07,
      "loss": 0.1133,
      "step": 12763
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9801530838012695,
      "learning_rate": 1.8692869414474292e-07,
      "loss": 0.8045,
      "step": 12764
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.882974147796631,
      "learning_rate": 1.8661425132682654e-07,
      "loss": 0.1055,
      "step": 12765
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.905172109603882,
      "learning_rate": 1.8630006817285275e-07,
      "loss": 0.6526,
      "step": 12766
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.347677230834961,
      "learning_rate": 1.8598614469976904e-07,
      "loss": 0.0796,
      "step": 12767
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.676992416381836,
      "learning_rate": 1.8567248092451086e-07,
      "loss": 0.801,
      "step": 12768
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.2914350032806396,
      "learning_rate": 1.8535907686399846e-07,
      "loss": 0.185,
      "step": 12769
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.2687413692474365,
      "learning_rate": 1.8504593253514059e-07,
      "loss": 0.1028,
      "step": 12770
    },
    {
      "epoch": 0.92,
      "grad_norm": 15.131204605102539,
      "learning_rate": 1.8473304795482816e-07,
      "loss": 0.6704,
      "step": 12771
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.675586223602295,
      "learning_rate": 1.8442042313994157e-07,
      "loss": 0.1984,
      "step": 12772
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.321180820465088,
      "learning_rate": 1.8410805810734512e-07,
      "loss": 0.7262,
      "step": 12773
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.101781845092773,
      "learning_rate": 1.8379595287388974e-07,
      "loss": 0.6436,
      "step": 12774
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1083269119262695,
      "learning_rate": 1.8348410745641144e-07,
      "loss": 0.0978,
      "step": 12775
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9718868732452393,
      "learning_rate": 1.8317252187173505e-07,
      "loss": 0.0664,
      "step": 12776
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0716893672943115,
      "learning_rate": 1.828611961366683e-07,
      "loss": 0.107,
      "step": 12777
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7403706312179565,
      "learning_rate": 1.8255013026800607e-07,
      "loss": 0.0442,
      "step": 12778
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.289683818817139,
      "learning_rate": 1.8223932428252933e-07,
      "loss": 0.5339,
      "step": 12779
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.484226703643799,
      "learning_rate": 1.8192877819700473e-07,
      "loss": 0.0956,
      "step": 12780
    },
    {
      "epoch": 0.92,
      "grad_norm": 26.18611717224121,
      "learning_rate": 1.8161849202818493e-07,
      "loss": 0.9715,
      "step": 12781
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.3955764770507812,
      "learning_rate": 1.813084657928099e-07,
      "loss": 0.1645,
      "step": 12782
    },
    {
      "epoch": 0.92,
      "grad_norm": 13.384496688842773,
      "learning_rate": 1.8099869950760295e-07,
      "loss": 0.617,
      "step": 12783
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.48822021484375,
      "learning_rate": 1.806891931892746e-07,
      "loss": 0.3626,
      "step": 12784
    },
    {
      "epoch": 0.92,
      "grad_norm": 31.47344970703125,
      "learning_rate": 1.8037994685452254e-07,
      "loss": 0.3473,
      "step": 12785
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3471195697784424,
      "learning_rate": 1.800709605200296e-07,
      "loss": 0.092,
      "step": 12786
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.1993274688720703,
      "learning_rate": 1.7976223420246297e-07,
      "loss": 0.1789,
      "step": 12787
    },
    {
      "epoch": 0.92,
      "grad_norm": 21.457067489624023,
      "learning_rate": 1.794537679184788e-07,
      "loss": 0.5731,
      "step": 12788
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.477878093719482,
      "learning_rate": 1.7914556168471709e-07,
      "loss": 0.7871,
      "step": 12789
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.575947761535645,
      "learning_rate": 1.7883761551780398e-07,
      "loss": 0.6372,
      "step": 12790
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.581319570541382,
      "learning_rate": 1.785299294343529e-07,
      "loss": 0.0934,
      "step": 12791
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.618710517883301,
      "learning_rate": 1.782225034509616e-07,
      "loss": 0.7237,
      "step": 12792
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3032162189483643,
      "learning_rate": 1.7791533758421463e-07,
      "loss": 0.0769,
      "step": 12793
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.275019645690918,
      "learning_rate": 1.7760843185068378e-07,
      "loss": 0.0628,
      "step": 12794
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.298391342163086,
      "learning_rate": 1.773017862669235e-07,
      "loss": 0.7465,
      "step": 12795
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9649007320404053,
      "learning_rate": 1.7699540084947674e-07,
      "loss": 0.0672,
      "step": 12796
    },
    {
      "epoch": 0.92,
      "grad_norm": 13.260149002075195,
      "learning_rate": 1.7668927561487193e-07,
      "loss": 0.6753,
      "step": 12797
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.09126353263855,
      "learning_rate": 1.7638341057962415e-07,
      "loss": 0.0432,
      "step": 12798
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.116616249084473,
      "learning_rate": 1.7607780576023193e-07,
      "loss": 0.7037,
      "step": 12799
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9776651859283447,
      "learning_rate": 1.757724611731837e-07,
      "loss": 0.8522,
      "step": 12800
    },
    {
      "epoch": 0.92,
      "grad_norm": 19.158849716186523,
      "learning_rate": 1.754673768349502e-07,
      "loss": 0.8737,
      "step": 12801
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.804460525512695,
      "learning_rate": 1.7516255276198934e-07,
      "loss": 0.4726,
      "step": 12802
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.19618821144104,
      "learning_rate": 1.748579889707458e-07,
      "loss": 0.086,
      "step": 12803
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9091362953186035,
      "learning_rate": 1.7455368547765028e-07,
      "loss": 0.0788,
      "step": 12804
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.150677442550659,
      "learning_rate": 1.74249642299118e-07,
      "loss": 0.1646,
      "step": 12805
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.291748285293579,
      "learning_rate": 1.7394585945155083e-07,
      "loss": 0.1533,
      "step": 12806
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9123833179473877,
      "learning_rate": 1.736423369513368e-07,
      "loss": 0.6892,
      "step": 12807
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.631887912750244,
      "learning_rate": 1.7333907481484947e-07,
      "loss": 0.6082,
      "step": 12808
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9917333126068115,
      "learning_rate": 1.7303607305845016e-07,
      "loss": 0.0539,
      "step": 12809
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.35100793838501,
      "learning_rate": 1.7273333169848306e-07,
      "loss": 0.3819,
      "step": 12810
    },
    {
      "epoch": 0.92,
      "grad_norm": 14.322869300842285,
      "learning_rate": 1.724308507512812e-07,
      "loss": 0.4151,
      "step": 12811
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1275224685668945,
      "learning_rate": 1.721286302331604e-07,
      "loss": 0.0842,
      "step": 12812
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.198307514190674,
      "learning_rate": 1.7182667016042655e-07,
      "loss": 0.7365,
      "step": 12813
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.637413501739502,
      "learning_rate": 1.7152497054936767e-07,
      "loss": 0.649,
      "step": 12814
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.9210054874420166,
      "learning_rate": 1.712235314162597e-07,
      "loss": 0.0647,
      "step": 12815
    },
    {
      "epoch": 0.92,
      "grad_norm": 46.27165985107422,
      "learning_rate": 1.7092235277736513e-07,
      "loss": 0.7555,
      "step": 12816
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.1729581356048584,
      "learning_rate": 1.7062143464892988e-07,
      "loss": 0.2236,
      "step": 12817
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.576704978942871,
      "learning_rate": 1.7032077704718875e-07,
      "loss": 0.7725,
      "step": 12818
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6147048473358154,
      "learning_rate": 1.7002037998835985e-07,
      "loss": 0.1053,
      "step": 12819
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.4413082599639893,
      "learning_rate": 1.6972024348864912e-07,
      "loss": 0.1391,
      "step": 12820
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9648520946502686,
      "learning_rate": 1.6942036756424695e-07,
      "loss": 0.0544,
      "step": 12821
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0905911922454834,
      "learning_rate": 1.6912075223133205e-07,
      "loss": 0.162,
      "step": 12822
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.7878355979919434,
      "learning_rate": 1.6882139750606652e-07,
      "loss": 0.0832,
      "step": 12823
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.734154462814331,
      "learning_rate": 1.6852230340459907e-07,
      "loss": 0.6338,
      "step": 12824
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.594605445861816,
      "learning_rate": 1.6822346994306572e-07,
      "loss": 0.3476,
      "step": 12825
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.536750793457031,
      "learning_rate": 1.679248971375874e-07,
      "loss": 0.3683,
      "step": 12826
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0678184032440186,
      "learning_rate": 1.676265850042691e-07,
      "loss": 0.0817,
      "step": 12827
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.074467420578003,
      "learning_rate": 1.673285335592062e-07,
      "loss": 0.1513,
      "step": 12828
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.111496925354004,
      "learning_rate": 1.6703074281847587e-07,
      "loss": 0.0858,
      "step": 12829
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.59211802482605,
      "learning_rate": 1.6673321279814357e-07,
      "loss": 0.1321,
      "step": 12830
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.051573753356934,
      "learning_rate": 1.664359435142593e-07,
      "loss": 0.777,
      "step": 12831
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.191254138946533,
      "learning_rate": 1.661389349828596e-07,
      "loss": 0.5243,
      "step": 12832
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.635666847229004,
      "learning_rate": 1.6584218721996726e-07,
      "loss": 0.1587,
      "step": 12833
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.326504945755005,
      "learning_rate": 1.6554570024159057e-07,
      "loss": 0.1757,
      "step": 12834
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.757550239562988,
      "learning_rate": 1.6524947406372448e-07,
      "loss": 0.2244,
      "step": 12835
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.938387870788574,
      "learning_rate": 1.6495350870234794e-07,
      "loss": 0.1518,
      "step": 12836
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6055538654327393,
      "learning_rate": 1.6465780417342814e-07,
      "loss": 0.0908,
      "step": 12837
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9888346195220947,
      "learning_rate": 1.643623604929173e-07,
      "loss": 0.8797,
      "step": 12838
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.452075481414795,
      "learning_rate": 1.6406717767675385e-07,
      "loss": 0.5353,
      "step": 12839
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.726178169250488,
      "learning_rate": 1.6377225574085998e-07,
      "loss": 0.3873,
      "step": 12840
    },
    {
      "epoch": 0.92,
      "grad_norm": 26.543716430664062,
      "learning_rate": 1.6347759470114744e-07,
      "loss": 0.8397,
      "step": 12841
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.224232196807861,
      "learning_rate": 1.631831945735113e-07,
      "loss": 0.6875,
      "step": 12842
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.434643030166626,
      "learning_rate": 1.6288905537383382e-07,
      "loss": 0.1166,
      "step": 12843
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.608961582183838,
      "learning_rate": 1.6259517711798178e-07,
      "loss": 0.1427,
      "step": 12844
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.308377981185913,
      "learning_rate": 1.6230155982180916e-07,
      "loss": 0.1317,
      "step": 12845
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.415524959564209,
      "learning_rate": 1.6200820350115665e-07,
      "loss": 0.1344,
      "step": 12846
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.5696022510528564,
      "learning_rate": 1.617151081718482e-07,
      "loss": 0.1401,
      "step": 12847
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.78400993347168,
      "learning_rate": 1.6142227384969566e-07,
      "loss": 0.4959,
      "step": 12848
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.596754312515259,
      "learning_rate": 1.6112970055049636e-07,
      "loss": 0.1337,
      "step": 12849
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.350309371948242,
      "learning_rate": 1.6083738829003383e-07,
      "loss": 0.3649,
      "step": 12850
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.003285884857178,
      "learning_rate": 1.6054533708407704e-07,
      "loss": 0.807,
      "step": 12851
    },
    {
      "epoch": 0.92,
      "grad_norm": 30.748937606811523,
      "learning_rate": 1.602535469483807e-07,
      "loss": 0.8435,
      "step": 12852
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1251699924468994,
      "learning_rate": 1.599620178986866e-07,
      "loss": 0.0841,
      "step": 12853
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.1460819244384766,
      "learning_rate": 1.5967074995072107e-07,
      "loss": 0.2351,
      "step": 12854
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.2183408737182617,
      "learning_rate": 1.5937974312019655e-07,
      "loss": 0.2024,
      "step": 12855
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.655822277069092,
      "learning_rate": 1.5908899742281213e-07,
      "loss": 0.1259,
      "step": 12856
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7564555406570435,
      "learning_rate": 1.587985128742531e-07,
      "loss": 0.0367,
      "step": 12857
    },
    {
      "epoch": 0.92,
      "grad_norm": 21.437742233276367,
      "learning_rate": 1.5850828949018794e-07,
      "loss": 0.6399,
      "step": 12858
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.205332040786743,
      "learning_rate": 1.5821832728627584e-07,
      "loss": 0.1338,
      "step": 12859
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9792676568031311,
      "learning_rate": 1.5792862627815763e-07,
      "loss": 0.0106,
      "step": 12860
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.559157133102417,
      "learning_rate": 1.5763918648146082e-07,
      "loss": 0.1474,
      "step": 12861
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.768516540527344,
      "learning_rate": 1.5735000791180176e-07,
      "loss": 0.1596,
      "step": 12862
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.912954807281494,
      "learning_rate": 1.5706109058477915e-07,
      "loss": 0.7138,
      "step": 12863
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.940822124481201,
      "learning_rate": 1.5677243451597823e-07,
      "loss": 0.4003,
      "step": 12864
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.271266460418701,
      "learning_rate": 1.564840397209727e-07,
      "loss": 0.9192,
      "step": 12865
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1112565994262695,
      "learning_rate": 1.5619590621531954e-07,
      "loss": 0.0801,
      "step": 12866
    },
    {
      "epoch": 0.92,
      "grad_norm": 14.391988754272461,
      "learning_rate": 1.5590803401456244e-07,
      "loss": 0.6848,
      "step": 12867
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.223499774932861,
      "learning_rate": 1.556204231342312e-07,
      "loss": 0.1236,
      "step": 12868
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.113768577575684,
      "learning_rate": 1.5533307358984117e-07,
      "loss": 0.8099,
      "step": 12869
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.9041647911071777,
      "learning_rate": 1.5504598539689274e-07,
      "loss": 0.1439,
      "step": 12870
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.169433116912842,
      "learning_rate": 1.547591585708752e-07,
      "loss": 0.7733,
      "step": 12871
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.656888484954834,
      "learning_rate": 1.544725931272606e-07,
      "loss": 0.2868,
      "step": 12872
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.739014625549316,
      "learning_rate": 1.5418628908150767e-07,
      "loss": 0.7211,
      "step": 12873
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.8792943954467773,
      "learning_rate": 1.5390024644906297e-07,
      "loss": 0.1468,
      "step": 12874
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.1328301429748535,
      "learning_rate": 1.536144652453564e-07,
      "loss": 0.2131,
      "step": 12875
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.028299331665039,
      "learning_rate": 1.5332894548580447e-07,
      "loss": 0.1644,
      "step": 12876
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.021205902099609,
      "learning_rate": 1.5304368718580987e-07,
      "loss": 0.8715,
      "step": 12877
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5364307165145874,
      "learning_rate": 1.5275869036076251e-07,
      "loss": 0.0778,
      "step": 12878
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.049877643585205,
      "learning_rate": 1.524739550260357e-07,
      "loss": 0.2459,
      "step": 12879
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.3344292640686035,
      "learning_rate": 1.521894811969904e-07,
      "loss": 0.7385,
      "step": 12880
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.209510564804077,
      "learning_rate": 1.5190526888897217e-07,
      "loss": 0.1615,
      "step": 12881
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6010630130767822,
      "learning_rate": 1.5162131811731374e-07,
      "loss": 0.1733,
      "step": 12882
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1342484951019287,
      "learning_rate": 1.513376288973334e-07,
      "loss": 0.0654,
      "step": 12883
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.3361947536468506,
      "learning_rate": 1.5105420124433445e-07,
      "loss": 0.1891,
      "step": 12884
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0019896030426025,
      "learning_rate": 1.5077103517360746e-07,
      "loss": 0.2114,
      "step": 12885
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0255391597747803,
      "learning_rate": 1.5048813070042688e-07,
      "loss": 0.1546,
      "step": 12886
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.396771430969238,
      "learning_rate": 1.5020548784005608e-07,
      "loss": 0.4529,
      "step": 12887
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.280394554138184,
      "learning_rate": 1.4992310660774167e-07,
      "loss": 0.328,
      "step": 12888
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.089386224746704,
      "learning_rate": 1.4964098701871653e-07,
      "loss": 0.055,
      "step": 12889
    },
    {
      "epoch": 0.92,
      "grad_norm": 22.210254669189453,
      "learning_rate": 1.4935912908820128e-07,
      "loss": 0.6427,
      "step": 12890
    },
    {
      "epoch": 0.92,
      "grad_norm": 15.335257530212402,
      "learning_rate": 1.490775328314009e-07,
      "loss": 1.0899,
      "step": 12891
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.295315742492676,
      "learning_rate": 1.4879619826350444e-07,
      "loss": 0.9195,
      "step": 12892
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.19699478149414,
      "learning_rate": 1.4851512539969138e-07,
      "loss": 0.4007,
      "step": 12893
    },
    {
      "epoch": 0.92,
      "grad_norm": 13.391433715820312,
      "learning_rate": 1.482343142551229e-07,
      "loss": 0.8751,
      "step": 12894
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.343207836151123,
      "learning_rate": 1.4795376484494805e-07,
      "loss": 0.0919,
      "step": 12895
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.945608139038086,
      "learning_rate": 1.4767347718430191e-07,
      "loss": 0.4898,
      "step": 12896
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.277078628540039,
      "learning_rate": 1.4739345128830463e-07,
      "loss": 0.0781,
      "step": 12897
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.320361137390137,
      "learning_rate": 1.4711368717206186e-07,
      "loss": 0.5896,
      "step": 12898
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.127122402191162,
      "learning_rate": 1.4683418485066715e-07,
      "loss": 0.4627,
      "step": 12899
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.703243255615234,
      "learning_rate": 1.4655494433919838e-07,
      "loss": 0.5751,
      "step": 12900
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.403802394866943,
      "learning_rate": 1.4627596565271796e-07,
      "loss": 0.4334,
      "step": 12901
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.83001971244812,
      "learning_rate": 1.4599724880627775e-07,
      "loss": 0.0821,
      "step": 12902
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.15604019165039,
      "learning_rate": 1.4571879381491293e-07,
      "loss": 0.4998,
      "step": 12903
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.5080184936523438,
      "learning_rate": 1.4544060069364374e-07,
      "loss": 0.1492,
      "step": 12904
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.048548936843872,
      "learning_rate": 1.451626694574787e-07,
      "loss": 0.0525,
      "step": 12905
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4312262535095215,
      "learning_rate": 1.4488500012141195e-07,
      "loss": 0.1039,
      "step": 12906
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.72113561630249,
      "learning_rate": 1.4460759270042035e-07,
      "loss": 0.8343,
      "step": 12907
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3582966327667236,
      "learning_rate": 1.44330447209472e-07,
      "loss": 0.1314,
      "step": 12908
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.9140706062316895,
      "learning_rate": 1.4405356366351543e-07,
      "loss": 0.14,
      "step": 12909
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.622995853424072,
      "learning_rate": 1.4377694207748817e-07,
      "loss": 0.287,
      "step": 12910
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.7440505027771,
      "learning_rate": 1.4350058246631382e-07,
      "loss": 0.7078,
      "step": 12911
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2740938663482666,
      "learning_rate": 1.4322448484489993e-07,
      "loss": 0.0686,
      "step": 12912
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.427385330200195,
      "learning_rate": 1.4294864922814122e-07,
      "loss": 0.7883,
      "step": 12913
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.185608386993408,
      "learning_rate": 1.4267307563091747e-07,
      "loss": 0.0511,
      "step": 12914
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.606632709503174,
      "learning_rate": 1.423977640680968e-07,
      "loss": 0.8492,
      "step": 12915
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.792485237121582,
      "learning_rate": 1.4212271455452786e-07,
      "loss": 0.581,
      "step": 12916
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.7049849033355713,
      "learning_rate": 1.4184792710505158e-07,
      "loss": 0.1763,
      "step": 12917
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4444752931594849,
      "learning_rate": 1.4157340173448997e-07,
      "loss": 0.0372,
      "step": 12918
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7752832174301147,
      "learning_rate": 1.4129913845765396e-07,
      "loss": 0.0327,
      "step": 12919
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.979910373687744,
      "learning_rate": 1.4102513728933677e-07,
      "loss": 0.1993,
      "step": 12920
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.892622947692871,
      "learning_rate": 1.407513982443226e-07,
      "loss": 0.6101,
      "step": 12921
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6465721130371094,
      "learning_rate": 1.4047792133737693e-07,
      "loss": 0.1077,
      "step": 12922
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.699289321899414,
      "learning_rate": 1.4020470658325235e-07,
      "loss": 0.9145,
      "step": 12923
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.933144569396973,
      "learning_rate": 1.3993175399668934e-07,
      "loss": 0.2983,
      "step": 12924
    },
    {
      "epoch": 0.93,
      "grad_norm": 24.40021324157715,
      "learning_rate": 1.3965906359241165e-07,
      "loss": 0.702,
      "step": 12925
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.558142900466919,
      "learning_rate": 1.3938663538512976e-07,
      "loss": 0.1809,
      "step": 12926
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.668822288513184,
      "learning_rate": 1.391144693895413e-07,
      "loss": 0.2003,
      "step": 12927
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.9159340858459473,
      "learning_rate": 1.388425656203274e-07,
      "loss": 0.0821,
      "step": 12928
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.088735818862915,
      "learning_rate": 1.3857092409215567e-07,
      "loss": 0.0519,
      "step": 12929
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.705566167831421,
      "learning_rate": 1.3829954481968223e-07,
      "loss": 0.1961,
      "step": 12930
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.947824716567993,
      "learning_rate": 1.3802842781754533e-07,
      "loss": 0.1159,
      "step": 12931
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.625913143157959,
      "learning_rate": 1.3775757310037107e-07,
      "loss": 0.5929,
      "step": 12932
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.665903091430664,
      "learning_rate": 1.3748698068277166e-07,
      "loss": 0.1105,
      "step": 12933
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4868760108947754,
      "learning_rate": 1.3721665057934375e-07,
      "loss": 0.0861,
      "step": 12934
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.633039951324463,
      "learning_rate": 1.369465828046701e-07,
      "loss": 0.6899,
      "step": 12935
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.803468704223633,
      "learning_rate": 1.3667677737332187e-07,
      "loss": 0.6308,
      "step": 12936
    },
    {
      "epoch": 0.93,
      "grad_norm": 18.536237716674805,
      "learning_rate": 1.3640723429985238e-07,
      "loss": 0.5259,
      "step": 12937
    },
    {
      "epoch": 0.93,
      "grad_norm": 40.06260681152344,
      "learning_rate": 1.3613795359880278e-07,
      "loss": 0.9707,
      "step": 12938
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0927793979644775,
      "learning_rate": 1.3586893528470036e-07,
      "loss": 0.108,
      "step": 12939
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.20261812210083,
      "learning_rate": 1.3560017937205682e-07,
      "loss": 0.0701,
      "step": 12940
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.04629898071289,
      "learning_rate": 1.353316858753706e-07,
      "loss": 0.5266,
      "step": 12941
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5685555934906006,
      "learning_rate": 1.350634548091262e-07,
      "loss": 0.1579,
      "step": 12942
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.0507190227508545,
      "learning_rate": 1.3479548618779425e-07,
      "loss": 0.0789,
      "step": 12943
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8390347957611084,
      "learning_rate": 1.3452778002582934e-07,
      "loss": 0.0534,
      "step": 12944
    },
    {
      "epoch": 0.93,
      "grad_norm": 17.439136505126953,
      "learning_rate": 1.3426033633767377e-07,
      "loss": 0.5156,
      "step": 12945
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.652090311050415,
      "learning_rate": 1.3399315513775602e-07,
      "loss": 0.1309,
      "step": 12946
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.865962982177734,
      "learning_rate": 1.3372623644048844e-07,
      "loss": 0.5016,
      "step": 12947
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.41171932220459,
      "learning_rate": 1.3345958026027005e-07,
      "loss": 0.7308,
      "step": 12948
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4895286560058594,
      "learning_rate": 1.3319318661148662e-07,
      "loss": 0.1317,
      "step": 12949
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2733089923858643,
      "learning_rate": 1.3292705550850937e-07,
      "loss": 0.0906,
      "step": 12950
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.957812547683716,
      "learning_rate": 1.3266118696569407e-07,
      "loss": 0.1331,
      "step": 12951
    },
    {
      "epoch": 0.93,
      "grad_norm": 13.412612915039062,
      "learning_rate": 1.3239558099738426e-07,
      "loss": 0.3865,
      "step": 12952
    },
    {
      "epoch": 0.93,
      "grad_norm": 14.540708541870117,
      "learning_rate": 1.3213023761790788e-07,
      "loss": 0.6252,
      "step": 12953
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3161286115646362,
      "learning_rate": 1.3186515684157907e-07,
      "loss": 0.0283,
      "step": 12954
    },
    {
      "epoch": 0.93,
      "grad_norm": 13.557589530944824,
      "learning_rate": 1.3160033868269863e-07,
      "loss": 0.7078,
      "step": 12955
    },
    {
      "epoch": 0.93,
      "grad_norm": 15.883333206176758,
      "learning_rate": 1.3133578315555174e-07,
      "loss": 0.9025,
      "step": 12956
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.087353229522705,
      "learning_rate": 1.310714902744098e-07,
      "loss": 0.9002,
      "step": 12957
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.6972527503967285,
      "learning_rate": 1.3080746005353194e-07,
      "loss": 0.1374,
      "step": 12958
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.573799133300781,
      "learning_rate": 1.3054369250716071e-07,
      "loss": 0.3914,
      "step": 12959
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.2320241928100586,
      "learning_rate": 1.3028018764952465e-07,
      "loss": 0.2358,
      "step": 12960
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9705901145935059,
      "learning_rate": 1.3001694549484023e-07,
      "loss": 0.0582,
      "step": 12961
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.951171398162842,
      "learning_rate": 1.2975396605730773e-07,
      "loss": 0.1996,
      "step": 12962
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.342470169067383,
      "learning_rate": 1.2949124935111356e-07,
      "loss": 0.0814,
      "step": 12963
    },
    {
      "epoch": 0.93,
      "grad_norm": 13.947547912597656,
      "learning_rate": 1.2922879539043086e-07,
      "loss": 0.5271,
      "step": 12964
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.241338729858398,
      "learning_rate": 1.2896660418941775e-07,
      "loss": 0.3407,
      "step": 12965
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.06025505065918,
      "learning_rate": 1.287046757622179e-07,
      "loss": 0.6626,
      "step": 12966
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.372912883758545,
      "learning_rate": 1.2844301012296278e-07,
      "loss": 0.3481,
      "step": 12967
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.8454430103302,
      "learning_rate": 1.2818160728576722e-07,
      "loss": 0.156,
      "step": 12968
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.436758041381836,
      "learning_rate": 1.2792046726473218e-07,
      "loss": 0.2162,
      "step": 12969
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.481576919555664,
      "learning_rate": 1.2765959007394692e-07,
      "loss": 0.8355,
      "step": 12970
    },
    {
      "epoch": 0.93,
      "grad_norm": 20.780529022216797,
      "learning_rate": 1.2739897572748404e-07,
      "loss": 0.5142,
      "step": 12971
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.289109468460083,
      "learning_rate": 1.2713862423940237e-07,
      "loss": 0.1184,
      "step": 12972
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.154112815856934,
      "learning_rate": 1.268785356237473e-07,
      "loss": 0.6857,
      "step": 12973
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.450296401977539,
      "learning_rate": 1.2661870989454983e-07,
      "loss": 0.7313,
      "step": 12974
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.377169609069824,
      "learning_rate": 1.2635914706582598e-07,
      "loss": 0.3118,
      "step": 12975
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.756705284118652,
      "learning_rate": 1.2609984715157842e-07,
      "loss": 0.5789,
      "step": 12976
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.283008337020874,
      "learning_rate": 1.2584081016579486e-07,
      "loss": 0.0722,
      "step": 12977
    },
    {
      "epoch": 0.93,
      "grad_norm": 34.808406829833984,
      "learning_rate": 1.2558203612245022e-07,
      "loss": 0.8693,
      "step": 12978
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.1702721118927,
      "learning_rate": 1.2532352503550393e-07,
      "loss": 0.2771,
      "step": 12979
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.908658504486084,
      "learning_rate": 1.250652769189026e-07,
      "loss": 0.1332,
      "step": 12980
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.342245101928711,
      "learning_rate": 1.2480729178657558e-07,
      "loss": 0.0712,
      "step": 12981
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.038143157958984,
      "learning_rate": 1.2454956965244236e-07,
      "loss": 0.5801,
      "step": 12982
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.210253715515137,
      "learning_rate": 1.242921105304057e-07,
      "loss": 0.5594,
      "step": 12983
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.05456280708313,
      "learning_rate": 1.2403491443435332e-07,
      "loss": 0.1417,
      "step": 12984
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.982702255249023,
      "learning_rate": 1.2377798137816087e-07,
      "loss": 0.2852,
      "step": 12985
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.048628807067871,
      "learning_rate": 1.2352131137568945e-07,
      "loss": 0.0734,
      "step": 12986
    },
    {
      "epoch": 0.93,
      "grad_norm": 12.81161880493164,
      "learning_rate": 1.2326490444078465e-07,
      "loss": 0.6614,
      "step": 12987
    },
    {
      "epoch": 0.93,
      "grad_norm": 19.339279174804688,
      "learning_rate": 1.2300876058727874e-07,
      "loss": 1.3645,
      "step": 12988
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7399803400039673,
      "learning_rate": 1.2275287982898953e-07,
      "loss": 0.0654,
      "step": 12989
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.32582426071167,
      "learning_rate": 1.2249726217972047e-07,
      "loss": 0.1248,
      "step": 12990
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.529010772705078,
      "learning_rate": 1.2224190765326272e-07,
      "loss": 0.0996,
      "step": 12991
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.29383659362793,
      "learning_rate": 1.219868162633908e-07,
      "loss": 0.5018,
      "step": 12992
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.068143367767334,
      "learning_rate": 1.2173198802386542e-07,
      "loss": 0.1231,
      "step": 12993
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.41317081451416,
      "learning_rate": 1.214774229484339e-07,
      "loss": 0.3886,
      "step": 12994
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.220174551010132,
      "learning_rate": 1.2122312105082967e-07,
      "loss": 0.0664,
      "step": 12995
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.997152328491211,
      "learning_rate": 1.2096908234477066e-07,
      "loss": 0.6718,
      "step": 12996
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.318728446960449,
      "learning_rate": 1.2071530684396148e-07,
      "loss": 0.6873,
      "step": 12997
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.026515007019043,
      "learning_rate": 1.2046179456209229e-07,
      "loss": 0.7878,
      "step": 12998
    },
    {
      "epoch": 0.93,
      "grad_norm": 28.543994903564453,
      "learning_rate": 1.202085455128399e-07,
      "loss": 0.6149,
      "step": 12999
    },
    {
      "epoch": 0.93,
      "grad_norm": 35.575992584228516,
      "learning_rate": 1.199555597098645e-07,
      "loss": 0.7006,
      "step": 13000
    },
    {
      "epoch": 0.93,
      "grad_norm": 22.730215072631836,
      "learning_rate": 1.1970283716681519e-07,
      "loss": 0.5355,
      "step": 13001
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.85345721244812,
      "learning_rate": 1.194503778973255e-07,
      "loss": 0.1266,
      "step": 13002
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.641480445861816,
      "learning_rate": 1.1919818191501286e-07,
      "loss": 0.597,
      "step": 13003
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.481074810028076,
      "learning_rate": 1.1894624923348418e-07,
      "loss": 0.0841,
      "step": 13004
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.887509822845459,
      "learning_rate": 1.1869457986632915e-07,
      "loss": 0.1022,
      "step": 13005
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.019429922103882,
      "learning_rate": 1.1844317382712467e-07,
      "loss": 0.1059,
      "step": 13006
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.620206832885742,
      "learning_rate": 1.1819203112943378e-07,
      "loss": 0.4464,
      "step": 13007
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.704540252685547,
      "learning_rate": 1.1794115178680398e-07,
      "loss": 0.8548,
      "step": 13008
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.134341239929199,
      "learning_rate": 1.1769053581276946e-07,
      "loss": 0.2111,
      "step": 13009
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.823184967041016,
      "learning_rate": 1.1744018322084993e-07,
      "loss": 0.854,
      "step": 13010
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.8310606479644775,
      "learning_rate": 1.1719009402455183e-07,
      "loss": 0.1096,
      "step": 13011
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.8120956420898438,
      "learning_rate": 1.1694026823736438e-07,
      "loss": 0.1934,
      "step": 13012
    },
    {
      "epoch": 0.93,
      "grad_norm": 25.782556533813477,
      "learning_rate": 1.1669070587276676e-07,
      "loss": 1.0536,
      "step": 13013
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.4017333984375,
      "learning_rate": 1.1644140694422101e-07,
      "loss": 0.2146,
      "step": 13014
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.015871524810791,
      "learning_rate": 1.1619237146517637e-07,
      "loss": 0.7972,
      "step": 13015
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.266488552093506,
      "learning_rate": 1.1594359944906652e-07,
      "loss": 0.0935,
      "step": 13016
    },
    {
      "epoch": 0.93,
      "grad_norm": 13.275615692138672,
      "learning_rate": 1.1569509090931296e-07,
      "loss": 0.75,
      "step": 13017
    },
    {
      "epoch": 0.93,
      "grad_norm": 13.540071487426758,
      "learning_rate": 1.1544684585932053e-07,
      "loss": 0.4388,
      "step": 13018
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.865823745727539,
      "learning_rate": 1.151988643124824e-07,
      "loss": 0.5581,
      "step": 13019
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0200307369232178,
      "learning_rate": 1.1495114628217507e-07,
      "loss": 0.0784,
      "step": 13020
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9860665798187256,
      "learning_rate": 1.1470369178176289e-07,
      "loss": 0.0735,
      "step": 13021
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4648470878601074,
      "learning_rate": 1.1445650082459403e-07,
      "loss": 0.1044,
      "step": 13022
    },
    {
      "epoch": 0.93,
      "grad_norm": 14.574954986572266,
      "learning_rate": 1.1420957342400507e-07,
      "loss": 0.562,
      "step": 13023
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6980698108673096,
      "learning_rate": 1.1396290959331535e-07,
      "loss": 0.6767,
      "step": 13024
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9037754535675049,
      "learning_rate": 1.13716509345832e-07,
      "loss": 0.0417,
      "step": 13025
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.175448417663574,
      "learning_rate": 1.1347037269484772e-07,
      "loss": 0.0822,
      "step": 13026
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.6018521785736084,
      "learning_rate": 1.1322449965364024e-07,
      "loss": 0.1688,
      "step": 13027
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.90427303314209,
      "learning_rate": 1.1297889023547281e-07,
      "loss": 0.7974,
      "step": 13028
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.901002883911133,
      "learning_rate": 1.1273354445359652e-07,
      "loss": 0.5772,
      "step": 13029
    },
    {
      "epoch": 0.93,
      "grad_norm": 15.100754737854004,
      "learning_rate": 1.1248846232124633e-07,
      "loss": 0.74,
      "step": 13030
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.506491661071777,
      "learning_rate": 1.1224364385164276e-07,
      "loss": 0.4034,
      "step": 13031
    },
    {
      "epoch": 0.93,
      "grad_norm": 19.160776138305664,
      "learning_rate": 1.1199908905799418e-07,
      "loss": 0.6904,
      "step": 13032
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4793739318847656,
      "learning_rate": 1.117547979534922e-07,
      "loss": 0.1437,
      "step": 13033
    },
    {
      "epoch": 0.93,
      "grad_norm": 24.788036346435547,
      "learning_rate": 1.1151077055131576e-07,
      "loss": 0.3302,
      "step": 13034
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.6022682189941406,
      "learning_rate": 1.1126700686462988e-07,
      "loss": 0.1265,
      "step": 13035
    },
    {
      "epoch": 0.93,
      "grad_norm": 15.017518043518066,
      "learning_rate": 1.1102350690658403e-07,
      "loss": 0.7199,
      "step": 13036
    },
    {
      "epoch": 0.93,
      "grad_norm": 14.185675621032715,
      "learning_rate": 1.1078027069031383e-07,
      "loss": 0.4641,
      "step": 13037
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.440200090408325,
      "learning_rate": 1.1053729822894155e-07,
      "loss": 0.1162,
      "step": 13038
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2982583045959473,
      "learning_rate": 1.1029458953557449e-07,
      "loss": 0.078,
      "step": 13039
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.029545783996582,
      "learning_rate": 1.1005214462330494e-07,
      "loss": 0.4906,
      "step": 13040
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4425368309020996,
      "learning_rate": 1.0980996350521356e-07,
      "loss": 0.0725,
      "step": 13041
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.62953782081604,
      "learning_rate": 1.0956804619436434e-07,
      "loss": 0.1524,
      "step": 13042
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.9240782260894775,
      "learning_rate": 1.0932639270380685e-07,
      "loss": 0.0715,
      "step": 13043
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.835528373718262,
      "learning_rate": 1.0908500304657899e-07,
      "loss": 0.8703,
      "step": 13044
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.9533803462982178,
      "learning_rate": 1.08843877235702e-07,
      "loss": 0.2284,
      "step": 13045
    },
    {
      "epoch": 0.94,
      "grad_norm": 17.00788116455078,
      "learning_rate": 1.0860301528418383e-07,
      "loss": 1.1566,
      "step": 13046
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.303116798400879,
      "learning_rate": 1.0836241720501849e-07,
      "loss": 0.826,
      "step": 13047
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.5351691246032715,
      "learning_rate": 1.0812208301118454e-07,
      "loss": 0.19,
      "step": 13048
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.88619327545166,
      "learning_rate": 1.0788201271564713e-07,
      "loss": 0.3569,
      "step": 13049
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.021333694458008,
      "learning_rate": 1.0764220633135758e-07,
      "loss": 0.5897,
      "step": 13050
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8046648502349854,
      "learning_rate": 1.0740266387125275e-07,
      "loss": 0.1172,
      "step": 13051
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.23323917388916,
      "learning_rate": 1.07163385348254e-07,
      "loss": 0.696,
      "step": 13052
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.232234477996826,
      "learning_rate": 1.0692437077527096e-07,
      "loss": 0.1471,
      "step": 13053
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.757112503051758,
      "learning_rate": 1.0668562016519668e-07,
      "loss": 0.6233,
      "step": 13054
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.850725173950195,
      "learning_rate": 1.0644713353091085e-07,
      "loss": 0.917,
      "step": 13055
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.127445697784424,
      "learning_rate": 1.0620891088527873e-07,
      "loss": 0.0565,
      "step": 13056
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6569701433181763,
      "learning_rate": 1.0597095224115228e-07,
      "loss": 0.0206,
      "step": 13057
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.4068009853363037,
      "learning_rate": 1.0573325761136788e-07,
      "loss": 0.183,
      "step": 13058
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.96481466293335,
      "learning_rate": 1.0549582700874861e-07,
      "loss": 0.5149,
      "step": 13059
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.160403728485107,
      "learning_rate": 1.0525866044610311e-07,
      "loss": 0.7275,
      "step": 13060
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.909512996673584,
      "learning_rate": 1.0502175793622449e-07,
      "loss": 0.141,
      "step": 13061
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.009239196777344,
      "learning_rate": 1.0478511949189362e-07,
      "loss": 0.9735,
      "step": 13062
    },
    {
      "epoch": 0.94,
      "grad_norm": 17.800580978393555,
      "learning_rate": 1.0454874512587698e-07,
      "loss": 1.6439,
      "step": 13063
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.6132283210754395,
      "learning_rate": 1.0431263485092436e-07,
      "loss": 0.8403,
      "step": 13064
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.493630886077881,
      "learning_rate": 1.0407678867977389e-07,
      "loss": 0.6998,
      "step": 13065
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.5474653244018555,
      "learning_rate": 1.038412066251493e-07,
      "loss": 0.1166,
      "step": 13066
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.6726951599121094,
      "learning_rate": 1.0360588869975819e-07,
      "loss": 0.6682,
      "step": 13067
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.1274502277374268,
      "learning_rate": 1.033708349162954e-07,
      "loss": 0.1514,
      "step": 13068
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.35213851928711,
      "learning_rate": 1.0313604528744192e-07,
      "loss": 0.7947,
      "step": 13069
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.9848411083221436,
      "learning_rate": 1.0290151982586261e-07,
      "loss": 0.133,
      "step": 13070
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1642262935638428,
      "learning_rate": 1.0266725854421012e-07,
      "loss": 0.0347,
      "step": 13071
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.826761245727539,
      "learning_rate": 1.0243326145512156e-07,
      "loss": 0.688,
      "step": 13072
    },
    {
      "epoch": 0.94,
      "grad_norm": 19.9191951751709,
      "learning_rate": 1.0219952857122017e-07,
      "loss": 0.4601,
      "step": 13073
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3232946395874023,
      "learning_rate": 1.0196605990511477e-07,
      "loss": 0.0894,
      "step": 13074
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.92310094833374,
      "learning_rate": 1.0173285546940081e-07,
      "loss": 0.2842,
      "step": 13075
    },
    {
      "epoch": 0.94,
      "grad_norm": 15.801265716552734,
      "learning_rate": 1.0149991527665826e-07,
      "loss": 1.1118,
      "step": 13076
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4642691612243652,
      "learning_rate": 1.0126723933945315e-07,
      "loss": 0.0891,
      "step": 13077
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8494834899902344,
      "learning_rate": 1.0103482767033768e-07,
      "loss": 0.1515,
      "step": 13078
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.404594421386719,
      "learning_rate": 1.0080268028185014e-07,
      "loss": 0.3632,
      "step": 13079
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.1165103912353516,
      "learning_rate": 1.005707971865133e-07,
      "loss": 0.091,
      "step": 13080
    },
    {
      "epoch": 0.94,
      "grad_norm": 13.736527442932129,
      "learning_rate": 1.003391783968366e-07,
      "loss": 1.0666,
      "step": 13081
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.496191024780273,
      "learning_rate": 1.0010782392531504e-07,
      "loss": 0.5075,
      "step": 13082
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.072310209274292,
      "learning_rate": 9.987673378442919e-08,
      "loss": 0.071,
      "step": 13083
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.90114164352417,
      "learning_rate": 9.96459079866463e-08,
      "loss": 0.0614,
      "step": 13084
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.220375061035156,
      "learning_rate": 9.941534654441699e-08,
      "loss": 0.4651,
      "step": 13085
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.657890796661377,
      "learning_rate": 9.918504947018014e-08,
      "loss": 0.1014,
      "step": 13086
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.69984769821167,
      "learning_rate": 9.895501677635921e-08,
      "loss": 0.2063,
      "step": 13087
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3908255100250244,
      "learning_rate": 9.872524847536424e-08,
      "loss": 0.0904,
      "step": 13088
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.199155807495117,
      "learning_rate": 9.849574457958866e-08,
      "loss": 0.1614,
      "step": 13089
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.1377103328704834,
      "learning_rate": 9.826650510141533e-08,
      "loss": 0.1518,
      "step": 13090
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.099535942077637,
      "learning_rate": 9.803753005320993e-08,
      "loss": 0.7795,
      "step": 13091
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.5784170627593994,
      "learning_rate": 9.780881944732479e-08,
      "loss": 0.2601,
      "step": 13092
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.5550785064697266,
      "learning_rate": 9.758037329609782e-08,
      "loss": 0.1036,
      "step": 13093
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.321197271347046,
      "learning_rate": 9.73521916118536e-08,
      "loss": 0.0892,
      "step": 13094
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.728045701980591,
      "learning_rate": 9.712427440690064e-08,
      "loss": 0.1465,
      "step": 13095
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.21798038482666,
      "learning_rate": 9.689662169353464e-08,
      "loss": 0.8062,
      "step": 13096
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.90134048461914,
      "learning_rate": 9.66692334840369e-08,
      "loss": 0.4885,
      "step": 13097
    },
    {
      "epoch": 0.94,
      "grad_norm": 27.00072479248047,
      "learning_rate": 9.64421097906737e-08,
      "loss": 0.7371,
      "step": 13098
    },
    {
      "epoch": 0.94,
      "grad_norm": 22.03737449645996,
      "learning_rate": 9.621525062569748e-08,
      "loss": 0.7437,
      "step": 13099
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.73809814453125,
      "learning_rate": 9.598865600134732e-08,
      "loss": 0.4508,
      "step": 13100
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.519134521484375,
      "learning_rate": 9.576232592984624e-08,
      "loss": 0.6542,
      "step": 13101
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.9748172760009766,
      "learning_rate": 9.553626042340336e-08,
      "loss": 0.6412,
      "step": 13102
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.023220062255859,
      "learning_rate": 9.531045949421558e-08,
      "loss": 0.8766,
      "step": 13103
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3879315853118896,
      "learning_rate": 9.508492315446372e-08,
      "loss": 0.0939,
      "step": 13104
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.647602558135986,
      "learning_rate": 9.485965141631304e-08,
      "loss": 0.7261,
      "step": 13105
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.557012557983398,
      "learning_rate": 9.463464429191771e-08,
      "loss": 0.6174,
      "step": 13106
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4553468227386475,
      "learning_rate": 9.440990179341581e-08,
      "loss": 0.113,
      "step": 13107
    },
    {
      "epoch": 0.94,
      "grad_norm": 15.878726959228516,
      "learning_rate": 9.418542393293095e-08,
      "loss": 1.0673,
      "step": 13108
    },
    {
      "epoch": 0.94,
      "grad_norm": 19.029571533203125,
      "learning_rate": 9.396121072257236e-08,
      "loss": 0.8463,
      "step": 13109
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.360919952392578,
      "learning_rate": 9.373726217443701e-08,
      "loss": 0.4393,
      "step": 13110
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.995266914367676,
      "learning_rate": 9.351357830060415e-08,
      "loss": 0.173,
      "step": 13111
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.128240585327148,
      "learning_rate": 9.329015911314243e-08,
      "loss": 0.2884,
      "step": 13112
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.289731979370117,
      "learning_rate": 9.306700462410334e-08,
      "loss": 0.9597,
      "step": 13113
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.226785659790039,
      "learning_rate": 9.284411484552558e-08,
      "loss": 0.0965,
      "step": 13114
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.068336009979248,
      "learning_rate": 9.262148978943342e-08,
      "loss": 0.0758,
      "step": 13115
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.889547109603882,
      "learning_rate": 9.239912946783613e-08,
      "loss": 0.1102,
      "step": 13116
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.169978141784668,
      "learning_rate": 9.217703389272969e-08,
      "loss": 0.4124,
      "step": 13117
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8654425144195557,
      "learning_rate": 9.195520307609507e-08,
      "loss": 0.0899,
      "step": 13118
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.1361894607543945,
      "learning_rate": 9.17336370298999e-08,
      "loss": 0.6042,
      "step": 13119
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.6458351612091064,
      "learning_rate": 9.151233576609575e-08,
      "loss": 0.1452,
      "step": 13120
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.3438658714294434,
      "learning_rate": 9.129129929662139e-08,
      "loss": 0.1043,
      "step": 13121
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.823746681213379,
      "learning_rate": 9.107052763340118e-08,
      "loss": 0.6397,
      "step": 13122
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.883572816848755,
      "learning_rate": 9.085002078834449e-08,
      "loss": 0.7372,
      "step": 13123
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.715028285980225,
      "learning_rate": 9.062977877334677e-08,
      "loss": 0.4959,
      "step": 13124
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.792936086654663,
      "learning_rate": 9.040980160029023e-08,
      "loss": 0.0443,
      "step": 13125
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.033125877380371,
      "learning_rate": 9.019008928104034e-08,
      "loss": 0.632,
      "step": 13126
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.6958417892456055,
      "learning_rate": 8.997064182745096e-08,
      "loss": 0.8287,
      "step": 13127
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0152041912078857,
      "learning_rate": 8.975145925135986e-08,
      "loss": 0.0736,
      "step": 13128
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.023953914642334,
      "learning_rate": 8.953254156459145e-08,
      "loss": 0.0907,
      "step": 13129
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1590800285339355,
      "learning_rate": 8.931388877895464e-08,
      "loss": 0.0772,
      "step": 13130
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.836409568786621,
      "learning_rate": 8.909550090624607e-08,
      "loss": 0.4232,
      "step": 13131
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2835893630981445,
      "learning_rate": 8.887737795824692e-08,
      "loss": 0.0713,
      "step": 13132
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.218532085418701,
      "learning_rate": 8.865951994672273e-08,
      "loss": 0.1286,
      "step": 13133
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.275142192840576,
      "learning_rate": 8.844192688342745e-08,
      "loss": 0.0714,
      "step": 13134
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.684027910232544,
      "learning_rate": 8.822459878009893e-08,
      "loss": 0.7349,
      "step": 13135
    },
    {
      "epoch": 0.94,
      "grad_norm": 13.794271469116211,
      "learning_rate": 8.800753564846109e-08,
      "loss": 0.5143,
      "step": 13136
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.219805717468262,
      "learning_rate": 8.779073750022405e-08,
      "loss": 0.6588,
      "step": 13137
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.185880661010742,
      "learning_rate": 8.757420434708285e-08,
      "loss": 0.1742,
      "step": 13138
    },
    {
      "epoch": 0.94,
      "grad_norm": 12.566263198852539,
      "learning_rate": 8.735793620071875e-08,
      "loss": 0.6165,
      "step": 13139
    },
    {
      "epoch": 0.94,
      "grad_norm": 25.174131393432617,
      "learning_rate": 8.714193307279851e-08,
      "loss": 0.6865,
      "step": 13140
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.772921562194824,
      "learning_rate": 8.692619497497501e-08,
      "loss": 0.4418,
      "step": 13141
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8158948421478271,
      "learning_rate": 8.671072191888619e-08,
      "loss": 0.0525,
      "step": 13142
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.423324108123779,
      "learning_rate": 8.649551391615607e-08,
      "loss": 0.8129,
      "step": 13143
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.831901788711548,
      "learning_rate": 8.628057097839482e-08,
      "loss": 0.1691,
      "step": 13144
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.242443084716797,
      "learning_rate": 8.606589311719705e-08,
      "loss": 0.3637,
      "step": 13145
    },
    {
      "epoch": 0.94,
      "grad_norm": 14.57268238067627,
      "learning_rate": 8.585148034414403e-08,
      "loss": 0.4914,
      "step": 13146
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.50260591506958,
      "learning_rate": 8.563733267080265e-08,
      "loss": 0.1236,
      "step": 13147
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.687458038330078,
      "learning_rate": 8.542345010872477e-08,
      "loss": 0.5665,
      "step": 13148
    },
    {
      "epoch": 0.94,
      "grad_norm": 17.757776260375977,
      "learning_rate": 8.520983266944949e-08,
      "loss": 0.6527,
      "step": 13149
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.574676990509033,
      "learning_rate": 8.499648036450036e-08,
      "loss": 0.1345,
      "step": 13150
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.74813985824585,
      "learning_rate": 8.478339320538653e-08,
      "loss": 0.7712,
      "step": 13151
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.324835777282715,
      "learning_rate": 8.457057120360379e-08,
      "loss": 0.4599,
      "step": 13152
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.7190306186676025,
      "learning_rate": 8.435801437063296e-08,
      "loss": 0.1326,
      "step": 13153
    },
    {
      "epoch": 0.94,
      "grad_norm": 16.254568099975586,
      "learning_rate": 8.414572271793986e-08,
      "loss": 0.4598,
      "step": 13154
    },
    {
      "epoch": 0.94,
      "grad_norm": 12.420778274536133,
      "learning_rate": 8.393369625697811e-08,
      "loss": 0.7855,
      "step": 13155
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.368011951446533,
      "learning_rate": 8.372193499918524e-08,
      "loss": 0.0906,
      "step": 13156
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6615548133850098,
      "learning_rate": 8.351043895598432e-08,
      "loss": 0.1014,
      "step": 13157
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.328801155090332,
      "learning_rate": 8.329920813878567e-08,
      "loss": 0.7938,
      "step": 13158
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.3946609497070312,
      "learning_rate": 8.308824255898407e-08,
      "loss": 0.2029,
      "step": 13159
    },
    {
      "epoch": 0.94,
      "grad_norm": 18.145992279052734,
      "learning_rate": 8.287754222795985e-08,
      "loss": 0.8047,
      "step": 13160
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.290095329284668,
      "learning_rate": 8.266710715708059e-08,
      "loss": 0.9098,
      "step": 13161
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.642944574356079,
      "learning_rate": 8.245693735769721e-08,
      "loss": 0.6746,
      "step": 13162
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2317819595336914,
      "learning_rate": 8.224703284114787e-08,
      "loss": 0.0754,
      "step": 13163
    },
    {
      "epoch": 0.94,
      "grad_norm": 20.73026466369629,
      "learning_rate": 8.203739361875684e-08,
      "loss": 1.088,
      "step": 13164
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3531813621520996,
      "learning_rate": 8.182801970183341e-08,
      "loss": 0.1068,
      "step": 13165
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.946152448654175,
      "learning_rate": 8.16189111016713e-08,
      "loss": 0.189,
      "step": 13166
    },
    {
      "epoch": 0.94,
      "grad_norm": 14.078112602233887,
      "learning_rate": 8.141006782955207e-08,
      "loss": 0.7095,
      "step": 13167
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6947169303894043,
      "learning_rate": 8.120148989674225e-08,
      "loss": 0.0963,
      "step": 13168
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.047933578491211,
      "learning_rate": 8.099317731449286e-08,
      "loss": 0.6299,
      "step": 13169
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.194098711013794,
      "learning_rate": 8.07851300940421e-08,
      "loss": 0.0316,
      "step": 13170
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.6446125507354736,
      "learning_rate": 8.057734824661379e-08,
      "loss": 0.6437,
      "step": 13171
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6956329345703125,
      "learning_rate": 8.036983178341617e-08,
      "loss": 0.0951,
      "step": 13172
    },
    {
      "epoch": 0.94,
      "grad_norm": 13.680745124816895,
      "learning_rate": 8.016258071564364e-08,
      "loss": 0.4034,
      "step": 13173
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.642326831817627,
      "learning_rate": 7.995559505447836e-08,
      "loss": 0.5224,
      "step": 13174
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.1581954956054688,
      "learning_rate": 7.974887481108472e-08,
      "loss": 0.204,
      "step": 13175
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.570911169052124,
      "learning_rate": 7.95424199966155e-08,
      "loss": 0.0969,
      "step": 13176
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.137040138244629,
      "learning_rate": 7.93362306222073e-08,
      "loss": 0.1875,
      "step": 13177
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.979308605194092,
      "learning_rate": 7.913030669898402e-08,
      "loss": 0.7011,
      "step": 13178
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.887205600738525,
      "learning_rate": 7.892464823805401e-08,
      "loss": 0.4222,
      "step": 13179
    },
    {
      "epoch": 0.95,
      "grad_norm": 50.75909423828125,
      "learning_rate": 7.871925525051283e-08,
      "loss": 0.5747,
      "step": 13180
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.273439645767212,
      "learning_rate": 7.851412774743938e-08,
      "loss": 0.0516,
      "step": 13181
    },
    {
      "epoch": 0.95,
      "grad_norm": 35.85977554321289,
      "learning_rate": 7.830926573989928e-08,
      "loss": 0.3737,
      "step": 13182
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.090315341949463,
      "learning_rate": 7.810466923894478e-08,
      "loss": 0.2055,
      "step": 13183
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.297146320343018,
      "learning_rate": 7.790033825561372e-08,
      "loss": 0.8334,
      "step": 13184
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.513716697692871,
      "learning_rate": 7.769627280092728e-08,
      "loss": 0.806,
      "step": 13185
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.301250457763672,
      "learning_rate": 7.749247288589557e-08,
      "loss": 0.9742,
      "step": 13186
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0091278553009033,
      "learning_rate": 7.728893852151253e-08,
      "loss": 0.0712,
      "step": 13187
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.5276832580566406,
      "learning_rate": 7.708566971875663e-08,
      "loss": 0.0946,
      "step": 13188
    },
    {
      "epoch": 0.95,
      "grad_norm": 20.18840217590332,
      "learning_rate": 7.688266648859577e-08,
      "loss": 0.4697,
      "step": 13189
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.927290439605713,
      "learning_rate": 7.66799288419795e-08,
      "loss": 0.6866,
      "step": 13190
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.844334363937378,
      "learning_rate": 7.647745678984519e-08,
      "loss": 0.2813,
      "step": 13191
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.729759693145752,
      "learning_rate": 7.627525034311634e-08,
      "loss": 0.554,
      "step": 13192
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.000827789306641,
      "learning_rate": 7.607330951270031e-08,
      "loss": 0.2893,
      "step": 13193
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.257803440093994,
      "learning_rate": 7.587163430949008e-08,
      "loss": 0.0873,
      "step": 13194
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.510326862335205,
      "learning_rate": 7.567022474436691e-08,
      "loss": 0.1918,
      "step": 13195
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.4095945358276367,
      "learning_rate": 7.546908082819604e-08,
      "loss": 0.154,
      "step": 13196
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3556432723999023,
      "learning_rate": 7.52682025718271e-08,
      "loss": 0.0996,
      "step": 13197
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.554915428161621,
      "learning_rate": 7.506758998609809e-08,
      "loss": 0.5249,
      "step": 13198
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7764298915863037,
      "learning_rate": 7.486724308183036e-08,
      "loss": 0.7696,
      "step": 13199
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.135210037231445,
      "learning_rate": 7.466716186983247e-08,
      "loss": 0.7353,
      "step": 13200
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2575061321258545,
      "learning_rate": 7.446734636089747e-08,
      "loss": 0.0874,
      "step": 13201
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.593588352203369,
      "learning_rate": 7.426779656580507e-08,
      "loss": 0.6942,
      "step": 13202
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.955426216125488,
      "learning_rate": 7.406851249532054e-08,
      "loss": 0.3318,
      "step": 13203
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3654959201812744,
      "learning_rate": 7.386949416019418e-08,
      "loss": 0.091,
      "step": 13204
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.541278123855591,
      "learning_rate": 7.367074157116183e-08,
      "loss": 0.0885,
      "step": 13205
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.46201229095459,
      "learning_rate": 7.347225473894549e-08,
      "loss": 0.7726,
      "step": 13206
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.862241268157959,
      "learning_rate": 7.327403367425323e-08,
      "loss": 0.3047,
      "step": 13207
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.254202365875244,
      "learning_rate": 7.307607838777875e-08,
      "loss": 0.7023,
      "step": 13208
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.9241042137146,
      "learning_rate": 7.287838889020016e-08,
      "loss": 0.4244,
      "step": 13209
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4222967624664307,
      "learning_rate": 7.268096519218171e-08,
      "loss": 0.087,
      "step": 13210
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1778502464294434,
      "learning_rate": 7.248380730437487e-08,
      "loss": 0.0644,
      "step": 13211
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.539955139160156,
      "learning_rate": 7.228691523741505e-08,
      "loss": 0.4792,
      "step": 13212
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.633544445037842,
      "learning_rate": 7.209028900192317e-08,
      "loss": 0.0859,
      "step": 13213
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.227051734924316,
      "learning_rate": 7.189392860850741e-08,
      "loss": 0.3905,
      "step": 13214
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.61845064163208,
      "learning_rate": 7.169783406776043e-08,
      "loss": 0.1246,
      "step": 13215
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.084618330001831,
      "learning_rate": 7.150200539026042e-08,
      "loss": 0.1774,
      "step": 13216
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.755073547363281,
      "learning_rate": 7.130644258657227e-08,
      "loss": 0.6048,
      "step": 13217
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.904087066650391,
      "learning_rate": 7.111114566724531e-08,
      "loss": 0.8261,
      "step": 13218
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.525852680206299,
      "learning_rate": 7.091611464281444e-08,
      "loss": 0.0855,
      "step": 13219
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4985854625701904,
      "learning_rate": 7.072134952380238e-08,
      "loss": 0.089,
      "step": 13220
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4107871055603027,
      "learning_rate": 7.052685032071516e-08,
      "loss": 0.1169,
      "step": 13221
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.078543663024902,
      "learning_rate": 7.033261704404493e-08,
      "loss": 0.4624,
      "step": 13222
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.4424796104431152,
      "learning_rate": 7.013864970427053e-08,
      "loss": 0.1663,
      "step": 13223
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.5843353271484375,
      "learning_rate": 6.994494831185528e-08,
      "loss": 0.1093,
      "step": 13224
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.755157709121704,
      "learning_rate": 6.975151287724858e-08,
      "loss": 0.0468,
      "step": 13225
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.980064392089844,
      "learning_rate": 6.955834341088596e-08,
      "loss": 0.4118,
      "step": 13226
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1481988430023193,
      "learning_rate": 6.9365439923188e-08,
      "loss": 0.103,
      "step": 13227
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.334554672241211,
      "learning_rate": 6.917280242456137e-08,
      "loss": 0.5241,
      "step": 13228
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.4847702980041504,
      "learning_rate": 6.898043092539719e-08,
      "loss": 0.1802,
      "step": 13229
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.081577777862549,
      "learning_rate": 6.87883254360744e-08,
      "loss": 0.8699,
      "step": 13230
    },
    {
      "epoch": 0.95,
      "grad_norm": 23.4083251953125,
      "learning_rate": 6.859648596695522e-08,
      "loss": 0.7435,
      "step": 13231
    },
    {
      "epoch": 0.95,
      "grad_norm": 13.042851448059082,
      "learning_rate": 6.840491252838976e-08,
      "loss": 0.8335,
      "step": 13232
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.557686805725098,
      "learning_rate": 6.821360513071251e-08,
      "loss": 0.5418,
      "step": 13233
    },
    {
      "epoch": 0.95,
      "grad_norm": 28.112056732177734,
      "learning_rate": 6.802256378424244e-08,
      "loss": 0.7937,
      "step": 13234
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7712898254394531,
      "learning_rate": 6.78317884992874e-08,
      "loss": 0.0412,
      "step": 13235
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9052045345306396,
      "learning_rate": 6.76412792861375e-08,
      "loss": 0.0668,
      "step": 13236
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.21299934387207,
      "learning_rate": 6.74510361550712e-08,
      "loss": 0.7498,
      "step": 13237
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.524225234985352,
      "learning_rate": 6.726105911635028e-08,
      "loss": 1.063,
      "step": 13238
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.682020425796509,
      "learning_rate": 6.707134818022377e-08,
      "loss": 0.0643,
      "step": 13239
    },
    {
      "epoch": 0.95,
      "grad_norm": 14.441481590270996,
      "learning_rate": 6.688190335692623e-08,
      "loss": 0.6195,
      "step": 13240
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.526546001434326,
      "learning_rate": 6.669272465667675e-08,
      "loss": 0.1043,
      "step": 13241
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.528806209564209,
      "learning_rate": 6.65038120896816e-08,
      "loss": 0.1529,
      "step": 13242
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.003218412399292,
      "learning_rate": 6.631516566613094e-08,
      "loss": 0.1228,
      "step": 13243
    },
    {
      "epoch": 0.95,
      "grad_norm": 20.171411514282227,
      "learning_rate": 6.612678539620277e-08,
      "loss": 1.0445,
      "step": 13244
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.5769453048706055,
      "learning_rate": 6.593867129005838e-08,
      "loss": 0.1219,
      "step": 13245
    },
    {
      "epoch": 0.95,
      "grad_norm": 18.253692626953125,
      "learning_rate": 6.575082335784632e-08,
      "loss": 0.3007,
      "step": 13246
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.219906330108643,
      "learning_rate": 6.556324160970018e-08,
      "loss": 0.8383,
      "step": 13247
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2568442821502686,
      "learning_rate": 6.537592605573905e-08,
      "loss": 0.0855,
      "step": 13248
    },
    {
      "epoch": 0.95,
      "grad_norm": 13.329076766967773,
      "learning_rate": 6.518887670606877e-08,
      "loss": 0.8545,
      "step": 13249
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.570000648498535,
      "learning_rate": 6.500209357077847e-08,
      "loss": 0.4419,
      "step": 13250
    },
    {
      "epoch": 0.95,
      "grad_norm": 19.385786056518555,
      "learning_rate": 6.481557665994564e-08,
      "loss": 0.5621,
      "step": 13251
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.032224655151367,
      "learning_rate": 6.46293259836328e-08,
      "loss": 0.6576,
      "step": 13252
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.241868019104004,
      "learning_rate": 6.44433415518847e-08,
      "loss": 0.8176,
      "step": 13253
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6182936429977417,
      "learning_rate": 6.425762337473717e-08,
      "loss": 0.0819,
      "step": 13254
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.5446834564208984,
      "learning_rate": 6.407217146220835e-08,
      "loss": 0.1096,
      "step": 13255
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.4276998043060303,
      "learning_rate": 6.38869858243013e-08,
      "loss": 0.1899,
      "step": 13256
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.8482260704040527,
      "learning_rate": 6.370206647100807e-08,
      "loss": 0.1113,
      "step": 13257
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4775760173797607,
      "learning_rate": 6.351741341230288e-08,
      "loss": 0.0854,
      "step": 13258
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3664722442626953,
      "learning_rate": 6.333302665814778e-08,
      "loss": 0.0901,
      "step": 13259
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.491835594177246,
      "learning_rate": 6.314890621848979e-08,
      "loss": 0.1057,
      "step": 13260
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0735485553741455,
      "learning_rate": 6.296505210326153e-08,
      "loss": 0.0538,
      "step": 13261
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.096113681793213,
      "learning_rate": 6.278146432238063e-08,
      "loss": 0.8211,
      "step": 13262
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.789889812469482,
      "learning_rate": 6.25981428857514e-08,
      "loss": 0.3,
      "step": 13263
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0753931999206543,
      "learning_rate": 6.241508780326366e-08,
      "loss": 0.0711,
      "step": 13264
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.288551330566406,
      "learning_rate": 6.223229908479178e-08,
      "loss": 0.4526,
      "step": 13265
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.2031779289245605,
      "learning_rate": 6.204977674019675e-08,
      "loss": 0.0679,
      "step": 13266
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.029527425765991,
      "learning_rate": 6.186752077932567e-08,
      "loss": 0.1962,
      "step": 13267
    },
    {
      "epoch": 0.95,
      "grad_norm": 18.015121459960938,
      "learning_rate": 6.168553121200904e-08,
      "loss": 0.6235,
      "step": 13268
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7559436559677124,
      "learning_rate": 6.150380804806622e-08,
      "loss": 0.0524,
      "step": 13269
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.413970470428467,
      "learning_rate": 6.132235129729991e-08,
      "loss": 0.1241,
      "step": 13270
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2038490772247314,
      "learning_rate": 6.114116096949785e-08,
      "loss": 0.0677,
      "step": 13271
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.963517427444458,
      "learning_rate": 6.096023707443666e-08,
      "loss": 0.697,
      "step": 13272
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.806461215019226,
      "learning_rate": 6.077957962187464e-08,
      "loss": 0.0483,
      "step": 13273
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.833909034729004,
      "learning_rate": 6.0599188621559e-08,
      "loss": 0.8181,
      "step": 13274
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.428664207458496,
      "learning_rate": 6.041906408321973e-08,
      "loss": 0.9841,
      "step": 13275
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.361494064331055,
      "learning_rate": 6.023920601657574e-08,
      "loss": 0.3723,
      "step": 13276
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.3998212814331055,
      "learning_rate": 6.005961443132757e-08,
      "loss": 0.1781,
      "step": 13277
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.249256134033203,
      "learning_rate": 5.988028933716473e-08,
      "loss": 0.1365,
      "step": 13278
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.408634662628174,
      "learning_rate": 5.97012307437611e-08,
      "loss": 0.3544,
      "step": 13279
    },
    {
      "epoch": 0.95,
      "grad_norm": 14.56265926361084,
      "learning_rate": 5.952243866077623e-08,
      "loss": 0.9883,
      "step": 13280
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.133974313735962,
      "learning_rate": 5.934391309785403e-08,
      "loss": 0.0719,
      "step": 13281
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.586527109146118,
      "learning_rate": 5.9165654064627374e-08,
      "loss": 0.1002,
      "step": 13282
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.8087334632873535,
      "learning_rate": 5.898766157071134e-08,
      "loss": 0.1396,
      "step": 13283
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.908201217651367,
      "learning_rate": 5.880993562570769e-08,
      "loss": 0.6742,
      "step": 13284
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.446937084197998,
      "learning_rate": 5.8632476239205424e-08,
      "loss": 0.0795,
      "step": 13285
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.837292432785034,
      "learning_rate": 5.845528342077633e-08,
      "loss": 0.1454,
      "step": 13286
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.548674583435059,
      "learning_rate": 5.827835717997998e-08,
      "loss": 0.866,
      "step": 13287
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.2412450313568115,
      "learning_rate": 5.810169752636152e-08,
      "loss": 0.1777,
      "step": 13288
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.8535943031311035,
      "learning_rate": 5.7925304469449994e-08,
      "loss": 0.5166,
      "step": 13289
    },
    {
      "epoch": 0.95,
      "grad_norm": 15.959070205688477,
      "learning_rate": 5.774917801876112e-08,
      "loss": 1.0072,
      "step": 13290
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.929064989089966,
      "learning_rate": 5.75733181837973e-08,
      "loss": 0.1578,
      "step": 13291
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.4174442291259766,
      "learning_rate": 5.739772497404483e-08,
      "loss": 0.1363,
      "step": 13292
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.696290016174316,
      "learning_rate": 5.722239839897559e-08,
      "loss": 0.8556,
      "step": 13293
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7439379692077637,
      "learning_rate": 5.704733846804922e-08,
      "loss": 0.739,
      "step": 13294
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.500216484069824,
      "learning_rate": 5.6872545190708727e-08,
      "loss": 0.1052,
      "step": 13295
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0136749744415283,
      "learning_rate": 5.669801857638324e-08,
      "loss": 0.0563,
      "step": 13296
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.117849349975586,
      "learning_rate": 5.652375863448856e-08,
      "loss": 0.5784,
      "step": 13297
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1609835624694824,
      "learning_rate": 5.634976537442494e-08,
      "loss": 0.0568,
      "step": 13298
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.657191276550293,
      "learning_rate": 5.6176038805578756e-08,
      "loss": 0.0917,
      "step": 13299
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.9964165687561035,
      "learning_rate": 5.6002578937321974e-08,
      "loss": 0.115,
      "step": 13300
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.7394843101501465,
      "learning_rate": 5.582938577901209e-08,
      "loss": 0.1819,
      "step": 13301
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.2779030799865723,
      "learning_rate": 5.56564593399922e-08,
      "loss": 0.1066,
      "step": 13302
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.2788519859313965,
      "learning_rate": 5.548379962959094e-08,
      "loss": 0.2661,
      "step": 13303
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.24526309967041,
      "learning_rate": 5.531140665712254e-08,
      "loss": 0.3034,
      "step": 13304
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.12748384475708,
      "learning_rate": 5.5139280431886765e-08,
      "loss": 0.8313,
      "step": 13305
    },
    {
      "epoch": 0.95,
      "grad_norm": 13.997440338134766,
      "learning_rate": 5.496742096317009e-08,
      "loss": 0.8039,
      "step": 13306
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.792363405227661,
      "learning_rate": 5.479582826024288e-08,
      "loss": 0.0845,
      "step": 13307
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0011584758758545,
      "learning_rate": 5.462450233236161e-08,
      "loss": 0.1009,
      "step": 13308
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.49119234085083,
      "learning_rate": 5.445344318876999e-08,
      "loss": 0.9315,
      "step": 13309
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.160655975341797,
      "learning_rate": 5.428265083869455e-08,
      "loss": 0.5653,
      "step": 13310
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.3864240646362305,
      "learning_rate": 5.411212529135012e-08,
      "loss": 0.7888,
      "step": 13311
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.122159957885742,
      "learning_rate": 5.394186655593492e-08,
      "loss": 0.8757,
      "step": 13312
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2276551723480225,
      "learning_rate": 5.377187464163436e-08,
      "loss": 0.0669,
      "step": 13313
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.909273147583008,
      "learning_rate": 5.360214955761833e-08,
      "loss": 0.3245,
      "step": 13314
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.712085723876953,
      "learning_rate": 5.343269131304396e-08,
      "loss": 0.5931,
      "step": 13315
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.687019348144531,
      "learning_rate": 5.326349991705171e-08,
      "loss": 0.7991,
      "step": 13316
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.654457688331604,
      "learning_rate": 5.309457537876928e-08,
      "loss": 0.0405,
      "step": 13317
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3013100624084473,
      "learning_rate": 5.292591770730882e-08,
      "loss": 0.0673,
      "step": 13318
    },
    {
      "epoch": 0.96,
      "grad_norm": 26.361095428466797,
      "learning_rate": 5.2757526911770275e-08,
      "loss": 0.9809,
      "step": 13319
    },
    {
      "epoch": 0.96,
      "grad_norm": 17.83018684387207,
      "learning_rate": 5.2589403001236385e-08,
      "loss": 1.2071,
      "step": 13320
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.479372978210449,
      "learning_rate": 5.242154598477711e-08,
      "loss": 0.1332,
      "step": 13321
    },
    {
      "epoch": 0.96,
      "grad_norm": 18.98883056640625,
      "learning_rate": 5.225395587144799e-08,
      "loss": 1.5688,
      "step": 13322
    },
    {
      "epoch": 0.96,
      "grad_norm": 15.240880012512207,
      "learning_rate": 5.208663267029013e-08,
      "loss": 0.7197,
      "step": 13323
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.536976337432861,
      "learning_rate": 5.191957639032852e-08,
      "loss": 0.6849,
      "step": 13324
    },
    {
      "epoch": 0.96,
      "grad_norm": 14.810187339782715,
      "learning_rate": 5.1752787040577625e-08,
      "loss": 0.4691,
      "step": 13325
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.5889954566955566,
      "learning_rate": 5.158626463003302e-08,
      "loss": 0.1295,
      "step": 13326
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.6209661960601807,
      "learning_rate": 5.142000916767864e-08,
      "loss": 0.2357,
      "step": 13327
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4186432361602783,
      "learning_rate": 5.125402066248342e-08,
      "loss": 0.0921,
      "step": 13328
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.566723346710205,
      "learning_rate": 5.1088299123401854e-08,
      "loss": 0.9984,
      "step": 13329
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.979346513748169,
      "learning_rate": 5.0922844559373466e-08,
      "loss": 0.0861,
      "step": 13330
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.674711227416992,
      "learning_rate": 5.0757656979325e-08,
      "loss": 0.5707,
      "step": 13331
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.862525701522827,
      "learning_rate": 5.059273639216655e-08,
      "loss": 0.1161,
      "step": 13332
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.132038116455078,
      "learning_rate": 5.0428082806795454e-08,
      "loss": 0.2713,
      "step": 13333
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.206965446472168,
      "learning_rate": 5.026369623209459e-08,
      "loss": 0.0845,
      "step": 13334
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.487266778945923,
      "learning_rate": 5.009957667693133e-08,
      "loss": 0.0641,
      "step": 13335
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8706490993499756,
      "learning_rate": 4.9935724150159125e-08,
      "loss": 0.0802,
      "step": 13336
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.557749032974243,
      "learning_rate": 4.97721386606187e-08,
      "loss": 0.2446,
      "step": 13337
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.78349494934082,
      "learning_rate": 4.960882021713298e-08,
      "loss": 0.6677,
      "step": 13338
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.954891204833984,
      "learning_rate": 4.944576882851271e-08,
      "loss": 0.9114,
      "step": 13339
    },
    {
      "epoch": 0.96,
      "grad_norm": 19.414329528808594,
      "learning_rate": 4.9282984503554735e-08,
      "loss": 0.7109,
      "step": 13340
    },
    {
      "epoch": 0.96,
      "grad_norm": 15.215154647827148,
      "learning_rate": 4.912046725104092e-08,
      "loss": 0.5666,
      "step": 13341
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.9473135471343994,
      "learning_rate": 4.895821707973702e-08,
      "loss": 0.7348,
      "step": 13342
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.374669075012207,
      "learning_rate": 4.879623399839662e-08,
      "loss": 0.0996,
      "step": 13343
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.6218559741973877,
      "learning_rate": 4.8634518015758244e-08,
      "loss": 0.1974,
      "step": 13344
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.913919448852539,
      "learning_rate": 4.847306914054606e-08,
      "loss": 0.4072,
      "step": 13345
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8724513053894043,
      "learning_rate": 4.8311887381468636e-08,
      "loss": 0.0954,
      "step": 13346
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.5922329425811768,
      "learning_rate": 4.8150972747221804e-08,
      "loss": 0.1204,
      "step": 13347
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.1132588386535645,
      "learning_rate": 4.799032524648639e-08,
      "loss": 0.537,
      "step": 13348
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.120995283126831,
      "learning_rate": 4.7829944887928804e-08,
      "loss": 0.1721,
      "step": 13349
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.2057275772094727,
      "learning_rate": 4.766983168020045e-08,
      "loss": 0.1264,
      "step": 13350
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.08419418334961,
      "learning_rate": 4.750998563193887e-08,
      "loss": 0.4239,
      "step": 13351
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.68772554397583,
      "learning_rate": 4.735040675176772e-08,
      "loss": 0.8514,
      "step": 13352
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.612338066101074,
      "learning_rate": 4.7191095048295664e-08,
      "loss": 0.9805,
      "step": 13353
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.800552368164062,
      "learning_rate": 4.7032050530116394e-08,
      "loss": 0.7239,
      "step": 13354
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4168739318847656,
      "learning_rate": 4.6873273205809144e-08,
      "loss": 0.0456,
      "step": 13355
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.6886372566223145,
      "learning_rate": 4.6714763083940964e-08,
      "loss": 0.6495,
      "step": 13356
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2921810150146484,
      "learning_rate": 4.655652017306222e-08,
      "loss": 0.1256,
      "step": 13357
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.211258888244629,
      "learning_rate": 4.6398544481708885e-08,
      "loss": 0.9088,
      "step": 13358
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.941694259643555,
      "learning_rate": 4.6240836018404124e-08,
      "loss": 0.4882,
      "step": 13359
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.185328960418701,
      "learning_rate": 4.608339479165558e-08,
      "loss": 0.1108,
      "step": 13360
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.6922733783721924,
      "learning_rate": 4.59262208099559e-08,
      "loss": 0.1679,
      "step": 13361
    },
    {
      "epoch": 0.96,
      "grad_norm": 26.925029754638672,
      "learning_rate": 4.5769314081784424e-08,
      "loss": 0.5352,
      "step": 13362
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.312771797180176,
      "learning_rate": 4.561267461560548e-08,
      "loss": 0.3507,
      "step": 13363
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.9888458251953125,
      "learning_rate": 4.545630241986954e-08,
      "loss": 0.1788,
      "step": 13364
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.123178005218506,
      "learning_rate": 4.5300197503012066e-08,
      "loss": 0.5817,
      "step": 13365
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.453211545944214,
      "learning_rate": 4.514435987345411e-08,
      "loss": 0.1094,
      "step": 13366
    },
    {
      "epoch": 0.96,
      "grad_norm": 14.067482948303223,
      "learning_rate": 4.498878953960284e-08,
      "loss": 0.6177,
      "step": 13367
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.424760580062866,
      "learning_rate": 4.483348650985098e-08,
      "loss": 0.1682,
      "step": 13368
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.663593292236328,
      "learning_rate": 4.467845079257627e-08,
      "loss": 0.7338,
      "step": 13369
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.981987953186035,
      "learning_rate": 4.452368239614202e-08,
      "loss": 0.6676,
      "step": 13370
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.064502716064453,
      "learning_rate": 4.436918132889767e-08,
      "loss": 0.551,
      "step": 13371
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.32633113861084,
      "learning_rate": 4.421494759917766e-08,
      "loss": 0.052,
      "step": 13372
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.052866458892822,
      "learning_rate": 4.406098121530311e-08,
      "loss": 0.8139,
      "step": 13373
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.749582290649414,
      "learning_rate": 4.390728218557905e-08,
      "loss": 0.1725,
      "step": 13374
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.344928026199341,
      "learning_rate": 4.375385051829717e-08,
      "loss": 0.0906,
      "step": 13375
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.676229238510132,
      "learning_rate": 4.3600686221734743e-08,
      "loss": 0.1738,
      "step": 13376
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3149256706237793,
      "learning_rate": 4.344778930415461e-08,
      "loss": 0.0404,
      "step": 13377
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.039262294769287,
      "learning_rate": 4.329515977380405e-08,
      "loss": 0.0365,
      "step": 13378
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.0025243759155273,
      "learning_rate": 4.31427976389176e-08,
      "loss": 0.1122,
      "step": 13379
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.7091763019561768,
      "learning_rate": 4.2990702907714787e-08,
      "loss": 0.7076,
      "step": 13380
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.620807647705078,
      "learning_rate": 4.2838875588400166e-08,
      "loss": 0.1147,
      "step": 13381
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.141193389892578,
      "learning_rate": 4.26873156891644e-08,
      "loss": 0.0783,
      "step": 13382
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.126354455947876,
      "learning_rate": 4.253602321818373e-08,
      "loss": 0.2317,
      "step": 13383
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.664473056793213,
      "learning_rate": 4.2384998183618854e-08,
      "loss": 0.7284,
      "step": 13384
    },
    {
      "epoch": 0.96,
      "grad_norm": 14.932379722595215,
      "learning_rate": 4.223424059361825e-08,
      "loss": 0.7749,
      "step": 13385
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.852616310119629,
      "learning_rate": 4.20837504563143e-08,
      "loss": 0.6207,
      "step": 13386
    },
    {
      "epoch": 0.96,
      "grad_norm": 14.38532543182373,
      "learning_rate": 4.193352777982551e-08,
      "loss": 0.8234,
      "step": 13387
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.9620914459228516,
      "learning_rate": 4.178357257225485e-08,
      "loss": 0.7939,
      "step": 13388
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.83671498298645,
      "learning_rate": 4.163388484169306e-08,
      "loss": 0.1949,
      "step": 13389
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.160528659820557,
      "learning_rate": 4.1484464596214804e-08,
      "loss": 0.1683,
      "step": 13390
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4549999237060547,
      "learning_rate": 4.1335311843880286e-08,
      "loss": 0.1551,
      "step": 13391
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.821384906768799,
      "learning_rate": 4.118642659273642e-08,
      "loss": 0.6004,
      "step": 13392
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.7048332691192627,
      "learning_rate": 4.1037808850814544e-08,
      "loss": 0.8309,
      "step": 13393
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.750955581665039,
      "learning_rate": 4.088945862613214e-08,
      "loss": 0.1778,
      "step": 13394
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.99651050567627,
      "learning_rate": 4.074137592669225e-08,
      "loss": 0.7496,
      "step": 13395
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.196304798126221,
      "learning_rate": 4.059356076048348e-08,
      "loss": 0.3267,
      "step": 13396
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.939056396484375,
      "learning_rate": 4.044601313548002e-08,
      "loss": 0.2175,
      "step": 13397
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7863903045654297,
      "learning_rate": 4.0298733059640496e-08,
      "loss": 0.0701,
      "step": 13398
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.21479606628418,
      "learning_rate": 4.015172054091132e-08,
      "loss": 0.6512,
      "step": 13399
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.956916332244873,
      "learning_rate": 4.0004975587222824e-08,
      "loss": 0.0837,
      "step": 13400
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.499246597290039,
      "learning_rate": 3.98584982064909e-08,
      "loss": 0.6173,
      "step": 13401
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.691965103149414,
      "learning_rate": 3.971228840661812e-08,
      "loss": 0.2926,
      "step": 13402
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.8912100791931152,
      "learning_rate": 3.956634619549149e-08,
      "loss": 0.0831,
      "step": 13403
    },
    {
      "epoch": 0.96,
      "grad_norm": 14.170405387878418,
      "learning_rate": 3.942067158098417e-08,
      "loss": 0.9617,
      "step": 13404
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.830087184906006,
      "learning_rate": 3.927526457095488e-08,
      "loss": 0.1328,
      "step": 13405
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9911149740219116,
      "learning_rate": 3.913012517324788e-08,
      "loss": 0.0284,
      "step": 13406
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.47891902923584,
      "learning_rate": 3.898525339569248e-08,
      "loss": 0.9287,
      "step": 13407
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.836817979812622,
      "learning_rate": 3.884064924610464e-08,
      "loss": 0.0806,
      "step": 13408
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.272183895111084,
      "learning_rate": 3.86963127322848e-08,
      "loss": 0.7789,
      "step": 13409
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.437607765197754,
      "learning_rate": 3.855224386201895e-08,
      "loss": 0.297,
      "step": 13410
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.034572124481201,
      "learning_rate": 3.840844264307919e-08,
      "loss": 0.1064,
      "step": 13411
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.60036563873291,
      "learning_rate": 3.826490908322378e-08,
      "loss": 0.755,
      "step": 13412
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.800389766693115,
      "learning_rate": 3.812164319019485e-08,
      "loss": 0.4367,
      "step": 13413
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.8452348709106445,
      "learning_rate": 3.797864497172232e-08,
      "loss": 0.0558,
      "step": 13414
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.322871208190918,
      "learning_rate": 3.7835914435518925e-08,
      "loss": 0.1653,
      "step": 13415
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.310945510864258,
      "learning_rate": 3.769345158928517e-08,
      "loss": 0.2705,
      "step": 13416
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8824195861816406,
      "learning_rate": 3.755125644070656e-08,
      "loss": 0.0745,
      "step": 13417
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.312656879425049,
      "learning_rate": 3.74093289974542e-08,
      "loss": 0.7798,
      "step": 13418
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.959392547607422,
      "learning_rate": 3.7267669267183616e-08,
      "loss": 0.6356,
      "step": 13419
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.764313697814941,
      "learning_rate": 3.7126277257537036e-08,
      "loss": 0.4567,
      "step": 13420
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.281177282333374,
      "learning_rate": 3.698515297614336e-08,
      "loss": 0.2034,
      "step": 13421
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.9164795875549316,
      "learning_rate": 3.6844296430613714e-08,
      "loss": 0.7065,
      "step": 13422
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8352638483047485,
      "learning_rate": 3.670370762854758e-08,
      "loss": 0.0967,
      "step": 13423
    },
    {
      "epoch": 0.96,
      "grad_norm": 49.79658889770508,
      "learning_rate": 3.6563386577530005e-08,
      "loss": 0.5849,
      "step": 13424
    },
    {
      "epoch": 0.96,
      "grad_norm": 17.74705696105957,
      "learning_rate": 3.642333328512992e-08,
      "loss": 1.4046,
      "step": 13425
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.277052640914917,
      "learning_rate": 3.628354775890242e-08,
      "loss": 0.0934,
      "step": 13426
    },
    {
      "epoch": 0.96,
      "grad_norm": 23.5953426361084,
      "learning_rate": 3.614403000638922e-08,
      "loss": 1.2687,
      "step": 13427
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.113100528717041,
      "learning_rate": 3.6004780035116537e-08,
      "loss": 0.8485,
      "step": 13428
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.6616199016571045,
      "learning_rate": 3.586579785259559e-08,
      "loss": 0.0708,
      "step": 13429
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.294409990310669,
      "learning_rate": 3.572708346632481e-08,
      "loss": 0.1103,
      "step": 13430
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8666231632232666,
      "learning_rate": 3.5588636883787664e-08,
      "loss": 0.0678,
      "step": 13431
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.228299140930176,
      "learning_rate": 3.5450458112451515e-08,
      "loss": 0.8546,
      "step": 13432
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.052384376525879,
      "learning_rate": 3.531254715977206e-08,
      "loss": 0.5618,
      "step": 13433
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.5325093269348145,
      "learning_rate": 3.5174904033187795e-08,
      "loss": 0.796,
      "step": 13434
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.6751723289489746,
      "learning_rate": 3.5037528740124446e-08,
      "loss": 0.0889,
      "step": 13435
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.6788716316223145,
      "learning_rate": 3.490042128799331e-08,
      "loss": 0.0921,
      "step": 13436
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.66918659210205,
      "learning_rate": 3.476358168419069e-08,
      "loss": 0.5666,
      "step": 13437
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.168299913406372,
      "learning_rate": 3.462700993609791e-08,
      "loss": 0.0827,
      "step": 13438
    },
    {
      "epoch": 0.96,
      "grad_norm": 28.553390502929688,
      "learning_rate": 3.4490706051082955e-08,
      "loss": 1.3856,
      "step": 13439
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.900330543518066,
      "learning_rate": 3.435467003649939e-08,
      "loss": 0.614,
      "step": 13440
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.143917083740234,
      "learning_rate": 3.4218901899685245e-08,
      "loss": 0.708,
      "step": 13441
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.9354233741760254,
      "learning_rate": 3.408340164796464e-08,
      "loss": 0.746,
      "step": 13442
    },
    {
      "epoch": 0.96,
      "grad_norm": 17.143247604370117,
      "learning_rate": 3.3948169288647856e-08,
      "loss": 1.2011,
      "step": 13443
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.278285503387451,
      "learning_rate": 3.3813204829029055e-08,
      "loss": 0.0802,
      "step": 13444
    },
    {
      "epoch": 0.96,
      "grad_norm": 21.77707290649414,
      "learning_rate": 3.3678508276390744e-08,
      "loss": 0.7126,
      "step": 13445
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2321040630340576,
      "learning_rate": 3.3544079637998214e-08,
      "loss": 0.0583,
      "step": 13446
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3541383743286133,
      "learning_rate": 3.34099189211029e-08,
      "loss": 0.1273,
      "step": 13447
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.1170284748077393,
      "learning_rate": 3.327602613294345e-08,
      "loss": 0.1229,
      "step": 13448
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.532699584960938,
      "learning_rate": 3.314240128074242e-08,
      "loss": 0.4706,
      "step": 13449
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.941819190979004,
      "learning_rate": 3.300904437170793e-08,
      "loss": 0.5535,
      "step": 13450
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.895922660827637,
      "learning_rate": 3.2875955413034234e-08,
      "loss": 0.3578,
      "step": 13451
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8856840133666992,
      "learning_rate": 3.274313441190169e-08,
      "loss": 0.0703,
      "step": 13452
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.692486763000488,
      "learning_rate": 3.2610581375474574e-08,
      "loss": 0.3604,
      "step": 13453
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.9268062114715576,
      "learning_rate": 3.247829631090438e-08,
      "loss": 0.184,
      "step": 13454
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.4358859062194824,
      "learning_rate": 3.234627922532652e-08,
      "loss": 0.1692,
      "step": 13455
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.358294486999512,
      "learning_rate": 3.221453012586362e-08,
      "loss": 0.815,
      "step": 13456
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4992945194244385,
      "learning_rate": 3.208304901962278e-08,
      "loss": 0.1478,
      "step": 13457
    },
    {
      "epoch": 0.97,
      "grad_norm": 12.914836883544922,
      "learning_rate": 3.195183591369666e-08,
      "loss": 0.744,
      "step": 13458
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.243896007537842,
      "learning_rate": 3.182089081516404e-08,
      "loss": 0.5141,
      "step": 13459
    },
    {
      "epoch": 0.97,
      "grad_norm": 12.041922569274902,
      "learning_rate": 3.1690213731088715e-08,
      "loss": 0.5512,
      "step": 13460
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.838679075241089,
      "learning_rate": 3.15598046685206e-08,
      "loss": 0.1316,
      "step": 13461
    },
    {
      "epoch": 0.97,
      "grad_norm": 12.015164375305176,
      "learning_rate": 3.142966363449407e-08,
      "loss": 0.5851,
      "step": 13462
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1234192848205566,
      "learning_rate": 3.129979063602962e-08,
      "loss": 0.0859,
      "step": 13463
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8765714168548584,
      "learning_rate": 3.117018568013497e-08,
      "loss": 0.0692,
      "step": 13464
    },
    {
      "epoch": 0.97,
      "grad_norm": 12.919534683227539,
      "learning_rate": 3.104084877380009e-08,
      "loss": 1.0213,
      "step": 13465
    },
    {
      "epoch": 0.97,
      "grad_norm": 13.862587928771973,
      "learning_rate": 3.091177992400329e-08,
      "loss": 0.2917,
      "step": 13466
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.8777220249176025,
      "learning_rate": 3.078297913770678e-08,
      "loss": 0.149,
      "step": 13467
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.124418020248413,
      "learning_rate": 3.0654446421858884e-08,
      "loss": 0.0696,
      "step": 13468
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4514424800872803,
      "learning_rate": 3.05261817833935e-08,
      "loss": 0.0581,
      "step": 13469
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.827178478240967,
      "learning_rate": 3.039818522923066e-08,
      "loss": 0.3614,
      "step": 13470
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.80055046081543,
      "learning_rate": 3.027045676627427e-08,
      "loss": 0.3423,
      "step": 13471
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.163187265396118,
      "learning_rate": 3.014299640141549e-08,
      "loss": 0.0628,
      "step": 13472
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.848196506500244,
      "learning_rate": 3.0015804141529934e-08,
      "loss": 0.6669,
      "step": 13473
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.219280958175659,
      "learning_rate": 2.988887999347989e-08,
      "loss": 0.106,
      "step": 13474
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.017998695373535,
      "learning_rate": 2.9762223964110993e-08,
      "loss": 0.7404,
      "step": 13475
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.311643123626709,
      "learning_rate": 2.9635836060257774e-08,
      "loss": 0.0886,
      "step": 13476
    },
    {
      "epoch": 0.97,
      "grad_norm": 13.174318313598633,
      "learning_rate": 2.9509716288736445e-08,
      "loss": 0.4917,
      "step": 13477
    },
    {
      "epoch": 0.97,
      "grad_norm": 18.463390350341797,
      "learning_rate": 2.9383864656352123e-08,
      "loss": 0.6534,
      "step": 13478
    },
    {
      "epoch": 0.97,
      "grad_norm": 20.048160552978516,
      "learning_rate": 2.9258281169893266e-08,
      "loss": 0.5885,
      "step": 13479
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.451717853546143,
      "learning_rate": 2.9132965836135007e-08,
      "loss": 0.8302,
      "step": 13480
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.93934440612793,
      "learning_rate": 2.9007918661837496e-08,
      "loss": 0.8027,
      "step": 13481
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.468745231628418,
      "learning_rate": 2.888313965374645e-08,
      "loss": 0.827,
      "step": 13482
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.291973114013672,
      "learning_rate": 2.8758628818593704e-08,
      "loss": 0.88,
      "step": 13483
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.42560076713562,
      "learning_rate": 2.8634386163094997e-08,
      "loss": 0.1416,
      "step": 13484
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.123434066772461,
      "learning_rate": 2.8510411693953855e-08,
      "loss": 0.0499,
      "step": 13485
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.1751275062561035,
      "learning_rate": 2.8386705417858264e-08,
      "loss": 0.1721,
      "step": 13486
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.354491710662842,
      "learning_rate": 2.826326734148066e-08,
      "loss": 0.1579,
      "step": 13487
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.814018726348877,
      "learning_rate": 2.8140097471481276e-08,
      "loss": 0.118,
      "step": 13488
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.656723976135254,
      "learning_rate": 2.8017195814503684e-08,
      "loss": 0.704,
      "step": 13489
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6315340995788574,
      "learning_rate": 2.7894562377179247e-08,
      "loss": 0.1057,
      "step": 13490
    },
    {
      "epoch": 0.97,
      "grad_norm": 15.474284172058105,
      "learning_rate": 2.7772197166121563e-08,
      "loss": 0.6624,
      "step": 13491
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.419720649719238,
      "learning_rate": 2.7650100187933683e-08,
      "loss": 0.3576,
      "step": 13492
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4375267028808594,
      "learning_rate": 2.7528271449201448e-08,
      "loss": 0.0803,
      "step": 13493
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.8584938049316406,
      "learning_rate": 2.7406710956497383e-08,
      "loss": 0.7311,
      "step": 13494
    },
    {
      "epoch": 0.97,
      "grad_norm": 13.117058753967285,
      "learning_rate": 2.7285418716378466e-08,
      "loss": 0.8642,
      "step": 13495
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.5583086013793945,
      "learning_rate": 2.7164394735388344e-08,
      "loss": 0.1892,
      "step": 13496
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4905760288238525,
      "learning_rate": 2.7043639020056246e-08,
      "loss": 0.0936,
      "step": 13497
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.021834373474121,
      "learning_rate": 2.69231515768964e-08,
      "loss": 0.1908,
      "step": 13498
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.457416534423828,
      "learning_rate": 2.6802932412408057e-08,
      "loss": 0.6093,
      "step": 13499
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.2332277297973633,
      "learning_rate": 2.6682981533077133e-08,
      "loss": 0.2233,
      "step": 13500
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.654815435409546,
      "learning_rate": 2.6563298945374015e-08,
      "loss": 0.6885,
      "step": 13501
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.580261707305908,
      "learning_rate": 2.644388465575576e-08,
      "loss": 0.1395,
      "step": 13502
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.839755058288574,
      "learning_rate": 2.6324738670663874e-08,
      "loss": 0.517,
      "step": 13503
    },
    {
      "epoch": 0.97,
      "grad_norm": 23.677078247070312,
      "learning_rate": 2.6205860996526e-08,
      "loss": 0.5438,
      "step": 13504
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.584489822387695,
      "learning_rate": 2.608725163975534e-08,
      "loss": 0.9097,
      "step": 13505
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.693812370300293,
      "learning_rate": 2.5968910606749553e-08,
      "loss": 0.1419,
      "step": 13506
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.164010047912598,
      "learning_rate": 2.585083790389409e-08,
      "loss": 0.4389,
      "step": 13507
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.9444990158081055,
      "learning_rate": 2.5733033537557185e-08,
      "loss": 0.7756,
      "step": 13508
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.9850354194641113,
      "learning_rate": 2.5615497514094313e-08,
      "loss": 0.1553,
      "step": 13509
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.8252298831939697,
      "learning_rate": 2.5498229839847065e-08,
      "loss": 0.6851,
      "step": 13510
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.070123195648193,
      "learning_rate": 2.5381230521140387e-08,
      "loss": 0.6416,
      "step": 13511
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.8301846981048584,
      "learning_rate": 2.5264499564285892e-08,
      "loss": 0.2298,
      "step": 13512
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.4191484451293945,
      "learning_rate": 2.514803697558188e-08,
      "loss": 0.8107,
      "step": 13513
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7600051164627075,
      "learning_rate": 2.5031842761309988e-08,
      "loss": 0.0299,
      "step": 13514
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.1659884452819824,
      "learning_rate": 2.4915916927739093e-08,
      "loss": 0.1576,
      "step": 13515
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.071124792098999,
      "learning_rate": 2.480025948112308e-08,
      "loss": 0.0696,
      "step": 13516
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3867084980010986,
      "learning_rate": 2.468487042770029e-08,
      "loss": 0.1086,
      "step": 13517
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.0427098274230957,
      "learning_rate": 2.456974977369686e-08,
      "loss": 0.1117,
      "step": 13518
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.947258472442627,
      "learning_rate": 2.4454897525322263e-08,
      "loss": 0.3138,
      "step": 13519
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.9380838871002197,
      "learning_rate": 2.43403136887721e-08,
      "loss": 0.7187,
      "step": 13520
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.8356618881225586,
      "learning_rate": 2.422599827022809e-08,
      "loss": 0.2988,
      "step": 13521
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4651317596435547,
      "learning_rate": 2.4111951275857527e-08,
      "loss": 0.1766,
      "step": 13522
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.697182655334473,
      "learning_rate": 2.399817271181215e-08,
      "loss": 0.3653,
      "step": 13523
    },
    {
      "epoch": 0.97,
      "grad_norm": 18.28174591064453,
      "learning_rate": 2.388466258423039e-08,
      "loss": 1.0294,
      "step": 13524
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9403491020202637,
      "learning_rate": 2.377142089923512e-08,
      "loss": 0.0626,
      "step": 13525
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7847627401351929,
      "learning_rate": 2.36584476629359e-08,
      "loss": 0.0548,
      "step": 13526
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.102972984313965,
      "learning_rate": 2.3545742881426748e-08,
      "loss": 0.1398,
      "step": 13527
    },
    {
      "epoch": 0.97,
      "grad_norm": 34.45656204223633,
      "learning_rate": 2.343330656078835e-08,
      "loss": 0.6859,
      "step": 13528
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.1587233543396,
      "learning_rate": 2.33211387070853e-08,
      "loss": 0.8186,
      "step": 13529
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.6591215133667,
      "learning_rate": 2.3209239326369427e-08,
      "loss": 0.4858,
      "step": 13530
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.195538282394409,
      "learning_rate": 2.3097608424676456e-08,
      "loss": 0.0792,
      "step": 13531
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.3917949199676514,
      "learning_rate": 2.2986246008029343e-08,
      "loss": 0.1625,
      "step": 13532
    },
    {
      "epoch": 0.97,
      "grad_norm": 26.831878662109375,
      "learning_rate": 2.287515208243496e-08,
      "loss": 1.3758,
      "step": 13533
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.645841598510742,
      "learning_rate": 2.2764326653886836e-08,
      "loss": 0.145,
      "step": 13534
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.906815528869629,
      "learning_rate": 2.2653769728362974e-08,
      "loss": 0.4837,
      "step": 13535
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.538771152496338,
      "learning_rate": 2.2543481311828595e-08,
      "loss": 0.1341,
      "step": 13536
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.287670135498047,
      "learning_rate": 2.2433461410232282e-08,
      "loss": 0.397,
      "step": 13537
    },
    {
      "epoch": 0.97,
      "grad_norm": 23.54510498046875,
      "learning_rate": 2.2323710029509836e-08,
      "loss": 0.8921,
      "step": 13538
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.5220835208892822,
      "learning_rate": 2.2214227175581525e-08,
      "loss": 0.1462,
      "step": 13539
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9478697776794434,
      "learning_rate": 2.2105012854353736e-08,
      "loss": 0.0771,
      "step": 13540
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.0227739810943604,
      "learning_rate": 2.1996067071718417e-08,
      "loss": 0.1492,
      "step": 13541
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.20666790008545,
      "learning_rate": 2.188738983355254e-08,
      "loss": 0.3377,
      "step": 13542
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.476851463317871,
      "learning_rate": 2.1778981145718635e-08,
      "loss": 0.1156,
      "step": 13543
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.958674907684326,
      "learning_rate": 2.1670841014064802e-08,
      "loss": 0.1108,
      "step": 13544
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.2065088748931885,
      "learning_rate": 2.1562969444425263e-08,
      "loss": 0.1533,
      "step": 13545
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.552970886230469,
      "learning_rate": 2.145536644261925e-08,
      "loss": 0.5784,
      "step": 13546
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.5332748889923096,
      "learning_rate": 2.134803201445157e-08,
      "loss": 0.1814,
      "step": 13547
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.338387489318848,
      "learning_rate": 2.124096616571203e-08,
      "loss": 0.4137,
      "step": 13548
    },
    {
      "epoch": 0.97,
      "grad_norm": 30.444793701171875,
      "learning_rate": 2.1134168902176565e-08,
      "loss": 0.6898,
      "step": 13549
    },
    {
      "epoch": 0.97,
      "grad_norm": 15.122658729553223,
      "learning_rate": 2.102764022960724e-08,
      "loss": 0.7364,
      "step": 13550
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6042330265045166,
      "learning_rate": 2.0921380153750004e-08,
      "loss": 0.1263,
      "step": 13551
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.7930445671081543,
      "learning_rate": 2.08153886803375e-08,
      "loss": 0.093,
      "step": 13552
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.08522891998291,
      "learning_rate": 2.0709665815087375e-08,
      "loss": 0.5469,
      "step": 13553
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4084012508392334,
      "learning_rate": 2.0604211563703402e-08,
      "loss": 0.1807,
      "step": 13554
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.166205883026123,
      "learning_rate": 2.0499025931873807e-08,
      "loss": 0.7817,
      "step": 13555
    },
    {
      "epoch": 0.97,
      "grad_norm": 15.939569473266602,
      "learning_rate": 2.039410892527349e-08,
      "loss": 0.7377,
      "step": 13556
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6630516052246094,
      "learning_rate": 2.028946054956238e-08,
      "loss": 0.0604,
      "step": 13557
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.345234394073486,
      "learning_rate": 2.0185080810385393e-08,
      "loss": 0.3123,
      "step": 13558
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.347875118255615,
      "learning_rate": 2.0080969713374142e-08,
      "loss": 0.4796,
      "step": 13559
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.086322784423828,
      "learning_rate": 1.9977127264144136e-08,
      "loss": 0.7663,
      "step": 13560
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.2039496898651123,
      "learning_rate": 1.9873553468297557e-08,
      "loss": 0.105,
      "step": 13561
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.579442024230957,
      "learning_rate": 1.977024833142216e-08,
      "loss": 0.5227,
      "step": 13562
    },
    {
      "epoch": 0.97,
      "grad_norm": 12.737347602844238,
      "learning_rate": 1.9667211859090707e-08,
      "loss": 0.602,
      "step": 13563
    },
    {
      "epoch": 0.97,
      "grad_norm": 13.465506553649902,
      "learning_rate": 1.9564444056861532e-08,
      "loss": 0.5116,
      "step": 13564
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.401611089706421,
      "learning_rate": 1.9461944930278533e-08,
      "loss": 0.0665,
      "step": 13565
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.505577802658081,
      "learning_rate": 1.9359714484871726e-08,
      "loss": 0.2683,
      "step": 13566
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.9254956245422363,
      "learning_rate": 1.9257752726155042e-08,
      "loss": 0.1304,
      "step": 13567
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0273098945617676,
      "learning_rate": 1.9156059659629633e-08,
      "loss": 0.1012,
      "step": 13568
    },
    {
      "epoch": 0.97,
      "grad_norm": 15.66772747039795,
      "learning_rate": 1.9054635290781663e-08,
      "loss": 1.0009,
      "step": 13569
    },
    {
      "epoch": 0.97,
      "grad_norm": 16.61053466796875,
      "learning_rate": 1.8953479625081762e-08,
      "loss": 0.5127,
      "step": 13570
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.431931257247925,
      "learning_rate": 1.8852592667987225e-08,
      "loss": 0.1429,
      "step": 13571
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.788093090057373,
      "learning_rate": 1.875197442494092e-08,
      "loss": 0.1706,
      "step": 13572
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.254063606262207,
      "learning_rate": 1.8651624901370736e-08,
      "loss": 0.7614,
      "step": 13573
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6531283855438232,
      "learning_rate": 1.8551544102689556e-08,
      "loss": 0.1158,
      "step": 13574
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.5662355422973633,
      "learning_rate": 1.8451732034296956e-08,
      "loss": 0.1166,
      "step": 13575
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.5917887687683105,
      "learning_rate": 1.8352188701576957e-08,
      "loss": 0.0397,
      "step": 13576
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.183804512023926,
      "learning_rate": 1.8252914109900265e-08,
      "loss": 0.0723,
      "step": 13577
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2674942016601562,
      "learning_rate": 1.8153908264622045e-08,
      "loss": 0.1263,
      "step": 13578
    },
    {
      "epoch": 0.97,
      "grad_norm": 20.444719314575195,
      "learning_rate": 1.8055171171082464e-08,
      "loss": 0.7794,
      "step": 13579
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.070065498352051,
      "learning_rate": 1.795670283460893e-08,
      "loss": 0.1787,
      "step": 13580
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.478021621704102,
      "learning_rate": 1.785850326051386e-08,
      "loss": 0.7803,
      "step": 13581
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.331636428833008,
      "learning_rate": 1.776057245409357e-08,
      "loss": 0.2421,
      "step": 13582
    },
    {
      "epoch": 0.97,
      "grad_norm": 27.183176040649414,
      "learning_rate": 1.7662910420631617e-08,
      "loss": 0.7333,
      "step": 13583
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.451833248138428,
      "learning_rate": 1.756551716539656e-08,
      "loss": 0.6394,
      "step": 13584
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4901890754699707,
      "learning_rate": 1.7468392693641978e-08,
      "loss": 0.103,
      "step": 13585
    },
    {
      "epoch": 0.97,
      "grad_norm": 17.422639846801758,
      "learning_rate": 1.7371537010608118e-08,
      "loss": 0.5808,
      "step": 13586
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.941697120666504,
      "learning_rate": 1.727495012151914e-08,
      "loss": 0.6332,
      "step": 13587
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.3960089683532715,
      "learning_rate": 1.7178632031586427e-08,
      "loss": 0.436,
      "step": 13588
    },
    {
      "epoch": 0.97,
      "grad_norm": 17.352184295654297,
      "learning_rate": 1.708258274600527e-08,
      "loss": 0.7131,
      "step": 13589
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.1497390270233154,
      "learning_rate": 1.6986802269957637e-08,
      "loss": 0.1735,
      "step": 13590
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.743584632873535,
      "learning_rate": 1.68912906086105e-08,
      "loss": 0.6654,
      "step": 13591
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.575355529785156,
      "learning_rate": 1.6796047767115298e-08,
      "loss": 1.0389,
      "step": 13592
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.478656768798828,
      "learning_rate": 1.6701073750611807e-08,
      "loss": 0.5712,
      "step": 13593
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.956381797790527,
      "learning_rate": 1.660636856422204e-08,
      "loss": 0.8393,
      "step": 13594
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3869800567626953,
      "learning_rate": 1.6511932213055247e-08,
      "loss": 0.1467,
      "step": 13595
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2151997089385986,
      "learning_rate": 1.6417764702206797e-08,
      "loss": 0.0791,
      "step": 13596
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7428278923034668,
      "learning_rate": 1.6323866036755954e-08,
      "loss": 0.0276,
      "step": 13597
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0845232009887695,
      "learning_rate": 1.6230236221768114e-08,
      "loss": 0.0721,
      "step": 13598
    },
    {
      "epoch": 0.98,
      "grad_norm": 14.383596420288086,
      "learning_rate": 1.6136875262294238e-08,
      "loss": 0.6562,
      "step": 13599
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.94759464263916,
      "learning_rate": 1.604378316337141e-08,
      "loss": 0.4236,
      "step": 13600
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.800736427307129,
      "learning_rate": 1.595095993002116e-08,
      "loss": 0.7538,
      "step": 13601
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0176477432250977,
      "learning_rate": 1.5858405567250602e-08,
      "loss": 0.045,
      "step": 13602
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.230690002441406,
      "learning_rate": 1.5766120080053516e-08,
      "loss": 0.5235,
      "step": 13603
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.5342047214508057,
      "learning_rate": 1.5674103473407586e-08,
      "loss": 0.1458,
      "step": 13604
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.911174774169922,
      "learning_rate": 1.558235575227718e-08,
      "loss": 0.9012,
      "step": 13605
    },
    {
      "epoch": 0.98,
      "grad_norm": 15.299276351928711,
      "learning_rate": 1.5490876921611664e-08,
      "loss": 0.3875,
      "step": 13606
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.01854944229126,
      "learning_rate": 1.5399666986345983e-08,
      "loss": 0.3816,
      "step": 13607
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1227009296417236,
      "learning_rate": 1.5308725951400095e-08,
      "loss": 0.075,
      "step": 13608
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.413987159729004,
      "learning_rate": 1.5218053821680624e-08,
      "loss": 0.51,
      "step": 13609
    },
    {
      "epoch": 0.98,
      "grad_norm": 13.584870338439941,
      "learning_rate": 1.5127650602078658e-08,
      "loss": 0.7089,
      "step": 13610
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0707638263702393,
      "learning_rate": 1.503751629747141e-08,
      "loss": 0.0793,
      "step": 13611
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.326313018798828,
      "learning_rate": 1.49476509127211e-08,
      "loss": 0.0954,
      "step": 13612
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.307718753814697,
      "learning_rate": 1.4858054452675519e-08,
      "loss": 0.746,
      "step": 13613
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9362801313400269,
      "learning_rate": 1.4768726922168019e-08,
      "loss": 0.0778,
      "step": 13614
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9979145526885986,
      "learning_rate": 1.4679668326017527e-08,
      "loss": 0.219,
      "step": 13615
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.502131938934326,
      "learning_rate": 1.459087866902853e-08,
      "loss": 0.6643,
      "step": 13616
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.4641900062561035,
      "learning_rate": 1.4502357955991086e-08,
      "loss": 0.0697,
      "step": 13617
    },
    {
      "epoch": 0.98,
      "grad_norm": 12.209504127502441,
      "learning_rate": 1.4414106191679711e-08,
      "loss": 0.3178,
      "step": 13618
    },
    {
      "epoch": 0.98,
      "grad_norm": 13.061986923217773,
      "learning_rate": 1.4326123380856149e-08,
      "loss": 0.674,
      "step": 13619
    },
    {
      "epoch": 0.98,
      "grad_norm": 12.966331481933594,
      "learning_rate": 1.4238409528266606e-08,
      "loss": 0.2616,
      "step": 13620
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.802252769470215,
      "learning_rate": 1.4150964638642296e-08,
      "loss": 1.1512,
      "step": 13621
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.568675994873047,
      "learning_rate": 1.4063788716701665e-08,
      "loss": 0.8455,
      "step": 13622
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.563505172729492,
      "learning_rate": 1.3976881767145955e-08,
      "loss": 0.6358,
      "step": 13623
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.410744667053223,
      "learning_rate": 1.3890243794665303e-08,
      "loss": 0.7839,
      "step": 13624
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.614613652229309,
      "learning_rate": 1.3803874803932083e-08,
      "loss": 0.0302,
      "step": 13625
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9011497497558594,
      "learning_rate": 1.3717774799605898e-08,
      "loss": 0.0619,
      "step": 13626
    },
    {
      "epoch": 0.98,
      "grad_norm": 12.85091495513916,
      "learning_rate": 1.3631943786331925e-08,
      "loss": 0.6005,
      "step": 13627
    },
    {
      "epoch": 0.98,
      "grad_norm": 17.435504913330078,
      "learning_rate": 1.3546381768739791e-08,
      "loss": 0.446,
      "step": 13628
    },
    {
      "epoch": 0.98,
      "grad_norm": 14.837423324584961,
      "learning_rate": 1.3461088751445806e-08,
      "loss": 0.5064,
      "step": 13629
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.453209161758423,
      "learning_rate": 1.3376064739051287e-08,
      "loss": 0.0541,
      "step": 13630
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.7635281085968018,
      "learning_rate": 1.3291309736142566e-08,
      "loss": 0.1106,
      "step": 13631
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.659672260284424,
      "learning_rate": 1.3206823747291542e-08,
      "loss": 0.8819,
      "step": 13632
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.131722927093506,
      "learning_rate": 1.312260677705679e-08,
      "loss": 0.1792,
      "step": 13633
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3942151069641113,
      "learning_rate": 1.3038658829980788e-08,
      "loss": 0.0945,
      "step": 13634
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.694629669189453,
      "learning_rate": 1.295497991059269e-08,
      "loss": 1.0013,
      "step": 13635
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.54229736328125,
      "learning_rate": 1.287157002340611e-08,
      "loss": 0.5793,
      "step": 13636
    },
    {
      "epoch": 0.98,
      "grad_norm": 18.277883529663086,
      "learning_rate": 1.2788429172921336e-08,
      "loss": 0.548,
      "step": 13637
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.278923988342285,
      "learning_rate": 1.270555736362311e-08,
      "loss": 0.677,
      "step": 13638
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3369548320770264,
      "learning_rate": 1.2622954599982307e-08,
      "loss": 0.0853,
      "step": 13639
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.719616174697876,
      "learning_rate": 1.2540620886454801e-08,
      "loss": 0.082,
      "step": 13640
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.409100532531738,
      "learning_rate": 1.2458556227482044e-08,
      "loss": 0.7584,
      "step": 13641
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.6556942462921143,
      "learning_rate": 1.2376760627491047e-08,
      "loss": 0.1521,
      "step": 13642
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.673576354980469,
      "learning_rate": 1.2295234090894947e-08,
      "loss": 0.8113,
      "step": 13643
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.956605434417725,
      "learning_rate": 1.2213976622091895e-08,
      "loss": 0.8114,
      "step": 13644
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1702284812927246,
      "learning_rate": 1.2132988225464492e-08,
      "loss": 0.0757,
      "step": 13645
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.766162157058716,
      "learning_rate": 1.2052268905382025e-08,
      "loss": 0.1217,
      "step": 13646
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.608735084533691,
      "learning_rate": 1.1971818666199898e-08,
      "loss": 0.5215,
      "step": 13647
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.6163911819458,
      "learning_rate": 1.189163751225686e-08,
      "loss": 0.4151,
      "step": 13648
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.68365478515625,
      "learning_rate": 1.1811725447879453e-08,
      "loss": 0.7131,
      "step": 13649
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.63632345199585,
      "learning_rate": 1.1732082477378114e-08,
      "loss": 0.5152,
      "step": 13650
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2002856731414795,
      "learning_rate": 1.1652708605048857e-08,
      "loss": 0.0671,
      "step": 13651
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.144331455230713,
      "learning_rate": 1.1573603835174363e-08,
      "loss": 0.0813,
      "step": 13652
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.2715678215026855,
      "learning_rate": 1.1494768172021776e-08,
      "loss": 0.165,
      "step": 13653
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.193839073181152,
      "learning_rate": 1.1416201619843802e-08,
      "loss": 0.864,
      "step": 13654
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.2494537830352783,
      "learning_rate": 1.1337904182879278e-08,
      "loss": 0.0838,
      "step": 13655
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.127577781677246,
      "learning_rate": 1.1259875865350932e-08,
      "loss": 0.5582,
      "step": 13656
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.8672235012054443,
      "learning_rate": 1.118211667146929e-08,
      "loss": 0.1486,
      "step": 13657
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.628582000732422,
      "learning_rate": 1.1104626605428769e-08,
      "loss": 0.1004,
      "step": 13658
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.7583765983581543,
      "learning_rate": 1.1027405671409919e-08,
      "loss": 0.1136,
      "step": 13659
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.670872211456299,
      "learning_rate": 1.0950453873577738e-08,
      "loss": 0.3444,
      "step": 13660
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.873518943786621,
      "learning_rate": 1.0873771216084461e-08,
      "loss": 0.3609,
      "step": 13661
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.8963582515716553,
      "learning_rate": 1.0797357703066225e-08,
      "loss": 0.7639,
      "step": 13662
    },
    {
      "epoch": 0.98,
      "grad_norm": 18.89845848083496,
      "learning_rate": 1.0721213338645287e-08,
      "loss": 1.1545,
      "step": 13663
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.5454282760620117,
      "learning_rate": 1.0645338126929473e-08,
      "loss": 0.0996,
      "step": 13664
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.7189886569976807,
      "learning_rate": 1.0569732072012174e-08,
      "loss": 0.1036,
      "step": 13665
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.643820762634277,
      "learning_rate": 1.0494395177971239e-08,
      "loss": 0.7571,
      "step": 13666
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1379642486572266,
      "learning_rate": 1.0419327448871752e-08,
      "loss": 0.08,
      "step": 13667
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.217761993408203,
      "learning_rate": 1.034452888876325e-08,
      "loss": 0.9153,
      "step": 13668
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.4416074752807617,
      "learning_rate": 1.0269999501680284e-08,
      "loss": 0.0931,
      "step": 13669
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.335707902908325,
      "learning_rate": 1.0195739291643526e-08,
      "loss": 0.0423,
      "step": 13670
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.7003023624420166,
      "learning_rate": 1.0121748262659215e-08,
      "loss": 0.1583,
      "step": 13671
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.062264919281006,
      "learning_rate": 1.0048026418719159e-08,
      "loss": 0.0574,
      "step": 13672
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.470688581466675,
      "learning_rate": 9.974573763799621e-09,
      "loss": 0.6269,
      "step": 13673
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.85768461227417,
      "learning_rate": 9.9013903018641e-09,
      "loss": 0.1079,
      "step": 13674
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.182077407836914,
      "learning_rate": 9.828476036859436e-09,
      "loss": 0.7223,
      "step": 13675
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.21134614944458,
      "learning_rate": 9.755830972720259e-09,
      "loss": 0.0604,
      "step": 13676
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.2425696849823,
      "learning_rate": 9.683455113364548e-09,
      "loss": 0.1059,
      "step": 13677
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.987929821014404,
      "learning_rate": 9.611348462696956e-09,
      "loss": 0.5366,
      "step": 13678
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.494348526000977,
      "learning_rate": 9.539511024607706e-09,
      "loss": 0.9777,
      "step": 13679
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.560645580291748,
      "learning_rate": 9.46794280297203e-09,
      "loss": 0.1343,
      "step": 13680
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.6957437992095947,
      "learning_rate": 9.396643801650174e-09,
      "loss": 0.7051,
      "step": 13681
    },
    {
      "epoch": 0.98,
      "grad_norm": 17.329374313354492,
      "learning_rate": 9.32561402448906e-09,
      "loss": 0.708,
      "step": 13682
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.810399055480957,
      "learning_rate": 9.254853475320623e-09,
      "loss": 0.3726,
      "step": 13683
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.5332493782043457,
      "learning_rate": 9.184362157961812e-09,
      "loss": 0.0827,
      "step": 13684
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3045601844787598,
      "learning_rate": 9.11414007621514e-09,
      "loss": 0.5779,
      "step": 13685
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3400344848632812,
      "learning_rate": 9.04418723386924e-09,
      "loss": 0.0733,
      "step": 13686
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.420848846435547,
      "learning_rate": 8.974503634697762e-09,
      "loss": 0.0893,
      "step": 13687
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.171825647354126,
      "learning_rate": 8.905089282459923e-09,
      "loss": 0.0818,
      "step": 13688
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.373344421386719,
      "learning_rate": 8.835944180899947e-09,
      "loss": 0.3776,
      "step": 13689
    },
    {
      "epoch": 0.98,
      "grad_norm": 12.257493019104004,
      "learning_rate": 8.767068333748185e-09,
      "loss": 0.6988,
      "step": 13690
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.626416206359863,
      "learning_rate": 8.698461744719999e-09,
      "loss": 0.6113,
      "step": 13691
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.7334365844726562,
      "learning_rate": 8.630124417516871e-09,
      "loss": 0.1354,
      "step": 13692
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9877426624298096,
      "learning_rate": 8.562056355825298e-09,
      "loss": 0.182,
      "step": 13693
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.924535751342773,
      "learning_rate": 8.494257563317343e-09,
      "loss": 0.5003,
      "step": 13694
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.487610816955566,
      "learning_rate": 8.426728043650079e-09,
      "loss": 0.4744,
      "step": 13695
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.583172082901001,
      "learning_rate": 8.359467800466703e-09,
      "loss": 0.6618,
      "step": 13696
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.1789040565490723,
      "learning_rate": 8.292476837395425e-09,
      "loss": 0.1288,
      "step": 13697
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.8448646068573,
      "learning_rate": 8.225755158050574e-09,
      "loss": 0.0947,
      "step": 13698
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.732783079147339,
      "learning_rate": 8.159302766031497e-09,
      "loss": 0.1166,
      "step": 13699
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.6630704402923584,
      "learning_rate": 8.093119664922544e-09,
      "loss": 0.0934,
      "step": 13700
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1390907764434814,
      "learning_rate": 8.02720585829475e-09,
      "loss": 0.0936,
      "step": 13701
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.2340402603149414,
      "learning_rate": 7.961561349703607e-09,
      "loss": 0.1853,
      "step": 13702
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.141426086425781,
      "learning_rate": 7.896186142690166e-09,
      "loss": 0.8266,
      "step": 13703
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8425699472427368,
      "learning_rate": 7.831080240781053e-09,
      "loss": 0.0782,
      "step": 13704
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.757672309875488,
      "learning_rate": 7.766243647489013e-09,
      "loss": 0.7278,
      "step": 13705
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.2221860885620117,
      "learning_rate": 7.701676366311805e-09,
      "loss": 0.08,
      "step": 13706
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.348750352859497,
      "learning_rate": 7.637378400732198e-09,
      "loss": 0.1487,
      "step": 13707
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2980780601501465,
      "learning_rate": 7.57334975421853e-09,
      "loss": 0.1137,
      "step": 13708
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.913100481033325,
      "learning_rate": 7.509590430225255e-09,
      "loss": 0.0999,
      "step": 13709
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1919238567352295,
      "learning_rate": 7.446100432192404e-09,
      "loss": 0.0675,
      "step": 13710
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.7017862796783447,
      "learning_rate": 7.382879763544459e-09,
      "loss": 0.144,
      "step": 13711
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.369493246078491,
      "learning_rate": 7.319928427692025e-09,
      "loss": 0.1474,
      "step": 13712
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.970982074737549,
      "learning_rate": 7.2572464280312725e-09,
      "loss": 0.0876,
      "step": 13713
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.6750545501708984,
      "learning_rate": 7.194833767943388e-09,
      "loss": 0.7161,
      "step": 13714
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.8390228748321533,
      "learning_rate": 7.132690450795121e-09,
      "loss": 0.1667,
      "step": 13715
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3978397846221924,
      "learning_rate": 7.070816479939347e-09,
      "loss": 0.2254,
      "step": 13716
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.917428016662598,
      "learning_rate": 7.009211858713949e-09,
      "loss": 0.5902,
      "step": 13717
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.243250846862793,
      "learning_rate": 6.947876590442382e-09,
      "loss": 0.0879,
      "step": 13718
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.600072383880615,
      "learning_rate": 6.886810678432554e-09,
      "loss": 0.3145,
      "step": 13719
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.06687068939209,
      "learning_rate": 6.826014125979608e-09,
      "loss": 0.5738,
      "step": 13720
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.8919451236724854,
      "learning_rate": 6.765486936362586e-09,
      "loss": 0.177,
      "step": 13721
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8921620845794678,
      "learning_rate": 6.7052291128477665e-09,
      "loss": 0.0477,
      "step": 13722
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.181661367416382,
      "learning_rate": 6.645240658684771e-09,
      "loss": 0.0578,
      "step": 13723
    },
    {
      "epoch": 0.98,
      "grad_norm": 17.475008010864258,
      "learning_rate": 6.585521577110454e-09,
      "loss": 0.4871,
      "step": 13724
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.791187286376953,
      "learning_rate": 6.526071871345574e-09,
      "loss": 0.8047,
      "step": 13725
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3147947788238525,
      "learning_rate": 6.466891544598675e-09,
      "loss": 0.093,
      "step": 13726
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.0729217529296875,
      "learning_rate": 6.407980600060537e-09,
      "loss": 0.1713,
      "step": 13727
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.262119293212891,
      "learning_rate": 6.34933904091084e-09,
      "loss": 0.3351,
      "step": 13728
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.928440093994141,
      "learning_rate": 6.2909668703120545e-09,
      "loss": 0.8062,
      "step": 13729
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.719803810119629,
      "learning_rate": 6.232864091412772e-09,
      "loss": 0.6457,
      "step": 13730
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.9992222785949707,
      "learning_rate": 6.175030707348817e-09,
      "loss": 0.6677,
      "step": 13731
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.117166996002197,
      "learning_rate": 6.117466721238807e-09,
      "loss": 0.8022,
      "step": 13732
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9956028461456299,
      "learning_rate": 6.0601721361891464e-09,
      "loss": 0.0798,
      "step": 13733
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.101947784423828,
      "learning_rate": 6.003146955289585e-09,
      "loss": 0.0867,
      "step": 13734
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.984440803527832,
      "learning_rate": 5.946391181617106e-09,
      "loss": 0.5205,
      "step": 13735
    },
    {
      "epoch": 0.98,
      "grad_norm": 23.639013290405273,
      "learning_rate": 5.889904818233705e-09,
      "loss": 0.6578,
      "step": 13736
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.998614549636841,
      "learning_rate": 5.833687868185833e-09,
      "loss": 0.1212,
      "step": 13737
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.301118850708008,
      "learning_rate": 5.7777403345066206e-09,
      "loss": 0.8633,
      "step": 13738
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9747647047042847,
      "learning_rate": 5.722062220214209e-09,
      "loss": 0.0743,
      "step": 13739
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.750796318054199,
      "learning_rate": 5.6666535283117495e-09,
      "loss": 0.7083,
      "step": 13740
    },
    {
      "epoch": 0.99,
      "grad_norm": 24.579769134521484,
      "learning_rate": 5.611514261789075e-09,
      "loss": 0.4714,
      "step": 13741
    },
    {
      "epoch": 0.99,
      "grad_norm": 15.359683990478516,
      "learning_rate": 5.556644423620472e-09,
      "loss": 0.5446,
      "step": 13742
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7298510074615479,
      "learning_rate": 5.502044016765795e-09,
      "loss": 0.0445,
      "step": 13743
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7242867946624756,
      "learning_rate": 5.447713044170466e-09,
      "loss": 0.2034,
      "step": 13744
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.88454270362854,
      "learning_rate": 5.393651508766029e-09,
      "loss": 0.8559,
      "step": 13745
    },
    {
      "epoch": 0.99,
      "grad_norm": 16.889101028442383,
      "learning_rate": 5.339859413468485e-09,
      "loss": 1.2398,
      "step": 13746
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.253692150115967,
      "learning_rate": 5.2863367611794005e-09,
      "loss": 0.1241,
      "step": 13747
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.816033124923706,
      "learning_rate": 5.233083554786466e-09,
      "loss": 0.5647,
      "step": 13748
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.962874412536621,
      "learning_rate": 5.18009979716294e-09,
      "loss": 0.1338,
      "step": 13749
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.515859842300415,
      "learning_rate": 5.127385491165981e-09,
      "loss": 0.063,
      "step": 13750
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8865690231323242,
      "learning_rate": 5.074940639639981e-09,
      "loss": 0.0505,
      "step": 13751
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.542832612991333,
      "learning_rate": 5.022765245414341e-09,
      "loss": 0.1585,
      "step": 13752
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.485781192779541,
      "learning_rate": 4.970859311303478e-09,
      "loss": 0.1388,
      "step": 13753
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.725495457649231,
      "learning_rate": 4.919222840107929e-09,
      "loss": 0.0556,
      "step": 13754
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.4797635078430176,
      "learning_rate": 4.867855834612134e-09,
      "loss": 0.1204,
      "step": 13755
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2729439735412598,
      "learning_rate": 4.8167582975883195e-09,
      "loss": 0.151,
      "step": 13756
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.053509712219238,
      "learning_rate": 4.7659302317926144e-09,
      "loss": 0.6728,
      "step": 13757
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.209465980529785,
      "learning_rate": 4.71537163996727e-09,
      "loss": 0.8204,
      "step": 13758
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.5028235912323,
      "learning_rate": 4.665082524838993e-09,
      "loss": 0.1014,
      "step": 13759
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.699234962463379,
      "learning_rate": 4.615062889121169e-09,
      "loss": 0.8562,
      "step": 13760
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1621224880218506,
      "learning_rate": 4.565312735512195e-09,
      "loss": 0.0762,
      "step": 13761
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.6439368724823,
      "learning_rate": 4.515832066696035e-09,
      "loss": 0.0864,
      "step": 13762
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.7439374923706055,
      "learning_rate": 4.466620885341666e-09,
      "loss": 0.6057,
      "step": 13763
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.076844215393066,
      "learning_rate": 4.4176791941041855e-09,
      "loss": 0.3948,
      "step": 13764
    },
    {
      "epoch": 0.99,
      "grad_norm": 13.18912124633789,
      "learning_rate": 4.369006995623148e-09,
      "loss": 0.6884,
      "step": 13765
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.235793113708496,
      "learning_rate": 4.3206042925253436e-09,
      "loss": 0.0878,
      "step": 13766
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.4657187461853027,
      "learning_rate": 4.272471087420904e-09,
      "loss": 0.729,
      "step": 13767
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.161090850830078,
      "learning_rate": 4.224607382906642e-09,
      "loss": 0.6749,
      "step": 13768
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.9949655532836914,
      "learning_rate": 4.1770131815649376e-09,
      "loss": 0.0994,
      "step": 13769
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.86389684677124,
      "learning_rate": 4.129688485963179e-09,
      "loss": 0.4568,
      "step": 13770
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.3400585651397705,
      "learning_rate": 4.082633298654326e-09,
      "loss": 0.1094,
      "step": 13771
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.8705387115478516,
      "learning_rate": 4.035847622176347e-09,
      "loss": 0.109,
      "step": 13772
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.128385543823242,
      "learning_rate": 3.989331459054446e-09,
      "loss": 0.038,
      "step": 13773
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.286087036132812,
      "learning_rate": 3.943084811796616e-09,
      "loss": 0.8424,
      "step": 13774
    },
    {
      "epoch": 0.99,
      "grad_norm": 12.794748306274414,
      "learning_rate": 3.897107682898638e-09,
      "loss": 0.38,
      "step": 13775
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.680763244628906,
      "learning_rate": 3.851400074840195e-09,
      "loss": 0.3181,
      "step": 13776
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2583985328674316,
      "learning_rate": 3.805961990087093e-09,
      "loss": 0.0579,
      "step": 13777
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.531047821044922,
      "learning_rate": 3.760793431091258e-09,
      "loss": 1.1397,
      "step": 13778
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7882952690124512,
      "learning_rate": 3.715894400288522e-09,
      "loss": 0.0564,
      "step": 13779
    },
    {
      "epoch": 0.99,
      "grad_norm": 20.41424560546875,
      "learning_rate": 3.671264900100835e-09,
      "loss": 1.1185,
      "step": 13780
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.5370371341705322,
      "learning_rate": 3.6269049329368256e-09,
      "loss": 0.1005,
      "step": 13781
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.836493492126465,
      "learning_rate": 3.5828145011890247e-09,
      "loss": 0.1866,
      "step": 13782
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0662710666656494,
      "learning_rate": 3.5389936072360853e-09,
      "loss": 0.0893,
      "step": 13783
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.9502553939819336,
      "learning_rate": 3.495442253441117e-09,
      "loss": 0.6331,
      "step": 13784
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4814884662628174,
      "learning_rate": 3.452160442155017e-09,
      "loss": 0.0591,
      "step": 13785
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7927656173706055,
      "learning_rate": 3.409148175711474e-09,
      "loss": 0.1385,
      "step": 13786
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0345304012298584,
      "learning_rate": 3.36640545643141e-09,
      "loss": 0.0542,
      "step": 13787
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.098893642425537,
      "learning_rate": 3.323932286620202e-09,
      "loss": 0.7899,
      "step": 13788
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.803431510925293,
      "learning_rate": 3.2817286685699057e-09,
      "loss": 0.1201,
      "step": 13789
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.830631971359253,
      "learning_rate": 3.2397946045559237e-09,
      "loss": 0.1304,
      "step": 13790
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.5684962272644043,
      "learning_rate": 3.1981300968420002e-09,
      "loss": 0.0727,
      "step": 13791
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.149471282958984,
      "learning_rate": 3.156735147674672e-09,
      "loss": 0.5658,
      "step": 13792
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.83180046081543,
      "learning_rate": 3.1156097592877077e-09,
      "loss": 0.566,
      "step": 13793
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.9470160007476807,
      "learning_rate": 3.0747539338993326e-09,
      "loss": 0.0892,
      "step": 13794
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.399060249328613,
      "learning_rate": 3.0341676737133397e-09,
      "loss": 0.9138,
      "step": 13795
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0310585498809814,
      "learning_rate": 2.9938509809196435e-09,
      "loss": 0.0594,
      "step": 13796
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.262603759765625,
      "learning_rate": 2.9538038576926166e-09,
      "loss": 0.6674,
      "step": 13797
    },
    {
      "epoch": 0.99,
      "grad_norm": 19.712434768676758,
      "learning_rate": 2.9140263061938624e-09,
      "loss": 0.6901,
      "step": 13798
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.8724253177642822,
      "learning_rate": 2.874518328567777e-09,
      "loss": 0.1282,
      "step": 13799
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.236027240753174,
      "learning_rate": 2.8352799269459884e-09,
      "loss": 0.7935,
      "step": 13800
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.749547004699707,
      "learning_rate": 2.7963111034462476e-09,
      "loss": 0.1333,
      "step": 13801
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.475844383239746,
      "learning_rate": 2.757611860169651e-09,
      "loss": 0.0624,
      "step": 13802
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.672628402709961,
      "learning_rate": 2.7191821992045285e-09,
      "loss": 0.8098,
      "step": 13803
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.626749038696289,
      "learning_rate": 2.6810221226236666e-09,
      "loss": 0.747,
      "step": 13804
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9151126146316528,
      "learning_rate": 2.643131632485418e-09,
      "loss": 0.0446,
      "step": 13805
    },
    {
      "epoch": 0.99,
      "grad_norm": 20.887409210205078,
      "learning_rate": 2.605510730834815e-09,
      "loss": 0.7658,
      "step": 13806
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.580851554870605,
      "learning_rate": 2.5681594197007888e-09,
      "loss": 0.7967,
      "step": 13807
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.686222076416016,
      "learning_rate": 2.5310777010978395e-09,
      "loss": 0.3708,
      "step": 13808
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.701013565063477,
      "learning_rate": 2.494265577026589e-09,
      "loss": 0.4722,
      "step": 13809
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.588240623474121,
      "learning_rate": 2.4577230494737814e-09,
      "loss": 0.8349,
      "step": 13810
    },
    {
      "epoch": 0.99,
      "grad_norm": 14.16977596282959,
      "learning_rate": 2.4214501204095076e-09,
      "loss": 0.77,
      "step": 13811
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.773021936416626,
      "learning_rate": 2.3854467917916457e-09,
      "loss": 0.3007,
      "step": 13812
    },
    {
      "epoch": 0.99,
      "grad_norm": 57.15290832519531,
      "learning_rate": 2.3497130655614208e-09,
      "loss": 0.6548,
      "step": 13813
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8936787843704224,
      "learning_rate": 2.3142489436472906e-09,
      "loss": 0.0934,
      "step": 13814
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.2696754932403564,
      "learning_rate": 2.2790544279621686e-09,
      "loss": 0.0889,
      "step": 13815
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.3581349849700928,
      "learning_rate": 2.2441295204050916e-09,
      "loss": 0.1535,
      "step": 13816
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.104617118835449,
      "learning_rate": 2.2094742228589984e-09,
      "loss": 0.2131,
      "step": 13817
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0834619998931885,
      "learning_rate": 2.175088537194614e-09,
      "loss": 0.0496,
      "step": 13818
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.173983097076416,
      "learning_rate": 2.1409724652660114e-09,
      "loss": 0.8199,
      "step": 13819
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.771076679229736,
      "learning_rate": 2.10712600891394e-09,
      "loss": 0.8516,
      "step": 13820
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.993013620376587,
      "learning_rate": 2.073549169964717e-09,
      "loss": 0.187,
      "step": 13821
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.8815982341766357,
      "learning_rate": 2.040241950229671e-09,
      "loss": 0.6452,
      "step": 13822
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.708486318588257,
      "learning_rate": 2.007204351504588e-09,
      "loss": 0.1185,
      "step": 13823
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1553127765655518,
      "learning_rate": 1.974436375572486e-09,
      "loss": 0.0868,
      "step": 13824
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.807889223098755,
      "learning_rate": 1.9419380242013954e-09,
      "loss": 0.1822,
      "step": 13825
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.723636627197266,
      "learning_rate": 1.9097092991438025e-09,
      "loss": 0.6749,
      "step": 13826
    },
    {
      "epoch": 0.99,
      "grad_norm": 16.118452072143555,
      "learning_rate": 1.8777502021383174e-09,
      "loss": 0.6878,
      "step": 13827
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.575288772583008,
      "learning_rate": 1.8460607349096715e-09,
      "loss": 0.738,
      "step": 13828
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2864036560058594,
      "learning_rate": 1.8146408991670528e-09,
      "loss": 0.1348,
      "step": 13829
    },
    {
      "epoch": 0.99,
      "grad_norm": 12.638891220092773,
      "learning_rate": 1.7834906966052169e-09,
      "loss": 0.3341,
      "step": 13830
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.1756579875946045,
      "learning_rate": 1.7526101289050411e-09,
      "loss": 0.1129,
      "step": 13831
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2995598316192627,
      "learning_rate": 1.7219991977318606e-09,
      "loss": 0.1069,
      "step": 13832
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0358097553253174,
      "learning_rate": 1.6916579047371317e-09,
      "loss": 0.1498,
      "step": 13833
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.324230670928955,
      "learning_rate": 1.6615862515584336e-09,
      "loss": 0.1579,
      "step": 13834
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.799437522888184,
      "learning_rate": 1.6317842398172468e-09,
      "loss": 0.6294,
      "step": 13835
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.502458572387695,
      "learning_rate": 1.6022518711211742e-09,
      "loss": 0.8059,
      "step": 13836
    },
    {
      "epoch": 0.99,
      "grad_norm": 13.53696346282959,
      "learning_rate": 1.5729891470633863e-09,
      "loss": 0.6521,
      "step": 13837
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.0530829429626465,
      "learning_rate": 1.5439960692231747e-09,
      "loss": 0.6908,
      "step": 13838
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1590118408203125,
      "learning_rate": 1.5152726391642891e-09,
      "loss": 0.0857,
      "step": 13839
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.240720748901367,
      "learning_rate": 1.4868188584360454e-09,
      "loss": 0.1167,
      "step": 13840
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7745473384857178,
      "learning_rate": 1.4586347285733272e-09,
      "loss": 0.1581,
      "step": 13841
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.898955821990967,
      "learning_rate": 1.4307202510971396e-09,
      "loss": 0.8394,
      "step": 13842
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.918333053588867,
      "learning_rate": 1.4030754275129454e-09,
      "loss": 0.3676,
      "step": 13843
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7091848850250244,
      "learning_rate": 1.3757002593117742e-09,
      "loss": 0.1143,
      "step": 13844
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.2341277599334717,
      "learning_rate": 1.3485947479707773e-09,
      "loss": 0.0665,
      "step": 13845
    },
    {
      "epoch": 0.99,
      "grad_norm": 22.178180694580078,
      "learning_rate": 1.3217588949526738e-09,
      "loss": 0.5348,
      "step": 13846
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.1698737144470215,
      "learning_rate": 1.2951927017040845e-09,
      "loss": 0.6329,
      "step": 13847
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.304154872894287,
      "learning_rate": 1.268896169658862e-09,
      "loss": 0.1127,
      "step": 13848
    },
    {
      "epoch": 0.99,
      "grad_norm": 14.027241706848145,
      "learning_rate": 1.2428693002358716e-09,
      "loss": 0.8104,
      "step": 13849
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.9803571701049805,
      "learning_rate": 1.2171120948384352e-09,
      "loss": 0.1465,
      "step": 13850
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.9876394271850586,
      "learning_rate": 1.1916245548559967e-09,
      "loss": 0.188,
      "step": 13851
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7726457118988037,
      "learning_rate": 1.166406681664678e-09,
      "loss": 0.0842,
      "step": 13852
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.572612285614014,
      "learning_rate": 1.1414584766233916e-09,
      "loss": 0.6102,
      "step": 13853
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.635838508605957,
      "learning_rate": 1.1167799410793934e-09,
      "loss": 0.8531,
      "step": 13854
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.527207374572754,
      "learning_rate": 1.0923710763627304e-09,
      "loss": 1.0701,
      "step": 13855
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.314077615737915,
      "learning_rate": 1.0682318837906824e-09,
      "loss": 0.0992,
      "step": 13856
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.726486682891846,
      "learning_rate": 1.0443623646660961e-09,
      "loss": 0.8369,
      "step": 13857
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.465792179107666,
      "learning_rate": 1.02076252027572e-09,
      "loss": 0.1042,
      "step": 13858
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.578271865844727,
      "learning_rate": 9.974323518929795e-10,
      "loss": 0.8839,
      "step": 13859
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.0544610023498535,
      "learning_rate": 9.743718607768682e-10,
      "loss": 0.5742,
      "step": 13860
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.922205686569214,
      "learning_rate": 9.515810481708353e-10,
      "loss": 0.102,
      "step": 13861
    },
    {
      "epoch": 0.99,
      "grad_norm": 12.982231140136719,
      "learning_rate": 9.290599153044533e-10,
      "loss": 0.541,
      "step": 13862
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.044224739074707,
      "learning_rate": 9.06808463393416e-10,
      "loss": 0.7354,
      "step": 13863
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.5100128650665283,
      "learning_rate": 8.848266936367645e-10,
      "loss": 0.0998,
      "step": 13864
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.9293482303619385,
      "learning_rate": 8.63114607221327e-10,
      "loss": 0.9334,
      "step": 13865
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.18876838684082,
      "learning_rate": 8.416722053178339e-10,
      "loss": 0.5159,
      "step": 13866
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.377115249633789,
      "learning_rate": 8.204994890836926e-10,
      "loss": 0.2513,
      "step": 13867
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.6696789264678955,
      "learning_rate": 7.995964596602124e-10,
      "loss": 0.1136,
      "step": 13868
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.253706455230713,
      "learning_rate": 7.789631181759349e-10,
      "loss": 0.1151,
      "step": 13869
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.562439441680908,
      "learning_rate": 7.585994657433038e-10,
      "loss": 0.2314,
      "step": 13870
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.123188972473145,
      "learning_rate": 7.385055034608846e-10,
      "loss": 0.4053,
      "step": 13871
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.029839515686035,
      "learning_rate": 7.186812324133651e-10,
      "loss": 0.4725,
      "step": 13872
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.6218433380126953,
      "learning_rate": 6.991266536693353e-10,
      "loss": 0.1385,
      "step": 13873
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.613529682159424,
      "learning_rate": 6.798417682840619e-10,
      "loss": 0.7473,
      "step": 13874
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.6862828731536865,
      "learning_rate": 6.608265772978239e-10,
      "loss": 0.145,
      "step": 13875
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.779024124145508,
      "learning_rate": 6.420810817364676e-10,
      "loss": 0.7064,
      "step": 13876
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.3685226440429688,
      "learning_rate": 6.236052826114058e-10,
      "loss": 0.0495,
      "step": 13877
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.276646852493286,
      "learning_rate": 6.05399180919064e-10,
      "loss": 0.2293,
      "step": 13878
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.224526405334473,
      "learning_rate": 5.874627776419895e-10,
      "loss": 0.686,
      "step": 13879
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0033342838287354,
      "learning_rate": 5.697960737471864e-10,
      "loss": 0.0569,
      "step": 13880
    },
    {
      "epoch": 1.0,
      "grad_norm": 13.194517135620117,
      "learning_rate": 5.523990701883364e-10,
      "loss": 0.5843,
      "step": 13881
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4211230278015137,
      "learning_rate": 5.352717679030228e-10,
      "loss": 0.0408,
      "step": 13882
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.803957939147949,
      "learning_rate": 5.184141678160614e-10,
      "loss": 0.4674,
      "step": 13883
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.841874122619629,
      "learning_rate": 5.01826270836725e-10,
      "loss": 0.3644,
      "step": 13884
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.86722469329834,
      "learning_rate": 4.855080778598531e-10,
      "loss": 0.4871,
      "step": 13885
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.890770435333252,
      "learning_rate": 4.694595897652976e-10,
      "loss": 0.0852,
      "step": 13886
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.157498359680176,
      "learning_rate": 4.536808074190324e-10,
      "loss": 0.453,
      "step": 13887
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.048357009887695,
      "learning_rate": 4.3817173167315374e-10,
      "loss": 0.4229,
      "step": 13888
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.064136028289795,
      "learning_rate": 4.229323633625493e-10,
      "loss": 0.7542,
      "step": 13889
    },
    {
      "epoch": 1.0,
      "grad_norm": 12.088311195373535,
      "learning_rate": 4.079627033110045e-10,
      "loss": 0.2775,
      "step": 13890
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.869192361831665,
      "learning_rate": 3.932627523250965e-10,
      "loss": 0.0569,
      "step": 13891
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.18622088432312,
      "learning_rate": 3.788325111980795e-10,
      "loss": 0.2516,
      "step": 13892
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.068212985992432,
      "learning_rate": 3.6467198070877506e-10,
      "loss": 0.4813,
      "step": 13893
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2136919498443604,
      "learning_rate": 3.507811616204615e-10,
      "loss": 0.0905,
      "step": 13894
    },
    {
      "epoch": 1.0,
      "grad_norm": 19.986127853393555,
      "learning_rate": 3.3716005468309443e-10,
      "loss": 0.662,
      "step": 13895
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.22369647026062,
      "learning_rate": 3.238086606310864e-10,
      "loss": 0.1626,
      "step": 13896
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.756117343902588,
      "learning_rate": 3.1072698018441703e-10,
      "loss": 0.0674,
      "step": 13897
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.85107707977295,
      "learning_rate": 2.979150140497433e-10,
      "loss": 0.5962,
      "step": 13898
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8278937339782715,
      "learning_rate": 2.8537276291706883e-10,
      "loss": 0.051,
      "step": 13899
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.744239091873169,
      "learning_rate": 2.7310022746418473e-10,
      "loss": 0.1746,
      "step": 13900
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.797567844390869,
      "learning_rate": 2.6109740835222887e-10,
      "loss": 0.3448,
      "step": 13901
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.357065200805664,
      "learning_rate": 2.493643062290163e-10,
      "loss": 0.4548,
      "step": 13902
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.018855571746826,
      "learning_rate": 2.379009217273742e-10,
      "loss": 0.8455,
      "step": 13903
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.527125358581543,
      "learning_rate": 2.2670725546625194e-10,
      "loss": 0.451,
      "step": 13904
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.8594071865081787,
      "learning_rate": 2.1578330804905566e-10,
      "loss": 0.1078,
      "step": 13905
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.328782320022583,
      "learning_rate": 2.0512908006475873e-10,
      "loss": 0.097,
      "step": 13906
    },
    {
      "epoch": 1.0,
      "grad_norm": 12.111705780029297,
      "learning_rate": 1.9474457208901176e-10,
      "loss": 0.4084,
      "step": 13907
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.526254892349243,
      "learning_rate": 1.8462978468081205e-10,
      "loss": 0.5363,
      "step": 13908
    },
    {
      "epoch": 1.0,
      "grad_norm": 12.88279914855957,
      "learning_rate": 1.7478471838694445e-10,
      "loss": 0.7832,
      "step": 13909
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.323476791381836,
      "learning_rate": 1.6520937373809555e-10,
      "loss": 0.079,
      "step": 13910
    },
    {
      "epoch": 1.0,
      "grad_norm": 12.518742561340332,
      "learning_rate": 1.5590375125107416e-10,
      "loss": 0.3449,
      "step": 13911
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8589764833450317,
      "learning_rate": 1.46867851427146e-10,
      "loss": 0.0502,
      "step": 13912
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.757720470428467,
      "learning_rate": 1.3810167475425406e-10,
      "loss": 0.7961,
      "step": 13913
    },
    {
      "epoch": 1.0,
      "grad_norm": 13.133455276489258,
      "learning_rate": 1.2960522170535338e-10,
      "loss": 0.4957,
      "step": 13914
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.7996506690979004,
      "learning_rate": 1.2137849273896607e-10,
      "loss": 0.6543,
      "step": 13915
    },
    {
      "epoch": 1.0,
      "grad_norm": 12.255483627319336,
      "learning_rate": 1.134214882980711e-10,
      "loss": 0.6595,
      "step": 13916
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.679261684417725,
      "learning_rate": 1.0573420881287988e-10,
      "loss": 0.3747,
      "step": 13917
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.749101400375366,
      "learning_rate": 9.831665469750562e-11,
      "loss": 0.1368,
      "step": 13918
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.084415912628174,
      "learning_rate": 9.116882635273883e-11,
      "loss": 0.3627,
      "step": 13919
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.350583076477051,
      "learning_rate": 8.429072416327177e-11,
      "loss": 0.1113,
      "step": 13920
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.9422025680542,
      "learning_rate": 7.768234850102919e-11,
      "loss": 0.4736,
      "step": 13921
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.467819213867188,
      "learning_rate": 7.134369972183753e-11,
      "loss": 0.3032,
      "step": 13922
    },
    {
      "epoch": 1.0,
      "grad_norm": 56.603919982910156,
      "learning_rate": 6.527477816764549e-11,
      "loss": 0.809,
      "step": 13923
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.478342294692993,
      "learning_rate": 5.947558416652399e-11,
      "loss": 0.0453,
      "step": 13924
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.9368531703948975,
      "learning_rate": 5.3946118030445694e-11,
      "loss": 0.8108,
      "step": 13925
    },
    {
      "epoch": 1.0,
      "grad_norm": 17.14094352722168,
      "learning_rate": 4.86863800586157e-11,
      "loss": 1.5697,
      "step": 13926
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.5758285522460938,
      "learning_rate": 4.369637053414089e-11,
      "loss": 0.1539,
      "step": 13927
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6312686204910278,
      "learning_rate": 3.897608972625033e-11,
      "loss": 0.0403,
      "step": 13928
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.252440452575684,
      "learning_rate": 3.452553788974022e-11,
      "loss": 0.7092,
      "step": 13929
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.2212042808532715,
      "learning_rate": 3.0344715264973845e-11,
      "loss": 0.7826,
      "step": 13930
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.6590139865875244,
      "learning_rate": 2.643362207677136e-11,
      "loss": 0.1082,
      "step": 13931
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1459927558898926,
      "learning_rate": 2.279225853663025e-11,
      "loss": 0.0478,
      "step": 13932
    },
    {
      "epoch": 1.0,
      "grad_norm": 18.570865631103516,
      "learning_rate": 1.942062484106e-11,
      "loss": 0.504,
      "step": 13933
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.025468826293945,
      "learning_rate": 1.6318721172137174e-11,
      "loss": 0.7234,
      "step": 13934
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.866817831993103,
      "learning_rate": 1.3486547696395235e-11,
      "loss": 0.0612,
      "step": 13935
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.9637932777404785,
      "learning_rate": 1.0924104567044957e-11,
      "loss": 0.7388,
      "step": 13936
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.9696133136749268,
      "learning_rate": 8.631391922309107e-12,
      "loss": 0.1108,
      "step": 13937
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.0580410957336426,
      "learning_rate": 6.608409886532663e-12,
      "loss": 0.0718,
      "step": 13938
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.836336851119995,
      "learning_rate": 4.855158567407259e-12,
      "loss": 0.1228,
      "step": 13939
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.46092700958252,
      "learning_rate": 3.3716380609671863e-12,
      "loss": 0.337,
      "step": 13940
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.9942779541015625,
      "learning_rate": 2.15784844603828e-12,
      "loss": 0.4261,
      "step": 13941
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.491347312927246,
      "learning_rate": 1.2137897892339211e-12,
      "loss": 0.3683,
      "step": 13942
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.8074116706848145,
      "learning_rate": 5.394621405141465e-13,
      "loss": 0.1758,
      "step": 13943
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.5910024642944336,
      "learning_rate": 1.348655370714269e-13,
      "loss": 0.1206,
      "step": 13944
    },
    {
      "epoch": 1.0,
      "step": 13944,
      "total_flos": 8.731981544790426e+17,
      "train_loss": 0.554974593159826,
      "train_runtime": 56531.5487,
      "train_samples_per_second": 3.947,
      "train_steps_per_second": 0.247
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 13945,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 3000,
  "total_flos": 8.731981544790426e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}