{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.977622377622378,
  "eval_steps": 500,
  "global_step": 1428,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004195804195804196,
      "grad_norm": 717.93115234375,
      "learning_rate": 5.0000000000000004e-08,
      "loss": 2.7062,
      "step": 1
    },
    {
      "epoch": 0.008391608391608392,
      "grad_norm": 700.4446411132812,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 2.7012,
      "step": 2
    },
    {
      "epoch": 0.012587412587412588,
      "grad_norm": 1356.7574462890625,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 2.7212,
      "step": 3
    },
    {
      "epoch": 0.016783216783216783,
      "grad_norm": 899.5720825195312,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 2.7709,
      "step": 4
    },
    {
      "epoch": 0.02097902097902098,
      "grad_norm": 875.39892578125,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 2.7834,
      "step": 5
    },
    {
      "epoch": 0.025174825174825177,
      "grad_norm": 925.267822265625,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 2.7798,
      "step": 6
    },
    {
      "epoch": 0.02937062937062937,
      "grad_norm": 71.72831726074219,
      "learning_rate": 3.5000000000000004e-07,
      "loss": 2.774,
      "step": 7
    },
    {
      "epoch": 0.033566433566433566,
      "grad_norm": 569.9453735351562,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 2.7603,
      "step": 8
    },
    {
      "epoch": 0.03776223776223776,
      "grad_norm": 384.319580078125,
      "learning_rate": 4.5000000000000003e-07,
      "loss": 2.7579,
      "step": 9
    },
    {
      "epoch": 0.04195804195804196,
      "grad_norm": 599.7899169921875,
      "learning_rate": 5.000000000000001e-07,
      "loss": 2.709,
      "step": 10
    },
    {
      "epoch": 0.046153846153846156,
      "grad_norm": 566.8663940429688,
      "learning_rate": 5.5e-07,
      "loss": 2.6726,
      "step": 11
    },
    {
      "epoch": 0.05034965034965035,
      "grad_norm": 917.609619140625,
      "learning_rate": 6.000000000000001e-07,
      "loss": 2.6404,
      "step": 12
    },
    {
      "epoch": 0.05454545454545454,
      "grad_norm": 1377.7421875,
      "learning_rate": 6.5e-07,
      "loss": 2.6995,
      "step": 13
    },
    {
      "epoch": 0.05874125874125874,
      "grad_norm": 1760.7694091796875,
      "learning_rate": 7.000000000000001e-07,
      "loss": 2.6683,
      "step": 14
    },
    {
      "epoch": 0.06293706293706294,
      "grad_norm": 1408.8568115234375,
      "learning_rate": 7.5e-07,
      "loss": 2.4465,
      "step": 15
    },
    {
      "epoch": 0.06713286713286713,
      "grad_norm": 140.02610778808594,
      "learning_rate": 8.000000000000001e-07,
      "loss": 2.4905,
      "step": 16
    },
    {
      "epoch": 0.07132867132867132,
      "grad_norm": 243.03379821777344,
      "learning_rate": 8.500000000000001e-07,
      "loss": 2.5882,
      "step": 17
    },
    {
      "epoch": 0.07552447552447553,
      "grad_norm": 1498.6507568359375,
      "learning_rate": 9.000000000000001e-07,
      "loss": 2.4565,
      "step": 18
    },
    {
      "epoch": 0.07972027972027972,
      "grad_norm": 1193.4425048828125,
      "learning_rate": 9.500000000000001e-07,
      "loss": 2.4394,
      "step": 19
    },
    {
      "epoch": 0.08391608391608392,
      "grad_norm": 1025.7431640625,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 2.3618,
      "step": 20
    },
    {
      "epoch": 0.08811188811188811,
      "grad_norm": 194.40928649902344,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 2.2295,
      "step": 21
    },
    {
      "epoch": 0.09230769230769231,
      "grad_norm": 509.2408752441406,
      "learning_rate": 1.1e-06,
      "loss": 2.1498,
      "step": 22
    },
    {
      "epoch": 0.0965034965034965,
      "grad_norm": 177.00662231445312,
      "learning_rate": 1.1500000000000002e-06,
      "loss": 2.136,
      "step": 23
    },
    {
      "epoch": 0.1006993006993007,
      "grad_norm": 266.7733459472656,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 2.0781,
      "step": 24
    },
    {
      "epoch": 0.1048951048951049,
      "grad_norm": 515.4574584960938,
      "learning_rate": 1.25e-06,
      "loss": 2.1255,
      "step": 25
    },
    {
      "epoch": 0.10909090909090909,
      "grad_norm": 242.0486297607422,
      "learning_rate": 1.3e-06,
      "loss": 1.9491,
      "step": 26
    },
    {
      "epoch": 0.11328671328671329,
      "grad_norm": 226.7819366455078,
      "learning_rate": 1.3500000000000002e-06,
      "loss": 1.7425,
      "step": 27
    },
    {
      "epoch": 0.11748251748251748,
      "grad_norm": 657.841064453125,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 1.7396,
      "step": 28
    },
    {
      "epoch": 0.12167832167832168,
      "grad_norm": 2119.27197265625,
      "learning_rate": 1.45e-06,
      "loss": 1.6205,
      "step": 29
    },
    {
      "epoch": 0.1258741258741259,
      "grad_norm": 2379.28857421875,
      "learning_rate": 1.5e-06,
      "loss": 1.5446,
      "step": 30
    },
    {
      "epoch": 0.13006993006993006,
      "grad_norm": 306.68927001953125,
      "learning_rate": 1.5500000000000002e-06,
      "loss": 1.4194,
      "step": 31
    },
    {
      "epoch": 0.13426573426573427,
      "grad_norm": 653.746337890625,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 1.3444,
      "step": 32
    },
    {
      "epoch": 0.13846153846153847,
      "grad_norm": 886.5042114257812,
      "learning_rate": 1.6500000000000003e-06,
      "loss": 1.2742,
      "step": 33
    },
    {
      "epoch": 0.14265734265734265,
      "grad_norm": 852.1470947265625,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 1.1457,
      "step": 34
    },
    {
      "epoch": 0.14685314685314685,
      "grad_norm": 1664.3746337890625,
      "learning_rate": 1.75e-06,
      "loss": 1.164,
      "step": 35
    },
    {
      "epoch": 0.15104895104895105,
      "grad_norm": 1804.9078369140625,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 1.1269,
      "step": 36
    },
    {
      "epoch": 0.15524475524475526,
      "grad_norm": 167.39125061035156,
      "learning_rate": 1.85e-06,
      "loss": 0.9994,
      "step": 37
    },
    {
      "epoch": 0.15944055944055943,
      "grad_norm": 397.78350830078125,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 0.8804,
      "step": 38
    },
    {
      "epoch": 0.16363636363636364,
      "grad_norm": 1123.1029052734375,
      "learning_rate": 1.9500000000000004e-06,
      "loss": 0.8312,
      "step": 39
    },
    {
      "epoch": 0.16783216783216784,
      "grad_norm": 1532.86474609375,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.7654,
      "step": 40
    },
    {
      "epoch": 0.17202797202797201,
      "grad_norm": 755.2657470703125,
      "learning_rate": 2.05e-06,
      "loss": 0.6198,
      "step": 41
    },
    {
      "epoch": 0.17622377622377622,
      "grad_norm": 1052.9649658203125,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.5466,
      "step": 42
    },
    {
      "epoch": 0.18041958041958042,
      "grad_norm": 219.14976501464844,
      "learning_rate": 2.15e-06,
      "loss": 0.4735,
      "step": 43
    },
    {
      "epoch": 0.18461538461538463,
      "grad_norm": 1271.767333984375,
      "learning_rate": 2.2e-06,
      "loss": 0.4669,
      "step": 44
    },
    {
      "epoch": 0.1888111888111888,
      "grad_norm": 1254.658935546875,
      "learning_rate": 2.25e-06,
      "loss": 0.3965,
      "step": 45
    },
    {
      "epoch": 0.193006993006993,
      "grad_norm": 927.441162109375,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 0.3704,
      "step": 46
    },
    {
      "epoch": 0.1972027972027972,
      "grad_norm": 424.604736328125,
      "learning_rate": 2.35e-06,
      "loss": 0.2998,
      "step": 47
    },
    {
      "epoch": 0.2013986013986014,
      "grad_norm": 562.2470703125,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.2485,
      "step": 48
    },
    {
      "epoch": 0.2055944055944056,
      "grad_norm": 619.4100341796875,
      "learning_rate": 2.4500000000000003e-06,
      "loss": 0.2549,
      "step": 49
    },
    {
      "epoch": 0.2097902097902098,
      "grad_norm": 282.3805847167969,
      "learning_rate": 2.5e-06,
      "loss": 0.2081,
      "step": 50
    },
    {
      "epoch": 0.213986013986014,
      "grad_norm": 530.0814208984375,
      "learning_rate": 2.55e-06,
      "loss": 0.229,
      "step": 51
    },
    {
      "epoch": 0.21818181818181817,
      "grad_norm": 315.1905212402344,
      "learning_rate": 2.6e-06,
      "loss": 0.201,
      "step": 52
    },
    {
      "epoch": 0.22237762237762237,
      "grad_norm": 363.47918701171875,
      "learning_rate": 2.6500000000000005e-06,
      "loss": 0.1797,
      "step": 53
    },
    {
      "epoch": 0.22657342657342658,
      "grad_norm": 485.228271484375,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 0.1743,
      "step": 54
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 332.8292236328125,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.1804,
      "step": 55
    },
    {
      "epoch": 0.23496503496503496,
      "grad_norm": 386.52447509765625,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.1663,
      "step": 56
    },
    {
      "epoch": 0.23916083916083916,
      "grad_norm": 368.46307373046875,
      "learning_rate": 2.85e-06,
      "loss": 0.1784,
      "step": 57
    },
    {
      "epoch": 0.24335664335664337,
      "grad_norm": 336.0594787597656,
      "learning_rate": 2.9e-06,
      "loss": 0.1637,
      "step": 58
    },
    {
      "epoch": 0.24755244755244754,
      "grad_norm": 34.440250396728516,
      "learning_rate": 2.95e-06,
      "loss": 0.1392,
      "step": 59
    },
    {
      "epoch": 0.2517482517482518,
      "grad_norm": 1019.0678100585938,
      "learning_rate": 3e-06,
      "loss": 0.1957,
      "step": 60
    },
    {
      "epoch": 0.25594405594405595,
      "grad_norm": 1919.9234619140625,
      "learning_rate": 3.05e-06,
      "loss": 0.2649,
      "step": 61
    },
    {
      "epoch": 0.2601398601398601,
      "grad_norm": 1502.3551025390625,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 0.255,
      "step": 62
    },
    {
      "epoch": 0.26433566433566436,
      "grad_norm": 880.5719604492188,
      "learning_rate": 3.1500000000000003e-06,
      "loss": 0.1904,
      "step": 63
    },
    {
      "epoch": 0.26853146853146853,
      "grad_norm": 140.78419494628906,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.1275,
      "step": 64
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 280.4155578613281,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.1373,
      "step": 65
    },
    {
      "epoch": 0.27692307692307694,
      "grad_norm": 337.9070739746094,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.1642,
      "step": 66
    },
    {
      "epoch": 0.2811188811188811,
      "grad_norm": 331.3836669921875,
      "learning_rate": 3.3500000000000005e-06,
      "loss": 0.152,
      "step": 67
    },
    {
      "epoch": 0.2853146853146853,
      "grad_norm": 184.1329803466797,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.1101,
      "step": 68
    },
    {
      "epoch": 0.2895104895104895,
      "grad_norm": 3.4701905250549316,
      "learning_rate": 3.45e-06,
      "loss": 0.1073,
      "step": 69
    },
    {
      "epoch": 0.2937062937062937,
      "grad_norm": 244.07217407226562,
      "learning_rate": 3.5e-06,
      "loss": 0.1375,
      "step": 70
    },
    {
      "epoch": 0.29790209790209793,
      "grad_norm": 304.42498779296875,
      "learning_rate": 3.5500000000000003e-06,
      "loss": 0.1409,
      "step": 71
    },
    {
      "epoch": 0.3020979020979021,
      "grad_norm": 216.21617126464844,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.1348,
      "step": 72
    },
    {
      "epoch": 0.3062937062937063,
      "grad_norm": 105.47506713867188,
      "learning_rate": 3.65e-06,
      "loss": 0.113,
      "step": 73
    },
    {
      "epoch": 0.3104895104895105,
      "grad_norm": 75.4766845703125,
      "learning_rate": 3.7e-06,
      "loss": 0.1107,
      "step": 74
    },
    {
      "epoch": 0.3146853146853147,
      "grad_norm": 48.33282470703125,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.1177,
      "step": 75
    },
    {
      "epoch": 0.31888111888111886,
      "grad_norm": 9.441969871520996,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.1258,
      "step": 76
    },
    {
      "epoch": 0.3230769230769231,
      "grad_norm": 45.23430252075195,
      "learning_rate": 3.85e-06,
      "loss": 0.1016,
      "step": 77
    },
    {
      "epoch": 0.32727272727272727,
      "grad_norm": 53.305484771728516,
      "learning_rate": 3.900000000000001e-06,
      "loss": 0.1054,
      "step": 78
    },
    {
      "epoch": 0.33146853146853145,
      "grad_norm": 27.8891544342041,
      "learning_rate": 3.95e-06,
      "loss": 0.1066,
      "step": 79
    },
    {
      "epoch": 0.3356643356643357,
      "grad_norm": 18.898664474487305,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.1036,
      "step": 80
    },
    {
      "epoch": 0.33986013986013985,
      "grad_norm": 30.54612159729004,
      "learning_rate": 4.05e-06,
      "loss": 0.1044,
      "step": 81
    },
    {
      "epoch": 0.34405594405594403,
      "grad_norm": 23.064117431640625,
      "learning_rate": 4.1e-06,
      "loss": 0.1082,
      "step": 82
    },
    {
      "epoch": 0.34825174825174826,
      "grad_norm": 77.5650863647461,
      "learning_rate": 4.15e-06,
      "loss": 0.1079,
      "step": 83
    },
    {
      "epoch": 0.35244755244755244,
      "grad_norm": 81.39286041259766,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.0984,
      "step": 84
    },
    {
      "epoch": 0.35664335664335667,
      "grad_norm": 40.70750045776367,
      "learning_rate": 4.25e-06,
      "loss": 0.1047,
      "step": 85
    },
    {
      "epoch": 0.36083916083916084,
      "grad_norm": 8.890117645263672,
      "learning_rate": 4.3e-06,
      "loss": 0.0978,
      "step": 86
    },
    {
      "epoch": 0.365034965034965,
      "grad_norm": 13.848636627197266,
      "learning_rate": 4.350000000000001e-06,
      "loss": 0.101,
      "step": 87
    },
    {
      "epoch": 0.36923076923076925,
      "grad_norm": 21.611827850341797,
      "learning_rate": 4.4e-06,
      "loss": 0.0901,
      "step": 88
    },
    {
      "epoch": 0.3734265734265734,
      "grad_norm": 22.858949661254883,
      "learning_rate": 4.450000000000001e-06,
      "loss": 0.0958,
      "step": 89
    },
    {
      "epoch": 0.3776223776223776,
      "grad_norm": 3.473940372467041,
      "learning_rate": 4.5e-06,
      "loss": 0.0921,
      "step": 90
    },
    {
      "epoch": 0.38181818181818183,
      "grad_norm": 9.609582901000977,
      "learning_rate": 4.5500000000000005e-06,
      "loss": 0.1034,
      "step": 91
    },
    {
      "epoch": 0.386013986013986,
      "grad_norm": 8.03201961517334,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.0928,
      "step": 92
    },
    {
      "epoch": 0.3902097902097902,
      "grad_norm": 1.6049108505249023,
      "learning_rate": 4.65e-06,
      "loss": 0.0867,
      "step": 93
    },
    {
      "epoch": 0.3944055944055944,
      "grad_norm": 6.763339996337891,
      "learning_rate": 4.7e-06,
      "loss": 0.0911,
      "step": 94
    },
    {
      "epoch": 0.3986013986013986,
      "grad_norm": 7.147759914398193,
      "learning_rate": 4.75e-06,
      "loss": 0.1057,
      "step": 95
    },
    {
      "epoch": 0.4027972027972028,
      "grad_norm": 7.845475673675537,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.1026,
      "step": 96
    },
    {
      "epoch": 0.406993006993007,
      "grad_norm": 30.894298553466797,
      "learning_rate": 4.85e-06,
      "loss": 0.0894,
      "step": 97
    },
    {
      "epoch": 0.4111888111888112,
      "grad_norm": 51.0138053894043,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.1011,
      "step": 98
    },
    {
      "epoch": 0.4153846153846154,
      "grad_norm": 45.050018310546875,
      "learning_rate": 4.95e-06,
      "loss": 0.0943,
      "step": 99
    },
    {
      "epoch": 0.4195804195804196,
      "grad_norm": 9.880765914916992,
      "learning_rate": 5e-06,
      "loss": 0.0923,
      "step": 100
    },
    {
      "epoch": 0.42377622377622376,
      "grad_norm": 1.3705830574035645,
      "learning_rate": 4.999993004586259e-06,
      "loss": 0.0967,
      "step": 101
    },
    {
      "epoch": 0.427972027972028,
      "grad_norm": 4.682475566864014,
      "learning_rate": 4.999972018384183e-06,
      "loss": 0.0916,
      "step": 102
    },
    {
      "epoch": 0.43216783216783217,
      "grad_norm": 3.074434518814087,
      "learning_rate": 4.9999370415112185e-06,
      "loss": 0.0864,
      "step": 103
    },
    {
      "epoch": 0.43636363636363634,
      "grad_norm": 1.248001217842102,
      "learning_rate": 4.999888074163108e-06,
      "loss": 0.0944,
      "step": 104
    },
    {
      "epoch": 0.4405594405594406,
      "grad_norm": 2.949608325958252,
      "learning_rate": 4.999825116613888e-06,
      "loss": 0.0878,
      "step": 105
    },
    {
      "epoch": 0.44475524475524475,
      "grad_norm": 22.338565826416016,
      "learning_rate": 4.999748169215891e-06,
      "loss": 0.0848,
      "step": 106
    },
    {
      "epoch": 0.4489510489510489,
      "grad_norm": 33.91677474975586,
      "learning_rate": 4.999657232399739e-06,
      "loss": 0.0959,
      "step": 107
    },
    {
      "epoch": 0.45314685314685316,
      "grad_norm": 53.919132232666016,
      "learning_rate": 4.999552306674345e-06,
      "loss": 0.0909,
      "step": 108
    },
    {
      "epoch": 0.45734265734265733,
      "grad_norm": 28.08509063720703,
      "learning_rate": 4.999433392626908e-06,
      "loss": 0.0839,
      "step": 109
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 6.412335395812988,
      "learning_rate": 4.999300490922911e-06,
      "loss": 0.0823,
      "step": 110
    },
    {
      "epoch": 0.46573426573426574,
      "grad_norm": 2.714561700820923,
      "learning_rate": 4.999153602306115e-06,
      "loss": 0.0949,
      "step": 111
    },
    {
      "epoch": 0.4699300699300699,
      "grad_norm": 1.5083353519439697,
      "learning_rate": 4.998992727598557e-06,
      "loss": 0.0856,
      "step": 112
    },
    {
      "epoch": 0.47412587412587415,
      "grad_norm": 1.422822117805481,
      "learning_rate": 4.998817867700546e-06,
      "loss": 0.0837,
      "step": 113
    },
    {
      "epoch": 0.4783216783216783,
      "grad_norm": 7.532190799713135,
      "learning_rate": 4.998629023590656e-06,
      "loss": 0.0901,
      "step": 114
    },
    {
      "epoch": 0.4825174825174825,
      "grad_norm": 28.832021713256836,
      "learning_rate": 4.998426196325721e-06,
      "loss": 0.0852,
      "step": 115
    },
    {
      "epoch": 0.48671328671328673,
      "grad_norm": 13.580469131469727,
      "learning_rate": 4.998209387040829e-06,
      "loss": 0.0879,
      "step": 116
    },
    {
      "epoch": 0.4909090909090909,
      "grad_norm": 11.114592552185059,
      "learning_rate": 4.9979785969493165e-06,
      "loss": 0.0819,
      "step": 117
    },
    {
      "epoch": 0.4951048951048951,
      "grad_norm": 1.150378942489624,
      "learning_rate": 4.9977338273427625e-06,
      "loss": 0.0773,
      "step": 118
    },
    {
      "epoch": 0.4993006993006993,
      "grad_norm": 1.427876353263855,
      "learning_rate": 4.997475079590977e-06,
      "loss": 0.0848,
      "step": 119
    },
    {
      "epoch": 0.5034965034965035,
      "grad_norm": 1.115370750427246,
      "learning_rate": 4.9972023551419995e-06,
      "loss": 0.0766,
      "step": 120
    },
    {
      "epoch": 0.5076923076923077,
      "grad_norm": 1.1270095109939575,
      "learning_rate": 4.996915655522086e-06,
      "loss": 0.0833,
      "step": 121
    },
    {
      "epoch": 0.5118881118881119,
      "grad_norm": 1.219711184501648,
      "learning_rate": 4.9966149823357e-06,
      "loss": 0.0881,
      "step": 122
    },
    {
      "epoch": 0.5160839160839161,
      "grad_norm": 0.8859412670135498,
      "learning_rate": 4.996300337265512e-06,
      "loss": 0.0817,
      "step": 123
    },
    {
      "epoch": 0.5202797202797202,
      "grad_norm": 0.5764117240905762,
      "learning_rate": 4.995971722072379e-06,
      "loss": 0.0792,
      "step": 124
    },
    {
      "epoch": 0.5244755244755245,
      "grad_norm": 0.8799321055412292,
      "learning_rate": 4.9956291385953385e-06,
      "loss": 0.0781,
      "step": 125
    },
    {
      "epoch": 0.5286713286713287,
      "grad_norm": 1.0168715715408325,
      "learning_rate": 4.9952725887516015e-06,
      "loss": 0.0797,
      "step": 126
    },
    {
      "epoch": 0.5328671328671328,
      "grad_norm": 1.323622465133667,
      "learning_rate": 4.99490207453654e-06,
      "loss": 0.0842,
      "step": 127
    },
    {
      "epoch": 0.5370629370629371,
      "grad_norm": 1.757987141609192,
      "learning_rate": 4.9945175980236745e-06,
      "loss": 0.0766,
      "step": 128
    },
    {
      "epoch": 0.5412587412587413,
      "grad_norm": 2.3605685234069824,
      "learning_rate": 4.994119161364663e-06,
      "loss": 0.0772,
      "step": 129
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 1.0868511199951172,
      "learning_rate": 4.993706766789287e-06,
      "loss": 0.075,
      "step": 130
    },
    {
      "epoch": 0.5496503496503496,
      "grad_norm": 5.12212610244751,
      "learning_rate": 4.993280416605444e-06,
      "loss": 0.0778,
      "step": 131
    },
    {
      "epoch": 0.5538461538461539,
      "grad_norm": 13.817222595214844,
      "learning_rate": 4.992840113199131e-06,
      "loss": 0.0776,
      "step": 132
    },
    {
      "epoch": 0.558041958041958,
      "grad_norm": 1.0479037761688232,
      "learning_rate": 4.992385859034431e-06,
      "loss": 0.0833,
      "step": 133
    },
    {
      "epoch": 0.5622377622377622,
      "grad_norm": 4.102362155914307,
      "learning_rate": 4.991917656653501e-06,
      "loss": 0.0789,
      "step": 134
    },
    {
      "epoch": 0.5664335664335665,
      "grad_norm": 3.706054210662842,
      "learning_rate": 4.991435508676557e-06,
      "loss": 0.0703,
      "step": 135
    },
    {
      "epoch": 0.5706293706293706,
      "grad_norm": 0.9211217761039734,
      "learning_rate": 4.990939417801859e-06,
      "loss": 0.0832,
      "step": 136
    },
    {
      "epoch": 0.5748251748251748,
      "grad_norm": 1.108568787574768,
      "learning_rate": 4.990429386805694e-06,
      "loss": 0.0745,
      "step": 137
    },
    {
      "epoch": 0.579020979020979,
      "grad_norm": 0.732056736946106,
      "learning_rate": 4.989905418542366e-06,
      "loss": 0.0745,
      "step": 138
    },
    {
      "epoch": 0.5832167832167832,
      "grad_norm": 0.5949814319610596,
      "learning_rate": 4.989367515944173e-06,
      "loss": 0.0809,
      "step": 139
    },
    {
      "epoch": 0.5874125874125874,
      "grad_norm": 1.0817596912384033,
      "learning_rate": 4.988815682021398e-06,
      "loss": 0.0698,
      "step": 140
    },
    {
      "epoch": 0.5916083916083916,
      "grad_norm": 1.5362236499786377,
      "learning_rate": 4.988249919862285e-06,
      "loss": 0.073,
      "step": 141
    },
    {
      "epoch": 0.5958041958041959,
      "grad_norm": 1.4847773313522339,
      "learning_rate": 4.987670232633027e-06,
      "loss": 0.0765,
      "step": 142
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7822750806808472,
      "learning_rate": 4.987076623577745e-06,
      "loss": 0.0733,
      "step": 143
    },
    {
      "epoch": 0.6041958041958042,
      "grad_norm": 0.9540371894836426,
      "learning_rate": 4.986469096018472e-06,
      "loss": 0.0752,
      "step": 144
    },
    {
      "epoch": 0.6083916083916084,
      "grad_norm": 1.355391263961792,
      "learning_rate": 4.9858476533551335e-06,
      "loss": 0.0707,
      "step": 145
    },
    {
      "epoch": 0.6125874125874126,
      "grad_norm": 0.9062367081642151,
      "learning_rate": 4.985212299065528e-06,
      "loss": 0.0748,
      "step": 146
    },
    {
      "epoch": 0.6167832167832168,
      "grad_norm": 0.7668821811676025,
      "learning_rate": 4.984563036705309e-06,
      "loss": 0.0667,
      "step": 147
    },
    {
      "epoch": 0.620979020979021,
      "grad_norm": 0.7187187671661377,
      "learning_rate": 4.983899869907963e-06,
      "loss": 0.0719,
      "step": 148
    },
    {
      "epoch": 0.6251748251748251,
      "grad_norm": 1.0374109745025635,
      "learning_rate": 4.9832228023847905e-06,
      "loss": 0.0724,
      "step": 149
    },
    {
      "epoch": 0.6293706293706294,
      "grad_norm": 1.6743290424346924,
      "learning_rate": 4.982531837924887e-06,
      "loss": 0.0745,
      "step": 150
    },
    {
      "epoch": 0.6335664335664336,
      "grad_norm": 0.8888541460037231,
      "learning_rate": 4.981826980395117e-06,
      "loss": 0.0683,
      "step": 151
    },
    {
      "epoch": 0.6377622377622377,
      "grad_norm": 0.8322030305862427,
      "learning_rate": 4.981108233740096e-06,
      "loss": 0.066,
      "step": 152
    },
    {
      "epoch": 0.641958041958042,
      "grad_norm": 1.1838620901107788,
      "learning_rate": 4.98037560198217e-06,
      "loss": 0.0815,
      "step": 153
    },
    {
      "epoch": 0.6461538461538462,
      "grad_norm": 1.9783540964126587,
      "learning_rate": 4.979629089221387e-06,
      "loss": 0.0732,
      "step": 154
    },
    {
      "epoch": 0.6503496503496503,
      "grad_norm": 1.0773838758468628,
      "learning_rate": 4.978868699635482e-06,
      "loss": 0.0635,
      "step": 155
    },
    {
      "epoch": 0.6545454545454545,
      "grad_norm": 1.0689246654510498,
      "learning_rate": 4.978094437479843e-06,
      "loss": 0.0705,
      "step": 156
    },
    {
      "epoch": 0.6587412587412588,
      "grad_norm": 0.9070070385932922,
      "learning_rate": 4.977306307087502e-06,
      "loss": 0.0628,
      "step": 157
    },
    {
      "epoch": 0.6629370629370629,
      "grad_norm": 2.045474052429199,
      "learning_rate": 4.976504312869093e-06,
      "loss": 0.0645,
      "step": 158
    },
    {
      "epoch": 0.6671328671328671,
      "grad_norm": 4.762063503265381,
      "learning_rate": 4.975688459312844e-06,
      "loss": 0.0661,
      "step": 159
    },
    {
      "epoch": 0.6713286713286714,
      "grad_norm": 10.791966438293457,
      "learning_rate": 4.97485875098454e-06,
      "loss": 0.0657,
      "step": 160
    },
    {
      "epoch": 0.6755244755244755,
      "grad_norm": 1.7596170902252197,
      "learning_rate": 4.974015192527505e-06,
      "loss": 0.075,
      "step": 161
    },
    {
      "epoch": 0.6797202797202797,
      "grad_norm": 2.43743896484375,
      "learning_rate": 4.97315778866257e-06,
      "loss": 0.0622,
      "step": 162
    },
    {
      "epoch": 0.6839160839160839,
      "grad_norm": 1.6388510465621948,
      "learning_rate": 4.972286544188052e-06,
      "loss": 0.0601,
      "step": 163
    },
    {
      "epoch": 0.6881118881118881,
      "grad_norm": 0.8130991458892822,
      "learning_rate": 4.971401463979722e-06,
      "loss": 0.0607,
      "step": 164
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 0.8717385530471802,
      "learning_rate": 4.970502552990782e-06,
      "loss": 0.0706,
      "step": 165
    },
    {
      "epoch": 0.6965034965034965,
      "grad_norm": 0.8151756525039673,
      "learning_rate": 4.969589816251837e-06,
      "loss": 0.0628,
      "step": 166
    },
    {
      "epoch": 0.7006993006993008,
      "grad_norm": 0.7532489895820618,
      "learning_rate": 4.968663258870862e-06,
      "loss": 0.0687,
      "step": 167
    },
    {
      "epoch": 0.7048951048951049,
      "grad_norm": 0.5839558839797974,
      "learning_rate": 4.967722886033181e-06,
      "loss": 0.0554,
      "step": 168
    },
    {
      "epoch": 0.7090909090909091,
      "grad_norm": 0.6098427772521973,
      "learning_rate": 4.966768703001429e-06,
      "loss": 0.0639,
      "step": 169
    },
    {
      "epoch": 0.7132867132867133,
      "grad_norm": 0.9210315942764282,
      "learning_rate": 4.965800715115531e-06,
      "loss": 0.0704,
      "step": 170
    },
    {
      "epoch": 0.7174825174825175,
      "grad_norm": 0.9615612626075745,
      "learning_rate": 4.964818927792669e-06,
      "loss": 0.068,
      "step": 171
    },
    {
      "epoch": 0.7216783216783217,
      "grad_norm": 0.7204346060752869,
      "learning_rate": 4.963823346527249e-06,
      "loss": 0.0663,
      "step": 172
    },
    {
      "epoch": 0.7258741258741259,
      "grad_norm": 0.7109172344207764,
      "learning_rate": 4.962813976890872e-06,
      "loss": 0.0603,
      "step": 173
    },
    {
      "epoch": 0.73006993006993,
      "grad_norm": 0.9478862881660461,
      "learning_rate": 4.961790824532306e-06,
      "loss": 0.061,
      "step": 174
    },
    {
      "epoch": 0.7342657342657343,
      "grad_norm": 0.7901272773742676,
      "learning_rate": 4.96075389517745e-06,
      "loss": 0.0572,
      "step": 175
    },
    {
      "epoch": 0.7384615384615385,
      "grad_norm": 0.7225059270858765,
      "learning_rate": 4.959703194629304e-06,
      "loss": 0.0567,
      "step": 176
    },
    {
      "epoch": 0.7426573426573426,
      "grad_norm": 0.693332850933075,
      "learning_rate": 4.958638728767936e-06,
      "loss": 0.0545,
      "step": 177
    },
    {
      "epoch": 0.7468531468531469,
      "grad_norm": 0.834145724773407,
      "learning_rate": 4.95756050355045e-06,
      "loss": 0.0671,
      "step": 178
    },
    {
      "epoch": 0.7510489510489511,
      "grad_norm": 1.00062894821167,
      "learning_rate": 4.956468525010948e-06,
      "loss": 0.0539,
      "step": 179
    },
    {
      "epoch": 0.7552447552447552,
      "grad_norm": 1.1205754280090332,
      "learning_rate": 4.955362799260507e-06,
      "loss": 0.0529,
      "step": 180
    },
    {
      "epoch": 0.7594405594405594,
      "grad_norm": 2.0870769023895264,
      "learning_rate": 4.954243332487133e-06,
      "loss": 0.059,
      "step": 181
    },
    {
      "epoch": 0.7636363636363637,
      "grad_norm": 2.168609619140625,
      "learning_rate": 4.953110130955733e-06,
      "loss": 0.0534,
      "step": 182
    },
    {
      "epoch": 0.7678321678321678,
      "grad_norm": 1.5437854528427124,
      "learning_rate": 4.9519632010080765e-06,
      "loss": 0.0472,
      "step": 183
    },
    {
      "epoch": 0.772027972027972,
      "grad_norm": 2.0716335773468018,
      "learning_rate": 4.950802549062764e-06,
      "loss": 0.0605,
      "step": 184
    },
    {
      "epoch": 0.7762237762237763,
      "grad_norm": 1.5331146717071533,
      "learning_rate": 4.9496281816151895e-06,
      "loss": 0.0615,
      "step": 185
    },
    {
      "epoch": 0.7804195804195804,
      "grad_norm": 0.8838733434677124,
      "learning_rate": 4.948440105237499e-06,
      "loss": 0.0494,
      "step": 186
    },
    {
      "epoch": 0.7846153846153846,
      "grad_norm": 0.8043892979621887,
      "learning_rate": 4.947238326578563e-06,
      "loss": 0.0568,
      "step": 187
    },
    {
      "epoch": 0.7888111888111888,
      "grad_norm": 1.4767056703567505,
      "learning_rate": 4.946022852363932e-06,
      "loss": 0.0565,
      "step": 188
    },
    {
      "epoch": 0.793006993006993,
      "grad_norm": 1.083964467048645,
      "learning_rate": 4.944793689395802e-06,
      "loss": 0.0582,
      "step": 189
    },
    {
      "epoch": 0.7972027972027972,
      "grad_norm": 0.99649977684021,
      "learning_rate": 4.943550844552978e-06,
      "loss": 0.048,
      "step": 190
    },
    {
      "epoch": 0.8013986013986014,
      "grad_norm": 0.674566924571991,
      "learning_rate": 4.9422943247908275e-06,
      "loss": 0.0485,
      "step": 191
    },
    {
      "epoch": 0.8055944055944056,
      "grad_norm": 1.1796026229858398,
      "learning_rate": 4.9410241371412525e-06,
      "loss": 0.0501,
      "step": 192
    },
    {
      "epoch": 0.8097902097902098,
      "grad_norm": 0.9794552326202393,
      "learning_rate": 4.939740288712644e-06,
      "loss": 0.0527,
      "step": 193
    },
    {
      "epoch": 0.813986013986014,
      "grad_norm": 0.8760250210762024,
      "learning_rate": 4.938442786689843e-06,
      "loss": 0.0539,
      "step": 194
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 1.3841289281845093,
      "learning_rate": 4.9371316383341e-06,
      "loss": 0.0543,
      "step": 195
    },
    {
      "epoch": 0.8223776223776224,
      "grad_norm": 1.4383552074432373,
      "learning_rate": 4.935806850983034e-06,
      "loss": 0.0479,
      "step": 196
    },
    {
      "epoch": 0.8265734265734266,
      "grad_norm": 1.7970709800720215,
      "learning_rate": 4.934468432050594e-06,
      "loss": 0.049,
      "step": 197
    },
    {
      "epoch": 0.8307692307692308,
      "grad_norm": 1.0703388452529907,
      "learning_rate": 4.933116389027017e-06,
      "loss": 0.0527,
      "step": 198
    },
    {
      "epoch": 0.8349650349650349,
      "grad_norm": 3.151549816131592,
      "learning_rate": 4.931750729478781e-06,
      "loss": 0.0529,
      "step": 199
    },
    {
      "epoch": 0.8391608391608392,
      "grad_norm": 1.772434949874878,
      "learning_rate": 4.9303714610485705e-06,
      "loss": 0.0535,
      "step": 200
    },
    {
      "epoch": 0.8433566433566434,
      "grad_norm": 0.8801355361938477,
      "learning_rate": 4.928978591455228e-06,
      "loss": 0.0504,
      "step": 201
    },
    {
      "epoch": 0.8475524475524475,
      "grad_norm": 1.4067658185958862,
      "learning_rate": 4.9275721284937115e-06,
      "loss": 0.049,
      "step": 202
    },
    {
      "epoch": 0.8517482517482518,
      "grad_norm": 1.0001884698867798,
      "learning_rate": 4.926152080035055e-06,
      "loss": 0.0614,
      "step": 203
    },
    {
      "epoch": 0.855944055944056,
      "grad_norm": 1.1527674198150635,
      "learning_rate": 4.924718454026318e-06,
      "loss": 0.0479,
      "step": 204
    },
    {
      "epoch": 0.8601398601398601,
      "grad_norm": 1.1766809225082397,
      "learning_rate": 4.923271258490547e-06,
      "loss": 0.0482,
      "step": 205
    },
    {
      "epoch": 0.8643356643356643,
      "grad_norm": 0.8541592359542847,
      "learning_rate": 4.921810501526728e-06,
      "loss": 0.0458,
      "step": 206
    },
    {
      "epoch": 0.8685314685314686,
      "grad_norm": 1.3551443815231323,
      "learning_rate": 4.9203361913097394e-06,
      "loss": 0.0497,
      "step": 207
    },
    {
      "epoch": 0.8727272727272727,
      "grad_norm": 1.417004108428955,
      "learning_rate": 4.918848336090309e-06,
      "loss": 0.0472,
      "step": 208
    },
    {
      "epoch": 0.8769230769230769,
      "grad_norm": 1.3648598194122314,
      "learning_rate": 4.917346944194969e-06,
      "loss": 0.0415,
      "step": 209
    },
    {
      "epoch": 0.8811188811188811,
      "grad_norm": 1.0727978944778442,
      "learning_rate": 4.915832024026002e-06,
      "loss": 0.0421,
      "step": 210
    },
    {
      "epoch": 0.8853146853146853,
      "grad_norm": 2.974808692932129,
      "learning_rate": 4.914303584061405e-06,
      "loss": 0.0474,
      "step": 211
    },
    {
      "epoch": 0.8895104895104895,
      "grad_norm": 1.6205039024353027,
      "learning_rate": 4.912761632854834e-06,
      "loss": 0.0445,
      "step": 212
    },
    {
      "epoch": 0.8937062937062937,
      "grad_norm": 4.179663181304932,
      "learning_rate": 4.911206179035557e-06,
      "loss": 0.0438,
      "step": 213
    },
    {
      "epoch": 0.8979020979020979,
      "grad_norm": 10.701528549194336,
      "learning_rate": 4.90963723130841e-06,
      "loss": 0.0399,
      "step": 214
    },
    {
      "epoch": 0.9020979020979021,
      "grad_norm": 11.126282691955566,
      "learning_rate": 4.908054798453744e-06,
      "loss": 0.0499,
      "step": 215
    },
    {
      "epoch": 0.9062937062937063,
      "grad_norm": 8.534737586975098,
      "learning_rate": 4.906458889327375e-06,
      "loss": 0.0445,
      "step": 216
    },
    {
      "epoch": 0.9104895104895104,
      "grad_norm": 1.3229663372039795,
      "learning_rate": 4.90484951286054e-06,
      "loss": 0.0495,
      "step": 217
    },
    {
      "epoch": 0.9146853146853147,
      "grad_norm": 1.9636162519454956,
      "learning_rate": 4.903226678059842e-06,
      "loss": 0.0464,
      "step": 218
    },
    {
      "epoch": 0.9188811188811189,
      "grad_norm": 1.5941684246063232,
      "learning_rate": 4.901590394007202e-06,
      "loss": 0.0469,
      "step": 219
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 2.0621442794799805,
      "learning_rate": 4.899940669859807e-06,
      "loss": 0.0438,
      "step": 220
    },
    {
      "epoch": 0.9272727272727272,
      "grad_norm": 0.8177353739738464,
      "learning_rate": 4.89827751485006e-06,
      "loss": 0.0519,
      "step": 221
    },
    {
      "epoch": 0.9314685314685315,
      "grad_norm": 6.465226173400879,
      "learning_rate": 4.896600938285526e-06,
      "loss": 0.0424,
      "step": 222
    },
    {
      "epoch": 0.9356643356643357,
      "grad_norm": 2.424146890640259,
      "learning_rate": 4.8949109495488834e-06,
      "loss": 0.0444,
      "step": 223
    },
    {
      "epoch": 0.9398601398601398,
      "grad_norm": 3.0828588008880615,
      "learning_rate": 4.893207558097867e-06,
      "loss": 0.0402,
      "step": 224
    },
    {
      "epoch": 0.9440559440559441,
      "grad_norm": 0.6914602518081665,
      "learning_rate": 4.891490773465221e-06,
      "loss": 0.038,
      "step": 225
    },
    {
      "epoch": 0.9482517482517483,
      "grad_norm": 0.7816252112388611,
      "learning_rate": 4.8897606052586384e-06,
      "loss": 0.0419,
      "step": 226
    },
    {
      "epoch": 0.9524475524475524,
      "grad_norm": 1.139478325843811,
      "learning_rate": 4.8880170631607145e-06,
      "loss": 0.0402,
      "step": 227
    },
    {
      "epoch": 0.9566433566433566,
      "grad_norm": 0.9213783740997314,
      "learning_rate": 4.8862601569288885e-06,
      "loss": 0.0413,
      "step": 228
    },
    {
      "epoch": 0.9608391608391609,
      "grad_norm": 1.3993144035339355,
      "learning_rate": 4.8844898963953875e-06,
      "loss": 0.0383,
      "step": 229
    },
    {
      "epoch": 0.965034965034965,
      "grad_norm": 0.8406299352645874,
      "learning_rate": 4.8827062914671775e-06,
      "loss": 0.0345,
      "step": 230
    },
    {
      "epoch": 0.9692307692307692,
      "grad_norm": 1.1061815023422241,
      "learning_rate": 4.880909352125901e-06,
      "loss": 0.0411,
      "step": 231
    },
    {
      "epoch": 0.9734265734265735,
      "grad_norm": 1.6662583351135254,
      "learning_rate": 4.879099088427824e-06,
      "loss": 0.0376,
      "step": 232
    },
    {
      "epoch": 0.9776223776223776,
      "grad_norm": 1.7703438997268677,
      "learning_rate": 4.877275510503783e-06,
      "loss": 0.0412,
      "step": 233
    },
    {
      "epoch": 0.9818181818181818,
      "grad_norm": 0.8591011762619019,
      "learning_rate": 4.875438628559124e-06,
      "loss": 0.038,
      "step": 234
    },
    {
      "epoch": 0.986013986013986,
      "grad_norm": 1.0391316413879395,
      "learning_rate": 4.8735884528736444e-06,
      "loss": 0.0446,
      "step": 235
    },
    {
      "epoch": 0.9902097902097902,
      "grad_norm": 1.1077414751052856,
      "learning_rate": 4.871724993801541e-06,
      "loss": 0.037,
      "step": 236
    },
    {
      "epoch": 0.9944055944055944,
      "grad_norm": 1.3140100240707397,
      "learning_rate": 4.869848261771348e-06,
      "loss": 0.0379,
      "step": 237
    },
    {
      "epoch": 0.9986013986013986,
      "grad_norm": 0.7918304204940796,
      "learning_rate": 4.867958267285879e-06,
      "loss": 0.037,
      "step": 238
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7918304204940796,
      "learning_rate": 4.866055020922167e-06,
      "loss": 0.0281,
      "step": 239
    },
    {
      "epoch": 1.0041958041958041,
      "grad_norm": 1.9362425804138184,
      "learning_rate": 4.864138533331411e-06,
      "loss": 0.0259,
      "step": 240
    },
    {
      "epoch": 1.0083916083916085,
      "grad_norm": 1.20184326171875,
      "learning_rate": 4.862208815238908e-06,
      "loss": 0.0214,
      "step": 241
    },
    {
      "epoch": 1.0125874125874126,
      "grad_norm": 1.2968955039978027,
      "learning_rate": 4.860265877444001e-06,
      "loss": 0.0275,
      "step": 242
    },
    {
      "epoch": 1.0167832167832167,
      "grad_norm": 0.7607834935188293,
      "learning_rate": 4.85830973082001e-06,
      "loss": 0.0263,
      "step": 243
    },
    {
      "epoch": 1.020979020979021,
      "grad_norm": 0.9456799626350403,
      "learning_rate": 4.8563403863141825e-06,
      "loss": 0.0276,
      "step": 244
    },
    {
      "epoch": 1.0251748251748252,
      "grad_norm": 0.7519564032554626,
      "learning_rate": 4.854357854947619e-06,
      "loss": 0.0245,
      "step": 245
    },
    {
      "epoch": 1.0293706293706293,
      "grad_norm": 1.3622503280639648,
      "learning_rate": 4.852362147815225e-06,
      "loss": 0.0204,
      "step": 246
    },
    {
      "epoch": 1.0335664335664336,
      "grad_norm": 1.1832894086837769,
      "learning_rate": 4.850353276085634e-06,
      "loss": 0.0269,
      "step": 247
    },
    {
      "epoch": 1.0377622377622377,
      "grad_norm": 1.7349433898925781,
      "learning_rate": 4.84833125100116e-06,
      "loss": 0.0247,
      "step": 248
    },
    {
      "epoch": 1.0419580419580419,
      "grad_norm": 0.978853166103363,
      "learning_rate": 4.8462960838777234e-06,
      "loss": 0.0278,
      "step": 249
    },
    {
      "epoch": 1.0461538461538462,
      "grad_norm": 3.0965521335601807,
      "learning_rate": 4.844247786104794e-06,
      "loss": 0.0326,
      "step": 250
    },
    {
      "epoch": 1.0503496503496503,
      "grad_norm": 12.701726913452148,
      "learning_rate": 4.842186369145323e-06,
      "loss": 0.0275,
      "step": 251
    },
    {
      "epoch": 1.0545454545454545,
      "grad_norm": 1.9011986255645752,
      "learning_rate": 4.840111844535682e-06,
      "loss": 0.026,
      "step": 252
    },
    {
      "epoch": 1.0587412587412588,
      "grad_norm": 1.687672734260559,
      "learning_rate": 4.838024223885599e-06,
      "loss": 0.0247,
      "step": 253
    },
    {
      "epoch": 1.062937062937063,
      "grad_norm": 1.5581176280975342,
      "learning_rate": 4.835923518878088e-06,
      "loss": 0.0174,
      "step": 254
    },
    {
      "epoch": 1.067132867132867,
      "grad_norm": 0.980636715888977,
      "learning_rate": 4.833809741269392e-06,
      "loss": 0.0294,
      "step": 255
    },
    {
      "epoch": 1.0713286713286714,
      "grad_norm": 1.0523900985717773,
      "learning_rate": 4.8316829028889076e-06,
      "loss": 0.0241,
      "step": 256
    },
    {
      "epoch": 1.0755244755244755,
      "grad_norm": 1.6697787046432495,
      "learning_rate": 4.829543015639128e-06,
      "loss": 0.022,
      "step": 257
    },
    {
      "epoch": 1.0797202797202796,
      "grad_norm": 0.9570637941360474,
      "learning_rate": 4.82739009149557e-06,
      "loss": 0.0251,
      "step": 258
    },
    {
      "epoch": 1.083916083916084,
      "grad_norm": 1.8936893939971924,
      "learning_rate": 4.825224142506708e-06,
      "loss": 0.0229,
      "step": 259
    },
    {
      "epoch": 1.088111888111888,
      "grad_norm": 1.288090467453003,
      "learning_rate": 4.823045180793914e-06,
      "loss": 0.0258,
      "step": 260
    },
    {
      "epoch": 1.0923076923076924,
      "grad_norm": 1.9112778902053833,
      "learning_rate": 4.820853218551375e-06,
      "loss": 0.0206,
      "step": 261
    },
    {
      "epoch": 1.0965034965034965,
      "grad_norm": 1.1951854228973389,
      "learning_rate": 4.818648268046038e-06,
      "loss": 0.0201,
      "step": 262
    },
    {
      "epoch": 1.1006993006993007,
      "grad_norm": 12.678963661193848,
      "learning_rate": 4.816430341617536e-06,
      "loss": 0.021,
      "step": 263
    },
    {
      "epoch": 1.104895104895105,
      "grad_norm": 11.643608093261719,
      "learning_rate": 4.8141994516781196e-06,
      "loss": 0.0208,
      "step": 264
    },
    {
      "epoch": 1.1090909090909091,
      "grad_norm": 9.084321022033691,
      "learning_rate": 4.811955610712588e-06,
      "loss": 0.0236,
      "step": 265
    },
    {
      "epoch": 1.1132867132867132,
      "grad_norm": 1.2543950080871582,
      "learning_rate": 4.809698831278217e-06,
      "loss": 0.0224,
      "step": 266
    },
    {
      "epoch": 1.1174825174825176,
      "grad_norm": 1.0576941967010498,
      "learning_rate": 4.8074291260046925e-06,
      "loss": 0.0225,
      "step": 267
    },
    {
      "epoch": 1.1216783216783217,
      "grad_norm": 0.9116222262382507,
      "learning_rate": 4.805146507594034e-06,
      "loss": 0.0206,
      "step": 268
    },
    {
      "epoch": 1.1258741258741258,
      "grad_norm": 1.11386239528656,
      "learning_rate": 4.8028509888205325e-06,
      "loss": 0.0183,
      "step": 269
    },
    {
      "epoch": 1.1300699300699302,
      "grad_norm": 0.7633559703826904,
      "learning_rate": 4.800542582530668e-06,
      "loss": 0.0226,
      "step": 270
    },
    {
      "epoch": 1.1342657342657343,
      "grad_norm": 1.9914929866790771,
      "learning_rate": 4.79822130164305e-06,
      "loss": 0.0231,
      "step": 271
    },
    {
      "epoch": 1.1384615384615384,
      "grad_norm": 0.8240312337875366,
      "learning_rate": 4.7958871591483305e-06,
      "loss": 0.0158,
      "step": 272
    },
    {
      "epoch": 1.1426573426573428,
      "grad_norm": 0.8062412738800049,
      "learning_rate": 4.793540168109146e-06,
      "loss": 0.0165,
      "step": 273
    },
    {
      "epoch": 1.1468531468531469,
      "grad_norm": 0.8970699906349182,
      "learning_rate": 4.791180341660035e-06,
      "loss": 0.0223,
      "step": 274
    },
    {
      "epoch": 1.151048951048951,
      "grad_norm": 1.1959490776062012,
      "learning_rate": 4.788807693007367e-06,
      "loss": 0.0185,
      "step": 275
    },
    {
      "epoch": 1.1552447552447553,
      "grad_norm": 1.3478264808654785,
      "learning_rate": 4.786422235429269e-06,
      "loss": 0.0219,
      "step": 276
    },
    {
      "epoch": 1.1594405594405595,
      "grad_norm": 2.527574300765991,
      "learning_rate": 4.784023982275551e-06,
      "loss": 0.0171,
      "step": 277
    },
    {
      "epoch": 1.1636363636363636,
      "grad_norm": 2.0560972690582275,
      "learning_rate": 4.781612946967632e-06,
      "loss": 0.0232,
      "step": 278
    },
    {
      "epoch": 1.167832167832168,
      "grad_norm": 1.2283650636672974,
      "learning_rate": 4.779189142998463e-06,
      "loss": 0.0228,
      "step": 279
    },
    {
      "epoch": 1.172027972027972,
      "grad_norm": 1.0990374088287354,
      "learning_rate": 4.776752583932455e-06,
      "loss": 0.0207,
      "step": 280
    },
    {
      "epoch": 1.1762237762237762,
      "grad_norm": 0.762873113155365,
      "learning_rate": 4.774303283405396e-06,
      "loss": 0.0192,
      "step": 281
    },
    {
      "epoch": 1.1804195804195805,
      "grad_norm": 3.5389952659606934,
      "learning_rate": 4.771841255124385e-06,
      "loss": 0.0193,
      "step": 282
    },
    {
      "epoch": 1.1846153846153846,
      "grad_norm": 1.5018829107284546,
      "learning_rate": 4.769366512867747e-06,
      "loss": 0.0177,
      "step": 283
    },
    {
      "epoch": 1.1888111888111887,
      "grad_norm": 0.9355039596557617,
      "learning_rate": 4.766879070484957e-06,
      "loss": 0.0222,
      "step": 284
    },
    {
      "epoch": 1.193006993006993,
      "grad_norm": 0.9899689555168152,
      "learning_rate": 4.764378941896568e-06,
      "loss": 0.0237,
      "step": 285
    },
    {
      "epoch": 1.1972027972027972,
      "grad_norm": 0.7639883160591125,
      "learning_rate": 4.761866141094126e-06,
      "loss": 0.0206,
      "step": 286
    },
    {
      "epoch": 1.2013986013986013,
      "grad_norm": 0.9030625224113464,
      "learning_rate": 4.759340682140095e-06,
      "loss": 0.0147,
      "step": 287
    },
    {
      "epoch": 1.2055944055944057,
      "grad_norm": 1.505378246307373,
      "learning_rate": 4.756802579167781e-06,
      "loss": 0.0206,
      "step": 288
    },
    {
      "epoch": 1.2097902097902098,
      "grad_norm": 0.5766127109527588,
      "learning_rate": 4.754251846381248e-06,
      "loss": 0.0138,
      "step": 289
    },
    {
      "epoch": 1.213986013986014,
      "grad_norm": 0.937792181968689,
      "learning_rate": 4.75168849805524e-06,
      "loss": 0.0191,
      "step": 290
    },
    {
      "epoch": 1.2181818181818183,
      "grad_norm": 0.8431217074394226,
      "learning_rate": 4.749112548535104e-06,
      "loss": 0.0192,
      "step": 291
    },
    {
      "epoch": 1.2223776223776224,
      "grad_norm": 0.7842226624488831,
      "learning_rate": 4.746524012236706e-06,
      "loss": 0.0151,
      "step": 292
    },
    {
      "epoch": 1.2265734265734265,
      "grad_norm": 0.9969326853752136,
      "learning_rate": 4.743922903646351e-06,
      "loss": 0.0171,
      "step": 293
    },
    {
      "epoch": 1.2307692307692308,
      "grad_norm": 1.2452991008758545,
      "learning_rate": 4.741309237320703e-06,
      "loss": 0.0203,
      "step": 294
    },
    {
      "epoch": 1.234965034965035,
      "grad_norm": 0.9601305723190308,
      "learning_rate": 4.738683027886706e-06,
      "loss": 0.0144,
      "step": 295
    },
    {
      "epoch": 1.239160839160839,
      "grad_norm": 1.095999002456665,
      "learning_rate": 4.736044290041496e-06,
      "loss": 0.0136,
      "step": 296
    },
    {
      "epoch": 1.2433566433566434,
      "grad_norm": 0.9043452143669128,
      "learning_rate": 4.733393038552323e-06,
      "loss": 0.0216,
      "step": 297
    },
    {
      "epoch": 1.2475524475524475,
      "grad_norm": 1.330333948135376,
      "learning_rate": 4.730729288256468e-06,
      "loss": 0.0186,
      "step": 298
    },
    {
      "epoch": 1.2517482517482517,
      "grad_norm": 12.428945541381836,
      "learning_rate": 4.72805305406116e-06,
      "loss": 0.0224,
      "step": 299
    },
    {
      "epoch": 1.255944055944056,
      "grad_norm": 4.790683746337891,
      "learning_rate": 4.725364350943492e-06,
      "loss": 0.0165,
      "step": 300
    },
    {
      "epoch": 1.2601398601398601,
      "grad_norm": 4.931190490722656,
      "learning_rate": 4.722663193950336e-06,
      "loss": 0.0177,
      "step": 301
    },
    {
      "epoch": 1.2643356643356642,
      "grad_norm": 1.1879394054412842,
      "learning_rate": 4.719949598198258e-06,
      "loss": 0.0167,
      "step": 302
    },
    {
      "epoch": 1.2685314685314686,
      "grad_norm": 0.9889580607414246,
      "learning_rate": 4.717223578873442e-06,
      "loss": 0.0154,
      "step": 303
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.7459793090820312,
      "learning_rate": 4.714485151231593e-06,
      "loss": 0.0115,
      "step": 304
    },
    {
      "epoch": 1.2769230769230768,
      "grad_norm": 1.0093220472335815,
      "learning_rate": 4.711734330597857e-06,
      "loss": 0.014,
      "step": 305
    },
    {
      "epoch": 1.2811188811188812,
      "grad_norm": 0.701112687587738,
      "learning_rate": 4.708971132366739e-06,
      "loss": 0.017,
      "step": 306
    },
    {
      "epoch": 1.2853146853146853,
      "grad_norm": 1.1203210353851318,
      "learning_rate": 4.7061955720020096e-06,
      "loss": 0.0142,
      "step": 307
    },
    {
      "epoch": 1.2895104895104894,
      "grad_norm": 1.1437033414840698,
      "learning_rate": 4.703407665036622e-06,
      "loss": 0.0213,
      "step": 308
    },
    {
      "epoch": 1.2937062937062938,
      "grad_norm": 1.3569507598876953,
      "learning_rate": 4.70060742707263e-06,
      "loss": 0.0127,
      "step": 309
    },
    {
      "epoch": 1.2979020979020979,
      "grad_norm": 0.8850122690200806,
      "learning_rate": 4.697794873781089e-06,
      "loss": 0.0162,
      "step": 310
    },
    {
      "epoch": 1.302097902097902,
      "grad_norm": 0.7463717460632324,
      "learning_rate": 4.69497002090198e-06,
      "loss": 0.0063,
      "step": 311
    },
    {
      "epoch": 1.3062937062937063,
      "grad_norm": 1.432754397392273,
      "learning_rate": 4.692132884244113e-06,
      "loss": 0.0144,
      "step": 312
    },
    {
      "epoch": 1.3104895104895105,
      "grad_norm": 1.0831643342971802,
      "learning_rate": 4.689283479685045e-06,
      "loss": 0.009,
      "step": 313
    },
    {
      "epoch": 1.3146853146853146,
      "grad_norm": 0.908507227897644,
      "learning_rate": 4.686421823170987e-06,
      "loss": 0.0105,
      "step": 314
    },
    {
      "epoch": 1.318881118881119,
      "grad_norm": 0.829806923866272,
      "learning_rate": 4.683547930716716e-06,
      "loss": 0.0132,
      "step": 315
    },
    {
      "epoch": 1.323076923076923,
      "grad_norm": 0.7971309423446655,
      "learning_rate": 4.680661818405485e-06,
      "loss": 0.0129,
      "step": 316
    },
    {
      "epoch": 1.3272727272727272,
      "grad_norm": 1.5599275827407837,
      "learning_rate": 4.677763502388934e-06,
      "loss": 0.0139,
      "step": 317
    },
    {
      "epoch": 1.3314685314685315,
      "grad_norm": 0.7332860231399536,
      "learning_rate": 4.674852998886998e-06,
      "loss": 0.0104,
      "step": 318
    },
    {
      "epoch": 1.3356643356643356,
      "grad_norm": 2.5700080394744873,
      "learning_rate": 4.67193032418782e-06,
      "loss": 0.0132,
      "step": 319
    },
    {
      "epoch": 1.3398601398601397,
      "grad_norm": 1.2639752626419067,
      "learning_rate": 4.668995494647653e-06,
      "loss": 0.0119,
      "step": 320
    },
    {
      "epoch": 1.344055944055944,
      "grad_norm": 0.846431314945221,
      "learning_rate": 4.666048526690775e-06,
      "loss": 0.0135,
      "step": 321
    },
    {
      "epoch": 1.3482517482517482,
      "grad_norm": 0.9121400713920593,
      "learning_rate": 4.663089436809395e-06,
      "loss": 0.015,
      "step": 322
    },
    {
      "epoch": 1.3524475524475523,
      "grad_norm": 0.9713990092277527,
      "learning_rate": 4.660118241563558e-06,
      "loss": 0.0139,
      "step": 323
    },
    {
      "epoch": 1.3566433566433567,
      "grad_norm": 1.3189496994018555,
      "learning_rate": 4.657134957581057e-06,
      "loss": 0.0085,
      "step": 324
    },
    {
      "epoch": 1.3608391608391608,
      "grad_norm": 1.2847219705581665,
      "learning_rate": 4.654139601557337e-06,
      "loss": 0.0089,
      "step": 325
    },
    {
      "epoch": 1.365034965034965,
      "grad_norm": 0.8150692582130432,
      "learning_rate": 4.651132190255401e-06,
      "loss": 0.0081,
      "step": 326
    },
    {
      "epoch": 1.3692307692307693,
      "grad_norm": 1.006005048751831,
      "learning_rate": 4.648112740505718e-06,
      "loss": 0.0147,
      "step": 327
    },
    {
      "epoch": 1.3734265734265734,
      "grad_norm": 1.7578229904174805,
      "learning_rate": 4.645081269206128e-06,
      "loss": 0.0153,
      "step": 328
    },
    {
      "epoch": 1.3776223776223775,
      "grad_norm": 1.0592750310897827,
      "learning_rate": 4.642037793321749e-06,
      "loss": 0.014,
      "step": 329
    },
    {
      "epoch": 1.3818181818181818,
      "grad_norm": 0.9236828684806824,
      "learning_rate": 4.638982329884878e-06,
      "loss": 0.0117,
      "step": 330
    },
    {
      "epoch": 1.386013986013986,
      "grad_norm": 1.0400675535202026,
      "learning_rate": 4.635914895994901e-06,
      "loss": 0.0096,
      "step": 331
    },
    {
      "epoch": 1.39020979020979,
      "grad_norm": 1.1736772060394287,
      "learning_rate": 4.632835508818192e-06,
      "loss": 0.0121,
      "step": 332
    },
    {
      "epoch": 1.3944055944055944,
      "grad_norm": 1.1377946138381958,
      "learning_rate": 4.629744185588022e-06,
      "loss": 0.0102,
      "step": 333
    },
    {
      "epoch": 1.3986013986013985,
      "grad_norm": 0.7523559927940369,
      "learning_rate": 4.626640943604459e-06,
      "loss": 0.0095,
      "step": 334
    },
    {
      "epoch": 1.4027972027972029,
      "grad_norm": 0.6026727557182312,
      "learning_rate": 4.623525800234272e-06,
      "loss": 0.0088,
      "step": 335
    },
    {
      "epoch": 1.406993006993007,
      "grad_norm": 1.142449975013733,
      "learning_rate": 4.620398772910833e-06,
      "loss": 0.0095,
      "step": 336
    },
    {
      "epoch": 1.4111888111888111,
      "grad_norm": 0.7902323007583618,
      "learning_rate": 4.617259879134025e-06,
      "loss": 0.008,
      "step": 337
    },
    {
      "epoch": 1.4153846153846155,
      "grad_norm": 1.2716599702835083,
      "learning_rate": 4.614109136470133e-06,
      "loss": 0.0128,
      "step": 338
    },
    {
      "epoch": 1.4195804195804196,
      "grad_norm": 1.1127336025238037,
      "learning_rate": 4.610946562551759e-06,
      "loss": 0.0152,
      "step": 339
    },
    {
      "epoch": 1.4237762237762237,
      "grad_norm": 0.9858589172363281,
      "learning_rate": 4.607772175077712e-06,
      "loss": 0.0159,
      "step": 340
    },
    {
      "epoch": 1.427972027972028,
      "grad_norm": 1.2081674337387085,
      "learning_rate": 4.604585991812914e-06,
      "loss": 0.0094,
      "step": 341
    },
    {
      "epoch": 1.4321678321678322,
      "grad_norm": 1.1321561336517334,
      "learning_rate": 4.601388030588303e-06,
      "loss": 0.0118,
      "step": 342
    },
    {
      "epoch": 1.4363636363636363,
      "grad_norm": 0.6680831909179688,
      "learning_rate": 4.5981783093007275e-06,
      "loss": 0.0072,
      "step": 343
    },
    {
      "epoch": 1.4405594405594406,
      "grad_norm": 0.9412718415260315,
      "learning_rate": 4.59495684591285e-06,
      "loss": 0.0141,
      "step": 344
    },
    {
      "epoch": 1.4447552447552447,
      "grad_norm": 8.079177856445312,
      "learning_rate": 4.591723658453047e-06,
      "loss": 0.0097,
      "step": 345
    },
    {
      "epoch": 1.4489510489510489,
      "grad_norm": 8.037176132202148,
      "learning_rate": 4.588478765015304e-06,
      "loss": 0.0079,
      "step": 346
    },
    {
      "epoch": 1.4531468531468532,
      "grad_norm": 10.418330192565918,
      "learning_rate": 4.585222183759121e-06,
      "loss": 0.0114,
      "step": 347
    },
    {
      "epoch": 1.4573426573426573,
      "grad_norm": 6.077114582061768,
      "learning_rate": 4.581953932909403e-06,
      "loss": 0.0134,
      "step": 348
    },
    {
      "epoch": 1.4615384615384617,
      "grad_norm": 0.7302384972572327,
      "learning_rate": 4.578674030756364e-06,
      "loss": 0.0076,
      "step": 349
    },
    {
      "epoch": 1.4657342657342658,
      "grad_norm": 0.8345720171928406,
      "learning_rate": 4.575382495655421e-06,
      "loss": 0.0159,
      "step": 350
    },
    {
      "epoch": 1.46993006993007,
      "grad_norm": 0.6626363396644592,
      "learning_rate": 4.572079346027096e-06,
      "loss": 0.0099,
      "step": 351
    },
    {
      "epoch": 1.4741258741258743,
      "grad_norm": 1.0112382173538208,
      "learning_rate": 4.5687646003569055e-06,
      "loss": 0.0076,
      "step": 352
    },
    {
      "epoch": 1.4783216783216784,
      "grad_norm": 0.6920218467712402,
      "learning_rate": 4.565438277195265e-06,
      "loss": 0.0075,
      "step": 353
    },
    {
      "epoch": 1.4825174825174825,
      "grad_norm": 0.8465420007705688,
      "learning_rate": 4.562100395157379e-06,
      "loss": 0.0084,
      "step": 354
    },
    {
      "epoch": 1.4867132867132868,
      "grad_norm": 0.7478057742118835,
      "learning_rate": 4.5587509729231405e-06,
      "loss": 0.0129,
      "step": 355
    },
    {
      "epoch": 1.490909090909091,
      "grad_norm": 0.6042934656143188,
      "learning_rate": 4.555390029237026e-06,
      "loss": 0.0069,
      "step": 356
    },
    {
      "epoch": 1.495104895104895,
      "grad_norm": 0.7616173028945923,
      "learning_rate": 4.552017582907986e-06,
      "loss": 0.0103,
      "step": 357
    },
    {
      "epoch": 1.4993006993006994,
      "grad_norm": 0.7291771173477173,
      "learning_rate": 4.54863365280935e-06,
      "loss": 0.0139,
      "step": 358
    },
    {
      "epoch": 1.5034965034965035,
      "grad_norm": 1.2063847780227661,
      "learning_rate": 4.545238257878708e-06,
      "loss": 0.0064,
      "step": 359
    },
    {
      "epoch": 1.5076923076923077,
      "grad_norm": 0.37422144412994385,
      "learning_rate": 4.541831417117815e-06,
      "loss": 0.009,
      "step": 360
    },
    {
      "epoch": 1.511888111888112,
      "grad_norm": 0.6277711391448975,
      "learning_rate": 4.53841314959248e-06,
      "loss": 0.003,
      "step": 361
    },
    {
      "epoch": 1.5160839160839161,
      "grad_norm": 0.9660812020301819,
      "learning_rate": 4.534983474432458e-06,
      "loss": 0.0098,
      "step": 362
    },
    {
      "epoch": 1.5202797202797202,
      "grad_norm": 0.5692089200019836,
      "learning_rate": 4.5315424108313485e-06,
      "loss": 0.005,
      "step": 363
    },
    {
      "epoch": 1.5244755244755246,
      "grad_norm": 0.2833493649959564,
      "learning_rate": 4.528089978046481e-06,
      "loss": 0.004,
      "step": 364
    },
    {
      "epoch": 1.5286713286713287,
      "grad_norm": 0.731679379940033,
      "learning_rate": 4.524626195398812e-06,
      "loss": 0.0051,
      "step": 365
    },
    {
      "epoch": 1.5328671328671328,
      "grad_norm": 0.6426528692245483,
      "learning_rate": 4.521151082272817e-06,
      "loss": 0.0051,
      "step": 366
    },
    {
      "epoch": 1.5370629370629372,
      "grad_norm": 0.6782161593437195,
      "learning_rate": 4.5176646581163775e-06,
      "loss": 0.0075,
      "step": 367
    },
    {
      "epoch": 1.5412587412587413,
      "grad_norm": 0.8598656058311462,
      "learning_rate": 4.514166942440679e-06,
      "loss": 0.0119,
      "step": 368
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 1.1501063108444214,
      "learning_rate": 4.5106579548200935e-06,
      "loss": 0.0068,
      "step": 369
    },
    {
      "epoch": 1.5496503496503498,
      "grad_norm": 1.291435956954956,
      "learning_rate": 4.50713771489208e-06,
      "loss": 0.0066,
      "step": 370
    },
    {
      "epoch": 1.5538461538461539,
      "grad_norm": 0.6325899362564087,
      "learning_rate": 4.503606242357064e-06,
      "loss": 0.0046,
      "step": 371
    },
    {
      "epoch": 1.558041958041958,
      "grad_norm": 1.2787585258483887,
      "learning_rate": 4.5000635569783365e-06,
      "loss": 0.009,
      "step": 372
    },
    {
      "epoch": 1.5622377622377623,
      "grad_norm": 1.0738176107406616,
      "learning_rate": 4.496509678581936e-06,
      "loss": 0.0102,
      "step": 373
    },
    {
      "epoch": 1.5664335664335665,
      "grad_norm": 0.6169814467430115,
      "learning_rate": 4.492944627056544e-06,
      "loss": 0.0084,
      "step": 374
    },
    {
      "epoch": 1.5706293706293706,
      "grad_norm": 3.8595056533813477,
      "learning_rate": 4.4893684223533675e-06,
      "loss": 0.0061,
      "step": 375
    },
    {
      "epoch": 1.574825174825175,
      "grad_norm": 6.41073751449585,
      "learning_rate": 4.4857810844860325e-06,
      "loss": 0.0086,
      "step": 376
    },
    {
      "epoch": 1.579020979020979,
      "grad_norm": 12.252816200256348,
      "learning_rate": 4.482182633530468e-06,
      "loss": 0.0044,
      "step": 377
    },
    {
      "epoch": 1.5832167832167832,
      "grad_norm": 5.570111274719238,
      "learning_rate": 4.4785730896247985e-06,
      "loss": 0.0091,
      "step": 378
    },
    {
      "epoch": 1.5874125874125875,
      "grad_norm": 2.8494083881378174,
      "learning_rate": 4.474952472969224e-06,
      "loss": 0.0046,
      "step": 379
    },
    {
      "epoch": 1.5916083916083916,
      "grad_norm": 1.0975168943405151,
      "learning_rate": 4.471320803825915e-06,
      "loss": 0.0083,
      "step": 380
    },
    {
      "epoch": 1.5958041958041957,
      "grad_norm": 0.6368414163589478,
      "learning_rate": 4.467678102518894e-06,
      "loss": 0.01,
      "step": 381
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6025326251983643,
      "learning_rate": 4.464024389433924e-06,
      "loss": 0.0088,
      "step": 382
    },
    {
      "epoch": 1.6041958041958042,
      "grad_norm": 0.8672336339950562,
      "learning_rate": 4.4603596850183915e-06,
      "loss": 0.0079,
      "step": 383
    },
    {
      "epoch": 1.6083916083916083,
      "grad_norm": 0.7727923393249512,
      "learning_rate": 4.4566840097811956e-06,
      "loss": 0.0077,
      "step": 384
    },
    {
      "epoch": 1.6125874125874127,
      "grad_norm": 0.7475773096084595,
      "learning_rate": 4.452997384292633e-06,
      "loss": 0.0089,
      "step": 385
    },
    {
      "epoch": 1.6167832167832168,
      "grad_norm": 0.5794192552566528,
      "learning_rate": 4.449299829184278e-06,
      "loss": 0.0062,
      "step": 386
    },
    {
      "epoch": 1.620979020979021,
      "grad_norm": 0.7462018132209778,
      "learning_rate": 4.445591365148875e-06,
      "loss": 0.0073,
      "step": 387
    },
    {
      "epoch": 1.6251748251748253,
      "grad_norm": 0.5339632630348206,
      "learning_rate": 4.4418720129402145e-06,
      "loss": 0.0036,
      "step": 388
    },
    {
      "epoch": 1.6293706293706294,
      "grad_norm": 0.4537692666053772,
      "learning_rate": 4.438141793373023e-06,
      "loss": 0.0049,
      "step": 389
    },
    {
      "epoch": 1.6335664335664335,
      "grad_norm": 0.9906306266784668,
      "learning_rate": 4.434400727322844e-06,
      "loss": 0.0072,
      "step": 390
    },
    {
      "epoch": 1.6377622377622378,
      "grad_norm": 0.48505276441574097,
      "learning_rate": 4.430648835725922e-06,
      "loss": 0.0045,
      "step": 391
    },
    {
      "epoch": 1.641958041958042,
      "grad_norm": 0.6711535453796387,
      "learning_rate": 4.426886139579083e-06,
      "loss": 0.0045,
      "step": 392
    },
    {
      "epoch": 1.646153846153846,
      "grad_norm": 0.6675395369529724,
      "learning_rate": 4.423112659939622e-06,
      "loss": 0.0097,
      "step": 393
    },
    {
      "epoch": 1.6503496503496504,
      "grad_norm": 0.5513535141944885,
      "learning_rate": 4.419328417925177e-06,
      "loss": 0.0027,
      "step": 394
    },
    {
      "epoch": 1.6545454545454545,
      "grad_norm": 0.5774904489517212,
      "learning_rate": 4.415533434713621e-06,
      "loss": 0.0045,
      "step": 395
    },
    {
      "epoch": 1.6587412587412587,
      "grad_norm": 0.39901164174079895,
      "learning_rate": 4.411727731542937e-06,
      "loss": 0.0021,
      "step": 396
    },
    {
      "epoch": 1.662937062937063,
      "grad_norm": 0.526216447353363,
      "learning_rate": 4.407911329711097e-06,
      "loss": 0.0077,
      "step": 397
    },
    {
      "epoch": 1.6671328671328671,
      "grad_norm": 1.100061297416687,
      "learning_rate": 4.404084250575952e-06,
      "loss": 0.0065,
      "step": 398
    },
    {
      "epoch": 1.6713286713286712,
      "grad_norm": 0.8158161640167236,
      "learning_rate": 4.4002465155551014e-06,
      "loss": 0.005,
      "step": 399
    },
    {
      "epoch": 1.6755244755244756,
      "grad_norm": 0.4041314423084259,
      "learning_rate": 4.39639814612578e-06,
      "loss": 0.0024,
      "step": 400
    },
    {
      "epoch": 1.6797202797202797,
      "grad_norm": 0.3537557125091553,
      "learning_rate": 4.392539163824738e-06,
      "loss": 0.003,
      "step": 401
    },
    {
      "epoch": 1.6839160839160838,
      "grad_norm": 0.6756988763809204,
      "learning_rate": 4.388669590248119e-06,
      "loss": 0.0044,
      "step": 402
    },
    {
      "epoch": 1.6881118881118882,
      "grad_norm": 1.2168368101119995,
      "learning_rate": 4.3847894470513356e-06,
      "loss": 0.0032,
      "step": 403
    },
    {
      "epoch": 1.6923076923076923,
      "grad_norm": 1.1175090074539185,
      "learning_rate": 4.3808987559489536e-06,
      "loss": 0.0036,
      "step": 404
    },
    {
      "epoch": 1.6965034965034964,
      "grad_norm": 0.3256933093070984,
      "learning_rate": 4.376997538714569e-06,
      "loss": 0.0029,
      "step": 405
    },
    {
      "epoch": 1.7006993006993008,
      "grad_norm": 0.9790744781494141,
      "learning_rate": 4.373085817180684e-06,
      "loss": 0.0067,
      "step": 406
    },
    {
      "epoch": 1.7048951048951049,
      "grad_norm": 1.3782650232315063,
      "learning_rate": 4.3691636132385885e-06,
      "loss": 0.0051,
      "step": 407
    },
    {
      "epoch": 1.709090909090909,
      "grad_norm": 0.7676941752433777,
      "learning_rate": 4.365230948838232e-06,
      "loss": 0.0075,
      "step": 408
    },
    {
      "epoch": 1.7132867132867133,
      "grad_norm": 0.4800433814525604,
      "learning_rate": 4.361287845988107e-06,
      "loss": 0.0031,
      "step": 409
    },
    {
      "epoch": 1.7174825174825175,
      "grad_norm": 0.5705649852752686,
      "learning_rate": 4.357334326755123e-06,
      "loss": 0.0023,
      "step": 410
    },
    {
      "epoch": 1.7216783216783216,
      "grad_norm": 0.6476736664772034,
      "learning_rate": 4.353370413264481e-06,
      "loss": 0.0035,
      "step": 411
    },
    {
      "epoch": 1.725874125874126,
      "grad_norm": 0.29256880283355713,
      "learning_rate": 4.349396127699552e-06,
      "loss": 0.0031,
      "step": 412
    },
    {
      "epoch": 1.73006993006993,
      "grad_norm": 0.44642290472984314,
      "learning_rate": 4.345411492301757e-06,
      "loss": 0.0025,
      "step": 413
    },
    {
      "epoch": 1.7342657342657342,
      "grad_norm": 0.43764063715934753,
      "learning_rate": 4.341416529370431e-06,
      "loss": 0.0024,
      "step": 414
    },
    {
      "epoch": 1.7384615384615385,
      "grad_norm": 0.47648465633392334,
      "learning_rate": 4.337411261262711e-06,
      "loss": 0.0044,
      "step": 415
    },
    {
      "epoch": 1.7426573426573426,
      "grad_norm": 0.7295253872871399,
      "learning_rate": 4.3333957103934025e-06,
      "loss": 0.0062,
      "step": 416
    },
    {
      "epoch": 1.7468531468531467,
      "grad_norm": 0.30801764130592346,
      "learning_rate": 4.329369899234857e-06,
      "loss": 0.0018,
      "step": 417
    },
    {
      "epoch": 1.751048951048951,
      "grad_norm": 0.40857842564582825,
      "learning_rate": 4.325333850316846e-06,
      "loss": 0.0054,
      "step": 418
    },
    {
      "epoch": 1.7552447552447552,
      "grad_norm": 1.0663392543792725,
      "learning_rate": 4.321287586226436e-06,
      "loss": 0.0042,
      "step": 419
    },
    {
      "epoch": 1.7594405594405593,
      "grad_norm": 0.7497619986534119,
      "learning_rate": 4.317231129607859e-06,
      "loss": 0.0024,
      "step": 420
    },
    {
      "epoch": 1.7636363636363637,
      "grad_norm": 0.6514341235160828,
      "learning_rate": 4.313164503162391e-06,
      "loss": 0.005,
      "step": 421
    },
    {
      "epoch": 1.7678321678321678,
      "grad_norm": 0.7208667993545532,
      "learning_rate": 4.309087729648217e-06,
      "loss": 0.003,
      "step": 422
    },
    {
      "epoch": 1.772027972027972,
      "grad_norm": 0.31625956296920776,
      "learning_rate": 4.3050008318803125e-06,
      "loss": 0.0027,
      "step": 423
    },
    {
      "epoch": 1.7762237762237763,
      "grad_norm": 0.38829004764556885,
      "learning_rate": 4.30090383273031e-06,
      "loss": 0.0039,
      "step": 424
    },
    {
      "epoch": 1.7804195804195804,
      "grad_norm": 0.8225162029266357,
      "learning_rate": 4.2967967551263725e-06,
      "loss": 0.0048,
      "step": 425
    },
    {
      "epoch": 1.7846153846153845,
      "grad_norm": 0.3218310475349426,
      "learning_rate": 4.292679622053066e-06,
      "loss": 0.0042,
      "step": 426
    },
    {
      "epoch": 1.7888111888111888,
      "grad_norm": 0.9199247360229492,
      "learning_rate": 4.288552456551229e-06,
      "loss": 0.0013,
      "step": 427
    },
    {
      "epoch": 1.793006993006993,
      "grad_norm": 0.6360734105110168,
      "learning_rate": 4.2844152817178476e-06,
      "loss": 0.0016,
      "step": 428
    },
    {
      "epoch": 1.797202797202797,
      "grad_norm": 0.1691218614578247,
      "learning_rate": 4.28026812070592e-06,
      "loss": 0.0027,
      "step": 429
    },
    {
      "epoch": 1.8013986013986014,
      "grad_norm": 0.503375768661499,
      "learning_rate": 4.276110996724332e-06,
      "loss": 0.0026,
      "step": 430
    },
    {
      "epoch": 1.8055944055944058,
      "grad_norm": 0.2944371998310089,
      "learning_rate": 4.271943933037726e-06,
      "loss": 0.0043,
      "step": 431
    },
    {
      "epoch": 1.8097902097902097,
      "grad_norm": 0.6249594688415527,
      "learning_rate": 4.267766952966369e-06,
      "loss": 0.0015,
      "step": 432
    },
    {
      "epoch": 1.813986013986014,
      "grad_norm": 0.7547388076782227,
      "learning_rate": 4.263580079886025e-06,
      "loss": 0.0032,
      "step": 433
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.32078686356544495,
      "learning_rate": 4.259383337227821e-06,
      "loss": 0.0021,
      "step": 434
    },
    {
      "epoch": 1.8223776223776222,
      "grad_norm": 0.4883444905281067,
      "learning_rate": 4.255176748478117e-06,
      "loss": 0.0016,
      "step": 435
    },
    {
      "epoch": 1.8265734265734266,
      "grad_norm": 0.22250854969024658,
      "learning_rate": 4.2509603371783776e-06,
      "loss": 0.0014,
      "step": 436
    },
    {
      "epoch": 1.830769230769231,
      "grad_norm": 0.5906175971031189,
      "learning_rate": 4.246734126925035e-06,
      "loss": 0.0042,
      "step": 437
    },
    {
      "epoch": 1.8349650349650348,
      "grad_norm": 0.2544255256652832,
      "learning_rate": 4.242498141369361e-06,
      "loss": 0.0012,
      "step": 438
    },
    {
      "epoch": 1.8391608391608392,
      "grad_norm": 0.6039454340934753,
      "learning_rate": 4.238252404217334e-06,
      "loss": 0.0042,
      "step": 439
    },
    {
      "epoch": 1.8433566433566435,
      "grad_norm": 0.7017889618873596,
      "learning_rate": 4.233996939229502e-06,
      "loss": 0.0039,
      "step": 440
    },
    {
      "epoch": 1.8475524475524474,
      "grad_norm": 0.3206017017364502,
      "learning_rate": 4.229731770220858e-06,
      "loss": 0.0008,
      "step": 441
    },
    {
      "epoch": 1.8517482517482518,
      "grad_norm": 0.2899838984012604,
      "learning_rate": 4.225456921060698e-06,
      "loss": 0.0008,
      "step": 442
    },
    {
      "epoch": 1.855944055944056,
      "grad_norm": 0.23223742842674255,
      "learning_rate": 4.221172415672492e-06,
      "loss": 0.0005,
      "step": 443
    },
    {
      "epoch": 1.86013986013986,
      "grad_norm": 0.41323575377464294,
      "learning_rate": 4.216878278033753e-06,
      "loss": 0.001,
      "step": 444
    },
    {
      "epoch": 1.8643356643356643,
      "grad_norm": 0.3661796450614929,
      "learning_rate": 4.2125745321758945e-06,
      "loss": 0.0003,
      "step": 445
    },
    {
      "epoch": 1.8685314685314687,
      "grad_norm": 0.577558159828186,
      "learning_rate": 4.208261202184104e-06,
      "loss": 0.0008,
      "step": 446
    },
    {
      "epoch": 1.8727272727272726,
      "grad_norm": 0.3006725013256073,
      "learning_rate": 4.203938312197203e-06,
      "loss": 0.0012,
      "step": 447
    },
    {
      "epoch": 1.876923076923077,
      "grad_norm": 0.40426135063171387,
      "learning_rate": 4.199605886407515e-06,
      "loss": 0.0026,
      "step": 448
    },
    {
      "epoch": 1.8811188811188813,
      "grad_norm": 0.7835168242454529,
      "learning_rate": 4.195263949060728e-06,
      "loss": 0.0038,
      "step": 449
    },
    {
      "epoch": 1.8853146853146852,
      "grad_norm": 0.32763028144836426,
      "learning_rate": 4.190912524455762e-06,
      "loss": 0.0011,
      "step": 450
    },
    {
      "epoch": 1.8895104895104895,
      "grad_norm": 0.12388205528259277,
      "learning_rate": 4.186551636944628e-06,
      "loss": 0.0002,
      "step": 451
    },
    {
      "epoch": 1.8937062937062938,
      "grad_norm": 0.29386594891548157,
      "learning_rate": 4.1821813109322975e-06,
      "loss": 0.0009,
      "step": 452
    },
    {
      "epoch": 1.8979020979020977,
      "grad_norm": 0.8328471183776855,
      "learning_rate": 4.177801570876561e-06,
      "loss": 0.0008,
      "step": 453
    },
    {
      "epoch": 1.902097902097902,
      "grad_norm": 0.35669347643852234,
      "learning_rate": 4.1734124412878915e-06,
      "loss": 0.0006,
      "step": 454
    },
    {
      "epoch": 1.9062937062937064,
      "grad_norm": 0.20081250369548798,
      "learning_rate": 4.169013946729312e-06,
      "loss": 0.0009,
      "step": 455
    },
    {
      "epoch": 1.9104895104895103,
      "grad_norm": 0.4637051820755005,
      "learning_rate": 4.164606111816256e-06,
      "loss": 0.0017,
      "step": 456
    },
    {
      "epoch": 1.9146853146853147,
      "grad_norm": 0.38190585374832153,
      "learning_rate": 4.160188961216423e-06,
      "loss": 0.0005,
      "step": 457
    },
    {
      "epoch": 1.918881118881119,
      "grad_norm": 0.13845530152320862,
      "learning_rate": 4.155762519649654e-06,
      "loss": 0.0001,
      "step": 458
    },
    {
      "epoch": 1.9230769230769231,
      "grad_norm": 0.09175657480955124,
      "learning_rate": 4.151326811887778e-06,
      "loss": 0.0002,
      "step": 459
    },
    {
      "epoch": 1.9272727272727272,
      "grad_norm": 0.37238427996635437,
      "learning_rate": 4.146881862754485e-06,
      "loss": 0.0011,
      "step": 460
    },
    {
      "epoch": 1.9314685314685316,
      "grad_norm": 0.08203031122684479,
      "learning_rate": 4.142427697125181e-06,
      "loss": 0.0006,
      "step": 461
    },
    {
      "epoch": 1.9356643356643357,
      "grad_norm": 0.4510544240474701,
      "learning_rate": 4.137964339926852e-06,
      "loss": 0.0004,
      "step": 462
    },
    {
      "epoch": 1.9398601398601398,
      "grad_norm": 0.23816390335559845,
      "learning_rate": 4.133491816137921e-06,
      "loss": 0.0019,
      "step": 463
    },
    {
      "epoch": 1.9440559440559442,
      "grad_norm": 0.42791134119033813,
      "learning_rate": 4.129010150788112e-06,
      "loss": 0.0041,
      "step": 464
    },
    {
      "epoch": 1.9482517482517483,
      "grad_norm": 0.677318274974823,
      "learning_rate": 4.1245193689583085e-06,
      "loss": 0.0026,
      "step": 465
    },
    {
      "epoch": 1.9524475524475524,
      "grad_norm": 0.21399132907390594,
      "learning_rate": 4.120019495780412e-06,
      "loss": 0.0002,
      "step": 466
    },
    {
      "epoch": 1.9566433566433568,
      "grad_norm": 0.16479253768920898,
      "learning_rate": 4.115510556437201e-06,
      "loss": 0.0008,
      "step": 467
    },
    {
      "epoch": 1.9608391608391609,
      "grad_norm": 0.28853413462638855,
      "learning_rate": 4.110992576162193e-06,
      "loss": 0.0002,
      "step": 468
    },
    {
      "epoch": 1.965034965034965,
      "grad_norm": 0.273203581571579,
      "learning_rate": 4.106465580239502e-06,
      "loss": 0.0001,
      "step": 469
    },
    {
      "epoch": 1.9692307692307693,
      "grad_norm": 0.2718943655490875,
      "learning_rate": 4.101929594003694e-06,
      "loss": 0.0019,
      "step": 470
    },
    {
      "epoch": 1.9734265734265735,
      "grad_norm": 0.33305594325065613,
      "learning_rate": 4.09738464283965e-06,
      "loss": 0.0012,
      "step": 471
    },
    {
      "epoch": 1.9776223776223776,
      "grad_norm": 0.05372636765241623,
      "learning_rate": 4.092830752182423e-06,
      "loss": 0.0002,
      "step": 472
    },
    {
      "epoch": 1.981818181818182,
      "grad_norm": 0.054683368653059006,
      "learning_rate": 4.088267947517088e-06,
      "loss": 0.0003,
      "step": 473
    },
    {
      "epoch": 1.986013986013986,
      "grad_norm": 0.5915653109550476,
      "learning_rate": 4.083696254378615e-06,
      "loss": 0.0017,
      "step": 474
    },
    {
      "epoch": 1.9902097902097902,
      "grad_norm": 0.9250974059104919,
      "learning_rate": 4.079115698351709e-06,
      "loss": 0.0008,
      "step": 475
    },
    {
      "epoch": 1.9944055944055945,
      "grad_norm": 0.22745700180530548,
      "learning_rate": 4.074526305070679e-06,
      "loss": 0.0003,
      "step": 476
    },
    {
      "epoch": 1.9986013986013986,
      "grad_norm": 0.02354848012328148,
      "learning_rate": 4.069928100219287e-06,
      "loss": 0.0001,
      "step": 477
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.02354848012328148,
      "learning_rate": 4.065321109530612e-06,
      "loss": 0.0001,
      "step": 478
    },
    {
      "epoch": 2.0041958041958043,
      "grad_norm": 0.03878749534487724,
      "learning_rate": 4.060705358786897e-06,
      "loss": 0.0001,
      "step": 479
    },
    {
      "epoch": 2.0083916083916082,
      "grad_norm": 0.05548346787691116,
      "learning_rate": 4.056080873819412e-06,
      "loss": 0.0011,
      "step": 480
    },
    {
      "epoch": 2.0125874125874126,
      "grad_norm": 0.8392278552055359,
      "learning_rate": 4.051447680508304e-06,
      "loss": 0.0002,
      "step": 481
    },
    {
      "epoch": 2.016783216783217,
      "grad_norm": 0.02598468028008938,
      "learning_rate": 4.046805804782456e-06,
      "loss": 0.0001,
      "step": 482
    },
    {
      "epoch": 2.020979020979021,
      "grad_norm": 0.06524959951639175,
      "learning_rate": 4.042155272619345e-06,
      "loss": 0.0001,
      "step": 483
    },
    {
      "epoch": 2.025174825174825,
      "grad_norm": 0.09829440712928772,
      "learning_rate": 4.037496110044885e-06,
      "loss": 0.0003,
      "step": 484
    },
    {
      "epoch": 2.0293706293706295,
      "grad_norm": 0.03581927344202995,
      "learning_rate": 4.032828343133291e-06,
      "loss": 0.0001,
      "step": 485
    },
    {
      "epoch": 2.0335664335664334,
      "grad_norm": 0.12374325841665268,
      "learning_rate": 4.028151998006934e-06,
      "loss": 0.0015,
      "step": 486
    },
    {
      "epoch": 2.0377622377622377,
      "grad_norm": 0.1858854740858078,
      "learning_rate": 4.023467100836188e-06,
      "loss": 0.0011,
      "step": 487
    },
    {
      "epoch": 2.041958041958042,
      "grad_norm": 0.5057072043418884,
      "learning_rate": 4.018773677839289e-06,
      "loss": 0.0001,
      "step": 488
    },
    {
      "epoch": 2.046153846153846,
      "grad_norm": 0.1336149126291275,
      "learning_rate": 4.014071755282185e-06,
      "loss": 0.0003,
      "step": 489
    },
    {
      "epoch": 2.0503496503496503,
      "grad_norm": 0.7089390158653259,
      "learning_rate": 4.00936135947839e-06,
      "loss": 0.0001,
      "step": 490
    },
    {
      "epoch": 2.0545454545454547,
      "grad_norm": 0.14063000679016113,
      "learning_rate": 4.00464251678884e-06,
      "loss": 0.0001,
      "step": 491
    },
    {
      "epoch": 2.0587412587412586,
      "grad_norm": 0.09568943083286285,
      "learning_rate": 3.999915253621739e-06,
      "loss": 0.0006,
      "step": 492
    },
    {
      "epoch": 2.062937062937063,
      "grad_norm": 1.1405733823776245,
      "learning_rate": 3.995179596432418e-06,
      "loss": 0.0003,
      "step": 493
    },
    {
      "epoch": 2.0671328671328673,
      "grad_norm": 0.0644894391298294,
      "learning_rate": 3.99043557172318e-06,
      "loss": 0.0001,
      "step": 494
    },
    {
      "epoch": 2.071328671328671,
      "grad_norm": 0.008691688999533653,
      "learning_rate": 3.98568320604316e-06,
      "loss": 0.0001,
      "step": 495
    },
    {
      "epoch": 2.0755244755244755,
      "grad_norm": 0.22712311148643494,
      "learning_rate": 3.980922525988167e-06,
      "loss": 0.0004,
      "step": 496
    },
    {
      "epoch": 2.07972027972028,
      "grad_norm": 0.04054834693670273,
      "learning_rate": 3.976153558200543e-06,
      "loss": 0.0001,
      "step": 497
    },
    {
      "epoch": 2.0839160839160837,
      "grad_norm": 0.018998222425580025,
      "learning_rate": 3.971376329369011e-06,
      "loss": 0.0001,
      "step": 498
    },
    {
      "epoch": 2.088111888111888,
      "grad_norm": 0.018872397020459175,
      "learning_rate": 3.9665908662285235e-06,
      "loss": 0.0001,
      "step": 499
    },
    {
      "epoch": 2.0923076923076924,
      "grad_norm": 0.11350614577531815,
      "learning_rate": 3.961797195560118e-06,
      "loss": 0.0002,
      "step": 500
    },
    {
      "epoch": 2.0965034965034963,
      "grad_norm": 0.02821405790746212,
      "learning_rate": 3.956995344190762e-06,
      "loss": 0.0002,
      "step": 501
    },
    {
      "epoch": 2.1006993006993007,
      "grad_norm": 0.12058704346418381,
      "learning_rate": 3.952185338993202e-06,
      "loss": 0.0001,
      "step": 502
    },
    {
      "epoch": 2.104895104895105,
      "grad_norm": 0.11725764721632004,
      "learning_rate": 3.947367206885823e-06,
      "loss": 0.0006,
      "step": 503
    },
    {
      "epoch": 2.109090909090909,
      "grad_norm": 0.750499963760376,
      "learning_rate": 3.942540974832486e-06,
      "loss": 0.0001,
      "step": 504
    },
    {
      "epoch": 2.1132867132867132,
      "grad_norm": 0.028665080666542053,
      "learning_rate": 3.937706669842382e-06,
      "loss": 0.0019,
      "step": 505
    },
    {
      "epoch": 2.1174825174825176,
      "grad_norm": 0.34487929940223694,
      "learning_rate": 3.932864318969882e-06,
      "loss": 0.0002,
      "step": 506
    },
    {
      "epoch": 2.1216783216783215,
      "grad_norm": 0.1372770220041275,
      "learning_rate": 3.928013949314386e-06,
      "loss": 0.0,
      "step": 507
    },
    {
      "epoch": 2.125874125874126,
      "grad_norm": 0.014907466247677803,
      "learning_rate": 3.9231555880201655e-06,
      "loss": 0.0,
      "step": 508
    },
    {
      "epoch": 2.13006993006993,
      "grad_norm": 0.32426658272743225,
      "learning_rate": 3.91828926227622e-06,
      "loss": 0.0029,
      "step": 509
    },
    {
      "epoch": 2.134265734265734,
      "grad_norm": 0.3833695948123932,
      "learning_rate": 3.913414999316118e-06,
      "loss": 0.0001,
      "step": 510
    },
    {
      "epoch": 2.1384615384615384,
      "grad_norm": 0.30836614966392517,
      "learning_rate": 3.90853282641785e-06,
      "loss": 0.0002,
      "step": 511
    },
    {
      "epoch": 2.1426573426573428,
      "grad_norm": 0.011546188965439796,
      "learning_rate": 3.903642770903671e-06,
      "loss": 0.0003,
      "step": 512
    },
    {
      "epoch": 2.1468531468531467,
      "grad_norm": 0.2811814248561859,
      "learning_rate": 3.8987448601399495e-06,
      "loss": 0.0004,
      "step": 513
    },
    {
      "epoch": 2.151048951048951,
      "grad_norm": 1.190117359161377,
      "learning_rate": 3.893839121537015e-06,
      "loss": 0.0,
      "step": 514
    },
    {
      "epoch": 2.1552447552447553,
      "grad_norm": 0.01484633982181549,
      "learning_rate": 3.888925582549006e-06,
      "loss": 0.0033,
      "step": 515
    },
    {
      "epoch": 2.1594405594405592,
      "grad_norm": 0.5208448171615601,
      "learning_rate": 3.884004270673711e-06,
      "loss": 0.0001,
      "step": 516
    },
    {
      "epoch": 2.1636363636363636,
      "grad_norm": 0.010496806353330612,
      "learning_rate": 3.879075213452422e-06,
      "loss": 0.0013,
      "step": 517
    },
    {
      "epoch": 2.167832167832168,
      "grad_norm": 1.2324620485305786,
      "learning_rate": 3.874138438469775e-06,
      "loss": 0.0002,
      "step": 518
    },
    {
      "epoch": 2.172027972027972,
      "grad_norm": 0.017091237008571625,
      "learning_rate": 3.869193973353595e-06,
      "loss": 0.0002,
      "step": 519
    },
    {
      "epoch": 2.176223776223776,
      "grad_norm": 0.13021555542945862,
      "learning_rate": 3.864241845774746e-06,
      "loss": 0.0,
      "step": 520
    },
    {
      "epoch": 2.1804195804195805,
      "grad_norm": 0.025510359555482864,
      "learning_rate": 3.859282083446973e-06,
      "loss": 0.0002,
      "step": 521
    },
    {
      "epoch": 2.184615384615385,
      "grad_norm": 0.10301810503005981,
      "learning_rate": 3.854314714126749e-06,
      "loss": 0.0068,
      "step": 522
    },
    {
      "epoch": 2.1888111888111887,
      "grad_norm": 0.34572768211364746,
      "learning_rate": 3.8493397656131145e-06,
      "loss": 0.0014,
      "step": 523
    },
    {
      "epoch": 2.193006993006993,
      "grad_norm": 0.054788462817668915,
      "learning_rate": 3.844357265747531e-06,
      "loss": 0.0001,
      "step": 524
    },
    {
      "epoch": 2.197202797202797,
      "grad_norm": 0.30857688188552856,
      "learning_rate": 3.839367242413714e-06,
      "loss": 0.0021,
      "step": 525
    },
    {
      "epoch": 2.2013986013986013,
      "grad_norm": 0.07169409841299057,
      "learning_rate": 3.834369723537487e-06,
      "loss": 0.0001,
      "step": 526
    },
    {
      "epoch": 2.2055944055944057,
      "grad_norm": 0.1514081507921219,
      "learning_rate": 3.8293647370866206e-06,
      "loss": 0.0019,
      "step": 527
    },
    {
      "epoch": 2.20979020979021,
      "grad_norm": 0.2388014942407608,
      "learning_rate": 3.8243523110706736e-06,
      "loss": 0.0001,
      "step": 528
    },
    {
      "epoch": 2.213986013986014,
      "grad_norm": 0.048661720007658005,
      "learning_rate": 3.819332473540844e-06,
      "loss": 0.0005,
      "step": 529
    },
    {
      "epoch": 2.2181818181818183,
      "grad_norm": 0.2739057242870331,
      "learning_rate": 3.814305252589801e-06,
      "loss": 0.0002,
      "step": 530
    },
    {
      "epoch": 2.222377622377622,
      "grad_norm": 0.1036054790019989,
      "learning_rate": 3.8092706763515386e-06,
      "loss": 0.0002,
      "step": 531
    },
    {
      "epoch": 2.2265734265734265,
      "grad_norm": 0.06266741454601288,
      "learning_rate": 3.8042287730012117e-06,
      "loss": 0.0006,
      "step": 532
    },
    {
      "epoch": 2.230769230769231,
      "grad_norm": 0.06776294857263565,
      "learning_rate": 3.7991795707549806e-06,
      "loss": 0.0006,
      "step": 533
    },
    {
      "epoch": 2.234965034965035,
      "grad_norm": 0.11559757590293884,
      "learning_rate": 3.7941230978698516e-06,
      "loss": 0.0003,
      "step": 534
    },
    {
      "epoch": 2.239160839160839,
      "grad_norm": 0.12881553173065186,
      "learning_rate": 3.7890593826435207e-06,
      "loss": 0.0033,
      "step": 535
    },
    {
      "epoch": 2.2433566433566434,
      "grad_norm": 0.02464696206152439,
      "learning_rate": 3.7839884534142157e-06,
      "loss": 0.0002,
      "step": 536
    },
    {
      "epoch": 2.2475524475524473,
      "grad_norm": 0.10295409709215164,
      "learning_rate": 3.7789103385605338e-06,
      "loss": 0.0002,
      "step": 537
    },
    {
      "epoch": 2.2517482517482517,
      "grad_norm": 0.04006408154964447,
      "learning_rate": 3.7738250665012854e-06,
      "loss": 0.0002,
      "step": 538
    },
    {
      "epoch": 2.255944055944056,
      "grad_norm": 0.06407276540994644,
      "learning_rate": 3.768732665695338e-06,
      "loss": 0.0003,
      "step": 539
    },
    {
      "epoch": 2.2601398601398603,
      "grad_norm": 0.06269339472055435,
      "learning_rate": 3.7636331646414524e-06,
      "loss": 0.0002,
      "step": 540
    },
    {
      "epoch": 2.2643356643356642,
      "grad_norm": 0.11418299376964569,
      "learning_rate": 3.7585265918781234e-06,
      "loss": 0.0003,
      "step": 541
    },
    {
      "epoch": 2.2685314685314686,
      "grad_norm": 0.049903202801942825,
      "learning_rate": 3.7534129759834214e-06,
      "loss": 0.0001,
      "step": 542
    },
    {
      "epoch": 2.2727272727272725,
      "grad_norm": 0.027284186333417892,
      "learning_rate": 3.7482923455748354e-06,
      "loss": 0.0001,
      "step": 543
    },
    {
      "epoch": 2.276923076923077,
      "grad_norm": 0.0335276797413826,
      "learning_rate": 3.7431647293091076e-06,
      "loss": 0.0001,
      "step": 544
    },
    {
      "epoch": 2.281118881118881,
      "grad_norm": 0.016256438568234444,
      "learning_rate": 3.7380301558820745e-06,
      "loss": 0.0002,
      "step": 545
    },
    {
      "epoch": 2.2853146853146855,
      "grad_norm": 0.09327496588230133,
      "learning_rate": 3.7328886540285103e-06,
      "loss": 0.0002,
      "step": 546
    },
    {
      "epoch": 2.2895104895104894,
      "grad_norm": 0.05653909221291542,
      "learning_rate": 3.72774025252196e-06,
      "loss": 0.0008,
      "step": 547
    },
    {
      "epoch": 2.2937062937062938,
      "grad_norm": 1.1632578372955322,
      "learning_rate": 3.7225849801745835e-06,
      "loss": 0.0002,
      "step": 548
    },
    {
      "epoch": 2.297902097902098,
      "grad_norm": 0.14743748307228088,
      "learning_rate": 3.7174228658369905e-06,
      "loss": 0.0001,
      "step": 549
    },
    {
      "epoch": 2.302097902097902,
      "grad_norm": 0.01920212060213089,
      "learning_rate": 3.7122539383980818e-06,
      "loss": 0.0001,
      "step": 550
    },
    {
      "epoch": 2.3062937062937063,
      "grad_norm": 0.0132295535877347,
      "learning_rate": 3.7070782267848853e-06,
      "loss": 0.0001,
      "step": 551
    },
    {
      "epoch": 2.3104895104895107,
      "grad_norm": 0.01497499831020832,
      "learning_rate": 3.701895759962397e-06,
      "loss": 0.0001,
      "step": 552
    },
    {
      "epoch": 2.3146853146853146,
      "grad_norm": 0.0332457534968853,
      "learning_rate": 3.6967065669334168e-06,
      "loss": 0.0001,
      "step": 553
    },
    {
      "epoch": 2.318881118881119,
      "grad_norm": 0.2429577261209488,
      "learning_rate": 3.6915106767383873e-06,
      "loss": 0.0002,
      "step": 554
    },
    {
      "epoch": 2.3230769230769233,
      "grad_norm": 0.011774536222219467,
      "learning_rate": 3.686308118455228e-06,
      "loss": 0.0,
      "step": 555
    },
    {
      "epoch": 2.327272727272727,
      "grad_norm": 0.08883197605609894,
      "learning_rate": 3.6810989211991777e-06,
      "loss": 0.0001,
      "step": 556
    },
    {
      "epoch": 2.3314685314685315,
      "grad_norm": 0.019479207694530487,
      "learning_rate": 3.6758831141226294e-06,
      "loss": 0.0001,
      "step": 557
    },
    {
      "epoch": 2.335664335664336,
      "grad_norm": 0.024236135184764862,
      "learning_rate": 3.6706607264149647e-06,
      "loss": 0.0,
      "step": 558
    },
    {
      "epoch": 2.3398601398601397,
      "grad_norm": 0.006734573282301426,
      "learning_rate": 3.6654317873023936e-06,
      "loss": 0.0001,
      "step": 559
    },
    {
      "epoch": 2.344055944055944,
      "grad_norm": 0.02075507491827011,
      "learning_rate": 3.6601963260477923e-06,
      "loss": 0.0001,
      "step": 560
    },
    {
      "epoch": 2.3482517482517484,
      "grad_norm": 0.20709136128425598,
      "learning_rate": 3.6549543719505327e-06,
      "loss": 0.0011,
      "step": 561
    },
    {
      "epoch": 2.3524475524475523,
      "grad_norm": 0.006221509538590908,
      "learning_rate": 3.649705954346325e-06,
      "loss": 0.0,
      "step": 562
    },
    {
      "epoch": 2.3566433566433567,
      "grad_norm": 0.011554291471838951,
      "learning_rate": 3.6444511026070523e-06,
      "loss": 0.0001,
      "step": 563
    },
    {
      "epoch": 2.360839160839161,
      "grad_norm": 0.03156081587076187,
      "learning_rate": 3.6391898461406045e-06,
      "loss": 0.0,
      "step": 564
    },
    {
      "epoch": 2.365034965034965,
      "grad_norm": 0.005950749386101961,
      "learning_rate": 3.6339222143907125e-06,
      "loss": 0.0,
      "step": 565
    },
    {
      "epoch": 2.3692307692307693,
      "grad_norm": 0.011972339823842049,
      "learning_rate": 3.6286482368367893e-06,
      "loss": 0.0001,
      "step": 566
    },
    {
      "epoch": 2.3734265734265736,
      "grad_norm": 0.01295622531324625,
      "learning_rate": 3.6233679429937576e-06,
      "loss": 0.0,
      "step": 567
    },
    {
      "epoch": 2.3776223776223775,
      "grad_norm": 0.009713024832308292,
      "learning_rate": 3.6180813624118898e-06,
      "loss": 0.0,
      "step": 568
    },
    {
      "epoch": 2.381818181818182,
      "grad_norm": 0.013629626482725143,
      "learning_rate": 3.61278852467664e-06,
      "loss": 0.0,
      "step": 569
    },
    {
      "epoch": 2.386013986013986,
      "grad_norm": 0.040028758347034454,
      "learning_rate": 3.607489459408481e-06,
      "loss": 0.0001,
      "step": 570
    },
    {
      "epoch": 2.39020979020979,
      "grad_norm": 0.009488624520599842,
      "learning_rate": 3.602184196262736e-06,
      "loss": 0.0001,
      "step": 571
    },
    {
      "epoch": 2.3944055944055944,
      "grad_norm": 0.3338744044303894,
      "learning_rate": 3.5968727649294134e-06,
      "loss": 0.0001,
      "step": 572
    },
    {
      "epoch": 2.3986013986013988,
      "grad_norm": 0.009633059613406658,
      "learning_rate": 3.5915551951330403e-06,
      "loss": 0.0001,
      "step": 573
    },
    {
      "epoch": 2.4027972027972027,
      "grad_norm": 0.1984877586364746,
      "learning_rate": 3.5862315166324986e-06,
      "loss": 0.0,
      "step": 574
    },
    {
      "epoch": 2.406993006993007,
      "grad_norm": 0.016029132530093193,
      "learning_rate": 3.580901759220854e-06,
      "loss": 0.0,
      "step": 575
    },
    {
      "epoch": 2.4111888111888113,
      "grad_norm": 1.1584510803222656,
      "learning_rate": 3.575565952725193e-06,
      "loss": 0.0027,
      "step": 576
    },
    {
      "epoch": 2.4153846153846152,
      "grad_norm": 0.6399851441383362,
      "learning_rate": 3.570224127006456e-06,
      "loss": 0.0,
      "step": 577
    },
    {
      "epoch": 2.4195804195804196,
      "grad_norm": 0.020673168823122978,
      "learning_rate": 3.5648763119592687e-06,
      "loss": 0.0,
      "step": 578
    },
    {
      "epoch": 2.423776223776224,
      "grad_norm": 0.006884521339088678,
      "learning_rate": 3.5595225375117713e-06,
      "loss": 0.0014,
      "step": 579
    },
    {
      "epoch": 2.427972027972028,
      "grad_norm": 0.30941614508628845,
      "learning_rate": 3.55416283362546e-06,
      "loss": 0.0001,
      "step": 580
    },
    {
      "epoch": 2.432167832167832,
      "grad_norm": 0.05363042280077934,
      "learning_rate": 3.5487972302950115e-06,
      "loss": 0.0,
      "step": 581
    },
    {
      "epoch": 2.4363636363636365,
      "grad_norm": 0.0040152608416974545,
      "learning_rate": 3.5434257575481167e-06,
      "loss": 0.0,
      "step": 582
    },
    {
      "epoch": 2.4405594405594404,
      "grad_norm": 0.014275682158768177,
      "learning_rate": 3.538048445445316e-06,
      "loss": 0.0,
      "step": 583
    },
    {
      "epoch": 2.4447552447552447,
      "grad_norm": 0.0028259342070668936,
      "learning_rate": 3.5326653240798283e-06,
      "loss": 0.0,
      "step": 584
    },
    {
      "epoch": 2.448951048951049,
      "grad_norm": 0.009732532314956188,
      "learning_rate": 3.5272764235773802e-06,
      "loss": 0.001,
      "step": 585
    },
    {
      "epoch": 2.453146853146853,
      "grad_norm": 0.28131428360939026,
      "learning_rate": 3.5218817740960453e-06,
      "loss": 0.0001,
      "step": 586
    },
    {
      "epoch": 2.4573426573426573,
      "grad_norm": 0.05070802569389343,
      "learning_rate": 3.5164814058260667e-06,
      "loss": 0.0028,
      "step": 587
    },
    {
      "epoch": 2.4615384615384617,
      "grad_norm": 2.6187479496002197,
      "learning_rate": 3.5110753489896924e-06,
      "loss": 0.0005,
      "step": 588
    },
    {
      "epoch": 2.4657342657342656,
      "grad_norm": 0.6809223294258118,
      "learning_rate": 3.5056636338410066e-06,
      "loss": 0.0,
      "step": 589
    },
    {
      "epoch": 2.46993006993007,
      "grad_norm": 0.005053817760199308,
      "learning_rate": 3.500246290665758e-06,
      "loss": 0.0,
      "step": 590
    },
    {
      "epoch": 2.4741258741258743,
      "grad_norm": 0.013348731212317944,
      "learning_rate": 3.4948233497811917e-06,
      "loss": 0.0,
      "step": 591
    },
    {
      "epoch": 2.478321678321678,
      "grad_norm": 0.008500329218804836,
      "learning_rate": 3.4893948415358803e-06,
      "loss": 0.0,
      "step": 592
    },
    {
      "epoch": 2.4825174825174825,
      "grad_norm": 0.009890018962323666,
      "learning_rate": 3.483960796309552e-06,
      "loss": 0.0,
      "step": 593
    },
    {
      "epoch": 2.486713286713287,
      "grad_norm": 0.009032574482262135,
      "learning_rate": 3.4785212445129244e-06,
      "loss": 0.0001,
      "step": 594
    },
    {
      "epoch": 2.4909090909090907,
      "grad_norm": 0.05060562491416931,
      "learning_rate": 3.4730762165875287e-06,
      "loss": 0.0,
      "step": 595
    },
    {
      "epoch": 2.495104895104895,
      "grad_norm": 0.010769933462142944,
      "learning_rate": 3.4676257430055438e-06,
      "loss": 0.0001,
      "step": 596
    },
    {
      "epoch": 2.4993006993006994,
      "grad_norm": 0.11530755460262299,
      "learning_rate": 3.462169854269622e-06,
      "loss": 0.0,
      "step": 597
    },
    {
      "epoch": 2.5034965034965033,
      "grad_norm": 0.07504540681838989,
      "learning_rate": 3.4567085809127247e-06,
      "loss": 0.0001,
      "step": 598
    },
    {
      "epoch": 2.5076923076923077,
      "grad_norm": 0.010455475188791752,
      "learning_rate": 3.4512419534979445e-06,
      "loss": 0.0001,
      "step": 599
    },
    {
      "epoch": 2.511888111888112,
      "grad_norm": 0.01390789169818163,
      "learning_rate": 3.4457700026183378e-06,
      "loss": 0.0001,
      "step": 600
    },
    {
      "epoch": 2.5160839160839163,
      "grad_norm": 0.04701072350144386,
      "learning_rate": 3.440292758896754e-06,
      "loss": 0.0001,
      "step": 601
    },
    {
      "epoch": 2.5202797202797202,
      "grad_norm": 0.06625508517026901,
      "learning_rate": 3.43481025298566e-06,
      "loss": 0.0001,
      "step": 602
    },
    {
      "epoch": 2.5244755244755246,
      "grad_norm": 0.03539646044373512,
      "learning_rate": 3.429322515566975e-06,
      "loss": 0.0,
      "step": 603
    },
    {
      "epoch": 2.5286713286713285,
      "grad_norm": 0.011965765617787838,
      "learning_rate": 3.4238295773518924e-06,
      "loss": 0.0001,
      "step": 604
    },
    {
      "epoch": 2.532867132867133,
      "grad_norm": 0.04724963381886482,
      "learning_rate": 3.4183314690807147e-06,
      "loss": 0.0001,
      "step": 605
    },
    {
      "epoch": 2.537062937062937,
      "grad_norm": 0.13822439312934875,
      "learning_rate": 3.412828221522674e-06,
      "loss": 0.0001,
      "step": 606
    },
    {
      "epoch": 2.5412587412587415,
      "grad_norm": 0.010348084382712841,
      "learning_rate": 3.407319865475766e-06,
      "loss": 0.0,
      "step": 607
    },
    {
      "epoch": 2.5454545454545454,
      "grad_norm": 0.00970398262143135,
      "learning_rate": 3.4018064317665745e-06,
      "loss": 0.0,
      "step": 608
    },
    {
      "epoch": 2.5496503496503498,
      "grad_norm": 0.028090260922908783,
      "learning_rate": 3.3962879512500983e-06,
      "loss": 0.0,
      "step": 609
    },
    {
      "epoch": 2.5538461538461537,
      "grad_norm": 0.009530160576105118,
      "learning_rate": 3.390764454809582e-06,
      "loss": 0.0001,
      "step": 610
    },
    {
      "epoch": 2.558041958041958,
      "grad_norm": 0.03673725947737694,
      "learning_rate": 3.3852359733563393e-06,
      "loss": 0.0,
      "step": 611
    },
    {
      "epoch": 2.5622377622377623,
      "grad_norm": 0.02584812417626381,
      "learning_rate": 3.3797025378295826e-06,
      "loss": 0.0,
      "step": 612
    },
    {
      "epoch": 2.5664335664335667,
      "grad_norm": 0.0064214663580060005,
      "learning_rate": 3.37416417919625e-06,
      "loss": 0.0,
      "step": 613
    },
    {
      "epoch": 2.5706293706293706,
      "grad_norm": 0.005574128590524197,
      "learning_rate": 3.3686209284508264e-06,
      "loss": 0.0018,
      "step": 614
    },
    {
      "epoch": 2.574825174825175,
      "grad_norm": 1.5216443538665771,
      "learning_rate": 3.3630728166151807e-06,
      "loss": 0.0003,
      "step": 615
    },
    {
      "epoch": 2.579020979020979,
      "grad_norm": 0.03951020538806915,
      "learning_rate": 3.357519874738382e-06,
      "loss": 0.0006,
      "step": 616
    },
    {
      "epoch": 2.583216783216783,
      "grad_norm": 1.124062180519104,
      "learning_rate": 3.3519621338965306e-06,
      "loss": 0.0017,
      "step": 617
    },
    {
      "epoch": 2.5874125874125875,
      "grad_norm": 0.19897794723510742,
      "learning_rate": 3.346399625192584e-06,
      "loss": 0.0036,
      "step": 618
    },
    {
      "epoch": 2.591608391608392,
      "grad_norm": 0.2012002319097519,
      "learning_rate": 3.3408323797561832e-06,
      "loss": 0.0001,
      "step": 619
    },
    {
      "epoch": 2.5958041958041957,
      "grad_norm": 0.008610494434833527,
      "learning_rate": 3.3352604287434752e-06,
      "loss": 0.0,
      "step": 620
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.15457655489444733,
      "learning_rate": 3.3296838033369415e-06,
      "loss": 0.0005,
      "step": 621
    },
    {
      "epoch": 2.604195804195804,
      "grad_norm": 0.04072263091802597,
      "learning_rate": 3.3241025347452257e-06,
      "loss": 0.0,
      "step": 622
    },
    {
      "epoch": 2.6083916083916083,
      "grad_norm": 0.005995282903313637,
      "learning_rate": 3.318516654202953e-06,
      "loss": 0.0,
      "step": 623
    },
    {
      "epoch": 2.6125874125874127,
      "grad_norm": 0.028927762061357498,
      "learning_rate": 3.31292619297056e-06,
      "loss": 0.0001,
      "step": 624
    },
    {
      "epoch": 2.616783216783217,
      "grad_norm": 0.00831018015742302,
      "learning_rate": 3.3073311823341176e-06,
      "loss": 0.0,
      "step": 625
    },
    {
      "epoch": 2.620979020979021,
      "grad_norm": 0.011540360748767853,
      "learning_rate": 3.301731653605158e-06,
      "loss": 0.0012,
      "step": 626
    },
    {
      "epoch": 2.6251748251748253,
      "grad_norm": 0.07628694921731949,
      "learning_rate": 3.296127638120497e-06,
      "loss": 0.0001,
      "step": 627
    },
    {
      "epoch": 2.629370629370629,
      "grad_norm": 0.014886665157973766,
      "learning_rate": 3.29051916724206e-06,
      "loss": 0.0014,
      "step": 628
    },
    {
      "epoch": 2.6335664335664335,
      "grad_norm": 0.35742446780204773,
      "learning_rate": 3.284906272356707e-06,
      "loss": 0.0,
      "step": 629
    },
    {
      "epoch": 2.637762237762238,
      "grad_norm": 0.011443364433944225,
      "learning_rate": 3.2792889848760556e-06,
      "loss": 0.0001,
      "step": 630
    },
    {
      "epoch": 2.641958041958042,
      "grad_norm": 0.00856731366366148,
      "learning_rate": 3.2736673362363046e-06,
      "loss": 0.0,
      "step": 631
    },
    {
      "epoch": 2.646153846153846,
      "grad_norm": 0.011218360625207424,
      "learning_rate": 3.2680413578980623e-06,
      "loss": 0.0002,
      "step": 632
    },
    {
      "epoch": 2.6503496503496504,
      "grad_norm": 0.053541265428066254,
      "learning_rate": 3.2624110813461647e-06,
      "loss": 0.0001,
      "step": 633
    },
    {
      "epoch": 2.6545454545454543,
      "grad_norm": 0.06373050063848495,
      "learning_rate": 3.2567765380895027e-06,
      "loss": 0.0002,
      "step": 634
    },
    {
      "epoch": 2.6587412587412587,
      "grad_norm": 0.08870777487754822,
      "learning_rate": 3.2511377596608453e-06,
      "loss": 0.0001,
      "step": 635
    },
    {
      "epoch": 2.662937062937063,
      "grad_norm": 0.019052699208259583,
      "learning_rate": 3.245494777616664e-06,
      "loss": 0.0001,
      "step": 636
    },
    {
      "epoch": 2.6671328671328673,
      "grad_norm": 0.17224816977977753,
      "learning_rate": 3.2398476235369536e-06,
      "loss": 0.0012,
      "step": 637
    },
    {
      "epoch": 2.6713286713286712,
      "grad_norm": 0.016559280455112457,
      "learning_rate": 3.2341963290250565e-06,
      "loss": 0.0001,
      "step": 638
    },
    {
      "epoch": 2.6755244755244756,
      "grad_norm": 0.13853605091571808,
      "learning_rate": 3.2285409257074883e-06,
      "loss": 0.0005,
      "step": 639
    },
    {
      "epoch": 2.6797202797202795,
      "grad_norm": 1.7352495193481445,
      "learning_rate": 3.2228814452337587e-06,
      "loss": 0.0001,
      "step": 640
    },
    {
      "epoch": 2.683916083916084,
      "grad_norm": 0.01987098902463913,
      "learning_rate": 3.2172179192761922e-06,
      "loss": 0.0007,
      "step": 641
    },
    {
      "epoch": 2.688111888111888,
      "grad_norm": 0.1731719821691513,
      "learning_rate": 3.2115503795297547e-06,
      "loss": 0.0,
      "step": 642
    },
    {
      "epoch": 2.6923076923076925,
      "grad_norm": 0.011574595235288143,
      "learning_rate": 3.205878857711876e-06,
      "loss": 0.0001,
      "step": 643
    },
    {
      "epoch": 2.6965034965034964,
      "grad_norm": 0.01981521025300026,
      "learning_rate": 3.2002033855622683e-06,
      "loss": 0.0001,
      "step": 644
    },
    {
      "epoch": 2.7006993006993008,
      "grad_norm": 0.01182253472507,
      "learning_rate": 3.1945239948427513e-06,
      "loss": 0.0001,
      "step": 645
    },
    {
      "epoch": 2.7048951048951047,
      "grad_norm": 0.02396792732179165,
      "learning_rate": 3.1888407173370773e-06,
      "loss": 0.0001,
      "step": 646
    },
    {
      "epoch": 2.709090909090909,
      "grad_norm": 0.01550604309886694,
      "learning_rate": 3.1831535848507477e-06,
      "loss": 0.0001,
      "step": 647
    },
    {
      "epoch": 2.7132867132867133,
      "grad_norm": 0.013106832280755043,
      "learning_rate": 3.177462629210838e-06,
      "loss": 0.0,
      "step": 648
    },
    {
      "epoch": 2.7174825174825177,
      "grad_norm": 0.009590541943907738,
      "learning_rate": 3.1717678822658193e-06,
      "loss": 0.0001,
      "step": 649
    },
    {
      "epoch": 2.7216783216783216,
      "grad_norm": 0.031126253306865692,
      "learning_rate": 3.166069375885381e-06,
      "loss": 0.0001,
      "step": 650
    },
    {
      "epoch": 2.725874125874126,
      "grad_norm": 0.010289938189089298,
      "learning_rate": 3.160367141960251e-06,
      "loss": 0.0001,
      "step": 651
    },
    {
      "epoch": 2.73006993006993,
      "grad_norm": 0.010946341790258884,
      "learning_rate": 3.154661212402017e-06,
      "loss": 0.0,
      "step": 652
    },
    {
      "epoch": 2.734265734265734,
      "grad_norm": 0.14648909866809845,
      "learning_rate": 3.14895161914295e-06,
      "loss": 0.0001,
      "step": 653
    },
    {
      "epoch": 2.7384615384615385,
      "grad_norm": 0.04171716421842575,
      "learning_rate": 3.1432383941358247e-06,
      "loss": 0.0001,
      "step": 654
    },
    {
      "epoch": 2.742657342657343,
      "grad_norm": 0.032476991415023804,
      "learning_rate": 3.1375215693537377e-06,
      "loss": 0.0,
      "step": 655
    },
    {
      "epoch": 2.7468531468531467,
      "grad_norm": 0.11922305822372437,
      "learning_rate": 3.131801176789934e-06,
      "loss": 0.0006,
      "step": 656
    },
    {
      "epoch": 2.751048951048951,
      "grad_norm": 0.021055785939097404,
      "learning_rate": 3.126077248457622e-06,
      "loss": 0.0,
      "step": 657
    },
    {
      "epoch": 2.755244755244755,
      "grad_norm": 0.00565719697624445,
      "learning_rate": 3.120349816389801e-06,
      "loss": 0.0,
      "step": 658
    },
    {
      "epoch": 2.7594405594405593,
      "grad_norm": 0.012458412908017635,
      "learning_rate": 3.114618912639075e-06,
      "loss": 0.0,
      "step": 659
    },
    {
      "epoch": 2.7636363636363637,
      "grad_norm": 0.013393380679190159,
      "learning_rate": 3.1088845692774798e-06,
      "loss": 0.0001,
      "step": 660
    },
    {
      "epoch": 2.767832167832168,
      "grad_norm": 0.05774718150496483,
      "learning_rate": 3.1031468183962994e-06,
      "loss": 0.0,
      "step": 661
    },
    {
      "epoch": 2.772027972027972,
      "grad_norm": 0.004843105562031269,
      "learning_rate": 3.0974056921058843e-06,
      "loss": 0.0001,
      "step": 662
    },
    {
      "epoch": 2.7762237762237763,
      "grad_norm": 0.08403902500867844,
      "learning_rate": 3.091661222535479e-06,
      "loss": 0.0,
      "step": 663
    },
    {
      "epoch": 2.78041958041958,
      "grad_norm": 0.00799650140106678,
      "learning_rate": 3.0859134418330373e-06,
      "loss": 0.0,
      "step": 664
    },
    {
      "epoch": 2.7846153846153845,
      "grad_norm": 0.005614469293504953,
      "learning_rate": 3.0801623821650417e-06,
      "loss": 0.0,
      "step": 665
    },
    {
      "epoch": 2.788811188811189,
      "grad_norm": 0.023028664290905,
      "learning_rate": 3.0744080757163247e-06,
      "loss": 0.0001,
      "step": 666
    },
    {
      "epoch": 2.793006993006993,
      "grad_norm": 0.00975822750478983,
      "learning_rate": 3.068650554689892e-06,
      "loss": 0.0,
      "step": 667
    },
    {
      "epoch": 2.797202797202797,
      "grad_norm": 0.013870530761778355,
      "learning_rate": 3.0628898513067357e-06,
      "loss": 0.0,
      "step": 668
    },
    {
      "epoch": 2.8013986013986014,
      "grad_norm": 0.05066269636154175,
      "learning_rate": 3.0571259978056576e-06,
      "loss": 0.0,
      "step": 669
    },
    {
      "epoch": 2.8055944055944058,
      "grad_norm": 0.0033949597273021936,
      "learning_rate": 3.051359026443092e-06,
      "loss": 0.0,
      "step": 670
    },
    {
      "epoch": 2.8097902097902097,
      "grad_norm": 0.009085940197110176,
      "learning_rate": 3.045588969492918e-06,
      "loss": 0.0001,
      "step": 671
    },
    {
      "epoch": 2.813986013986014,
      "grad_norm": 0.00574359530583024,
      "learning_rate": 3.0398158592462847e-06,
      "loss": 0.0003,
      "step": 672
    },
    {
      "epoch": 2.8181818181818183,
      "grad_norm": 0.13129757344722748,
      "learning_rate": 3.0340397280114275e-06,
      "loss": 0.0001,
      "step": 673
    },
    {
      "epoch": 2.8223776223776222,
      "grad_norm": 0.01526966318488121,
      "learning_rate": 3.028260608113489e-06,
      "loss": 0.0,
      "step": 674
    },
    {
      "epoch": 2.8265734265734266,
      "grad_norm": 0.005011808127164841,
      "learning_rate": 3.0224785318943368e-06,
      "loss": 0.0,
      "step": 675
    },
    {
      "epoch": 2.830769230769231,
      "grad_norm": 0.0032294627744704485,
      "learning_rate": 3.0166935317123824e-06,
      "loss": 0.0,
      "step": 676
    },
    {
      "epoch": 2.834965034965035,
      "grad_norm": 0.00804353691637516,
      "learning_rate": 3.0109056399424034e-06,
      "loss": 0.0,
      "step": 677
    },
    {
      "epoch": 2.839160839160839,
      "grad_norm": 0.003976212814450264,
      "learning_rate": 3.0051148889753568e-06,
      "loss": 0.0,
      "step": 678
    },
    {
      "epoch": 2.8433566433566435,
      "grad_norm": 0.004317536484450102,
      "learning_rate": 2.999321311218203e-06,
      "loss": 0.0,
      "step": 679
    },
    {
      "epoch": 2.8475524475524474,
      "grad_norm": 0.017628613859415054,
      "learning_rate": 2.9935249390937184e-06,
      "loss": 0.0,
      "step": 680
    },
    {
      "epoch": 2.8517482517482518,
      "grad_norm": 0.005905550438910723,
      "learning_rate": 2.9877258050403214e-06,
      "loss": 0.0004,
      "step": 681
    },
    {
      "epoch": 2.855944055944056,
      "grad_norm": 0.17101143300533295,
      "learning_rate": 2.981923941511885e-06,
      "loss": 0.0,
      "step": 682
    },
    {
      "epoch": 2.86013986013986,
      "grad_norm": 0.007147667463868856,
      "learning_rate": 2.9761193809775584e-06,
      "loss": 0.0,
      "step": 683
    },
    {
      "epoch": 2.8643356643356643,
      "grad_norm": 0.0024307058192789555,
      "learning_rate": 2.970312155921585e-06,
      "loss": 0.0,
      "step": 684
    },
    {
      "epoch": 2.8685314685314687,
      "grad_norm": 0.0035131166223436594,
      "learning_rate": 2.964502298843116e-06,
      "loss": 0.0,
      "step": 685
    },
    {
      "epoch": 2.8727272727272726,
      "grad_norm": 0.003971125464886427,
      "learning_rate": 2.9586898422560355e-06,
      "loss": 0.0,
      "step": 686
    },
    {
      "epoch": 2.876923076923077,
      "grad_norm": 0.00530631560832262,
      "learning_rate": 2.9528748186887746e-06,
      "loss": 0.0,
      "step": 687
    },
    {
      "epoch": 2.8811188811188813,
      "grad_norm": 0.005819481331855059,
      "learning_rate": 2.9470572606841295e-06,
      "loss": 0.0,
      "step": 688
    },
    {
      "epoch": 2.885314685314685,
      "grad_norm": 0.001991645200178027,
      "learning_rate": 2.9412372007990813e-06,
      "loss": 0.0,
      "step": 689
    },
    {
      "epoch": 2.8895104895104895,
      "grad_norm": 0.006197805982083082,
      "learning_rate": 2.9354146716046107e-06,
      "loss": 0.0,
      "step": 690
    },
    {
      "epoch": 2.893706293706294,
      "grad_norm": 0.0028611517045646906,
      "learning_rate": 2.9295897056855187e-06,
      "loss": 0.0,
      "step": 691
    },
    {
      "epoch": 2.8979020979020977,
      "grad_norm": 0.0034895988646894693,
      "learning_rate": 2.9237623356402423e-06,
      "loss": 0.0,
      "step": 692
    },
    {
      "epoch": 2.902097902097902,
      "grad_norm": 0.003276653354987502,
      "learning_rate": 2.917932594080673e-06,
      "loss": 0.0,
      "step": 693
    },
    {
      "epoch": 2.9062937062937064,
      "grad_norm": 0.003894565626978874,
      "learning_rate": 2.912100513631974e-06,
      "loss": 0.0,
      "step": 694
    },
    {
      "epoch": 2.9104895104895103,
      "grad_norm": 0.0043608578853309155,
      "learning_rate": 2.906266126932399e-06,
      "loss": 0.0,
      "step": 695
    },
    {
      "epoch": 2.9146853146853147,
      "grad_norm": 0.004705101251602173,
      "learning_rate": 2.900429466633107e-06,
      "loss": 0.0002,
      "step": 696
    },
    {
      "epoch": 2.918881118881119,
      "grad_norm": 0.059318650513887405,
      "learning_rate": 2.894590565397979e-06,
      "loss": 0.0,
      "step": 697
    },
    {
      "epoch": 2.9230769230769234,
      "grad_norm": 0.011291037313640118,
      "learning_rate": 2.88874945590344e-06,
      "loss": 0.0,
      "step": 698
    },
    {
      "epoch": 2.9272727272727272,
      "grad_norm": 0.0038987151347100735,
      "learning_rate": 2.8829061708382735e-06,
      "loss": 0.0,
      "step": 699
    },
    {
      "epoch": 2.9314685314685316,
      "grad_norm": 0.007601252291351557,
      "learning_rate": 2.8770607429034352e-06,
      "loss": 0.0,
      "step": 700
    },
    {
      "epoch": 2.9356643356643355,
      "grad_norm": 0.0019491576822474599,
      "learning_rate": 2.8712132048118745e-06,
      "loss": 0.0,
      "step": 701
    },
    {
      "epoch": 2.93986013986014,
      "grad_norm": 0.0018927466589957476,
      "learning_rate": 2.8653635892883517e-06,
      "loss": 0.0,
      "step": 702
    },
    {
      "epoch": 2.944055944055944,
      "grad_norm": 0.0038515932392328978,
      "learning_rate": 2.85951192906925e-06,
      "loss": 0.0,
      "step": 703
    },
    {
      "epoch": 2.9482517482517485,
      "grad_norm": 0.0016372023383155465,
      "learning_rate": 2.8536582569023964e-06,
      "loss": 0.0,
      "step": 704
    },
    {
      "epoch": 2.9524475524475524,
      "grad_norm": 0.0038650059141218662,
      "learning_rate": 2.8478026055468793e-06,
      "loss": 0.0,
      "step": 705
    },
    {
      "epoch": 2.9566433566433568,
      "grad_norm": 0.003746791509911418,
      "learning_rate": 2.841945007772862e-06,
      "loss": 0.0,
      "step": 706
    },
    {
      "epoch": 2.9608391608391607,
      "grad_norm": 0.007748388685286045,
      "learning_rate": 2.8360854963613986e-06,
      "loss": 0.0,
      "step": 707
    },
    {
      "epoch": 2.965034965034965,
      "grad_norm": 0.00320612033829093,
      "learning_rate": 2.8302241041042564e-06,
      "loss": 0.0,
      "step": 708
    },
    {
      "epoch": 2.9692307692307693,
      "grad_norm": 0.0047491490840911865,
      "learning_rate": 2.824360863803725e-06,
      "loss": 0.0,
      "step": 709
    },
    {
      "epoch": 2.9734265734265737,
      "grad_norm": 0.0035452062729746103,
      "learning_rate": 2.818495808272439e-06,
      "loss": 0.0,
      "step": 710
    },
    {
      "epoch": 2.9776223776223776,
      "grad_norm": 0.010304715484380722,
      "learning_rate": 2.8126289703331895e-06,
      "loss": 0.0,
      "step": 711
    },
    {
      "epoch": 2.981818181818182,
      "grad_norm": 0.0020293130073696375,
      "learning_rate": 2.8067603828187446e-06,
      "loss": 0.0,
      "step": 712
    },
    {
      "epoch": 2.986013986013986,
      "grad_norm": 0.0017530432669445872,
      "learning_rate": 2.8008900785716612e-06,
      "loss": 0.0,
      "step": 713
    },
    {
      "epoch": 2.99020979020979,
      "grad_norm": 0.002541259629651904,
      "learning_rate": 2.795018090444106e-06,
      "loss": 0.0,
      "step": 714
    },
    {
      "epoch": 2.9944055944055945,
      "grad_norm": 0.013668973930180073,
      "learning_rate": 2.789144451297668e-06,
      "loss": 0.0,
      "step": 715
    },
    {
      "epoch": 2.998601398601399,
      "grad_norm": 0.02957094833254814,
      "learning_rate": 2.7832691940031755e-06,
      "loss": 0.0,
      "step": 716
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.24756021797657013,
      "learning_rate": 2.777392351440513e-06,
      "loss": 0.0003,
      "step": 717
    },
    {
      "epoch": 3.0041958041958043,
      "grad_norm": 0.0016691182972863317,
      "learning_rate": 2.771513956498436e-06,
      "loss": 0.0,
      "step": 718
    },
    {
      "epoch": 3.0083916083916082,
      "grad_norm": 0.0028321805875748396,
      "learning_rate": 2.76563404207439e-06,
      "loss": 0.0,
      "step": 719
    },
    {
      "epoch": 3.0125874125874126,
      "grad_norm": 0.0023459219373762608,
      "learning_rate": 2.759752641074322e-06,
      "loss": 0.0,
      "step": 720
    },
    {
      "epoch": 3.016783216783217,
      "grad_norm": 0.0050096469931304455,
      "learning_rate": 2.7538697864124973e-06,
      "loss": 0.0,
      "step": 721
    },
    {
      "epoch": 3.020979020979021,
      "grad_norm": 0.0029662717133760452,
      "learning_rate": 2.747985511011319e-06,
      "loss": 0.0,
      "step": 722
    },
    {
      "epoch": 3.025174825174825,
      "grad_norm": 0.0019866821821779013,
      "learning_rate": 2.7420998478011394e-06,
      "loss": 0.0,
      "step": 723
    },
    {
      "epoch": 3.0293706293706295,
      "grad_norm": 0.00763437245041132,
      "learning_rate": 2.7362128297200784e-06,
      "loss": 0.0,
      "step": 724
    },
    {
      "epoch": 3.0335664335664334,
      "grad_norm": 0.002217785222455859,
      "learning_rate": 2.730324489713837e-06,
      "loss": 0.0,
      "step": 725
    },
    {
      "epoch": 3.0377622377622377,
      "grad_norm": 0.002127251587808132,
      "learning_rate": 2.7244348607355163e-06,
      "loss": 0.0,
      "step": 726
    },
    {
      "epoch": 3.041958041958042,
      "grad_norm": 0.00227255723439157,
      "learning_rate": 2.718543975745428e-06,
      "loss": 0.0,
      "step": 727
    },
    {
      "epoch": 3.046153846153846,
      "grad_norm": 0.002066431799903512,
      "learning_rate": 2.712651867710914e-06,
      "loss": 0.0,
      "step": 728
    },
    {
      "epoch": 3.0503496503496503,
      "grad_norm": 0.0024342311080545187,
      "learning_rate": 2.706758569606163e-06,
      "loss": 0.0,
      "step": 729
    },
    {
      "epoch": 3.0545454545454547,
      "grad_norm": 0.001150021213106811,
      "learning_rate": 2.700864114412021e-06,
      "loss": 0.0,
      "step": 730
    },
    {
      "epoch": 3.0587412587412586,
      "grad_norm": 0.0014957587700337172,
      "learning_rate": 2.6949685351158097e-06,
      "loss": 0.0,
      "step": 731
    },
    {
      "epoch": 3.062937062937063,
      "grad_norm": 0.002742597833275795,
      "learning_rate": 2.6890718647111424e-06,
      "loss": 0.0,
      "step": 732
    },
    {
      "epoch": 3.0671328671328673,
      "grad_norm": 0.0019520387286320329,
      "learning_rate": 2.6831741361977388e-06,
      "loss": 0.0,
      "step": 733
    },
    {
      "epoch": 3.071328671328671,
      "grad_norm": 0.001996438018977642,
      "learning_rate": 2.6772753825812395e-06,
      "loss": 0.0,
      "step": 734
    },
    {
      "epoch": 3.0755244755244755,
      "grad_norm": 0.07907167822122574,
      "learning_rate": 2.671375636873022e-06,
      "loss": 0.0004,
      "step": 735
    },
    {
      "epoch": 3.07972027972028,
      "grad_norm": 0.06860248744487762,
      "learning_rate": 2.665474932090017e-06,
      "loss": 0.0003,
      "step": 736
    },
    {
      "epoch": 3.0839160839160837,
      "grad_norm": 0.0016444770153611898,
      "learning_rate": 2.659573301254521e-06,
      "loss": 0.0,
      "step": 737
    },
    {
      "epoch": 3.088111888111888,
      "grad_norm": 0.0016304998425766826,
      "learning_rate": 2.6536707773940133e-06,
      "loss": 0.0,
      "step": 738
    },
    {
      "epoch": 3.0923076923076924,
      "grad_norm": 0.0031452891416847706,
      "learning_rate": 2.6477673935409716e-06,
      "loss": 0.0,
      "step": 739
    },
    {
      "epoch": 3.0965034965034963,
      "grad_norm": 0.0016455184668302536,
      "learning_rate": 2.6418631827326857e-06,
      "loss": 0.0,
      "step": 740
    },
    {
      "epoch": 3.1006993006993007,
      "grad_norm": 0.007033637724816799,
      "learning_rate": 2.6359581780110732e-06,
      "loss": 0.0,
      "step": 741
    },
    {
      "epoch": 3.104895104895105,
      "grad_norm": 0.0015307165449485183,
      "learning_rate": 2.6300524124224947e-06,
      "loss": 0.0,
      "step": 742
    },
    {
      "epoch": 3.109090909090909,
      "grad_norm": 0.002308042487129569,
      "learning_rate": 2.6241459190175707e-06,
      "loss": 0.0,
      "step": 743
    },
    {
      "epoch": 3.1132867132867132,
      "grad_norm": 0.0023787126410752535,
      "learning_rate": 2.6182387308509927e-06,
      "loss": 0.0,
      "step": 744
    },
    {
      "epoch": 3.1174825174825176,
      "grad_norm": 0.022157154977321625,
      "learning_rate": 2.6123308809813395e-06,
      "loss": 0.0,
      "step": 745
    },
    {
      "epoch": 3.1216783216783215,
      "grad_norm": 0.0076769436709582806,
      "learning_rate": 2.606422402470896e-06,
      "loss": 0.0,
      "step": 746
    },
    {
      "epoch": 3.125874125874126,
      "grad_norm": 0.026079457253217697,
      "learning_rate": 2.6005133283854635e-06,
      "loss": 0.0,
      "step": 747
    },
    {
      "epoch": 3.13006993006993,
      "grad_norm": 0.0022884586360305548,
      "learning_rate": 2.5946036917941765e-06,
      "loss": 0.0,
      "step": 748
    },
    {
      "epoch": 3.134265734265734,
      "grad_norm": 0.002465897938236594,
      "learning_rate": 2.5886935257693156e-06,
      "loss": 0.0,
      "step": 749
    },
    {
      "epoch": 3.1384615384615384,
      "grad_norm": 0.0031883539631962776,
      "learning_rate": 2.582782863386129e-06,
      "loss": 0.0,
      "step": 750
    },
    {
      "epoch": 3.1426573426573428,
      "grad_norm": 0.0015031610382720828,
      "learning_rate": 2.5768717377226382e-06,
      "loss": 0.0,
      "step": 751
    },
    {
      "epoch": 3.1468531468531467,
      "grad_norm": 0.0016049172263592482,
      "learning_rate": 2.570960181859458e-06,
      "loss": 0.0,
      "step": 752
    },
    {
      "epoch": 3.151048951048951,
      "grad_norm": 0.0021165150683373213,
      "learning_rate": 2.5650482288796137e-06,
      "loss": 0.0,
      "step": 753
    },
    {
      "epoch": 3.1552447552447553,
      "grad_norm": 0.006701579783111811,
      "learning_rate": 2.5591359118683495e-06,
      "loss": 0.0,
      "step": 754
    },
    {
      "epoch": 3.1594405594405592,
      "grad_norm": 0.0013220134424045682,
      "learning_rate": 2.5532232639129497e-06,
      "loss": 0.0,
      "step": 755
    },
    {
      "epoch": 3.1636363636363636,
      "grad_norm": 0.0018707640701904893,
      "learning_rate": 2.547310318102548e-06,
      "loss": 0.0,
      "step": 756
    },
    {
      "epoch": 3.167832167832168,
      "grad_norm": 0.005933595821261406,
      "learning_rate": 2.541397107527947e-06,
      "loss": 0.0,
      "step": 757
    },
    {
      "epoch": 3.172027972027972,
      "grad_norm": 0.0023362054489552975,
      "learning_rate": 2.5354836652814297e-06,
      "loss": 0.0,
      "step": 758
    },
    {
      "epoch": 3.176223776223776,
      "grad_norm": 0.0035263069439679384,
      "learning_rate": 2.529570024456576e-06,
      "loss": 0.0,
      "step": 759
    },
    {
      "epoch": 3.1804195804195805,
      "grad_norm": 0.002927372930571437,
      "learning_rate": 2.5236562181480794e-06,
      "loss": 0.0,
      "step": 760
    },
    {
      "epoch": 3.184615384615385,
      "grad_norm": 0.0018356060609221458,
      "learning_rate": 2.5177422794515556e-06,
      "loss": 0.0,
      "step": 761
    },
    {
      "epoch": 3.1888111888111887,
      "grad_norm": 0.0017849173163995147,
      "learning_rate": 2.511828241463364e-06,
      "loss": 0.0,
      "step": 762
    },
    {
      "epoch": 3.193006993006993,
      "grad_norm": 0.0028886182699352503,
      "learning_rate": 2.505914137280417e-06,
      "loss": 0.0,
      "step": 763
    },
    {
      "epoch": 3.197202797202797,
      "grad_norm": 0.0028084160294383764,
      "learning_rate": 2.5e-06,
      "loss": 0.0,
      "step": 764
    },
    {
      "epoch": 3.2013986013986013,
      "grad_norm": 0.003333339234814048,
      "learning_rate": 2.4940858627195843e-06,
      "loss": 0.0,
      "step": 765
    },
    {
      "epoch": 3.2055944055944057,
      "grad_norm": 0.0017387164989486337,
      "learning_rate": 2.488171758536637e-06,
      "loss": 0.0,
      "step": 766
    },
    {
      "epoch": 3.20979020979021,
      "grad_norm": 0.0017168617341667414,
      "learning_rate": 2.482257720548445e-06,
      "loss": 0.0,
      "step": 767
    },
    {
      "epoch": 3.213986013986014,
      "grad_norm": 0.0027419906109571457,
      "learning_rate": 2.4763437818519205e-06,
      "loss": 0.0,
      "step": 768
    },
    {
      "epoch": 3.2181818181818183,
      "grad_norm": 0.0020276198629289865,
      "learning_rate": 2.4704299755434234e-06,
      "loss": 0.0,
      "step": 769
    },
    {
      "epoch": 3.222377622377622,
      "grad_norm": 0.0021964339539408684,
      "learning_rate": 2.464516334718571e-06,
      "loss": 0.0,
      "step": 770
    },
    {
      "epoch": 3.2265734265734265,
      "grad_norm": 0.0021198950707912445,
      "learning_rate": 2.4586028924720544e-06,
      "loss": 0.0,
      "step": 771
    },
    {
      "epoch": 3.230769230769231,
      "grad_norm": 0.002706088125705719,
      "learning_rate": 2.4526896818974534e-06,
      "loss": 0.0,
      "step": 772
    },
    {
      "epoch": 3.234965034965035,
      "grad_norm": 0.00930150505155325,
      "learning_rate": 2.446776736087051e-06,
      "loss": 0.0,
      "step": 773
    },
    {
      "epoch": 3.239160839160839,
      "grad_norm": 0.006174617446959019,
      "learning_rate": 2.440864088131651e-06,
      "loss": 0.0,
      "step": 774
    },
    {
      "epoch": 3.2433566433566434,
      "grad_norm": 0.003733342979103327,
      "learning_rate": 2.434951771120387e-06,
      "loss": 0.0,
      "step": 775
    },
    {
      "epoch": 3.2475524475524473,
      "grad_norm": 0.0016013550339266658,
      "learning_rate": 2.429039818140543e-06,
      "loss": 0.0,
      "step": 776
    },
    {
      "epoch": 3.2517482517482517,
      "grad_norm": 0.0020765475928783417,
      "learning_rate": 2.4231282622773626e-06,
      "loss": 0.0,
      "step": 777
    },
    {
      "epoch": 3.255944055944056,
      "grad_norm": 0.001071031205356121,
      "learning_rate": 2.4172171366138717e-06,
      "loss": 0.0,
      "step": 778
    },
    {
      "epoch": 3.2601398601398603,
      "grad_norm": 0.002402965445071459,
      "learning_rate": 2.411306474230685e-06,
      "loss": 0.0,
      "step": 779
    },
    {
      "epoch": 3.2643356643356642,
      "grad_norm": 0.0023972857743501663,
      "learning_rate": 2.405396308205825e-06,
      "loss": 0.0,
      "step": 780
    },
    {
      "epoch": 3.2685314685314686,
      "grad_norm": 0.004906000103801489,
      "learning_rate": 2.3994866716145365e-06,
      "loss": 0.0,
      "step": 781
    },
    {
      "epoch": 3.2727272727272725,
      "grad_norm": 0.0016749349888414145,
      "learning_rate": 2.3935775975291053e-06,
      "loss": 0.0,
      "step": 782
    },
    {
      "epoch": 3.276923076923077,
      "grad_norm": 0.0026932235341519117,
      "learning_rate": 2.3876691190186613e-06,
      "loss": 0.0,
      "step": 783
    },
    {
      "epoch": 3.281118881118881,
      "grad_norm": 0.0012396976817399263,
      "learning_rate": 2.381761269149009e-06,
      "loss": 0.0,
      "step": 784
    },
    {
      "epoch": 3.2853146853146855,
      "grad_norm": 0.0025537756737321615,
      "learning_rate": 2.3758540809824306e-06,
      "loss": 0.0,
      "step": 785
    },
    {
      "epoch": 3.2895104895104894,
      "grad_norm": 0.0018863689620047808,
      "learning_rate": 2.369947587577506e-06,
      "loss": 0.0,
      "step": 786
    },
    {
      "epoch": 3.2937062937062938,
      "grad_norm": 0.0036140403244644403,
      "learning_rate": 2.3640418219889276e-06,
      "loss": 0.0,
      "step": 787
    },
    {
      "epoch": 3.297902097902098,
      "grad_norm": 0.0013448689132928848,
      "learning_rate": 2.358136817267315e-06,
      "loss": 0.0,
      "step": 788
    },
    {
      "epoch": 3.302097902097902,
      "grad_norm": 0.0014878468355163932,
      "learning_rate": 2.352232606459029e-06,
      "loss": 0.0,
      "step": 789
    },
    {
      "epoch": 3.3062937062937063,
      "grad_norm": 0.0017818931955844164,
      "learning_rate": 2.346329222605987e-06,
      "loss": 0.0,
      "step": 790
    },
    {
      "epoch": 3.3104895104895107,
      "grad_norm": 0.002789202146232128,
      "learning_rate": 2.340426698745479e-06,
      "loss": 0.0,
      "step": 791
    },
    {
      "epoch": 3.3146853146853146,
      "grad_norm": 0.0018016956746578217,
      "learning_rate": 2.334525067909983e-06,
      "loss": 0.0,
      "step": 792
    },
    {
      "epoch": 3.318881118881119,
      "grad_norm": 0.0025564406532794237,
      "learning_rate": 2.3286243631269783e-06,
      "loss": 0.0,
      "step": 793
    },
    {
      "epoch": 3.3230769230769233,
      "grad_norm": 0.0012296305503696203,
      "learning_rate": 2.3227246174187617e-06,
      "loss": 0.0,
      "step": 794
    },
    {
      "epoch": 3.327272727272727,
      "grad_norm": 0.002000252716243267,
      "learning_rate": 2.3168258638022625e-06,
      "loss": 0.0,
      "step": 795
    },
    {
      "epoch": 3.3314685314685315,
      "grad_norm": 0.002654314273968339,
      "learning_rate": 2.3109281352888593e-06,
      "loss": 0.0,
      "step": 796
    },
    {
      "epoch": 3.335664335664336,
      "grad_norm": 0.002474359702318907,
      "learning_rate": 2.3050314648841916e-06,
      "loss": 0.0,
      "step": 797
    },
    {
      "epoch": 3.3398601398601397,
      "grad_norm": 0.11288738995790482,
      "learning_rate": 2.29913588558798e-06,
      "loss": 0.0001,
      "step": 798
    },
    {
      "epoch": 3.344055944055944,
      "grad_norm": 0.0026657001581043005,
      "learning_rate": 2.2932414303938378e-06,
      "loss": 0.0,
      "step": 799
    },
    {
      "epoch": 3.3482517482517484,
      "grad_norm": 0.002544724615290761,
      "learning_rate": 2.2873481322890866e-06,
      "loss": 0.0,
      "step": 800
    },
    {
      "epoch": 3.3524475524475523,
      "grad_norm": 0.007572051137685776,
      "learning_rate": 2.2814560242545726e-06,
      "loss": 0.0,
      "step": 801
    },
    {
      "epoch": 3.3566433566433567,
      "grad_norm": 0.14846807718276978,
      "learning_rate": 2.275565139264484e-06,
      "loss": 0.0001,
      "step": 802
    },
    {
      "epoch": 3.360839160839161,
      "grad_norm": 0.001332366606220603,
      "learning_rate": 2.2696755102861632e-06,
      "loss": 0.0,
      "step": 803
    },
    {
      "epoch": 3.365034965034965,
      "grad_norm": 0.0018846419407054782,
      "learning_rate": 2.263787170279922e-06,
      "loss": 0.0,
      "step": 804
    },
    {
      "epoch": 3.3692307692307693,
      "grad_norm": 0.0015413407236337662,
      "learning_rate": 2.2579001521988606e-06,
      "loss": 0.0,
      "step": 805
    },
    {
      "epoch": 3.3734265734265736,
      "grad_norm": 0.0012465774780139327,
      "learning_rate": 2.252014488988682e-06,
      "loss": 0.0,
      "step": 806
    },
    {
      "epoch": 3.3776223776223775,
      "grad_norm": 0.0022036691661924124,
      "learning_rate": 2.2461302135875035e-06,
      "loss": 0.0,
      "step": 807
    },
    {
      "epoch": 3.381818181818182,
      "grad_norm": 0.0017414194298908114,
      "learning_rate": 2.2402473589256793e-06,
      "loss": 0.0,
      "step": 808
    },
    {
      "epoch": 3.386013986013986,
      "grad_norm": 0.0017003871034830809,
      "learning_rate": 2.2343659579256107e-06,
      "loss": 0.0,
      "step": 809
    },
    {
      "epoch": 3.39020979020979,
      "grad_norm": 0.0076058306731283665,
      "learning_rate": 2.228486043501565e-06,
      "loss": 0.0,
      "step": 810
    },
    {
      "epoch": 3.3944055944055944,
      "grad_norm": 0.05593888461589813,
      "learning_rate": 2.222607648559488e-06,
      "loss": 0.0002,
      "step": 811
    },
    {
      "epoch": 3.3986013986013988,
      "grad_norm": 0.008083322085440159,
      "learning_rate": 2.2167308059968258e-06,
      "loss": 0.0,
      "step": 812
    },
    {
      "epoch": 3.4027972027972027,
      "grad_norm": 0.0025677771773189306,
      "learning_rate": 2.210855548702333e-06,
      "loss": 0.0,
      "step": 813
    },
    {
      "epoch": 3.406993006993007,
      "grad_norm": 0.0031788419000804424,
      "learning_rate": 2.2049819095558946e-06,
      "loss": 0.0,
      "step": 814
    },
    {
      "epoch": 3.4111888111888113,
      "grad_norm": 0.0016516821924597025,
      "learning_rate": 2.1991099214283388e-06,
      "loss": 0.0,
      "step": 815
    },
    {
      "epoch": 3.4153846153846152,
      "grad_norm": 0.001721789943985641,
      "learning_rate": 2.193239617181256e-06,
      "loss": 0.0,
      "step": 816
    },
    {
      "epoch": 3.4195804195804196,
      "grad_norm": 0.0013714516535401344,
      "learning_rate": 2.1873710296668104e-06,
      "loss": 0.0,
      "step": 817
    },
    {
      "epoch": 3.423776223776224,
      "grad_norm": 0.002976557705551386,
      "learning_rate": 2.181504191727562e-06,
      "loss": 0.0,
      "step": 818
    },
    {
      "epoch": 3.427972027972028,
      "grad_norm": 0.002027187030762434,
      "learning_rate": 2.175639136196276e-06,
      "loss": 0.0,
      "step": 819
    },
    {
      "epoch": 3.432167832167832,
      "grad_norm": 0.0014425985282287002,
      "learning_rate": 2.169775895895745e-06,
      "loss": 0.0,
      "step": 820
    },
    {
      "epoch": 3.4363636363636365,
      "grad_norm": 0.0011664831545203924,
      "learning_rate": 2.1639145036386018e-06,
      "loss": 0.0,
      "step": 821
    },
    {
      "epoch": 3.4405594405594404,
      "grad_norm": 0.0009184313821606338,
      "learning_rate": 2.158054992227139e-06,
      "loss": 0.0,
      "step": 822
    },
    {
      "epoch": 3.4447552447552447,
      "grad_norm": 0.0018825981533154845,
      "learning_rate": 2.152197394453121e-06,
      "loss": 0.0,
      "step": 823
    },
    {
      "epoch": 3.448951048951049,
      "grad_norm": 0.008687013760209084,
      "learning_rate": 2.146341743097604e-06,
      "loss": 0.0,
      "step": 824
    },
    {
      "epoch": 3.453146853146853,
      "grad_norm": 0.0038241585716605186,
      "learning_rate": 2.1404880709307505e-06,
      "loss": 0.0,
      "step": 825
    },
    {
      "epoch": 3.4573426573426573,
      "grad_norm": 0.003121076850220561,
      "learning_rate": 2.134636410711649e-06,
      "loss": 0.0,
      "step": 826
    },
    {
      "epoch": 3.4615384615384617,
      "grad_norm": 0.0028751338832080364,
      "learning_rate": 2.128786795188126e-06,
      "loss": 0.0,
      "step": 827
    },
    {
      "epoch": 3.4657342657342656,
      "grad_norm": 0.0017557827522978187,
      "learning_rate": 2.1229392570965656e-06,
      "loss": 0.0,
      "step": 828
    },
    {
      "epoch": 3.46993006993007,
      "grad_norm": 0.0017682121833786368,
      "learning_rate": 2.117093829161727e-06,
      "loss": 0.0,
      "step": 829
    },
    {
      "epoch": 3.4741258741258743,
      "grad_norm": 0.0014107830356806517,
      "learning_rate": 2.1112505440965607e-06,
      "loss": 0.0,
      "step": 830
    },
    {
      "epoch": 3.478321678321678,
      "grad_norm": 0.0039533344097435474,
      "learning_rate": 2.1054094346020222e-06,
      "loss": 0.0,
      "step": 831
    },
    {
      "epoch": 3.4825174825174825,
      "grad_norm": 0.001848749234341085,
      "learning_rate": 2.0995705333668948e-06,
      "loss": 0.0,
      "step": 832
    },
    {
      "epoch": 3.486713286713287,
      "grad_norm": 0.0019147819839417934,
      "learning_rate": 2.093733873067602e-06,
      "loss": 0.0,
      "step": 833
    },
    {
      "epoch": 3.4909090909090907,
      "grad_norm": 0.0016443360364064574,
      "learning_rate": 2.0878994863680265e-06,
      "loss": 0.0,
      "step": 834
    },
    {
      "epoch": 3.495104895104895,
      "grad_norm": 0.0023553299251943827,
      "learning_rate": 2.082067405919328e-06,
      "loss": 0.0,
      "step": 835
    },
    {
      "epoch": 3.4993006993006994,
      "grad_norm": 0.0018669285345822573,
      "learning_rate": 2.0762376643597586e-06,
      "loss": 0.0,
      "step": 836
    },
    {
      "epoch": 3.5034965034965033,
      "grad_norm": 0.0031040417961776257,
      "learning_rate": 2.070410294314482e-06,
      "loss": 0.0,
      "step": 837
    },
    {
      "epoch": 3.5076923076923077,
      "grad_norm": 0.0009603937505744398,
      "learning_rate": 2.0645853283953897e-06,
      "loss": 0.0,
      "step": 838
    },
    {
      "epoch": 3.511888111888112,
      "grad_norm": 0.0028102989308536053,
      "learning_rate": 2.058762799200919e-06,
      "loss": 0.0,
      "step": 839
    },
    {
      "epoch": 3.5160839160839163,
      "grad_norm": 0.0012625858653336763,
      "learning_rate": 2.0529427393158704e-06,
      "loss": 0.0,
      "step": 840
    },
    {
      "epoch": 3.5202797202797202,
      "grad_norm": 0.0019511563004925847,
      "learning_rate": 2.0471251813112258e-06,
      "loss": 0.0,
      "step": 841
    },
    {
      "epoch": 3.5244755244755246,
      "grad_norm": 0.002669193549081683,
      "learning_rate": 2.0413101577439657e-06,
      "loss": 0.0,
      "step": 842
    },
    {
      "epoch": 3.5286713286713285,
      "grad_norm": 0.002437007147818804,
      "learning_rate": 2.035497701156885e-06,
      "loss": 0.0,
      "step": 843
    },
    {
      "epoch": 3.532867132867133,
      "grad_norm": 0.0011534279910847545,
      "learning_rate": 2.0296878440784164e-06,
      "loss": 0.0,
      "step": 844
    },
    {
      "epoch": 3.537062937062937,
      "grad_norm": 0.001191779039800167,
      "learning_rate": 2.023880619022442e-06,
      "loss": 0.0,
      "step": 845
    },
    {
      "epoch": 3.5412587412587415,
      "grad_norm": 0.003263662103563547,
      "learning_rate": 2.0180760584881154e-06,
      "loss": 0.0,
      "step": 846
    },
    {
      "epoch": 3.5454545454545454,
      "grad_norm": 0.001550830784253776,
      "learning_rate": 2.01227419495968e-06,
      "loss": 0.0,
      "step": 847
    },
    {
      "epoch": 3.5496503496503498,
      "grad_norm": 0.03212891146540642,
      "learning_rate": 2.006475060906283e-06,
      "loss": 0.0001,
      "step": 848
    },
    {
      "epoch": 3.5538461538461537,
      "grad_norm": 0.0014326617820188403,
      "learning_rate": 2.0006786887817982e-06,
      "loss": 0.0,
      "step": 849
    },
    {
      "epoch": 3.558041958041958,
      "grad_norm": 0.001659504254348576,
      "learning_rate": 1.994885111024643e-06,
      "loss": 0.0,
      "step": 850
    },
    {
      "epoch": 3.5622377622377623,
      "grad_norm": 0.0011762354988604784,
      "learning_rate": 1.989094360057597e-06,
      "loss": 0.0,
      "step": 851
    },
    {
      "epoch": 3.5664335664335667,
      "grad_norm": 0.001663403119891882,
      "learning_rate": 1.9833064682876175e-06,
      "loss": 0.0,
      "step": 852
    },
    {
      "epoch": 3.5706293706293706,
      "grad_norm": 0.00426329905167222,
      "learning_rate": 1.9775214681056645e-06,
      "loss": 0.0,
      "step": 853
    },
    {
      "epoch": 3.574825174825175,
      "grad_norm": 0.001822963822633028,
      "learning_rate": 1.9717393918865123e-06,
      "loss": 0.0,
      "step": 854
    },
    {
      "epoch": 3.579020979020979,
      "grad_norm": 0.0025887812953442335,
      "learning_rate": 1.9659602719885737e-06,
      "loss": 0.0,
      "step": 855
    },
    {
      "epoch": 3.583216783216783,
      "grad_norm": 0.000913949275854975,
      "learning_rate": 1.9601841407537157e-06,
      "loss": 0.0,
      "step": 856
    },
    {
      "epoch": 3.5874125874125875,
      "grad_norm": 0.019461164250969887,
      "learning_rate": 1.9544110305070825e-06,
      "loss": 0.0001,
      "step": 857
    },
    {
      "epoch": 3.591608391608392,
      "grad_norm": 0.0011764619266614318,
      "learning_rate": 1.9486409735569086e-06,
      "loss": 0.0,
      "step": 858
    },
    {
      "epoch": 3.5958041958041957,
      "grad_norm": 0.0018363435519859195,
      "learning_rate": 1.942874002194343e-06,
      "loss": 0.0,
      "step": 859
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.003464712994173169,
      "learning_rate": 1.937110148693265e-06,
      "loss": 0.0,
      "step": 860
    },
    {
      "epoch": 3.604195804195804,
      "grad_norm": 0.07246309518814087,
      "learning_rate": 1.9313494453101082e-06,
      "loss": 0.0001,
      "step": 861
    },
    {
      "epoch": 3.6083916083916083,
      "grad_norm": 0.004755206871777773,
      "learning_rate": 1.9255919242836753e-06,
      "loss": 0.0,
      "step": 862
    },
    {
      "epoch": 3.6125874125874127,
      "grad_norm": 0.001071130158379674,
      "learning_rate": 1.919837617834959e-06,
      "loss": 0.0,
      "step": 863
    },
    {
      "epoch": 3.616783216783217,
      "grad_norm": 0.0016157825011759996,
      "learning_rate": 1.9140865581669627e-06,
      "loss": 0.0,
      "step": 864
    },
    {
      "epoch": 3.620979020979021,
      "grad_norm": 0.001602882519364357,
      "learning_rate": 1.9083387774645217e-06,
      "loss": 0.0,
      "step": 865
    },
    {
      "epoch": 3.6251748251748253,
      "grad_norm": 0.001598101924173534,
      "learning_rate": 1.9025943078941168e-06,
      "loss": 0.0,
      "step": 866
    },
    {
      "epoch": 3.629370629370629,
      "grad_norm": 0.0013639685930684209,
      "learning_rate": 1.8968531816037019e-06,
      "loss": 0.0,
      "step": 867
    },
    {
      "epoch": 3.6335664335664335,
      "grad_norm": 0.003929173108190298,
      "learning_rate": 1.8911154307225204e-06,
      "loss": 0.0,
      "step": 868
    },
    {
      "epoch": 3.637762237762238,
      "grad_norm": 0.001529913628473878,
      "learning_rate": 1.8853810873609255e-06,
      "loss": 0.0,
      "step": 869
    },
    {
      "epoch": 3.641958041958042,
      "grad_norm": 0.001963794231414795,
      "learning_rate": 1.8796501836102e-06,
      "loss": 0.0,
      "step": 870
    },
    {
      "epoch": 3.646153846153846,
      "grad_norm": 0.0018717608181759715,
      "learning_rate": 1.8739227515423785e-06,
      "loss": 0.0,
      "step": 871
    },
    {
      "epoch": 3.6503496503496504,
      "grad_norm": 0.021375982090830803,
      "learning_rate": 1.8681988232100674e-06,
      "loss": 0.0001,
      "step": 872
    },
    {
      "epoch": 3.6545454545454543,
      "grad_norm": 0.0023632431402802467,
      "learning_rate": 1.862478430646263e-06,
      "loss": 0.0,
      "step": 873
    },
    {
      "epoch": 3.6587412587412587,
      "grad_norm": 0.001687430078163743,
      "learning_rate": 1.8567616058641755e-06,
      "loss": 0.0,
      "step": 874
    },
    {
      "epoch": 3.662937062937063,
      "grad_norm": 0.0182957760989666,
      "learning_rate": 1.8510483808570494e-06,
      "loss": 0.0,
      "step": 875
    },
    {
      "epoch": 3.6671328671328673,
      "grad_norm": 0.028998883441090584,
      "learning_rate": 1.8453387875979834e-06,
      "loss": 0.0001,
      "step": 876
    },
    {
      "epoch": 3.6713286713286712,
      "grad_norm": 0.0009737509535625577,
      "learning_rate": 1.8396328580397505e-06,
      "loss": 0.0,
      "step": 877
    },
    {
      "epoch": 3.6755244755244756,
      "grad_norm": 0.001062242779880762,
      "learning_rate": 1.8339306241146201e-06,
      "loss": 0.0,
      "step": 878
    },
    {
      "epoch": 3.6797202797202795,
      "grad_norm": 0.0013671009801328182,
      "learning_rate": 1.8282321177341822e-06,
      "loss": 0.0,
      "step": 879
    },
    {
      "epoch": 3.683916083916084,
      "grad_norm": 0.00197093328461051,
      "learning_rate": 1.822537370789163e-06,
      "loss": 0.0,
      "step": 880
    },
    {
      "epoch": 3.688111888111888,
      "grad_norm": 0.0012111988617107272,
      "learning_rate": 1.8168464151492532e-06,
      "loss": 0.0,
      "step": 881
    },
    {
      "epoch": 3.6923076923076925,
      "grad_norm": 0.0013270566705614328,
      "learning_rate": 1.8111592826629234e-06,
      "loss": 0.0,
      "step": 882
    },
    {
      "epoch": 3.6965034965034964,
      "grad_norm": 0.001210037269629538,
      "learning_rate": 1.8054760051572496e-06,
      "loss": 0.0,
      "step": 883
    },
    {
      "epoch": 3.7006993006993008,
      "grad_norm": 0.0012341660913079977,
      "learning_rate": 1.7997966144377328e-06,
      "loss": 0.0,
      "step": 884
    },
    {
      "epoch": 3.7048951048951047,
      "grad_norm": 0.0013252273201942444,
      "learning_rate": 1.7941211422881248e-06,
      "loss": 0.0,
      "step": 885
    },
    {
      "epoch": 3.709090909090909,
      "grad_norm": 0.0009930897504091263,
      "learning_rate": 1.7884496204702457e-06,
      "loss": 0.0,
      "step": 886
    },
    {
      "epoch": 3.7132867132867133,
      "grad_norm": 0.0031181543599814177,
      "learning_rate": 1.7827820807238086e-06,
      "loss": 0.0,
      "step": 887
    },
    {
      "epoch": 3.7174825174825177,
      "grad_norm": 0.004060108680278063,
      "learning_rate": 1.7771185547662417e-06,
      "loss": 0.0,
      "step": 888
    },
    {
      "epoch": 3.7216783216783216,
      "grad_norm": 0.001300285803154111,
      "learning_rate": 1.7714590742925125e-06,
      "loss": 0.0,
      "step": 889
    },
    {
      "epoch": 3.725874125874126,
      "grad_norm": 0.001757982186973095,
      "learning_rate": 1.7658036709749443e-06,
      "loss": 0.0,
      "step": 890
    },
    {
      "epoch": 3.73006993006993,
      "grad_norm": 0.00146966811735183,
      "learning_rate": 1.7601523764630476e-06,
      "loss": 0.0,
      "step": 891
    },
    {
      "epoch": 3.734265734265734,
      "grad_norm": 0.002285903552547097,
      "learning_rate": 1.754505222383337e-06,
      "loss": 0.0,
      "step": 892
    },
    {
      "epoch": 3.7384615384615385,
      "grad_norm": 0.0011265241773799062,
      "learning_rate": 1.7488622403391553e-06,
      "loss": 0.0,
      "step": 893
    },
    {
      "epoch": 3.742657342657343,
      "grad_norm": 0.001024533761665225,
      "learning_rate": 1.743223461910498e-06,
      "loss": 0.0,
      "step": 894
    },
    {
      "epoch": 3.7468531468531467,
      "grad_norm": 0.0020864696707576513,
      "learning_rate": 1.7375889186538359e-06,
      "loss": 0.0,
      "step": 895
    },
    {
      "epoch": 3.751048951048951,
      "grad_norm": 0.0016020084731280804,
      "learning_rate": 1.7319586421019383e-06,
      "loss": 0.0,
      "step": 896
    },
    {
      "epoch": 3.755244755244755,
      "grad_norm": 0.00147394766099751,
      "learning_rate": 1.7263326637636956e-06,
      "loss": 0.0,
      "step": 897
    },
    {
      "epoch": 3.7594405594405593,
      "grad_norm": 0.002340656006708741,
      "learning_rate": 1.7207110151239448e-06,
      "loss": 0.0,
      "step": 898
    },
    {
      "epoch": 3.7636363636363637,
      "grad_norm": 0.0011478533269837499,
      "learning_rate": 1.715093727643293e-06,
      "loss": 0.0,
      "step": 899
    },
    {
      "epoch": 3.767832167832168,
      "grad_norm": 0.001313065062277019,
      "learning_rate": 1.7094808327579401e-06,
      "loss": 0.0,
      "step": 900
    },
    {
      "epoch": 3.772027972027972,
      "grad_norm": 0.0018346544820815325,
      "learning_rate": 1.7038723618795039e-06,
      "loss": 0.0,
      "step": 901
    },
    {
      "epoch": 3.7762237762237763,
      "grad_norm": 0.0009255964541807771,
      "learning_rate": 1.6982683463948432e-06,
      "loss": 0.0,
      "step": 902
    },
    {
      "epoch": 3.78041958041958,
      "grad_norm": 0.0014225435443222523,
      "learning_rate": 1.6926688176658837e-06,
      "loss": 0.0,
      "step": 903
    },
    {
      "epoch": 3.7846153846153845,
      "grad_norm": 0.003851379966363311,
      "learning_rate": 1.6870738070294412e-06,
      "loss": 0.0,
      "step": 904
    },
    {
      "epoch": 3.788811188811189,
      "grad_norm": 0.0012000198476016521,
      "learning_rate": 1.681483345797048e-06,
      "loss": 0.0,
      "step": 905
    },
    {
      "epoch": 3.793006993006993,
      "grad_norm": 0.001237357035279274,
      "learning_rate": 1.675897465254775e-06,
      "loss": 0.0,
      "step": 906
    },
    {
      "epoch": 3.797202797202797,
      "grad_norm": 0.002991842105984688,
      "learning_rate": 1.670316196663059e-06,
      "loss": 0.0,
      "step": 907
    },
    {
      "epoch": 3.8013986013986014,
      "grad_norm": 0.0042267004027962685,
      "learning_rate": 1.6647395712565256e-06,
      "loss": 0.0,
      "step": 908
    },
    {
      "epoch": 3.8055944055944058,
      "grad_norm": 0.003248338121920824,
      "learning_rate": 1.6591676202438174e-06,
      "loss": 0.0,
      "step": 909
    },
    {
      "epoch": 3.8097902097902097,
      "grad_norm": 0.0012438372941687703,
      "learning_rate": 1.6536003748074162e-06,
      "loss": 0.0,
      "step": 910
    },
    {
      "epoch": 3.813986013986014,
      "grad_norm": 0.0011140950955450535,
      "learning_rate": 1.6480378661034702e-06,
      "loss": 0.0,
      "step": 911
    },
    {
      "epoch": 3.8181818181818183,
      "grad_norm": 0.0018570018000900745,
      "learning_rate": 1.6424801252616186e-06,
      "loss": 0.0,
      "step": 912
    },
    {
      "epoch": 3.8223776223776222,
      "grad_norm": 0.00169821095187217,
      "learning_rate": 1.6369271833848204e-06,
      "loss": 0.0,
      "step": 913
    },
    {
      "epoch": 3.8265734265734266,
      "grad_norm": 0.0017073426861315966,
      "learning_rate": 1.631379071549174e-06,
      "loss": 0.0,
      "step": 914
    },
    {
      "epoch": 3.830769230769231,
      "grad_norm": 0.0022255678195506334,
      "learning_rate": 1.6258358208037511e-06,
      "loss": 0.0,
      "step": 915
    },
    {
      "epoch": 3.834965034965035,
      "grad_norm": 0.002181857591494918,
      "learning_rate": 1.6202974621704176e-06,
      "loss": 0.0,
      "step": 916
    },
    {
      "epoch": 3.839160839160839,
      "grad_norm": 0.015438455156981945,
      "learning_rate": 1.6147640266436615e-06,
      "loss": 0.0,
      "step": 917
    },
    {
      "epoch": 3.8433566433566435,
      "grad_norm": 0.0012607112294062972,
      "learning_rate": 1.6092355451904184e-06,
      "loss": 0.0,
      "step": 918
    },
    {
      "epoch": 3.8475524475524474,
      "grad_norm": 0.0028839819133281708,
      "learning_rate": 1.6037120487499025e-06,
      "loss": 0.0,
      "step": 919
    },
    {
      "epoch": 3.8517482517482518,
      "grad_norm": 0.0011634485563263297,
      "learning_rate": 1.5981935682334266e-06,
      "loss": 0.0,
      "step": 920
    },
    {
      "epoch": 3.855944055944056,
      "grad_norm": 0.0018964925548061728,
      "learning_rate": 1.5926801345242346e-06,
      "loss": 0.0,
      "step": 921
    },
    {
      "epoch": 3.86013986013986,
      "grad_norm": 0.0011763435322791338,
      "learning_rate": 1.587171778477326e-06,
      "loss": 0.0,
      "step": 922
    },
    {
      "epoch": 3.8643356643356643,
      "grad_norm": 0.0017115079099312425,
      "learning_rate": 1.581668530919286e-06,
      "loss": 0.0,
      "step": 923
    },
    {
      "epoch": 3.8685314685314687,
      "grad_norm": 0.0008064614958129823,
      "learning_rate": 1.5761704226481078e-06,
      "loss": 0.0,
      "step": 924
    },
    {
      "epoch": 3.8727272727272726,
      "grad_norm": 0.001312964828684926,
      "learning_rate": 1.5706774844330264e-06,
      "loss": 0.0,
      "step": 925
    },
    {
      "epoch": 3.876923076923077,
      "grad_norm": 0.00406665587797761,
      "learning_rate": 1.5651897470143412e-06,
      "loss": 0.0,
      "step": 926
    },
    {
      "epoch": 3.8811188811188813,
      "grad_norm": 0.001838710973970592,
      "learning_rate": 1.5597072411032472e-06,
      "loss": 0.0,
      "step": 927
    },
    {
      "epoch": 3.885314685314685,
      "grad_norm": 0.0010559471556916833,
      "learning_rate": 1.5542299973816626e-06,
      "loss": 0.0,
      "step": 928
    },
    {
      "epoch": 3.8895104895104895,
      "grad_norm": 0.000948031956795603,
      "learning_rate": 1.5487580465020557e-06,
      "loss": 0.0,
      "step": 929
    },
    {
      "epoch": 3.893706293706294,
      "grad_norm": 0.002021998865529895,
      "learning_rate": 1.5432914190872757e-06,
      "loss": 0.0,
      "step": 930
    },
    {
      "epoch": 3.8979020979020977,
      "grad_norm": 0.0011419267393648624,
      "learning_rate": 1.5378301457303784e-06,
      "loss": 0.0,
      "step": 931
    },
    {
      "epoch": 3.902097902097902,
      "grad_norm": 0.001520116231404245,
      "learning_rate": 1.5323742569944573e-06,
      "loss": 0.0,
      "step": 932
    },
    {
      "epoch": 3.9062937062937064,
      "grad_norm": 0.0015044461470097303,
      "learning_rate": 1.5269237834124715e-06,
      "loss": 0.0,
      "step": 933
    },
    {
      "epoch": 3.9104895104895103,
      "grad_norm": 0.019327526912093163,
      "learning_rate": 1.5214787554870752e-06,
      "loss": 0.0,
      "step": 934
    },
    {
      "epoch": 3.9146853146853147,
      "grad_norm": 0.0015376959927380085,
      "learning_rate": 1.5160392036904475e-06,
      "loss": 0.0,
      "step": 935
    },
    {
      "epoch": 3.918881118881119,
      "grad_norm": 0.002472449792549014,
      "learning_rate": 1.5106051584641208e-06,
      "loss": 0.0,
      "step": 936
    },
    {
      "epoch": 3.9230769230769234,
      "grad_norm": 0.007587876636534929,
      "learning_rate": 1.5051766502188092e-06,
      "loss": 0.0,
      "step": 937
    },
    {
      "epoch": 3.9272727272727272,
      "grad_norm": 0.0012112654512748122,
      "learning_rate": 1.4997537093342436e-06,
      "loss": 0.0,
      "step": 938
    },
    {
      "epoch": 3.9314685314685316,
      "grad_norm": 0.0017228553770110011,
      "learning_rate": 1.4943363661589938e-06,
      "loss": 0.0,
      "step": 939
    },
    {
      "epoch": 3.9356643356643355,
      "grad_norm": 0.013345488347113132,
      "learning_rate": 1.4889246510103078e-06,
      "loss": 0.0,
      "step": 940
    },
    {
      "epoch": 3.93986013986014,
      "grad_norm": 0.001575305126607418,
      "learning_rate": 1.483518594173934e-06,
      "loss": 0.0,
      "step": 941
    },
    {
      "epoch": 3.944055944055944,
      "grad_norm": 0.0012752519687637687,
      "learning_rate": 1.4781182259039553e-06,
      "loss": 0.0,
      "step": 942
    },
    {
      "epoch": 3.9482517482517485,
      "grad_norm": 0.0010486049577593803,
      "learning_rate": 1.47272357642262e-06,
      "loss": 0.0,
      "step": 943
    },
    {
      "epoch": 3.9524475524475524,
      "grad_norm": 0.00167929963208735,
      "learning_rate": 1.4673346759201728e-06,
      "loss": 0.0,
      "step": 944
    },
    {
      "epoch": 3.9566433566433568,
      "grad_norm": 0.015494981780648232,
      "learning_rate": 1.4619515545546848e-06,
      "loss": 0.0,
      "step": 945
    },
    {
      "epoch": 3.9608391608391607,
      "grad_norm": 0.0038782304618507624,
      "learning_rate": 1.4565742424518841e-06,
      "loss": 0.0,
      "step": 946
    },
    {
      "epoch": 3.965034965034965,
      "grad_norm": 0.001402477384544909,
      "learning_rate": 1.451202769704989e-06,
      "loss": 0.0,
      "step": 947
    },
    {
      "epoch": 3.9692307692307693,
      "grad_norm": 0.0014286726946011186,
      "learning_rate": 1.44583716637454e-06,
      "loss": 0.0,
      "step": 948
    },
    {
      "epoch": 3.9734265734265737,
      "grad_norm": 0.0015969790983945131,
      "learning_rate": 1.4404774624882296e-06,
      "loss": 0.0,
      "step": 949
    },
    {
      "epoch": 3.9776223776223776,
      "grad_norm": 0.0028078528121113777,
      "learning_rate": 1.4351236880407321e-06,
      "loss": 0.0,
      "step": 950
    },
    {
      "epoch": 3.981818181818182,
      "grad_norm": 0.0021153613924980164,
      "learning_rate": 1.4297758729935434e-06,
      "loss": 0.0,
      "step": 951
    },
    {
      "epoch": 3.986013986013986,
      "grad_norm": 0.0014967190800234675,
      "learning_rate": 1.4244340472748076e-06,
      "loss": 0.0,
      "step": 952
    },
    {
      "epoch": 3.99020979020979,
      "grad_norm": 0.0012663602828979492,
      "learning_rate": 1.4190982407791467e-06,
      "loss": 0.0,
      "step": 953
    },
    {
      "epoch": 3.9944055944055945,
      "grad_norm": 0.0009616782190278172,
      "learning_rate": 1.413768483367503e-06,
      "loss": 0.0,
      "step": 954
    },
    {
      "epoch": 3.998601398601399,
      "grad_norm": 0.0010298120323568583,
      "learning_rate": 1.4084448048669603e-06,
      "loss": 0.0,
      "step": 955
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.0010298120323568583,
      "learning_rate": 1.403127235070587e-06,
      "loss": 0.0,
      "step": 956
    },
    {
      "epoch": 4.004195804195804,
      "grad_norm": 0.003489186055958271,
      "learning_rate": 1.3978158037372647e-06,
      "loss": 0.0,
      "step": 957
    },
    {
      "epoch": 4.008391608391609,
      "grad_norm": 0.0008752505527809262,
      "learning_rate": 1.392510540591519e-06,
      "loss": 0.0,
      "step": 958
    },
    {
      "epoch": 4.012587412587413,
      "grad_norm": 0.0013871409464627504,
      "learning_rate": 1.3872114753233597e-06,
      "loss": 0.0,
      "step": 959
    },
    {
      "epoch": 4.0167832167832165,
      "grad_norm": 0.0018275025067850947,
      "learning_rate": 1.381918637588112e-06,
      "loss": 0.0,
      "step": 960
    },
    {
      "epoch": 4.020979020979021,
      "grad_norm": 0.002547784009948373,
      "learning_rate": 1.3766320570062435e-06,
      "loss": 0.0,
      "step": 961
    },
    {
      "epoch": 4.025174825174825,
      "grad_norm": 0.0011057290248572826,
      "learning_rate": 1.3713517631632111e-06,
      "loss": 0.0,
      "step": 962
    },
    {
      "epoch": 4.0293706293706295,
      "grad_norm": 0.0012969310628250241,
      "learning_rate": 1.3660777856092883e-06,
      "loss": 0.0,
      "step": 963
    },
    {
      "epoch": 4.033566433566434,
      "grad_norm": 0.0012018791167065501,
      "learning_rate": 1.3608101538593965e-06,
      "loss": 0.0,
      "step": 964
    },
    {
      "epoch": 4.037762237762238,
      "grad_norm": 0.0015174384461715817,
      "learning_rate": 1.3555488973929477e-06,
      "loss": 0.0,
      "step": 965
    },
    {
      "epoch": 4.041958041958042,
      "grad_norm": 0.0011106351157650352,
      "learning_rate": 1.3502940456536756e-06,
      "loss": 0.0,
      "step": 966
    },
    {
      "epoch": 4.046153846153846,
      "grad_norm": 0.0013968718703836203,
      "learning_rate": 1.3450456280494682e-06,
      "loss": 0.0,
      "step": 967
    },
    {
      "epoch": 4.05034965034965,
      "grad_norm": 0.001489020069129765,
      "learning_rate": 1.3398036739522088e-06,
      "loss": 0.0,
      "step": 968
    },
    {
      "epoch": 4.054545454545455,
      "grad_norm": 0.03593435510993004,
      "learning_rate": 1.3345682126976062e-06,
      "loss": 0.0,
      "step": 969
    },
    {
      "epoch": 4.058741258741259,
      "grad_norm": 0.001545711187645793,
      "learning_rate": 1.3293392735850355e-06,
      "loss": 0.0,
      "step": 970
    },
    {
      "epoch": 4.062937062937063,
      "grad_norm": 0.0011249794624745846,
      "learning_rate": 1.3241168858773714e-06,
      "loss": 0.0,
      "step": 971
    },
    {
      "epoch": 4.067132867132867,
      "grad_norm": 0.0008237747824750841,
      "learning_rate": 1.3189010788008234e-06,
      "loss": 0.0,
      "step": 972
    },
    {
      "epoch": 4.071328671328671,
      "grad_norm": 0.0009013775270432234,
      "learning_rate": 1.313691881544773e-06,
      "loss": 0.0,
      "step": 973
    },
    {
      "epoch": 4.0755244755244755,
      "grad_norm": 0.0011539814295247197,
      "learning_rate": 1.3084893232616141e-06,
      "loss": 0.0,
      "step": 974
    },
    {
      "epoch": 4.07972027972028,
      "grad_norm": 0.001654982566833496,
      "learning_rate": 1.3032934330665837e-06,
      "loss": 0.0,
      "step": 975
    },
    {
      "epoch": 4.083916083916084,
      "grad_norm": 0.004433086141943932,
      "learning_rate": 1.2981042400376032e-06,
      "loss": 0.0,
      "step": 976
    },
    {
      "epoch": 4.0881118881118885,
      "grad_norm": 0.0009525130735710263,
      "learning_rate": 1.292921773215116e-06,
      "loss": 0.0,
      "step": 977
    },
    {
      "epoch": 4.092307692307692,
      "grad_norm": 0.001205323962494731,
      "learning_rate": 1.2877460616019195e-06,
      "loss": 0.0,
      "step": 978
    },
    {
      "epoch": 4.096503496503496,
      "grad_norm": 0.0012073866091668606,
      "learning_rate": 1.2825771341630097e-06,
      "loss": 0.0,
      "step": 979
    },
    {
      "epoch": 4.100699300699301,
      "grad_norm": 0.0025041645858436823,
      "learning_rate": 1.277415019825417e-06,
      "loss": 0.0,
      "step": 980
    },
    {
      "epoch": 4.104895104895105,
      "grad_norm": 0.0014390015276148915,
      "learning_rate": 1.2722597474780398e-06,
      "loss": 0.0,
      "step": 981
    },
    {
      "epoch": 4.109090909090909,
      "grad_norm": 0.0010887256357818842,
      "learning_rate": 1.2671113459714903e-06,
      "loss": 0.0,
      "step": 982
    },
    {
      "epoch": 4.113286713286714,
      "grad_norm": 0.014471828937530518,
      "learning_rate": 1.2619698441179257e-06,
      "loss": 0.0,
      "step": 983
    },
    {
      "epoch": 4.117482517482517,
      "grad_norm": 0.008815598674118519,
      "learning_rate": 1.2568352706908937e-06,
      "loss": 0.0,
      "step": 984
    },
    {
      "epoch": 4.1216783216783215,
      "grad_norm": 0.0013799811713397503,
      "learning_rate": 1.251707654425165e-06,
      "loss": 0.0,
      "step": 985
    },
    {
      "epoch": 4.125874125874126,
      "grad_norm": 0.0017953617498278618,
      "learning_rate": 1.2465870240165794e-06,
      "loss": 0.0,
      "step": 986
    },
    {
      "epoch": 4.13006993006993,
      "grad_norm": 0.0014298532623797655,
      "learning_rate": 1.2414734081218777e-06,
      "loss": 0.0,
      "step": 987
    },
    {
      "epoch": 4.1342657342657345,
      "grad_norm": 0.0007059157942421734,
      "learning_rate": 1.2363668353585486e-06,
      "loss": 0.0,
      "step": 988
    },
    {
      "epoch": 4.138461538461539,
      "grad_norm": 0.0006820279522798955,
      "learning_rate": 1.2312673343046621e-06,
      "loss": 0.0,
      "step": 989
    },
    {
      "epoch": 4.142657342657342,
      "grad_norm": 0.0010997310746461153,
      "learning_rate": 1.2261749334987146e-06,
      "loss": 0.0,
      "step": 990
    },
    {
      "epoch": 4.146853146853147,
      "grad_norm": 0.0015070467488840222,
      "learning_rate": 1.2210896614394677e-06,
      "loss": 0.0,
      "step": 991
    },
    {
      "epoch": 4.151048951048951,
      "grad_norm": 0.0010309036588296294,
      "learning_rate": 1.216011546585785e-06,
      "loss": 0.0,
      "step": 992
    },
    {
      "epoch": 4.155244755244755,
      "grad_norm": 0.0009461237932555377,
      "learning_rate": 1.2109406173564789e-06,
      "loss": 0.0,
      "step": 993
    },
    {
      "epoch": 4.15944055944056,
      "grad_norm": 0.0016318035777658224,
      "learning_rate": 1.205876902130149e-06,
      "loss": 0.0,
      "step": 994
    },
    {
      "epoch": 4.163636363636364,
      "grad_norm": 0.0008777921902947128,
      "learning_rate": 1.2008204292450198e-06,
      "loss": 0.0,
      "step": 995
    },
    {
      "epoch": 4.1678321678321675,
      "grad_norm": 0.0019420183962211013,
      "learning_rate": 1.195771226998789e-06,
      "loss": 0.0,
      "step": 996
    },
    {
      "epoch": 4.172027972027972,
      "grad_norm": 0.005027747247368097,
      "learning_rate": 1.1907293236484626e-06,
      "loss": 0.0,
      "step": 997
    },
    {
      "epoch": 4.176223776223776,
      "grad_norm": 0.0006222941447049379,
      "learning_rate": 1.1856947474102e-06,
      "loss": 0.0,
      "step": 998
    },
    {
      "epoch": 4.1804195804195805,
      "grad_norm": 0.0012766474392265081,
      "learning_rate": 1.1806675264591572e-06,
      "loss": 0.0,
      "step": 999
    },
    {
      "epoch": 4.184615384615385,
      "grad_norm": 0.0016877582529559731,
      "learning_rate": 1.1756476889293269e-06,
      "loss": 0.0,
      "step": 1000
    },
    {
      "epoch": 4.188811188811189,
      "grad_norm": 0.00160312675870955,
      "learning_rate": 1.1706352629133803e-06,
      "loss": 0.0,
      "step": 1001
    },
    {
      "epoch": 4.193006993006993,
      "grad_norm": 0.0018009484047070146,
      "learning_rate": 1.165630276462514e-06,
      "loss": 0.0,
      "step": 1002
    },
    {
      "epoch": 4.197202797202797,
      "grad_norm": 0.002118004485964775,
      "learning_rate": 1.1606327575862869e-06,
      "loss": 0.0,
      "step": 1003
    },
    {
      "epoch": 4.201398601398601,
      "grad_norm": 0.0013233154313638806,
      "learning_rate": 1.1556427342524698e-06,
      "loss": 0.0,
      "step": 1004
    },
    {
      "epoch": 4.205594405594406,
      "grad_norm": 0.001025038305670023,
      "learning_rate": 1.150660234386886e-06,
      "loss": 0.0,
      "step": 1005
    },
    {
      "epoch": 4.20979020979021,
      "grad_norm": 0.001148061826825142,
      "learning_rate": 1.1456852858732513e-06,
      "loss": 0.0,
      "step": 1006
    },
    {
      "epoch": 4.213986013986014,
      "grad_norm": 0.0018501273589208722,
      "learning_rate": 1.1407179165530267e-06,
      "loss": 0.0,
      "step": 1007
    },
    {
      "epoch": 4.218181818181818,
      "grad_norm": 0.00721105607226491,
      "learning_rate": 1.1357581542252555e-06,
      "loss": 0.0,
      "step": 1008
    },
    {
      "epoch": 4.222377622377622,
      "grad_norm": 0.0010389365488663316,
      "learning_rate": 1.1308060266464063e-06,
      "loss": 0.0,
      "step": 1009
    },
    {
      "epoch": 4.2265734265734265,
      "grad_norm": 0.001098422915674746,
      "learning_rate": 1.1258615615302258e-06,
      "loss": 0.0,
      "step": 1010
    },
    {
      "epoch": 4.230769230769231,
      "grad_norm": 0.0013185304123908281,
      "learning_rate": 1.1209247865475783e-06,
      "loss": 0.0,
      "step": 1011
    },
    {
      "epoch": 4.234965034965035,
      "grad_norm": 0.0021497008856385946,
      "learning_rate": 1.1159957293262888e-06,
      "loss": 0.0,
      "step": 1012
    },
    {
      "epoch": 4.2391608391608395,
      "grad_norm": 0.006119229830801487,
      "learning_rate": 1.1110744174509952e-06,
      "loss": 0.0,
      "step": 1013
    },
    {
      "epoch": 4.243356643356643,
      "grad_norm": 0.0013496100436896086,
      "learning_rate": 1.1061608784629855e-06,
      "loss": 0.0,
      "step": 1014
    },
    {
      "epoch": 4.247552447552447,
      "grad_norm": 0.0011937256203964353,
      "learning_rate": 1.101255139860051e-06,
      "loss": 0.0,
      "step": 1015
    },
    {
      "epoch": 4.251748251748252,
      "grad_norm": 0.0011465779971331358,
      "learning_rate": 1.0963572290963298e-06,
      "loss": 0.0,
      "step": 1016
    },
    {
      "epoch": 4.255944055944056,
      "grad_norm": 0.001504244515672326,
      "learning_rate": 1.0914671735821497e-06,
      "loss": 0.0,
      "step": 1017
    },
    {
      "epoch": 4.26013986013986,
      "grad_norm": 0.003262688172981143,
      "learning_rate": 1.0865850006838814e-06,
      "loss": 0.0,
      "step": 1018
    },
    {
      "epoch": 4.264335664335665,
      "grad_norm": 0.004067919682711363,
      "learning_rate": 1.0817107377237814e-06,
      "loss": 0.0,
      "step": 1019
    },
    {
      "epoch": 4.268531468531468,
      "grad_norm": 0.000898704573046416,
      "learning_rate": 1.0768444119798357e-06,
      "loss": 0.0,
      "step": 1020
    },
    {
      "epoch": 4.2727272727272725,
      "grad_norm": 0.001670478261075914,
      "learning_rate": 1.0719860506856152e-06,
      "loss": 0.0,
      "step": 1021
    },
    {
      "epoch": 4.276923076923077,
      "grad_norm": 0.0009978491580113769,
      "learning_rate": 1.0671356810301189e-06,
      "loss": 0.0,
      "step": 1022
    },
    {
      "epoch": 4.281118881118881,
      "grad_norm": 0.0020964746363461018,
      "learning_rate": 1.0622933301576192e-06,
      "loss": 0.0,
      "step": 1023
    },
    {
      "epoch": 4.2853146853146855,
      "grad_norm": 0.0010049011325463653,
      "learning_rate": 1.0574590251675145e-06,
      "loss": 0.0,
      "step": 1024
    },
    {
      "epoch": 4.28951048951049,
      "grad_norm": 0.0009352843626402318,
      "learning_rate": 1.0526327931141774e-06,
      "loss": 0.0,
      "step": 1025
    },
    {
      "epoch": 4.293706293706293,
      "grad_norm": 0.001520221820101142,
      "learning_rate": 1.047814661006798e-06,
      "loss": 0.0,
      "step": 1026
    },
    {
      "epoch": 4.297902097902098,
      "grad_norm": 0.022821085527539253,
      "learning_rate": 1.0430046558092399e-06,
      "loss": 0.0,
      "step": 1027
    },
    {
      "epoch": 4.302097902097902,
      "grad_norm": 0.0010403190972283483,
      "learning_rate": 1.0382028044398823e-06,
      "loss": 0.0,
      "step": 1028
    },
    {
      "epoch": 4.306293706293706,
      "grad_norm": 0.001574207446537912,
      "learning_rate": 1.0334091337714763e-06,
      "loss": 0.0,
      "step": 1029
    },
    {
      "epoch": 4.310489510489511,
      "grad_norm": 0.0016617587534710765,
      "learning_rate": 1.0286236706309899e-06,
      "loss": 0.0,
      "step": 1030
    },
    {
      "epoch": 4.314685314685315,
      "grad_norm": 0.0016201343387365341,
      "learning_rate": 1.0238464417994583e-06,
      "loss": 0.0,
      "step": 1031
    },
    {
      "epoch": 4.3188811188811185,
      "grad_norm": 0.0011412663152441382,
      "learning_rate": 1.0190774740118343e-06,
      "loss": 0.0,
      "step": 1032
    },
    {
      "epoch": 4.323076923076923,
      "grad_norm": 0.010682007297873497,
      "learning_rate": 1.0143167939568419e-06,
      "loss": 0.0,
      "step": 1033
    },
    {
      "epoch": 4.327272727272727,
      "grad_norm": 0.0012214913731440902,
      "learning_rate": 1.0095644282768205e-06,
      "loss": 0.0,
      "step": 1034
    },
    {
      "epoch": 4.3314685314685315,
      "grad_norm": 0.0019182339310646057,
      "learning_rate": 1.0048204035675827e-06,
      "loss": 0.0,
      "step": 1035
    },
    {
      "epoch": 4.335664335664336,
      "grad_norm": 0.0014221406308934093,
      "learning_rate": 1.0000847463782615e-06,
      "loss": 0.0,
      "step": 1036
    },
    {
      "epoch": 4.33986013986014,
      "grad_norm": 0.0013618165394291282,
      "learning_rate": 9.953574832111603e-07,
      "loss": 0.0,
      "step": 1037
    },
    {
      "epoch": 4.344055944055944,
      "grad_norm": 0.0011814313475042582,
      "learning_rate": 9.906386405216097e-07,
      "loss": 0.0,
      "step": 1038
    },
    {
      "epoch": 4.348251748251748,
      "grad_norm": 0.006232825573533773,
      "learning_rate": 9.85928244717816e-07,
      "loss": 0.0,
      "step": 1039
    },
    {
      "epoch": 4.352447552447552,
      "grad_norm": 0.0011092665372416377,
      "learning_rate": 9.812263221607114e-07,
      "loss": 0.0,
      "step": 1040
    },
    {
      "epoch": 4.356643356643357,
      "grad_norm": 0.0014064816059544683,
      "learning_rate": 9.765328991638127e-07,
      "loss": 0.0,
      "step": 1041
    },
    {
      "epoch": 4.360839160839161,
      "grad_norm": 0.001561799319460988,
      "learning_rate": 9.718480019930662e-07,
      "loss": 0.0,
      "step": 1042
    },
    {
      "epoch": 4.365034965034965,
      "grad_norm": 0.0013349915388971567,
      "learning_rate": 9.671716568667097e-07,
      "loss": 0.0,
      "step": 1043
    },
    {
      "epoch": 4.36923076923077,
      "grad_norm": 0.001316425041295588,
      "learning_rate": 9.625038899551162e-07,
      "loss": 0.0,
      "step": 1044
    },
    {
      "epoch": 4.373426573426573,
      "grad_norm": 0.0009382697171531618,
      "learning_rate": 9.57844727380656e-07,
      "loss": 0.0,
      "step": 1045
    },
    {
      "epoch": 4.3776223776223775,
      "grad_norm": 0.001024338649585843,
      "learning_rate": 9.531941952175436e-07,
      "loss": 0.0,
      "step": 1046
    },
    {
      "epoch": 4.381818181818182,
      "grad_norm": 0.000992637942545116,
      "learning_rate": 9.485523194916976e-07,
      "loss": 0.0,
      "step": 1047
    },
    {
      "epoch": 4.386013986013986,
      "grad_norm": 0.0009175704326480627,
      "learning_rate": 9.439191261805894e-07,
      "loss": 0.0,
      "step": 1048
    },
    {
      "epoch": 4.3902097902097905,
      "grad_norm": 0.0016189605230465531,
      "learning_rate": 9.392946412131035e-07,
      "loss": 0.0,
      "step": 1049
    },
    {
      "epoch": 4.394405594405594,
      "grad_norm": 0.001867191749624908,
      "learning_rate": 9.34678890469389e-07,
      "loss": 0.0,
      "step": 1050
    },
    {
      "epoch": 4.398601398601398,
      "grad_norm": 0.002096191281452775,
      "learning_rate": 9.300718997807131e-07,
      "loss": 0.0,
      "step": 1051
    },
    {
      "epoch": 4.402797202797203,
      "grad_norm": 0.0013373218243941665,
      "learning_rate": 9.254736949293216e-07,
      "loss": 0.0,
      "step": 1052
    },
    {
      "epoch": 4.406993006993007,
      "grad_norm": 0.001026992336846888,
      "learning_rate": 9.208843016482916e-07,
      "loss": 0.0,
      "step": 1053
    },
    {
      "epoch": 4.411188811188811,
      "grad_norm": 0.0026033087633550167,
      "learning_rate": 9.163037456213852e-07,
      "loss": 0.0,
      "step": 1054
    },
    {
      "epoch": 4.415384615384616,
      "grad_norm": 0.0006945374771021307,
      "learning_rate": 9.117320524829121e-07,
      "loss": 0.0,
      "step": 1055
    },
    {
      "epoch": 4.41958041958042,
      "grad_norm": 0.010211163200438023,
      "learning_rate": 9.07169247817579e-07,
      "loss": 0.0,
      "step": 1056
    },
    {
      "epoch": 4.4237762237762235,
      "grad_norm": 0.0011074518552049994,
      "learning_rate": 9.026153571603505e-07,
      "loss": 0.0,
      "step": 1057
    },
    {
      "epoch": 4.427972027972028,
      "grad_norm": 0.0015874669188633561,
      "learning_rate": 8.980704059963067e-07,
      "loss": 0.0,
      "step": 1058
    },
    {
      "epoch": 4.432167832167832,
      "grad_norm": 0.0010024852817878127,
      "learning_rate": 8.935344197604992e-07,
      "loss": 0.0,
      "step": 1059
    },
    {
      "epoch": 4.4363636363636365,
      "grad_norm": 0.005343414377421141,
      "learning_rate": 8.890074238378074e-07,
      "loss": 0.0,
      "step": 1060
    },
    {
      "epoch": 4.440559440559441,
      "grad_norm": 0.0009140478214249015,
      "learning_rate": 8.844894435628001e-07,
      "loss": 0.0,
      "step": 1061
    },
    {
      "epoch": 4.444755244755244,
      "grad_norm": 0.0012009950587525964,
      "learning_rate": 8.799805042195891e-07,
      "loss": 0.0,
      "step": 1062
    },
    {
      "epoch": 4.448951048951049,
      "grad_norm": 0.0012144141364842653,
      "learning_rate": 8.754806310416917e-07,
      "loss": 0.0,
      "step": 1063
    },
    {
      "epoch": 4.453146853146853,
      "grad_norm": 0.0014837405178695917,
      "learning_rate": 8.709898492118885e-07,
      "loss": 0.0,
      "step": 1064
    },
    {
      "epoch": 4.457342657342657,
      "grad_norm": 0.0014103974681347609,
      "learning_rate": 8.665081838620796e-07,
      "loss": 0.0,
      "step": 1065
    },
    {
      "epoch": 4.461538461538462,
      "grad_norm": 0.0033163889311254025,
      "learning_rate": 8.620356600731486e-07,
      "loss": 0.0,
      "step": 1066
    },
    {
      "epoch": 4.465734265734266,
      "grad_norm": 0.001850529690273106,
      "learning_rate": 8.575723028748203e-07,
      "loss": 0.0,
      "step": 1067
    },
    {
      "epoch": 4.46993006993007,
      "grad_norm": 0.0007387638324871659,
      "learning_rate": 8.531181372455161e-07,
      "loss": 0.0,
      "step": 1068
    },
    {
      "epoch": 4.474125874125874,
      "grad_norm": 0.0007748849457129836,
      "learning_rate": 8.486731881122226e-07,
      "loss": 0.0,
      "step": 1069
    },
    {
      "epoch": 4.478321678321678,
      "grad_norm": 0.0011792995501309633,
      "learning_rate": 8.442374803503469e-07,
      "loss": 0.0,
      "step": 1070
    },
    {
      "epoch": 4.4825174825174825,
      "grad_norm": 0.0012669803109019995,
      "learning_rate": 8.398110387835767e-07,
      "loss": 0.0,
      "step": 1071
    },
    {
      "epoch": 4.486713286713287,
      "grad_norm": 0.0021643047221004963,
      "learning_rate": 8.353938881837445e-07,
      "loss": 0.0,
      "step": 1072
    },
    {
      "epoch": 4.490909090909091,
      "grad_norm": 0.0021073156967759132,
      "learning_rate": 8.309860532706882e-07,
      "loss": 0.0,
      "step": 1073
    },
    {
      "epoch": 4.495104895104895,
      "grad_norm": 0.0010203431593254209,
      "learning_rate": 8.265875587121092e-07,
      "loss": 0.0,
      "step": 1074
    },
    {
      "epoch": 4.499300699300699,
      "grad_norm": 0.0025379681028425694,
      "learning_rate": 8.221984291234405e-07,
      "loss": 0.0,
      "step": 1075
    },
    {
      "epoch": 4.503496503496503,
      "grad_norm": 0.0014055216452106833,
      "learning_rate": 8.178186890677029e-07,
      "loss": 0.0,
      "step": 1076
    },
    {
      "epoch": 4.507692307692308,
      "grad_norm": 0.0011124424636363983,
      "learning_rate": 8.134483630553713e-07,
      "loss": 0.0,
      "step": 1077
    },
    {
      "epoch": 4.511888111888112,
      "grad_norm": 0.002580048283562064,
      "learning_rate": 8.090874755442383e-07,
      "loss": 0.0,
      "step": 1078
    },
    {
      "epoch": 4.516083916083916,
      "grad_norm": 0.0009783831192180514,
      "learning_rate": 8.047360509392726e-07,
      "loss": 0.0,
      "step": 1079
    },
    {
      "epoch": 4.520279720279721,
      "grad_norm": 0.011658071540296078,
      "learning_rate": 8.003941135924859e-07,
      "loss": 0.0,
      "step": 1080
    },
    {
      "epoch": 4.524475524475524,
      "grad_norm": 0.0013083721278235316,
      "learning_rate": 7.960616878027983e-07,
      "loss": 0.0,
      "step": 1081
    },
    {
      "epoch": 4.5286713286713285,
      "grad_norm": 0.0011040378594771028,
      "learning_rate": 7.917387978158969e-07,
      "loss": 0.0,
      "step": 1082
    },
    {
      "epoch": 4.532867132867133,
      "grad_norm": 0.0010877397144213319,
      "learning_rate": 7.874254678241056e-07,
      "loss": 0.0,
      "step": 1083
    },
    {
      "epoch": 4.537062937062937,
      "grad_norm": 0.0010763180907815695,
      "learning_rate": 7.83121721966248e-07,
      "loss": 0.0,
      "step": 1084
    },
    {
      "epoch": 4.5412587412587415,
      "grad_norm": 0.011900283396244049,
      "learning_rate": 7.788275843275081e-07,
      "loss": 0.0,
      "step": 1085
    },
    {
      "epoch": 4.545454545454545,
      "grad_norm": 0.0011615053517743945,
      "learning_rate": 7.745430789393027e-07,
      "loss": 0.0,
      "step": 1086
    },
    {
      "epoch": 4.549650349650349,
      "grad_norm": 0.001175162149593234,
      "learning_rate": 7.702682297791431e-07,
      "loss": 0.0,
      "step": 1087
    },
    {
      "epoch": 4.553846153846154,
      "grad_norm": 0.001497325487434864,
      "learning_rate": 7.66003060770498e-07,
      "loss": 0.0,
      "step": 1088
    },
    {
      "epoch": 4.558041958041958,
      "grad_norm": 0.0015857350081205368,
      "learning_rate": 7.61747595782667e-07,
      "loss": 0.0,
      "step": 1089
    },
    {
      "epoch": 4.562237762237762,
      "grad_norm": 0.005172734148800373,
      "learning_rate": 7.57501858630639e-07,
      "loss": 0.0,
      "step": 1090
    },
    {
      "epoch": 4.566433566433567,
      "grad_norm": 0.002228137105703354,
      "learning_rate": 7.532658730749656e-07,
      "loss": 0.0,
      "step": 1091
    },
    {
      "epoch": 4.570629370629371,
      "grad_norm": 0.0007612527697347105,
      "learning_rate": 7.490396628216237e-07,
      "loss": 0.0,
      "step": 1092
    },
    {
      "epoch": 4.5748251748251745,
      "grad_norm": 0.0014022511895745993,
      "learning_rate": 7.448232515218839e-07,
      "loss": 0.0,
      "step": 1093
    },
    {
      "epoch": 4.579020979020979,
      "grad_norm": 0.0009858199628069997,
      "learning_rate": 7.406166627721803e-07,
      "loss": 0.0,
      "step": 1094
    },
    {
      "epoch": 4.583216783216783,
      "grad_norm": 0.001173322438262403,
      "learning_rate": 7.364199201139763e-07,
      "loss": 0.0,
      "step": 1095
    },
    {
      "epoch": 4.5874125874125875,
      "grad_norm": 0.0010235118679702282,
      "learning_rate": 7.322330470336314e-07,
      "loss": 0.0,
      "step": 1096
    },
    {
      "epoch": 4.591608391608392,
      "grad_norm": 0.0013280995190143585,
      "learning_rate": 7.280560669622744e-07,
      "loss": 0.0,
      "step": 1097
    },
    {
      "epoch": 4.595804195804196,
      "grad_norm": 0.0009987136581912637,
      "learning_rate": 7.238890032756684e-07,
      "loss": 0.0,
      "step": 1098
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.0014701707987114787,
      "learning_rate": 7.197318792940805e-07,
      "loss": 0.0,
      "step": 1099
    },
    {
      "epoch": 4.604195804195804,
      "grad_norm": 0.001174332108348608,
      "learning_rate": 7.155847182821524e-07,
      "loss": 0.0,
      "step": 1100
    },
    {
      "epoch": 4.608391608391608,
      "grad_norm": 0.004871551413089037,
      "learning_rate": 7.114475434487708e-07,
      "loss": 0.0,
      "step": 1101
    },
    {
      "epoch": 4.612587412587413,
      "grad_norm": 0.0011388111161068082,
      "learning_rate": 7.073203779469348e-07,
      "loss": 0.0,
      "step": 1102
    },
    {
      "epoch": 4.616783216783217,
      "grad_norm": 0.002202955773100257,
      "learning_rate": 7.032032448736279e-07,
      "loss": 0.0,
      "step": 1103
    },
    {
      "epoch": 4.620979020979021,
      "grad_norm": 0.001086483127437532,
      "learning_rate": 6.990961672696908e-07,
      "loss": 0.0,
      "step": 1104
    },
    {
      "epoch": 4.625174825174825,
      "grad_norm": 0.0015999553725123405,
      "learning_rate": 6.949991681196877e-07,
      "loss": 0.0,
      "step": 1105
    },
    {
      "epoch": 4.629370629370629,
      "grad_norm": 0.002830302342772484,
      "learning_rate": 6.909122703517838e-07,
      "loss": 0.0,
      "step": 1106
    },
    {
      "epoch": 4.6335664335664335,
      "grad_norm": 0.0018011541105806828,
      "learning_rate": 6.868354968376098e-07,
      "loss": 0.0,
      "step": 1107
    },
    {
      "epoch": 4.637762237762238,
      "grad_norm": 0.001498807454481721,
      "learning_rate": 6.827688703921407e-07,
      "loss": 0.0,
      "step": 1108
    },
    {
      "epoch": 4.641958041958042,
      "grad_norm": 0.0012186354724690318,
      "learning_rate": 6.787124137735648e-07,
      "loss": 0.0,
      "step": 1109
    },
    {
      "epoch": 4.6461538461538465,
      "grad_norm": 0.002188766608014703,
      "learning_rate": 6.746661496831544e-07,
      "loss": 0.0,
      "step": 1110
    },
    {
      "epoch": 4.65034965034965,
      "grad_norm": 0.0014885258860886097,
      "learning_rate": 6.706301007651433e-07,
      "loss": 0.0,
      "step": 1111
    },
    {
      "epoch": 4.654545454545454,
      "grad_norm": 0.0007323445752263069,
      "learning_rate": 6.666042896065983e-07,
      "loss": 0.0,
      "step": 1112
    },
    {
      "epoch": 4.658741258741259,
      "grad_norm": 0.001076756278052926,
      "learning_rate": 6.625887387372893e-07,
      "loss": 0.0,
      "step": 1113
    },
    {
      "epoch": 4.662937062937063,
      "grad_norm": 0.0010681330459192395,
      "learning_rate": 6.585834706295696e-07,
      "loss": 0.0,
      "step": 1114
    },
    {
      "epoch": 4.667132867132867,
      "grad_norm": 0.0008571094367653131,
      "learning_rate": 6.545885076982445e-07,
      "loss": 0.0,
      "step": 1115
    },
    {
      "epoch": 4.671328671328672,
      "grad_norm": 0.0020040576346218586,
      "learning_rate": 6.506038723004484e-07,
      "loss": 0.0,
      "step": 1116
    },
    {
      "epoch": 4.675524475524475,
      "grad_norm": 0.0008787247934378684,
      "learning_rate": 6.466295867355205e-07,
      "loss": 0.0,
      "step": 1117
    },
    {
      "epoch": 4.6797202797202795,
      "grad_norm": 0.0008860119269229472,
      "learning_rate": 6.426656732448786e-07,
      "loss": 0.0,
      "step": 1118
    },
    {
      "epoch": 4.683916083916084,
      "grad_norm": 0.001538944197818637,
      "learning_rate": 6.387121540118938e-07,
      "loss": 0.0,
      "step": 1119
    },
    {
      "epoch": 4.688111888111888,
      "grad_norm": 0.0013904342195019126,
      "learning_rate": 6.347690511617693e-07,
      "loss": 0.0,
      "step": 1120
    },
    {
      "epoch": 4.6923076923076925,
      "grad_norm": 0.0016476798336952925,
      "learning_rate": 6.308363867614125e-07,
      "loss": 0.0,
      "step": 1121
    },
    {
      "epoch": 4.696503496503497,
      "grad_norm": 0.00075963931158185,
      "learning_rate": 6.269141828193159e-07,
      "loss": 0.0,
      "step": 1122
    },
    {
      "epoch": 4.7006993006993,
      "grad_norm": 0.0023602042347192764,
      "learning_rate": 6.230024612854318e-07,
      "loss": 0.0,
      "step": 1123
    },
    {
      "epoch": 4.704895104895105,
      "grad_norm": 0.0020317654125392437,
      "learning_rate": 6.191012440510469e-07,
      "loss": 0.0,
      "step": 1124
    },
    {
      "epoch": 4.709090909090909,
      "grad_norm": 0.0010020819026976824,
      "learning_rate": 6.152105529486648e-07,
      "loss": 0.0,
      "step": 1125
    },
    {
      "epoch": 4.713286713286713,
      "grad_norm": 0.0010115219047293067,
      "learning_rate": 6.113304097518824e-07,
      "loss": 0.0,
      "step": 1126
    },
    {
      "epoch": 4.717482517482518,
      "grad_norm": 0.0015433626249432564,
      "learning_rate": 6.074608361752624e-07,
      "loss": 0.0,
      "step": 1127
    },
    {
      "epoch": 4.721678321678322,
      "grad_norm": 0.001206689397804439,
      "learning_rate": 6.036018538742208e-07,
      "loss": 0.0,
      "step": 1128
    },
    {
      "epoch": 4.725874125874126,
      "grad_norm": 0.0007945495308376849,
      "learning_rate": 5.997534844449001e-07,
      "loss": 0.0,
      "step": 1129
    },
    {
      "epoch": 4.73006993006993,
      "grad_norm": 0.000998255331069231,
      "learning_rate": 5.959157494240489e-07,
      "loss": 0.0,
      "step": 1130
    },
    {
      "epoch": 4.734265734265734,
      "grad_norm": 0.00796382874250412,
      "learning_rate": 5.920886702889025e-07,
      "loss": 0.0,
      "step": 1131
    },
    {
      "epoch": 4.7384615384615385,
      "grad_norm": 0.0013102483935654163,
      "learning_rate": 5.882722684570638e-07,
      "loss": 0.0,
      "step": 1132
    },
    {
      "epoch": 4.742657342657343,
      "grad_norm": 0.0009221647051163018,
      "learning_rate": 5.844665652863787e-07,
      "loss": 0.0,
      "step": 1133
    },
    {
      "epoch": 4.746853146853147,
      "grad_norm": 0.0013080157805234194,
      "learning_rate": 5.806715820748238e-07,
      "loss": 0.0,
      "step": 1134
    },
    {
      "epoch": 4.751048951048951,
      "grad_norm": 0.0033610165119171143,
      "learning_rate": 5.768873400603791e-07,
      "loss": 0.0,
      "step": 1135
    },
    {
      "epoch": 4.755244755244755,
      "grad_norm": 0.0011778888292610645,
      "learning_rate": 5.731138604209169e-07,
      "loss": 0.0,
      "step": 1136
    },
    {
      "epoch": 4.759440559440559,
      "grad_norm": 0.001562577672302723,
      "learning_rate": 5.693511642740787e-07,
      "loss": 0.0,
      "step": 1137
    },
    {
      "epoch": 4.763636363636364,
      "grad_norm": 0.0014231757959350944,
      "learning_rate": 5.655992726771567e-07,
      "loss": 0.0,
      "step": 1138
    },
    {
      "epoch": 4.767832167832168,
      "grad_norm": 0.0014265241334214807,
      "learning_rate": 5.618582066269776e-07,
      "loss": 0.0,
      "step": 1139
    },
    {
      "epoch": 4.772027972027972,
      "grad_norm": 0.001964756054803729,
      "learning_rate": 5.581279870597866e-07,
      "loss": 0.0,
      "step": 1140
    },
    {
      "epoch": 4.776223776223777,
      "grad_norm": 0.002261451678350568,
      "learning_rate": 5.54408634851126e-07,
      "loss": 0.0,
      "step": 1141
    },
    {
      "epoch": 4.78041958041958,
      "grad_norm": 0.000675290182698518,
      "learning_rate": 5.50700170815722e-07,
      "loss": 0.0,
      "step": 1142
    },
    {
      "epoch": 4.7846153846153845,
      "grad_norm": 0.0014971964992582798,
      "learning_rate": 5.47002615707368e-07,
      "loss": 0.0,
      "step": 1143
    },
    {
      "epoch": 4.788811188811189,
      "grad_norm": 0.0009386117453686893,
      "learning_rate": 5.433159902188043e-07,
      "loss": 0.0,
      "step": 1144
    },
    {
      "epoch": 4.793006993006993,
      "grad_norm": 0.0013478758046403527,
      "learning_rate": 5.396403149816087e-07,
      "loss": 0.0,
      "step": 1145
    },
    {
      "epoch": 4.7972027972027975,
      "grad_norm": 0.00110398989636451,
      "learning_rate": 5.359756105660765e-07,
      "loss": 0.0,
      "step": 1146
    },
    {
      "epoch": 4.801398601398601,
      "grad_norm": 0.0027916990220546722,
      "learning_rate": 5.323218974811059e-07,
      "loss": 0.0,
      "step": 1147
    },
    {
      "epoch": 4.805594405594405,
      "grad_norm": 0.001024013152346015,
      "learning_rate": 5.286791961740855e-07,
      "loss": 0.0,
      "step": 1148
    },
    {
      "epoch": 4.80979020979021,
      "grad_norm": 0.0017076724907383323,
      "learning_rate": 5.250475270307767e-07,
      "loss": 0.0,
      "step": 1149
    },
    {
      "epoch": 4.813986013986014,
      "grad_norm": 0.0008102352148853242,
      "learning_rate": 5.214269103752029e-07,
      "loss": 0.0,
      "step": 1150
    },
    {
      "epoch": 4.818181818181818,
      "grad_norm": 0.011365428566932678,
      "learning_rate": 5.178173664695324e-07,
      "loss": 0.0,
      "step": 1151
    },
    {
      "epoch": 4.822377622377623,
      "grad_norm": 0.001112837577238679,
      "learning_rate": 5.142189155139685e-07,
      "loss": 0.0,
      "step": 1152
    },
    {
      "epoch": 4.826573426573427,
      "grad_norm": 0.010056550614535809,
      "learning_rate": 5.106315776466328e-07,
      "loss": 0.0,
      "step": 1153
    },
    {
      "epoch": 4.8307692307692305,
      "grad_norm": 0.00160472362767905,
      "learning_rate": 5.070553729434566e-07,
      "loss": 0.0,
      "step": 1154
    },
    {
      "epoch": 4.834965034965035,
      "grad_norm": 0.0013296622782945633,
      "learning_rate": 5.034903214180639e-07,
      "loss": 0.0,
      "step": 1155
    },
    {
      "epoch": 4.839160839160839,
      "grad_norm": 0.001238483702763915,
      "learning_rate": 4.999364430216639e-07,
      "loss": 0.0,
      "step": 1156
    },
    {
      "epoch": 4.8433566433566435,
      "grad_norm": 0.0007860551122575998,
      "learning_rate": 4.963937576429365e-07,
      "loss": 0.0,
      "step": 1157
    },
    {
      "epoch": 4.847552447552448,
      "grad_norm": 0.001160689047537744,
      "learning_rate": 4.928622851079206e-07,
      "loss": 0.0,
      "step": 1158
    },
    {
      "epoch": 4.851748251748251,
      "grad_norm": 0.008290434256196022,
      "learning_rate": 4.893420451799064e-07,
      "loss": 0.0,
      "step": 1159
    },
    {
      "epoch": 4.855944055944056,
      "grad_norm": 0.0008265050128102303,
      "learning_rate": 4.85833057559322e-07,
      "loss": 0.0,
      "step": 1160
    },
    {
      "epoch": 4.86013986013986,
      "grad_norm": 0.0015449656639248133,
      "learning_rate": 4.823353418836224e-07,
      "loss": 0.0,
      "step": 1161
    },
    {
      "epoch": 4.864335664335664,
      "grad_norm": 0.0012514643603935838,
      "learning_rate": 4.788489177271835e-07,
      "loss": 0.0,
      "step": 1162
    },
    {
      "epoch": 4.868531468531469,
      "grad_norm": 0.001488383044488728,
      "learning_rate": 4.7537380460118855e-07,
      "loss": 0.0,
      "step": 1163
    },
    {
      "epoch": 4.872727272727273,
      "grad_norm": 0.0014867460122331977,
      "learning_rate": 4.719100219535194e-07,
      "loss": 0.0,
      "step": 1164
    },
    {
      "epoch": 4.876923076923077,
      "grad_norm": 0.0016748507041484118,
      "learning_rate": 4.684575891686516e-07,
      "loss": 0.0,
      "step": 1165
    },
    {
      "epoch": 4.881118881118881,
      "grad_norm": 0.001467471127398312,
      "learning_rate": 4.6501652556754205e-07,
      "loss": 0.0,
      "step": 1166
    },
    {
      "epoch": 4.885314685314685,
      "grad_norm": 0.001478524529375136,
      "learning_rate": 4.615868504075208e-07,
      "loss": 0.0,
      "step": 1167
    },
    {
      "epoch": 4.8895104895104895,
      "grad_norm": 0.0011108319740742445,
      "learning_rate": 4.581685828821858e-07,
      "loss": 0.0,
      "step": 1168
    },
    {
      "epoch": 4.893706293706294,
      "grad_norm": 0.0022319292183965445,
      "learning_rate": 4.547617421212927e-07,
      "loss": 0.0,
      "step": 1169
    },
    {
      "epoch": 4.897902097902098,
      "grad_norm": 0.002263713860884309,
      "learning_rate": 4.513663471906507e-07,
      "loss": 0.0,
      "step": 1170
    },
    {
      "epoch": 4.902097902097902,
      "grad_norm": 0.0012681003427132964,
      "learning_rate": 4.4798241709201414e-07,
      "loss": 0.0,
      "step": 1171
    },
    {
      "epoch": 4.906293706293706,
      "grad_norm": 0.0012149532558396459,
      "learning_rate": 4.4460997076297504e-07,
      "loss": 0.0,
      "step": 1172
    },
    {
      "epoch": 4.91048951048951,
      "grad_norm": 0.005214029457420111,
      "learning_rate": 4.412490270768594e-07,
      "loss": 0.0,
      "step": 1173
    },
    {
      "epoch": 4.914685314685315,
      "grad_norm": 0.0010257532121613622,
      "learning_rate": 4.3789960484262204e-07,
      "loss": 0.0,
      "step": 1174
    },
    {
      "epoch": 4.918881118881119,
      "grad_norm": 0.001326566212810576,
      "learning_rate": 4.345617228047358e-07,
      "loss": 0.0,
      "step": 1175
    },
    {
      "epoch": 4.923076923076923,
      "grad_norm": 0.0014746327651664615,
      "learning_rate": 4.3123539964309486e-07,
      "loss": 0.0,
      "step": 1176
    },
    {
      "epoch": 4.927272727272728,
      "grad_norm": 0.0014187521301209927,
      "learning_rate": 4.279206539729053e-07,
      "loss": 0.0,
      "step": 1177
    },
    {
      "epoch": 4.931468531468531,
      "grad_norm": 0.0012298322981223464,
      "learning_rate": 4.2461750434457935e-07,
      "loss": 0.0,
      "step": 1178
    },
    {
      "epoch": 4.9356643356643355,
      "grad_norm": 0.0014907042495906353,
      "learning_rate": 4.2132596924363666e-07,
      "loss": 0.0,
      "step": 1179
    },
    {
      "epoch": 4.93986013986014,
      "grad_norm": 0.0011598754208534956,
      "learning_rate": 4.180460670905978e-07,
      "loss": 0.0,
      "step": 1180
    },
    {
      "epoch": 4.944055944055944,
      "grad_norm": 0.0008537492831237614,
      "learning_rate": 4.147778162408794e-07,
      "loss": 0.0,
      "step": 1181
    },
    {
      "epoch": 4.9482517482517485,
      "grad_norm": 0.0012970289681106806,
      "learning_rate": 4.1152123498469613e-07,
      "loss": 0.0,
      "step": 1182
    },
    {
      "epoch": 4.952447552447552,
      "grad_norm": 0.0010976853081956506,
      "learning_rate": 4.0827634154695344e-07,
      "loss": 0.0,
      "step": 1183
    },
    {
      "epoch": 4.956643356643356,
      "grad_norm": 0.0011723351199179888,
      "learning_rate": 4.0504315408714993e-07,
      "loss": 0.0,
      "step": 1184
    },
    {
      "epoch": 4.960839160839161,
      "grad_norm": 0.004449024796485901,
      "learning_rate": 4.018216906992739e-07,
      "loss": 0.0,
      "step": 1185
    },
    {
      "epoch": 4.965034965034965,
      "grad_norm": 0.003084977390244603,
      "learning_rate": 3.9861196941169797e-07,
      "loss": 0.0,
      "step": 1186
    },
    {
      "epoch": 4.969230769230769,
      "grad_norm": 0.002362647792324424,
      "learning_rate": 3.9541400818708665e-07,
      "loss": 0.0,
      "step": 1187
    },
    {
      "epoch": 4.973426573426574,
      "grad_norm": 0.001029789331369102,
      "learning_rate": 3.922278249222894e-07,
      "loss": 0.0,
      "step": 1188
    },
    {
      "epoch": 4.977622377622378,
      "grad_norm": 0.0013352007372304797,
      "learning_rate": 3.890534374482416e-07,
      "loss": 0.0,
      "step": 1189
    },
    {
      "epoch": 4.9818181818181815,
      "grad_norm": 0.004397594835609198,
      "learning_rate": 3.8589086352986694e-07,
      "loss": 0.0,
      "step": 1190
    },
    {
      "epoch": 4.986013986013986,
      "grad_norm": 0.0008433952461928129,
      "learning_rate": 3.8274012086597616e-07,
      "loss": 0.0,
      "step": 1191
    },
    {
      "epoch": 4.99020979020979,
      "grad_norm": 0.0018399981781840324,
      "learning_rate": 3.796012270891672e-07,
      "loss": 0.0,
      "step": 1192
    },
    {
      "epoch": 4.9944055944055945,
      "grad_norm": 0.0007988702272996306,
      "learning_rate": 3.7647419976572924e-07,
      "loss": 0.0,
      "step": 1193
    },
    {
      "epoch": 4.998601398601399,
      "grad_norm": 0.0023097116500139236,
      "learning_rate": 3.7335905639554164e-07,
      "loss": 0.0,
      "step": 1194
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.0023097116500139236,
      "learning_rate": 3.702558144119781e-07,
      "loss": 0.0,
      "step": 1195
    },
    {
      "epoch": 5.004195804195804,
      "grad_norm": 0.005584676284343004,
      "learning_rate": 3.671644911818084e-07,
      "loss": 0.0,
      "step": 1196
    },
    {
      "epoch": 5.008391608391609,
      "grad_norm": 0.0013620021054521203,
      "learning_rate": 3.640851040051002e-07,
      "loss": 0.0,
      "step": 1197
    },
    {
      "epoch": 5.012587412587413,
      "grad_norm": 0.0012699882499873638,
      "learning_rate": 3.610176701151225e-07,
      "loss": 0.0,
      "step": 1198
    },
    {
      "epoch": 5.0167832167832165,
      "grad_norm": 0.000982654164545238,
      "learning_rate": 3.5796220667825233e-07,
      "loss": 0.0,
      "step": 1199
    },
    {
      "epoch": 5.020979020979021,
      "grad_norm": 0.0024986863136291504,
      "learning_rate": 3.549187307938726e-07,
      "loss": 0.0,
      "step": 1200
    },
    {
      "epoch": 5.025174825174825,
      "grad_norm": 0.0011792376171797514,
      "learning_rate": 3.5188725949428265e-07,
      "loss": 0.0,
      "step": 1201
    },
    {
      "epoch": 5.0293706293706295,
      "grad_norm": 0.0009689652360975742,
      "learning_rate": 3.4886780974459995e-07,
      "loss": 0.0,
      "step": 1202
    },
    {
      "epoch": 5.033566433566434,
      "grad_norm": 0.0011087720049545169,
      "learning_rate": 3.458603984426634e-07,
      "loss": 0.0,
      "step": 1203
    },
    {
      "epoch": 5.037762237762238,
      "grad_norm": 0.001034510089084506,
      "learning_rate": 3.4286504241894283e-07,
      "loss": 0.0,
      "step": 1204
    },
    {
      "epoch": 5.041958041958042,
      "grad_norm": 0.0034064892679452896,
      "learning_rate": 3.398817584364422e-07,
      "loss": 0.0,
      "step": 1205
    },
    {
      "epoch": 5.046153846153846,
      "grad_norm": 0.0007158444495871663,
      "learning_rate": 3.3691056319060554e-07,
      "loss": 0.0,
      "step": 1206
    },
    {
      "epoch": 5.05034965034965,
      "grad_norm": 0.0010681343264877796,
      "learning_rate": 3.339514733092253e-07,
      "loss": 0.0,
      "step": 1207
    },
    {
      "epoch": 5.054545454545455,
      "grad_norm": 0.001441069645807147,
      "learning_rate": 3.310045053523475e-07,
      "loss": 0.0,
      "step": 1208
    },
    {
      "epoch": 5.058741258741259,
      "grad_norm": 0.0009510577656328678,
      "learning_rate": 3.2806967581218097e-07,
      "loss": 0.0,
      "step": 1209
    },
    {
      "epoch": 5.062937062937063,
      "grad_norm": 0.0013143954565748572,
      "learning_rate": 3.2514700111300203e-07,
      "loss": 0.0,
      "step": 1210
    },
    {
      "epoch": 5.067132867132867,
      "grad_norm": 0.0010633381316438317,
      "learning_rate": 3.22236497611067e-07,
      "loss": 0.0,
      "step": 1211
    },
    {
      "epoch": 5.071328671328671,
      "grad_norm": 0.0013499686028808355,
      "learning_rate": 3.1933818159451566e-07,
      "loss": 0.0,
      "step": 1212
    },
    {
      "epoch": 5.0755244755244755,
      "grad_norm": 0.0009298166842199862,
      "learning_rate": 3.164520692832848e-07,
      "loss": 0.0,
      "step": 1213
    },
    {
      "epoch": 5.07972027972028,
      "grad_norm": 0.0006035981350578368,
      "learning_rate": 3.135781768290133e-07,
      "loss": 0.0,
      "step": 1214
    },
    {
      "epoch": 5.083916083916084,
      "grad_norm": 0.0008695109281688929,
      "learning_rate": 3.1071652031495506e-07,
      "loss": 0.0,
      "step": 1215
    },
    {
      "epoch": 5.0881118881118885,
      "grad_norm": 0.0008600199944339693,
      "learning_rate": 3.078671157558877e-07,
      "loss": 0.0,
      "step": 1216
    },
    {
      "epoch": 5.092307692307692,
      "grad_norm": 0.005410914309322834,
      "learning_rate": 3.050299790980207e-07,
      "loss": 0.0,
      "step": 1217
    },
    {
      "epoch": 5.096503496503496,
      "grad_norm": 0.0013415181310847402,
      "learning_rate": 3.02205126218911e-07,
      "loss": 0.0,
      "step": 1218
    },
    {
      "epoch": 5.100699300699301,
      "grad_norm": 0.0018672473961487412,
      "learning_rate": 2.993925729273706e-07,
      "loss": 0.0,
      "step": 1219
    },
    {
      "epoch": 5.104895104895105,
      "grad_norm": 0.0026037986390292645,
      "learning_rate": 2.965923349633779e-07,
      "loss": 0.0,
      "step": 1220
    },
    {
      "epoch": 5.109090909090909,
      "grad_norm": 0.0013551546726375818,
      "learning_rate": 2.938044279979918e-07,
      "loss": 0.0,
      "step": 1221
    },
    {
      "epoch": 5.113286713286714,
      "grad_norm": 0.0010574861662462354,
      "learning_rate": 2.910288676332626e-07,
      "loss": 0.0,
      "step": 1222
    },
    {
      "epoch": 5.117482517482517,
      "grad_norm": 0.001598590170033276,
      "learning_rate": 2.882656694021438e-07,
      "loss": 0.0,
      "step": 1223
    },
    {
      "epoch": 5.1216783216783215,
      "grad_norm": 0.0009886042680591345,
      "learning_rate": 2.8551484876840815e-07,
      "loss": 0.0,
      "step": 1224
    },
    {
      "epoch": 5.125874125874126,
      "grad_norm": 0.0018374572973698378,
      "learning_rate": 2.827764211265588e-07,
      "loss": 0.0,
      "step": 1225
    },
    {
      "epoch": 5.13006993006993,
      "grad_norm": 0.0009750261669978499,
      "learning_rate": 2.800504018017422e-07,
      "loss": 0.0,
      "step": 1226
    },
    {
      "epoch": 5.1342657342657345,
      "grad_norm": 0.0011703410418704152,
      "learning_rate": 2.773368060496659e-07,
      "loss": 0.0,
      "step": 1227
    },
    {
      "epoch": 5.138461538461539,
      "grad_norm": 0.001050062826834619,
      "learning_rate": 2.7463564905650855e-07,
      "loss": 0.0,
      "step": 1228
    },
    {
      "epoch": 5.142657342657342,
      "grad_norm": 0.0010128063149750233,
      "learning_rate": 2.7194694593883965e-07,
      "loss": 0.0,
      "step": 1229
    },
    {
      "epoch": 5.146853146853147,
      "grad_norm": 0.000786403426900506,
      "learning_rate": 2.6927071174353226e-07,
      "loss": 0.0,
      "step": 1230
    },
    {
      "epoch": 5.151048951048951,
      "grad_norm": 0.008308904245495796,
      "learning_rate": 2.666069614476777e-07,
      "loss": 0.0,
      "step": 1231
    },
    {
      "epoch": 5.155244755244755,
      "grad_norm": 0.0020032585598528385,
      "learning_rate": 2.639557099585047e-07,
      "loss": 0.0,
      "step": 1232
    },
    {
      "epoch": 5.15944055944056,
      "grad_norm": 0.0011452429462224245,
      "learning_rate": 2.613169721132952e-07,
      "loss": 0.0,
      "step": 1233
    },
    {
      "epoch": 5.163636363636364,
      "grad_norm": 0.0012882297160103917,
      "learning_rate": 2.5869076267929756e-07,
      "loss": 0.0,
      "step": 1234
    },
    {
      "epoch": 5.1678321678321675,
      "grad_norm": 0.0014330644626170397,
      "learning_rate": 2.560770963536499e-07,
      "loss": 0.0,
      "step": 1235
    },
    {
      "epoch": 5.172027972027972,
      "grad_norm": 0.0009334956994280219,
      "learning_rate": 2.53475987763295e-07,
      "loss": 0.0,
      "step": 1236
    },
    {
      "epoch": 5.176223776223776,
      "grad_norm": 0.001475191325880587,
      "learning_rate": 2.5088745146489595e-07,
      "loss": 0.0,
      "step": 1237
    },
    {
      "epoch": 5.1804195804195805,
      "grad_norm": 0.0014340244233608246,
      "learning_rate": 2.483115019447596e-07,
      "loss": 0.0,
      "step": 1238
    },
    {
      "epoch": 5.184615384615385,
      "grad_norm": 0.0022010444663465023,
      "learning_rate": 2.4574815361875255e-07,
      "loss": 0.0,
      "step": 1239
    },
    {
      "epoch": 5.188811188811189,
      "grad_norm": 0.0010593435727059841,
      "learning_rate": 2.431974208322191e-07,
      "loss": 0.0,
      "step": 1240
    },
    {
      "epoch": 5.193006993006993,
      "grad_norm": 0.0009170479024760425,
      "learning_rate": 2.406593178599057e-07,
      "loss": 0.0,
      "step": 1241
    },
    {
      "epoch": 5.197202797202797,
      "grad_norm": 0.0016952977748587728,
      "learning_rate": 2.3813385890587516e-07,
      "loss": 0.0,
      "step": 1242
    },
    {
      "epoch": 5.201398601398601,
      "grad_norm": 0.000966139486990869,
      "learning_rate": 2.3562105810343266e-07,
      "loss": 0.0,
      "step": 1243
    },
    {
      "epoch": 5.205594405594406,
      "grad_norm": 0.005574371665716171,
      "learning_rate": 2.3312092951504357e-07,
      "loss": 0.0,
      "step": 1244
    },
    {
      "epoch": 5.20979020979021,
      "grad_norm": 0.0009455638355575502,
      "learning_rate": 2.3063348713225437e-07,
      "loss": 0.0,
      "step": 1245
    },
    {
      "epoch": 5.213986013986014,
      "grad_norm": 0.001704404130578041,
      "learning_rate": 2.2815874487561536e-07,
      "loss": 0.0,
      "step": 1246
    },
    {
      "epoch": 5.218181818181818,
      "grad_norm": 0.0019490857375785708,
      "learning_rate": 2.2569671659460473e-07,
      "loss": 0.0,
      "step": 1247
    },
    {
      "epoch": 5.222377622377622,
      "grad_norm": 0.0025993373710662127,
      "learning_rate": 2.2324741606754629e-07,
      "loss": 0.0,
      "step": 1248
    },
    {
      "epoch": 5.2265734265734265,
      "grad_norm": 0.0008124838350340724,
      "learning_rate": 2.2081085700153714e-07,
      "loss": 0.0,
      "step": 1249
    },
    {
      "epoch": 5.230769230769231,
      "grad_norm": 0.0008490126929245889,
      "learning_rate": 2.1838705303236896e-07,
      "loss": 0.0,
      "step": 1250
    },
    {
      "epoch": 5.234965034965035,
      "grad_norm": 0.001284430967643857,
      "learning_rate": 2.1597601772444966e-07,
      "loss": 0.0,
      "step": 1251
    },
    {
      "epoch": 5.2391608391608395,
      "grad_norm": 0.0009174643200822175,
      "learning_rate": 2.135777645707318e-07,
      "loss": 0.0,
      "step": 1252
    },
    {
      "epoch": 5.243356643356643,
      "grad_norm": 0.001712885918095708,
      "learning_rate": 2.111923069926339e-07,
      "loss": 0.0,
      "step": 1253
    },
    {
      "epoch": 5.247552447552447,
      "grad_norm": 0.0006240654038265347,
      "learning_rate": 2.0881965833996516e-07,
      "loss": 0.0,
      "step": 1254
    },
    {
      "epoch": 5.251748251748252,
      "grad_norm": 0.0008200556039810181,
      "learning_rate": 2.0645983189085423e-07,
      "loss": 0.0,
      "step": 1255
    },
    {
      "epoch": 5.255944055944056,
      "grad_norm": 0.0013281999854370952,
      "learning_rate": 2.041128408516696e-07,
      "loss": 0.0,
      "step": 1256
    },
    {
      "epoch": 5.26013986013986,
      "grad_norm": 0.009047938510775566,
      "learning_rate": 2.0177869835695113e-07,
      "loss": 0.0,
      "step": 1257
    },
    {
      "epoch": 5.264335664335665,
      "grad_norm": 0.00105593993794173,
      "learning_rate": 1.99457417469332e-07,
      "loss": 0.0,
      "step": 1258
    },
    {
      "epoch": 5.268531468531468,
      "grad_norm": 0.0008916796650737524,
      "learning_rate": 1.9714901117946855e-07,
      "loss": 0.0,
      "step": 1259
    },
    {
      "epoch": 5.2727272727272725,
      "grad_norm": 0.0008890751632861793,
      "learning_rate": 1.9485349240596613e-07,
      "loss": 0.0,
      "step": 1260
    },
    {
      "epoch": 5.276923076923077,
      "grad_norm": 0.007282212842255831,
      "learning_rate": 1.925708739953086e-07,
      "loss": 0.0,
      "step": 1261
    },
    {
      "epoch": 5.281118881118881,
      "grad_norm": 0.007486447226256132,
      "learning_rate": 1.9030116872178317e-07,
      "loss": 0.0,
      "step": 1262
    },
    {
      "epoch": 5.2853146853146855,
      "grad_norm": 0.0010757662821561098,
      "learning_rate": 1.8804438928741204e-07,
      "loss": 0.0,
      "step": 1263
    },
    {
      "epoch": 5.28951048951049,
      "grad_norm": 0.0028176503255963326,
      "learning_rate": 1.8580054832188055e-07,
      "loss": 0.0,
      "step": 1264
    },
    {
      "epoch": 5.293706293706293,
      "grad_norm": 0.0008915740181691945,
      "learning_rate": 1.835696583824645e-07,
      "loss": 0.0,
      "step": 1265
    },
    {
      "epoch": 5.297902097902098,
      "grad_norm": 0.0011043963022530079,
      "learning_rate": 1.813517319539629e-07,
      "loss": 0.0,
      "step": 1266
    },
    {
      "epoch": 5.302097902097902,
      "grad_norm": 0.001800728146918118,
      "learning_rate": 1.7914678144862614e-07,
      "loss": 0.0,
      "step": 1267
    },
    {
      "epoch": 5.306293706293706,
      "grad_norm": 0.0013883200008422136,
      "learning_rate": 1.7695481920608716e-07,
      "loss": 0.0,
      "step": 1268
    },
    {
      "epoch": 5.310489510489511,
      "grad_norm": 0.0012809536419808865,
      "learning_rate": 1.7477585749329173e-07,
      "loss": 0.0,
      "step": 1269
    },
    {
      "epoch": 5.314685314685315,
      "grad_norm": 0.0021919989958405495,
      "learning_rate": 1.7260990850443138e-07,
      "loss": 0.0,
      "step": 1270
    },
    {
      "epoch": 5.3188811188811185,
      "grad_norm": 0.0026385996025055647,
      "learning_rate": 1.7045698436087304e-07,
      "loss": 0.0,
      "step": 1271
    },
    {
      "epoch": 5.323076923076923,
      "grad_norm": 0.0012904665200039744,
      "learning_rate": 1.683170971110934e-07,
      "loss": 0.0,
      "step": 1272
    },
    {
      "epoch": 5.327272727272727,
      "grad_norm": 0.000998994568362832,
      "learning_rate": 1.6619025873060913e-07,
      "loss": 0.0,
      "step": 1273
    },
    {
      "epoch": 5.3314685314685315,
      "grad_norm": 0.0008117191027849913,
      "learning_rate": 1.64076481121912e-07,
      "loss": 0.0,
      "step": 1274
    },
    {
      "epoch": 5.335664335664336,
      "grad_norm": 0.0020949828904122114,
      "learning_rate": 1.6197577611440167e-07,
      "loss": 0.0,
      "step": 1275
    },
    {
      "epoch": 5.33986013986014,
      "grad_norm": 0.0011048150481656194,
      "learning_rate": 1.5988815546431807e-07,
      "loss": 0.0,
      "step": 1276
    },
    {
      "epoch": 5.344055944055944,
      "grad_norm": 0.001492594718001783,
      "learning_rate": 1.5781363085467765e-07,
      "loss": 0.0,
      "step": 1277
    },
    {
      "epoch": 5.348251748251748,
      "grad_norm": 0.0009185225935652852,
      "learning_rate": 1.5575221389520695e-07,
      "loss": 0.0,
      "step": 1278
    },
    {
      "epoch": 5.352447552447552,
      "grad_norm": 0.001103386515751481,
      "learning_rate": 1.5370391612227699e-07,
      "loss": 0.0,
      "step": 1279
    },
    {
      "epoch": 5.356643356643357,
      "grad_norm": 0.0013922671787440777,
      "learning_rate": 1.5166874899884054e-07,
      "loss": 0.0,
      "step": 1280
    },
    {
      "epoch": 5.360839160839161,
      "grad_norm": 0.006303776986896992,
      "learning_rate": 1.4964672391436674e-07,
      "loss": 0.0,
      "step": 1281
    },
    {
      "epoch": 5.365034965034965,
      "grad_norm": 0.005603837314993143,
      "learning_rate": 1.4763785218477627e-07,
      "loss": 0.0,
      "step": 1282
    },
    {
      "epoch": 5.36923076923077,
      "grad_norm": 0.0010654004290699959,
      "learning_rate": 1.4564214505238094e-07,
      "loss": 0.0,
      "step": 1283
    },
    {
      "epoch": 5.373426573426573,
      "grad_norm": 0.0015981359174475074,
      "learning_rate": 1.4365961368581844e-07,
      "loss": 0.0,
      "step": 1284
    },
    {
      "epoch": 5.3776223776223775,
      "grad_norm": 0.0013174727791920304,
      "learning_rate": 1.4169026917999012e-07,
      "loss": 0.0,
      "step": 1285
    },
    {
      "epoch": 5.381818181818182,
      "grad_norm": 0.000877175887580961,
      "learning_rate": 1.3973412255600033e-07,
      "loss": 0.0,
      "step": 1286
    },
    {
      "epoch": 5.386013986013986,
      "grad_norm": 0.0012549633393064141,
      "learning_rate": 1.3779118476109243e-07,
      "loss": 0.0,
      "step": 1287
    },
    {
      "epoch": 5.3902097902097905,
      "grad_norm": 0.002990401117131114,
      "learning_rate": 1.3586146666858923e-07,
      "loss": 0.0,
      "step": 1288
    },
    {
      "epoch": 5.394405594405594,
      "grad_norm": 0.0010732028167694807,
      "learning_rate": 1.3394497907783299e-07,
      "loss": 0.0,
      "step": 1289
    },
    {
      "epoch": 5.398601398601398,
      "grad_norm": 0.0017518067033961415,
      "learning_rate": 1.3204173271412159e-07,
      "loss": 0.0,
      "step": 1290
    },
    {
      "epoch": 5.402797202797203,
      "grad_norm": 0.0016910296399146318,
      "learning_rate": 1.3015173822865185e-07,
      "loss": 0.0,
      "step": 1291
    },
    {
      "epoch": 5.406993006993007,
      "grad_norm": 0.0012405765010043979,
      "learning_rate": 1.2827500619845918e-07,
      "loss": 0.0,
      "step": 1292
    },
    {
      "epoch": 5.411188811188811,
      "grad_norm": 0.0008711732225492597,
      "learning_rate": 1.2641154712635607e-07,
      "loss": 0.0,
      "step": 1293
    },
    {
      "epoch": 5.415384615384616,
      "grad_norm": 0.0008044462883844972,
      "learning_rate": 1.2456137144087694e-07,
      "loss": 0.0,
      "step": 1294
    },
    {
      "epoch": 5.41958041958042,
      "grad_norm": 0.0011185897747054696,
      "learning_rate": 1.2272448949621764e-07,
      "loss": 0.0,
      "step": 1295
    },
    {
      "epoch": 5.4237762237762235,
      "grad_norm": 0.0013221855042502284,
      "learning_rate": 1.2090091157217653e-07,
      "loss": 0.0,
      "step": 1296
    },
    {
      "epoch": 5.427972027972028,
      "grad_norm": 0.001719593652524054,
      "learning_rate": 1.1909064787409996e-07,
      "loss": 0.0,
      "step": 1297
    },
    {
      "epoch": 5.432167832167832,
      "grad_norm": 0.0010209758765995502,
      "learning_rate": 1.1729370853282296e-07,
      "loss": 0.0,
      "step": 1298
    },
    {
      "epoch": 5.4363636363636365,
      "grad_norm": 0.006935483776032925,
      "learning_rate": 1.1551010360461246e-07,
      "loss": 0.0,
      "step": 1299
    },
    {
      "epoch": 5.440559440559441,
      "grad_norm": 0.001246898784302175,
      "learning_rate": 1.137398430711123e-07,
      "loss": 0.0,
      "step": 1300
    },
    {
      "epoch": 5.444755244755244,
      "grad_norm": 0.000980127602815628,
      "learning_rate": 1.119829368392858e-07,
      "loss": 0.0,
      "step": 1301
    },
    {
      "epoch": 5.448951048951049,
      "grad_norm": 0.0006885117618367076,
      "learning_rate": 1.1023939474136187e-07,
      "loss": 0.0,
      "step": 1302
    },
    {
      "epoch": 5.453146853146853,
      "grad_norm": 0.0010247151367366314,
      "learning_rate": 1.0850922653478008e-07,
      "loss": 0.0,
      "step": 1303
    },
    {
      "epoch": 5.457342657342657,
      "grad_norm": 0.0008010207675397396,
      "learning_rate": 1.0679244190213378e-07,
      "loss": 0.0,
      "step": 1304
    },
    {
      "epoch": 5.461538461538462,
      "grad_norm": 0.0008371576550416648,
      "learning_rate": 1.0508905045111762e-07,
      "loss": 0.0,
      "step": 1305
    },
    {
      "epoch": 5.465734265734266,
      "grad_norm": 0.0007785236812196672,
      "learning_rate": 1.033990617144745e-07,
      "loss": 0.0,
      "step": 1306
    },
    {
      "epoch": 5.46993006993007,
      "grad_norm": 0.002055832650512457,
      "learning_rate": 1.0172248514994044e-07,
      "loss": 0.0,
      "step": 1307
    },
    {
      "epoch": 5.474125874125874,
      "grad_norm": 0.0009061041055247188,
      "learning_rate": 1.0005933014019309e-07,
      "loss": 0.0,
      "step": 1308
    },
    {
      "epoch": 5.478321678321678,
      "grad_norm": 0.0016957665793597698,
      "learning_rate": 9.840960599279858e-08,
      "loss": 0.0,
      "step": 1309
    },
    {
      "epoch": 5.4825174825174825,
      "grad_norm": 0.00188540352974087,
      "learning_rate": 9.677332194015843e-08,
      "loss": 0.0,
      "step": 1310
    },
    {
      "epoch": 5.486713286713287,
      "grad_norm": 0.0013081313809379935,
      "learning_rate": 9.515048713946067e-08,
      "loss": 0.0,
      "step": 1311
    },
    {
      "epoch": 5.490909090909091,
      "grad_norm": 0.0010523265227675438,
      "learning_rate": 9.354111067262584e-08,
      "loss": 0.0,
      "step": 1312
    },
    {
      "epoch": 5.495104895104895,
      "grad_norm": 0.0013150300364941359,
      "learning_rate": 9.194520154625686e-08,
      "loss": 0.0,
      "step": 1313
    },
    {
      "epoch": 5.499300699300699,
      "grad_norm": 0.0023728753440082073,
      "learning_rate": 9.036276869159006e-08,
      "loss": 0.0,
      "step": 1314
    },
    {
      "epoch": 5.503496503496503,
      "grad_norm": 0.0010280892020091414,
      "learning_rate": 8.879382096444289e-08,
      "loss": 0.0,
      "step": 1315
    },
    {
      "epoch": 5.507692307692308,
      "grad_norm": 0.00130606594029814,
      "learning_rate": 8.723836714516681e-08,
      "loss": 0.0,
      "step": 1316
    },
    {
      "epoch": 5.511888111888112,
      "grad_norm": 0.007057744078338146,
      "learning_rate": 8.569641593859562e-08,
      "loss": 0.0,
      "step": 1317
    },
    {
      "epoch": 5.516083916083916,
      "grad_norm": 0.003100609639659524,
      "learning_rate": 8.416797597399884e-08,
      "loss": 0.0,
      "step": 1318
    },
    {
      "epoch": 5.520279720279721,
      "grad_norm": 0.0007680396665818989,
      "learning_rate": 8.265305580503257e-08,
      "loss": 0.0,
      "step": 1319
    },
    {
      "epoch": 5.524475524475524,
      "grad_norm": 0.0011286045191809535,
      "learning_rate": 8.115166390969126e-08,
      "loss": 0.0,
      "step": 1320
    },
    {
      "epoch": 5.5286713286713285,
      "grad_norm": 0.001259137992747128,
      "learning_rate": 7.966380869026096e-08,
      "loss": 0.0,
      "step": 1321
    },
    {
      "epoch": 5.532867132867133,
      "grad_norm": 0.001644861651584506,
      "learning_rate": 7.818949847327228e-08,
      "loss": 0.0,
      "step": 1322
    },
    {
      "epoch": 5.537062937062937,
      "grad_norm": 0.0015183102805167437,
      "learning_rate": 7.672874150945281e-08,
      "loss": 0.0,
      "step": 1323
    },
    {
      "epoch": 5.5412587412587415,
      "grad_norm": 0.001018814160488546,
      "learning_rate": 7.528154597368192e-08,
      "loss": 0.0,
      "step": 1324
    },
    {
      "epoch": 5.545454545454545,
      "grad_norm": 0.0013377416180446744,
      "learning_rate": 7.384791996494528e-08,
      "loss": 0.0,
      "step": 1325
    },
    {
      "epoch": 5.549650349650349,
      "grad_norm": 0.0012961754109710455,
      "learning_rate": 7.242787150628844e-08,
      "loss": 0.0,
      "step": 1326
    },
    {
      "epoch": 5.553846153846154,
      "grad_norm": 0.0014105831505730748,
      "learning_rate": 7.102140854477219e-08,
      "loss": 0.0,
      "step": 1327
    },
    {
      "epoch": 5.558041958041958,
      "grad_norm": 0.0014379629865288734,
      "learning_rate": 6.962853895142924e-08,
      "loss": 0.0,
      "step": 1328
    },
    {
      "epoch": 5.562237762237762,
      "grad_norm": 0.000870969204697758,
      "learning_rate": 6.824927052121899e-08,
      "loss": 0.0,
      "step": 1329
    },
    {
      "epoch": 5.566433566433567,
      "grad_norm": 0.0022240409161895514,
      "learning_rate": 6.68836109729834e-08,
      "loss": 0.0,
      "step": 1330
    },
    {
      "epoch": 5.570629370629371,
      "grad_norm": 0.0012337393127381802,
      "learning_rate": 6.55315679494059e-08,
      "loss": 0.0,
      "step": 1331
    },
    {
      "epoch": 5.5748251748251745,
      "grad_norm": 0.0007441772031597793,
      "learning_rate": 6.419314901696671e-08,
      "loss": 0.0,
      "step": 1332
    },
    {
      "epoch": 5.579020979020979,
      "grad_norm": 0.0015356371877714992,
      "learning_rate": 6.286836166590094e-08,
      "loss": 0.0,
      "step": 1333
    },
    {
      "epoch": 5.583216783216783,
      "grad_norm": 0.000988760031759739,
      "learning_rate": 6.155721331015746e-08,
      "loss": 0.0,
      "step": 1334
    },
    {
      "epoch": 5.5874125874125875,
      "grad_norm": 0.0012449473142623901,
      "learning_rate": 6.025971128735598e-08,
      "loss": 0.0,
      "step": 1335
    },
    {
      "epoch": 5.591608391608392,
      "grad_norm": 0.0008410963346250355,
      "learning_rate": 5.897586285874751e-08,
      "loss": 0.0,
      "step": 1336
    },
    {
      "epoch": 5.595804195804196,
      "grad_norm": 0.0009110424434766173,
      "learning_rate": 5.770567520917314e-08,
      "loss": 0.0,
      "step": 1337
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.0008415644406341016,
      "learning_rate": 5.644915544702284e-08,
      "loss": 0.0,
      "step": 1338
    },
    {
      "epoch": 5.604195804195804,
      "grad_norm": 0.0009318139054812491,
      "learning_rate": 5.520631060419751e-08,
      "loss": 0.0,
      "step": 1339
    },
    {
      "epoch": 5.608391608391608,
      "grad_norm": 0.0009331358014605939,
      "learning_rate": 5.3977147636068425e-08,
      "loss": 0.0,
      "step": 1340
    },
    {
      "epoch": 5.612587412587413,
      "grad_norm": 0.0013923910446465015,
      "learning_rate": 5.276167342143756e-08,
      "loss": 0.0,
      "step": 1341
    },
    {
      "epoch": 5.616783216783217,
      "grad_norm": 0.0010665234876796603,
      "learning_rate": 5.1559894762501507e-08,
      "loss": 0.0,
      "step": 1342
    },
    {
      "epoch": 5.620979020979021,
      "grad_norm": 0.0026989055331796408,
      "learning_rate": 5.037181838481148e-08,
      "loss": 0.0,
      "step": 1343
    },
    {
      "epoch": 5.625174825174825,
      "grad_norm": 0.0009420453570783138,
      "learning_rate": 4.919745093723588e-08,
      "loss": 0.0,
      "step": 1344
    },
    {
      "epoch": 5.629370629370629,
      "grad_norm": 0.0004940045764669776,
      "learning_rate": 4.8036798991923925e-08,
      "loss": 0.0,
      "step": 1345
    },
    {
      "epoch": 5.6335664335664335,
      "grad_norm": 0.0010841360781341791,
      "learning_rate": 4.68898690442679e-08,
      "loss": 0.0,
      "step": 1346
    },
    {
      "epoch": 5.637762237762238,
      "grad_norm": 0.0011570312781259418,
      "learning_rate": 4.575666751286734e-08,
      "loss": 0.0,
      "step": 1347
    },
    {
      "epoch": 5.641958041958042,
      "grad_norm": 0.001236400450579822,
      "learning_rate": 4.4637200739493514e-08,
      "loss": 0.0,
      "step": 1348
    },
    {
      "epoch": 5.6461538461538465,
      "grad_norm": 0.0014157688710838556,
      "learning_rate": 4.353147498905225e-08,
      "loss": 0.0,
      "step": 1349
    },
    {
      "epoch": 5.65034965034965,
      "grad_norm": 0.0011891762260347605,
      "learning_rate": 4.243949644955142e-08,
      "loss": 0.0,
      "step": 1350
    },
    {
      "epoch": 5.654545454545454,
      "grad_norm": 0.0014961018459871411,
      "learning_rate": 4.136127123206463e-08,
      "loss": 0.0,
      "step": 1351
    },
    {
      "epoch": 5.658741258741259,
      "grad_norm": 0.001079764449968934,
      "learning_rate": 4.0296805370696466e-08,
      "loss": 0.0,
      "step": 1352
    },
    {
      "epoch": 5.662937062937063,
      "grad_norm": 0.0010065160458907485,
      "learning_rate": 3.9246104822550366e-08,
      "loss": 0.0,
      "step": 1353
    },
    {
      "epoch": 5.667132867132867,
      "grad_norm": 0.001042189891450107,
      "learning_rate": 3.8209175467694424e-08,
      "loss": 0.0,
      "step": 1354
    },
    {
      "epoch": 5.671328671328672,
      "grad_norm": 0.0013273860095068812,
      "learning_rate": 3.7186023109128665e-08,
      "loss": 0.0,
      "step": 1355
    },
    {
      "epoch": 5.675524475524475,
      "grad_norm": 0.0012103270273655653,
      "learning_rate": 3.617665347275201e-08,
      "loss": 0.0,
      "step": 1356
    },
    {
      "epoch": 5.6797202797202795,
      "grad_norm": 0.005653926637023687,
      "learning_rate": 3.518107220733175e-08,
      "loss": 0.0,
      "step": 1357
    },
    {
      "epoch": 5.683916083916084,
      "grad_norm": 0.0015714728506281972,
      "learning_rate": 3.419928488446911e-08,
      "loss": 0.0,
      "step": 1358
    },
    {
      "epoch": 5.688111888111888,
      "grad_norm": 0.001443857909180224,
      "learning_rate": 3.323129699857208e-08,
      "loss": 0.0,
      "step": 1359
    },
    {
      "epoch": 5.6923076923076925,
      "grad_norm": 0.001631380058825016,
      "learning_rate": 3.227711396682015e-08,
      "loss": 0.0,
      "step": 1360
    },
    {
      "epoch": 5.696503496503497,
      "grad_norm": 0.0007834993302822113,
      "learning_rate": 3.1336741129138205e-08,
      "loss": 0.0,
      "step": 1361
    },
    {
      "epoch": 5.7006993006993,
      "grad_norm": 0.0018269442953169346,
      "learning_rate": 3.041018374816351e-08,
      "loss": 0.0,
      "step": 1362
    },
    {
      "epoch": 5.704895104895105,
      "grad_norm": 0.0029521749820560217,
      "learning_rate": 2.949744700921825e-08,
      "loss": 0.0,
      "step": 1363
    },
    {
      "epoch": 5.709090909090909,
      "grad_norm": 0.0009765252470970154,
      "learning_rate": 2.8598536020278678e-08,
      "loss": 0.0,
      "step": 1364
    },
    {
      "epoch": 5.713286713286713,
      "grad_norm": 0.001315471832640469,
      "learning_rate": 2.7713455811948787e-08,
      "loss": 0.0,
      "step": 1365
    },
    {
      "epoch": 5.717482517482518,
      "grad_norm": 0.0014739949256181717,
      "learning_rate": 2.6842211337430034e-08,
      "loss": 0.0,
      "step": 1366
    },
    {
      "epoch": 5.721678321678322,
      "grad_norm": 0.011172263883054256,
      "learning_rate": 2.5984807472494988e-08,
      "loss": 0.0,
      "step": 1367
    },
    {
      "epoch": 5.725874125874126,
      "grad_norm": 0.000764565949793905,
      "learning_rate": 2.5141249015459833e-08,
      "loss": 0.0,
      "step": 1368
    },
    {
      "epoch": 5.73006993006993,
      "grad_norm": 0.0014840587973594666,
      "learning_rate": 2.4311540687156354e-08,
      "loss": 0.0,
      "step": 1369
    },
    {
      "epoch": 5.734265734265734,
      "grad_norm": 0.0008991289068944752,
      "learning_rate": 2.349568713090722e-08,
      "loss": 0.0,
      "step": 1370
    },
    {
      "epoch": 5.7384615384615385,
      "grad_norm": 0.0009909298969432712,
      "learning_rate": 2.269369291249879e-08,
      "loss": 0.0,
      "step": 1371
    },
    {
      "epoch": 5.742657342657343,
      "grad_norm": 0.001297612558118999,
      "learning_rate": 2.1905562520156686e-08,
      "loss": 0.0,
      "step": 1372
    },
    {
      "epoch": 5.746853146853147,
      "grad_norm": 0.0012095985002815723,
      "learning_rate": 2.113130036451888e-08,
      "loss": 0.0,
      "step": 1373
    },
    {
      "epoch": 5.751048951048951,
      "grad_norm": 0.0010022992501035333,
      "learning_rate": 2.0370910778612917e-08,
      "loss": 0.0,
      "step": 1374
    },
    {
      "epoch": 5.755244755244755,
      "grad_norm": 0.0022376845590770245,
      "learning_rate": 1.962439801783067e-08,
      "loss": 0.0,
      "step": 1375
    },
    {
      "epoch": 5.759440559440559,
      "grad_norm": 0.0014023073017597198,
      "learning_rate": 1.8891766259904188e-08,
      "loss": 0.0,
      "step": 1376
    },
    {
      "epoch": 5.763636363636364,
      "grad_norm": 0.00174057693220675,
      "learning_rate": 1.8173019604884045e-08,
      "loss": 0.0,
      "step": 1377
    },
    {
      "epoch": 5.767832167832168,
      "grad_norm": 0.0021597202867269516,
      "learning_rate": 1.7468162075113816e-08,
      "loss": 0.0,
      "step": 1378
    },
    {
      "epoch": 5.772027972027972,
      "grad_norm": 0.0009314701310358942,
      "learning_rate": 1.6777197615209795e-08,
      "loss": 0.0,
      "step": 1379
    },
    {
      "epoch": 5.776223776223777,
      "grad_norm": 0.0011890660971403122,
      "learning_rate": 1.6100130092037704e-08,
      "loss": 0.0,
      "step": 1380
    },
    {
      "epoch": 5.78041958041958,
      "grad_norm": 0.0009535514982417226,
      "learning_rate": 1.5436963294691576e-08,
      "loss": 0.0,
      "step": 1381
    },
    {
      "epoch": 5.7846153846153845,
      "grad_norm": 0.0011502953711897135,
      "learning_rate": 1.4787700934472405e-08,
      "loss": 0.0,
      "step": 1382
    },
    {
      "epoch": 5.788811188811189,
      "grad_norm": 0.002636754186823964,
      "learning_rate": 1.4152346644867032e-08,
      "loss": 0.0,
      "step": 1383
    },
    {
      "epoch": 5.793006993006993,
      "grad_norm": 0.0017057901713997126,
      "learning_rate": 1.3530903981528454e-08,
      "loss": 0.0,
      "step": 1384
    },
    {
      "epoch": 5.7972027972027975,
      "grad_norm": 0.0013505290262401104,
      "learning_rate": 1.2923376422255551e-08,
      "loss": 0.0,
      "step": 1385
    },
    {
      "epoch": 5.801398601398601,
      "grad_norm": 0.0009337167139165103,
      "learning_rate": 1.2329767366973666e-08,
      "loss": 0.0,
      "step": 1386
    },
    {
      "epoch": 5.805594405594405,
      "grad_norm": 0.0009492154931649566,
      "learning_rate": 1.1750080137715447e-08,
      "loss": 0.0,
      "step": 1387
    },
    {
      "epoch": 5.80979020979021,
      "grad_norm": 0.0010318229906260967,
      "learning_rate": 1.118431797860281e-08,
      "loss": 0.0,
      "step": 1388
    },
    {
      "epoch": 5.813986013986014,
      "grad_norm": 0.0021571647375822067,
      "learning_rate": 1.0632484055827507e-08,
      "loss": 0.0,
      "step": 1389
    },
    {
      "epoch": 5.818181818181818,
      "grad_norm": 0.0012594949221238494,
      "learning_rate": 1.0094581457635033e-08,
      "loss": 0.0,
      "step": 1390
    },
    {
      "epoch": 5.822377622377623,
      "grad_norm": 0.002224020892754197,
      "learning_rate": 9.570613194306578e-09,
      "loss": 0.0,
      "step": 1391
    },
    {
      "epoch": 5.826573426573427,
      "grad_norm": 0.0010083214147016406,
      "learning_rate": 9.060582198141822e-09,
      "loss": 0.0,
      "step": 1392
    },
    {
      "epoch": 5.8307692307692305,
      "grad_norm": 0.0014112305361777544,
      "learning_rate": 8.564491323443113e-09,
      "loss": 0.0,
      "step": 1393
    },
    {
      "epoch": 5.834965034965035,
      "grad_norm": 0.0020075817592442036,
      "learning_rate": 8.082343346499099e-09,
      "loss": 0.0,
      "step": 1394
    },
    {
      "epoch": 5.839160839160839,
      "grad_norm": 0.0009844913147389889,
      "learning_rate": 7.61414096556945e-09,
      "loss": 0.0,
      "step": 1395
    },
    {
      "epoch": 5.8433566433566435,
      "grad_norm": 0.007462687324732542,
      "learning_rate": 7.159886800869875e-09,
      "loss": 0.0,
      "step": 1396
    },
    {
      "epoch": 5.847552447552448,
      "grad_norm": 0.003451945725828409,
      "learning_rate": 6.7195833945568616e-09,
      "loss": 0.0,
      "step": 1397
    },
    {
      "epoch": 5.851748251748251,
      "grad_norm": 0.0009963244665414095,
      "learning_rate": 6.293233210713789e-09,
      "loss": 0.0,
      "step": 1398
    },
    {
      "epoch": 5.855944055944056,
      "grad_norm": 0.0010306105250492692,
      "learning_rate": 5.880838635337893e-09,
      "loss": 0.0,
      "step": 1399
    },
    {
      "epoch": 5.86013986013986,
      "grad_norm": 0.001379998167976737,
      "learning_rate": 5.482401976325269e-09,
      "loss": 0.0,
      "step": 1400
    },
    {
      "epoch": 5.864335664335664,
      "grad_norm": 0.0010511106811463833,
      "learning_rate": 5.0979254634594964e-09,
      "loss": 0.0,
      "step": 1401
    },
    {
      "epoch": 5.868531468531469,
      "grad_norm": 0.0007628606399521232,
      "learning_rate": 4.727411248398317e-09,
      "loss": 0.0,
      "step": 1402
    },
    {
      "epoch": 5.872727272727273,
      "grad_norm": 0.001114350976422429,
      "learning_rate": 4.3708614046622525e-09,
      "loss": 0.0,
      "step": 1403
    },
    {
      "epoch": 5.876923076923077,
      "grad_norm": 0.004994195885956287,
      "learning_rate": 4.028277927621838e-09,
      "loss": 0.0,
      "step": 1404
    },
    {
      "epoch": 5.881118881118881,
      "grad_norm": 0.001528345514088869,
      "learning_rate": 3.6996627344879077e-09,
      "loss": 0.0,
      "step": 1405
    },
    {
      "epoch": 5.885314685314685,
      "grad_norm": 0.0007531394367106259,
      "learning_rate": 3.385017664299661e-09,
      "loss": 0.0,
      "step": 1406
    },
    {
      "epoch": 5.8895104895104895,
      "grad_norm": 0.004105644766241312,
      "learning_rate": 3.0843444779152243e-09,
      "loss": 0.0,
      "step": 1407
    },
    {
      "epoch": 5.893706293706294,
      "grad_norm": 0.000797791057266295,
      "learning_rate": 2.7976448580008252e-09,
      "loss": 0.0,
      "step": 1408
    },
    {
      "epoch": 5.897902097902098,
      "grad_norm": 0.000901344115845859,
      "learning_rate": 2.524920409023024e-09,
      "loss": 0.0,
      "step": 1409
    },
    {
      "epoch": 5.902097902097902,
      "grad_norm": 0.0015011667273938656,
      "learning_rate": 2.266172657237886e-09,
      "loss": 0.0,
      "step": 1410
    },
    {
      "epoch": 5.906293706293706,
      "grad_norm": 0.014325911179184914,
      "learning_rate": 2.0214030506834905e-09,
      "loss": 0.0,
      "step": 1411
    },
    {
      "epoch": 5.91048951048951,
      "grad_norm": 0.001023322925902903,
      "learning_rate": 1.7906129591713228e-09,
      "loss": 0.0,
      "step": 1412
    },
    {
      "epoch": 5.914685314685315,
      "grad_norm": 0.0012225008103996515,
      "learning_rate": 1.573803674279617e-09,
      "loss": 0.0,
      "step": 1413
    },
    {
      "epoch": 5.918881118881119,
      "grad_norm": 0.0013718855334445834,
      "learning_rate": 1.3709764093441935e-09,
      "loss": 0.0,
      "step": 1414
    },
    {
      "epoch": 5.923076923076923,
      "grad_norm": 0.0012083190958946943,
      "learning_rate": 1.1821322994542972e-09,
      "loss": 0.0,
      "step": 1415
    },
    {
      "epoch": 5.927272727272728,
      "grad_norm": 0.0006525355856865644,
      "learning_rate": 1.0072724014437153e-09,
      "loss": 0.0,
      "step": 1416
    },
    {
      "epoch": 5.931468531468531,
      "grad_norm": 0.0012734682532027364,
      "learning_rate": 8.463976938860585e-10,
      "loss": 0.0,
      "step": 1417
    },
    {
      "epoch": 5.9356643356643355,
      "grad_norm": 0.001441239845007658,
      "learning_rate": 6.995090770900437e-10,
      "loss": 0.0,
      "step": 1418
    },
    {
      "epoch": 5.93986013986014,
      "grad_norm": 0.0017710952088236809,
      "learning_rate": 5.666073730925536e-10,
      "loss": 0.0,
      "step": 1419
    },
    {
      "epoch": 5.944055944055944,
      "grad_norm": 0.0017363233491778374,
      "learning_rate": 4.476933256555849e-10,
      "loss": 0.0,
      "step": 1420
    },
    {
      "epoch": 5.9482517482517485,
      "grad_norm": 0.0013834848068654537,
      "learning_rate": 3.4276760026152877e-10,
      "loss": 0.0,
      "step": 1421
    },
    {
      "epoch": 5.952447552447552,
      "grad_norm": 0.0011294902069494128,
      "learning_rate": 2.518307841095635e-10,
      "loss": 0.0,
      "step": 1422
    },
    {
      "epoch": 5.956643356643356,
      "grad_norm": 0.0010773580288514495,
      "learning_rate": 1.7488338611232337e-10,
      "loss": 0.0,
      "step": 1423
    },
    {
      "epoch": 5.960839160839161,
      "grad_norm": 0.0027081803418695927,
      "learning_rate": 1.1192583689256797e-10,
      "loss": 0.0,
      "step": 1424
    },
    {
      "epoch": 5.965034965034965,
      "grad_norm": 0.0011895291972905397,
      "learning_rate": 6.295848878179467e-11,
      "loss": 0.0,
      "step": 1425
    },
    {
      "epoch": 5.969230769230769,
      "grad_norm": 0.001616847119294107,
      "learning_rate": 2.798161581718528e-11,
      "loss": 0.0,
      "step": 1426
    },
    {
      "epoch": 5.973426573426574,
      "grad_norm": 0.0007913967710919678,
      "learning_rate": 6.9954137416061715e-12,
      "loss": 0.0,
      "step": 1427
    },
    {
      "epoch": 5.977622377622378,
      "grad_norm": 0.0012839620467275381,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 1428
    }
  ],
  "logging_steps": 1,
  "max_steps": 1428,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 238,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.5964447470125056e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}